J. Phys. Chem. 1992,96,9561-9564 Recently it has been reported that a proton-transfer reaction in ethanol is supprewd at temperature8 below 77 K.16 Thus, it may be that H30+ ions are not formed by reaction 15 at very low temperatures but probably by reaction 13. The two alternative mechanisms for the formation of excited hydroxyl radicals by 0ions or H atoms arc only tentative. To determine the mechanism, it would be necessary to study the behavior of trapped species produced by radiolysis of ice at low temperatures. In conclusion, the new strong emission from high-energyelectron-irradiated ice has been observed for the first time at ultralow temperatures. The emission is probably from excited hydroxyl radicals. The result may stimulate further studies on excited species produced by irradiation of ice.
Acknowledgment. This work was supported in part by a Grant-in-Aid for Scientific Research from the Japanese Ministry of Education, Science and Culture.
References and Notes
-
(1) Quickenden, T. I.; Trotman, S.M.; Sangster, D. F.1.Chem. Phys. 1982, 77, 3790. Related papers on emissions from ice are cited therein. (2) Merkel, P. B.; Hamill, W. H. J. Chem. Phys. 1971,54, 1695.
9561
(3) Buxton, G. V.; GiUii, H. A.; K l w n , N. V. Cun. J. Chem. 1977.55, 2385. (4) Trotman, S. M.;Quickenden, T. I.; Sangstcr, D. F. J. Chem. Phys. 1986,85,2555. ( 5 ) Vernon, C. F.;Matich, A. J.; Quickenden, T. I.; Sangster, D. F. J. Phys. Chem. 1991,95,7313. (6) The information on one-shot single-photon counting b available from Hamamatsu Photonics Co., Electron Tube Division (314-5, Shimokanzo, Toyooka-village, Iwata-gun, Shizuoka-ken, 438-01, Japan). (7) Kawahta, K.; Nagata, Y.; Okabe, S.; I(imura, N.; Tsumori, K.; Kawanishi, M.;Buxton, G. V.; Salmon,G. A. J. Chem. Phys. 1982,77,3884. (8) Quickenden, T. I.; Matich, A. J.; Bakker, M. G.; Freeman, C. G.; Sangster, D. F.J. Chem. Phys. 1991, 95, 8843. (9) Johnson, J. E.;Moulton, G. C. J. Chem. Phys. 1978, 69, 3108. (10) Maria, H. J.; McGlynn, S.P. J. Chem. Phys. 1970, 52, 3402. (1 1) Prince, R. H.; Sears,G. N.; Morgan, F. J. J . Chem. Phys. 1976,61, 3918. (12) Siegel, S.;Baum, L. H.; Skolnik, S.;Flournoy, J. M.J. Chem. Phys. 1960.32, 1249. (13) Wu, Z.; Gillis, H. A.; Klasscn, N. V.; Tcather, G. G. J. Chem. Phys. 1903, 78, 2449. (14) S ~ I O I M M., C. R. J . Chem. Soc., Chem. Commun. 1980,675. (15) Flournoy, J. M.; Baum, L. H.; Siegcl, S. J. Chem. Phys. 1962, 36, ***A
LLL7.
(16) Miyazaki, T.; Shih, T.; Fueki, K.; Kamiya, Y. J. Phys. Chem. 1991, 95, 9115.
Femtosecond Solvation Dynamics following Electron Attachment at the Water-CH,CI Interface. A Molecular Dynamics Study Daniel A. Rose and Ilan Benjamin* Department of Chemistry. University of California, Santa Cruz, California 95064 (Received: July 20, 1992)
We use molecular dynamics Calculations to teat a recent suggestion that the femtaseoond solvation dynamics following dissociation at the condensed phascvacuum interface can be probed by a measurement of the kinetic energy of the dissociation product ejected from the surface. We find that the energy released upon solvation is deposited in bath modes and is not camed out by the dissociation product.
I. Intmdu&'011 Over the last decade, developments of picosecond and femtod timemlved spectn>eoopl'c methods' have made it possible to probe the return to equilibrium of the solvent polarization following a sudden change in the solute charge distribution. The combination of experimental techniques,13 analytical theories," and simulations"' has provided a molecularly detailed picture of this important process. The basic motivation for these studies has been to understand solvent effects on ultrafast reactions in solution. As ultrafast measurements of charge-transfer dynamics and other reactions are beginning to be carried out at fluid interface~,'f'~an extension of the study of solvation dynamics to these systems is desirable. This is important not only for elucidating the role of the solvent in fast interfacial reactions but also for understanding solvation dynamics in general, since the unique electrical properties of interface? may present more stringent tests of solvationdynamics models. The detection of solvation dynamics at interfa- presents a formidable technical problem. As a result, studies of solvation dynamics at interfaces have been limited to very few theoretical phenomenological and simulations. 1 6 ~ 1 In a recent interesting experiment, Cowin and his -workers have prarented'" an intriguing new approach to studying solvation dynamics at interfaces. In their experiment, a layer of methyl chloride (CH3Cl)adsorbed on one layer or several layers of water (as well as other materials), which is supported by a nickel surface, is irradiated with a laser light. The photon is ejecting an electron from the metal surface, which is subecqwntly attached to a CH3Cl
'
molecule, producing CH3Cl-. The CH3Cl- promptly dissociates to Cl- and CHI. The CH3 is detected using %nwf-flight'' mass spectroscopy. The C1- is never observed and assumed solvated at the interface. The most striking result of the experiment is that the kinetic energy of the CH, product ( 4 . 5 ev) is much grentcr than the energy observed when the electron-attachment expsrimCnt is carried out in the gas phase. This large effect has been attributed to a shift in the CH3Cl- repulsive surface due to the solvation of the Cl- ion. The prompt femtosecond solvation of the Cl- is acoompanied by a loss of about 3 eV, and it is postulated that some part of this energy is transferred to the CH3. If this interpretation is indeed correct, then one can study solvation dynamics on the femtosecond time scale using a technique that potentially can provide a level of control and detail similar to that established in gas-phase chemical physics. The purpose of this paper is to test this interpretation using classical molecular dynamics trajectories. We limit out8chrcs here to the issue of whether the CH3 can carry some or most of the energy released by the prompt solvation of the C1- product. Previous studies of solvation dynamics of CT at water interfaces1417 (as well as similar calculations in bulk waterlo) indeed dcmonstratal that a large part of the few electronvolts of solvation energy is released on the same time scale as a typical prompt dissociation proceaP ( l a than 100 fs). However, it is also necessary to have a coupling mechanism which will enable the ejecting CH3 to carry some of the solvation energy. The calculations described below clearly show that a classical solvation model can not explain the observed excess energy of the CH3 radicals. The energy released upon the solvation of the Cl-
0022-365419212096-9561S03.00/0 @ 1992 American Chemical Society
9562 The Journal of Physical Chemistry, Vol. 96, No. 23, 1992 TABLE I: P a r s "
Rose and Benjamin
for the Potential Eaergy Frmeths
Intramolecular Potential for CH&I (Equation '0s 1.776 382 A
1)
0.929 968 A-I 234.525 kcal/mol
4
Intramolecular Potential for CH3CI- (Equation V. 0.692 kcal/mol A2 3.459 kcal/mol 4.6
Q2
2)
A-I
Lennard-Jones Parameters for Nonbonded Interactions (Equations 3 and 4) atom 0, A C, kcal/mol e Q9
CH3 CI
c10 H
3.98 3.16 3.93 3.17 2.81
0.114
0.5 0.832 0.155 0.017
0.227 -0.227
Pt
-1 -0.82 0.41
is almost entirely deposited into bath modes. Although the parameters of the CH,CI--bath interactions can be adjusted to increase the energy of the ejected CH,, this excess energy is still smaller than what is observed. In addition, the source of this excess CH, energy is the size difference between the Cl- and the C1 atom of CH3CI and not directly related to solvation. Our calculations therefore suggest that one needs to look for other explanations of the excess energy of the CH3 fragment.
II. Potentids and Methods The system we study includes three layers of 196 water mol-
ecules adsorbed on the (100) face of a Pt surface (assembled using 216 Pt atoms) and a monolayer of 30 CH3Cl molecules adsorbed at the water-vacuum interface. The surfaces are perpendicular to the Z axis, and the system cross section is 23.54 X 23.54 A. Although the experiment of Cowin and co-workers was carried out using nickel as the electron source,l*the only role of the metal surface in our calculations is to support the water layers. We choose Pt because we have already used the water-Pt surface to study solvation, and there are accurate water-Pt potentials available. In addition, the structure of the water adsorbed on the Pt is quite similar to its structure on nickel; the water wets both surfaces, creating an ice-like stru~ture.'~ The water is modeled using a flexible SPC (simple point charge) modelZoincluding the intramolecular potential of Kuchitsu and Morino.Z1 The Pt-Pt interactions are described using a nearest-neighbor harmonic potentiaLZ2For Pt-H20, we use the potential developed by Spohr and Heinzing~r.~' We use the Morse potential for the intramolecular CH3Cl vibration
(parameters given in Table I), and a simple exponential repulsive potential for CH3CI-:
where r is the intramolecular distance and rq is the equilibrium distance of the ground-state CH3CI. The parameters V, and A2 are uniquely determined from the electron affinity of C1- (3.613 eV)24and of CH3CI (0.18 eV)25and the dissociation energy of CH,CI (3.637 eV).ZSThe parameter a2,which determines the steepness of the repulsive curve, is not known, but we estimate it from the observed energy bandwidth (0.08 eV) for the electron-attachment cross section in the gas phase:' using a simple Franck4hndon procedure. The CH3Cl-CH3Cland CH$I-H20 interactions are modeled using Lennard-Jones + Coulomb potentials
where rj, is the distance between atom i (electric charge Qj)and
0.00
0.16
0.08
J
0.24
PH,0&3)
Flgm 1. Atomic density profiles of the CH3CIand H 2 0 adsorbed layers. The solid lines are for the oxygen and chlorine, and the dotted lines are for the hydrogen and the methyl group.
atom j (charge Qj) on different molecules. The parameters ei, and ujj are determined from the combination rules 'jj
= ('j'j)'/2
uij = !/*(u,
+ u,)
(4)
The parameters ti and aiare given in Table I (CH3 is described as one united atom). The partial charges on CH3Cl are chosen to reproduce the CH3CI dipole moment. The interaction of CH, and C1- with CH3Cl and with H 2 0 is also described using Lennard-Jones Coulomb potentials. The parameters that enter all the potentials described in this paragraph are listed in Table I. The system is prepared by equilibrating six layers of water molecules adsorbed on the metal. Some of the water molecules evaporate during the equilibration, and some are subsequently removed,leaving approximately three layers of water. The CH$l layer is formed by depositing one molecule at a time at a temperature of 150 K. Excess CH3CI is then removed, leaving only one monolayer. Finally, the system is equilibrated at 50 K for 250 ps. A schematic representation of the system, together with the density profiles of the different atoms in the system, is shown in Figure 1. To study the electron-attachment process, we select one of the CH3Cl molecules, transfer it to the repulsive potential curve of CH3CI-, and switch the charge on the CH3 to zero and on the C1 to -1. The trajectory is followed for 1 ps. This is repeated with new initial momenta and for each of the 30 CH3Clmolecules. Thus, we are able to obtain a high level of accuracy by averaging over all these trajectories. To better understand the role of the initial orientation of the dissociating CH3Cl molecule, the trajectories that correspond to all the CH3Cl molecules in a given orientation are separately averaged, as explained below. The integration of the trajectories is done at constant volume and energy, using the velocity version of the Verlet algorithm with an integration time step of 0.5 fs. The standard deviation in total energy is about 0.0014%.
+
III. Results and Mseassion An analysis of the structure of the CH3Cllayer shows that the distribution of initial CH3Cl orientations is quite broad due to the disordered water structure at the interface. Since it is expected that the c ~ l g g ytransfer to the CH, radical will depend on its initial orientation, we perform the averaging over trajectorieswith similar initial orientations of the dissociating CH3Cl molecule. Specifically, if B is the angle between the interface normal and the Cl-CH3 vector (so B 0 corresponds to the CH, radical pointed toward the vacuum), then we consider a given orientation as 'up" if cus B > 'I3,as 'parallel" if -I/, < cos B < I/,, and as 'down" if cos B < J/,, Figure 2 shows that, as expected, the C1- product is promptly solvated, releasing a large amount of energy. The Cl- is mostly stabilized by the water, although quite a significant percentage
-
The Journal of Physical Chemistry, Vol. 96, No. 23, 1992 9563
Femtosecond Solvation Dynamics
(kcalhol)
-60
H20 - CH3Cl - 10
20
0 0.0
0.25
0.5
0.75
0.25
0.0
1.0
t (PS)
0.50 0.75
1.0
t (PS)
Figure 2. Average time-dependent change in the electrostaticsolvation
energy of Cl- following electron attachmentto a CH3C1at the vacuumwater/CH$l interface. The average is Over all trajectories for which the initial CH3Clorientation is “up” (as defined in the text). Shown are the separate contributionsfrom the solvation of C1- by the H20and the CH3Cllayers.
Figure 4. Average time-dependent change in the H20(inter and intra)
potential energy, the CH3Cl(inter and intra) potential energy, and the H20CH3Clpotential energy following electron attachment to a CH3CI in the “up” orientation at the vacuum-water/CH,Cl interface. The dotted line gives the change in the temperature of the system (values shown on the right axis). \ -10
-20.
AE
-
-4 0.0
0.25
0.50
0.75
1.0
t (PS)
Figure 3. Average time-dependent change in the kinetic energy of the CHI ejected at the interface (solid lines) and in the gas phase (dashed
line), following electron attachment to a CH3Clmolecule. The three solid lies (a-c) correspond to the initial orientationof the digpociatingCH3C1 molecule being “up”, ‘parallel”, and “down”,respectively, as defined in the text. The dotted line is the average CH3-Cl- potential energy. of the solvation energy (25%) is due to the interaction with neighboring CH3Cl molecules. The results in Figure 2 are presented for the case where the dissociating CH3Cl molecule is in the up orientation (the CH3 is ‘facing” the vacuum), but very similar results are obtained in the other two orientations. No matter what the initial orientation of the dissociating CH3Cl molecule is, the C1- never escapes into the vacuum. Figure 3 shows that the CHI fragment does not carry any of this solvation energy. In this figure, the change in the kinetic energy of the CH3 fragment (relative to the initial Boltzmann distribution) as a function of time for the three different orientations is compared with the same process carried out under identical conditions in the gas phase. We also show the average CH3-Cl- potential energy. Clearly, the CH3 fragment carries less energy than the amount deposited in the gas phase. The time scale for the dissociation is such that in about 75 fs the CH3 and Clfragments do not interact, and the CH3 reaches its maximum velocity. Depending on its orientation, some of its kinetic energy is lost to the bath. Where does the energy released upon solvation of the Cl- go? Figure 4 demonstrates that the energy is almost entirely deposited in bath potential energy: water intramolecular vibrations and water nonbonded (hydrogen ‘bonds”) energy (60%);water-CH3Cl potential energy (20%); CH3C1intermolecular and vibrational
(kcal/mol)-30 ,
\
Y C1-- CH3C1
-50}
1 0.0
0.25
0.5
0.75
0
t (PS)
Figure 5. Same as in Figure 2 but with the C1- dissociation product having a 1-A larger van der Waals radius.
potential energy (less than 10%);and the rest is kinetic energy of the bath modes, as is evident from the increase in the temperature of the system. Our conclusion is that although the solvation of the Cl- is indeed accompanied by a large amount of energy release, most of this energy goes into the bath and is not transferred to the ejected CH3 radical. One possible mechanism for energy transfer to the CH3 radical that needs to be explored is the effect of the increase in the size of the C1 atom of the dissociating CH3C1 upon electron attachment. As a result of this size increase, the Cl- ion m a y find itself on the repulsive side of the Cl--M Lennard-Jones potential, where M is any bath atom (for example, CH3 of a neighboring CH3Cl molecule). The momentum transfer in the Cl--M recoil may be transferred to the ejecting CH3 radical. The results presented in Figures 2 4 were obtained using a change of about 25% in the size of C1 upon electron attachment (from 3.16 to 3.93 A). To determine how sensitive the results to this increase are, we have repeated the calculationspresented above with a new size for the C1- which is significantly larger (4.93 A). The results of these new calculations are shown in Figures 5-7, which are analogous to Figures 2 4 (with the “normal” Cl- size). These figures show that now the CH3 radical gains more energy than in the gas phase, but the total energy released (an average of about 0.2 eV in the most favorable orientation) is still significantly ltss than the one measured in the experiment. Note that the kinetic energy release of the CH3radical when the CH3Cl- is in the up orientation, shown in Figure 6,is the result of an averaging of 80 trajectories. The insert shows the distribution of final CH3 kinetic energies to be
Rose and Benjamin
9564 The Journal of Physical Chemistry, Vol. 96, No. 23, I992
4
t 1 2 3 4 5 6 7 8
0'"
3.0
0.25
L
0.50
0.75
1.0
(PSI
Figure 6. Same as in Figure 3 but with the C1- dissociation product having a 1-A larger van der Waals radius. The insert shows the distribution of fml kinetic energies of the CHI product when the initial orientation of the dissociating CH&I molecule is 9 ~ p " .
knowledge of the CH3-Cl- repulsive potential is not complete, but the only unknown parameter-the steepness of this potentialwhich we choose, is probably not too far from reality. (Obviously, the total amount of energy released in the gas phase is independent of the choice of this parameter.) Another approximate feature of our calculation is the assumption of sudden charge switching from the initial charge on CH3Cl to the final charges. A more reasonable (but quite artificial) choice would be to have a charge-switching function interpolating between the two limits. However, this is expected to further slow down the ejected CH3, as the partial charge carried by this fragment in the very early stages of the dissociation will tend to be solvated by the bath. We conclude that an explanation for the excess energy of the CH3 must be found elsewhere. One possible mechanism that our classical model is not capable of elucidating is the possibility that the energy change for the reaction CH3C1 e- CH3Cl- in the condensed phase is larger than the one in the gas phase due to the change in the electronic structure of the solvated CH3C1. As a result, more energetic electrons would be required to create the anion,and some of this energy may be available for the departing CH3 radical. There is some evidence for a large increase in the negative electron affinity of CHpCl in the condensed phase from quantum mechanical calc~~lations.~~' However, these are not very reliable, as they significantly overestimate the same quantity in the gas phase.** Another source of difficulty is the correct description of the CH3Cl- dissociation. Quantum mechanical calculations predict that the dissociation process may involve a coupling between the CH3 rocking mode and the reaction coorMore work is needed to shed light on all of these issues.
+
-
Acknowledgment. The work has been supported by Grants
NSF (CHE-9015106) and ACS-PRF (22862-G2). Registry No. Chloromethane, 74-87-3; water, 7732-18-5.
Refennces lad N o h I
0.0
0.25
0.50
0.75
1.0
t (PS)
Figure 7. Same as in Figure 4 but with the C1- dissociation product having a 1-A larger van der Waals radius.
quite broad, but still the largest kinetic energy observed is only 7.2 kcal/mol (0.3 eV). The increase in the energy of the methyl radical upon h m s h g the size differencebetween the Cl- product and the parent C1 atom is purely kinematic and does not represent an outcome of ionic solvation. In fact, as Figure 5 demonstrates, the solvation energy is less in this case than in the case where the size of the C1- is normal (Figure 2). Again, the solvation energy is balanced by an increase in the energy of the bath, as shown in Figure 7. (Note the change in scale between Figures 2-4 and 5-7.)
Iv. coaclrasioas The calculation^ presented above show that a classical solvation model cannot be used to explain the large energy d i s w l in the ejected CH3 radical upon electron attachment at the condensed phase-vacuum interface. Although the other product of the electron attachment, the C1- ion, is promptly solvated, the large energy released goes entirely into the bath. Within the classical model, our calculations involve several approximations, and some of the potential energy functions are not known very accurately. Nevertheless, this is not expected to deet the main amclusions. For example, the significant increase we allow in the size of the C1 atom upon electron attachment (in some of the trajectories) probably yields the upper limit to the energy deposited in the CH3fragment due to this mechanism. Our
(1) Fleming, G. R. Chemical Applications of Ultrafast Spectroscopy; Oxford University: New York, 1986. (2) Simon, J. D. Ace. Chem. Res. 1988, 21, 128. (3) Barbara, P. F.; Jamba, W. Ado. Phorochem. 1990,15, 1. (4) Zwan, G. van der; Hyncs, J. T. J. Phys. Chem. 1985,89,4181. ( 5 ) Wolynes, P . G . J. Chem. Phys. 1987,86,5133. (6) Rips, I.; Klafter, J.; Jortner, J. J. Chem. Phys. 1988,89, 4288. (7) Bagchi, 9. Annu. R N . Phys. Chem. 1989,40, 115. ( 8 ) Raincri, F.0.;Zhou, Y.; Friedman, H. L.; Stell, G. Chem. Phys. 1991, 152. 201. (9) Karim, 0. A.; Haymet, A. D. J.; Banet, M.J.; Simon, J. D. J. Phys. Chem. 1988,92,3391. (10) Maronctlli, M.;Fleming, G. R. J. Chem. Phys. 1988, 89, 5044. 1111 Carter. E. A.: Hmes. J. T. J. Chem. Phvs. 1991.91. 5961. (12) Sitzmann, E.-V.;%nthal, K. 9. J. Chem. Phy;. 1989, 90,2831. (13) Meecb, S. R.; Yoahira, K. Chem. Phys. Lerr. 1990, 174, 423. (14) Zwan, G. van der; Mazo, R. M. J. Chem. Phys. 1985, 82, 3344. (15) Urbakh, M.; Klafter, J. J. Phys. Chem. 1992,%, 3480. (16) Benjamin, I. J. Chem. Phys. 1991, 95, 3698. (17) Row, D.A.; Ecnjamin, I. J . Chem. Phys. 1991,95,6856. (18) Kwini, M.;Iedcma, M.J.; Cowin, J. P.; Gilton, T. L. J. Phys. Chem. 1992, %, 2795. (19) Thisl, P. A.; Madey, T. E. Surf.Scf. Rep. 1987, 7, 211. (20) BacnQen, H. J. C.; Poatma, J. P. M.;Gunateren, W. F. van; Her~I
mans,J. In Intermolecular Forces; Pullman, B., Eds.;D. Rcidel: Dordncht,
1981; p 331. (21) Kuchitau, K.; MoMo, Y.Bull. Chem. Soc. Jpn. 1965,38,814. (22) Black, J. E.; Bopp, P. Surf. Sci. lW, 140,275. (23) Spohr, E.; Heinzinger. K.Brr. Bunsen-Ges. Phys. Chem. 1988,92, 1358.
(24) Hotop, H.; Linekrger, W. C. J . Phys. Chrm. Ref.Data 1985, 14, 731. (25) Petrovic, Z. L.; Wang, W. C.; Lee, L. C. J. Chem. Phys. 1989,W. 3145. (26) Canadell, E.; Karafdoglou, P.; Salem, L. J. Am. Chrm. Soc. 1980, 102, 855. (27) Clark,T. Faraday Discuss. Chem. Soc. I%, 78, 203. (28) M , R.;&snardi, F.;Bottoni, A.; Robb, M.A.; Taddei, F. Chem. Phys. Lcrr. 1!989,161,79. (29) Tada, T.; Yoshimura, R. J . .4m. Chem. Soc. 1992,114, 1593.
