Delayed electron emission from higher fullerenes Cn, n .ltoreq. 96

Delayed electron emission from higher fullerenes Cn, n .ltoreq. 96, following excitation at 1064 nm. Keith W. Kennedy, and Olof Echt. J. Phys. Chem. ,...
0 downloads 0 Views 506KB Size
J. Phys. Chem. 1993,97, 7088-7091

7088

Delayed Electron Emission from Higher Fullerenes C,, (n 5 96) following Excitation at 1064 nm Keith W. Kennedy and Olof Echt’ Physics Department, University of New Hampshire, Durham, New Hampshire 03824-3568 Received: March 30. 1993

Excitation of gas-phase fullerenes with laser pulses at 1064 nm yields large amounts of higher fullerene ions C,+, 72 5 n I96, even though their neutral precursors are only minor contaminants in the vapor over the toluene extract of soot. The ions are particularly prominent in the size distribution of delayed ions recorded at low fluence. Comparison with data obtained a t 532 nm suggests that the size-dependent variations in the onset of photoabsorption are responsible for the observed phenomena. We propose that delayed ionization may be utilized to distinguish between higher fullerenes and other, more loosely bound, carbon aggregates.

Introduction Recently, there has been a surge of interest in delayed ionization of large molecules and clusters in the gas phase.l Refractory metal clusters as small as Ta4 or Ws exhibit delayed ionization several microseconds after multiphoton excitation in the visible or near UV.293 c 6 0 and C ~yield O copious amounts of ions as late as 100 ps after photoexcitation in the wavelength range 212-532 nm.68 Similar phenomena have been observed for collisionally excited Cm,9JO for c60- following collisional excitation or photoexcitation,IlJ and for very large positively charged fullerenes.13 Recently, we could show that all even-sized fragments of the “magic” fullerenes c 6 0 and c70 undergo delayed ionization, for sizes n 1 36.7 Most authors have viewed these processes as the molecular analogue of thermionicemission; i.e., it is assumed that thermally activated electron emission occurs after complete energy randomi~ation.2-~,9-13Others have questioned the assumption that the energy is fully equipartioned between the electron and nuclear degrees of freedom, at least as far as Cm and c 7 0 are concerned.l.*J4J5 In this contribution, we report on another remarkable observation pertaining to fullerenes: following low-fluence photoexcitation in the IR a t 1064 nm, the yield of delayed cluster ions consists predominantly of higher fullerenes 72 In I96,although their neutral precursors are only minor contaminants in the vapor phase above the toluene-soluble soot extract. The laser fluence dependence of the delayed ion yield is of very low order for some of the higher fullerenes, even though no less than roughly 30 photons need to be absorbed by their neutral precursors. These results prove that c60 and C70 are by no means special as far as delayed ionization is concerned. We also point out a possible application of our findings: if large carbon clusters are produced in the gas phase by, e.g., clustercluster collisions, it is important to distinguish between genuine fullerenes and other, more loosely bound, aggregates. A search for delayed ions after photoexcitation a t 1064 nm will accomplish that task. Experimental Section The fullerene sample, prepared by toluene extraction from soot formed by the Kratschmer-Huffman method,I6 was obtained from Polygon Enterprises. It is placed into a resistively heated copper crucible with a l-mm orifice and outgassed at about 200 OC for several hours in a vacuum of 2 X 10-7 Torr. During the experiment, the oven temperature ranges from 480 to 600 OC, corresponding to equilibrium vapor pressures of about 0.3-1 2 mTorr for c60 and slightly less for C70.1~We have also used a

* To whom correspondence should be addressed.

quartz microbalance to determine the flux of particles emerging from the oven. As a result, we find that the number density of particles at the point of photoexcitation is 2 X 1O1Ocm-3, for a temperatureof 600O C . Taken together with the fact that similar findings are obtained at reduced oven temperature, we conclude that ion-molecule reactions do not play a significant role in the present study. The molecular beam emerging from this effusive source is collimated and intersected by light from a mildly focused (beam diameter 1.1 mm) Q-switched Nd:YAG laser in the first gap of a Wiley-McLaren ion source.l* The laser operates at any one of its harmonics (i.e., at 1064,532,355,or 266 nm), and the pulse duration is 5-8 ns, with a repetition rate of 50 Hz. The energy per laser pulse is controlled by varying the time delay between the flash-lamp pulse and the Q-switch trigger. The time-offlight mass spectrometer, whose axis is collinear with the molecular beam, may be operated in the conventional mode, with static extraction fields. In this case, the tail of a mass peak immediately reveals the presence of delayed i0ns.3.~ However, these “prompt” spectra are difficult to analyze if more than one delayed ion species is present. Alternatively, the potentials applied to the two plates which define the extraction region may be switched rapidly by up to 8 kV, such that only those ions that are formed within a welldefined time window contribute to the mass peaks in the “delayed” mass spectra. This technique is a modification of the doublepulse technique introduced by Leisner et al.2 A detailed description of our version has been published elsewhere.6 In short, all delayed ions of a given mass, formed during the sampling time [tbl, rex], will arrive simultaneously at the detector. Ions formed earlier are totally rejected. Ions formed after t,, (measured with respect to the laser pulse) will still arrive at the detector, but with a corresponding delay. The strength of the method stems from the fact that delayed ions of different masses are no longer superimposed in the timeof-flight spectra. We have succeeded in detecting all even-sized fragments of Cm and c 7 0 as small as c 3 6 in delayed spectra, and delayed formation of C a + after photoexcitation at 266 nm has been traced over 100 p s . 6 ~ ~ Results Figure 1 presents three “prompt” time-of-flight mass spectra, recorded with static extraction fields. The top two spectra were recorded at 532nm but different pulse energies, while the bottom spectrum was recorded at 1064nm. The oven temperature ranged from 480 to 600 OC,but from previous experiments7 we know that this parameter does not significantly affect the relative intensities in the spectra. Also note that all spectra are individually normalized.

0022-365419312097-7088%04.00/0 0 1993 American Chemical Society

Delayed Electron Emission from Higher Fullerenes

The Journal of Physical Chemistry, Vol. 97, No. 27, 1993 7089 532 nm 65 mJ [2, 4 P

78 h

I

a4

h

v)

.-c

v)

.e B

C

h I\

x .-c v)

532 nm 0.5 mJ

1

3

I

C

a

.-C C .-0

a4

A d b & 60

20

25

30

96

35

time of flight (ps)

Figure 1. ‘Prompt” time-of-flight mass spectra, recorded with static

extraction field. Photon wavelengths and energies per laser pulse are indicated in the figure; oven temperature ranges from 480 OC (middle spectrum) to about 600 OC (top and bottom spectra). Figure 1 demonstrates a striking result: whereas fullerenes larger than C70+ are barely discernible at 532 nm (nor a t 355 or 266 nm), they are easily identified at 1064 nm in “prompt” spectra up to Cs4+,and with some effort up to n = 92. We also note that low-fluence irradiation at 532 nm strongly favors C70+ versus Cm+, but the higher fullerenes are not enhanced. Figure 2 presents delayed mass spectra, recorded with a sampling window of [2,4 ps], at 532 nm with high and low fluence (top and middle spectra, respectively) and a t 1064 nm with relatively low fluence (bottom). These spectra were recorded shortly after each other and with an identical oven temperature of 600 O C ; hence, we are sure they all sample exactly the same vapor phase. By and large, these delayed spectra exhibit similar features as the prompt spectra: fullerenes larger than C70+ are barely discernible at 532, 355, or 266 nm, and low fluence a t 532 nm strongly favors C70+, but none of the higher fullerenes, relative to Cm+. Most strikingly, however, under low-fluence excitation at 1064 nm we observe copious amounts of higher fullerene ions ranging up to c96+. In fact, the intensity of c78+ exceeds that of C70+, and all the higher fullerenes are at least as intense as Cm+. None of our “prompt” spectra recorded a t 1064 nm show the fullerenes beyond C84+ with such clarity. Note that all spectra shown in Figure 2 are averaged over approximately 2000 laser shots. At increased sensitivity we are able to identify clusters up to C84+ in the top spectrum (532 nm, high fluence, see trace with enhanced sensitivity), while the larger ones, if present, are buried in an intense background. Reduced fluence at 532 nmdoes improve the signal-to-background (middle spectrum), but the total intensity decreases; only C78+ can be identified under these conditions. A cautionary note may be in order here: we cannot determine the neutral precursors of the observed ions, but the toluene extract of soot is known to contain, a t least, c76, C84, CW,and c94.I9

time of flight (ps)

Figure 2. “Delayed“ time-of-flight mass spectra, sampling ions formed in the time window 2 ps I t I 4 ps after the laser pulse. Wavelengths

and energiesper pulse areindicated,all other conditions are kept identical. Oven temperature is 600 O C .

h

.--w

v)

10‘

3

e

W 92

A 4 04

v

c3a

7a ti 70 H 60

0

.-ex, io5

.-0

.

d

1 Figure 3. Yield of delayed ions Cn+,sampled in the window 2 ps I t I 4 ps, versus laser energy per pulse, for cluster size n = 60,70,78,84,92. Photon wavelength is 1064 nm, oven temperature is 600 OC.

Other fullerenes (C78, c82, and Cga)have been identified using CS2 as a solvent.20 This compares reasonably well with the increased abundance of c78, (284, and perhaps C92 and the abrupt termination of the fullerene series a t (296 in our spectra. Similar features have also been observed in electron attachment and electron impact ionization studiesz1 Figure 3 presents the ion yield of some representative fullerenes, sampled within the window of [2,4 ps] after photoexcitation a t 1064 nm, as a function of laser pulse energy. The fluences range from 1 to 6 J/cm2. The ion yield was obtained by integrating

7090

Kennedy and Echt

The Journal of Physical Chemistry, Vol. 97, No. 27, 1993 1064 nm 30 mJ

100

h

-w

a,

.--N

0

E0 v

2! .-a, x

A-A a4 Q-C ?a H 70 >t)( 66 H 60

c

.-0 lo-'

i-t

2

56

4

6

a

10

delay ( F ) Figure 4. Yield of delayed ions Cn+,sampled over a window of 2-ps duration, versus delay, for cluster sizes n = 56, 60, 66, 70, 78, 84. All curves are individually normalized at 2 ps. Photon wavelength is 1064 nm, oven temperature is 600 OC.

over the appropriate peak areas in the (delayed) time-of-flight mass spectra. The data reveal that, at low fluence, the yield of all the higher fullerenes exceeds that of Cm+. This is related to their unusually low fluence dependence. Taking just the first two data points for each size, we obtain slopes of about 7 for Cm, 4 for c78, 2 for C84, and less than 1 for (292. We note, though, that this study has shortcomings: the fluenceis not uniform across the (Gaussian-like) laser beam profile, and more measurements at lower fluence would have been desirable. The yields of c60t and C70+ indicate that saturation occurs: the yields actually start to decrease, probably due to increased fragmentation of the neutrals into c 6 8 , C58, etc. Finally, in Figure 4, we plot the time dependence of the ion yield of various clusters, following photoexcitation at 1064 nm with 30 mJ per laser pulse. Each data point represents the yield integrated over a sampling window of 2-ps duration; hence these curves do suffer from some distortion (they would exhibit more curvature if sampled with a narrower window).7 Still, the trend is obvious: the ion yield of the higher fullerenes decays slightly less rapidly than that of c 6 0 while it decays even more rapidly for fragments of c 6 0 and C70. This latter observation is in agreement with an earlier study performed at 266 nm?

Discussion The delayed yield of higher fullerene ions exceeds that of C60' if the gas-phase sample is irradiated at 1064 nm with low laser fluence (see Figure 2, bottom spectrum, and Figure 3). In contrast, these species are barely discernible under irradiation at 532,355, or 266 nm,6,' or in electron impact ionization ~ p e c t r a . t ~These .~~ latter results are in agreement with the notion that the vapor phase over toluene-extractable soot is composed almost entirely of C6o and C70, although the extract itself probably contains a few percent of higher fullerenes ranging up to C96 or c98.19'24 Hence, the higher neutral fullerenes autoionize much more efficiently than C70 and, in particular, C a r if irradiated at 1064 nm at low fluence. Before discussing the possible origin of these findings, it is helpful to summarize the behavior of Cm, the only fullerene for which delayed ionization has been thoroughly studied so far.4-8.15 Photoexcitation of c 6 0 with laser pulses of about 10% duration25 and wavelengths ranging from 212 to 532 nm results in large yields of delayed Cm+. The nonexponentially decaying

yield has been traced over 100 ~ s . 7 3The fraction of "prompt" Cm+, formed during or shortly after the laser pulse, is small compared to the total yield. The effective rate constant for a major fraction of all observable ions is on the order of IO5 s-l. If the emission process is viewed as vibrational autoionization from fully equilibrated Cm or, in other words, as the analogue of thermionic emission, the corresponding vibrational excitation energy may be estimated from statistical theories to be about 40 eV, corresponding to an effective temperature of slightly less than 4000 K.13J4.26 This requires an absorption of about 30 photons at 1064 nm, if the oven temperature is 600 OC. The assumption of complete equilibration has been questioned, with very general arguments, and with arguments specific to Cm and C,0.1914,15 One such specific argument is related to the existence of a long-lived triplet state which is rapidly populated with near-unity quantum efficiency through intersystem crossing from the optically accessible (singlet) ~tates.2~Absorption of further photons would then proceed within the triplet manifold. In this scenario, the activation energy for thermal electron emission, and hence the required excitation energy, would be less than the above estimates by some 20%. Another argument is related to the existence of a competing decay channel, namely, unimolecular fragmentation of C,, into Cw2 C2. The rate of this reaction is expected to be orders of magnitude higher than the rate of ionization if the dissociation energy is less than the ionization energy.lJ4 Some experiments suggest that Cm in its electronic ground state meets this condition,14*2*hence providing further support to the triplet state scenario. However, the dissociation energy of Cm is still a controversial issue.1lJ9-31 These considerations suggest two possible explanations for the phenomena observed in the present study. First, the onset of electronic absorption of Cm and C70 in solution is around 650 nm.22 This onset is shifted more and more into the near-infrared for higher fullerenes (different isomers may feature different onsets); absorption of CW and c 9 6 extends well beyond 1000 nm.1930 Hence, photoexcitation at 1064 nm would be much more efficient for the higher fullerenes, the order of laser fluence dependence might be reduced, and the ion yield would saturate at lower fluences. Figure 3, however, indicates that the yield of higher fullerene ions does not saturate earlier than that of C a and C70. Note that the higher fullerenes probably need to absorb a slightly higher number of photons than Cm in order to exhibit the same emission rate: the ionization energy IE(n) will decrease with increasing n,32 thus reducing the required effective temperature. This will, however, be more than outweighed by the (approximately linear) increase in the specificheat. We estimate that the higher fullerenes need to absorb 3040 photons at 1064 nm if the oven temperature is 600 OC. Again, this estimate assumes full energy equilibration. Second, while IE(n) decreases with increasing n, the cohesive energy E,h(n) per atom increases toward the bulk value of graphite. However, the dissociation rate is controlled by 0, = E,h(n) - E,h(n - 2) - E&,@), Le., by the dissociation energy with respect to loss of C2. According to tight-bindingcalculations, D m significantly exceeds D, for all n up to, at least, n = 94.31 Hence, it is not immediatelyclear in what direction the postulated competition between delayed ionization and dissociation will change within the investigated size range. A significant change in favor of ionization would not only increase the ion yield, but it would also be accompanied by a reduction in the observed effective rate constant. Figure 4 does show such an effect, but it is very small. Note that, in the absence of competing decay channels, the observed rate constant would also tend to decrease because larger clusters offer more degrees of freedom to absorb the excess energy. However, none of the arguments in this paragraph would explain the dramatic differences between irradiation at 1064 and 532 nm.

+

Delayed Electron Emission from Higher Fullerenes

The Journal of Physical Chemistry, Vol. 97, No. 27, 1993 7091

Summarizing this section, the higher fullerenes feature a large (5) Wurz, P.; Lykke, K. R. J. Phys. Chem. 1992,96, 10129. (6) Walder, G.; Echt, 0.In Clusters undFullerent-s; Kumar, V., Martin, yield of delayed ions if excited at 1064 nm. Comparison with T. P., Tosatti, E., Eds.; World Scientific: Singapore, 1992; p 297. data obtained at 532 nm suggests that the size-dependent (7) Walder, G.; Kennedy, K. W.;Echt, 0. Z . Phys. D, in prcse. variations in the onset of photoabsorption are responsible for the (8) Locpfe, M.; Siegmann, H. C.; Sattler, K. Z . Phys. D, in press. (9) Wan, Z.; Christian, J. F.; Anderson, S.L. Phys. Rev. Lerr. 1992,69, observed phenomena. Furthermore, if long-lived triplet states 1352. do indeed play a key role in thermally activated electron emission (10) Yeretzian, C.; Whetten, R. L. Z . Phys. 1992, 024, 199. of C ~ and O C70, then one needs to postulate that a similar process (11) Beck, R. D.;St. John, P.; Alvarez, M. M.; Diederich, F.; Whetten, R. L. J. Phys. Chem. 1991, 95, 8402. enhances theemission rate of higher fullerenes as well. The excited (12) Wang, L.-S.; Conceicao, J.; Jin, C., Smalley, R. E. Chem. Phys. Leu. state needs to be long-lived (no less than -165 s) up to about 1991, 182, 5. 3000 K, where fullerenes are predicted to undergo rapid (13) Maruyama, S.;Lee, M. Y.; Haufler, R. E.; Chai, Y.; Smalley, R. E. isomerization.33 Z . Phys. 1991, 019,409. (14) Sandler, P.; Lifshitz, C.; Klots, C. E. Chem. Phys. Lett. 1992, 200, Finally, we point out a potential application of the phenomena 445. discussed above. Growth of larger carbon clusters has recently (15) Ding, D.; Compton, R. N.; Haufler, R. E.; Klots, C. E. J . Phys. been accomplished in a dense, laser-heated vapor of C ~ and O ~ ~Chem. 1993,97,2500. (16) KrHtschmer, W.; Lamb, L. D.; Fostiropoulos, K.; Huffman, D. R. through energeticC&&collisions.35 In either case, it is essential Nuture 1990,347,354. to demonstrate that the synthesized species are, indeed, tightly (17) Mathews, C. K.; Baba, M. Sai; Narasimhan, T. S. Lakshmi; bound, and not merely van der Waals bound. This can be, and Balasubramanian, R.; Sivaraman, N.; Srinivasan, T. G.; Rao, P. R. Vasudwa J . Phys. Chem. 1992,96,3566. Abrefah, J.; Olander, D. R.; Balooch, M.; has been, done by scattering the clusters off a ~urface,3~ but the Siekhaus, W. J. Appl. Phys.Lert. 1992,60,1313.Pan,C.;Chandrasekharaiah, approach requires an elaborate experimental setup. M. S.;Agan, D.; Hauge, R. H.; Margrave, J. L. J . Phys. Chem. 1992,96, Alternatively, one can make use of the fact that vibrational 6752. (18) Wiley, W. C.;McLaren, I. H. Rev. Sci. Insrrum. 1955,26, 1150. autoionization will be quenched for weakly bound species, as (19) Diederich, F.; Whetten, R. L. Acc. Chem. Res. 1992, 25, 119. explained above. The cohesive energy of fullerite (the condensed Diederich, F.;Ettl, R.; Rubin, Y.; Whetten, R. L.; Beck, R.; Alvarez, M.;Anz, phase of fullerenes) is only 1.5 eV per Cw; the dissociationenergy S.;Sensharma, D.; Wudl, F.; Khemani, K. C.; Koch, A. Science 1991, 252, of the dimer C&a has been calculated to be 0.28 eV,36 which 548. (20) Kikuchi,K.;Nakahara,N.; Wakabayashi,T.;Honda, M.;Matsumiya, is 20 times less than the ionizationenergy! Hence, whereas higher H.;Moriwaki,T.;Su~,S.;Shiromaru,H.;Saito, K.; Yamauchi, K.; Ikcmoto, fullerenes (at least up to n = 96) can be efficiently transformed I.; Achiba, Y. Chem. Phys. Lert. 1992, 188,177. into delayed ions if irradiated at 1064nm, aggregates of fullerenes, (21) Ben-Amotz, D.; Cooks, R. Graham; Dejarme, L.; Gunderson, J. C.; Hoke, 11, S.H.; Kahr, B.; Payne, G. L.; Wood, J. M. Chem. Phys. Lett. 1991, (C,Jm, either would not absorb (if n = 60) or would fragment 183, 149. rapidly without ever reaching the temperature of 3000-4000 K (22) Ajie, H.; Alvarez, M. M.; Anz, S.J.; Beck, R. D.; Diederich, F.; required for delayed ionization. Fostiropoulos, K.; Huffman, D. R.; KrHtschmer, W.; Rubin, Y.; Schriver, K. E.; Sensharma, D.; Whetten, R. L. J. Phys. Chcm. 1990,94, 8630. It remains to be shown experimentally that fullerenes beyond (23) Luffer, D. R.; Schram, K. H. Rup. Comm. Mass Specrrom. 1990, n = 96, vaporized either from raw soot or from suitable soot 4, 552. can, in fact, be efficiently photoexcited at 1064 nm. In (24) Smart, C.; Eldridge, B.; Reuter, W.; Zimmermann, J. A,; Creasy, W. any event, detection of delayed ions in a time-of-flight mass R.; Rivera, N., Ruoff, R. S.Chem. Phys. Lerr. 1992, 188,171. (25) Delayed ionization is absent for sub-picosecond laser excitation at spectrometer is straightforward. We also add that cluster fusion 248 nm: Zhang, Y.; Spiith, M.; KrHtschmer, W.; Stuke, M. Z . Phys. 1992, experiments in the gas phase usually involve cluster cations. In D25, 195. this case, delayed ionization is disfavored with respect to (26) Klots, C. E. Chem. Phys. Letr. 1991, 186,73. (27) Palit, D. K.; Sapre, A. V.; Mittal, J. P.; Rao, C. N. R. Chem. Phys. dissociation due to the higher ionization energy of cations. Lerr. 1992, 195,1. Lee, M.; Song, 0. K.; Seo, J. C.; Kim, D.; Suh, Y.D.; Nevertheless, delayed ionization of carbon cluster cations has Jin.S. M.: Kim. S. K. Chem. Phvs. Lett. 1992.196.325 and references therein. been observed for sizes n = 150 and beyond.13 (28) Radi, P.P.;Hsu, M. f.;Rincon, M: E.; Kempcr, P. R.; Bowers, M.

Acknowledgment. We gratefully acknowledge stimulating correspondence with Dr. C. E. Klots. References and Notes (1) Schlag, E. W.; Levine, R. D. J . Phys. Chem. 1992, 96, 10608. Remacle, F.; Levine, R. D. Phys. Leu. 1993,A173, 284. (2) Leisner, T.; Athanassenas, K.; Kandler, 0.; Krei.de, D.; Recknagel, E.; Echt, 0. Mot. Res. Soc. Symp. Proc. 1991, 206, 259. Leisner, T.; Athanasscnas, K.; Echt, 0.;Kreisle, D.; Recknagel, E. In Clwrer Models for Surfuce and Bulk Phenomenu, NATO AS1 Serica B.; Paccioni, G., Bagus, P. S.,Parmigiani, F., Eds.;Plenum Pres: New York, 1992;Vol. 293,p 51. (3) Amrein,A.;Simpson,R.;Hackett,P.J. Chem.Phys. 1991,95,1781. (4) Campbell, E. E. B.; Ulmer, G.; Hertel, I. V. Phys. Rev. Lett. 1991, 67, 1986;Z . Phys. 1992,D24, 81.

T. Chem. Phys. Lert. 1990,174,223. (29) Foltin, M.; Lezius, M.; Scheier, P.; MHrk, T. D. J. Chem. Phys., in press. (30) Stanton. R. E. J. Phvs. Chem. 1992. 96. 111. (31) Zhang, B.L.;Xu,C.H.; Wang,C. Z.';Chan,C.T.;Ho,K. M.Phys. Rev. 1992,846,7333. (32) The work function of graphite is 5.0eV, while the ionization energy of Cm is 7.6eV: Zimmerman, J. A.; Eyler, J. R.; Bach, S. B. H.; McElvany, S.W. J. Chem. Phys. 1991,94,3556. (33) Jing, X.;Chelikowsky, J. R. Phys. Rev. 1992, B46, 15503. (34) Yeretzian, C.; Hansen, K.; Diederich, F.; Whetten, R. L. Nuture 1992,359, 44. (35) Campbell, E. E. B.; Schyja, V.; Ehlich, R.; Hertel, I. V. Phys. Rev. Lert. 1993,70,263. (36) Girifalco, L. A. J . Phys. Chem. 1992, 96,858. (37) Creasy, W. R.; Zimmerman, J. A.;Ruoff, R. S . J . Phys. Chem. 1993, 97,973.