II. COLLISION PROCESSES INVOLVING CHEMICAL REACTIONS Vibrational Excitation in Photochemical Processes BY N. BASCO* AND R. G. W. NORRISH Dept. of Physical Chemistry, University of Cambridge Received 26th January, 1962 Various ways of producing disequilibrated populations of vibrationally excited molecules and (1) vibrational excitation of the ground state involving prior electronic excitation ; (2) photolysis followed by reactions of the type, radicals are described. The main processes considered are : A+ BCD +AB* + CD ; (3) direct production of vibrationally excited molecules in reactions of the type, ABC+/iv-+AB*+C. The energy distributions resulting from these types of reactions, the study of vibration-translation and vibration-vibration energy transfer in these systems, and the chemical properties of the vibra- tionally excited species are also discussed.The techniques of flash photolysis and kinetic spectroscopy make it possible (in theory at least) to study the physical and chemical properties of vibrationally excited species and the nature of the processes in which they are produced. This paper describes to what extent these possibilities have so far been explored. Experimentally, a suitable molecule in the presence of an inert diluent (usually nitrogen) is subjected to an intense flash of visible and U.V. radiation. This is of 20-50 p sec duration and is produced by discharging 1500-5000 J, stored in a bank of condensers at 10-20 kV, through an inert gas (&,A) in a quartz tube. The absorption spectra of the products are then taken at various known time-intervals measured in microseconds by means of a second much less powerful but even shorter flash, and recorded photographically. In the systems to be considered, these spectra reveal the presence of molecules or radicals possessing up to 20 quanta of vibrational energy though in all other respects at room temperature.Since it is not normally possible, at room temperature, to detect the presence of molecules even with only one vibrational quantum, the degree to which the vibrational levels are overpopulated is evidently extremely marked. EXCITATION INVOLVING AN EXCITED ELECTRONIC STATE The simplest way of producing some vibrationally excited molecules would be by flashing with infra-red radiation, but this does not yet seem to have been done.A somewhat similar method is available for diatomic molecules with discrete intense absorption in the visible or U.V. The molecule is electronically excited and then, either by fluorescence or by collision quenching, returns to various vibrational levels of its ground electronic state according to the Franck-Condon principle. In suitable * Department of Chemistry, University of Sheffield, Sheffield, 10. 99100 VIBRATIONAL EXCITATION cases, the population of the first few excited levels can be interested several hundred- or thousand-fold by this method-sufficient to allow direct observation of the excited species. This type of excitation has been achieved in the case of nitric oxide 1 which, when flashed in the presence of a large excess of inert gas (to ensure isothermal conditions), undergoes the following reactions : electronic excitation fluorescence NO.X ~ I I ( U = O)+h-+NO. A~X(V = 0,1,2) NO. A ~ Z ( U = 0,1,2)+NO. X~II(U = 0,1:2. . .)+hv NO. X ~ I I ( U = ~)+NO.X~IT(U = 0)+2NO. X~II(U = 1). or collisional quenching These lead, almost exclusively, to a several hundred-fold overpopulation of the first vibrational level of the ground state, the relaxation of which could then be followed quantitatively in the presence of various other molecules. An extension of this method makes possible the vibrational excitation of free radicals which are produced by photolysis and then excited in a similar way to that described above. Both the photolysis and the subsequent excitation can be achieved by the same flash and, provided the radical is sufficiently stable and absorbs radia- tion strongly, several vibrational levels can be sufficiently overpopulated to permit observation.An example of this type is the excitation of the cyanogen radical produced in the photolysis of cyanogen and the cyanogen halides.;! Following the photolysis, in which the radicals are produced with no vibrational excitation, the CN radical under- goes electronic excitation to the B2Z and A2Il states. On returning to the ground state (X2I=), about 15 % will possess one or more vibrational quanta. The radical absorbs so strongly that this process can be repeated many times during the period of the photoflash and each time the number of vibrationally excited radicals is in- creased. Thus, theoretically, an infinite vibrational temperature can be produced in the absence of dissociation or relaxation. In practice, only the first six excited levels can be observed directly because of the experimental difficulty in resolving the close sequences which make up the violet system of CN.For the same reason, quantitative measurements are only possible up to the fourth excited level, but the relative populations of these five levels can be measured as functions of, for example, time 2nd pressure. These measurements show that the theoretically predicted distribution is in fact closely realized in that the levels which can be observed are found to be approximately equally populated. Evidence that higher Bevels are also occupied is provided by the observation that, under certain conditions, the popula- tions of the fourth excited level (and probably of the fifth also) can greatly exceed that of the lower levels.This situation arises as a result of a partial relaxation from the predicted distribution and the fact that up to 90 % (or even more) of the radicals are in various vibrationally excited levels opens up the possibility of studying their properties. This may be contrasted with nitric oxide where only about 5 % of the molecules are observed to be excited and virtually all of these to only the first excited level. The production of vibrationally excited carbon monosulphide in the flash plioto- Iysis of carbon disulphide 3 is, very-probably, another example of this type though, so far, other mechanisms have not been excluded.The sulphur atoms produced at the same time give rise to sulphur (S2) which is also observed to be vibrationally excited. The recombination of atoms theoretically provides a method for the production of vibrationally excited molecules with energy up to the dissociation limit. The experimental difficulty in producing atoms in a simple system of flash photolysis and NO . A2X(v = 0,1,2) + M-NO . X2II(u = 0,1,2 . . .) resonance exchangeN. BASCO AND R . G. W. NORRISH 101 the vibrationally excited molecules produced does not seem to have been In principle, the presence of these vibrationally excited species could be short time delays with, for example, bromine and iodine. in detecting over come. detected at REACTIONS BETWEEN ATOMS AND MOLECULES There is a large group of reactions of the type A+ BCD+AB* + CD in which the exothermic reaction between an atom and a molecule produces a vibra- tionally excited molecule AB*.In some cases the exothermicity is provided by the electronic energy of the atom.4 This class of reaction, proposed by McGrath and Norrish 5 has been reviewed by Polanyi,6 Smith7 and ourselves 8 and it is only necessary here to mention two of the outstanding problems of interest. The most striking features of these reactions are the very large degree of excitation observed and their apparent specificity, in that only the molecule AB containing the newly formed bond has been observed to be vibrationally excited. In all reactions, however, there were good reasons why CD* might have escaped detection even if it were produced.In the last year, two exceptions to this have been reported and both support the original conclusion. We 9*1* studied the reaction under conditions where NO* would have been detected had it constituted about 5 % or possibly less of the total NO produced, but could detect none. C1+ ClNO+ Cl2 + NO A similar reaction, viz., H + ClNO+HCl+ NO was studied by Cashion and Polanyill They observed emission in the infra-red from both products, but that from NO* was <l/lOth of the total. Since, because of the relative slowness of vibrational relaxation, vibrational disequilibrium should result from any exothermic reaction, these results show that equipartition is not the rule in reactions of this type. The extent to which the molecule AB can be vibrationally excited is not also finally settled, though the evidence strongly suggests that at least some molecules are produced with the entire exothermicity of the reaction in the form of vibration.In fact, in one case, 0; from N02,89 12 the vibrational excitation has been observed to exceed the exothermicity. The most interesting problem of whether all the molecules are initially produced with the maximum degree of excitation as proposed by Folanyi 6 or whether there is a finite probability of individual molecules being produced with any degree of excitation as suggested by us 8 still remains open. Of course, because of relaxation and the limited time resolution of present techniques, a distribution of energy among all vibrational levels will be observed in practice, whatever the initial distribution.Improved techniques, with increased sensitivity allowing the use of much lower pressures, will be required before the question is finally settled, and some progress is already being made in that direction. The fact that the relative populations of excited levels have not been observed to change greatly with time, suggests to us that the initial distribution, in those systems, is one in which all levels are populated in the initial reaction with an increasing probability of excitation towards lower levels. This implies that the distribution observed in flash photolysis experiments may be fairly close to the initial distribution unless a very much more rapid relaxation process occurs before observation is possible. This could only be resonance exchange and even if this occurred on every collision, it is doubtful whether it could lead to a very marked change in the distribution in the time available.The effect of nitric102 VIBRATIONAL EXCITATION oxide on NO* and oxygen on 0,' has been studied 10y12 and the results seem to support the view that, generally, the concentration of excited species is likely to be too small for resonance reactions to be of such importance in comparison with the other mechanisms. DIRECT VIBRATIONAL EXCITATION DURING PHOTOLYSIS In principle, whenever a molecule dissociates following the absorption of radiation of shorter wavelength than that corresponding to the strength of the bond broken, the fragments may carry part of the excess energy in the form of vibration. This possibility is in fact so obvious that it may be surprising how little real evidence there is to substantiate it.The presence of vibrational energy in a molecule can sometimes be inferred from the exceptional chemical reactivity of the molecule itself or of products formed from it. That this reactivity is due to vibrational rather than translational energy is deduced from its ability to survive a relatively large number of collisions. In this way, evidence has been obtained that the methyl radical produced in the photolysis of methyl iodide with 1849 1$ radiation,l3 the methylene radical produced from diazomethane 14 and several halogen substituted methyl radicals produced from various po ly- halomet hanes 15 arc vibrat ionally excited.As far as the primary process is concerned, the evidence which can be obtained in this way, though valuable, is necessarily indirect and limited. To study the precise extent of the vibrational excitation in various molecules and its variation with the wavelength of the radiation absorbed as well as the corresponding change in chemical reactivity it is highly desirable to observe the excited species directly. This will, of course, be essential where the degree of excitation is insufficient for chemical reaction and this may very often be the case. As we have seen, there are many examples of the direct observation of vibrationally excited species in flash photolysis experiments but in only three of these has it been suggested that they could have been produced in the primary photolytic process. These are the reactions 16,179 99 10 CHZCO + hv+CO(v<2) + CH2 NOCl (or NOBr) + hv+NO(v< 11) + Cl(or Br) To these we may add, for further consideration, the reactions 0, + hv+O; + 0 cs, + IZV-,CS* + s in which the production of the vibrational species observed is ascribed to other mechanisms.33 18 In all of these reactions, several mechanisms are possible and it is difficult to prove beyond doubt which is in fact responsible in a particular case.Only for the nitrosyl halides and cyanogen and the cyanogen halides have serious attempts been made to do this. When nitrosyl chloride or bromide is flash-photolyzed, nitric oxide molecules with up to 11 quanta of vibrational energy are observed in absorption, though the system is maintained at room temperature by the addition of a large (up to 7000-fold) excess of inert gas.Both compounds absorb strongly in the U.V. and thus, with a quartz apparatus, up to 110-120 kcal/mole of excess energy are available. The results show that up to at least 55 kcal of this can appear in vibration and it seems probable that with improved experimental techniques the presence of even moreN . BASCO AND R. G. W. NORRISH 103 highly excited molecules will be revealed. In these systems there are good expei- mental and theoretical reasons for believing that the excited NO is produced directly in the photolytic process. The most interesting alternative is a sequence of reactions involving a singlet and a triplet excited state of NOCl(and NOBr) and the 411 state of NO.This mechanism cannot in fact be entirely excluded, but, using light filters, it was possible to show that it restricts the energy of the 4l-I state to not more than about 3.5eV above the ground state. Most of the present theoretical and experi- mental evidence points to a value 1 eV higher than this. The main experimental difficulty with this system is the decreasing absorption coefficient of the nitrosyl halides towards longer wavelengths, so that, at present, radiation below about 2600 A must be used in order to produce a detectable amount of decomposition. When this restriction is removed (and an improvement in sensi- tivity by a factor of ten is already practicable) it should be possible to study the variation of the degree of excitation with wavelength and, at the same time, to resolve the question of whether the 4l-I state of NO is involved.Regarding the other possible examples of this type, the most general alternative mechanism to be excluded is the electronic excitation process This very probably occurs in the case of CS* (as suggested by the authors 3) so that demonstrating the simultaneous occurrence of another mechanism may be difficult. Apparently the least complicated is ketene where the photoflash radiation transmitted by quartz is not capable of exciting CO. The case of SO* is of interest in another important connection in that an alternative mechanism involves its production as an excited SO molecule in the reaction This was one of the few possible sources of evidence against the postulate that only the AB bond is excited.o(%) + 0, -+ 0: + o2 0 + so,-,o2 +so*. The reaction has for some time now been one of the most striking examples of its type. However, it now seems necessary, in the light of our work on the nitrosyl halides, to reconsider the mechanism by which vibrationally excited oxygen molecules are produced in the ozone system. Ozone absorbs U.V. radiation strongly with a maximum extinction coefficient around 2500A. At this wavelength, 90 kcal/mole of excess energy are available over that required to produce an atom and a molecule of oxygen both in their ground electronic states. At the effective limit of transparency of the quartz apparatus, the excess energy is about 125 kcal/mole-sufficient in fact to dissociate the oxygen molecule, while the highest degree of vibrational excitation so far observed (20 quanta 8 ) corresponds to only 76 kcal.Thus, on purely energetic grounds, the direct reaction could easily explain the experimental facts. Arguments against this mechanism and favouring the reaction between the oxygen atom and the ozone molecule may be summarized as follows. (i) The small change in the bond length.18 This would apply equally to the pro- duction of NO* from NOCl and NOBr. In our view, an overall change in bond length is not a necessary condition though it may favour vibrational excitation. (ii) It has been shown beyond doubt that the photolysis of ozone with U.V. radia- tion produces 1D oxygen atoms3 18 It would therefore be necessary to postulate an additional primary process. 0 3 + h v + 0 2 .3z;(~%0)+03~1 04 VIBRATIONAL EXCITATION (iii) The suppression of the 0; spectrum by the addition of a sufficient excess of H20, H2, HCl, CH4 and NH3 and the appearance of the OH and OH* spectrum 4 , 1 8 is well explained by the competing reactions and O'D + HR-, OH* + R O'D+ 0 3 + 0; + 0,. This implies that the second reaction is almost exclusively respomble for the pro- duction of 0;. Though for this to be a complete proof, the possibilities that RH, R or reaction products are efficient at removing 0; should also be considered. The reaction scheme 0% + 03-' 0; + o2 O:(v> 17)f 03+20, + 0 ' D 0 'D 4- H2 0 -+20H OH+ 03-+H02 + 0, HO2 + 03+20, +OH explains the observed quantum yields for the photolysis of dry and wet ozone with visible and U.V. radiation.(iv) The fact that, under certain conditions, there is an optimum flash energy for the production of the strongest 0; spectrum12 as is to be expected if 0 2 is produced in the reaction between an oxygen atom and an ozone molecule. The rate of this reaction evidentIy depends on the concentrations of both species and, obviously, if all the ozone is destroyed by the flash itself, there is none left to react with the oxygen atoms. A much more detailed study of the variation of the observed 0; concentra- tion with flash energy at various ozone pressures is required to test this argument ; but there are various complications and the mathematical treatment of the results will be difficult. (v) The production of 0; when NO2 or C102 are flashed is accounted for by the reactions 0 +N02-+O: +NO By analogy, the reaction of an oxygen atom with ozone should likewise produce 0; and the existence of many other reactions of a similar type strongly supports this mechanism.We conclude that there is no experimental support for the idea that the reaction 0 + c102-+ 0; + cio. 0, + hv+O: + 0 contributes significantly to the observed production of 0;. In our study of the photolysis of cyanogen and the cyanogen halides,2 the pos- sibility that the vibrationally excited CN radical observed was produced directly was investigated in detail. The compounds all absorb radiation of the shortest wave- length available and the excess energy lies between 40 and 90 kcal/mole-again sufficient to account for the degree of excitation observed. Using various light filters, we were able to show that the production of CN* in the process where R is either CN, Br or I, if it occurred at all, accounted for not more than 6 % of the total CN* observed.This may be compared to NOCl where, under certain conditions, about 50 % of the nitric oxide is observed to be excited. It appears that the direct production of vibrationally excited species may not, after all, be so obvious RCN+hv-+R+CN . X2C(v>O)N. BASCO AND R. G. W. NORRISH 105 and we believe it merits much more attention both from the theoretical and from the experimental points of view. We have already mentioned most of the known examples of the exceptional chemical reactivity of vibrationally excited species.13-15 The outstanding example of reactivity towards other molecules is the reaction O;(u> 17)+ 03+202 + OID postulated by McGrath and Norrish to explain their observations and the high quantum yield for ozone decomposition by U.V.radiation. Evidence for the reaction NO*(o> 8)+ NOC1+2NO + C1 has been sought by Wayne and ourselves,l9 but without success. Further information on this aspect would be of great interest. BEHAVIOUR OF VIBRATIONALLY EXCITED SPECIES The opportunities for vibrational relaxation studies provided by the systems described have so far been little exploited. Lipscomb, Norrish and Thrush20 followed the decay of the fifth, sixth and seventh vibrational levels of the excited oxygen produced in the flash photolysis of NO2 and C102. From half-life measure- ments they found that about 2,000collisions between 0; and C10 were required to remove a vibrational quantum.In a detailed theoretical paper, Schuler 21 showed that the analysis of relaxation data in terms of half-lives loses its meaning when more than two levels are involved in the relaxation process. To obtain information on energy transfer in multilevel systems, it is therefore necessary to follow in detail the time behaviour of the population of several (and preferably all) the energy levels. In many systems it may be difficult to make complete studies of this kind and it is therefore worth pointing out that reasonably accurate relaxation data could be ob- tained more simply from those multilevel systems in which the populations of the vibrational levels decrease towards higher levels. Even where it is not possible to measure the relative populations of two or three adjacent levels, it should be possible to detect changes in the distribution.If the change in the relative populations is small over a period of one or two apparent half-lives, then the required conditions hold. To illustrate this argument, consider a system in which the populations of the adjacent vibrational levels A, By C . . . are X,O, Xg, X,O . . . at time t = 0 and Xa, Xb, Xc . . . at time t and where A is the highest level present. Assuming step- wise degradation with equal probabilities for each level, it can be shown that the relative populations after n half-lives of level A are given by . . . etc. and can easily be calculated. It is immediately obvious from this that the larger the ratios Xg/Xz and X z / X z , the less the distribution changes with time and the closer the measured half-lives are to the true values.The error will not greatly exceed an order of magnitude provided that these ratios are greater than unity and the result may be very much better than this for higher values. Applying these calculations to the relaxation of 0; in the C102 and NO2 systems, we see that the fact that the half-lives for the seventh, sixth and fifth levels were not greatly different is inconsistent with the estimate that the levels were equally populated.106 VIBRATIONAL EXCITATION In fact, when the population of the Mth level had fallen to half its initial value, the ratio of the populations of the fifth and seventh levels (i.e. Xc/Xa) would have been about seven compared to the initial value of unity.While it is possible that an estimate of the relative concentrations could have been considerably in error, it seems much less likely that a '7-fold change in relative concentrations could have been missed. We conclude that the seventh, sixth and fifth levels were probably populated to an increasing extent and that the relaxation results need not have been in error by much more than a factor of two. On Schuler's analysis the error is by a factor of ten. Re-investigating the NO2 system, we 87 12 showed that all levels up to the thirteenth are populated but the previous treatment 20, which assumed that higher levels than the eighth were absent and that the population of the eighth itself was negligibly small, can easily be extended to cover this situation.However, our tentative estimate of the populations suggests that they decrease by about a factor of three towards higher levels and this, if confirmed, would support the argument presented above. A detailed account of the relaxation of nitric oxide studied by kinetic spectroscopy has been published,l the relaxation of the cyanogen radical has been studied2 and direct evidence for resonance energy transfer has been obtained. The reaction has already been mentioned,l and the similar reaction NO(u = n)+NO(u = O)-+NO(u = n-l)+NO(Y = 1) has been shown to occur in the NOCl system when nitric oxide is added3 10 Finally, when NO+CO are flashed the presence of CO(u = I) very probably arising from the reaction can be detected in the vacuum u.v.22 Quantitative measurements on this and the NO* + N2 exchange reaction are in satisfactory agreement with the theory of Herzfeld and Litovitz.23 Summarizing, one need only say that serious studies of the various problems discussed in this paper have virtually only just begun and that very much interesting work remains to be done. NO(u = 2) + NO(u = 0)+2NO(u = I) NO(u = l)+CO(u = O)+NO(u = O)+CO(U = 1) 1 Basco, Callear and Norrish, Proc. Roy. SOC. A, 1961,260,459. 2 Basco, Nicholas, Norrish and Vickers, to be published. 3 Callear and Norrish, Nature, 1960, 188, 53. 4 Basco and Norrish, Proc. Roy. SOC. A , 1961,260, 293. 5 McGrath and Norrish, 2. physik. Chem., 1958, 15, 245. 6 Polanyi, J. Chem. Physics, 1959, 31, 1338. 8 Basco and Norrish, Can. J. Chem., 1960,38,1769. 9 Basco and Norrish, Nature, 1961, 189,455. 7 Smith, J. Chem. Physics, 1959,31, 1352. 10 Basco and Norrish, Proc. Roy. SOC. A, in press. 11 Cashion and Polanyi, J. Chem. Physics, 1961,35, 600. 12 Basco and Norrish, unpublished results. 13 Harris and Willard, J. Amer. Chem. SOC., 1954,76,4678. 14 Frey, Proc. Roy. SOC. A, 1959, 250,409. 15 Simons and Yarwood, Trans. Faradar SOC., 1961, 57, 2167. 16 Norrish and Oldershaw, Proc. Roy. SOC. A, 1958,249,498. 17 Herzberg, Proc. Chem. SOC., 1959, 116. 18 McGrath and Norrish, Proc. Roy. SOC. A , 1957, 242,265. 19 Basco, Norrish and Wayne, to be published. 20 Lipscomb, Norrish and Thrush, Proc. Roy. SOC. A, 1956,233,455. 21 Schuler, J. Physic. Chem., 1957, 61, 849. 23 Herzfeld and Litovitz, Absorption and Dispersion of Ultrasonic Waves (Academic Press. New 22 Basco, Callear and Norrish, to be published. York and London, 1959).