J. Phys. Chem. 1995,99, 15733-15737
15733
Femtosecond, Velocity-Gating of Complex Structures in Solvent Cages P. Y. Cheng, D. Zhong, and Ahmed H. Zewail” Arthur Amos Noyes Laboratory of Chemical Physics, Califomia Institute of Technology, Pasadena, Califomia 91125 Received: August 16, I99P
The method of femtosecond velocity-gating in complex systems is reported with an application to the caging problem in a pure solvent cluster of iodobenzenes (n = 2-8). Two structural forms of iodine atoms in the clusters are identified based on the two distinct product recoil velocity distributions and their temporal behaviors: One is the “uncaged” iodine atoms that exhibit ultrafast dissociations (-1.4 ps) and high recoil velocities; the other is the “caged” iodine atoms that escape very slowly (up to 300 ps) from the clusters with very low kinetic energies. These solvation dynamics relate to the nature of the microscopic molecular structures.
Introduction Photofragment translational spectroscopy, which measures product kinetic-energy distribution (KED), provides valuable information on the nature of the exit channel and energy disposal in reactions (see, for example, reviews in refs 1 and 2). More recently, this approach has been applied to the photoinitiated dissociation and reactions in molecular demonstrating its potential in these systems. These studies have shown that the product KEDs from clusters are in general broader and less anisotropic than those from the monomers and are shifted to much lower energies. The behavior has been attributed to the collisional relaxation of the photoproducts escaping from the clusters or the condensed-phaseanalogy of the “cage effect”. Molecular dynamics simulation studies have elucidated this Recently, we reported on a two-dimensional, time-velocity correlation method in which the KED measurements are made with femtosecond (fs) time resolution.Io In this method, molecules of interest are dissociated by a fs laser, and the free reaction products are detected by another fs laser in a time-offlight mass spectrometer (TOF-MS) through resonance ionization. The KED measurement is accomplished by measuring the kinetic-energy resolved time-of-flight (KETOF) distribution of the ionized products. This allows the fs temporal evolution of the product KEDs to be observed. Reaction channels producing different product KEDs can then be separately monitored, even if their product identities are indistinguishable by conventional mass analyses. We have previously applied this method to the dissociation dynamics of iodobenzene and separated two channels for the C-I bond breakage based on their product KETOF distributions and temporal behaviors.IO Earlier, El-Sayed’s group” has shown -that the two channels in iodobenzene have different anisotropy parameters @) and they related the degree of the anisotropy to the time scale of energy redistribution. Both our real-time study and El-Sayed’s anisotropy study were made on the iodobenzene monomer. In this letter, we report the application of this fs KETOF method to a more complex system (iodobenzene clusters) illustrating the power of the time resolution in identifying the caging dynamics and the solvent structures. Dissociation of pure iodobenzene clusters yields product KEDs that are broader, less structured, and lower in energy. Temporal evolution of ‘Contribution no. 9123. Abstract published in Advance ACS Abstracts, October 15, 1995.
@
the product KEDs from the clusters is clearly revealed, and two vastly different channels are separated: One in which dissociation occurs in few picoseconds (ps) and the other in which dissociation takes up to hundreds of ps. We relate these two distinct channels two different solvation structures and we systematically use the velocity gating to examine the nature of caging in the different-sized structures (n = 2-8).
Experimental and Methodology The molecular beam apparatus and the fs laser system used in the present study have been described previously.12 Briefly, the molecular beam was produced by expanding iodobenzene vapor (70 “C, -10 Torr) with He gas (-800 Torr) through a pulsed valve and was skimmed before the intersection with the fs laser pulses in the extraction region of a TOF-MS. The pump laser pulse (275 nm) initiated the reaction and the probe laser pulse (304.7 nm), delayed in time, resonantly ionized the free I atom products through 2+1 REMPI. The resulting ions were detected by a dual microchannel plate (MCP) assembly. The molecular beam, TOF-MS axis and laser beams are mutually orthogonal. The pump laser polarization was parallel to the TOF-MS axis and perpendicular to the probe laser polarization. The TOF-MS is of the two-stage Wiley-McLaren type operated under the space-focusing condition. In traditional photofragment translational ~pectroscopy,‘~~ one measures the flight time of a neutral product from the photolysis region to a mass spectrometer, after the flight, where it is ionized and mass analyzed. In the present technique, the initiation of the reaction and ionization are made in the laser interaction region prior to the drift into the TOF field-free region. This use of a TOF-MS for measuring product velocity distributions has been discussed by many authors.5.I3-I7 For two fragments recoiling with the same speed lvzl along the TOF axis ( 2 ) but in opposite directions, their arrival times at the detector are separated by the “turn-around” time in the extraction filed region: At = 2m(v,(lqE,where m and q are the mass and charge of the ions, and E is the extraction electric field. Since the tum-around time is linearly proportional to lv,l, the observed TOF distribution is a direct measure of the projection of the recoil velocity distribution onto the TOF axis and is symmetric with respect to v, = 0. A discrimination aperture (5 mm diameter) is placed right in front of the MCP detector to reject those ions with high-velocity components perpendicular to the TOF axis (vX,J. Under our operating conditions, the flight time of I+ is centered at -24
0022-365419512099-15733$09.00/0 0 1995 American Chemical Society
15734 J. Phys. Chem., Vol. 99, No. 43, 1995
Letters
Theoretical Section Mass Spectra: (Iodobenrene),
As mentioned before, the KETOF distribution measured in this work is a one-dimensional projection of the full 3-dimensional product-velocity distribution onto the TOF axis. Analyses of the angular and velocity distributions have been discussed by several author^,^^^^-'^ and here we only briefly consider the key theoretical treatment. For fragments having the same kinetic energy, the KETOF distribution is dependent on their angular distribution P(B), which can be expressed as
n=2
where P2(x) = ' / 2 ( 3 x 2 - 1) is the second Legendre polynomial, 8 is the angle between the final recoil velocity ( Y ) , and the pump-laser polarization vector (E), and p is the recoil anisotropy parameter. If the molecule dissociates much faster than the rotational period, then p is given by
n=2
n-I
Case (a)
20
40
30
50
60
70
Time of Flight (ps)
Figure 1. TOF-MSspectra obtained in three regions of the gas pulse: (case a) the leading edge, (case b) slightly behind the leading edge, and (case c) central region of the gas pulse. In the case a spectrum, the 275 nm fs laser is used as the ionization laser. In the case b and c spectra, the laser wavelength was tuned to 285 nm to minimize ionic fragmentation. The reduction of the monomer intensity in the case b and c spectra is due to the lowering of the photon energy. The TOFMS was optimized for the dimer in the spectra shown in the lower panel, but was optimized for larger clusters in the spectrum taken under the case c condition shown in the upper panel.
ps. Theoretically, this means that ions with
1 vdisc = 110
m/s will not be detected because they will miss the discrimination aperture at the detection plane. For ions having recoil speed much greater than vdisc, the discrimination value, the aperture ensures the ions detected have recoil velocity nearly parallel to the TOF axis, i.e., lvzl IvI, making the observed v, distribution a good approximation to the speed distribution. Practically, the degree of discrimination is somewhat degraded in our experiments due to the space charge effect and blurring diffraction off the wire mesh grids. The I atom transients were obtained by monitoring the I+ ion signal with a boxcar gate integrator while scanning the pump-probe delay time. Gating experiments were performed by setting the boxcar gate to sample only a certain portion of the I+ KETOF distribution. A digital oscilloscope (LeCroy 9361, 2.5 GS/s) was used for recording mass spectra and KETOF distributions. Different regions of the gas pulse were sampled by varying the time delay between the fs laser pulses and the pulsed valve opening. This allows for selective excitation of different clustersize distributions without changing the beam conditions. As shown in Figure 1, iodobenzene monomer dominates the leading edge of the gas pulse, whereas the central region of the gas pulse contains significant amount of iodobenzene neat clusters. This selective excitation is also supported by the dramatic differences in the I atom velocity distributions and their temporal behaviors observed in these two regions of the gas pulse, as discussed below.
where xo is a fixed angle between the transition dipole moment @) of the molecule and the fragment final recoil velocity ( Y ) . For purely parallel (XO = 0) and perpendicular (XO = d 2 ) transitions, p = 2 and -1, respectively. If the dissociation is slower than the rotational period, the magnitude of p is reduced. In the laboratory coordinate system in which the z axis is of the TOF axis, the projection of the velocity distribution onto the z axis is
P(v,) = (1/2v,)[ 1 + rgP2(COS 5) P,(cos 0)l
(3)
where v, is the velocity component along the z axis, YO is the speed of the fragment, 6 is the angle between the z axis and the pump-laser polarization, and 0 = COS-~(Y,/Y~) is the angle between the z axis and the recoil velocity. If the pump-laser polarization is parallel to the TOF axis, as employed in this work, then the v, distribution reduces to
-
(4) The appearance of this v, distribution depends on the value of For a purely parallel transition @ = 2), the distribution is a parabola function truncated at v, = fvo. For a purely perpendicular transition (B = -I), it is an inverted parabola function truncated at v, = f v o . For p = 0, the v, distribution is constant between -YO and f v o . Several points are worth mentioning. First, if the time scale for bond breaking and caging is relatively long, the above treatment must take into account changes in ,L? during that time. Second, rotational dephasing is important and will depend on the initial temperature.'* Such dephasing becomes increasingly important as the reaction time gets longer. Third, in a full treatment, both the pump and probe electric-field orientations, relative to the transition moments of the parent and fragment, must be ~0nsidered.l~ Fourth, the presence of the discrimination aperture reduces the collection solid angle and improves the kinetic energy resolution because of the limitation on the v, distribution. l 3 Finally, it should be noted that caging can lead to a change in ,L?, in addition to the expected cage exit delay. This change in ,L? can give rise t o a temporal behavior in the gating
p.
-
Letters
J. Phys. Chem., Vol. 99, No. 43, 1995 15135
KETOF
Case (c)
.
2.4 ps
I
I
I
800
0
-800
VI
“s)
Figure 2. (Bottom): I+ KETOF distribution arising from the dissocia-
tion of iodobenzene monomer at a pump-probe delay time of -10 ps. (Top): I+ KETOF distributions observed in the cluster-rich region (case c) of the gas pulse at various pump-probe delay times. experiment. However, the velocity distribution and its temporal evolution and the detection of total versus gated ion signal can be used to separate the change in @ for real exit delay.
Results and Discussion The first absorption band of iodobenzene (320-240 nm) is composed of two types of transitions: One is the (n,a*) excitation, localized along the C-I bond, and the other is the (n,rc*)excitation of the benzene ring s y ~ t e m . ’ The ~ - ~dissocia~ tion dynamics following excitation in this wavelength region is complicated by the severe spectral overlaps of these two transitions. The (n,a*) states are similar to those of the A-band continuum observed in alkyl iodides which lead to a direct dissociation of the C-I bond in less than 400 fs.2’ The (n,n*) states are expected to exhibit predissociative behavior due to mixing with the isoenergetic (n,a*) states. El-Sayed and coworkers” measured the I atom product-velocity distributions and found that these two dissociative channels produce two very different distributions. On the basis of the anisotropy measurements, they have assigned a narrow high-velocity distribution to the (n,o*) direct dissociation and a broader low-velocity distribution to the (n,n*)predissociation. With the fs time-resolved KETOF technique, we have previously reported velocity distributions of I atom at various delay times following the t = 0 excitation.I0 The I+ KETOF distributions observed at long delay times (>2.5 ps) in our experiments are very similar to those observed by the El-Sayed group” using nanosecond lasers. A typical I atom KETOF distribution observed at -10 ps is shown in Figure 2 (case a). A narrow high-velocity distribution and a broad low-velocity one are clearly observed. The high-velocity component is peaked at 800 ic 50 d s , corresponding to a kinetic energy of -3400 cm-’ for the I atom fragments.
0
1
2
3
4
Time Delay (ps)
Figure 3. I atom transient obtained in the monomeddimer region (case b) of the gas pulses by gating the highest velocity component in the I+
KETOF distribution. The data are fitted to the sum of two delayed step functions convoluted with the pump-probe correlation. The first rise is due to the monomer (n = 1) and is found to be identical to the transient obtained in the monomer region. The second rise due to the dimer (n = 2) is delayed from the first one by -1 ps. The second delayed rise is reproducible, but the appearance of the entire transient depends on the gating and the molecular beam conditions. Transients obtained by gating the high- and low-velocitydistributionsreveal markedly different temporal behaviors.I0The high-velocity I atoms clearly exhibit a coherent wave packet motion manifested by the 400 fs delayed rise, consistent with the direct dissociation mechanism. Whereas, the low-velocity I atom transients are characterized by an exponential rise of about 600 fs, indicating an indirect predissociative process. To avoid any clustering, the KETOF distribution shown in Figure 2 (case a) was produced using an effusive beam of iodobenzene with no He carrier gas. Identical KETOF distributions can be observed under the supersonic beam condition (with 800 Torr of He) when the fs lasers sample the leading edge of the gas pulse. Under both conditions, only iodobenzene monomer is present in the mass spectra, as shown in Figure 1 (case a). When the fs lasers are moved from the leading edge toward the central region of the gas pulses, iodobenzene cluster ions gradually appear in the mass spectra (Figure 1, cases b and c). In the meantime, the I+ KETOF distributions also gradually become less structured and finally turn into those shown in Figure 2 (case c) when the fs laser pulses reach the central region of the gas pulse. Between the leading edge and the central region of the gas pulse, the KETOF distributions exhibit mixed characteristics of the two extreme conditions. The change in KETOF distributions is related to the iodobenzene clusters as it is clearly accompanied with the appearance of the cluster peaks in the mass spectra. In the following discussions, we divide the gas pulse into three regions: the “monomer” (case a, leading edge of the gas pulse), the “monomer/dimer” (case b), and the “cluster-rich” (case c, central portion of the gas pulse) regions. These limits are characterized based on the mass spectra observed in each region (see Figure 1). Slightly behind the leading edge of the pulse, there exists a region where only the iodobenzene monomer and dimer can be seen in the mass spectra (Figure 1, case b). The KETOF distribution observed in this monomeddimer region is mostly from the monomer, but is distorted due to the presence of the dimer. By gating the highest velocity component under this condition, we obtain the transient shown in Figure 3. The initial rise is identical to the I atom transient obtained in the monomer region. A second component, delayed by -1.4 ps from the zero-of-time, is also observed due to the iodobenzene dimers.
15736 J. Phys. Chem., Vol. 99, No. 43, 1995
d , ,
,
,
,
0
3
6
0
3
6
xi00 ,
Letters
I
9
0
3
6
9
9
0
3
6
9
Time Dealy (ps)
Figure 4. I atom transients obtained by gating the high-velocity (right panels) and low-velocity (left panels) components of the I+ KETOF distribution under the monomer (case a) and cluster-rich (case c) conditions. Note the different time scales in the two long-time transients.
This observation indicates that the coherent wave packet motion of the C-I direct dissociation observed in the iodobenzene monomerlo remains intact in the dimer but is greatly slowed down by the presence of the other iodobenzene molecule. In the monomer, the appearance of the free I atom is govemed by the relative velocity of the two fragments falling apart. We have shown that it takes 400 fs for the two fragments to escape from the long-range vdW interaction such that the free I atoms can be detected. For the iodobenzene dimer, the further delayed appearance of the free I atom can be rationalized by the following two effects. First, the relative velocity of the two dissociating fragments can be slowed down due to the additional vdW potential originating from the undissociated iodobenzene. Second and more importantly, the relative velocity between the departing I atom and the undissociated iodobenzene is now much slower, making it spend more time to escape from the vdW potential of the undissociated iodobenzene. Figure 2 (case c) shows the KETOF distributions observed in the cluster-rich region. The mass spectra observed in this region indicate that a broad distribution of iodobenzene clusters (n = 2-8) is sampled. Comparison with the monomer KETOF distribution (case a) suggests that the monomer contribution is quite small. Also shown in Figure 2 is the dramatic temporal evolution of the KETOF distributions as the pump-probe delay time increases. Two distinct types of distributions are readily identified from this time-resolved KETOF experiment. One is a high-velocity bifurcated distribution similar in shape to the monomer KETOF distribution but is broader and less structured. By gating the peak region of this distribution, while the pumpprobe delay is scanned, we obtained transients that are dominated by this distribution. As shown in Figure 4 (right case c),
this cluster high-velocity distribution grows in rapidly and can be characterized by a delayed singlf-exponential rise of 1.4 ps. The similarities between this cluster high-velocity component and the monomer in terms of their velocity distributions and temporal behaviors suggest that some monomer-like "uncaged" I atoms are present in the clusters. These I atoms dissociate from the clusters without direct interactions with the rest of the system and are expected to result in a rapid escape and little translational energy loss. The velocity distributions, although less structured, clearly reveal that the averaged kinetic energy of the I atoms released from the clusters is comparable to that from the monomer. This indicates that the initial kinematics does not change significantly for these uncaged I atoms, and the gated signal is mostly due to the direct dissociation channel. The additional vdW potentials in the clusters are expected to decrease the available energy for the dissociation, consistent with the slightly lower peak velocity (smaller splitting) observed in the cluster KETOF distributions. The presence of the additional vdW potential also slows down the I atom departure in time, as manifested by the delayed rise observed in the monomeddimer transient shown in Figure 3. In larger clusters (n > 2), the uncaged I atoms are expected to be further slowed down because of the deeper vdW wells and different structures. We extrapolate that the delayed exponential rise of 1.4 ps observed in the cluster high-velocity I atom transient shown in Figure 4 is an averaged result of several delayed rises originating from various sizes of clusters (n = 2-8). At long pump-probe delay times, an additional component is clearly observed in the KETOF distributions (Figure 2, case c). Subtraction of the early time (-5 ps) distribution from the latter ones reveals a Gaussian-like distribution centered at v, = 0. Transients obtained by gating this general region exhibit biexponential behaviors with the slower rise being as long as 365 ps (Figure 4,left panels). By analogy the initial fast rise (2.3 ps) is due to the predissociation of uncaged I atoms in the clusters following the (n,n*) excitation. Again, the dissociation is slowed down because of the presence of the additional vdW potential. Nevertheless, the much slower rise of the Gaussianlike velocity distribution centered at v, = 0 is a strong indication that a very different form of I atoms is also present in the clusters. We consider this much slower component as evidence to "caged" I atoms that immediately undergo collisions with other iodobenzene molecules following dissociation. A rough calculation shows that -95% of the translational energy of an iodine atom can be lost by a single collinear collision with a point mass having the mass of an iodobenzene molecule. Therefore, even though it may depend on the impact parameters, because of the finite size of the iodobenzene molecule, the collisional translational energy relaxation of the I atom in the clusters should be facile. In larger clusters, these caged I atoms can be trapped for long periods of time, allowing the kinetic energy of the I atoms to be distributed among the cage modes. The energy transfer can soften the solvent cage and leads to the escape of I atoms with very low kinetic energy. These qualitative descriptions in terms of the product velocity distribution and its temporal behavior are in complete accord with the cluster low-velocity distribution. Upon reducing the degree of clustering, the slower component gradually decreases and its rise time becomes shorter. We have taken two low-velocity I atom transients somewhere between the cluster-rich and the monomeddimer regions and obtained rise times of 160 and 55 ps. At the leading edge of the gas pulse, the low-velocity I atom transient becomes eventually
J. Phys. Chem., Vol. 99, No. 43, 1995 15737
Letters identical to the monomer transient shown in Figure 4 (left, case a). These observations indicate that the residence times of the caged I atoms are longer in larger clusters, consistent with the caging dependence on size. Moreover, they also suggest that the fraction of the caged I atoms increases with the cluster size. The latter conclusion provides important implications to the structures of the iodobenzene clusters. The two structural forms of I atoms present in the iodobenzene clusters can be rationalized by the fact that the dispersive interactions between 1-1, I-phenyl, and phenyl-phenyl are relatively strong and comparable. The relatively large dipole moment of iodobenzeneZ2(-1.7 D) may also play an important role through electrostatic interactions. In an iodobenzene cluster of a certain size, some C-I bonds may point directly toward a phenyl ring or an I atom, while some may not. Isomeric structures may also exist because of the comparable interaction energies between the three pairs mentioned above.
Conclusion The results presented here demonstrate the usefulness of adding a new dimension (time) to translational spectroscopy. Reaction dynamics can be examined not only by the analysis of product kinetic-energy distributions but also by their temporal evolution for a specific velocity distribution. The reverse method of gating at different times allows one to separate the dynamics of different structures. It was shown here that two structural forms of I atoms are present in the iodobenzene clusters (n = 2-8). The uncaged I atoms exhibit slightly perturbed monomer-like dynamics and reflect the effect of the vdW forces to the departing fragments. On the other hand, the caged I atoms show dramatically different dissociation dynamics due to a caging structure. Such distinct structures have also been found in the theoretical study of HF in Are9 Structures of dimers exhibit, surprisingly, coherent caging phenomena. Further theoretical studies of the molecular dynamics in different-sized structures are underway.
Acknowledgment. This work was supported by the National Science Foundation.
References and Notes (1) Riley, S. J.; Wilson, K. R. Faraday Discuss. Chem. SOC.1972, 53, 132. (2) Wodtke, A. M.; Lee, Y. T. In Molecular Photodissociation Dynamics; Ashfold, M. N. R., Baggott, J. E., Eds.; Royal Society of Chemistry: London, 1987; Chapter 2. Bersohn, R. Ibid., Chapter 1. (3) Segall, J.; Wen, Y.; Wittig, C.; Garcia-Vela, A,; Gerber, R. B. Chem. Phys. Lett. 1993,207, 504. Jaques, C.; Valachovic, L.; Ionov, S.; Bohmer, E.; Wen, Y.; Segall, J.; Wittig, C. J. Chem. SOC.,Faraday Trans. 1993, 89, 1419. (4) Young, M. A. J. Chem. Phys. 1995, 102, 7925. (5) Ogorzalek Loo, R.; Hall, G. E.; Haem, H. P.; Houston, P. L. J . Phys. Chem. 1988, 92, 5. (6) Syage, J. A. Chem. Phys. Lett., submitted. (7) Alimi, R.; Gerber, R. B. Phys. Rev. Lett. 1990, 64, 1453. (8) Garcia-Vela, A.; Gerber, R. B.; Imre, D. G.; Valentini, J. J. Chem. Phys. Lett. 1993, 202, 473. (9) Schroder, T.; Schinke, R.; Liu, S.; BaEiC, Z.; Moskowitz, J. W. J . Chem. Phys., submitted. (10) Cheng, P. Y.; Zhong, D.; Zewail, A. H. Chem. Phys. Lett. 1995, 237, 399. (11) Hwang, H. J.; El-Sayed, M. A. J. Chem. Phys. 1992, 96, 856. Freitas, J. E.; Hwang, H. J.; El-Sayed, M. A. J . Phys. Chem. 1993, 97, 12481. (12) Zewail, A. H. Femrochemisrry-Ultrafastdynamics of the chemical bond World Scientific: Singapore, 1994, and articles therein. (13) Hwang, H. J.; Griffiths, J.; El-Sayed, M. A. Int. J . Mass Spectrom. Ion Processes 1994, 131, 265. (14) Penn, S. M.; Hayden, C. C.; Carlson Muyskens, K. J.; Crim, F. F. J . Chem. Phys. 1988, 89, 2909. (15) Hertz, R. A,; Syage, J. A. J . Chem. Phys. 1994, 100, 9265. (16) Mons, M.; Dimicoli, I. J . Chem. Phys. 1989, 90, 4037. (17) Baumert, T.; Pedersen, S.; Zewail, A. H. J . Phys. Chem. 1993,97, 12447. (18) Baskin, J. S.; Zewail, A. H. J. Phys. Chem. 1994, 98, 3337 and references therein. (19) Dunn, T. M.; Iredale, T. J . Chem. SOC.1952, 1592. (20) Freedman, A.; Yang, S. C.; Kawasaki, M.; Bersohn, R. J . Chem. Phys. 1980, 72, 1028. (21) Knee, J. L.; Khundkar, L. R.; Zewail, A. H. J . Chem. Phys. 1985, 83, 1996. (22) Hurdis, E. C.; Smyth, C. P. J . Am. Chem. SOC.1942, 64, 2212. JP9524385