摘要:

Faraday Discuss. 1998 109 165»181 Neutralñneutral reactions at the temperatures of interstellar clouds and C3H6 Delphine Chastaing Philip L. James Ian R. Sims and Ian W. M. Smith School of Chemistry T he University of Birmingham Edgbaston Birmingham UK B15 2T T with O2 C2H2 C2H4 and C at temperatures 2H C2H2 radicals are generated by photolysis of A CRESU (Cineç tique de Reç action en Ecoulement Supersonique Uniforme) apparatus has been used to measure rate constants for the reactions of the ethynyl radical (C from 295 down to 15 K. C 2H) at 193 nm using an ArF excimer laser and reaction rates are determined by observing the chemiluminescence from CH(A 2*) which is generated in a minor channel of the reaction between C2H O2 . radicals and The rate constants for all four reactions increase as the temperature is lowered and those for reactions with the unsaturated hydrocarbons exceed 10~10 cm3 molecule~1 s~1 at all temperatures below 100 K.The results con–rm that C2H radicals like CN radicals react rapidly with unsaturated hydrocarbons at very low temperatures and in both cases the radical replaces an H atom in the unsaturated molecule. It therefore seems likely that combinations of these reactions play a major role in forming species with long C chains and in synthesising the cyanopolyynes in dense interstellar clouds. Rate coefficients for reactions of C2H O2 C2H2 C2H4 radicals with down to 15 K 3H6 Since 1992 through a collaboration between our group at the University of Birmingham and that of Bertrand Rowe in Rennes rate coefficients have been measured for almost 30 elementary reactions involving electrically neutral species at temperatures down to as low as 13 K.1h12 The experiments employ a CRESU (Cineç tique de Reç action en Ecoulement Supersonique Uniforme) apparatus in which the expansion of a gas mixture through a Laval nozzle provides a cold uniform and relatively dense —ow of gas with a temperature which is de–ned by the design of the convergent»divergent nozzle.Free radicals can be generated by pulsed laser photolysis (PLP) if a suitable precursor is included at low mole fraction in the —owing gas. In favourable cases the rate at which these radicals react can be observed using a second pulsed laser. It is tuned to a frequency absorbed by the radical so that laser-induced —uorescence (LIF) is generated with an intensity which is proportional to the concentration of the radical at the instant when the second laser is –red.Kinetic information is then obtained by scanning the time delay between the pulses from the photolysis and probe lasers. All the kinetic measurements on elementary reactions at ultra-low temperatures which have been reported so far have been performed on the CRESU apparatus at Rennes. Recently however a similar apparatus has been completed in Birmingham. It was –rst used to measure detailed rates of rovibronic energy transfer in collisions involving NO(X 2%) by an IR»UV double-resonance technique.13h15 The present paper reports the –rst measurements on elementary reactions undertaken in this CRESU apparatus.165 166 Neutral»neutral reactions at the temperatures of IS clouds The CRESU technique is currently the only one available which enables the rate coefficients of neutral»neutral reactions to be measured at the temperatures which are present in dense interstellar clouds. The demonstration that a number of reactions involving only neutral species are very rapid at these ultra-low temperatures has cast doubt16,17 on the previously held view that neutral»neutral reactions play at most a minor role in the synthesis of interstellar molecules. Nearly all the rate coefficients that have been reported so far are for reactions of the three free radicals CN OH and CH. Although all three species have been identi–ed in interstellar sources their reactions were selected as the –rst to be studied in a CRESU apparatus more for experimental reasons as they are ideally suited to the application of the PLP»LIF technique than for their perceived importance in the chemistry of interstellar clouds.In this paper we report the –rst kinetic measurements on reactions of the ethynyl radical C2H. Moreover in the General Discussion we report preliminary results on some reactions of carbon atoms in their electronic ground state C(3PJ). The reactions of these species with alkenes and alkynes have been targeted precisely because they might lead to carbon accretion in interstellar clouds ; that is to the generation of species containing long carbon chains. species with the longest carbon chains which have been unequivocally identi–ed in dense The higher members of the series of cyanopolyynes H(C)2nCN with nO5 are the interstellar clouds.18,19 Attempts have been made20h22 to explain their synthesis via routes involving ion»molecule reactions in what are sometimes referred to as standard models.Herbst and Leung,22 for example suggested three ion»molecule pathways that might contribute to the formation of the higher cyanopolyynes in interstellar clouds ; (i) routes involving reactions of smaller cyanopolyynes with ions such as C2H2 ` (ii) mechanisms in which complex hydrocarbon ions react with sources of nitrogen such as CN HCN and N atoms and (iii) reactions of organo-nitrogen ions with hydrocarbon neutrals. Although their calculations22 reproduced the variation in cyanopolyyne abundances with carbon chain length moderately well the absolute abundances were seriously underestimated.At the Faraday Symposium on Chemistry in the Interstellar Medium in December 1992 two of the present authors suggested23 that long-chain cyanopolyynes might be generated by sequences of neutral»neutral reactions involving CN and C2H radicals with polyynes and smaller cyanopolyynes; i.e. (1a) (1b) (2) CN]H(C) C2H]H(C)2nH]H(C)2n`2H]H C2H]H(C)2nCN]H(C)2n`2CN]H 2nH]H(C)2nCN]H rather than by the mechanisms involving ion»molecule reactions that had been proposed previously.20h22 The bases for this proposal were that (i) the reactions of both radicals with C2H2 were known to be rapid at room temperature (ii) the reaction of CN (iii) CN and C with C2H2 had been shown2 to occur at essentially the collision rate below 100 K and 2H are isoelectronic so they might be expected to exhibit similar kinetic behaviour.Based on some early results from studies of neutral»neutral reactions in a CRESU apparatus Herbst et al.16,17 investigated the predicted eÜects on molecular abundances of including fast neutral»neutral reactions in their models of interstellar clouds. The eÜects were dramatic not least on the abundances of the larger organic molecules which were sharply reduced. However it should be emphasised as it was by Herbst et al. that only a small fraction of the neutral»neutral reactions which they assumed to be fast had been shown to be so in laboratory experiments at very low temperatures.Subsequently Winstanley and Nejad24 have investigated the chemistry of cyanopolyynes in TMC-1 (Taurus molecular cloud 1) using time-dependent models and three diÜerent reaction 167 D. Chastaing et al. networks one of which included the rapid neutral»neutral chemistry introduced by Herbst and co-workers.16,17 In the context of the present paper the most important conclusions of Winstanley and Nejad24 were (a) that there are two major chemical regimes ion»molecule chemistry dominating the formation of cyanopolyynes at early times (O104 years) with cyanopolyynes being formed at longer times chie—y in the neutral»neutral reactions (1) and (2) suggested by Smith and Sims,23 and (b) that there was a real need for more kinetic data for neutral»neutral reactions at very low temperatures.In addition we note that CherchneÜ et al.25 have included reactions (1) and (2) in calculations designed to examine the formation of cyanopolyynes in the circumstellar envelope of IRC]10216. They conclude that in this environment the reactions of C3N C and radicals with 5N C2H2 were more likely to create the higher cyanopolyynes than sequences of reactions like those represented by eqn. (1) and (2). Our earlier proposal regarding the possible importance of reactions of C2H distributed over a number of vibrational levels especially those l bending mode as well as in the low-lying A2% Ideally in any kinetics experiments on reactions of C provided a stimulus for us to study the reactions of these radicals with unsaturated hydrocarbons.This paper reports the –rst such measurements at temperatures down to 15 K. The method adopted is that originally employed by Renlund et al.26 and exploited by others notably Peeters and co-workers,27h29 and most recently Thiesemann and Taatjes.30 The progress of reactions of C2H is followed by observing chemiluminescence from CH(A 2*]X2%) which is generated in a minor channel of the reaction between C2H radicals and O2 .28 As in most other kinetics work we have employed photodissociation of C2H2 with an ArF excimer laser at 193 nm to generate the ethynyl radicals. A potential problem in such experiments is that the photodissociation of C2H2 at this wavelength produces C associated with excitation of the 2H electronically excited state.This distribution has been characterised using forms of pho- 2 tofragment spectroscopy,31,32 observation of the IR emission from C2H by FTIR spectroscopy, 33 and measurements of the LIF spectrum of the C2H radicals.34h37 By observing time-resolved IR emission Shokoohi et al.38 determined the phenomenological rates of collisional quenching of both C H(A 2%) C and vibrationally excited by a 2 variety of collision partners but only at room temperature. 2H 2H conditions should be is relaxed rapidly employed such that any electronically or vibrationally excited C relative to the removal of the radicals by chemical reaction. In conventional cell experi- 2H ments like those of Peeters and co-workers27h29 and those in which the kinetic decays of C are followed by transient IR absorption,39h44 relatively high total pressures can be used and substantial concentrations of efficient relaxants such as SF can be included 2H in the gas mixture.In CRESU experiments the total gas density is more circumscribed 6 and additional relaxants can only be added in limited concentration. Consequently as one means of examining the eÜect of vibrational excitation of the C2H on our kinetic results some preliminary experiments have been performed in which we have observed the LIF spectrum of C2H and how it changes as the time delay between the pulses from the photolysis laser and the probe laser is altered. Experimental A description of the CRESU method as applied to neutral»neutral reactions has been given by Sims et al.1b Full speci–cations of the Birmingham CRESU apparatus have also been given recently.15 Here we concentrate on those features of our experiments which are speci–c to kinetic experiments on the reactions of C2H radicals.The heart of the CRESU apparatus is an axisymmetric Laval nozzle. It is mounted on a reservoir which is equipped with a perforated Te—on disc to ensure laminar —ow and good mixing of the gas streams entering the reservoir. Although the gas reservoir is 168 Neutral»neutral reactions at the temperatures of IS clouds Fig. 1 Schematic diagram of the CRESU apparatus at the University of Birmingham adapted for the study of C2H kinetics by a chemiluminescent marker method jacketed permitting cryogenic cooling,14,15 this facility was not made use of in the present experiments.All the temperatures in the gas —ows were achieved by isentropic expansion of the gas mixture prepared in the reservoir through the nozzle and into the main chamber. This expansion generates a supersonic —ow of gas in which the Mach number the temperature the density of the gas the mole fraction of the co-reagent and the velocity of the gas stream are constant along the axis of the —ow. using the 193 nm Several nozzles were employed in the present work each providing a particular temperature and density for the selected carrier gas. Experiments were carried out in He Ar and N All the nozzles were characterised by impact-pressure measurements.1b In addition the temperatures provided by some nozzles were con–rmed by recording the 2 .spectra of bands in the A2&`»X2% system of NO and matching these with simulated spectra.14b The temperatures determined from the relative intensities of the rotational lines agreed with the values inferred from the impact-pressure measurements. C2H C2H2 radicals were created by the pulsed laser photolysis of radiation from one of two similar excimer lasers (Lambda-Physik Compex 102 or Compex 202) operating on ArF. The beam from this laser entered the CRESU apparatus through a Brewster angle window passed through another such window mounted on the back of the reservoir and then propagated through the throat of the Laval nozzle and along the axis of the —ow before leaving the vacuum chamber via a third Brewsterangle window.In the kinetic experiments which are reported here the progress of reaction was monitored by observing (A 2*]X2%) chemiluminescence from electronically excited CH which is formed as a minor product of the reaction between C2H O and 2 .28 Because the radiative lifetime of CH in the A2* state is short (470 ns),45 its concentration rapidly reaches a steady-state value which is directly proportional both to the intensity of the emission and to the concentration of the C2H radical (see below for further discussion). The chemiluminescent emission from CH(A 2*) passed through a narrow band –lter (peak transmission at 428 nm; FWHM bandwidth of 10 nm) which transmitted *v\0 bands in the A2*»X2% system and the variation of emission intensity with time following the pulse from the photolysis laser was recorded using a photomultiplier tube.Signals from the photomultiplier were passed to a 200 MHz digital storage oscilloscope (Tektronix model TDS350) which averaged emission traces from 128 laser shots before transferring the resultant data to a PC. Typically this procedure was repeated 169 D. Chastaing et al. four times so that the traces which were analysed were usually the result of 512 individual experiments. In addition under some conditions it was necessary to perform a background subtraction to allow for the eÜects of scattered light. The background trace was recorded by omitting O from the gas mixture leaving all other conditions the same. Fig. 1 shows a schematic diagram of the CRESU»chemiluminescence apparatus.2 2H4 In experiments where LIF spectra were recorded the probe radiation was provided by a frequency-doubled dye laser (LAS LDL205) pumped by an Nd YAG laser (Spectron SL805) at 355 nm. Fluorescence was gathered by an optically fast telescopemirror combination mounted inside the main vacuum chamber. The same experimental arrangement was used for recording how the intensity of LIF signals varied with time delay between the photolysis and probe lasers. The signals were accumulated processed and analysed by the same procedures as before.14,15 The —ows of C C C2H2 O2 ( any additional co-reagent or and the carrier were taken directly from the cylinders (all from Air Products) and 3H6) gas (He Ar or N2) regulated by means of mass-—ow controllers (MKS).The C2H2 was passed through a trap immersed in cooled methanol at [40 °C to remove any residual acetone resulting from the storage method. Knowledge of the total gas density from Pitot-tube measurements and of the individual gas —ows enabled the concentrations of the minor gases in the supersonic —ows which are needed to determine the rate coefficients to be calculated. (3) radicals then reacted with both (4) C (5) C2H]C2H2 ]products 2H]O2 ]products (5a) 2H2 (i) rad Results Our –rst and most extensive experiments have been carried out on mixtures containing relatively small concentrations of both C2H2 and O diluted in the carrier gas (He Ar or N 2 2) for which the particular nozzle was designed. Ethynyl radicals were generated by the photolysis of a small fraction of the C2H2 by the 193 nm output from the excimer laser These C C2H2]hl(j\193 nm)]C2H]H 2H C2H2 and O2 Reaction (5) proceeds via a number of channels.Branching ratios are not completely established even at room temperature. However one minor channel produces CH(A 2*) C2H]O2 ]CH(A 2*)]CO2 for which Devriendt et al.28 have measured the rate coefficient to be k5a\(3.6^1.4) ]10~14 cm3 molecule~1 s~1 at 295 K corresponding to a fractional yield of CH(A 2*) of k5a/k5\1.1]10~3. We observed the intensity of the (A 2*»X2% *v\0) chemiluminescence as a function of time following photolysis of C typical traces. The kinetics of this emission in experiments of the present kind have been in the presence of O2 .Fig. 2 shows two discussed by Van Look and Peeters.27 In brief the intensity is proportional to the instantaneous concentration of CH(A 2*) which rapidly reaches a value determined by the steady-state equation k5a[C2H][O2]\krad[CH(A 2*)] because the mean radiative lifetime of CH(A 2*) i.e. 1/k is only 470 ns.45 When the concentrations of C and O are much greater than the concentration of C2H the 2 2H2 170 Neutral»neutral reactions at the temperatures of IS clouds 2H2]\3.2]1014 Fig. 2 Traces of chemiluminescent intensity vs. time without background subtraction –tted after an initial delay to single exponential decays (smooth curves). Experimental conditions (a) T \112 K [Ar]\2.79]1016 molecule cm~3 [C2H2]\3.7]1012 molecule cm~3 [O2]\4.3 ]1014 molecule cm~3; (b) T \25 K [He]\4.4]1016 molecule cm~3 [C molecule cm~3 [O2]\4.4]1014 molecule cm~3.latter can be written as (ii) [C2H]t\[C2H]0 exp([k1st t) where (iii) I k1st\k4[C2H2]]k5 [O2]. Hence cl(t)P[CH(A 2*)]t\Mk5a[O2]/kradN[C2H]0 exp([k1st t) Icl(t) decays exponentially with a pseudo-–rst-order and O2 . The traces of and the chemiluminescent intensity rate coefficient which depends on the concentrations of C chemiluminescent intensity shown in Fig. 2 like all others which we have recorded 2H2 exhibit this predicted behaviour. They were –tted using a non-linear least-squares algoat a particular temperature. rithm to yield values of k1st for each gas mixture on which experiments were carried out k and k5 series of experi- To establish values of the second-order rate coefficients ments were performed in which the concentration of either4C2H2 or O was varied whilst the other concentration was kept constant.The results of one experiment of each 2 D. Chastaing et al. second-order rate constant for the reaction of C Fig. 3 Pseudo-–rst-order decay constants k1st vs. reagent concentration [R] at T \25 K [He]\4.44]1016 molecule cm~3 and (a) [O2]\2.22]1014 molecule cm~3 R\C2H2 ; (b) [C2H2]\2.7]1014 molecule cm~3 R\O2 . In each case the gradient of the line yields the 2H with R. with kind are displayed in Fig. 3 where k is plotted against [C2H2] [ 2] withO constant 1st and against C [O2] [ with constant. The gradients of these lines yield the values of 2H2] k and k5 respectively.Our complete results for reactions (4) and (5) are summarised in Tables 1 and 2. The temperature dependence of the rate coefficients for these reactions of 4 C2H C2H2 and O are displayed in Fig. 4 and 5. 2 As mentioned earlier a problem with the present experimental method compounded 2H C2H2 was not observed directly is that photolysis of at 193 nm radicals with appreciable electronic and vibrational excitation. Moreover radical. by the fact that C produces C it is difficult to perform experiments in the CRESU apparatus in which relaxation of this 2H excitation is clearly much faster than the removal of C2H in reactions (4) and (5). However we have performed a number of measurements that taken together imply that the rates of reactions (4) and (5) are essentially independent of the degree of excitation in the C The –rst piece of evidence that the rates of reactions (4) and (5) are not materially 2H reagent comes the observation that the chemilumineat diÜerent rates at least at room temperature.38 Furthermore three series of aÜected by excitation in the C2H scent signals from CH(A 2*) accurately –t single exponentials in all three carrier gases although He Ar and N are known to quench electronically and vibrationally excited C 2 2H 171 172 Table 1 Rate coefficients for the reaction between C2H C2H2 radicals and at temperatures from 295 to 15 K [M]/1016 ]/1014 no.of [O k/10~10 cm3 [C 2 2H2]/ T /K M molecule cm~3 molecule cm~3 1014 molecule cm~3 points molecule~1 s~1 295 Ar 295 Ar 295 CO He Ar He Ar 149 112 63 54 39 39 N N2 2 He He He 25 25 25 He He He He He 15 15 15 15 15 a All errors quoted are ^tp statistical error where t is the appropriate value of Studentœs tdistribution for the 95% point.b Values in bold are weighted averages of a number of measurements at the same temperature. Table 2 Rate coefficients for the reaction between C2H O radicals and at temperatures from 295 2 [M]/1016 2H2]/1013 [C k/10~11 cm3 [O2]/ no. of T /K M molecule cm~3 molecule cm~3 1014 molecule cm~3 points molecule~1 s~1 295 112 54 Ar 149 He Ar 63 He Ar 39 N2 Ar 27 25 He 15 He 15 He 15 He 15 He 15 He a All errors quoted are ^tp statistical error where t is the appropriate value of Studentœs tdistribution for the 95% point.b Values in bold are weighted averages of a number of measurements at the same temperature. Neutral»neutral reactions at the temperatures of IS clouds 21 60 58 23.26 16.0 15.6 1.1 1.6 1.6 5.0 2.1 4.6 1.44 2.79 2.08 5.36 3.33 3.33 2.2 4.4 6.7 4.44 4.44 4.44 4.7 2.4 2.4 0.95 0.95 5.05 5.05 5.05 5.05 5.05 21 6.5 0.37 1.9 0.6 11 0.25 27 11.6 5.7 5.7 23.26 1.44 2.79 2.08 5.36 3.33 5.22 4.44 5.05 5.05 5.05 2.8 1.16 5.05 5.05 10 8 10 1.3»12.9 1.8»12 1.8»14 10 9 10 10 10 12 0.1»1.3 0.38»3.36 0.19»1.93 0.07»0.75 0.36»1.20 0.36»1.20 10 10 10 0.54»5.33 0.54»5.31 0.53»5.28 7 10 10 6 9 0.06»0.57 0.06»0.58 0.06»0.58 0.06»0.46 0.12»0.58 to 15 K 10 13 10 10 10 10 14 10 8 10 10 2.3»22.7 0.22»2.80 0.62»6.10 0.32»3.14 1.23»12.1 0.6»6 1.23»17.9 0.90»8.83 0.98»7.72 0.95»9.35 0.94»9.27 9 10 1.88»9.28 0.95»9.38 1.08^0.03a 0.94^0.08 1.05^0.16 1.06ª0.03b 1.05^0.06 1.01^0.02 1.27^0.04 1.76^0.02 2.54^0.42 2.00^0.20 2.10ª0.18 1.62^0.03 1.68^0.07 1.76^0.08 1.64ª0.03 2.34^1.11 2.12^0.44 2.37^0.85 2.31^0.51 2.69^1.04 2.27ª0.29 4.02^0.06a 2.89^0.16 2.80^0.10 3.88^0.39 5.43^0.12 7.18^0.58 7.56^0.26 7.68^0.41 7.60^0.58 11.41^0.63 10.01^0.26 10.97^0.20 10.38^2.51 10.49ª0.15b 173 D.Chastaing et al. Fig. 4 Rate coefficient k for the reaction of C2H C with plotted on a log»log scale against 4 temperature. The –lled symbols show the results of this work (Ö) Ar carrier gas; (=) N and (>) 2H2 2 He. The results of Peeters et al.27 using a comparable chemiluminescent technique are shown as (K) while the results of Pedersen et al.43a obtained by an IR laser absorption technique are shown as (»). In addition the combined cell and CRESU results of Sims et al.2 for the reaction of CN with C2H2 are shown as (L). experiments were performed at 25 K (see Table 1). In each series a diÜerent constant concentration of O was present and the concentration of C2H2 was varied.Although the O concentrations were changed by a factor of three the derived values of k were in 2 2 excellent agreement. 4 In addition we have performed a limited number of experiments in which CO has been added to the gas mixture to act as an efficient relaxant for vibrationally excited C2H. CO was chosen partly because it can be added in reasonably high concentrations in CRESU expansions without aÜecting the uniformity of the supersonic —ow and partly because it is likely to relax vibrationally excited C2H very eÜectively. The reason for this assertion is that C2H and CO undergo an association reaction which under usual laboratory conditions,41 proceeds via energised collision complexes that can be stabilised by 2 Fig.5 Rate coefficients k for the reaction of C2H O with plotted on a log»log scale against 5 temperature. The –lled symbols show the results of this work (Ö) Ar carrier gas; (=) N and (>) 2 He. The results of Peeters et al.27 using a comparable chemiluminescent technique are shown as (K) while the results of Opansky et al.44 obtained by an IR laser absorption technique are shown as (»). In addition the combined cell and CRESU results of Sims et al.1 for the reaction of CN with O are shown as (L). 2 T /K 295 112 54 39 39 27 27 15 15 15 a All errors quoted are ^tp statistical error where t is the appropriate value of Studentœs t-distribution for the 95% point. b Values in bold are weighted averages of a number of measurements at the same temperature.Table 3 2 Rate coefficients for the reactions between C2 radicals and H4 at H C temperatures from 295 to 15 K [M]/1016 molecule cm~3 M 22 Ar 2.79 Ar 5.36 Ar N2 N2 3.33 3.33 4.65 4.65 He He 5.05 5.05 5.05 He He He ]/1014 [O2 molecule cm~3 [C2H2]/1013 molecule cm~3 11 26 3.1 3.7 6.2 1.5 3.0 1.5 3.6 1.8 4.7 2.34 5.6 3.9 2.1 1.9 9.8 3.6 2.3 1.2 [C2H4]/ 1014 molecule cm~3 6.3»49.8 0.7»7.1 0.36»5.7 0.35»3.5 0.35»1.7 1.1»10.7 0.53»5.3 0.55»5.5 0.55»5.5 0.55»5.5 Neutral»neutral reactions at the temperatures of IS clouds k/10~10 cm3 molecule~1 s~1 no.of points 0.99^0.05a 8 1.59^0.004 10 2.30^0.015 9 1.71^0.58 1.76^1.21 1.72ª0.52b 10 9 2.32^0.07 2.31^0.22 2.32ª0.07 10 10 2.10^0.39 1.83^0.28 2.56^0.28 10 10 10 2.17ª0.17 174 T /K 295 112 54 39 39 27 27 15 15 15 a All errors quoted are ^tp statistical error where t is the appropriate value of Studentœs t-distribution for the 95% point. b Values in bold are weighted averages of a number of measurements at the same temperature. Table 4 3 Rate coefficients for the reactions between C2 radicals and H6 at H C temperatures from 295 to 15 K [M]/1016 molecule cm~3 M 22 Ar 2.79 Ar 5.36 Ar N2 N2 3.33 3.33 4.65 4.65 He He 5.05 5.05 5.05 He He He ]/1014 [O2 molecule cm~3 [C2H2]/1013 molecule cm~3 19.6 23 3.1 3.7 2.5 3.7 3.1 4.6 3.7 3.6 4.7 2.34 5.6 3.9 2.4 2.1 2.1 3.0 3.6 3.6 [C3H6]/ 1014 molecule cm~3 0.77»12.3 0.5»4.5 1.0»10.0 0.24»0.98 0.5»8.78 0.75»7.5 0.37»3.73 0.4»2.34 1.33»27.2 1.33»9.72 175 D.Chastaing et al. k/10~10 cm3 molecule~1 s~1 no. of points 1.84^0.03a 10 2.35^0.09 10 3.39^0.13 10 7.09^4.43 8.04^2.48 7.81ª2.16b 6 15 2.15^0.14 3.01^0.24 2.34ª0.12 10 10 1.36^0.45 0.81^0.47 2.00^1.52 9 18 12 1.14ª0.32 176 Neutral»neutral reactions at the temperatures of IS clouds k Fig. 6 Rate coefficients plotted on a log»log scale against for the reaction of C2H C2H4 with temperature.The –lled symbols show the results of this work (Ö) Ar carrier gas; (=) N and (>) 6a 2 He. The results of Opansky and Leone43b obtained by an IR laser absorption technique are shown as (»). In addition the combined cell and CRESU results of Sims et al.2 for the reaction of CN with C2H4 are shown as open circles (L). resulting from its formation. collision with a third-body gas. There are numerous examples of how the formation of such complexes can lead to very eÜective vibrational relaxation.46 We have carried out experiments on the kinetics of reaction (4) at 295 K with CO in place of one of the usual diluent gases (see Table 1). The result of the experiment was unaÜected by this change. Since the concentration of CO was 1.56]1017 molecule cm~3 and the rate constant for relaxation by CO is likely by analogy with similar systems,46 to exceed 10~11 cm3 molecule~1 s~1 the rate of relaxation should exceed that of reaction by roughly two orders of magnitude in these experiments.However the derived rate constant was the same within experimental error as those determined in the presence of argon. Our conclusion is that the rate constant for reaction (4) is unchanged by the internal excitation of the C The reactions of C 2H 2H C2H4 with and C3H6 have been investigated in series of experiments in which diÜerent concentrations of these additional reagents were added to mixtures containing constant concentrations of C2H2 and O2 . Under these circumstances decay of the chemiluminescent signal from CH(A 2*) continues to decay exponentially but now the pseudo-–rst-order rate constant is given by (iv) k1st\k4[C2H2]]k5[O2]]k6[alkene] where [alkene] is the concentration of C2H4 or C3H6 investigation and the rate constant k is for one or other of the reactions present in the mixture under (6a) 6 C2H]C2H4 ]products (6b) C2H]C3H6 ]products The values of the second-order rate coefficients for these reactions could be determined k from the gradient of plots of vs and .[alkene] in experiments where [C2H2] [O 1st were kept constant. These results are given in Tables 3 and 4 and the dependence of the 2] rate coefficients on temperature is shown in Fig. 6 and 7. Although experiments were performed on the reaction of C2H C with at 39 K in N2 rate coefficients have not been quoted as the results appeared to show evidence of being aÜected by the formation 3H6 of van der Waals complexes.177 D. Chastaing et al. 2 k Fig. 7 Rate coefficients plotted on a log»log scale against for the reaction of C2H C3H6 with temperature. The –lled symbols show the results of this work (Ö) Ar carrier gas; (=) N and (>) 6b He. The cell results of Sims et al.2 for the reaction of CN with C3H6 are shown as (L). 2 ref. methoda k5/10~11 cm3 molecule~1 s~1 3.1 FP-UVAbs(C2H4) 47(a) 47(b) PLP-CL(CH*) 5 2 3.2 1.5 26(a) 39 30 4.2 2.9 PLP-CL(CH*) PLP-IRAbs(C PLP-IRAbs(C2H) 2H) 41 48 2H4) 49 3.3 1.3 1.5 1.3 PLP-UVAbs(H) FP-UVAbs(C PLP-IRAbs(C PLP-IRAbs(C2H) PLP-CL(CH*) 2H) 43(a) 44 42 27 this work PLP-CL(CH*) 3.3 4.02ª0.06 Discussion Rate coefficients for the reactions of ethynyl radicals with C2H2 and O have been measured several times before at room temperature by a variety of methods.Table 5 2 compares these previously measured values of k and k with the rate coefficients that 4 we have determined in the present work at room temperature using the pulsed laser 5 photolysis (PLP) chemiluminescence (CL) technique. The good agreement that is found increases our con–dence in the application of this PLP»CL method at low temperatures. In addition in Fig. 4 and 5 we show the temperature-dependent results of Leone and co-workers,43a,44 and Van Look and Peeters27 for these two reactions.Until the present measurement there were no data for either of these reactions below 170 K. The magnitudes of the rate coefficients for the reactions of C2H C with 2H2 C2H4 and C3H6 their lack of any dependence on pressure and their increase as the tem- Table 5 Comparison of rate coefficients for the reactions of C2H C2H2 with and O from the present work with those from previous measurements k4/10~10 cm3 molecule~1 s~1 1.9 1.3 1.08ª0.03 a FP » —ash photolysis ; UVAbs(X) » ultraviolet absorption on the speci–ed (product) species ; PLP » pulsed laser photolysis ; IRAbs(C2H) » IR absorption on C2H using a colour centre laser ; CL(CH*) » chemiluminescence from CH(A 2*) as in the present work. 178 Neutral»neutral reactions at the temperatures of IS clouds perature is lowered as well as thermochemical considerations all point to these reactions proceeding via initial formation of an energised complex when the radical attacks a p-orbital in the unsaturated hydrocarbon.The energised complex then rapidly loses a H atom so that the overall reactions can be represented as CxHy]C2H]Cx`2Hy]H 2H]C2H2 C Laufer and co-workers47 observed 4H2 (7) This proposal parallels that made2 for the reactions of CN radicals with unsaturated hydrocarbons where again the overwhelming evidence is that the radical displaces a H atom in the molecular reagent. In con–rmation of this mechanism represented by eqn. (7) we note that in studies of C and Shin and Michael48 monitored H atoms in both cases by UV absorption.The kinetics of the reaction between C2H O and again suggest that reaction proceeds via the initial formation of a transient energised complex in this case a substituted 2 peroxyl radical over a potential-energy surface on which there is no barrier between reagents and complex. This is undoubtedly the case for the isoelectronic reaction between CN and O2 .50,51 The major products of that reaction are NCO]O although there is evidence for the existence of a minor channel leading to CO]NO.51 In the reaction between C2H O2 and a number of pathways are possible on thermochemical grounds.44 Although several products have been identi–ed in previous studies the branching ratios into diÜerent channels have not been conclusively established.As our results shed no light on this difficult problem we do not discuss it further. If all the reactions which we have studied do indeed occur via transient addition complexes which are accessed from the reagents by motion across a monotonically attractive potential-energy surface then the rates will be determined by ìcaptureœ on the long-range potential acting between the reagents.52,53 If this is the case then the same reactions involving vibrationally excited species will occur across vibrationally adiabatic potentials which are parallel at long range to that correlating with reagents in their ground vibrational states. Consequently one should expect46,54 that the rate of capture and hence of reaction will be essentially independent of the vibrational state of the C radical in agreement with the experimental evidence mentioned earlier.2H As stated in the Introduction a major motivation for the present work was our earlier proposal23 that the kinetics of reactions of C2H should be similar to those of CN and therefore that a number of such reactions should contribute to the chemistry of cold interstellar clouds where C2H and CN are among the most abundant free radicals. In Fig. 4»7 we compare the rate coefficients obtained in the present work with those measured previously2 for the corresponding reactions of CN. It can be seen clearly that our latest data con–rm the earlier hypothesis. For reactions with the unsaturated hydrocarbons the rates of the ethynyl radicals are somewhat lower than those of CN.This diÜerence may re—ect either the slight increase in structural complexity in going from CN to C2H C2H or the lower dipole moment of which will mean that the long-range potential between it and a given molecule will be somewhat less strongly attractive than that between CN and the same molecule. The –nding that C2H reacts rapidly with simple unsaturated hydrocarbons lends strong support to the proposal that reactions like (1) and (2) could play an important role in the chemistry of cold interstellar clouds especially in the synthesis of cyanopolyynes. The ability of free radicals to add to unsaturated hydrocarbons has been discussed in the past55 in terms of a high electron affinity of the radical and a low ionisation energy of the molecule facilitating reaction.As longer chain alkynes have lower ionisation energies than C2H2 ,56 it seems reasonable to suppose that their reactions with C2H will also be rapid at ultra-low temperatures. Experimental con–rmation of this suggestion will be one future aim of work in our laboratory. at comparably rapid rates with unsaturated hydrocarbons it is possible to re-examine Now that our experiments have demonstrated that CN and C2H radicals both react the neutral»neutral reactions which might impact on the concentrations of cyano-179 polyynes in interstellar clouds. In addition to the formation of H(C)2nCN by reaction (2) CN]H(C)2nH]H(C)2nCN]H it is reasonable to suppose that this molecule can also be formed by and lost by and or C2H]H(C)2nCN]H(C)2n`2CN]H C radicals are at an abundance of 8]10~8 relative to total hydrogen,22,57 one of the most common molecular species observed in the cold dark cloud TMC-1.By coin- 2H cidence the abundance of CN (3]10~8)22,57 is ca. three times smaller so that the pseudo-–rst-order rate constants for the C2H]C2H2 reaction i.e. k4[C2H] and that CN]C for i.e. k2(n/1)[CN] are almost exactly the same and it seems likely that this relationship will hold approximately for higher alkynes. Observations of the same 2H2 source show that the abundances along the sequence HC CN HC CN HC6CN and 2 4 HC CN decrease by a factor of 2»3.5 at each step along the sequence.22 Unfortunately there is no observational information about the abundances of the non-polar molecules 8 H(C)2nH NC(C)2nCN.Winstanley and Nejad24 have suggested that the sequence of reactions starting with C2H]HCN]HC2CN]H followed by reactions of type (10) might also provide a route to the synthesis of cyanopolyynes. As far as we aware there is no experimental information available about the kinetics of the reactions of CN and C2H H2CN radicals withC or with cyanopolyynes. However recent ab initio calculations by Fukuzawa and Osamura16 indicate that there is a signi–cant barrier to the addition of C2H to HCN whereas calculations at the same level of theory con–rm the absence of any barrier in the reaction of CN radicals with C2H2 . The –nding of a barrier for reaction (11) is consistent with the kinetic information for the isoelectronic reaction The rate constant for reaction (12) at 300K is only k12\3.3]10~14 cm3 molecule~1 s~1 and the temperature dependence of k12 between 300 and 740 K is consistent with an Ea\13.5 kJ mol~1 (i.e.Ea/R\1625 activation energy of K).58,59 Hence the evidence is that neither reaction (11) nor reaction (12) can occur at signi–cant rates at the temperatures encountered in interstellar clouds. It therefore seems likely that although CN and C2H radicals can react rapidly at ultra-low temperatures with alkenes and alkynes by initially adding to a double or triple bond addition to a CN bond is an activated process and will not occur at signi–cant rates at the temperatures of interstellar clouds. This conclusion is consistent with the idea that such additions are promoted by electron withdrawal by the approaching radical and will be easier the lower the ionisation energy of the unsaturated molecule.The ionisation energy of HCN is 13.8 eV compared with 11.4 eV for C Summary and conclusions We have measured the rate coefficients for reactions of the ethynyl radical with O and C C 2 2H2 C2H4 3H6 coefficients have been reported for any of these reactions at temperatures below 150 K. at temperatures as low as 15 K. This is the –rst time that rate D. Chastaing et al. (2) (8) C2H]H(C)2n~2CN]H(C)2nCN]H (9) CN]H(C)2nCN]NC(C)2nCN]H (10) (11) (12) CN]HCN]NCCN]H 2H2 .60 180 As the C Neutral»neutral reactions at the temperatures of IS clouds All these reactions are rapid at these low temperatures with rate coefficients for those of C with the unsaturated hydrocarbons in excess of 10~10 cm3 molecule~1 s~1 in general a factor of 2»3 smaller than those for the corresponding reactions of the isoelec- 2H tronic CN radical.2H radical is relatively abundant in cold dark interstellar clouds the present results suggest that the reactions of this radical as well as those of CN should be included in chemical models of these regions of interstellar space. In particular and in keeping with an earlier proposal,23 it seems that combinations of reaction of CN and C2H with alkynes and cyanoalkynes probably play a major role in the formation and reactions of cyanopolyynes. We hope that the data reported here will enable such reactions to be incorporated into future models of interstellar clouds with values of the rate coefficients in which one can have some con–dence.Studies of the reactions of C2H radicals are continuing in our laboratory. We are grateful to EPSRC for a substantial research grant to construct the CRESU apparatus and for an Advanced Fellowship to one of us (I.R.S.). We also thank EPSRC and the EU (under its TMR programme) for studentships (P.L.J. and D.C. respectively). The EPSRC Laser Loan Pool at the Rutherford»Appleton laboratory provided one of the excimer lasers for which we express our thanks. We are also very grateful to Dr Bertrand Rowe and his colleagues at Rennes for valuable advice and discussions and to Stuart Arkless Mark Cheshire and Steve West for their skilled technical assistance. References 1 (a) I.R. Sims J-L. QueÜelec A. Defrance C. Rebrion-Rowe D. Travers B. R. Rowe and I. W. M. Smith J. Chem. Phys. 1992 97 8798; (b) I. R. Sims J-L. QueÜelec A. Defrance C. Rebrion-Rowe D. Travers P. Bocherel B. R. Rowe and I. W. M. Smith J. Chem. Phys. 1994 100 4229. 2 I. R. Sims J-L. QueÜelec A. Defrance D. Travers B. R. Rowe L. Herbert J. Karthaé user and I. W. M. Smith Chem. Phys. L ett. 1993 211 461. 3 I. R. Sims P. Bocherel A. Defrance D. Travers B. R. Rowe and I. W. M. Smith J. Chem. Soc. Faraday T rans. 1994 90 1473. 4 I. R. Sims I. W. M. Smith D. C. Clary P. Bocherel and B. R. Rowe J. Chem. Phys. 1994 101 1748. 5 P. Sharkey I. R. Sims I. W. M. Smith P. Bocherel and B. R. Rowe J. Chem. Soc. Faraday T rans. 1994 90 3609. 6 I. R. Sims and I. W. M.Smith Annu. Rev. Phys. Chem. 1995 46 109. 7 P. Bocherel L. B. Herbert B. R. Rowe I. R. Sims I. W. M. Smith and D. Travers J. Phys. Chem. 1996 8 L. Herbert I. R. Sims I. W. M. Smith D. W. A. Stewart A. Canosa and B. R. Rowe J. Phys. Chem. 9 R. A. Brownsword A. Canosa B. R. Rowe I. R. Sims I. W. M. Smith D. W. A. Stewart A. C. 100 3063. 1996 100 14928. Symonds and D. Travers J. Chem. Phys. 1997 106 7662. 10 I. W. M. Smith IAU Symposium 178 Molecules in Astrophysics Probes and Processes Kluwer Dordrecht 1997 p. 253. 11 A. Canosa I. R. Sims D. Travers I. W. M. Smith and B. R. Rowe Astron. Astrophys. 1997 323 644. 12 S. O. Le Picard A. Canosa D. Travers D. Chastaing B. R. Rowe and T. Stoecklin J. Phys. Chem. A 1997 101 9988. 13 P. L. James I. R. Sims and I.W. M. Smith Chem. Phys. L ett. 1997 272 412. 14 P. L. James I. R. Sims and I. W. M. Smith Chem. Phys. L ett. 1997 276 423. 15 P. L. James I. R. Sims I. W. M. Smith M. H. Alexander and M. Yang J. Chem. Phys. 1998 in press. 16 E. Herbst H-H. Lee D. A. Howe and T. J. Millar Mon. Not. R. Astronom. Soc. 1994 268 335. 17 R. P. A. Bettens H-H. Lee and E. Herbst Astrophys. J. 1995 443 664. 18 E. Herbst Annu. Rev. Phys. Chem. 1995 46 27. 19 M. B. Bell P. A. Feldman M. J. Travers M. C. McCarthy C. A. Gottleib and P. Thaddeus Astrophys. J. 1997 483 L61. 20 G. Winnewisser and E. Herbst T op. Curr. Chem. 1987 139 119. 21 D. K. Bohme S. Wlodek and A. B. Raksit Can. J. Chem. 1987 65 2057. 22 E. Herbst and C. M. Leung Astrophys. J. Suppl. 1989 69 271. 23 I. W.M. Smith and I. R. Sims J. Chem. Soc. Faraday T rans. 1993 89 2166. 24 N. Winstanley and L. A. M. Nejad Astrophys. Space Sci. 1996 240 13. 25 I. CherchneÜ A. E. Glassgold and G. A. Mamon Astrophys. J. 1993 410 188. 181 D. Chastaing et al. 26 (a) A. M. Renlund F. Shokoohi H. Reisler and C. Wittig Chem. Phys. L ett. 1981 84 293; (b) A. M. Renlund F. Shokoohi H. Reisler and C. Wittig J. Phys. Chem. 1982 86 4165. 27 H. Van Look and J. Peeters J. Phys. Chem. 1995 99 16284. 28 K. Devriendt H. Van Look B. Ceursters and J. Peeters Chem. Phys. L ett. 1996 261 450. 29 (a) J. Peeters H. Van Look and B. Ceursters J. Phys. Chem. 1996 100 15124; (b) K. Devriendt and J. Peeters J. Phys. Chem. A 1997 101 2546. 30 H. Thiesemann and C. A. Taatjes Chem. Phys. L ett. 1997 270 580.31 (a) A. M. Wodtke and Y. T. Lee J. Phys. Chem. 1985 89 4744; (b) B. A. Balko J. Zhang and Y. T. Lee J. Chem. Phys. 1991 94 7958. 32 S. H. S. Wilson C. L. Reed D. H. Mordaunt M. N. R. Ashfold and M. Kawasaki Bull. Chem. Soc. Jpn. 1996 69 71. 33 T. R. Fletcher and S. R. Leone J. Chem. Phys. 1989 90 871. 34 Y-C. Hsu J. J-M. Lin D. Papousek and J-J. Tsai J. Chem. Phys. 1993 98 6691. 35 Y-C. Hsu P. R. Wang M-C. Yang D. Papousek Y-T. Chen and W-Y. Chiang Chem. Phys. L ett. 1992 190 507. 36 Y-C. Hsu Y-J. Shiu and C. M. Lin J. Chem. Phys. 1995 103 5919. 37 Y-C. Hsu F-T. Chen L-C. Chou and Y-J. Shiu J. Chem. Phys. 1996 105 9153. 38 F. Shokoohi T. A. Watson H. Reisler F. Kong A. M. Renlund and C. Wittig J. Phys. Chem. 1986 90 5695. 39 J. W. Stephens J.L. Hall H. Solka W-B. Yan R. F. Curl and G. P. Glass J. Phys. Chem. 1987 91 5740. 40 D. R. Lander K. G. Unfried J. W. Stephens G. P. Glass and R. F. Curl J. Phys. Chem. 1989 93 4109. 41 D. R. Lander K. G. Unfried G. P. Glass and R. F. Curl J. Phys. Chem. 1990 94 7759. 42 S. K. Farhat C. L. Morter and G. P. Glass J. Phys. Chem. 1993 97 12789. 43 (a) J. O. P. Pedersen B. J. Opansky and S. R. Leone J. Phys. Chem. 1993 97 6822; (b) B. J. Opansky and S. R. Leone J. Phys. Chem. 1996 100 19904. 44 B. J. Opansky P. W. Seakins J. O. P. Pedersen and S. R. Leone J. Phys. Chem. 1993 97 8583. 45 J. E. Butler J. W. Fleming L. P. Goss and M. C. Lin Chem. Phys. 1981 56 355. 46 I. W. M. Smith J. Chem. Soc. Faraday T rans. 1997 93 3741. 47 (a) A. H. Laufer and A. M. Bass J. Phys. Chem. 1979 83 310; (b) A. H. Laufer and R. Lechleider J. Phys. Chem. 1984 88 66. 48 K. S. Shin and J. V. Michael J. Phys. Chem. 1991 95 5864. 49 M. Koshi K. Fukada K. Kamiya and H. Matsui J. Phys. Chem. 1992 96 9839. 50 S. J. Klippenstein and Y-W. Kim J. Chem. Phys. 1993 99 5790. 51 I. W. M. Smith in T he Chemical Kinetics and Dynamics of Small Radicals ed. K. Liu and A. Wagner World Scienti–c Singapore 1995 p. 214. 52 D. C. Clary Annu. Rev. Phys. Chem. 1990 41 61. 53 I. W. M. Smith Int. J. Mass Spectrom. Ion Proc. 1995 149/150 231. 54 I. W. M. Smith Berichte Bunsen-Ges Phys. Chem. 1997 101 516. 55 E. Martinez B. Caban8 as A. Aranda J. Albaladijo and R. P. Wayne J. Chem. Soc. Faraday T rans. 56 S. G. Lias J. E. Bartmess J. F. Liebmann J. L. Holmes R. D. Levin and W. G. Mallard J. Phys. Chem. 1997 93 2043 and references therein. Ref. Data 1988 17 Suppl. 1. 57 W. M. Irvine P. F. Goldsmith and A. Hjalmarson in Interstellar Processes ed. D. J. Hollenbach and H. A. Thronson Jr. Reidel Dordrecht 1987 p. 561. 58 D. L. Yang T. Yu M. C. Lin and C. F. Melius J. Chem. Phys. 1992 97 222. 59 D. L. Yang and M. C. Lin in ref. 51 p. 164. 60 Handbook of Chemistry and Physics ed. D. R. Lide CRC Press Boca Raton FL 71st edn. 1990. Paper 8/00495A; Received 19th January 1998

ISSN:1359-6640    

DOI:10.1039/a800495a

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Faraday Discuss. 1998 109 183»204 Combined crossed molecular beams and ab initio investigation of the formation of carbon-bearing molecules in the interstellar medium via neutralñneutral reactions R. I. Kaiser,a§ C. Ochsenfeld,bî D. Stranges,c° M. Head-Gordon,b“ Y. T. Leed** a Academia Sinica Institute of Atomic and Molecular Sciences 1 Section 4 Roosevelt Rd.,T aipei 116 T aiwan ROC and Department of Physics T echnical University Chemnitz-Zwickau 09107 Chemnitz Germany b Department of Chemistry University of California Berkeley California 94720 USA and Chemical Sciences Division L awrence Berkeley National L aboratory Berkeley California 94720 USA c Dipartimento di Chimica (NEC) Universita L a Sapienza Piazzale A. Moro 5 00185 Rome Italy and Centro Stadi T ermodinamica Chimica Alte T emperature CNR 00185 Rome Italy d Academia Sinica Institute of Atomic and Molecular Sciences 1 Section 4 Roosevelt Rd.T aipei 116 T aiwan ROC The crossed molecular beams reaction of atomic carbon C(3P with hydrogen sul–de H j) 2S allene H CCCH2 C2H3 the vinyl radical and deuteroacetylene C HD have been studied at diÜerent collision energies up to 42.2 2 2 kJ mol~1 and combined with high level ab initio calculations. All reactions are barrier-less and are dominated by a carbon»hydrogen exchange to form thioformyl (HCS) butatrienyl (HCCCCH2) C3H2 isomer(s) and deuteriated tricarbon hydride(s). This carbon»hydrogen replacement channel represents a one-step alternative reaction pathway to competing ion»molecule reactions to form complex carbon-bearing molecules in the interstellar medium as well as in the out—ow of carbon stars.183 1 Introduction One of the most fundamental questions in astrochemistry is the formation of molecules in the interstellar medium (ISM). To tackle this still unresolved puzzle we have –rst to familiarize ourselves with the physical conditions and the distribution of matter in distinct interstellar environments before we can elucidate well de–ned mechanisms to synthesize interstellar molecules and radicals. The ISM contains ca. 10% of the mass of our galaxy and consists of gas (99%) and 0.1»0.2 lm ellipsoidal-shaped grain particles (1%) with averaged number densities of 1 H atom cm~3 and 10~11 grains cm~3 respectively.1 Its chemical composition is dominated by hydrogen and helium (93.38% H 6.49% He) whereas biogenic elements oxygen carbon and nitrogen contribute 0.11% (O C NB8 3 1).2 All remaining elements furnish only 0.02%.Although the interstellar dust component embodies only 1% these predominantly silicate- and carbonaceous-based grain nuclei play a key role in the § E-mail kaiser=po.iams.sinica.edu.tw î E-mail chi=alcatraz.cchem.berkeley.edu ° E-mail dstranges=axrma.uniromal.it “ E-mail mhg=alcatraz.cchem.berkeley.edu ** E-mail ytlee=gate.sinica.edu.tw 184 Formation of carbon-bearing molecules in the ISM samples with MeV particles leads to a broad product spectrum of 4 formation of new molecules. Deep in the interior of dense clouds grain particles eÜectively shield newly synthesized molecules in the gas phase from the destructive external galactic UV radiation –eld.In addition these grains hold typical temperatures of 10 K in dark clouds.3 Molecules radicals and atoms from the gas phase are accreted on grain surfaces resulting in an icy mantle up to 0.1 lm thick. Here solid mixtures containing H OCS OCN~ 2O CO CH OH NH3 H2S CH4 H2CO H2 CO and have been 3 2 identi–ed unambiguously via IR spectroscopy.4 This ice mantle is altered chemically by the cosmic-ray-induced internal UV radiation present even in the deep interior of dense clouds5 and through charged particles such as protons (p H`) and helium nuclei (aparticles He2`)6 of the galactic cosmic radiation –eld. This combined photon and particle bombardment leads to the synthesis of new molecules in the solid state.7 Since typical carbon»hydrogen and carbon»carbon bond strengths in organic molecules range between 3 and 10 eV cosmic ray particles are too energetic to form stable chemical bonds as implanted into the icy mantle.However upon interaction with the solid target each cosmic ray particle releases its excess energy to the target atoms in successive collisions via elastic and inelastic interactions.8 Here the elastic process transfers energy to the nuclei of the target atoms igniting primary knock-on particles (PKOs; –rst generation of knock-on particles) if this amount is larger than the binding energy of the atom. MeV a-particles for example generate carbon PKOs with kinetic energies up to 10 keV. These knock-on particles can transfer their energy in consecutive encounters to the target atoms resulting in a collision cascade of secondary tertiary etc.knock-on atoms. Moderated to ca. 1»10 eV the so-called chemical energy range these atoms are not in thermal equilibrium with the 10 K target (hence non-equilibrium or suprathermal particles) and react with the target molecules via elementary steps of bond insertion addition to double or triple bonds and hydrogen abstraction.9 Irradiating solid CH C2H2 C2H4 and synthesized species such as atomic and molecular hydrogen H and H2 CHn (n\1»4) (n\1»6) C C (n\4»8) larger alkanes and alkenes with up to 18 carbon atoms 2Hn 3Hn as well as polycyclic aromatic hydrocarbons (PAHs) up to coronene.10 The power of these suprathermal reactions to form new molecules at temperatures even as low as 10 K is based in their ability to overcome reaction barriers in the entrance channel easily since suprathermal species can impart their excess kinetic energy into the transition state of the reaction.Even endothermic reactions are feasible if the energy de–cit can be covered by a suprathermal reactant extending the synthetic power of this reaction class beyond thermal reactions and diÜusion controlled chemistry on interstellar grains. These unique aspects of suprathermal reactions result in reaction rate constants up to 16 orders of magnitude larger than their thermal counterparts even at temperatures as low as 10 K.11 Once molecules are formed on interstellar grains a consecutive grain heating by a young stellar object embedded inside dense clouds followed by equilibrium sublimation as well as explosive molecule ejections from grains storing a critical concentration of radicals can redistribute these molecules into the gas phase.12 Besides the solid state molecules can also be synthesized in the gas phase of the ISM.Table 1 lists those species detected so far most of them thermally unstable and extremely reactive in terrestrial laboratories.13 On average 97% of all species exist as neutrals whereas only 3% are ions. These molecules radicals and ions are not distributed homogeneously but are con–ned to distinct interstellar environments such as interstellar clouds hot cores and circumstellar envelopes of e.g. dying carbon and oxygen stars.1,2 DiÜuse (hot) clouds hold number densities n of ca.10 molecules cm~3 and average translational temperatures T B100 K whereas dense (cold dark molecular) clouds are characterized by n\102»106 molecules cm~3 and T \10»40 K. Hot cores of molecular clouds have typical number densities up to 107 molecules cm~3 and temperatures reaching up to 200 K. Molecules in the out—ow of carbon stars contribute only a minor amount to the interstellar matter but temperatures can rise up to 4000 K Table 1 Classi–cation of neutral molecules radicals and ions detected in the ISM ì ? œ indicates an uncertain assignment; bold typed species are only detected in circumstellar environments 1st atom group 4 2nd atom 1st period 2nd period CC CSi SiC 1st period 2nd Period 1st atom group 4 2nd atom OC OSi 1st period 2nd period SC SSi 1st period 2nd Period CH4 SiH4 C2H4 CH4 185 R.I. Kaiser et al. diatomic molecules H2 group 5 group 4 group 6 group 5 group 4 group 6 group 5 NP NO NS CO CS NC NSi CN CP PN PC SiN SiO SiS group 6 group 6 group 5 OS ON SO SN halides and pseudohalides NaCN HCl KCl MgNC AlCl NaCl MgCN AlF hydrides NH3 H H2O CH NH OH SiH 2S CH2 NH2 2( ?) closed-shell hydroarbons C2H2 CH3CyCH CH3CyCwCyCH ring molecules SiC2 C3H C3H2 C2H4O long-chain molecules CH3w(CyC)nwH n HC Cn Hw(CyC) w(CyC) nwCN nwCN CH3w(CyC)nwCN C C H n n 2 S O Cn n\1 2 n\0»8 n\2 3 5 n\0»4 n\0 1 n\0 1 2 n\3 4 n\1 2 3 5( ?) n\1 2 3 5 n\4 CnSi HCCNC HNCCC 186 CO CH3wNH2 in the outer photosphere14 and a more complex chemistry is anticipated.Comparing this data with that from terrestrial laboratories it is worth mentioning that number densities even in the ìdensestœ interstellar clouds of 106 molecules cm~3 and T \40 K are equivalent to 5]10~12 mbar which compares to the best ultra-high vacuum (UHV) condition in terrestrial laboratories. However despite large fractional abundances of e.g. C up to 10~8 3H C3H2 and relative to atomic hydrogen well de–ned synthetic mechanisms even of these abundant Formation of carbon-bearing molecules in the ISM Table 1 Continued structural isomers l-C HNC 3 3 H H2 c-C3H2 c-C HCN 3H 3CN CH3NC CH MgCN HCO HCCCN HCO` l-C MgNC HOC HCCNC HOC` C CH3wOH oxygen- and carbon-containing neutral molecules HCOOH CH3wCOOH 2H5wOH H2CxO CH3wCHO HCOOCH3 HOC HOC H2CxCxO HCCwCHO CH3wCOwCH3 CO C CH3wOwCH3 2O C3O C5O(?) 2 sulfur- and carbon-containing neutral molecules H2CxS CS CH3wSH nitrogen- and carbon-containing neutral molecules HCN C2H3wCN C2H5wCN H2NwCN CHwCN CH2wCN CH3CN H H2CxNH 2CxN other neutral molecules OCS N2O HNO HNxCxO HNxCxS SO2 HCOwNH2H2NwCH2wCOOH ions SO` HOC` CS` HNN` CH` HCO` HCNH` H3 ` H3O` CO` HCS` HOCO` H2COH`HC CH2D` 3NH` C60 `( ?) 187 R.I.Kaiser et al. C(3P species have not yet been fully resolved. Since the kinetic energy of interstellar species is con–ned to typically 0.8 kJ mol~1 (diÜuse clouds) and 0.08 kJ mol~1 (dark molecular clouds) on average gas-phase reactions under thermodynamical equilibrium conditions (a) must be exothermic or only slightly endothermic (b) should exhibit little or no entrance barriers and (c) must involve only two-body collisions. A three-body reaction occurs only once in a few 109 years and can be neglected since mean interstellar cloud lifetimes are ca. 106 years. Early chemical equilibrium models of interstellar chemistry satisfy these criteria and focus on ion»molecule reactions radiative associations and dissociative recombination between cations and electrons to advance interstellar chemistry.15 This approach however involves multiple reaction chains with subsequent collisions and often cannot reproduce observed structural isomer ratios isotopic enrichments and number densities of extremely abundant radicals e.g. those of C3H and C The inclusion of alternative one-step exothermic neutral»neutral reactions into chemical models of the circumstellar envelope surrounding the carbon star 3H2 . IRC]10216 and the dark cloud TMC-1 occurred only gradually,16 predominantly because entrance barriers were assumed to hinder this reaction class. However the ad hoc postulation of spin conservation simple thermochemistry and the lack of information on reaction products clearly demonstrate the urgency of systematic laboratory examinations probing detailed chemical dynamics and reaction products of neutral» neutral encounters in the gas phase of the ISM.Recently Husain and co-workers investigated rate constants of C(3Pj) with unsatuj) rated hydrocarbons monitoring the decay kinetics of at room temperature.17 These bulk experiments indicated that the reactions proceed with second-order kinetics are barrier-less and rapid (k\10~10»10~9 cm3 s~1) within orbiting limits.18 Likewise kinetic studies of neutral»neutral reactions involving OH CN and CH radicals at ultralow temperatures revealed rate constants of ca. (1»6]10~10 cm3 s~1 with maxima between 50 and 70 K and only a slight decrease as the temperature falls to 13 K.19 However despite extremely valuable kinetic data it was not possible to probe the reaction products experimentally.These limitations clearly indicate the urgency for systematic laboratory studies to identify the reaction products of neutral»neutral encounters relevant to interstellar gas-phase chemistry. What experimental technique is suitable to investigate these gas-phase reactions ? First experiments must be performed under single-collision conditions. This means that in a binary reaction A]BC][ABC]*]AB]C one species A reacts only with one species BC without collisional stabilization or successive reaction of the initially formed [ABC]* complex (exclusion of three-body reactions). This requirement guarantees that the nascent reaction product undergoes no secondary reaction. Secondly highly unstable and reactive radicals often with unknown spectroscopic properties have to be probed.Hence the majority of interesting interstellar radicals such as long-chain carbon chains cummulenes and sulfur-containing species cannot be sampled via optical detection schemes such as laser-induced —uorescence (LIF) and resonance-enhanced multiphoton ionization (REMPI) and a ìuniversalœ detector is crucial. Finally we have to take into consideration that a great variety of structural isomers can contribute to the reaction product (for example the simple formula C4H5 has 25 local minima). Here the knowledge of detailed chemical dynamics of a reaction can be employed to elucidate the product isomer(s). In our experiments all these requirements are ful–lled using the crossed molecular beams technique with a universal mass spectrometric detector,20 cf.Section 2 for a detailed description. This set-up represents a versatile tool to (a) study reaction products under single-collision conditions without three-body reactions ; (b) investigate the chemical dynamics of neutral»neutral reactions and (c) identify distinct structural isomers relevant to interstellar chemistry under well de–ned collision energies. To allow for an explicit assignment of reaction mechanisms and products it is often crucial to combine 188 Formation of carbon-bearing molecules in the ISM j) C2H2 C2H4 CH3 CCH pro- ethylene methylacetylene 3H3(2B2) reaction (2)] butatrienyl24 crossed molecular beams experiments with high level ab initio calculations for structure and energetics of possible intermediate collision complexes as well as reaction energies.21 Recently we initiated these experiments in our laboratory elucidating the chemical dynamics and reaction products of exothermic atom»molecule and atom»radical reactions of C(3P with acetylene pylene C3H6 C and the propargyl radical These investigations provided 3H3 .collision-energy dependent (8.8»45.0 kJ mol~1) doubly diÜerential cross-sections to interstellar linear/cyclic tricarbon hydride22 [l/c-C3H (2%1@2/2B2) reaction (1)] and to hitherto unobserved interstellar propargyl23 [C [n-C4H3(2A@) reaction (3)] methylpropargyl [C4H5(2B2/2AA) reaction (4)] and (1) C(3P diacetylene26 [C4H2(1&g `) reaction (5)] j)]C2H2(1&g `)]l-C3H(2%1@2)]H(2S1@2) C(3P ]c-C3H(2B2)]H(2S1@2) C(3Pj)]C2H4(1Ag)]C3H3(2B2)]H(2S1@2) j)]CH3CCH(1A1)]n-C4H3(2A@)]H(2S1@2) reservoir,27 and formation of j) C3H via neutral» (2) (3) C(3Pj)]C3H6(1A@)]C4H5(2B2/2AA)]H(2S1@2) (4) C(3Pj)]C3H3(2B2)]C4H2(1&g `)]H(2S1@2) (5) The identi–cation of this carbon»hydrogen exchange under single-collision conditions demonstrated the importance of a one-step pathway to form free hydrocarbon radicals and closed-shell hydrocarbons through entrance-barrier-free reactions in interstellar environments.Further the cyclic and linear C3H isomers have both been identi- –ed around the dark molecular cloud TMC-1 and the carbon star IRC]10216. In dark clouds typical ratios of the cyclic versus the linear isomer are near unity but decrease to 0.2^0.1 around the carbon star.In particular the circumstellar shell of IRC]10216 contains a C C(3P 2H2 as well as a neutral reaction very likely takes place. A common acetylene precursor to interstellar c/l-C radicals via atom»neutral reaction with(3P can explain these astronomically 3H Cj) observed isomer ratios.28 The present paper extends these crossed molecular beams investigations and presents results on the reactions of atomic carbon with hydrogen sul–de (H S) allene Supplementary ab initio (H CCCH 2 2) vinyl radical (C2H3) and deuteroacetylene C HD. 2 calculations were performed. Here the reaction of carbon atoms with the closed-shell 2 sulfur-containing molecule hydrogen sul–de is the simplest organosulfur reaction to form carbon»sulfur-containing molecules and radicals in the interstellar medium.The second experiment investigates the chemical dynamics and products of the reaction of ground-state atomic carbon with allene and compares these –ndings with reaction (3) studied recently in our laboratory. These studies allow an explicit identi–cation of the product isomer and outline the necessity to include distinct reactant as well as product isomers in chemical reaction networks modelling the chemistry in interstellar environments. Further we investigated the reaction of atomic carbon with the vinyl radical to study the formation of interstellar C isomers via C 3H2 3H3 reactive intermediates. Finally the reaction of C(3P with to investigate the deuterium isotope eÜect on the formation of interstellar l/c-C j) C2HD is closely related to reaction (1) and allows us 3D.2 Experimental The reactive scattering experiments are carried out in a universal crossed molecular beam apparatus described in detail in ref. 29. Fig. 1 shows a schematic top view of the R. I. Kaiser et al. Fig. 1 Schematic top view of the crossed molecular beams set-up set-up. The 35A crossed molecular beams machine consists of two source chambers (10~4 mbar) a stainless-steel scattering chamber (10~7 mbar) and a rotatable diÜerentially pumped quadrupole mass spectrometric detector (10~11 mbar). The supersonic carbon beam is generated in the primary source via laser ablation of graphite.30 Here the 30 Hz 30»65 mJ 266 nm output of a Spectra Physics GCR-270-30 Nd YAG laser is focused on a rotating carbon rod.Ablated carbon atoms in their 3P electronic ground j state are seeded into neon or helium gas released by a Proch»Trickl pulsed valve.31 A four-slot chopper wheel is mounted 40 mm after the ablation zone to select a 9.0 ls segment of the seeded carbon beam. This segment crosses a second pulsed reactant beam under a well de–ned collision energy at 90° under single collision conditions in the interaction region of the scattering chamber. Neat and seeded mixtures of H2S H CCCH2 C2HD and are held at 1 atm backing pressure. C2H3 2 was generated by 193 nm photolyses of 10% C2H3Br precursor seeded in helium carrier gas. 2 Reactively scattered species are monitored using a diÜerentially pumped quadrupole mass spectrometer rotatable in the plane of the beams with respect to the interaction region.DiÜerentially pumped regions I and II reduce the gas load from the main chamber whereas region III contains the Brink-type electron impact ionizer,32 surrounded by a liquid-nitrogen cold shield the quadrupole mass –lter and the Daly-type scintillation particle detector.33 Despite this diÜerential pumping set-up molecules desorbed from wall surfaces lying on a straight line with the electron impact ionizer (straight-through-molecules) cannot be avoided since the mean free path of these species is of the order of 103 m compared to maximum detector dimensions of ca. 1 m. To reduce these straight-through-molecules a copper plate is attached to a two-stage closed cycle helium refrigerator and cooled to ca.10 K. Since the copper shield is located between the two skimmers and the scattering region the ionizer ìviewsœ a cooled surface from which only H and He desorb at 10 K. The velocity distribution of the products is determined recording the time-of-—ight (TOF) spectra at diÜerent laboratory angles H between [25° and 75° with respect to 189 190 Formation of carbon-bearing molecules in the ISM the carbon beam. In this TOF mode the mass spectrometric controller is set at a constant mass to charge ratio (m/z) and records the time-dependent number density of reactively scattered species at this m/z value at a constant laboratory angle H I(H t). If we integrate the TOF spectra at diÜerent laboratory angles we obtain the intensity distribution in the laboratory reference frame (LAB).(I) g\ø(A)[ø(BC) (II) 3 Data analysis In the previous section we described the crossed molecular beams set-up. Now we present our method of analysis of the laboratory data. For the physical interpretation of the reactive scattering data it is necessary to transform the laboratory data into the center-of-mass (CM) system cf Fig. 2. An observer in the laboratory frame notices that the CM moves with velocity ø(CM). However if this observer dwells at the CM the CM is at rest. Fig. 2 shows the relation between both reference frames. In the experiment a beam of species A with a laboratory velocity ø(A) crosses a beam of species BC with a laboratory velocity ø(BC) at 90° giving the relative velocity of A with respect to BC In the laboratory system the CM frame moves with velocity ø(CM) calculated with the masses of the reactants m(A) and m(BC) ø(CM)\m(A)ø(A)]m(BC)ø(BC)]/[m(A)]m(BC)] The CM velocity vector divides g into two parts the velocity of A and BC in the CM frame u(A) and u(BC) respectively.The magnitude of these vectors is inversly proportional to the mass ratio of the reactants. To convert the laboratory data to the CM system we use a forward-convolution routine to –t the TOF spectra I(H t) at diÜerent laboratory angles H and the product angular distribution in the laboratory frame (LAB).34,35 This procedure initially guesses the angular —ux distribution T (h) and the P(E translational energy —ux distribution in the CM frame.Here h de–nes the scattering angle in the CM system measured from the A beam and E the CM translational T) T energy. Then TOF spectra and LAB distribution were calculated from these T (h) and P(ET) accounting for the velocity and angular spread of both beams the detector acceptance angle and the ionizer length. Both T (h) and P(E reasonable –t of the experimental data is achieved. T) are re–ned iteratively until a Fig. 2 Relation between the LAB and CM reference frames 191 R. I. Kaiser et al. In detail four transformation steps are necessary. First we transform the I(H t) time domain to the velocity domain recalling that (III) I(H t) dt\I(H v) dv and (IV) dv dt\[ v L 2 with the velocity v and ionizer length L . Putting eqn. (IV) into (III) gives (V) I(H t)\[I(H v)v2/L Secondly the detector analyses number density whereas T (h) and P(E distributions.Hence we have to transform a number density distribution (molecules T) represent —ux cm~3) to a —ux distribution (molecules cm~1 s~1). Here the product of I(H v)v is nothing else but a —ux distribution in the laboratory frame de–ned as p(H v) yielding (VI) I(H t)\[p(H v)v/L Thirdly we have to transform p(H v) to the CM —ux distribution p(h u) with the velocity u of the product. The v2/u2-transformation Jacobian can be derived by considering the cross-section proportional —ux per unit time in a solid angle to be constant in the laboratory and CM frames (VII) p(H v) *X\p(h u) *u where *X is the solid angle sustained by the detector aperture *A in the laboratory and the solid angle *u in the CM frame.Recalling the de–nition of a solid angle i.e. dX\dA/r2 with the de–ning aperture of area dA at a distance r from the interaction region and *X\*A/(vt)2 and *u\*A/(ut)2 we put these equations into eqn. (VII) to obtain the transformation Jacobian. Now we can modify eqn. (VI) to (VIII) I(H t)\[p(h u)v3/L /u2 The fourth step transforms the velocity distribution to the translational energy distribution using energy and momentum conservation with k6 \mAB(mAB/mAB/mC]1) and the masses of the products AB and C to give (IX) p(h u)\p(h ET)k6 u yielding I(H t)\[p(h ET)k6 v3/L /u (X) Here p(h ET) is the double diÜerential cross-section in the CM reference frame. p(h ET) is proportional to T(h) and P(E (XI) T) hence p(h ET)\C]T (h)P(ET) are normalized C is obtained by integrating with a constant C.Since T (h) and P(ET) p(h ET) (XII) T)T (h)sin h dh dr dET\C over E n h r (the p(E angle )\Paround = P2n Pthe P( relative velocity vector g) and ET 0 0 0 This identi–es the constant C as the integrated reaction cross-section of the bimolecular reaction A]BC]AB]C at a collision energy E. Hence the –nal relation between the TOF spectra at a laboratory angle H I(H t) and the iteratively re–ned CM —ux distributions T (h) and P(E with the constant C is given by T) (XIII) I(H t)\C T(h)P(ET)v3/u 192 4 Results and Discussion 4.1 C(3Pj) + H2S isomer was observed at m/z\46. The crossed beams experiments were performed at two diÜerent collision energies of 16.7 and 42.4 kJ mol~1.Fig. 3 and 4 show the laboratory angular distributions as well as TOF spectra of the reactive scattering signal at m/z\45 (HCS/HSC) at a selected collision energy of 42.4 kJ mol~1. TOF spectra at m/z\44 show the identical shapes as m/z\45 indicating that HCS` fragments partly to CS` in the electron impact ionizer and that no CS is formed in our experiments. Further no radiative association to any H2CS C(3P Fig. 3 Top HCS product laboratory angular distribution of the reaction j)]H2S(1A1) at a collision energy 42.4 kJ mol~1. Filled circles and 1p error bars indicate experimental data the solid lines the calculated distribution and C.M. the CM angle. Bottom Corresponding velocity C LAB frame.The inner circle stands for the maximum CM recoil velocity of the HSC isomer the vector diagram. v and vH2S indicate the velocities of the carbon and hydrogen sul–de beam in the outer circle for the HCS in the CM frame assuming all available energy channels into translational energy of the products. The solid lines point to distinct laboratory angles whose TOFs are Formation of carbon-bearing molecules in the ISM shown in Fig. 4. 193 R. I. Kaiser et al. Fig. 4 Normalized TOF data of HCS at m/z\45 at a collision energy of 42.4 kJ mol~1. (L) Experimental data (»») the –t. Now we investigate the chemical dynamics of the reaction to unravel the intermecomplex( es) as well as product HCS isomer(s). The experimentally found diate H2CS high-energy cut-oÜs of 208 and 232 kJ mol~1 agree very well with the sum of recent ab initio reaction energy for the HCS isomer36 and the relative collision energies i.e.201 and 226 kJ mol~1. The less stable HSC isomer is expected to show cut-oÜs at 35.0 and 60.7 kJ mol~1 and hence can be excluded as a major contribution to our reactive scattering signal. The shape of the CM angular distributions (Fig. 5) depends strongly on the collision energy Ec . As E is increased T (h) changes from an isotropic forward» backward symmetric to a more forward scattered distribution. This –nding suggests c only one reaction channel following indirect reactive scattering dynamics through a complex formation. At lower collision energy the fragmenting H2CS isomer has a lifetime longer than its rotational period but as the collision energy rises to 42.4 kJ mol~1 37 the lifetime of the complex is reduced to less than one rotational period.2SC The identi–cation of the HCS product clearly excludes decomposing singlet or triplet H 1/2 cf. Fig. 6 since an SwH bond rupture would yield solely the HSC isomer. Further the experimentally found T (h) shows a forward peaking with respect to the carbon beam higher collision energy. This requires that the incorporated carbon atom 194 Formation of carbon-bearing molecules in the ISM Fig. 5 CM angular —ux distributions (bottom) and translational energy —ux distributions (top) for the reaction at peak collision energies of 16.7 (»») and 42.4 kJ mol~1 (» » ») C(3Pj)]H2S(X1A1) and the leaving H atom must be located on opposite sites of the rotational axes.Based on the calculated ab initio geometries of triplet and singlet H2CS 6/7 no rotation axis ful–lls this requirement. Hence thioformaldehyde can also be excluded as the decomposing complex. Based on these conclusions thiohydroxycarbenes 3 4 and 5 are the only remaining intermediates. Each of these complexes can rotate around the B/C axis to account for the forward peaked CM angular distribution yielding HCS and H in the –nal bond rupture. The dynamics leading to the thiohydroxycarbene itself are governed by an addition of C(3Pj) H2S to form triplet 2,2-dihydrothiocarbonyl 2. A direct to insertion into the SwH bond of H2S to yield triplet thiohydroxycarbene 3 can likely be ruled out considering the symmetry-forbidden nature.Therefore one expects a signi–- cant entrance barrier much larger than our lowest collision energy. Owing to the heavy sulfur atom and the narrow singlet»triplet gap of 1 and 2 intersystem crossing (ISC) to 1 might occur followed by an H migration to 4/5. Alternatively 2 could undergo H migration to 3 and subsequent ISC to 4/5. 4.2 C(3Pj) + H2CCCH2 The reactive scattering experiments were performed at two collision energies of 19.6 and 39.3 kJ mol~1. In strong analogy to reactions (1)»(5) the carbon»hydrogen exchange 195 R. I. Kaiser et al. 2CS C(3P and HCS isomers j)]H2S(X1A1) reaction and ab initio structures Fig. 6 Schematic energy level diagram of the of calculated H 4H4 collision complex survives under single-collision (m/z\52) could be detected.This result demonchannel also dominates the product distributions cf. Fig. 7 and 8. The reactive scattering signal is only observed at m/z\51 i.e. C4H3 . TOF spectra were recorded at lower m/z values 50»48 but show identical patterns. This –nding indicates that the signal at these m/z ratios originates from cracking of the parent in the ionizer. In addition no radiative associations to C strates that no internally excited C conditions in our experiments as well as in the ISM. 4H4 Emax\215 kJ mol~1 and 240 kJ mol~1 at our to maximum translational energies of Best –ts of TOF spectra and LAB distributions are achieved with P(ET)s extending lower and higher collision energy respectively (Fig. 9). This high-energy cut-oÜ can be employed to identify the product isomers if their energetics are well separated.Within the error limits data are consistent with the formation of n-C4H3 at both collision energies since the high-energy cut-oÜs are expected at 213.6 kJ mol~1 at lower and 233.3 kJ mol~1 at higher collision energy. At lower collision energy the T (h) is symmetric around n/2. As the collision energy rises the T (h) shape changes to a more forward scattered distribution i.e. increasing intensity at 0°. In strong analogy to the reaction of C(3Pj) H2S these –ndings with suggest a reduced lifetime of the decomposing C4H4 complex as the collision energy rises. Complex formation takes place but the well-depth along the reaction coordinate is too shallow to allow multiple rotations and the complex decomposes with a random lifetime distribution before one full rotation elapses.We now investigate the chemical dynamics of the reaction. The identi–cation of the n-C4H3 isomer strongly suggests that C(3Pj) attacks the p-bond in allene to form a substituted triplet cyclopropylidene intermediate 1 cf. Fig. 10 rotating in the molecular plane which contains the four carbon atoms. 1 undergoes a subsequent ring opening to triplet butatrienylidene 2 followed by a CwH bond rupture to yield atomic hydrogen and n-C4H3 or a hydrogen shift to methylpropargylene 3 prior to decomposition of 3 to n-C4H3]H. To distinguish between these two possibilities we re-ran the crossed 196 Formation of carbon-bearing molecules in the ISM indicate the velocities of the carbon and allene beam in the LAB C(3P Fig.7 Top C4H3 product LAB angular distribution of the reaction j)]H2CCCH2 at a collision energy 39.3 kJ mol~1. Filled circles and 1p error bars indicate experimental data the solid lines the calculated distribution and C.M. the CM angle. Bottom Corresponding velocity vector diagram. ø and ø C frame. The circle stands for the maximum CM recoil velocity of the n-C C3H4 available energy channels into translational energy of the products. The solid lines point to dis- 4H3 isomer assuming all tinct laboratory angles whose TOFs are shown in Fig. 8. molecular beams experiments of C(3Pj) with methylacetylene at the same collision energies as those obtained with allene. Our data show that at all angles the ratio of the intensity of the integrated TOFs at m/z\51 50 and 49 is identical at the higher collision energy for both C(3Pj)]allene and methylacetylene reactions i.e.0.3 (m/z\ 51) 1.0 (m/z\50) 0.5 (m/z\49). Both LAB distributions are also identical. These data strongly indicate that at higher collision energy the decomposing C4H4 complex and the reaction intermediate are the same in both reactions. Since triplet methylpropargylene was assigned as the decomposing complex in reaction (2) we conclude that the chemical dynamics of allene reacting with C(3Pj) are initiated by an attack to the alkenic p-bond to form a triplet cyclopropylidene derivative followed by a ring opening to 2 a H atom migration to 3 and –nal CwH bond rupture to the n-C4H3 isomer. However at lower collision energy ratios of m/z\51 to 50 to 49 of both the methylacetylene and allene reaction with atomic carbon do not match.Hence we must conclude that their chemical dynamics are diÜerent. We pointed out earlier that a second isomer might contribute to the reactive scattering signal of the reaction C/CH CCH at m/z\51.24 A detailed analysis shows that a second cyclic C4H3 isomer 4 5 or 6 is 3 formed as well as at lower collision energy n-C4H3 .38 197 R. I. Kaiser et al. Fig. 8 Normalized TOF data of C4H3 at m/z\51 at a collision energy of 39.3 kJ mol~1. (L) Experimental data (»») the –t. 4.3 C(3Pj) + C2H3 A reactive scattering signal was only observed at m/z\38 i.e. C3H2 . Owing to the limited signal-to-noise ratio and the necessary background subtraction procedure we can present only one TOF recorded at the CM angle of 53° Fig.11. One must keep in mind that it took ca. 9 months to modify and optimize the crossed molecular beams machine for this two-laser experiment to maximize the number density in the C2H3 beam to obtain this reactive scattering signal. Currently the crossed molecular beam machine is undergoing a major re–t to an oil-free operation. This will reduce the background and enhance the signal-to-noise level thus reducing the data accumulation time by a factor of ca. 10. With these experimental improvements this experiment will continue in the near future. Although we have not yet been able to record a full angular distribution the identi–- 3H2 under single-collision conditions underlines the potential importance of isomers in the ISM.Hence at this stage we can only outline 2 and/or ring opens to the propargyl radical 3. 3 either loses an H cation of C this reaction to form C all feasible reaction pathways without resolving the actual one(s) cf. Fig. 12. Atomic 3H2 carbon can attack the alkenic p bond to yield a cyclic C3H3 isomer 1. Depending on its lifetime and the chemical reaction dynamics 1 undergoes CwH bond cleavage to form singlet/triplet C3H2 198 Formation of carbon-bearing molecules in the ISM Fig. 9 CM angular —ux distribution (bottom) and translational energy —ux distribution (top) for the reaction at a selected energy of 39.3 kJ mol~1 isomer(s) formed remains to be resolved. atom at the acetylenic carbon atom to form singlet/triplet vinylidenecarbene H CCC 4 2 or at the alkenic carbon atom to yield triplet/singlet propargylene 5.Based on the energetics all three isomers can be formed in either the singlet or triplet state ; the nature of the C3H2 C(3Pj)]H2CCCH2 n-C4H3 are under investigation.53 C(3Pj)]H2CCCH2 reaction. The equilibrium structures of 3 Fig. 10 Simpli–ed scheme of the and 199 R. I. Kaiser et al. Fig. 11 TOF spectra of C3H2 at m/z\38 at the CM angle of 53° 4.4 C(3Pj) + C2HD Our ab initio calculations show that the isotopic substitution of H versus D shows a profound eÜect on the energetics of the title reaction.39 Earlier investigations revealed that formation of both the c-C3H l-C3 H and isomers are exothermic by 8.6 and 1.3 kJ mol~1 respectively C(3Pj)]C2H2(1&g `)]l-C3H(2%1@2)]H(2S1@2 3 ) ; *RH°\[1.3 kJ mol~1 *RH°\[8.6 kJ mol~1 (6) ]c-C3H(2B2)]H(2S1@2) ; Substituting one H atom by D gives the following reaction energies C(3Pj)]C2DH(1&)]l-C3D(2%1@2)]H(2S1@2) ; *RH°\[0.2 kJ mol~1 * ]l-C3H(2%1@2)]D(2S1@2) ; *RH°\]5.9 kJ mol~1 ]c-C3D(2B2)]H(2S1@2) ; RH°\[9.2 kJ mol~1 *RH°\[1.4 kJ mol~1 (7) ]c-C3H(2B2)]D(2S1@2) ; Crossed molecular beams experiments are underway.Fig. 12 Schematic pathways to distinct C3H2 isomers via C3H3 intermediates 200 Formation of carbon-bearing molecules in the ISM j) C2S:42 Very recently experimental as well as theoretical rotational constants of the HCS radical have been obtained.43 These will be applied to search for the HCS radical in the ISM.Methylacetylene CH CCH has been widely observed in dark molecular clouds such as OMC-1 and TMC-1 in high fractional abundances between (4»6)]10~9 cm~3 by microwave spectroscopy.44 A second C isomer allene (H CCCH2) holds no permanent electric dipole moment and hence shows no rotational spectrum. Although H CCCH should be detectable via IR spectroscopy in the circumstellar shell of e.g. the carbon star IRC]10216 this isomer has escaped any astronomical identi–cation so far. Despite this failure the allene isomer is strongly expected to be present in dark molecular clouds as well as in the out—ow of carbon stars and hence was included in chemical reaction networks modelling time-dependent chemistry in these extraterrestrial environments16 and in astrochemical databases.45 However owing to insufficient laboratory data these models cannot predict the formation of distinct structural isomers and hence are unable to account for diÜerent chemical reactivities of allene versus methylacetylene.Therefore the explicit identi–cation of the n-C4H3 isomer as well as a second cyclic isomer in the reaction of C(3P include distinct reactant as well as product isomers in chemical reaction network model- j) with methylacetylene outline the necessity to ling of the chemistry in interstellar environments. 2 5 Astrophysical implications 5.1 C(3Pj) + H2S C(3P The j)»H2S system represents the prototype reaction of ubiquitous interstellar carbon atoms with the simplest saturated sulfur molecule hydrogen sul–de to synthesize sulfur-containing species via a single atom»neutral collision.The insights in the chemical dynamics of this reaction reveal an important pathway to hitherto astronomically unobserved HCS. The thermodynamically less stable HSC isomer could not be detected in our experiments and upper limits show a maximum contribution of 10% to the reactive scattering signal. Further this reaction does not form CS through H elimination. H2S is ubiquitous in the ISM and has been observed for example in molecular clouds TMC-1 and OMC-1 toward the star-forming region SgrB2 and around the circumstellar envelope of the carbon star IRC]10216.40,41 Upon reaction of C(3Pj) with hydrogen sul–de the very –rst CwS bond is formed. HCS could react with C(3P to form the astronomically observed 5.2 C(3Pj) + H2CCCH2 3 2 2 2 5.3 C(3Pj) + C2H3 Singlet cyclopropenylidene hereafter referred to as c-C was detected in 1985 via 3H2 microwave spectroscopy in the ISM.46 Subsequent quantitative surveys indicated that c-C3H2 is one of the most abundant molecules in interstellar environments such as dark clouds TMC-1 Oph A Ori A and SgrB2 and the carbon star IRC]10216 holding fractional abundances up to 10~8 molecules cm~3.47 In diÜuse clouds cyclopropenylidene is depleted by a factor of ca. 100.48 A second C3H2 isomer singlet vinylidenecarbene H CCC was discovered six years later by Cernicharo et al. towards TMC-1.49 Compared with cyclopropenylidene its fractional abundance is only 1»2%. C(3P C(3Pj)]H2S]HCS]H j)]HCS]C2S]H (8) (9) 2 3H4 201 R.I. Kaiser et al. Most surprisingly however a third isomer triplet propargylene although more stable than vinylidenecarbene has never been observed in the ISM. However despite high number densities of c-C3H2 the formation mechanism of this cyclic molecule has never been resolved either experimentally or theoretically. Typical ion»molecule reaction networks postulate elaborate multiple ion»molecule reactions :50 C2H2]C`]l/c-C3H`]H 3H`]H2 ]c-C3H3 `]hl l/c-C l/c-C3H3 `]e]l/c-C3H2]H C2H4]C`]c-C3H3 `]H ]c-C3H2 `]H2 ]l-C3H2 `]H2 C]cosmic ray]C`]e (10) (11) (12) (13) (14) (15) (16) l/c-C3H3 `]e]l/c-C3H2]H (17) These approaches however neither reproduced fractional abundances isomer-ratios 3H2 isomers.of C-C vs. H CCC nor accounted for high deuterium enrichment observed in c-C3HD c-C3H2 i.e. an observed value of 0.08 in TMC-1 vs. 0.015 obtained in chemi- 2 vs. cal models. Hence the reaction of atomic carbon with the vinyl radical can replace the ion»molecule based four- to –ve-step synthesis through a single reactive encounter to form C3H2 cannot be 5.4 C(3Pj) + C2HD The reaction of C(3P with deuterium isotope eÜect on the formation of l/c-C j) C2HD is closely related to reaction (1) and investigates the 3D. Although the deuteriated isomers have never been observed in the ISM Yamamoto et al. suggested that at least the c-C3D radical should be present and be observable towards TMC-1 in the microwave region.51 Recent crossed beam investigations combined with ab initio calculations of reaction (1) showed that the synthesis of c-C is exothermic by 8.6 kJ mol~1 com- 3H pared with synthesis of the l-C isomer that is exothermic by only 1.3 kJ mol~1.52 The substitution of one H atom by a D atom in acetylene changes the zero-point vibration 3H energy and hence the reaction energy.Based on our ab initio calculations the formation of l-C3H C2DH from covered by the reactantsœ average translational energy of ca. 0.08 kJ mol~1 in cold molecular clouds and only l-C3D can be formed. Both reaction pathways to the cyclic isomer however are exothermic. These –ndings should be re—ected in prospective astronomical surveys of the fractional abundances of l-C vs. 3H l-C3 D toward dark clouds.Our results strongly suggest enhanced deuterium enrichment in the linear isomer versus the cyclic one. In warmer interstellar environments such as the out—ow of carbon stars the reaction endothermicity to l-C3H of only 5.9 kJ mol~1 could be compensated by the enhanced averaged translational temperature of the reactants. Hence compared with cold clouds the isotopic enrichment is expected to be less pronounced. 6 Conclusions The crossed molecular beams technique and ab initio calculations have been established as a universal and powerful tool to investigate neutral»neutral reactions of potential importance to interstellar chemistry under well de–ned reaction conditions. All C(3Pj) reactions with unsaturated hydrocarbons and H2S studied so far are barrier-less and are dominated by a carbon»hydrogen exchange channel.Based on the CM angular —ux 202 Formation of carbon-bearing molecules in the ISM distribution T (h) and CM translational energy —ux distribution P(ET) the crossed molecular beams approach with a universal detector is able to distinguish between distinct hydrocarbon and thiohydrocarbon product isomers. This carbon»hydrogen exchange channel represents an alternative pathway to competing ion»molecule reactions. Further it clearly underlines that not only are reaction rate constants important to model interstellar chemistry but also that the inclusion of distinct structural isomers into these interstellar reaction networks is equally important. This versatile concept can be used further to predict the formation of these radicals in interstellar environments.If regions of high fractional abundances of atomic carbon overlapping with those of the second reactant radical/molecule are identi–ed then the reaction takes place in these environments. Since none of the species except l-C H/c-C3H have been detected in the ISM our results should encourage astronomical search for these hitherto unobserved 3 radicals. In addition no radiative association takes place under single-collision conditions. If the reactions studied here take place on interstellar grains the collision complexes involved could be stabilized. For example the reaction of C(3P with j) H2S on interstellar grains might resolve the anticorrelation of H2CS and H2S in carbon-rich dark clouds TMC-1.Since H2S is formed on interstellar grains implanted carbon atoms from the gas phase very likely react to give a thiohydroxycarbene intermediate. Its lifetime is expected to be longer in a solid matrix as compared with our crossed beam experiments and a second H-migration to thioformaldehyde might take place. The work presented so far is just the –rst steps towards a better understanding of the importance of neutral»neutral reactions in contrast to ion»molecule reactions in the formation of molecules and radicals in extraterrestrial environments. The chemical dynamics of atom»radical and radical»radical reactions in the ISM are completely unknown. Both reaction classes however are expected to have a profound impact on chemistry in interstellar and hydrocarbon-rich planetary environments at very low temperatures down to 10 K reactive encounters between C(3Pj) CH C2H and open-shell hydrocarbons such as CH C2H C2H3 prototype reactions proceeding without any barrier in the entrance channel.Therefore and C3H3 radicals are thought to resemble these reactions are strongly expected to form complex species even in the coldest known interstellar clouds where the average kinetic energy of reactant molecules is ca. 0.08 kJ mol~1 and will be studied in the future. We will keep you informed. R.I.K. is indebted to the Deutsche Forschungsgemeinschaft (DFG) and Academia Sinica. Institute of Atomic and Molecular Sciences (IAMS) for a Habilitation fellowship (IIC1-Ka1081/3-1). Support from Prof.D. Gerlich (Technical University Chemnitz Germany) and Prof. Y. T. Lee (Academia Sinica Taiwan) is gratefully acknowledged. C.O. acknowledges –nancial support by a DFG postdoctoral fellowship. Special thanks to Dr. I. Hahndorf (IAMS) and Mr. S. Harich (IAMS) for comments on this manuscript. This work was partly supported by the Director Office of Energy Research Office of Basic Energy Sciences Chemical Sciences Division of the U.S. Department of Energy under Contract No. DE-AC03-76SF00098. References 1 Interstellar Processes ed. D. J. Hollenbach and H. A. Thronson Reidel Dordrecht 1987. 2 H. Scheffler and H. Elsaé sser Physics of the Galaxy and Interstellar Matter Springer Berlin 1988. 3 J. M. Greenberg Astron. Astrophys. 1971 12 240; A. G. G. M. Tielens and L.J. Allamandola in ref. 1 p. 397. 4 D. C. B. Whittet in Dust and Chemistry in Astronomy ed. T. J. Millar and D. A. Williams Institute of Physics Bristol 1993 p. 1; B. Schmitt in Molecules and Grains in Space ed. I. Nenner AIP Press New York 1994 p. 735; W. A. Schutte et al. Astron. Astrophys. 1996 309 633; A. G. G. M. Tielens L. J. Allamandola and S. A. Sandford in Solid-State Astrophysics ed. E. Bussoletti et al. North-Holland Amsterdam 1991 p. 29; J. E. Chiar A. J. Adamson T. H. Kerr and D. C. B. Whittet Astrophys. J., 203 R. I. Kaiser et al. 1995 426 240; R. J. A. Grim and J. M. Greenberg Astrophys. J. L ett. 1987 321 91; J. H. Lacy F. Baas L. J. Allamandola S. E. Persson P. J. McGregor C. J. Lonsdale T. R. Geballe and C. E. P. van de Bult Astrophys.J. 1984 276 543; S. C. Tegler D. A. Weintraub T. W. Rettig Y. J. Pendleton D. C. B. Whittet and C. A. Kulesa Astrophys. J. 1995 439 279; M. E. Palumbo A. G. G. M. Tielens and A. T. Tokunaga Astrophys. J. 1995 449 674. 5 L. B. dœHendecourt and L. J. Allamandola Astron. Astrophys. Suppl. 1986 64 453; L. J. Allamandola S. A. Sandford and G. J. Valero Icarus 1988 76 225; R. J. A. Grim J. M. Greenberg M. S. de Groot F. Baas W. A. Schutte and B. Schmitt Astron. Astrophys. Suppl. 1989 78 191; O. M. Shalabiea and J. M. Greenberg Astron. Astrophys. 1994 290 266. 6 R. E. Johnson Energetic Charged Particle Interactions with Atmospheres and Surfaces Springer Berlin 1990; L. J. Lanzerotti and R. E. Johnson in Ion Beam Modi–cations of Insulators ed. P. G. W. Mazzoldi and E.Arnold Elsevier Amsterdam 1987 p. 631. 7 R. E. Johnson L. J. Lanzerotti and W. L. Brown Adv. Space Res. 1984 4 41; J. Geiss et al. in COSPAR Colloquia Series ed. E. Marsch and R. Schwenn Pergamon New York 1992 vol. 3 p. 20 and references therein. 8 K. Roessler in Solid-State Astrophysics ed. E. Bussoletti and G. Strazzulla North-Holland Amsterdam 1991 p. 454. 9 G. Stoé cklin Chemie heiêer Atome VCH Weinheim 1991. 10 R. I. Kaiser and K. Roessler Astrophys. J. 1997 475 144; R. I. Kaiser G. Eich A. Gabrysch and K. Roessler Astrophys. J. 1997 484 487; R. I. Kaiser and K. Roessler Astrophys. J. 1997 submitted. 11 M. Heyl Report Jué l-2409 1990. 12 W. A. Schutte and J. M. Greenberg Astron. Astrophys. 1991 244 190 and references therein ; R.I. Kaiser G. Eich A. Gabrysch and K. Roessler Astrophys. J. 1997 484 487. 13 E. L. O. Bakes T he Astrochemical Evolution of the Interstellar Medium Twin Press Vledder 1997. 14 Z. K. Alksne A. K. Alksnis and U. K. Dzervitis Properties of Galactic Carbon Stars Orbit Malabar 1991. 15 E. Herbst and W. Klemperer Astrophys. J. 1973 185 505; E. Herbst N. G. Adams and D. Smith Astrophys. J. 1984 285 618; G. Winnewisser and E. Herbst T opic Curr. Chem. 1987 121. 16 T. J. Millar and E. Herbst Astron. Astrophys. 1994 288 561; I. CherchneÜ and A. E. Glassgold Astrophys. J. L ett. 1993 419 41; E. Herbst H. H. Lee D. A. Rowe and T. J. Millar Mon. Not. R. Astron. Soc. 1994 268 335. 17 D. Husain J. Chem. Soc. Faraday T rans. 1993 89 2164; D. C. Clary N. Haider D. Husain and M.Kabir Astrophys. J. 1994 422 416. 18 R. D. Levine and R. B. Bernstein Molecular Reaction Dynamics and Chemical Reactivity Oxford University Press Oxford 1987. 19 I. R. Sims et al. J. Chem. Phys. 1994 100 4229; I. R. Sims et al. J. Chem. Phys. 1992 97 8798; I. R. Sims et al. Chem. Phys. L ett. 1993 211 461; I. R. Sims et al. J. Chem. Soc. Faraday T rans. 1994 90 1473. 20 Y. T. Lee Science 1987 236 793. 21 C. Ochsenfeld R. I. Kaiser A. G. Suits Y. T. Lee and M. Head-Gordon J. Chem. Phys. 1997 106 4141. 22 R. I. Kaiser C. Ochsenfeld M. Head-Gordon Y. T. Lee and A. G. Suits Science 1996 274 1508. 23 R. I. Kaiser Y. T. Lee and A. G. Suits J. Chem. Phys. 1996 105 8705. 24 R. I. Kaiser D. Stranges Y. T. Lee and A. G. Suits J. Chem. Phys. 1996 105 8721.25 R. I. Kaiser D. Stranges H. M. Bevsek Y. T. Lee and A. G. Suits J. Chem. Phys. 1997 106 4945. 26 R. I. Kaiser W. Sun A. G. Suits and Y. T. Lee J. Chem. Phys. 1997 107 8713. 27 J. Keene K. Young T. G. Phillips and T. H. Bué ttgenbach Astrophys. J. L ett. 1993 415 131. 28 R. I. Kaiser C. Ochsenfeld M. Head-Gordon Y. T. Lee and A. G. Suits Science 1996 274 1508. 29 Y. T. Lee J. D. McDonald P. R. LeBreton and D. R. Herschbach Rev. Sci. Instrum. 1969 40 1402. 30 R. I. Kaiser and A. G. Suits Rev. Sci. Instrum. 1995 66 5405. 31 D. Proch and T. Trickl Rev. Sci. Instrum. 1989 60 713. 32 G. O. Brink Rev. Sci. Instrum. 1966 37 857. 33 N. R. Daly Rev. Sci. Instrum. 1960 31 264. 34 M. S. Weiss PhD Thesis 1986 University of California Berkeley. 35 M. Vernon PhD Thesis 1981 University of California Berkeley.36 R. I. Kaiser C. Ochsenfeld M. Head-Gordon and Y. T. Lee Science submitted. 37 W. B. Miller S. A. Safron and D. R. Herschbach Discuss. Faraday Soc. 1967 44 108 291. 38 R. I. Kaiser A. Mebel and Y. T. Lee J. Chem. Phys. to be submitted. 39 The zero-point corrections have been calculated for 1-C3D C2HD at the CCSD(T) (coupled cluster singles and doubles with a perturbative treatment of triple excitations) approximation using a TZP (triple zeta polarization) basis set. For c-C3D the EOMIP-CCSD/TZP (equation of motion CCSD for ionized states) approach was used (R. J. Bartlett and J. F. Stanton in Reviews of Computational Chemistry ed. K. B. Lipkowitz and D. B. Boyd VCH New York 1990 p. 65. 40 P. Thaddeus et al.Astrophys. J. L ett. 1997 176 73. 204 Formation of carbon-bearing molecules in the ISM 41 Y. C. Minh L. M. Ziurys W. M. Irvine and D. McGonagle Astrophys. J. 1991 366 192. 42 E. Herbst personal communication 1997. 43 H. Habara S. Yamamoto C. Ochsenfeld M. Head-Gordon R. I. Kaiser and Y. T. Lee J. Chem. Phys. to be submitted. 44 E. C. Sutton R. Peng W. C. Danchi P. A. Jaminet G. Sandell and A. P. G. Russel Astrophys. J. Suppl. 1995 97 455; F. Combes G. Wlodarczak P. Encrenaz and C. Laurent Astron. Astrophys. 1992 253 L29. 45 T. J. Millar P. R. A. Farquhar and K. Willacy Astron. Astrophys. Suppl. 1997 121 139. 46 P. Thaddeus J. M. Vritilek and C. A. Gottlieb Astrophys. J. L ett. 1985 299 63; H. E. Matthews and W. M. Irvine Astrophys. J. L ett. 1985 298 61. 47 S. Green Astrophys. J. 1980 240 962; B. E. Turner Astrophys. J. L ett. 1989 347 L39; P. Thaddeus J. M. Vrtilek and C. A. Gottlieb Astrophys. J. L ett. 1985; 299 63; T. B. Kuiper J. B. Whiteoak R. S. Peng W. L. Peters III and J. E. Reynolds Astrophys. J. L ett. 1993 416 33. 48 B. E. Turner L. J. Rickard and L. P. Xu 1989 344 292. 49 J. Cernicharo et al. Astrophys. J. L ett. 1991 368 39. 50 S. C. Madden in Chemistry in Space ed. J. M. Greenberg and V. Pirronello Kluwer Dordrecht 1991 p. 437 and references therein. 51 S. Yamamoto and S. Saito Astrophys. J. L ett. 1990 363 13. 52 C. Ochsenfeld R. I. Kaiser A. G. Suits Y. T. Lee and M. Head-Gordon J. Chem. Phys. 1997 106 4141. 53 A. Mebel personal communication 1997. Paper 8/00077H; Received 2nd January 1998

ISSN:1359-6640    

DOI:10.1039/a800077h

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Faraday Discuss. 1998 109 205»216 Chemical and physical evolution of dark clouds Molecular spectral line survey toward TMC-1 Masatoshi Ohishia and Norio Kaifub a Nobeyama Radio Observatory Nobeyama Minamimaki Mimanisaku Nagano 384-1305 Japan b Subaru Observatory National Astronomical Observatory of Japan Hilo HI 96720 USA Interstellar dark clouds are known as the source of star and planetary system formation and the study of chemical composition in dark clouds is important for understanding the process of evolution of matter in the universe toward the planets and to life. The results of the –rst molecular spectral line survey toward a typical dark cloud TMC-1 (Taurus Molecular Cloud-1) have been reported. The observations were completed in the frequency range 8800»50 000 MHz and the results of detailed data analysis are presented.This is the –rst complete spectral line survey toward a cold dark cloud ever made. We used the 45 m mm-wave telescope and a very large acousto-optical radiospectrometer with 32 000 output channels of the Nobeyama Radio Observatory NAOJ. We detected 404 lines from 38 mol- HNCCC HCCNC HCCCNH` HCCCHO CH ecules including 11 new molecules such as C6H CCO CCCO CCS CCCS 2CN them are short-lived organic compounds unknown before our detection. We and cyclic C3H all of also detected three new isotopomers CC34S CCC34S and HDCS. These results provide the basic general composition of cold dark clouds for the –rst time. We also discuss chemical and physical evolution of dark clouds.Cold dark clouds are formation sites for low-mass stars and planetary systems. These TKB10 K n(H clouds often contain several dense cores with 2)B104»105 cm~3 and the mass of one to a few M_ . Such physical conditions together with the lack of embedded high-luminosity sources make the cold dark molecular clouds ideal testing sites for models of gas-phase chemistry. TMC-1 (Taurus Molecular Cloud-1) is a famous cold dark cloud with a distance of 140 pc and is a part of the Heiles Cloud 2. It is well known that many molecular species have been observed toward TMC-1 and modellers of chemical reaction networks sometimes compile observed molecular abundances toward TMC-1 from various telescopes. Because the line width is narrow *vB0.5 km s~1 observers had to use high-frequency resolution spectrometers and thus narrow total bandwidths.Furthermore because the lines are so weak typically 0.1»1 K they need to integrate for a long time. These have prevented observers from performing unbiased spectral line surveys toward cold dark clouds. Also only a small portion of centimeter and millimeter wave regions were observed toward dark clouds. The 45 m radiotelescope of the Nobeyama Radio Observatory National Astronomical Observatory of Japan is equipped with 32 000 channel high-resolution acoustooptical radio spectrometers (AOS-Hs) covering 320 MHz at maximum. Combining the antennaœs large collecting area and the wide total bandwidth of the spectrometers have enabled us to make an unbiased spectral line survey toward cold dark cloud to 205 206 Chemical and physical evolution of dark clouds Fig.1 Compressed spectrum from 8800 to 50 000 MHz toward TMC-1 (T A * is the antenna temperature corrected for atmospheric attenuation and ohmic loss of the antenna) derive systematic molecular abundances in a typical dark cloud TMC-1. We also expected to detect several unknown materials through the unbiased spectral line survey. In this paper we report the result of the –rst spectral line survey toward TMC-1 discuss the excitation of molecules abundances and isotopic ratios brie—y and review the chemical and physical evolution of dark clouds. Observations The observations were made from April 1986 to January 1997 using the 45 m radio telescope of the Nobeyama Radio Observatory.The observed position is the cyanopolyyne peak of TMC-1 and the coordinate is a(1950.0)\04h 38m 38.s6 d(1950.0)\25°35@45A. The frequency coverage is from 8800 MHz to 50 000 MHz. We used four cooled HEMT receivers and an SIS receiver to cover the whole frequency region. The system temperatures were ca. 100 K between 8800 and 26 000 MHz ca. 120 K between 26 000 and 33 000 MHz and ca. 300 K above 33 000 MHz respectively. The backend was a set of acousto-optical spectrometers (AOS) with a frequency resolution of 37 kHz and a channel spacing of 20 kHz. Therefore our data set consists of 2 060 000 data points. Since a single spectrometer has a bandwidth of 40 MHz we used 16 spectrometers at maximum and the maximum frequency coverage of a single observation was 320 MHz.The pointing was checked by observing the SiO maser in NML Tau. The beam sizes ranged from 200 arcseconds at 8800 MHz to 36 arcseconds at 50 000 MHz. The observed lines are narrow (*vB0.5 km s~1) and weak. Therefore we needed a long integration time typically 1»2 h ON position in a single frequency coverage. Because there are many receivers operating in the receiver cabin of the 45 m telescope we sometimes experienced weak birdies which look like real signals. To distinguish birdies and real molecular lines we repeated observations a few times at the same fre-207 M. Ohishi and N. Kaifu Fig. 2 Sample spectrum from 45 000 to 46 000 MHz quency region by changing LO frequencies slightly. Thus we achieved fairly uniform rms noise levels of ca.10 mK at around 10 000 MHz and ca. 20 mK at around 40 000 MHz. Among the weaker lines there are many carbon-chain mol- Data and Results Initial line identi–cations were made by referring to published laboratory papers. We found that some line frequencies were not listed in these laboratory papers and we calculated the frequencies using the molecular constants in the papers. Finally we revised the molecular constants by reanalyzing laboratory and astronomical frequencies simultaneously. The revised molecular constants will be published elsewhere. Fig. 1 shows the whole spectrum toward the cyanopolyyne peak of TMC-1 in the range 8800»50 000 MHz. There are several constant-frequency spacing patterns from linear molecules such as HC3N HC5N H7N. and C The strongest lines seen in the frequency range are HC3N.208 Chemical and physical evolution of dark clouds ecules like C4H C5H C6H and and some asymmetric organic molecules. In total we detected 404 lines from 38 species. Therefore the line density (number of lines detected/ frequency width) is 404/41.2\9.8 lines per GHz. In Fig. 2 we show a sample of line rich region at around 45 000 MHz and Fig. 3 is a sample of line-by-line display. It is seen that the line width is narrow ca. 0.5»0.6 km s~1 and the radial velocity is around 5.8 km s~1 for most of the lines. It is known that there are two velocity components toward the cyanopolyyne peak 5.6 and 6.0 km s~1. For optically thick lines such as CS two components are blended while for some lines e.g.CCO and CCCO the 5.6 km s~1 component is dominant. For cyanopolyynes HC3N HC5N HC7N H9N and C and polyynes C3H C4H C4H C5H C and the 6.0 km s~1 component is dominant. This suggests that cyanopolyynes and polyynes may have common formation processes. 6H Fig. 3 Sample of line-by-line display 209 M. Ohishi and N. Kaifu In TMC-1 49 molecules have been detected so far. Therefore we have observed 78% of the molecular species known in TMC-1. The other 13 species are small and abundant molecules such as CO HCN and HCO` whose fundamental transition frequencies fall above 50 000 MHz and less abundant species (H CN HC etc.,) which were detected 2 through recent very sensitive observations. These statistics are shown in Table 1. 11N 6H HC3NH` CH2 N 3H. All molecules except for HCCCHO were short-lived organic Eleven molecules out of 38 were detected by the 45 m telescope during our spectral line survey C CCO CCCO CCS CCCS HCCNC HNCCC HCCCHO and c-C compounds unknown before our detection and all these molecules have carbon-chain backbones.Most of the detected molecules toward TMC-1 have carbon-chain molecular Table 1 Molecules detected toward dark cloud TMC-1 species number of detected isotopic lines detected rarer isotopomers number of detected lines 4 1 0 3 1 2 6 0 0 1 3 0 0 1 0 5 0 0 3H 6 0 0 5H 16 0 CS HCS` SO OCS NH HNCO C C l -C 3 4H 26 0 C C6H 34 0 3N 12 0 2 1 0 0 0 0 03 0 3N 4 0 0 11 87 5 1 1 54 00 2 1 0 00 7 HC HC CCO CCCO CCS CCCS HC HCCNC HNCCC HC HC3NH` 2 0 5N 46 6 7N 34 0 9N 16 0 30 000 0 0 4 0 0 0 12 1 3 HH 2C3 2C4 HH 2CO CCO 22 CS 0 1 0 38 2 5 8 H CH CH2CN 3CN 33 CH CCCN CH CCH 3 CH CH3C4H 8 0 3OH 2 1 3 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0 0 6 72 CH CHO HCCCHO CH CHCN 12 1 c-C2 c-C3H 6 0 0 3H2 9 2 38 molecules 404 1 20 unidenti–ed 210 Chemical and physical evolution of dark clouds backbones and it is clear that the carbon-chain molecules are dominant at the cyanopolyyne peak of TMC-1.It is recognized that the carbon-chain molecules are the fourth form of carbon in our universe after diamond amorphous carbon and fullerene.2CO 21 1»21 2 Finally we would like to note that there are two absorption lines ; H at 14 488.479 MHz and C3H2 22 0»21 1 at 21 587.395 MHz and there is only one con- –rmed unidenti–ed line U49536. This means that most of the fairly abundant fairly large molecules have already been detected. Molecules which might be detected in the future below 50 000 MHz would be less abundant species i.e. longer molecules and very short-lived radicals or very exotic ones. Of course new molecules would be detected in higher frequency regions because small basic molecules have transitions in short millimeter and submillimeter wave regions. (1) TB\[Jl(Tex)[Jl(TBB)][1[exp([q)] (2) Jm(T )\Ah k l expAkT hl [1D~1 BC B (3) DB2AN cm (mol) ~2 BAkm * s v ~1B~1 Q S f (l Tex) q\1.248]10~13Ak (4) B B f (l Tex)\CexpAT hl [1DexpAkT Eu ex 7N 7N.Discussion Excitation analysis T The excitation temperature and the column densities of each molecule N(mol) in TMC-1 were derived from the observed lines by adjusting the excitation temperature ex and the column density by a least-squares –t. The equations used are as follows where T In these equations T and represent the excitation temperature and the cosmic ex BB background radiation temperature respectively and l is the transition frequency. We assumed the Boltzmann population distribution with temperature Tex and derived the total column density N(mol) from the optical depth q by using the following relations and ex where S is the line strength Q is the partition function and k is the permanent electric E stands for the upper-level energy of the transition and *v for the dipole moment.FWHM of the line. The partition function of each molecule was calculated for energy u levels Eu\100 cm~1. Before applying the above analyses we had to correct for beam dilution by assuming a source size of 160 arcseconds to convert antenna temperatures to brightness temperatures. For several species whose number of detected lines are too small to apply the least-squares analyses we assumed an excitation temperature of 5 K. Furthermore we used optically thin lines when possible and assumed terrestrial isotopic ratios to derive column densities of normal species.Fig. 4 shows an example of the least-squares local thermodynamic equilibrium (LTE) analysis for HC We have detected 34 transitions of HC and most of the data are well explained by a single excitation temperature. This means that the assumption of the LTE condition is a good approximation of the level distribution of HC in TMC-1. As 7N M. Ohishi and N. Kaifu Fig. 4 Result of the least-squares LTE analysis for HC7N is seen from this –gure the excitation temperature is 8.3 K which is higher than the ìnormalœ excitation temperature of 5»6 K. We also derived excitation temperatures for other linear carbon-chain molecules. Fig. 5 summarizes the result. There seems a clear trend that the excitation temperature is higher for longer carbon-chain molecules except Fig.5 Plot of excitation temperatures as a function of number of carbon atoms. It is seen that the excitation temperature approaches the gas kinetic temperature as the chain grows. 211 212 Chemical and physical evolution of dark clouds N(H2)\1]1022 cm~2. Fig. 6 Histogram of column density of observed molecular species. The column density can be converted into fractional abundance by dividing by for HC Our result is consistent with the result by Bell et al.1 for HC 3N. 9N they report an excitation temperature of HC to be ca. 11.5 K and our value is 11.1 K. This trend 9N may be explained as follows. Since the Einstein coefficients are proportional to l3k2 the radiative cooling rates remain small even for transitions with high J numbers.This means that the collisional processes become more dominant than radiative processes for longer molecules. This can reconcile the exceptional case for HC3N because of its high optical depths the photon trapping eÜect raises the excitation temperature of HC3N more than for the optically thin case. Therefore we conclude that the excitation temperature approaches the gas kinetic temperature as the chain grows and the kinetic temperature at the cyanopolyyne peak of TMC-1 is about 11 K. 213 M. Ohishi and N. Kaifu Column densities 2 Fig. 6 summarizes the column densities of observed species. The column densities can be N(H converted into fractional abundances by dividing by 2)\1]1022 cm~2. Column densities of many molecules have been revised compared with previous work (Ohishi et al.2).For example N(HC N) N(l-C3H) and N(C4H) increased by about one order of magnitude. This is due to the fact that we detected several weak hyper–ne components 3 and in the analyses we –tted all hyper–ne components simultaneously. For optically thin molecular lines column densities are consistent with previous work.2 As is already known carbon-chain molecules are quite abundant at the cyanopolyyne peak of TMC-1. Our observations have enhanced this fact. As far as we know the total column density of carbon-chain molecules is close to or above that of OH and is next to that of CO. In Fig. 6 CS is less abundant than CCS contrary to previous work. Because we used only J\1»0 data to derive N(CS) the value should be checked by observing other transitions.This also applies to some other molecules such as H CS. These new data should be used to develop further the gas-phase chemistry models of cold dark clouds. Isotopic ratios We detected several isotopomers of various molecules. In Table 2 we summarize isotopic abundance ratios in TMC-1 relative to solar ratios. The table includes four isotopes 13C 34S 33S and D. The isotopic ratios of 13C 34S and 33S are consistent with the solar value within a factor of two. On the other hand the isotopic ratio of D is much higher than the solar value by 1»2 orders of magnitude. Deuterium fractionation is very important for considering formation mechanisms of molecules. For example Minowa et al.3 have measured the laboratory spectrum of Table 2 Isotopic species ratios in TMC-1 species abundance radioa relative to solar ratio number of detected lines 1 2.3 1 1 8 1 1 0.5 6 1 1 0 1 1 1 13CS 1 C34S 1 C33S 1 1 34SO 1 CC34S 3 CCC34S 1 1 DCCCN 6 5 6 H13CCCN HC13CCN HCC13CN 6 3 6 DCCCCCN H13CCCCCN HC13CCCCN 6 6 4 HCC13CCCN HCCC13CCN HCCCC13CN 5 4 1 HDCS 230 c-C3HD c-C 500 2 2 3H2 (13C oÜ axis) a Abundance ratio\M([isotopic species]/[normal molecule])dark cloudN/ M([isotope atom]/[normal atom])solar valueN.214 Chemical and physical evolution of dark clouds 2 Fig. 7 Evolutionary track of dark cloud as a function of [[CCS]/[NH3]. Numbers along the solid line indicate chemical ìageœ of cloud in units of 105 yr.The number density of H was assumed to be 10~3 and TK\10 K. IRAS sources normally indicate that young stars have already formed in the cloud. HDCS and identi–ed it in TMC-1. They considered its formation pathway as follows. is considered to form with the following gas-phase reactions (I) (II) (III) This is 4 H2CS CH4]S`]CH3S`]H followed by CH3S`]e]H2CS]H The following neutral»neutral reaction is suggested to be a possible pathway CH3]S]H2CS]H Millar et al.4 predict a high deuterium fractionation both in CH and CH mainly caused by the results of the fractionation reaction 3 . (IV) CH3 `]HD]CH2D`]H2 The ratio of the deuterated species in CH and CH is 0.03 in their early-time model 4 with a new recombination rate.If H 3 2CS is produced through the reactions shown in eqn. (I) and (II) the deuterium fractionation in CH is transferred to H CS. In this case 4 the probability that the deuterium atom remains in H 2 2CS is one-half unless H atoms are preferentially detached in these reactions. Therefore the expected [HDCS]/[H CS] 2 via the reaction shown in eqn. (III) the expected [HDCS]/[H CS] ratio is 0.02. In the ratio would be 0.015 which is comparable to the ratio observed. If H2CS is produced 2 model of Howe and Millar,5 the ratios of the deuterated species of CH and CH are 4 0.036 and 0.037 respectively as long as the HD molecule is produced on dust grains. 3 These values are slightly higher than the Millar et al.4 values and hence the expected [HDCS]/[H CS] ratio would be slightly higher too.Therefore the deuterium fractiona- 2 215 M. Ohishi and N. Kaifu tion of HDCS in TMC-1 can be qualitatively explained on the basis of gas-phase reactions. Similar discussion would apply to other deuterated species and it is expected that the formation pathways for these molecules would be understood in more detail. Chemical evolution of dark cloud Followed by our detection Suzuki et al.6 surveyed the CCS molecule and NH in 49 dark cloud cores in the Taurus and the Ophiucus regions. They found that the abun- 3 dance ratio of CCS to NH is a good indicator of chemical and physical evolution of dark clouds. In Fig. 7 we show an evolutionary track diagram of dark cloud. Suzuki et 3 al.6 simulated a pseudo-time-dependent chemical network to study the formation N(H process of CCS.This –gure is based on the simulation for 2)\104 cm~3 and TK\ 10 K. The solid line shows the result of the simulation. The numbers along the line indicate cloud ìageœ in units of 105 yr. The –lled symbols are cores without Infrared Astronomical Satellite (IRAS) sources i.e. with no embedded protostars and the open symbols represent those with IRAS sources. It is quite natural that cores with embedded protostars are physically more evolved than those without protostars. It should be noted that most of the cores without an IRAS source are located in the top-left part of the –gure and those with an IRAS source are seen in the bottom-right. In the top-left part the ratio [CCS]/[NH3] is larger than that in the bottom-right.Furthermore the fractional abundance of CCS is much higher in the top-left. The simulation showed that CCS has a higher fractional abundance when the cloud ìageœ is around 105 yr whereas it decreases rapidly as the cloud evolves physically. Therefore we conclude that cores with high [CCS]/[NH3] ratio are chemically and physically ìyoungœ and those with low ratio are evolved. It is recognized that cores have density structure and are dynamically contracting. The above simulation assumed a uniform density and did not consider dynamical contraction. Therefore it is necessary to develop a more realistic simulation model to understand further physical and chemical evolution of dark cloud cores in the future.However we believe that the general trend of the evolution does not change greatly. Conclusions We have completed the –rst spectral line survey toward the cyanopolyyne peak of a cold dark cloud TMC-1 in the frequency range 8800»50 000 MHz. We detected 404 spectral lines from 38 molecular species. During the survey we discovered 11 new molecules most of which were unknown before our discoveries. Most of the observed species at the cyanopolyyne peak of TMC-1 are carbon-chain molecules and their derivatives. Therefore it is recognized that the carbon-chain is the fourth form of carbon in our universe. We have a reliable molecular abundance data set to understand further the gasphase chemistry in a quiescent molecular cloud. We have also found that the excitation temperature of long linear carbon-chain molecules approaches the gas kinetic temperature.We thank all the staÜ of the Nobeyama Radio Observatory for their assistance with our observations especially Shin-ichi Ishikawa who joined most of our observations. Keisuke Miyazawa provided very low noise HEMT receivers. Without their participation we could not have completed this survey. We acknowledge Kentarou Kawaguchi of the Nobeyama Radio Observatory Shuji Saito of the Institute for Molecular Science Satoshi Yamamoto of the University of Tokyo Yasuhiro Hirahara of Nagoya University and Shuro Takano of the University of Cologne. We also thank Chiaki Noumaru Sumiko Harasawa and Michiko Okuda for their contribution to the preparation of the –gures. 216 References 1 M. B. Bell P. A. Feldman and J. K. G. Watson 1998 in preparation. 2 M. Ohishi W. M. Irvine and N. Kaifu in Proceedings of the IAU symposium No. 150 Astrochemistry of Cosmic Phenomena ed. P. D. Singh 1992 p. 171. 3 H. Minowa M. Satake T. Hirota S. Yamamoto M. Ohishi and N. Kaifu Astrophys. J. L ett. 1997 491 L63. 4 T. J. Millar A. Bennette and E. Herbst Astrophys. J. 1989 340 906. 5 D. A. Howe and T. J. Millar Mon. Not. R. Astron. Soc. 1993 262 868. 6 H. Suzuki S. Yamamoto M. Ohishi N. Kaifu S. Ishikawa Y. Hirahara and S. Takano Astrophys. J. 1992 392 551. Chemical and physical evolution of dark clouds Paper 8/01058G; Received 5th February 1998

ISSN:1359-6640    

DOI:10.1039/a801058g

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Faraday Discuss. 1998 109 217»256 General Discussion Introducing his paper Dr Sorokin said In our paper the photon mechanism shown in Fig. 4 was proposed to explain selective enhancement of the k5797 and k6379 DIB intensities. In this mechanism coherent VUV Stokes-wave emission was assumed to occur on the transition B0-0P3. Although spectral evidence for such an emission band appears to be present in the low-resolution Copernicus VUV spectrum of b Ori A (see Fig. 11 of our paper) a prominent emission band peaking at the exact B0-0P3 wavelength (1115.9 ”) is not seen in the high-resolution Copernicus VUV spectrum of the same star that is shown in ref. 27. Therefore it has become necessary for us to reexamine this situation and to try to –nd another mechanism capable of ìfeedingœ molecules to virtual quantum levels surrounding J@\2 of B0.One such mechanism shown in Fig. 1 here appears to possess a number of conceptual advantages. In this new scheme virtual quantum levels surrounding J@\2 of B0 are strongly fed from an excited vibrational level (JA\1 of X2) by Ly-a photons trapped by elastic scattering in the cloud of our model. (The radiative strength of B0-2R1 is approximately fourteen times that of B0-0P3.) Molecules are driven from Fig. 1 New photon mechanism proposed for exciting virtual quantum levels around J@\2 of B0 217 218 General Discussion JA\1 of X0 to JA\1 of X2 as part of the four-wave mixing process shown on the right. Spectral evidence of the existence of components of this four-wave mixing process is indicated in the –gure.With this new scheme one can also accurately account for the ìnotchesœ that are observed on the k5797 and k6379 DIB pro–les. Fig. 2 shows the very high resolution spectra of the k5797 DIB pro–le that are presented in ref. 13. (Since the existence of the k5797 ìnotchœ was –rst clearly revealed in this study and since four of the six authors of this work hailed from Nottingham University it seems appropriate that this spectral feature be designated ìNottingham Notchœ !). The explanation for the notches in the scheme in Fig. 1 relies upon the fact that the narrow spectral portion of the Ly-a radiation that is exactly resonant with the ìDoppler coreœ of the B0-2R1 transition can in principle be directly absorbed in three-photon transitions originating from JA\1 of X0 and terminating on J@\2 of B0.This would in turn leave less Ly-a radiation at these frequencies available for feeding virtual quantum levels of the J@\2 B0 intermediate state in the inverse Raman absorption mixing processes proposed in the text as the sources of the k5797 and k6379 DIBs. A direct consequence of this diminished Ly-a —ux resonant with the H B0-2R1 Doppler core should be the appearance of notches in the spectral pro–les of these DIBs. According to this scheme the width of the ìNottingham 2 Notchœ in any line-of-sight provides an accurate measurement of the H Doppler width in the cloud in which the DIBs are actually produced. 2 The intrinsic pro–les of most narrow DIBs that have been carefully studied are devoid of sharp spectral structure.An example is the very strong narrow DIB at 7224 ”. We have very recently assigned this DIB to the inverse Raman process shown in Fig. 3. In Fig. 3 the ca. 10.75 lm broadband IR radiation which ìfeeds œ virtual quantum levels around J@\5 of B11 is considered to be one of the two waves generated in a simultaneous two-photon emission process [EF7 J\4]][EF4 J\4]. Since the pro–le of this IR radiation should be spectrally smooth the spectral pro–le of the 7224 ” inverse Raman absorption band should also be smooth as is observed.1 Also shown in Fig. 3 is the same transition to which the strong DIB at 6284 ” was assigned in the text. However the optical pathway shown in Fig. 3 feeding the quantum level EF4 J\4 seems more convincing to us than the one proposed in our paper (see Not.R. Astron. Soc. 1995 277 L41). Fig. 2 k5797 DIB spectral pro–les from ref. 13 of our paper (reproduced with kind permission from P. J. Sarre J. R. Miles T. H. Kerr R. E. Hibbins S. J. Fossey and W. B. Somerville Mon. 219 General Discussion Fig. 3 Proposed assignment for k7224 DIB Fig. 23). Also shown in Fig. 3 are the transitions to which we have recently assigned two more DIBs. All four DIB assignments shown in Fig. 3 again illustrate a main theme that has evolved from our Faraday Discussion paper namely if a detectable inverse Raman absorption is to occur on a given H inter-Rydberg transition a ìpush»pull œ situation must prevail as a necessary condition. If the lower level of the transition happens to be a 2 gerade quantum level there must both be present a coherent light wave strongly populating this level and a broadband coherent light wave spectrally centered about a downward transition from the upper level.If the lower level is an ungerade quantum level there must exist both a broadband coherent light wave ìfeedingœ virtual quantum levels surrounding this level as well as another coherent light wave actually originating from the upper level. 1 Herbig and Soderblom Astrophys. J. 1982 252 610. Dr Ubachs said Following up on the laboratory observation of coincidences between pronounced inter-Rydberg transitions in H with –ve diÜuse interstellar bands shown in Fig. 5 of our paper we present some preliminary results from a multichannel 2 quantum defect theory (MQDT) calculation performed with E.Reinhold Ch. Jungen and S. C. Ross. As shown in the energy diagram of Fig. 4 here the –ve coincident resonances originate from the J\1(f) J\1(e) and J\2(e) levels of the C1%u v\5 and v\6 vibronic states which couple via strong transitions in the XUV-domain to the lowest para- and ortho-levels in the X1&g ` v\0 electronic ground state of molecular hydrogen. Values for C-state energies are obtained from a high-resolution study of Reinhold et al.1 As for the terminal levels of the two-photon absorption only one autoionizing level for both para- and ortho-hydrogen is involved (see Fig. 4). In the MQDT calculations it is established that the level in para-hydrogen at 124 701.9 cm~1 (above X1&g ` v\0 J\0) is a J\0(e) level of (]) parity while the 220 General Discussion does not coincide with a documented DIB.2 Fig. 4 Energy level diagram not to scale pertaining to the –ve coincidences between observed inter-Rydberg transitions in H and DIBs. A sixth resonance originating in v\5 J\1(e) C1%u level in ortho-hydrogen at 124 752.5 cm~1 is a J\1(e) level of ([) parity. The terminal levels are Rydberg states predominantly consisting of an X2&g ` v\6 core with a 3s Rydberg electron. Also some 3ds v\7 character is mixed in. The MQDT calculations con–rm that for both para- and ortho-hydrogen there exists only one strong resonance in a frequency span of 1000 cm~1 namely the ones coinciding with the resonances observed in the double-resonance laser excitation study.The oscillator strength of transitions originating in C1%u v\6 is found to be very high. These two conditions strong and lone transitions from a speci–c state are of importance for an explanatory model for the DIBs. 1 E. Reinhold W. Hogervorst and W. Ubachs J. Mol. Spectrosc. 1996 180 156. Prof. Sarre said We have just published1 high-resolution and high signal-to-noise spectra of the k5797 diÜuse interstellar band recorded along three lines of sight using the AAT and KPNO telescopes (see Fig. 5). This has revealed a considerable amount of structure including clear evidence for a narrow absorption ìspikeœ near 5796.9 ”. The wavelength of the molecular hydrogen transition quoted by the authors is 5796.96 ” which is close to our reported rest wavelengths of 5796.942 and 5796.945 ” towards l Sgr and f Per which are deduced by reference to the K I line.Under the current proposal of molecular hydrogen as a carrier of diÜuse bands the entire k5797 pro–le including its shape width and –ne structure arises from a single R(2) rovibronic line. My suggestion is that the problem is turned around and rather we should consider what is established and use this to predict the spectrum which would be expected for molecular hydrogen bathed in UV light to provide the initial excitation with subsequent absorp-221 General Discussion Fig. 5 Ultra-high-resolution spectra of k5797 (a) toward l Sgr (AAT) and f Per (KPNO) and (b) toward f Oph (AAT). The vertical line indicates the position of the ìspikeœ.Reproduced with permission from T. H. Kerr R. E. Hibbins S. J. Fossey J. R. Miles and P. J. Sarre Astrophys. J. 1998 495 941. tion of visible photons. This predicted spectrum can then be sought through observations at high-sensitivity and high-resolution. It is conceivable that the sharp ìspikeœ might be due to H as this represents only a very small part of the k5797 band but at present it is hard to see how the rest of the observed band could arise from this tran- 2 sition in molecular hydrogen. 1 T. H. Kerr R. E. Hibbins S. J. Fossey J. R. Miles and P. J. Sarre Astrophys J. 1998 495 941. Dr Sorokin Dr Glownia and Dr Ubachs commented After struggling to –nd a reasonable explanation for the ìNottingham Notchœ we learned that Prof. Sarre and colleagues have very recently reported new re–ned measurements of the k5797 DIB pro–le1 which show the detailed spectral structure of this DIB to be rather diÜerent from that described in their previous study.The main ìnotchœ (feature b in Fig. 2 of their new study) is still present but it appears shifted to slightly longer wavelengths (ca. 5797.0 ”) in the rest frame spectrum adopted in the new study. Immediately bordering the main notch on the high-frequency side there now appears a remarkably sharp ìspikeœ of increased DIB absorption (feature 0 in Fig. 2 of their new study). There are also evident smaller notches (i.e. decreased DIB absorption) at a and c. We have tried to think of suitable explanations for this newly revealed spectral structure based upon the scheme shown in Fig.1. One possibility is that the trapped Ly-a continuum in the cloud has had small ì bites œ taken out of it by impurity atoms present somewhere between the H2-containing cloud and the bright illuminating star. Such ìabsorptionsœ (i.e. elastic scatterings) would have to occur very close to the B0-2R1 frequency 82 155.21 cm~1 (1217.208 ”). A check of simple atoms such as OI NI CI etc. reveals no lines in this vicinity. None of the tabulated absorption lines of CO were also found to match. The fact that the Nottingham group observed almost identical spectral pro–les for the k5797 DIB in three diÜerent lines-of-sight (with the use of two widely separated telescopes) even with regard to the sharp features just mentioned strongly argues against such an ìimpurityœ eÜect.One might think generally that if the spectral structure is intrinsically due to molecular hydrogen there necessarily would have to exist other H transitions located very 2 222 General Discussion near to 82 155.21 cm~1. A check of ref. 28 in our paper presented at this meeting reveals that there are none. However there remains one H2-based explanation which appears to be able to account for both the main ìnotchœ at ” ca. 5797.0 and the sharp ìspikeœ at ca. 5796.94 ”. This is to assume that together these features constitute what is known in spectroscopy as a Fano pro–le. In 1935 Fano2 –rst interpreted some discrete structure in the absorption spectra of Ar and Kr above the ionization potential as being due to the interaction of discrete states and an underlying ionization continuum.Viewing Fig. 1 one realizes that in the present case broadband Ly-a radiation can also either be absorbed in a transition to a ì discrete œ state (i.e. B0 J@\2) or in a ìcontinuumœ that spectrally surrounds the transition B0-2R1. The cross-section of this continuum would be proportional to the product of light intensities existing in the cloud at ca. 17 245 cm~1 ca. 964 cm~1 and ca. 610 cm~1. Typically a Fano pro–le shows an increase in absorption above the continuum level on one side of the discrete state resonance and a (non-symmetrical) decrease in absorption below the continuum level on the other side of the resonance. A spectral feature of this type present in the Ly-a absorption crosssection would produce a similar feature (reversed in sign with respect to wavelength) in the DIB spectral pro–le.We note that the air wavelength of the inter-Rydberg transition between EF12 J\3 and B0 J@\2 is 5796.96 ” in between the main ìnotchœ and the ìspikeœ in the newly measured k5797 DIB pro–le as one would expect from such a model. The above interpretation implies of course that a similar sharp Fano resonance structural feature must also be present in the spectral pro–le of the k6379 DIB. To our knowledge no truly very high resolution spectral measurements on this DIB have yet been performed although in both ref. 14 and 15 of our paper presented at this meeting the presence of a de–nite ìnotchœ on this DIB was recorded. In Fig. 6 here we show our assignment for yet another DIB the one at 6376 ”.From the general argument here being presented it would appear that this narrow DIB should likewise display –ne structure bearing a Fano pro–le of some kind. However since the equivalent width of the k6376 DIB is roughly –ve times less than that of the k5797 DIB very high-resolution spectral measurements may be difficult to perform here. Fig. 6 Proposed mechanism for k6376 DIB General Discussion 223 There is actually a severe intensity anomaly connected with the k6376 assignment shown in Fig. 6 which is instructive to discuss. Despite the fact that the calculated radiative transition strength of B1-2P5 is twice that of B0-2R1 that its oÜset from the peak of the trapped Ly-a radiation spectrum is –ve times less that the upper state for the inter-Rydberg transition to which the k6376 DIB is assigned is the same as for the k5797 DIB and that the calculated ATM for the (EF12 J\3)7(B1 J@\4) transition is about six times greater than the calculated ATM for the (EF12 J\3)7(B0 J@\2) transition the equivalent width of the k6376 DIB is as noted above roughly –ve times less than that of the k5797 DIB.A possible explanation for this anomaly is provided by assuming that the coupling between the trapped Ly-a radiation and the transition B1-2P5 is strong enough so that the Ly-a radiation is able to excite a broadband SRS process from the X2 JA\5 quantum level to some other X-state quantum level. This would lead to decreased absorption of ” ca. 6376 photons via the inverse Raman absorption process shown in Fig.6 since a fraction of the Ly-a photons necessary for this process would be utilized in pumping the SRS process. In addition the eÜect analyzed in ref. 19 of our paper namely the drastic reduction in two-photon absorption intensity that can occur in a three-level system when the intermediate state is radiatively coupled to a fourth state might also play a role here. At present the only evidence we have of the quantum level (X2 JA\5) being involved in an n-wave mixing process is the intense VUV emission band dominating the 1062 ” »1080 spectral region in the high-resolution Copernicus b Ori A spectrum of ref. 27 in our paper. This band is the same band assigned as B9-2R1 in Fig. 12 and 14 of our paper. However the high-resolution spectrum of ref.27 de–nitely shows the peak occurring at 1078.25 ” not at the wavelength (1078.40 ”) of B9-2R1. The only nearby H ” ” and we have re-assigned the 1078.25 peak to transition is B10-2R5 at 1078.27 this transition. Again this illustrates the necessity for having extreme wavelength accu- 2 racy in seeking to interpret astrophysical data with the H DIBs model. On the basis of ref. 27 we have already had to change many of the VUV emission band assignments 2 made in our paper. However although we seem to be continually proposing new excitation mechanisms for the DIBs the actual assignments we have made of many narrow DIBs to H inter-Rydberg transitions have not changed that much. To us this is simply a re—ection of the accuracy and precision with which astronomers have characterized 2 the total DIB spectrum.It remains a remarkable template for evaluating astrophysical theories. 1 T. H. Kerr R. E. Hibbins S. J. Fossey J. R. Miles and P. J. Sarre Astrophys. J. 1998 495 941. 2 U. Fano Nuovo Cimento 1935 12 156. Dr Schutte said Since ISO and IRTF spectroscopy of the diÜuse galactic emission in the mid-infrared region of the spectrum has become possible. These data show the well known speci–c UIR emission bands at 6.2 7.7 8.6 and 11.2 lm with no indication of additional features.1 The UIR features have been identi–ed with various vibrational modes of polycyclic aromatic hydrocarbons.2 Where would the energy absorbed by H2 and/or carbon-chains in the visual and possibly ultraviolet part of the spectrum be reemitted ? Would carbon chains show speci–c infrared features corresponding to the characteristic vibrational modes of the chain structure ? Would the presence of the H /carbon chain emission still be consistent with the ISO observations ? 2 1 K.Mattila D. Lemke L. K. Haikala et al. Astron. Astrophys. 1996 315 L35; T. Onaka I. Yamamuna T. Tanabeç T. L. Roellig and L. Yuen 1996 Publ. Astron. Soc. Pac. 1996 48 L59. 2 A. Leç ger and J. L. Puget Astron. Astrophys. 1984 137 L5; L. J. Allamandola A. G. G. M. Tielens and J. R. Barker Astrophys. J. 1985 290 L25. Prof. Maier responded The UIR features are the result of an overlap of IR transitions of many species. Carbon chains would contribute like other molecules to the IR 224 General Discussion emission.Our study of the electronic spectrum of C7~ in the visible has concerned itself with merely one such chain but numerous other including hydrogen nitrogen and oxygen containing derivatives would also contribute to IR emission. Prof. Thaddeus said I have only one comment to make on the wonderful work of Prof. Maier and his group it is too good to be wrong. In the long literature on the interstellar diÜuse bands no identi–cation has been made which is remotely as convincing as the assignment here of C7~. Dr Schutte has asked where the energy absorbed by a DIB ends up»how is it rerradiated ? It is rerradiated by the entire vibrational spectrum of a carbon chain but mainly by the bond stretches CwC and CwH and so blends into the fuzz of the UIR bands between 3 and 8 lm.Dr Sorokin Dr Glownia and Dr Ubachs commented Our current H DIBs model 2 does indeed predict coherent generation of broadband (10»100 cm~1 bandwidth) VUV and IR radiation at several frequencies. This generation is viewed as occurring in each of the (two-photon) steps comprised by the 2n-wave mixing processes discussed in our paper. For example the four-wave mixing process shown in Fig. 18 of our paper would predict generation of coherent IR light at ca. 6.2 lm occurring as Stokes-wave emission in a broadband stimulated Raman scattering (SRS) process pumped by incoherent VUV light spectrally centered on the transition C4-0R0 at 929.5 ”. As explained at the beginning of our paper one expects an enormous enhancement of the VUV photon density that is nearly resonant with C4-0R0 to result from the complete dominance of elastic (Rayleigh) scattering over inelastic scattering (i.e.ìabsorptionœ) in the cloud of our model. The wave mixing process shown in Fig. 18 also predicts generation of coherent broadband IR light in a band centered at ca. 8.4 lm and generation of a similar bandwidth of coherent VUV light spectrally centered on the transition B13-0R0 at 954.4 ”. In the process shown in Fig. 18 absorption of a photon from the incoherent VUV pumping –eld and coherent emission of the three photons that are produced are to be visualized as occurring simultaneously with total photon energy and total photon momentum being rigorously conserved in the overall process. The latter condition can be satis–ed only if the four waves in Fig.18 are co-propagating collinearly. The scheme shown in Fig. 18 of our paper was suggested by the apparent presence of an H2 VUV emission peak at ” ca. 954.4 (see Fig. 19) and by the favorable calculated transition strengths shown in Fig. 18. A subsequent more careful examination of the spectrum shown in Fig. 19 revealed that the peak actually occurs at 954.47 ” and that it is apparently necessary to correct all wavelengths in this range by adding ca. 0.07 ”. This would position the peak at ” ca. 954.54 (104 763 cm~1)»about 14 cm~1 less than the B13-0R0 frequency (104 777 cm~1). The correct assignment of the emission peak here being considered could thus more likely be the transition D0-2R3 with frequency at 104 763 cm~1.(The strong transition B25-2R3 is oÜset by only 7 cm~1 from Ly-a which would allow eÜective depopulation of X2 JA\3). In this case the scheme of Fig. 18 becomes irrelevant. This example illustrates once more the crucial need for obtaining astrophysical spectra that possess high signal-to-noise ratios are taken at the maximum spectral resolutions that are technically available and are precisely calibrated in terms of rest frame wavelengths. Of the archived VUV spectra analyzed in our paper the highresolution Copernicus VUV spectrum of b Ori A presented in ref. 27 turns out to be superior to all the others in these three respects. The one serious drawback of the archived Copernicus spectra we have examined is the fact that the detectors evidently ceased to respond at wavelengths shorter than ca.1000 ”. In the text of our original H DIBs paper1 we went to extremes in ìassigningœ most of the known UIR (UIB) emission bands as Stokes-waves in H SRS processes that a 2 priori would be expected to have strong transition probabilities. However in the ìNote 2 225 General Discussion Added in Manuscriptœ included in the same paper we subsequently strongly cautioned the reader against accepting these assignments. At the present time we simply must conservatively speculate that there is perhaps some component of the UIR emission band spectrum that is observed in some lines-of-sight that may be due either to H SRS Stokes-wave generation or to parametric oscillation (the second two-photon step shown 2 in the scheme of Fig.18). Wavelengths of infrared emission bands that might plausibly be generated with the H DIBs model can be seen in several of the other –gures contained in our Faraday Discussion paper or in our earlier comment about our own 2 paper. From Fig. 15 of our paper one would predict broadband IR generation at 10.86 lm and 3.47 lm. From Fig. 21 IR generation at 8.45 lm is predicted. From the scheme in Fig. 24 IR generation at 9.64 lm seems most likely although generation at 11.88 lm is also predicted. From Fig. 3 earlier IR generation at 13.4 10.87 10.75 and 8.70 lm is likely. To try to –nd any real evidence of such broadband H IR emissions one might consider carefully probing lines-of-sight to unreddened bright young stars. 2 1 P. P. Sorokin and J.H. Glownia Astrophys. J. 1996 473 900. Dr Somerville communicated Astronomers are not irrationally prejudiced against the H model for diÜuse interstellar bands but concerned about serious astrophysical implications of the model of which the authors seem not to be aware. One difficulty with 2 a circumstellar model for the DIBs pointed out before is that DIB strengths are rather well correlated with the interstellar dust abundance as measured by the reddening E(B[V ). Sorokin has responded to this with an ingenuity that can only be admired by arguing that because the DIB carrier is circumstellar so must the dust be actually producing scattering of starlight that leads to DIB formation so that as observed DIB strength increases with the amount of dust.The problem with this is that it is not only the diÜuse bands that are correlated with E(B[V ) but so is everything else associated with what we call the diÜuse interstellar medium along the line-of-sight including atoms as well as other molecules. All the evidence indicates that the gas and dust are well mixed»not just correlations but for example the observed depletions of gas-phase atoms onto the dust grains. Properties of this gas are not consistent with it being bathed in a strong circumstellar ultraviolet —ux. It is of low density clumped into clouds and must be distributed in space. Several strong DIBs including k4428 5797 5780 and 6283 correlate well with the dust and cannot therefore be primarily of circumstellar origin. However correlation studies have been carried out only for a relatively small number of the strongest features and the diÜuse bands do not vary all together.1 It is therefore not impossible that some of the weaker absorptions observed so far in a few sightlines may be produced only in circumstellar regions.1 W. B. Somerville in T he DiÜuse Interstellar Bands ed. A. G. G. M. Tielens and T. P. Snow Kluwer Dordrecht 1995 p. 83. Dr Sorokin Dr Glownia and Dr Ubachs communicated In our H nonlinear DIBs model it appears no longer necessary to assume that the H -containing cloud is circum- 2 2 stellar. If a thin ìplanarœ cloud containing at least ca. 104 cm~3 H molecules and as many H atoms were located ca. 10 pc from an O- or B-type star the enhancement in the 2 near-resonant VUV photon density might well be high enough for sufficient SRS Stokeswave gain to develop in some direction lying in the ìplaneœ of the cloud for coherent lightwave generation to occur particularly if the cloud subtended a large solid angle about the star.Such a geometry is not totally unlike some of the models that were considered for the H -containing cloud seen toward f Oph see for example ref. 1. In the line-of-sight to this star velocity shift measurements show that it is possible to associate 2 all of the observed absorption lines of known species with one (or more) of six diÜerent 226 General Discussion velocity clouds. It thus would appear that towards f Oph at least the interstellar gas can hardly be viewed as being ìwell mixedœ. 1 D. C. Morton Astrophys.J. 1975 197 85. Prof. Krelowski commented It seems necessary to create good criteria for the identi- –cation of interstellar features. This task requires a close co-operation between experimentalists and observers. It is apparently impossible to compare astrophysical and matrix spectra as inside any possible wavelength inaccuracy caused by the matrix shift we observe up to several tens of unidenti–ed features. It is interesting to observers whether they should expect similar pro–les and/or intensity ratios of the features originating in one even very complicated molecule. Concerning the proposed sequence of the C7~ chain the observed features do not obey the above-mentioned criteria they barely match the wavelengths. Also the violet/ blue features from the other electronic transition of C7~ are not observed.This creates doubts concerning the proposed identi–cation. Prof. Maier responded There seems to be a misunderstanding. We are not comparing astrophysical and matrix spectra. We present a gas-phase electronic spectrum of C7~ measured in the laboratory at rotational temperatures below 100 K (see Fig. 5 of our paper) and compare the band wavelengths with the DIB values given in the compilation of Jenniskens and Deç sert.1 These wavelengths match very well as can be seen in Table 3 of the paper. The error limits given for the laboratory measurements (^0.2 nm) are deliberately conservative»the band maxima have been determined to ^0.05 nm. The reason for this is stated in the paper»the band maxima will shift with temperature.The intensities in the gas-phase spectrum presented are not reliable because of saturation by the lasers (leading to band broadening). The absorption spectrum of C7~ measured in a neon matrix gives a more realistic impression of the relative intensities. located at 493 nm in the gas-phase but all the bands are much broader (2»3 nm) than C7~ does show another transition to the blue B 2%»X2%. The origin band is for the A2%»X2% transition due to intramolecular processes. These corresponding DIBs would be comparably broad and most of them happen to lie in regions where strong stellar or telluric lines are present ; i.e. origin band near 493 nm 30 1 near 480 nm 10 1 near 448 nm. 1 P. Jenniskens and F-X. Deç sert Astron.Astrophys. Suppl. 1994 106 39. Dr Halasinski communicated The gas-phase experimental measurements reported by Prof. Maier for carbon chains represent an exciting step in the experimental approach to the DIB problem. The relatively small shift measured between the neon matrix and the gas-phase spectrum of C7~ is very encouraging. It tends to support the recent consensus on carbon-bearing molecules and ions as carriers of the DIBs. We would like to add that in addition to the proposed carriers discussed in this session several PAH cations have been proposed as possible carriers for the DIBs as has been amply discussed in the literature.1,2 The comparisons in Table 1 made in neon matrices do provide as good a match to several DIBs as those of the carbon chains in neon.3 Thus we wish to point out that in view of the experimental results reported by Prof.Maier gas-phase PAH cations are also expected to provide the same quality of match with some of the DIBs. 1 F. Salama C. Joblin and L. J. Allamandola Planet Space Sci. 1995 43 1165. 2 F. Salama E. L. O. Bakes L. J. Allamandola and A. G. G. M. Tielens Astrophys. J. 1996 458 621. 3 P. Freigovel J. Fulara and J. P. Maier Astrophys. J. 1994 431 L151. Dr Krelowski said Large molecules are very likely to produce spectra with one strong and several much weaker features. This general picture may –t the observational Table 1 Comparison of DIBs with PAH cation bands PAHs isolated in neon matrices (reprinted from Planetary Space Sci. 1995 43 1165 with permission from Elsevier PAH` pyrene (C16H10 `) 1-methylpyrene (CH3wC16H9 `) 4-methylpyrene (CH3wC16H9 `) naphthalene (C10H8 `) phenanthrene (C14H10 ` ) tetracene (C18H12 `) benzo(ghi)perylene (C22H12 `) coronene (C24H12 `) results as we observe about 20 strong DIBs accompanied by hundreds of very weak features.We managed (Galazutdimov Fulara Musaev and I) to extract a sequence of features (6614 6520 6150 and 5923) ; the –rst one is strong the others very weak. Their intensity ratios and pro–les are always the same. It may be the spectrum of one carbon chain. Concerning the H hypothesis (1) The 5797 and 6379 DIBs assumed to share the same ground level are of variable intensity ratio (HD 183143 vs. HD 41117); (2) DIBs 2 are observed in spectra of F G super-giants at high galactic latitudes.227 General Discussion Science). DIBs/nm jpeak/nm 442.9 439.5 (443.0 in Ar) 442.9 444.2 482.4 758.1 (457.7) 482.8 757.6 674.1 652.0 674.2 652.0 857.2 898.3 856.8 864.8 864.7 503.9 ( ?) 758.1 ; 758.6 755.8 ( ?) ; 756.2 793.5 (prob.) 502.2 758.4 755.2 794.3 459.5 946.6 459.0 946.5 228 General Discussion Dr Sorokin Dr Glownia and Dr Ubachs commented On the basis of the H DIBs 2 theory an observed variable intensity ratio for the k5797 and k6379 DIBs might be explained in the following way (see Fig. 1 earlier). A variable intensity ratio would imply that the coherent lightwave emission step originating from (EF12 J\3) follows a path that is separate from the one by which (EF10 J\1) is populated.In the –gure for simplicity we have shown these paths to be the same. In the ìpush»pull œ schemes for each of these DIBs only the ìpushœ component is evidently the same! Dr Lynas-Gray commented Sorokin and Glownia1 match the wavelengths of over 70 diÜuse interstellar bands with those of inter-Rydberg transitions in H2 ; these are resonantly enhanced simultaneous two-photon absorption from an early-type star by H2 in a tenuous nearby cloud. Sorokin et al.2 attempt to substantiate the Sorokin and Glownia model1 by identifying ìemissionœ features in Copernicus and ORFEUS-1 ultraviolet spectra of early-type stars with both simulated Raman scattering (SRS) and fourwave parametric oscillations (FWPO) in H2 .Some of the identi–cations suggested by Sorokin et al.2 in the reddened B5 Ia supergiant g CMa are indicated by vertical bars in Fig. 7 where part of the scaled Copernicus spectrum of g CMa is plotted as a thin line ; from left to right the H SRS and FWPO features which Sorokin et al. advocate are B9-2R1 B9-2R2 B4-7P1 B9-2P2 B12-3P1 B8-2R1 B8-2R2 B8-2P2 B16-4P4 B8-2P4 2 B7-2R1 B7-2P1 B0-0R0 & B0-0R1 and B0-0P1. Cassinelli et al.3 make a detailed study of the B2 II giant e CMa; part of the Copernicus spectrum for this star is also shown in Fig. 7 as the heavy line ; it can be seen that the SRS and FWPO features which Sorokin et al.2 claim to identify in g CMa are also present in the spectrum of e CMa. At 600 ” e CMa is the brightest star in the sky and Cassinelli et al.3 establish a remarkably low neutral hydrogen column density of about 1]1018 cm~2.In the essential absence of neutral hydrogen it is unlikely that an H2 cloud could form in the vicinity of e CMa and the labeled features in the –gure are consequently unlikely to be SRS and FWPO transitions in H2 ; because these also occur in g CMa and other stars studied by Sorokin et al.,2 their identi–cations must be in doubt. It is of course well established that line blanketing in the ultraviolet spectra of early-type stars means that only absorption features are observed; this is absolutely essential to any attempt to model the entire energy distribution as Cassinelli et al.3 discuss in the case of e CMa. 1 P. P. Sorokin and J.H. Glownia Astrophys. J. 1996 473 900. Fig. 7 Comparison of Copernicus spectra for g CMa (thin line) and e CMa (thick line) in 1070 Aé \j\1120 Aé . Further details are given in the text above. 229 General Discussion 2 P. P. Sorokin J. H. Glownia and W. Ubachs Faraday Discuss. 1998 109 137. 3 J. P. Cassinelli D. H. Cohen J. J. MacFarlane J. E. Drew A. E. Lynas-Gray M. G. Hoare J. V. Vallerga B. Y. Welsh P. W. Vedder I. Hubeny and T. Lanz Astrophys. J. 1995 438 932. Dr Sorokin Dr Glownia and Dr Ubachs responded Viewed together as a whole the comments by Dr Lynas-Gray would appear to be very damaging to our H DIBs model which strongly relies upon the assumption that coherent oscillations produced in 2 various wave mixing processes driven by broadband SRS transitions are occurring in H -containing clouds near bright young stars.In our reply we consider the two main thrusts of his arguments separately. 2 We certainly agree with Dr Lynas-Gray that the EUVE spectroscopy measurements of the B2 II giant e CMa to which he refers in his comment (Cassinelli et al.) indicate a remarkably low H-atom column density of about 1]1018 cm2 in the line-of-sight to this star and that consequently it is unlikely that an H cloud could form in the vicinity of this star. However we diÜer with Dr Lynas-Grayœs conclusion that the VUV 2 ìemission bandsœ seen in a low-resolution Copernicus VUV spectrum of e CMa are exactly the same as those we assigned as emission bands coherently generated by SRS and FWPO processes in an H cloud near the B5 Ia supergiant g CMa one of the mid-/late-B supergiants considered in our Faraday Discussion paper.The evidence for 2 Dr Lynas-Grayœs conclusion is based upon a comparison of the low-resolution Copernicus VUV spectra of these two stars which are shown superimposed in the Fig. 7. We refer now to this –gure. The vertical bar on the left hand side marks a very prominent ìemission bandœ seen in the four mid-/late-B supergiant spectra considered in our paper. However as explained in our reply to the comment by Prof. Sarre on the basis of the high-resolution Copernicus b Ori A spectrum shown in ref. 27 we now assign this band to the transition B10-2R5 rather than B9-2R1. From Fig. 7 shown by Dr Lynas-Gray it is not at all apparent that at this wavelength there is any corresponding e CMa emission band rising signi–cantly above the continuum level for that star.Interestingly enough in neither the high-resolution Copernicus VUV spectrum of the B2 IV star c Peg,1 nor in the highresolution Copernicus VUV spectrum of the B3 IV star i Herculis2 is there likewise any sign of emission at 1078.27 ” rising signi–cantly above the continuum level. In our paper we focussed almost entirely on the VUV spectra of B-type supergiants because these were the ones that most clearly seemed to show evidence of emission bands. Corresponding to the position of the fourth vertical bar drawn in Fig. 7 is a somewhat less intense emission band that we now assign as D2-7P4 (1081.43 ”) rather than as B9-2P2 (1081.51 ”) since the band appears in the high-resolution Copernicus spectrum of b Ori A at 1081.45 ”.There appears to exist an emission band that slightly exceeds the continuum level at this wavelength in the e CMa spectrum shown in Dr Lynas-Grayœs –gure but it appears to be much broader than the g CMa band. When one examines the high-resolution spectrum of c Peg at ca. 1081.45 ” one sees only a remarkably —at continuum emission with almost no stellar or interstellar absorption lines. In i Her there is likewise no emission above the continuum level at this wavelength. One should also be aware of the fact that very strong stellar absorptions de–- nitely are seen in both ” ” e CMa and g CMa at ca. 1077.2 (S III) ca. 1084.0 (N II partially interstellar) and ca. 1084.6 ” (N II partially interstellar).These strong common absorptions have the eÜect of making the two superimposed spectra more similar in appearance. The position of the sixth vertical bar in the –gure marks one of the very strongest emission bands appearing at ” ca. 1090.84 in the high-resolution spectrum of b Ori A shown in ref. 27. We have re-assigned this band from B8-2R1 (1090.75 ”) to B20-5R3 (1090.84 ”). No especially prominent emission in excess of the continuum occurs here in the case of e CMa. In this wavelength region the spectrum of c Peg is even quieter, 230 General Discussion showing only modest absorption lines on a relatively —at continuum. No emission above the continuum level is seen at this wavelength in i Her. Similar comments apply to the coincidences marked by the other vertical bars.We conclude that the VUV spectrum of e CMa does not show the signature characteristics that we have interpreted in our paper as arising from H coherent emission. For reasons unknown to us these characteristics seem only to be found in B-type supergiants and 2 some B0 giants. The –gure included in Dr Lynas-Grayœs comment does aÜord us an opportunity to make a correction to our paper. The spectra of both stars shown in this –gure display very strong stellar absorption bands at ” ” ca. 1108.3 (S II Si III) ca. 1110.0 (Si III) and ca ” . 1113.2 (Si III). We have come to realize that these three bands are pervasive in the VUV spectra of B-type stars and that two of them must undoubtedly mostly account for the absorptions labeled B0-0R0 B0-0R1 and B0-0P1 in Fig.7 of our paper. The argument for non-linearity presented at the beginning of Section 4 of our paper had been based upon the deduction made from a comparison of Fig. 7 and 8 that the observed absorption strengths of the B0-0 components relative to those of B4-0 were much greater than they theoretically should be. By comparing the B4-0 components with those of B1-0 (Fig. 8) the argument for non-linearity can still probably be made since there are no known stellar absorptions overlapping the latter and the calculated B4-0 transition probabilities are roughly four times those of B1-0. 1 J. B. Rogerson Astrophys. J. Suppl. Ser. 1985 57 751. 2 W. L. Upson and J. B. Rogerson Astrophys. J. Suppl. Ser. 1980 42 175.Prof. Snow said I would like to suggest that we should hear nothing more about the hypothesis of DIB formation by non-linear radiative processes in molecular hydrogen until the following astronomical problems have been successfully addressed (1) The DIBs do not exhibit dependence on stellar type ; even late B stars have strong bands» this argues against any dependence on radiation –eld intensity. (2) DIBs correlate with origin. (3) The DIBs tend to be weak not enhanced in regions with high FUV radiation distance and other IS parameters such as EB~V which argues against any circumstellar –elds which argues against radiative processes in general. (4) The DIBs do not correlate well with H2 ; the correlation with atomic hydrogen is better»this argues against an origin related directly to molecular hydrogen.(5) The two-photon process is physically unrealistic radiative relaxation of excited states of H is far more rapid (about 10~9 s timescale) than the photon arrival time in any realistic environments (photon arrival 2 times are ca. 1 per year in the diÜuse ISM a factor of ca. 1016 too slow to support this model!)»this eÜectively rules out any two-photon process even in circumstellar environments. (6) Copernicus observations show little or no H in excited states yet there should be a large equilibrium population of excited H in the two-photon model. 2 (7) The expected ultraviolet emission lines from circumstellar nebulae have not been 2 observed»this argues strongly against the photon-trapping hypothesis that has been invoked to explain how the UV radiation –eld could be sufficiently intense to satisfy the requirements of this model.(8) DIBs have been detected in lines-of-sight toward cool stars (spectral types F and G; I refer here to a poster presented at this meeting by Drs Zacs and Schmidt). This is probably fatal to the entire hypothesis because these lines-ofsight have virtually no UV photons available to excite H2 . The suggestion that the DIBs might be formed by non-linear optical eÜects in molecular hydrogen has been an interesting and worthwhile contribution but unfortunately further examination of the idea has found a number of very serious objections and inconsistencies represented by the list above. It is now incumbent upon the authors of the hypothesis to withdraw from further promotion of the model unless and until they can provide realistic responses to this list of challenges.231 General Discussion Dr Sorokin Dr Glownia and Dr Ubachs responded As a scientist who has long been actively involved with both astronomical observations and laboratory experiments aimed at revealing the nature of the diÜuse interstellar band (DIB) carriers it is natural for Prof. Snow to demand that we try to provide some type of rational argument to show that the unusual non-linear lightwave mechanisms [i.e. broadband stimulated Raman scattering (SRS) and four-wave parametric oscillation (FWPO)] we invoke in our H DIBs model could in principle occur in an astrophysical environment. Since Prof. Snow is here primarily concerned with there being orders-of-magnitude too little 2 VUV pump light to drive the non-linear mechanisms at the (0.01»10 pc) cloud-to-star distances assumed in our model we will begin our reply with a simple argument which we believe points the way out of this dilemma.We will then try to address some of Prof. Snowœs speci–c questions. n per cm3. Consider a thin spherical shell region bounded by two spheres with nearly equal radii that are centered about a bright star. (For the sake of de–niteness let us assume that the shell has a radius ca. 1 pc and a thickness ca. 0.01 pc.) Let *l de–ne a small bandwidth of VUV light centered around let us say the Ly-a resonance line of H atoms. Designate by S the total number of photons emitted by the star per second in *l the bandwidth *l.We will calculate the steady state values of n(r) the number of photons per cm3 in the bandwidth *l at a radius r lying within the shell for two separate cases (a) an empty shell and (b) a shell that is uniformly –lled with H atoms at a density H In the –rst case it is trivial to show that na(r)\s*l/4pr2c. For the second case let us start by assuming a value for n of 104 per cm3. Since an atom in such a gas would collide with another one at an average rate of ca. 1 collision every two years we assume H this gas to be a collisionless medium. For VUV photons having frequencies that lie within *l one can entirely neglect inelastic scattering (i.e. ìabsorptionœ). All scattering of these photons is elastic (i.e. resonance Rayleigh scattering).Therefore the outward —ow of such VUV photons through the thin spherical shell will be governed by Fickœs –rst law of diÜusion. At a distance r lying within the shell one has (1) [4nr2D dn dr\S*l where D is the diÜusion coefficient equal to (1/3)gc. Here g is the mean free path equal to 1/J2 pnH . The quantity p is the elastic scattering cross-section. For simplicity we take it to be the maximum value it can have j2/2p. It thus follows from eqn. (1) that within the spherical shell (2) nb(r)\ S*l 4prD For j\1216 ” the mean free path g is ca. 3]106 cm. Since the nearly resonant n photon density enhancement is given by b(r)/na(r)\3r/g at r\1 pc this ratio is ca. 3]1012! The biggest signi–cance of eqn. (2) is probably the fact that the nearly resonant VUV photon density varies as 1/r not as 1/r2.This means that the Stokes-wave gain over a full great circle path in a spherical shell of H gas would be independent of the radius of the shell. Thus the thin H -containing cloud of our model is really not required to be 2 extremely close to the illuminating star. The total Stokes-wave gain existing in a path 2 comprising a complete great circle in a thin gas-–lled spherical shell of radius 10 pc would be the same as that existing in a great circle path in a shell with radius 1 pc for example. It should be possible actually to calculate the gain for resonantly enhanced broadband IR or VUV Stokes-wave generation over a great circle for the case represent-232 General Discussion ed by eqn (2) but we have not done so yet.The methodology for performing such calculations was derived more than 20 years ago. (See for example ref. 1.) We now address some of the speci–c objections to the H DIBs model listed by Prof. Snow. (1) If the DIBs do not depend upon radiation –eld intensity then why are vir- 2 tually all of the stars listed in Table 3 of Herbigœs pioneering study of diÜuse interstellar bands2 either O- or B-type stars ? Of the 10 early A-stars listed by Herbig in this table which are the only exceptions to the above statement all happen to be supergiants. In exciting the various non-linear processes in our model the important quantity is the total amount of radiation emitted by the star in narrow wavelength regions in the VUV spanning the strongest (X]B C) H resonances and the Lyman lines of H atoms.Obviously the amount of radiation emitted depends both upon a starœs spectral type 2 and upon its radius. An A0 supergiant has a surface emitting area that is about 30 times greater than that of a B0 V star for example. (2) We are fully aware of the fact that DIB intensities in a given line-of-sight are strongly correlated with the amount of reddening for this fact based upon the H non-linear DIBs model. This explanation does not (i.e. EB~V) observed in the same line-of-sight. In our paper we propose an explanation 2 require the H2-containing cloud of our model to be circumstellar. It only requires that there be dust present in the cloud. We do not fully understand Prof. Snowœs statement that DIBs correlate with interstellar distance.To us it would seem that a star that is viewed along a heavily reddened line-of-sight is not necessarily a very distant star. There may simply be present an intervening dust cloud. The B1 Ib halo star (d\1.7 kpc) HD 214080 discussed at length in our paper is relatively distant from us as far as standard (E DIB stars go yet its line-of-sight is only slightly reddened B~V\0.07). Comparatively weak DIBs are seen in this star.3 (3) We are not knowledgeable enough about astronomical measurements of DIBs performed in speci–c lines-of-sight to comment with any real conviction on Prof. Snowœs statement that DIBs tend to be weak not enhanced in regions of high VUV radiation –elds. However one might question the validity of this remark on the basis of one speci–c paper in which the environmental dependence of DIBs was studied.4 In probing the strengths of two DIBs in various lines-of-sight in the Orion region Ehrenfreund and Jenniskens did indeed –nd them to be absent or much weaker in lines-of-sight to H II regions which are regions immediately adjacent to stars in which all of the hydrogen is ionized and in which therefore there should be no molecular hydrogen.However in other lines-of-sight that probed various portions of the Orion Giant Molecular Cloud they found that while in the lighter portions of the cloud the DIB intensities did increase with increasing reddening in the darker portions of the cloud the DIB intensities decreased with increasing reddening.The scientists concluded that the determining factor for the decrease of DIB strength was the attenuated VUV –eld in dark clouds. (4) A universal correlation of DIB strength with atomic hydrogen is predicted by the version of the H DIBs model discussed in our paper. In Fig. 20 the 2 secondary broadband SRS process involving the quantum level (BA1 J@\1) as resonant intermediate state is shown pumped by Ly-b photons the latter assumed to be present in the H -containing cloud at enormously enhanced densities via the mechanism of elastic scattering [i.e. the mechanism implied in the derivation of eqn. (1) and (2) above]. 2 The Stokes-wave radiation generated in this process results in the appearance of the DIB at ca. 5780 ” through the inverse Raman absorption step pictured in Fig.21. Likewise Ly-a radiation is instrumental in producing the k5797 and k6379 DIBs in the scheme shown in Fig. 1 earlier. With regard to the statement asserting that there is no correlation of DIB intensities with H column densities in various lines-of-sight one should be somewhat cautious. Consider the ideal case of a thin spherical shell of hydrogen gas (both atomic and 2 molecular) centered about a bright star that was assumed in the derivation of eqn. (1) and (2) above. This geometry should represent an ideal situation for generating strong DIB intensities according to the non-linear model discussed in our paper (provided that 233 General Discussion the shell also contains some dust). One pictures the light waves involved in the various coherent 2n-wave mixing processes discussed propagating in great circles lying within the ìplaneœ of the thin H -containing cloud.One would expect to see strong Ly-a and 2 Ly-b ìabsorptionsœ in the line-of-sight because these radiations are truly absorbed in pumping various steps of the 2n-wave mixing processes. For the same reason one would also expect to see ìabsorptionsœ around the stronger (X]B C) H absorption lines. But what about the weaker X]B transitions e.g. B0-0R0 B0-0R1 B0-0P1 B1-0R0 2 B1-0R1 B1-0P1 B2-0R0 B2-0R1 B2-0P1 . . . etc. ? In our model SRS pumping would not likely occur on these transitions so no non-linear absorptions would occur here. We have argued all along that linear inelastic scattering does not occur in our model. There is only linear elastic scattering.However the spherical symmetry of the thin H2- containing cloud here assumed implies that no attenuation of light should be evident in viewing these weaker X]B transitions in the line-of-sight ! If one were to estimate the H column density via only these weaker transitions one would conclude that the H 2 column density is zero. 2 In response to points (5) and (6) at the very basis of our H non-linear DIBs model is the fundamental assumption that excitation via single-photon transitions (i.e. inelastic 2 scattering) does not occur at all. In this model all the excitations occur to gerade state quantum levels in simultaneous two-photon steps. Our model does not involve direct one-photon absorptions originating from excited-state levels that need to be populated.The only vibrationally excited quantum levels of the X-state that are populated via two-photon transitions in the various 2n-wave mixing processes we invoke are ones that can be depopulated by secondary SRS processes pumped by Ly-a Ly-b Ly-c . . . etc. radiation. Basically the various 2n-wave mixing processes act on the H molecules 2 located in the very lowest energy quantum levels (i.e. JA\0 and 1 of X0) in such a way that they remain in these same quantum levels. (7) The main thrust of Section 4 of our paper is that we feel evidence of VUV emission bands generated in 2n-wave mixing processes exists in the VUV spectra of mid-/late-B supergiants. (8) At the Faraday Discussion meeting we did not actually see the poster paper presented by Drs Zacs and Schmidt.However later in a private conversation with us at the meeting Dr Ehrenfreund commented that the DIBs reside in the interstellar medium towards stars and that on 1 Kpc distance the eÜects of the circumstellar environment is often negligible. Only when the overall line-of-sight conditions and geometry towards the star are known can information on the excitation mechanism of the DIB carriers be obtained. To observe DIBs in cold stars is however very difficult due to strong stellar contamination but apparently not impossible in very reddended targets. 1 V. S. Letokhov and V. P. Chebotayev Nonlinear L aser Spectroscopy Springer-Verlag Berlin Heidelberg New York 1977 ch. 5. 2 G. H. Herbig Astrophys. J. 1975 196 129. 3 G.H. Herbig Astrophys. J. 1993 407 142. 4 P. Ehrenfreund and P. Jenniskens in T he DiÜuse Interstellar Bands ed. A. G. G. M. Tielens and T. P. Snow Kluwer Dordrecht Boston London 1995. Dr Zacó s and Dr Schmidt communicated We have presented (on a poster) the results of our search for diÜuse bands in the spectra of three post-AGB candidate stars (IRAS19244]1115 IRAS19114]0002 IRAS22272]5435) carried out in the frame of a more extensive study of the photospheric abundances of protoplanetary nebulae (PPN) candidates. DIBs in the spectrum of IRAS19114]0002 (\HD179821) have been identi–ed. The optical spectrum of HD179821 (spectral type F5I other authors give G5I) shows that it is a high-luminosity object and the presence of a detached cool dust shell suggest that it has experienced mass loss in the recent past.The observed spectrum (Fig. 8) [thick line ; wavelengths corrected for the stellar heliocentric radial velocity (RV) of ]88.4 km s~1] and synthetic (calculated) (thin line) spectrum of HD179821 around 234 General Discussion Fig. 8 Observed (óó) and calculated (»») spectra of HD 179821 around DIBs at 5780 and 5797 Aé DIBs at 5780 equivalent widths (EW) A \230 mAé and 5797 é (160 mAé ) are shown. The synthetic spectrum was calculated in LTE using Kuruczœs models and atmospheric parameters (and abundance) derived by Zacs et al.2 The presence of DIBs is evident. Note the strange pro–le of the DIB at 5780 ” if compared with the intrinsic pro–le of this DIB given by Krelowski and Schmidt.3 Although there is a moderate contribution (15 mAé ) of the stellar Fe II absorption on the red wing the presence of the circumstellar component DCB (RV\]46 km s~1 EWB80 mAé ) is possible.Note that NaD lines consist of four absorption components with radial velocities ([11 ]42 ]64 ]107) diÜerent from the mean photospheric velocity (]88.4 km s~1). 1 Kurucz SAO CD-ROM no. 13 1993. 2 L. Zacó s V. G. Klochkova V. E. Panchuk and R. Spelmanis Mon. Not. R. Astron. Soc. 1996 282 1171. 3 J. Krelowski and M. Schmidt Astrophys. J. 1997 477 209. Dr Sorokin Dr Glownia and Dr Ubachs responded We refer to the last item of our response to Prof. Snowœs comments. From the actual comment (and Fig. 8) submitted by Drs Zacó s and Schmidt it certainly would appear that the k5780 and k5797 DIBs are seen in the line-of-sight to HD 179821 an F5 Ia star.Prof. Glinski commented We have been studying the relatively regular Herbig DIB group near 6800 ” as it may be due to perpendicular rovibronic bands of a molecule or ion of the form CH2X.1 The regular spacing of about 18 cm~1 and the intensity alternation is entirely consistent with the rotational constants and ortho/para ratio corresponding to the HwCwH moiety. We have been modelling the spectrum with various carbon chains as the X. In Fig. 9 we show one of our preliminary models of the perpendicular band group for H2CxCxCxCN which could also be an ion (Schulz et al.2) We use the recently measured BA rotational constants for H2C5 of McCarthy et al.3 and the estimable AA A@ and B@ constants to synthesize the spectrum.We note that there is relatively little leeway for varying the A and B constants in these types of molecules because they are rather rigid. The required K-level temperature could present a problem unless the molecules were an ion. We are continuing to re–ne our models of these groups of bands along with their single parallel band counterparts. These multiple bands will serve as a strong test of any spectroscopic –t. Our result supports the idea that a family of carbon chain molecules may account for many of the weak DIBs in accord with the work of Prof. Maierœs group. We would like to learn from Prof. Maier if the rotational temperature in the C7~ bands are sufficiently broad in order to –ll in the DIB lines. The modelled width of the C7~ DIBs will 235 General Discussion Fig.9 Synthetic rovibronic spectrum that was obtained using the following parameters AA\9.40 BA\0.070 A@\9.20 and B@\0.068 ; T position of the synthetic spectrum is not de–nite. The Herbig spectrum is reproduced with kind k\600 K T f\300 K. Absolute wavelength permission from Astrophys. J. 1988 331 999. The two bands represented as arrowheads at the right are positioned where shown in the work of Herbig and Leka (Astrophys. J. 1991 382 193). yield the rotational excitation temperature of that non-polar ion in those environments and that could be quite warm. 1 R. J. Glinski and J. A. Nuth Publ. Astron. Soc. Pac. 1995 107 453. 2 S. A. Schulz J. E. King and R. J. Glinski in preparation. 3 M. C. McCarthy M. J.Travers A. Kovacs C. A. Gottlieb and P. Thaddeus Astrophys. J. Suppl. Ser. 1997 113 105. Prof. Maier responded The width of the bands in the electronic spectrum of C7~ is very sensitive to the laser powers used in the two colour excitation/detachment approach because of saturation eÜects. The narrowest band-width we obtained for the origin band is 0.02 nm which is less than the 0.1 nm FWHM of the 627.01 nm DIB. The rotational temperature in our gas-phase measurements is expected to be less than 100 K. Dr Ehrenfreund commented The development of DIB research in recent years indicates that most DIB carriers could be large carbon-bearing molecules which reside ubiquitously in the interstellar gas.1,2 The –rst detection of substructures in the pro–le of several DIBs indicated the molecular nature of some DIB carriers3,4 Foing and Ehrenfreund5 observed two DIBs at 9577 and 9632 ” as –rst evidence for C60 ` the largest molecule ever detected in space.A new reference target for DIB studies was recently detected which shows the strongest DIBs ever measured and allows de–nition of the DIBs in several categories which respond in a totally diÜerent way to the local environment. ” 6 Cami et al.7 completed a survey of DIB correlations over 4000 which showed that most of the DIB carriers are undergoing photo-ionization and that all measured DIBs do originate from diÜerent carriers. The spectra of PAH and fullerene cations measured in a neon matrix carbon chains measured in the gas phase and theoretical calculation of the non-linear H2-DIB model have all shown striking coincidences with some diÜuse bands.Another approach is to study the complete DIB spectrum in diÜerent interstellar and circumstellar regions and to relate the line-of-sight conditions directly to the formation/evolution and destruction properties of DIB carrier molecules. Additional observations of spectral molecular features such as CH CH` CN (as well as atomic lines Ca I Ca II Na I) reveal variations of physical parameters of the interstellar environment and can constrain the chemistry ionization balance metallicity and electron density in the circumstellar and interstellar environment. Those observations show currently up to ca. 200 DIBs in dense and cold environments as well as in UV dominated regions.Their central wavelength is extremely constant. The band strength of the 236 General Discussion strongest DIBs (such as the 5780 and 5797 ” DIBs which are measured towards more than 200 sources) does not change more than a factor ca. 2. Even the high resolution pro–le of the 6613 ” DIB which shows a characteristic triple peak displays only slight changes in diÜerent environments. The DIB strength varies with the total HI column density which results in a decline of DIB strength with depth in the cloud where the H concentration rises. The so called ìskin-eÜectœ can be explained by a concentration of 2 DIB carriers in the surface layers of dense clouds.1 The relative DIB strength W / E(B[V) seems to re—ect an interplay between ionization and recombination and destruction of the DIB carrier molecules.These recent observational results favour molecules such as PAH and fullerene cations as the most interesting DIB carrier molecules. PAH cations ful–l the criteria of abundance and stability and show in general only one strong band per molecule in the visible range. Salama et al.2 have discussed the ìone DIB»one PAHœ hypothesis assuming that a limited number (150»200) of stable PAH cations can be responsible for the known DIB spectrum. Recent measurements of hydrogenated dehydrogenated PAHs as well as PAH anions show that those species have a very diÜerent spectrum in the optical range compared to PAH cations. The complexity of visible absorption which would be created by the diÜerent transient forms of PAHs during their evolution such as hydrogenated and dehydrogenated PAHs dications or fragments is not manifested in the current DIB spectrum.There is a possibility that all those transient PAH states may be responsible for the very weak DIBs. The grouping of DIBs has recently been studied by Krelowski et al.8 and indicated that certain combinations of strong and very weak DIBs close by may be formed by the same carriers. The carbon chain hypothesis as well as the non-linear H2-DIB model show and predict a large number of DIBs. Due to the high density of known DIBs in the optical region there is a substantial risk of purely accidental coincidences. For such species a valid selection mechanism must be found which is compatible with interstellar line-of-sight conditions.To summarise the DIB spectrum follows rather simple rules and shows apart from a strong consistency in wavelength position and pro–le structure a strong dependence on the UV –eld. These important criteria have to be respected when searching for the carrier molecules. 1 G. H. Herbig Ann. Rev. Astron. Astrophys. 1995 33 19. 2 F. Salama E. L. O. Bakes L. J. Allamandola and A. G. G. M. Tielens Astrophys. J. 1996 458 621. 3 P. J. Sarre J. R. Miles T. H. Kerr R. E. Hibbins S. J. Fossey W. B. Somerville Mon. Not. R. Astron. Soc. 1995 277 L41. 4 P. Ehrenfreund and B. H. Foing Astron. Astrophys. 1996 307 L25. 5 B. H. Foing and P. Ehrenfreund Astron. Astrophys. 1997 317 L59. 6 P. Ehrenfreund J. Cami E. Dartois and B. H. Foing Astron. Astrophys.1997 318 L28. 7 J. Cami P. Sonnentrucker P. Ehrenfreund and B. H. Foing Astron. Astrophys. 1997 326 822. 8 J. Krelowski M. Schmidt and T. P. Snow Publ. Astron. Soc. Pac. 1997 109 1135. Prof. Snow ” commented The 4430 diÜuse band once the principal subject of many DIBs studies has been largely neglected in recent years. To date there have been no general studies of the 4430 ” DIB based on high-quality CCD spectra. This diÜuse band is the strongest of all the DIBs and also the broadest with the result that several stellar photospheric lines are superposed on it. Perhaps this has discouraged intensive studies of k4430 especially considering the fact that many DIBs in the red could be analyzed without confusion by stellar lines (an additional factor is the higher QE of CCD detectors in the red as compared with the blue).We (Massey Boyd and I) have undertaken a CCD-based survey of k4430 in a number of highly reddened O stars with the result that the band appears to be saturated in lines-of-sight with high extinction (the rate of growth of band strength with EB~V levels oÜ with increasing E narrow closely spaced individual lines which become saturated in high column-density B~V). This implies that k4430 may be composed of many 237 General Discussion lines-of-sight. If so this suggests that we should search for –ne structure within the band in less-reddened lines-of-sight where saturation will be less severe. In June 1997 we carried out observations of the 4430 ” band using the ultra-high Resolution Facility (UHRF) on the Anglo-Australian Telescope in lines-of-sight toward stars where –ne structure had previously been detected in one or more of the longerwavelength DIBs.At its resolving power of k/*k ca. 106 the UHRF provides coverage of about 2” ” at this wavelength (as compared with the DIBœs FWHM of about 30 ) so we chose several small ìwindowsœ within the k4430 band carefully selected to avoid contamination by stellar lines and obtained high-resolution spectra in those windows. The result is that no trace of –ne structure within the 4430 ” band was detected. The high photometric quality of the data (signal-to-noise ca. 100) rules out signi–cant –ne structure at a level above ca. 1% of the local continuum. This result was uniform for the –ve lines-of-sight that were observed.We conclude that the 4430 ” DIB is either a smooth continuous feature or that its –ne structure either has a scale length smaller than ” ” *k ca. 0.005 or else is broader than the ca. 2 bandpass that we observed (but in that case earlier searches for –ne structure such as the one by Herbig in 1966 would have revealed it). It is not clear what this result implies for the identi–cation of the k4430 carrier. A molecular origin is not ruled out but it is now seen to be viable only if internal conversion completely eliminates –ne structure or else the –ne structure lines are so closely spaced as to appear smoothly blended to a part in 106 in wavelength. But on the other hand a dust impurity carrier which would be expected to produce a smooth pro–le may also be viable.In this latter connection it is noteworthy that this particular DIB has not been scrutinized for signs of a solid-state carrier (e.g. through searches for spectropolarization structure) as have many of the longer-wavelength DIBs and therefore a solid-state carrier has not been ruled out. It appears at least possible that the 4430 ” DIB has a diÜerent origin than the sharper features that appear at longer wavelengths. Prof. Breç chignac commented I think that there are two points that have not been well addressed yet in the discussion. The –rst one is the stability of the DIBs carriers the second one is their abundance. As a matter of fact these two points may well be chemically connected in interstellar conditions. We heard about nice coincidences involving the photodetachment spectrum of C7~ and the DIBs spectrum and indeed there has been a large number of –rm detections of chains in the interstellar medium.But even when summing up the abundances of all of them it does not count for much relative to the total abundance of cosmic carbon in comparison to the amount estimated for PAHs through the —ux emitted in the UIR bands. Similarly the PAHs are expected to oÜer a larger stability against the UV radiation –eld. These two arguments were at the basis of the proposition of PAH cations as DIBs carriers candidates.1 We have set up in my laboratory in Orsay an experiment devoted to the search for the electronic spectra of gas phase cold PAH cations in the visible. Making large quantities of cold cations in the gas phase in order to measure absorption directly is very difficult.The principle of the technique we used consists of inducing a fragmentation of the cation consecutive to the resonant absorption of laser photons so that the change in the charge-to-mass ratio can be detected in a mass spectrometer. However it is well known that because of their stability PAHs (neutrals as well as cations) need a fairly large excitation energy before they begin to fragment on a reasonable timescale. This usually requires several laser photons which means high laser —ux and then a number of experimental difficulties associated to saturation eÜects power broadening nonresonant absorption and problems in controlling the number of photons involved in the process.We have used a trick consisting of attaching an argon atom to the PAH by forming van der Waals complexes in a molecular beam. This complex is photoionised by reso-238 General Discussion nant two-photon absorption in the near UV. Then the binding energy of the complex cation is so small (about 500 cm~1) that it will fragment for every single photon absorbed. The use of two pulsed lasers of diÜerent colours in the ionisation step allows us to control exactly the total energy and to produce complexes which are cold both rotationally and vibrationally. The technique has been tested –rst on a benzene derivative the 4-—uorostyrene and found to work. It was then applied to a larger aromatic hydrocarbon the —uorene molecule C13 H10 . The time-of-—ight spectrum shown in Fig.10 illustrates how the technique works. The wavelength of the –rst UV photon has been tuned to optimise the formation of the —uorene`»Ar (Fl`»Ar 2 mass of the Fl`»Ar complex and a larger peak at the mass of Fl` since neutral —uorene 2) complexes. Some signal can be seen at the is much more abundant but not formed efficiently. The pulsed visible laser inducing absorption by the cation is delayed with respect to the ionising lasers so that the Fl` fragment produced by ejection of the two argon atoms appears separated in time from the rest of the Fl` ions. The spectrum shown in the bottom trace of Fig. 11 has been obtained by monitoring the ion signal in the time window relative to this fragment while the laser wavelength is scanned.The middle trace is the spectrum obtained similarly after optimisation of the Fl`»Ar signal. The recording of these spectra required a relatively small energy (of the order of a few microjoules per pulse) for the visible laser since it corresponds to a single photon process. On the contrary the recording of the top trace after optimisation of the Fl` signal and adjusting the time window on the channels associated to the main fragment of the free ion i.e. the ion formed by the loss of a single H atom required a laser energy about two orders of magnitude larger. As a consequence the spectrum looks broader and presents less structure. It can be noted that the peak wavelength of this spectrum is found very close to 6284 ” which is the wavelength of one of the strongest DIBs.However all of the three spectra appear to be several times broader than the interstellar band. This raises the question of the origin of the observed width. Although these results are still preliminary and should be con–rmed we believe that we deal with the spectra of vibrationally and rotationally cold cations. The order of magnitude of the laser energies used to fragment the van der Waals complexes does not seem sufficient to produce signi–cant power broadening. Thus we should be in the presence of a case of lifetime broadening as a result of intramolecular dynamics eÜects. The fact that both argon atoms are ejected from the Fl`»Ar is an indication that an ultrafast electronic internal conversion within the Fl` moiety may have preceded the step of vibrational predissociation of the complex.If this 2 Fig. 10 Time-of-—ight spectrum showing the principle of the technique used to record the visible absorption spectra of cold PAH cations in the gas phase (see text) 239 General Discussion Fig. 11 Photodissociation spectra of the Fl` cation and its van der Waals complexes with one and two argon atoms physical picture is correct it rules out the Fl` cation from the possible candidates for the 6284 ” DIB carrier. Some of you may object to the fact that the technique gives access to the spectra of the van der Waals complexes rather than of the PAHs cations which is correct. But by using the van der Waals shift additivity rule known to be obeyed by these aromatic clusters,2 one can easily recover the spectrum of PAH` from the spectra of PAH`»Ar and PAH`»Ar produce electronic spectra for other gas phase PAH cations.This should help to 2 . In conclusion we hopefully believe that we are now in a position to progress in the testing of their suggestion as DIBs carriers. 1 A. Leç ger and L. dœHendecourt Astron. Astrophys. 1985 146 81; G. P. van der Zwet and L. J. Allamandola Astron. Astrophys. 1985 146 76. 2 P. Hermine P. Parneix B. Coutant F. G. Amar and Ph. Breç chignac Z. fué r Phys. D 1992 22 529. Prof. Thaddeus commented If the interstellar diÜuse bands are carried by large molecules and if some of these are unsymmetrical and polar as seems likely they will inevitably have low frequency radio lines. These potentially represent a new type of radio spectrum probably best observed in absorption against intense galactic continuum sources like Cas A.The trouble is that these lines will be quite weak and undoubtedly very numerous»a thicket of lines. Probably the best instruments with which to search for them are the newly resurfaced Arecibo telescope working at 1»10 GHz and the 100 m GBT under construction by NRAO. There are plans to undertake such searches with these instruments. In emission at slightly higher frequency (B30 GHz) these polar molecules may contribute to high latitude anisotropy in the cosmic microwave background claimed by radio astronomers as Webster has pointed out. 3N H5N and C the ground-state dipole Ms Heyl said On the basis of large-scale coupled cluster calculations and comparison with experimental values for HCN HC moments of HC C were established to be [4.82 and [5.20 D with an 7N H9N and 240 General Discussion estimated uncertainty of ca.0.02 D.1 In a similar way k (HC11N) 0 is found to be [5.47 D.2 The k values of cyanopolyynes HC2n`1N 0 up to n\5 with experimental values3h5 taken for species with n\0 1 and 2 are displayed graphically in Fig. 12. Pronounced non-linear behaviour is observed. 1 P. Botschwina and M. Horn J. Mol. Spectrosc. 1997 185 191. 2 P. Botschwina unpublished cited in M. B. Bell P. A. Feldmann M. J. Travers M. C. McCarthy G. A. Gottlieb and P. Thaddeus Astrophys. J. 1997 483 L61. 3 Landoldt»Boé rnstein Molecular constants Group II ed. G. Guelachvili Springer Berlin vol. 20 sub. vol.B1 1995. 4 R. L. DeLeon and J. S. Muenter J. Chem. Phys. 1984 80 3892. 5 A. J. Alexander H. W. Kroto and D. R. M. Walton J. Mol. Spectrosc. 1976 62 175. Prof. Irvine said I would like to ask at what length long carbon chains will begin to form cyclic isomers and whether that can be studied in the laboratory ? Prof. Herbst responded Although the relative energies of linear and cyclic carbon chain molecules depend to some extent on whether one is referring to charged or neutral molecules and whether one is referring to bare carbon chains or to more complex entities the general picture is as follows. For small molecules both linear and cyclic isomers exist and can be relatively close in energy. The diÜerent isomers can be produced by diÜerent sequences of reactions.As one gets beyond 10 carbon atoms in size the cyclic forms tend to become consistently lower in energy although straight chain species have been studied in the laboratory proving that kinetically controlled isomerization need not be rapid under all conditions. Experiments indicate however that isomerization of bare linear carbon chains to cyclic isomers eventually does occur for species with 10»20 carbon atoms.1 In our theoretical treatment we have estimated that spontaneous isomerization to cyclic molecules starts under interstellar conditions when the molecules contain somewhat more than 20 carbon atoms.2 We have shown that low temperature ion»molecule and neutral»neutral reactions operating in dense and diÜuse clouds can produce the following classes of molecules carbon chains]monocyclic rings]tricyclic rings B fullerenes PAHs on the other hand can only be formed at high temperatures most probably in carbon rich circumstellar envelopes.2n`1 species with n\0 1 and 2 N up to n\5 with experimental values taken for Fig. 12 Values of k for cyanopolyynes HC 0 241 General Discussion 1 G. von Helden N. G. Gotts and M. T. Bowers Nature (L ondon) 1993 363 60. 2 R. P. A. Bettens and E. Herbst. Int. J. Mass Spectrom. Ion Proc. 1995 149/150 321. Dr Takahashi said As a related work to radical ring-chain molecules in Prof. Thaddeusœs paper I would like to introduce an ab initio quantum chemical study on stable isomers of carbon chain molecules C (n\3»7) by us.1 We calculated most of all the nH stable isomers of CnH molecules and found that there is a general tendency that the second most stable isomers have 3-member-ring structures for odd n while n-memberrings for even n (see Fig.13). 3HC The relative energies calculated at the QCISD/6-31G** level are shown in Table 2. The ground electronic states of linear C are 2%. On the other 3H C5H C6H C7H and hand it is found that the diÜerence between 2& and 2% is very delicate for C4H. The 2B electronic states of 3-ring isomers of and C3H C5H C7H are while those of n-ring 2 2B 2A C As is now known,2h5 the 3-ring isomer of 4H C6H and are respectively and 1 C of 1. is nearly as stable as or slightly more stable than the linear one. The 3-ring isomers 5H C7H are less stable than linear ones by only 4 and 15 kcal mol~1 respec- and tively.The n-ring isomers of C are less stable by 23 and 15 kcal mol~1 4H C6H and respectively. We expect that the second most stable isomers especially of C5H C7H and might be detectable in interstellar clouds. We have calculated rotational constants dipole moments IR spectra etc. for them which would be useful for their detection. 1 J. Takahashi and A. Murakami in preparation. 2 J. Takahashi and K. Yamashita J. Chem. Phys. 1996 104 6613. 3 C. Ochsenfeld R. I. Kaiser A. G. Suits Y. T. Lee and M. Head-Gordon J. Chem. Phys. 1997 106 4141. 4 S. Ikuta J. Chem. Phys. 1997 106 4536. 5 K. Aoki K. Hashimoto and S. Ikuta Astrophys. J. submitted 1997. Fig. 13 Stable isomers of CnH molecules Table 2 Relative energies of C (n\3»7) isomers calculated nH by ab initio molecular orbital method at the QCISD/6-31g** level (kcal mol~1) C3H C4H C5H C6H C7H linear 2% linear 2& 3-ring 0 0 0 0 0 » [0.6 » » ]15 » n-ring ]2.9 ]30 ]15 » ]3.7 ]55 ]0.3 ]40 ]23 242 General Discussion Prof.Papoular said I wish someone would come up here to talk about free-—ying PAHs and give us such a convincing proof of their existence in space as Prof. Thaddeus did for long carbon chains ! Regarding the formation of the latter theorists seem to have a hard time synthesizing molecules larger than three or four atoms using exclusively gas-phase molecular reactions. On the other hand the number of long carbon chains identi–ed in space is increasing steadily and the largest size observed is already 19 atoms.How is this paradox to be resolved ? Noting that these chains are of the unsaturated carbon type (alternating single and triple bonds or successive CxC double bonds) I propose that they are essentially extracts from kerogen-type interstellar grains. This is justi–ed as follows. The SiC emission feature at 11.3 lm bears witness to the formation of carbonaceous grains around C-rich stars on the Red Giant Branch. Later in their evolution (on the Asymptotic Giant Branch) when their mass loss increases and their circumstellar shell becomes thicker this feature is replaced by near-IR features in extinction the ice feature near 3300 cm~1 and the aliphatic CwH feature around 2900 cm~1 (3.4 lm).The latter becomes prominent towards the Galactic Center1 which is known to harbour many of these stars. The 3.4 lm feature is also exhibited by meteorites where it is assigned to a notable kerogen-like carbonaceous component.2 Kerogen is an earthly mineral associated with oil beds and similar to the less evolved coals.3 Geologists describe it schematically as in Fig. 14 it is made up of a few benzenic ring clusters a number of isolated rings all with attached hydrogen and oxygen atoms or functional groups (mainly aliphatic CwH CH2 CH3 responsible for the 3.4 lm band) and embedded in a matrix of long undulating aliphatic (unsaturated) chains exactly like the molecules identi–ed by Thaddeus et al. in their paper. It is very tempting to imagine that in their journey in the harsh interstellar medium kerogenic grains from the stars are eroded into the observed long chains on the one hand and on the other hand into ring clusters which ultimately form the polycrystalline graphite responsible for the UV extinction bump at 217 nm.4 1 S.A. Sandford Y. J. Pendleton and L. J. Allamandola Astrophys. J. 1995 440 697. Fig. 14 Chemical models of a type 1 kerogen (reproduced with kind permission from Behar and Vandenbroncke Rev. Inst. Francais Peç trole 1986 41) ; (a) early stage of evolution (b) later stage (ageing in mild temperatures helps expel the aliphatic chains) 243 General Discussion 2 J. F. Kerridge in Carbon in the Galaxy ed. J. C. Tarter S. Chang and D. J. De Frees NASA Conference Publication 3061 1990 p.3. 3 See Kerogen ed B. Durand Editions Technip Paris 1980. 4 R. Papoular O. Guillois I. Nenner J. M. Perrin C. Reynaud and J-P. Sivan Planetary Space Sci. 1995 43 1987. Prof. Thaddeus opened the discussion of Dr Ohishiœs paper Do you think better sources than TMC-1 of large molecules can be found either in Taurus or elsewhere in the Galaxy? It would be peculiar if this somewhat obscure cloud core turned out to be the best in the entire system. Dr Ohishi responded It is a difficult question to answer. If I knew a better source than TMC-1 I would have observed it already. Even in the Taurus region there are several cores with high carbon chain abundances. Generally speaking there is of course a possibility that better sources exist between the Galactic Center and us.Therefore it is important to survey such sources. 7~ is unusual in that it is an anion. A polar carbon chain anion would Dr Webster commented John Maierœs suggested identi–cation of four diÜuse interstellar bands C show up in Dr Ohishiœs centimetre-wave survey of TMC-1 as an unidenti–ed line so the number of anionic species could be estimated or limited by the number of unidenti–ed lines. Is that number large or small ? Dr Ohishi responded At present we have only one con–rmed unidenti–ed line in our data. This line was observed three times to check its reliability. We have about 10 lines with less reliability and about 200 lines whose intensities are 2»3 p (noise —uctuation) levels. I donœt deny the possibility that new species could be found from these unidenti- –ed lines.Prof. Irvine commented It is perhaps useful to point out that in addition to the source-to-source abundance variations mentioned by Dr Ohishi there are de–nitely abundance gradients within speci–c interstellar clouds of which chemical modelers should be aware. One study of such variations in TMC-1 has been recently published by Pratap et al.1 An example for another cold dark cloud is shown in Fig. 15 where the emission maps for the cloud L134N are given. Although these intensity distributions have not yet been converted to abundance distributions it seems clear that there must be gradients in for example the relative abundances of N2H` and SO. 1 P. Pratap J. E. Dickens R. L. Snell R. P. Miralles E. A. Bergin W.M. Irvine and F. P. Schloerb Astrophys. J. 1997 486 862; J. E. Dickens W. M. Irvine R. L. Snell E. A. Bergin F. P. Schloerb P. Pratap and M. P. Miralles in preparation ; J. E. Dickens PhD Thesis University of Massachusetts 1998. Dr Palumbo asked Carbon-chain molecules have been detected both in the molecular cloud TMC-1 and in the circumstellar shell of the nearby carbon star IRC]10216. Do you think that the formation mechanism of these molecules in these diÜerent environments is the same? Prof. Herbst responded Current generations of gas-phase chemical models such as ours and those of the UMIST group are capable of reproducing most if not all observed abundances in the molecular cloud TMC-1 (ref. 1) and in the circumstellar shell of the carbon star IRC]10 216.2 Although these models both contain ion»molecule and neutral»neutral reactions the role of neutral»neutral reactions in the syntheses of complex molecules does seem to be more important in the circumstellar shell than in the molecular cloud.In addition photochemistry is far more important in the circumstellar model. In TMC-1 which is thought to be oxygen-rich inclusion of unstudied neutral» 244 General Discussion Fig. 15 Maps of the intensity of molecular transitions of the indicated molecular species over an extended region of the dark cloud L134N observed in the 3 mm band at the Five College Radio Astronomy Observatory of the University of Massachusetts (J. E. Dickens W. M. Irvine R. L. Snell E. A. Bergin F. P. Schloerb P. Pratap and M.P. Miralles in preparation ; J. E. Dickens PhD Thesis University of Massachusetts 1998.) 245 General Discussion neutral reactions involving atomic oxygen actually worsens agreement with observation. 1,3 1 R. Terzieva and E. Herbst Astrophys. J. 1998 501 207. 2 T. J. Millar and E. Herbst Astron. Astrophys. 1994 288 561. 3 R. P. A. Bettens H-H. Lee and E. Herbst Astrophys. J. 1995 443 664. Mr Markwick commented Recently we (Markwick et al.1) have been investigating the chemical eÜects on a model of TMC-1 of injecting some carbon-bearing species into the gas phase which would result from the evaporation of grain ice mantles. This evaporation could be triggered by an MHD Alveç n wave of the kind argued for by Charnley and Butner2 propagating down the ridge away from the IRAS source located towards the northwest.Brie—y the cloud model is run to steady state (108 yr) and then small quantities of CH and C 4 C2H2 2H4 are injected to simulate the ice mantle evaporation. The clock is reset and the model run for a further 106 yr. The results for some species are shown in Fig. 16. The main aim was to see whether this evaporation process could produce the chemical gradients observed along the TMC-1 ridge,3 and we found that it could. To calculate abundance gradients along the ridge the time ordinate has to be converted to distance along the ridge so a speed of 2 km s~1 is assumed for the wave. The zero time at the cyanopolyyne (CP) peak is arbitrary (we donœt know when the wave passed) and so it was chosen to be the point at which the cyanopolyyne species peak after injection (B3]105 yr).Fig. 16 Variation of fractional abundances of cyanopolyyne species CS and CCS with time after the evaporation of grain ice mantles Table 3 Observed fractional abundance of cyanopolyyne species at the CP peak together with the two peak values in the model. ìPeak 1œ is before mantle evaporation and ìPeak 2œ after. [a(b)\a]10b] peak 2 peak 1 observed HC HC HC HC 3N 8.0([08) 9.1([11) 6.2([09) 5N 6.0([09) 1.3([11) 2.4([09) 7N 1.0([09) 5.4([13) 1.3([10) 9N 7.0([10) 4.1([15) 6.7([12) 246 General Discussion The data presented here by Ohishi and Kaifu means that we can compare the abundances at the CP peak point in our model with observed values. There are two points of interest (1) The evaporation of ice mantles boosts the abundances of cyanopolyyne species by orders of magnitude (Table 3) suggesting that this process could be important for their formation in the quantities observed.(2) The ìunusualœ result of CCS being more abundant than CS at the CP peak can be obtained as seen in Fig. 16. In the model CCS is more abundant than CS between 1.6]101 and 7.9]105 yr after evaporation consistent with the adoption of 3]105 yr for the time at the CP peak. 1 A. J. Markwick T. J. Millar and S. B. Charnley in preparation. 2 S. B. Charnley and H. Butner 1995 in Circumstellar Matter ed. G. D. Watt Kluwer Dordrecht 1995 p. 443. 3 P. Pratap J. E. Dickens R. L. Snell M. P. Miralles E. A. Bergin W. M. Irvine and F.P. Schloerb Astrophys. J. 1997 486 862. Dr Ohishi responded This is an interesting and important comment on the chemistry occuring in dark cloud cores. As we have noted in the text the abundance of CS was calculated by using a single transition (J\1»0) the value quoted in the paper may contain a large uncertainty. Therefore we plan to improve the CS abundance toward the cyanopolyyne peak of TMC-1 in the near future. When a more reliable value has been obtained and it has been veri–ed that CS is less abundant than CCS your theoretical work gives a great impact on the interstellar chemistry that evaporation of molecules from grain mantles is also occuring even in dark clouds as is the case for hot cores. Miss Chastaing Mr L. James Dr Sims and Prof.Smith opened the discussion of Dr Kaiserœs paper You have presented an exciting study of the dynamics of C-atom reactions with small hydrocarbons studied in a crossed molecular beam apparatus at collision energies down to approximately 8 kJ mol~1. In your presentation you referred to the importance of possible small exit and entrance channel barriers in such reactions to the chemistry of dense interstellar clouds. However the average collision energy at the low temperatures prevailing in dense clouds is two orders-of-magnitude less than the lowest energy you were able to reach. Small barriers which may be in—uential in determining rate coefficients and product branching ratios at these temperatures (10»50 K) would be very unlikely to be inferred from your higher energy studies.Is there any way that you could extend these interesting studies to lower collision energies of more relevance to the chemistry of dense clouds ? Furthermore you mentioned the possibility of inferring rate coefficients for these reactions by estimating absolute cross sections. However not only is the estimation of absolute cross sections notoriously difficult,1 but their subsequent conversion to rate coefficients introduces further unknowns. Are you able to address these concerns in order to provide absolute rate coefficients for interstellar chemistry ? 1 See for example G. Scoles Atomic and Molecular Beam Methods ed. G. Scoles D. E. Bassi U. Buck and D. Laineç Oxford University Press New York and Oxford 1988 vol. 1 p. 4. Dr Kaiser responded Since our crossed beams machine has two sources –xed at 90° it is extremely difficult to go to collision energies less than 8 kJ mol~1.If we push it to the lower limit we might go down to about 4 kJ mol~1. But thatœs it. As stated small barriers might in—uence the reactions at lower energies such as present in cold molecular clouds but we cannot investigate these low energy processes with our machine. I agree that the branching ratio to e.g. 1-C3H c-C3 H vs. depends strongly on the collision energy and that it is tricky to extrapolate branching ratios from our studies but the prime directive of our experiments is to investigate the chemical dynamics reaction products and the potential energy surface of these reactions. Both experiments and high level ab initio calculations can then be combined to get information on possible branch-247 General Discussion ing ratios relevant to cold molecular clouds.However based on the chemical dynamics the reaction products investigated over a wide range in our crossed beam experiments should be present in the ISM as well if small entrance barriers are absent. Currently the group of Prof. Head-Gordon at UC Berkeley is extending the ab initio studies of the reaction C]C2H2 ]C3H]H C to 3HD]C]C3H/C3D]D/H including transition state calculations. Hence these calculations should give the –nal answer if there are any small entrance/exit barriers and the in—uence on the reaction pathway. Regarding the cross sections of course it is extremely difficult but it is feasible.Itœs only a question of time but sooner or later it will work. If the absolute reactive cross sections are known we can convert them to collision energy dependent rate constants k(E). We cannot give temperature dependent rate constants. Even an approximation would be too shaky. reactions reported in this Discussion.3 2H CJ)(3P atoms were produced by the 193 nm Miss Chastaing Mr James Dr Sims and Prof. Smith commented Kaiser et al.1 have in their article highlighted the importance of C-atom reactions with neutral co-reagents especially unsaturated hydrocarbons in interstellar chemistry. Husain and co-workers have measured the rate coefficients of a wide range of such reactions at room temperature and in collaboration with Clary have related these to a simple electrostatic capture model.2 We have very recently performed measurements of the rate coefficients of C(3PJ) C with 2H2 C2H4 and O at temperatures from 295 K down to 27 K within 2 the ultracold environment of the Birmingham CRESU apparatus,3 using a chemiluminescent marker technique resembling in principle that described by us in our study of C excimer laser photolysis of carbon suboxide C3O2 ,4 and detected by adding a constant excess of NO to the gas —ow and observing the chemiluminescence resulting from the reaction 2 (1) C]NO2 ]CO]NO Some proportion of the NO is formed in the B 2% electronically excited state,5 and observation of the NO b(B 2%»X2%; v@\0]vA\8) emission via an interference –lter centred at 320 nm (10 nm FWHM) gave a signal proportional to the C-atom concentration.Our preliminary results for the reactions of C atoms with C2H2 C2H4 and O at temperatures between 295 and 27 K are shown in Fig. 17»19 and the results of non- 2 linear least-squares –tting to a rate coefficient expression of the form k\A (T /298 K)n are given in the –gure captions. It should be emphasised that these are approximate expressions (estimated con–dence ^20%) which should only be used in the measured 2 ; temperature. The –lled symbols show the results of this work Ö denotes Ar carrier gas; = N Fig. 17 Rate coefficients k for the reaction of C with C2H2 plotted on a log»log scale against and > He. The room temperature result of Naider and Husain6 is shown as L while »» shows the –tted expression k\2.84]10~10 (T /298 K)~0.11 cm3 molecule~1 s~1.248 General Discussion 2 ; temperature. The –lled symbols show the results of this work Ö denotes Ar carrier gas; = N Fig. 18 Rate coefficients k for the reaction of C with C2H4 plotted on a log»log scale against and > He. The room temperature result of Naider and Husain6 is shown as L while »» shows the –tted expression k\3.00]10~10 (T /298 K)~0.13 cm3 molecule~1 s~1. temperature range 27»295 K. We note that while our room temperature rate coefficients are somewhat higher than those measured by Husain and co-workers6,7 (by ca. 30»40%) in the case of C]O we are in near exact agreement with the recent laser- 2 based determination of this rate coefficient by Becker et al.8 We plan on making measurements at lower temperatures and with a wider range of reaction partners in the very near future as well as further measurements using a complementary technique involving direct detection of C atoms by vacuum ultraviolet laserinduced —uorescence.1 R. I. Kaiser C. Ochsenfeld D. Stranges M. Head-Gordon and Y. T. Lee Faraday Discuss. 1998 109 183. 2 See D. C. Clary N. Haider D. Husain and M. Kabir Astrophys. J. 1994 422 416 and references contained therein. 3 D. Chastaing P. L. James I. R. Sims and I. W. M. Smith Faraday Discuss. 1998 109 000. 4 D. J. Anderson and R. N. Rosenfeld J. Chem. Phys. 1991 94 7857. 5 G. Dorthe J. Caille S. Burdenski P. Caubet M. Costes and G. Nouchi J. Chem. Phys. 1985 82 2313. 6 N. Naider and D. Husain J. Photochem. Photobiol. A Chem. 1993 70 119.7 D. Husain and A. N. Young J. Chem. Soc. Faraday T rans. 2 1975 71 525. 8 K. H. Becker K. J. Brockmann and P. Wiesen J. Chem. Soc. Faraday T rans. 2 1988 84 455. Prof. Clary commented The elegant molecular beam experiments of Kaiser and co-workers have quite high initial collision energies which are much higher than those and 2 ; Fig. 19 Rate coefficients k for the reaction of C with O plotted on a log»log scale against temperature. The –lled symbols show the results of this work Ö denotes Ar carrier gas; = N 2 > as He. The room temperature result of Husain and Young7 is shown asL and that of Becker et al. K while »» shows the –tted expression k\4.68]10~11 (T /298 K)~0.40 cm3 molecule~1 s~1. 249 General Discussion expected for interstellar clouds.The ab initio calculations for some of their reactions give some products that are almost equal in energy to that of the reactants (e.g. C ]C2H2 ]linear[C3H]H). If there was a small exit channel barrier in the potential surface for producing these particular products then they might not be produced efficiently in the conditions of the interstellar medium. What are the prospects for Dr Kaiser and his colleagues who do electronic structure calculations for studying this point in more detail ? Dr Kaiser responded Currently Prof. Head-Gordonœs group at UC Berkeley is C]C2H2 ]C3H]H C and 3HD 3D]D/H including transition state calculations. These investigations extending the ab initio studies of the reaction ]C]C H/C should resolve the question of whether there are any entrance/exit transition states.3 Prof. Clary said The interesting experimental results of Prof. Smith on C2H tions and Dr Sims on C atom reactions provide an excellent test of simple predictions reacwe made1 on the temperature dependence of rate constants for C atom reactions with alkenes and alkynes. A simple capture theory predicted that the reaction rate constant was proportional to N1@3T 1@6 where N is the number of carbon atoms in the organic molecule. Systematic experiments undertaken by Husain and co-workers showed the predicted N dependence to work well.1 The new experiments by Smith and Sims and co-workers give a very weak temperature dependence for the reaction rate constants at lower temperatures and this also –ts in quite well with the prediction from simple capture theory.1 D. C. Clary N. Haider D. Husain and M. Kabir Astrophys. J. 1994 422 416. Prof. Herbst said The temperature dependence of radical»neutral reactions depends CN]C both on long-range and short-range forces. For 2H]C2H2 simple 2H2 and C capture calculations yield a T 1@6 dependence for the rate coefficient in disagreement with the measured inverse temperature dependence for temperatures above 50 K. A phase space calculation for CN]C2H2 using a small exit channel barrier reproduces the experimental temperature dependence.1 1 D. E. Woon and E. Herbst Astrophys. J. 1997 477 204. Prof. Smith said I should like to respond to two comments made by Prof. Herbst. My –rst remark relates to the term ìcaptureœ or ìcapture theoryœ.I believe that two points are now generally agreed in regard to the factors which control the rate coefficients of neutral»neutral reactions proceeding over potential energy surfaces without barriers. First the rates are always determined by ìcaptureœ ; that is by the magnitude of the reactive —ux through some crucial region of the attractive potential between the reagents. However it is generally not sufficient to consider only the attractive potential in the limit of long-range which is determined by the dispersion forces and by interaction of non-symmetric charge distributions on the reagents. Capture theories based on such potentials generally provide an upper limit to the rate of capture which is approached only at very low temperatures.At higher temperatures the transition state region moves to smaller separations and the magnitude of the reactive —ux or capture falls. It is such eÜects which cause the rate coefficients of many of the neutral»neutral reactions that have been measured down to ca. 20 K to fall with increasing temperature. Therefore the principal diÜerence between diÜerent theoretical estimates is in the potential which is used. My second comment relates to what might be termed the ìwish list œ of reactions for which it would be desirable to have low temperature rate coefficients. It would certainly be very worthwhile if astrochemical modellers advise experimental and theoretical physical chemists about which processes have particularly strong in—uences on observed 250 General Discussion abundances in particular astronomical regions so that they can be studied particularly intensively.However it has to be realised that there is little prospect of some reactions being studied experimentally by techniques that are currently available. It is for this reason that the combined eÜorts of experimentalists and theoreticians are so important. Finally in this respect I wonder if it would be useful to attempt to set up a body of experts to review data for astronomical modelling along the lines that has been adopted so successfully for atmospheric chemistry. Miss Terzieva (with Prof. Herbst) commented Starting with our New Standard Model1 we introduced a number of neutral»neutral reactions studied at temperatures below 25 K or close analogs to such reactions including the reactions just described by Prof.Smith. I will refer to this model as Proveit 1 (see Table 4). In a second model called Table 4 Modi–cations to neutral»neutral rates and models ref.a updates to the NSM Proveit 1 n\2 4 6 and 8 CN]C CN]C CN]C nH2 ]HCn`1N]H 2H4 ]C2H3]HCN 2H4 ]C3H3N]H 2 m\2 3 . . . 7 2 2 3 3 3 5 3 4 CN]O CH]H C2H]CmH2 ]Cm`2H2]H C C C 2H]O2 ]CH]CO2 2H]O2 ]H]CO]CO 2H]O2 ]HCO]CO 2 ]O]OCN 2 ]CH3 reaction deleted CN]O2 ]CO]NO 5 p\2 3 . . . 8 Proveit 2 C]CpH2 ]Cp`1H]H a Full references are given with the comment. total N Fig. 20 Molecular production efficiency agree/Ntotal where Nagree is the number of species with calculated abundances within one order of magnitude of the observed abundances and N is the total number of species detected in TMC-1 is plotted vs.time for the three models discussed. 251 General Discussion Proveit 2 we also included the reaction between C and C2H2 and analogs and adopted the room temperature rate constant not knowing about the study that Dr Sims just described. From Fig. 20 one can see that the addition of reaction classes well studied at low temperature (Proveit 1) maintains the level of agreement with the observations in TMC-1. The introduction of reaction classes which have not been studied at low temperatures and the extrapolations people make are critical for the models. So the negative conclusions concerning models with rapid neutral»neutral reactions may be overly harsh.1 R. P. A. Bettens H-H. Lee and E. Herbst Astrophys. J. 1995 443 664. 2 I. R. Sims J-L. QueÜelec D. Travers B. R. Rowe L. B. Herbert J. Karthaeuser and I. W. M. Smith Chem. Phys. L ett. 1993 211 461. 3 D. Chastaing P. L. James I. R. Sims and I. W. M. Smith Faraday Discuss. 1998 109 165. 4 B. R. Rowe A. Canosa and I. R. Sims J. Chem. Soc. Faraday T rans 1993 89 2193 5 R. A. Brownsword I. R. Sims I. W. M. Smith D. W. A. Stewart A. Canosa and B. R. Rowe Astrophys. J. 1997 485 195. 6 N. Haider and D. Husain J. Photochem. Photobiol. A 1993 70 119. Dr Costes and Prof. Naulin said Laboratory studies of neutral»neutral atomexchange reactions (A]BC]AB]C) are essential for the comprehension of interstellar chemistry.In this context the contributions of the CRESU apparatus at Rennes1 and Birmingham2 are particularly relevant. Kaiser et al.3 have also demonstrated the bene–ts that are gained by crossed beam experiments which yield detection of reaction products in single collision conditions. However as noted by these authors themselves the coldest known interstellar clouds have kinetic energy of reactant molecules that can be as low as 0.08 kJ mol~1 which is considerably lower than the collision energies obtainable in their own experiments. A\740»3000 ms~1) with good velocity resolution (*lA/lA\ We recently designed in Bordeaux a crossed beam machine especially –tted to approach such very low collision energies. This new apparatus which employs laserinduced —uorescence detection has common features with the experiment of Kaiser et al.3 It thus has two diÜerentially pumped pulsed supersonic beam sources and also uses laser ablation to generate beams of refractory species such as carbon atoms.A large range of velocities (l 0.08»0.13) is obtainable for the A atom beam. A more restricted range (lBC\700»1400 ms~1 *lBC/*lBC\0.10»0.20) is obtainable for the BC molecular beam but the whole molecular beam source assembly is rotatable under vacuum with respect to the atom beam. Collisions at beam intersection angles varying between 22.5 and 90° with a higher limit that can be extended to 145° in some favourable cases are achieved. As a result an exceptionally large range of relative translational energies etr is accessible with the machine including those relevant to the conditions of the interstellar medium.The machine has been –rst tested with the Al(2PJ)]O2(X 3&g~)]AlO(X 2&`) reaction.4 Experiments performed with Al beams of the same velocity dis- ]O(3PJ) tribution but with diÜerent populations of spin»orbit states (obtained with a diÜerent carrier gas) have revealed substantial diÜerences in the Al spin»orbit reactivity. Hence Al(2P3@2) energy sampled (6.8 meV). The diÜerence in reactivity reduces when increasing collision reacts six times less than the ground-state Al(2P1@2) at the lowest collision energy and cancels at the highest energies (200»250 meV). Such an experiment demonstrates the importance that spin»orbit eÜects could have under conditions relevant to the interstellar chemistry.The translational energy dependence of the reaction cross section was also determined for the ground 2P state. It could be –tted between 6 and (etr)~0.82. Averaging over a Maxwell distribution of veloci- 200 meV by a dependency in 1@2 ties yields a temperature dependance of the rate coefficient in T ~0.32 in very good agreement with recent CRESU experiments at Rennes.5 252 General Discussion Fig. 21 Log»log plot of the detailed reactive cross section (in arbitrary units) of the C]NO]CN]O reaction for CN(X 2&` lA\2 JA\18) vs. relative translational energy A quick run was also performed very recently on the C(3PJ) photon laser-induced —uorescence scheme at 280 nm and CN radicals were probed by ]NO(X2%r)]CN(X 2&`)]O(3PJ) reaction.Carbon atoms were detected by a two LIF on the (B 2&`^X2&`) *l\[2 sequence around 460 nm. Fig. 21 which plots the dependence of the detailed reaction cross section for CN(X 2&` lA\2 JA\18) between 4.4 and 56 meV demonstrates that this is another neutral»neutral reaction without an energy barrier. 1 S. D. Le Picard A. Canosa and B. R. Rowe Faraday Discuss. 1998 109 poster. 2 D. Chastaing P. L. James I. R. Sims and I. W. M. Smith Faraday Discuss. 1998 109 165. 3 R. I. Kaiser C. Ochsenfeld D. Stranges M. Head-Gordon and Y. T. Lee Faraday Discuss. 1998 109 183. 4 M. Costes and C. Naulin unpublished work. 5 S. D. Le Picard A. Canosa D. Travers D. Chastaing B. R. Rowe and T. Stoecklin J. Phys.Chem. 1997 101 9988. Prof. Casavecchia and co-workers§ commented In relation to the interesting work of Kaiser et al. on combined crossed molecular beams and ab initio investigation of carbon-bearing molecules in the interstellar medium via neutral»neutral reactions we would like to report on similar studies on the formation of nitrogen-bearing molecules via neutral»neutral reactions. In our laboratory during the last few years we have been investigating in crossed molecular beams a series of reactions of potential relevance in astrochemistry such as those of oxygen atoms with H2S H and and those of OH 2 radicals with H and CO.1 Very recently we have been investigating reactions of nitro- 2 N(2D)]C2H2 reaction at a collision energy (E of 9.5 kcal mol~1 Fig.22 Laboratory angular distribution of the HCCN product (detected at m/z\38) from the c) 253 General Discussion with N(2D) [the reaction of N(4S) with C2H2 C2H4 as well as that with is gen atoms in the electronically excited 2D state N(2D) with some simple unsaturated hydrocarbons as possible one-step neutral processes for the formation of nitriles in the atmosphere of Titan. We have carried out product angular and velocity distribution measurements under single-collision conditions in crossed beam experiments2 with mass spectrometric detection using supersonic beams of N atoms generated by radiofrequency discharge3 in high pressure N2»rare gas mixtures [a magnetic analysis4 by a Stern»Gerlach magnet has given the following electronic state beam composition N(4S) 72% N(2D) 21% N(2P) 7%].These are the –rst successful reactive scattering experiments involving atomic nitrogen. From product angular (see Fig. 22) and velocity distribution measurements at diÜerent collision energies we conclude that HCCN (cyanomethylene) formation is the major reaction pathway and arises from the reaction of C2H2 nearly thermoneutral and has high activation energy,5 while the possible reactions of N(2P) are expected to be slow on the basis of simple adiabatic electronic energy correlation diagrams]. The reaction is found to proceed via formation of a long-lived complex at low collision energy (3.1 kcal mol~1) and via an osculating complex (i.e. a complex whose lifetime is comparable to its rotational period) at higher collision energy (9.5 kcal mol~1).2 Very recently Takayanagi and co-workers6 have carried out high quality electronic N(2D)]C2H2 reaction.Their work shows that N(2D) attacks the intermediate radical complex by (1,2) H-shift and ring opening or to a some- 2H2 .2,6 This suggests that 2H2 . structure calculations on the geometry and energy of the possible reaction intermediates and products of the two C atoms of acetylene forming a cyclic intermediate which can lead to the very stable CH what less stable linear HNCCH intermediate by simple ring opening. With the support 2CN of these calculations we conclude that HCCN formation proceeds through the cyanomethyl (H CCN) complex intermediate (see Fig. 23).2 The internally hot H CCN 2 2 radical will then dissociate under single collision conditions to HCCN]H by CwH bond cleavage.The HCCN radical is found to be highly internally excited. Although formation of the cyclic-HCCN isomer cannot be ruled out theoretical calculations indicate that cyclic triplet HCCN (more stable than the singlet form by about 40 kcal mol~1)6,7 does not correlate adiabatically with N(2D)]C formation of cyclic HCCN does not occur readily in the reaction N(2D)]C esting similarities as well as diÜerences are noted with respect to the results obtained by 2H2 . Inter- Kaiser et al.8 for the similar reaction C(3P)]C Our results strongly support the hypothesis made about ten years ago by Yung9 about the origin and role of HCCN radicals to explain the Voyager observations on Titan and suggest that the N(2D)]C2H2 reaction may well be the –rst step in the Fig.23 Schematic energy level and correlation diagram for the N(2D)]C2H2 ]HCCN]H reaction (based on theoretical calculations by Takayanagi and co-workers2,6,10) 254 General Discussion reaction leading presumably to CH3CN (acetonitrile) formation a species also formation of nitriles and should be included in chemical reaction networks modeling of the Titan atmosphere. The H-displacement pathway has also been observed in the study of the N(2D) ]C detected in Titanœs atmosphere. Studies such as these are expected to play an important 2H4 role in understanding data of the Cassini probe a spacecraft with the scope of analyzing Titanœs atmosphere within the Huygens»Cassini mission.1 M. Alagia N. Balucani L. Cartechini P. Casavecchia and G. G. Volpi in IAU Symposium No. 178 ìMolecules in Astrophysics Probes and Processesœ ed. E. F. van Dishoeck Kluwer Dordrecht 1997 pp. 271»280. 2 M. Alagia N. Balucani L. Cartechini P. Casavecchia and G. G. Volpi in preparation. 3 M. Alagia N. Balucani P. Casavecchia D. Stranges and G. G. Volpi J. Chem. Soc. Faraday T rans. 1995 91 575. 4 M. Alagia V. Aquilanti D. Ascenzi N. Balucani D. Cappelletti L. Cartechini P. Casavecchia F. Pirani G. Sanchini and G. G. Volpi Isr. J. Chem. 1997 37 329. 5 J. V. Michael Chem. Phys. L ett. 1980 76 129. 6 T. Takayanagi personal communication. 7 N. Goldberg A. Fiedler and H. Schwarz J. Phys. Chem. 1995 99 15327. 8 R. I. Kaiser C. Ochsenfeld M. Head-Gordon Y.T. Lee and A. G. Suits Science 1996 274 1506. 9 Y. L. Yung Icarus 1987 72 468. 10 T. Takayanagi Y. Kurosaki K. Misawa M. Sugiura Y. Kobayashi K. Sato and S. Tsunashima J. Phys. Chem. A. 1998 in press. § M. Alagia N. Balucani L. Cartechini and G. G. Volpi. Prof. Casavecchia and co-workers§ communicated In relation to the paper presented by Kaiser et al. and speci–cally to the study of the reaction C(3P)]H2S]HCS ]H we would like to introduce the results we obtained by using the same experimental technique (crossed molecular beams with mass spectrometric detection) on the O(3P)]H closely related reaction 2S]HSO]H which may also be of astrophysical relevance. In spite of the strong similarity between the two reacting atoms C(3P) and O(3P)»both having two unpaired 2p electrons but with the noteworthy diÜerence that C(3P) has an empty p orbital while O(3P) has a fully occupied p orbital»a comparison between the results by Kaiser et al.and our results points to diÜerent behaviours of the two reacting systems. According to Kaiserœs suggestion the reaction C(3P) carbonyl; owing to the small gap between the singlet and the triplet states of both 2,2- ]H2S]HCS]H H occurs by addition of C(3P) to 2S to form triplet 2,2-dihydrothiodihydrothiocarbonyl and thiohydroxycarbene the occurrence of intersystem crossing (ISC) before and/or after H-migration is called into play to account for the experimental –nding of the HCS isomer product.1 An argument in favour of ISC is the presence of a to the hydrogen atom displacement channel Fig.24 Schematic energy level and correlation diagram for the O(3P 1D)]H2S reaction relative General Discussion 255 sulfur atom whose relatively large atomic weight may make the spin conservation rule less rigid. Interestingly our results on the dynamics of O(3P)]H without invoking the occurrence of ISC to the singlet surface. Indeed we studied the 2S may be interpreted reaction dynamics of both the ground 3P and the –rst electronically excited 1D states of atomic oxygen under the same experimental conditions and for diÜerent values of the initial collision energy; this has permitted us to compare directly the dynamical behavior of the two electronic states.2h5 These studies were made possible by the capability of generating by high-pressure and high-power radio-frequency discharge continuous supersonic beams of atomic oxygen containing in addition to O(3P) also a small percentage of O(1D).6,7 Fig.24 depicts the energy level and correlation diagram for the triplet and singlet reaction leading to H-displacement. It should be noted that the triplet reaction can only lead to formation of the HSO isomer following addition of oxygen to the sulfur atom while the singlet reaction can lead to both HSO and HOS isomers following O(1D) insertion into the HS bond or O(1D) addition to the sulfur (the more exoergic by about 3.5 kcal mol~1,8 HSO]H channel is expected to be favored). In Fig. 25 the center-of-mass (cm) functions derived from measurements of angular and velocity distributions of the HSO(HOS) product at a collision energy of 8.15 kcal mol~1 are shown.Notable diÜerences between the cm functions for the reactions of the two states are present (1) the angular distribution for O(1D)]H2S is almost symmetric with respect to h\90° with a slight preference for forward (h\0°) scattering while that for O(3P)]H2S is strongly anisotropic with the backward direction (h\180°) favored ; (2) the translational energy distribution for O(1D)]H2S peaks at a value which corresponds to B20% of the total available energy while that for O(3P)]H2S peaks at a value close to B80%. Similar results have been found in a wide range of collision energies (up to 12 kcal mol~1) and have been interpreted in terms of two diÜerent reaction mechanisms.3,5 The reaction of O(1D) proceeds through the formation of a complex whose average lifetime is longer than (at Ec\3.4»4.8 kcal mol~1) or a fraction Fig.25 Upper Center-of-mass HSO product angular distributions from the O(1D) (»») and 2S at a collision energy of 8.15 kcal mol~1. Lower Product trans- O(3P) (» » ») reaction with H lational energy distributions for the same reactions plotted vs. the fraction of energy in translation. 256 General Discussion E (at c\6.7»12 kcal mol~1) of the complex rotational period ; the complex can either be the thioperoxide HOSH formed following the O(1D) insertion into one of the HwS bonds or the sulfoxide H SO which can be formed after isomerization of the thioperoxide or directly by addition of the oxygen atom to sulfur (see Fig. 24). On the 2 contrary the reaction of O(3P) with H2S follows a direct rebound mechanism on the triplet surface with a strong energy release in relative motion of the two moieties. In conclusion the dynamics followed by the triplet and singlet reactions are remarkably diÜerent implying no evidence of ISC between the surfaces of the two electronic states. If fact if a nonadiabatic transition occurred on the entrance channel from the triplet to the singlet potential energy surface (PES) the dynamics of HSO (HOS) formation from O(3P) would re—ect the in—uence of the deep well characterizing the singlet intermediate but the experimental results do not indicate this. The diÜerent behavior of the C( and 3P)]H2S O(3P)]H2S reactions can be attributed to the signi–cant diÜerences in the corresponding PES and particularly to the fact that no stable triplet intermediate exists for O(3P)]H2S as pointed out by detailed theoretical calculations.8 1 R. I. Kaiser C. Ochsenfeld M. Head-Gordon and Y. T. Lee Science 1998 279 1181. 2 N. Balucani P. Casavecchia D. Stranges and G. G. Volpi J. Chem. Phys. 1991 94 8611. 3 M. Alagia N. Balucani P. Casavecchia D. Stranges and G. G. Volpi J. Chem. Soc. Faraday T rans. 1995 91 575. 4 N. Balucani P. Casavecchia D. Stranges and G. G. Volpi Chem. Phys. L ett. 1993 211 469. 5 P. Casavecchia N. Balucani M. Alagia L. Cartechini and G. G. Volpi Acc. Chem. Res. 1988 submitted. 6 M. Alagia V. Aquilanti D. Ascenzi N. Balucani D. Cappelletti L. Cartechini P. Casavecchia F. Pirani G. Sanchini and G. G. Volpi Isr. J. Chem. 1997 37 329. 7 P. Casavecchia N. Balucani and G. G. Volpi in Advanced Series in Physical Chemistry. T he Chemical Dynamics and Kinetics of Small Radicals ed. K. Liu and A. Wagner World Scienti–c Singapore 1995 ch. 9. 8 A. Goumri J-D. R. Rocha D. Laakso C. E. Smith and P. Marshall J. Chem. Phys. 1984 101 9405; 1985 102 161. § N. Balucani L. Beneventi and G. G. Volpi.

ISSN:1359-6640    

DOI:10.1039/FD109217

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Faraday Discuss. 1998 109 257»266 Molecules in harsh environments John H. Black Onsala Space Observatory Chalmers University of T echnology S-439 92 Onsala Sweden In the past it has been conventional to view interstellar molecules as probes of cold low-energy environments. We now –nd many instances of molecules —ourishing in the presence of energetic radiation high temperatures and superthermal components of the molecular speed distribution. This study centres on reactive molecular ions as speci–c tracers of such harsh physical environments. Reactive ions such as H2 ` CH` and CO` constitute a special class because they are destroyed on virtually every collision with the most abundant species H H and e~. The abundance and excitation of CH` and CO` are discussed with reference to recent observations of 2 nebular boundary layers.The hydrogen molecular ions H2 ` and H3 ` achieve high abundances in X-irradiated molecular gas such as that in the hypothesized molecular tori surrounding active galactic nuclei. A population inversion is predicted in the (J K)\(4 4)](3 1) transition of H and the conditions required for maser ampli–cation are discussed. 3 ` SO` HCS` HOCO` H3 ` HOC` H3O` Reactive molecular ions in interstellar gas Several molecular ions are observed in interstellar gas (Table 1). These and other unseen ions are thought to play important roles in driving the gas-phase chemistry indeed their existence is clear proof that ion»molecule reactions are responsible for at least part of the remarkably complex interstellar chemistry.Interstellar clouds are predominantly neutral ionized fractions range from ca. 10~4 in regions that are translucent to ultraviolet starlight down to much lower values 10~9»10~7 in the centres of thick dark clouds where the source of ionization is penetrating cosmic rays. With two exceptions the abundances of the observed ions can be explained satisfactorily given what is known about the —uxes of ionizing photons and cosmic rays and the reactions with neutrals and electrons that remove these ions. Two of the observed species CH` and CO` are members of a special class of reactive molecular ions which are destroyed on virtually Table 1 Identi–ed molecular ions in interstellar gas CH` CO` H2D`(tent.) HCO` HCNH` HC3NH` N2H` H COH` 2 257 258 Molecules in harsh environments every collision with the most abundant neutral species H (1) (2) CH`]H]C`]H2 CO`]H]H`]CO and H2 (3) (4) CH`]H2 ]CH2 `]H CO`]H2 ]HCO`]H In addition both CH` and CO` are removed rapidly by dissociative recombination with electrons.The short chemical lifetimes of CH` and CO` make it very difficult to account for their observable abundances in the interstellar gas. CH` is widely observed through its absorption lines superimposed on the visible spectra of background stars.1,2 Its abundance in diÜuse molecular clouds [CH`]/[H2]P10~8 has presented a challenge for all theories of interstellar chemistry. Brie—y the gas-phase chemical sources of the observed CH in diÜuse molecular clouds fail by factors of 100 or more to explain the comparable abundance of the ion CH` because it is destroyed so much more rapidly than the neutral.It has been proposed that the endoergic reaction (5) C`]H2 ]CH`]H[(0.398 eV) is the main source of interstellar CH` but in a disturbed component of the gas. The ubiquity of CH` in the diÜuse interstellar medium thus requires that the energetic reactants must be very widespread. The nature of the disturbed component is debated. Molecular shock waves that heat some gas to temperatures T [1000 K can satisfy the energetic requirements of CH` formation but the predicted separations in Doppler velocity between the quiescent CH and the disturbed CH` are usually not observed.1h4 The superthermal component of interstellar gas that is associated with widespread turbulence might supply the energetic collisions of reaction (5) but the nature of the turbulence and its chemical eÜects are still controversial.5h8 A new chapter has been added to the story of CH` through the recent discovery of several pure rotational transitions in emission in the far-IR spectrum of the planetary nebula NGC 7027 which has a substantial molecular envelope.9 This is an example of an environment that is hazardous for molecules of any sort as it lies at the boundary of T a nebula of gas that has been heated to a kinetic temperature kB104 K and is highly T ionized by the radiation of a star of surface temperature *\1.4]105 K.Further analysis of the far-IR spectra of NGC 7027 obtained with the Infrared Space Observatory (ISO) has revealed rotational transitions of CH and has placed an interesting upper limit on the intensity of the J\1]0 line of HeH` another reactive ion.10 An early report of the detection of a single transition of interstellar CO` in the Orion molecular cloud11 was dismissed owing to confusion with emission lines of other well identi–ed species.More than a decade later a convincing identi–cation of this reactive molecular ion was made in NGC 7027 and in the molecular gas associated with M 17 an ionized nebula and star-forming region.12 CO` has since been measured by others in several rotational transitions (N\2]1 and 3]2) in additional sources,13h15 including the ìOrion Barœ that marks the interface between the nebula and a concentration of molecular gas.All the sources of CO` have in common a close association with a hot star. The detection in the re—ection nebula NGC 7023 is unusual because the B-type exciting star is too cool to maintain a substantial ionized nebula.15 It would appear natural to explain the CO` through known reactions in the photon-dominated regions (PDRs) at the nebular/molecular interfaces,16 but quantitative models of the Orion Bar fail by factors of 10»100 to account for the observed line intensities.13 There is a very recent report of high-N rotational lines of CO` in the far-IR spectrum of IRAS 259 J. H. Black The triatomic hydrogen molecular ion H 16293[2422 a young stellar object that drives a bipolar molecular out—ow deep inside the Ophiuchus molecular cloud.17 3 ` has long been believed to play a central to form H2 ` is followed by role in interstellar chemistry.Oka18 originally suggested that interstellar H3 ` might be observed through absorption lines of its l fundamental vibration»rotation band superimposed on the spectra of deeply buried IR stars. These lines remained just beyond the 2 sensitivity of astronomical IR spectrometers19,20 until Geballe and Oka21 reported the discovery of H3 ` in absorption toward two deeply buried young stellar objects. This ion has recently been observed in several additional molecular clouds.22,23 The ionization of H2 H2 `]H]H2]H` and H2 `]H2 ]H3 `]H except where the ionization is high enough that dissociative recombination competes (6) (7) (8) H2 `]e~]H]H All three of these processes are rapid and thus H2 ` is another member of the class of reactive molecular ions that are destroyed on virtually every collision.In a molecular gas where the ratio of number densities n(H2)/n(H)A1 reaction (7) is the principal outcome of the ionization of H2 H so that 3 ` results from almost all ionization events. The removal of H3 ` is dominated in the interstellar medium by reactions with the most abundant neutrals containing carbon and oxygen notably (9) (10) H3 `]O]OH`]H2 H3 `]CO]HCO`]H2 and in regions of relatively high ionization by dissociative recombination (11a) (11b) H3 `]e]H]H]H H3 `]e]H2]H. Consequently the number density of interstellar H3 ` is simple to estimate (I) n(H3 `)\x(e)k11]8]10~10x(O) eff ]1.7]10~9x(CO)]… … … cm~3 f is the eÜective rate (s~1 per H2) H at which ionization of leads to formation 2 where f 3 `eff x(i)\n(i)/n(H of H is the fractional abundance of species i\e~ O CO etc.and The determination of the rate coefficient of dissociative recombination [reaction (11)] 2) has a long and interesting history but experimental results seem –nally to have conk verged on a value 11B1]10~7(300/T )1@2 cm3 s~1 for ground-state ions.24 As is apparent from eqn. (I) the concentration of interstellar H3 ` is predicted to be independent of the total hydrogen density and directly related to the eÜective ionizing rate. 3 ` in interstellar molecular regions.Owing to so reactive that it leads immediately to H In the preceding paragraph H2 ` was quickly dispatched as a transient intermediary its high reactivity the possibility of observing H2 ` astronomically has been largely ignored aside from some discussions of its interesting hyper–ne-structure spectrum in the microwave region.25h28 Following the same arguments that lie behind eqn. (I) we estimate a concentration of H2 ` (II) n(H2 `)\n(H)k6]n(H2)k7]n(e)k8 B4]108feff cm~3 fn(H2) 260 Molecules in harsh environments in a molecular region where x(e)@1. Where the gaseous oxygen abundance is typically x(O)]x(CO)]… … …B5]10~4 we thus expect (III) n(H3 `)/n(H2 `)B5000 In the following sections the observability of the molecular ions is analysed in order to illustrate their uses as probes of disturbed molecular environments.Coupled chemistry and excitation of reactive ions In the neutral interstellar medium where H and H are the most abundant species the CH` and CO`) will be destroyed on virtually 2 ` ions that react rapidly with them (H 2 every collision reactions (1)»(4) and (6) and (7) are known to proceed at nearly the gas-kinetic rates. Rotationally inelastic collisions with H and H are no faster than the reactive processes and therefore can never fully thermalize the rotational population 2 distributions of the reactive ions. Indeed if such an ion is formed with kinetic energy much greater than the mean thermal energy elastic collisions with H or H may be unable to thermalize its translational motions.These points suggest several interesting 2 consequences for the observed properties of the reactive ions. First radiative processes may compete with J-dependent formation destruction and collisional excitation processes in establishing non-thermal population distributions. Secondly if radiative processes do not completely control the population distributions then these may retain some memory of the energetics of the formation process. Thirdly if elastic collisions are not signi–cantly faster than reactive collisions then non-Maxwellian (or super-thermal) speed distributions may be observable through line shapes. Finally if the ions are highly excited their reactivities may be aÜected. Taken together these remarks suggest that the excitation and abundances must be considered together in a coupled way if accurate abundances and physical parameters are to be derived from astronomical spectra.We have been developing techniques for describing self-consistently the formation destruction and excitation of the reactive molecular ions. Some of these considerations are illustrated below for the example of nebular CH`. Further details will be published elsewhere.29 Nebular CHë The discovery of several pure rotational lines J\2»1 3»2 4»3 5»4 and 6»5 of CH` in emission in the far-IR spectrum of the planetary nebula NGC 7027 provides an interesting test case.9 The authors show that the CH` has a nearly uniform rotational excita- T tion temperature ex\150 K which means that it radiates as though in equilibrium at this temperature.They make the conventional assumption that inelastic collisions are rapid enough to achieve such thermalization in competition with the radiative rates ; T that is they assume a kinetic temperature k\Tex . Because the spontaneous transition A probabilities A21\0.061 s~1 to A65\ of the observed transitions lie in the range J{J_ 1.87 s~1 the density of neutral collision partners would need to be greater than Aul/qulB109 cm~3 even if the collision rate coefficients were as large as qulB10~9 cm3 s~1. If electron collisions dominated at rates qul e B10~7 cm3 s~1 then electron densities greater than 107 cm~3 would be required. The labels u and l refer to the upper and lower states respectively. Such densities are orders of magnitude higher than have been inferred for NGC 7027 by other means.As suggested above however it is very unlikely that non-reactive inelastic collisions can thermalize rotational populations in CH` because the reactive collisions that destroy it are so rapid. Another important factor in the excitation of CH` was not taken into account fully. Observations of NGC 7027 show that the intensity of continuous radiation is quite high in the same wavelength l8 261 J. H. Black lB region in which the lines appear. f 30,31 The observed continuum —uxes can be used to Il\fl/X averaged over the solid place lower limits on the internal intensity of radiation l angle subtended by the nebula XB2.4]10~9 sr. This continuous spectrum in the wavelength range 50»350 lm is consistent with thermal emission by dust particles that have a frequency-dependent emissivity Pl at Tdust\74 K.The intensity of this contin- IlPlBl(Tdust) where B uous spectrum can be expressed l(T ) is the Planck function. By coincidence the function l(T \74) has very closely the same frequency dependence as Bl(T \150) over the frequency range of interest for CH`. In other words if radiative processes are faster than all the collisional processes the CH` molecules will equilibrate T to a temperature ex\150 K that corresponds to the spectral shape of the radiation that they absorb most strongly. Indeed the average internal intensity of far-IR radiation in NGC 7027 produces absorption rates of at least 10~4 s~1 in each of the observed transitions.The true intensity may be somewhat higher if the CH` molecules and the emitting dust particles occupy the same small fraction of the projected size of the nebula. It can be shown that the visible and near-IR photons in NGC 7027 are less important in pumping CH` through its electronic and vibrational transitions. If we suppose that the observed CH` arises in the neutral molecular envelope at the boundary of the ionized nebula,32 its destruction is dominated by reactions (1) and (3). This permits a formation rate of CH` to be estimated from the observed emission line intensities. A preliminary estimate of the production rate of CH` is B3]10~10 s~1 per hydrogen for a destruction rate of ca. 10~5 s~1 which suggests a fractional abundance [CH`]/[H]2H as large as 2]B3]10~5.In combination with our analysis of the radiative excitation this implies that each CH` ion can radiate 10 or more line photons during its lifetime. This must be a more efficient source of excitation than inelastic collisions whose eÜect will always be limited by the rapid reactive processes. In the future it will be of interest to try to make a detailed accounting of the CH` formation processes that might satisfy these requirements. COë in nebular boundaries Similar considerations apply to the analysis of the abundance and excitation of CO` which has been observed in the boundary regions of several nebulae.12h15 The observed features of the mm-wave CO` emission lines are interesting the line widths are typically narrow 1 to 2 km s~1 (Doppler full-width at half-maximum) and the rotational excitation temperatures appear to be low (TexB30 K).The line widths and corresponding three-dimensional velocity dispersions p suggest that the mean kinetic energy of CO` is mp2/2B0.25 eV. If the destruction of CO` is so rapid that its translational motions do not become thermalized one would expect the mean kinetic energy to be comparable to the kinetic energy upon formation. In that case the narrow line widths might indicate that CO` is not formed by a highly exoergic process. Preliminary calculations of the rotational excitation of CO` including the formation and destruction rates explicitly suggest that the rotational excitation can remain subthermal (Tex\Tk) even where the density is high by interstellar standards.It will be interesting to pursue this analysis further to see if the observations provide any important clues to identifying the speci–c processes that form CO` in detectable amounts. Abundance and excitation of H3 ë ul Thanks to the eÜorts of spectroscopists and molecular theorists the vibration»rotation spectrum of H3 ` is known and its distortion-induced pure rotational transitions are accurately predicted.33h36 In the calculations described below the theoretical transition frequencies (cm~1) and spontaneous transition probabilities Aul (s~1) have been 262 (a) mm-wave maser transition JA K@ J@ 1 3 4 4 (b) Selected far-IR rotational lines JA K@ J@ 2 2 1 1 0 1 3 1 2 3 2 4 1 3 2 3 3 2 1 1 4 2 3 2 0 2 1 3 2 5 2 4 3 3 4 3 JA U@ G@ J@ 4 3 1 2 3 3 2 2 1 1 1 1 2 0 [1 3 1 1 1 2 1 2 1 1 4 3 1 2 0 3 1 2 1 3 2 3 2 4 4 3 3 3 3 2 2 2 4 3 3 2 4 3 4 3 2 1 [1 6 2 5 2 6 2 5 2 In this table depth q.Table 2 Strong emission lines of H3 ` in a model of Cygnus A KA l/GHz KA j/lm (c) Selected IR vibration»rotation lines KA j/lm f is the integrated line —ux and f is the —ux in the continuum at the same frequency. The line c entries for the maser transition include the rotational excitation temperature Tex and the line-centre optical taken from the recent database of Neale et al.36 Cross-sections for the collision-induced transitions are unknown; therefore illustrative values of the rate coefficients for downward transitions induced by collisions with H2 qul\10~10 to 10~9 cm3 s~1 have been assumed for all transitions that are radiatively allowed.Similar rate coefficients have been adopted for reactive collisions with H2 that change the nuclear spin species (ortho%para) subject to *J\^1 within the vibrational ground state. In order to illustrate the intensities of H3 ` emission lines that might be possible we n(H consider an X-irradiated molecular gas with a number density and a kinetic temperature T Molecules in harsh environments Tk\100 K T q ex/K 217.853 [70.3 [0.000 063 Tk\100 K fline/W m~2 2.11([21) 1.48([20) 2.08([21) 4.49([20) 2.92([20) 1.04([19) 2.05([19) 2.72([20) 149.8186 146.9352 104.8327 95.0788 70.5048 52.4480 49.6194 36.5355 30.7248 1.36([19) Tk\100 K fline/W m~2 2.75([19) 2.30([19) 3.54([19) 3.95([19) 3.03([19) 3.56([19) 3.97([19) 3.27([19) 3.51([19) 8.77([20) 2.10([19) 1.62([19) 4.5097 4.5083 4.3562 4.3555 4.3554 4.3498 4.2155 4.2037 3.9711 2.1944 2.1380 2.0407 2.0259 1.27([19) 2)\5]104 cm~3 kP100 K.A speci–c instance is the suspected molecular fline/fc 0.000 083 fline/fc 0.072 0.497 0.053 1.06 0.537 1.50 2.83 0.293 1.27 fline/fc 0.532 0.444 0.666 0.742 0.570 0.668 0.726 0.598 0.612 0.094 0.220 0.164 0.128 Tk\1000 K q fline/fc Tex/K 0.069 [14.5 [0.036 Tk\1000 K fline/fc fline/W m~2 5.76([20) 2.86([20) 2.47([19) 2.98([20) 3.13([18) 1.92([18) 1.05([17) 4.72([18) 0.623 0.304 1.99 0.222 18.3 8.80 46.0 16.1 11.0 3.72([18) Tk\1000 K fline/fc fline/W m~2 1.80([18) 1.72([18) 1.03([18) 9.37([19) 1.30([18) 1.28([18) 7.69([19) 8.99([19) 1.09([18) 4.59([19) 2.93([19) 6.74([19) 1.10 1.05 0.614 0.557 0.771 0.759 0.445 0.519 0.598 0.156 0.097 0.215 0.189 5.96([19) 263 J.H. Black TkP100 K. The corresponding ionization rate in eqn. (I) torus surrounding the active nucleus of the radio galaxy Cygnus A.The X-ray lumi- L nosity of the central source is XB1038 W; therefore gas at a distance R\300 pc (B1000 light-years\9]1018 m) from this source will have an eÜective ionization parameter meffB10~3 which is a ratio of photon density to gas density in the nomenclature of Maloney et al.37 Such a value of meff permits the gas to be mostly molecular and supports a temperature effB10~12 n(H3 `)B0.1 cm~3 f and (II) is s~1 at the adopted density. In this situation [eqn. (I)] and x(e~)B10~4 so that reaction (11) controls the removal of H3 `. If the N total hydrogen column density is H\2/n(H2)dr\1023 cm~2 then a column density as large as N(H3 `)\1017 cm~2 is possible. The temperature depends on details of the cooling processes and the abundances of elements such as carbon nitrogen and oxygen; T therefore a sample of line intensities is presented for two limiting values k\100 and 1000 K.The continuum radiation of the central source from microwave to X-ray wavelengths is known.38 The coupled equations of statistical equilibrium and radiative trans- —uxes are presented for a few of the strongest lines in Table 2. Thermal Doppler line fer have been solved for 354 radiative transitions involving 77 levels of H3 `. Predicted widths have been assumed. This may be a useful assumption for a small parcel of the molecular gas but will represent poorly the kinematical line broadening of a rotating torus in the centre of an active galaxy. Are the tabulated —uxes measurable? The tabulated line/continuum ratios for Cygnus A are all at least 0.5.Thus if the continuum radiation can be detected spectroscopically with resolution comparable to the linewidth then the line will be detected even at low signal-to-noise ratios. The column density of H3 ` may be overestimated of course but the predicted —uxes in Table 2 illustrate what might be expected in such an extremely harsh molecular environment. There are several interesting features of the excitation of H3 ` near an AGN like Tk\100 K the kinetic temperature that characterizes the coll and 2l vibrational bands T 2 radB200 to 350 K. 2 Tk\1000 K the collisions contribute relatively more to the vibrational excita- Cygnus A. First the results for the two temperatures shown in Table 2 represent quite diÜerent regimes. At lisional rates is smaller than the radiation brightness temperature in the model cloud at the wavelengths 2 to 5 lm of the Thus the excitation will tend to be radiation dominated.At the higher kinetic temperature tion. Secondly several pairs of levels show persistent population inversions. The most interesting of these arises in the (J K)\(4 4)](3 1) rotational transition which is the transition of very lowest frequency among all the low-lying states. It is easy to see why there is a population inversion. Fig. 1 shows the rotational energy level structure for all states of H3 ` that lie within 1050 cm~1 of the (1 1) ground level. There is only one possible radiative transition out of (4 4) namely the (4 4)](3 1) transition itself for which A\2.78]10~9 s~1 according to Miller and Tennyson,35 while Pan and Oka33 calculated A\4.1]10~9 s~1.In contrast there are two transitions at much higher frequencies out of the (3 1) state with a combined transition probability of 3.77]10~5. As a result the lower state of the (4 4)](3 1) transition will always be depopulated much faster than the upper state at low densities as long as there is sufficient excitation to reach these states which lie at *E/kB600 K above (1 1) the upper state will tend to be relatively overpopulated. The inversion will be quenched by collisions at H densities 2 of the order of 106 cm~3 or higher when both states are depopulated by collisions rather than by radiative transitions. The inversion is a necessary condition for an astrophysical maser.The other requirement is a sufficient column density per line width to achieve ampli–cation. In the illustrative model presented here we –nd inversion but negligible ampli–cation at Tk\100 K. However at Tk\1000 K the ampli–cation factor (gain) is already 1.04 and it would of course grow non-linearly for slightly larger column densities. Therefore a new astrophysical maser is predicted in the (J K)\(4 4)](3 1) rotational transition of H3 `. The frequency of this transition was originally calculated to be 7.178 cm~1 by Pan and Oka33 while the more elaborate methods used 264 Molecules in harsh environments Fig. 1 Energy level diagram shows the lowest rotational levels of H3 ` with energy E(J K) in cm~1 displayed vs. the K quantum number. The value of the rotational quantum number is indicated alongside each level.Dashed lines indicate the single transition out of the (J K)\(4 4) level and the two possible transitions out of the (3 1) level. by Neale et al.36 yield 7.267 cm~1 or 217.85 GHz. There is an urgent need for a more accurate transition frequency if radioastronomical searches for this line are to be unambiguous there are several identi–ed interstellar molecules with observed lines that lie within 40 MHz of this frequency. Even in the absence of ampli–cation this transition may be detectable in interstellar gas that is exposed to high ionizing —uxes. Two other pairs of levels will sometimes have inverted populations (5 4)[(4 1) with a transition at 104.8 lm and (2 1)[(2 2) with a transition at 146.9 lm.Further the highly metastable (3 3) and (5 5) states can build up excessive populations as pointed out by Pan and Oka. Is H2 ë The diatomic hydrogen ion has two handicaps relative to H3 ` (1) its vibrational and rotational spectrum consists of even less probable electric quadrupole transitions and (2) it will usually be less abundant than the triatomic ion [cf. eqn. (III)]. If H2 ` is produced solely by ionization of H2 then the concentration in state (v J) can be expressed (IV) observable? f vJ(H2 `)B 1.5]10~9n(H2)];v_ J_A(vJ]vAJA) cm~3 n eff n(H2)e(v J) 265 J. H. Black where e(v J) is the fraction of ionizing events that yields H2 ` in state (v J). At the density considered above in the discussion of H3 ` n(H2)\5]104 cm~3 H2 `(v J) will be destroyed by reaction (7) ca.100 times faster than it can radiate through quadrupole transitions to lower states because the vibrational transition probabilities39 are of the order of 10~7 s~1. This implies that the total emission rate in H2 ` vibration»rotation lines will be ca. 1% of its production rate. Thus a characteristic —ux in a vibration» rotation line is (16) fB nvJ(H2 `)hcl8 A(vJ]vAJA)4nR2 dR W m~2 4nD2 where A(vJ]vAJA) is the spontaneous transition probability and hcl8 is the photon energy. In the molecular torus of Cygnus A described above where RB300 pc and dRB0.32 pc are the inner radius and thickness respectively and D is the distance to the 10~12 s~1BCe 0.075 (v J)DA2000 l8 cm~1BC3 A ] (vJ 10 ] ~7 vA s J ~1 A) DW m~2 (17) source, fB1.0]10~20A feff 2 ` lines at wavelengths 4»5 lm will be ca.100 times In other words the strongest H weaker than the strongest H3 ` lines listed in Table 2. Conclusions Several reactive molecular ions have been discussed as interesting probes of astrophysical environments that would seem hazardous to molecules in general. In the cases of CH` and CO` the reactive processes that destroy them are so rapid that their observable rotational excitation is probably coupled directly to their formation and removal. The discussion of H3 ` in X-irradiated molecular gas shows further remarkable features of that beautifully complicated ion. In particular it is predicted that a population inversion will commonly occur in the (J K)\(4 4)](3 1) rotational transition.If the abundance is sufficient an astrophysical maser will arise. The line frequency is poorly determined by the standards of mm-wave radio astronomy a determination of it in the laboratory would provide an interesting and important challenge. It would also be of great value to know something about the inelastic collisions between H and H3 ` that cause vibrational and rotational excitation in the ion. 2 This work has received generous support from Chalmers University of Technology. Note added in proof There is a recent discussion of the selection rules for reactive collisions of H and H3 ` in which the nuclear spin species change (D. Uy M. Cor- 2 donnier and T. Oka Phys. Rev.L ett. 1997 78 3844). This will aid in the description of realistic collision rates that are important in the rotational excitation of H3 `. The excitation of IR lines of H3 ` in an X-ray heated molecular gas has previously been discussed by B. T. Draine and D. T. Woods (Astrophys. J. 1990 363 464). Those authors reach somewhat diÜerent conclusions regarding the strongest lines from the results presented here (Table 2). References 1 R. Gredel E. F. van Dishoeck and J. H. Black Astron. Astrophys. 1993 269 477. 2 R. Gredel Astron. Astrophys. 1997 320 929. 3 I. A. Crawford Mon. Not. R. Astron. Soc. 1995 277 458. 4 P. Crane D. L. Lambert and Y. SheÜer Astrophys. J. Suppl. 1995 99 107. 5 M. Spaans PhD Thesis Leiden University 1995; M. Spaans J. H. Black and E.F. van Dishoeck 1997 unpublished results. 266 Molecules in harsh environments 6 W. W. Duley T. W. Hartquist A. Sternberg R. Wagenblast and D. A. Williams Mon. Not. R. Astron. Soc. 1992 255 463. 7 T. W. Hartquist J. E. Dyson and D. A. Williams Mon. Not. R. Astron. Soc. 1992 257 419. 8 E. Falgarone G. Pineau des Fore� ts and E. RoueÜ Astron. Astrophys. 1995 300 870. 9 J. Cernicharo X-W. Liu E. Gonzaç lez-Alfonso P. Cox M. J. Barlow T. Lim and B. M. Swinyard Astrophys. J. 1997 483 L65. 10 X-W. Liu M. J. Barlow A. Dalgarno J. Tennyson T. Lim B. M. Swinyard J. Cernicharo P. Cox J-P. Baluteau D. Peç quignot Nguyen-Q-Rieu R. J. Emery and P. E. Clegg Mon. Not. R. Astron. Soc. 1997 290 L71. 11 N. R. Erickson R. L. Snell R. B. Loren L. Mundy and R.L. Plambeck Astrophys. J. 1981 245 L83. 12 W. B. Latter C. K. Walker and P. R. Maloney Astrophys. J. 1993 419 L97. 13 D. Jansen M. Spaans M. Hogerheijde and E. F. van Dishoeck Astron. Astrophys. 1995 303 541. 14 H. Stoé rzer J. Stutzki and A. Sternberg Astron. Astrophys. 1995 296 L9. 15 A. Fuente and J. Martïç n-Pintado Astrophys. J. 1997 477 L107. 16 A. Sternberg and A. Dalgarno Astrophys. J. Suppl. 1995 99 565. 17 A. Rudolph C. Ceccarelli E. Caux M. Wol–re B. Nisini P. Saraceno and G. J. White Bull. Am. Astron. Soc. 1997 Meeting 191 abstract 12104. 18 T. Oka Phil. T rans. R. Soc. L ondon A 1981 303 543. 19 T. R. Geballe and T. Oka Astrophys. J. 1989 342 855. 20 J. H. Black E. F. van Dishoeck S. P. Willner and R. C. Woods Astrophys. J. 1990 358 459.21 T. R. Geballe and T. Oka Nature (L ondon) 1996 384 334. 22 T. R. Geballe B. J. McCall and T. Oka in Star Formation with the Infrared Space Observatory (Vol. 132 of the ASP Conference Series) ed. Joao Yun and Reneç Liseau Astronomical Society of the Paci–c San Francisco 1998 p. 72. 23 C. A. Kulesa and J. H. Black 1998 in preparation. 24 M. Larsson Annu. Rev. Phys. Chem. 1997 48 151. 25 K. B. JeÜerts A. A. Penzias D. F. Dickinson A. E. Lilley and H. Pen–eld Astrophys. J. 1968 154 389. 26 K. B. JeÜerts A. A. Penzias J. A. Ball D. F. Dickinson and A. E. Lilley Astrophys. J. 1970 159 L15. 27 P. M. Kalaghan PhD Thesis Harvard University 1972. 28 W. L. H. Shuter D. R. W. Williams S. R. Kulkarni and C. Heiles Astrophys. J. 1986 306 255. 29 J. Lindholm and J. H. Black 1998 in preparation. 30 X-W. Liu M. J. Barlow Nguyen-Q-Rieu Truong-Bach P. Cox D. Peç quignot P. E. Clegg B. M. Swinyard M. J. Griffin J. P. Baluteau T. Lim C. J. Skinner H. A. Smith P. A. R. Ade I. Furniss W. A. Towlson S. J. Unger K. J. King G. R. Davis M. Cohen R. J. Emery J. Fischer W. M. Glencross E. Caux M. A. Greenhouse C. Gry M. Joubert. D. Lorenzetti B. Nisini A. Omont R. Orfei P. Saraceno G. Serra H. J. Walker C. Armand M. Burgdorf A. Di Giorgio S. Molinari M. Price D. Texier S. Sidher and N. Trams Astron. Astrophys. 1996 315 L257. 31 G. R. Knapp G. Sandell and E. I. Robson Astrophys. J. Suppl. 1993 88 173. 32 J. R. Graham E. Serabyn T. M. Herbst K. Matthews G. Neugebauer and B. T. Soifer Astron. J. 1993 105 250. 33 F-S. Pan and T. Oka Astrophys. J. 1986 305 518. 34 J. Tennyson Rep. Prog. Phys. 1995 58 421. 35 S. Miller and J. Tennyson Astrophys. J. 1988 335 486. 36 L. Neale S. Miller and J. Tennyson Astrophys. J. 1996 464 516. 37 P. R. Maloney D. J. Hollenbach and A. G. G. M. Tielens Astphys. J. 1996 466 561. 38 J. H. Black in T he Molecular Astrophysics of Stars and Galaxies ed. T. W. Hartquist and D. A. Williams Oxford University Press Oxford 1998 in press. 39 A. G. Posen A. Dalgarno and J. M. Peek Atomic Data Nucl. Data T ables 1983 28 265. Paper 8/00591E; Received 21st January 1998

ISSN:1359-6640    

DOI:10.1039/a800591e

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Faraday Discuss. 1998 109 267»280 in dense and diÜuse clouds H 2 3 ë Benjamin J. McCall,a* Kenneth H. Hinkle,b Thomas R. Geballec and Takeshi Okaa a Department of Astronomy and Astrophysics Department of Chemistry and the Enrico Fermi Institute T he University of Chicago Chicago IL 60637 USA b National Optical Astronomy Observatories T ucson AZ 85726 USA c Joint Astronomy Centre University Park Hilo HI 96720 USA Interstellar H3 ` has been detected in dense as well as diÜuse clouds using three 3.7 lm infrared spectral lines of the l fundamental band. Column densities of H3 ` from (1.7»5.5)]1014 cm~2 have been measured in dense clouds in absorption against the infrared continua of the deeply embedded young stellar objects GL2136 W33A MonR2 IRS 3 GL961E and GL2591.Strong and broad H3 ` absorptions have been detected in dense and diÜuse clouds towards GC IRS 3 and GCS3-2 in the region of the galactic center. A large column density of H3 ` comparable to that of a dense cloud has been detected towards the visible star Cygnus OB2 No. 12 which has a line of sight that crosses mostly diÜuse clouds. The H3 ` chemistry of dense and diÜuse clouds are discussed using a very simple model. Some future projects and problems are discussed. 1 Background Protonated hydrogen H3 ` is the simplest stable polyatomic molecule and was discovered in 1911 by J. J. Thompson.1 It is the most abundant ion in hydrogen plasmas as initially discovered by A. J. Dempster.2 In 1925 Hogness and Lunn3 introduced the celebrated ion»neutral reaction (I) H2 `]H2 ]H3 `]H as the primary mechanism for H3 ` production.By the 1930s the predominance of H3 ` among cations in hydrogen plasmas was well established experimentally,4 and the systematic theoretical studies by Eyring Hirschfelder and others had explained the large cross-section5 and high exothermicity6 of reaction (I). Readers are referred to a review7 for more details of early works. 1.1 Interstellar H3 ë The 1961 paper by Martin et al.8 seems to be the –rst to suggest that H3 ` should be abundant in interstellar space. In 1970 Stecher and Williams9 discussed the production and destruction rates of interstellar H H3 ` concentration was reported by Solomon and Werner,10 who also took the decisive 3 `.The –rst numerical estimate of the interstellar step of introducing the cosmic ray as the major agent of ionization. Their estimate of the H3 ` fraction X(H3 `)4[H3 `]/[&H]B10~6 (where [&H] denotes the total number density of hydrogen atoms) can be contrasted to the value 10~8 derived in this paper. de 267 H 268 3 `)B0.4»1.0]10~6. Reaction (I) was 2 3 ` in dense and diÜuse clouds Jong11 did similar calculations and obtained X(H also used by Glassgold and Langer12 as the mechanism for cosmic ray heating of molecular clouds and by Watson13 in his theory of isotope fractionation in interstellar HD. In 1973 the science of interstellar H3 ` acquired a new dimension when Watson14 and Herbst and Klemperer15 independently proposed a network of ion»neutral reactions as the mechanism to generate the wide variety of simple molecules that had been observed in interstellar space by radioastronomers.16 This idea which was perhaps in—uenced by the millimeter wave detection of X-ogen by Buhl and Snyder17 and its subsequent identi–cation as HCO` by Klemperer,18 revealed that H3 ` plays a central role in interstellar chemistry.Because of the relatively low proton affinity of H (4.5 eV) H3 ` protonates practically all atoms and molecules through the general reaction (II) 2 H3 `]X]HX`]H2 (He Ne Ar N and O are notable exceptions). After protonation the HX` combines with other species through the reaction (III) HX`]Y]XY`]H and initiates a network of chemical reactions.The detailed numerical model calculation of such networks in the classic paper by Herbst and Klemperer15 explained many of the observed results. Their success triggered an avalanche of papers and reviews on interstellar chemistry based on the ion»neutral reaction scheme. While they are too numerous to cite many important papers can be traced from the references given in three works that were essential in the preparation of this discussion the paper by de Jong et al.19 on H3 ` chemistry the chemical model calculation for diÜuse clouds by van Dishoeck and Black,20 and the model for dense clouds by Lee et al.21 1.2 The search for H3 ë 2 ì It is likely that H3 ` is present in the interstellar medium since H2 ` ions must be formed from the H molecules present in the interstellar medium either by light absorption beyond 805 ” or by cosmic rays and since each H2 ` ion will upon collision with a neutral H2 H3 ` according to reaction (I).However the possibility of molecule immediately form detecting H3 ` in interstellar space depends on the discovery of a spectrum of this molecule in the laboratory.œ Gerhard Herzberg 196722 H3 ` in emission. Since H3 ` does not have well-bound electronic Herzberg thus attempted together with J. W. C. Johns to observe the infrared l fundamental band of 2 excited states,7 no ultraviolet or visible spectrum is expected. Its symmetric equilateral triangle structure also forbids a conventional rotational spectrum. Therefore the infrared active degenerate l band is the most straightforward way to detect interstellar H3 `.There were other proposals based on radioastronomy which is by far the most sensi- 2 tive method of detecting interstellar molecules. Salpeter and Malone23 pointed out the possibility of detecting H3 ` using its radio recombination lines which are slightly shifted from the H` recombination lines due to the diÜerence in reduced mass. The recombination lines of He` and C` were well known. An emission line feature was noted between the 85aH` and 85aHe` in NGC202424 but its frequency was not quite right.25 The detection of H3 ` using this technique is probably very difficult because of the low abundance of H3 ` in H II regions where recombination lines are strong. 2D`. The deuteration Another proposal26 was to detect the deuterated species H shift of the center of gravity from the center of charge produces an eÜective dipole moment of 0.6 D and makes the rotational spectrum active in the radio and far-infrared region.The abundance of H2D` is much higher than expected from the natural abun-269 B. J. McCall et al. The most straightforward way of searching for H dance of deuterium because of the efficient isotope fractionation –rst explained by Watson.13 27 A detection of H2D` emission at 372 GHz was reported28 in NGC2264 but was later negated.29 A more recent detection of an H2D` signal by Boreiko and Betz30 in absorption at 1370 GHz in IRc2 has a better signal-to-noise ratio though its authenticity has yet to be con–rmed. The search for interstellar H3 ` using its centrifugal distortion spectrum31 was noted32 and advocated by Draine and Woods33 for studies of high temperature objects such as the X-ray heated clouds NGC6240.3 ` became possible in 1980 when its infrared l band spectrum was discovered in the laboratory.34 An immediate attempt 3 ` in the Becklin»Neugebauer (BN) source in Orion using the FTIR spectro- to detect H 2 meter at the 4 m Mayall Telescope of the Kitt Peak National Observatory (KPNO) was unsuccessful.35 A search by two of the authors (T.R.G. and T.O.) was continued using a Fabry»Perot interferometer and a generation of cooled grating spectrometers (CGS) at the UK Infrared Telescope (UKIRT) on Mauna Kea during which negative results for several sources were published.36 The search was also attempted by many other groups and some of them published their inconclusive results.37h40 From 1988 to 1994 our observational work was diverted to studying H3 ` in plan- 2 etary ionospheres following the discovery of strong H3 ` emission in the auroral regions of Jupiter Saturn and Uranus,41 as well as the Comet SL-9 impact on Jupiter.During this time the resolution sensitivity and reliability of observational infrared spectrometers improved dramatically»a major factor in this development was the use of infrared detector arrays. In 1994 an infrared absorption line of H in NGC2024 was detected at 2 the NASA Infrared Telescope Facility (IRTF).42 This detection suggested that the sensitivity of IRTFœs CSHELL spectrometer had reached the point necessary for H3 ` detection since it was known35 that the ratio of the intensities of the H3 ` dipole transition and the H quadrupole transition (B109) just about cancelled the abundance ratio of H3 ` to H2 (B10~9).Our applications for observing time using CSHELL on IRTF were rejected for three consecutive terms and interstellar H3 ` was instead detected by the CGS4 spectrometer at UKIRT in 1996. Since then our observations have progressed yielding positive detections in dense clouds diÜuse clouds and in the region of the galactic center. spectrum 2 The H3 ë Since the details of the l fundamental band vibration»rotation spectrum were given in a recent Faraday Discussion,43 here we simply note two characteristics of the H3 ` 2 spectrum the vibrational frequency and the rotational level structure.Both of these characteristics have important consequences for the observation of interstellar H3 `. 2.1 Vibrational frequency When a proton is added to an H molecule the extra charge pushes the two protons away and the equilibrium interproton distance increases from 0.74” ” to 0.87 . The 2 vibrational frequency is reduced from 4161.2 cm~1 to l1\3178.3 cm~1 44 and l2\ 2521.3 cm l ~1.45 The infrared active band is located in a region free from spectral lines of ordinary molecules made from atoms with high cosmic abundance. The hydrogen 2 stretching vibrations of CwH NwH and OwH bonds are all much higher in frequency and even the high J P-branch lines of light molecules such as CH4 NH3 and H2O do not reach the 4 lm region.The hydrogen bending vibrations are all too low in frequency and their R-branches do not reach this region. The stretching vibrations of heavier elements such as CxO CyN and CyC are the closest to this region but their rotational structures do not extend much in frequency because of their larger moments of inertia. Fig. 1 which was adapted from Genzelœs review,46,47 shows the unique position 270 Fig. 1 Position of the that l band of H3 ` compared with other common molecular vibrations. Note H3 ` is relatively free of interference from spectral lines of ordinary molecules made from 2 atoms with high cosmic abundance. This –gure is adapted from Genzelœs review.46,47 of the H3 ` l2 band. This freedom from the spectra of other molecules is the reason for the extremely pure H3 ` emission spectrum of Jupiter reported by Maillard et al.48 More makes our ground-based observation relatively unhindered by interference with molimportantly for observations of interstellar H3 ` this favored location of the band origin ecules in the terrestrial atmosphere (L window).The only spectral lines that interfere with our observations are deuterium stretching vibrations notably of HDO and overtone and combination bands notably of the which are of course orders of magnitude weaker than the fundamental bands. Had the band appeared in the 3 lm region it would have been next to impossible to detect interstellar H3 ` from ground-based observatories. 2 2.2 Rotational level structure 2 Because of its small mass H3 ` has large rotational constants45 B\43.56 cm~1 and C\20.61 cm~1 and only the lowest few levels are signi–cantly populated in molecular clouds with temperatures of 10»100 K.The structure of the lowest rotational levels is shown in Fig. 2 where the energy scale is expressed in temperature (Kelvin). J is the rotational angular momentum quantum number and K is its projection onto the C3 symmetry axis. A special characteristic of this rotational structure is that the lowest level with J\K\0 (shown in Fig. 2 with a broken line) is not allowed by the Pauli exclusion principle. According to Diracœs statement of the Pauli principle,49,50 the total wavefunction must change sign when two protons are interchanged but must remain invariant when the three protons are cyclicly permuted.The wavefunction of the lowest rotational level is simply a constant and these conditions cannot be simultaneously satis–ed whether this rotational wavefunction is combined with the ortho nuclear spin function (in which all proton spins are parallel I\3 and the –rst condition is not satis–ed) or with the para nuclear spin function (in which one proton spin is antiparallel 2 I\1 and the second condition is not satis–ed). This leaves the J\1 K\1 level of para-H3 ` as the lowest ground rotational level. This and the next lowest level with J\1 K\0 of ortho-H3 ` which is higher than the ground level by 32.9 K are the only levels that are signi–cantly populated for temperatures of 5»50 K.These two levels contain nearly equal populations of molecules H3 ` in dense and diÜuse clouds l2]l4 band of CH4 , l 271 B. J. McCall et al. 2) Fig. 2 Structure of the lowest rotational levels of H3 `. Broken lines represent forbidden levels the bold line indicates a level with the ortho (I\3 spin modi–cation the thin lines indicate levels with the para (I\1 spin modi–cations. The (]) and ([) signs indicate the parity of the levels. 2) The transitions studied in interstellar space arise from the J\1 levels. since the higher spin statistical weight of ortho-H3 ` (gI\2I]1\4) than that of para- We thus have six spectral lines of comparable intensities at 30 K as shown in Fig. 3 two H3 ` (gI\2) is approximately compensated for by the Boltzmann factor exp ([32.9/T ).3 ` para-H3 ` ortho-H from [R(1,1)` R(1,1)~ Q(1,1) [R(1,0) and Q(1,0)] and four from and P(1,1)]. The existence of six lines makes the observation —exible»we may choose lines that are freest from the telluric interference depending on the weather and the Doppler shift of the night. Two of the spectral lines R(1,1)` of para-H3 ` and R(1,0) of l band of Fig. 3 Six available spectral lines of the H3 `. Note that R(1,1)` and R(1,0) form a 2 doublet with spacing of 0.321 cm~1 which is particularly useful for astronomical observations. The broken line marks the hypothetical position of the transition arising from the forbidden level J\K\0. This line would have an intensity four times that of the other strongest lines if it were allowed.These intensities are calculated for an assumed temperature of 30 K. H 272 3 ` in dense and diÜuse clouds ortho-H3 ` are separated by only 0.321 cm~1 and are particularly useful for the measurement of temperature and for con–rmation of detections. Had the lowest J\K\0 level been allowed the spectrum of H3 ` would be like an atomic spectrum since most of the molecules would be in the lowest level and the R(0,0) line would be the only strong line at the position shown with a broken arrow in Fig. 3. 3 Observed results Observations of interstellar H3 ` have so far been conducted using three infrared spectrometers the CGS4 at UKIRT with spectral resolution RB20 000 the Phoenix spectrometer at KPNO with RB60 000 and the CSHELL at NASA IRTF with RB20 000.All of them have produced positive results. Interstellar H3 ` has been found in gravitationally bound dense clouds with high density ([&H]B103»105 cm~3) as well as in unbound diÜuse clouds with low density (10»103 cm~3). 3.1 H3 ë in dense clouds The –rst spectra of interstellar H3 ` were detected towards the young stellar objects (YSOs) GL2136 and W33A which are deeply embedded in dense molecular clouds.51 These spectra were obtained with CGS4 at UKIRT on the nights of April 25 June 10 and July 15 1996. These YSOs were chosen because of their infrared brightness and because of their large column densities of foreground gas. In addition it was thought proton hop reaction (II). Strong absorptions of solid CO frozen on dust grains have advantageous52 to use carbon depleted clouds where H3 ` is destroyed less by the been reported53,54 although the depletions might not be large.55 observed signal to noise ratios of the absorption lines were by no means great but the The R(1,0)»R(1,1)` doublet of H3 ` mentioned earlier was used for the detection.The Doppler shift of the doublet lines due to the earthœs orbital motion from April 25 to July 15 convinced us that the signals were genuine (see Fig. 4). 3 ` in dense clouds towards three Subsequent observations revealed interstellar H other YSOs MonR2 IRS 3 and GL961E (February 11»14 1997 at UKIRT) and Fig. 4 Spectra of the R(1,0)»R(1,1)` doublet of H UKIRT along the line of sight to GL2136. The upper trace was obtained on 29 April 1996 while 3 ` obtained with the CGS4 spectrometer at the lower trace was obtained on 15 July 1996.The observed Doppler shift of the doublet (marked with arrows) between the two dates matches that expected from the Earthœs orbital motion providing convincing evidence that the doublet is genuine. 273 B. J. McCall et al. GL2591 (July 11»12 1997 at UKIRT). The observed equivalent widths W yield the l H3 ` (1) column densities using W formula l4 the P[ standard *I(l)/I(l)] dl\(8p3l/3hc)No l o2 3 ` where o l o2 is the square of the transition dipole moment. The spectral lines of ortho- H3 ` and para-H3 ` give their column densities No and Np separately and their sum N(H column density gives the total H 3 `). The ratio of No and Np gives the temperature of the clouds using the standard formula (2) e(~*E@kT)\2 e(~32.9@T) No\ go Np gp These results are summarized in Table 1.We have also studied the infrared sources GL490 GL989 LkHa 101 MonR2 IRS 2 M17 IRS 1 S140 IRS 1 W3 IRS 5 Elias 29 NGC2024 IRS 2 and BN. So far our data reduction has not provided evidence of column densities at the level of ca. 2»3]1014 cm~2 but careful reprocessing of these was particularly surprising in view of the large column density of H reported in the spectra continues. The lack of strong H3 ` absorption towards NGC2024 IRS 2 and BN 2 former42 and the observed richness of molecules in the latter. We believe that these non-detections are not due to the absence of H3 ` in the clouds but are simply due to the short column length of the clouds in front of the source (see the discussion in Section 4.1).More details of our study of dense clouds will be published elsewhere.56 3.2 H3 ë in diÜuse clouds During our survey of H3 ` in dense clouds we observed strong and broad H3 ` absorption signals in the direction of the infrared sources GC IRS 3 and GCS3-2 (July 11»12 UKIRT) in the region near the galactic center. These sources are thought to be 8 kpc away and their lines of sight may cross several clouds both dense and diÜuse. Indeed McFadzean et al.57 reported observational evidence for two components in the extinction the 3.0 lm ice absorption (a signature of dense clouds) and the 3.4 lm hydrocarbon absorption (a signature of diÜuse clouds).More details of our galactic center observations will be published separately.58 The galactic center results led us to try Cygnus OB2 No. 12 a visible star with high extinction discovered in 1954.59 It is generally believed that this star is obscured largely by diÜuse low density clouds containing little molecular material.60 We clearly observed Table 1 Positions and derived column densities and temperatures for H3 ` sources position infrared source T /K d (1950) a (1950) N(H3 `)/(1014 cm~2)a 35^4 30^6 24^4 24^5 dense clouds GL2136 W33A MonR2 IRS 3 GL961E GL2591 18 19 36.6 18 11 44.2 06 05 21.8 06 31 59.1 20 27 35.8 3.6^0.6 5.5^1.9 2.1^0.7 1.7^0.7 2.0^1.0b [13 31 40 [17 52 56 [06 22 26 ]04 15 10 ]40 01 14 20 30 53.4 diÜuse clouds Cyg OB2 No.12 20^4 3.8^0.5 ]41 03 52 a Statistical uncertainties (3p) are quoted in parentheses but systematic errors are difficult to estimate and might be larger. b Estimated systematic uncertainty is given for GL2591 as this spectrum is not yet fully reduced. H 274 3 ` in dense and diÜuse clouds 3 ` was obtained with CGS4 at UKIRT on 11 July 1997. Fig. 5 Spectra of the line of sight towards the visible star Cygnus OB2 No. 12. The left trace showing the R(1,0)»R(1,1)` doublet of H The right trace showing the R(1,1)~ line of H3 ` was obtained with the new Phoenix spectrometer at KPNO on 17 September 1997. The high frequency interference in the CGS4 spectrum near 3.6675 lm is due to the removal of a telluric CH absorption line.4 10 ^111 and 422 ^423) made the obserthe H3 ` R(1,0)»R,(1,1)` doublet (July 11»12 UKIRT) and the R(1,1)~ singlet (September 15»17 KPNO).61 Because of the high humidity in Arizona in September strong and wide telluric HDO lines (l band 1 vation of the doublet impossible but the singlet (which has only CH lines nearby) was 2 clearly observed. The observed spectrum is shown in Fig. 5. Using the observed equiva- 4 lent widths of the lines and eqn. (1) we obtain the remarkable result that the column density of H3 ` in the direction of Cygnus OB2 No. 12 is (3.8^0.5)]1014 cm~2 comparable to that of the dense clouds listed in Table 1. van Dishoeck and Black20,62 reported their extensive chemical model calculation in diÜuse clouds and predicted high column densities of H3 ` but their calculation was based on an extremely small electron recombination rate constant which has since been demonstrated to be too low by more than three orders of magnitude.63 Calculations given in the next section show that the large column density of H3 ` in the diÜuse clouds towards Cygnus OB2 No.12 is due not to a high number density of H3 ` but simply to a long column length. Our calculation is much cruder than that of van Dishoeck and Black but is essentially the same as far as the H3 ` chemistry is concerned except that a revised recombination rate is used. 4 H chemistry 3 ë A very attractive aspect of H3 ` as a molecular astronomical probe is its simple chemistry.The simplicity of the chemistry allows us to make relatively simple and reliable arguments about the H3 ` number densities and other astrophysical quantities. In the following we give a crude order of magnitude discussion of its chemistry ; more detailed chemical model calculations such as those given by Lee et al.21 and by van Dishoeck and Black20 are of course desirable for more accurate discussions. 4.1 H3 ë chemistry in dense clouds In cold dense clouds which are protected from star radiation H3 ` is produced almost 2 exclusively from the cosmic ray (CR) ionization of H to H2 ` CR (IV) H »»»’ H2 `]e~ 2 275 B. J. McCall et al. cm~3 yield an followed by the ion»neutral reaction (I). Reaction (I) is many orders of magnitude more rapid than reaction (IV) and the production rate is governed by the rate of reaction (IV) [H i.e.f[H2]. The cosmic ray ionization rate fB10~17 s~1 and H2 number density 2]B104 H3 ` production rate of ca. 10~13 cm~3 s~1. H3 ` is destroyed predominantly by the proton hop reaction (II). Equating the production and destruction rates we have the steady-state equation (3) f[H2]\; kx[H3 `][X] x where k is the rate constant for reaction (II). Since CO is the most abundant molecule x in dense clouds we neglect the terms of the other atoms and molecules in eqn. (3) and obtain the H3 ` number density [H (4) 3 `]\k f [CO] [H2] CO Since the ratio [H ]/[CO]B104 is approximately constant over a wide variety of molecular parameters,21 this shows that [H3 `] is constant.Using the Langevin rate64 2 3 `]B10~4 cm~3. The observed cm3 s~1 we obtain [H COB10~9 k 3 ` column H density of 3]10~14 cm~2 (see Table 1) gives a typical eÜective column length of B1 pc. The most serious omission in this discussion is the neglect of X terms other than CO from eqn. (3). In the model calculations of Lee et al.,21 the abundance of O is predicted to be comparable to that of CO. Inclusion of the O term will reduce the [H3 `] by ca. 30% since the rate constant k is about 1/2.5 of kCO .64 The neglect of the electron O term X\e~ in eqn. (3) also has to be addressed since the recombination rate constant k is larger than kCO by more than two orders of magnitude (see Section 4.2). However e the model calculations of Lee et al.21 show that this correction is signi–cant only in clouds with high metallicity where the electron concentration is increased by the ionization of alkali and alkaline-earth metals with low work functions.The lack of accurate measurements of k (and for that matter even of kCO) is also a source of error and more laboratory studies are awaited. O However all these corrections will be small compared to the large uncertainty in f. We hope that our H3 ` measurements will help further constrain this important parameter. 4.2 H3 ë chemistry in diÜuse clouds In diÜuse clouds where the number density is low (10»103 cm~3) and visible light passes through cosmic ray ionization followed by reaction (I) is again the primary mechanism for H3 ` production.Photoionization of H is not eÜective because the cloud contains 2 abundant atomic H atoms whose ionization potential (13.6 eV) is lower than that of H2 (15.4 eV). recombination because of the high number density of electrons created by photoioniza- The main destruction mechanism of H3 ` in diÜuse clouds is expected to be electron tion of carbon (the carbon atom has the lowest ionization potential 11.3 eV of any abundant species). We assume for simplicity that all carbon atoms which are not depleted onto dust grains are ionized and that all electrons come from the ionization of carbon atoms i.e. [e~]\[C`]\[&C] where [&C] denotes the total number density of carbon atoms. The solution of the steady state equation is then [H (5) 3 `]\k f [H [e~ 2 ] ] number e indicating that the H3 ` number density is constant also in diÜuse clouds.Using cm3 s~1 63 and eB10~7 3 ` k ]/[e~]B[H ]/[&C]B104 we obtain an H [H2 2 H 276 3 ` in dense and diÜuse clouds density of [H3 `]B10~6 cm~3 smaller than that of dense clouds by two orders of magnitude. Thus the same H3 ` column density as in dense clouds (3]1014 cm~2) implies an eÜective path length (L ) that is longer by two orders of magnitude L D100 pc. This path length is very likely composed of several diÜuse clouds rather than a single cloud. There is a major uncertainty in the above estimates apart from that of f. Unlike k with Langevin rates which are independent of temperature,65,66 k varies sige\ 4.6]10~6/T 0.65 cm3 s~1 as determined k e other x ni–cantly at low temperature.If we use from the storage ring experiment of Sundstroé m et al.,67 and assume T B30 K k is e closer to 10~6 cm3 s~1 and L B1 kpc. In addition we have not considered direct photodissociation of H3 `. This is thought to be slow68 but more theoretical and experimental studies are certainly needed. 4.3 Intermediate case The above two analyses for the extreme cases can be generalized to the intermediate case where the destruction rates of H3 ` by CO and by electrons are comparable. We assume that all carbon atoms in the gas phase are either in the form of C` or CO that is [&C]\[C`]][CO] where [&C] denotes the total number density of carbon atoms in any gaseous form. Other carbon species (atomic C CO etc.) can be included in [CO] since they all have Langevin rates for the proton hop reaction (II).We 2 CH4 have (6) [H k 3 `]\f 2 f [ [ & & H] C] e(1[a)]kCO aD C 1 where f is the fraction of hydrogen atoms in molecular form f42[H2]/[&H] and a is the fraction of carbon atoms in molecular form a4[CO]/[&C]. For derivation of this formula and further discussions of the total number density [&H] and path length L of the cloud see McCall et al.61 5 Future prospects Our observations have established that interstellar H3 ` exists with sufficient abundance to be observable from ground-based observatories both in dense and diÜuse clouds. In fact we –nd it easier to observe H3 ` absorption lines than H infrared absorption lines.3 ` is not only a powerful probe for the study of plasma activities of astrono- Perhaps H 2 mical objects but also a most convenient probe for the detection of hydrogenic molecular species. In the spirit of this conference we speculate in this section on some possible developments in the immediate future. 5.1 Future observations From ground-based observatories H3 ` will be found in many other sources. For dense clouds the observations will give information of the depth of the embedded YSO and for diÜuse clouds they will give the dimension of the clouds. For a source like the Quintuplet near the galactic center where many infrared sources are positioned within a narrow angle of sight some type of ìmappingœ such as radioastronomers do might be possible.This will be most efficiently done when the Phoenix spectrometer is moved to the Cerro Tololo Interamerican Observatory next year (1999). The expected advent of larger telescopes with high-resolution infrared spectrometers such as Gemini and Subaru and the installation of a high-resolution spectrometer at Keck will allow us to observe much fainter infrared sources. H3 ` will be observed in a great many more 277 B. J. McCall et al. objects with higher spectral resolution. We may not have to wait many years before H3 ` is observed in extragalactic objects. 3 ` emission lines observed in planetary ionospheres.7 5.2 H3 ë emission Observing the infrared spectrum of H3 ` in emission is an interesting possibility.One remembers the strong and pure H The strongest H quadrupole emission line S (1) (with a spontaneous emission lifetime 2 1 of 7]106 s 69) is observed with large signal to noise ratios in planetary nebulae,70 extragalactic superluminous objects,71 and many other objects even using low resolution spectrometers. In order to evaluate the prospects for detecting H3 ` emission we make a rough estimate of the ratio of intensities for H3 ` emission IH3` H and 2 emission IH2 (7) \[H3 `] kH3` I I H H 3 2 ` H2 kH3` and 3 ` should be roughly [H3 `]/[H2]B10~8 which one 2 [H kH2 2] k where are rate constants for collisional pumping from v\0 to v\1. We estimate that the abundance of H would think would make the ratio small.However we must consider the diÜerences in the vibrational pumping mechanisms. The collisional excitation of H by H2 (V) H2]H2 ]H2*]H2 is performed by a weak physical interaction in which the translational energy of H2 must be converted to vibrational energy (V»T transfer) during the short time of the encounter. Resonant V»V transfer cannot contribute since the number of H remains 2* the same in the ìreactionœ H2*]H2 ]H2]H2*. action On the other hand the excitation of H3 ` is performed by a strong chemical inter- (VI) H3 `]H2 ]H3 `*]H2 where asterisks signify vibrational excitation. In this case the molecules attract each other by the Langevin force form an activated complex (H5 `)* and then dissociate into H 2 3 `* H and . This reaction is known to have a Langevin rate from a deuterium experiment64 and a recent experiment of spin modi–cation.72 The branching ratio of reaction (VI) to form H3 ` or H3 `* is not known but we assume that it is not much diÜerent from 1 1.Then using the approximate equality between the Langevin rate and the rate of rotational energy transfer (R»R),73 and the rule of thumb74 kVhT/kRhRB10~5 k we obtain H3`/kH2B105. If we use experimental and theoretical v@\1]0 deexcitation rates75 and the principle of detailed balancing we –nd that kH3`/kH2B 105»106 for T \2000»1000 K. IH3`/IH2B10~3»10~2. This is a minimum value and 2 Thus we obtain an estimate of will be larger for a molecular cloud with a density higher than the critical density.76 In such a high-density environment the H3 ` intensity will be increased due to the faster collisional pumping (as H3 ` has a spontaneous emission time of only B10 ms77) but the H intensity will be limited by the slower rate of spontaneous emission.Even if IH3`/IH2B10~3 H the detection of 3 ` emission is a realistic prospect in view of the extremely high observed signal to noise ratios of H emission (Z1000 at low resolution). 2 5.3 H3 ë as an interstellar agent Interstellar H3 ` not only plays the central role of the protonator to initiate a network of chain reactions but also performs other essential functions of interstellar chemistry. For H 278 3 ` in dense and diÜuse clouds ortho-H to para-H through the proton hop 2 2 example it will mediate conversion of reaction (VII) H3 `]H3 2 ]H2]H3 2H` and proton exchange reaction (VIII) H3 `]H3 2 ]H2H3 `]HH3 This scrambling of protons will thermalize spin modi–cations.The actual efficiency of this mechanism should be calculated using the nuclear modi–cation branching ratios theoretically predicted by Quack78 and recently experimentally demonstrated.72 These processes must be much more efficient than the mechanism proposed earlier79 (IX) H`]o-H2 ]H`]p-H2 both because of the higher abundance of H3 ` and the higher rate constant of reactions (VII) and (VIII) than (IX). Klemperer and Miller80 have recently proposed that the strong CO Cameron band emission (a ” 3%]X1&`) from 1850»2600 observed in the Red Rectangle nebula81,82 might be due to a chemical pumping of CO by H3 ` through the reactions (X) H3 `]CO]HCO`]H2 and where the asterisk signi–es CO in the a 3% excited state.In diÜuse clouds where [e~]A [CO] the second reaction is much faster than the –rst and the rate for CO* excitation is kCO[H3 `][CO]B10~26 cm~3 s~1 (here again we neglect the branching ratio between CO and CO*). Glinski et al.81 proposes the direct electron pumping HCO`]e~]CO*]H (XI) CO]e~*]CO*]e~ (XII) 3 ` is the agent for the emission especially for the to be the main mechanism where e~* signi–es electrons at high energy B8»12 eV. The rate of this process is k[CO][e~*] where the rate constant of excitation k is B10~8 cm3 s~1. Thus the relative efficiency of the H3 ` pumping of Klemperer and Miller and the electron pumping of Glinski et al.depends on the relative magnitudes of [H3 `] and 10[e~*]. It is quite probable that H rotationally cold emission core. We have pro–ted from discussions with W. Klemperer and K. Takayanagi. B.J.M. is supported by the Fannie and John Hertz Foundation. The University of Chicago portion of this work has been supported by NSF grant PHYS-9722691 and NASA grant NAG5-4070. References 1 J. J. Thompson Philos. Mag. 1911 21 225. 2 A. J. Dempster Philos. Mag. 1916 31 438. 3 T. R. Hogness and E. G. Lunn Phys. Rev. 1925 26 44. 4 H. D. Smyth Rev. Mod. Phys. 1931 3 347. 5 H. Eyring J. O. Hirschfelder and H. S. Taylor J. Chem. Phys. 1936 4 479. 6 J. O. Hirschfelder J. Chem. Phys. 1938 6 795. 7 T. Oka in Molecular Ions Spectroscopy Structure and Chemistry ed.T. A. Miller and V. E. Bondybey North Holland New York 1983 p. 73. 8 D. W. Martin E. W. McDaniel and M. L. Meeks Astrophys. J. 1961 134 1012. 9 T. P. Stecher and D. A. Williams Astrophys. L ett. 1970 7 59. 10 P. M. Solomon and M. W. Werner Astrophys. J. 1971 165 41. 279 B. J. McCall et al. 11 T. de Jong Astron. Astrophys. 1972 20 263. We assume that his Case (A) is closer to reality than Case (B). 12 A. E. Glassgold and W. D. Langer Astrophys. J. 1973 179 L147. 13 W. D. Watson Astrophys. J. 1973 182 L73. 14 W. D. Watson Astrophys. J. 1973 183 L17. 15 E. Herbst and W. Klemperer Astrophys. J. 1973 185 505. 16 D. M. Rank C. H. Townes and W. J. Welch Science 1971 174 1083. 17 D. Buhl and L.E. Snyder Nature (L ondon) 1970 228 267. 18 W. Klemperer Nature (L ondon) 1970 227 1230. 19 T. de Jong A. Dalgarno and W. Boland Astron. Astrophys. 1980 91 68. 20 E. F. van Dishoeck and J. H. Black Astrophys. J. Suppl. 1986 62 109. 21 H.-H. Lee R. P. A. Bettens and E. Herbst Astron. Astrophys. Suppl. 1996 119 111. 22 G. Herzberg T rans. R. Soc. Can. 1967 V 3. 23 E. E. Salpeter and R. C. Malone Astrophys. J. 1971 167 27. 24 J. M. MacLeod L. H. Doherty and L. A. Higgs Astron. Astrophys. 1975 42 195. 25 T. Oka personal communication. 26 A. Dalgarno E. Herbst S. Novick and W. Klemperer Astrophys. J. 1973 183 L131. 27 W. D. Watson Rev. Mod. Phys. 1976 48 513. 28 T. G. Phillips G. A. Blake J. Keene R. C. Woods and E. Churchwell Astrophys. J. 1985 294 L45.29 E. F. van Dishoeck T. G. Phillips J. Keene and G. A. Blake Astron. Astrophys. 1992 261 L13. 30 R. T. Boreiko and A. L. Betz Astrophys. J. 1993 405 L39. 31 J. K. G. Watson J. Mol. Spectrosc. 1971 40 536. 32 F.-S. Pan and T. Oka Astrophys. J. 1986 305 518. 33 B. T. Draine and D. T. Woods Astrophys. J. 1990 363 464. 34 T. Oka Phys. Rev. L ett. 1980 45 531. 35 T. Oka Philos. T rans. R. Soc. L ondon A 1981 303 543. 36 T. R. Geballe and T. Oka Astrophys. J. 1989 342 855. 37 P. G. Burton E. von Nagy-Felsobuki and G. Doherty Chem. Phys. L ett. 1984 104 323. 38 J. H. Black E. F. van Dishoeck S. P. Willner and R. C. Woods Astrophys. J. 1990 358 459. 39 J. Tennyson S. Miller and H. Schild J. Chem. Soc. Faraday T rans. 1993 89 2155. 40 J. P. Maillard Spectrochim.Acta Part A 1995 51 1105. 41 See for review T. Oka Rev. Mod. Phys. 1992 64 1141. 42 J. H. Lacy R. Knacke T. R. Geballe and A. T. Tokunaga Astrophys. J. 1994 428 L69. 43 T. Oka and M.-F. Jagod J. Chem. Soc. Faraday T rans. 1993 89 2147. 44 W. Ketterle M.-P. Messmer and H. Walther Europhys. L ett. 1989 8 333. 45 J. K. G. Watson S. C. Foster A. R. W. McKellar P. Bernath T. Amano F.-S. Pan M. W. Crofton A. J. Altman and T. Oka Can. J. Phys. 1984 62 1875. 46 R. Genzel in T he Galactic Interstellar Medium ed. D. Pfenniger and P. Bartholdi Springer-Verlag Berlin 1992 p. 275. 47 L. J. Allamandola in Galactic and Extragalactic Infrared Spectroscopy ed. M. F. Kessler and J. P. Phillips Reidel Dordrecht 1984 p. 5. 48 J.-P. Maillard P. Drossart J. K. G.Watson S. J. Kim and J. Caldwell Astrophys. J. 1990 363 L37. 49 P. A. M. Dirac Proc. R. Soc. L ondon A 1926 112 661. 50 For a pedagogical explanation of the relation between the Pauli principle and relativity see S. Tomonaga T he Story of Spin Univ Chicago Press Chicago IL 1998. 51 T. R. Geballe and T. Oka Nature (L ondon) 1996 384 334. 52 S. Lepp A. Dalgarno and A. Sternberg Astrophys. J. 1987 321 383. 53 T. R. Geballe F. Baas J. M. Greenberg and W. Schutte Astron. Astrophys. 1985 146 L6. 54 W. A. Schutte P. A. Gerakines T. R. Geballe E. F. van Dishoeck and J. M. Greenberg Astron. Astrophys. 1996 309 633. 55 G. F. Mitchell J.-P. Maillard M. Allen R. Beer and K. Belcourt Astrophys. J. 1990 363 554. 56 B. J. McCall T. R. Geballe K. H. Hinkle and T. Oka manuscript in preparation.57 A. D. McFadzean D. C. B. Whittet A. J. Longmore M. F. Bode and A. J. Adamson Mon. Not. R. Astron. Soc. 1989 241 873. 58 T. R. Geballe B. J. McCall K. H. Hinkle and T. Oka manuscript in preparation. 59 W. W. Morgan H. L. Johnson and N. G. Roman Publ. Astron. Soc. Pac. 1954 66 85. 60 D. C. B. Whittet A. C. A. Boogert P. A. Gerakines W. Schutte A. G. G. M. Tielens Th. de Graauw T. Prusti E. F. van Dishoeck P. R. Wesselius and C. M. Wright Astrophys. J. 1997 490 729. 61 B. J. McCall T. R. Geballe K. H. Hinkle and T. Oka Science 1998 279 1910. 62 J. H. Black in IAU Symp. 120 Astrochemistry ed. M. S. Vardya and S. P. Tarafdar Dordrecht Reidel 1987 p. 217. 63 T. Amano Astrophys. J. 1988 329 L121. 64 V. G. Anicich and W. T. Huntress Jr.Astrophys. J. Suppl. 1986 62 553. 65 J. C. Maxwell Philos. T rans. R. Soc. 1879 170 231. H 280 3 ` in dense and diÜuse clouds 66 M. P. Langevin Ann. Chim. Phys. 1905 5 245. 67 G. Sundstroé m J. R. Mowat H. Danared S. Datz L. Brostroé m A. Filevich A. Kaé llberg S. Mannervik K. G. Rensfelt P. Sigray M. af Ugglas and M. Larsson Science 1994 263 785. 68 E. F. van Dishoeck in IAU Symp. 120 Astrochemistry ed. M. S. Vardya and S. P. Trafdar Dordrecht Reidel 1986 p. 51. 69 J. H. Black and A. Dalgarno Astrophys. J. 1976 203 132. 70 See for example H. A. Thronson Jr. Astrophys. J. 1981 248 984. 71 See for example T. R. Geballe Can. J. Phys. 1994 72 782. 72 D. Uy M. Cordonnier and T. Oka Phys. Rev. L ett. 1997 78 3844. 73 T. Oka Adv. Atom. Mol. Phys. 1973 9 127. 74 W. H. Flygare Acc. Chem. Res. 1968 1 121. 75 M. Cacciatore M. Capitelli and G. D. Billing Chem. Phys. L ett. 1989 157 305. 76 J. M. Shull and D. J. Hollenbach Astrophys. J. 1978 220 525. 77 G. D. Carney and R. N. Porter J. Chem. Phys. 1976 65 3547. 78 M. Quack Mol. Phys. 1977 34 477. 79 A. Dalgarno J. H. Black and J. C. Weisheit Astrophys. L ett. 1973 14 77. 80 W. Klemperer and A. Miller personal communication. 81 R. J. Glinski J. A. Nuth M. D. Reese and M. L. Sitko Astrophys. J. 1996 467 L109. 82 R. J. Glinski J. T. Lauroesch M. D. Reese and M. L. Sitko Astrophys. J. 1997 490 826. Paper 8/00655E; Received 23rd January 1998

ISSN:1359-6640    

DOI:10.1039/a800655e

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Faraday Discuss. 1998 109 281»301 Molecular evolution in planet-forming circumstellar disks 1 Introduction Stars are formed by the gravitational collapse of molecular cloud cores. The circumstellar disk is a natural byproduct of collapse of a rotating cloud core. Radio IR and optical observations have recently revealed the existence of circumstellar disks around young stellar objects called T Tauri stars (ref. 1 and 2 and references therein). A variety of evidence suggests that at least 50% of low-mass young stars are surrounded by disks. These disks are considered to be the formation sites of the planetary systems and are thus called protoplanetary disks. Theoretical study of the evolution of molecular abundance in protoplanetary disks is important for various reasons.First it will show directly the composition of the bodies § Present address Department of Physics The Ohio State University Columbus OH 43210 USA. Y. Aikawa,a§ T. Umebayashi,b T. Nakanoc and S. Miyamaa a National Astronomical Observatory of Japan Mitaka T okyo 181 Japan b Computing Service Center Y amagata University Y amagata 900 Japan c Nobeyama Radio Observatory National Astronomical Observatory Nobeyama Minamisaku Nagano 384-13 Japan We have investigated the evolution of molecular abundance in circumstellar disks around young low-mass stars which are considered to be the formation sites of planetary systems. Adopting the standard accretion disk model we investigated molecular evolution mainly in the accretion phase. In the region of surface density less than 102 g cm~2 (distance from the star Z10 AU) cosmic rays are barely attenuated even in the midplane of the disk and produce chemically active ions such as He` and H considerable amount of CO and N 3 `.We found that a 2 the initial dominant components of the disk is transformed into CO2 CH4 NH3 and HCN through reactions with these ions. Where the temperature is low enough for these products to freeze onto grains they are selectively ìlocked upœ and accumulate in the ice mantle. As the matter accretes towards inner warmer regions the ice mantle evaporates. The desorbed molecules such as CH4 are transformed into larger and less volatile molecules by reactions in the gas phase. The molecular abundance both in the gas phase and in the ice mantle depends crucially on the temperature and thus varies signi–cantly with the distance from the central star.If the ionization rate and the grain size in the disk are the same as those in molecular clouds the timescale of the molecular evolution in which CO and N are transformed into other molecules is ca. 106 years slightly less than the life time of the disk. The timescale of molecular evolu- 2 tion is less for higher ionization rates and greater for lower ionization rates or larger grain size. We have compared our results with the molecular composition in comets the most primitive bodies in our solar system. The molecular abundance derived from our model reproduces the coexistence of oxidized ice and reduced ice as observed in comets.Our model also suggests that comets formed in diÜerent regions of the disk will have diÜerent molecular compositions. 281 282 Molecular evolution in planet-forming circumstellar disks in the planetary systems. Chemical composition determines some basic properties of the planets such as density thermal history and composition of their atmospheres. Knowledge of the chemical composition at each radius of the disk is helpful in order to understand and predict the individual properties of planets formed there. Secondly it is indispensable for the observational study of the protoplanetary disks via molecular lines. Owing to the recent development of radio telescopes observational study of protoplanetary disks is now making great progress. For example emission lines of molecules such as CO CN and HCO` have recently been detected for some T Tauri stars.3h6 Aperture-synthesized images directly revealed the distribution of CO molecules around some T Tauri stars in Taurus molecular clouds with spatial resolution of a few hundreds of AU.7h11 Moreover submillimetre arrays will enable us to observe disks with spatial resolution of a few tens of AU in the near future.It is expected that these observations will tell us precisely about the structure and evolution of the disk i.e. we will determine the initial conditions and the process of planet formation. However observations of molecular lines need help from a theoretical study of molecular evolution. For example the amount of the gaseous component in the disk is estimated by observations of molecules other than the main component H2 H since 2 can only emit photons at the very inner region of the disk (R[0.1 AU) where the temperature is high enough to excite the H molecule.Thus knowledge of the abundance of the molecules relative to hydrogen is essential. 2 Thirdly molecular abundance can be a useful probe in investigating the formation processes of our own solar system. For example we may be able to reduce the processes places and epochs of the formation of primitive bodies such as comets by comparing their molecular composition with theoretical results on the distribution and evolution of molecular abundance in the disk. and Some theoretical studies have been carried out on the molecular abundance in protoplanetary disks.The primary works by Lewis and his colleagues investigated the composition in the disk assuming thermochemical equilibrium. Considering the pressure and temperature distribution in the solar nebula they concluded that most of the carbon is in the form of CO at T Z700 K while it is present as CH at lower tem- 4 perature. Similarly most of the nitrogen is in the form of N at T Z300 K but is present 2 as NH at lower temperature. Hence the disk is CO- and N -rich for radius R[1 AU 3 2 while it is CH RZ1 AU.12,13 4- NH3 -rich for Prinn and his colleagues (ref. 14 and references therein) pointed out however that at low temperatures the reaction rates are so small that thermochemical equilibrium cannot be achieved within the dynamical timescale of the disk.Instead of the chemical equilibrium model they proposed the so-called ì kinetic inhibition (K1) modelœ which assumes that the matter at diÜerent radii is well mixed by turbulent mixing. Their basic idea is that the molecular abundance in the matter —owing outwards is ìquenchedœ when its temperature has decreased to a value below which the timescale for chemical reaction is larger than the dynamical timescale [Fig. 1(a)]. Adopting the dynamical timescale of the disk of ca. 1013 s they concluded that the molecular abundance was quenched at 850»1500 K (R\1 AU) and that CO and N were the dominant components of the solar nebula. 2 In this paper we investigate the molecular abundance in the protoplanetary disks making three improvements. First we take into account ionization by cosmic rays and subsequent ion»molecule reactions which were neglected previously.In the region of radius RZseveral AU the disk is partially ionized since the column density of the disk is less than the attenuation length of cosmic rays ca. 96 g cm~2.15 The reaction rates of ion»molecule reactions are much faster than those of neutral»neutral reactions especially at low temperatures. Thus ion»molecule reactions should be efficient in the outer region of the disk which was previously considered to be chemically inactive. Secondly we adopt the standard accretion disk model [Fig. 1(b)]. In the KI model it was assumed that the matter in the outer region is transported from the region inside the quench 283 Y . Aikawa et al. Fig. 1 Schematic illustrations of the disk models assumed in (a) KI model14 and (b) this work.In the KI model it was assumed that the material in diÜerent radii is well mixed. The molecular abundance is quenched at the ìquench radiusœ as the matter —ow outwards (see text). In contrast we adopt the steady-accretion disk model where the matter —ows inwards. We also take into account ionization by cosmic rays. radius at R\1 AU however it is not clear whether such outward —ow exists in the disk considering the conservation of angular momentum. According to the standard accretion disk model of Lynden-Bell and Pringle,16 most of the material loses its angular momentum due to viscosity and accretes towards the central star in the circumstellar disk. Then for any value of R most of the matter is from the outer region.Observations at UV and optical wavelengths suggest that the protoplanetary disks are viscous accretion disks. IR spectral slopes of the spectral energy distribution (SED) of the disks are similar to those of the accretion disk models described by Lynden-Bell and Pringle16 (Bertout et al.17). Thirdly we investigate the evolution of molecular abundance by solving directly the reaction equations. This enables us to obtain the molecular abundance fully time dependently whereas previous studies obtained only the steady state values. In Section 2 we describe the models of the disk and the reaction network. Numerical results on the evolution and distribution of molecular abundance are reported in Section 3. The dependence of our results on some physical parameters such as cosmic ray —ux and grain size are discussed in Section 4.In Section 5 we compare our results with the composition of comets. 2 Model 2.1 Disk model Here we describe the theoretical model of the structure of protoplanetary disks. The excess of UV radiation and intense Ha emission lines indicate that the disks are accreting towards the central star at ages[several]106 years. Thus we adopt the standard accretion disk model with steady —ow.16,18 Assuming that the accretion rate and the disk structure do not change with 0 ) lR(R time \ M the column B 1[AR * density 1@2D of the disk R is given by (1) C 3n R where l is the viscosity M0 is the mass accretion rate R is the radius of the central star * and R is the distance from the central star ; we adopt the cylindrical coordinates (R U Z).The origin of the viscosity is not well known although there are some candidates 284 Molecular evolution in planet-forming circumstellar disks such as turbulence caused by the shearing instability of a weakly magnetized disk.19,20 Owing to the lack of knowledge of the viscosity in the disk we adopt the a-prescription of viscosity given by Shakura and Sunyaev,21 (2) l\acsH cm2 s~1 The sound velocity c and the scale height of the disk H are given by s (3) cs\Sk kT m cm s~1 (4) H\S2 kT H R3 cm km GM H m the mass of a hydrogen atom G the gravitational constant andM the mass * * where k is the Boltzmann constant T the temperature k the mean molecular weight of the gas H of the central star.We adopt k\2.34 M *\1M_ and R *\3R_ where R_ and M_ M0 are the radius and the mass of the sun. The accretion rate and a are determined from observational data. The UV excesses indicate that the accretion rate is M0 \10~8»10~7 M_ year~1. This corresponds to aB10~2 considering the disk mass (MB10~2 M_) estimated from the SED in IR wavelengths.22 The number density of the hydrogen nuclei in the disk is then given by (5) nH\0.756] 2Hm R H assuming the solar abundance.23 The accretion velocity is given by (6) vR\ M0 2nRR cm s~1 Next we consider the distribution of temperature in the disk. For simplicity we assume that the disk is locally isothermal in the Z-direction because heat is easily transported in the Z-direction by thermal radiation of the dust particles.The temperature distribution is determined by the balance between the photon —ux from the central star gravitational energy released by the viscous accretion and thermal radiation of the disk. Ruden and Pollack24 formulated the temperature distribution in the accreting protoplanetary disk ; [1[exp([qIR)]pT 4\ 4n L R * 2 1 2 3n R * ]R d d R C4 B AR AH RBD]3GM * M0 1[AR *B1@2D 8nR3 R C (7) where p is the Stefan»Boltzmann constant and L * is the luminosity of the central star. The –rst term of the right-hand side represents the heating by the central star and the second term represents the heating by viscous accretion.We adopt L *\1L _ where L _ is the luminosity of the present sun. The left-hand side of eqn. (7) represents the thermal radiation of the disk. The optical depth q of the disk in IR wavelengths is given by (8) IR qIR\iR 30 KB2 cm2 g~1 where i is the absorption coefficient. i\1] The Aabsorption T coefficient i is assumed to be (9) from Krué gel and Siebenmorgen,25 who calculated the absorption coefficient of various kinds of dust particles with Mie theory. Y . Aikawa et al. 285 The structure of the accreting protoplanetary disk is given by solving eqn. (1)»(9). M0 \1]10~8 M Adopting year~1 a\0.01 and L *\1L _ we obtain the tem- _ perature distribution as shown in Fig. 2(a). In the region of radius RZ10 AU with which we are concerned (see below) the accretion energy is much less than the radiation energy from the central star and the –rst term dominates in the right-hand side of eqn.(7). Then the temperature is nearly proportional to R~3@7 as long as the disk is optically thick to the thermal radiation of the disk itself in IR wavelengths.26 At larger radius RZ100 AU where the disk is optically thin to the thermal radiation the temperature is almost independent of the distance from the central star. Fig. 2(b) shows the distribution of the column density R. The column density is proportional to R~15@14 at RB10 AU where the disk is optically thick to the IR radiation while it is proportional to R~3@2 at RZ100 AU where the disk is optically thin. Note that the cosmic rays penetrate the disk at RZ2 AU since the column density of the disk is smaller than the attenuation length of the cosmic rays ca.96 g cm~2.15 Thus ionization by cosmic rays should be taken into account in investigating the molecular evolution in this region. v The matter accretes ca. 102 AU in 106 R . Fig. 2(c) shows the accretion velocity years. Finally Fig. 2(d) shows the number density of the hydrogen nuclei n in the disk. H 2.2 Reaction network In protoplanetary disks chemical equilibrium will not be achieved within the dynamical timescale of the disk owing to the low density and low temperature. Thus we investigate the evolution of molecular abundance in the disk by solving the reaction equations. The M_ Fig. 2 Distribution of (a) temperature (b) column density (c) accretion (radial) velocity and (d) number density of hydrogen nuclei in the steady accretion disk model.We assume M0 \1]10~8 year~1 a\0.01 and L *\1L _. 286 Molecular evolution in planet-forming circumstellar disks basic reaction equation is (10) j dx(i) dt \; aij x( j)]; bijk x( j)x(k)nH]; cijkl x( j)x(k)x(l)nH nH jk jkl the time. The –rst term on the right-hand side of eqn. (10) represents reactions with where x(i) is the abundance x(i)4n(i)/nH n(i) is the number density of species i and t is external particles such as ionization by cosmic rays. The second term represents twobody reactions and the third term represents three-body reactions in which species i is formed by the reaction of species j and k with the excess energy carried away by species l.We adopt reaction rates in the gas phase which are essentially the same as the UMIST94 data base.27,28 We also take into account some three-body reactions from the reaction network adopted in atmospheric chemistry.29 Unfortunately however reaction rate coefficients for the three-body reactions are still scarce. Thus we consider the region of the density given by the number of hydrogen n nuclei H[1012 cm~3 which corresponds to RZ10 AU in our disk model [Fig. 2(d)]. With a typical rate coefficient kB10~30 cm6 s~1 for the three-body reactions we have in this density range kn(H2)[10~18 cm3 s~1 even for the most abundant species H2 as a third body. This is less than the characteristic rate coefficient ca. 10~17 cm3 s~1 for radiative association which is the slowest of the two-body reactions.Thus the threebody reactions are not important in this density range. The attenuation length of cosmic rays 96 g cm~2,15 is much larger than the column density of the disk at RZ10 AU. Thus we adopt the ionization rate in the molecular cloud f\1.3]10~17 s~1 which is derived from the observed abundance of ions in molecular clouds.30 Although we take into account ionization and dissociation by interstellar and stellar UV radiation they do not aÜect the results. The UV radiation is attenuated by grains in a thin surface layer of the disk (RB10~3 g cm~2). In addition to the reactions in the gas phase we also take into account reactions on the grain surfaces formation of H molecules recombination of ions and electrons formation of an ice mantle due to the adsorption of molecules and thermal desorption 2 of the ice mantle.We assume that the sticking probability of a gas particle when it collides with a grain is 0.3.31 The desorption rate is determined from experimental data32,33 and theoretical estimates.34 For simplicity and clarity we do not consider other chemical reactions on grains surfaces because detailed mechanisms and reactionrate coefficients are not well-understood. The abundance of elements that we adopt in this study is shown in Table 1 which is the abundance in the interstellar medium in which heavy elements are depleted to some extent relative to solar abundance forming grain cores (see e.g. ref. 35). We determine the initial molecular abundance of the protoplanetary disks referring to the abundance in the molecular clouds since the star and the protoplanetary disks Table 1 Abundances of elements and grainsa abundance element abundance element Mg Si S Fe H He C N grain 1.09([6) 9.74([7) 9.14([6) 2.74([7) 6.24([3)c O Na 1.00 7.0([2)b 7.86([5) 2.47([5) 1.80([4) 2.25([7) a The abundance is by number except for grains.b a(b) means a]10b. c The abundance of grain cores by mass relative to hydrogen. 287 Y . Aikawa et al. are formed in the molecular clouds. In this paper we assume that initially all carbon is in the form of CO the remaining oxygen is locked up in the water ice mantle and nitrogen is present as N Calculation is performed up to 3]106 years which corre- 2 .sponds to the typical life time of the accretion phase.36 3 `. H Through reaction with 3 ` CO is transformed into 3 Numerical results In this section we show the results of the numerical calculations on molecular evolution. We have introduced two new processes which might aÜect molecular evolution in the protoplanetary disks ; ionization by cosmic rays and mass accretion towards the central star. We have found that their eÜect on molecular evolution is complex. Hence we –rst describe molecular evolution in the disk ignoring accretion towards the central star in order to understand the eÜect of ionization clearly. We then show the evolution and distribution of molecular abundance in the accretion disk and describe the eÜect of mass accretion on the evolution of molecular abundance.3.1 Evolution of molecular abundance in the disk without accretion We have found that there are two types of reaction which critically determine molecular evolution production of ions by cosmic rays and formation of the ice mantle. As examples we show the molecular evolution in two representative regions ; R\90 and 10 AU. nH\2.11]109 cm~3 and the temperature is T \20.2 K. Initially all carbon 4 3.1.1 C- N- O-chemistry at R= 90 AU. Fig. 3(a) shows the time variation of the abundance of carbon-bearing molecules relative to hydrogen at R\90 AU where the density is is assumed to be in the form of CO. With time CO ice HCN ice CH ice and H2CO ice are produced. The main formation processes of these molecules are shown in Fig.4. 2 First a fraction of H molecules is ionized by the cosmic-rays to produce H2 ` ions which reacts with H to form H 2 2 Fig. 3 Evolution of molecular abundance of (a) carbon-bearing molecules (b) nitrogen- and n oxygen-bearing molecules. Density H\2.11]109 cm~3 and temperature T \20.2 K which correspond to R\90 AU in our disk model. Initially all the carbon is in the form of CO nitrogen is in the form of N and remaining oxygen is in the form of water ice. 2 288 Molecular evolution in planet-forming circumstellar disks Fig. 4 (a) Main reactions in carbon reaction-network. G([) indicates grain-surface recombination ; (rad) indicates radiative association. 2 HCO` some of which becomes HCO via grain-surface recombination.Reactions of HCO with O and N –nally form CO and HCN in the gas phase. Although there is neither O atom nor N atom initially these are extracted from CO and N molecules via 2 ion»molecule reactions. Reaction of HCO with HCO produces H CO. 2Helium is also ionized by the cosmic rays to form He` ion. The helium ion He` also reacts with CO to 2 produce C`. Radiative association and H-atom abstraction from H transforms C` into CH5 ` C which recombines with electrons to produceH4 . 4 Owing to the high density and low temperature these products are adsorbed onto grain surfaces in a timescale of ca. 10 years and accumulate in the ice mantle. Formation of the ice mantle acts as a ìsinkœ in the chemical reaction network; some fraction of the products might be destroyed and transformed into other molecules if they were in the gas phase.For example CH reacts with C` to form larger hydrocarbons at higher temperatures where CH is easily desorbed (see below). 4 Fig. 3(b) shows the time variation of the abundance of nitrogen- and oxygen-bearing With time HCN 2 2. N Finally H4 ` is turned into NH3 via dissociative recombination molecules. Initially all nitrogen is assumed to be in the form of N ice and NH ice become dominant. The main formation process of NH is as follows. 2 . 3 reacts with He` to form N` which is transformed into NH4 ` by repeated H atom- N 3 abstraction from H or grain-surface recombination. 3.1.2 C- N- O-chemistry at R= 10 AU. Fig. 5(a) shows the evolution of carbonbearing molecules at R\10 AU where the density is nH\5.89]1011 cm~3 and the temperature is 37.7 K.Since the temperature is so high HCO and CH cannot accumulate in the ice mantle. They are transformed into CO 4 2 H2CO and larger hydrocarbons by gas-phase reactions. 2 Fig. 5(b) shows the evolution of nitrogen- and oxygen-bearing molecules. Owing to the high temperatures CN NO O and OH cannot accumulate in the ice mantle and 2 are transformed into less volatile molecules such as HCN NO and H2O. 3.2 Evolution of molecular abundance in the accretion disk In the previous section we have shown the evolution of molecular abundance at a –xed temperature and density in order to understand the chemical reaction network clearly. 289 Y . Aikawa et al.which correspond to R\10 AU in our disk model. Other details are as in Fig. 3. Fig. 5 Evolution of molecular abundance of (a) carbon-bearing molecules and (b) nitrogen- and oxygen-bearing molecules. The density nH\5.89]1011 cm~3 and temperature T \37.7 K 3 Here we take into account mass accretion in the protoplanetary disks. We consider –rst the disk with radius RB1000 AU. The —uid cell at each radius accretes towards the central star with accretion velocity given by eqn. (6). The temperature and density in the —uid cell given by eqn. (7) and (5) change with time as the matter accretes towards the central star. As an example we show the evolution of molecular abundance in the matter which accretes from R\405 AU to R\10 AU in 3]106 years.First we show the evolution at tO2.4]106 years. Fig. 6(a) shows the temperature of the matter and the distance from the central star as a function of time. The matter accretes from R\405 AU to RB55 AU in 2.4]106 years and the temperature is almost constant during this period. Fig. 6(b)»(d) show the evolution of C- N- and O-bearing molecules. Since the temperature is almost constant T B20 K the evolution is very similar to that described in the previous section. CO is destroyed by ions such as He` and H3 ` formed by cosmic-ray ionization. The carbon is transformed to CO2 HCN and hydrocarbons by reactions in the gas-phase and accumulates in the ice mantle. N is also destroyed by the He` ion and transformed into e.g. NH and HCN. 2 Since the evolution at tP2.4]106 years is too complicated to be shown with the HFormaldehyde is produced by the reaction time on a log-scale it is shown separately in Fig.7(a)»(d). The temperature of the matter and the distance from the central star is shown in Fig. 7(a). Matter accretes from 55 AU to 10 AU and the temperature rises rapidly during this period. As the matter is transformed into a warmer region some species are desorbed from the ice mantle. For example CH is desorbed at around 2.85]106 years (RB20 AU) since its sublimation temperature is ca. 28 K. Then CH is transformed into larger hydrocarbons such as 4 by ion»molecule reactions in the gas phase. CO exhibits complex C 4 2H2 and C3H3 behaviour. Its abundance decreases once in ca. 2.5]106 years ; CO is transformed into other carbon-bearing molecules such as and H2CO CO2 .However CO increases again at tB2.9]106 years (RB20 AU). The reason is that CO is formed from radicals such as HCO and OH which are desorbed from the ice mantle. CO is also produced from CH3 C4. H2CO which is formed from 290 Molecular evolution in planet-forming circumstellar disks Fig. 6 Evolution of molecular abundance in the matter which accretes from R\405 to ca. 55 AU in 2.4]106 years ; (a) distance from the central star and the temperature of the matter as a function of time and (b)»(d) evolution of C- N- and O-bearing molecules. HCO]HCO]H increase as their precursor molecules (CN and NO) sublime. 2CO]CO at tB2.9]106 years. Similarly OCN and NO2 ice 3.3 Distribution of molecules in the accretion disk We perform the same calculation as described in the previous subsection for the matter which accretes from R\436 to 20 AU from 467 to 30 AU from 497 to 40 AU from 291 Y .Aikawa et al. Fig. 7 Evolution of molecular abundance in the matter which accretes from RB55 to 10 AU in t\2.4]106»3]106 years ; (a) distance from the central star and the temperature of the matter as a function of time and (b)»(d) evolution of C- N- and O-bearing molecules. 525 to 50 AU from 553 to 60 AU from 580 to 70 AU and from 631 to 90 AU respectively. Thus we get the distribution of molecular abundance at t\3.0]106 years. Fig. 8(a)»(c) show the distribution of C- N- and O-bearing molecules. The vertical bar shows the sublimation front where the temperature is equal to the sublimation 292 Molecular evolution in planet-forming circumstellar disks on the bar in units of K.Fig. 8 Distribution of molecular abundance in the accretion disk at t\3]106 year. The vertical bar shows the sublimation front of the molecules labelled. The sublimation temperature is labelled are more abundant at temperature. The sublimation temperature is labelled on the bar in units of K. We can see that the molecular abundance is almost the same at RZ50 AU where the temperature is nearly constant T B20 K. Since molecular evolution is critically dependent on temperature the molecular abundance is similar in regions of similar temperature. At R[50 AU the temperature is higher in the inner region and the molecular abundance is quite diÜerent from that in the outer radius e.g.CH is less abundant at R[20 AU 4 than in the outer region. Instead larger hydrocarbons such as C2H2 and C3H3 formed from CH R[20 AU than in the outer region. 4 , 293 Y . Aikawa et al. 2 4 Discussion 4.1 Uncertainties in the grain-surface recombination of HCOë One of the merits in solving directly the reaction equations is that we can –nd out the main reactions which determine the molecular evolution from amongst thousands of reactions. Analysing the reaction network we have found a key reaction which critically determines the transformation rate of CO to CO and H2CO grain-surface recombination of HCO`. It is normally assumed that grain-surface recombination occurs without dissociation if the newly formed neutral species is stable.Thus HCO` would be neutralized to HCO by depositing excess energy to the grain surface instead of being dissociated to H and CO. CO and H2CO are then formed from HCO. In the protoplanetary disk the ratio of negatively charged-grain/electron is larger than in the molecular clouds because of the higher density. Thus grain-surface recombination is more important in the protoplanetary disk. Unfortunately however there is no direct measurement or calculation at present on whether the grain-surface recombination of HCO` is dissociative or not. Thus we perform the calculation considering another option H and CO are produced by grainsurface recombination of HCO`. 2 In Fig. 9 we compare the evolution of molecular abundance in the matter accreting from R\580 to 70 AU in the two cases ; (a) HCO` recombines to form HCO and (b) HCO` recombines to form CO and H atom.Formation of CO and HCN is considerably suppressed and thus CO is abundant at t\3]106 years in the dissociative case 2 Fig. 9 Evolution of molecular abundance in the matter which accretes from R\580 to 70 AU. The product of the grain-surface recombination of HCO` is HCO in (a) while it is CO]H in (b). 294 Molecular evolution in planet-forming circumstellar disks (b). The abundance of CH4 N3 hydrocarbons andHare not much aÜected since they are formed through other reaction networks. Fig. 10 shows the distribution of molecules at t\3]106 years in the case of dissociative recombination. Among the carbon-bearing molecules CO is the most abundant through CO R\10»90 AU.The abundances of CO2 H HCN and whose precursor is HCO are less than those in Fig. 8 by about an order of magnitude. On the other hand 2 distribution of other molecules such as CH4 N3 hydrocarbons andHare similar to those in Fig. 8 since they are formed through other reaction networks which are irrelevant to HCO. the same as in Fig. 8. Fig. 10 Distribution of molecular abundance in the accretion disk at t\3]106 years. The product of the grain-surface recombination of HCO` is assumed to be CO]H. Other details are 295 Y . Aikawa et al. Since the measurement or the quantum calculation on grain-surface recombination is beyond the scope of this paper we have assumed that HCO` recombines to form 2 H2CO HCN and and a HCO. This gives an upper limit on the abundance of CO lower limit on the abundance of CO.4.2 EÜect of varying ionization rate In the previous section we have found that the main cause for the molecular evolution is ionization by cosmic rays in the outer region of the protoplanetary disks. Since the attenuation length of cosmic rays is much larger than the column density of the disk at RZ10 AU we adopted the same ionization rate f\1.3]10~17 s~1 as in the molecular clouds. However there are some uncertainties in this value. For example ionization by cosmic rays can be as low as fB10~19 s~1 if the low-energy cosmic rays are scattered by magnetized stellar wind such as the solar wind although the precise value is strongly dependent on physical parameters of the stellar wind such as the strength of the magnetic –eld.Cosmic rays might be scattered by magnetic –elds in the disk if the disk is magnetized and turbulent.37 On the other hand there is also a possibility of a larger ionization rate. Since supernovae are considered to be a source of cosmic rays the ionization rate in the protoplanetary disk can be considerably larger if the disk is formed near a supernova. The evidence for live 26Al in Allende refractory inclusions suggests the possibility that the solar system was formed near the supernova.38 As well as cosmic rays radioactive elements can be a source of ionization. The abundance of 26Mg in Allende refractory inclusions indicates that the ionization rate via the decay of 26Al was several]10~19 s~1 in the solar nebula.15 Thus we investigate the eÜect of varying ionization rate in the range 10~18Of/s~1O10~16.is less in the case of the lower ionization rate. As an example we show the evolution of molecular abundance in matter which accretes from R\580»70 AU in Fig. 11. The ionization rate f is (a) 1.3]10~18 (b) 1.3]10~17 and (c) 1.3]10~16 s~1. We can see that molecular evolution is critically dependent on the ionization rate. The abundance of molecules such as CO2 CH4 and NH are smaller by a factor of 3»10 for the lower ionization rate (a) at t\3]106 years. The formation of these molecules is triggered by the reaction of CO and N with 3 He` and H3 `. H Since He` and are formed by cosmic-ray ionization the abundance 2 of these ions is proportional to the ionization rate.Thus the formation rate of CH and 3 ` NH 4 3 Fig. 11 Evolution of molecular abundance in the matter which accretes from R\580 to 70 AU. The ionization rate is (a) 1.3]10~18 (b) 1.3]10~17 and (c) 1.3]10~16 s~1. 296 Molecular evolution in planet-forming circumstellar disks In the case of a higher ionization rate (c) the abundance of CH4 hydrocarbon and NH is larger than for case (b) while the abundance of CO2 H2CO HCN and is not. The latter molecules are formed from HCO which is produced via the grain-surface 3 recombination of HCO`. Although the reaction of CO with H3 ` is accelerated owing to the higher abundance of H3 ` the eÜect is somewhat compensated by the dissociative recombination of HCO` because the abundance ratio of electron/negatively chargedgrain is larger in the case of the higher ionization rate (c).The ionization rate fB10~18 s~1 would be almost the lower limit in the solar nebula considering the abundance of 26Mg in Allende refractory inclusions and the present cosmic-ray —ux in the heliosphere (ref. 15 and references therein). In this lower limit the abundance of CH ni/nHB10~6 and CO is 4 CO2 N3 HCN andHis as low as n the most dominant component CO/nH\10~4 among C-bearing molecules at t\3]106 years. Since the ionization rate is an important parameter which determines the evolution of molecular abundance in the disk we propose that it should be estimated from the observation of ions in the protoplanetary disks. 4.3 EÜect of grain growth So far we have assumed for simplicity that the dust/gas mass ratio is ca.6.24]10~3 and the grain radius is a\10~5 cm which is a typical value for interstellar grains (see e.g. ref. 39). However these values may change as the dust grains coagulate and sediment towards the midplane of the disk. We investigate the eÜect of sedimentation and growth of grains on molecular evolution in the disk. Fig. 12 shows the molecular evolution of matter which accretes from R\580 to 70 AU where the grain radius is (a) 10~5 (b) 10~4 (c) 10~3 cm. The dusk/gas mass ratio is assumed to be the same as that in interstellar clouds. In the case of larger grains the abundances of CO ice H2CO ice and HCN ice are smaller by one or two orders of 2 magnitude at t\3.0]106 years than in for case (a). The reason is that their abundance is critically dependent on the rate of the grain-surface recombination of HCO` which is signi–cantly less in the case of larger grains.The geometrical total cross-section of the grain particles na2n (grain) is inversely proportional to the grain radius. Moreover the Coulomb force at the grain surface which attracts the ions is less for the larger grains. On the other hand the abundance of CH ice and NH ice at t\3]106 years is 4 similar for case (a) and case (b) and smaller by a factor of ca. 5 in case (c). The adsorp- 3 Fig. 12 Evolution of molecular abundance in the matter which accretes from R\580 to 70 AU. The grain radius is (a) 1]10~5 (b) 1]10~4 and (c) 1]10~3 cm. 297 Y . Aikawa et al. tion timescale is dependent on the total cross-section of the grains and is smaller by two orders of magnitudes for case (c) than for case (a).Thus more CH and NH is destroyed 4 and transformed into other molecules before it is adsorbed onto grains in the case of the 3 larger grains (c). In the accretion phase the timescale of the grain»grain collision is qcol\Mna2vther n(grain)N~1 \1.6]106A20 KB1@2A a T B 10~5 cmB5@2A6.24] m 10~3BA108 n cm~3 years (11) H where m is the dust/gas mass ratio and nH\108 cm~3 corresponds to RB200 AU in our disk model. Micrometre-sized grains exhibit a thermal (Brownian) velocity spectrum because they couple efficiently to the smallest scales of the turbulence in which the gas —ow will be laminar.40,41 The matter which is in the region of radius R\10»90 AU at t\3]106 years is in the region of RZ200 AU for the –rst 1]106 years (Fig.6). Thus grains in the accreting matter cannot be much larger than aB10~5 cm and the molecular evolution is not much aÜected by the growth of grains at t[1]106 years. As the matter accretes towards the inner regions the timescale of the grain-growth become T q less colB ; 5]106A20 KB1@2A10~4 a cmB5@2A6.24] m 10~3BA1010 n cm~3 years (12) B 2 H2CO HCN and would be suppressed H n where H\1010 cm~3 corresponds to RB50 AU in our disk model. If the sticking probability in a grain»grain collision is unity grains in the accreting matter can be as large as aB10~4 cm and the formation of CO in the last 2]106 years. Thus the abundance of HCN and CO at t\3]106 years i/nHB10~6 in the region of R\10»90 AU.The abundance of CH4 n can be as low as 2 and NH are little aÜected by the grain growth as long as the grain radius is a[10~4 cm. 3 In the post-accretion phase the growth of grains would be accelerated. Once the disk becomes quiet enough grains begin to sediment towards the midplane of the disk. The sedimentation velocity of the grains is determined by the balance between gas-drag force and the vertical component of the stellar gravity and is proportional to the grain radius. Thus larger grains sink faster and collide with smaller grains sinking more slowly. According to Nakagawa et al.42 the collision time by this process at height Z is tcolB 3nm 2 H Z tK (13) ZBA R B1.2]104A6.24]10~3BAH 50 AUB3@2 years m where t is the Keplerian timescale.This timescale is independent of the grain size. Hence the timescale of molecular evolution would be much greater in the post-accretion K phase than in the accretion phase if the sticking probability of grain»grain collision is Z10~2 although the eÜect is somewhat compensated by the larger dust/gas ratio in the midplane owing to sedimentation of the dust grains. 5 Application 5.1 Comparison with the chemical composition in comets It is widely accepted that comets are one of the most pristine objects in the solar system preserving a large amount of volatile material such as water. They are now considered 298 to be the remnants of icy planetesimals formed in the outer region (RZ20 AU) of the solar system.In this section we compare the molecular abundance in our disk model with that in comets. Table 2 shows the molecular abundance in recent comets.43h45 The molecular abundance is expressed as a ratio relative to water ice which is the most dominant component in comets. One of the most important characteristics of the molecular composition in comets is the coexistence of oxidized ice (e.g. CO and CO2) and reduced ice (NH and CH 3 In previous studies of the molecular evolution in the solar nebula the formation of 4) are produced from CO which initially accretes from molecular clouds via2 4/H2O NH3/H2O CO2 /H2O areZa few reduced species are invoked to be catalytic reactions such as Fischer»Tropsch-type reactions or reactions in the sub-nebula around forming Jovian planets (ref.47 and references therein). On the other hand we have found that the coexistence is reproduced in the protoplanetary disk by considering the cosmic-ray ionization in the disk. Both CO ion»molecule reactions. The ratios CH and CH4 percent at tZ106 years which is consistent with the composition in comets although the precise value is somewhat dependent on the physical parameters in the disk (Section 4). CO and N would also coexist in the ice mantle even in the region of radius R[40 2 AU if they are physically trapped in water ice,48 although this eÜect is not included in our present calculation. There is another scenario on the origin of cometary ice ; cometary ice is considered to be unprocessed interstellar matter.43,46 Indeed our model shows that interstellar ice contributes to the molecular abundance in comets as long as it is not desorbed e.g.most of the water ice in our model is of interstellar origin. However note that our model molecule H CO2O H2CO CO CH2 4 NH3 HCN N 3OH CH 2 a The abundance is by number. Molecular evolution in planet-forming circumstellar disks as can be seen from Table 2.46 Table 2 Volatile abundance in recent comets relative abundancea 100 ca. 7»8 20 2 1»3 0»5 0.04 0.1 3 \0.2»1.2 0»2 \0.2 1.5»4.5 0.7 0.1»0.3 1»2 0.25 0.1 \0.02 0.03»0.2 ca. 0.02 ca. 1 ca. 1 ca. 5 object Comet P/Halley West (1976 VI) Brad–eld (1979 X) Austin (1990 V) Comet P/Halley Levy (1990 XX) Austin (1990 V) Comet P/Halley Comet P/Halley Comet P/Halley Levy (1990 XX) Wilson (1987 VII) Hyakutake Comet P/Halley Comet P/Halley Hyakutake Comet P/Halley Comet P/Halley several comets Comet P/Halley Comet P/Halley Levy (1990 XX) Austin (1990 V) and comments variable if a parent molecule if a parent molecule IR obs.(Vega 1 IKS) ground-based IR obs. Giotto IMS; model dependent ground-based IR obs. airborne IR obs. variable ; based on obs. of NH2 Giotto IMS variable ; ground-based radio Giotto IMS ground-based N2 ` emission Giotto NMS and IMS 299 Y . Aikawa et al. predicts that many molecules are produced in the protoplanetary disks in addition to those in the interstellar component.Our model also shows that the grain size and the cosmic-ray —ux critically determine the production rate of molecules in the protoplanetary disk. Precise observation of cometary ice and interstellar ice will tell us how much was produced in the protoplanetary disk and thus will give a constraint on the cosmic-ray —ux ionization state and timescale of grain growth in the solar nebula. 5.2 Chemical variation and formation site of comets are more abundant in In the observations of comets an array of evidence is accumulating in support of the idea that comets exhibit internuclear heterogeneity. The ratio of production rates for molecules varies strongly from comet to comet (Table 2). Although the variation may also be related to reprocessing after formation these eÜects must be considered along with cosmogonic diÜerences i.e.the molecular abundance of each comet must be dependent on where it was formed. Indeed our model shows that the molecular abundance varies with distance from the central star. Hence we can discuss the possibility of estimating the formation site of each comet from its molecular composition. First we can make use of the fact that the abundance of molecules in ice changes drastically in the region of their sublimation temperature (sublimation front). For example CH is abundant in comets formed at RZ20 AU but depleted in comets 4 formed at R[20 AU (Fig. 8). The abundance of CO shows similar behaviour; being almost constant at the outer radius and decreasing at RB40 AU where the temperature is equal to the sublimation temperature (ca.20 K). Secondly our model shows that some species such as C3H3 increase inwards inside the sublimation front of HCO the inner region. Since they are formed from precursor species such as CO and CH their abundance increases inside the sublimation front of precursor molecule. For 4 example C R[20 AU. Thus comets with more C 3H3 increases inwards at 3H3 would have been formed further inside the inner region than those with less C3H3 . Similarly CO (R[20 AU). and H 2Note that the latter dependence of molecular abundance on the formation site can 2CO for example. only be obtained from the chemical reactions in the disk. If all the volatiles in comets were unprocessed interstellar matter the molecular abundance in the comets would depend on the formation site only as –rst described ; the comets with more C3H3 would have been formed further into the outer region than those with less C3H3 According to AœHearn et al.,49 comets which were formed at the Kuiper belt are depleted in carbon-chain species such as C and C compared with comets which were 2 formed in the region of Uranus and Neptune.They observed 85 comets over a period of 3 17 years and investigated the molecular production rate of each comet. It was found that a certain group of comets called ìJupiterœs familyœ is depleted in carbon-chain species such as C and C3 by a factor of 20 compared with other groups. They argue that most Jupiter-family comets are expected to have come originally from the Kuiper belt beyond 2 Neptune and Pluto while most of other group of comets such as the Halley-family and long-period comets are expected to have come originally from the region of Uranus and Neptune according to the statistical study on the orbital evolution.Since the large hydrocarbons C2H2 and C3H3 can be the parent molecule for these carbon-chain molecules this observational result might suggest that the comets are made not only from interstellar matter but also from matter processed in the solar nebula. 6 Summary Evolution of molecular abundance in protoplanetary disks has been investigated. In the region of radius RZseveral AU the column density of the disk is less than the attenuation length of cosmic rays (ca.96 g). Ionization by cosmic rays plays an important role 300 Molecular evolution in planet-forming circumstellar disks in molecular evolution in these regions. A considerable fraction of CO which has accreted from the interstellar clouds is destroyed by ions and –nally transformed into CO2 HCN H2CO and CH4 . In the region where the temperature is low enough for these products to freeze onto grains they are adsorbed onto grains in a short timescale because of the high density in the disk. The products then accumulate in the ice mantle. Similarly N is transformed into HCN and NH3 . If the ionization rate and the grain size in the disk is the same as in molecular clouds the timescale of the molecular evolution in which CO and N are transformed into 2 other molecules is ca.106 years and is slightly less than the life time of the disk. The timescale of molecular evolution is less for the higher ionization rate while it is larger for lower ionization rate or larger grain-size. 2 The molecular distribution is strongly dependent on the temperature distribution. are abundant. The molecular abundance derived from our model is compared with that in comets. must have been formed in the region of RZ20 AU while comets with abundant For example CH ice is abundant in the region of radius Z20 AU where the temperature is lower than its sublimation temperature (ca. 25 K). In the inner region CH is 4 desorbed and larger hydrocarbons formed from CH 4 4 We have found that our model reproduces the coexistence of oxidized ice and reduced ice which is observed in comets.It is suggested that the formation site of each comet can be estimated from its molecular abundance. For example comets with abundant CH carbon-chain molecules are formed in the inner region. Detailed observation of molecu- 4 lar abundance in comets will give a constraint on the cosmic-ray —ux or the timescale of the grain-growth in the early stages of our solar system because molecular abundance in the disk is dependent on the ionization rate and grain size. References 1 C. R. OœDell and W. Zheng Astrophys. J. 1994 436 194. 2 A. I. Sargent and S. V. W. Beckwith Astrophys. Space Sci. 1994 212 181. 3 M. F. Skrutskie R. L. Snell K. M. Strom S. E. Strom S. Edwards Y. Fukui A. Mizuno M.Hayashi and N. Ohashi Astrophys. J. 1993 409 422. 4 S. Guilloteau and A. Dutrey Astron. Astrophys. 1994 291 L23. 5 T. Handa S. M. Miyama T. Yamashita T. Omodaka Y. Kitamura M. Hayashi T. Onishi R. L. Snell S. Strom K. Strom M. F. Skrutskie S. Edwards N. Ohashi K. Sunada M. Saito Y. Fukui A. Mizuno J. Watanabe and H. Kataza Astrophys. J. 1995 449 894. 6 A. Dutrey S. Guilloteau and M. Gueç lin Astron. Astrophys. 1997 317 L55. 7 R. Kawabe M. Ishiguro T. Omodaka Y. Kitamura and S. M. Miyama Astrophys. J. 1993 404 L63. 8 D. W. Koerner A. I. Sargent and S. V. W. Beckwith Icarus 1993 106 2. 9 A. Dutrey S. Guilloteau and M. Simon Astron. Astrophys. 1994 286 149. 10 D. W. Koerner and A. I. Sargent Astron. J. 1995 109 2138. 11 M. Saito R. Kawabe M. Ishiguru S.M. Miyama M. Hayashi T. Handa Y. Kitamura and T. Omodaka Astrophys. J. 1995 453 384. 12 J. S. Lewis Science 1974 186 440. 13 J. S. Lewis S. S. Barshay and B. Noyes Icarus 1979 37 190. 14 R. G. Prinn in Protostars and Planets III ed. E. H. Levy and J. I. Lunine University of Arizona Press 1993 Tucson 1005. 15 T. Umebayashi and T. Nakano Publ. Astron. Soc. Jpn. 1981 33 617. 16 D. Lynden-Bell and J. E. Pringle Mon. Not. R. Astron. Soc. 1974 168 603. 17 C. Bertout G. Basri and J. Bouvier Astrophys. J. 1988 330 350. 18 J. Frank A. King and D. Raine in Accretion Power in Astrophysics Cambridge University Press 1992 p. 77. 19 S. A. Balbus and J. F. Hawley Astrophys. J. 1991 376 214. 20 J. F. Hawley and S. A. Balbus Astrophys. J. 1991 376 223. 21 N.I. Shakura and R. A. Sunyaev Astron. Astrophys. 1973 24 37. 22 G. Basri and C. Bertout in ref. 14 p. 543. 23 A. G. W. Cameron Space Sci. Rev. 1973 15 121. 24 S. P. Ruden and J. B. Pollack Astrophys. J. 1991 375 740. 25 E. Krué gel and R. Siebenmorgen Astron. Astrophys. 1994 288 929. 26 T. Kusaka T. Nakano and Hayashi Prog. T heor. Phys. 1970 44 1580. 301 Y . Aikawa et al. 27 T. J. Millar J. M. C. Rawlings A. Bennett P. D. Brown and S. B. Charnley Astron. Astrophys. Suppl. Ser. 1991 87 585. 28 P. R. A. Farquhar and T. J. Millar CCP7 Newsletter 1993 18 6. 29 G. Brasseur and S. Solomon in Aeronomy of the Middle Atmosphere ed. G. Brasseur and S. Solomon Reidel Dordrecht 1986. 30 S. Lepp in T he Astrochemistry of Cosmic Phenomena ed. P. D. Singh Kluwer Dordrecht 1992 p.471. 31 D. A. Williams in Dust and Chemistry in Astronomy ed. T. J. Millar and D. A. Williams Institute of Physics London 1993 p. 143. 32 Y. Yamamoto N. Nakagawa and Y. Fukui Astron. Astrophys. 1983 122 171. 33 S. A. Sandford and L. J. Allamandola Astrophys. J. 1993 417 815. 34 T. I. Hasegwa and E. Herbst Mon. Not. R. Astron. Soc. 1993 261 83. 35 D. C. Morton Astrophys. J. 1974 193 L35. 36 K. M. Strom S. E. Strom S. Edwards S. Cabrit and M. F. Skrutskie Astron. J. 1989 97 1451. 37 A. Z. Dolginov and T. F. Stepinski Astrophys. J. 1994 427 377. 38 D. D. Clayton Nature (L ondon) 1985 315 633. 39 J. S. Mathis W. Rumpl and K. H. Nordsieck Astrophys. J. 1977 217 425. 40 S. J. Weidenschilling and J. N. Cuzzi in ref. 14 p. 1031. 41 A. Chokshi A. G. G. M. Tielens and D. Hollenbach Astrophys. J. 1993 407 806. 42 Y. Nakagawa K. Nakazawa and C. Hayashi Icarus 1981 45 517. 43 M. J. Mumma P. R. Weissmen and S. A. Stern in ref. 14 p. 1177. 44 H. A. Weaver P. D. Feldman M. F. AœHearn C. Arpigny J. C. Brandt C. E. Randall M. A. Disanti 46 T. Yamamoto in Comets in the Post-Halley Era ed. R. L. Newburn Jr. et al. Kluwer Academic M. J. Mumma N. Dello Russo D. X. Xie M. Fomenkova and K. Magee-Sauer IAU Circular 1996 6374. 45 D. Lis J. Keene K. Young T. Phillips E. Bergin P. Goldsmith D. Bockelee-Morvan J. Crovisier D. Gautier A. Wootten D. Despois T. Owen J. Geophys B. Butler P. Palmer D. Yeomans and D. W. E. Green IAU Circular 1996 6362. Dordrecht 1991 361. 47 R. G. Prinn and Fegley Jr. in Origin and Evolution of Planetary and Satellite Atmospheres ed. S. K. Atreya J. B. Pollack and M. S. Matthews University of Arizona Press Tucson 1989 78. 48 A. Bar-Nun G. Hertman D. Laufer and M. L. Rappaport Icarus 1985 63 317. 49 M. F. AœHearn R. L. Mills D. G. Schleicher D. J. Osip and P. V. Birch Icarus 1995 118 223. Paper 8/00258D; Received 8th January 1998

ISSN:1359-6640    

DOI:10.1039/a800258d

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Faraday Discuss. 1998 109 303»319 Inorganic dust formation in astrophysical environments Hans-Peter Gaila,§ and Erwin Sedlmayrb,î a Institut f ué r T heoretische Astrophysik Universitaé t Heidelberg T iergartenstr. 15 D-69121 Heidelberg Germany b Institut f ué r Astronomie und Astrophysik T echnische Universitaé t Berlin Hardenbergstr. 36 D-10623 Berlin Germany 2 The problem of inorganic dust formation is discussed for the conditions of cool late-type giants and supergiants having an oxygen rich element composition. Particular emphasis is put on the formation of the primary condensates i.e. the –rst kind of grains forming from the gas phase. By explorative chemical equilibrium discussion it has been shown that titanium oxides especially TiO2 molecules are the most probable candidates to nucleate –rst.Though SiO is by far the most abundant reacting oxygen bearing species it can usually be ruled out at high temperatures because eÜective condensation takes place only below 600 K under typical M star conditions. The situation regarding the nucleation of aluminium bearing molecules to form corundum is not as clear. Though corundum has an extremely high stability the constituting aluminium oxides are much less stable at temperatures above the onset of TiO condensation. Therefore it might be argued that corundum is also not likely to constitute the –rst condensate in the out—ows of M stars. 1 Introduction Generally the cosmic dust complex is discussed in view of three fundamental aspects regarding (1) the molecule»solid transition i.e.the formation of macroscopic specimens out of the gas phase (2) grain processing i.e. the modi–cation of already existing clusters or dust particles by growth erosion sputtering coagulation etc. and (3) the role of the various grain components in the realistic modeling of both the actual appearance and the secular evolution of real astrophysical objects. In this paper we focus on the aspect of primary grain condensation in order to gain insight into the –rst nucleating molecules in question their speci–c condensation conditions required and on the general local conditions provided by the dust forming objects. In this respect winds of late type giants and supergiants turned out to be those objects contributing by far the largest fraction of stellar material and hence of newly formed dust ejected into the interstellar medium.Owing to its high bond energy the CO molecule can only be dissociated by rather energetic photons. In the absence of strong dissociating UV radiation –elds nearly all carbon (oxygen) is therefore consumed by CO formation in the case of oxygen (carbon) rich compositions with the consequence that in the case of M stars (C stars) practically no additional carbon (oxygen) compounds can be formed. This fact does not only determine the type of chemistry developing in such stellar out—ows but also distinguishes those species from which the primary condensate –rst forms out of the gas phase. § E-mail gail=ita.uni-heidelberg.de î E-mail sedlmayr=physik.tu-berlin.de 303 304 Inorganic dust formation in astrophysical environments While for cool carbon rich out—ows considerable progress has been achieved in the detailed understanding and consistent modeling of the grain formation process during the last decade1h3 a comparable description of the formation of inorganic dust in oxygen rich situations is still not at hand.Though the underlying criteria controlling circumstellar primary condensate formation [grains should be formed from (i) abundant species (ii) molecules not blocked by high bond energies and (iii) species resulting in high temperature condensates allowing the product to survive the often hostile environmental conditions] apply for both situations the corresponding key molecules as well as the detailed nucleation mechanisms are basically diÜerent.While carbon grain formation in C star winds proceeds via a considerably well de–ned pathway along which more and more complex aromatic hydrocarbons evolve which –nally result in macroscopic carbon grains the situation in M giant out—ows is not as clear. With regard to the expected condensates detailed inspection of the data and the chemical composition con–rms oxides as the basic nucleation species (see also Fig. 1). In M giant out—ows SiO is by far the most abundant reactive oxygen bearing compound thus suggesting this to be the –rst species to condense out of the gas phase. Calculation of the nucleation rate however indicates that efficient SiO nucleation under M star shell conditions only takes place at temperatures well below 600 K (e.g.Gail and Sedlmayr4) a value only compatible with observations in the case of the earliest M stars having only a very low mass loss rate. Only in these cases might the SiO molecule be the –rst condensing species. This conclusion however cannot be true for M giants of later spectral type having considerably larger mass loss rates and exhibiting a temperature of the inner edge of the dust shell as high as 1000 K a fact which de–nitely rules out SiO molecules as the –rst species to condense out of the gas phase. Owing to the low stability of the resulting condensate phosphorus oxide phosphorus oxide can also be ruled out as a key condensing species and all other species less abundant than TiO can be ruled out because of their very low concentration hence only the abundant aluminium oxides and the very stable titanium oxides remain as potentially possible primary condensing species.In order not to veil the fundamental principles and selection rules governing the speci–c molecule»solid transitions by the explicit complications of heterogeneous nucle- Fig. 1 Bond energies of particularly strongly bound diatomic molecules of abundant elements relevant for dust formation plotted against the abundance of the less abundant of its constituents (bond energies from Kide,5 abundances from Anders and Grevesse6) 305 H. P. Gail and E. Sedlmayr ation theory we focus our investigation on the discussion of equilibrium situations described by the law of mass action. In this way we expect to gain insight into both the relevant primary condensates emerging and the importance of the various formation processes to be taken into account in more realistic future approaches.for comparison the dissociation limit of H 2 Refractory condensates of low-abundance elements Studies of equilibrium condensation in solar abundance element mixtures7,8 have shown that the –rst condensates appearing in a cooling gas with standard cosmic element abundances are compounds of aluminium titanium and zirconium. From these elements only Al and Ti have a sufficiently high element abundance that formation of Al or Ti bearing clusters may be responsible for the formation of the seed nuclei in stellar out- —ows which are required for the growth of macroscopic dust grains.Let us consider what might be the outcome of a condensation process based on these two elements. 2.1 Titanium compounds Solid titanium compounds are formed under low pressure conditions as they are encountered in a circumstellar condensation zone (p[10~2 dyn cm~2 °) at temperatures below ca. 1500 K. Since CO and SiO have extremely high bond energies (1076.5 and 799.6 kJ mol~1) at such low temperatures the silicon and the carbon are bound in SiO and CO molecules respectively. The formation of these two molecules consumes an amount of the oxygen corresponding to the abundance of C and Si. The remaining oxygen at low temperatures forms H2O with the abundant hydrogen and the hydrogen itself is associated to H molecules. The equilibrium curve where one half of the H has associated to H is shown in Fig.2. We have to consider thus the formation 2 of titanium compounds from a gas mixture with this composition. Additionally there 2 exist a lot of less abundant compounds of the abundant elements and also compounds of other elements but these are of no interest for the problem of formation of solid titanium compounds. Fig. 2 Dissociation limits for Ti molecules and stability limits for the formation of solid Ti and for the formation of solid Ti compounds from molecules from the gas phase. The dashed line shows 2 . ° 1 dyn cm~2\0.1 Pa. 306 Inorganic dust formation in astrophysical environments Owing to the high bond energy of the TiO molecule (672.4 kJ mol~1) the titanium forms this molecule already at rather high temperatures in the atmospheres of K and M stars.Other titanium molecules have much less bond energy with the result that the formation of TiO consumes all the available Ti. Fig. 2 shows the equilibrium curve where half of the Ti is bound in TiO. Below this Ti is nearly completely bound in TiO. Since TiO is chemically not yet saturated a second O atom can be added to TiO to form TiO but because of the much lower bond energy of the second atom (248.4 kJ mol~1) this occurs at a much lower temperature. Fig. 2 shows the equilibrium curve where the 2 abundance of TiO equals that of TiO calculated for chemical equilibrium in a gas of standard cosmic element composition. Above this curve the TiO molecules dominate in 2 the gas phase below this curve the TiO molecules dominate.Other Ti compounds in the gas phase have negligible abundances. Hence we have to consider the formation of 2 titanium compounds from these two precursor molecules. For pressure»temperature conditions where the oxygen not bound in CO or SiO is bound in H2O the chemical equilibrium abundances of TiO and TiO in the gas phase can be determined by the following simpli–ed calculation. Consider the reaction 2 TiO]H2O]TiO2]H2 At chemical equilibrium the partial pressures of the molecules involved in the reaction satisfy p pTiO2 pH2 TiO pH2O \exp([*ºGE)\Kp P is used for the total pressure and for the –ctitious pressure of hydrogen nuclei were they all present as free atoms; p is used for the partial pressure of all species present in the gas phase.Data for calculating the change *ºGE of the Gibbs free energy for the reaction is taken from Sharp and Huebner.8 If hydrogen is nearly completely associated to H we have for the –ctitious partial pressure of hydrogen nuclei 2 (1) PH\ 1]2eHe 2P and for the partial pressure of water and hydrogen molecules (2) 2 pH2O\(eO[eC[eSi)PH pH2\12 PH By assuming that the fraction c of the titanium is bound in TiO molecules while the fraction 1[c is bound in TiO molecules we obtain (1[c)2(eO[eC[eSi) \Kp c and from this (3) c\ 1]2(e 2(eO[eC[eSi)Kp O[eC[eSi)Kp The variation of c with T is shown in Fig. 3. This expression is used below to calculate the fraction of the gas phase Ti bound in TiO2 . 2.1.1 Solid titanium oxides.The titanium oxides all are stable up to very high temperatures and are prime candidates for appearing as the –rst condensate in circumstellar out—ow. Several diÜerent oxides of titanium exist. These compounds are ionic crystals in which Ti appears as Ti3` or Ti4` and sometimes as the Ti2` cation. We consider here the following oxides TiO TiO2 Ti2O3 Ti3O5 and Ti4O7 and determine their stability limits against condensation in a standard cosmic element mixture. 307 H. P. Gail and E. Sedlmayr Fig. 3 Fraction c of the titanium in the gas phase bound in TiO molecules under conditions 2 The remaining fraction 1[c of Ti where the oxygen not bound into CO or SiO is bound in H is bound in TiO molecules. 2O. (g)]H2 ]TiO(s)]H2O Consider the formation of solid TiO from the gas phase.At the temperatures where solid TiO becomes thermodynamically stable at circumstellar pressure conditions most of the Ti has associated to TiO2 . A possible formation reaction of solid TiO then is TiO2 Since the hydrogen is almost completely in H and the oxygen not bound in CO or SiO at the relevant pressures and temperatures is bound in water vapour the partial pres- 2 p sures of the gas phase species are H2\12 PH pH2O\(eO[eC[eSi)PH and pTiO2\(1 [f )ceTiPH . Here we have assumed that the fraction f of all Ti is bound in solid TiO while the remaining fraction 1[f stays in the gas phase from which the fraction c action actually is in TiO2 . In chemical equilibrium the gas phase species satisfy the law of mass (4) \exp([*ºGE/RT )\K [TiO(s)] p p pH2O TiO2 pH2 from which we obtain (5) P\(eO[eC[eSi)(1]2eHe) (1[f )ceTiKp[TiO2(s)] This de–nes a curve in the P»T plane along which just the fraction f of all Ti has condensed into solid TiO.The curve with f\0 de–nes the limit of stability of the solid. This is shown in Fig. 2. Next we consider the formation of solid TiO from the gas phase according to the 2 reaction TiO(g)]TiO (s) 2 For the equilibrium state between TiO molecules from the gas phase and the TiO2 solid we have for the partial pressure of TiO molecules 2 2 1 (6) \exp([*ºGE/RT )\K [TiO (s)] p 2 pTiO2 and from this it follows (7) P\ 2(1[f )ce 1]2eHe TiKp[TiO2(s)] Here we have assumed that the fraction f of Ti is condensed into solid TiO while the 2 remaining fraction 1[f stays in the gas phase.c is the fraction of the gas phase Ti 308 Inorganic dust formation in astrophysical environments can be formed from gas phase species for instance by the which is in TiO2 . The limit curve f\0 of stability of solid TiO is shown in Fig. 2. At 2 pressures around P\10~4 dyn cm~2 this limit curve occurs above and to the left of the limit curve of stability of solid TiO. Thus solid TiO is more stable than solid TiO under conditions encountered in the condensation zone of a circumstellar dust shell and 2 the solid TiO will not be formed in chemical equilibrium. The titanium oxide Ti2O3 reaction 2TiO2(g)]H2 ]Ti2O3(s)]H2O We obtain similarly as above (8) P2\ (1]2eHe)2(eO[eC[eSi) 2(1[f )2c2eTi 2 Kp[Ti2O3(s)] where we now have assumed that the fraction f of Ti is condensed into solid Ti2O3 while the remaining fraction 1[f stays in the gas phase.The limit curve f\0 of the stability of solid Ti2O3 is shown in Fig. 2. It intersects the limit curve of stability of solid TiO2 just in the pressure region around P\10~4 encountered in the condensation zone of circumstellar dust shells. Thus both oxides of Ti would coexist in some temperature region in chemical equilibrium. (g)]H2 ]Ti3O5(s)]H2O The oxide Ti3O5 can be formed from the gas phase for instance by the reaction 3TiO2 and we obtain (9) P3\ (1]2eHe)3(eO]eC[eSi) 4(1[f )3c3eTi 3 Kp[Ti3O5(s)] 2O3 . where we again have assumed that the fraction f of Ti is condensed into solid Ti3O5 while the remaining fraction 1[f stays in the gas phase.The limit curve f\0 of stability of solid Ti3O5 is shown in Fig. 2. It is very close to the stability limits of TiO and Ti 2 The oxide Ti4O7 can be formed from the gas phase for instance by the reaction 4TiO2(g)]H2 ]Ti4O7(s)]H2O and we obtain (10) P4\ (1]2eHe)4(eO[eC[eSi) 8(1[f )4c4eTi 4 Kp[Ti4O7(s)] is shown in Fig. 2. where again it is assumed that the fraction f of Ti is condensed into solid Ti4O7 while the remaining fraction 1[f stays in the gas phase. The limit curve f\0 of stability of solid Ti4O7 In the pressure region below 10~2 dyn cm~2 in the condensation zone of circumstellar dust shells this stability limit is slightly above the stability limits of the other oxides ; hence Ti4O7 is the most stable titanium oxide in chemical equilibrium under conditions encountered in circumstellar dust shells and only this one would be formed in an equilibrium state.The diÜerences in the stability of the oxides TiO2 Ti2 O3 Ti3 O4 and Ti4O7 however are small. The composition diÜerences merely result from the replacement of some Ti4` cations by Ti3` cations. The most abundant titanium bearing molecule from the gas phase in this region is TiO and since in a real condensation process in a cooling gas the solid has to be formed from the TiO molecules we can expect 2 therefore that at the onset of condensation of titanium oxides small clusters and par- 2 309 H. P. Gail and E. Sedlmayr ticles of TiO are formed perhaps with some of the Ti cations being only double or triple instead of fourfold charged ions.This condensation mechanism is especially attrac- 2 tive since it does not require any complicated surface chemistry to operate in the formation of such clusters but simply the step by step addition of a molecular species which is fairly abundant in the gas phase. 2.1.2 Other titanium compounds. From calculations of condensation sequences (e.g. Lattimer et al.7 and Sharp and Huebner8) it is known that some other titanium compounds are stable at rather high temperatures. The titanium»calcium compound CaTiO (perovskite) for instance usually appears –rst in cooling sequences based on chemical equilibrium. Since the Ca does not form molecules of high bond energy it is 3 present in the gas phase as the free atom.The perovskite may then be formed from the gas phase for instance by the reaction TiO2]Ca]H2O]CaTiO3]H2 If the fraction f of the titanium is bound in perovskite and the remaining fraction is present as free TiO molecules we obtain 2 (11) (1]2e P2\ He)2 8(1[f )ceTi(eCa[feTi)(eO[eC[eSi)Kp[CaTiO3(s)] This de–nes a curve in the P»T plane where just the fraction f of the Ti is condensed into perovskite. The value f\0 de–nes the stability limit of the solid. Fig. 4 shows the limit curve f\0. Obviously perovskite is stable up to higher temperatures than the titanium oxides. which are also in the gas phase is present as the free atom the magnesium titanate Titanium forms with Mg the compounds MgTiO and MgTi 3 stable up to quite high temperatures.Since magnesium above the stability limit of for- 2O5 sterite (Mg SiO 2 5) shows for comparison the dissociation limit of H Fig. 4 Dissociation limits for Ti molecules and stability limits for the formation of solid Ti and for the formation of some solid Ti compounds from molecules from the gas phase. The dashed line 2 . 310 Inorganic dust formation in astrophysical environments MgTiO could be formed for instance in the reaction 3 Mg]TiO2]H2O]MgTiO3(s)]H2 If neither Ti nor Mg is bound in any other solid compound and if the fraction f of the Ti is condensed into MgTiO we obtain 3 (12) P2\ 23(e (1]2eHe)2 Mg[feTi)(1[f )ceTi(eO[eC[eSi)Kp(MgTiO3) could be formed from the gas phase by the reaction MgTi2O5 Mg]2TiO2]H2O]MgTi2O5(s)]H2 Analogously as above we obtain for the chemical equilibrium state (13) P3\ 24(e (1]2eHe)3 Mg[12 feTi)(1[f )2c2eTi 2 (eO[eC[eSi)Kp(MgTi2O5) Eqn.(12) and (13) de–ne a curve in the P»T plane along which just the fraction f of the Ti is bound in MgTiO or MgTi respectively. f\0 corresponds to the stability 3 limits of these solids which are shown in Fig. 4. Both magnesium titanates are less stable 2O5 than the titanium oxides and thus cannot be formed by direct condensation from the gas phase in chemical equilibrium. We have also considered the Ti metal and solid TiN and solid TiC as possible condensates. We do not present details but only show the stability limits of these compounds in Fig. 4. Obviously Ti does not condense into one of these solids in a circumstellar environment.principle the possibility exists that the same kind of mixed Ca»Ti-oxide cluster might be Since perovskite becomes stable at a somewhat higher temperature than TiO2 in the –rst nucleating species in stellar winds. In any case any formation mechanism for such particles necessarily requires some mechanism which adsorbs a Ca atom from the gas phase and at the same time by some surface reaction with water molecules catches an oxygen atom. This makes this process kinetically much less favourable than the simple step by step growth which is possible for the formation of TiO clusters. 2 2.2 The aluminiumñcalcium complex 2Si2O8). (cf. Fig. 3 of Sharp and Huebner8). The oxygen not bound in solid compounds is From the calculations of the equilibrium composition of solid/gas mixtures for oxygen rich element mixtures by Grossman9 and e.g.Saxena and Ericksson10 and Sharp and Huebner8 it is known that the following solid aluminium and aluminium»calcium compounds may exist in thermochemical equilibrium at temperatures of the order of 1000 K encountered in the condensation zone of circumstellar shells. These are (i) aluminium oxide or corundum (Al2O3) (ii) melilite a solid solution of gehlenite (Al2Ca2SiO7) and a- kermanite (MgSiCa SiO7) (iii) spinel (MgAl2O4) (iv) diopside (CaMgSi2O6) and (v) anorthite (CaAl 2 The most stable of these compounds is corundum due to its extremely high energy of formation. Molecules formed from Al and O on the other hand do not have a particular high bond energy.At temperatures above 1000 K the most abundant Al compounds in the gas phase are AlO and Al2O and at a somewhat higher temperature most of the Al is present as the free atom in the gas phase. At the low pressures relevant for circumstellar dust shells the corundum becomes the most stable aluminium compound at temperatures well below ca. 1400 K. The most abundant molecule in the gas phase then is Al2O 311 H. P. Gail and E. Sedlmayr bound in H2O CO and SiO. Calcium and magnesium are present as the free atoms. We consider the formation of aluminium and calcium compounds from this mixture (cf. also Gail11). 2.2.1 Corundum. Consider the formation of corundum from gas phase species by the reaction Al2O]2H2O]Al2O3(s)]2H2 If the fraction f of the aluminium is condensed in corundum then the partial pressure of partial pressure of H2O is Al2O is pAl2O\12 (1[f )eAlPH and if no other condensates than corundum exist then the librium the partial pressures of the gas phase species satisfy pH2O\(eO[eC[eSi[32 feAl)PH .In thermodynamic equi- (14) \exp([*ºGE/RT )\K (Al2O3) p p pH2 2 Al2O pH2O 2 We obtain (15) P\ 22(1[f )e 1]2eHe Al(eO[eC[eSi[32 feAl)2Kp(Al2O3) Fig. 5 shows the stability limit f\0 of corundum in the P»T plane. 2.2.2 Melilite. The most stable aluminium»calcium compound is gehlenite (Ca which forms a solid solution with a- kermanite (e.g. Grossman,9 Sharp and 2Al2SiO7) Huebner8). Since the a- ckermanite forms only a small admixture of at most a few percent at low pressures (cf.Fig. 15 in Saxena and Ericksson10) we neglect this component. The gehlenite becomes stable at temperatures where the aluminium in an equilibrium state is and 2 Fig. 5 Stability limit of corundum and stability limits for conversion between some compounds of aluminium with calcium and magnesium. The dashed lines show the dissociation limit of H the stability limit of the most refractory of the silicates forsterite (Mg SiO4). 2 312 Inorganic dust formation in astrophysical environments condensed into corundum. Then gehlenite can be formed only by conversion of corundum into gehlenite. A possible reaction is Al2O3(s)]2Ca]SiO]3H2O]Ca2Al2SiO7(s)]3H2 If f is the fraction of the Al condensed into corundum and g the fraction of the Ca condensed into gehlenite the partial pressures of the molecules in the gas phase are ( p eAl2O\12 [(1[f )eAl[geCa]PH pCa\(1[g)eCaPH pSiO\(eSi[12 geCa)PH and pH2O\ O[eC[eSi[32 feAl[62 geCa)PH .“ From the law of mass action we obtain (16) (1]2e P3\ He)3 26(1[g)2eCa 2 (eSi[12 geCa)(eO[eC[eSi[32 feAl[3geCa)3Kp(gehl) The fraction f of the aluminium condensed into corundum is given by eqn.(15) which reads in the present case as (17) P\ 22(e 1]2eHe Al[feAl[geCa)(eO[eC[eSi[32 feAl[3geCa)Kp(cor) Since the calcium abundance is less than that of aluminium the formation of gehlenite consumes always only part of the aluminium; the remaining fraction of Al remains as corundum. Eqn. (16) and (17) determine the degree of condensation of Al and Ca into corundum and gehlenite for given P and T .The limit curve g\0 of stability is shown in Fig. 5. 2.2.3 Spinel. With decreasing temperature magnesium spinel MgAl2O4 more stable than corundum. A possible conversion reaction of corundum into spinel is becomes Al2O3(s)]Mg]H2O]MgAl2O4(s)]H2 Assume that a fraction f of the aluminium is –rst condensed into corundum and that a fraction x of this is converted into spinel. At the same time part of the aluminium is bound in gehlenite. The partial pressures of the gas phase species are pMg\[eMg[12 (1 [x) fe H2O\MeO[eC[eSi[[32 (1[x)]2x] feAl[3geCaNPH . From the law of mass action for the partial pressures of the gas phase species we obtain Al]PH and p P\ 4Me (1]2e Mg[[(1[x)/2] feAlNMeO[eC[eSi[[(3]x)/2] feAl[3geCaNKp(spin) (18) He) This equation and eqn.(15) for corundum which reads in the present case as (19) P2\ 25(1[f )2e (1]2eHe)2 Al 2 MeO[eC[eSi[[(3]x)/2] feAl[3geCaN3Kp(cor) and eqn. (16) for the equilibrium between corundum and gehlenite which is (20) (1]2e P3\ He)3 26(1[g)2eCa 2 (eSi[12 geCa)(eO[eC[eSi[32 feAl[3geCa)3Kp(gehl) for –xed P form a system of three equations for the four unknown quantities T f g and x. They determine thus for –xed f or x a family of curves of constant f or x in the P»T plane. The upper limit curve x\0 of stability for spinel and the lower limit f\0 of “ 6 The factor takes into account that one of the oxygen atoms in gehlenite is already counted by subtracting the full Si abundance.2 313 H. P. Gail and E. Sedlmayr existence of corundum diÜer only by ca. 2 K (for –xed P). The transition from corundum to spinel thus occurs nearly discontinuously in chemical equilibrium. The limit curve f\0 for the disappearance of corundum is shown in Fig. 5. A possible reaction for the 2O6). tends to be bound more stably in diopside (CaMgSi 2.2.4 Diopside. The calcium bound in gehlenite (Al2Ca2SiO7) at lower temperatures conversion of gehlenite into diopside is SiO\[eSi[2f[12 (1[f )eCa]PH p 2O In chemical equilibrium the gas phase species satisfy pH2O\[eO are and Al2Ca2SiO7(s)]3Mg]3SiO]6H2O]2CaMgSi2O6(s)]MgAl2O4]6H2 If f denotes the fraction of the calcium bound in diopside and if the only other abundant calcium compound is gehlenite then the partial pressure of magnesium atoms in the gas phase is pMg\MeMg[feCa[12 [eAl[(1[f )eCa]NPH since the calcium is either bound in diopside or gehlenite and the aluminium is bound either in gehlenite or spinel.The partial pressures of SiO and H [e the law of mass action and from this we obtain C[eSi[(1]3f)eCa[2eAl]PH . P6\ 212Me (1]2eHe)6 Mg[[(3f[1)/2]eCa[12 eAl)3 1 (21) ]MeSi[[(3f]1)/2]eCaN3[eO[eC[eSi[(1]3f )eCa[2eAl]6Kp(diops) This de–nes a curve in the P»T plane along which the fraction f of the calcium is bound in diopside while the remaining fraction of the Ca is bound in gehlenite. At the same time the Al not bound in gehlenite is bound in spinel. f\0 de–nes the upper stability limit of diopside above and to the left of which no diopside exists while f\1 de–nes the lower stability limit of gehlenite below and to the right of which no gehlenite exists.p The transition between the two extreme cases occurs for –xed P within a temperature interval of only a few degrees where both solids coexist.The transition between the two Ca bearing compounds occurs nearly discontinuously at a sharp transition temperature. The limit for conversion of gehlenite into diopside is shown in Fig. 5 which also shows the limit curve of stability of forsterite (Mg SiO4). The stability limit of forsterite 2 is calculated as in Gail and Sedlmay.4 From Fig. 5 one easily recognises that the limit occurs at nearly the same temperature but always slightly above that where forsterite starts forming with decreasing temperature.This justi–es our previous assumption that most of the Mg is present as free atoms in the gas phase. 2.2.5 Anorthite. The aluminium bound in spinel tends to form at lower temperatures the more stable aluminium»calcium compound anorthite (CaAl2Si2O8) by conversion of spinel and part of the diopside (cf. Grossman9 and Lattimer et al.7). A possible reaction for the formation of anorthite is MgAl2O4(s)]CaMgSi2O6(s)]H2O]SiO]CaAl2Si2O8(s)]Mg2SiO4(s)]H2 We do not present the details of our calculation but only show in Fig. 5 the stability of anorthite. In accordance with the –ndings of Sharp and Huebner8 that anorthite does not form in their calculation at temperatures above ca. 1000 K the conversion of spinel into anorthite occurs only at rather low temperatures well below the stability limit of forsterite.This makes it rather unlikely that anorthite can be formed in circumstellar p The upper stability limit of gehlenite is determined by eqn. 16. 314 Inorganic dust formation in astrophysical environments shells since if aluminium grains are formed in the out—ow they will serve as growth centres for silicate condensation once the silicates become stable. This will shield the aluminium grain from surface reactions of the type shown above required in the conversion to anorthite. 3 Formation of seed nuclei The discussion in Gail and Sedlmay4 has shown that the formation of seed nuclei for subsequent growth of dust grains is not possible by nucleation of abundant gas phase species bearing the abundant dust forming elements Si Fe and Mg.The onset of dust formation would occur in this case at a signi–cantly lower temperature than that which is derived from the IR emission from circumstellar dust shells. In this contribution we explore the possibility that dust formation is triggered by cluster formation from molecules bearing elements of lower abundance than Si Fe or Mg. Fig. 1 (cf. also Gail and Sedlmay4) suggests that there are two especially favourable candidates (i) aluminium compounds because of their high stability and since Al has an only slightly lower abundance than Si Fe and Mg and (ii) titanium compounds because of the exceptional high bond energy of titanium oxide molecules and of solid titanium compounds which favours condensation at high temperatures despite a low Ti abundance.3.1 Nucleation of titanium oxide In view of the discussion in Section 2.1 the stability limits of titanium oxides occur at pressure»temperature conditions where the titanium contained in the gas phase is nearly completely bound in the TiO molecule which may condense into a solid of the same stoichiometric composition. Though solid TiO is not the most stable solid titanium 2 2 condensate it is only marginably less stable than Ti4O7 the most stable of the oxides. The kinetically most simple process then would be homogeneous nucleation of TiO by the sequence of reactions 2 TiO »»»’ (TiO `TiO2 2 `TiO2 `TiO2 2)2 »»»’ (TiO2)3 »»»’ (TiO2)4 »»»’ … … … `TiO2 where a stepwise addition of TiO leads to formation of clusters of increasing size which may serve later as centres for heterogeneous growth by addition of diÜerent gas phase 2 species.Here we concentrate on this simple mechanism and explore whether this process possibly triggers the observed dust formation around M stars. The titanium compounds have a strong ionic bonding character (e.g. Pauling12 and Evans13). The structure and bond energies of such compounds can be successfully modelled by considering the electrostatic forces between ions and induced dipoles plus an empirical repulsion potential for instance of the Born»Mayer type. This is well known from solid state physics and has recently been applied to model MgO clusters by Koé hler et al.14 Since results of such calculations for bond energies and cluster structures for TiO clusters are not yet available we proceed in the following simpli–ed way (1) Since 2 the electronegativity diÜerence of 2.0 between Ti and O (Pauling12) nearly equals the electronegativity diÜerence of 2.1 between Na and Cl the bond properties of TiO clusters are determined by Coulomb forces between ions as for NaCl.We assume then that 2 the dependence of the total binding energy Eb(N) of a cluster formed from N TiO2 molecules on the cluster size N is similar to that of NaCl clusters.** The size dependence of the bond energy of one molecule in the cluster for NaCl is taken from Martin.15 It ** We do not scale the bond energy of TiO clusters with our previous results for MgO clusters (Koé hler et 2 al.13) since the bonding in MgO has a signi–cant covalent contribution for small clusters.315 H. P. Gail and E. Sedlmayr can well be approximated by the following interpolation formula (22) b(N)\E= N]0.3 E N2 where E= is the bond energy per NaCl molecule in solid NaCl. We use eqn. (22) to clusters of size N from the bond energy of E calculate the bond energy of TiO2 =\6.62 eV per TiO2 approximated by that of a homogeneous sphere of volume V molecule in the solid. (2) The moments of inertia I of the clusters are (23) I\5 2 MA3 4 V nB2@3 and V \NV0 . The volume V of the monomer TiO in the solid is 0 2 m\NmTiO2 with calculated as (24) V0\ m o TiO2 D oD\4.32 g cm~3 is the bulk density of solid TiO2 . (3) The vibrational contributions to the partition function are estimated following the approach in John.16 The Debye temperature of TiO is HD\750 K (Landolt»Boé rnstein17).The corresponding Debye wave- 2 kH number of the solid lattice vibrations is D/(hc)\521 cm~1. The vibrational wavenumbers of the TiO molecule are l\207 935 and 962 cm~1 (Chase et al.18). These were converted to a ìDebye frequencyœ of the molecule according to the method 2 described by Mu l é ller et al.19 The frequency of the ith mode of a cluster with f vibra- i tional modes and a Debye wavenumber l can be approximated by (Mué ller et al.19) (25) li\ADi [ f 12 B1@3lD From this we calculate an average l for TiO from the vibrational wavenumbers. We D lD\1050 cm~1. The Debye wavenumber choose for our further calculations a value of 2 of a cluster of size N is determined by interpolation between the extreme values of the solid and the molecule by (26) lD(N)\lD solid] lD molecule N1@3 [lD solid l With this value of D(N) and the approximate frequencies calculated from eqn.(25) the vibrational contribution to the entropy and the internal energy of the clusters is calculated. These data are used to calculate the thermodynamic functions for TiO clusters as is 2 described for the case of MgO clusters by Koé hler et al.14 The thermodynamic functions are then used to calculate the nucleation rate J by the method explained in detail by Koé hler et al.14 The sticking coefficient in the calculation of J is chosen as a\1 since * * already the dimer Ti2O4 contains six atoms which makes a radiative stabilisation of the collision complex very likely.* /NH The result is shown in Fig. 6 where lines of constant rate of formation of seed nuclei per hydrogen nucleus i.e. J are plotted in the P»T plane. For comparison the –gure shows the P»T loci of the sonic point in a stationary stellar wind for four diÜerent values of the mass loss rate M0 between the lowest and highest observed values. The sharp turn oÜ and nearly vertical run of the lines of constant nucleation rate in the P»T plane occur when the critical cluster size becomes the monomer. The nucleation rate is then determined by the rate of TiO collisions. 2 316 Inorganic dust formation in astrophysical environments 2 J Fig. 6 Nucleation rate per hydrogen nucleus for TiO clusters from * /NH\10~19 to J * /NH\ 2 10~16 seed nuclei per hydrogen nucleus in steps of a factor 10 (full line).The dotted lines show the pressure and temperature at the sonic point of a stationary wind for mass loss rates of M0 \10~4 M yr~1. The dashed lines show the stability limit of forsterite (Mg SiO4) (upper) and where one _ half of the silicon is condensed into forsterite (lower). As we clearly see the onset of nucleation occurs between ca. 1050 K for stellar winds with a low mass loss rate of M0 [10~7 M_ yr~1 at the lower end of the scale of observed mass loss rates and nearly 1200 K for winds with the highest observed mass loss rates of M0 Z10~4 M_ yr~1. The formation of TiO clusters occurs at a temperature ca.30 K higher than the stability limit of forsterite which is the most stable 2 one of all silicon»magnesium»iron compounds in a circumstellar shell. Clusters of titanium oxide thus can serve as seed nuclei on which the silicate compounds condense once they become stable in the out—owing and cooling wind. Such a two step dust formation process [(1) formation of seed nuclei bearing some less abundant elements at a temperature somewhat higher than the stability limit of silicates which do not emit observable quantities of IR radiation and (2) massive formation of condensed silicates at their stability limit by growth on a diÜerent kind of seed nuclei] is just what is required to match the observation of the onset of massive IR emission from silicate dust at a temperature of roughly 1000 K at the inner edge of the zone of formation of observable dust.Titanium oxide clusters thus might well be the seed nuclei for circumstellar dust formation. 3.2 Nucleation of aluminium oxide Having the highest sublimation temperature among all the minerals in question corundum is naturally expected to emerge as the –rst condensate in the cooling out—ows of oxygen rich giants and supergiants.20 In order to determine the gas phase abundance of Al2O3 the molecular properties of Al2O3 (bond energy vibrational frequencies moments of inertia) have been calculated by quantum mechanical ab initio calculations (MP II) and the thermodynamic functions have been derived.2 From this we calculated the gas phase abundance of Al2O3 .Fig. 7 shows for a solar element mixture contour lines of constant supersaturation ratio S of Al vapour pressure over a —at corundum 2O3 317 H. P. Gail and E. Sedlmayr K. The out—ow velocity is v\1 km s~1 corresponding to the inner edge of the Fig. 7 Contour lines of constant supersaturation ratio S of Al2O3 lines represent typical pressure and temperature model structures of a stationary out—ow for three in the P»T plane. The dotted diÜerent mass loss rates M0 \10~7 M 10~6 10~5 yr~1. The stellar parameters are L *\104 _ T *\2500 dust shell. The crosses indicate the radial position r/R L _ in the wind. * surface is a pressure»temperature plane. One observes a steep increase of the supersaturation ratio S within a relatively narrow temperature interval already at rather high temperatures which suggests an early condensation of corundum from the gas in the out—ow.The chemical equilibrium calculations for the interesting temperature range yield however almost vanishing local concentrations of the Al molecules (cf. Fig. 8). The strong increase of the supersaturation ratio therefore is solely caused by the 2O3 extraordinary stability of corundum i.e. by the extremely small vapour pressure and thus does not imply a correspondingly high nucleation rate of Al2O3 monomers. For 2O3 Fig. 8 Chemical equilibrium abundances of some gaseous Al/O species in a typical stationary out—ow with solar element abundances. The dashed line shows the temperature decline of the wind model (cf. Fig. 7). The thermodynamic functions are taken from the JANAF thermochemical tables (Chase et al.18) except for the data of Al which are obtained from ab initio quantum mechanical calculations (cf.Chang et al.21). 318 Inorganic dust formation in astrophysical environments this reason homogeneous nucleation of molecular Al2O3 to form corundum out of the gas phase does not seem likely to constitute a favourable process for producing the primary condensate in a stellar out—ow. 2O Al2O2 H2O). etc. and other oxygen bearing molecules (such as OH or Inspection Despite this corundum clusters nevertheless could in principle be formed by some heteromolecular process involving abundant aluminium bearing oxides such as AlOH Al of Fig. 8 however shows that the aluminium oxides and hydroxides are abundant only at temperatures below ca.1150 K but at this temperature TiO seed grains are already present. Consequently aluminium compounds should certainly play a major role in 2 astrophysical grain growth as expected but might be ruled out as the –rst condensates in circumstellar out—ows. 4 Concluding remarks We have considered the problem that nucleation process might be responsible for the onset of dust formation in circumstellar shells around stars with a standard (i.e. oxygen rich) cosmic element composition and a high mass loss rate. At low mass loss rates the dust formation can be initiated by SiO nucleation4,22 but at high mass loss rates this is ruled out by much too low a condensation temperature. Having also ruled out nucleation by molecular species of the abundant elements Si Mg and Fe from the gas phase as primary condensates (cf.Gail and Sedlmayr4) we consider possible nucleation processes involving gas phase species of less abundant elements. The homogeneous nucleation of TiO turns out to be a process that –ts the requirement of a condensation temperature of 1000 K or slightly above as it is required by 2 observations of IR emission from circumstellar dust shells. It also oÜers an explanation for the observation that obviously two diÜerent nucleation mechanisms operate in stellar winds of low and high mass loss rate respectively because of quite diÜerent temperatures for the onset of dust formation in both cases the rather low abundance of Ti makes nucleation by TiO inefficient at low densities in the stellar wind but nucleation of SiO remains possible down to much lower densities.Thus we have formation of 2 seed nuclei by TiO at T Z1000 K for high densities and formation of seed nuclei by 2 SiO at T B500 K for low densities. The present calculation of the nucleation rate is based on estimates of the basic cluster properties the accuracy of which cannot easily be judged. A –nal conclusion as to whether TiO clusters really are (one of) the primary condensates in circumstellar shells of M stars requires more accurate calculations of bond energies vibrational frequencies 2 and moments of inertia. Such calculations are presently performed at our institutes by ab initio quantum mechanical calculations for small clusters and by using semi-empirical potentials such as those described by Koé hler et al.14 for large clusters.The homogeneous nucleation of Al2O3 in a stellar wind is not possible. The abundance of the Al2O3 molecule in the gas phase is much too small. Condensation of aluminium by a heterogeneous growth process involving more abundant Al bearing gas phase species cannot be ruled out at present since the required data for clusters built from more than two Al atoms are not yet available. These cannot be estimated by a simple procedure as in the case of titanium since the aluminium oxygen bond energy is rather weak at the molecular level despite the strong bonding of the solid. This shows that Al»O clusters are not pure ion clusters. For Al»O clusters with more than two Al atoms presently quantum mechanical ab initio calculations are performed at our institutes in order to generate the data for calculating the rates of more complicated nucleation paths for aluminium oxide clusters.In any case even if it turns out that aluminium oxide is not the primary condensate as it seems likely to be the case at present the aluminium can condense in circumstellar shells as a dust species of its own because of the temperature gap between the formation 319 H. P. Gail and E. Sedlmayr of TiO seed nuclei and the growth of silicate dust on these particles. Since solid aluminium becomes stable at a temperature approximately 200 K above the condensation 2 temperature of silicates aluminium oxide can condense onto the seed nuclei prior to silicate formation.In fact from analysis of abundance ratios for oxygen and aluminium isotopes in meteoritic material it is known that corundum (Al2O3) grains of de–nitely circumstellar origin have been present in the protoplanetary disk and spinel (MgAl2O4) material (e.g. Hutcheon et al.,23 Huss et al.24 and Nittler et al.25) The results presented in this contribution give for the –rst time a de–nite indication of how dust formation in circumstellar shells of M stars may really operate it is most likely that dust formation starts with formation of TiO clusters which serve as growth centers both for aluminium oxides and silicates. 2 We acknowledge the assistance of M. John Kyung Sook Jeong and A. B. C. Patzer in part of the calculations and in the preparation of some of the –gures.Paper 7/09290C; Received 24th December 1997 References 1 H-P. Gail and E. Sedlmayr in Physical Processes in Interstellar Clouds ed. G. E. Mor–ll and M. Scholer Reidel Dordrecht 1987 p. 275. 2 E. Sedlmayr in IAU Colloquium 146 Molecules in the Stellar Environment L ecture Notes in Physics ed. U. G. Joé rgensen Springer Berlin 1994 vol. 428 p. 163. 3 E. Sedlmayr and C. Dominik Space Science Reviews 1995 73 211. 4 H-P. Gail and E. Sedlmayr in T he Molecular Astrophysics of Stars and Galaxies»A V olume Honouring Alexander Dalgarno ed. T. W. Hartquist and D. A. Williams Oxford University Press London in press. 5 R. D. Lide CRC Handbook of Chemistry and Physics 76th edn. CRC Press Boca Raton 1995. 6 E. Anders and N. Grevesse Geochim.Cosmochim. Acta 1989 53 197. 7 J. M. Lattimer D. N. Schramm and L. Grossmann Astrophys. J. 1978 219 230. 8 C. M. Sharp and W. F. Hué bner Astrophysical J. Suppl. 1990 72 417. 9 L. Grossman Geochim. Cosmochim. Acta 1972 36 597. 10 S. K. Saxena and G. Ericksson in Chemistry and Physics of T errestrial Planets ed. S. K. Saxena Springer New York 1986 p. 30. 11 H-P. Gail Astron. Astrophys. 1998 332 1099. 12 L. Pauling T he Nature of the Chemical Bond and the Structure of Molecules and Crystals Cornell University Press 1960. 13 R. C. Evans An Introduction to Crystal Chemistry Cambridge University Press Cambridge 2nd edn. 1966. 14 T. M. Koé hler H-P. Gail and E. Sedlmayr Astron. Astroph. 1996 320 553. 15 T. P. Martin Phys. Rep. 1983 95 167. 16 M. John Bildung von Eisen-Clustern in kué hlen Sternwinden Diploma thesis Technical University Berlin 17 L andolt»Boé rnstein Numerical data and functional relationships in science and technology New Series ed. 18 M. Chase C. Davies J. Downey D. Frurip R. MacDonald and A. Syverud ìJANAF Thermochemical 1995. K-H. Hellwege and O. Madelung Springer Berlin vol. 17g 1984. Tables 3rd edn.œ J. Phys. Chem. Ref. Data 1985 vol. 14 suppl. no. 1. 19 H. Mué ller H-G. Frische and L. Skala in Clusters of Atoms and Molecules ed. H. Haberland Springer Berlin 1995 p. 114. 20 E. E. Salpeter Ann. Rev. Astron. Astrophys. 1977 15 267. 21 Ch. Chang A. B. C. Patzer E. Sedlmayr and D. Sué zle Z. Phys. D 1998 submitted. 22 J. A. Nuth and B. Donn J. Chem. Phys. 1982 77(5) 2639. 23 I. D. Hutcheon G. R. Huss A. J. Fahey and G. J. Wasserburg Astrophysical J. 1994 425 L97. 24 G. R. Huss A. J. Fahey R. Gallino and G. J. Wasserburg Astrophysical J. 1994 430 L81. 25 L. R. Nittler C. M. Alexander X. Gao R. M. Walker and E. K. Zinner Nature (L ondon) 1994 370 443.

ISSN:1359-6640    

DOI:10.1039/a709290c

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Faraday Discuss. 1998 109 321»334 Opacity of TiO from a coupled electronic state calculation parametrized by ab initio and experimental data David W. Schwenke M.S. 230-3 NASA Ames Research Center MoÜett Field CA 94035-1000 USA We have carried out calculations of ro-vibrational energy levels of the 13 lowest electronic states of TiO. The dominant couplings between the various states are included with the coupling parameters and potential parameters optimized to match experimental energy levels. We have also performed ab initio electronic structure calculations of the spin»orbit and rotation»orbit couplings to verify that the physical results are obtained from the optimization. Our wavefunctions were used to predict the intensities of both well characterized and unobserved forbidden bands.1 Introduction The electronic spectroscopy of the TiO molecule is very important for the study of many types of stars and consequently there have been a large number of experimental studies of the various bands. The electronic states of importance to the present work are given in Fig. 1 and the bands of interest are the a band (C^X) the b band (c^a) the c band (A^X) the c@ band (B^X) the d band (b^a) the z band (b^d) the f^a band and the e band (E^X). Recently J‘rgensen1 has reviewed the literature and produced a line list for use in astrophysical simulations. Since then there have been a number of new experimental studies2h6 of line positions and a theoretical determination7 of the transition moments. These data should allow production of a much more accurate line list.An improvement on the work of J‘rgensen,1 which includes some of the new data is the new calculations of Plez.8 In these calculations to compute intensities the transition moments of LanghoÜ7 are used in conjunction with numerically determined rovibrational wavefunctions using potential-energy curves determined from experimental spectroscopic parameters. The line positions were obtained from the spectroscopic parameters. Using this line list Plez8 found quite good agreement between experiment and his simulations of cool stars for most bands. In the present work we include couplings between the various electronic states. The only previous work in this area has been the experimental determination of the c»C spin»orbit coupling;4 these two states strongly perturb each other.9 The energy levels of all other states can be accurately reproduced by spectroscopic parameters without explicitly invoking coupling however there is coupling among the various electronic states.The most obvious manifestation of the coupling is strong perturbations as are present due to cC coupling and the lambda-doubling of the % states. Other more subtle indications are the anomalous spin»orbit splittings of the triplets and the distortion of potential parameters from their physical meanings; for example in their analysis of the b 1% state Galehouse et al.10 give separate values of B for each parity. In addition there can be transitions between normally forbidden states. Another potential advantage e 321 322 Opacity of TiO from coupled electronic state calculation Fig.1 Electronic states of TiO considered in the present work. Singlets are a b c d f g and h and triplets are X A B C D and E. The states g and h are named in this work. of a calculation coupling the electronic states is that part of the variation with vibrational quantum number is essentially due to the change in Franck»Condon factor so it may be possible to represent the Hamiltonian accurately with fewer parameters or to extrapolate more reliably to uncharacterized levels. Since most states can be represented without couplings it is clear that introducing empirical couplings introduces a certain amount of ambiguity. To reduce this we have attempted to compute the dominant coupling terms from ab initio electronic structure calculations.We have found however that we can perform reliable calculations only for the low-lying levels. We only consider the bound states given in Fig. 1. An extreme complication for the electronic structure calculations is that there are repulsive states crossing the bound states and in general we should consider couplings to them as well however the Franck»Condon factors will be very small so we will neglect these in our calculations. Our calculations will not yield perfect agreement with experiment for several reasons. One is the inclusion of only a small number of electronic states. Another is errors in the electronic structure calculations including the neglect of smaller relativistic corrections to the Hamiltonian.Finally there can be errors in the parameters determining the potential-energy curves. These are mostly obtained from experimental analysis. We attempt to correct for all these de–ciencies in our calculations by optimizing the parameters in our Hamiltonian. Our long term goal is to reproduce all experimental data accurately while using physically reasonable couplings. In this paper we report on the current status of our calculations and demonstrate the unique features of the spectrum we have obtained. 323 D. W . Schwenke 2 Electronic structure calculations In the present work we use the same one-electronic basis sets as used by LanghoÜ.7 State-averaged complete active space self-consistent –eld (SA-CASSCF) calculations were carried out to determine molecular orbitals using the MOLPRO program.11 The (4331) and (5331) active spaces of LanghoÜ were used.Matrix elements of the operators L were obtained in the course of these calculations. These matrix elements determine x the rotation»orbit interaction. The spin»orbit interaction was computed using the SIRIUS/ABACUS12 program package with the same molecular orbitals and CASSCF space as the MOLPRO calculations. For the present work the couplings were assumed to be independent of the bond length and were evaluated at a bond length of 3.2 a0 which is near the minima of the various potential-energy curves. There are two difficulties with the present calculations. The –rst is the determination of the phase of the coupling matrix elements.There is no guarantee that the wavefunction phases will be the same between MOLPRO and SIRIUS/ABACUS or be the same as the bond length changes or if the CAS space or the states averaged changed. We tried to eliminate this problem by performing approximate calculations of the matrix elements assuming that there were two electrons and the wavefunction could be written as a single determinant in terms of a single s p and d function. The one-electron matrix elements were parametrized to reproduce the results of the ab initio calculations approximately. This worked well except when con–guration mixing produced –nite matrix elements while the approximate calculations produced zero. These approximate electronic wavefunctions were also used to derive the rotation»electronic Hamiltonian matrix.A second difficulty with the calculations is that it was not feasible to use a large enough CAS space so that all the states could be simultaneously well represented in a single SA-CASSCF calculation. Thus in our calculations the couplings were determined by averaging over the smallest possible number of states. This means that the couplings of the higher states are much less accurate than the lower levels because more roots are required. For couplings between states of the same spin we used the option in MOLPRO to converge on states of given K. However this option only works when a CSF basis is used and MOLPRO can state average over singlets and triplets only when a determinant basis is used. This causes most difficulties with the % and ' symmetries.In particular the relative weight of the rp and dp con–gurations in the b 1% state was very sensitive to the states averaged. Thus evaluating accurate Bb coupling is not possible. This also indicates that part of the diÜerence in opacity between theory and experiment for the d band (b^a) and z band (b^d) observed by Plez8 may be due to inaccuracies in the theoretical transition moment. 3 Ro-vibrational calculations We parametrized the potential-energy curves by spectroscopic constants. An RKR procedure was then used to generate the curves used in the calculations. We generated RKR potentials by representing G(v) and B(v) for non-integral v using the spectroscopic con- Te ue uexe u Be ae eye and stants c and and integrals we use are given by eqn.(4)»(5) of ref. 13. These were evaluated using the e respectively. The forms of the RKR trapezoidal rule. A departure from the usual procedure is that we evaluated the turning points for non-integral v speci–cally every 0.1. This greatly reduces the error introduced in interpolating the potential. The potential minimum was computed from Be . Typically we go up to v\35 interpolate using a spline and extrapolate by assuming a linear form. Since we are interested in states with v\35 this introduces negligible error. For the non-linear least squares we have analytically evaluated the derivatives of the RKR potentials with respect to the spectroscopic parameters. In our calculations we used the masses 29156.94642 m for O and 87403.76317 m for 48Ti where m is the electron mass.e e e We convert energy units using the 1 cm~1\4.556335]10~6 Eh . 324 Opacity of TiO from coupled electronic state calculation The initial values of the spectroscopic constants were taken either from the literature or when reliable values were not available estimates were obtained from ab initio electronic structure calculations. These spectroscopic constants were then re–ned in the course of a non-linear weighted least-squares –t to experimental data. In our –ts we used energy levels deduced from experiment. We included all levels we could obtain with JO50 and to speed up the calculations we only used even J. The energy levels were determined as follows From the work of Gustavsson et al.,14 we have energy levels for the X state for v\0»5 and from the work of Hocking et al.15 and Amiot et al.,2 we have energy levels for the B state for v\0 1.To obtain the other energy levels we combined the X state data with data from the E^X (e) band,16 the C^X a transitions,3 the b^a (z) bands,10 the b^a (d) bands,5 the f^a bands,17 the c^a bands,18 and the A^X (c) and C^X (a) bands.19 We then repeatedly passed through the data and solved for all energy levels possible. Because we included all available data there are numerous estimates of most energy levels. We took these estimates computed the average and root-mean-square (RMS) deviation from the average and if the RMS deviation was less than 0.1 cm~1 the average was taken to be the energy of that level. If the RMS deviation is larger than 0.1 cm~1 we assumed that there had been a typographic error misassignment or experimental error associated with that level and we did not use that level in subsequent calculations.We also require at least two determinations in order to keep a level. By this means we obtained data for the a state v\0»2 the d state v\0»5 the b state v\0»4 the c state v\0»2 the f state v\0»2 the A state v\0»5 and the C state v\0»7. There are also data for the c^a transition for vA\v@\3,18 but there are not sufficient other data to give more than the energy diÜerence between these levels. Thus we did not include these data in our –t. Since the diÜerent states have diÜering amounts of data and the diÜerent levels have diÜerent RMS deviations we gave each energy a non-unit weight.Within each electronic state the weight is 1/max (0.01 cm~1 RMS deviation). We then renormalized the weights so that the sum of the squares for each state was the same. The calculations were performed using a Hundœs case (a) basis set. Thus the coupled electronic states are speci–ed by n S K and X with n diÜerentiating states with the same S K and X where S is the total electron spin quantum number K is the absolute value of the projection of the total electronic orbital angular momentum onto the internuclear axis and X is the absolute value of R]K where R is the projection of the total electron spin angular momentum onto the internuclear axis. The coupled equations are parametrized by J the total angular momentum quantum number and P which speci–es the & parity.The states only occur when P([1)J\]1. The potential energy and vibra- 0 B tional kinetic energy are diagonal in n S K and X the centrifugal potential is diagonal in n S and K and is given by 2kR2 MdXX{[J(J]1)]S(S]1)]2KX[K2[2X2] 1 [dXX{B1[S(S]1)[(X[K)(X@[K@)]1@2[J(J]1)[XX@]1@2N (1) the rotation»orbit interaction is diagonal in S and is given by dKK{B1MdXX{[2(1]dX0)]1@2[[J(J]1)[XX@]1@2 Sn o L x o n@T 2kR2 ][dXX{B1(1]dKX dK{X{ dmin(X X{)0)1@2]P([1)JdXX{Y1 dmin(X X{)0]N (2) with Sn o L x o n@T an electronic matrix element. The spin»orbit interaction is diagonal in X and is given by 325 D. W . Schwenke (3) An dnn{ dSS{ dKK{(X[K)]Hnn{[1]dX0 dmin(K K{)0]1@2 with A and Hnn@ spin»orbit matrix elements. We also occasionally included a phenom- n enological spin»spin coupling parameter and this is diagonal in all electronic quantum numbers and is expressed as (4) c(X[K)2 To compute line strengths we require radial integrals over the vibrational wavefunctions and transition moment and angular integrals over the remaining coordinates.The angular integrals are diagonal in S and non-zero only if PP1\[1. In the present work 1 X J@ @B ( we [1) took X[(2 angular J]1)(2J integrals @]1)]1@2 toGd be X~KX{~K{A J [X ] (1]d 1 K0 ] dX d 0)1@2(1]dK{0 dX{0)1@2[dX~KK X {~ [ X X {A@ X J [X 1 [X@ X J@ @BP([1)S`JH (5) X0 dX{0 dK0 dK{0 The vibrational basis functions used were determined by performing a –nite diÜerence boundary value method calculation for each electronic state.101.04 104.81 0.0 0.0 97.33 199.55 101.30 105.08 0.1 0.0 97.42 199.54 101.30 105.08 0.0 0.0 97.61 199.54 101.30 108.08 0.0 0.0 97.43 199.55 101.30 0.0 0.0 0.0 100.86 199.55 4 Energy level –tting results In this section we discuss our optimizations of the parameters for the various states. We couple groups of states together and we discuss each group separately. 4.1 X3D a 1D d 1Rë and E3P states We –rst consider the lowest lying states. These all have electronic structures dominated by two open shell electrons in a sigma orbital and a d (X and a) p (E) or r (d) orbital either triplet or singlet coupled.These states have weak coupling to the higher states because of both their electronic character and the relatively large energy separations. We now explore in depth the manifestations of couplings between these states. The X3* state is well represented by Hundœs case (a) basis functions for low J. Its diÜerent spin components are split by ca. 100 cm~1 but the central (X\2) level is ca. 3 cm~1 lower than expected. This was termed ìanomalousœ spin»orbit splitting when –rst observed. This was attributed by Kovaç cs20 to spin»orbit coupling to the a 1* state which lowers the X\2 level and spin»spin interaction which raises the X\1 and 3 levels. Empirically it is not possible to distinguish the two eÜects. In Table 1 we show splittings computed using various values of the spin»orbit interaction.The –rst two columns show the results with and without Xa coupling using the ab initio matrix elements and the next two columns show the same results using couplings optimized so as to reproduce experiment (i.e. the fourth column gives the experimental results of Gustavsson et al.14). The ab initio results have the right trends but the matrix elements are Table 1 X3* splittings for v\0 J\10 HXX so HXa so HXX ss 112.5 116.7 0.0 0.0 107.98 221.82 112.5 0.0 0.0 0.0 111.99 221.82 E(X\1)[E(X\1) E(X\2)[E(X\1) E(X\3)[E(X\1) Energies in cm~1. 326 Opacity of TiO from coupled electronic state calculation too large. Now the spin»spin interaction does not alter the diÜerence between the X\3 and X\1 levels so if we assume that the spin»orbit matrix elements are oÜ by the same factor we can estimate the size of the spin»spin interaction.The factor is 0.9004 and the results obtained by just scaling are given in column 5 and the results including a spin» spin interaction of 0.1 cm~1 are given in column 6. This reproduces column 4 very well. However the a 1* state is not the only state that can spin»orbit couple to the X3* state. The most strongly coupled triplet state is E 3% and since it too is well represented by Hundœs case (a) basis functions for low J it will shift the X\1 and X\2 levels down. We have carried out calculations coupling the X a and E states together using the ab initio coupling between the E and X and E and a states and scaling the X state spin»orbit splitting and Xa spin»orbit coupling by the same factor (0.8981) to reproduce the diÜerence between the X\3 and X\1 levels.This then requires a spin»spin coupling parameter of [0.05 cm~1 to match experiment. Now the determination of the spin»spin coupling parameter is very approximate. The assumption of a uniform scaling parameter is not necessarily realistic and there are other small eÜects neglected.21 What we can say with reasonable certainty is that the majority of the shift in the energy levels is due to spin»orbit coupling and one probably makes negligible error by adjusting the spin»orbit matrix elements to make up for the missing couplings. This is the strategy we are currently pursuing. Thus by coupling the X a and E states together we can well represent the X and a levels but we still need to account for the K-doubling of the E state.This is due to coupling to a & state and the fact that the coupling is essentially constant as a function of J strongly suggests that the perturber be a singlet since a triplet would have coupling that increases with J. Thus we assumed that the d state caused the K-doubling and we simultaneously optimized the parameters for all four states. The –nal parameters are given in Table 2. In this –t the unweighted RMS errors were 0.014 cm~1 for the X state 0.015 cm~1 for the a state 0.024 cm~1 for the E state and 0.007 cm~1 for the d state. u and ue xe for the E state. The data of e\924.2^4.5 and ue xe\5.1^1.0) but we were not able to obtain u and ue xe chosen It is worth commenting on the values of e Simard and Hackett are only for v\0 and limited J (we use energies only up to J\12) so u and ue xe are not well determined.Approximate values were obtained by Linton e and Broida,22 (u good –ts using their values. Thus we repeatedly performed –ts with from normal distributions having the mean and standard deviations of the results of e Linton and Broida. After a dozen or so trials we obtained the results given in Table 2. The values we obtain are well within the error bars given by Linton and Broida. Table 2 Optimized parameters for the X a E and d states d 1&` E3% a 1* X3* 11 868.349 922.259 5.2453 3346.726 1018.267 4.4885 B 5559.070 1023.059 4.8886 0.002 614 0.549 373 9 0.003 343 36 3.806 1009.163 4.5632 [0.003 559 0.535 392 7 0.003 022 85 T ue uu e xe e e ye e 0.517 386 5 0.005 248 92 0.537 787 5 0.002 970 41 [0.000 005 112 [0.000 010 83 101.113 104.443 ae c K H eA 83.552 64.200 [226.662 44.782 1.110 H Ha X d f f f SX o L x o ?T Energies in cm~1.D. W . Schwenke 327 It is interesting to compare the spin»orbit splitting and various coupling matrix elements determined from the optimization and the values obtained from the ab initio electronic structure calculations. The X state splitting and Xa couplings are given in Table 1 and are close to the optimized values. The E state splitting is computed as 73.7 cm~1 the EX spin»orbit coupling as 57.4 cm~1 and rotation»orbit coupling as 0.96 which are also in good agreement with the optimized values.The Ea coupling 58.9 cm~1 and Ed coupling 13.4 cm~1 are not computed as reliably. The Ea coupling parameter may not be well determined by the existing data. Since these four states are coupled forbidden transitions such as d^X and E^a are possible. These will be discussed in a later section. It is interesting to note that the lowest 1% state (b) is predominately for pd occupation,7 and so the close coupling between the X and a states does not occur between the E state and a singlet p state of rp occupation. 4.2 A3U and g 1C states The electronic structure of the A state is predominantly described by pd occupation and this state is relatively well isolated from the other triplets ; it can only couple directly to the X and C states and they are fairly distant energetically.The singlet deltas are also distant but a 1! state lies nearby which has predominantly d2 occupation. We call this the g state. Since there is no experimental information about the g state one can consider not including it but rather introducing a phenomenological spin»spin parameter to account for its presence. We have found good –ts both with and without the g state but we prefer not to include the phenomenological parameter. T and B were The –nal parameters for these states are given in Table 3. Only e optimized for the g state while the other parameters were estimated from electronic e structure calculations.The A state is the only one which requires radial dependence in the spin»orbit splitting. With these parameters we obtained an RMS error of 0.044 cm~1 for the A state which is quite satisfactory. 4.3 B3P b 1P D3Ró and h 1Rë states We next consider the B 3% state and its most strongly coupled partners. The B and b states are predominantly of pd character and the D and h states are predominantly of d2 character. Experimentally only the E3% b1% and B3% states exhibit K-doubling with the B 3% state having by far the most interesting doubling splitting. In Fig. 2 we Table 3 Optimized parameters for the A and g states g 1! A3' 15 810.089 931.08 3.83 T ue B uu e xe e 14 169.057 867.482 3.8335 [0.011 85 0.507 313 6 0.003 175 77 0.513 144 3 0.002 797 e e ye K ac e e A [0.000 005 708 173.955[64.218(R[R0)/R0 57.005 3.2 a0 R HAf 0 Energies in cm~1.328 Opacity of TiO from coupled electronic state calculation Fig. 2 K doubling splitting for B 3% state from four diÜerent calculations. (See text for details.) show the K-doubling for the v\0 level as a function of J obtained from various calculations. The curves centred on zero agree qualitatively with experiment.15 K-doubling is caused by coupling to states whose Hamiltonian matrix elements depend on the overall parity P. These are the &0 B states and there are several candidates. At energies below the B 3% state there is the d 1&` state and the D3&~ state and at higher energies the next 1&` state which we label h as well as other states ignored in the present calculations.The B state is not well described by Hundœs case (a) basis functions therefore X is not a good quantum number. However we will label the levels by X1 with X1 \0 denoting the lowest-energy sublevel of the three X1 \2 the highest and X1 \1 the middle-energy sublevel. Now the D3&~ state is coupled to the B 3% state by both spin»orbit and rotation» orbit interactions while the singlets are coupled only by spin»orbit interaction to the B state. This is important because the rotation»orbit interaction grows with J while the spin»orbit interaction remains constant. Thus we expect the large-J K-doubling to be controlled by the D3&~ state.In the curves in Fig. 2 centred on 10 cm~1 we show the K-doubling computed coupling only the B and D states via rotation»orbit interaction shifted upwards by 10 cm~1. The splitting is very small at low J but then the magnitude of the splitting increases monotonically as J increases and is nearly the same for all three sublevels with experimental results although this limit has not been reached by J\100. X1 \0 2 negative and X1 \1 positive. This is the same as the large-J limit of the The next set of lines in Fig. 2 centred on 7 cm~1 show the K-doubling computed coupling only the B and D states via spin»orbit interaction shifted upwards by 7 cm~1. The X1 \0 sublevel has signi–cant negative splitting at low J while the other two show 329 D. W . Schwenke very small splitting.As J increases the magnitude of the splitting of the X1 \0 sublevel decreases while the splitting for the other two sublevels becomes negative. The low-J splitting is diÜerent from what is seen experimentally the X1 splitting needs to be shifted upward by ca. 2 cm~1. With the lines in Fig. 2 centred on 4 cm~1 we show the K-doubling splitting obtained coupling only the B and D states via both rotation»orbit and spin»orbit interaction shifted upwards by 4 cm~1. This shows many features in common with the lowest set of curves which match experiment but there is a notable diÜerence at low J where the X1 \0 sublevel has negative splitting instead of positive splitting and crosses zero at JB30 rather than being always positive until large J.It is not possible to change the sign of the splitting of the X1 \0 level by changing the relative phases of the spin»orbit and rotation»orbit interaction with the D state. What is required is a state with energy higher than the B state and furthermore coupled strongly enough so that the spin»orbit interaction with the D and d states is overcome and the net result is a positive splitting of the X1 \0 at low J. We have coupled the second 1&` state to the B state as well as including the B and D state coupling and the results are shown as the lines centred on 0 in Fig. 2. These results are in qualitative agreement with experiment. To obtain quantitative agreement with experiment it is necessary to couple the other states and to optimize the coupling constants further.In summary the K-doubling of the B 3% state at low J is primarily due to spin»orbit coupling to the second 1&` state which is higher in energy while at high J it is primarily due to rotation»orbit coupling to the D3&~ state which is lower in energy. Next we turn to the b state. Its K-doubling is smaller than that of the B state and increases with J. This suggests that it is due to spin»orbit interactions with the B state since the B state splitting increase with J or to rotation»orbit coupling to a 1&` state. We include coupling to the h state but not to the d state since the predominant electronic con–gurations diÜer by a double excitation (r2 vs. pd) which will make the coupling very small. In Table 4 we give the optimized parameters for these states.In order to improve the description of the K-doubling we –t separately the average and diÜerence between the energy levels which diÜer only by the parity and the diÜerence energies are weighted ten times as much as the average. The RMS errors computed using the actual energies rather than the average/diÜerence were 0.743 cm~1 for the B state and 0.054 cm~1 for the b state. In these –ts we –xed T for the h state at its ab initio value obtained from an internally contracted multi-reference con–guration-interaction calculation using the e Table 4 Optimized parameters for the B b D and h states h 1&` D3&~ b 1% B3% 17 564.880 852.117 5.613 6 12 202.658 977.429 3.857 0 16 219.387 867.093 1.852 3 T ue uu e xe e 14 663.772 4 919.485 4.205 6 [0.015 92 0.513 630 1 0.002 710 11 0.488 143 0 0.005 457 56 0.489 527 2 0.002 816 00 0.506 220 9 0.002 978 2 Ba e e ye [0.000 054 02 19.740 42.404 c K H eA e [39.772 25.168 [41.760 [65.316 [42.404 1.336 6 D B f f HS B o L x o ?T [0.982 5 Sb o L x o ?T Energies in cm~1.330 Opacity of TiO from coupled electronic state calculation e Be u were varied and we included the determination of this states position and *G1@2 4331 reference space but optimized the remaining parameters. For the D state T and e by Barnes et al.6 in the –t although with reduced weight since these data are not of the same accuracy as the B and b state data.We match these values within 0.3 cm~1 which is quite satisfactory. The higher order parameters for the D state were –xed at values estimated from electronic structure calculations. is somewhat too small. This is not particularly surprising since data for v\0 1 The errors for the B state are somewhat larger than one would hope for and we have tried various methods to reduce them. One also expects that our optimized value of ue xe primarily determine only two of the three parameters Te ue and ue xe . Perhaps the most important de–ciency in our calculations is the neglect of the coupling to the C state. We have included this coupling in trial calculations but did not obtain signi–- u and u cantly improved results. This may be because the values of exe are not close enough to the correct values to give good overlap with the vibrational levels of the C e state.4.4 C3D c 1U and f 1D states The C state is known to be the subject of several perturbations. Explicit interaction with the c state is required for a compact description of the c state and Amiot et al.4 have determined an estimate of the coupling between these levels 83.85 cm~1. Phillips and Davis23 deduced approximate parameters of a perturber which in retrospect is the f state. Thus at the very least we need to couple these three states. These states are predominantly described by r@d occupation (C and f ) and pd (c). In Table 5 we give the parameters for these states obtained from our optimization. We include a phenomenological spin»spin parameter for the C state to try and account for other electronic states.The RMS errors were 1.582 cm~1 for the C state 0.119 cm~1 for the c state and 0.099 cm~1 for the f state. Note the good agreement between our spin»orbit coupling parameter between the C and c states and that obtained by Amiot et al. It seems clear from the quality of the –ts that an important perturber to the C state has not been included. The errors come mostly from v[4 in the X\1 sublevel. We have included additional states such as the B b D and h manifold and the A state but have not obtained signi–cantly improved results. There are several possible solutions to this dilemma. One is that coupling to the lower states can indeed resolve the diÜerences but we have just not hit upon the correct parameters yet.This is a real problem for the ab initio calculations are just not reliable in this regime; it is not possible to perform Table 5 Optimized parameters for the C c and f states f 1* c 1' C 3* 22 482.170 873.060 2.0870 21 226.587 919.634 4.3273 19 430.264 837.631 4.3177 [0.03 146 0.489 305 0 0.002 667 13 0.503 704 9 0.002 763 13 0.524 734 8 0.003 289 38 [0.000 070 99 99.817 83.123 u T e u ue Ba e e ye e xe c " H A e e 86.323 0.0071 Cf Sc o L x o ?T c [1.996 Energies in cm~1. 331 D. W . Schwenke calculations with a large enough CAS space to describe well all the states with a single set of orbitals. Another solution might be coupling to higher electronic states.There do not appear to be any strongly bound electronic states nearby that we have not included but there are many repulsive states that cross the C state. In their ab initio calculations of the spin»orbit coupling in the X2% state of OH LanghoÜ et al.24 found that repulsive states were increasingly important as R increased. Since higher-v samples have larger R this might be an explanation. Finally we consider the possibility of experimental error. The data we are using19 are not that recent and we have observed two disturbing things First there appear lines with X[J which is not physically possible. (This is not limited to the C^X bands.) Thus either these lines are misassigned or the line positions are formulae generated rather than being actual experimental data.The second is that one can deduce the X-state energies from the A^X and C^X transitions for J higher than that used to start the energy level calculation. We have done this using X-state levels with JO5 and compared the results we obtain for higher J with the results of Gustavsson et al.14 The agreement is not particularly good by J\50 the deviations are ]0.06 cm~1 for X\1 [0.44 cm~1 for X\2 and]0.27 cm~1 for X\3 all v\0. This may be an indication of a problem with the present scheme for computing the energy levels. If there is a problem with the C-state data it is likely that the A is also in difficulty since together the c and a bands give consistent X-state energies. In contrast to these errors we have compared the C-state energies obtained just using the data of Kaledin et al.3 with those obtained just using the Berkeley data and –nd very good agreement.These data are however limited to JO26 v\2 and X\3. 5 Opacities We have computed the line strengths from transitions from the X a D and E states to all states using the transitions moments of LanghoÜ7 and the ro-vibrational wavefunctions determined in the present work. We used J\0»300 and used vibrational basis functions with v up to 19. We also included the X a d E and D dipole moments so that IR transitions are also computed. The dipole moments were obtained from ab initio electronic structure calculations. The X state used a 4221 active space and internally contracted multi-reference con–guration interaction which is similar to the techniques used for the transition moments except that a smaller active space is used and only the X state was computed.A similar calculation was carried out for the a state. For the d E and D states the dipole moment was computed at the CAS-SCF level. The ED and BD transition moments were also computed at the CAS-SCF level. The X a and d state dipole moments were computed at many bond lengths using a –ne grid while the E and D dipole moments and ED and BD transition moments were evaluated only at 2.8 3.0 3.2 and 3.4 a0 . Phasing the transition moments is a concern in the present calculations because the phases need to be consistent with the Hamiltonian coupling matrix elements. Furthermore the x y transition moments are skew-symmetric.We were not able to assign unambiguously the phases in the present work because we could not compute accurate Hamiltonian coupling matrix elements. Our best estimate is that EX XA XB XC ED EB EC BD ab ac af and db are all positive at 3.0 a0 . In Fig. 3 we give the absorption spectrum computed at 4000 K. This was computed using a database of 45]106 lines. The spectrum is dominated by the A^X C^X and B^X transitions. In Fig. 4 we give all the bands separately so that one can see the forbidden bands. We assign electronic quantum numbers by determining the state having the maximum contribution to a level so in cases where there is strong mixing, 332 Fig. 4 Absorption spectrum of TiO at 3000 K decomposed into individual bands Opacity of TiO from coupled electronic state calculation Fig.3 Absorption spectrum of TiO at 3000 K 333 D. W . Schwenke the labels are not necessarily very meaningful. In Fig. 4(a) we show the transitions from the ground electronic state to other triplets. We observe that the forbidden D3&~^X3* transition is almost as strong as the allowed E 3%^X3* transition although the former lies in the same frequency range as the much stronger B 3%^X3* band. In Fig. 4(b) we show transitions from the ground electronic state to the singlets. The strongest of the forbidden bands is the c 1'^X3* transition but there are also transitions to b 1% and h 1&` due to coupling to the B 3% state and transitions to the g 1! state due to coupling with the A3' state and transitions to the a 1* and d 1&` states due to the X3* a1* d1&` and E3% state coupling.In Fig. 4(c) we show the transitions originating from the a 1* state. The strongest forbidden transition is the C3*^a 1* transition. Finally in Fig. 4(d) we show the transitions originating in the d 1&` and E3% states. It would be interesting to have experimental observations of the intensities of some of the forbidden bands for they would be very sensitive probes to the coupling matrix elements in our calculations. A detailed line list will be made available over the world wide web.§ 6 Conclusions We have performed calculations of the ro-vibrational energy levels of TiO in which the various electronic states are directly coupled via spin»orbit and rotation»orbit interactions.The interaction parameters were determined from ab initio electronic structure calculations followed by adjustment in an attempt to match all experimental data. States which are not strongly coupled (X a E d and A) are very well represented. The remaining singlets are also well described but the two triplets B and C which have strong couplings to other states are not well represented. Paradoxically one problem with the B state is there are not enough experimental data only v\0 and v\1 are characterized. The B state is also strongly coupled to two states (D h) for which there are essentially no experimental data which complicates their description. Apart from the A state spin»orbit splitting we use radial-independent couplings. Our ab initio electronic structure calculations con–rm this for the low-lying states but one may make the conjecture that radial dependence is important to describe the C state well because we include data up to v\7.We plan to pursue this as well as to investigate the role of Band C-state couplings in future work. We represent several states well but have trouble with the C state most likely because we are not including enough coupling to diÜerent states. Some of the states have surprisingly little experimental data which causes some difficulties. The principal lack is in vibrational levels of the B state. An important consideration for all states is that data over a wide range of J are required to ensure uniqueness in the –t. This is because diÜerent mixings of the electronic states for instance a basis which diagonalizes the Hamiltonian for –xed X which would be appropriate for a full relativistic electronic structure calculation are only distinguished by the dependence of their couplings on J K and X.Our calculations yield predictions of band intensities for both well characterized bands and new unobserved forbidden bands. Since many of the electronic states are coupled the usual selection rules regarding S and K are no longer rigorous. Thus we see a 1*^X3% at ca. 3500 cm~1. The forbidden bands will be weak but if they occur in a region where no other TiO transitions occur then they might be observable. Another transition of interest is g 1!^X3* at ca. 15 000 cm~1 which is close to the 6500 ” region where Plez8 –nds he is missing opacity from his simulation.It would be interesting to see if this forbidden transition is the cause of the discrepancy. § http //ccf.arc.nasa.gov/Ddschwenke 334 Opacity of TiO from coupled electronic state calculation Paper 8/00070K; Received 2nd January 1998 References 1 U.G. J‘rgensen Astron. Astrophys. 1994 284 179. 2 C. Amiot E. Azaroual P. Luc and R. Vetter J. Chem. Phys. 1995 102 4375. 3 L. A. Kaledin J. E. McCord and M. C. Heaven J. Mol. Spectrosc. 1995 173 499. 4 C. Amiot M. Cheikh P. Luc and R. Vetter J. Mol. Spectrosc. 1996 179 159. 5 R. S. Ram P. F. Bernath and L. Wallace Astrophys. J. Supp. 1996 107 443. 6 M. Barnes A. J. Merer and G. F. Metha J. Mol. Spectrosc. 1997 181 180. 7 S. R. LanghoÜ Astrophys. J. 1997 481 1007.8 B. Plez Astron. Astrophys. submitted. 9 J. G. Phillips Astrophys. J. Supp. 1973 26 313. 10 D. C. Galehouse J. W. Brault and S. P. Davis Astrophys. J. Supp. 1980 42 241. 11 MOLPRO 96 is a package of ab initio programs written by H.-J. Werner and P. J. Knowles with contributions from J. Almloé f R. D. Amos M. J. O. Deegan S. T. Elbert C. Hampel W. Meyer K. Peterson R. Pitzer A. J. Stone and P. R. Taylor. 12 T. Helgaker P. R. Taylor K. Ruud O. Vahtras and H. Koch ììHERMIT a molecular integral programœœ ; H. J. Jensen and H. Agren ììSIRIUS an MCSCF programœœ ; O. Vahtras H ”gren P. J‘rgensen H. J. Aa. Jensen T. Helgaker and J. Olsen J. Chem. Phys. 1992 96 2118. 13 S. Weissman J. T. Vanderslice and R. Battino J. Chem. Phys. 1963 39 2226. 14 T. Gustavsson C. Amiot and J. Verge` s J. Mol. Spectrosc. 1991 145 56. 15 W. H. Hocking M. C. L. Gerry and A. J. Merer Can. J. Phys. 1979 57 54. 16 B. Simard and P. A. Hackett J. Mol. Spectrosc. 1991 148 128. 17 G. R. Brandes and D. C. Galehouse J. Mol. Spectrosc. 1985 109 345. 18 C. Linton J. Mol. Spectrosc. 1974 50 235. 19 J. G. Phillips and S. P. Davis TiO band analyses 4100»9000 ” Dept. of Astronomy and Physics Univ. of California Berkeley. 20 I. Kovaç cs J. Mol. Spectrosc. 1965 18 229. 21 R. McWeeny Methods of Molecular Quantum Mechanics Academic Press London 2nd edn. 1992. 22 C. Linton and H. P. Broida J. Mol. Spectrosc. 1977 64 382. 23 J. G. Phillips and S. P. Davis Astrophys. J. 1972 175 583. 24 S. R. LanghoÜ M. L. Sink R. H. Pritchard and C. W. Kern J. Mol. Spectrosc. 1982 96 200.

ISSN:1359-6640    

DOI:10.1039/a800070k

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Faraday Discuss. 1998 109 335»347 Excitation processes for the emission of the unidenti–ed IR bands 1 Introduction To our knowledge no individual molecule has yet been shown to carry all the ubiquitous and closely correlated UIR bands. Hoyle and Wickramasinghe1 attempted to solve the problem by mathematically adding the spectra of 115 independent PAH molecules. The result however only bears a slight resemblance to typical interstellar spectra which exhibit narrow bands with high contrast (weak continuum). Recently the ISO a satellite dedicated to interstellar and galactic research was launched with four IR instruments on board. Three of these have good mid-IR spectral resolution capabilities designed to analyse the UIR bands and hopefully resolve whatever molecular structure could be present.None could be detected and this result is considered to be very difficult to reconcile with the PAH hypothesis.2 Another Japanese IR satellite IRTS has shown that the near-IR CwH-stretching band observed in emission near the Galactic Plane carries an important aliphatic component, 3 reminiscent of the dominant feature observed in absorption towards IRS 6E Olivier Guillois Gilles Ledoux Ire` ne Nenner Renaud Papoular* and Ceç cile Reynaud CEA CE Saclay SPAM 91191 Gif-sur-Y vette Cedex France To our knowledge no individual polycyclic aromatic hydrocarbon (PAH) molecule has yet been shown to carry all the ubiquitous and closely correlated unidenti–ed IR (UIR) bands. More generally no IR space observatory (ISO) instrument was able to split these bands into the narrower signatures expected from molecules.Again the Japanese IRTS satellite has shown that the near-IR CH-stretching band observed near the galactic plane carries an important aliphatic component alien to aromatic molecules. On the other hand a number of solid-state disordered more or less hydrogenated carbonaceous materials have been shown to carry all the UIR bands. Moreover their emission spectrum computed under thermal equilibrium with typical circumstellar radiation –elds was shown to –t satisfactorily and in detail the observed near- and mid-IR spectra of a number of planetary nebulae (PNe) and proto-planetary nebulae (PPNe). However the interstellar radiation –eld (ISRF) is not strong enough to heat this solid-state dust to the equilibrium temperatures required for emission observed from the galactic plane and re—ection nebulae.Even stochastic heating is insufficient in these cases because nanometric grains are too small to retain the desirable optical properties of the bulk. One way out of this deadlock is to look for other excitation mechanisms. Noting that spatial maxima of UIR emission usually occur in photodissociation regions between ionized and molecular gas maxima where atomic H is most abundant we have set up to study the interactions of atomic hydrogen with solid-state carbon surfaces. We consider in particular a mechanism in which upon collision with a grain the potential energy of recombination carried by an atom is delivered to one of the functional groups which will emit a UIR band.Such an IR chemiluminescence is found to be possible in principle. A laboratory experiment dedicated to the quest of IR chemiluminescence has been built and is described. 335 336 Excitation processes for the emission of UIR bands near the Galactic Center.4 Both features are alien to the aromatic molecules of the PAH conjecture. Support for the molecular hypothesis now stems from the belief that the only possible mechanism for the excitation of the bands in the IS medium is stochastic heating which requires that the emitters be very small aggregates of atoms irradiated by far-UV photons. EÜorts are therefore being made using PAHs and other precursors to synthesize in the laboratory aggregates large enough to reproduce the UIR bands but not too large to respond adequately to stochastic heating.However Sellgren5 has long been drawing attention to another problem with this model in several instances UIR spectra are observed similar to the usual ones although the photons available from the illuminating stars are mainly visible. The latest and most telling instance of this is given by Uchida et al.6 Consider now the opposite models bulk solid-state carbonaceous grains made of hydrogenated amorphous carbon (HAC) quenched carbonaceous compounds (QCC) or natural coals. Guillois et al.7 showed that coals with various degrees of hydrogenation (and hence aromaticity) as found in mines have bulk optical properties which provide satisfactory spectral –ts to celestial observations.In the case of strongly illuminated dust such as in PPNe or PNe radiative equilibrium heating was shown to produce agreement not only with observed band shapes and contrasts but also with absolute intensities. Two major difficulties however stand in the way of applying this model to the nearand mid-IR emission of re—ection nebulae (RNe) and photo-dissociation regions (PDRs). First the radiation illuminating such objects is usually insufficient to sustain equilibrium grain temperatures high enough to excite the UIR bands as observed. Second the underlying continuum emitted by these objects is so much reduced (with respect to the bands)8 as to be incompatible with the emissivity spectra of the carbonaceous materials considered above.The latter difficulty obviously cannot be overcome by stochastic heating (which involves the same emissivities as equilibrium heating). Neither can the former. For in order to have the right spectroscopic features carbonaceous materials must also have intermediate-range order as well as short-range order. This sets a lower limit to the size of the grain for the latter to retain the optical properties of the bulk it is of the order of 30 ” i.e. ca. 2000 atoms. This is much larger than the upper limit set for efficient stochastic heating.9 What then is the way out of this deadlock? Clearly not enough was done in the past to explore other excitation processes for vibrational emission. This paper considers theoretically and experimentally the possibility that the interaction of atomic hydrogen with solid carbonaceous grains can provide a convenient excitation mechanism.The reason for this interest in the –rst place is the observation that the spatial maximum of this emission normally occurs in the PDR region between ionized and molecular gas maxima i.e. where hydrogen atoms are dominant and more precisely near the restricted region where H vibrational radiation is observed. This is best illustrated by 2 the edge-on Orion Bar.10 Another hint to the role of atomic hydrogen in exciting the UIR emission is the detection of the latter at high galactic latitudes by the IRTS satellite :11 the intensity of emission decreases quickly with increasing latitude but its spectral pro–le remains unchanged. Concurrently it is observed that contrary to H molecules which are con- –ned near the Galactic plane the density of H atoms extends farther than 200 pc on 2 either side,12 although it too decreases with increasing latitude.Also as it happens recombination of H atoms on grains is presently the only candidate mechanism for the formation of the abundant hydrogen molecules. In this paper we explore the following hypothetical scenario. Upon hitting a grain atoms recombine either together or with H or C atoms in the grain (exothermic 337 O. Guillois et al. reactions). Part of the corresponding bonding energy is then deposited in selected nuclear oscillators which are thus set in vibration. If the vibration is adequately con–ned (localized) these oscillators eventually emit the desired near- and mid-IR bands.In this scenario contrary to photonic excitation the available energy is not deposited in the electronic bath to be subsequently distributed among the nuclear oscillators. Rather it is directly delivered to the oscillators of interest this mechanism should be energetically more efficient and at the same time it does not excite the undesired continuum (due to the electrons) and thus overcomes both difficulties mentioned above. Moreover in accordance with observations,11 the spectral shape does not depend on the ambient –eld intensity. Since the atomic H population is maintained by the ambient UV –eld against recombination a necessary condition for the above scenario is that the energy carried by this –eld be at least as large as that carried by the UIR bands.In Sections 2 and 3 below this is shown to be the case on the basis of astronomical observations. Section 4 then discusses the details of the interaction of the atoms with the grains recombination into H molecules or CwH functional groups energy deposition in the grain (closely related to the former) energy con–nement and –nally IR emission. The probability of each 2 process is assessed in the light of laboratory data. The results of this analysis are encouraging enough to start an experiment dedicated to the detection and measurements of these processes. This is described in Section 5. 2 Geometry and spatial distributions Generally and qualitatively when a cloud is illuminated by a source of far UV (FUV) the outer front layers of the cloud are ionized and form an HII shell.Between the latter and the molecular core an intermediate shell is formed in which the hydrogen is mostly dissociated (HI) and which is called a PDR. Most of the earlier theoretical treatments of PDRs have assumed steady-state and uniform distribution of nuclei and dust densities throughout which considerably simpli–es the analysis.13 However recent analyses14,15 suggest that the steady-state often does not hold because early-type stars light up quickly and send out fast shocks into the nearby cloud. In such cases the dust is only present in the molecular regions. Fig. 1(a) schematically represents at a given typical evolutionary phase the corresponding one-dimensional spatial distributions of the material species in the HI and H shells surrounding the HII region.The latter is not relevant to the present discussion and therefore is not rep- 2 resented. This diagram seems to comply with observations of famous PDRs such as Orion and M17. In particular (a) neither the IR continuum nor dust band emission Fig. 1 Schematic representation of 1-D spatial distributions of particle number densities (HI H2 slightly in front of the H peak.19 H (FUV FVIS) along the radius from the illuminating star (assumed to be 2*) and of radiative —uxes on the left-hand side) to the molecular core (on the right-hand side). The UIR emission peaks 2* 338 Excitation processes for the emission of UIR bands UV occurs in most of the HI region but both increase steeply in intensity near the molecular edge then decrease more slowly into the molecular cloud;10,16h19 (b) no dust absorption feature (such as the 10 lm band of silicates) appears in the HI region ; (c) the maximum quadrupolar vibrational H emission (v1]0) occurs just beyond the IR maximum 2 while n(HI) decreases and the molecular densities increase steeply.16,18,19 Consider now the radiation (W m~2 Hz~1) falling upon the PDR from the left.Part F of it (W m~2) is hard enough to be absorbed by molecules (11»13.6 eV; see ref. 20) ; UVFVIS(W the rest m~2) is only absorbed by dust. No photon harder than 13.6 eV is available since it would have ionized the H atoms. The spatial variations of F and F are schematically represented in Fig. 1(b) and justi–ed as follows. VISLet n1 n2 nT and ng respectively represent the number densities of atomic molecular nuclei (or total) hydrogen and dust grains.Let a be a typical grain radius and o (g cm~3) the dust bulk density. Assume a constant ratio g of dust to nuclei masses. Then (1) g\4 3 na3 mH o n n T g T\n1]2n2 . For instance taking g\3]10~3 where m is the atomic mass of H and n (carbon-rich dust) a\10~6 cm and o\2 g cm~3 we get ng/nTB7]10~10. H In order of magnitude the optical cross-section of the grains is given by pg\na22aa\2naa3 where a (cm~1) is the absorbance of the bulk grain material (of the order of 105) and 2aa the optical efficiency. As for the H molecule its average absorption cross-section over the UV range can 2 be estimated from the work of Dalgarno and Stephens21 and Jura22 to be p2\2 ]10~17 cm2 assuming the IS photon —ux to be uniform over the spectral range de–ned above.The ratio of molecular to dust UV absorbances is therefore (2) SUV\ n n 2 p2\5]104 n n 2T F g pg Now in the HI region n2/nT@1 so that most of the incoming UV photons would have been absorbed by dust had this been present in the normal mass concentration [eqn. (1)]. Since observations suggest that dust is excluded from this region (see above) the radiative —uxes remain constant throughout except near the border of the H region (Fig. 1). There they begin to encounter dust and hence are attenuated. This allows 2 recombination of atoms to start. As a consequence the number of molecules increases and the UV radiation is further attenuated.The ensuing catastrophic process quickly eliminates most of the H atoms and the gas becomes essentially molecular i.e. i n n 2BnT S eqn. (2) and UVA1. Hence the steep fall in FUV to the right through molecular absorption while VIS which is only absorbed by dust decays more slowly. 3 Energetic budget Part of the irradiation energy goes into heating dust to temperatures which in the cases considered here do not exceed ca. 100 K. This energy cannot therefore contribute to the UIR emission. The rest of the irradiation energy goes into dissociation of the hydrogen molecules. The latter proceeds through a two-step process (see ref. 21) an electronic photoexcitation followed by radiative dissociation into slow H atoms in the IR ground state.However each atom is endowed with a potential energy of 2.25 eV with respect to the H molecule in its ground state equal to half the dissociation energy. Barring threebody recombination in the gas phase in view of the low gas densities involved this 2 energy can only be transferred to the grains upon recombination thereon. Fig. 1 shows 339 O. Guillois et al. that this can only occur around the thin transition region between HI and H2 where H is absorbed. 2 UV * Dalgarno and Stephens21 have shown that not every UV absorption by an H F is con–ned and where essentially all molecule terminates in the dissociation of the latter this only happens with an average 2 probability of ca. 25%. However there is no indication that the rest of the energy absorbed is efficiently converted into kinetic or excitation energy of the molecules the sum of these does not exceed ca.0.1 eV. In most cases absorption is followed by emission of another UV photon and relaxation to the electronic ground state of the molecule. Since the latter photon is most probably absorbed by another molecule near by rather than by a dust grain this cycle can be repeated a number of times in a sort of resonance trapping until dissociation occurs. We shall therefore assume as a zero-order approximation that all F is converted into atomic H potential energy through molecular dissociation. In assuming this approximation note that in the HI region part of the UV extreme UV of the star is converted into Lyman lines which can also contribute to the dissociation of H2 .Finally in non-equilibrium conditions the ionization and dissociation fronts may merge together,14,15 so that photons harder than 13.6 eV also become available for dissociation. g be the overall energetic efficiency of the putative excitation process i.e. the Let probability that upon hitting a grain surface the H potential energy is converted into e UIR emission over 4n sr. The brightness of the PDR seen at an angle h to the x-axis is then (3) Bir\ 4 g n e cos FUV h where cos h takes into account the fact that the emitting layer is optically thin in the IR B so that its brightness (W m~2 sr~1) is proportional to the length of material along the sight line. IR Note the simplicity of this result which is independent of the cloud parameters; it is a consequence of the double assumption made here namely that the UV radiation is absorbed only by H molecules and that statistical equilibrium obtains between dissociation and recombination.2 F and B B7.7\5]104 MJy sr~1\2]10~3 W m~2 sr~1 (for a g is ca. 0.2. What do the available observations tell us about UV the ISO satellite have been condensed into curves relating these quantities for each IR ? Recent results from mid-IR band.23 It turns out that (1) B is roughly proportional to F IR former is integrated over a given UIR band and the latter over the useful spectral range UV where the for molecular absorption. Eqn. (3) is in functional agreement with this observational result. (2) In the case of M17 for instance the stellar irradiation is cU/2B0.25 W m~2 (where the energy density UB104 eV cm~3 and c is the velocity of light) and the intensity of the 7.7 lm band is bandwidth *l\4]1012 Hz).Thus B7.7/FUVB8]10~3 sr~1 and from eqn. (3) g7.7B0.07 if an ìaverage eÜectiveœ value of 2~1@2 is taken for cos h. Since the total in-band energy in the UIR radiation is ca. 3 times the energy in the 7.7 lm band the minimum overall efficiency While this shows that enough potential energy is available in principle it also e implies that once an H atom is trapped at the grain surface the probability of successive bonding vibration excitation and IR emission should be quite high. This is a severe constraint but not out of reach in principle as will be seen in the next section dealing with this issue in more detail.Although the argument above was made for dense and highly illuminated PDRs it can also apply to dense ìcirruses œ weakly illuminated by the IS RF. The hydrogen behaviour in such objects has been theorized by Hollenbach et al.24 and Jura.25 If the cloud density is high enough for appreciable UV extinction to occur and be followed by steep hydrogen recombination the general picture is the same as above. If this is not the 340 Excitation processes for the emission of UIR bands case the IR emission no longer obeys eqn. (3) but depends on the cloud density and volume. Moreover contrary to the case of illumination by a single powerful and known associated star it becomes difficult to take into account all the neighbouring stars in estimating the eÜective illumination of the cloud.22 This is why it is preferable for quantitative purposes to consider the dense and strongly illuminated PDRs.Having ascertained above that the potential energy of the atomic hydrogen gas is in principle sufficient to fuel the UIR emission we now turn to the detailed analysis of the possible mechanisms by which this energy can be transferred to the grains and thus excite the vibrations of the solid state nucleic oscillators. 4 Details of hydrogen interactions with grains A rich variety of chemical reactions and energy transfer processes occur at gas/solid interfaces and recent technical progress has fostered renewed interest and vigorous research in this –eld.26h31 It must be clear that we are not considering here gas condensation or physisorption which involve weak forces such as van der Waalsœ or other ìphysicalœ forces and therefore small energy exchanges (O0.1 eV).We are interested rather in chemical processes such as chemisorption in which energy exchanges are ca. 1 eV. For the present purposes it is of interest to distinguish (1) catalytic recombination of gas particles and (2) chemical reactions between gas and surface atoms.32 Both types involve energy exchanges between gas and solid and each can be subdivided into subclasses (see for instance ref. 33). The interaction is favoured when atomic gases (radicals) are involved because of the presence of unpaired electrons. In particular when atomic H impinges on a carbon surface a wealth of reactions occur with the bulk material (see review by Kuppers34) which can be grossly classi–ed as hydrogenation H abstraction (or dehydrogenation) H-induced erosion (extraction of various molecular carbonaceous species) and reactions with physisorbed hydrocarbon molecules.It should also be remembered that VIS»UV irradiation may modify the course of these reactions.35,36 Finally even molecules can interact with surfaces as in dissociative chemisorption.30 While the literature on these subjects is huge it is mostly very speci–c to the particular combination of gas and solid considered. In the present state of the art it is not easy to extract laws general enough to predict behaviours in diÜerent astronomical environments. We have therefore to make educated guesses and most importantly do laboratory work.4.1 Recombination of atoms on grain surfaces (kinetics) The physics of surface recombination has traditionally been described by two model processes (1) the Langmuir»Hinshelwood (L»H) mechanism in which two gas atoms independently strike and stick (accommodate) to the surface then wander randomly (diÜuse) until they collide and recombine together. (2) The Eley»Rideal (E»R) (or direct) mechanism in which only one gas atom is initially bound to the surface and recombination occurs by direct (head-on) impact of a second gas atom on the former before any accomodation of the latter to the surface has taken place.30,37,38 In recent years however a more general process has been introduced,39 which reconciles the other two and most gas»surface experiments.This is the precursor- or trapping-mediated process in which after striking a surface a gas particle is partially accomodated (i.e. comes nearly to thermal equilibrium with the surface) and is hence momentarily trapped; it can then explore the surface population to form a precursor compound and –nally relax to yield a stable molecule. Clearly the L»H and E»R processes can be considered as opposite limiting cases of the precursor process depending on the life-time of the intermediate stages (a speci–c character of the reactive system 341 O. Guillois et al. under consideration). If this time is short enough with respect to the inverse of the arrival rate of impinging particles so that a steady state can set in then the reaction rate is proportional to the gas particles in—ux (as is always the case in the E»R mechanism).This also appears to hold for H formation in space where the gas atoms in—ux is n1 and the rate of formation per unit volume is therefore written as proportional to 2 Rn with R nearly constant and of order 3]10~17 cm3 s~1 on average.40 We shall hence adopt the precursor paradigm which moreover is not limited to catalytic heter- 1nT ogeneous recombinations and applies to numerous other gas»surface reactions which may be of interest to us. We shall now use the observational constraints on R to obtain information on the detailed recombination mechanism on grains. If v is a typical H atom velocity then the n1v1/4 and using eqn. (1) the rate of recombination per cm3 —ux impinging on a grain is 1 can also be written (4) 4 1 n1v14na2 3 4 gm na H 3o nT c where c is the recombination probability per impact.Equating eqn. (4) to Rn1nT we get (5) v a 1 g \4 3 R m o B5]107 s~1 H For a gas temperature of 1000 K,12 independent of n and nT . 1 v1B5]105 cm s~1 and (6) cg a B100 cm~1 Rd\na2ng/nT; hence a\2]10~6 cm and the required cB0.06. Since cO1 by de–nition the grain radius cannot exceed ca. 3]10~5 cm. This argument can be considered as a ìmeasurementœ of the grain-size range. Alternatively eqn. (6) may help identify the composition and structure of the relevant grains. Indeed Spitzer,40 from general extinction considerations deduced the value of 10~21 cm2 for R ical grains d the mean projected area of dust grains per H atom.Assuming spher- Now c has been measured directly by King and Wise41 for H on pure carbon –lms. (7) B They found it to vary with surface temperature T according to c\0.38 expA[2300 RT where R\1.98 cal °C~1 mol~1 and T is in K. At the lowest temperature used in this work ca. 300 K c is already down to 0.01. Indirectly from the work of Balooch and Olander,42 who measured the sticking coefficient of H on the prism plane of pyrolytic graphite to be 0.02 we infer again that the recombination probability is too small. We are therefore led to consider two cures to this situation (a) The grains may be highly porous so that the eÜective surface presented to the atoms is much larger than the projected one; like for instance activated charcoal which presents an eÜective surface for adsorption of up to 2.5]107 cm2 g~1,43 as compared with 1.5]105 for –lled spherical grains of radius 10~5 cm.Pirronello et al.44 reach the same conclusion from an experimental study of the recombination of H on olivine. (b) A chemical reaction speci–c to amorphous hydrogenated carbons (a-C :H) exposed to H radicals could have a much higher c because of the presence of a large population of H atoms in the bulk some of them possibly trapped interstitially (as opposed to chemically bound to carbon atoms45). Sugai et al.,45 for instance observed a linear dependence of the DH molecule release rate on the —ux of D` ions bombarding a-C:H –lms. They interpreted this as abstraction of H by D followed by HD recombination in the bulk (but near the surface) 342 Excitation processes for the emission of UIR bands and subsequent molecular diÜusion to the surface and out.From their results we infer cB0.5. In this experiment the energy of the in-falling D ions was as high as 300 eV and one may wonder if similar reactions would occur at lower energies such as encountered in the interstellar medium (ISM). However Harris and Weiner46 and Gat and Angus47 con–rm this possibility by measuring recombination probabilities cP0.1 in carbon –lm deposition reactors (TgasB1000 K or about the temperature expected in PDRs). It should be stressed that in this type of reaction at the steady-state for each H abstracted by an incident atom to form a molecule another incident atom is trapped in the solid sample and makes a bond with the lattice,48 wherein some energy is deposited.Thus both for catalytic recombination and surface chemical abstraction a close relation is established between H recombination and a possible chemiluminescence. 4.2 Energy exchange upon recombination (dynamics) etc. are extracted from the target.50,51 Upon recombination of the atoms part of their dissociation energy (exothermicity) is given up to the solid and to the newly formed molecule (translational vibrational and rotational energies). Although high levels of vibrational and rotational excitations have been observed in the nascent molecules they do not carry all the available energy and neither does the translational component.For example Rettner37 has measured the energy partition in the case of H atoms striking a gold surface covered with adsorbed chlorine to produce HCl molecules. About half the heat of reaction (2.3 eV) is carried away by these molecules 26% translational 14% vibrational (mostly v\1) and 5% rotational. The rest of the exothermicity is delivered to the solid lattice through singleand multi-phonon excitation or electron»hole (e»h) pair creation. The transfer of a fraction of the incident atomic potential energy to the phonon bath of the target is evinced by the temperature rise of the latter as measured by a simple thermocouple.49 In the case of hydrocarbon targets the presence of abundant low-mass particles to functional groups comprising H atoms.As a consequence even heavier molecules (bound H atoms) increases the probability of energy transfer from the light nascent H2 such as CH4 C2H2 Another observed outlet for exothermicity is visible chemiluminescence.35,52h60 In this case however the details of the energy channelling are not understood. It has been suggested that some of the energy carried away by the newly formed molecules may be transferred by collisions to the lattice atoms or molecules which may thus be excited to a higher electronic state through the movement of the electron clouds which accompany the vibrating nuclei of the nascent molecules. The high mobility of the latter in the bulk should favour this pathway. Another possible pathway in metals is e»h excitation.35,57,61 Whatever the case may be a notable fraction of the exothermicity is observed to be available for excitation of molecular vibrations so that IR chemiluminescence appears to be another possible outcome of the energy transfer at least in principle. The experimental search for such an emission is precisely the subject of the present work. In the gas phase it has already been abundantly observed. In the case of gas»solid interactions the obvious obstacle to such an emission is the strength of the non-radiative (n-r) deexcitation processes as compared with the weakness of radiative (r) de-excitation. This issue is taken up next. 4.3 Excitation energy con–nement (localization) Most of the available data on vibrational relaxation is relative to molecules attached to regular (crystalline) solid lattices.61,62 In general the corresponding n-r relaxation times range between 10~12 and 10~6 s while IR radiative times range between 10~3 and 1 s making for a very low IR conversion efficiency.In several instances however vibra-343 O. Guillois et al. tional —uorescence decay times of several ms were observed upon selective vibrational excitation of functional groups by IR lasers Koch et al.63 (CN~ on alkali halides) Yang et al.64 (CN~ on CsCl); Chang et al.65 (CO on NaCl). The main n-r relaxation route in non-metals is emission of lattice phonons (heat conduction and heating of the bulk). However the probability of this process can be considerably reduced in the following circumstances (a) There is a large energy mismatch between the oscillation of interest and the lattice characteristic phonons.Relaxation then requires the improbable simultaneous emission of several phonons.66 In this respect the shorter wavelength UIRs (CwH stretches) are privileged as their wavenumbers are much larger than those of the carbon lattice (mostly below 1000 cm~1). (b) The lattice is disordered (chemically topologically or substitutionally) or highly porous. In this case the very notion of phonon is questionable for some vibrational modes because scattering hinders their propagation.67 The concept of localization of the vibration is therefore introduced.68 Indeed it appears from molecular dynamics calculations69 and from neutron scattering measurements,70 that some vibrations can be shared by only a limited number of atoms around the excited bond (spatial con–nement).DiÜusion of energy to the bulk is hindered and the lifetime of the vibration lengthened as a consequence. In coals which are good dust analogues disorder shows up in the very low thermal conductivity as compared with graphite and in very long proton spin»lattice relaxation times (up to 500 ms; see ref. 71). (c) The bulk temperature is low which reduces n-r relaxation by stimulated emission of phonons.72 This is illustrated by the usual decrease of thermal conductivity at lower temperatures. The physics of electronic energy relaxation in metals and semiconductors may serve as an example of the factors involved as it was studied –rst and foremost and because of the similarity between phonons and electron waves.These factors are discussed at length in the literature with regard to silicon-based materials. In porous silicon (p-Si) high porosity gives rise to quantum con–nement and consequently high (red) photoluminescence yields even at room temperature and even higher at lower temperatures. Typical con–nement lengths are a few nanometres.73 In amorphous hydrogenated silicon (a-Si :H) disorder entails localization.74 This was beautifully demonstrated75 by monitoring the photoluminescnce spectrum of a sample of a-Si:H while its porosity was progressively increased. The con–nement length was of the same order as above but was independent of porosity. Another interesting feature of porous silicon (p-Si) is that it is often hydrogenated (passivated) ìby constructionœ to –ll the inevitable dangling bonds which would otherwise provide efficient channels for n-r relaxation.Many SiwH bonds are therefore available whose characteristic frequency of ca. 2200 cm~1 provides a good opportunity to look in the laboratory for solid-state IR chemiluminescence under atomic H bombardment. 4.4 Global emission efficiency g e We are now in a position to compare quantitatively the requirements of astronomical observations with the physics of energy conversion. We have seen that the latter comprizes three steps recombination energy deposition and IR emission. Since no energy loss process for H atoms and no formation mechanism of H molecules are presently contemplated other than recombination of atoms on dust grains all the FUV –eld 2 energy expended in H dissociation must end up in the grains.The global emission efficiency de–ned by eqn. (3) can then be written as 2 (8) ge\gex gd gr where gex is that fraction of the dissociating photon energy which is made available to d the grain by the atoms in chemical form (exothermicity) ; g is that fraction of the exothermicity which goes into the excitation of molecular vibrations of interest and g is r 344 Excitation processes for the emission of UIR bands the energetic branching ratio to the corresponding IR radiation as opposed to n-r relaxation after excitation. g was deduced from observations23 to be ca. 0.2 while from Section g g dB0.75 so that from eqn. (8) must reach ca. r In Section 2 gexB0.4 4.2 e and we may hope that 0.7.Now g can be expressed as r gr\q nvr q nvr]qr qr~1 and q~1 nvr ). Hence qnvr should be of the same q i.e. 10 ms for CwH bondings.76 As shown in Section 4 such a where q and qnvr are respectively the life-times of vibrational excitations against IR radiation (in any one of the UIR bands) and non-radiative relaxation to the lattice (the r rates of the corresponding processes are order of magnitude as long n-r time although exceptional in metals and regular lattices is more probable for r hydrogen vibrators in disordered structures with low-energy lattice phonons (low Debye frequency). However obtaining an adequate g remains the hardest part of the present challenge and can only be taken up by investigating various compounds and structures.r The next section describes the apparatus we have set up for this purpose. 2 5 Experimental The experiment is designed to observe a solid sample surface in IR and/or VIS spectroscopy while it is held in vacuum and irradiated with H atoms and/or electrons and FUV photons. Its temperature can be lowered to 77 K or raised to 500 °C. The sample sits at the centre of an eight-port vacuum chamber made of steel (Fig. 2). The functions and parts are as follows Pumping port A; 110 l s~1 molecular pump; chamber pressure limit \ 10~6 Torr. Cooling port B; vertical copper rod with lower end immersed in liquid nitrogen ; the sample chip or pellet sits on the rodœs bevelled upper end inclined at 45° to the vertical ; the rod is rotatable about its axis.Heating electric coil wound around the upper end of the copper rod. H dosing port C; the atomic H source comprises a vertical cylindrical watercooled Pyrex glass chamber in which an electric discharge is maintained in hydrogen gas at a pressure between 0.05 and 5 Torr by 20 W of rf (27 Mhz) fed by a thyristor generator through a copper coil.77 The dissociated gas diÜuses downwards into the main vessel through a capillary 2 mm in diameter whose lower end is set at a few millimetres above the sample. The inner walls of the Pyrex vessel are lined with a thin layer of orthophosphoric acid which inhibits the recombination of H atoms. This source delivers up to 1018 particles s~1 the atomic fraction of which may reach 75%.FUV irradiation port D; YAG laser (1.06 lm) followed by two doubling stages ; the FUV (2660 ”) output is ca. 1 W in 6 ns pulses 45 mJ each at a repetition rate of 20 Hz. Before entering the main chamber the FUV beam is attenuated as desired by Schott glass –lters (UG11) and a calibrated rotating polarizer»depolarizer system. Electron dosing port E; electron gun 450 eV 150 lA with two pairs of orthogonal deviation plates. Observation port F equipped with a CaF window; (a) InSb/HgCdTe spectrophotometer equipped with circular variable –lters (2.5»15 lm); sensitivity 2]10~13 W Hz~1@2; (b) a three-grating spectrometer with VIS»near-IR photomultipliers. An intermediate lens collects sample radiation from an area up to 1 mm2 into a solid angle of up to 5 msr.The estimated lower limit of the conversion efficiency g3.3 detectable with this system is 2]10~6 for the 3.3 lm band. Miscellaneous Gauges port G. Thermocouple port H; chromel»alumel; sensitive to 0.05 °C variation in the sample temperature. Using a copper sample under known FUV —ux the response of the system is found to be 0.25 °C mW~1. 345 O. Guillois et al. Fig. 2 Schematic representation of the experimental set-up. See text (Section 5). The whole set up was proven to be operational overall by irradiating a copper chip with a known hydrogen —ux from the source and deducing the recombination probability c from the sample temperature increase. Assuming each recombining atom delivers 2.25 eV to the sample c is found to be 0.25 as expected.32 6 Conclusion Analysis of available data on PDRs led us to suspect that atomic hydrogen plays a leading role in the excitation of UIR bands.General arguments about dissociation and recombination of H show that the quantity of dissociation energy stored in the atoms is indeed larger than that which is emitted in the UIR bands. This suggests a mechanism wherein bands are excited upon direct collisions of H particles with the functional groups to which these bands are assigned. As a consequence the shape of the emitted IR (vibrational) spectrum is independent of the illuminating radiation. However since there are only as many H atoms as can be liberated through H dissociation by the FUV –eld the total energy in the UIR bands from a given nebula should be proportional to the 2 available FUV energy at the surface of that nebula.Both expected results are in agreement with observations of weakly illuminated nebulae such as RNe. 346 Excitation processes for the emission of UIR bands The problem with this scenario is to –nd the right structure of hydrogenated dust material for which non-radiative relaxation rates do not exceed the radiative rates i.e. which will con–ne the vibrational energy until an IR photon is emitted. Examples of vibrational life-times as long as the required several ms are already found in the literature but the reason for them is not yet quite clear. In the case of the much studied silicon-based materials disorder or high porosity are known to cause long con–nement times. These –ndings encouraged us to build and describe an experiment dedicated to the detection of IR vibrational emission of various solid surfaces irradiated by H atoms.The results will be reported as available. Note added in proof. Another way of estimating the required overall emission efficiency g (Section 3) is provided by the energetic budgets drawn up by K. Sellgren L. Allaman- e dola J. Bregman M. Werner and D. Wooden (Astrophys. J. 1980 299 446) for a number of nebulae. They indicate that the UIBs carry ca. 1»2% of the total incident radiative energy. For a star of temperature 3]104 K the FUV carries a fraction (0.13) of the total ; hence geB0.1. References 1 F. Hoyle and N. Wickramasinghe in T he T heory of Cosmic Grains Kluwer Dordrecht 1991. 2 J. L. Puget IAU Conference Kyoto August 1997 to appear in Highlights of Astronomy 1998.3 M. Tanaka T. Matsumoto H. Murakanu M. Kawada M. Noda and S. Matsuura Publ. Astron. Soc. Jpn. 1996 48 L53. 4 S. Sandford Y. Pendleton and L. Allamandola Astrophys. J. 1995 440 697. 5 K. Sellgren L. Luan and M. Werner Astrophys. J. 1990 359 384. 6 K. Uchida K. Sellgren and M. Werner Astrophys. J. L ett. in press. 7 O. Guillois I. Nenner R. Papoular and C. Reynaud Astrophys. J. 1996 464 810. 8 R. Laurejs et al. Astron. Astrophys. 1996 315 L313 and other examples in the same issue. 9 W. Schutte A. Tielens and L. Allamandola Astrophys. J. 1995 415 397. 10 P. Roche and D. Aitken Mon. Not. R. Astron. Soc. 1989 236 485. 11 T. Onaka I. Yamamura T. Tanabeç T. Roellig and L. Yuen Publ.Astron. Soc. Jpn. 1996 48 L59. 12 K. Imamura and Y. Sofue Astron. Astrophys. 1997 319 1. 13 D. Hollenbach and A. Tielens Annu. Rev. Astron. Astrophys. 1997 35 179. 14 O. Goldschmidt and A. Sternberg Astrophys. J. 1995 439 256. 15 F. Bertoldi and B. Draine Astrophys. J. 1996 458 222. 16 A. Tielens M. Meixner P. van der Werf J. Bregman J. Tauber J. Stutzki and D. Rank Science 1993 262 86. 17 J. Bregman K. Larson D. Rank and P. Temi Astrophys. J. 1994 423 326. 18 P. van der Werf J. Stutzky A. Sternberg and A. Krabbe Astron. Astrophys. 1996 313 633. 19 G. Sloan J. Bregman T. Geballe L. Allamandola and C. Woodward Astrophys. J. 1997 474 735. 20 B. Draine Astrophys. J. Suppl. 1978 36 595; K. Lang Astrophysical Formulae 1990. 21 A. Dalgarno and T. Stephens Astrophys.J. L ett. 1970 160 L107. 22 M. Jura Astrophys. J. 1974 191 375. 23 F. Boulanger Conference on Gas»surface and gas»aggregate reactivities 1997 St-MichellœObservatoire ; also personal communication. 24 D. Hollenbach M. Werner and E. Salpeter Astrophys. J. 1971 163 165. 25 M. Jura Astrophys. J. 1975 197 581. 26 C. Arumainayagam and R. Madix Progr. Surf. Sci. 1991 38 1. 27 A. Winkler and K. Rendulik Int. Rev. Phys. Chem. 1992 11 101. 28 S. Lombardo and A. Bell Surf. Sci. Rep. 1991 13 1. 29 Dynamics of Gas»Surface Interactions ed. C. Rettner and M. Ashfold Royal Society of Chemistry Cambridge 1991. 30 C. Rettner D. Auerbach J. Tully and A. Kleyn J. Chem. Phys. 1996 100 13021. 31 R. Masel Principles of Adsorption and Reaction on Solid Surfaces Wiley New York 1996.32 H. Wise and B. Wood Adv. Atom. Mol. Phys. 1963 3 291. 33 A. De Pristo in ref. 29 p. 47. 34 J. Kuppers Surf. Sci. Rep. 1995 22 249. 35 T. Wolkenstein Electronic Processes on Semi-conductor Surfaces during Chemisorption Consultants Bureau New York 1991. 36 J. C. Polanyi and H. Rieley in ref. 29 p. 329. 347 O. Guillois et al. 37 C. Rettner J. Chem. Phys. 1994 101 1529. 38 C. Rettner and D. Auerbach J. Chem. Phys. 1996 104 2732. 39 J. Harris and B. Kasemo Surf. Sci. 1981 105 L281. 40 L. Spitzer Jr. Physical Processes in the Interstellar Medium Wiley New York 1978 p. 125. 41 A. B. King and H. Wise J. Phys. Chem. 1963 67 1163. 42 M. Balooch and D. Olander J. Chem. Phys. 1975 63 4772. 43 S. Dushman Scienti–c Foundations of V acuum Science and T echnology J.Wiley New York 2nd edn. 1966. 44 V. Pirronello Chi Liu L. Shen and G. Vidali Astrophys. J. L ett. 1997 475 L69. 45 H. Sugai S. Yoshida and H. Toyoda Appl. Phys. L ett. 1989 54 1412. 46 S. J. Harris and A. Weiner J. Appl. Phys. 1993 74 1022. 47 R. Gat and J. Angus J. Appl. Phys. 1993 74 5981. 48 C. Lutterloh A. Schenk J. Biener B. Winter and J. Kuppers Surf. Sci. 1994 316 L1039. 49 W. Smith J. Chem. Phys. 1943 11 110. 50 E. Vietzke and V. Philipps Fusion T echnol. 1989 15 108. 51 R. Rye Surf. Sci. 1977 69 653. 52 K. Sancier W. Fredericks and H. Wise J. Chem. Phys. 1959 30 1355. 53 K. Sancier W. Fredericks and H. Wise J. Chem. Phys. 1962 37 854; 860. 54 V. Shatrov V. Gordon A. Ponomarev and V. Talœroze Doklady Akad.Nauka SSSR (Phys. Chem.) 1971 197 400. 55 V. Styrov ZH. ET F Pisma Red. 1972 15 242. 56 R. Kucher I. Opeida and I. Dumbai Khimiya T ver. T opl. 1975 9 53. 57 V. Styrov Izv. Akad. Nauka SSSR 1987 51 524. 58 V. Grankin Zh. Prikl. Spek. 1996 63 444. 59 L. Konig I. Rabin W. Schulze and G. Ertl Science 1996 274 1353. 60 J. M. Mestdagh M. A. Gaveau C. Gee O. Sublemontier and J. P. Visticot Int. Rev. Chem. Phys. 1997 16 215. 61 R. Cavanagh E. Heilweil and J. Stephenson Surf. Sci. 1994 299/300 643. 62 E. Heilweil M. Casassa R. Cavanagh and J. Stephenson Annu. Rev. Phys. Chem. 1989 40 143. 63 K. Koch Y. Yang and F. Luty Phys. Rev. B 1984 29 5840. 64 Y. Yang W. v. d. Osten and F. Luty Phys. Rev. B 1985 32 2724. 65 H. C. Chang and G. Ewing J. Chem. Phys.1990 94 7635. 66 G. Ewing Acc. Chem. Res. 1992 25 292. 67 I. Pocsic Physica A 1993 201 34. 68 R. J. Bell Rep. Progr. Phys. 1972 35 1315; P. Dean Rev. Mod. Phys. 1972 44 127; T. Nakayama K. Yakubo and R. Orbach Rev. Mod. Phys. 1994 66 381. 69 D. Payton and W. Visscher Phys. Rev. 1967 154 802; 156 1032. 70 U. Buchenau C. Pecarroman R. Zorn and B. Frick Phys. Rev. L ett. 1996 77 659. 71 D. van Krevelen Coal Elsevier Amsterdam 3rd edn. 1993. 72 G. Korzeniewski E. Hood and H. Metiu J. V ac. Sci. T echnol. 1982 20 594. 73 Porous Silicon ed. Z. Feng and R. Tsu World Science Singapore 1994; Porous Silicon Science and T echnology ed. J. C. Vial and J. Derrien Springer Berlin 1995. 74 T he Physics of Hydrogenated Amorphous Silicon (II) ed. J. Joannopoulos and G. Lucovski Springer Verlag Berlin 1984 p. 304. 75 I. Solomon R. Wehrsphon J. N. Chalaziel and F. Ozanam in Amorphous and Microcrystalline Semiconductors ICAMS 17 Budapest 1997. 76 A. Radzig and B. Smirnov Reference Data on Atoms Molecules and Ions Springer Series Chem. Phys. 31 Springer Berlin 1985. 77 J. Slevin and W. Sterling Rev. Sci. Instrum. 1981 52 1780. Paper 8/00068I; Received 2nd January 1998

ISSN:1359-6640    

DOI:10.1039/a800068i

出版商:RSC    

年代:1998

数据来源: RSC