|Home | About | Journals | Submit | Contact Us | Français|
Metallic nitride fullerenes (MNFs) and oxometallic fullerenes (OMFs) react quickly with an array of Lewis acids. Empty-cage fullerenes are largely unreactive under conditions used in this study. The reactivity order is Sc4O2@Ih-C80 > Sc3N@C78 > Sc3N@C68 > Sc3N@D5h-C80 > Sc3N@Ih-C80. Manipulations of Lewis acids, molar ratios and kinetic differences within the family of OMF and MNF metallofullerenes are demonstrated in a selective precipitation scheme, which can be used either alone for purifying Sc3N@Ih-C80 or combined with a final HPLC pass for Sc4O2@Ih-C80, Sc3N@D5h-C80, Sc3N@C68, or Sc3N@C78. The purification process is scalable. Analysis of experimental rate constants versus electrochemical band gap explains the order of reactivity among the OMF and MNFs.
Metallic Nitride Fullerenes (MNFs, e.g., Sc3N@Ih-C80) consist of a trimetallic nitride cluster trapped within the carbon housing of fullerene cages.1-5 In contrast, the newly discovered OxoMetallic Fullerenes (OMFs) consist of encapsulated metal oxide clusters within fullerene cages.6,7 For the OMFs, little is known with regard to their chemical reactivity. The dominant representatives of MNF and OMF compounds are Sc3N@Ih-C80 and Sc4O2@Ih-C80, respectively. Of the former, Sc3N@D5h-C80 is a structural isomer of minor abundance.8,9
The separation of C60 from other empty-cages fullerenes (e.g., C70) using Lewis acids has been previously described10 based on C60 being the more inert species. For our metallofullerene system, we hypothesize a dramatic increase in reactivity for MNFs and OMFs based on their cages having formal charges of -6 versus the neutrality of empty-cage fullerenes.11
The major hurdle to MNF and OMF experimentation is poor availability of isomerically pure samples. This paucity of materials is due to inefficient separation technologies which historically include classical HPLC methods. Recent reports of non-chromatographic methods for isolating MNFs are based on their resistance to reaction with solid supports such as cyclopentadiene immobilized on Merrifield resin12 or aminocapped silica.13,14 A support free method of separating MNFs from empty-cage fullerenes has been achieved using molten 9-methylanthracene.15 Another approach to separate MNFs is an electrochemical, oxidation-based differentiation between Ih and D5h isomers of Sc3N@C80.16 An alternative method to isolate non-MNF metallofullerenes (e.g., Gd@C70, Gd@C82) from empty-cage fullerenes exploits differences in solubility and redox reactivity.17
Herein we report the reactivity of OMF and MNF as new Lewis bases and their selective complexation with Lewis acids. Subsequent manipulation of kinetic differences between these species can be used in development of a new separation scheme, which permits isolation of individual OMF and MNF compounds.
Soots containing OMF and MNFs were prepared using a cylindrical electric-arc reactor as previously described.18 Cored rods were then packed with Cu (Cerac) or Cu(NO3)2·2.5 H2O (Aldrich) and vaporized using the CAPTEAR process6,19 for enhanced yield of OMF and MNF compounds. The soot was extracted with CS2 or o-xylene, filtered, and the solvent was removed under reduced pressure to furnish a dried extract, which was washed with diethyl ether or acetone. Soot extracts were weighed and characterized by HPLC to determine the type and amount of fullerene material present. HPLC peak areas were obtained using standard chromatographic integration software (Vernier, Logger Pro). HPLC separations were as follows: PYE column (4.6 mm × 250 mm), flow rates of 0.3 mL/min toluene or 0.5 mL/min xylenes as the mobile phase, and UV detection at 360 nm.
For comparison of C60, C70, and Sc3N@Ih-C80 reactivities, 1.0 mg of each fullerene type was dissolved in 3 mL of carbon disulfide. To each solution was added 40 mg of AlCl3. Time lapse photography was used to monitor and compare the loss of color (i.e., removal of fullerene from solution via precipitation of the fullerene-Lewis acid complex). After 2 min of reaction time the solution in the vial containing Sc3N@Ih-C80 was colorless.
7.1 mg MgCl2, 10 mg AlCl3, 12 mg FeCl3 or 20 mg AlBr3 were added to four stirring solutions containing 15 mg each of fullerene extract dissolved in 150 mL carbon disulfide. Equimolar amounts of Lewis acids were used to compare the speeds of reactions, which were allowed to proceed for a minimum of 3 minutes. Reaction mixtures having slower kinetics were monitored for longer times.
A 1 g sample of extract (~1.2 mmol fullerenes) was dissolved in 1 L of carbon disulfide. While stirring, 217 mg FeCl3 was added to the fullerene solution. Aliquots of the reaction mixture were taken and analyzed via HPLC, whose conditions were 0.3 mL/min toluene mobile phase, 360 nm UV detection, and 10 μL injection onto a 4.6 mm × 10 mm PYE column. Conversion from peak area to molarity was performed via use of extinction coefficients and standardized samples of purified fullerenes as previously described.20
To a 2 mg fullerene sample enriched in OMF and MNFs was added 15 mL CS2. While stirring, 10 mg of AlCl3 was added to generate a reaction mixture, from which aliquots were analyzed as described in experimental section C.1.
Extract solutions containing ~1.3 g of fullerenes dissolved in 500 mL of carbon disulfide were prepared. To each of these 3 solutions was added separately, while stirring, 1.75 g AlBr3, 240 mg FeCl3, and 198 mg AlCl3. The reactions were allowed to proceed a minimum of 3 hours. The reaction mixtures were filtered, and the precipitate contained primarily OMF and MNF fullerenes complexed to the Lewis acid. Upon addition of ice water, sodium bicarbonate and carbon disulfide to the solid material remaining on the paper filter from a Buchner funnel, these fullerenes were released from the complex and readily dissolved in the CS2 layer (i.e., bottom layer in a separatory funnel). After several washes with deionized water, this CS2 fullerene solution was filtered by membrane filtration. Solvent was removed via rotary evaporation, and the solid material (i.e., recovered fullerenes) was ether-washed, dried and weighed. The filtrate from the reaction mixture was also washed with water and sodium bicarbonate as described above. The masses of dried fullerenes obtained from the filtrate and precipitate were added and compared to the original extract mass for percent recovery calculations.
982 mg of Sc fullerene extract was dissolved in 500 mL carbon disulfide. While stirring, 340 mg AlCl3 was added to this solution. Reaction progress was monitored by loss of HPLC peak area for Sc4O2@Ih-C80. The reaction was stopped at 44 hours. Upon filtration the precipitate containing Sc4O2@Ih-C80 OMF and contaminant MNFs was treated as described in experimental section D. A second step using 45 mg of this OMF and MNF enriched fullerene material dissolved in 250 mL carbon disulfide was used. While stirring, 135 mg of AlCl3 was added. The reaction time was 4 h 40 min, at which time the resulting precipitate was processed as described in experimental section D to recover the enriched Sc4O2@Ih-C80 fullerene sample.
An extract solution was prepared by dissolving ~1 g fullerenes in 500 mL of CS2. To this extract solution was added 340 mg AlCl3. The reaction proceeded for 44 h, at which time the collected precipitate was processed as described in experimental section D.
This experiment used the filtrate obtained from step 1 (see AlCl3 chemistry, Figure 6e). To the filtrate, with stirring, was added 150 mg FeCl3. After 70 minutes of reaction time, the Ih isomer of Sc3N@C80 had complexed with the Lewis acid and was precipitated from solution. Upon filtration, the collected precipitate was processed as described in experimental section D to obtain 27 mg of isomerically purified Sc3N@Ih-C80 (Figure 6f).
An extract solution of 1340 mg fullerenes was dissolved in 500 mL carbon disulfide. While stirring, 245 mg AlCl3 was added. After 21 h, the reaction mixture was filtered to remove OMF and MNF contaminants of Sc3N@C68, Sc3N@C78, and Sc3N@D5h-C80. The filtrate, containing 1.072 g fullerenes, of which Sc3N@Ih-C80 is the primary metallofullerene in addition to empty-cage fullerenes (e.g., C60, C70, C84), was diluted to 1 L with CS2. To this solution, while stirring, was added 217 mg FeCl3 to precipitate the Ih isomer of Sc3N@Ih-C80. After stirring for 55 minutes, the Sc3N@Ih-C80 complex was precipitated from solution. The reaction mixture was filtered, and the precipitate was processed as described in experimental section D. An isolated sample of 0.107 g of Sc3N@Ih-C80 was obtained.
For comparing reactivity differences between empty-cage fullerenes and metallofullerenes having a C806- cage (e.g., OMFs, MNFs), an experiment was performed in which 1.0 mg each of C60 (1.4 μmol), C70 (1.2 μmol), and Sc3N@Ih-C80 (0.9 μmol) was dissolved in 3 mL carbon disulfide with a large molar excess of AlCl3 (40 mg, 300 μmol). Results from Figure 1 demonstrate removal of Sc3N@Ih-C80 within 2 minutes. In contrast, empty-cage fullerenes C60 and C70 were more resistant to complexation and precipitation.
Given this difference in reactivity, we expanded the palette of Lewis acids. Weaker Lewis acids such as MgCl2 were nonresponsive under our experimental conditions, with the stronger Lewis acids of AlCl3, FeCl3 and AlBr3 being much more reactive to MNFs. For these experiments, equimolar ratios of Lewis acids were used for direct comparison. Assuming pseudo-first order kinetics, graphs obtained from the log of fullerene concentration versus time (Figure 2) resulted in linear plots, from which kobs rate data was readily obtained. These rate constants for reaction of Sc3N@Ih-C80 with Lewis acids were 0.605 min-1 (AlBr3), 0.302 min-1 (FeCl3), 0.0124 min-1 (AlCl3), and ~0 min-1 (MgCl2). The data indicate that AlBr3 and FeCl3 react much more quickly with MNFs than does AlCl3.
It is well known that the Ih isomer of Sc3N@C80 is less reactive than Sc3N@D5h-C80 for other types of reactions (i.e., non-Lewis acid reactions) such as cycloadditions.21-26 To determine whether a similar trend occurs with Lewis acids, an experiment was performed with a stronger Lewis acid such as FeCl3 to ensure sufficient reaction with both isomers. Aliquots at arbitrary times were collected to monitor loss of peak area for both Sc3N@C80 isomers from solution. Using 1st order kinetics of uptake (Figure 3a), the ratio of kobs (Sc3N@D5h-C80) to kobs (Sc3N@Ih-C80) was 1.6. Our finding of Sc3N@D5h-C80 being more reactive than Sc3N@Ih-C80 to Lewis acids is consistent with literature reports of Sc3N@D5h-C80 being more reactive than Sc3N@Ih-C80.21-26 The data in Figure 3b clearly shows isomeric purity can be achieved at only 55 minutes of reaction time. This more rapid removal (55 min) is compared with the 13 h of reaction time for Sc3N@D5h-C80 removal with the SAFA process, which uses aminosilica to selectively bind reactive fullerene and metallofullerene species.14
For reactivity comparisons among other MNF and OMF metallofullerenes, we selected AlCl3 based on its slower reaction kinetics. Of particular interest was probing the reactivity of the OMF Sc4O2@Ih-C80 species relative to MNF compounds such as Sc3N@C68, Sc3N@C78, Sc3N@D5h-C80, and Sc3N@Ih-C80. For these experiments, 2 mg of a fullerene sample enriched in these compounds was dissolved in 15 mL of CS2. While stirring, 10 mg of AlCl3 was added to generate a reaction mixture from which aliquots were taken at arbitrary times to monitor fullerenes remaining in solution (i.e., fullerenes not bound and precipitated by the Lewis acid). The amount of C60 and C70 remaining in solution is relatively constant. The logarithm of fullerene concentration was plotted as a function of time and 1st order kinetics were observed. Results comparing the reactivities of OMF and MNFs are summarized in Table 1. The significance of these results is the notion that one could manipulate these reactivity differences and develop a new method for purifying these metallofullerenes.
The variation in rate constants among these metallofullerenes may be related to the electrochemical (EC) band gap. When our experimental kinetic data (i.e., kobs) from Table 1 is plotted versus published electrochemical data for C70,27,28 Sc3N@Ih-C80,9,16,29 Sc3N@D5h-C80,29 Sc3N@C68,30 and Sc3N@C78,31 a correlation can be made as shown in Figure 4. Based on this proportionality between rate constant and band gap, a predicted EC band gap of ~1V for Sc4O2@Ih-C80 can be made.
Several key issues include the scalability and ability to release the fullerene from the precipitated Lewis acid complex. To address these concerns, several gram scale reactions were performed using three different types of Lewis acids. Solutions of ~1300 mg Sc-fullerene extract in 500 mL CS2 were prepared for reaction with 1.75 g AlBr3, 240 mg FeCl3, or 198 mg AlCl3. After 3 hours of reaction time, the reaction mixture was filtered to yield a precipitate, to which was added ice water, sodium bicarbonate, and CS2. Under these conditions, OMF and MNF fullerenes were released from the Lewis acid-fullerene complexes and dissolved in CS2. As a representative example, HPLC chromatograms from the FeCl3 experiments are shown in Figure 5. The HPLC results indicate selective precipitation and removal of OMF and MNFs from the unreacted empty-cage fullerenes remaining in solution (Figure 5b). The OMF and MNFs are readily recovered from the precipitate (Figure 5c) using the procedure in Experimental section D. Fullerene recoveries were 83-86%, regardless of which Lewis acid was utilized. Results from these scale-up and recovery experiments are provided in Table 2.
Moving to other species beyond the more chemically inert Sc3N@Ih-C80, it would be advantageous to develop and optimize this new method of selective Lewis acid precipitation toward the more reactive MNFs and OMFs. The advantage of determining the reactivity order of OMF and MNFs is the ability to subsequently manipulate the kinetics such that the more reactive species can be precipitated, at which time the reaction can be stopped, thereby leaving the majority of the more chemically inert species still in solution (Figure 6a). To demonstrate this concept, 982 mg of fullerene extract obtained from the vaporization of Sc2O3 packed graphite rods were dissolved in 500 mL CS2. While stirring, 340 mg of AlCl3 were added. Reaction progress (i.e., fullerene loss from solution) was monitored at arbitrary times. Mass spectral data (supporting information) indicates Sc4O2@Ih-C80 is the first of the OMF and MNF species to be precipitated from solution (t = 20 h). This result is consistent with the reactivity comparisons described above (e.g., Sc4O2@Ih-C80 > Sc3N@C78 > Sc3N@C68 > Sc3N@D5h-C80 > Sc3N@Ih-C80). Reaction beyond 20 h to 44 h results in further precipitation of Sc3N@C78, Sc3N@C68, Sc3N@D5h-C80, and a small quantity of Sc3N@Ih-C80 as shown in Figure 6b. A 2nd step with Lewis acid chemistry to enrich this sample in Sc4O2@Ih-C80 also utilizes AlCl3. Shown in Figures 6c, 6d are the HPLC and MALDI mass spectrum for the fullerene recovered precipitate obtained after 4 h 40 min for a reaction mixture of a 45 mg of the sample from Figure 6b, 250 mL CS2 and 135 mg AlCl3. Based on the MALDI data (Figure 6d), Sc4O2@Ih-C80 is the dominant species.
Reaction conditions for isolating enriched samples of Sc3N@D5h-C80 involve stirring a fullerene solution of ~1 g fullerene extract in 500 mL CS2 with 340 mg AlCl3 for 44 h. The data in Figure 6b clearly indicate isolation of an enriched fraction of Sc3N@D5h-C80 (50 mg sample). Note that the dominant peak in the HPLC chromatogram (Figure 6b) is the D5h isomer of Sc3N@C80. With the overwhelming majority of fundamental science focusing on the Ih isomer, a benefit of this Lewis acid approach is the ability to obtain samples in which the dominant species is Sc3N@D5h-C80. If desired, a final HPLC pass of this sample would yield a purified sample of Sc3N@D5h-C80, and HPLC details for isolating Sc3N@D5h-C80 have been published elsewhere.8,9
For isolation of isomerically pure Sc3N@Ih-C80, the filtrate obtained after AlCl3 chemistry (Step 1, Figure 6a) can be mixed with 150 mg FeCl3 as shown in Figure 6e. With other MNFs previously precipitated as described in E.1, this filtrate contains empty-cage fullerenes along with Sc3N@Ih-C80 as the predominant metallofullerene. Hence, precipitation of this MNF via reaction with FeCl3 (70 min) and subsequent workup results in isomerically purified Sc3N@Ih-C80 as confirmed by HPLC. (Figure 6f). To determine the purity of Sc3N@Ih-C80 relative to other fullerenes, a MALDI mass spectrum (Figure 6g) indicates a sample of >99% metallofullerene purity.
With Sc4O2@Ih-C80 OMF being a “bookend” compound with the highest reactivity, one can manipulate the kinetics in favor of the other “bookend” species, the more inert Sc3N@Ih-C80. Isolation of the Ih isomer of Sc3N@C80 is accomplished by monitoring loss of peak areas for the other MNFs and OMF species (i.e., selective precipitation of Sc4O2@Ih-C80, Sc3N@C78, Sc3N@C68, and Sc3N@D5h-C80) and stopping the reaction when these contaminant peaks disappear from the HPLC plot. Optimal chromatographic conditions include use of a PYE (pyrenyl-ethyl) stationary phase with slow flow rates such that elution of the D5h and Ih isomers of Sc3N@C80 are at least 60 min. Since Sc3N@D5h-C80 is adjacent to Sc3N@Ih-C80 in the reactivity trend for Lewis acids, monitoring loss of peak area for the D5h isomer is critical for knowledge on when to cease the reaction. Upon loss of HPLC peak area for the D5h isomer, the filtrate contains only unreacted empty-cage fullerenes (e.g., C60, C70, C76, C78, C84) and predominantly Sc3N@Ih-C80. The Sc3N@Ih-C80 can then be rapidly precipitated via reaction with a stronger Lewis acid (e.g., AlBr3, FeCl3). An overview of the sequential separation of metallofullerenes is provided in Scheme 1.
To demonstrate this concept in Scheme 1 and isolate large quantities of Sc3N@Ih-C80, an extract solution of 1340 mg fullerenes was dissolved in 500 mL carbon disulfide. While stirring, 245 mg AlCl3 was added. After 21 h, the reaction mixture was filtered to remove OMF and MNF contaminants of Sc4O2@Ih-C80, Sc3N@C68, Sc3N@C78, and Sc3N@D5h-C80. The filtrate (1.072 g fullerenes), of which Sc3N@Ih-C80 is the primary metallofullerene, was diluted to 1 L with CS2. To this stirring solution, was added 217 mg FeCl3 to precipitate the Sc3N@Ih-C80. After 55 min of reaction, Sc3N@Ih-C80 was precipitated from solution. The precipitate containing Sc3N@Ih-C80 complexed to FeCl3 was processed as described in experimental section D. Upon solvent removal and drying, 0.107 g of purified Sc3N@Ih-C80 was obtained. The resulting HPLC trace and MALDI mass spectrum for this isolated sample are of equivalent purity to those shown in Figure 6 f,g.
OxoMetallic fullerene (OMF) and metallic nitride fullerene (MNF) endohedral metallofullerenes react quickly with Lewis acids. The empty-cage fullerenes are largely unreactive under the molar ratios used in this study. The reactivity order is Sc4O2@Ih-C80 > Sc3N@C78 > Sc3N@C68 > Sc3N@D5h-C80 > Sc3N@Ih-C80. Graphical analysis of experimental rate constants versus electrochemical band gap explains the order of reactivity among the OMF and MNFs. Manipulation of Lewis acids and kinetic differences result in a selective precipitation scheme, which can be used alone for Sc3N@Ih-C80 or combined with a final HPLC pass for Sc4O2@Ih-C80, Sc3N@D5h-C80, Sc3N@C68, Sc3N@C78. The purification process is scalable. Efforts to expand this approach to other homometallic nitride fullerenes (e.g., Gd3N@C80) and mixed-metal nitride fullerenes (e.g., LaSc2N@C80) are underway.
Prof. Stevenson thanks the NSF CAREER (CHE-0547988), Lucas Research Foundation, and MALDI instrumentation grant NSF 0619455 for financial assistance. Prof. Phillips thanks the NSF CAREER (CHE-0847481) and NIH R15AG028408. MAM thanks the NSF GRFP program.