Pyrolytic production of fullerenes

ELSEVIER

Synthetic Met& 77 (1996) 17-22

Pyrolytic production of fullerenes Colin Crowley a, Roger Taylor a, Harold W. Kt-oto a, David R.M. Walton a,Pei-Chao Cheng b, Lawrence T. Scott b a School of Chemistry and Molecular Sciences, Universiry of Sussex, Brighton, Sussex BNl9QJ, b Merkert Chemistry Center, Boston College, Chestnut Hill, MA 02167-3860, USA

UK

Abstract Recently,we haveshownthat [ 601fullerene(C,,) may beproducedby pyrolysisof the bowl-shaped corannulene moleculeundera wide rangeof conditions.Both naphthalene andits dimerbenzo[k] fluoranthene maybe employedasprecursorsto CeO. Keywords:

Fullerene; Pyrolysis

1. Introduction As result of the 1985 discovery of the fullerene series by Kroto et al. [ 11, and their subsequent generation in macroscopic quantities by Kratschmer et al. [2] in 1990, much attention has been focused on the exploitation of the properties of these novel molecules. While the ‘classical ’ methods of fullerene generation have proved to be extremely facile, they do suffer from a number of drawbacks. The most important of these is the fact that a mixture of fullerenes is produced, and separation into individual components is a labour-intensive process. If a process that was tuneable to produce a specific fullerene could be developed then this might open up the world of fullerene science to a wider community and significantly reduce the costs of fullerene production. With these aims in mind a pyrolytic route to fullerene generation was sought. Polycyclic aromatic hydrocarbons (PAHs) have long been suspected to be precursors of fullerenes [ 31. Grosser and Hirsch [4] have detected fullerene-related PAH in the arc generator where the inert carrier gas has been poisoned or doped with a hydrogen- or chlorine-containing species. The work carried out at Sussex has been concentrated on two major research fronts: 1. The production of Cd,, and higher fullerenes by the pyrolysis of relevant PAH. 2. Elucidation of the mechanism by which fullerenes are formed in these processes, using chromatographic and spectroscopic techniques. To date, both C,, and CT0have been generated successfully by the pyrolysis of three PAHs (and some halogenated deriv0379~6779/96/$15.00 0 1996 Elsevier Science S.A. All rights reserved

atives). The most widely studied molecule of the three has been naphthalene. This was successfully shown to be a fullerene precursor by Taylor et al. [ 51 in 1993. We now show that corannulene and benzo [ k] fluoranthene are also fullerene precursors [6]. Several variables have been considered in

this work and, in particular, the effect of transition metal catalysis on the process of fullerene formation has been tested.

2. Experimental 2.1. Naphthalene, I-bromonaphthalene, l-chloronaphthaleneand benzo[k]@oranthene

Approximately 1 g of sample (50 mg of benzo[klfluoranthene) was placed in the sample well of a quartz pyrolysis tube (300 X 25 mm) situated in a programmable Carbolite tube furnace (Fig. 1). The apparatus consists of a glass spiral cooled in a solid CO* trap and four acetone bubblers to scrub the exhaust gas of any residual PAH. Argon was passed through the apparatus at a rate of 15 mlimin (as measured on a flowmeter in front of the apparatus) and the furnace brought up to operating temperature (600-1200 “C) . Where transition metal catalysis was employed, approximately 0.5 g of, the finely divided metal powder was spread along the heated region of the quartz tube. The sample was introduced into the heated zone over a period of lo-15 min by gently volatilizing with a hand-held butane-air blowtorch. Heating was discontinued when all of the sample had been introduced. The greyiblack exit vapours were condensed in

18

C. Crowley

et al. /Synthetic

Metals

77 (1996)

17-22

the solid carbon-dioxide-cooled spiral, or were dissolved in the acetone bubblers resulting in a coloured solution, On cooling to room temperature all glassware was ultrasonicated with acetone, filtered and washed with carbon disulfide. Generally, both solutions showed a reddish brown colour with very intense blue-green fluorescence. Samples of all solutions were evaporated to dryness and analysed by electron impact mass spectrometry.

pjanzTubekd22mm.)

2.2. Coranndene

Fig. 1, Experimental apparatus.

The details of the synthesis of corannulene have been dealt with elsewhere [7] and will not be repeated here. Approximately 1 mg of corannulene was placed in a 9 mm i.d. quartz tube, and the tube evacuated to a pressure of 0.05 mmHg. The tube was filled with argon gas and re-evacuated. The evacuation-fill cycle was repeated three times in order to remove residual oxygen. The system was allowed to equilibrate and maintained at 0.05 mmHg for 5 min before sealing with a butane-oxygen gas torch. The sealed ampoule was l.OES (4

9.714 9.2E4 6.7E4 8.2E4 7.lE4 7.1E4 6.6E4 6.1E4 5.684 5.1E4 4.6E4 4.1E4 3.6E4 LlE4

2.3El @I

2.1E7 2.OE7

7

1.9E7 l.BE? 1.7E7 1 1.6E7 1.5E7 1.4E.l 1.2E7 1.1E7 l.OE7 9.OE.5 7.9~6 6.8~6 5.6~6 4.5C6 3.4E6 2.3E.5 l.lE6

Fig. 2. Mass spectra for (a) naphthalene pyrolysis and (b) naphthalene/nickel pyrolysis at 1000 “C (note: increase in intensity of CM signal relative to the spectrum of the uncatalysed work above).

C. Crorvley

et al. /Synthetic

Metals

77 (1996)

17-22

19

2.3. Product analysis

Extracts from both the naphthalene and benzo[k] fluoranthene pyrolyses were Soxhlet extracted with HPLC grade acetonitrile, before concentration to approximately 2 ml and filtration through a 2 p,rn ashless filter. Separation conditions consisted of a Spherisorb 5 p,rn PAH column (250 X 10 mm), 500 p,l sample loop and various acetonitrile/ water gradients. The chromatography system used was a Gilson programmable semi-preparative, two-solvent system with fixed wavelength UV detector. A detection wavelength of 254 nm was chosen for all analysis work. Mass spectrometry was carried out on a VG Autospec instrument under electron impact conditions (70 eV) . Ultraviolet spectra were recorded on a Phillips PU 8720 scanning UV-Vis spectrophotometer. All compounds were identified by comparison with authentic samples and published spectra. All chemicals and solvents were used as supplied by Aldrich and Acres Chemica.

(b)

Fig. 3. (a) Benzo[k]fluoranthene; benzo [ I] aceanthrylene.

(b) tribenzo[c,io]triphenylene;

(c)

3. Discussion

placed in a 300X25 mm tube furnace and heated to the desired temperature (6OC-1200 “C). The ampoule was held at the chosen temperature before being rapidly cooled to room temperature. The ampoule contents were analysed by electron impact mass spectrometry.

Assignment1 1000 oC

1. 2.

of maJor (acetonitrile

peaks for naphthalane extracts).

Assignments mad@ by mart spectrometry spectra and rattntlon Umw far authsnlic Cannot be assigned to a dibenzoblphenylane

pyrolysis

3.1. Naphthalene and its l-halo derivatives

The pyrolysis of naphthalene has been studied for almost a century. Taylor et al. [5] showed that the pyrolysis of naphthalene in a silica tube at 1000 “C under an inert atmos-

at

and comparison with the U.V. sample. or benzoscesnthrylene species. n

-

1

2

34

5

6

\

7

8

9

10 12

14

11 13 Fig. 4. HPLC chromatogram for major acetonitrile-soluble products of naphthalene pyrolysis. Conditions: 43-100% MeCN, 60 min, flow 4 mlimin. A = 254 nm.

20

C. Crowley

et al. /Synthetic

Metals

77 (1996) 17-22

5.4%

(4

1

5.1E6 4.8E6 4.6~6 4.3E.5 4.OEb 3.8~6 3.5E6 3.286 3.OE6 2.766 2.4E6 2.1E6 1.9E6 1.6E6 1.3E6 l.lE6

720

8.1ES 5.4ES 2.7E5

(b) Fig. 5. (a) Mass spectrum for benzo [k] fluoranthene pyrolysis at 900 “C. (b) Proposed trimerization of benzo [kl fluoranthene (or an isomeric compound).

phere produces C60. We have shown (Fig. 2(a) ) that C,, is formed by stepwise condensation of C,, fragments, via a host of stable PAH with 10n (II = 2-6) carbon atoms and hydrogenated perimeters. Badger et al. [S] have shown that the pyrolysis of naphthalene proceeds via naphthalene radicals which readily combine, dehydrogenate and condense to give larger PAH, i.e., Ci0H8 * CZ0Hi4 * CZ0Hi2, etc. With respect to fullerene formation it has been suggested that benzo [ k] fluoranthene (dimeric naphthalene, Fig. 3(a) ) is likely to be a key intermediate. Volhardt [ 91 suggested that the last planar precursor to Cc0 is likely to be tribenzo [ c&o] triphenylene (trimeric naphthalene, Fig. 3 (b) ) . This molecule should dehydrogenate to form a bowl-shaped semi-buckminsterfullerene molecule, which would dehydrogenate and dimerize to form CcO. However, Erickson and Milliken [ lo] have recently indicated that attempts to induce this transformation by laser pyrolysis have proved unsuccessful. The observation of species corresponding to tetramerit and pentameric naphthalene in various stages of condensation, and also that of CT0 by Taylor in the mass spectra of the pyrolysis products is indicative of a patchwork mode of formation, in addition to any possible dimerization or trimerization of a particular fragment in the formation of C60. We have found that over the range studied (600-1200 “C), 1000 “C is the optimum temperature for C,, formation. A typical mass spectrum for naphthalene pyrolysis is shown in Fig. 2(a). While no direct attempts weremade to determine

the exact yields of fullerenes obtained, it is expected that yields are typically of the order of 1%. In an attempt to improve the yields of CeOwe have pyrolysed both the I-chloro and l-bromo derivatives of naphthalene. In theory these should give a higher yield of CGOas a result of the ease in breaking either a C-Cl bond (A E = - 34 kJ/mol) or a C-Br bond (AE= - 40 kJ/mol), compared to breaking a C-H bond. In both cases, we have observed C6a in the mass spectrum, but the yields appear to be lower. It is likely that this is due to the increased ease of formation of perylenes and perylene derivatives by Scholl 1,1’:8,8’ condensation reactions, competing with the 1,2’ reactions which are conducive to fullerene formation. The weakness of the C-Br bond (E= 289 kJ/mol) makes octabromonaphthalene the ideal starting material. We have attempted to form endohedral derivatives of C,, by conducting the pyrolysis in the presence of finely divided transition metals, as well as rarer metals such as palladium. While no endohedral derivatives were formed, higher yields of CeO(based on comparisons of total ion currents) were obtained when nickel powder or a 10% dispersion of palladium on charcoal was used. The latter is likely to be a direct result of well-known dehydrogenation procedures, but the increased yields obtained from use of nickel (Fig. 2(b) ) have yet to be adequately explained. In an attempt to identify, isolate and analyse the key intermediates in the fullerene formation process, the combined acetone and carbon disulfide extracts were evaporated to dry-

C. Crowley

et al. /Synthetic

-x20.00

Metals

77 (1996)

17-22

21

q

4 .SE6 4.3E.s

(4

4.OE6 3.8.x.5 3.6E6 3.4E6 3.1E6 1.9E6 2.7E6 ?.5E6 i.2E6 !."E6 .8E6 .6E6 .3E6 lE6 ,OE5 .7E5

5 g&

GO ,I,/ ,j,,/,,j,,, ,,,, ,,,, ,,,, ,,,, ,,,, ,,,, ,,,, ,,,, ,,,, ,,,,

0

2 0

460

660

8bO

10'00

1200

14100

16100

1dOO

2&O

2ziOO

24'00

26100

28100

879 La@C6o

,O

.SES ,2E5 .OEO m/L .l.lE4 .1,1E4 .l.OE4 -9.483 .8.9E3 .8.3E3 .7.8E3 .7.2E3 .6.7E3 .6.1E3 .5.6E3 .S.OE3 .4.4E3 .3.9E3 .3.323 .2.883 .2.2E3 .1.7E3 .l.lE3 .5.6E2 .O.OEO m/r

Fig. 6. Mass spectra for (a) corannulene pyrolysis at 1200 “C and (b) corannulene/lanthanum oxide pyrolysis.

ness and Soxhlet fractionated according to solubility in the following solvents: acetonitrile, ethyl acetate, dichloromethane and hexane. Separation of the various components of the acetonitrile extracts by reverse phase HPLC resulted in the isolation of 37 compounds. A typical chromatogram of the major species is shown in Fig. 4. It is possible that the other fractions will also yield a similar number of compounds. To date it has not been possible to identify the majority of these compounds due to the scarcity of reference compounds, reference spectra and the extremely small sample sizes intermediate/precursor is involved. One possible benzo [ I] aceanthrylene (Fig. 3 (c) ) , resulting from a thermal isomerization of a dimeric naphthalene species, which is analogous to that observed in fluoranthenes by Scott and Roelofs [ 111. This may account for the currently unidentified ion at mlz = 252 a.m.u. Whilst we have as yet been unable to identify all components, a significant advance on previous work has been made. Lang et al. [ 121 identified only eight components and suggested the structure for two more. It is note-

worthy that none of the four trimers identified or suggested in this work are conducive to the formation of the most stable isomer of C6e. 3.2. Benzo[k]Juoranthene

Benzo [ k] fluoranthene was chosen as a possible precursor as a result of (a) its presence (lo-15 wt.%) in the extracts of the naphthalene pyrolysis and (b) the presence of its carbon framework on the surface of &. Pyrolysis of this compound produces Cc,, (Fig. 5(a) ) and the stepwise mechanism of formation is indicated by the presence of successively more condensed dimers in the 502498 a.m.u. region of the mass spectrum. Trimerization in a manner (Fig. 5 (b) ) which obeys the isolated pentagon rule may be used to explain the ion patternin the 720-752 a.m.u. region of the mass spectrum. Trimerization, via single bonds only, yields a species with a mass of 752 a.m.u. Further condensation leads to a species which is sufficiently nonplanar to allow overlap of hydrogen

22

C. Crowley

et al. /Synthetic

atoms, followed by very rapid dehydrocyclization to give the desired fullerene. Trimerization in such a manner is in complete agreement with the observed oxidation of benzo[k] fluoranthene at the 7 and 12 positions in the first instance, followed by the 9 and 10 positions [ 131. Reverse phase HPLC analysis of the acetonitrile soluble products indicates that 46 species are present. Identification of these species and correlation with those found in the naphthalene work are currently in progress.

Metals

77 (1996) 17-22

As in the earlier naphthalene work, the effect of metals was investigated. Our findings indicate that, unlike the naphthalene results, cobalt powder and 10% palladium/carbon appear to have the most beneficial effect on the C6,, yield (likely to be approximately 1%). Studies with lanthanum( III) oxide show evidence of lanthanum encapsulation (m/z = 859) within the fullerene nucleus (Fig. 6(b) ) .

References 3.3. Corannulene

Corannulene was one of the first PAHs to be discussed in relation to fullerene formation. Its synthesis by Barth and Lawton [ 141 led to Osawa’s [ 151 original conjecture that C6,, may exist. Examination of the fullerene surface indicates that 12 corannulene units can be found. Various groups have suggested that corannulene may be a fullerene precursor [ 161. Evidence of the link between corannulene and C60 has been provided by (a) the detection of corannulene and its protonated form in sooting flames [ 171, (b) formation of decachlorocorannulene [ 41 in an arc generator in which the inert gas had been doped with chlorine gas, and (c) the detection of both corannulene and C,,, as well as benzo[k] fluoranthene in the Allende meteorite [IS]. Owing to the scarcity of material, little work had been done on the use of corannulene as a fullerene precursor. We have shown (Fig. 6 (a) ) that corannulene may be used to produce C6,, (and C,,) over a wide range of temperatures, pressures and reaction times. The mechanism of fullerene formation from corannulene is not well understood. It is not possible to form ChOby simple trimerization or patchwork growth processes, followed by dehydrocyclization, as in the other molecules studied. Several mechanisms have been proposed. The simplest and most probable mechanism involves the isomerization of corannulene [ 191 to a species similar to that shown in Fig. 3 (c) , followed by trimerization and dehydrogenation to give C6,,. This mechanism has recently been supported by the independent findings from theoretical studies carried out by Terrones [ 201 and by Taylor et al. [ 211, which indicate that the bowl is the most stable isomer of C,, at temperatures of up to 4500 K. This would appear to contradict the theories of Rubin et al. [ 221, McElvaney et al. [ 231 and Heath [ 241 who propose that a monocyclic ring structure is likely to be the most stable CZOisomer.

[ 1] H.W. Kroto, J.R. Heath, SC. O’Brien, R.F. Curl ‘and R.E. Smalley, Nafure, 318 (1985) 162. [2] W. Kratschmer, L.D. Lamb, K. Fostiropoulos and D.R. Huffman, Nature, 347 (1990) 254. [3] H. Schwarz,Angew. Chem., Inr. Ed. En& 32 (1993) 1412. [4] T. Grosser and A. Hirsch, Angew. Chem., Int. Ed, Engl,. 32 (1993) 1340. [5] R. Taylor, G.J. Langley, H.W. Kroto and D.R,M. Walton, Nature, 366 (1993) 728. [6] C. Crowley, H.W. Kroto, R. Taylor, D.R.M. Walton, Pei-Chao Cheng and L.T. Scott, submitted for publication. [7] L.T. Scott, M.M. Hashemi, D.T. Meyerand H.B. Warren,J. Am. Chem. Sot., 113 (1991) 7082. [8] G.M. Badger, SD. Jolad and T.M. Spotswood, Aust. J. Chem., 17 (1964) 771. [9] K.P.C. Volhardt, personal communication, [lo] MS. Erickson and J. Milliken, personal communication. [ 1I] L.T. Scott and N.H. Roelofs, J. Am. Cite/n. Sot., 109 (1987) 5461. [ 121 K.F. Lang, H. Buffleb and J. Kalowy, Clrern. Ber., 90 ( 1957) 2888. [13] E. Clar, Polycyclic Hydrocarbons, 2 (1964) 312. [ 143 W.E. Barth and R.G. Lawton, J. Am. Chem. Sot., 93 (1971) 1730. [ 151 E. Osawa, Philos. Trans. R. Sot. Londorl, Ser. A, 343 (1970) 1. [ 163 A.D. Haymet, J. Am. Chem. Sqc., 108 (1986) 319; A. Borchardt, A. Fuchicello, K.V. Kilway, K.K. Balridge and J.S. Siegel, J. Am. Chem. sot., 114 (1992) 1921. [17] J. Ahrens, M. Bachmann, Th. Baum, J. Griesheimer, R. Kovacs, P. Weilmiinster and K.H. Homann, brr. J. Mass Spectrom. ion. Proc., I38 (1994) 133; A.L. La Fleur, J.B. Howard, J.A. Marr and T. Yadav, J. Phys. Chem., 97 (1993) 13 539. [ IS] L. Becker, J.L. Bada, R.E. Winans and T.E. Bunch, Nature, 372 (1994) 507. [ 191 T.M. Chang, A. Naim, S.N. Ahmed, G. Goodloe and P.B. Shevlin, J. Am. Chem. SOL, 114 (1992) 7603. [20] H. Terrones, Fullerene Sci. Technol., 3 (1995) 107. [21] P.R. Taylor, E. Bylaska, J.H. Weare and R. Kawai, Cl,em. Phys. Lett., 235 (1995) 558. [22] Y. Rubin, M. Kahr, C.B. Knobler, F. Diederich ‘and CL. Wilkins, J. Am. Chem. Sot., 113 (1991) 495. [23] SW. McElvaney, M.M. Ross, N.S. Goroff and F. Diederich, Science, 259 (1993) 1594. [24] J.R. Heath, Spectroscopy, 5 (1990) 36.