High–yield reactive extraction of giant fullerenes from soot Frank Beer, Andreas Gu�gel,* Kai Martin, Joachim Ra�der and Klaus Mu�llen* Max-Planck-Institut fu� r Polymerforschung, Ackermannweg 10, D-55128 Mainz, Germany By the common Soxhlet extraction with 1,2,4-trichlorobenzene a mere 8 mass % of virgin fullerene soot can be dissolved. The extracted soot was subjected to a reactive extraction with 5-hexadecanamido-1,3-dihydro-2-benzothiophene 2,2-dioxide (4), an ortho-quinodimethane precursor.Through an irreversible Diels–Alder cycloaddition an additional 12 mass %was solubilized. Mass spectrometry, vapour pressure osmometry and elemental analysis indicate that the soluble material consists of multiple adducts of fullerenes C60–C418. Since the discovery of the fullerenes by Kroto et al.1 and the Results and Discussion preparation of macroscopic quantities of C60 by Kra�tschmer All experiments described in the following were performed et al.,2,3 the scientific community has been interested in fullerene with fullerene soot produced by the Kra�tschmer–Human soot.This material is a new kind of amorphous carbon process.2,3 Prior to the experiments the soot was exhaustively consisting of globular and irregularly shaped carbon structures Soxhlet extracted with either toluene (8.1 mass % soluble as well as stacks of bent and planar carbon sheets with dierent material) or successively with toluene and 1,2,4-trichloro- sizes and curvatures.4 Considerable amounts of C60 and larger benzene (8.4 mass % soluble material).carbon clusters are embedded in this insoluble black fullerene Our first experiments were aimed at the reactive extraction matrix.Consequently, many reports have dealt with the extrac- of soot with modifying agents which form thermally unstable tion of fullerene soot using high-boiling solvents.5–12 With adducts with fullerenes. This approach would allow us to 1,2,4-trichlorobenzene as solvent Diederich et al.isolated fuller- cleave the modifying agent after the extraction and thus enes up to C216.5 Ruo and co-workers reported that up to to obtain unsubstituted fullerenes. Unfortunately, both the 37% of virgin fullerene soot could be extracted using 1,2,4- extraction with anthracene-9-methanol and with 2- trichlorobenzene.6 However, the latter results are not backed trimethylsilyloxybutadiene which are both known to form up by elemental analyses so it is unclear whether the extracted reversible Diels–Alder adducts15 with fullerenes in boiling material consists only of all-carbon molecules.Nevertheless, toluene yielded only negligible amounts of soluble material. these studies5,6,7,12 agree on the distribution of the fullerenes Realizing that the thermal stability of the adducts is obvi- in the extracts, and carbon clusters up to a molecular mass of ously a crucial condition for successful reactive extractions we 2500 have been detected by time of flight mass spectrometry.turned towards ortho-quinodimethanes 2 as modifying agents. Several questions remain to be answered: (i ) How much of ortho-Quinodimethanes can be generated by thermal the fullerene soot is actually soluble? (ii ) What is the fullerene extrusion of sulfur dioxide from 1,3-dihydro-2-benzothiophene content of the virgin fullerene soot? (iii ) Which fullerenes are 2,2-dioxide (1) and its derivatives above 200°C (Scheme 1).most abundant in the soot? (iv) Can the catalytic activity of They are very reactive enophiles, which eagerly form [4+2] extracted fullerene soot (e.g.towards the conversion of methane cycloaddition products 3 with fullerenes giving high yields as into higher hydrocarbons) be traced back to residual (not well.16,17 These adducts are both thermally stable and highly extracted) fullerenes or is it actually a property of the pure soluble in common solvents due to their conformational soot?13 mobility (cyclohexene ring inversion).Answering these questions requires a method for the com- Refluxing 1,2,4-trichlorobenzene-extracted fullerene soot plete extraction of the soot. Reactive extraction seems to be (500 mg) together with 175 mg (1.04 mmol) 1,3-dihydro-2- especially well suited to this purpose. In this method, the benzothiophene 2,2-dioxide (1) (Scheme 1) in 1,2,4-trichloro- extractable compounds are solubilized via reaction with a benzene for 24 h (Table 1, Entry 1) followed by filtration from suitable partner in the liquid phase and are thus easily transinsoluble material aorded a black solution.This solution was ferred into the liquid phase.14 again filtered through a glass frit (pore size 10–16 mm) to avoid Herein, we describe the reactive extraction of fullerene soot contamination with small, insoluble particles of soot.with ortho-quinodimethanes as modifying agents. The influence Afterwards the solution was evaporated and the residue sus- of the structure of the modifying reagent on the extraction pended in ethanol and sonicated for 5 min. Unreacted 1, yield and on the composition of the extracted material will be byproducts and residual solvent were dissolved in ethanol and addressed and the results will be compared with common extraction methods. could be easily separated by filtration.The black powdery Scheme 1 Generation of ortho-quinodimethane 2 from 1,3-dihydro-2-benzothiophene 2,2-dioxide (1) and its cycloaddition reaction with C60 J. Mater. Chem., 1997, 7(8), 1327–1330 1327Table 1 Entries; compounds 1, 4 used for reactive extraction or solvent used for extraction; amount of extracted soot; extraction yields; elemental analyses of extracted material; fullerene distributions determined by LD–TOF MS; Mn (g mol-1) determined by vapour pressure osmometry yield of extracted extraction/ fullerene vapour reactive adducts or elemental analysis yield yield pressure extraction carbon fullerenes CFull CFull fullerenes LD–TOFd tosmometry entry with soot /mg C (%) H (%) N (%) H/N (%) /mg (%)c MS Mn [g mol-1]e 1a 1 (175 mg) 500 mg 45 92.36 2.74 <0.2 >13 59.7 27 5.4 C60–C344 2590 2a 1 (88 mg) from 1 9 90.34 2.63 <0.3 >8 59.0 5 1.0 C60–C396 — 3a 4 (440 mg) 500 mg 164 83.40 7.59 2.70 2.82 27.8 46 9.2 C60–C400 6610 4a 4 (220 mg) from 3 46 82.84 6.91 2.70 2.56 27.2 13 2.6 C60–C418 6480 5a 4 (330 mg) 52 mg 24 83.33 6.85 2.84 2.41 24.9 6 11.5 C60–C418 — 6a 4 (100 mg) 500 mg 51 85.72 6.40 2.50 2.56 27.7 14 2.8 C60–C348 3770 7a 4 (100 mg) 1000 mg 56 85.46 7.04 2.50 2.82 34.0 19 1.9 C60–C308 4010 8a 4 (440 mg) 10 g 113 90.79 4.08 1.45 2.81 61.0 69 0.7 C60–C350 1930 9 toluene 10 g 810 99.10 0.27 0.22 1.23 99.1 810 8.1 C60–C180 — soxhlet 10b 1,2,4- 30 g 81 93.68 0.71 <0.1 >7.1 93.7 76 0.25 C60–C192 — trichlorobenzene soxhlet 11b 1,2,4- 10 g 26 96.46 0.40 <0.1 >4 96.5 25 0.25 C60–C196 — trichlorobenzene reflux 12b quinoline 25 g 389 87.63 2.09 2.28 0.92 70.0 272 1.1 C60–C260 — soxhlet a1,2.4-Trichlorobenzene-extracted soot.bToluene-extracted soot. cCalculated with respect to soot.dPositive ion LD–TOF MS. eVapour pressure osmometry in tetrahydrofuran (measured at 30°C). filtration residue was washed several times with ethanol and mass ratio can be determined by elemental analysis and must dried under vacuum. Finally, a yield of 45 mg of mat- have a value of 2.8 for the formed adducts assuming that pure erial was obtained, which readily dissolved in CHCl3 or addition products of 4 and carbon clusters are formed.Again tetrahydrofuran (THF). According to its H5C ratio, which the sums of proportions of elements C, H, N and S determined was determined by elemental analysis, this corresponds to a by elemental analyses are only 96–98%. As mentioned above yield of fullerenes of 5.4 mass % (calculated with respect to we assign the missing amount of 2–4% to oxygen resulting soot).The sum of the proportions of elements C, H, N and S from epoxidation products of fullerenes. determined by elemental analysis is only 95%. We assume that Treatment of 1,2,4-trichlorobenzene-extracted fullerene soot the residual amount consists of oxygen which has reacted w with 4 (440 mg, 1.04 mmol) according to the above the fullerenes.It is well known that epoxidation of fullerenes procedure led to the isolation of 164 mg of soluble material in solution is induced by UV irradation18 or heating19 in the (Table 1, Entry 3). The elemental analysis of this material presence of oxygen. showed a H5N ratio of 2.8 which proved that the sample Laser desorption time-of-flight (LD–TOF) mass spectra of consisted entirely of modified carbon clusters. The total amount this sample showed a distribution of unmodified fullerenes from of carbon clusters extracted from the soot was calculated to C60 to C344, due to retro-Diels–Alder reaction of the formed be 9.2 mass %.This is considerably higher than the 5.4 mass % adducts during the ionization process. Vapour pressure which was achieved with the less flexible 1.From this it follows osmometry which gave a number-average molecular mass of that the higher the solubility of the adducts the higher the Mn=2590 g mol-1 also indicated that the soluble material extraction yields. consisted of Diels–Alder adducts of high molecular mass As described before for 1, a second extraction of the soot fullerenes (giant fullerenes).Additional soluble fullerene mate- with 4 yielded another 2.6 mass % of soluble material (Table 1, rial was obtained when the once-extracted soot was subjected Entry 4). It follows that approximately 12 mass % of the soot to a second reactive extraction. This yielded another 1 mass % can be extracted by this approach. In comparison, the total of soluble material. The LD–TOF mass spectra of this sample extraction of a small amount of soot (52 mg) within one step showed slightly higher fullerenes up to C396.with an extremely large excess of 4 (330 mg, 0.69 mmol) yielded This experiment clearly showed that ortho-quinodimethanes 11.5% soluble material (Table 1, Entry 5). are very well suited for reactive extraction. Obviously, the The soluble material which was received from the first solubility of the adducts is the crucial point for the additional reactive extraction (Table 1, Entry 3) was fully characterized yield and the distribution of extractable material.We therefore by means of mass spectrometry, vapour pressure osmometry synthesized the ortho-quinodimethane precursor 5-hexadecan- and thermogravimetric analysis.Positive as well as negative amido-1,3-dihydro-2-benzothiophene 2,2-dioxide (4) carrying ion LD–TOF mass spectra again show, due to retro-Diels– a long flexible alkyl chain. Alder reactions of the adducts during the ionization process, a distribution of even numbered unmodified carbon clusters (Fig. 1). The positive ion mass spectrum shows fullerenes up to mass ca. 4800 corresponding to C400 (Fig. 1, top panel). 170 peaks can be assigned to fullerenes with good signal to noise ratio. The molecular mass of fullerenes in the negative ion mass spectrum is somewhat lower compared to that in the positive ion mass spectrum. Strong signals of C60, C70 and C84 The fixed H5N mass ratio of 2.8 in 4 allows a more precise can be observed (Fig. 1, bottom panel). 125 peaks can be calculation of the product composition and can serve as an indicator for the purity of the extracted products. The H5N counted for fullerenes with good signal to noise ratio. Ruo 1328 J. Mater. Chem., 1997, 7(8), 1327–1330evident that the soluble material consists of fullerenes C60–C400 which are multiply functionalized with ortho-quinodimethanes. However, isolation of single adducts from the adduct mixture with HPLC was not achieved due to the unavoidable formation of multiple adducts and the corresponding regioisomers; the soluble material consists of more than 1000 compounds.Further experiments with dierent stoichiometries show that both the yields of fullerenes and the composition of the extracted material can be influenced. Reaction of 500 mg of soot with 100 mg (0.23 mmol) of 4 (Table 1, Entry 6) yielded 51 mg of fullerene adducts (fullerene content: 27.7%) which corresponds to 2.8 mass % pure fullerenes (calculated with respect to soot).In comparison, reaction of 1000 mg of soot with 100 mg (0.23 mmol) of 4 (Table 1, Entry 7) yielded 56 mg of fullerene adducts (fullerene content: 34%) which is equivalent to 1.9 mass % pure fullerenes.These results show that increasing soot to reagent ratios lead both to diminished yields of fullerenes and increasing fullerene content of the extracted material. With that, the number of attached ortho-quinodimethanes is also decreasing. The latter result is supported by the reaction of 440 mg (1.04 mmol) of 4 with a very large excess of soot (10 g) (Table 1, Entry 8) since this extraction yields a highly fullereneenriched sample with a fullerene content of 61 mass %.It can be concluded that the reactive extraction proceeds in three separate steps: (i ) fullerenes which are firmly embedded Fig. 1 (a) Positive and (b) negative ion LD–TOF MS of fullerene sample from reactive extraction with 5-hexadecanamido-1,3-dihydro- in the soot matrix and are not extractable by conventional 2-benzothiophene 2,2-dioxide 4 Soxhlet extraction react with ortho–quinodimethanes, (ii ) the modified and therefore soluble fullerenes dissolve in 1,2,4- trichlorobenzene, and (iii ) by further reaction of excess ortho- et al.discuss some reasons for dierences in negative and quinodimethanes with dissolved fullerene adducts multiple positive ion mass spectra.5 They suggest that dierent ioniz- adducts are formed.ation probabilities for positively and negatively charged carbon The advantage of the described reactive extraction becomes clusters in LD–TOF MS experiments lead to significant dier- evident by comparing it with conventional Soxhlet extraction. ences in the distributions of fullerenes. Field-desorption (FD) The yield of fullerenes from conventional Soxhlet extraction MS is a well-known technique for detecting fullerenes and of virgin soot with toluene is found to be 8.1% (Table 1, Entry fullerene adducts without fragmentation during the ionization 9).A subsequent extraction with 1,2,4-trichlorobenzene yields process. FD MS of this sample showed, apart from multi- only additional 0.25% fullerenes (Table 1, Entry 10) and with adducts of C60 and C70, mono- and bis-adducts of the recently quinoline 1.1% (Table 1, Entry 12).Furthermore, elemental isolated and characterized C80 (see Experimental).20 analysis of the quinoline extract indicates that the sample There are also significant dierences in the fullerene distri- contains large, not removable amounts of impurities due to butions of samples received from conventional and reactive decomposition of the solvent.Hence, the yields of reactive extractions. Fig. 2 shows the positive ion LD–TOF mass extraction are superior to those of conventional extraction. spectrum of pure fullerenes from 1,2,4-trichlorobenzene extract (Table 1, Entry 10). Fullerenes up to mass 2100 (C192) can be Conclusions detected.The FD mass spectrum of this sample indicates the existence of fullerenes with more than 100 carbon atoms too Giant fullerenes which are barely soluble in common solvents (see Experimental). and are firmly embedded in the solid fullerene soot matrix can It follows that reactive extraction enables the extraction of be eciently functionalized by means of reactive extraction fullerenes with approximately double the molecular mass com- with ortho-quinodimethanes. The fullerenes are thus made pared to conventional extraction. soluble in 1,2,4-trichlorobenzene and are extractable in very The number-average molecular mass Mn of the sample from high, so far unprecedented, yields. entry 3 was determined by vapour pressure osmometry as Commercially available fullerene soot which yields 8.4mass% 6610 g mol-1.Thermogravimetric analysis showed that these of soluble fullerenes by entire extraction with common solvents fullerene adducts are thermally stable up to at least 310°C aords additional 11.8 mass % of soluble material (funcwhich is in accordance with results obtained for ortho-quinodi- tionalized fullerenes C60–C418) by reactive extraction with an ortho-quinodimethane precursor (4).The above experiments methane adducts of C60.21 From the analytical results it is indicate that the conformational flexibility of the ortho-quinodimethanes is the key factor for the amount of extracted material. Therefore, further experiments will make use of ortho-quinodimethanes substituted with highly branched (extremely flexible) groups. Our studies show that the amount and the composition of the extracted material can be significantly influenced by diering the ratio of fullerene soot to modifying agent.The entirely extracted soot will be utilized to study if the catalytic activity of fullerene soot is due to the soot itself or comes from included redox active fullerenes.Experimental Carbon soot: for all experiments, samples of the same soot were used. The soot was prepared by arc synthesis according Fig. 2 Positive ion LD–TOF MS of 1,2,4-trichlorobenzene-extracted fullerenes to the Kra�tschmer–Human2,3 process and was provided by J. Mater. Chem., 1997, 7(8), 1327–1330 1329Hoechst AG. The soot was previously extracted with toluene ethanol and sonicated for 5 min.After filtration, washing with ethanol and drying a black powder was obtained. to remove C60 and C70 (approximately 8.1 mass %). LD–TOF MS was performed using a Bruker Reflux mass Extraction of the soot under reflux (Table 1, Entry 11) spectrometer with a 337 nm N2 laser. Laser power was adjusted to be the lowest level at which an ion signal was observed.The soot was placed in a flask and 200 ml of the solvent was A VG Instruments ZAB 2–SE–FPD mass spectrometer was added. After refluxing for 24 h, the suspension was filtered. used for recording FD mass spectra. The solution was evaporated and the residue suspended in 150 ml ethanol and sonicated for 5 min. Filtration, washing 1,3-Dihydro-2-benzothiophene 2,2-dioxide (1) with ethanol and drying gave a black powder.The reagent was synthesized as described by Cava and Deana.22 Positive ion FD MS of crude material from entry 10 dH (300 MHz, [2H6]acetone, 21°C) 4.35 (s, 4H, CH2), 7.29–7.40 (m, 4H, CH). dC (75 MHz, CDCl3, 28°C) 47.3 (CH2), arom. C C60 720.6 (95%), C70 840.7 (100%), C76 911.7 (15%), C78 935.6 atoms: 126, 129 and 131.5. Positive ion FD MS: 168 ([M+], (31%), C82 983.7 (13%), C84 1009.9 (14%), C86 1031.6 (21%), 100%). Mp 150–152°C.C88 1056.5 (22%), C90 1079.9 (21%), C92 1104.5 (10%), C94 1127.0 (8%), C96 1152.9 (20%), C98 1177.9 (5%), C100 1201.7 (11%), C106 1272.2 (11%), C110 1321.7 (7%), C112 1343.9 (8%), 5-Hexadecanamido-1,3-dihydro-2-benzothiophene 2,2-dioxide (4) C114 1367.7 (17%), C116 1392.4 (16%), C118 1418.9 (9%), C120 1,3-Dihydro-5-amino-2-benzothiophene 2,2-dioxide23 (2.87 g; 1439.1 (22%), C124 1489.1 (22%), C136 1633.5 (12%). 15.7 mmol) and triethylamine (2.1 g; 20 mmol) were dissolved in 1,4-dioxane (250 ml). Hexadecanoyl chloride (4.73 g; This research was supported by the ‘Bundesministerium fu�r 17.3 mmol) was added dropwise at room temp. over 20 min. Bildung und Forschung’ (grant number: 13N6665/8). We thank The solution was stirred under reflux for 20 min.The solution the Hoechst AG for providing fullerene soot. was cooled to room temp. and then added to H2O (300 ml), giving a white precipitate. The precipitate was collected by References filtration, washed with H2O and Et2O and dried in vacuo. Yield: 5.75 g (13.6 mmol; 87%). dH (500 MHz, C2D2Cl4, 100°C) 1 H. W.Kroto, J. R. Heath, S. C. O’Brien, R. F. Curl and R. E. Smalley, Nature, 1985, 318, 162. 0.87 (t, 3H, CH3), 1.26 (br, 24H, CH2), 1.66 (m, 2H, CH2), 2 W. Kra�tschmer, K. Fostiropoulos and D. R. Human, Chem. Phys. 2.34 (t, 2H, CH2), 4.25 (s, 2H, CH2), 4.27 (s, 2H, CH2), 7.13 (s, L ett., 1990, 347, 167. 1H, CH), 7.18 (d, 1H, CH), 7.3 (d, 1H, CH), 7.65 (s, 1H, NH). 3 W. Kra�tschmer, L.D. Lamb, K. Fostiropoulos and D. R. Human, dC (125 MHz, C2D2Cl4, 100°C) 14.2 (CH3), aliph. CH2: 22.80, Nature, 1990, 347, 354. 25.64, 29.63 (several signals overlapping), 32.08, 37.85, 57.08 4 H. Werner, D. Herein, J. Blo�cker, B. Henschke, U. Tegtmeyer, T. Schedel-Niedrig, M. Keil, A. M. Bradshaw and R. Schlo�gl, (CH2) and 57.63 (CH2), arom. CH: 117.71, 120.64 and 126.76, Chem.Phys. L ett., 1992, 194, 62. quart. arom. C atoms: 126.78, 132.0 and 139.03, 171.0 (CNO). 5 F. Diederich, R. Ettl, Y. Rubin, R. L. Whetten, R. Beck, M. Alvarez, Positive ion FD MS: 421.2 ([M+] 100%). Elemental analysis: S. Anz, D. Sens-harma, F. Wudl, K. C. Khemani and A. Koch, C24H39NO3S, calc. C 68.52%, H 9.29%, N 3.29%, S 7.47%. Science, 1992, 252, 548. Found C 68.37%, H 9.32%, N 3.32%, S 7.6%.Mp 152–154°C. 6 C. Smart, B. Eldridge, W. Reuter, J. A. Zimmerman, W. R. Creasy, N. Riviera and R. S. Ruo, Chem. Phys. L ett., 1992, 188, 171. 7 W. R. Creasy, J. A. Zimmerman and R. S. Ruo, J. Phys. Chem., General procedure for reactive extractions with 1 and 4 1993, 97, 973. (Table 1, Entry 1–8) 8 H. Shinohara, H. Sato, Y. Saito, M. Takayama, A. Izuoka and T.Sugawara, J. Phys. Chem., 1991, 95, 8449. All reactions were carried out under a dry, oxygen-free argon 9 H. Shinohara, H. Sato, Y. Saito, A. Izuoka, T. Sugawara, H. Ito, atmosphere. Soot and 1 or 4 were added to 150 ml of 1,2,4- T. Sakurai and T. Matsuo, Rapid Commun. Mass. Spectrom., 1992, trichlorobenzene. Refluxing for 24 h followed by filtration from 6, 413. 10 D. H. Parker, P.Wurz, K. Chatterjee, K. R. Lykke, J. E. Hunt, insoluble material aorded a black solution. The solution was M. J. Pellin, J. C. Hemminger, D. M. Gruen and L. M. Stock, evaporated and the residue suspended in 150 ml of ethanol J. Am. Chem. Soc., 1991, 113, 7499. and sonicated for 5 min. Unreacted 1 or 4 and residual solvent 11 D. H. Parker, K. Chatterjee, P. Wurz, K. L. Lykke, M.J. Pellin were dissolved in ethanol while a suspension of the reaction and L. M. Stock, Carbon, 1992, 30, 1167. products was formed. Filtration, washing with ethanol and 12 K. R. Lykke, D. H. Parker and P. Wurz, Int. J.Mass Spectrom. Ion drying gave a black powder. Processes, 1994, 138,147. 13 A. S. Hirschon, H. -J. Wu, R. B. Wilson and R. Malhotra, J. Phys. Chem., 1995, 99, 17483.Positive ion FD MS of crude material obtained from entry 3 14 E. Schlichting, W. Halwachs and K. Schu�gerl, Chem. Eng. Commun., 1987, 51, 193. C60-monoadduct 1078.4 ([M+], 25%), C60-bisadduct 1435.8 15 Yi. -Zh. An, G. A. Ellis, A. L. Viado and Y. Rubin, J. Org. Chem., ([M+], 35%), C60-trisadduct 1791.9 ([M+], 45%), C60-tetra- 1995, 60, 6353. adduct 2151.4 ([M+], 100%), C60-pentaadduct 2508.6 ([M+], 16 P.Belik, A. Gu�gel, J. Spickermann and K. Mu�llen, Angew. Chem., 45%), C70-monoadduct 1197.6 ([M+], 35%), C70-bisadduct 1993, 105, 95; Angew. Chem., Int. Ed. Engl., 1993, 32, 78. 17 B. Illescas, N. Martin, C. Seoane, P. de la Cruz, F. Langa and 1555.5 ([M+], 70%), C70-trisadduct 1912.5 ([M+], 55%), C70- F.Wudl, T etrahedron L ett., 1995, 36, 8307. tetraadduct 2270.2 ([M+], 45%), C70-pentaadduct 2627.3 18 J. M. Wood, B. Kahr, S. H. Hoke, L. Dejarme, R. G. Cooks and ([M+], 10%), C80-monoadduct 1317.1 ([M+], 20%), C80- D. Ben-Amotz, J. Am. Chem. Soc., 1991, 113, 5907. bisadduct 1674.8 ([M+], 15%). 19 A. M. Vasallo, L. S. K. Pang, P. A. Cole-Clarke and M. A. Wilson, J. Am. Chem. Soc., 1991, 113, 7820. 20 F. H. Hennrich, R. H. Michel, A. Fischer, S. Richard-Schneider, Thermogravimetric analysis of the extract from entry 3 S. Gilb, M. M. Kappes, D. Fuchs, M. Bu�rk, K. Kobayashi and S. Nagase, Angew. Chem., 1996, 108, 1839. (Mass 7.5 mg, heating rate 10 K min-1, N2 atmosphere) 21 A. Gu�gel, A. Kraus, J. Spickermann, P. Belik and K. Mu�llen, 313–560°C (-54.23%, maximum at 408.3°C, residue 3.43 mg). Angew. Chem., 1994, 106, 601; Angew. Chem., Int. Ed. Engl., 1994, 33, 559. Soxhlet extractions of the soot (Table 1, Entries 9, 10, 12) 22 M. P. Cava and A. A. Deana, J. Am. Chem. Soc., 1959, 81, 4266. 23 M. Walter, A. Gu�gel, J. Spickermann, P. Belik, A. Kraus and The soot was placed in the thimble of a Soxhlet extractor. The K. Mu�llen, Fullerene Sci. T echnol., 1996, 4, 101. extractions were carried out for 24 h. After filtration, the solution was evaporated and the Paper 6/08186J; Received 10th February, 1997 1330 J. Mater. Chem., 1997, 7(8), 1327–1330