年代:1977 |
|
|
Volume 73 issue 1
|
|
221. |
Kinetics of activated chemisorption. Part 3.—Amount and distribution of adsorbate at varying temperatures and pressures |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 73,
Issue 1,
1977,
Page 1943-1950
Chaim Aharoni,
Preview
|
PDF (556KB)
|
|
摘要:
Kinetics of Activated ChemisorptionPart 3.-Amount and Distribution of Adsorbate atVarying Temperatures and PressuresBY CHAIM AHARONI"Department of Chemical Engineering, Technion-Israel Institute of Technology,Haifa, IsraelMOSHE UNGARISHhhclear Research Centre, Nahal Soreq, Yavne, IsraelReceived 13th April, 1977ANDThe heterogeneous-surface model applied in Part 2 to the kinetics, is now applied to the effectsof temperature and pressure on the amount adsorbed. I t predicts isobars with a maximum thatis displaced towards higher temperatures at higher pressures. The low-temperature, pseudo-equilibrium isotherms obey the Freundlich equation. Increase in temperature or decrease inpressure produce desorption followed by readsorption. Temperature-programmed desorption givesone peak when heating is slow and two peaks when heating is rapid.When isotopes are successivelyadsorbed, the isotope adsorbed last is desorbed at lower temperature. The distribution of theadsorbate, according to the regions of various energy on the adsorbent, depends on temperature,pressure and time and is given by a discontinuous function with a maximum.In activated chemisorption the plot of the reciprocal of the rate against the timeis generally S-shaped; the convex part of the plot corresponds to the Banghamequation and the region around the inflection point corresponds to the Elovichequation. These characteristics are satisfied by homogeneous-surface and byheterogeneous-surface models (Parts 1 and 2). ' 9 The heterogeneous-surface modelsuggested in Part 2 comprises sonie specific assumptions concerning the energy ofadsorption and the equilibrium conditions and, therefore, its validity can be testednot only in relation to the kinetics, but also in relation to other aspects of thechemisorption process.In this paper and in Part 4 we examine the implicationsof this model as to the effects of pressure and temperature on the amounts adsorbedand on the heats of adsorption. In Part 5 we discuss the rate, equilibrium, andheat of adsorption on partially covered surfaces.The assumptions of the model arc summarised as follows : (a) The surface hasan enthalpy of adsorption that varies from Ho to H,. We consider it as an arrayof homogeneous regions with an enthalpy H varying within the range d H and anumber of sites nH(H).(b) The activation energy E, at the region characterised byHand at time t, is related to the value of H and to the coverage at that time, yt/nH, byEJRT = clH/RT+ln(gy,/nH +y> (1)where a, g and y are constants andgyJn, % y. (c) Equilibrium is attained at a regionwhen y t becomes equal to yes given b1944 KI NET1 C S 0 F A C TI V A TE D C HEMI SO R P TI 0 Nwhere k, is a pressure dependent constant. At time t the regions with H < Ht(t)are at equilibrium. (a) The regions with H > H,, take part in the adsorption process,and the distribution of the adsorbate is such that they have the same activation energyat any time.where ko is a pressure dependent constant andThe overall rate of adsorption is given bydqldt = k,Nt exp (- E,/RT) (3)N , = rm nH dH.Ht(4)DISTRIBUTION OF THE ADSORBATEThe model differs from other heterogeneous-surface models found in the literature,in that it implies a coverage-energy distribution y(H), basically different from thesite-energy distribution nH(H), and varying with temperature, pressure and time.The models in the literature are generally either " equilibrium models " or " kineticmodels ".In equilibrium models, the sites have different adsorption energies, andthere are no rate effects influencing the distribution of the adsorbate : the adsorbatefills the high energy sites before the low energy ones and y ( H ) is similar to n,(H) athigh Hand zero at low H. In the kinetic models, the sites have different activationenergies and adsorptian is remote from equilibrium : the adsorbate fills the lowerenergy sites before the higher energy ones and y(E) is similar to n(E) at low E andzero at high E.The model we are considering, leads to a coverage against energydistribution, different from the site against energy distribution because it assumesthat the energy of adsorption and the energy of activation vary in the same direction.Coverage is low at the low energy regions (low H and low E), because equilibriumis easily attained, and it is also low at the high energy regions (high H and high E )because adsorption is slow. The coverage is high at some intermediate energyregions.The coverage against energy distribution implied by the model is given by eqn (1)and (2).Eqn (2) gives yes/nH as an explicit function of H for the regions in whichequilibrium is attained. The parameters T and k, that appear in this equation, showthat the distribution varies with temperature and pressure (assuming that k , is afunction of p ) . The distribution for the regions that have not attained equilibriumis given by eqn (1). However, it is convenient to replace Et in this equation by anexpression in which the time appears directly. This is done by noting that eqn (3)becomes eqn ( 5 ) when applied to a single regionCombining eqn (1) and ( 5 ) and rearranging givesdyt/dt = konH exp (- E,/RT). ( 5 )wherey t / h = A exp (- aH/2RT)A = (2k0/g)*tt.Eqn (6) shows that the distribution varies with temperature, pressure and time andalso depends on the parameter a.(ko is assumed to be a function of the pressureand, therefore, A is a function of p and t ) .yes/nH and yt/nH, calculated according to eqn (2) and (6) are plotted against Hfor various k,, A , RT and a (fig. 1). His expressed in the same arbitrary units as RT,and its lower and upper values, Ho = 0 and H, = 30 are chosen arbitrarily. Eachset of values of the parameters k,, A and RT represents a set of conditions p , T and tC . AHARONI AND M . UNGARISH 1945and determines a pair of curves ye,/nH and yr/nH intersecting at H,. The yes/nHcurve is valid for H < H, and the ytinH curve for H > Ht. As expected, coverageis low at low H and at high H, and the yeq/nH and yJnH curves form peaks withmaximum at Ht.For a given a, the position height and width of the peaks dependon the parameters k,, A and RT, i.e., on the experimental conditions p , T and t.When H, is equal to or greater than H,, all the regions are at equilibrium, and thedistribution is given by a Y e q / n H curve extending from Ho to H,,, with a maximum atHm.HFIG. 1.-Distribution of the adsorbate. Ascending curves are plots of yeq/nH, for various RT andke(p). Descending curves are plots of yt/nH for various RT, A(p, t ) and u. u = 2, unless specifiedotherwise. Hand RTare in the same arbitrary units. Ho = 0 and H m = 30 are arbitrarily chosen.The curves are extrapolated into the imaginary region H > Hm, (dotted lines).It follows that at any given temperature and pressure, the surface has a limitednumber of ’‘ active sites ” capable of participating significantly in the adsorptionprocess. The active regions are those that possess Hand E of adequate magnitude.When the pressure and temperature are varied, other regions become the active ones.The amount adsorbed at any set of k,, A and RT is given bywhich becomes at full equilibriumFor the particular site against energy distribution nH constant, q is proportional tothe area under the appropriate yes/nH and yt/nH curves.Eqn (7) can also be solvedanalytically. Combining it with eqn (2) and (6) and integrating gives :q/nHRT = ke[exp (Ht/RT) -exp (HO/RT)I -k(2A/a)[exp (- H,a/2RT) -exp (- Hma/2RT)]. (9)Ht is eliminated from eqn (9) by noting that yes/nH = ytinH at H = Ht.IntroducingH = H, in eqn (2) and (6), equating yeP/nH and yt/nH and rearranging givesHt = [2RT/(a+2)] In (Alk,). (101946 KINETICS OF ACTIVATED CHEMISORPTIONCombining eqn (9) and (10) givesq/nBRT = [(a$2)/a]k~'("t2)1A[2/(a+2)1-k, exp (H,/RT)-( 2 4 ~ ) exp (- aHJ2RT). (1 1)(12)Eqn (8) is solved in a similar way, using eqn (2). It givesqeq/nHRT = k , b p (HdRT) -exp (Ho/RT)I.ISOBARS AND ISOTHERMSThe isobars and isotherms one measures experimentally may refer entirely orpartially to a state of pseudo-equilibrium in which the amount adsorbed is significantlylower than the amount that would have been adsorbed if true equilibrium wereattained. Eqn (3) shows that the rate of adsorption is determined by the two variables0.X0.152 0.10CF0.05C I I , I2 3 4 ' 5RTFIG.2.-Calculated isobars. Plots of ginH against AT for various ke and A and for M = 2, Ho = 0and Hm = 30 [according to eqn (11) and (12)].N, and E, that are both functions of the time. In a run the rate becomes immeasurablysmall either if Nt becomes sufficiently small or if EJRT becomes sufficiently large,the former case corresponds to true equilibrium and the latter to pseudo-equilibrium.For runs at low temperature E,/RT is likely to become very large before Nt hasvanished and pseudo-equilibrium is likely to occur. For runs at high temperatureEJRT remains small all the time and true equilibrium is attainable.In this work we regard experimental isobars and isotherms as referring to seriesof runs at a finite constant time : q(T) at constant p and t , or q(p) at constant T and t.t is sufficiently large so that adsorption in all the runs of the series becomes veryslow before t is attained, either because true equilibrium is approached or becauseEJRT becomes very large.The amount adsorbed during each single run q(p, T, t C. AHARONI AND M. UNGARISH 1947corresponds to the area under the appropriate pair of yes/nH and y,/na curves or tothe area under the appropriate yeQ/nH curve from Ho to H,.Fig. 2 shows isobars for various values of k, and A , with a = 2, Ho = 0 andH, = 30. At the range of temperatures in which Ht < H,, equilibrium is notattained and q is calculated using eqn (11). The range Ht 2 H, correspondsto true equilibrium; eqn (12) is used in that range.The isobars in fig. 2 have amaximum. This property can also be deduced by considering fig. 1 : the areasunder the peaks for given k, and A and for various temperatures increase with thetemperature ; on the other hand the areas under the ye,/nH curves for full equilibriumdecrease with the temperature. Fig. 2 also shows that the maxima are displacedtowards higher values of T when k , is increased or when A is decreased, i.e., whenpressure is raised or when the runs are performed during shorter times.The isobars with the same k, and various A (same p and various t ) finally join acommon curve corresponding to true equilibrium, a state that cannot depend on t.It is noted, however, that the isobars do not join it as soon as they have attained theirmaximum, they have a descending part that does not correspond to true equilibrium.Referring again to fig.1, true equilibrium is established when HI H, so that allthe area under the yt/nH curve is beyond H,, but the isobar starts decreasing when apart of this area is beyond H, and compensates for the increase of the peak withtemperature. It follows from this consideration, that if in an experiment, the resultsshow that q decreases with T, this does not prove definitely that true equilibrium isattained.The experimental isobars reported in the literature are generally in agreement withthe model. They often have a maximum 5-7 and this maximum is often displacedtowards higher temperatures at higher pressures.' However, more complex isobarswith a maximum preceded by a minimum or with two maxima are also known5The model predicts a single maximum for surfaces with a simple site-energy distribu-tion, such as the linear distribution discussed above.Complex distributions, andpresence of presorbed impurities lead to complex isobars (see Part 5). The formof the isobars also depends on the experimental procedure (see below).The isotherms predicted by the model, are always increasing functions whetherthey refer to pseudo-equilibrium or to true equilibrium. A simple analyticalexpression can be derived, for the particular pseudo-equilibrium state in which Ht isremote from Ho and from H,. We assume exp (HJRT) 9 exp (Ho/RT) andexp (- crHt/2RT) >> (- ctHm/2RT) and introduce these assumptions in eqn (9).Theequation corresponding to eqn (1 1) isWe also assumeandwhere k:, k; and A' are constants independent of the pressure, eqn (13) becomes :whereEqn (16) is a pseudo-equilibrium isotherm of Freundlich type, and it is applicableat the conditions in which the isobar is ascending and linear. It is noted that theisotherm for full equilibrium obtained by combining eqn (12) and (141, is one inq = n,RT[(a +2)/.]ky"/'"+ 2)JA[2/("+2)1. (13)k, = k:p (14)(15)(16)A = (2kot/g)* = (2kApt/g)* = A'p*q = KRTp[(a+ I)/(a+2)IK = nH[(ol+ 2)/cr]k;r"/""+ 2)IAW("+ 2)1*1-61948 KINETICS OF ACTIVATED CHEMISORPTIOKwhich q increases linearly with p. This results from the simplifying assumptionsnH NN nH--y, = nH-yeq implied in eqn (2) and (4).The model assumes thatthe total number of available sites decreases as regions with increasing H attainequilibrium and that the number of non-occupied sites in a region remains constantand equal to nH. These simplifying assumptions are useful for treating pseudo-equilibrium and they are also valid for true equilibrium at low pressure, but they arenot valid for true equilibrium at high pressures.The relation between coverage and pressure implied by eqn (I 6), viz., log q cc logpfor constant T and t, is obeyed by various systems. Cimino et aL9 found that thisrelation applies to the adsorption of hydrogen cm zinc oxide. We found that italso applies to the data of Burwell and Taylor lo for the system hydrogen +chromia;the plots of log q against log p for various temperatures and times are linear andparallel.We also obtained linear plots with the data of Low and Taylor for theadsorption of hydrogen on ruthenium +alumina; the measurements in that case werenot at constant pressure and we used the initial pressures. However, the linearityof the plot of log q against logp does not imply that the conditions nH constant and lcoand k, proportional top, are necessarily valid, and, therefore, one cannot use eqn (16)for calculating a without further tests. For estimating the distribution nH, and theother parameters one should simultaneously analyse experimental data concerningthe kinetics,' the amounts adsorbed at equilibrium and pseudo-equilibrium, and theheats of ad~orption.~ The equation for each of these properties contains severalparameters, but the parameters are common for all the properties.ADSORPTION AND DESORPTION INDUCED B Y CHANGES OF TEMPERATURE A N DPRESSUREEach pair of yeq/nH and p,/nH curves in fig.1 corresponds to a run performed atgiven constant p and T during a time t. If the temperature or the pressure is changedduring a run after a certain distribution has established itself, redistribution is inducedinvolving desorption from certain regions and adsorption on other regions. Theseprocesses are rate-dependent, and, consequently, the final distribution is not necessarilysimilar to the one obtained when all the run is performed at the final pressure andtemperature. The isobars depicted in fig.2 refer to an experimental procedure inwhich a fresh clean sample of adsorbent is used for each temperature. They maydiffer when other experimental procedures are used. Variation of the isobar withthe experimental procedure is reported in the literature.'The study of the effects induced by changes of temperature and pressure is aconvenient means for assessing the heterogeneity of surfaces. Results obtained withvarious experimental procedures are found in the literature. We examine theapplicability of the model to some of them.1. THE EXPERIMENTAL WORK OF T A Y L O R . ~ ~ The adsorbent containing ananiount of adsorbate, is submitted to sudden increase of temperature or decrease ofpressure, this results in desorption followed by readsorption.Referring to fig. 1, weconsider the pseudo-equilibrium state defined by RT = 1, k, = and A = 1000.The temperature is suddenly increased to RT = 2. This change induces desorptionof an amount corresponding to the area between the three curves RT = 1, k, = ;RT = 2, k, = On the other hand as the rate ofadsorption increases, the system ceases to be at pseudo-equilibrium. If we neglectthe adsorbate in the area under the curves RT = 2, k, = and RT = 1, A = 1000,we can assume that a new adsorption run has started and expect the new state ofand RT = 1, A = 1000C . AHARONI AND M. UNGARISH 1949pseudo-equilibrium to occur at the time corresponding to A = 1000 starting fromthe moment the change of temperature is made. The amount readsorbed, untilpseudo-equilibrium is re-established is given by the area under the curves RT = 2,k, = If weassume further that the rate of desorption (from low energy sites) is initially greaterthan the rate of adsorption (at higher energy sites) we conclude that the sudden changeof temperature should give desorption followed by readsorption, in agreement withthe experimental findings.However,readsorption is observed only if pressure lowering is performed when the system isremote from pseudo-equilibrium as the rate of readsorption at the lower pressureis necessarily smaller than the rate of adsorption at the original higher pressure.We consider for instance a system at RT = 1, k, = and A = 100 and a reductionof the pressure to k, = The adsorbate corresponding to the area betweenthe curves k , = is removed, but readsorption proceeds in thearea under the curve k, = beyond the curve A = 100.and RT = 2, A = 1000 and it is greater than the amount desorbed.Lowering of the pressure produces a similar effect [see also ref.(13)].and k, =2. THE “ DIFFERENTIAL ISOTOPIC METHOD ” OF KEIER AND ROGINSKI.14Hydrogen and deuterium are successively adsorbed on the same sample of adsorbent;the isotope introduced first is partially desorbed by evacuation, before the secondisotope is introduced. The adsorbent is finally submitted to increasing temperaturesand the composition of the desorbed gases is determined. The results show thatthe gas desorbed at low temperature mainly contains the isotope, adsorbed last, astemperature increases the proportion of the isotope adsorbed first, increases and itbecomes the only gas evolved at high temperature.We refer again to fig. 1, and assume that hydrogen is the isotope introducedfirst and that its distribution is given by the curves RT = 1, k, = and RT = 1,A = 1000.The pressure is subsequently reduced to k, = the distribution ofthe hydrogen that remains on the surface is now defined by the yes/nH curve withk, = and by a yr/nH curve with A > 1000, say A = 1100. Deuterium isadsorbed at k, = its adsorption progresses according to yt/na curves withincreasing A, but the surface available is the one represented by the area borderedby the curves k, = and k, = (unless A becomes > 1100). When heatingis finally applied the ye& curve, initially at RT = 1, k, = is displaced to theright, see for instance the curve RT = 1.1, k, = and the area to the left of thiscurve at each new position is depleted.Considering the areas occupied by hydrogenand by deuterium and considering the positions of the yeq/nH curves at increasing T,one finds that the isotope adsorbed last should be desorbed first in agreement withthe experimental results.3. TEMPERATURE-PROGRAMMED DESORPTION (TPD). An amount of gas ispresorbed, the temperature is raised at a controlled rate, and the amount of gasdesorbed is plotted against the temperature. Many experimental works andtheoretical discussions concerning this technique are found in the literature. 5 *According to the model, the results of TPD are determined by the relation betweenthe rate of heating and the rates of adsorption and desorption.When heating isrelatively rapid, desorption and readsorption occur as in Taylor’s experimentdiscussed above. However, readsorption does not take place as a peak correspondingto a defined temperature (RT = 2 in the example above) ; it is spread over a widerrange of high energy regions and, as time passes and the temperature continues toincrease, the readsorbate is displaced further into regions of increasing energy1950 KINETICS OF ACTIVATED CHEMISORPTIONDesorption of the readsorbate finally occurs when true equilibrium is attained, andthe adsorption capacity starts decreasing with further increase of the temperature.The plot of the amount desorbed against the temperature has two peaks: a lowtemperature peak corresponding to the desorption of a part of the adsorbate thatwas in the region bordered by RT = 1, k, = and RT = 1, A = 1000 and ahigh temperature peak corresponding to that part of the adsorbate that is readsorbedduring the heating process itself and redesorbed at equilibrium.When the rate of heating is low as compared with the rates of adsorption anddesorption, the first peak does not appear because the desorbed gas has enough timeto be readsorbed before it leaves the adsorbent. A single peak results, the high-temperature one.The discussion shows that the number of peaks in TPD is not necessarily equivalentto a number of “adsorption types ”.A site-energy distribution assumed to becontinuous and monotonous may lead to two peaks. More complex results areexpected for surfaces containing presorbed impurities (see Part 5).In conclusion, the present paper shows that the heterogeneous-surface modeldeveloped in order to explain the kinetic isotherm is applicable to a few otherphenomena, and forms the basis of a more comprehensive theory of chemisorption.More applications of the model are examined in Parts 4 and 5.C. Aharoni and M. Ungarish, J.C.S. Faraday I, 1976,72,400.C. Aharoni and M. Ungarish, J.C.S. Faraday Z, 1977,13,456.C . Aharoni and M. Ungarish, J.C.S. Faraday I, to be submitted.C. Aharoni and M. Ungarish, J.C.S. Faraday I, to be submitted.0. Hayward and B. M. W. Trapnell, Chemisorption (Butterworth, London, 2nd edn, 1964),p. 13 and pp. 68-71.J. Haber and F. S . Stone, Trans. Faraday SOC., 1963, 59, 192.A. F. Benton and T. A. White, J. Amer. Chem. Soc., 1930, 52,2325.A. Cimino, E. Cipollini, E. Molinari, G . Liuti and L. Manes, Gazzetta, 1960, 90, 91.lo R. L. Burwell and H. S . Taylor, J. Amer. Chem. Soc., 1936,58, 697.M. J. D. Low and H. A. Taylor, Canad. J. Chem., 1959,31, 544.l2 H. S . Taylor, Adv. Catalysis, 1948, 1, 1.l3 M. J. D. Low, J. Amer. Chem. Soc., 1965, 87,7.l4 N. P. Keier and S. Z . Roginski, Zzvest. Akad. Nairk S.S.S.R., OtdeE Khim. Nauk, 1950, 27, 38.Is R. J. Cvetanovic and Y. Amenomiya, Ado. Catalysis, 1967, 17, 103.M. Smutek, S. Cerny and F. Buzek, Adv. Catalysis, 1975, 24, 343.’ A. Kowalska, Bull. Acad. Polon. Sci., Ser. Sci. Chini., 1974, 22, 17.(PAPER 7/622
ISSN:0300-9599
DOI:10.1039/F19777301943
出版商:RSC
年代:1977
数据来源: RSC
|
222. |
Conductometric investigation of complexes of trifluoroacetic acid with hexaoxa-“crown” ethers |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 73,
Issue 1,
1977,
Page 1951-1957
Nehemia Nae,
Preview
|
PDF (418KB)
|
|
摘要:
Conductometric Investigation of Complexes ofTrifluoroacetic Acid with Hexaoxa-" Crown " EthersBY NEHEMIA NAE AND JOSEPH JAGUR-GRODZINSKI*Department of Plastics Research, The Weizmann Institute of Science, Rehovot, IsraelReceived 29th September, 1976The complexation of trifluoroacetic acid with two hexaoxa-" crown " ethers (18/CR/6 andDCC/A) in l,2-dichloroethane was investigated conductometrically. The equilibrium constants ofcomplex formation are of the same order of magnitude for the two ethers investigated (-1.9 x lo3and -0.9 x lo3 dm3 mol-I, respectively).Triple ions seem to represent the dominant species responsible for the high conductivity of theinvestigated systems. The overall ionization constant was found to be higher for the DCC/A thanfor the 18/CR/6 complex.The conductivity of the investigated solutions was found to be adversely affected by the additionof small amounts of water.This unusual behaviour has been attributed to the fact that the hydrateof the complexed acid cannot participate in the formation of triple ions.It was recently reported that hexaoxa-" crown " ethers act in 1,Zdichloroethane(DCE) as effective proton binding agents.'" Such binding is due to the interactionof the etheral oxygens of the macrocyclic ring with a proton. Apparently,l thoseoxygens which are separated in the macrocyclic ring by 5 atoms (-CH2CH2-O-CH2CH2- segments) are mainly involved :CR A-, H+The ionization of the CRH+, A- complex is enhanced on steric grounds by theseparation of the ions by the segments of the macrocyclic ring which protrude abovethe 0 .. . H . . . 0 plane.Highly conductive solutions of p-toluene sulphonic acid and picric acid in DCEwere thus obtained. la Analysis of their conductivities (in the concentration rangeof 10-4-10-3 mol dm-3) revealed a straight forward dissociation mechanism :K DCRH*, A- + CRH+ + A-.Preliminary experiments conducted in our laboratory with the trifluoroacetic acid(FTAH) complexes indicated that the ionization process may be more complex inthese systems. In the case of a tetra-alkyl-ammonium salt of trifluoroacetic acid,1951952 CONDUCTOMETRIC INVESTIGATIONthe concentration dependence of the conductivities of its solution in acetonitrile hasbeen attributed to triple ion formation, a phenomenon which was first discussed byFuoss and K r a ~ s .~A conductometric investigation of the TFAH-" crown " systems in DCE was,therefore, undertaken and its results are reported in the present paper.EXPERIMENTAL1,4,7,10,13,16-hexaoxa-cyclo-octadecene (18/CR/6) (PCR Inc.) was purified from tracesof water by azeotropic distillation with benzene. The cis-syn-cis isomer of perhydrodibenzo-[b,k] 1,4,7,10,13,16-hexaoxa-cyclo-octadecin (DCC/A) was isolated from the commercialproduct, (Aldrich, technical) and purified as described e1sewhere.lTrifluoroacetic acid (TFAH) (Fluka, CP) was used without further purification.1,2-dichloroethane (DCE) (Frutarom, CP) purified as previously described was driedbefore use on a silica column (Merck, Kieselgel 60).It had IC < ohm-' cm-I and itcontained < 5 xA Hewlett Packard 7602A Research Gas chromatograph with a " poro pack " type Q 33column was used to check for the presence of traces of water. It has been found that inall solutions investigated the concentration of water was <Stock solutions of TFAH and 18/CR/6 in DCE containing 4~ 1W2 mol dmn3 HzOwere used for a controlled addition of small amounts of water.Conductance measurements were performed at a frequency of 1 kHz. A General radio1680-A, Automatic Capacitance Bridge Assembly and a 0.435 cm-I conductivity cell, oil-thermostated at 25.0+O.l0C, were used. Each value was verified by at least two independentexperiments.molar water.mol dm-3.RESULTS AND DISCUSSIONA large increase in conductivity is noted upon addition of DCC/A to a solutionto TFAH in DCE (cf.table 1A). The most dramatic effect on the conductivity ofthe solution is caused by the first equivalent of DCC/A added and the conductivityTABLE 1.-cONDUCTIVITIES OF SOLmONS OF TFAH IN DCEAT 25°C IN THE PRESENCE OF DCC/AA. concentration of TFAH constant (0.4 mmol dm-3)DCC/A/nmoldm-3 none 0.40 0.63 1.26 1.92 2.90 3.50 3.69Jcx 106/ <0.003 2.85 3.46 3.95 4.09 4.16 4.16 4.13ohm-l cm-I $0.03 $0.04 k0.09 k0.06 $0.03 k0.04 +0.05B. concentration of DCC/A constant (3.5 mmol dm-3)TFAH/mol dm-3 0.04 0.08 0.12 0.16 0.32 0.40A/ohm-' cm-z 10.45 10.25 10.66 10.56 10.44 10.40mol-l $0.25 k0.20 $0.22 $0.18 k0.09 kO.11reaches its " plateau " value at a DCC/A TFAH molar ratio of about 5 : 1.Thechanges in conductivity may be attributed to the formation of an easily ionizablecomp1ex.l" The equilibrium constant of its formation is apparently quite high,since a five-fold excess of DCC/A is sufficient to convert all of TFAH into the complexform. It is also evident from data listed in table IB that the equivalent conductance,A, of the complex is virtually constant in the concentration range investigated.However, only a fraction of the complexed acid dissociates into conductive species,as indicated by the relatively low value of AN . NAE AND J . JAGUR-GRODZINSKI 1953An additional complication of the system, which could be due to the dimerizationof TFAH, can be ruled out. Although, TFAH is known 4* to form dimers, theirfraction in DCE, at concentrations used in our experiments, is GO.1 % [calculationsbased on data given in ref.(511. The formation of conductive species must begoverned, therefore, by an overall equilibrium in which two ions are generated bytwo neutral molecules :A+B + R++Q-, K. (1)A system, in which A and B actually refer to two distinctive species (SbCl, and acetylchloride) and eqn (1) represents the reaction mechanism, has recently been describedby Nuyken and Plesch.6When A and B refer to the same species, eqn (2) is obtained instead of eqn (I),2A + R++Q-, K (2)For example, the mechanism of ionization of organic chlorides in SbC13 was described ’in terms of eqn (2). One may suspect, however, that the actual mechanism may bemore complex, since it could be inferred that SbC13, which was used as a solvent,may also be active in formation of intermediates. In any case, in most of the systemsdescribed in the literature 2 * 3 , 8-10 in which equilibrium of the type given in eqn (2)has been implicitly involved, it was due to a two step process.We are referringhere to the generation of a triple ion and of a single ion from two ion pairs. Inall such systems the triple ion formation (which is due to the ion-dipole interactions)is superposed on the normal dissociation step.(3)NamelyA-, C+ + A- + C’, K DorA- + A-, C+ + (A,C)-, Kf.C++A-, C+ -+ (AC,)+, K;.(4’)(4”)For cases in which both eqn (4’) and eqn (4”) must be considered, the term “ bilateraltriple ion formation ” has been coined, while the term “ unilateral triple ion forma-tion ” has been used for systems in which only one of the two eqn (4) is of importance.8bFor a triple ion formation based on conjugation of a parent acid with its anion, theterm “ homoconjugation ” has been coined by Kolthoff.’O Coetzee discussed thehomoconjugation of various acids in acetonitrile.In many systems, both the triple ion and the normal dissociation mode must beconsidered simultaneously in order to account for the actual concentrations of theconductive species.Mathematical treatments dealing with such situations have beenworked out by several author^.^. *, On the other hand, for a system in which theconcentration of ionic species due to the normal mode of dissociation is negligiblysmall, as compared with that of the concentration of species generated according toeqn (4), the mathematical treatment discussed by Nuyken and Plesch may beapplied if no additional equilibria are involved.The last condition is not fulfilledin our case. On top of the ionogenic equilibria, we have to consider a complexationreaction, which preceeds the ionization steps.The following scheme seems to account for OUT experimental findings :TFAH + CR + C, Kc ( 5 )(6)(7)C + TFA- + CRH+, KDTFA-+ C + [(TFA),CRH]-, K1954 CONDUCTOMETRIC INVESTIGATIONwhere the complex, C, is the ion pair, TFA-, CRH*, of a structure similar to thoseof complexed acids described in ref. (la). The other mode of formation of tripleions, based on an interaction of the complex C, with the " crowned " proton, CRH+,may be ruled out on steric grounds.Let us now denote as H , the initial concentration of trifluoroacetic acid and as2d the concentration of ionic species at equilibrium.Hence,2d = [CRH"] + [TFA-] + [ {(TFA),CRH)-].(8)However, [TFA-] in eqn (8) can be neglected, if [TFA-] < [{(TFA),CRH)-I. Thiscondition is fulfilled when &[c] & 1 (it follows from eqn (7) that [((TFA),CRH)-I/[TFA-] = KT[C]).Accordingly, under such conditions :It follows from eqn (5)-(7) and (9) that :d N [CRH"] N [{(TFA)zCRH)-]. (9)andIt also follows from eqn (6) and (7) that[C]/d = (Kc&)-+.Let us further denote the experimentally determined conductivity, K , divided by theterm d as A'.Thus,A' = K X 103/d.(1 3)Note that the term given by eqn (13) represents the equivalent conductance of thatfraction of the complexed acid which has been dissociated into ions [cf. discussionon page 250 of ref. (12)].It has been pointed out in the previous section, that Kc must be fairly large.Hence, it cannot be accurately determined from experiments in which [CRIo 9 Ho.The experimental set up was chosen, therefore, so as to have [CR], = Ho.An algebraic rearrangement of eqn (10)-(13) yields, under such conditions :For Kc CQ, the second term in eqn (14) vanishes and it reduces toNote that eqn (15) is identical to that derived by Nuyken and Plesch for theircase I1 [see our eqn (2)]. In their notation K is identical to our K d T , Co with Hoand AT with our A' (they neglect the dependence of A, on concentration).The results of conductivity measurements for the TFAH+DCC/A and for theTFAH+18/CR/6 systems (at Ho = [CR],) are tabulated in table 2.Values of A'have been calculated by computer iteration procedure, analogous to Fuoss treatment,'using the basic Onsager relationship :A' = AO-(UIAO$C(~) JdN. NAE AND J . JAGUR-GRODZINSKI 1955Substitution of d by IC x 103/A' yields :where p = alA, fa,.A' = A, +/3 Ki 103 [ A ~ - ~ ~ * 1 0 ~ ( A , - ~ ~ * 1 0 ~ ( . . .(. . ,)ei.. .)-i)-i]-*TABLE 2.-cONDUCTIVITES AT 25°C OF DCE SOLUTIONS CONTAINING EQUAL MOLAR AMOUNTSOF TFAH AND " CROWN " ETHERA. " crown "-ether : DCC/ATFAH -k DCC/AIcx 106/ohm-' cm-' 0.509k0.005 1.13+0.02 2.85k0.03 4.98k0.04 8.8950.02/mmol dm-3 0.115 0.2 0.4 0.6 1 .oB." crown "-ether : 18/CR/6TFAH + 18/CR/6K X 106/ohm-1 cm-l 0.495f0.003 0.711+0.006 0.949k0.006 2.35k0.03 7.375k0.03/mmol dm-3 0.24 0.32 0.40 0.80 2.0A, = 42 was used in the computations. This value is based on the assumption,2,' = 20. [The value of 21 has been giveninref. (la) as A$ of the "crowned"proton1.Thus, eqn (14) can be solved. The resulting plots of H,(A'/Ic)~ against (A'/K)-i areshown in fig. 1. Values of Kc and KDKT calcuated from these plots are as follows :system &/dm3 mol-1 KDKTDCC/A+ TFAH 869 3.918/CR/6+TFAH 1914 0.0481 1 I0.15- ---I I I5 10 I5( K * / K * X 10-8) X lo3FIG. 1.-Plot of Ho(A'/~)*lO-~ against (A'/K)-* x 106 for TFAH solutions in DCE containingequivalent amounts of DDC.A or 18/CR/6.S TFAH+18/CR/6. Concentrations from 0.24 to2.0 mmol dm-3, 0 TFAH+DCC/A.The linearity of plots shown in fig. 1 indicates that even at concentrations as lowas niol dm-3, the assumption that [TFA-] 4 [(TFA),CRH]-, underlying thederivation of eqn (12), remains valid. Derivation of the dissociation constant, K,,Concentrations from 0.115 to 1.0 mmol dm-31956 CONDUCTOMETRIC INVESTIGATIONfrom conductance measurements in such systems is obviously impractical. However,an estimate of KD w 10-8-10-9 mol dm-3 could be made by considering the polar-K d T values in conjunction with this estimated value of K,, yield KT x 107-10 ' dm3mol-'. Such high values of KT are apparently due to the low dielectric constant ofthe medium and to the weak interactions between a free TFA- anion and the DCEsolvent.In fact, for HF in water,s KT is only 0.33 dm3 mol-1 and the homo-conjugation constants for various organic acids in acetonitrile l1 are only - lo4 dm3mol-l.However, the low dielectric constant and the weak anion-solvent interactionscannot explain the entire phenomenon, since in the same concentration range,complexes and salts of other acids do not form triple ions in DCE.l. l4The high localization of the negative charge in the anion and its small size, whichmake possible a close approach to two TFA-anions from both sides of the protonated" crown " molecules, may play a decisive role in our case.The intrinsic tendency of the TFA- to form triple ions is indicated by the factthat it is not restricted to rhe present system only.2 Moreover, it apparently alsooccurs in DCE in the absence of " crown '' molecules ; foi 10-50 mmol dm-' TFAH,conductivities increase from (5 to 25) x These values are too lowto allow an exact calculation of KDKT, but an estimate of KDKT - (3-4) x canbe made.On the other hand, the most unusual effect of water on the conductivities of theinvestigated solutions (see following section) indicates that the easy accessibility ofthe complexed acid by the TFA- anions may be of critical importance for its abilityto conjugate with them.Let us examine at this stage the consistency of the derived values of equilibriumconstants with our assumption KT[C] 9 1.For the concentration range used in ourexperiments, the calculated Kc and KDKT values yield [C] > 5 x mol dm-3.KT values have been estimated as at least lo7 dm3 mol-'.ohm-' cm-l.Hence KT[c] > 500 % 1.EFFECT OF WATER ON THE CONDUCTIVITY OF THE COMPLEXAs may be seen from data in table 3, the addition of water to the TFAH+ " crown "complex in DCE causes a decrease in the conductivity of the solution.A similardecrease is also observed when methanol or ethanol is added instead of water. Onthe other hand, when water is added to TFAH solution in DCE, which does notTABLE 3.-cONDUCTIVITY OF TFAH SOLUTIONS IN DCE AT 25°C IN THE PRESENCE OF SMALLAMOUNTS OF WATERHzO x lOz/mol dm-3 <0.01 1 2 3 4 -1TFA 0.011 - 0.015 - 0.018 0.023/2.2 x lo-* mol dm-3TFA+18/CR/6 K X 106/ohm-' cm-' 1.31 6.52 5.66 4.93 4.34 3.35/0.2 x mol dm-3 k0.02 F0.03 k0.03 k0.02 k0.04 (1.86)** water saturated solutioncontain a " crown " ether, the conductivity of such a solution is increased. Formationof 1 : 1 and 2 : 1 hydrates of TFAH in CCl, has been demonstrated by i.r.spectro-scopyAccordingly, the following explanation may be proposed for this most unusualeffect of water on the conductivity of TFAH " crown " complexes in DCE.and a similar behaviour may be expected in DCEN. NAE AND .I. JAGUR-GRODZINSKI 1957Addition of water to the TFA-, (CRH)+ complex leads to the formation of ahydrate,KH 0H f , TFA+HzO + HzO, Hf, TFA-0(16)(where denotes a macrocyclic ring). Evidently, approach of a second TFA- anionto the hydrate foimed [cf.eqn (16)] will be blocked by the attached H,O molecule.Other directions of approach are blocked by segments of the macrocyclic ring.Hence the triple ion cannot be formed.Of course, the hydrated complex may dissociate into free ions ;KbHzO, (CRH)+, TFA- f HzO, (CRH)++TFA- or (CRH)-+(TFA)-, HzO. (17)Although Kb may be much larger than KD, K& evidently remains much smallerthan KDKT. This means that loss in conductive species due to restrictions imposedon triple ion formation is not compensated by the more extensive dissociation of thehydrated complex. As a net result, one observes the unusual decrease in conductivitywhen water or alcohol are added to a solution of an electrolyte in a low dielectricconstant aprotic solvent.(a) N. Nae and J. Jagur-Grodzinski, J. Amer. Chem. Soc., 1977, 99,498 ; (b) E. Shchori andJ. Jagur-Grodzinski, J. Amer. Chem. Soc., 1972, 94, 7957.(a) R. M. Fuoss and C . A. Kraus, J. Amer. Chem. SOC., 1933,55,2387 ; (b) C. A. Kraus andR. M. Fuoss, J. Amer. Chem. SOC., 1933, 55, 21, 476.F. Thyrion and D. Decroocq, Compt. rend., 1965, 260,2797.T. L. Stevens, Ph.D. Dissertation (University of Oklahoma, 1968).0. Nuyken and P. H. Plesch, Chem. and Ind., 1973, 379. ’ A. G. Davies and E. C. Baughan, J. Chem. SOC., 1961, 1711.(a) C. B. Wooster, J. Amer. Chem. Soc., 1938, 60, 1609 ; (b) 1937, 59, 377.C. M. French and I. G. Roe, Trans. Faraday SOC., 1953, 49, 314.lo I. M. Kolthoff and M. K. Chantooni, J. Amer. Chem. SOC., 1965,87,1004.J. F. Coetzee, in Ionic Reactions in Acetonitrile, Progress in Physical Organic Chemistry, ed.Stzeitweiser and R. W. Taft (Interscience, 1976), vol 4, p. 54, 76.ia E. Shchori and J. Jagur-Grodzinski, Israel J. Chem., 1973, 11, 243.l3 R. M. Fuoss and F. Accascina, Electrolytic Conductance (Interscience, 1959).l4 (a) L. M. Tucker and C. A. Kraus, J. Amer. Chem. SOC., 1947,69,454; (b) J. T. Denison andl5 J. De Villepin, A. Lautie and M. L. Josien, Annales Chim., 1966, 365.* G. A. Forcier and J. W. Olver, Electrochim. Acta, 1970, 15, 1609.J. B. Ramsey, J. Amer. Chem. SOC., 1955, 77, 2615.PAPER 611828
ISSN:0300-9599
DOI:10.1039/F19777301951
出版商:RSC
年代:1977
数据来源: RSC
|
223. |
Luminescence following laser excitation of trapped electrons in mixed-solute aqueous glasses |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 73,
Issue 1,
1977,
Page 1958-1971
Tuan Q. Nguyen,
Preview
|
PDF (1039KB)
|
|
摘要:
Luminescence Following Laser Excitation of Trapped Electronsin Mixed-Solute Aqueous GlassesBY TUAN Q. NGUYEN AND DAVID c. WALKER*Chemistry Department, University of British Columbia, Vancouver, Canada V6T 1 W5Received 28th February, 1977Based on measurements of the luminescence produced when infrared-absorbing electron states(ei;) at 77 K decay, it is contended that this electron state is normally to be expected in irradiatedaqueous media and is not anomalous to those few specific solutes in whose glasses it has been found.Rather, it is destroyed, or its formation is inhibited, by the majority of solutes which are commonlyused to make aqueous glasses. This conclusion comes from studies of mixed-solute glasses,specifically using MgClz glasses with CaCl,, Mg(C10& or ethylene glycol added as second solutes.Though all three second solutes inhibit the luminescence from the MgClz glass, quite differentmechanisms seem to be involved in each case, as shown by the time-dependence of the luminescenceintensity.It is suggested that el; converts spontaneously to e,iswhen ethylene glycol (or 2-propanol)is present because it inhibits the luminescent process without destroying eE. These species wereproduced in the experiments by photoexcitation at 694 nrn of stable trapped electrons formed byy-radiolysis at 77 K.Buxton, Gillis and Klassen ’ recently discovered the formation of an infrared-absorbing electron state (ei,) during pulse radiolysis in a few specific aqueous systemsat -76 K. This electron state was short-lived at 76 K and, in general, it did nottransform to the stable visible-absorbing trapped electron state ( e d as it disappeared.Instead it gave rise to luminescence centred around 410 nm.ei, was found in onlythree out of ten of the aqueous glasses that were investigated, specifically in 9.5 moldm-3 LiCl, 2.5 (and 4) mol dm-3 MgC12 and in 7.2 mol dm-3 ethylene glycol (sameas 50 : 50, ethylene glycol : water, by volume) D20 glasses. No ei, was detectablein D,O glasses made from 4moldm-3 Mg(CIO&, Mg(CH3C02), and CaCl,8 mol dm-3 NaClO,, 7 mol dm-3 HCOONa and NaOD, 10 mol dm-3 KOD and5 mol dm-3 KzC03, whereas all these media readily produce stable trapped electronsabsorbing in the range 500-600 nm when irradiated at low temperatures.Theinfrared-absorbing species was also found in pure D,O crystals at 76 K.’found that ei, was produced when trappedelectrons in LiCl and MgC12 glasses were bleached by a pulse of light at 694 m. Bythis method too, the formation of ei, was not observable (at least 10-fold smaller)in 10 mol15m-~ OH- and 5 mol dm-3 K2C03 glasses, so an analogous specificity wasfound with regard to the glass-forming electrolyte used.in theirpost-pulse loss of absorption in the red tail and in their photobleaching character-istics 6* ’ as the solute is interchanged ; but in general, the glass-forming solute seemsto be inert and the trap of the electron can be regarded as being essentially water-Is ei, specific to a few specific solutes (like LiCl and MgCl’) and to ice,or is it normal to water itself, the majority of common solutes used either destroyingit by chemical reaction or preventing its formation by structural changes to theglasses ?1958Subsequently Gillis and WalkerThere are some differences in the yield of e;, and in their A,,, values,’T.Q. NGUYEN AND D. C. WALKER 1959To resolve this problem we have studied ei, in mixed-solute aqueous glasses.Most of the results we report here used a solution of 3.1 mol dm-3 MgCl, as themediuin in which ei, was formed. This solution was then mixed, in various propor-tions with 3.1 mol ~ l m - ~ CaCI, o r 4.1 mol dm-3 Mg(C104), solutions to make mixed-solute glasses in which the cation or anion only was changed without a significantchange in the total concentration of ions.Other mixtures reported consisted of3.1 mol dm-3 MgC1, with various concentrations of ethylene glycol or 2-propanol.Ethylene glycol glasses were found t o support ei,, as seen through their absorptionband, but did not produce luminescence as they decayed,, so the contrast betweenCaCl,, Mg(C104), and ethylene glycol as second solutes in MgCl, glasses shouldbe revealing.EXPERIMENTALThe experimental method is analogous to that discussed previ~usly.~ Trapped electrons,which were produced by y-radiolysis of the glasses at 77 K, were subsequently illuminatedby individual pulses of 694nm light from a Q-switched ruby laser, and the resultingluminescence was monitored at 450 nm over the time-scalehad shown that the rate of change in concentration of ei, could be observed by measuring theintensity of the broad emission band centred at -410 nm, because this emission is producedas a result of ei, undergoing spur decay reactions.The rate of the luminescent step seemsto be governed by the rate at which e; tunnels to a geminate reactante2.Stock solutions of 3.1 MgCl,, 3.1 CaCI, and 4.1 mol dm-3 Mg(CI04), were made fromreagent grade chemicals and doubly distilled water. To some MgC12 solutions neat reagentgrade ethylene glycol or 2-propanol was added. All samples were deoxygenated by thoroughbubbling with Ar gas prior to being frozen at 77K when plunged into liquid nitrogen.Laser-excitation experiments were performed on two types of samples. One was containedin 1 cm2 cross-section, transparent, high-purity quartz cells ; the other type of sampleconsisted of a number of approximately spherical pellets of the glass obtained by depositingpreviously deoxygenated droplets of solution directly into liquid nitrogen.The essentialdifference between these two samples lay in the fact that in the former case the quartz ceilitself wasy-irradiated and then exposed to laser pulses which resulted in a weak but observableluminescence ; whereas in the latter case the pellets were y-irradiated in one Dewar vesselthen transferred to an unirradiated container for laser-excitation. In the latter procedurethere was no background lun~inescence, so this method was invariably used with systemswhere the luminescence from e, was weak. Samples (either in quartz cells or as pellets)were y-irradiated in the dark at 77 K with a dose sufficient to produce a concentration ofev;, corresponding to a predetermined absorbance at 694 nm, then transferred in the dark tothe laser photolysis Dewar.Only nearly crack-free glasses were used.Since the yield and spectra of e;, in the various glasses differed it was necessary to firstmeasure the appropriate absorption spectra and the absorbance at 694 nm per unit dose.These were taken with y-irradiated samples in quartz cells contained in a Dewar vessel withoptical windows which was placed in the absorption compartment of a Cary 14 spectro-pha tometer.A Korad K1 ruby laser system, which was Q-switched with a Pockels cell, providedindividual 20 ns pulses (FWHM) of light at 694 nm to bleach the samples of trapped electrons.The laser pulse was attenuated for most experiments by a factor of 100 or 300 using CuSO4solutions.An optical train, similar to that used previ~usly,~ picked up some of theluminescence which was emitted perpendicularly to the laser beam as the sample was excitedby a laser pulse.Variations in the laser light intensity from pulse-to-pulse were recorded on a Hadronpower meter by deflecting a small fraction of the light pulse from a quartz plate ahead ofthe attenuators. All data were appropriately normalised to a constant light flux using thesemeasurements. Except during the determination of the spectrum, an interference-typefilter transmitting at 450 nm with a FWHM of 50 nm was used for wavelength selection infront of the detector. This filter was backed by a 694 nm 100 % reflecting mirror plus ato 2 s.The previous stud1960 LASER EXCITATION OF TRAPPED ELECTRONScorning colour glass # 5-56 filter to eliminate scattered laser light. 450 nni was used becausethe emission has -70 % of its maximum intensity at that wavelength and it correspondsclosely to the maximum sensitivity of the whole detection train. A IP28 photomultiplierwas used as the detector and a Tektronix 7904 oscilloscope with 7A16 amplifier displayedthe luminescence signal as a function of time. A wide range of anode loads (50 to lo6 a)were used with the photomultiplier to adjust the sensitivity, care being taken to ensure thatthe response time of the detection system was always <* of one division of the selectedoscilloscope sweep speed.A cell holder ensured that one sample could be exchanged with the next without alteringthe efficiency of the optical train, thereby allowing a direct comparison of luminescenceintensities.No attempt was made to evaluate the quantum yield for luminescence. Heliumgas was gently bubbled through the top of the liquid nitrogen in the photolysis Dewar tominimise bubble formation which could cause light scattering, so the temperature wasprobable one or two degrees <77 K.Emission spectra were measured between 350 and 550 nm on time scales ranging from100 ns to 10 ms and were corrected for laser pulse variation by use of a reference emissionbeam at 450 nm. Emission perpendicularly to the left of the laser beam was directed ontothe detection system described above and set at 450 nm.Emission to the right was collectedby a lens system, directed through a Bausch and Lomb 4 m monochromator and thence to aHamamatsu R213 photomultiplier, in an arrangement similar to that previously calibratedfor its overall spectral ~ensitivity.~ Both the reference and variable-wavelength signals wererecorded concurrently on a dual-trace amplifier of the oscilloscope (7A26) at several sweepspeeds and the spectrum composed for various times from the corrected signals, normalisedthrough the reference signal.RESULTSABSORPTION SPECTRA OF e,;, I N VARIOUS GLASSESFig. I shows the stable absorption spectra of various glasses made from single- ormixed-solutes when y-irradiated at 77 K to various doses to give the same absorbancei 0 .G 0.6 I0.5 i01 .o300 400 500 600 6t4 800 3&J 400 500 600 6b4 800wavelength/nmFIG. 1.-Absorption spectra of stable trapped electrons (e&), and of CI, (at 350 nm), in variousy-irradiated aqueous glasses at -76 K. 1. 3.1 mol dn1-3 MgCIZ, 2. 4.1 rnol dm-3 Mg(C10&, 3.50 : 50 mixture by volume of 1 and 2,4. 3.1 mol dm-3 CaCI2, 5.50 : 50 mixture by volume of 1 and 4,6.2.5 mol Samples were irradiated to differentdoses so as to give the same absorbance at 694nm. (Only curve 6 was corrected for scatter andreflection by the cell using the bleached sample as a blank, the rest used air as reference in thespectrophotorneter).MgCI2 + 1.9 mol dm-3 ethylene glycol mixtureT.Q. NGUYEN A N D D . C. WALKER 1961at 694 nm. The bands peaking at 350 nm arise from ClT and those in the range540 to 630 arise from trapped electrons, eGs. It is evident from the 350 nm peaks of1 and 4 that the radiation yields of e;, vary as well as the 3.,,, for e;, for single-soluteglasses, as is already known.' Mixtures of the solutes, however, produce glasseswhose yield of e& and band maxima change monotonically with composition.Table 1 gives the pertinent data for several of the glasses used.Whereas the variation in yield can readily be compensated for by adjustment ofthe radiation dose, the small changes in absorption band position raise a slight problemfor these experiments in which the photolysis light is of a fixed wavelength.Of thevarious possible normalisation procedures, we have chosen to make a comparison ofi he luminescence from samples which all had the same absorbance at 694 nm due toev&.TABLE 1 .-POSITION OF ABSORPTION MAXIMA AND THE PRODUCT OF G VALUE AND EXTINCTIONCOEFFICIENT AT 694nm AND AT THE BAND MAXLMA FOR e;, IN SEVERAL AQUEOUS GLASSES 1MgClz 3.1 630 7.2 5.980 % MgC12+20 % CaCI, 2.5 + 0.6 620 13.2 8.650 % MgCL + 50 % CaClz 1.5f1.5 600 18.2 10.420 % MgCL+ 80 % CaCl, 0.6+2.5 590 24.1 14.43.1 590 22.7 13.5concentrationsolutes imol dm-3 Ama,&t20/nm 10-3 G E ~ s x . ~ ~ 10-3 G ~ s 9 4 - p t80 % MgC12+20 % Mg(C104)Z 2.5+0.8 590 13.6 8.150 7; MgClz+5O % Mg(C104)Z 1.5f2.0 5 80 20.0 9.520 % MgClzf80 % Mg(C104)Z 0.6f3.3 550 27.7 7.7Mg(C104)z 4.1 540 24.1 4.5 - 5.9 90 % MgClzf10 % EG* 2.8 f0.7 -75 % MgClzf25 % EG 2.3 + 1.8 630 12.2 7.760 % MgC1~+40 % EG 1.9+2.9 630 15.0 9.9* EG means ethylene glycol; t p is the density of the glass ; $ measured - 15 min afterirradiation.Since we are comparing the effects of two solutes, one of which produces ei; byphotolysis and the other normally does not, we have to try to normalise them bysetting conditions under which they have an equal opportunity to produce the samenumber of ei;.We selected a fixed absorbance at 694 nm as the best such conditionfor the following reasons : (i) it ensures that the same number of laser photons areabsorbed by each sample ; (ii) during the laser pulse there will be the same absorbancegradient down the cell as a result of bleaching,'O so that corrections due to a changinggeometry of the effective luminescent light source will not be necessary; and (iii) therelative number of e;i, and their expected relative quantum yield for photo-inducedelectron release at 694 nm should have partially compensatory effects, therebytending to equalise the number of ei; which could be produced per absorbed photon.[This latter point can be seen for pure MgCl, and Mg(C10,)' glasses by referenceto fig.1. For doses corresponding to equal absorbance at 694 nm there will befev+er e,i, in MgCl, since cbg4 is larger in that glass. But for MgCl, the expectationis that its extinction coefficient at 6 9 4 m arises from a higher proportion of bound-free(as compared with bound-bound) character, thereby producing a higher quantumyield for photo-release at that wavelength, because 694 nm is nearer to A,,, on thered side.]'This normalisation procedure will only affect the direct comparison of theluminescence intensities, and then with an error which is unlikely to exceed 50 %1962 LASER EXCITATION OF TRAPPED ELECTRONSand it will not affect the fractional change with time, which, it transpires, forms thebasis for some rather prominent differences and the overall conclusions.EMISSION SPECTRAFig.2 shows typical dual-trace oscillograms with the reference at 450 nm as thelower trace and the variable wavelength as the upper trace, both at the same sweepspeed. The ratio of variable/reference signals when plotted against the wavelengthof the variable gives a normalised response spectrum, normalised for fluctuations inthe laser pulse intensity and for changes in the absorbance of the sample from pulse350 550wavelengthlnmFIG.2.-Observed emission as a function of wavelength at various times after laser-excitation ofe,i, in 9.5 mol dm-3 LiCl glasses. 0,100 ns ; @, 20 ps ; A, 10 ms. (This spectrum was norrnalisedat 450 n m for variations in the laser intensity from pulse to pulse ; but it has not been corrected forthe spectral sensitivity of the detection system so is not an emission spectrum.)Insert : Shows oscilloscope traces of the emission at the reference wavelength 450 nm and at475 nm, as measured concurrently. Vertical : 450 nm at 50 mV per div, 475 nm at 10 niV per div ;horizontal : 20 ps per div.to pulse, but not corrected for the spectral sensitivity of the optical detection systemon the variable side, which is similar to that given elsewhere but - 10 nm to theblue due to fewer light filters.Spectra obtained for 9.5 mol dm-3 LiCl glasses gave the data points shown infig. 2 for times of 100 ns, 20 ,us and 10 ms.These fit relatively well to the spectrumfound previously for 9.5 mol dm-3 LiCl glasses.2* There was no discernible changeduring the time period from 100 ns to 10 ms. This means that the " bulge " and thT . Q . NGUYEN AND D. C. WALKER 1963I oc l i t regions of the luminescence3 have similar spectra, which in turn implies thatthey have a common origin, although different electron states could be involved.MgC12 +CaC12 MIXTURES3.1 mol d ~ n - ~ stock solutions of MgCl, and CaC1, were appropriately combinedto give mixtures consisting of 0,20,50,80 and 100 % CaCl, solution.Each mixturegave a good transparent glass at 77 K. The intensity of luminescence from laser-excitation of e,i, is these glasses was quite different as was its change with time.(c) (4FIG. 3.-Oscilloscope traces showing the time-dependence of the luminescence at 450 nm and theinfrared absorption from laser-excited trapped electrons in various glasses at 76 K. (a) Luminescencefrom : 1.3.1 rnol dm-3 MgC1, glass, 2.3.1 mol dm-3 CaCl, glass. Vertical, 50 mV per div, horizontal,SO ns per div. (b) Luminescence from : 1. 3.1 mol dm-3 MgClz glass, 2.2.5 mol dm-3 MgCoz +0.8 no1 dm-3 Mg (C10& glass, 3. 0.6 mol dm-3 MgC12+3.3 mol dm-3(C104)2 glass. Vertical,50 mV per div, horizontal, 20 ns per div. Trace 4 shows the ruby laser pulse on the same time scale. (c)Luminescence from : 1. 3.1 rnol dm-3 MgCl, glass, 2. 1.6 rnol dm-3 MgCI2+2.1 rnol dm-3 Mg(C104),, 3. 2.5 rnol dm-3 MgClz + 1.9 mol ethylene glycol. Vertical, 50 mV per div, horizontal1 ps per div. (d) Absorption at 1340-1780 nm : traces 1 to 3 are for the same mixtures as 1 to 3 in (c).Vertical, 20 mV per div (with I. = 1.5 V), horizontal 1 ps per div.Oscilloscope traces for 100 % MgCl, and 100 % CaCI, are compared in fig. 3(a)over the first 400 ns. Emission intensities during and immediately after the pulse aredisplayed in fig.3(b), this time for 100 % MgClz, 80 % MgCI, and 20 % MgC12 inMg(C10J2 glasses. They show that the luminescence intensity reaches its maximumvalue after the laser pulse. The fluorescence lifetime can therefore be estimatedfrom the luminescence build-up time to be of the order of 20 ns. For times longerthan - lo-’ s, the rate controlling step is evidently not the fluorescent process but isgoverned, instead, by the rate of formation of the fluorescent species. On the slowertraces, such as fig. 3(a), a very different change of intensity with time is observed indifferent glasses, due to their influence on this rate of formation1964 LASER EXCITATION OF TRAPPED ELECTRONSI 0 ' -7 -5 -3 -1 0 +llog (tiss)(b)Fig. 4.-See caption oppositeT. Q.NGUYEN AND D. C . WALKER 1965By taking oscilloscope traces at different, but overlapping, sweep speeds, thechange of light intensity with time can be evaluated covering several decades of time.For convenience these are presented as log-log plots. When the intensities werequite small the pellet method was used so that corrections arising from luminescencefrom cold irradiated quartz were unnecessary. Fig. 4(a) shows log I against log tplots for 100 % MgC1, solution and three mixtures with CaCl,. As before the100 % MgClz showed a positive " bulge " for times < s but otherwise a goodstraight line of slope - 1.0+0.05. Similar linear log-log plots of slope - 1 werefound at times > s for the mixtures, though the intensity at any time after lo-' swas much lower the richer the mixture was in CaCl,.(4 (b)FIG.5.-(a) Normalised luminescence intensity as a function of solute composition from MgC12+CaCl2 mixtures at various times after the laser pulse. l / l ~ ~ is the ratio of luminescence in themixture at a given time divided by the luminescence from 3.1 mol dm-3 MgC12 glass at that same time.0 , At 100 ns after the laser pulse, showing typical standard deviation errors ; 0, at 1 ps after thelaser pulse ; C!, at 100 ps after the laser pulse ; A, at 10 ms after the laser pulse. (b) Normalisedaccumulated light emission as a function of solute composition in MgC12+CaC12 and MgC12+Ng(C104)z mixed glasses at various times after the laser pulse. The ordinate is the ratio of thearea up to time t under the oscilloscope trace for a mixture divided by that for 3.1 mol dm-3 MgClz.0, During 1 ps after the laser pulse for MgC12+ CaC12 ; 0, during 100 ps after the laser pulse forMgCl2 + CaC12 ; A, during 10 ms after the laser pulse for MgC12+CaC12 ; e, during 1 ps afterthe laser pulse for MgC12+Mg(C104)2 ; m, during 100 ps after the laser pulse for MgC12+Mg(C104)2 ; A, during 10 p s after the laser pulse for MgC12 + Mg(C104)2.Curves 1 to 5 of fig.4(b) show the same data as fig. 4(a) but this time presented asthe integrated light intensity up to time t (JiZdt) plotted against log t. These datawere obtained from the measured areas under the oscilloscope traces. For compari-son, analogous data from MgC12 + Mg(C104), mixtures are also given in fig.4(b).FIG. 4.-Plots of the luminescence intensity at 450 nm as a function of time after laser excitation ofe,i, at 76 K in various aqueous glasses. (a) Shown as logarithm of the luminescence intensity [allin the same (arbitrary) units] against the logarithm of the time/s. Curves 1, 2, 3, 4 and 5 are for100 %, 80 %, 50 %, 20 % and 0 % respectively by volume of 3.1 mol dm-3 MgCl2 in 3.1 mol dm-3CaC12 solutions. (6) Shown as the integrated intensity to time tui Idtlarbitrary units) against log(rls). Curves 1 to 5 are as in (a) above. Curves 6,7 and 8 are for MgC12+ Mg(C104)2 mixtures at,respectively, 20 %, 50 % and 80 % Mg(C104)2 solution by volume (corresponding to fig. 6). [Theslopes for the linear parts of the curves in fig. 4(a) are : la t-" where n = 1.03 for curve 1, n = 1.04for 2, n = 1.05 for 3, n = 1.06 for 4, and 1.05 for 5.1966 LASER EXCITATION OF TRAPPED ELECTRONS20-2+en0 --4-6-8log @Is)FIG. 6.-Graph of logarithm of the luminescence intensity at 450 nm against the logarithm of thetime after the laser pulse in MgC12 + Mg(C104)2 and MgClz + ethylene glycol mixtures.Curves1 and 6-8 correspond to mixtures made from 3.1 rnol dmd3 MgClz and 4.1 mol dm-3 Mg(C104)Zsolutions containing 0, 20, 50 and 80 % by volume of the latter. Curve 9 is for 2.8 mol dm-3MgC123-0.7 mol dm-3 ethylene glycol. Curve 10 is for 2.3 mol dm-3 MgC1, + 1.8 mol dm-3ethylene glycol. Curve 11 is for 1.9 mol dm-3 MgClz + 2.9 mol dm-3 ethylene glycol. [The slopesfor the linear parts of these curves (>lo-“ s) are : la t-’J where n = 1.03 for curve 1, ii = 1.12 for 6,n = 1.19 for 7, n = 1.46 for 8, TI = 1.10 for 9, n = 1.14 for 10 and n = 1.21 for 11.FIG.7.-Normalised luminescence intensity as a function of solute composition from MgClz +Mg(C104)2 mixtures at various times after the laser pulse. I/1mg is the ratio of the luminescencein the mixture at a given time divided by the luminescence from 3.1 mol dm-3 MgClz glass at thatsame time. 0, 100 ns after the laser pulse ; 0, 1 p s after the laser pulse ; 0, 100 ps after the laserpulse ; A, 10 ms after the laser pulseT. Q. NGUYEN AND D. C. WALKER 1967Fig. 5 shows the luminescence as a function of glass composition at various timesafter laser-excitation of e&. In fig. 5(a) this is presented as the normalised intensityat the specified time, whereas in fig.5(b) it is shown as the normalised accumulatedlight emission up to the specified time.Note that luminescence, and hence ei,, are found in 3.1 mol dm-3 CaCl, glasses,albeit in much smaller yield than in MgC1, glasses.MgCl, +Mg(ClO,), MIXTURESMixtures containing 0,20,50,80 and 100 % by volume of 4.1 mol dm-3 Mg(C104),in 3.1 mol dm-3 MgC1, were used. Since the glass-forming concentration of theperchlorate solution was higher than that of MgC12, the total ion concentration ofthe mixtures increased linearly from 3.1 to 4.1 mol dm-3, though this is not expectedto have a particularly significant effect on the results. Fig. 6 shows the log I againstlog t plots obtained for these mixtures and fig.4(b) gives the accumulated lightemission. I decreases faster than t-l for curves 6-8 of fig. 6. Not only is theluminescence intensity reduced by an increased C104 concentration, but the rate ofchange of intensity with time is also very much faster. This can be contrasted withthe MgCI2 + CaCl, mixtures. The composition dependence is also strikingly differentas revealed by the plots of fig. 7 at 100 ns, 1 ps, 100 p s and 10 ms. Rather than givinga roughly linear dependence on concentration, the C104 has a dominant inhibitingeffect even as a minor component.Again in contrast to the pure CaCI, glass, e,i, in 4.1 mol dm-3 Mg(C10&glasses produced no measurable luminescence at all, even when studied as pelletswhere the background emission was trivial.None was found during pulse radiolysisof Mg(C10,), glasses either.2ETHYLENE GLYCOL AND MIXTURES WITH MgC1, AND CaCI,Since aqueous ethylene glycol glasses yielded absorption due to ei, (but noluminescence upon decay) in pulse radioIysis,2 it was of interest to see what wouldhappen during laser excitation of e,i, in these glasses and in their mixtures with othersolutes.190% 50 100%MgCI, Ethyleneglycol( 7.2 MIFIG. 8.-Normalised luminescence intensity as a function of solute composition from MgCll +ethylene glycol mixtures at various times after the laser pulse. I/lMg as in fig. 7. 0, 100 ns afterthe laser pulse ; 0, 1 ps after the laser pulse ; 0, 100 ps after the laser pulse ; A, 10 ms after thelaser pulse1968 LASER EXCITATION OF TRAPPED ELECTRONSNo luminescence was detected to arise from excitation of trapped electrons at694 nm in 7.2 mol dm-3 ethylene glycol aqueous glasses, even when pellets were usedto give a negligible background emission.The absence of luminescence must havea different cause in ethylene glycol than it does in Mg(C104)2 glasses because theformer permits ei, formation whereas the latter does not.In order to see if the addition of C1- in an " inert " form (from CaCl, ratherthan MgCl, or LiCl) to ethylene glycol glasses might lead to luminescence as ei,decayed, mixtures of ethylene glycol + CaCl, aqueous glasses were examined.However, in no case was any significant luminescence observed following the bleachingof e& in these glasses.Since C1- is evidently not a necessary and adequate ingredientto give luminescence, the next step was to test the effect of MgClz.Various mixtures of 7.2 mol dm-3 ethylene glycol and 3.1 mol dm-3 MgC1,solutions were deoxygenated, frozen to 77 K, y-irradiated and laser-bleached in thestandard way. The results are given in fig. 6 and 8. They are seen to be qualitativelyanalogous to Mg(C104)2 + MgC12 mixtures, in that the higher the ethylene glycolcontent, the smaller the luminescence and the faster it decreased. For the 40 %mixture the luminescence intensity decayed much faster than 1 It.It was also found, using the method described previ~usly,~ that the absorption at1850 nm, due to ei,, in 3.6 mol dm-3 ethylene glycol + 1.7 r n ~ l d r n - ~ MgC12 D20 glasses(40/60 % mixed-solute) was quite comparable with that in 3.1 mol dm-3 MgCl, glass.These results imply two things.First, that the C1- ion is not a sufficient constituentto produce luminescence from ei, ; and secondly, ethylene glycol either inhibits theformation of, or reacts with, the luminescent state which is normally formed whenei, decays.MXXED AQUEOUS GLASSES CONTAINING 2-PROPANOLPure 2-propanol and some of its mixtures with aqueous 3.1 mol dm-3 MgC1,give good glasses at 77 K which support trapped electrons (eyis) after y-radiolysis,therefore various 2-propanol glasses were studied to compare with the effect ofethylene glycol.There was no luminescence from a pure 2-propanol glass but data similar to thatshown in fig. 6 and 8 was obtained from 3.2moldm-3 2-propano1+2.3 moldm-3MgC1, mixtures, so the effects seen in ethylene glycol are essentially mimicked by2-propanol. It suggests that a chemical effect of the OH group is involved ratherthan a particular structural property of ethylene glycol glasses.MgCl,+LiCl MIXTURESGlasses prepared from 9.5 mol dm-3 LEI, 3.1 mol dm-3 MgC12 and a mixtureof 4.8 mol dm-3 LiC1+ 1.6 mol dm-3 MgCl, all showed a similar intensity ofluminescence under the standardised conditions described.Furthermore, the changewith time showed I cc Z/t for times > s in all three cases. In all respects, then,the LiCl+ MgCl, mixture showed the same behaviour as did the single solute glasses.These results confirm that the simple process of dilution of the MgCl, does notnecessarily curtail luminescence.ABSORPTION BY e;A few experiments were performed as before on the absorption by ei, at 1850 nm(using an InAs detector with 60 ns time response) following laser-excitation of e&.Results were obtained with 3.1 mol dm-3 MgClz glasses and with 1.6 mol dm-T.Q . NGUYEN AND D. C . WALKER 1969MgC12 + 2.1 mol dm-3 Mg(C104), and 2.5 mol dm-3 MgC12 + 1.9 mol dm-3 ethyleneglycol mixtures. Typical results are shown in fig. 3(d) for 1/10 maximum ruby laserpulses. (With this laser intensity there were many more 694 nm photons in the pulsethan e,i, in the sample and the absorptions were maximised, even though the quantumyield falls-off under those conditions.) Some luminescence traces are given in fig. 3(c)on the same time-scale for the same glasses for comparison purposes.DISCUSSIONMECHANISMThe role of the second solute should be discussed within the framework of theluminescence mechanism.Based on previous discussions 2* we propose thefollowing sequence of events leads to luminescence, and its inhibition.r II(hv) I11 1V (tunnels V-R+e,i, -e,f-ei, R-*+R-+kvLStep I indicates the creation in the water through y-radiolysis of some stable trappedelectrons and unidentified species R. Their yields will change with solute. Step I1is the laser excitation of e,i, which, evidently, causes some electrons to become quasi-free (e2). Some of these subsequently become localised in shallow traps or suitablecavities to become ei, as signified by step 111. It has been shown that I K t-l and[ei,] K log t, which are in accord with the previously proposed decay mechanism ofei,; namely, by tunnelling to R within the same spur (geminate tunnelling, stepIV).2* The overall rate is given by -dei,/dt = k/t, due to a combination of theexponential distribution of separation distances and the exponential tunnellingefficiency.ll This implies that steps 11 and I11 do not seriously destroy the basicspur structure created by radiolysis.Step V corresponds to fluorescence of R-* andis rapid. From the traces of fig. 2 the lifetime of R-* was estimated to be N 2 x s.Consequently, for times > lo-' s, step IV will be the rate-controlling step.Luminescence, as found in MgC12 or LiCl glasses, requires all five steps I to Vof this mechanism.Its intensity should be reduced whenever a second solute indulgesin any of the competing processes, steps VI to X. A glass-forming solute whichreduces the availability of R during or after the radiolysis (step VI) will cause processIV to be slowed down. Process VII contributes to changes in the yield of stabletrapped electrons, but is unimportant here since constant absorbances at 694 nmwere used. Step VIII represents scavenging reactions of e,f reactions including trap-ping as e&. Loss of ei, by route IX will diminish the luminescence; but the time-dependence can take one of several forms. Consider just two of these. IX(a) iswhen some ei, are destroyed by reaction at their site of formation very rapidly(< s).This will diminish the luminescence without affecting the inverse timedependence. IX(b) is when the time-scales of IX and IV overlap; that is whenei, reacts, still by tunnelling, to a non-luminescing reaction partner which is distributedrandomly (the solute, an impurity or a suitable trapping site). The luminescenceintensity will fall-off at a greatly enhanced rate, eventually approaching an exponentialrate as IX(b) dominates over IV. Finally, processes indicated by X may be describedas competitive quenching reactions for R-*. They reduce the fluorescence intensitywithout altering its normalised time-dependence.Based on this mechanism how do the second solutes CaCl,, Mg(C104), andethylene glycol (or 2-propanol) inhibit the luminescence from MgC12 glasses ?\ VI \(694nm) VII \ VIII \.x \x4 4 4 1970 LASER EXCITATION OF TRAPPED ELECTRONS(i) CaC1,In MgCl,+CaClz mixtures the yield of ei, decreases relatively linearly withincreasing CaCl, concentration, suggesting that CaCl, interrupts step 111 by providingalternative paths VIII, such as to e,i, [G(e,i,) in CaCl, glass is - 3 times that in MgCl,initially12 or by inhibiting 111 through its structure-breaking propensity.In 3.1 moldm-3 CaCl, glasses there is still some luminescence ( N 10 % that found in 3.1 m01dm-~MgC1, glasses at t > s) so for CaCl, the difference is one of degree ratherthan kind.As a second solute in MgC1, glasses, the behaviour of perchlorate may be attributedto process IX; probably mainly to IX(b) in addition to initial loss of ei, throughIX(a). Infrared electrons may be reactive towards C104 or to impurities in theMg(ClO,), crystals used.The eventual dominance of IX(b) can account for thefall-off of luminescence greatly exceeding t - I .(iii) ETHYLENE GLYCOLSince ei, occurs in ethylene glycol glasses, processes VII, VIII and IX(a) areexonerated. The absence of luminescence therefore implicates either VI or X.Traces in fig. 3(c) and (d) indicate that the yield and lifetime of ei,, as revealed by theinfrared absorption, in 3.1 mol dm-3 MgCI, and 2.5 rnol dm-3 MgC12 + 1.9 rnol dm-3ethylene glycol glasses are comparable; yet the luminescence is much weaker fromthe latter glass and it falls off very much more rapidly than does the rate of loss ofabsorption.Step X merely competes with step V to reduce the lifetime of R-* anddiminish the luminescence intensity; but since V step is already fast and not rate-controlling, X cannot account for fig. 3(c). Consequently we infer that in glassescontaining ethylene glycol (or 2-propanol) reaction occurs with R (step VT), therebyreducing the yield of stable R species available for ei, to tunnel to. In this way theluminescence will also decrease rapidly. Instead of tunnelling to R, ei, may tunnelto trapping sites. Buxton et aZ., found that some of the ei, in ethylene glycol glasses(but not in LiCl or MgC1,) converted spontaneously to e&, by a process which theywere able to attribute to trap-to-trap tunnelling (rather than to dipolar relaxation1,).Transfer to e,i, will quickly dominate over tunnelling to R, because of the shortageof R, so that absorbance by ei, should greatly outlive the luminescence.Our basic proposal with regard to ethylene glycol as an additive is that it attacksR, and thereby provides an aqueous medium in which ei, can convert to e;,, albeitrelatively slowly and inefficiently.GENERAL CONCLUSIONSIt is the multifarious behaviour of these second solutes which leads us to thecontention that ei, is normally to be expected from quasi-free electrons in water,but that many commonly used glass-forming solutes inhibit its formation or destroy it.We infer the following effects for the second solutes.With LiCl there is no change ;with CaCI, the luminescence is diminished due to a dilution of the suitable infraredsites ; with Mg(C104)2 infrared electrons are very short-lived due to reaction ; andwith ethylene glycol R is destroyed so that ei, can convert to e&.The experiments have not provided information about the structure of ei, (exceptthat it is clearly not associated chemically with the specific presence of solvated C1-or Mg2+ ions), or on the identity of R-'$ (except in estimating its lifetime)T . Q . NGUYEN AND D . C . WALKER 1971We acknowledge the advice offered and interest shown by Dr. Hugh Gillis. Thefinancial support from the National Research Council of Canada is appreciated.G. V. Buxton, H. A. Gillis and N. V . Klassen, Chem. Phys. Letters, 1975, 32, 533.G. V. Buxton, H. A. Gillis and N. V. Klassen, Canad. J. Chem., 1976,54,367.H. A. Gillis and D. C. Walker, J. Chem. Phys., 1976,65,4590.B. G. Ershov and A. K. Pikaev, in Radiation Chemistry I, ed. R. F. Gould (American ChemicalSociety, Washington, 1968), p. 1.G. V. Buxton and K. G. Kemsley, J.C.S. Faraday I, 1975,71, 568.D. C. Walker and R. May, Int. J. Radiation Phys. and Chem., 1974, 6, 345.L. Kevan, J. Phys. Chem., 1972,76,3830.G . V. Buxton, F. C. R. Cattell and F. S. Dainton, Trans. Faraday SOC., 1971, 67, 687.9D. C. Walker, Canad. J. Chem., 1977, 55, 1987.'OD. C. Walker, in Electron-Solvent and Anion-Solvent Interactions, ed. L. Kevan and 3. C.Webster (Elsevier, Holland, 1976), p. 91.M. Tachiya and A. Mozumder, Chem. Phys. Letters, 1975,34, 77.l 2 H. Hase, M. Noda and T. Higashimura, J. Chem. Phys., 1971, 54,2975.(PAPER 7/343
ISSN:0300-9599
DOI:10.1039/F19777301958
出版商:RSC
年代:1977
数据来源: RSC
|
224. |
Structure and catalytic activity of CoχMg1–χAl2O4spinel solid solutions. Part 1.—Cation distribution of Co2+ions |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 73,
Issue 1,
1977,
Page 1972-1982
Carlo Angeletti,
Preview
|
PDF (937KB)
|
|
摘要:
Structure and Catalytic Activity of Co,Mgl -xA1204Spinel Solid SolutionsPart 1.-Cation Distribution of Co2+ IonsBY CARL0 ANGELETTI, FRANCO PEPE AND PIER0 PORTA”Centro di Studio su Struttura e Attivita Catalitica di Sistemi di Ossidi del C.N.R.,Istituto di Chimica Inorganica, University of Roma, Roma, ItalyReceived 28th February, 1977The cation distribution in CoA1204 and in several CoxMgl-xAlz04 solid solutions with x rangingfrom 0.05 to 0.5 has been studied by magnetic susceptibility, reflectance spectroscopy, lattice parametervariation and analysis of some X-ray reflections whose intensities are particularly sensitive to variationin cation positions.The results show that all specimens have the cobalt ions in a predominantly tetrahedral distributionand that there is a change towards a random cation distribution when the preparation temperatureis increased from 1073 to 1473 K.The fraction of Co2+ ions in tetrahedral sites, a, is equal to 0.82and 0.77 for the CoA1,04 spinel prepared at 1073 and 1473 K, respectively.The X-ray results show, moreover, that a does not vary significantly with composition for xvarying from 1.0 to 0.1. For the most dilute cobalt solid solution (x = 0.05), however, the valueof a is lower ; this trend is observed in both series of specimens prepared at different temperaturesand may be explained in terms of anion-anion contact effects.A large class of compounds having a wide range of chemical and physical propertiesof considerable technological importance is known to crystallise in the spinel structurewhich may simultaneously accommodate metal ions among the octahedral andtetrahedral sites available in the lattice.For a binary oxidic spinel containing divalent, X, and trivalent, Y, cations, twoextreme distributions of cations are possible : the “ normal ” X[Y2]04 and the“inverse” Y[XY]04 distribution,l where the ions in the octahedral sites are insquare brackets.Between these limiting cases intermediate distributions are possible,one being of particular interest, namely the “ random ” one X+Y+[XpY+]04.2The cation distribution in spinels has been experimentally proved to be anequilibrium function of temperature and A fine balance between theoctahedral and tetrahedral preferences of the ions concerned and other factors, suchas the ionic charge and the ionic radius,6 crystal and ligand field effect^,^" anionpolarisation,1° etc., contribute to the cation distribution in spinels.Moreover, bymaking use of the property of solid solution formation, gradual changes in the solidstate chemistry can be effected, other than with temperature and pressure, by varyingthe composition of the solution. Studies of cation distribution in the spinel structureare thus of considerable interest in solid state chemistry because they may allowinvestigation of the relative stabilities of ions in octahedral and tetrahedral coordina-tion and better understanding of the correlations between structure and physicaland chemical properties. The spinel system is also of interest in the study of itscatalytic properties, which can be discussed in terms of the symmetry adopted by thetransition metal ion.In this work the cation distribution in Co,Mg,-,Al,04 solid solutions has beenstudied by magnetic susceptibility, reflectance spectroscopy, lattice parameter variation197C.ANGELETTI, F . PEPE AND P. PORTA 1973and, mainly, by careful analysis of some X-ray reflections whose intensities areparticularly sensitive to variation in cation positions. Co,Mg,-,Al,O, is of interestbecause of the rather similar preferences of Co2+, Mg2+ and A13+ ions for octahedraland tetrahedral coordination. We describe the combined use of the four techniquesto evaluate the cobalt ion distribution and its dependence on temperature andcomposition.The powders, structurally characterised, were then utilised as catalyststowards the decomposition of nitrous oxide : Part 2 will deal with catalytic studies.EXPERIMENTALPREPARATION AND CHEMICAL ANALYSISPure MgA1204, pure CoA1204 and 4 Co,Mgl+41204 solid solutions with x = 0.05,0.1,0.25,0.5 were prepared by soaking AI2O3 with cobalt and magnesium nitrates in stoichio-metric quantities; the soaked mass for each composition was dried at 383 K and ground,heated in air at 873 R for 2 h in order to decompose the nitrates, carefully re-ground, andpressed into pellets (at a pressure of -8000 kg cm-2). The compounds were sintered inair at 1473 K for 100 h ; for each composition one batch of pellets was quenched in waterfrom 1473 K, while a second batch was cooled to 1073 K and equilibrated at this temperaturefor 50 h before quenching in water.This procedure gave 6 specimens each of which wasprepared at two different temperatures.TABLE 1 .-C0,Mg,-,A1204 SPECIMENS WITH NOMINAL AND EXPERIMENTAL COBALT CONTENTS(xnom, xexp), LATTICE PARAMETERS a, MAGNETIC MOMENTS p, WEISS CONSTANTS 8 AND EXPERI-MENTAL X-RAY DIFFRACTION INTENSITY RATIOS 1400/1220 AND 1400/1422designationSAMSAMCo 5SAMCo 10SAMCo 25SAMCo 50SAC0designationSAMSAMCo 5SAMCo 10SAMCo 25SAMCo 50SAC0Xnom-0.050.100.250.501 .ooXnom-0.050.100.250.501 .ooXCXP-0.0480.0970.2420.493-Xexp-0.0480.0970.2420.493-1473 KalA p1B.M. I~ooIJz~o 1400/I422- I - - 8.08318.0842 4.80 10 1.64 5.278.0849 4.77 35 1.46 4.658.0869 4.80 47 1.17 3.678.0941 4.76 65 0.71 2.218.1051 4.82 104 0.32 1.161073 KalA P/B-M- -elK 1400/1220 1400/1422I I - - 8.083 18.0844 4.72 8 1.62 5.178.0865 4.68 21 1.42 4.588.0885 4.71 30 1.09 3.528,0950 4.71 49 0.63 2.048.1058 4.76 113 0.25 0.90Samples containing magnesium and cobalt only are labelled SAM and SACo, respectively.The solid solutions are designated SAMCo.The numbers after the letters give the nominalconcentration of cobalt atoms with respect to 100 (Mg+Co) atoms. The samples wereanalysed for the cobalt content; a small quantity of compound was fused with KHS04in a platinum crucible, the cooled melt dissolved in dilute HCl and then filtered. The cobaltwas then determined by atomic absorption techniques with a Perkin-Elmer apparatus.Table 1 reports the list of the prepared specimens and the results of the analyses.X-RAY INVESTIGATIONS AND CALCULATION OF INTENSITIESIron-filtered Co Ka radiation was used to investigate all of the compounds both for thelattice parameters and for the intensity measurements.For all specimens, X-ray diffractionpatterns showed no lines other than those belonging to the cubic spinel structure1974For precise determination of lattice parameters, a Philips camera (114.6 mm diameter)with the asymmetric Strawnanis film mounting method was used. For all samples thereflections were very sharp. All al reflections in the region of 8 = 60-90" were read bymeans of a Norelco measuring device with an accuracy of 0.005 cm and used in constructingNelson-Riley plots for extrapolation to 6 = 90".During the exposure (-8 h) the cameratemperature was observed to vary within only 1-2 K. For each specimen the derived valueof the lattice parameter a was corrected for the thermal expansion coefficient, 6 . 5 ~ lW5 AK-l.12 The values reported in table 1 are referred to 294 K. The error in a is -2 x lW4 A.The integrated intensities of the 220, 400 and 422 reflections were measured by pulsecounting with a Philips high-angle goniometer spectrometer using a flat-plate sample holderand a scanning rate of 28 = 0.25" min-'. All specimens were run twice and after correctingfor background a mean value of the intensity was taken. The mean I4Oo/I220 and 1400/1422observed intensity ratios obtained from these experiments are summarised in table 1.These reflections and their relative intensity ratios were selected since they provide thebest basis for determining the cation distribution in Co,Mgl-,A1204 by a comparison ofexperimental and calculated values.In fact, according to Bertaut l3 and to Weil, Bertautand Bochirol,14 the best information on cation distribution in spinels is achieved by comparingexperimental and theoretical intensity ratios for reflections whose intensities : (i) do notdiffer strongly, (ii) are independent of (or only slightly dependent on) the oxygen parameter u,and (iii) vary inversely with the inversion parameter. Reflections of this type are, for thesystem under investigation, those of the classes (c) and (e) of table 2, presented below.Co,Mg,-,Al,O, SOLI D s OLU TI ON sTABLE 2.-sCATTERING FACTOR COMBINATIONS (FOR U = if) FOR THE X-RAY REFLECTIONS INTHE SPINEL STRUCTURE (FOR EXPLANATION SEE TEXT)class V+k2+12 example P FO F1a 32n 440 12 2fY +fx+ 4f0, 0b 16nS-11 311 24 f y + f x l J z - (Jz- 1)c 16(2n+l) 400 6 2fY -fx + 4f0, 4d 32nS-12 222 8 2fY - 4fox 2e 16n+8 220,422 12,24 fx -2f 16nf3 111 8 fy-fxt J2- (JT+ 1)To determine the cation distribution and its variation, if any, with x it is then necessaryto calculate for each composition the 1400/1220 and intensity ratios expected forgiven arrangements of the three cations and compare them with the experimental intensityratios.The method of calculations carried out for C O ~ M ~ ~ - ~ A ~ ~ O ~ solid solutions isdescribed in the following paragraphs.For the calculation of the relative integrated intensity, I, of a given diffraction reflectionfrom powder specimens, as observed in a diflractometer with a flat-plate sample holder,the following formula is valid :15Ihki = IF/,2,* p LP (1)where F = structure factor, P = multiplicity factor, Lp = Lorentz-polarisation factor.It may be observed that the calculated integrated intensities are valid at 0 K, and sincethe observed values are obtained at room temperature a suitable correction to eqn (1) is inprinciple necessary for precise comparison.However, since the spinels are high-meltingcompounds, the thermal vibration of the atoms at room temperature should have a negligibleeffect ; this has been confirmed 3a for NiA1204 and other spinels.Hence in our intensitycalculations no temperature correction was deemed necessary.According to Bertaut,13 for a spinel X2+Y:fO4 F may be written in the form :F = Fo+%AFl (2)where FO is the structure factor for a normal spinel, I. is a parameter describing the degreeof inversion and corresponds to the fraction of trivalent Y3+ ions on tetrahedral sites (Aequals zero and 0.5 for a normal and inverse spinel, respectively), A is the difference betweeC. ANGELETTI, F . PEPE AND P . PORTA 1975the atomic Scattering factors of the divalent (XI and trivalent (Y) cations, and Fl is a numericalfactor. AAFl is therefore the correction due to inversion. For the reflections selected, thevalues of P, Fo and Fl (together with the classes to which the reflections belong) are givenin table 2, in whichf,,f, andf,, refer to the atomic scattering factors of the divalent, trivalentand oxygen ions, respectively, and n is an integer.All other classes of reflection are alsoreported in table 2.For a spinel containing only two types of cation, X and Y , the cation distribution isconveniently described by the general formula :XI-2 1y22cx21y2-2210.4 (3)where the ions in the octahedral sites are in square brackets, and where A is the fraction oftrivalent cations in tetrahedral sites. However, because the solid solutions contain more1 1.9 11.81.71.60 P(\ 0L?2 1.51.41.31.20.00.20.40.60.81 .o0.2 0.4 0.6 0.8 1.0sFIG. 1.-X-ray diffraction intensity ratio 1400/1220 in Coo.lMgo.9AlzOq.Full lines, calculatedvalues for different values of the fractions of cobalt (a) and magnesium (P) ions in tetrahedral Asites ; dashed lines, experimental values for the specimen quenched from 1473 and 1073 K.than two cations, it is not sufficient to define only one inversion parameter 2. For a spinelsuch as Co,Mg1-,A1,O4 it is convenient to characterise the cation distribution by the generalformula :where a,ively.(4) it follows thatand 2 are the fraction of Co2+, Mg2+ and A13+ ions on tetrahedral sites, respec-This amounts to the addition of only one new parameter, since from eqn (3) and22 = l-ax-B(l-x). ( 5 1976 Co,Mg,-,Al,O, SOL ID SOL u TI o N sThe intensities were computed with a Univac 1108 computer at Rome Universityaccording to eqn (1) and (2) with the following input data incorporated in the program :(i) the molar composition of the spinel ; (ii) the value of u, the oxygen parameter ; (iii) thescattering factors for Co2+, Mg2+, A13+ and 02- ; (iv) the Lorentz-polarisation correction ;(v) the real part of anomalous dispersion correction for cobalt.The value of 0.387 was used for u, based on the values found for both MgAI2O4 andCoA1204 end members.16* International Crystallographic Tables were used for thescattering factors and Lp corrections, except that the value for the scattering factor for 02-was taken from Suzuki.18The computer program was designed to calculate intensity ratios in 0.1 intervals of botha and P.This is equivalent to providing the intensity ratios for 100 different values of Afor each spinel. The results were displayed for each solid solution by constructing fromthe 100 computed values a family of curves of intensity ratio against p(0 < p < 1) with avarying in steps of 0.1 from 0 to 1. As an example we illustrate in fig. 1 the calculatedcurves for the 1400/1220 intensity ratio for the composition x = 0.1. The experimentalresults are shown as dashed lines.MAGNETIC SUSCEPTIBILITYMagnetic susceptibilities were measured using the Gouy method over the range oftemperature 100 to 295 K. Correction was made for the diamagnetism of the sample;the susceptibility of the MgA1204, measured for this purpose, was x = - 0.303 x Acheck was made that the susceptibilities were independent of magnetic field strength.Thevalues of the Curie constant C and hence of the magnetic moment ,u were taken for allspecimens as angular coefficients from 1 /xat. against T plots.REFLECTANCE SPECTRAReflectance spectra in the range 350-2500 nm (28 500-4000 cm-I) were recorded on aBeckmann DK-1 spectrophotometer with a standard reflectance attachment, againstMgAI2O4 as reference.RESULTSLATTICE PARAMETERSThe lattice parameter a of the cubic unit cell of each of the spinels studied isgiven in table 1.There is an increase in lattice parameter with increasing cobalt content. Thevalues of a for the specimens quenched from 1473 K are lower than those of thecorresponding specimens quenched from 1073 K.The variation of a with cobaltfraction x is linear.X-R A Y D I FF R A c T I o N I N TE N s I TI E sIn principle the comparison of the experimental and calculated results for twodifferent intensity ratios is sufficient to determine both a and B (fractions of Co2+and Mg2+ ions on tetrahedral sites, respectively), and thus the ion distribution isuniquely determined. For many spinel solid solutions this would be true in practice.In Co,Mg,-,A1,O4, however, two of the cations (Mg2+ and A13+) have such similarscattering factors that the exchange of them between tetrahedral and octahedral sites(a constant, /3 varying 0 -+ 1) produces very little change in intensity ratio. A morecomplete treatment of this behaviour has been given in a previous a r t i ~ l e .~ It shouldbe remembered that MgA1204 was formerly regarded as a normal spinel (p = l),both on theoretical grounds and on the basis of neutron diffraction analysis.16 Morerecently, however, results from n.m.r.l and infrared spectroscopy 2o have pointeddecisively to a slightly inverted distribution for this spinel. Re-examination bC . ANGELETTI, F . PEPE AND P. PORTA 1977neutron diffraction has confirmed that MgA1204 is partly inverse and the fraction Bto Mg2+ ions in tetrahedral sites has been estimated around 0.9. By taking /? = 0.9,table 3 shows the results of a of our comparison between observed and calculatedintensity ratios. It may be recalled that for our purposes it is of interest to find afrend of a towards x and T rather than the absolute values of a, which are moredifficult to achieve in powder solid solutions.Fig. 2(a) and (b) show that the trendof a does not vary if one takes plausible values of p other than 0.9; in fact for the1473 K quenched samples, where p can be supposed to decrease for the effect ofentropy, the trend of a for p = 0.8 is similar to that observed for p = 0.9, althoughthe absolute values of a are slightly different.The mean values of a (for p = 0.9) are shown graphically in fig. 2(c). The errorson the mean a value, evaluated on the basis of those derived from both the experimentaland calculated inten~ities,~ are estimated to be 10 %.TABLE 3 .-CATION DISTRIBUTION IN C O x M g ~ - ~ A l ~ O ~ SOLID SOLUTIONS '1473 Kdesignation aSAM -SAMCo 5 0.51SAMCo 10 0.75SAMCo 25 0.78SAMCo 50 0.775SAC0 0.77B = 0.8a-x p(1-x)- 0.80.0255 0.760.075 0.720.195 0.600.3875 0.400.77 -1073 KB = 0.9designation z a.x &1-x)0.9 SAM - ISAMCo 5 0.69 0.035 0.855SAMCo 10 0.80 0.080 0.810SAMCo 25 0.82 0.205 0.675SAMCo 50 0.82 0.410 0.45SAC0 0.82 0.82 -[ax + S( 1 - 410.80.78550.7950.7950.78 70.775l a x i P( 1 - x)!0.90.890.890.850.860.82p = 0.901 a.x p(1-x)0.9 - -0.60 0.030 0.8550.75 0.075 0.8100.77 0.193 0.6750.77 0.385 0.4500.77 0.77 -0.90.8850.8850.8680.8350.77a a = fraction of Co2+ ions on tetrahedral A sites ; CA-x = total amount of Co2+ ions on A sites ;/3 = fraction of MgZ* ions on A sites ; j3(1 - x ) = total amount of Mg2+ ions on A sites ; [ax+/3(1 -x)J = total amount of divalent ions (Co2++ Mg2+) on A sites.The main conclusions which can be drawn are as follows : (i) Co2+ ions primarilyoccupy tetrahedral sites, but there is some octahedral cobalt in all specimens.(ii)For a given cobalt concentration in the solid solution, there is less cobalt in tetrahedralsites in the case of samples quenched from 1473 K. (iii) Within each series at differenttemperatures the value of a is nearly constant from x = 1 to x = 0.1 ; below thisvalue of x, a tendency for Co2+ ions to occupy fewer tetrahedral sites is shown.This trend is true if p is assumed to be constant in all the solid solutions.MAGNETIC SUSCEPTIBILITYThe results of the magnetic measurements are presented in table 1.The completeset of results shows that the magnetic moments are all in the range to be anticipatedfor a cobalt distribution which is predominantly tetrahedral (pLtet = 4.6 B.M.).Furthermore, there is a consistently higher magnetic moment for the specimensquenched from 1473 K than for those quenched from 1073 K (increased octahedralcobalt with increasing quenching temperature)1978 Co,Mg,-,Al,O, SOLID SOL UT I o N sThe Weiss constant, 8, of the Curie-Weiss law x = C/(T+8) taken from theintercept on the plots of l/xat against T, was found, as expected, to increase withincreasing cobalt content ; its values, reported in table 1, range from - - 10 K toabout - 110 K.0.60.5(b) 0.90.80.7-OL0.60.5(c) 0.90.8-a 3.70.60.5II,------ ---____,_._._._._._.-.-.-.-.-.Iii0.1 0.25 0.5X1.aFIG.2.-Fraction of cobalt ions in tetrahedral sites (or) plotted against cobalt content (x) inCoxMgl-,Alz04 spinels ; (a) and (b) : obtained for different values of the fraction of magnesiumions in tetrahedral sites (p), (- - -) j3 = 1.0 ; (-) p = 0.9 ; (- - -) /3 = 0.8. (a) 1473 K ; (b)1073 K. (c) Obtained for ,!? = 0.9 : (0) 1073 K, (A) 1473 K.REFLECTANCE SPECTRAThe tetrahedral bands dominate the spectra, and this is also due to the lessforbidden character of tetrahedral absorption relative to octahedral absorption. Weobserved, as illustrated in fig. 3 for the SAMCo 10-1473 K specimen, a broad band(a) in the region around 4300 cm-l, a group of 3 bands at 6000-7400 cm-i (bi, b2, b3),a shoulder at 9,300 cm-I (c), a group of 3 bands at 15 300-17 800 cm-l (d,, d2, d3),and 3 bands in the region 20 300-24 400 cm-' (e,L 9).Our spectra accorded veryclosely with those reported in the literature.22* 23 Band (a) is an envelope of theunresolved bands corresponding to the tetrahedral 4A2(F) + 4T2(F) transition ;bands (bl, bz, b3) are associated with the tetrahedral 4Az(F) -P 4T1(F) transitionC. ANGELETTI, F . PEPE AND P. PORTA 1979bands ( d l , d2, d3) are the tetrahedral 4A2(F) + 4T1(P) transitions and (e,f, g) corres-pond to the tetrahedral spin-forbidden 4A2(F) -+ 2T(G) transitions. The shoulderat 9300 cm-l (c) can be attributed to the octahedral 4T1,(F) + 4T2,(F) transition.All bands moved to lower energies with increasing cobalt content, in agreementwith the expansion of the cell volume at the increase of x (see values of lattice para-meters a in table l), and hence with the increase of cation-anion distance. Inspectionof the relative intensities of the tetrahedral bands with respect to the shoulder attributedto octahedral cobalt as a function of cobalt content revealed no variations.Incontrast to this, the comparison of band intensities for samples at equal cobaltcontent but prepared at different temperatules revealed higher band intensities for1.0 1IiI28.5 20 15 10 7.5 5 4t i r 1 I ' 1 I 1 1350 400 500 600 700 700 1000 1500 2000 2500 nmwavelength /nmFIG. 3.-Reflectance spectrum of the Coo.lMgo.,Alz04 specimen quenched from 1473 K.specimens quenched from 1073 K than for those quenched from 1473 K, The onlyexception to this feature is given by the shoulder at 9300cm-1 which seems to beless pronounced at lower temperatures.This behaviour is an indication of morecobalt tetrahedral occupation at lower quenching temperatures.DISCUSSIONThe distribution of cations among the available octahedral and tetrahedral sitesof the spinel structure is influenced by various energy terms, such as the Madelungenergy, the Born repulsive energy, anion polarisation and the individual site-preferenceenergy. It has been shown by Blasse that for spinels containing divalent andtrivalent cations in the absence of the site-preference energy the " normal " distribu-tion, i.e.that structure in which all trivalent cations are on the octahedral sites and1-61980 Co,Mg,-,Al,O, SOLID SOL u TIO N sall divalent cations on the tetrahedral sites, is favoured by the Madelung energy andanion polarisation. When the octahedral Crystal Field Stabilization Energy (CFSE)of divalent cations makes a significant contribution to the total lattice energy or whentrivalent cations exhibit a strong preference for tetrahedral coordination, the“ inverse ” distribution is favoured. In aluminium spinels all compounds butNiA1204 (Octahedral CFSE for Ni2+ ions equal to about 84 kJ 1110l-l)~ have accord-ingly been found “ normal ” or nearly “ normal ~pinels.~ CoA1204 and MgA1204are cases where there is no pronounced site-preference energy to contribute to thetotal lattice energy and hence they are nearly “ normal ’’ spinels.l7.21In “ normal ” aluminium spinels, with small A13+ ions in large octahedral sites,the radius ratio A13+/02- is such (0.39) that there is anion-anion contact with a highrepulsion term. Unit cells are thus expected to be larger for “normal” spinelsthan for “ inverse ” spinels when comparing structures containing ions with similarradii.Another method of looking for changes in the distribution of paramagnetic ions,such as Co2+, between octahedral and tetrahedral sites, is magnetic susceptibility.In fact for Co2+ in tetrahedral field (ground state 4A2) the effective magnetic momentpeff. = ,us.o. (1-4A/10 Dq), where A is the spin-orbit coupling constant, Dq is thecrystal field strength and pss0.is the spin-only value of the magnetic moment (forCo2+ps.,. = 3.9B.M.).24 The experimental values of peff. for Co2+ ions in thetetrahedral configuration has been found around 4.6 B.M. in Co,Zn,-,O solidThe ground state of Co2+ in an octahedral field is, instead, an orbitaltriplet 4T1, and its effective magnetic moment is accordingly appreciably increasedby the orbital contribution. Experiment has shown that p = 5.25 B.M. for Co2+ions in octahedral configuration (Co,Mgl-,0).26 The values found by us (seetable 1) for the magnetic moment thus indicate that the cobalt coordination in theCo,Mgl-,A1204 spinels is predominantly tetrahedral.All the methods used by us have shown a definite trend towards more randomcobalt distribution with increasing temperature at constant x composition.Thelattice parameters (table 1) are lower in the 1473 K quenched specimens, whichindicates that the spinel as a whole becomes more inverted.6 This effect can be dueto randomisation of Co2+ and/or Mg2+ ions but, considering (i) that the latticeparameter of MgA1204 does not change with temperature, (ii) that the reflectancespectra show lower intensities on the tetrahedral bands of cobalt and a morepronounced octahedral shoulder with increasing firing temperature and (iii) that thevalues of the magnetic moment due to Co2+ ions are slightly but systematicallyhigher for the specimens quenched at 1473 K, we believe that the variation of thelattice parameter a with temperature (at constant x composition) is due to randomisa-tion of the cobalt ions only.The X-ray intensity results demonstrate conclusivelythat the cobalt distribution is influenced by the temperature treatment ; with increasingquenching temperature there is a definite trend towards more octahedral cobaltoccupation and hence towards more “random” spinels [fig. 2(c)]. It should berecalled that the completely “ random ” spinel (for compounds containing divalentand trivalent cations) corresponds to -67 % of divalent cations in the octahedralsites. The observed effect of the temperature on the cation distribution is indeedto be expected for any spinel in which the enthalpy difference between normal andinverted states of a component ion is small and which, at low temperatures, ispredominantly normal in that species. The interchange enthalpy for the equilibriumX,”,t +Y,”,‘: +Let us now consider the relation between cation distribution and composition x.Within each series of Co,Mgl-,A1204 solid solutions at constant temperatures (1473+Y:e. in CoAl,O, is - + 54.39 kJ r n ~ l - ’ .~ C. ANGELETTI, F. PEPE AND P. PORTA 1981and 1073 K) the expansion of the lattice on going from MgA1204 to CoA1204 (table 1)is explained in terms of substitution of smaller Mg2+ ions (Goldschmidt ionic radiusof 0.78 A) with larger Co2+ ions (Goldschmidt ionic radius of 0.82 A). It has beensaid before that the a value is also affected by the cation distribution, but the experi-mental evidence (fig. 4) is for an increase of inversion towards more cobalt contentwhich should bring about a decrease of a.Hence, the cation distribution is in thiscase not responsible for the increase in a.The reflectance spectra show a shift of all the observed bands towards lowerenergies with increasing cobalt content in agreement with the expansion of the unitcell volume. The values of p, all in the range to be expected for cobalt beingpredominantly tetrahedral, do not indicate any substantial variation of cobaltdistribution with x.0.70. I 025 Q5 1.0XFIG. 4.-Total amount of divalent (Co2++ Mg2+) ions on tetrahedral A sites plotted against cobaltcontent (x) in CoxMg,-,AlzO4 solid solutions : (0) 1073 K, (V) 1473 K.However, the influence of composition on the cobalt distribution may be derivedfrom the X-ray intensity measurements.On the assumption that the magnesiumdistribution remains constant (p = 0.9) within all the solid solution series, we observethat the Co2+ tetrahedral occupation is nearly constant from CoA1204 to x = 0.1[fig. 2(c)]. A decrease in cobalt tetrahedral occupation, down to 0.69 and to 0.60for the two 1073 and 1473 K series respectively, has been observed when x = 0.05.We are well aware that the X-ray results are more affected by errors for the moredilute cobalt specimens and thus it is more difficult for these compositions to drawgood absolute values of cation distribution; it may be noted, however, that a slightdecrease in the cobalt tetrahedral occupation already becomes visible for spinelswith x = 0.1 (table 3).We suggest that the influence of composition on cationdistribution is brought about in this case by anion-anion contact effects.It may be recalled that calculations made by Gorter 28 and by Hafner 29 of thecell edge a and of the oxygen parameter u on a model consisting of a dense packingof rigid spheres conclude that there is anion-anion contact, and thus repulsion andstructure destabilisation, if there are too many small cations in the octahedral sites.This happens, of course, when no other energy terms, such as the octahedral site-preference energy, predominate.In our spinels, on going from CoA1204 to more dilute cobalt solid solutions, thestructure becomes more normal as a whale, i.e. the total amount of Me2+ in A site1982 Co,Mg,-,Al,O, SOLID SOLUTIONSis increasing (fig.4); this implies that the octahedral sites are contracted and theadjacent tetrahedral sites are expanded because more and more smaller A13+ ionsare replacing the larger divalent ions on the octahedral sites. This cannot continuebecause the anion-anion contact effect becomes a relevant factor to destabilise thestructure. Electrostatic energy tends to increase, and hence structure stabilisationoccurs, if, as observed, a higher fraction of larger Co2+ ions interchanges with smallerA13+ ions in the octahedral sites. Mg2+ ions are supposed, in this process, to be" anchored " at the tetrahedral sites (p = constant) more than the Co2+ ions becauseof their smaller octahedral ~reference.~It should be added that the process proposed here when x is small does not involvesuch a large total amount of Co2+ ions as to be detectable by lattice parameter ormagnetic moment variations.A variation of cation distribution with composition has also been observed forother spinel solid solutions studied by us, such as Ni,Mgl-,A1204,5 Ni,Znl-,A120,,CuxMgl-xA1204, CoGa,Al,-,04.*The authors thank Prof. A. Cimino for critical reading of the manuscript ; theyalso thank Mr. G. Minelli for assistance in some of the experiments.T. F. Barth and E. Posqiak, 2. Krist., 1932, 82, 325.F. Machatschki, 2. Krist., 1932, 82, 348.(a) R. K. Datta and R. Roy, J. Amer. Ceram. SOC., 1967,50,578 ; (b) R. K. Datta and R. Roy,Amer. Mineral., 1968, 53, 1456.H. Schmaluied, 2. phys. Chem. (N.F.), 1961, 28, 203.P. Porta, F. S. Stone and R. G. Turner, J. Solid State Chem., 1974, 11, 135.E. J. W. Verwey and E. L. Heilmann, J. Chem. Phys., 1947, 15, 174.D. S. McClure, J. Phys. Chem. Solids, 1957, 3, 311.G. Blasse, Philips Res. Rept. Suppl. No. 3, 1964.A. Cimino, Chimica e Industria, 1974, 56, 27.' J. D. Dunitz and L. E. Orgel, J. Phys. Chem. Solids, 1957, 3, 20.lo J. Smit, F. K. Lotgering and R. P. van Stapele, J. Phys. SOC. Japan, 1962, 17, B-1, 268.l2 R. J. Beak and R. L. Cook, J. Amer. Ceram. SOC., 1957,40,279.l3 E. F. Bertaut, Compt. rend., 1950,230,213.l4 L. Weil, E. F. Bertaut and L. Bochirol, J. Phys. Radium, 1950,11,208.l 5 M. J. Buerger, Crystal Structure Analysis (J. Wiley, New York, 1960).l6 G. E. Bacon, Acta Cryst., 1952, 5, 684.l7 H. Furuhashi, M. Inagaki and S. Naka, J. Inorg. Nuclear Chem., 1973,35, 3009.l8 T. Suzuki, Acta Cryst., 1960, 13, 279.l 9 E. Brun, S. Hafner, P. Hartmann and F. Laves, Naturwiss., 1960, 47, 277.2o S. Hafner and F. Laves, 2. Krist., 1961, 115, 321.21 E. Stroll, P. Fischer, W. Halg and G. Maier, J. Phys., 1964, 25, 447.22 M. Drifford and P. Rigny, Compt. rend., 1966, 263, 180.23 0. Schmitz, Du Mont, H. Brokopf and K. Burhardt, 2. anorg. Cherrz., 1958,295, 7 .24 C. J. Ballhausen, Introduction to Lwand Field Theory (McGraw Hill, New York, 1962).25 F. Pepe, M. Schiavello and G. Ferraris, J. Solid State Chem., 1975, 12, 63.26 A. Cimino, M. LoJacono, P. Porta and M. Viligi, 2. Phys. Chem. (N.F.), 1970, 70, 166.27 A. Navrotsky and 0. J. Kleppa, J. Inorg. Nuclear Chem., 1967,29,2701.28 E. W. Gorter, PhiEips Res. Rept., 1954, 9, 295.29 S. Hafner, Schweiz. Min. petr. Mitt., 1960, 40, 207.30 F. Pepe, P. Porta and M. Schiavello, Proceedings of the 8th Int. Symp. of the React. of Solids(Gothenburg, Sweden, June 1976), preprint 1P27.(PAPER 7/353
ISSN:0300-9599
DOI:10.1039/F19777301972
出版商:RSC
年代:1977
数据来源: RSC
|
225. |
Van der Waals interaction between mica surfaces: comparison of theory and experiment |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 73,
Issue 1,
1977,
Page 1983-1987
John Gregory,
Preview
|
PDF (363KB)
|
|
摘要:
van der Waals Interaction between Mica Surfaces :Comparison of Theory and ExperimentBY JOHN GREGORYDepartment of Civil and Municipal Engineering,University College London, Gower Street, London WClE 6BTReceiced 7th March, 1977Published experimental results for the interaction of crossed mica cylinders are compared withtheoretical predictions based on a simple London-Hamaker approach, using only one dispersionfrequency in the ultraviolet. Empirical expressions of Overbeek are used to calculate the retardedinteraction. Over the whole range of separation distances (2-1 30 nm), satisfactory agreementbetween theory and experiment is found.The experimental measurements of dispersion (van der Waals) attraction betweencrossed mica cylinders by Tabor and Winterton and Israelachvili and Tabor areof great value for comparison with theoretical predictions.Unlike several previousmeasurements, which were made at comparatively large separation distances so thatonly the “ fully-retarded ” force was measured, the data of Tabor extend to separationsof the mica surfaces as small as 1.5 nm (in the unretarded region). Since the measure-ments span the region between unretarded and fully-retarded interaction, the result sprovide a rather sensitive test for current theories of dispersion forces.Some recent attempts have been made 3-5 to analyse the results in terms ofLifshitz theory.6 Most emphasis has been on the “power law index”, n, whichappears in the expression for the force per unit area, P, between parallel flat platesseparated by a distance D :P = -BID”.(1)For very close approach, where retardation is unimportant theory gives n = 3and for large values of D, in the fully-retarded region, n = 4. The experimentalresults clearly show the transition between these two values. Comparisons of n fromtheory and experiment have so far shown little agreement, the experimental valuesindicating a rather smaller retardation effect than that predicted from Lifshitz theory.However, for a number of reasons, such comparisons may be misleading (see Whiteet al.).5A more reasonable test is to compare directly the predicted force with the experi-mental values at a number of separations. This is not entirely straightforward sincethe experiments do not directly give the force between crossed mica cylinders(geometrically equivalent to a sphere-flat plate interaction), but the results can berelated to the force between flat plates at the same separation.Such a procedure isonly valid when the separation distance is very much less than the radius of thecylinders, so that the Deryagin approximation7 can be used. Since the cylinderradius was of the order of 1 cm and the separation distance not more than 130 nm,this approximation would be justified in all cases.1981984 V A N DER WAALS INTERACTXONFor small distances a “jump ” method was used, in which the distance betweenmica cylinders was reduced until the surfaces jumped into contact. The uppercylinder was mounted on an adjustable cantilever spring and, at the critical jumpdistance, the gradients of the van der Waals force and the spring force must be equal.The technique was to measure the critical jump distance, Do, as a function of springstiffness, K.For cylinders of radius R, the spring stiffness can be related to the forcebetween parallel flat plates separated by the jump distance, P(Do) :2nP(Do) = K/R. (2)The experimental results of Tabor and Winterton and Israelachvili and Taborare plotted in fig. 1 as Do against K/R. The former results show rather more scatter,but extend further into the retarded region. The results can now be compared withthe theoretical force between mica plates, using eqn (2). Although it is possible touse the Lifshitz t~eatment,~’~ extensive dielectric data are required for mica, whichare subject to some uncertainty.While there is no doubt about the essential correct-ness of Lifshitz theory, many computations may be in error because of the choice ofincorrect representations of dielectric behaviour.I o2 to3 to4 I o5 lo6 I 0’(K/R)/N m-2FIG. 1.--“ Jump distance ’’ results of Tabor and Winterton ( x ) and Israelachvili and Tabor (0).The full line represents calculated values based on the Overbeek expression for the retarded inter-action between plates, eqn (4), and the dashed line shows the corresponding unretarded behaviour.The values used for Hamaker constant and characteristic wavelength for mica are A = 1.13 xJ and h = 82.5 nm, both obtained from optical dispersion data.A much simpler approach is based on the assumptions of pairwise additivity ofintermolecular forces and of a single important dispersion frequency. The method,based on the work of London and Hamaker,g has been discussed previously loand leads to the following expression for the Hamaker constant.A = 27hvC -(>) n t -64 ni+2 (3)where h is Planck‘s constant, v, the characteristic dispersion frequency and no thelimiting refractive indexJ. GREGORYDispersion data for mica l1 give no = 1.581 and v, = 3.63 xThe retarded interaction between plates can be calculated fromHamaker constant from eqn (3) then becomes A = 1.13 x 10-19 J.formulae of Overbeek l 2 for two regions of separation :3A D < - *2n ’1985015 Hz.Thethe empirical1(4)where 2. is the “ characteristic wavelength ” for the interaction.Eqn (4) and (5) are based on Overbeek’s empirical expressions for retarded inter-molecular attraction,l which are convenient representations of the original resultof Casimir and P01der.l~ When D < A, eqn (4) gives the l / D 3 force dependencecharacteristic of unretarded interaction.For very large separations, D P A, eqn ( 5 )gives the 1 /D4 dependence expected for the fully-retarded force.If it is assumed that the characteristic wavelength corresponds with the dispersionfrequency (ie. A = c/v,), then 2. = 82.5 nm and this value, together with the value ofA for mica quoted above, can be inserted into eqn (4) and (5) to give the force betweenflat plates as a function of separation distance. Thus, the value of K/R correspondingto a particular jump distance can be calculated from eqn (2).The full line in fig. Ishows the results of such calculations. [In fact, only eqn (4) was used in thesecalculations since 3A/2n N 40 nm and all of the experimental jump distances werebelow this value.] The dashed line in fig. 1 shows the unretarded results, calculatedfrom only the leading term in eqn (4).Considering the extremely simple model employed and the fact that only opticaldispersion data were used in the calculations, the agreement between theory andexperiment is remarkably good. The Hamaker constant derived from experimentalresults in the unretarded region ( D < 10 nm) was 1.353-0.15 x J, which is alittle higher than the London-Hamaker value of 1.13 x 1O-l’ J calculated fromdispersion data, although the difference is not much greater than experimentaluncertainty.It is likely that some fortuitous compensation of errors has occurred,since the pairwise additivity assumption over-estimates the interaction, whereas theneglect of contributions from infrared and microwave frequency bands would leadto an under-estimate. (In neither case should the error be more than -20 %.)Nevertheless, it appears that retardation, which becomes significant at separationsgreater than -10nm, is well accounted for by eqn (4) which involves only onecharacteristic wavelength in the ultraviolet.used a “resonance” technique, in which one cylinder was vibrated at a knownfrequency, v, and the distance from the second cylinder was adjusted until the vibrationinduced in the latter was exactly 90” out of phase with the inducing vibration.Underthese conditions, the second cylinder was vibrating at its resonant frequency v(D),which must be the same as the forcing frequency v. For a series of forcing frequenciesthe separation distances at which resonance occurred were determined. It has beenshown 2 9 that :For larger separations of the mica surfaces (> 15 nm) Israelachvili and Taborv(o)2 = v:[I-,-] 2nRP(D)where v,, is the natural frequency of the cylinder at infinite separation and P(D) isthe force per unit area between flat plates separated by a distance D1 9 8 6 VAN DER WAALS INTERACTIONA change in frequency Av corresponds to a change in separation distance A Dand these changes are related by :v(D)Av - nRvc dP(D) - - ---AD K ao a(7)[There is an incorrect factor 2 in the corresponding expression of White et aZ.,’eqn (1.6), but this is corrected in their subsequent expressions and does not affectthe computations.]were given as a plot of log D against log [AD/v(D)Av],as in fig.2. For comparison with the Overbeek force expressions, eqn (4) and (5),these must be differentiated to give aP/dD as a function of D. The results can thenbe combined with the known values of K, R and v, to give [AD/v(D)Av] from eqn (7).The full line in fig. 2 shows the results of this procedure. The dashed line representsthe limiting (fully-retarded) behaviour.Experimental resultsI 1 I tI I I I I 1lo2 .lo’ I lo-’ 1 o-2 I 0-3rAW(m.1FIG.2.--“ Resonance ” results of Israelachvili and Tabor. The full line is calculated from eqn (4)and (5) and the dashed line represents the fully-retarded behaviour. Values of A and h as in fig. 1 .For the resonance results in fig. 2 the agreement between experiment and simpletheory is even better than for the jump results. In the fully-retarded region (> 50 nm)the coefficient of l/D4 in eqn (1) was estimated to be 0.97k0.06 x J m. Thecorresponding value from eqn (5) is 1 . 1 6 x J m. The experimental transitionfrom partially- to fully-retarded behaviour is wellmatched by the Overbeek expressions.It has been shown that the experimental results for dispersion interaction betweencrossed mica cylinders over a wide range of separations (2-130 nm) agree well withcalculations based on a very simple London-Hamaker approach using only onedispersion frequency in the ultraviolet. This apparent agreement is difficult toreconcile with the conclusion of White et aL5 that U.V.terms do not contributesignificantly to the interaction at separations greater than - 5 nm.Note added in proofi Calculations of van der Waals forces for mica, based on Lifshitz theory andusing extensive dielectric data, have recently been reported by Chan and Richmond.” For shortrange interaction the results were considerably lower than Tabor’s experimental data,ln2 a though forlong range interaction the agreement was much betterJ . GREGORY 1987D. Tabor and R. H. S. Winterton, Proc. Roy. SOC. A, 1969,312,435.J. N. Israelachvili and D. Tabor, Proc. Roy. SOC. A, 1972, 331, 19.P. Richmond and B. W. Ninham, J. Colloid Interface Sci., 1972, 40, 406.K. B. Lodge, J. Colloid Interface Sci., 1975, 50, 462.L. R. White, J. N. Israelachvili and B. W. Ninham, J.C.S. Furaday I, 1976, 72, 2526.E. M. Lifshitz, Sou. Phys. JETP, 1956,2,73.B. V. Deryagin, Kolloid Z., 1934, 69, 155.F. London, Z . Physik, 1930,63,245.H. C. Hamaker, Physica, 1937,4, 1058.lo J. Gregory, Adv. Colloid Interface Sci., 1969, 2, 396.l1 Landolt-Bornstein, Zahlenwerte und Funktionen (Springer-Verlag, Berlin, 6th edn, 1962),l2 J. Th. G. Overbeek, quoted by A. van Silfhout, Proc. Kon. Akad. Wetensch. 23, 1966, 69, 501.l 3 J. Th. G. Overbeek, in Colloid Science, ed. H. R. Kruyt (Elsevier, Amsterdam, 1952), vol. I,l4 H. B. G. Casimir and D. Polder, Phys. Reu., 1948,73, 360.l5 I?. Chan and P. Richmond, Proc. Roy. SOC. A , 1977, 353, 163.vol. 11, part 8.p. 266.(PAPER 7/398
ISSN:0300-9599
DOI:10.1039/F19777301983
出版商:RSC
年代:1977
数据来源: RSC
|
226. |
Infrared study of the adsorption of [2H4]acetic acid on to rutile |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 73,
Issue 1,
1977,
Page 1988-1997
David M. Griffiths,
Preview
|
PDF (930KB)
|
|
摘要:
Infrared Study of the Adsorption of[2H4]Acetic Acid on to RutileBY DAVID M. GRIFFITHS AND COLIN H. ROCHESTER*Chemistry Department, The University, Nottingham NG7 2RDReceived 4th April, 1976Infrared spectroscopic studies are reported of the adsorption of [2H4]acetic acid onto three samplesof rutile which differed in their extents either of hydration or of reduction. Acetic acid was associ-atively adsorbed on Lewis acidic electron acceptor Ti4+ sites in the rutile surface: this mode ofadsorption was prevented by the preadsorption of water or by reduction of rutile in hydrogen.Weakly adsorbed acetic acid dimers appeared on the surface of rutile which had been pretreated inoxygen but not on reduced rutile. Chemisorption of acetic acid generated water, surface deuteroxylgroups and [2H3]acetate ions predominantly in a highly symmetrical bidentate chelate configuration.Both the presence of adsorbed water and prereduction of rutile in hydrogen hindered the chemi-sorption of acetic acid on rutile.Information concerning the character of acidic sites on the surface of rutile hasbeen gained by infrared studies of the adsorption of pyridineSimilar studies of the adsorption of carbon dioxide,2* 4*and hydrogen chloride 6 * have helped to establish the character of the titaniumdioxide surface. More recently additional information concerning the relativereactivity of surface functional groups acting as adsorption sites has been gainedby the carefully controlled addition of increasing amounts of an adsorbate.s Ex-posure of rutile to hexafluoroacetone vapour in equilibrium with the solid or liquidketone in a cryostat, gave infrared spectra which enabled the interactions betweenhexafluoroacetone and hydrogen-bond donor, electron acceptor, electron donor andnucleophilic OH- or 02- surface sites to be distinguished.8 This paper reportssimilar studies of the adsorption on rutile of acetic acid, chosen because of its potentialability to act as a Brsnsted acid, or as a donor or acceptor of hydrogen bonds, or as adonor of electrons (Lewis base). Studies of the effects of reduction in hydrogenupon the surface properties of rutile 8 * For reasons givenbefore rutile surfaces were fully deuterated and, in order to avoid complicationsdue to isotopic exchange reactions, deuterated acetic acid CD,COOD was used asadsorbate.and ammonia.2*sulphur dioxide,6 phenolhave been continued.EXPERIMENTALDetails of the source of rutile (surface area 30.3 m2 g-l) and the pretreatment proceduresto which the oxide was subjected are given elsewhere.8* [2H4]A~eti~ acid (CIBA) of isotopicpurity > 99.5 atom % D was degassed before use.Infrared spectra were recorded asbefore and [2H4]acetic acid was admitted to the infrared cell from a c r y o ~ t a t . ~ ~ lo Rutilewas subjected to three different series of pretreatments before exposure to [2H4]acetic acidvapour. The resulting oxide discs corresponded to the rutile samples A, B and C, the inter-actions of which with hexafluoroacetone had also been studied.8 The surfaces of all threesamples were fully deuterated.Samples A and B had both been heated in oxygen butwhereas sample A was subsequently evacuated at 673 K before being cooled to room tem-perature, sample B was equilibrated with saturated deuterium oxide vapour before evacuation198D. M. GRIFFITHS AND C . H. ROCHESTER 1989at room temperature. Sample C had been heated in hydrogen before evacuation at 673 Kand being cooled to room temperature. Full preparative details for samples A, B and Chave already been described.*RESULTSRUTILE SAMPLE AThe infrared spectrum of rutile sample A contained only weak bands at 2720 cm-'due to surface deuteroxyl groups and at - 2500 cm-l (broad) due to a low residualconcentration of strongly adsorbed molecular deuterium oxide [fig.1 (a)]. Theaddition of [2H4]acetic acid led to the appearance of a maximum at 1685cm-'which with increasing pressures (- 3-10 N m-2) of [2H,]acetic acid broadened[fig. l(b),(c)] and finally gave a narrower maximum at 1670 cm-l with a shoulder at2700 23 00 1600 140010080604 0200wavenumber Icm-'FIG. 1 .-Spectra of rutile sample A. (a) Initial surface, (6)-(e) after contact with increasing pressuresof t2H4]acetic acid vapour (BT) and evacuation (BT), (f) in contact with vapour before evacuation(1.5 h) gave spectrum (e).1685 cm-I [fig. l(d)]. Although never clearly resolved it appeared that there were twobands at 1685 and 1670 cm-l. These bands may be assigned to the C=O stretchingvibrations of [2H4'Jacetic acid molecules associatively adsorbed through coordinativeinteractions between the carbonyl oxygen atoms and two types of surface electronacceptor site.These sites were probably incompletely coordinated Ti4+ ions possiblyin the (100) and (101) surface planes of r ~ t i l e . ~ * Two types of Lewis acidic electronacceptor site on rutile have previously been distinguished by infrared studies of theadsorption of ammonia and ~yridine.'-~Chemisorption of C2H4]acetic acid on rutile sample A led to the appearance oftwo narrow infrared bands at 1480 and 1440cm-l due to surface [2H,]acetatespecies. A broad but weaker general increase in absorption intensity also occurredin the range 1510-1 580 cm-l. No maxima were clearly resolved in this range althoug1990 ACETIC ACID ON RUTILEat least two bands at - 1530 and - 1570 cm-1 were certainly present.The genera-tion of surface [2H3]acetate species was accompanied by the appearance of infraredbands at 2745, 2720(sh), 2690, 2660, 2610 and 2530 cm-1 [fig. l(d)] due to surfacedeuteroxyl groups (and adsorbed molecular deuterium oxide, responsible for a broadband centred at - 251Ocm-l). Redeuteroxylation of the oxide surface in the pre-sence of [2H4]acetic acid [fig. l(a)-(d)] followed exactly the same pattern as re-deuteroxylation of rutile with deuterium oxide vapour.3of [2H,]acetic acid to sampleA caused the progressive displacement of coordinatively bound adsorbate moleculesfrom the rutile surface. The bands at 1685 and 1670cm-l became weaker [fig.l(d), (e) and 2(a)] as the intensities of the bands at 1440 and 1480 cm-I continued toincrease.Bands also appeared at 1320 and 1365cm-l [fig. 2(a), (b)] and may beassigned to C-0 stretching vibrations of adsorbed carboxylate species.Admission of high vapour pressures (> 27 N10080?Q402002700 2300 1800 1600 14.00wavenumberlcm-FIG. 2.-Continuation of fig. 1. (a), (b) Further doses of [2H4]acetic acid followed by evacuation(BT), (c) sample in contact with vapour. Evacuation (1.5 h) gave spectrum (b), ( d ) after saturationwith [2H4]acetic acid vapour and evacuation at a series of increasing temperatures with a final tem-perature of 616 K (1.8 h), (e) subsequent evacuation (643 K, 2.3 h).At low pressures (< 27 N m-2) of [2H4]acetic acid [fig.l(b)-(d)] evacuation of theinfrared cell after contact between rutile and adsorbate showed no changes in infraredspectra. The species responsible for the observed infrared bands could not bedesorbed by evacuation at ambient temperatures. However, after contact betweenrutile sample A and higher pressures ( N 27-100 N m-2) of E2H4]acetic acid evacuationproduced changes in spectra which indicated the presence of weakly adsorbed specieson the oxide surface. Adsorbed [2H,]acetic acid molecules formed deuterium bondswith surface deuteroxyl groups. Evacuation reduced the intensity of the broadband centred at N 2530 cm-l, in part due to perturbed deuteroxyl groups, and led toincrases in the intensities of the bands at 2710 and 2690 cm-1 due to unperturbedsurface deuteroxyl groups [fig.l(e), (f)]. A band at 1670 cm-1 was weakened byevacuation of the oxide after exposure to ['H,]acetic acid [fig. l(e), df)]. At higheD. M. GRIFFITHS AND C. H. ROCHESTER 1991adsorbate pressures (> 250 N m-2) this effect was magnified [fig. 2(b), (c)] and, inthe presence of adsorbate, spectra contained two bands at 1670 (sh) and 1655 cm-'[fig. 2(c)]. The band at 1670 cm-1 differed in behaviour from the maximum at thesame position which appeared at low adsorbate pressures and which was assigned tocoordinatively bound [2H4]acetic acid molecules. The bands at 1670 and 1655 cm-1in spectra of oxide in contact with higher vapour pressures of [2H4]acetic acid areassigned to weakly adsorbed [2H4]acetic acid dimers.l The general decrease inintensity of absorption in the range 2200-2550 cm-l which accompanied the loss ofthe bands at 1670 and 1655 cm-' [fig.2(b), (c)] may also be ascribed to the desorptionof dimer from the surface. After evacuation at beam temperature a residual shoulderremained at 1650 cm-l [fig. 2(6)] suggesting that some dimer or a form of stronglydeuterium bonded [2H,]acetic acid molecules was retained on the surface. A weakband also persisted at 1725 cm-1 [fig. 2(b)].Infrared bands at 2720 and - 2690 cm-l [fig. 1(6)-(d)] due to isolated surfacedeuteroxyl groups which were initially generated when C2H4]acetic acid adsorbedon rutile sample A were removed from spectra as the surface concentration of acetatespecies was further increased.A residual band at 2540 cm-l [fig. 2(6)] suggests thatbridged deuteroxyl groups on the (1 lo> planes of rutile 9 * l 2 were not completelydisplaced by adsorbed acetate species. A band at 2600 cm-1 [fig. 2(b), (c)] is assignedto the OD-stretching vibration of surface deuteroxyl groups on rutile.After treatment of sample A with t2H,]acetic acid [fig. 2(b)] the infrared cell wasevacuated with the oxide disc at a series of progressively increasing temperatures.Heat treatment at 388K caused the disappearance of the shoulder at 1650cm-Ialthough after evacuation at 436 K a narrower and more intense maximum appearedat 1630 cm-l. The latter band in turn was weakened by evacuation at 488 K anddisappeared following heat treatment at 523 K. At the same temperature the overallintensity of absorption in the range 1500-1600 cm-I was reduced and distinct maximabecame apparent at 1510 and 1540 em-'.At increasing temperatures up to 616 Kthe majority of the adsorbed surface acetate species, including those responsible forthe strong maxima at It440 and 1480 cm-I, were desorbed. Infrared bands remainedat 1510,1470 and 1430 cm-l [fig. 2(d)]. The loss of surface acetates was accompaniedby the disappearance of all infrared bands due to the OD-stretching vibrations ofsurface deuteroxyl groups. Heat treatment at 643 K completed the loss of deuteroxylgroups and acetate species from the oxide surface [fig. 2(e)]. Mass spectra of thedesorbed vapours showed prominant maxima at rn/e values of 18 (OD and CD,),44 (CO,), 46 (CD3C0 and COOD) and 64 (CD,COOD) showing that the maindesorptionproducts were C2H,]acetic acid and carbon dioxide.Maxima at 98, 1 14 and132 were due to higher molecular weight products.RUTILE SAMPLE BThe infrared spectrum of rutile sample B contained bands at 2745, 2710(sh),2690,2660,2610 and 2530 cm-l [fig. 3(a)] the assignment of which to surface deuter-oxyl groups has been discussed el~ewhere.~ The adsorption of [,H,]acetic acidinitially led to a reduction in the intensity of the band at 2690 cm-' and a concomitantincrease in the overall intensity of absorption in the range 2300-2680cm-I [fig.3(u), (b)]. The bands at 2660 and 2610 cm-1 became more intense. These spectro-scopic changes were due to the generation of deuterium oxide which remained onthe rutile surface in the presence [2H4]acetic acid vapour.The deuterium oxide wasdesorbed by evacuation of the sample at ambient temperature. Resulting spectrain the 2300-2800 cm-1 region were similar, apart from a slight decrease in the intensit1992 ACETIC ACID ON RUTILEof the band at 2690 cm-', to the starting spectrum for sample B even after appreciableadsorption of C2H4]acetic acid had taken place [fig. 3(a), (41. The latter was indicatedby the presence in spectra of bands at 1525 (sh), 1480, 1440 and 1365 cm-1 due tovibrations of surface [2H3]acetate complexes. However after contact between rutilesample B and higher vapour pressures of [2H4]acetic acid the infrared bands at 2745,2710 and 2690 cm-l, due to isolated deuteroxyl groups, progressively disappearedas the surface concentration of acetate species increased [fig.3(d),(f)]. The final2700 2300 1600 1400wavenumber/cm-'FIG. 3.-Spectra of rutile sample B. (a) Initial surface, (6) in contact with [2H4]acetic acid vapour,(c) after further doses of adsorbate and evacuation (BT), (d) further doses and evacuation (BT), (e) incontact with higher vapour pressure of adsorbate, cf) subsequent evacuation (BT).spectra [fig. 3Cf)l were similar to those recorded after exposure of rutile sample A tohigh pressures of [2H4]acetic acid vapour [fig. 2(b)I except that more molecularlyadsorbed deuterium oxide (broad band centred at - 251Ocm-l) was retained bysample B. Also, in the presence of C2H,]acetic acid vapour, the band at 1655 cm-l[fig. 2(c)] assigned to adsorbed [2H,]acetic acid dimer, was accompanied by a narrowband at 1690 cm-1 [fig.3(e)]. This band may be assigned to deuterium bonded[2H,]acetic acid molecules, possibly dimer, adsorbed through interactions with surfacedeuterium oxide.RUTILE SAMPLE CResults for the adsorption of [2H,]acetic acid onto the reduced rutile sample Care shown in fig. 4. [2H3]acetate species were formed on reduced rutile and led to theappearance of infrared bands at 1365, 1440, 1480 and 1510-1580 (broad shoulder)cm-l. Initially the surface was to some extent redeuteroxylated [fig. 4(b)] althoughin the final stages of adsorption the deuteroxyl groups reacted and the infrared bandsat 2720 and 2690 cm-l disappeared [fig.4(f)]. A band at 1630 cm-1 in spectra ofsample C [fig. 4(e), (f)] was only observed in spectra of sample A after the latter hadbeen exposed to r2H4]acetic acid and subsequently evacuated at 388 K. Spectra ofsample C also exhibited a weak band at 1735 cm-l. In the presence of the highesD. M. GRIFFITHS AND C. H. ROCHESTER 1993vapour pressures of ['H,]acetic acid studied sample C failed to give spectra with bandsat 1655, 1670, or 1690 cm-l which could be assigned to weakly adsorbed deuteriumbonded or dimeric [2H4]acetic acid species.2900 27001 I I I 1 I1800 1600 1400wavenumber /errFIG. 4.-Spectra of rutile sample C. (a) Initial surface, (b)-(d; in contact with increasing pressuresof [2H4]acetic acid vapour, (e) cf) after contact with high pressures of adsorbate and evacuation (BT).DISCUSSIONInfrared bands at 1670 and 1685 cm-' [fig.l(b)-(e)] in spectra of ['H,]acetic acidadsorbed on rutile sample A were assigned to adsorbate molecules coordinativelyliganded to Ti4+ ion sites in the rutile surface. The shift Avco = - 100 cm-l fromthe corresponding band position for ['H,]acetic acid in the vapour phase suggeststhat the acid molecules were fairly strongly adsorbed which is consistent with thefact that they could not be desorbed by evacuation at ambient temperature. Aswould be expected from electronic considerations hexafluoroacetone was more weaklyadsorbed (shift Avco smaller and considerably weaker band due to the carbonylstretching vibration of coordinatively adsorbed species) than [2H4]acetic acid.However, as for hexafluoroacetone,* l2H4]acetic acid did not displace coordinativelyliganded deuterium oxide from Ti4+ ion sites.The infrared bands at 1670 and 1685cm-1 did not appear in spectra of ['H4]acetic acid adsorbed onto rutile sample Bwhich had been equilibrated with saturated deuterium oxide vapour before exposureto the acid. The existence of coordinatively bound water on rutile has been attributedto the existence of incompletely coordinated Ti4+ ions in the {loo) and { 101) surfaceplanes of the oxide.g* l2 Deuterium oxide was more strongly adsorbed on the Ti4+ion (Lewis acidic 1-3) sites than either hexafluoroacetone * or E2H,]acetic acid. TheLewis acidity of the surface of rutile was decreased by reduction of the oxide inhydrogen.Adsorption of [2H4]acetic acid onto the reduced rutile sample C gave noinfrared evidence for adsorbed ['H4]acetic acid molecules liganded to surface Ti4+ion sites. Reduction of rutile causes the removal of 0'- ions and the generationof Ti3+ ions on the surface planes of the oxide.14-16 The inability of the reducedrutile to adsorb ['H4]acetic acid through coordinative interactions involving Ti41994 ACETIC ACID ON RUTILEion sites parallels the marked decrease, caused by reduction of rutile in hydr~gen,~in the uptake by rutile of associatively liganded water or deuterium oxide molecules.[2H4]Acetic acid dimers were adsorbed onto rutile discs and could be desorbedby evacuation at ambient temperature. The carbonyl (C=O)-stretching vibrationof dimer in the vapour phase gives an infrared band at 1725 cm-l.l3 For liquid[2H4]acetic acid the band is at 1708 The existence of bands at 1655, 1670and 1690cm-' in spectra [fig.2(c) and 3(e)J of adsorbed species shows that thecarbonyl stretching vibrations of the diniers were further perturbed by the interactionswith the rutile surface. The band at 1690 cm-I was most intense for rutile sample Bwhich had been pre-equilibrated with deuterium oxide and is therefore assigned todimers which were involved in deuterium bonding interactions with adsorbeddeuterium oxide molecules. Dimers responsible for the bands at 1670 and 1655 cm-Iwere probably more strongly adsorbed since they could not be completely removed[fig. 2(b) and 3(f)] by evacuation at beam temperature (- 318 K).Weakly electronaccepting Ti4+ ion sites were possibly involved. The analogy can be drawn betweenthe bands at 1670 and 1655 cm-I due to dimers weakly adsorbed onto two types ofTi4+ ion site, and the bands at 1685 and 1670 cm-l due to more strongly adsorbedE2H4]acetic acid monomer liganded to strongly electron accepting (Lewis acidic) Ti4+sites. The latter sites were destroyed by the subsequent chemisorption of [2€-14]aceticacid to give [2H,]acetate surface species. [2H4]A~eti~ acid dimers were not adsorbedon rutile which had been reduced in hydrogen. This is consistent with the loss ofLewis acidity of rutile caused by reduction, and with the absence of adsorbed mole-cular deuterium oxide on the reduced sample C after treatment with C2H,]acetic acid.A previous study of the adsorption of acetic acid on rutile reported infraredbands at 1460 and 1545 cm-I which were assigned to surface acetate ions.17 Thepressure of acetic acid vapour to which the rutile was exposed (1.3 kN m-2) wasconsiderably greater than the pressures for which results are presented here.Infra-red bands due to surface acetate ions formed when acetic acid adsorbs on to anatasehave been observed at 1400 and 1530 cm-',18 or at 1410 and 1555 cm-1.2 Corres-ponding bands at 1424 and 1581 cm-I and at 1425 and 1530 cm-' have been reportedfor acetic acid adsorbed onto magnesium oxide and tin(1v) oxide 2o respectively.Acetic acid is adsorbed on alumina to give two types of acetate species with infraredbands at 1465 and 1590 cm-I and at 1420 and 1560 cm-1.21 The infrared bandsbetween 1400 and 1600 cm-l reported here are similarly assigned to several types ofC2H3]acetate species formed on the surface of rutile by the chemisorption of [2H4]-acetic acid.The most prominant maxima at 1440 and 1480 cm-I are assigned to thesymmetric and asymmetric (COO)-stretching vibrations respectively of one form ofsurfxe acetate ions which constitute the predominant adsorption product. Thespectra of rutile treated with C2H4]acetic acid and evacuated at high temperatures[fig. 2(d)] suggest that a maximum at 1430 cm-l, usually masked by the strong bandat 1440 cm-I, may be assigned to the symmetric (COO)-stretching vibration of surfaceacetate ions for which the corresponding asymmetric (COO)-stretching vibrationgave a maximum at 1510 cm-l.Both rutile samples A and C showed more intenseshoulders at - 1570 cm-I than sample B for which the band at 1480 cm-l was alsorelatively weaker than the band at 1440 cm-I. It would appear therefore that theband at - 1570 cm-I is possibly associated with a maximum at - 1480 cm-I whichis masked by the more intense band at the same position. The analysis involvinginfrared bands in the 1500-1600cm-1 region of the spectrum must be tentativebecause, apart from during desorption experiments, the bands merged into a broadshoulder on the maximum at 148Ocm-l, and could not be resolved into distinctmaxima. The spectra do, however, indicate the probable existence of three distincD. M .GRIFFITHS AND C . H . ROCHESTER 1995structural types of C2H,]acetate ion on rutile following treatment of the oxide with[ H4] a ce t ic acid.The difference Avco between the positions of the infrared bands due to the asym-metric and symmetric (COO)-stretching vibrations in transition metal acetates isgenerally > 100 cn1-1.22-24 For the [2H3]acetate ion in solution Avco is 139 CM--'.~~The values of Avco deduced from the present assignments of infrared bands for[2H3]acetate species on rutile are 90 cm-I (maxima at 1480 and 1570 cm-l), 80 cm-l(1430, 151Ocm-l) and 40cm-l (1440, 1480cm-'). These low values of Avco arematched by a similar value of 85 cm-I for trifluoroacetate ions formed on rutile bythe chemisorption of hexafluoroacetone.' Values of Avco below 80 cm-1 and downto 33 cm-I have been reported for a series of carboxylato complexes of ruthenium (11),the structure of which involved a symmetrical bidentate chelate arrangement of thecarboxylate groups.26-28 The values of Avco are greater for monodentate carboxylatocomplexes 29 than for bidentate complexes.27 Reports of spectra of complexes withthe molecular formula TiC13(CH3C00) differY3O* 31 although one complex of un-determined structure gave infrared bands at 1410 and 1480 cm-' (Avco = 70 ~ m - ' ) . ~ ~Values of Avco for cyclopentadienyltitanium(II1) carboxylates were sensitive to themode of coordination of the carboxylate groups.32 For unidentate carboxylatessplittings of up - 300cm-I were observed.For the dimers of C~T~(CH,COO)~and C ~ T ~ ( C ~ H ~ C O O ) ~ , in which the carboxylate groups were bridged by simultaneouscoordination to two titanium ions, the splitting Avco was 170 cm-I (bands at 1425and 1595 cm-I). The bidentate chelate structure of carboxylate groups in the com-plexes cp2Ti(CH3COO) and cp2Ti(C2H,COO) gave splittings of 60 cm-I (bands at1460 and 1520 cm-l) and 50 cm-l (1465, 1515 cm-l) respectively. By analogy withthe results for titanium(rr1) 32 and ruthenium@) 26-29 complexes the infrared bandsdue to [2H4]acetic acid adsorbed on rutile may be assigned to C2H3]acetate ionspresent on the surface in the chelating bidentate configuration(1) rather than in thebridged bidentate configuration(I1) or the unidentate configuration(Ii1).Despite al i I 6- 0-Ti4'7 3o=cI 0-ITi4'warning about the general validity of correlations between A Y ~ ~ and the structure ofcarboxylate ligands 24 the present conclusion would appear to be particularly reason-able for the predominant surface acetate species, for which Avco = 40 cm-l.The formation of C2H3]acetate ions on rutile by the adsorption of [2H4]acetic acid(1)CD3COOD +OD-+ CD,COO-+D20 (2)CD3COOD + 02- + CD,COO-+ OD-will proceed through reactions (1) and (2) which also lead to the generation of surfacedeuteroxide ions and deuterium oxide. Rutile samples A and C retained only lowresidual surface concentrations of deuteroxyl groups after evacuation at 673 K butwere redeuteroxylated in the presence of E2H,]acetic acid.Sample A retained someadsorbed molecular deuterium oxide after evacuation at beam temperature wherea1996 ACETIC ACID ON RUTILEspectra of the reduced rutile sample C gave no evidence for the retention of deuteriumoxide which must therefore have been desorbed. This result is consistent with theobservation that deuterium oxide is much less readily associatively adsorbed onreduced rutile than on rutile that has been preheated in ~ x y g e n . ~ Row A and row Bdeuteroxyl groups on the (1 10) surface planes of rutile 9 9 l2 differed in their reactivitytowards C2H4]acetic acid. Row A deuteroxyl groups reacted with [2H4]acetic acid[reaction (2)] to give C2H3]acetate ions and deuterium oxide which was dcsorbed.Row B deuteroxyl groups, which were responsible for the infrared band at 2530 cm-lin spectra [fig.3(a)] of rutile equilibrated with deuterium oxide at ambient temperature,were apparently less reactive towards [2H4]acetic acid. Bridged row B deuteroxylgroups on rutile have previously been found to be unreactive towards hexafluoro-acetone adsorbate.8 The adsorption of [2H4]acetic acid on rutile generated [reaction(l)] deuteroxyl groups giving an infrared band at 2600 cm-I [fig. 2(c) and 3(f)]. Inthe presence of deuterium oxide rutile gives an infrared band at 2610cm-I (OH-equivalent 3520 cm-') which has tentatively been assigned to deuteroxyl groups onthe { 1001 planes of rutile which are involved in deuterium bonding interactions withadjacent 02- or OD- ions or with physisorbed deuterium oxide molecules.In thesame way the band at 2600 cm-l may be tentatively attributed to deuteroxyl groupswhich are perturbed to some extent by interactions with adjacent [2H3]acetate specieson the oxide surface. In general the formation of ['H,]acetate species (judged bythe intensities of the infrared bands at 1440 and 1480 cm-1 for a particular cryostatvoltage) occurred less readily on sample B than on sample A. The presence of ahigh surface coverage of molecularly adsorbed deuterium oxide on rutile hindered thechemisorption of [2H4]acetic acid on the oxide. The formation of [2H3]acetatespecies also occurred less readily on the reduced rutile sample C than on rutile sampleA which had not been preheated in hydrogen.Thermal activation led to the desorption of [2H3]acetate species from the rutilesurface.Mass spectrometric analysis confirmed that a major desorption productwas r2H4]acetic acid. The desorption process involved the reverse of reactions (1)and (2). An infrared band at 1630 cm-1 was observed in the early stages of desorp-tion and could be assigned to unidentate C2H3]acetate ions. Unidentate acetate ionsin the titanium@) complex c~,T~(CH~COO)~ give a strong infrared band at - 1630formed at the surface by further reaction of acetate groups. The appearance ofcarbon dioxide in the desorption products might have involved the intermediateformation of surface carbonate or could have arisen from direct decarboxylation of[2H,]acetate groups. In either case the species responsible for the band at 1630 cm-Iwas also formed by the adsorption of [2H4]acetic acid onto reduced rutile sample Cat beam temperature.The formation of trifluoroacetate groups by the adsorptionof hexafluoroacetone occurred more readily on reduced rutile than on rutile whichhad been preheated in oxygen.8 By analogy it might be expected that carbonatespecies would be more readily formed from carboxylato groups on the surface ofreduced rutile. The bands at 1630 and 1725 cm-I in spectra of [2H4]acetic acid onreduced rutile [fig. 4 0 1 could therefore be due to vibrations of bidentate and bridgedcarbonate species 34 respectively. However the assignment of both bands tounidentate carboxylate species cannot be ruled out. Qn this interpretation the bandat 1630 cm-I would be due to ionic unidentate [2H,]acetate species 32 whereas thespecies responsible for the band at 1725 cm-' would be more covalent in character.A maximum at 1715 cm-l in spectra of acetic acid adsorbed on germania has beenassigned to the (C=Q)-stretching vibration of surface ester groups.33 The occurrenceof peaks at mfe values of 98, 114 and 132 in mass spectra of the vapours desorbedcm-l .3 2 Alternatively the band could be assigned to bidentate carbonate species 34* 3D. M. GRIFFITHS AND C . H . ROCHESTER 1997from rutile treated with [2H,]acetic acid was probably due to the formation of [’H6]-phloroglucinol by the condensation and cyclization of [2H,]acetic acid molecules orsurface [2H,]acetate groups.The authors thank Tioxide International Ltd.for the award of a studentship(to D. M. G.).G. D. Parfitt, J. Ramsbotham and C. H. Rochester, Trans. Favaduy SOC., 1971, 67, 1500.M. Primet, P. Pichat and M.-V. Mathieu, J. Phys. Chem., 1971, 75, 1221.G. D. Parfitt, J. Ramsbotham and C. H. Rochester, Truns. Faruduy SOC., 1971,67, 841.D. J. C . Yates, J. Phys. Chem., 1961, 65, 746.P. Jackson and G. D. Parfitt, J.C.S. Furuduy I, 1972,68,896.P. Jones and J. A. Hockey, Truns. Furuduy SOC., 1971, 67,2669.G. D. Parfitt, J. Ramsbotham and C. H. Rochester, Trans. Furaduy SOC., 1971, 67, 3100.D. M. Griffiths and C. H. Rochester, J.C.S. Furaday I, in press.D. M. Griffiths and C. H. Rochester, J.C.S. Faruday I, 1977, 73, 1510.lo D. R. Ashmead and R. Rudham, Chem. and Ind., 1962,401.l 1 M.Haurie and A. Novak, J. Chim. phys., 1965,62,137.l2 P. Jones and J. A. Hockey, Trans. Furuday Soc., 1971, 67, 2679.l3 M. Haurie and A. Novak, J. Chim. phys., 1965, 62, 146.l4 P. G, Gravelle, F. Juillet, P. Meriaudeau and S. J. Teichner, Disc. Faruduy SOC., 1971,52, 140.l5 P. C. Richardson, R. Rudham, A. D. Tullett and K. P. WagstafF, J.C.S. Furuduy I, 1972, 68,l6 R. D. Iyengar and M. Codell, Adv. Colloid Interface Sci., 1972, 3, 365.l7 A. V. Kiselev and A. V. Uvarov, Surfuce Sci., 1967, 6, 399.l8 R. Flaig-Baumann, M. Herrmann and H. P. Boehm, 2. anorg. Chem., 1970,372,296.l9 M. J. D. Low, H. Jacobs and N. Takezawa, Water, Air and SoiZPuZlution, 1973, 2, 61.2o E. W. Thornton and P. G. Harrison, J.C.S. Furaday I, 1975, 71,2468.21 S. Hayachi, T. Takenaka and R. Gotoh, Bull. Inst. Chem. Res., Kyoto Uniu., 1969, 47, 378.22 A. I. Grigor’ev, Russ. J. Inorg. Chem., 1963, 8, 409.23 A. I. Grigor’ev and V. N. Maksimov, Russ. J. Inorg. Chem., 1964,9, 580.24 D. A. Edwards and R. N. Hayward, Canad. J. Chem., 3968,46, 3443.2 5 K. Ito and H. J. Bernstein, Canad. J. Chem., 1956, 34, 170.26 D. Rose, J. D. Gilbert, R. P. Richardson and G. Wilkinson, J. Chem. SOC. (A), 1969,2610.27 S. D. Robinson and M. F. Uttley, J.C.S. Dalton, 1973, 1912.28 R. J. Young and G. Wilkinson, J.C.S. Dalton, 1976, 719.29 T. A. Stephenson, S. M. Morehouse, A. R. Powell, J. P. Heffer and G. Wilkinson, J, Chem.30 Y. A. Lysenko and L. I. Khokhlova, Russ. J. Inorg. Chem., 1974, 19, 690.31 J. Amaudrut and C. Devin, Bull. SOC. Chim. France, 1975, 1933.32 R. S. P. Coutts, R. L. Martin and P. C. Wailes, Austral. J. Chem., 1973, 26, 941.33 J. C. McManus and M. J. D. Low, J. Phys. Chem., 1968,72,2378.34 L. H. Little, Infrured Spectra of Adsorbed Species (Academic, London, 1966), p. 83.35 J. V. Evans and T. L. Whateley, Trans. Furuday SOC., 1967, 63,2769.2203.Soc., 1965, 3632.(PAPER 7/584
ISSN:0300-9599
DOI:10.1039/F19777301988
出版商:RSC
年代:1977
数据来源: RSC
|
227. |
Theory of compartmentalised free-radical polymerisation reactions. Part 1 |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 73,
Issue 1,
1977,
Page 1998-2009
David T. Birtwistle,
Preview
|
PDF (878KB)
|
|
摘要:
Theory of Compartmentalised Free- RadicalReactionsPart 1Polymerisati OMBY DAVID T. BIRTWISTLE AND DAVID C. BLACKLEY*Department of Mathematics and the National College of Rubber Technology,The Polytechnic of North London, Holloway Road, London N7 8DBReceived 4th April, 1977Explicit analytic solutions are given for nr, the number of reaction loci per unit volume containingY radicals, as a function of time, t, for the case of a seeded emulsion polymerisation which fulfilsthe following conditions : (1) radicals enter the reaction loci from a contiguous external phase at aconstant rate ; (2) the only significant processes which result in loss of radical activity from reactionloci are kinetically of first order with respect to the concentration of radicals in the loci ; (3) radicalslost by diffusion from loci to the external phase are not available for re-initiation ; (4) the volumeof the reaction loci is miform and does not increase significantly as polymerisation proceeds ; ( 5 )no nucleation of new loci takes place ; (6) no reduction in the total number of reaction loci occurs,e.g., through agglomeration.The general expression obtained for nr iswhere N is the total number of reaction loci per unit volunie of reaction system, u is the average rateof entry of radicals into a single locus, and k characterises the rate of loss of radical activity by first-order processes.This result is obtained by first deriving the following generating function for thelocus populations :Y((, t ) = Nexp { ~(e-1)(1-e-k')}where f is an auxiliary variable.The implications of the result for ndt) for (1) the approach to thesteady state, (2) the average number of radicals per locus, and (3) the rate of polymerisation arediscussed. It appears that the nr form a time-dependent Poisson distribution with respect to the Y,and that the parameter of the distribution at any instant is the average number of radicals per locusat that instant.1. INTRODUCTIONIt is generally recognised that the free-radical emulsion polymerisation of amonomer which is sparingly soluble in water occurs in two main ~ t a g e s , ~ ' ~ namely :(i) a stage during which particles are nucleated; and (ii) a stage during which theexisting particles grow at the expense of monomer, but no new particles are generated.It is convenient further fo subdivide the post-nucleation stage into two sub-stages :(a) a stage in which unreacted monomer is present as a separate phase as well as inthe reaction loci ; and (b) a stage in which unreacted monomer is present only in thereaction loci.This paper is concerned with the kinetics of free-radical polymerisation during thefirst post-nucleation stage, and, more specifically, with a situation where a fixednumber of reaction loci has been introduced into the reaction system by means of aseed latex.The model which is assumed is described in detail below. Our purpose199D. T. BIRTWISTLE AND D . C. BLACKLEY 1999is to present, and discuss the implications of, an alternative solution to the problemwhich Gilbert and Napper have recently attempted to s01ve.~ Since deriving oursolution, we have become aware that the same solution was obtained independentlyby Weiss and D i s h ~ n .~ The principal justification for publishing a further paperupon this solution is that we wish to draw attention to certain interesting implicationsof the solution. In addition, we also give a clear statement of the type of physicalreaction system to which the solution can be expected to apply; this is omitted byWeiss and Dishon.It will appear subsequently that, under certain conditions at least, the solutionobtained by Weiss and Dishon and by ourselves gives results which are numericallyindistinguishable from those given by the solution of Gilbert and Napper. Theprincipal disadvantage of the Gilbert-Napper solution is not its lack of precision butits lack of convenience. Unlike the new solution, it does not appear to be capableof being expressed in closed analytical form, and this inevitably hinders discussionof the implications of the solution.2.MODEL ASSUMEDThe reaction model assumed is one in which free-radical polymerisation iscompartmentalised within a fixed number of reaction loci, all of similar volume, whichare dispersed in an external phase. More specifically, the reaction loci are envisagedas being small particles of polymer which have been formed by, say, a previousemulsion polymerisation. No nucleation of new reaction loci occurs as polymerisa-tion proceeds, and the number of loci is not reduced by processes such as particleagglomeration. Monomer droplets are present as a separate phase, so that theconcentration of monomer within the reaction loci can be taken as constant throughoutthe reaction.Initially there are no free radicals present in the system, so that nopolymerisation is occurring. Then, at zero time, radicals begin to be generated inthe contiguous external phase at a constant rate. Radicals immediately begin toenter reaction loci at a constant rate, and thus the rate of acquisition of radicals by asingle locus is kinetically of zero order with respect to the concentration of radicalswithin the locus. Once a radical enters a reaction Iocus, it initiates a chain polymerisa-tion reaction which continues until the activity of the radical is lost.The modelassumed, therefore, corresponds to a seeded emulsion polymerisation in the presence ofa sufficient excess of monomer to form a separate droplet phase, and under conditionssuch that no nucleation of new particles occurs. It will appear subsequently thatthe analysis is further restricted to reactions in which the volume of the particlesincreases only slightly as the reaction proceeds; this condition will be realised in areaction for which the size of initial seed is large and the extent of polymerisation issmall.It is required to calculate as a function of time the numbers of particles whichcontain 0, 1, 2, . . . propagating radicals, and thereby to examine the nature of thetheoretical approach to the steady state (assuming, of course, that a steady stateexists for the model assumed).In principle, radical activity can be lost from reaction loci by two types of reaction :(i) Those which are kinetically of the first order in the radical concentration withinthe reaction locus.Included in this class will be loss of activity by termination throughreaction with monomer, and loss of activity by diffusion out of the reaction locusinto the external phase. (ii) Those which are kinetically of the second order in theradical concentration within the reaction locus. The most obvious of these isbimolecular mutual termination between propagating radicals2000 COMPARTMENTALISED POLYMER REACTIONSLike Gilbert and Napper and also Weiss and Dishon we make the simplifyingassumption that radical-loss reactions of the second type are of negligible occurrencerelative to those of the first type.The model as assumed in this paper, therefore,applies only to those seeded emulsion polymerisation reactions in which radical lossfrom loci occurs mainly by, say, reaction with monomer, or by diffusion back intothe external phase. It is further assumed that radicals which are lost from reactionloci by diffusion into the external phase are not able subsequently to re-enter thereaction loci and reinitiate polymerisation. They are not, therefore, regarded asbeing, as it were, added to the " bank " of radicals in the external phase available forentry into reaction loci.3. DIFFERENTIAL DIFFERENCE EQUATIONS TO BE SOLVEDLet ni be the number of reaction loci per unit volume of reaction system whichcontain i radicals.Until the steady state is attained, nl is a function of time, t ;this can be emphasised by writing as ni(t). Let N be the total number of reaction lociper unit volume of reaction system. Thus N = ni and does not vary with time,although the individual nl may. p denotes the overall rate of entry of radicals intoall the loci contained in unit volume, so that p/Nis the average rate of entry of radicalsinto each single locus. This latter quantity is conveniently denoted by 0.The equations to be solved in the most general case are obtained by modifyingthe recurrence relationship of Smith and Ewart ' in such a way as to allow for thepossibility that the populations for the different classes of reaction locus may not bestationary. The result for the rate of change of ni with time can be put in the form :00i = Odni kt - = (ni- - ni)a+ ((i + l ) n i + - inJk + ((i +2)(i + l)ni+ 2 - i(i- 1)ni)-dt 0(1)for i = 1,2,3, .. ., and for the special case i = 0dno - = -nn,a+nlk+2n2-. ktdt UIn these equations, v is the volume of the reaction locus, k, is the rate coefficient forthe mutual termination of radicals (in suitable units), and k is a composite constantwhich quantifies the rate at which radicals are lost from reaction loci by first-orderprocesses. If radicals are lost only by diffusion into the external phase, then k canbe expressed in the form koa/v, where a is the surface area of the reaction locus andko is a constant which expresses the tendency of radicals to diffuse across unit areaof the surface of the reaction locus when radicals are present in the locus at unitconcentration.It is not, however, necessary to enquire too closely into the exactphysical meaning of the constant k in order to solve the problem, although closerenquiry may, of course, be necessary when it is desired to apply the solution toparticular reaction systems. Insofar as the values of ko, a and v do not altersignificantly as the reaction proceeds, the quantity k will remain unaltered. Notethat the requirement that neither a nor v alters (strictly that a/v does not alter) impliesthat the initial size of the seed is large, and that relatively little growth occurs as aconsequence of polymerisation.The assumption that k,/v is constant also implieslarge seed and small growth, as well as invariance of k, as reaction proceeds;however, these latter assumptions are of no relevance for the present analysis, becausethe term in which k , occurs is subsequently to be set equal to zeroD. T. BIRTWISTLE AND D. C . BLACKLEY 200 1The boundary conditions for the problem are as follows :(i) no(0) = N (3)(ii) n,(O) = nz(0) = . . . = 0. (4)Furthermore, for all t2 ni(t) = n,(O) = N .i = O4. SOLUTION FOR CASE WHERE ki = 0Like Stockmayer and subsequently Weiss and Dishon,’ we seek a generatingfunctionwhich is such that, when expanded as a power series in the auxiliary variable {, thecoefficient of t i is ni. Unlike Stockmayer, our generating function is time-dependent,since we are concerned with the non-steady state, whereas Stockmayer’s concernwas with the special case of the steady state.The following properties of Y(5, t) are immediately obvious from its definition :(i) Y(1, t) = N (7)for all t, since, by hypothesis, N is not time-dependent.Thus the time-dependencyof Y(t, t) should disappear when the substitution 5 = 1 is made.for r = 0, 1, 2, . . .03(iii) (g) = C ini(t>.< = 1 i = l(9)Thus (aY!/at),=l gives the total number of radicals in all the reaction loci containedin unit volume of the reaction system at time t. The average number of radicalsper locus at time t is therefore given byTo be acceptable, the generating function must be such that (arY/a<‘)t=1 isfinite for all r.This follows because physically these quantities represent the sumsof products of numbers of loci and numbers of radicals.To convert the set of differential difference equation (1) and (2) into a differentialequation in Y, each equation for dni/dt is multiplied by ti, and then all the equationsso obtained are summed. The result isWk 2 iniSi+X 2 (i+2)(i+l)r1,+~t~-~ i(i-l)nici (11)i = O i = O i = Owhere x = kJu. Eqn (2) (for dno/dt) is included in eqn (1 1) if it is understood thann-, = 0. Noting that Znici = Y, Cni-15 = tY, Z(i+l)ni+l(i = aY/at, Xinisi 2002 COMPARTMENTALISED POLYMER REACTIONS<aY/a(, X(i+2)(i+ l)ni+2(i = i32Y/d(2 and X i ( i - l)niti = (2i?2Y/a<2, where in eachcase the summations cover all possible values of i, eqn (11) transforms to[It may be noted that Stockmayer's differential equation for his generating functionfollows immediately as a special case of eqn (12) by putting aY/at = 0 (corresponding,of course, to the steady state).The result iswhich is identical with eqn (4) of Stockmayer's paper, apart from differences innot at ion.]The solution of eqn (12) for the case where x = 0 is (see Appendix)Y((, t ) = N exp $<-l)(l-e-kt) . (14) (" I Evaluation of aY/a( and aY/at from eqn (14), and substitution in eqn (12) for thecase where x = 0, confirms that the result obtained for Y((, t ) is a solution of thegiven partial differential equation for Y(<, t ) . It will be noted that this result forY?(t, t ) meets the requirement that the time-dependence should disappear when thesubstitution ( = 1 is made.Indeed, it is clear that Y(1, t ) = N, which is consistentwith the definition of Y(<, t ) .From this result for Y(<, t ) , it inmediately follows thatPutting t = co gives the respective populations of the various types of loci when thesteady state has been achieved. The result is5. DISCUSSION OF SOLUTION FOR CASE WHERE kt= 0(i) COMPARISON WITH SOLUTION OF GILBERT AND NAPPEROur solution for iZr(G0) is identical with that of Gilbert and Napper, if allowanceis made for the differences in notation. However, our results for the variation of theloci populations in the approach to the steady state appear to differ slightly from thoseof Gilbert and Napper.They quote their results as rather cumbersome power seriesin ckt. corresponds to thea2 of Gilbert and Napper, and we seek expansions in ascending powers of 5 for theexpressions for n,(t)/N.For convenience, write e-kt = 5, and a/k = p ; thenFrom the set of equations (15), we have for no(t)This compares withP2C2 p3c3 = (q 1+~[+-+-+. . .) N 2! 3D . T. BIRTWISTLE AND D. C . BLACKLBY 2003for Gilbert and Napper’s expansion expressed in the present notation. The twodiffer in that our series has the factor e-fi in place of (1 -8/2)2 in Gilbert and Napper’sseries. These two factors are almost the same if B < 1.For nl(t), our result giveswhilst for n2(t) we obtainThese expansions are again identical with those of Gilbert and Napper, except thatours contain the factor e-’? in place of their (1 -p/2)2.(ii) PREDICTED VARIATION OF nr(t) WITH tOur result for no(t) predicts that it will fall monotonically with increasing t , fromits initial value of no(0) = N to its steady-state value of no(Go) = Ne-‘’k.This canbe shown by considering either the structure of the expression for no(t), or the signof the predicted dno(t)/dt. A uniform decrease in nO(t) with increasing time is whatwould be expected from intuitive physical considerations.The prediction does not appear to be so simple in the case of nr(t) when r isother than zero. Examination of the general expression for dn,(t)/dt shows that(c1) initially it is zero, (b) in the early stages of the reaction it is positive, (c) in the laterstages of the reaction it may become negative, and ( d ) eventually it becomes zero.The precise nature of the variation of dnr(t)/dt depends upon r and upon the valuesof the constants 0 and k.Clearly, if drtr(t)/dt is positive in the early stages of thereaction, and later becomes negative, then nr(t) must pass through a maximum as thereaction proceeds from the initial stage to the steady state. Examination of thegeneral expression for dnr(t)/dt shows that a maximum in rtr(t) as a function of texists if r < a/k.(iii) PREDICTION FOR TOTAL RADICAL POPULATION AND AVERAGE NUMBER OF= 1 gives the following result for the total radical popula-RADICALS PER REACTION LOCUSEvaluating dY/d( attion contained in N reaction loci at time t :Q Win,(t) = N-(l-e-kf).ki = 1The steady-state value is N(a/k). The prediction is that the total radical populationwill increase monotonically with increasing time, approaching its steady-state valueat a rate which decreases uniformly and exponentially with time. It thus appearsthat, whereas some of the n, may not increase uniformly with increasing time, thetotal population of radicals does increase uniformly2004 COMPARTMENTALISED POLYMER REACTIONSThe prediction for the average number of radicals per reaction locus is obviouslythe steady-state value being a/k. This result for the average number of radicalsper locus as a function of time of reaction enables the expressions for the locus popula-tions to be put in forms which give the populations as functions of i ( t ) only.Substituting from eqn (21) into the set (15) gives the interesting set of relationshipsThus it appears that the aggregate of the proportions of reaction loci which containvarious numbers of radicals forms a time-dependent Poisson distribution with respectto the number of contained radicals.The parameter of the distribution at anyinstant is the average number of radicals per locus at that instant.*(iv) PREDICTION FOR VARIATION OF RATE OF POLYMERISATION WITH TIME OFREACTIONThe rate of conversion of monomer molecules to polymer is given bydM a3- = kp[M] C inidt i = l(23)if the various quantities are expressed in appropriate units,8 where M denotes thenumber of monomer molecules reacted, [MI denotes the molar concentration ofmonomer at the reaction locus, and kp denotes the rate coefficient for the propagationreaction.Substituting for Z i q from eqn (20) and integrating (assuming that [MI isconstant) gives the following result for the number of molecules of monomer whichare polymerised in N loci in time t :dM(t) = k,[M]N (kt +e-kt- 1).kEqn (24) gives the predicted shape for the conversion-time curve for a seededemulsion polymerisation reaction which conforms to the model assumed for thepresent analysis. Note that at short times this equation predicts thatM(t) 2 k,[M]Ng.*t2. (25)Thus, like Gilbert and NapperY4 we predict that the extent of conversion will varyquadratically with time over the early stages of the reaction. It is of considerableinterest to note that Gilbert and Napper are able to cite results for the seededemulsion polymerisation of vinyl acetate at 40°C which show that M(t) is indeeddirectly proportional to t 2 up to - 50 % conversion [they actually show JM(t)as directly proportional to t ] .It is also of interest to note that eqn (25) predicts that,in the early stages, the extent of conversion does not depend upon k (which quantifiesthe rate at which radicals are lost from reaction loci by first-order processes), butonly upon Q (which is the average rate of entry of radicals into a single reaction locus).* Note added in proof: These conclusions, and several of the others which have been noted above,also follow immediately from the observation that, according to eqn (14), Y(E,, t ) / N is the frequency-generating function for a Poisson distribution whose parameter (and therefore mean) is (o/k)(l- e-k')D.T . BIRTWISTLE AND D. C . BLACKLEY 2005Eqn (25) provides a possible basis for obtaining an estimate of ON from measure-ments of M(t) as a function oft over the early stages of the reaction. M(t) plottedagainst t2 should give a straight line through the origin. Insofar as k, and [MI areknown, the slope of this line can be used to obtain aN. lnsofar as N is also known,it is then possible to obtain an estimate of cr itself. The value obtained can then becompared with the rate of production of radicals per reaction locus as inferred fromthe known or assumed decomposition kinetics of the initiator.1.0301.020 30.980 k6 I I I I0 0.5 1.0 1.5 2.0tFIG.1 .-Fractional locus-population generating function, Y([, t ) / N , as a function of the auxiliaryvariable, 6, for various values of time, t, taking 0 = 1 x sdl. The figuresappended to the curves give the values of t in seconds.s-' and k = 5 xWe have applied this procedure to the results for vinyl acetate cited by Gilbertand Napper in their fig. 2. Taking their values of kp = 1860 dm3 mol-1 s-l, and[MI = 6.0 mol dnr3, we estimate ON = 4.2 x loll dm-3 s-l. This compares wellwith Gilbert and Napper's own estimate of 4.6 x 1011 dm-3 s-l . Taking their valueof 1.2 x 1015 dm-3 s-l for the rate of production of primary free radicals by decom-position of the initiator, our calculations confirm the apparent remarkably lowefficiency of radical capture by the loci in the vinyl acetate system.In principle, it should also be possible to obtain an estimate of k from measure-ments of rate of polymerisation as a function of time over the early stages of a seededpolymerisation.Substituting foi Cini from eqn (20) into eqn (23) gives dM/dt as 2006 COMPARTMENTALISED POLYMER REACTIONSfunction of t.and rearranging, givesExpanding the exponential in the resultant expression as far as t2,1 dM kt 1- - N -k,[M]Not dt - 2over the early stages of the reaction. If ON is known, then the left-hand side ofeqn (26) can be calculated for various times of reaction. It should be directlyproportioinal to t, the slope of the relationship permitting k to be estimated.(V) NUMERICAL PREDICTIONSIn order to illustrate the numerical consequences of the theory given, we have takens-latthe same values of CT and k as did Gilbert and N a ~ p e r , ~ namely, CT = 1 xand k = 5 x s-l.The variation of the generating function "(5, t ) witht x 10-31sFIG. 2.-Fractional locus populations, nr(t)/N, as a function of time, t, for r = 0, 1 and 2, takingu = 1 x s-l. The ordinates for r = 0 are no(t)/N; those for r = 1 are40nl(t)/N; those for r = 2 are 2000n2(t)/N.s-' and k = 5 xvarious times of reaction is shown in fig. 1. What has been plotted is the quantityY(5, t ) / N , since the dependence of Y on N is of incidental significance only. Itappears that the variation of Y(<, t ) / N with 5 over the range shown (0 < 5 < 2) isalmost linear.The absence of appreciable curvature in the region { = 0 implies,of course, that, for the chosen values of o and k, most of the loci contain either noneor one radical; the proportion which contains two or more radicals is a negligiblefraction of the whole. A second feature of interest shown by the curves for Y(c, t ) / Nas a function of 5 at various values of t is the tendency to coalesce into a single curveas t gets large ; this is, of course, a consequence of the steady state being approached,with "(5, t ) becoming independent oft. Also of interest is the nature of the approachof W(<, t ) as a function of < to the steady-state Y(5, 00) as a function of 5, as tincreases. The intercept at 5 = 0 decreases uniformly as t increases, but the slopeat 4: = 0 increases uniformly with t in such a way that all the curves pass through thepoint (1,l).This, of course, implies that the proportion of loci containing no radicalD. T. BIRTWISTLE AND D. C. BLACKLEY 2007NU0Xh 2t x 10-31sFIG. 3.-Average number of radicals per locus, Z(t), as a function of time, t, taking u = 1 x s-'and k = 5 x lod4 s-l.decreases uniformly as time increases, whereas the number of loci containing oneradical increases uniformly as time increases.Fig. 2 shows n,(t)/N as a function of t for Y = 0, 1 and 2. In order to facilitatecomparison with the predictions of Gilbert and Napper's theory, these curves havebeen plotted in a manner analogous to the curves of fig. 1 of their paper.The closesimilarity between our curves and those of Gilbert and Napper is immediately evident.This is a consequence of the fact that, for the chosen values o f 0 and k, the value ofe-=lk is indistinguishable from that of (1 - 0 / 2 k ) ~ . The curves of fig. 2 show that,for the chosen values of CT and k, (i) the steady state is attained after a reaction timet x 10-31sFIG. 4.-Conversion of monomer to polymer in arbitrary units, M(r), as a function of time, t, takingu = 1 x S-' and k = 5 x s-'2008 COMPARTMENTALISED POLYMER REACTIONSof approximately lo4 s, (ii) at all times, the number of loci which contain two or moreradicals is a negligible fraction of the whole, and (iii) nl(t) and n z ( t ) increase unjformlyas t increases.Regarding (iii), this is to be expected in view of the particular valuesof 0 and k which have been chosen for the purposes of illustration. For these values,there can be no possibility of a maximum occurring in any of the n,(t), since alk forthis case is 2 x ; there is therefore no non-zero integer value of r such thatFig. 3 shows i ( t ) as a function oft. Again, the manner of approach to the steadystate is clearly evident. It should be noted that, for the chosen values of a and k,even in the steady state, the average number of radicals per locus falls far short ofthe Smith-Ewart “ Case 2 ” value of 0.5,The curve for rate of polymerisation as a function o f t will have the same form asfig. 3 if [MI and kp remain constant as reaction proceeds.The prediction forconversion as a function of time is shown in fig. 4. For the chosen values of Q andk, marked deviation from linearity occurs only in the range t < 5 x lo3 s. Attentionis drawn to the similarity in shape between the curve of fig. 4 and the conversion-time relationship shown by Gilbert and Napper for the seeded emulsion polymerisa-tion of vinyl acetate at 40°C.r < ojk.APPENDIXSOLUTION OF EQN (12) FOR CASE WHERE X = 0To solve eqn (12) for the case where x = 0, we use the method of separation ofvariables. The function Y(g, t ) is assumed to be of the form E(5). T(t) where S(t)is a function of 5 only, and T(t) is a function of t only. Then if 93t denotes the . .composite differential operat orthe eqn (12) with x = 0 becomes1 aT 1T d t o -- = ,9@.Since the left-hand side is a function of t only, and the right-hand side a function of 5only, it follows that eqn (27) can only be true if both sides are in fact independent oft and 5, i.e., both are equal to some constant A.Thus the problem becomes one ofsolving the two ordinary differential equations1 dTT dt= A --and(29)(30)where A is a constant whose value may depend upon that of A. The solution ofeqn (29) isc = ~ ( 1 - <)-W eatlk (31)where B is a second constant whose value may also depend upon that of A. Thus aparticular solution of eqn (12), for a particular value of A, is the function(32) I+@, t, A) = c(a)(i-t)-a’k eAt+ur/D. T . BIRTWISTLE AND D. C . BLACKLEY 2009where C is a composite constant whose possible dependence upon A is emphasisedby writing it as C(L). Since eqn (12) is a linear partial differential equation, it followsthat the complete solution is the sum of all the particular solutions, i.e., the completesolution of eqn (12) is the sum of all the functions $(t, t, A) for allphysically-acceptablevalues of A.Thus~ ( t , t ) = C $(t, t , a) = C c(~)(I - t)-’lJk eAt+6Slk. (33)12 aWe now argue that the only physically-acceptable values for A are zero and thosewhich make -L/k a positive integer. (The value zero will, of course, define thesteady state). Our reasoning is that it is only these values of 3, which ensure that Yand all its partial derivatives with respect to t remain finite when the substitution < = 1 is made. That this is so becomes clear if the successive partial differentialcoefficients with respect to 4: of the function on the extreme right-hand side of eqn (33)are written down.Put - A/k = j so that the acceptable values of j are 0, 1 , 2, . . ., and also A = - kj.Thento ~ ( 5 , t) = C C(j)(l- 5)’ e-kjt+oglk , (34)j = Owhere the constant is written asThe boundary conditions for tf j = OC ( j ) , since it can clearly be regarded as a function ofj.= 0 give Y(5, 0) = N, and thusC( j)( 1 - c ) j earlk = N . (35)This equation can be re-arranged to giveEquating coefficients of (1 - t)j givesSubstituting this result for C(j) in eqn (34) and performing the summation over allintegral values for j from 0 to co gives for Y(t, t ) the result embodied in eqn (14).It may be noted that the result for the steady-state populations can also be obtaineddirectly by solving the equations1 dT 1T dt a-- = ;g@) = Rfor the special case A = 0.J. L. Gardon, J. Polymer Sci. A-1, 1968, 6, 623.J. L. Gardon, Brit. Polymer J., 1970, 2, 1.D. C. Blackley, Emulsion Polymerisation (Applied Science, London, 1975), chap. 3-5.R. G. Gilbert and D. H. Napper, J.C.S. Faraday I, 1974, 70, 391.G. H. Weiss and M. Dishon, J.C.S. Faraday I, 1976,72, 1342.W. H. Stockmayer, J. Polymer Sci., 1957, 24, 314.D. C. Blackley, Emulsion Polymerisation (Applied Science, London, 1975), p. 95.‘ W. V. Smith and R. H. Ewart, J. Chem. Phys., 1948,16, 592.(PAPER 7/585
ISSN:0300-9599
DOI:10.1039/F19777301998
出版商:RSC
年代:1977
数据来源: RSC
|
228. |
Photolysis of 3-methyl-3-chlorodiazirine |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 73,
Issue 1,
1977,
Page 2010-2024
Henry M. Frey,
Preview
|
PDF (1092KB)
|
|
摘要:
Photolysis of 3-Methyl-3-chlorodiazirineBY HENRY M. FREY* AND DAVID E. PENNYDepartment of Chemistry, University of Reading,Whiteknights, Reading RG6 2ADReceived 27th April, 1977The photolysis of 3-methyl-3-chlorodiazirine has been investigated in the gas phase at pressures(with added bath gases) up to 6000 Torr (800 kPa). The principal products were vinyl chloride,nitrogen, acetylene, hydrogen chloride and 1,l-dichloroethane. The experimental results indicatethat the vinyl chloride is formed with a very wide energy spread; at low pressures most of itdecomposes to yield acetylene and HCl. Evidence is presented that the dichloroethane resultsfrom the reaction of methylchlorodiazomethane, formed by photoisomerization from the diazirine,with the HC1 produced in the system.There is no evidence that trapping by HCl of methylchloro-carbene, formed as the primary decomposition product of the electronically excited methylchloro-diazirine, can be more than a minor reaction pathway.The pyrolysis and photolysis of 3-methyl-3-chlorodiazirine have been reportedpreviously. However, the photolytic studies have not been detailed or extensiveand no quantum yield data have been reported.The pyrolytic products are exclusively vinyl chloride and nitrogen and, as in thecase of the thermal decomposition of many other diazirines, can be rationalised interms of a mechanism involving either the intermediate formation of a carbene,followed by rapid intramolecular hydrogen transfer or a concerted process. Thephotolytic mechanism may be similar but, because the energy of the light quanta areconsiderably greater than the energy of activation for the diazirine decomposition,the resulting vinyl chloride contains sufficient internal energy to allow further decom-position to acetylene and hydrogen chloride.With a monochromatic light sourcethere is the possibility of the production of essentially monoenergetic vinyl chloridewhich would then decompose with a unique rate constant. Such a situation wouldbe experimentally simple to confirm and study. Further, application of the Rice,Ramsperger, Kassel, Marcus (RRKM) theory would allow the energy of activationfor the hydrogen chloride elimination reaction to be evaluated; it should also bepossible to measure the collisional efficiency of intermolecular energy transfer fromthe vinyl chloride.If, however, the process following the absorption of light resultsin the decomposition of the excited diazirine molecule to yield a carbene and nitrogenwith a range of energies, then this should be experimentally observable. This wouldlead to the possibility of deconvoluting the experimental data to gain informationabout the distribution of the " excess " energy, due to the overall exothermicity ofthe process, between the two fragments. Realistic information of this kind is almostnon-existent for complex molecules.EXPERIMENTAL3-methyl-3-chlorodiazirine was prepared by the oxidation of acetamidine hydrochloridewith sodium hypochlorite in dimethyl sulphoxide s ~ l u t i o n .~ Impurities were chloro-methanes, chiefly methyl chloride, which was removed by pumping at - 78°C. The purified201H. M. FREY AND D. E. PENNY 201 1material m7as stored under vacuum in a trap maintained at -778°C. One explosion occurredwhen using the purified material during a trap-to-trap distillation (possibly being carried outtoo rapidly). Since diazirine explosions appear to be in the nature of detonations they canbe quite destructive at short range. Great caution should always be exercised wheneverquantities in excess of a few milligrams are handled. In the present work the storage andtransfer vessels, which could contain up to -0.5 g of the methylchlorodiazirine were housedinside a Perspex box.Photolyses were carried out in cylindrical glass cells using 325 nm radiation (Cadmiumion laser) 337 nm radiation (nitrogen laser) and radiation around 350 nm (mercury superpressure arc with high intensity monochromator). The cadmium ion laser (Spectra Physics)had a beam of 1.5 mm diameter and a power of - 15 mW (measured using a HewlettPackard model 8330 AJ8334A radiant flux meter system).The high intensity mono-chromator (Bausch and Lomb) employed fixed slits to give a band path (at half width) of7.5 nm. Since 3-methyl-3-chlorodiazirine has its most intense peak at 354 nm, the mercuryarc, monochromator combination, was set for radiation in this region.Pressures between 0.1 and 6000 Torr were used employing cells of approximate volumes2, 30 and 90 cm3 depending upon the pressure.At the low end of the pressure range pure3-methyl-3-chlorodiazirine was photolysed ; in the higher region nitrogen and sulphurhexafluoride were used as inert additives to increase the total pressure. A glass spiral gaugewas used to measure pressures up to 760 Torr. At higher pressures, mixtures of the diazirineand sulphur hexafluoride were frozen into the photolysis cell and the pressures calculatedfrom the various volumes of cells and mixture vessels.The experimental procedure consisted of photolysing a known pressure of the diazirinefor a given time, followed by gas chromatographic analysis using a Perkin Elmer F11instrument equipped with a gas sampling valve and a Flame Ionization detector. Signalsfrom the detector were integrated electronically using either a Perkin Elmer D2 or a HewlettPackard 3380 A instrument.Separation of reactant and photolytic products was achievedusing a 5.5 mx 2.2 mm column packed with OPN on Poracil C at 55°C.Quantum yield determinations for all diazirine decompositions present difficultiesbecause of their very sharply banded absorption spectra. This makes a highly mono-chromatic light source essential. In the present work this requirement was easily met byusing the cadmium ion laser. However, at 325 nm we were unable to find a suitable gasphase system to use as a chemical actinometer and we elected to make essentially absolutemeasurement by determining the energy entering and leaving a micro-photolysis cell usingthe Hewlett Packard 8330A flux meter.RESULTSOverall quantum yields for decomposition of the 3-methyl-3-chlorodiazirine weredetermined at room temperature and with the 325 nm radiation.Over a considerablerange of pressures the quantum yield for decomposition did not vary with pressure.In one experiment employing a photolysis cell of 2.38 cm3 containing 59.6 Torr ofthe diazirine, the absorption of 2.61 x lo1' photons led to the decomposition of2.51 x lo1* molecules of the diazirine (yielding a quantum yield for decompositionof 0.96). From experiments of this type we obtain #dec = 0.945k0.02.PH 0 T 0 LY S I S OF 3-ME TH Y L - 3 - C H L ORODI A 2 I RI N EThe major products of photolysis detected by gas chromatography were acetylene,vinyl chloride and 1,l-dichloroethane. A trace of ethylene (<1 % of the totalproducts) was detected.The product ratios were found to be independent ofpercentage decomposition of the diazirine at any particular pressure ; consequentlyphotolyses were carried out for times sufficient to produce adequate samples foranalysis .1-62012 PHOTOLYSIS OF 3-METHYL-3-CHLORODIAZIRINE0.12 o c-I I 1 1 I I I 1pressure/TorrFIG. 1.-Product distribution for photolysis using 354nm radiation and SF6 as bath gas. (a)--/cl, (b) CH3CHC12, (c) C2H2.pressure/TorrFIG. 2.-Ratio of vinyl chloride to acetylene as a function of pressure : 325 nm radiation, SF6 andN2 as bath gases. (a) SF6, (6) pure compound, (c) N2FIG.H. M. FREY AND D. E. PENNY 2013pressure-' /Torr-'3.-Ratio of acetylene to vinyl chloride as function of reciprocal ofradiation, Nz as bath gas.the pressure : 325 nm0 100 200 300 LOO 500 600 700 BOOpressure/TorrFIG.4.-Variation of vinyl chloride to acetylene ratio as function of pressure for different photolyticradiation : SF6 bath gas. (a) 354, (b) 337, (c) 325 nm2014 PHOTOLYSIS OF 3-METHYL-3-CHLORODIAZIRINEThe addition of oxygen and ethylene in separate runs had no effect on the acetyleneto vinyl chloride ratio ; no new products were observed.The addition of ammonia or acetic acid (vapour) both markedly reduced theyield of 1, l-dichloroethane. The addition of ammonia (reaction mixture diazirine13 Torr, ammonia 17 Torr, SF6 470 Torr) completely suppressed the production ofI ,l-dichloroethane, whereas acetic acid at the same concentration halved the expectedyield.65% 3.CI x x21I I 1 I 1 I Ipressure/Torr0, 10 Torr diazirine + nitrogen.0 20 LO 60 80 100 120 Ill0FIG.5.-Yield of 1 ,I-dichloroethane as function of pressure. 0, Pure methylchlorodiazirine,Plots of the product distribution and product ratios as a function of pressure in thepresence and absence of inert diluents (bath gases) and at various wavelengthsare shown in fig. 1 to 5. The use of initial pressure as the abscissa is justified at highpressures of added bath gases, since in this region there is only a very minor pressurechange during photolysis ; it is probably also justified, certainly within the limits ofour experimental error at low pressures since the photolysis products, acetylene andvinyl chloride (and nitrogen) are likely to be less effective colliders than the methyl-chlorodiazirine itself.(The yields of products are calculated in all cases ignoringthe yields of HCl and N2).DISCUSSIONA simple scheme to account for the formation of the principal photolysis productsis as follows H . M, FREY A N D D . E . P E N N YIIC1 NC1 NN/\Cl N C1\ */* ClC1 N - >:Cl+ N2+ --+ \-N=N /-c1c1+ >:c1c1*-/ - --+ = + HCl2015(1)>: +HC1 + CH3CHC12* (7)c1+ - > =N=N + HCl ___+ CH3CHC12+N2 (8)CH3CHC12* --+ -/'l - + HCI. (10)c1CH3CHC12* + M __+ CH3CHC12 + M (9)According to this scheme the initial decomposition of the diazirine yields methyl-chlorocarbene and nitrogen. The quantum yield measurements make it likely thatvirtually all electronically excited diazirine molecules decompose, with few if anyreturning to the ground state.Reaction (4) is a characteristic reaction of all highercarbenes and yields in this instance vibrationally excited vinyl chloride which caneither decompose to acetylene and hydrogen chloride or be collisionally stabilised.If reactions (l), (2) and (4)-(6) (which are the major reaction pathways) werethe only ones occurring and if the excited vinyl chloride was essentially monoenergetic,then a plot of [vinylchloride]/[acetylene] against pressure should be linear. Further,if the strong collision assumption is made, the value of k6 [i.e., k(E) the specific rateconstant for molecules with energy El could be determined from the slope of the line.From fig.2-4 it is apparent that the appropriate plots are not linear. Some of thisnon-linearity is undoubtedly due to the neglect of other steps in the reactionmechanism, but the major cause of curvature in plots of this type is due to energydispersion, which results in a range of values for k(E).t Indicates electronic excitation ; * indicates vibrational excitation2016 P HOTOLY SIS OF 3 -METHYL- 3 -CH LOR0 D I A Z IRI NECurvature has been observed previously for the methyl chlorodiazirine system (2b)and also in other systems involving the production of excited intermediates byphotodissociation, e.g., in the photolysis of linear and cyclic azo compounds andcyclob~tanones.~ In some of these cases it has been possible to fit the curves usingeither a Statistical or Gaussian distribution function for the energy dispersion.Forthe former case the total excess energy content of the excited species, before fragmen-tation, is partitioned between the photofragments according to the ratio of theirstate densities, calculated from their vibrational frequencies. In the latter case amost probable energy is selected to fix the peak of the distribution and the standarddeviation varied until the experimental curve is matched. However, no one functionhas been found which fits the experimental curves throughout a large pressure range.A further cause of energy dispersion additional to any initial particular distributionresults from the failure of the strong collision assumption.If energy is removedfrom the vibrationally excited species in relatively small amounts (say a few kJ permol per collision) then an increased spread of energies will be produced. In systemswhere there is reason to believe that vibrationally excited species were formed initiallyin an essentially monoenergetic state, curvature of the appropriate plots has beenascribed to this cause. While in the present work both effects will operate they doso in an opposite sense and hence the curvature observed underestimates the initialenergy dispersion.It is apparent from an examination of fig. 4 that the energy content of the vinylchloride depends on the wavelength of the photolytic radiation. At any particularpressure above about 50 Torr more of the vinyl chloride is stabilised when 354 nmradiation rather than 325nm radiation is used for the photolysis.This showsunambiguously that some of the photon energy in excess of the minimum required fordecomposition must find its way into the carbene, and thence to the vinyl chloride.The failure of the strong collision hypothesis can be directly deduced by aninspection of fig. 2. Here it can be seen that SF6 is appreciably more efficient atstabilising the vinyl chloride molecules than is nitrogen, indeed it appears to be moreefficient than the methylchlorodiazirine. Thus even if SF6 is a " strong " colliderthe other molecules cannot be.Because of uncertainties about some parts of the reaction mechanism we do notbelieve an attempt to analyse the results in the necessary detail to obtain a completedescription of the initial energy distribution function or functions is justified at thisstage.(We hope that an extension of the experimental work under conditions whichwill eliminate the interference of minor reaction pathways will yield results that couldbe deconvoluted to give the required energy distribution). Some information can,however, be obtained relatively easily and it may be related to the energy spread ofthe vinyl chloride molecules.Provided attention is restricted to certain well defined pressure regions it isrelatively easy to calculate values for k6. At pressures around an atmosphere (say300-900 Torr), plots of the ratio of vinyl chloride to acetylene against pressure arereasonably linear [these are usually referred to as S/D plots, i.e., stabilised molecules(S) to decomposed molecules (D)].The data plotted in fig. 4 yield values (in thispressure region) for k6 of 5.6 x lo9, 9.5 x lo9 and 1.4 x lolo s-' for photolyticradiation of 354, 337 and 325 nm respectively. These values were obtained assumingthat SF6 acts as a strong collider with a collision number of 1 x lo7 TomM1 s-'.A similar calculation (see fig. 2) yields a value for k6 (using 325 nm radiation) of3.6 x 1O1O s-' when nitrogen is the bath gas. This value for k6 is more than twicethat calculated from experiments with SF6 as the bath gas. Obviously nitrogencannot be a strong collider. A simple, unrealistic model would be one in which oH . M. FKEY AND D .E. PENNY 2017average only two collisions in five with nitrogen were effective in stabilising the vinylchloride molecules. A possibly more realistic model would be that mentionedearlier, where every collision removes some energy but not sufficient to deactivate(in one collision) a highly excited molecule.The data obtained using SF6 and 354 nm radiation are more extensive than thoserelating to other bath gases or photolytic wavelengths. Thus from the resultspresented in fig. 1 we may calculate the value of k 6 at pressures from 3000-6000 Torr.In this region k6 is found to be 1 . 4 ~ 10" S-' (compared with 5 . 6 ~ lo9 s-' atN 700 Torr). Comparison with the results obtained with the other photolyticradiation at lower pressures suggests that using 325 nm radiation would yield vinylchloride molecules with values for k6 at these high pressures of about 3.5 x 10" s-I.In order t o obtain the corresponding values of k6 at low pressures it is convenientto plot D/S against the reciprocal of the pressure as shown in fig.3. From thelinear portion (corresponding to pressures below 1 Ton) a value for k6 of only3.8 x lo6 s-' is obtained. That the rate constant is greater at very high pressuresthan at intermediate pressures is consistent with a fairly wide spread of energies inthe vinyl chloride molecules. At high pressures, only those molecules with largeexcesses of energy and correspondingly high values for k(E) are able to decomposeand hence contribute to k6. At low pressures those molecules with much smallerenergy excesses and hence relatively small values of k(E) contribute to the overallrate constant and the value of k6 declines.On the basis of an RRKM calculation(see Appendix), using a value for Eo of 301 bJ mol-1 (72 kcal mol-') the higher andlower rate constants viz., 3.5 x 10" and 3.8 x lo6 s-l respectively correspond tovalues of excess energy in the vinyl chloride of 520-3 18 kJ mob-' (1 24-76 kcal mol-l).This should be compared with rather crude estimates based on thermochemicalarguments of a possible range of 542 to 383 kJ mol-' (130 to 92 kcal mol-I). Theseexcess energies take the average energy content of vinyl chloride molecules at " roomtemperature '' to be zero and should not be confused with energies in excess of theminimum required for reaction.Even though the thermochemical estimates are crude it is extremely improbablethat they can be appreciably in error at the minimum end of the range.Hence thediscrepancy between 376 and 318 kJ mol-' appears to be greater than the likelysum of errors in the estimation and calculation. The difference may be rationalisedin terms of two possible situations (or a combinalion of both of them). If the vinylchloride undergoes multistep deactivation then even molecules formed initially with376 kJ mol-1 (or more) may eventually undergo decomposition after having lostsufficient energy by collision to have energies barely above the threshold values.Alternatively, there may be another route to vinyl chloride other than directly fromthe carbene.One such route is given in the suggested reaction mechanism [reaction(lo)] and will be further discussed later in this paper.FORMATION OF THE DIAZOCOMPOUNDThe formation of diazomethane as a product in the photolysis of diazirine wasreported and ascribed to a photoisomerization process5 However, such a mode offormation leading to a stabilised product seems improbable since most estimates 6* 'of the heats of formation of diazomethane and diazirine make the latter less stable.Hence, after formation, the diazomethane molecules would decompose very rapidly.Our estimate of its maximum lifetime (see Appendix) is - s ; this implies that< 1 % would have been stabilised in the experiments of Amrich and Bell, yet a 20 %yield was reported. Recent work by Lahmani has resolved this difficulty.Sh2018 PHOTOLYSIS OF 3-METHYL-3-CHLORODIAZIRINEconcludes that diazomethane is not a photoisomerization product but is formed bythe reactionN/, /ICH2 + CH2 ) I + CH2N2 + CH2.\ iiNIt was for this reason that reaction (3) was suggested in our mechanism.It is important to note that the above discussion does not rule out a photo-isomerization pathway, but that such a route could not have led to the observeddiazomethane. However, these arguments do not necessarily apply to substituteddiazirines (since the lifetimes of the resulting diazocompounds would be much longer)and indeed there is experimental evidence for the photoizomerization pathway insuch systems in the liquid phase.g0.3I 1 1 1 1 1 1 1 1 1 1 1 1 110 20 30 40 50 60 70 80 90 100 I!O 120 130 I10pressure/TorrFIG. 6.-Ratio of vinyl chloride to acetylene at low pressures : 325 nrn radiation.We must, therefore, add the following steps to our reaction mechanism :N.F>=&=N* --+ CH,=CHC1+N2.CiWhile we show the formation of the vibrationally excited diazocompound from theelectronically excited methylchlorodiazirine, it could well be that an electronicallyexcited diazocompound is formed first.For reasons mentioned earlier the diazo-compound will decompose unless it is colIisionally stabilised.Those molecules of methylchlorodiazomethane which are initially stabilised caH. M. FREY AND D. E. FENNY 201 9react with the HCl produced in the system by the decomposition of vinyl chloride,to yield 1,l-dichloroethane.FORMATION OF ~,~-DICHLOROETHANEOur reaction mechanism shows two possible precursors for the dichloroethane,viz., the methylchlorocarbene and the methylchlorodiazomethane,10 both reactingwith the hydrogen chloride generated in the system.These possibilities are supportedby the effects produced by the addition of ammonia and acetic acid (vapour) toreaction mixtures before the start of the photolyses. The addition of excess ammonia,which removes HCl as it is produced, results in the complete suppression of theformation of the dichloroethane. The addition of acetic acid reduced the yield ofdichloroethane. Obviously in the absence of HCl no dichloroethane can be formedby the suggested mechanism ; the reduction in yield in the presence of acetic acid isthe result of competition between the acetic acid and the HCl for reaction with eitherthe carbene or the diazocompound. If the dichloroethane arise only from reactionof a ground state methylchlorodiazomethane with the HCl, then our experimentsallow us to conclude that the rate constant for reaction with HC1 is nearly 10 timesthat for reaction with acetic acid.If the dichloroethane resulted entirely from reaction of HCl with the carbene,then at low pressures most of it would decompose into vinyl chloride and HCl.Our thermochemical estimates together with RRKM calculations would lead to amaximum lifetime of -5 x IO-" s with a possible value as small as 2 x s.Even in the former case there would be little stabilisation of the dichloroethane atpressures below an atmosphere.However, only for the minimum lifetime valuewould there be virtually no stabilisation at even our highest experimental pressures.We are forced to conclude that the trapping of the carbene to form 1,l-dichloroethanecannot be more than a minor pathway under most of our experimental conditions.Even accepting that the major reaction pathway to the dichloroethane involvesthe diazocompound, there remain some uncertainties. If the reaction only involves" thermalised " methylchlorodiazomethane, the most probable situation, then theresulting dichloroethane is unlikely to undergo further decomposition. Under thesecircumstances the variation in yield of the dichloroethane with pressure must beattributed to reactions (12) and (13), i.e., to the effect of pressure on the yield of thediazocompound.At high pressures where all or most of the diazocompound willbe stabilised, the yield of dichloroethane cannot exceed (but can equal) the yield ofHCl, which is of course equal to the yield of acetylene. Inspection of fig. 1 showsthat at total pressures of over 4000 Torr the yield of 1,l-dichloroethane is apparentlyslightly greater than the acetylene yield (though beginning to decrease with it). Webelieve this is an experimental artefact; both yields are the same. It is extremelydifficult to obtain accurate comparative calibration factors for an early peak (acetylene)and a late peak on a chromatogram when the former occurs immediately after anenormously overloaded peak (SF,).It seems clear that at pressures above 4000 Torrall the HCl formed in the system is being removed by reaction to form dichloroethane.Since at -4000 Ton dichloroethane constitutes 14 % of the reaction products, thenat least this fraction of the reaction must go via the diazocompound (neglecting thepossible minor carbene pathway). This would imply that at, say, 500 Torr (dichloro-ethane yield 8.4 %) at least 40 % of the initially formed diazocompound decomposedvia reaction (13) before being stabilised. If one uses the yields of dichloroethane asa function of pressure in this way to obtain values for the rate constant for decom-position of the methylchlorodiazomethane then, as in the case of vinyl chloride, 2020 PHOTOLYSIS OF 3-METHYL-3-CHLORODIAZIRINElarge range of values is obtained.This may again indicate a large energy spreadproduced by the initial mode of formation possibly accentuated by a weak collisionaldeactivation process.LOW PRESSURE RESULTSWe have already noted that at low pressures the apparent values of k6 appearunreasonably small. We have also noted the possibility that in this region reactions(12) and (13) may both be important. A simplified treatment of the appropriateportion of the reaction scheme (now excluding all bimolecular reactions of the carbene)is as follows :C=N=N*./ClIf we assume that at low pressures the fate of any stabilised diazocompound is toyield 1,l-dichloroethane (by reaction with HCl) then it is easy to show thatHence an extrapolation of a plot of the ratio of vinyl chloride to acetylene againstpressure to zero pressure should yield the value of (1 -cc)/a as the intercept.Thisplot is shown for pure diazirine photolysis (325nm) in the pressure range 10 to140 Torr in fig. 6.The plot is surprisingly linear, which suggests that in this pressure rangek6 % k,[M] and k13 % k,,[M]. From the intercept (0.38) we calculate, on the basisof eqn (i) that the photoisomerization pathway represents over 27 % of the primaryprocess. Of course the value of a may depend on the wavelength of the light usedand is, therefore, not necessarily the same for 354nm. It does suggest, however,that the maximum yield of 1,l-dichloroethane observed in the high pressure photolysisis determined only by the HCl present. (Further experiments which could havesettled this point unambiguously could not be carried out before this study had tobe terminated).Thus overa pressure range from 10 to 6000 Torr the value of k6 changes by a factor of - 5.This would not appear unreasonable and completely interpretable in terms of aninitial wide energy distribution.Unfortunately an examination of the data atpressures below 10 Torr makes our simple interpretation suspect.From the slope of the plot we obtain a value for k6 of 8.1 x lo9 s-l.VERY LOW PRESSURE RESULTSResults obtained below 10 Torr (diazirine and diazirine + N, mixtures) using325 nm radiation show a further apparent drop in the calculated value of k6.A plot of the type shown in fig.6 is linear from 1.5 to 10 Torr. It yields anintercept (0.295) which would make the photoisomerization pathway only 23 % ofthe primary process and from the slope of the plot k6 = 9.1 x lo8 s-l, a factor 9 time2021smaller than the " low pressure " value. Mow 1.5 Torr the ratio of vinyl chlorideto acetylene shows a further drop and at -0.2 Torr the calculated value of k6 dropsto the very low value of -4x lo6 s-I mentioned earlier in this paper.We believe that at these very low pressures the acetylene-forming reaction can-not possibly be from the reaction sequence (4)-(6). The most obvious otherpossible source of acetylene is from the vinyl chloride produced via reaction (13).In our mechanism we have shown the decomposition of the vibrationally exciteddiazocompound as yielding " stabilised " vinyl chloride which would be consistentwith thermal studies.However, the thermochemistry of the reaction makes it obvious that if anappreciable fraction of the initial excitation energy in the diazirine remains in thediazocompound and is not removed before it decomposes, then the vinyl chloridecould be chemically activated.It does appear possible that this route accounts forthe results observed at pressures of a few Torr and less.In the low to very-low pressure region the situation may be further complicatedby the possibility that some 1 , 1-dichloroethane is formed by the reaction of vibration-ally excited methylchlorodiazomethane and HC1. Such energy rich dichloroethanecould be another source of vinyl chloride.H.M. FREY AND D. E. PENNYCONCLUSIONSOn the basis of thermochemical estimates and RRKM calculations we have shownthat most if not all the 1,l-dichloroethane must arise by reaction between thernethylchlorodiazomethane and HC1. If the diazo compound arose only by reaction(3), as in the analogous case of the formation of diazomethane from diazirine, thenits yield would be proportional to the methylchlorodiazirine pressure. Now inspec-tion of fig. 5 shows that the yield of 1,l-dichloroethane is only a little greater in thephotolysis of the pure diazirine compared with a mixture of methylchlorodiazirinewith nitrogen at the same total pressure. This small difference can be easily accountedfor in terms of the lower efficiency of nitrogen compared with the diazirine atstabilising the methylchlorodiazomethane.Even if this were not so, the resultsshow clearly that onZy a minor percentage of the methylchlorodiazomethane canarise via reaction (3).Our analysis of the system suggests that the reaction steps of major importanceare therefore (I), (2), (4)-(6), (8) and (11)-(13). While we cannot rule out steps(3) and (7) they can only be of minor importance.We have deliberately avoided a discussion of the possible relationship betweenthe fission and isomerization pathways of the excited methylchlorodiazirine moleculesand their spin states. It is hoped that photolyses which we intend to carry out inthe future in the presence of additives will help to answer this question as well asallow some of the uncertainties of our present interpretation to be removed.Itshould then be possible to attempt the deconvolution of the experimental data toyield accurate values for the spread of rate constants and hence the partitioning ofenergy after the primary light absorption process.We thank the S.R.C. for a grant which made this work possible. We are alsograteful to a referee for his most useful remarks.APPENDIXRRKM C AL C U L AT1 0 N SSpecific rate constants k(E) for the decomposition of monoenergetic vinyl chloride,The high pressure Arrhenius 1 ,I -dichloroethane and diazomethane have been calculated2022 PHOTOLYSIS OF 3-METHYL-3-CHLORODIAZIRINEparameters were used to obtain values of the critical energy and make vibrational assignmentsof the activated complexes.Densities of states were determined using the Whitten-Rabinovi tch approximat ion.VINYL CHLORIDEThe high pressure Arrhenius parameters given by Cadman and Engelbrecht were used '[log (A/s-') = 13.83 and E, = 302 kJ mol-I]. The vibrational frequencies (in cm-l) ofthe molecule l2 and activated complex are given below. The bracketed numbers are thedegeneracies used, when different from unity.molecule coinplex3121 1030 2200 8003086 941 3086 8963030 896 3030 7201608 720 1758 12401369 620 1000 1701279 395 1279Activated complex frequencies were assigned with the help of a comparison with thosesuggested for the decomposition of activated vinyl fluoride.' The carbon chlorine stretch(720cm-l) was taken as the reaction co-ordinate.Values of k(E) as a function of E+,(E*-Eo) are shown. Half pressures (S/D = 1) are calculated using 2 = lo7 Torr-I s-'.E+/kJ mol-1 k(E)is-1 P3/Torr42 5.8 x 107 684 7.3 x lo8 73126 3 . 9 ~ 109 390167 1 . 2 ~ lozo 1240209 3.1 x 10'' 310025 1 6.4 x lo1* 64001,l -D I c H LOR OETH AN EThe Arrhenius parameters of Hartmann, Xeydtmann and Rinck l4 were used (log(A/s-l) = 13.5; Ea = 224 kJ mol-I). The frequencies used both for the molecule andcomplex are essentially those of Hassler and Setser.molecule complex2953 (4) 2933 (3) 857 (4)1362(5) 1434 63 51044(3) 1330 500677 (2) 1200 (3) 300350 (2) 934 193257 (2)E+/kJ mol-1 k(E)/s-1 P+/Torr42 1 x 107 1126 2~ 109 200209 2 .7 ~ 10'' 2 700293 1 . 4 ~ 1011 14 000377 4 . 2 ~ 10'' 42 000DIAZOMETHANETwo sets of high pressure Arrhenius parameters have been reported for the thermaldecomposition of diazomethane 16* l7 viz., log(A/s-l) = 13.0, Ea = 146 kJ mol-' anH . M. FREY AND D. E. PENNY 2023log(A/s-') = 12.95 and E, = 134 kJ mol-l. The vibrational assignment for the moleculeis that given by Moore and Pimentel.18 The carbon nitrogen asymmetric stretch has beentaken as the reaction coordinate.molecule complex3077 1109 3077 4663184 564 3184 4212102 421 2102 4061414 406 14141170 1 I09P /Tom + E+/kJ mol-1 k( E) Is- 142 2.2x lo1* 2 20084 1.2x 12 OOO126 3.4x 10'l 34 000167 6 . 9 ~ 10" 69 00025 1 1 . 7 ~ loi2 1 . 7 ~ lo5209 1 . 2 ~ i o l z 1 .2 ~ 105Note the pre-exponential factor appears improbably low for a dissociation reaction andthere may be a compensation effect in the Arrhenius parameters. A higher " A " factorwill result in an increase in the calculated values of k(E).THERMOCHEMICAL ESTIMATESThe heats of formation of the diazirines are in some doubt. If we take the value ofLaufer and Okabe for diazirine itself of 255 kJ mol-1 we may make appropriate adjust-ments for the methyl group and the chlorine atom (on the basis of group additivity) andobtain a value for the methylchlorodiazirine of about 209 kJ mol-l. A similar value(234 kJ mol-l) has recently been obtained by TyIer. l9 In an analogous manner we estimatethe heat of formation of methylchlorocarbene as 377 kJmol-l.Other values of AHf thatare used have been tabulated.'' (It should be noted that with Tyler's, dichloroethaneformed from the diazo compound may have more than the critical decomposition energy).EXCITED VINYL CHLORIDEMaximum and minimum energy contents can be estimated by assuming either that all thephoton energy is retained by the carbene and is, therefore, available to the vinyl chloridemolecule, or that the carbene carries no excess energy, when the vinyl chloride will then beenergised by the exothermicity of the reaction plus the activation energy of the 1,2 H transfer(in the carbene). Thisenergyof activation has been assumed to be 42 kJmol-l. The twovalues obtained in this way are 542 and 383 kJ mo1-1 respectively.EXCITED 1 ,1-D IC H LO ROE THANEA minimum energy content for the 1,l-dichloroethane may be calculated by assuming itto be formed by the reaction of HCl with thermalised methylchlorodiazomethane.A muchhigher energy content will be obtained if the dichloroethane is formed from the methyl-chlorocarbene and HCI. The values obtained are 247 and 414 kJ mol-l. The " high "value assumes that the carbene itself carried no excess energy. If we make allowance forthis possibility, then some dichloroethane molecules could have excess energies as high as615 kJmol-l. It should further be noted that the minimum value is unrealistic in that itassumes that the reaction of the diazocompound with HCl has a zero energy of activation.M. R. Bridge, H. M. Frey and M. T. H. Liu, J. Chem. SOC. A, 1969, 91.(a) R. A. Moss and A. Mamantov, J. Amer. Chern. Soc., 1970, 92, 6951; (b) P. Cadman,W. J. Engelbrecht, S. Lotz and S. W. J. Van der Merwe, J. S. African Chem. Inst., 1974, 27,149; (c) W. E. Jones, J. S . Wasson and M. T. H. Liu, J. Photochem., 1976,S, 193, 3112024 PHOTOLYSIS OF 3-METHYL-3-CHLORODIAZIRINEW. H. Graham, J. Amer. Chem. Soc., 1965, 87,4396."(a) G. 0. Pritchard and F. M. Servedio, Int. J. Chem. Kin., 1975, 7, 99; (6) F. H. Dorer,J. Phys. Chem., 1969, 73, 3109; 1970, 74, 1142; (c) T. F. Thomas, C. I. Sutin and C. Steel,J. Amer. Chem. SOC., 1967, 89, 5107; (d) R. J. Campbell and E. W. Schlag, J. Amer. Chem.SOC., 1967, 89, 5103.M. J. Amrich and J. A. Bell, J. Amer. Chem. SOC., 1964, 86, 292.A. H. Laufer and H. Okabe, J. Amer. Chem. Sac., 1971,93,4137. ' A. H. Laufer and H. Okabe, J. Phys. Chem., 1972,76,3504.* F. Lahmani, J. Phys. Chem., 1976,80,2623.R. A. G. Smith and J. R. Knowles, J. Amer. Chem. SOC., 1973, 95, 5072; J.C.S. Perkin II,1975, 686.l o J. C. Hassler and D. W. Setser, J. Amer. Chem. SOC., 1965, 87, 3793.'l P. Cadman and W. J. Engelbrecht, Chem. Cumm., 1970,453.l2 C. W. Gullikson and J. R. Nielson, J. MoZ. Spectr., 1957, 1, 158.l 3 E. Tschinkow-Roux and S . Kodama, J. Chem. Phys., 1969,50,5297.l4 H. Hartmann, H. Heydtmann and G. Rinck, 2. phys. Chem. (Frankfurt), 1961,28,71.J. C. Hassler and D. W. Setser, J. Chem. Phys., 1966,45, 3246.l6 D. W. Setser and B. S. Rabinovitch, Canad. J. Chem., 1962,40, 1425.l7 W. J. Dunning and C. C. McCain, J. Chem. Suc. B, 1966,68.l* C. B. Moore and G. C. Pimentel, J. Chem. Phys., 1964,40,342.l9 W. H. Archer and B. J. Tyler, J.C.S. Faraday I, 1976, 72,1448.zoD. R. Stull, E. F. Westrum and G. C. Sinke, m e Chemical Thermodynamics of OrganicCompoundr (Wiley, N.Y., 1969).(PAPER 7/708
ISSN:0300-9599
DOI:10.1039/F19777302010
出版商:RSC
年代:1977
数据来源: RSC
|
229. |
Reaction of oxygen atoms with nitromethane and with nitroethane |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 73,
Issue 1,
1977,
Page 2025-2030
Leo F. Salter,
Preview
|
PDF (454KB)
|
|
摘要:
Reaction of Oxygen Atoms with Nitromethaneand with NitroethaneBY LEO F. SALTER AND BRIAN A. THRUSH"Department of Physical Chemistry, University of Cambridge,Lensfield Road, Cambridge CB2 1EPReceived 10th June, 1977The reaction of oxygen atoms with nitromethane and nitroethane has been studied in a dischargeflow system at temperatures between 294 and 473 K. The initial attack involves abstraction of ahydrogen atom, for which the rate expressions are shown to be 10(10'18*0*27) exp (-22.44 1.8 kJmol-l)/RTdm3 mol-' s-' and 10(10*50*0.96) exp (-22.7+ 7.0 kJ m~l-~)/RTdm~ mol-' s-l respec-tively.Nitromethane and nitroethane have a variety of uses l* and the thermal decom-positions of nitrocompounds exhibit many points of experimental and theoreticalintere~t.~ Following earlier studies on alkyl nitrites and alkyl nitrate^,^ dischargeflow methods have been used to study the reaction of oxygen atoms with nitromethaneand nitroethane.Such experiments provide useful information about the mechanismof reaction of nitrocompounds in pyrolysis and combustion, where free radicalspredominate.EXPERIMENTALPROCEDUREThe apparatus and procedure were as reported previously.4* Infrared spectra wererun on a Perkin-Elmer Model 257 spectrometer, using a 10 cm path length, and gas analysiswas performed with a Perkin-Elmer F-1 1 chromatograph using a flame-ionization detectorand a 12ft, +in. 0.d. stainless steel column, packed with 5 % Cabowax on Teflon 6 at333 or 363 K.Total pressures in the flow system were between 0.2 and 1.0 kPa.For the kinetic studies,concentrations of oxygen atoms and reactants were typically (1 -4) x mol dm-3 and< 3 x lo-* mol dm-3 respectively.PURIFICATION OF REAGENTSNitromethane was distilled under vacuum at 273 K, the initial and final fractions beingdiscarded. The liquid was then subjected to several freeze-pump-thaw cycles, followed bydirect pumping at room temperature prior to use. Nitroethane was purified similarly,except that an initial distillation at 196 K preceded that at 273 K. Comparison of theinfrared spectra of the compounds with tabutated spectra,6 and chromatographic analysis,indicated the impurity levels to be below 1 %.Nitric oxide (Matheson) was purified by passage through two traps of silica gel at dryice temperature. This removes all other oxides of nitr~gen.~ Nitrogen itself was readilyeliminated by distillation at 77 K.Nitrogen dioxide, oxygen and argon were purified as reported previously.4*2022026 0 ATOM REACTION WITH NITROALKANESRESULTSREACTION OF OXYGEN ATOMS WITH NITROMETHANEThe kinetic behaviour of the reaction of oxygen atoms with nitromethane resemblesthat of the reaction of oxygen atoms with methyl nitrite and methyl nitrate.5 Themeasured limiting reaction stoichiometry is close to 7 (oxygen atoms removed pernitromethane molecule) and decreases with increasing substrate flow due to reactionsbetween product radicals and nitromethane (fig.1).nitromethane/mol dm-3 xFXG. 1 .--Dependence of reaction stoichiometry on nitromethane concentration.time at 294 K ; [010 = 4.0 x lo-' mol ~Irn-~.9, 0.1 s reaction0, 0.23 s reaction time at 423 K; [010 = 0.9 xlo-' mol dm-3.On the basis of this and previous work,* two mechanisms appear possible, bothOne giving stoichiometries n = 7.These differ in the initial mode of attack.mechanism isO+CH3N02 -+ H3CO+N02 AH1 = - 125 kJ mol-' (1)O+H,CO 3 H2CO+OH AH2 = -335 kJ mol-l (2)0 + H,CO + HCO + OH (3) AH3 = -63 kJ mol-'O+HCO -+ CO+OH+ CQ2+HH+HCO + H,+COO+OH --+ 0 2 + HO+NO, --+ 02+N0and the other begins0 + CH3N02 -+ OH + CHzNO2O+ CH2N02 + H2CO +NO2and is also followed by steps (3) to (7).AHda = -362 kJ mol-lAHab = -470 kJ mol-'AH5 = -366 kJ mol-lAH6 = -71 kJ mol-1AH7 = - 192 kJ mol-'( 4 4(44( 5 )(6)(7)(8)(9)AH8 + AHg = - 460 kJ mol-1In both these mechanisms the two initial steps yield OH and NO2 which reactrapidly with 0, giving a stoichiometry of 4.The subsequent reaction of oxygenatoms with f~rmaldehyde,~ which is fifty times faster than the initial step, has ameasured stoichiometry of 2.8 giving an overall stoichiometry close to the measuredvalue of 7.The observed Arrhenius parameters for the initial reaction agree well with thosepredicted for the abstraction of hydrogen by atomic oxygen from a series of primarL. F. SALTER AND B . A . THRUSH 2027alkanes lo and are also virtually identical with those observed for the reaction ofoxygen atoms with methyl nitrite and methyl nitrate.5 It seems reasonable toconclude that reaction (8) is the initial step, with k8 given byk8 = 10,(10'18*0*27)exp( -22.4+ 1.8 kJ mo1-l)/RTdm3 mol-' s-l.The error limits correspond to two standard deviations from the straight line in fig.2and at 295 K this expression gives k 8 = 1.6 x lo6 dm3 mo1-l s-l ; in fair agreementwith the value of 1.9 (k0.3) x lo6 dm3 mol-1 s-l reported previously.*FIG. 2.-Arrhenius8.31104 KITfor initial attack of oxygen atoms on 0 nitromethane and 0 nitroethane.REACTION OF OXYGEN ATOMS WITH NITROETHANEThe observed stoichiometry (fig. 3) for the reaction of oxygen atoms with nitro-ethane approachesstoichiometry plota limiting value of - 11 at low substrate concentration. Thefor nitroethane rises more sharply towards the limit than thenitroethanelmol dm-3 xFIG.3.-Dependence of reaction stoichiometry on nitroethane concentration.time at 473 K ; [O], = 9.1 x lO-'mol dm-3.0, 0.2 s reaction9, 0.175 s reaction time at 333 K ; [O], = 5.2 xmol dm-3.corresponding plot for nitromethane ; this behaviour was also observed for thesimilar plots of ethyl nitrites and nitrates when compared to their methyl analogues ;the difference probably arises from the greater number of free radical species (e.g.,OH, H) produced from fragmentation of ethyl derivatives as compared with th2028 0 ATOM REACTION WITH NITROALKANESmethyl derivatives. As in the previous studies a slightly lower activation energy forthe ethyl derivative is also observed here, supporting the analogous mechanism0 + CH3CH2N02 + CHSCHNOZ +OH (10)0 + CHSCHN02 + CH3CHO + NO2 (1 1) ] AHlo + AHl = -493 kJ mol-1O+NO, + NO+OZ AH = - 192 kJ mol-l (7)O+OH + H+O2 AH = -71 kJ mol-1 (6)O+CH3CH0 + OH+CH3C0 AH = -63 kJmol-l (12)O+OH -+ OZ+H AH = -71 kJ mol-l (6)(1 3)O+CH3 + H2CO+H AH = -286 kJ mol-l (14)0 + CH3CO -+ CH3 + C02 AH = -430 kJ mol-1followed by reactions (3) to (6) between oxygen atoms and formaldehyde.Thismechanism gives an overall stoichiometry of n = 11 and the reactions of oxygenatoms with the intermediate molecules HkCO and CH3CH0 are both more than tentimes faster than the initial step at 300 K9* l1 Incorporating this stoichiometryinto the rate equation yields the Arrhenius plot shown in fig. 2 where the best straightline givesklo = 10(10*50*0*96) exp (-22.2f7 kJ mo1-l)/RTdm3 mol-l s-l.The error limits quoted are two standard deviations from the straight line in fig.2.PRODUCTSUnlike the alkyl nitrites and nitrates, the nitro compounds contain a C-N linkagewhich might partially survive free radical attack. An attempt was, therefore, madeto identify some of the products of the oxygen atom-nitroalkane reaction. It wasdifficult to obtain sufficient amounts of material for analysis whilst maintaining theconditions for which secondary reactions were minimized, i.e., low substrate con-centration and short reaction times. For gas chromatographic analysis a trap at196K was placed at the end of the flow tube and the product collected. Thereliability of these analyses, performed after some hours of continuous sampling,should be viewed with caution, since some subsequent reaction could have occurredin the cold trap.For nitromethane a portion of the gas-flow was bubbled through alkaline aqueousferrous sulphate and the " Prussian Blue " test for hydrogen cyanide performed.This sensitive test could not detect the presence of hydrogen cyanide amongst theproducts of reaction showing that hydrogen cyanide represents <0.1 % of thereaction products. A similar result was obtained with both infrared and gaschromatographic analysis of the condensed products.The absence of hydrogencyanide indicates that the C-N bond in nitromethane does not survive oxygen atomattack under these conditions and that no CH3N0 is formed, since this moleculeyields hydrogen cyanide via reaction (1 5) ;3(1 5 )It is also possible that the attack of oxygen atom on any nitrosomethane presentwould also yield hydrogen cyanide.Formaldehyde was not found in the gas chromatographic analysis of the products,but this is thought to reflect both the low steady state concentration of formaldehyde(kg - 50ks at 300 K) and the ease with which formaldehyde undergoes polymerizationCH3NO + CHZNOH + HCN+H,OL.F. SALTER AND B. A. THRUSH 2029on a cold surface [see, for instance, ref. (14)], rather than its absence as a primaryproduct. It is also possible that formaldehyde reacts with unconverted nitromethanein the condensed phase l 3 to give a variety of products in quantities too small forgas chromatographic detection.For nitroethane, gas chromatographic analysis of the collected products indicatedthat acetaldehyde was present as the dominant condensible product at short reactiontimes.No other products were found in quantities large enough for unambiguousidentification, in agreement with the suggested mechanisms which do not includeany other condensible products which would give a signal with flame ionizationdetection.As reported previously for nitromethane,8 a considerable acceleration in the rateof reaction of nitromethane, or nitroethane, and atomic oxygen is observed at highsubstrate concentrations and long reaction times.This results in complete consump-tion of the atomic oxygen and is almost certainly due to a chain reaction betweenoxygen atoms and the nitrocompounds.Under these conditions trace amounts oforganic products, but no hydrogen cyanide, were just detectable in the products fromthe nitromethane reaction, whilst for nitroethane the compounds methyl cyanide,methanol, ethanol and acetone were identified as products in addition to acetaldehyde.As expected, the product sampling technique permits only a qualitative examinationof the system, but the presence of methyl cyanide shows that under these extremeconditions C-N bond rupture is not involved in every reactive encounter of anitroethane molecule. The variety of organic products also suggests the occurrenceof a different mechanism from that in the " low substrate concentration-shortreaction time " regime used for the kinetic studies reported above.DISCUSSIONThe activation energies for hydrogen abstraction from paraffin hydrocarbons byatomic oxygen show a linear dependence on bond dissociation energy loEA = 0.36 [D(C--H)-343 kJ mol-l].If one assumes that this relationship also applies to the C-H bond in nitroalkanes,the energies of the C-H bonds broken in the primary processes (8) and (10) mustbe in the region of 405 kJ mol-I.As was found for alkyl nitrites and alkyl nitratesY5these are the energies expected for primary or secondary C-H bonds in alkanes;there is no evidence from the present work that the proximity of a nitro-group affectsthe C-H bond energy of an alkane. It should be noted that the observed pre-exponential factors for reactions (8) and (10) also agree well, within the error limits,with the values expected for an abstraction reaction by oxygen atoms involving threeprimary hydrogen and two secondary hydrogens respectively from an alkane. loIt has also been reported l 6 that hydroxyl radicals react with nitromethane andmethyl nitrite at similar rates [(5.5+0.6) x lo8 and (8.0+ 1.1) x lo8 dm3 mol-1 s-'respectively at 292 K].These rate coefficients are - 100 times greater than for thereaction of oxygen atoms with the same species as might be expected, since thereaction RH+OH 3 R+H20 is 71 kJ mol-' more exothermic than the reactionRH+O -+ R+OH, which is almost thermoneutral. In contrast, the reactionRO + OH + R + H02 is 230 kJ mol-' less exothermic than RO + 0 --+ R + O2 forany molecule and abstraction of 0 by OH from nitroalkanes would be endothermic.The relative reactivities of the nitroalkanes with 0 and with OH, therefore, providefurther evidence that abstraction of H is the primary mode of attack2030 0 ATOM REACTION W I T H NITROALKANESIn the thermal decomposition of mononitroalkanes, the presence of alkyl andalkoxy radicals, and their reactions with oxides of nitrogen, leads to a wide range ofprod~cts.~ Under the conditions used in the present experiments, the concentrationof atoms or radicals other than oxygen is kept very small; hence reactions occurchiefly with the oxygen atoms which are present in high concentration.The hydrogenabstraction reaction observed here should also apply to attack by hydrogen atoms,hydroxyl radicals and alkoxy radicals, but the suggested chain reaction which occursat high substrate concentrations and long reaction times, together with the formationof a variety of organic products, indicates that a different mechanism may aiso beimportant under such conditions.From a simple examination of the temperaturedependence of the disappearance of the emission at the end of the flow tube, thisalternative mechanism appears to become more important at higher temperaturesand is likely to play a role in the decomposition of nitroalkanes under the conditionspertaining in combustion and explosion.The research reported here was sponsored in part by the United States Governmentthrough the European Research Office.J. D. Rose, Proc. 11th Int. Congress Pure and Appl. Chem., (London, 1947), 291.J. L. Martin and P. J. Baker, Kirk-Othmer Encyl. Chem. Tech., (2nd edn), 1967, 13, 864.G. M. Nazin, G. B. Manelis and F. I. Dubovitskii, Uspekhi Khim., 1968,37, 1443.J. A. Davidson and B. A. Thrush, J.C.S. Faraday I, 1975,71,2413.L. F. Salter and B. A. Thrush, J.C.S. Faraday I, 1977, 73, 1098.R. H. Pierson, A. N. Fletcher and E. C. Gantz, Analyt. Chem., 1956, 28, 1218.I. M. Campbell and K. Goodman, Chem. Phys. Letters, 1975,34, 105.G. P. R. Mack and B. A. Thrush, J.C.S. Faraday I, 1973, 69, 208.G. P. R. Mack and B. A. Thrush, J.C.S. Faraday I, 1974,70, 187.p. 292.’ E. E. Hughes, J. Chem. Phys., 1961,35, 1531.l o J. T. Herron and R. E. Huie, J. Phys. Chem., 1969,73, 3327.l2 G. Lunge and H. R. Ambler, Technical Gas Analysis (Gurney and Jackson, London, 1934),l 3 A. L. Myerson and J. J. Chludzinski, J. Chromatog. Sci., 1975, 13, 554.l4 P. Gray, Proc. Roy. SOC. A, 1954, 221,462.l5 L. F. Fieser and M. Fieser, Aduanced Organic Chemistry (Reinhold, N.Y., 1968), p. 461.l6 I. M. Campbell and K. Goodman, Chem. Phys. Letters, 1975, 36, 382.(PAPER 71993
ISSN:0300-9599
DOI:10.1039/F19777302025
出版商:RSC
年代:1977
数据来源: RSC
|
230. |
Pulse radiolysis study of the reaction of solvated electrons with sulphur hexafluoride in methanolic solution |
|
Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases,
Volume 73,
Issue 1,
1977,
Page 2031-2035
David W. Johnson,
Preview
|
PDF (323KB)
|
|
摘要:
Pulse Radiolysis Study of the Reaction of SolvatedElectrons with Sulphur Hexafluoride in Methanolic SolutionBY DAVID w. JOHNSON? AND G. ARTHUR SALMON"University of Leeds, Cookridge Radiation Research Centre,Cookridge Hospital, Leeds LS16 6QBReceiued 13th January, 1977The rate constant for the reaction of e, with SFG in deaerated methanolic solution has beendetermined to be kf = (1.88 k 0.06) x loio dm3 mol-' s-I. The product of this reaction, SF5, hasan absorption spectrum with a maximum at 300 nm with E = 1300 dm3 mol-I cm-' and decaysby mixed first- and second-order kinetics.In a recent paper Bansal and Fessenden have observed oxidising properties ofa species generated by the reaction of the hydrated electron with sulphur hexafluoride[reaction (l)] in aqueous solutions.They assigned this species to be *SF, and havedetermined that it has a lifetime with respect to hydrolysis by the solvent [reaction (2)]of -6 ,us. A previous measurement by Asmus and Fendler indicated the lifetimeto be <1.5 ps.e,; + SF6 + *SF5 + F- (1)*SF,+H,O + SF,+F-+*OH+H+. (2)In the present work the spectrum and decay kinetics of *SF5 are studied in methanolwhere hydrolysis by the solvent is likely to be much slower.EXPERIMENTALThe techniques of pulse radiolysis employed in this laboratory have been describedprevio~sly.~~ Solutions of SF6 at less than atmospheric pressure were prepared by distil-lation of a measured quantity of SF6 into a flask containing deaerated methanol.to be3 . 3 2 ~ mol dm-3 (mm Hg)-', independent of temperature between 15 and 25°C.Thisvalue is in excellent agreement with that measured by Rzad and Fendler by gas chromato-graphic techniques.The solubility of SF6 was determined by the method of Markham and KobeRESULTS AND DISCUSSIONThe reaction of solvated electrons, e;, with SF, [reaction (3)] in neutral methanolicsolutions was followed by observing the decay of absorption oE e, at 580 nm insolutions with 1.22 x lod4 < [SF6]/mol dm-3 < 1.46 x In each case the decaysfollowed pseudo first-order kinetics and yielded k3 = (1.88+0.06) x lo1* dm3 mol-'s-l,e, +SF6 + *SF5 +F-. (3)In fig. 1 is shown the absorption spectrum induced in SF,-saturated solutions by0.2 ,us, 5.9 krad pulses corrected for the small decay of absorption occurring duringt Present address : Department of PhysicaI Chemistry, University of Bradford, Bradford BD7 1DP.203 2032the pulse.This spectrum is significantly different in both shape and magnitude fromthat of hydroxymethyl radicals, -CH,OH, observed in N20-saturated solutions. Atthe concentration of SF6 employed, the reaction of eH with SF, isessentiallycompletedduring the pulse since t3 21 1.5 ns. This was confirmed by the absence at the endof the pulse of any absorption at 580 nm due to e;. The effect of concentration ofREACTION OF SOLVATED ELECTRONS WITH SF,4 .O3.09 I ( 3 2.0'1.07- - 1 I I2.01.5n c;:? v1.0 a H0.5I I I I2 5 0 3 0 0 3 5 0 400h/nmFb. 1.-End-of-pulse spectrum (+, - - -) and the spectrum of *SF5 (0, -)* generated by a 0.2 ps,5.9 krad pulse in an SF6-saturated solution.SF6 on the radiation induced absorption was studied at 360 nm, where the absorptionof *CH20H is in~ignificant.~* * These data are given in table 1.Also the GET valuesat each [SF,] were plotted against, and found to be linearly related to, the yields offluoride-ion from y-irradiated SF6 solutions as measured by b a d and Fendler ti (seefig. 2). The fluoride-ion is thought to be formed in reaction (3) and by the subsequentTABLE 1 .-VARIATION OF Ge3 60 AND (1 /6)G(F-) WITH [SF,][SFtj]/lllOl dm-3 G8360 1/6G(F-) 31.22 x 700+ 30 0.91 . 2 2 ~ 1400f30 1.42 . 5 ~ 1650f50 2.1(0.73)§$ Data taken from ref, (6) ; 5 the value in brackets is corrected as described in the text.degradation of *SF5.The value of G(F-) obtained by Rzad and Fendler at thelowest concentration of SF6 (1.44 x loq3 mol dm-3) was corrected for the differencein rates of decay of e, in pulse- and y-irradiated methanol in the absence of a soluteby using data from ref. (1 1). The possibility that the absorption may be formed bythe reaction of SF6 with the other radicals generated in irradiated methanol (.CH20H,* Calculated as described in text. t GE is the product of the radiation chemical yield, G(X), of species X in units of molecules(109 ev)-l and its molar decadic extinction coefficient in units of dm3 mol-I cm-ID. W. JOHNSON AND G . A. SALMON 2033and CH30* loH*+CH30H -+ H,+*CH,OH (4)CH30- +CH30H + CH30H+CH20H. (5)CH30* and Ha) can be discounted on kinetic grounds.Both Hereact rapidly with the solvent forming CH20H [reactions (4) and (5)]The rate constants for these reactions, 1.6 x lo6 and 2.63 x lo5 dm3 mol-l s-lre~pectively,~- l o are such that reaction of He and CH30* with SF6 could only besignificant if the rates of these latter processes greatly exceeded the diffusion controlledrate in methanol. Furthermore, the limiting yield of fluoride ion at high concentra-tions of SF6 as indicated by a reciprocal-reciprocal plot extrapolation of the data intable 1, i.e. 1/6G(F-) = 2.3, can be accounted for entirely by reaction (1). If thereactions of He, CH30* or *CH20H with SF6 occurred to a significant extent thenmuch higher yields of fluoride would be expected with extrapolated yields at high[SF,] exceeding 1/6G(F-) = 5.63, the radical yield in argon saturated sol~tion.~ Forthese reasons we believe the absorption at 360 nm is due to -SF5 formed by reaction(3).Observation of the formation of *SFS using dilute solutions of SF6 provedimpossible owing to the large absorption of e; at 360 nm.I I I 1 1I I I I a.5 1.0 1.5 2.0116 G(F-)FIG. 2.-Comparison between pulse radiolysis yidd of *SF5 at 360 nm and the yield of F- formed,on y-radiolysis.?An estimate of the spectrum of *SF5 can be made by assuming that the totalyield of radicals is unaffected by the presence of SF6 and is equal to that in argon-saturated solutions ([G(R*)]*' = 5.63 +0.05).4 Errors due to the use of this assump-tion must be minor since although the radical yield in nitrous oxide solution isincreased to [G(R-)]N~~ = 6.45+0.05,4 the concentration of SF6 at saturation islower than that of N20 by a factor of six.SF6 reacts with e; in competition withthe reactions of the latter with the solvent and H+ [reactions (6) and (711, in each caseforming He which react rapidly with the solvent and form -CH20He;+H+ + He (6)(7) e; + CH30H + H-+ CH30-.t r-radiolysis data taken from ref. (6)2034 REACTION OF SOLVATED ELECTRONS WITH SF6Therefore, the total yield of radicals at the end of the pulse is given by eqn (I), whencethe absorption at wavelength A, [GE]~, is given by eqn (11)G(.SF5) + G(CH2OH) = [GfR*)]*' = 5.63 & 0.05 (1)Substitution of G(-SF5) = 1/6G(F-) = 2.1 k0.2 as measured by Rzad and Fendlerfor SF6-saturated solutions yields eqn (111)[GE]A- (3.53 $-0.2)&(*CHzOH),4*SF5), = 2.1 k0.2for the extinction coefficient of .SF,.The extinction coefficients so obtained and thecontribution of -SF5 to the end-of-pulse spectrum are shown in fig. 1. Values of&(*CHzOH)~ are taken from ref. (8). The value of E ( - S F ~ ) ~ ~ ~ may also be calculatedfrom the slope of fig. 2 which givesSince &(.CH20H)360 = (40k 1) dm3 mol-l cm-1 we obtain E ( . S F & ~ ~ = 750+120 dm3 mol-1 cm-I. This value, which utilises data at several concentrations ofSF6, is in excellent agreement with &('SFg)360 = 710_+20 dm3 mol-' Cm-' fromI&~SF,),~O-&(*CH~OH)~~~] = 710f 120 dm3 MO1-l Cm-l.fig. 1.time/pspulse.FIG. 3.-Decay of adsorption of *SFS in SF6-saturated methanol.A, 5.9 krad pulse ; 0,25.5 kradThe decay of absorption followed at 360 nm in SF,-saturated solutions was foundto fit linear plots of reciprocal optical density against time (see fig. 3), but the slopeswere dependent on dose with kobs/&('SF5)360 nm equal to 1.7 x lo7 and 9.8 x lo6 cm s-Iat doses per pulse of 5.9 and 25.5 krad respectively, i.e. kobs = 1.0 x 1 O 1 O and 6 x lo9dm3 mol-l s-l.In aqueous solution *SF5 decays by a pseudo first-order reaction with the solvent[reaction (2)], but a discrepancy exists between the two measurements of the rateconstant which have been made (>4.5 x lo5 and - 1.1 x lo5 s-l).l* The linearreciprocal optical density against time plots obtained in this work indicate that SF,.decays predominantly by a second order process.However, since the lines for thD. W. JOHNSON AND G . A . SALMON 2035different doses cannot be superimposed by a translation along the time axis we mustconclude that an additional second order process must be considered involving anothertransient species. We therefore propose that reactions (8) and (9) are mainly%SF5 + SF,+SF, (8)(9) *SFS + *CH20H 4 SF4 + HCHO + Hf + F-responsible for the decay of *SF5 ; however, the first-order reaction of *SF5 with thesolvent [reaction (lo)] is probably also contributing to the dependence of kobs on dose.*SF5 + CH30H -+ SF, + HF +*CH,OH. (10)The limited data available on the decay of -SF5 do not permit any realistic assessmentof kg, k9 and klo. Reaction (8) will not affect the yield of F- produced by electronsreacting with SF6 as measured by Rzad and Fendler since, at the dose-rates usedin y-radiolysis studies, the decay of -SF5 should take place entirely by the first-orderreaction (10).D. W. J. wishes to thank the S.R.C. for the award of a studentship.K. Bansal and R. H. Fessenden, J. Phys. Chem., 1976, 80,1743.K. D. Asmus and J. H. Fendler, J. Amer. Chem. SOC., 1970, 92,2625.D. W. Johnson and G. A. Salmon, J.C.S. Faraduy I, 1975, 71, 583.D. W. Johnson and G. A. Salmon, J.C.S. Faraduy I, 1977, 73, 256.A. A. Markham and K. A. Kobe, J, Amer. Chem. Soc., 1941,63,419.S . J. Rzad and J. H. Fendler, J. Chem. Phys., 1970, 52, 5395. ’ F. S. Dainton, G. A. Salmon and P. Wardman, Proc. Roy. SOC. A, 1969, 313, 1.* D. W. Johnson and G. A. Salmon, Canad. J. Chem., 1977, 55, 2030.M. Anbar, Farhataziz and A. B. Ross, NSRDS-NBS 46.D. W. Johnson, PkD. l7zesis (University of Leeds, 1976).lo F. S. Dainton, I. V. Janovska and G. A. Salmon, Proc. Roy. SOC. A, 1972,327, 305.(PAPER 71067
ISSN:0300-9599
DOI:10.1039/F19777302031
出版商:RSC
年代:1977
数据来源: RSC
|
|