Faraday Discuss. Chem. SOC., 1987, 84, 65-73 Laser-induced Fluorescence Study of the F + I, --+ IF + I Reaction Bertrand Girard, Nicholas Billy, Gerard Gouedard and Jacques Vigue" Laboratoire de Spectroscopie Hertzienne de 1' ENS,? 24, Rue Lhomond, 75231 Paris Cedex 05, France' The F+ I2 4 IF+ I reaction has been studied in a crossed-beam experiment. The IF rovibrational distribution has been measured with the laser-induced fluorescence detection technique. Many new and important results concern- ing the energy disposal in this reaction are thus obtained. Moreover, a bimodal rotational distribution is observed in several vibrational levels: this phenomenon seems to be due to migratory collisions. In this paper we present a crossed-beam study of the F+ I2 -+ IF+ I reactive collision.The I F product has been detected by the laser-induced fluorescence (1.i.f.) technique. In order to obtain a well resolved spectrum, we used C.W. single-frequency dye lasers. Consequently very long spectral scans were made. In this way, the relative population of each rovibrational level of I F was measured accurately on a large range of vibrational and rotational levels. In the present report, we describe mainly the results of this study, many details appear in other The F + I2 + IF + I Reaction We have chosen to study this particular reaction as it belongs to a family already well studied, the halogen atom-halogen molecule exchange reactions. For various spectro- scopic reasons (in particular predissociation) I F is the interhalogen for which 1.i.f.detection can be most efficiently applied, therefore the F+ I2 reaction is a natural choice. However, having I2 as the reaction partner (rather than ICl or IBr), creates some experimental problems: the intense I2 fluorescence makes it difficult (or even impossible) to detect several vibrational levels of I F (2 < v < 7); the I2 + F2 chemiluminescence creates an intense background if the fluorine atomic beam contains F2 in an appreciable amount (as is the case in our experiment). On the contrary, the previous knowledge of this reaction is important and this stimulated our interest. reported results of crossed beam studies of this reaction (and there are two papers in this Discussion also). It was then shown that a large fraction (ca. 8 5 % ) of the energy available to products was channelled into internal energy (vibrational energy of IF or excitation energy of the iodine atom in its 2P1,2 state).In 1975, the rate constant of this reaction was measured and found to be large' [ k = (4.3 * 1.1) x lo-'' cm3 s-'1. Several 1.i.f. studies were then made.6-9 The most complete study' gave evidence of a bimodal vibrational distribution with a very intense peak at v = 0 and a slightly inverted distribution peaking at high v ( u = 17). The v = 0 peak was interpreted as indirect evidence of the production of iodine atoms and it was concluded that this was the main branch of this reaction. However, three experiments''-'* were then made to detect these 2P,,2 iodine atoms. In the 1973 Discussion of the Faraday Society, two t Unit6 associke au CNRS UA n"18.65L.I.F. Study of F+ Iz+ IF+ I Table 1. Main parameters of the fluorine and iodine molecular beams (the densities refer to the beam crossing region; their values, as well as the iodine beam velocity, were estimated, but the fluorine beam velocity distribution was measured by the time-of-flight technique) fluorine beam iodine beams supersonic supersonic quasi-eff usive source pressure/mbar 3 60 40 1 source temperature/'C 700 125 120 velocity/ cm s-' 1.02 x lo5 3 x lo4 2.2 x lo4 nozzle diameter/mm 0.2 0.2 0.8 density/g cmF3 1.5 x 10" 1.7 x 1013 6 x lot2 parallel speed ratio Si, 7.6 not measured not measured Although not in good agreement with each other, they all prove that the production of I 'P,,* is at most a very minor branch of this reaction.Experiment a1 The Crossed Beam Apparatus The two molecular beams cross inside a large vacuum chamber and the laser beam illuminates the beam-crossing region. The atomic fluorine beam is supersonic and is produced by thermal dissociation of F2 seeded in argon (10-9OOh) in a nickel oven. This beam passes through a skimmer and a collimator, both differentially pumped. The iodine molecular beam is produced by a pure iodine expansion. In order to have a large iodine density in the beam-crossing region, the beam source is located in the main chamber and is only weakly collimated. Table 1 gives the main characteristics of these two beams. The experiments with the blue laser (A < 495 nm) have been performed with a quasi-effusive iodine beam, also described in table 1.With the supersonic iodine beam, the blue laser excited an intense fluorescence signal that we attribute to iodine molecule dimers (12)2 and this signal was considerably reduced when the supersonic iodine beam was replaced by the quasi-effective beam. Lasers and Wavelength Measurement The lasers are home-made C.W. single-frequency dye lasersI3 pumped by krypton ion lasers. They allow long continuous scans (up to 10cm-'). This is very useful as we have scanned very large spectral ranges. The laser wavenumber is measured by a home-made Michelson type lambdameter,I4 with an accuracy of the order of cm-'. This accurate value of the wavenumber is very useful for the identification of the lines observed. The laser beam is carried from the lasers to the beam machine (in the next room) thanks to a set of multidielectric mirrors and it is weakly focussed on the beam-crossing region by a two-lens system.Several dyes were used as well as different pump laser wavelengths to cover the studied spectral ranges. Table 2 shows this information and the levels that are detected in each spectral region.B. Girard, N. Billy, G. Gouidard and J. Vigue 67 Table 2. The different spectral ranges, the dyes used and the vibrational levels of IF that can be detected wavelength range/nm 470-500 700-789 837-873 dye Coumarin 102 LD 700 HITC vibrational levels 091 (2) 8-16 13-20 v) * c 0 500( 1 I I I 13198 13200 wavenumber/cm-' Fig. 1. A typical laser-induced fluorescence spectrum of IF. The fluorescence signal (in counts per 0.2 s channel) is plotted as a function of the laser wavenumber.The 1.i.f. lines appear on the background due to the I, + F2 chemiluminescence. Fluorescence Detection The fluorescence light is collected by a two-lens system, filtered and detected by a photomultiplier with a bialkali-metal photocathode. The filter is used to reject the laser scattered light and also the anti-Stokes fluorescence emitted by iodine molecules excited by the i.r. laser. The combined effects of the cathode sensitivity and of the filter transmission select the following spectral ranges: 300-600 nm with the i.r. laser, 500- 630 nm with the blue laser. The photomultiplier pulses are treated by a standard photon counting electronics. Experimental Procedure and Data Treatment Typical scanning speed was 5 x lo-' cm-' s-' so that a typical 5 cm-' long scan was recorded in 103s.On the average 20 such scans were made every working day and scanning the i.r. regions took 35 days of experiment. During a scan, several quantities are measured and stored in the memory of a microcomputer, for further processing. The most important quantities are obviously the fluorescence intensity, as given by the number of counts per 0.2 s channel and the laser wavenumber. The 1.i.f. lines appear on the intense background of the chemiluminescence due to the I,+F2 reaction (the most intense unblended 1.i.f. line is ca. lo4 counts s - '68 L.I.F. Study of F+ Iz+ IF+ I Table 3. The different contributions to the available energya term average value r.m.s. deviation /eV molecule-' /eV molecule-' difference of binding energies AD, collision kinetic energy iodine internal energy fluorine atom fine structure energy total energy 1.226 0.111 0.016 0.01 1 1.364 0 0.01 8 0.02 1 0.02 1 0.035 a The internal energy of the iodine beam was estimated using reasonable values of the rotational and vibrational temperatures ( T,,, = 100 K, Tvib = 200 K). For each term, the table gives the average value and its r.m.s.deviation. All quantities are in eVmolecule-'. and the chemiluminescence ca. 4 x lo4 counts s-'). We used the chemiluminescent back- ground as a measure of the product of the two beam intensities and we divided the line intensity by the value of the background intensity. This gave an efficient way of correcting the slow drifts of the iodine beam intensity.Fig. 1 presents a typical spectrum. The positions and intensities of the lines were measured automatically by the computer. After a treatment briefly described in the next part, the population of a level UJ was deduced from the intensities of the 1.i.f. lines starting from this level. Results In this part, we discuss first the problem of the available energy, then we present separately the results obtained with the i.r. laser and with the blue laser. The Energy Available in the F + I2 Collision There are four contributions to this energy: the difference of binding energies AD,, the collision kinetic energy T and the internal energies of the two reactants (rotational and vibrational energies for 12, fine structure for F). Table 3 gives for each quantity its average value and its r.m.s.deviation. There is an important dispersion of the total energy. However, the exact importance of the F2P1,2 state is unknown, as is the reactivity of this state, Map of the Detected Levels We followed each band throughout the spectrum and thus deduced the assignments of all the observed lines (ca. lo4 lines were recorded and analysed). This was necessary as the existing spectro~copy'~ of the B-X system was not sufficiently accurate, especially at high J values, to give the assignments. A new spectroscopy has been derived and will be published soon.I6 Fig. 2 .shows the map of the detected levels. For each vibrational level, the last rotational level observed has an energy close to 1.40 eV. The small fluctuations which appear are perhaps significant, but may be due also to the fact that our ability to detect a low-intensity line depends on many factors (local density of lines, absolute intensity of the bands etc.).The energy limit is well inside our estimate of the total energy. We have not included here the results concerning the u = 0 level discussed later. Saturation of the Excitation Process We have shown" that under the experimental conditions used here the excitation process is in the coherent saturation regime. This saturation regime is well understood andB. Girard, N. Billy, G. Goubdard and J. Vigub .- 69 Fig. 2. The map of the observed levels in the VJ plane. The levels that we have detected are shown by the heavy lines. The energy limiting curves correspond to an internal energy of IF equal to 1.40 eV (curve I ) or 0.46 eV (curve I*), the difference being equal to the excitation energy of the iodine atom.experiments are in good agreement with theory. The most important results are the following. The 1.i.f. signal is a linear function of the laser electric field, i.e. the square root of the laser power, and also of the dipole moment of the excitation transition (in absence of saturation it should be quadratic in these two quantities). The 1.i.f. signal per molecule is not a simple function of the time T spent by the molecule in the laser beam: this signal is linear in T if r T < 1 and proportional to if r T >> 1, where r is the total decay rate of the excited state. This decay is equal to the radiative decay rate r r a d for predissociation free levels: r r a d is equal to about 1.2x lo5 s-’ l 8 and the average time (7) is equal to (2-4) x lop6 s, so that r ( T ) < 1 and it is a reasonable approximation to consider that the 1.i.f.signal is linear in T and that, consequently, this is a measure of the population of the level uJ in the beam-crossing region. However, the time (7) depends on the level uJ through the differential cross- section and the recoil energy. In this paper we have not taken this effect into account. Distribution of the Population over the Rovibrational Levels uJ From the integrated intensities of the 1.i.f. lines, we have deduced the populations. This calculation” requires in particular the knowledge of the excitation transition dipole moment ( u’lDBx ( T ) ~ U ’ ’ ) ~ .We have calculated this quantity (the index J means that both wavefunctions were corrected for centrifugal distortion) using the DBX ( r ) function measured by Trautmann et al.”70 L.I.E Study Of F+I,+IF+I t lo t g O 8 50 100 150 f 19 f l 8 4 l6 + J Fig. 3. The rovibrational distribution deduced from the 1.i.f. line intensities. For each vibrational level v, the population n(v, J ) is plotted as a function of J. The population unit is arbitrary, but the same for all v (including v = 0 in fig. 5 ) . Each tick on the horizontal axis corresponds to a J increment of 10. We have verified that the population thus calculated was independent of the transition used for its detection if we except the following problems. There were three sudden changes of sensitivity during the recording time, probably due to misalignments.By comparing the line intensities that were recorded twice (before and after each change), we easily corrected the effects of these changes. The B state is affected by predissociation. The predissociation threshold was known in the vibrational levels ZJ = 8-10,’’ and we found evidence of the thresholds in the levels ZJ = 4-9 through the reduction of the 1.i.f. signal. Obviously we did not use the intensities of the lines involving predissociated levels for the calculation of populations. Finally the measurements involving the weakest bands are more noisy and may differ from the measurements using intense bands by up to 30%. This is partly due to the imperfect knowledge of the dipole matrix elements, the relative error being larger for the weak bands.B.Girard, N. Billy, G. GouCdard and J. ViguC h a v E: 10 15 20 71 V Fig. 4. The vibrational population n ( u ) as a function of u. Our results are plotted in ( a ) and the curve is a guide for the eye. The results of Trick1 and Wanner' are plotted in ( b ) and they are noticeably different. In both cases the population unit is arbitrary. T X --Y X' x xx/ \ . X x\ f I I I I 1 0 10 20 30 40 50 J Fig. 5. The rotational distribution measured in the u = 0 level. The curve is a Boltzmann distribution fitted to the data.72 L.I.F. Study of F+ I2+ IF+ I Fig. 3 presents the resulting population n (v, J ) as a function of J for all the vibrational levels u = 8 to 19. The rotational distributions presented here are clearly bimodal for several vibrational levels, v = 9 to 15: there is a plateau and a peak at high J.These distributions can also be used to calculate integral properties, as the vibrational popula- tion n( u ) shown in fig. 4. All these results will be discussed in the final part of this paper. The D = 0 Level The results concerning this level are quite different in nature and they are described in detail in ref. (1 ). We have detected 1.i.f. lines arising from the v = 0 state. The populations thus measured are plotted as a function of J in fig. 5. This population is well represented by a Boltzmann distribution with a temperature T,,, = 263 f 8 K. We compared directly the populations of the v = 0 and v = 13 levels, and the population units are the same on fig.3 and 5. After integration over J, we get n( v = 0)/ n( v = 13) = 1.7 f 0.7 a result very different from the one found by Trickl and Wanner9 for the same quantity (103 f 22). Moreover we proved that this signal is an artifact; this was shown directly by modulating the fluorine beam at ca. 1.2 kHz. All the direct collision signals (the chemiluminescence of the F2 + I2 reaction, 1.i.f. signals of the high vibrational levels) were totally modulated, as the fluorine beam, while the 1.i.f. signal of the v=O level was not. This proves that this signal is due to a cold IF vapour which persists after the extinction of the fluorine beam, at least for a few hundred microseconds. The origin of this cold vapour is not clear. It is probably due to surface effects.The direct reactive signal in v = 0-1 has been found to be under the detection limit of our experiment. Discussion First, our results on the v = 0 levels prove that the results previously on the low v levels ( v = 0, 1) are not directly due to reaction products and that these levels represent only a minor branch of this reaction. The main features of the energy disposal in the F+I, reaction are now understood. The first crossed beam experiments"" showed that only a small fraction (ca. 15%) of the available energy appears as recoil energy of the products. It is now clear, thanks to the results concerning the direct detection of I 2P1/2,10-12 that the electronic energy is a negligible fraction of the available energy. This result is contrary to a prediction made by Dinur et uZ.:~' applying the information-theoretic approach of Levine and Bernstein2l to the interpretation of the experimental data of ref.(3) and (4), they predicted a value of the order of 0.24-0.30 for the 2P1/2-2P3/2 branching ratio, which is considerably larger than the observed value. If we neglect completely the electronic energy, the internal energy is shared by vibrational and rotational energies. From our data we may calculate the fractions of energy disposed in vibration and in rotation, and deduce the fraction in translation as the complementary part. These results appear in table 4. They are not very accurate as we ignore the contributions of the levels that have not been detected here. Neverthe- less, the recoil energy is in good agreement with previous result^."^ As shown in fig.4, the vibrational distribution presents a broad peak centred on v = 13. Trickl and Wanner9 also observed a peak at high v, but the shape and locations of the two peaks are noticeably different; it is not possible to establish what is due to a real phenomenon (due for instance to the different collision energies) and what is a consequence of their way of analysing unresolved spectra. Finally, the bimodal rotational distribution deserves a comment. Similar distributions have been observed by J. C . Polanyi and co-workers. The most striking example hasB. Girard, N. Billy, G. Goukdard and J. Viguk Table 4. The vibrational and rotational average energies deduced from the rovibrational distribution presented in fig. 3 (9 s v d 19) and the recoil energy deduced from the knowledge of the average total energy 73 fraction of the type of energy average energy/eV total energy rotational vibrational recoil 0.19 0.94 0.23 0.14 0.69 0.17 been obtained in the H + ICl- HCl+ I reaction.22 The bimodality has been interpreted in this case as a consequence of the existence of two different families of paths on the potential-energy surface:13 a direct path, in which the H atom interacts only with the CI end of the ICI molecule, and a migratory path in which the H atom interacts with the I end first and, after a more or less complex trajectory, reacts with CI.The existence of migratory collisions has been observed by Fletcher and Whitehead24 in their trajectory calculation of the F+I, reaction.However, migratory trajectories were found to be statistically insignificant in another calculation made by Urrecha et al.2s on another type of potential-energy surface. The dependence of the importance of migration on the potential-energy surface is not surprising, but it proves that in the absence of further information on the potential-energy surface, we cannot be conclusive on the origin of the observed bimodality. Obviously , one can imagine other explanations, for instance the existence of several reactive potential-energy surfaces. This work was supported by the GRECO 87 ‘Dynamique des Rkactions Molkculaires’ (CNRS). References 1 2 3 B. Girard, N. Billy, G. Gouedard and J . Vigue, Chem. Phys. Left., 1987, 136, 101. B. Girard, N. Billy, G . Gouedard and J .Vigui, to be published. C. F. Carter, M. R. Levy, K. B. Woodall and R. Grice, Discuss. Faraday Soc., 1973, 55, 381; 385. 4 Y. C. Wong and Y. T. Lee, Discuss. Furadaj. Soc., 1973, 55, 383. 5 E. H . Appelman and M. A. A. Clyne, J. Chem. Soc., Faraday Trans. 1, 1975, 71, 2072. 6 R. J . Donovan, D. P. Fernie, M. A. D. Fluendy, R. M. Glen, A. G. A. Rae and J. R. Wheeler. Chem. 7 J. R. Wheeler, Thesis (University of Edinburgh, 1982). 8 T. Trickl and J . Wanner, J. Chem. fhys., 1981, 74, 6508. 9 T. Trickl and J . Wanner, J. Chem. Phys., 1983, 78, 6091. Phys. Lett., 1980, 69, 472. 10 B. S. Agrawalla, J . P. Singh and D. W. Setser, J. Chem. Phys., 1983, 79, 6416. 11 P. Das, T. Venkitachalam and R. Bersohn, J. Chem. Phys., 1984, 80, 4859. 12 H. Brunet, Ph. Chauvet, M. Mabru and L. Torchin, Chem. Phys. Lett., 1985, 117, 371. 13 F. Biraben and P. Labastie, Opt. Cornmun., 1982, 41, 49. 14 J. Vigue and B. Girard, Rev. Phys. Appl., 1986, 21, 463 and references therein. 15 T. Trickl and J . Wanner, J. Mol. Spectrosc., 1984, 104, 174. 16 G. Gouedard, N . Billy, B. Girard and J . Vigue, to be published. 17 N. Billy, B. Girard, G. Gouedard and J . Vigue, Mol. Phys., 1987, 61, 65. 18 M. A. A. Clyne and I . S. McDermid, .I. Chem. Soc., Faraday Trans. 2, 1978, 74, 1644. 19 M. Trautmann, J. Wanner, S. K. Zhou and C. R. Vidal, J. Chem. Phys., 1985, 82, 693. 20 U. Dinur, R. Kosloff, R. D. Levine and M. J. Berry, Chem. Phys. Lett., 1975, 34, 199. 21 A. Ben Shaul, R. D. Levine and R. B. Bernstein, J. Chem. Phys., 1972, 57, 5427. 22 M. A. Nazar, J. C. Polanyi and W. J . Skrlac, Chem. Phys. Lett., 1974, 29, 473. 23 J. C. Polanyi, J. L. Schreiber and W. J. Skrlac, Discuss. Faraday Soc., 1979, 67, 66. 24 I . W. Fletcher and J . C. Whitehead, J. Chem. Soc., Faraday Trans. 2, 1982, 78, 1165. 25 I . Urrecha, F. Castaiio and J. Iturbe, J. Chem. Soc., Faraday Trans. 2, 1986, 82, 1077. Received 18th May, 1987