Principale
Journal of the American Chemical Society Microscopic mechanism for diffusion and the rates of diffusioncontrolled reactions in simple...
Microscopic mechanism for diffusion and the rates of diffusioncontrolled reactions in simple liquid solvents
Emeis, Cornelis A., Fehder, P. L.Quanto ti piace questo libro?
Qual è la qualità del file?
Scarica il libro per la valutazione della qualità
Qual è la qualità dei file scaricati?
Volume:
92
Lingua:
english
Rivista:
Journal of the American Chemical Society
DOI:
10.1021/ja00711a009
Date:
April, 1970
File:
PDF, 870 KB
I tuoi tag:
 Per favore accedi al tuo account prima

Serve aiuto? Per favore leggi la nostra breve guida come inviare un libro a Kindle
Il file verrà inviato al tuo indirizzo email. Ci vogliono fino a 15 minuti prima di riceverlo.
Il file verrà inviato al tuo account Kindle. Ci vogliono fino a 15 minuti prima di riceverlo.
Nota: devi verificare ogni libro che desideri inviare al tuo Kindle. Controlla la tua email per l'email di verifica da Amazon Kindle.
Nota: devi verificare ogni libro che desideri inviare al tuo Kindle. Controlla la tua email per l'email di verifica da Amazon Kindle.
Booklist collegate
0 comments
Puoi scrivere una recensione del libro e condividere le tue esperienze. Gli altri lettori saranno sempre interessati alla tua opinione sui libri che hai letto. Che tu abbia amato o meno il libro, se dai i tuoi pensieri onesti e dettagliati, le persone troveranno nuovi libri adatti a loro.
1

2

2246 The Raman spectra were recorded photographically on a grating instrument constru2ted in this laboratory. A SpectraPhysics Model 125 HeNe laser was used for the excitation source, and the use of a SpectraPhysics Model 310 polarization rotator made the identification of the polarized lines a straightforward matter. Bv using a Corning filter, No. SPO98565,in front of the slit, it was poisible to obtain Raman spectra above 200 cml from the exciting line, although the maximum transmission of the Corning filter is not reached until 400 cm1. Several Rowland ghosts appear in the region between 200 and 400 cm1 but they are weak and do not interfere. On the other hand, the region below 200 cm1 is heavily populated with ghosts. A special Baird filter permitted one to obtain high transmission from about 40 to 200 cml while removing all the ghosts. However, the constant presence of lines at 63 and 91 cml was observed. Acknowledgment. This wLik was supported by the Advanced Research projects A~~~~~ through the IDL program* We are indebted to Dr* John Ferraro and’ Dr. Clarence Postmus of Argonne National Laboratories for making the highpressure facilities available to us and helping with the measurements, to D,.. Stanley Abramowitz of the National Bureau of a Of 6LioH, to Dr* v* Standards for Thornton of Phillips Petroleum Company for the polyethylene used for the farminfrared cells, and to Mr. Jim Saunders, Mr. Jack Fisher, and Mr. Arthur Curry for experimental help in this laboratory. The Microscopic Mechanism for Diffusion and the Rates of DiffusionControlled Reactions in Simple Liquid Solvents’ C. A. Emeis2 and P. L. Fehder Contribution No. 3945 from the Arthur Amos Noyes Laboratory of Chemical Physics, California Institute of Technology, Pasadena, California 91109. Received September 15, 1969 Abstract: Standard theoretical treatments of chemical reaction kinetics generally neglect any mean interaction potentials or “excluded volume” effects that might interfere with the relative diffusion of a pai; r of reactant mole cules in solution. Analyses of the computergenerated simulation data for a model dense fluid of LennardJones disks have shown that the microscopic mechanism for diffusion in simple liquids is largely “cooperative” in nature, and that shortrange correlations associated with this cooperative mechanism tend to slow the relative diffusion of pairs of molecules approaching to within threefour diameters of each other. In this paper we examine the impact of these results upon the theoretical prediction of diffusioncontrolled reaction rates and the physical interpretation of several other very fast chemical processes in solution. T he use of diffusion models to treat the kinetics of fast reactions in solution was first proposed by Smoluchowski3 and has more recently been reviewed by Noyesa4 Although this approach is widely used and frequently provides satisfactory orderofmagnitude predictions of rate constants, several fundamental difficulties remain. These difficulties appear to stem primarily from a lack of detailed information regarding the microscopic mechanism for diffusion in liquids. In particular, the manner in which this mechanism might affect the relative motions of molecules in a liquid is not well understood. In Smoluchowskitype treatments of chemical reaction kinetics, it is frequently assumed that the relative diffusion of molecules of two reactant species is described by a coefficient that is just the sum of the bulk diffusion coefficients for the two species in solution. This assumption is equivalent to a supposition that no correlation exists between the timedependent relative displacements of two solute molecules and their relative positions and motions at previous times. It is not altogether clear that this supposition is valid for molecules that are separated by only short distances, and (1) This work was supported in part by a grant from the National Science Foundation, No. GP7258. (2) Koninklijke/Shell Laboratorium, Amsterdam, Holland. (3) M. von Smoluchowski, 2.Phys. Chem., 92, 129 (1917). (4) R. M. Noyes in “Progress in Reaction Kinetics,” Vol. 1, G. Porter, Ed., Pergamon Press, New York, N. Y., 1961,p 128. Journal of the American Chemical Society indeed a deviating behaviorthe socalled “solvent cage effect”was long ago suggested by R a b i n ~ w i t c h . ~ The results obtained by Noyes and coworkers from their investigations of iodine atom recombination rates6*’ and the wavelength dependence of the quantum yield for iodine photodissociations10 in solution also suggest that these shortrange correlations may have a measurable effect upon the kinetics of certain very fast chemical processes. In order to obtain additional information regarding the microscopic structure and kinetics characteristic of simple liquids, one of the authors (P. L. F.) has recently completed a series of computer calculations simulating the dynamics of a twodimensional dense fluid of LennardJones disks. Two previous papers have presented the results from some preliminary analyses of the simulation data” and a detailed investigation of the mechanism for diffusion and relative diffusion in the model fluid.12 The purpose of this paper is to examine the ( 5 ) E. Rabinowitch, Trans. Faraday Soc., 33, 1225 (1937); W. c. Wood, i b d , 32, 1381 (1936). (6) H. Rosman and R. M. Noyes, J . Amer. Chem. Soc., 80, 2410 ( 1958). (7) R. M. Noyes, ibid., 86, 4529 (1964). (8) R. M. Noyes, ibid., 77,2042 (1955). (9) L.F. Meadows and R. M. Noyes, ibid., 82, 1872 (1960). (10) R. M. Noyes, 2.Bektrochem., 64, 153 (1960). (11) P. L. Fehder, J . Chem. Phys., SO, 2617 (1969). (12) P..L.Fehder, C.A. Emeis, and R. P. Futrelle, “The Microscopic Mechanism for SelfDiffusion and Relative Diffusion in Simple Liquids,” manuscript in preparation. 92:8 / April 22, 1970 2247 Figure 1. “Snapshot” of an instantaneous configuration of the model fluid in a liquidlike state. The particles are plotted with a diameter 6 ,the distance parameter in the LennardJones pair potential. impact of previous findings upon the treatment of diffusioncontrolled reaction kinetics in solution. Related topics, such as the solvent cage effect and the wavelength dependence of quantum yields for photodissociation in solution will also be discussed briefly. Diffusion in Simple Liquids Examination of graphical displays of the simulation data has led to several intuitively important observations regarding the microscopic character of simple liquids. of the instantaneous conFor example, figuration of the model system (see, e.g., Figure 1) provide evidence that the “excess” volume acquired by a liquid through thermal expansion is localized into relatively large, irregular “holes.” Although a number of theories of the liquid state1* have postulated the existence of holes in the microstructure of real liquids, the phenomenon observed differs from that suggested by the theoretical models in two significant ways: (i) the holes appearing in the model fluid bear no relationship to the size and shape of an individual fluid particle (that is, the holes do not appear as “vacancies” in an otherwise quasicrystalline structure), and (ii) comparison of snapshots for successive times shows that a given hole may persist in the same region of the fluid for secwell in excess of times of the order of 5 X the characteristic kinetic relaxation time (ca. 2.5 X 1013 sec) for the system. A more detailed analysis of the microscopic structure of the model fluid, and the relationship between this structure and the structure of real simple liquids, is presented elsewhere. l 5 (13) A detailed description of the graphical display techniques employed in analysis of the simulation data is provided in ref 11. (14) See, for example, H. Eyring and R. P. Marchi, J . Chem. Educ., 40,562 (1963), and the discussion in J. M. H. Levelt and E. G. D. Cohen In “Studies in Statistical Mechanics,” Vol. 2, J. deBoer and G. E. Uhlenbeck, Ed., NorthHolland Publishing Co., Amsterdam, Holland, 1964, p 178 ff. (15) P.L.Fehder,J. Chem. Phys., 52, 791 (1970). Figure 2. Trajectories of the particles in a liquidlike state of the model fluid. The small circles mark the initial positions of the particles, and the irregular lines extending therefrom the paths of the centers during the remainder of a 2 X IO**sec interval. The initial configuration also corresponds t o that shown in Figure 1. Plots such as those shown in Figure 2 of the trajectories of the particles in the model fluid provide some insight into the microscopic mechanism for diffusion in simple liquids. As can be seen in the figure, extensive diffusive migration isover a surprisingly long time intervallargely restricted to local groups of particles in the region of a hole. Furthermore, motion pictures created from the simulation data show that the local groups of long trajectories arise from a concerted migration of the particles involved, rather than from successive “jumps” or knockon collisions. Diffusion in the model fluid therefore proceeds by a mechanism that is largely cooperative in nature, and it is reasonable to assume that similar cooperative phenomena occur in real liquids. Although it is unlikely that cooperative phenomena of the sort observed in the model fluid would have a macroscopically discernible effect upon singlet l6 diffusion in real liquids, the shortrange correlations associated with these phenomena become more important when the relatiue diffusion of molecules in solution is examined. It is convenient to describe relative diffusion in terms of the motion of one molecule in a coordinate system fixed to the center of the other; in two dimensions, the relative diffusion tensor D R then has the form where D,,and DBBare the coefficients for radial and tangential diffusion, respectively, and the offdiagonal elements of the tensor vanish by symmetry. Consider a solution of two solute species, X and Y , in a solvent S. It is easily shown1* that, if the diffusive motions of the X and Y molecules are completely (16) We use the word “singlet” here to distinguish between the migration of individual molecules in solution and the relafiue diffusion of pairs of molecules. Emeis, Fehder 1 DiffusionControlled Reactions in SimpIe Lquid Solvents 2248 I 3 I I L I L r ( o ; d , ~ I Figure 3. Comparison of the mean potential $ ( r ) calculated for the liquidlike state of the model fluid shown in Figures 1 and 2 and the LennardJones pair potential used in the simulation calculations. uncorrelated, the coefficients describing the diffusion of X molecules relative to Y molecules (or vice versa) are just equal to the sum of the bulk diffusion coefficients for X and Y in S . In statistical terms,I7 this means that the average square of the timedependent displacements of an X molecule relative to a Y molecule (or vice oersa) is just the sum of the timedependent mean square displacements of the X and Y molecules taken separately. And by analogy, if no correlations exist between the motions of molecules in a onecomponent fluid, the coefficients D,,and Do8describing the relative diffusion of pairs of the molecules would both be equal to just twice the coefficient D,for selfdiffusion. Particles separated by large distances in the model fluid diffuse independently. But analysis of the simulation data has also shown1* that shortrange “cooperative” correlations slow the relative motions of molecules approaching to within about 34 diameters of each other. A similar phenomenon had previously been suggested by no ye^;^ in terms of chemical reaction kinetics in solution, this result implies that the standard Smoluchowskitype treatments may overestimate the frequency of reactantpair encountersand thus, overestimate the rates of socalled “diffusioncontrolled’’ reactions. Conversely, the computer results also indicate that encounter pairs will remain in close proximity appreciably longer than predicted by an “independent” diffusion model. In the case of reactions having a nonnegligible activation energy or those requiring a specific steric configuration of the reactant molecules, the depressed rate of reactant encounter may therefore be offset by an increased probability of reaction upon encounter. T o a good approximation, relative diffusion in solution may be treated in terms of the mean potential $(r) acting between two molecules when interactions with the surrounding solvent molecules are taken into account. A convenient form for the mean potential in a onecomponent liquid is obtained from the familiar radial distribution function g(r)18 (17) See, for example, R. Zwanzig, Annu. Rea. Phys. Chem., 16, 67 (1965). (18) P. A. Egelstaff, “An Introduction to the Liquid State,” Academic Press, New York, N. Y . , 1967, p 16. i 3 4 ~ J Figure 4. The empirically determined radial relative diffusion coefficient Drr(r)for the liquidlike state of the model fluid shown in Figures 1 and 2. The dashed line represents the usual assumption that D,, = 2 0 , . where kB is the Boltzmann constant and T the temperature. The radial distribution function for the model fluid has been discussed in detail elsewhere.I5 I n Figure 3, the mean potential obtained via eq 1 for a liquidlike state of the model fluid is shown in comparison with the LennardJones pair potential. used in the simulation calculations. Unlike the simple pair potential, $(r) exhibits a number of subsidiary maxima and minima corresponding to the first, second, . . . , etc., “shells” of neighbors surrounding a molecule in a liquid. Thus, in the mean potential model for relative diffusion, a molecule diffusing toward another molecule in solution must cross successively higher potential “barriers” before the two molecules come into direct contact. Numerical solutions of the twodimensional diffusion equation including +(r) have shown12however that the mean potential is not in itself sufficient to account for the relative diffusion phenomena observed in the model fluid; to obtain agreement with the simulation data it was also necessary to lower the value of D,,for pairs of particles separated by short distances. Although the precise physical meaning of this empirical variation in the relative diffusion coefficient is not entirely clear, we believe that it reflects the inability of a timeaveraged function like $(r) to account completely for the role played by transient geometric or “excluded volume” effects in the microscopic mechanism for diffusion at liquidlike densities. The twodimensional diffusion equation including both $(r) and a relative diffusion coefficient &(r) of the form shown in Figure 4 was found to reproduce quite accurately the relative diffusion phenomena observed in the model fluid. Reaction Kinetics in Solution In this section we obtain expressions describing the rates of socalled diffusioncontrolled reactions in twoand threedimensional solutions. The derivation follows closely that presented by Noyes,4 but is extended to take into account both the mean force and the functional &(r) discussed above. The physical reasoning Journal of the American Chemical Society / 92:8 / April 22, 1970 2249 in support of this treatment has been discussed in detail e l ~ e w h e r e ’ ~ ~and “ * ~will not be reproduced here. ThreeDimensional Solution. Consider again the solution of two solute species, X and Y, in solvent S. Let us assume that the X and Y molecules exert no longrange forces on each other, and that initially the molecules of each species are distributed randomly throughout S in the way they would be if the other species were not present. Furthermore, let us assume that at some zero time we can “turn on” a diffusionY P products in the solution. controlled reaction X We wish then to calculate the rate of the reaction at subsequent times. Very soon after the reaction is initiated, most of the X molecules that were near Y molecules at t = 0 will have reacted so that the concentration of Y molecules near a stillunreacted X will, on the average, be somewhat lower than the remaining bulk concentration of Y in the solution. This situation is then analogous to the existence of a concentration gradient in Y around the remaining X molecules. In most systems of chemical interest, a steadystate condition is quickly achieved such that the next flux @ of Y molecules toward X molecules along this gradient is the same at all distances away from the centers of the X molecules andis just sufficient to provide for the rate at which the X molecules react. If c(r) is the average concentration of Y at a distance r from the center of an X molecule, the flux of Y molecules through a sphere of radius r about an X is given by To a good approximation, [Y] in eq 7 may be equated with c( a). Comparison of eq 4 and 7 then shows that Substituting this result into eq 5 with r arranging, we obtain the final expression = p (9) + @ = 4ar2Drr(r)[’%) which differs from the expression obtained by Noyes4 in that the quantity (pD)*, defined in eq 6 at r = p, is calculated with reference to the mean potential U(r) and an rdependent coefficient Drr(r). For reactions of the type, X X + products, the righthand sides of eq 4 and 7 must be multiplied by a factor of 2, leading to + TwoDimensional Solution. If we attempt to carry out a similar derivation for the rate of reaction in a twodimensional solution, we quickly come upon a striking difference between the situations in two and three dimensions. For the sake of simplicity, let us first assume that D,, is independent of r, and that any interaction U(r) between X and Y molecules can be ignored. Then in two dimensions the steadystate condition is represented by kBT dr kc(p) where D,,(r) is the radial coefficient for the relative diffusion of X and Y molecules, and U(r) is the potential of the mean force acting on XY pairs in the solution. But in steady state, this net flux must be balanced by the rate at which Y molecules are depleted from solution by reaction kc(p) exp(U(p)/kBT) (4) where p is the XY distance at which reaction can occur, and k is the rate constant that would be observed were an equilibrium distribution of solute molecules maintained in the system. Combining eq 3 and 4 and solving for the steadystate concentration yields @ = c(r) = and re exp(  U(Y)/kgT)[C( m )  where the quantity (rD)* is given by The microscopic distribution c(r) is not accessible to direct experimental measurement. Instead, kinetics data are used to determine the macroscopic secondorder rate constant k’ based on the bulk concentration [Y] of Y @ = k’[Y] (7) (19) P. Debye, Trans. Electrochem. Soc., 82, 265 (1942). (20) F. C. Collins and G . E. Kimball, J . Colloid Sci., 4, 425 (1949). (21) F. C. Collins, ibid., 5,499 (1950). = 27rrDr,W r ) br Equation 11 is identical with the expression obtained for three dimensions by equating the righthand sides of eq 3 and 4, except that the factor 2 ~ for r the circumference of a circle appears in place of 4rr2, the surface area of a sphere. Integration of eq 11 yields the result 41) = + (k/2TDrr) In (rlp)l c(P)[~ (12) which indicates that c( a ) must be infinite if a steadystate condition is to be maintained. We conclude that diffusion in two dimensions does not provide a sufficient supply of inflowing Y molecules to sustain a steadystate concentration gradient. (The same conclusion obtains if the mean potential U(r) and variations in D,,(r) can be neglected for XY distances greater than some value R. Equations 11 and 12 are then valid for r > R,requiring that c( m ) be infinite for a steadystate condition to be achieved.) The concentration c(p)and hence the observed rate of reaction k’must therefore decrease monotonically with increasing time until reaction is complete. In contrast to eq 8, the timedependent macroscopic rate factor k’(t) for a diffusioncontrolled reaction in two dimensions is given by (13) for times t sufficiently short that the bulk concentration of Y (here approximated by c( a)) does not change appreciably. Solutions to eq 13 can then be obtained Emeis, Fehder / DiffusionControlled Reactions in Simple Liquid Solvents 2250 fixed value of D,, = 8.78 X 101Du2sec = 2 0 , for U(r > p ) = 0 (curve D), U(r) = $(r) (curve B), and for U(r) equal to the LennardJones pair potential cpLJ(r)(curve A). Curve C was obtained with U(r) = $(r) and D,,(r) as shown in Figure 4. Curve C therefore represents the most accurate evaluation of k’(t) for a diffusioncontrolled reaction in the model fluid. During the first stage of the reaction, the rate is primarily determined by the equilibrium ( t < 0) concentration of closely associated XY pairs provided by U(r) according to eq 15. But after a brief induction time these initial pairs are depleted from solution by reaction, and the rate thereafter is determined by the rate at which new encounter pairs are formed through diffusion. The LennardJones potential, which provides both the highest initial concentration of XY c pairs and the least resistance to relative diffusion, CL I yields the highest reaction rate (curve A) over the entire C 5 :1 I5 20 2 X 10“ sec interval spanned by Figure 5. Comt x 0“wc) parison of curves A and B shows the effect upon the Figure 5 . Plot of the timedependent macroscopic rate constants reaction rate of the barriers to relative diffusion profor a diffusioncontrolled reaction in two dimensions. The mathevided by the mean potential +(r), while curve C indimatical assumptions leading to each of the four curves are identified cates the additional lowering of the reaction rate that is in thetext. obtained when the “adjusted” coefficient Drr(r) is by numerical integration of the system of equations included in the calculations. ThreeDimensional Solution. Liquid argon at a temperature 108.18”K and density 1.261 g ~ m was  ~ used as a model for the reaction system in the threedimensional rate calculations. The radial distribution =  kc(p,t) exp[ U(p)/k,T] function g(r) for this state has been measured by bt Smelser,22and a tabulation of the function was kindly from the initial condition provided by that author. The LennacdJones potential parameters for argon are u = 3.405 A, e/kB = 119.80 c(r,O) = c( m ) exp[  U(Y)/kgT] (1 5 ) and the selfdiffusion coefficient for argon in this where (14a) is the diffusion equation in two dimensions D , = 4.37 X cm2 ~ e c  ’ . ~From ~ these state is and (14b) accounts for the depletion of Y molecules parameters a value k = 3.70 X 10” 1. mol’ sec’ is due to reaction. obtained for the equilibrium constant. The predicted values for k’ obtained from eq 6 and Results 9 for several combinations of U(r) and Drr(r)are listed Rate factors k’ for a diffusioncontrolled reaction in Table I. Standard theoretical treatments of chemX Y * products in two and threedimensional ical reaction kinetics in solution generally neglect any solutions were computed for a variety of combinations mean pair potentials or excluded volume effects that of D,,(r) and U(r). It was assumed that the species X might interfere with the relative diffusion of the reactant and Y distinguish themselves from the solvent only in molecules. To determine the effect of including a their ability to react with each other; otherwise, the reactantpair potential in the calculations, k’ was compotential U(r) and the relative diffusion coefficient puted for U(r > p) = 0, U(r) = +(r), and U(r) equal to D,, for an XY pair were assumed to be the same as for the LennardJones potential with D,, constant and equal a pair of solvent molecules. The “equilibrium” rate to 2 0 , for all XY distances. Comparison of the values constants k were calculated from two and threeobtained for U ( r ) = 0 and U ( r ) = $(r) shows that the dimensional kinetic gas theory with the assumption predicted rate of reaction in three dimensions is not that every collision would result in reaction. The changed appreciably when a quasirealistic interaction distance p at which reaction can occur was taken equal $ ( r ) is incorporated into the theory. like to the u parameter in the LennardJones pair potential. Although the magnitude of shortrange cooperative TwoDimensional Solution. The rate calculations correlations in the relative diffusion of molecules in for two dimensions were based on the liquidlike state is not known and is not presently accessible real liquids of the model fluid shown in Figures 1 and 2 and exto direct experimental measurement, some estimate of amined in detail in ref 12. The value of the selfdiffuthe effect these correlations would have upon the sion coefficient for this state is D , = 4.39 X 1010a2 is obtained by kinetics of diffusioncontrolled reactions sec’; the potential of the mean force $(r) is shown in calculating k’ under the assumption that Dr,(r) for Figure 3, and the empirically ,determined relative (22) S. Smelser, Ph.D. Thesis, California Institute of Technology, diffusion coefficient is that shown in Figure 4. Pasadena, Calif., 1969. To be made available through University The timedependent behavior of k ’ ( t ) for four Microfilms. (23) J. 0. Hirschfelder, C. F. Curtiss, and R. B. Bird, “Molecular different Drr(r)U(r) combinations is shown in Figure Theory of Gases and Liquids,” John Wiley & Sons, Inc., New York, 5 . In order to determine the effect of the interaction N. Y . , 1954, p 165. U ( r ) on the rate of reaction, k ’ ( t ) was calculated with a (24) J. Naghizadeh and S. A. Rice, J . Chem. Phys., 36,2710 (1962). + Journal of the American Chemical Society J 92:8 J April 22, 1970 2251 Table I. Calculated SteadyState Rate Constants for a DiffusionControlled Reaction in Liquid Argon (PO)*, U(r) DrXr) 0 +(r) LennardJones 0 20s 20s 20. As shown in Figure 4 *(r) LennardJones crna sec1 2.12 x 2.23 x 2.67 X 1.50 X 1.60 X 1.96 x 1012 1OI2 10l2 10l2 10l2 10l2 k’, 1. mol’ sec1 1.00 x 1.03 x 1.14 X 0.79 X 0.83 X 0.95 x 10’0 10’0 1Olo 1Olo 1O1O 1Olo the threedimensional solution varies as the ratio Drr(r)/2D, observed in the twodimensional model fluid, The final three entries in Table I show the effect of U(r) when the rate constant is calculated with a coefficient Drr(r) that decreases for small reactantpair separations as shown in Figure 4. For each of the three assumed forms for U(r), the predicted value for k‘ is lowered about 20% when the rdependent coefficient is included in the calculations. The small change in k’ that is obtained when $(r) is included in the calculations probably reflects the partial cancellation of two opposing effects. As in two dimensions, the barriers to relative diffusion presented by an oscillatory potential like $(r) would tend to decrease the rate of reaction. Yet any potential having an attractive component extending beyond r = p would tend to increase the minimum distance within which a pair of molecules would have to approach each other before reaction becomes probable, and hence would tend to increase the reaction rate. This latter effect is illustrated by the relatively large increase in k‘ that is obtained when U(r) is set equal to the LennardJones pair potential and either form of Drr(r) is assumed. A similar effect is also observed in two dimensions, as may be seen by comparing curves A and D in Figure 5 . The substantial decrease in the predicted value for k‘ that is obtained when an rdependent relative diffusion coefficient is incorporated into the calculations maywithin the framework of the Smoluchowski model for diffusioncontrolled reactions in solutionbe attributed to the fact that a functional form for Drr(r) like that shown in Figure 4 tends to slow the relative diffusion of a pair of reactant molecules just in the region where the gradient in c(r) is greatest. A more thorough analysis of this phenomenon would require an investigation of the steadystate concentration distributions that are established when various combinations of U(r) and Drr(r) are assumed. Discussion In light of the results presented here and in two previous papers, l 1 v 1 * we may draw several conclusions regarding the mechanical influence of the solvent upon the microscopic kinetics of simple chemical reactions in solution. Data obtained from the computer simulation of a model dense fluid of LennardJones disks has shown that diffusion in simple liquids may proceed by a mechanism that is, at the molecular level, largely cooperative in nature; and further, that this cooperative mechanism tends to retard the relative diffusion of molecules separated by short distances in the liquid. For solutions in which the solute and solvent molecules are physically similar, the average force acting between a pair of solute molecules may be approximated by the mean potential $(r) obtained from the experimentally accessible25 radial distribution function g(r) for the solvent. Our calculations have shown however that this timeaveraged mean potential does not provide a complete description of the transient “excluded volume” effects that apparently play an important role in relative diffusion phenomena at liquidlike densities. A more accurate description of relative diffusion in the twodimensional model fluid was obtained from a theoretical treatment that included both +(r) and rdependent relative diffusion coefficients. Unfortunately it is difficult to estimate, on the basis of the twodimensional simulation data alone, the relative importance of shortrange cooperative correlations in the mechanism for diffusion in real, threedimensional liquids. The presence of an additional degree of freedom would be expected to decrease the dynamic importance of excluded volume effects; yet the relative diffusion of two molecules in a threedimensional liquid must involve interactions with a much larger number of neighboring solvent molecules. Although direct experimental observation of relative diffusion phenomena in real liquids is not at present possible, analyses similar to those described in ref 12 of existing 27 simulation data for threedimensional model should yield some insight into the problem. The results presented in this paper indicate that the decrease in the coefficient describing the relative diffusion of reactantpairs separated by short distances is an important factor in determining the steadystate rate for a diffusioncontrolled reaction in real systems. The mean potential +(r) can nonetheless serve as a convenient intuitive device for interpreting a number of chemically important processes occurring in solution. In our investigation of diffusion in the simulated fluid,12 we observed that pairs of particles diffusing away from each other tended to become “trapped” momentarily in first, second, and third nearestneighbor positions. This phenomenon is reminiscent of the socalled “solvent cage effect,” and can to a first approximation be ascribed to the successive potential barriers to relative diffusion presented by $(r). Photodissociation of a molecular solute is another process that “samples” the microscopic structure and dynamics of the solvent in local regions of a solution. Noyes and Meadows9 have investigated the wavelength dependence of the quantum yield for photodissociation of molecular iodine in a number of nonreactive solvents and determined lo that the experimental results are not reproduced by a theoretical model that neglects the microscopic structure of the soluent surrounding the reaction site. For longer wavelengthssuch that the excess energy over that required to break the iodineiodine bond is smallthe simple “solvent continuum” theory is found to predict quantum yields larger than those observed experimentally; yet for progressively shorter wavelengths the observed quantum yield is found to increase more rapidly than predicted by the theory. (25) See, for example, ref 18 or H. H. Paalman and C . J. Pings, Reo. Mod. Phys., 35, 389 (1963). (26) A. Rahman, Phys. Reo., 136, A405 (1964); J . Chem. Phys., 45, 2585 (1966). (27) L. Verlet, Phys. Rev., 159, 98 (1967). Emeis, Fehder / DiffusionControlled Reactions in Simple Liquid Soluents 2252 On the basis of the model for relative diffusion phenomena presented in this paper, the experimental quantum yield data can be interpreted in terms of transient processes dependent upon the iodinesolvent interaction described by $(r) and longer lived processes dependent upon Dm(r). Immediately after dissociation, the separating iodine atoms encounter the barrier in $(r) between first and second nearestneighbor positions. If the excess energy provided by the exciting photon is small, the atoms are reflected from the barrier and recombine quickly; but if the excess energy is sufficient to permit the separating atoms to reach second nearestneighbor positions, the barrier will tend to keep them apart and thus prevent recombination. Furthermore, if Drr(r) increases with increasing r, pairs of atoms that initially achieve a large separation will diffuse away from each other more quickly, and thus be even less likely to recombine. In reality of course, the successive maxima in # ( I ) and the rdependence of &(r) are the result of interactions between the solute iodine atoms and surrounding solvent molecules, and are truly descriptive only of an equilibrium situation. Monchick28has however presented a theoretical treatment of photodissociation processes that includes an “effective” potential much like # ( I ) , although no theory incorporating both $(r) and an rdependent relative diffusion coefficient has previously been treated. Acknowledgments. We wish to thank Professor G. Wilse Robinson for his helpful discussions and his critical reading of this paper. One of the authors (C. A. E.) gratefully acknowledges the support of The Netherlands Organization for the Advancement of Pure Research (ZWO). (28) L. Monchick, J . Chem. Phys., 24, 381 (1956). Investigation of Singlet + Triplet and Singlet Singlet Transitions by Phosphorescence Excitation Spectroscopy. VIII. Santonins f Grace Marsh,” David R. Kearns,lb and Michael Fisch” Contribution f r o m the Department of Chemistry, University of California, Riverside, California 92502, and the Department of Chemistry, University of California, Irvine, California 92664. Received September 1.5, 1969 Abstract: The singlet + singlet and singlet + triplet transitions in asantonin (I), 6episantonin (TI), and 2bromosantonin (111) have been investigated at both 77 and 4.2”K. The results may be summarized as follows. (i) Polarization measurements demonstrate that vibronic coupling between the S(n,n) and S(n,n)states is the principal  source of intensity for the So S(n,n) transition. The magnitude of the electronically allowed contribution to the intensity appears to depend upon the molecular structure. (ii) The lowest excited triplet state in each case is identified as a 3(n,a) state. (iii) The 3(n,a) state has been observed in each compound just above (12001600 cm*) the 3 ( a , ~ state. ) The intensity in the So + T(n,a) transition of asantonin (emax 0.3) is attributed to strong spinorbit coupling with the l(n,n)state. (iv) Diffuseness observed in the So+ T(n,n) absorption spectra at 4.2”K is attributed to vibronic interaction between the a(n,a)state and nearly degenerate vibronic levels of the lower lying 3(a,a)state. These spectroscopic observations appear to have an important bearing on the interpretation of the asantonin photochemistry and the possible role of the 3(n,n) and 3(s,a)states in the excited state transformations. ince they were first identified in 1944,2 interest in triplet state molecules has grown, and this growth has been particularly rapid during the last few years. Of the many types of molecules which have been studied considerable attention has been given to organic ketones and aldehydes. 310 Surprisingly enough most S (1) (a) Riverside; (b) to whom correspondence should be addressed at Riverside; (c) Irvine. (2) (a) G . Lewis and M. Kasha, J . Amer. Chem. Soc., 66, 2100 (1944); (b) A. Terenin, Acta Physicochim. U R S S , 18, 210 (1943); Zh. Fiz. Khim., 18, l(1944). (3) R. Shimada and L. Goodman, J . Chem. Phys., 43, 2027 (1965). (4) J. M . Hollas, E. Gegorek, and L. Goodman, ibid., 49, 1745 (1968). ( 5 ) Y. Karida, H. Kasada, and T. Matamura, Spectrochim. Acta, 20, 1387 (1964.). (6) D. R. Kearns and W. A. Case, J . Amer. Chem. Soc., 88, 5087 ( 1966). (7) S . Dym, R. M. Hochstrasser, and M. Schafer J . Chem. Phys., 48, 646 (1968). ( 8 ) E. Eastwood and C. P. Snow, Proc. Roy. Soc., Ser. A , 149, 434 ( 1935 ) . (9) J. M. Hollas, Spectrochim. Acta, 19, 1425 (1963). Journal of the American Chemical Society 92:8 / April 22, 1970 of these studies have been concerned with aromatic carbonyl compounds and there have been relatively few studies of “simple” carbonylcontaining molecules such as enones and dienones.810 Because of this, and because of the current interest in their photochemical properties, we have started a comprehensive investigation of the spectroscopic properties of a wide variety of enones and dienones. Some of our earlier studies of enones have already been published,”! l 2 and more detailed results will be forthcoming.13 In the present paper we discuss results obtained with the following three crossconjugated dienones: asantonin (I), 6epi(10) G. Herzberg, “Electronic Spectra and Structure of Polyatomic Molecules,” D. Van Nostrand Co., Inc., Toronto, 1966. (11) D. R. Kearns, G. Marsh, and K. Schaffner, J . Chem. Phys., 49, 3316 (1968). ( 1 2 ) G . Marsh, D. R . Kearns, and K. Schaffner, Hela. Chim. Acta, 51, 1890 (1968). (13) G. Marsh, Ph.D. Thesis, University of California, Riverside, Cal., 1969.