J . Phys. Chem. 1992, 96, 5231-5234 (10) Mckenzie, D. R.; Davis, C. A.; Cockayne, D. J. H.; Muller, D. A.; Vassallo, A. M. Nature 1991, 355, 622-624. (11) Thurnauer, M. C.; Katz, J. J.; Norris, J. R. Proc. Narl. Acad. Sci. U.S.A. 1975, 72, 3270. (12) Since the spectrum is polarized to the extent that
the integral over the entire spectrum is practically zero, the disappearance of the individual parts of the spectrum are governed by TI and the decay of the carrier. The
5231
rate constant for the latter process is much smaller (Table I) so that the disappearance of the spectrum is determined almost entirely by T I . (13) For a uniform spin distribution and pure p-orbitals the estimated hyperfine coupling is 115/60 = 1.9 MHz = 0.68 G. This is a minimum value because of the curved surface the orbitals must contain some s-character. (14) Riibsam, M.; Dinse, K.-P.;Pliischau, M.; Fink, J.; Kratschmer, W.; Fostiropoulos, K.; Taliani, C. Submitted for publication.
Production and Characterization of Metallofullerenes Mark M. ROSS,* H. H. Nelson, John H. Callahan, and Stephen W. McElvany Code 61 lO/Chemistry Division, Naval Research Laboratory, Washington, D.C. 20375-5000 (Received: April 8, 1992; In Final Form: May 12, 1992) Negative ion/desorption chemical ionization mass spectrometry was used to characterize fullerenes with an encapsulated metal atom@), metallofullerenes (M,@C,), in arc-generated soot, pyridine extracts, and the extract residue. In agreement with results from other laboratories, the pyridine extracts of La203/graphitesoot contain mostly La@Cg2and La2@CCB0, in addition to the “pure” (empty) fullerenes. However, the raw soot and the extract residue contain a broader range of metallofullereneswith relative abundances different from those observed from the extract (e.g., abundant La@Cm,La@CTO, and L ~ @ J C , ~Arcing ). mixed-metal oxide-impregnated graphite rods yielded a mixed-metal dimetallofullerene (YLa@Cgo) and higher relative abundances of metallofullerenes with the lower ionization potential metal atom. The thermal desorption behavior and solubility in different solvents of the doped and undoped fullerenes indicate an interaction between the C, and La,@C, species. Finally, analysis of aqueous solutions of dried pyridine extracts of lanthana/graphite soot shows C, and La,@C,, which is consistent with the possible presence of metallofullerene/fullerene ionic complexes, (La,@C,)+C,,-.
Introduction One of the many interesting directions of fullerene research involves encapsulation of an atom(s) inside the fullerene cage to form metal/fullerene endohedral complexes, or metallofullerenes. This idea dates back to the original experiments by Smalley, Kroto, and co-workers’ in which laser photodissociation (“shrinkwrapping”) studies of metal-carbon cluster adducts provided strong evidence for the postulated cage fullerene structures. The availability of large quantities of fullerenes2 and improved fullerene production methods3 have advanced this field significantly. In addition to representing another new and unprecedented set of chemical species, the metallofullerenes have given rise to a unique chemical nomenclature, M@C,, where M is a metal atom (or M, for multiple atoms) encapsulated in a fullerene consisting of n carbon atoms. Several groups have used different methods to generate soot from metal compound-impregnated graphite rods and have shown that the soot contains metallofullerenes as well as fullerenes. For example, Smalley and co-workers4 have shown that La- and Y-encapsulated fullerenes (La,@C,, Y,@C,), in addition to fullerenes (C,), can be produced by laser vaporization of graphite rods impregnated with the appropriate metal oxide. Laser desorption mass spectrometric (LD/MS) analysis of a film of sublimed material from the soot generated by laser vaporization of a La203/graphite rod showed mostly La@C60, La@C7,, and h@c82 while extraction with toluene yielded mostly La@C,,. Similar results were obtained from studies of yttriumfullerene complexes, with the additional observation of an anomalous abundance of Y2@cg2 in the sublimed film.5 Johnson et a1.6 generated a large quantity of La@Cg2and used electron paramagnetic resonance (EPR) spectroscopy to show that La has a 3+ formal charge and the CS2a charge of 3-. Scandium-, yttrium-, and lanthanum-encapsulated fullerenes were also produced, extracted, and characterized by Shinohara and c o - ~ o r k e n , ~ who noted that pyridine and carbon disulfide are much better solvents than toluene for the metallofullerenes. In agreement with the previous studies, some metallofullerenes such as Y@Cm and Y@C70 were not observed in the solvent extracts. A dilanthanum fullerene, La2@Cmwas observed first by Whetten and co-workers8 to be very abundant in the laser desorption mass spectrum of the toluene extract of soot generated using the Kratschmer-Huffman method. This particular dimetallofullerene had not been observed using similar soot production and extraction methods, and it was speculated to be dependent on the amount of La203loaded into
the graphite rod. Until recently, our work has focused on the mechanism of metallofullerene formation by laser vaporization in vacuum. In these studies evidence was obtained for laser-induced coalescence reactions between fullerenes and metal oxide compounds to form a wide size distribution of mono- and dimetallof~llerenes.~These investigations showed that laser desorption/vaporization can produce species that are not initially present in the sample. Despite the significant effort and achievements several questions remain. For instance, it is not clear why certain stoichiometries of M,@C, are more abundant than others and, specifically, why particular dimetallofullerenes (e.g., La.@, and Y2@CE2)are not uniformly observed. The dramatic differences in the abundances of the metallofullerenes observed in the soot extracts versus those in a film sublimed from the soot are not understood, although they are likely due to differences in solubility or reactivity. The above and other experimental observations have led to speculations that the metallofullerenes could exist as ionic complexes, possibly with “pure” (undoped) fullerenes.5J0 The chemical nature of these unusual species must be determined before any potential materials application can be approached. We report here some new insights obtained from characterization of metallofullerenes from soot extracts and directly from the soot using desorption chemical ionization mass spectrometry.
Experimental Methods Soot containing fullerenes and metallofullerenes was generated by arcing metal oxide-impregnated graphite rods using methods that have been described in detail.3q4.8 In brief, 0.25-in.-diameter graphite rods (Wale Apparatus Co.) were drilled out (0.17in.-diameter by 2 in. deep) and fiued with a mixture of metal oxide, La203 (Fisher Scientific, 99%), Y203(Alfa Products, 99.99%), Sc203(Reacton, 99.99%) or a 1/1 mixture (by weight), and graphite cement (GC-HS, Dylon Industries). The resulting rod was approximately 20%metal oxide by weight. Before use, the cement was cured at 140 OC for 4 h and the rods were degassed by heating to 1000 OC under vacuum overnight. Two metal oxide-impregnated rods were used as the two electrodes in a “Smalley-type” water-wold contact arc r e a ~ t o r .The ~ resultant soot was collected and extracted for 2-8 h in pyridine using a Soxhlet extraction apparatus. All analyses of the soot, soot extracts, and extract residues were performed using negative ion/desorption chemical ionization mass spectrometry (NI/DCIMS) with a triple quadrupole mass
This article not subject to U S . Copyright. Published 1992 by the American Chemical Society
5232 The Journal of Physical Chemistry, Vol. 96, No. 13? 1992
Letters
I
"1
La o C,,
'70 I
c:
'L La@C,
Y@Cm
m/z
0lm
1050
& 1 loo
m/z
Figure 2. Methane negative ion/desorption chemical ionization mass spectrum of a pyridine extract of YzO,/LazO,/graphite soot.
capsulated fullerenes, there is essentially only one dilanthanum fullerene observed. Although this result is in contrast with other reports of extensive distributions of M,@C, observed using LD/MS,7,8 our laser (CO,) desorption analysis of this pyridine extract in a Fourier transform mass spectrometer (FI'MS) yielded results that were nearly identical with those from NI/DCIMS.l3 The limited size distributions of the extracted metallofullerenes led us to question whether other species are present in the raw soot. The lower panel of Figure 1 is the methane NI/DCI mass spectrum of the unextracted soot using the same desorption and ionization conditions that were used to obtain the mass spectrum 700 8w 1om 1lbo in the upper panel. This analysis shows that the soot contains some metallofullerenes, most notably (La@C,)- ( m / z 859) and m/z (La@C70)-( m / z 979), that are not observed in the extract, which Figure 1. Methane negative ion/desorption chemical ionization mass is in agreement with Smalley's LD/MS analysis of the sublimed spectrum of a pyridine extract of La20,/graphite soot in the upper panel film.43s A wide range of (La@C,)- (with even n = 60, 70-90, and of raw (unextracted) La20,/graphite soot in the lower panel. i.e., only those C, that can have structures with isolated pentagons) spectrometer (Finnigan TSQ-70). A 1-pL (or less) aliquot of the is observed, but with different relative abundances than those from pyridine extract solution or a suspension of the raw soot in toluene the extract. For example, the abundance of (La@C& is comwas deposited onto the wire tip of a probe, which was inserted parable to that of (La@C7J, which is about twice that of into the ion source containing a methane reagent bath gas. (La@C,,)-. In addition, (Laz@Cso)- is observed with a much Thermal desorption of species was achieved by rapid application lower relative abundance than that from the extract. The NI/ of an increasing current (50-1000 mA in a 0.5-1-min time period) DCIMS analysis of the extract residue yielded a mass spectrum to the wire, and ionization was accomplished by electron capture nearly identical to that of the soot in the lower panel of Figure to form negative ions. Mass spectra were acquired by scanning 1, except that there is no detectable L a , c i ~ C ~indicating ~, an the quadrupole over a specified m / z range every 0.5 s during the especially high extraction efficiency for this species. thermal desorption, which results in a profile of the abundance The extractable dimetallofullerenes were investigated further of any single m / z ion, or all ions in the range, as a function of through analysis of the soot generated from arcing a mixed time during the heating ramp. Y,O,/La,O,-impregnated graphite rod. Figure 2 shows the methane NI/DCI mass spectrum of the pyridine extract of the Results and Discussion mixed-metal (Y/La) oxide,lgcaphite soot. In addition to As discussed above, laser desorption mass spectrometry (Y@CSZ)-( m / z 1073), ( L a @ G J - ( m / z 1W, (Y,@C,)- ( m / z (LD/MS) has been extensively used to analyze the fullerenes or 1138), and (La2@Cso)-( m / z 1238) the mixed-metal dimetallometallofullerenes that are extracted or sublimed from ~ o o t . ~ - ~ fdlerene, (YLa@CW)-( m / z 1188), is observed. It should be noted While LD/MS can be used to analyze large, thermally labile that an anomalously high abundance of Y & c @ is nor observed. organic molecules, with fullerenes, under certain conditions as Once again, except for low abundances of (M,@JC,~)-there is mentioned above, it is possible that the laser can cause formation no distribution of dimetallofullerene species, unlike the monoof species that are not originally present in the a n a l ~ t e .The ~ metal-encapsulated fullerenes. It is also interesting to note that NI/DCIMS technique offers an alternative that has been shown although the graphite rod contained an approximate 1.5/ 1 mole to be very sensitive in the analysis of fullerenes and causes negratio of Y/La, there is a greater abundance of La-containing ligible fragmentation."Jz The upper panel of Figure 1 shows the species. It has been shown that negative ion chemical ionization methane NI/DCI mass spectrum of the pyridine extract of yields a different C60/C70ratio compared with positive ion electron Laz03/graphitesoot. This spectrum represents an average of 10 ionization (EI) analysis, possibly due to different electron capture spectra acquired at the peak of the desorption profile of the cross-sections.I2 However, in this work not only are similar size metallofullerenes, La,@ C,. The "pure" fullerenes, C,, observed metallofullerenes (e.g., Y @CS2vs La@C8,) compared, but the consist mostly of c60, C70, and lower abundances of higher positive ion E1 analysis showed the same trend of more La- than fullerenes (e.g., c76, c78, C84), with no detectable fragmentation. Y-containing species, also. In agreement with other studies, the (La@C8,)- ( m / z 1123) In a related experiment, analysis of Sc203/Y203/graphitesoot species is the most abundant lanthanum-containing fullerene, in showed predominately Y-containing species. These observations addition to lower abundances of (La@C,)-, with even n = 74-100 provide some support for the speculation that the ionization po(not all shown here), and (h@CsO)- ( m / z 1238). It is interesting tential (IP) of the metal atom may in part determine the efficiency that while there is a broad distribution of monolanthanum-enof metallofullerene f0rmati0n.I~ The IPSof La, Y, and Sc are
The Journal of Physical Chemistry, Vol. 96, No. 13, 1992 5233
Letters a) toluene extract C .,
m/z 720
P !
I
"1 I
/1 J
c60
x3 ?
I
LamC,,
C,"
b) pyridine extract
I
P
I m/r 7001250
A
I
Scan Number Figure 3. Thermal desorption profiles (ion abundance versus scan number) of selected ions produced by methane negative ion/desorption chemical ionization of (a) a toluene extract of Laz03/graphitesoot in the upper panel and (b) a pyridine extract of La2O3/graphitesoot in the lower panel. Each profile is normalized to the maximum abundance of the selected ion(s). The first component peaks at approximately 280 OC. the second at 350 OC, and the third at 500 OC.
5.58,6.2, and 6.54 eV, respectively, and the greater abundance of metallofullerenes with the lower IP metal suggests that ion/ molecule reactions in the arc plasma may be important in the formation of M,@C,. We are investigating this topic in greater detail. In attempts to separate the metallofullerenes from the empty fullerenes, different extraction solvents and procedures were investigated, with no success. However, some insight was gained into the chemical nature of the metallofullerenes. For instance, a lanthana-graphite soot sample was first sonicated a t room temperature in toluene and resulted in only fullerenes being extracted. The residue from the toluene extraction was Soxhlet extracted with pyridine, which yielded both fullerenes and metallofullerenes. In contrast, when the pyridine was then evaporated, the residue readily dissolved in toluene and the NI/DCIMS analysis of this solution showed the same C; and (La,@C,,)- with the same relative abundances as those in the original pyridine extract. The extraction with pyridine appears to change the solubility of the metallofullerenes in toluene, which is analogous to recent reports of the solubility of large fullerene~.'~These observations suggest that La@Cm and La@C70may be present in different chemical forms or bound in the soot differently than other extractable metallofullerenes. In pursuing this idea further, the thermal desorption profiles of C; and (La,@C,)- were examined. The upper panel of Figure 3 shows the ion abundance versus scan number (which is proportional to time and temperature) profiles, or ion chromatograms, of Cm-, C70-rand all ions in the range m / z 700-1250 during a I-min thermal desorption of a room temperature toluene extract of a lanthana-graphite soot sample. As stated previously, these extraction conditions do not yield metallofullerenes. It should be noted that these plots are single-component, nearly symmetrical profiles, which peak a t approximately 350 OC. The lower panel
mlz Figure 4. Methane negative ion/desorption chemical ionization mass spectrum of an aqueous solution of a dried pyridine extract of Laz03/ graphite soot.
of Figure 3 shows the profiles, using the same conditions as in the upper panel, for the same species and (La@&)- from the pyridine extract of a lanthana3raphite soot sample. In contrast to the toluene extract, the profiles for Cm-, C70-, and all ions have three components, the middle of which occurs at approximately the same time (temperature) as those in the upper panel. Upon inspection of the mass spectra, the first component was found to be made up of mostly Cm-, C70-, and adducts of these fullerenes that correspond to addition of oxygen atoms, CH,, OCH,,I6 and pyridine. It is interesting to note that this first component of the profile increases with the age of the solution and is not observed to the same extent from similar age toluene extracts of the same soot samples. In addition, this first component is not observed from the soot or residue samples, which show only the second two components. This indicates that reactions are occurring in the pyridine extract solutions. The third component of the profile is made up of C,- and (La,@C,)- and peaks a t approximately 500 'C. The fact that the middle component, composed of mostly C;, is easily distinguished from the third component, composed of (La,@C,)- and C;, suggests that there is some interaction between C, and La,@C,. Loosely bound complexes of La,@C,.C, might desorb at higher temperatures than C, but dissociate in the process, as no such complex ions were observed in these studies. If the metallofullerenes are in the form of ionic complexes with fullerenes, e.g., (Lax@Cn)+C;,sJo then these species might be soluble in more polar solvents such as water. To test this hypothesis, the pyridine extract of a lanthana-graphite soot sample was dried and added to water. Figure 4 shows the methane NI/DCI mass spectrum of this solution, which shows both C; and (La,@C,)-. In contrast, an attempt was made to dissolve a dried toluene extract containing mostly Cs0and C70, in water, yet no fullerenes were observed in the NI/DCI mass spectrum. In addition, the aqueous solution of the pyridine extract of the lanthana-graphite has a yellowish color, indicating dissolution of some species, which is supported by the mass spectrometric analysis.
Summary Negative ion/desorption chemical ionization mass spectrometry was used to analyze metallofullerenes provided by arcing metal oxide/graphite rods and provides a simple and sensitive alternative to laser desorption mass spectrometry. It was shown that although certain metallofullerenes, such as La@CB2and La2@Cso,are efficiently extracted by pyridine, the soot contains other La @ C,, most notably L ~ @ I Cand ~ ~La@C70,which may not be soluble in the solvents tried to date. Extensive distributions of monometallofullerenes are observed, in contrast to the dimetallofullerenes for which only the M2@CB0species are detected in high abundance. A mixed-metal dimetallofullerene was produced by arcing a mixed-metal oxide (Yz0,/La2O3)-impregnated graphite
J . Phys. Chem. 1992, 96, 5234-5231
5234
rod. Analyses of pyridine extracts of mixed YfLa- and ScfYgraphite soot showed greater abundances of metallofullerenes containing the lower ionization potential metal atom. This observation suggests that growth reactions between atomic ions and fullerene precursor species may be important in the arc plasma. However, there are other periodic properties of atoms or ions (e&, size) that might account for these observations. The thermal desorption profiles provided some indication of metallofullereneffullerene interactions, which is supported by solubility of C,.La,@C, species in water. These observations are consistent with the speculation that metallofullerenes exist as ionic complexes with fullerenes, (La,@C,)+C;. Acknowledgment. We thank JoAnn Milliken and Andy Baronavski for technical assistance. We are grateful to Rob Whetten for some initial lanthana-graphite soot extracts, Prof. H. Shinohara for providing preprints, and Rob Whetten, Rick Smalley, Mike Alford, and Francois Diederich for helpful discussions. We acknowledge the Office of Naval Research for support of this research. References and Notes ( I ) Heath, J . R.; Q’Brien, S. C.; Zhang, Q.; Liu, Y.; Curl, R. F.; Kroto, H . W.; Tittel, F. K.; Smalley, R. E. J. Am. Chem. SOC.1985, 107, 7779. Weiss, F. D.; Elkind, J. L.; OBrien, S. C.; Curl, R. F.; Smalley, R. E. J. Am. Chem. SOC.1988, 110, 4464.
(2) Kratschmer, W.; Lamb, L. D.; Fostiropoulos, K.; Huffman, D. R. Nature 1990, 347, 354. Kratschmer, W.; Fostiropoulos, K.; Huffman, D. R. Chem. Phys. Lett. 1990, 170, 167. (3) Haufler, R. E.; Chai, Y.; Chibante, L. P. F.; Conceicao, J.; Jin, Changming; Wang, Lai-Sheng; Maruyama, Shigeo; Smalley, R. E. Mater. Res. SOC.Symp. Proc. 1990, 206, 627. (4) Chai, Y.; Guo, T.; Jin, C.; Haufler, R. E.; Chibante, L. P. F.; Fure, J.; Wang, L.; Alford, J. M.; Smalley, R. E. J. Phys. Chem. 1991, 95, 7564. (5) Weaver, J. H.; Chai, Y.; Kroll, G.H.; Jin, C.; Ohno, T. R.; Haufler, R. E.; Guo, T.; Alford, J. M.; Conceicao, J.; Chibante, L. P. F.; Jain, A,; Palmer, G.;Smalley, R. E. Chem. Phys. Lett. 1992, 190, 460. (6) Johnson, R. E.; de Vries, M. S.; Salem, J.; Bethune, D. S.; Yannoni, C. S. Nature 1992, 355, 239. (7) Shinohara, H.; Sato, H.; Saito, Y.; Ohkohchi, M.; Ando, Y . J. Phys. Chem., submitted for publication. Shinohara, H.; Sato, H.; Ohkohchi, M.; Ando, Y.; Kodama, T.; Shida, T.; Kato, T.; Saito, Y. Nature, submitted for publication. (8) Alvarez, M. M.; Gillan, E. G.;Holczer, K.; Kaner, R. B.; Min, K.S.; Whetten, R. L. J . Phys. Chem. 1991, 95, 10561. (9) McElvany, S. W. J . Phys. Chem., in press. (IO) Diederich, F.; Rubin, Y. Angew. Chem., submitted for publication. ( 1 1) Ben-Amotz, D.; Cooks, R. G.;Dejarme, L.; Gunderson, J. C.; Hoke, S. H.; Kahr, B.; Payne, G. L.; Wood, J. M. Chem. Phys. Lett. 1991,183, 149. (12) McElvany, S. W.; Callahan, J. H . J. Phys. Chem. 1991, 95, 6186. (13) McElvany, S. W.; Callahan, J. H.; Ross, M. M. Unpublished results. (14) Smalley, R. E. Private communication. (15) Smart, C.; Eldridge, B.; Reuter, W.; Zimmerman, J. A.; Creasy, W. R.; Rivera, N.; Ruoff, R. S. Chem. Phys. Lett. 1992, 188, 171. ( I 6) Photochemically induced products of pure fullerenes in solution have been reported by: Wood, J. M.; Kahr, B.; Hoke, S. H.; Dejarme, L.; Cooks, R. G.;Ben-Amotz, D. J . Am. Chem. SOC.1991, 113, 5907.
Size Evolution of Solvent Vibrational Structure in Neutral Solute-( Solvent), Clusters: Benzene-(N,),, n = 1-32 Vincent A. Venturo, Patrick M. Maxton, and Peter M. Felker*qt Department of Chemistry, University of California, Los Angeles, California 90024- I569 (Received: April 16, 1992; In Final Form: May 6, 1992)
Size-selective Raman spectra of the N2 moieties in benzene-(N,), clusters (n = 1-32) have been measured by mass-selective ionization-loss stimulated Raman spectroscopy. The evolution of the spectra is interpreted in terms of the structure of the solvent N2 species about the benzene solute.
Introduction In the past decade numerous spectroscopic experiments on neutral solute-(solvent), clusters in supersonic molecular beams have been reported (e.g., refs 1-14). A major motivation for studying clusters of this type derives from the fact that such species can serve as model systems for the characterization of local structure and dynamics in real solutions. Aside from this, the species provide an opportunity to study many-body dynamics and finite-size effects in systems of weakly interacting species. Thus far, probably the most prolific and informative approach to the study of neutral solute-(solvent), clusters for n greater than three or four has been resonantly enhanced multiphoton ionization mass spectroscopy, which allows the measurement of size-selective vibronic spectra.’” Other important schemes involve measuring the infrared absorption of the clusters by bolometer detection’ or by ion-depletion methods.Ic A third approach consists of observing vibronic resonances by fluorescence excitation spectroscopy.*-” More recent experiments have employed vibronic hole-burning spectr~scopy’~-’~ and photoionization threshold measurement^'^ to characterize cluster level structures in a size-selective manner. In all of these studies the spectra that are measured correspond to transitions within the solute chromophore. From such observations properties of the cluster as a whole are then inferred. *To whom correspondence should be addressed.
’NSF Presidential Young Investigator, 1987-92. 0022-36S4/92/2096-5234$03.00/0
An obvious and potentially very powerful way to extend one’s knowledge of neutral solute-(solvent),, clusters beyond what can be learned from studying the effects of clustering on the solute chromophore is to study the solvent chromophores as well. Indeed, by this approach one might expect to obtain the most direct information about solvent structure and dynamics, as indicated by the results of infrared experiments on solvated-ion cluster^.'^ There are two major problems with such an approach, however. First, while it is advantageous to work with simple solvent species (rare gases, diatomics) so as to facilitate the synthesis of the clusters and the interpretation of the spectroscopic results, such species tend not to have readily accessible vibronic chromophores. Second, if the solvent species does have accessible electric-dipole-allowed transitions, then dipoledipole coupling between the solvent moieties will broaden those transitions significantly for all clusters except those corresponding to the lowest values of n. The aforementioned problems associated with probing solvent-localized resonances can be circumvented by using Raman spectroscopies rather than spectroscopies based on single-photon transitions. Even homonuclear diatomics have allowed Raman transitions. Moreover, in many cases a molecule excited in a Raman transition is only coupled to other molecules of its type by weak, higher-order multipole-multipole interactions. The advantages in cluster studies of these two features of the Raman effect have been amply demonstrated by Beck et a1.16 in their stimulated Raman loss experiments on large, homogeneous clusters of N,. Unfortunately, the stimulated Raman loss technique does 0 1992 American Chemical Society
