Dispersed Fluorescence Spectroscopy of the ZnC2H5Free Radical

JOURNAL OF MOLECULAR SPECTROSCOPY ARTICLE NO.

185, 48–53 (1997)

MS977361

Dispersed Fluorescence Spectroscopy of the ZnC2H5 Free Radical Simon J. Pooley and Andrew M. Ellis 1 Department of Chemistry, University of Leicester, University Road, Leicester LE1 7RH, United Kingdom Received March 7, 1997; in revised form May 28, 1997

Dispersed fluorescence spectra of the zinc monoethyl radical are presented. These have allowed vibrational frequencies of several modes in the ground electronic state of this molecule to be determined for the first time. In particular, substantial activity in both the Zn–C stretch and the Zn–C–C bend has been observed, with the frequencies for these modes being deduced as 387 and 180 cm01 , respectively. With the aid of the dispersed fluorescence data it has also proved possible to make reasonable assignments for several bands in the excitation spectrum. ZnC2H5 was prepared in these experiments by a dc electrical discharge through an argon/diethyl zinc mixture in a supersonic nozzle. q 1997 Academic Press

particularly the skeletal Zn–C stretching and Zn–C–C bending modes. The dispersed fluorescence spectra have also provided important evidence which can be used to make some assignments of bands in the excitation spectrum. These new assignments are also presented here.

INTRODUCTION

A number of reports of fluorescence spectra of the ZnCH3 and CdCH3 free radicals have been made in recent years (1– 10). The monoalkyls of zinc and cadmium are interesting for several reasons. First they are, by the standards of organometallic chemistry, relatively simple molecules which might serve as useful models of metal–ligand bonding in more complex systems. Recent high-resolution LIF studies of both molecules have provided detailed information on their structures and bonding (9, 10). A second point of interest in the monoalkyls of zinc and cadmium derives from their role in the decomposition of the corresponding dialkyl metals, a process used for metal deposition in the production of electronic and optical devices. Thermal decomposition is the most important route in the industrial production of zinc and cadmium films, but UV photolysis of the dialkyl metals is also a possibility (11). Because of this practical application, and because of their relative simplicity, dimethyl zinc and dimethyl cadmium have been the targets of several studies of photodissociation mechanisms (1, 5, 12–15). We recently reported the first spectroscopic observation of the ZnC2H5 radical, which we achieved using LIF spectroscopy following a dc discharge through diethyl zinc (16). The jet-cooled excitation spectrum showed strong absorption bands in the blue which were clearly attributable to ZnC2H5 . Prior to this work it had been suggested that ZnC2H5 was most probably unobservable by LIF spectroscopy due to excited state predissociation (13). Although the vibrationally resolved excitation spectrum that we observed had an excellent signal/noise ratio, none of the vibrational structure could be confidently assigned in our initial study. We present in this work the first dispersed fluorescence spectra of ZnC2H5 . These have allowed us to probe some of the vibrational structure in the electronic ground state, 1

EXPERIMENTAL

The experimental apparatus and procedure employed was similar to that described in detail in previous publications (16, 17) and so the description given here is deliberately brief. ZnC2H5 was prepared by discharge fragmentation of diethyl zinc (Aldrich). The precursor was subjected to freeze–pump–thaw cycles before each experiment to ensure that no volatile impurities were present. In order to obtain a jet-cooled spectrum, the discharge of diethyl zinc, seeded in argon carrier gas, took place in a single-channel discharge nozzle prior to expansion into a vacuum chamber. About 2 cm downstream of the nozzle the expanding gas was crossed by the beam from a Nd:YAG pumped dye laser. Excitation spectra were recorded by detecting the unfiltered fluorescence with a Hammamatsu R568 photomultiplier tube. Dispersed fluorescence spectra were obtained using a 0.27-m monochromator (Acton Research) also equipped with a Hammamatsu PMT (R928). The fluorescence signals were digitized and then fed to a computer on a shot-to-shot basis for software-driven gating, integration, and averaging. The resolution in the dispersed fluorescence spectra was approximately 20 cm01 and the estimated accuracy in the quoted band positions is {5 cm01 . RESULTS AND DISCUSSION

Preliminaries Figure 1 shows a broad survey of the laser excitation spectrum of ZnC2H5 . In addition to the relatively broad

To whom correspondence should be addressed. 48

0022-2852/97 $25.00 Copyright q 1997 by Academic Press All rights of reproduction in any form reserved.

AID

JMS 7361

/

6t1f$$$101

09-09-97 09:49:01

mspal

DISPERSED FLUORESCENCE OF ZnC2H5

FIG. 1. Laser excitation spectrum of ZnC2H5 . The band assignments were made using evidence from the dispersed fluorescence spectra recorded in the present work. See text for more details.

bands due to ZnC2H5 , sharp lines from ZnH are clearly identifiable in the spectrum. The strong band at 22 515 cm01 is the lowest energy transition seen for ZnC2H5 and is therefore assigned to the A˜ –X˜ 0 00 transition. We can elucidate the symmetries of the ground and excited electronic states of ZnC2H5 by correlation with ZnCH3 (which has C3£ symmetry). The ground electronic state of ZnCH3 is 2 A1 and therefore in ZnC2H5 , assuming Cs point group symmetry, we expect a 2 A * ground state. The upper state in the strong LIF transitions of ZnCH3 , a 2 E state, will split into 2 A * and 2 A 9 states in ZnC2H5 . Transitions to both of these excited states from the ground state are optically allowed but there was no evidence from our previous LIF study (16) which would allow us to determine which has the lower energy. In our earlier study we carried out ab initio calculations at the Hartree–Fock level (double z plus polarization basis set) on the ground electronic state of ZnC2H5 (16). The aim of these calculations was to obtain an approximate structure of ZnC2H5 , which is summarized diagrammatically in Fig. 2, and to predict vibrational frequencies. Since the structure observed in the excitation spectrum reflects excited rather than ground state vibrational intervals, the ab initio vibrational frequencies were necessarily of limited assistance in the assignment of the excitation spectrum. It would clearly be more appropriate to make comparisons with ground state vibrational observations and this is possible using the dispersed fluorescence data.

49

FIG. 2. Illustration of the equilibrium geometry of ZnC2H5 generated from the parameters reported in Ref. (16). Parameters not shown in the ˚ , rC – H3 Å 1.093 A ˚ , rC – H4 Å rC – H5 Å figure are rC – H1 Å rC – H2 Å 1.092 A ˚ , uZn – C – H1 Å 105.87, uH3 – C – H4 Å 107.07, and uH4 – C – H5 Å 107.17. 1.090 A

states lying close to the X˜ 2 A * ground state, the additional structure can be attributed to vibrational excitation in the X˜ state. To assign this vibrational structure we can draw on help from two sources, (i) comparison with ZnCH3 , and (ii) comparison with our previous ab initio predictions (16). Evidence from previous LIF studies of ZnCH3 (6), confirmed by ab initio calculations carried out by Jamorski and Dargelos (18), shows that the A˜ 2 E–X˜ 2 A1 transition can approximately be described as a zinc 4p R 4s transition of the unpaired electron. Since the bonding is largely unaltered by this process, diagonal Franck–Condon factors predominate in the A˜ –X˜ system. In fact significant Franck–Condon activity is seen in only two modes, the Zn–C stretch and the methyl umbrella. If we make the reasonable assumption that the ZnC2H5 spectrum also arises from a mainly metal-localized transition, then as a first approximation we need only consider those modes, particularly totally symmetric modes,

Dispersed Fluorescence Spectrum from Pumping A˜ –X˜ 0 00 Transition The dispersed fluorescence spectrum obtained on laser pumping the A˜ –X˜ origin transition at 22 515 cm01 is shown in Fig. 3. Several medium- and low-intensity bands are seen in addition to emission (and some scattering of laser light) at the excitation wavelength. Given that, by analogy with ZnCH3 , we expect no other optically accessible electronic

FIG. 3. Dispersed fluorescence spectrum obtained on laser excitation of the A˜ –X˜ origin transition (22 515 cm01 ).

Copyright q 1997 by Academic Press

AID

JMS 7361

/

6t1f$$$102

09-09-97 09:49:01

mspal

50

POOLEY AND ELLIS

TABLE 1 Calculateda and Observed Vibrational Frequencies of ZnC2H5

for which the vibrational potential is significantly altered by this process. Thus we would certainly expect to see some activity in the Zn – C stretching mode and also possibly some in the Zn – C – C bend. Other modes, such as the C – C stretch, should play a lesser role in the vibrational structure because of their remoteness from the chromophore. Of course our description of the normal modes in terms of simple motions such as a C – C stretch is only meant to convey the dominant character of a vibration. Comparison with the ab initio data from Ref. ( 16 ) allows us to make some specific assignments. Table 1 summarizes the calculated harmonic vibrational frequencies of the ground electronic state of ZnC2H5 . For reasons that will become clear later, we are interested mainly in the very low frequency modes in the molecule. Notice that the three lowest frequency modes are the Zn – C – C

bend ( n11 ) , the methyl torsion ( n18 ) , and the Zn – C stretch ( n10 ) ; of these n11 and n10 are totally symmetric. The two bands immediately to the red of the laser wavelength in Fig. 3 correspond to excitation of two relatively low frequency modes in the ground electronic state and the most obvious assignment is that they arise from the Zn–C stretch ( n10 ) and the Zn–C–C bend ( n11 ). According to the ab initio results in Table 1, n11 will have by far the lower frequency of these two modes. Consequently, the lowest frequency band in the dispersed fluorescence spectrum, which is 180 cm01 to the red of the laser frequency, is most likely due to the Zn–C–C bending mode. The ab initio calculations predict a harmonic frequency of 196 cm01 for this mode, a value remarkably close to the observed interval. The only other mode expected to possess such a low frequency is the methyl torsion ( n18 ). However, we can eliminate this from consideration for several reasons. First of all,

Copyright q 1997 by Academic Press

AID

JMS 7361

/

6t1f$$$102

09-09-97 09:49:01

mspal

DISPERSED FLUORESCENCE OF ZnC2H5

since the methyl torsion has a 9 symmetry, we would only expect transitions with Dn Å even for this mode, making the effective vibrational interval of the order 500 cm01 . This argument rests, of course, on the assumption that the 18 01 transition has not gained intensity by vibronic coupling. There are two further arguments we can invoke. The first is the substantial difference between the ab initio frequency for n18 and the observed vibrational interval. Second, we reiterate the point made earlier, namely that because of the metal-localized nature of the electronic transition we would in any case expect some Franck–Condon activity in the Zn– C–C bend and there are no alternative candidates in the spectrum. Thus we can confidently assign the 0180 cm01 band to the Zn–C–C bending fundamental. The more intense of the two low frequency bands, that at 0387 cm01 , is assigned to one quantum in the Zn–C stretch. The second member of the stretching progression, although quite weak, can also be identified in the spectrum, as shown in Fig. 3. As in the case of the skeletal bend, the experimental frequency for the Zn–C stretch is quite close to the ab initio prediction (see Table 1). There are other, weaker, bands in Fig. 3 which cannot be attributed to either the Zn–C stretch or Zn–C–C bend or their combinations. In particular, there is a trio of bands covering the range 900–1100 cm01 . Of the modes listed in Table 1, n7 , n8 , and n9 are all totally symmetric and may be responsible for these bands. We are unable to make a firm assignment, but give very tentative assignments in Fig. 3 and in Table 1 based solely on the frequency order predicted from the ab initio calculations. Implications for the Assignment of the Excitation Spectrum We have recorded dispersed fluorescence spectra by laser excitation of a number of other vibronic transitions in addition to the origin. These data, which are collected together in Figs. 4 and 5, confirm our assignment of the Zn–C stretching and Zn–C–C bending vibrational frequencies. They also provide valuable evidence which can be used to assign several bands in the excitation spectrum. Before considering these, it should be borne in mind, particularly when dealing with bands in close proximity, that substantial vibronic interactions are likely, especially in the upper vibronic states. We are forced, by the limited resolution of our experiments, to adopt what may sometimes be simplistic assignments in terms of the dominant vibronic component. We now deal with each band in turn. 22 760 cm01 . Figure 4a shows the dispersed fluorescence spectrum obtained by pumping the transition at 22 760 cm01 in the excitation spectrum. Two possible assignments for this transition were suggested in our earlier work (16), namely that it is due either to the origin transition of a second (B˜ –X˜ ) system or to excitation of the Zn–C–C bend in the A˜ state. The dispersed fluorescence spectrum in Fig. 4a

FIG. 4. Dispersed fluorescence spectra obtained by laser excitation of transitions at (a) 22 760, (b) 22 939, and (c) 22 990 cm01 . See text for details of the assignments of the excitation bands.

shows that the latter assignment is the correct one since we see a dramatic enhancement in one quantum of the bend in the ground electronic state. We can therefore locate the 11 1 level at 245 cm01 above the zero point level in the A˜ state. 22 939 and 22 951 cm01 . The strong transition at 22 939 cm01 yields a dispersed fluorescence spectrum (Fig. 4b) which appears to be dominated by a progression in n10 . The strongest band, which we assign to 10 11 , could alternatively be due to 11 22 . However, the latter assignment is unlikely given the substantial intensity of a second band at 0758 cm01 , which nicely fits as the second member in the stretching progression. Consequently, the 22 939 cm01 transition is assigned to the excitation of one quantum of the Zn–C stretch in the A˜ –X˜ manifold. The transition slightly further to the blue in the excitation spectrum, that at 22 951 cm01 , appears almost as a shoulder on the stronger 22 939 cm01 band. The dispersed fluorescence spectrum obtained by pumping this band is indistinguishable from that shown in Fig. 5a and so is not shown. Presumably the similarity is caused by virtually complete collisional depopulation of the excited vibronic level to the nearby 10 1 level. Consequently we cannot currently assign the 22 951 cm01 band. 22 990 cm01 . The dispersed fluorescence spectrum shown in Fig. 4c was obtained by exciting the 22 990 cm01 transition. At first sight it appears to be similar to that in Fig. 4b, but in fact there are two important differences. First, a substantially weaker band is observed in the 2n10 region in Fig. 4b. Even more significantly, the strongest band is shifted some 15 cm01 to the blue of the strongest band in Fig. 4a. Both of these factors indicate that the strongest feature in Fig. 4b is not due to excitation of the Zn–C stretch and so we must find an alternative assignment. Using the frequency of the Zn–C–C bend, n11 , in the ground electronic state deduced earlier, 2n11 would be expected at Ç360 cm01 , a value very close to the observed shift ( 0365 cm01 ) from

Copyright q 1997 by Academic Press

AID

JMS 7361

/

6t1f$$$102

09-09-97 09:49:01

51

mspal

52

POOLEY AND ELLIS

assignment to a combination transition, the 11 1010 10 transition. The dispersed fluorescence spectrum shows prominent excitation of one quantum of the Zn–C–C bending vibration and its combination with one quantum in the Zn–C stretch.

FIG. 5. Dispersed fluorescence spectra obtained by laser excitation of transitions at (a) 23 032, (b) 23 141, and (c) 23 526 cm01 . See text for details of the assignments of the excitation bands.

the excitation wavelength in Fig. 4b. Thus the main peak in the dispersed fluorescence spectrum is assigned to excitation of 2n11 and the 22 990 cm01 band is assigned to the A˜ –X˜ 11 20 transition. This is consistent with the 11 10 assignment made earlier since it would give a 11 2 –11 1 separation of 230 cm01 , slightly smaller than the 11 1 –11 0 value of 245 cm01 . As shown in Fig. 1, a plausible extrapolation of the n11 progression in the excitation spectrum to the next member, 11 30 , is also possible. Before we leave the 22 990 cm01 band, it should be noted that expanded scans show that it is clearly broader than the surrounding bands in the excitation spectrum. Thus while the A˜ –X˜ 11 20 transition may be the dominant contributor, there must be at least one further underlying transition. Our current data are insufficient to assign this additional transition. 23 032 cm01 . The dispersed fluorescence spectrum obtained by laser excitation of the 23 032 cm01 transition is shown in Fig. 5a. The prominent bands are readily assignable in terms of structure in n11 and n10 . Interestingly, the strongest band is at the laser excitation wavelength, more than 85% of which is genuine emission rather than scattered laser light (established by tuning the laser to a nearby off-resonance wavelength and re-recording the spectrum). Given that the most intense emission is back to the zero point level in the X˜ state, this suggests that the 23 032 cm01 transition may not involve a vibrationally excited upper electronic state and might instead be an electronic origin, presumably the B˜ –X˜ origin. However, the observed vibrational structure, which shows apparent excitation of 2n11 and n11 / n10 , is difficult to explain if the B˜ –X˜ 0 00 assignment is correct. Clearly this assignment should be regarded as tentative. 23 141 cm01 . The 23 141 cm01 band is a fairly weak band in the laser excitation spectrum of ZnC2H5 . The corresponding dispersed fluorescence spectrum, shown in Fig. 5b, is consistent with the 23 141 cm01 band being due to

23 526 cm01 . The final dispersed fluorescence spectrum we consider is that shown in Fig. 5c, which was obtained by excitation of the medium-intensity feature seen in the excitation spectrum at 23 526 cm01 . Excitation of this transition, which is 1011 cm01 above the A˜ –X˜ origin, gives rise to a strong band at 0985 cm01 in the dispersed fluorescence spectrum together with weaker features which are easily attributable to excitation of the Zn–C–C bend and Zn–C stretch. The 985 and 1011 cm01 intervals presumably represent ground and excited state fundamentals in a particular mode. Furthermore, because of the considerable intensity of the 23 526 cm01 band in the excitation spectrum, there is clear evidence for a significant change in equilibrium geometry in the direction of the normal coordinate. From the ab initio calculations there are three totally symmetric modes which are contenders, two of which, the C–C stretch ( n8 ) and the CH2 /CH3 wag ( n7 ), have predicted frequencies particularly close to the experimental values. Unfortunately, although the band is labeled as 8 10 in Fig. 1, there is no convincing evidence to favor n8 over n7 . Further work, perhaps using deuterated monoethyl zinc, is necessary to establish a firm assignment. Bonding and Structure We can draw a number of conclusions about the bonding and structure in ZnC2H5 from the combination of dispersed fluorescence data and new assignments of excitation transitions obtained in this work. We have established vibrational frequencies of the Zn–C stretching and Zn–C–C bending modes in both X˜ and A˜ states. As can be seen from Table 1, the stretching frequency increases from 387 to 424 cm01 on excitation from the X˜ to the A˜ state. The same trend has previously been observed for ZnCH3 , the frequency rising from 445 cm01 in the X˜ 2 A1 state to 460 cm01 in the A˜ 2E1/ 2 state (6). The lowering of the Zn–C stretching frequency in moving from ZnCH3 to ZnC2H5 is consistent with the increase in reduced mass being the dominant cause (in a crude calculation in which the methyl or ethyl group is treated as a point mass one expects a lowering of about 90 cm01 and we observe an actual lowering of 60 cm01 ). We are therefore led to the conclusion that the Zn–C force constants must therefore be, not surprisingly, very similar and this ties in with the known similarity in Zn–C bond dissociation enthalpies for ZnCH3 and ZnC2H5 (19). The observation of a progression in n11 indicates that there may be a significant change in the Zn–C–C bond angle on electronic excitation to the A˜ state. However, caution is warranted in drawing this conclusion since there will inevitably be some coupling of the Zn–C stretching and Zn–C– C bending motions. Thus, while the increase in frequency

Copyright q 1997 by Academic Press

AID

JMS 7361

/

6t1f$$$103

09-09-97 09:49:01

mspal

53

DISPERSED FLUORESCENCE OF ZnC2H5

of n11 from 180 to 245 cm01 on electronic excitation does suggest a stiffening in the bending coordinate, part of the frequency increase, and the Franck–Condon activity, may simply result from the inclusion of Zn–C stretching character in the mode. A rotationally resolved study will be necessary to quantify the effect of electronic excitation on the molecular geometry. CONCLUSIONS

The dispersed fluorescence spectra recorded in the present work have allowed us to obtain ground electronic state vibrational frequencies of ZnC2H5 . In addition these data have allowed us to make assignments, some firm and others tentative, of several bands in the excitation spectrum. The vibrational data that we now possess provide some information on the bonding and structural changes on electronic excitation. However, it should be possible to obtain far more detailed information from rotationally resolved spectra, including an assignment of the upper state symmetry for each band. Such spectra have recently been recorded in the laboratory of T. A. Miller and analysis of these is now in progress (20). ACKNOWLEDGMENTS The authors are grateful for financial support for this work from the UK Engineering and Physical Sciences Research Council (Grant K05399) and the University of Leicester (studentship for S.J.P.).

REFERENCES 1. C. F. Yu, F. Youngs, K. Tsukiyama, R. Bersohn, and J. Preses, J. Chem. Phys. 85, 1382 (1986). 2. A. Amirav and A. Penner, J. Chem. Phys. 90, 5232 (1989). 3. T. Ibuki, A. Hiraya, and K. Shobatake, J. Chem. Phys. 92, 2797 (1990). 4. R. L. Jackson, Chem. Phys. Lett. 174, 53 (1990). 5. R. L. Jackson, J. Chem. Phys. 92, 807 (1990). 6. E. S. J. Robles, A. M. Ellis, and T. A. Miller, Chem. Phys. Lett. 178, 185 (1991). 7. A. Penner, A. Amirav, and R. Bersohn, Chem. Phys. Lett. 176, 147 (1991). 8. A. M. Ellis, E. S. J. Robles, and T. A. Miller, Chem. Phys. Lett. 190, 599 (1992). 9. T. M. Cerny, X. Q. Tan, J. M. Williamson, E. S. J. Robles, A. M. Ellis, and T. A. Miller, J. Chem. Phys. 99, 9376 (1993). 10. X. Q. Tan, T. M. Cerny, J. M. Williamson, and T. A. Miller, J. Chem. Phys. 101, 6396 (1994). 11. R. M. Osgood, Annu. Rev. Phys. Chem. 34, 77 (1983). 12. C. Jonah, P. Chandra, and R. Bersohn, J. Chem. Phys. 55, 1903 (1971). 13. R. L. Jackson, J. Chem. Phys. 96, 5938 (1992). 14. J. A. Elias, P. J. Wisoff, and W. L. Wilson, J. Appl. Phys. 74, 6962 (1993). 15. P. F. Seidler, J. Phys. Chem. 98, 2095 (1994). 16. I. M. Povey, A. J. Bezant, G. K. Corlett, and A. M. Ellis, J. Phys. Chem. 98, 10427 (1994). 17. A. J. Bezant, D. D. Turner, G. Dormer, and A. M. Ellis, J. Chem. Soc., Faraday Trans. 92, 3023 (1996). 18. C. Jamorski and A. Dargelos, Chem. Phys. 164, 191 (1992). 19. R. L. Jackson, Chem. Phys. Lett. 163, 315 (1989). 20. S. J. Pooley, A. M. Ellis, C. Carter, J. M. Williamson, S. I. Panov, and T. A. Miller, work in progress.

Copyright q 1997 by Academic Press

AID

JMS 7361

/

6t1f$$$104

09-09-97 09:49:01

mspal