Introduction

Endohedral metallofullerenes (EMFs) exhibit properties that attract attention for use as contrast agents and other biomedical diagnostics, photovoltaics and materials1,2,3. EMFs are also of interest to the chemist, because the void within the cage acts as a nanoscale container for the study of remarkable structures and unusual systems4,5,6. Although first observed only days after the discovery of Buckminsterfullerene (C60)7,8, the mechanism of metallofullerene formation remains unclear despite nearly three decades of intensive research, demonstrating the challenge of achieving in situ mechanistic insight under representative EMF-generating conditions. EMFs are synthesized in the gas phase from high-temperature, high-density carbon vapour by evaporation of metal-doped graphite9. All production methods involve creation of a carbon plasma with co-evaporated metal in the presence of an inert gas such as helium7,10,11. Therefore, the method of energy input used to vapourize graphite and metal is independent of the primary reactions that result in metallofullerene formation. It is generally established that a plasma serves as the starting point for synthesis by the electric arc discharge and laser vapourization techniques7,10,12. In fact, a distinct carbon–metal vapour (plasma) plume is physically observable during graphite evaporation13. Notably, one of the most vigorously exploited EMFs, M@C82 (M=metal), has been macroscopically synthesized by the use of both methods since the dawn of metallofullerene science12,14.

Top-down formation proposals have been predominantly advanced to explain fullerene synthesis over the past decade. By means of quantum chemical dynamics simulations, Irle et al.15 concluded that spontaneous organization of giant carbon cages occurs first, after which C2 elimination subsequently produces smaller fullerenes such as C60. Recently, Chuvilin et al.16 demonstrated the transformation of graphene into a fullerene cage by electron beam irradiation in transmission electron microscopy studies. Very recently, Dorn and colleagues17 isolated and characterized a metallofullerene proposed to be a key intermediate in the top-down formation of EMFs from graphite. However, we have recently reported that empty cages, such as C60 and C70, may form through a different mechanism18.

Here we present an alternate route to the formation of metallofullerenes, performed under characteristic synthesis conditions. We show that most known metal-encapsulated carbon cage sizes assemble through a bottom-up formation process. The prototypical M@C82 metallofullerene is used to explicitly substantiate the nanocarbon reaction mechanism. More than 80 elements are individually investigated for mono-metallofullerene formation from metal-incorporated graphite to elucidate factors that control the process, in combination with density functional theory computations. Oxidation state of the encapsulated metal and charge transfer to the carbon cage are determinant factors that govern EMF formation.

Results

Molecular behaviour of Pr@C82 in evaporated graphite vapour

Pr@C82 is synthesized by evaporation of Pr-doped graphite by the use of the electric arc plasma discharge technique. Two metal-encapsulated carbon cages are isolated from the resulting soot by chemical separation19, and then isomerically purified by multistep high-performance liquid chromatography (HPLC). The recovered species Pr@C82 (I) and Pr@C82 (II)20, labelled based on HPLC elution order, exhibit cages C2v(9)-C82 and Cs(6)-C82. The Pr@C82 compounds are described by an ionic model, in which valence electrons from Pr are formally donated to the cage, to give the charge distribution Pr3+@C823−. These particular EMF cages are related by a single C2 loss to the prominent M@C80 cage (M=metal or cluster)4,21,22,23 and are linked to all smaller endohedral fullerene cage sizes by further C2 eliminations. By contrast, a single C2 insertion reaction distinguishes Pr@C82 from the larger M@C84 cage24,25,26, and by further C2 gain is connected to larger endohedrals27,28. Pr@C82 occupies an intermediate position between well-known larger and smaller EMFs and provides an ideal platform to study the mechanism of metallofullerene formation. Mono-metallofullerenes offer compelling experimental evidence, because there is no ambiguity in product cage size29,30 and the extent of charge transfer from metal to cage is well defined.

Figure 1 shows molecular cage behaviour and reactivity of Pr@C82 (I) under synthetic conditions that generate EMFs, namely at high temperature, in the presence of carbon evaporated from graphite, and at a low-pressure of He. Direct sampling of the chemical formation process is achieved by the use of a pulsed laser vapourization source18,31, analysed by high-resolution Fourier transform ion cyclotron resonance (FT-ICR) mass spectrometry32. Pr@C82 is unambiguously shown to undergo C2 insertion reactions to form Pr@C84 in high abundance, whereas a reverse reaction by C2 elimination to Pr@C80 essentially does not occur. All observed product EMFs display excellent agreement with calculated isotope distributions (inset in Fig. 1). Further, the results are consistent with previous investigations of medium-sized empty cage fullerenes (C60, C70, C76, C78, C84) and the endohedral, La2@C80 (ref. 18), which all exhibit growth into larger species without generation of smaller fullerenes in graphite vapour. As shown in Fig. 1, extensive C2 insertion reactions form EMFs significantly larger than Pr@C84, including giant metallofullerenes (>M@C100). The endohedral nature of the Pr@C2n (C2n≥C84) products is confirmed by dissociation of isolated formation products (Supplementary Fig. 1). Carbon nanostructures smaller than Pr@C82 are primarily empty cages that result from bottom-up growth of C80, along with low abundance Pr@C80 and Pr@C78. The species C80, Pr@C78 and Pr@C80 are attributed to result from an unavoidable interaction of the vapourization laser and the Pr@C82 starting material, as indicated by Supplementary Figs 2 and 3. Small M@C2n (C2n ≤76) species, although abundantly generated by vapourization of graphite, are entirely absent in Fig. 1. Further, we note that those small endohedrals have received little study, because they are generally not detectable by use of the common arc discharge technique, as shown in Supplementary Fig. 4. The second isomer, Pr@C82 (II), exhibits similar behaviour under EMF synthesis conditions. Moreover, we have confirmed bottom-up formation of medium-sized cages to larger EMFs with other elements encapsulated in M@C82, such as Y@C82 and Yb@C82.

Figure 1: Bottom-up formation of large and giant mono-EMFs from Pr@C2v-C82.
figure 1

(a) FT-ICR mass spectrum of cluster cations after reaction of Pr@C82 with carbon evaporated from graphite in a low-pressure He atmosphere and (b) reaction scheme, with possible structures of the larger EMFs. The starting material is shown in blue, whereas bottom-up formation products are shown in orange.

M@C2n formation from metal-doped graphite

Consequently, smaller compounds mark the start of EMF production from metal-doped graphite. Figure 2 shows fullerene synthesis products by laser evaporation of metal-incorporated graphite with representative elements from Groups 1, 2 and 3. All experiments are performed under similar formation conditions that result in a complex mixture of metallofullerenes and empty cages. EMFs that encapsulate a single metal atom (Supplementary Fig. 5) range from M@C28 to greater than M@C100 in some cases, and all experimental distributions are highly reproducible (see Supplementary Fig. 6). The smallest M@C2n that form for a given element under the present conditions are shown in Figs 2 and 3; however, the largest M@C2n generally extend past ~M@C84. Experimental parameters are selected to yield C60 as a dominant fullerene, thereby facilitating comparison of EMF with empty cage assembly and verifying consistent formation conditions among the large number of elements studied. In these strongly ionizing environments, positive ions are expected to be directly representative of neutral EMF distributions7. Supplementary Fig. 7 shows that calculated ionization potentials for Ca@C2n and Ti@C2n do not correspond to signal magnitude, providing additional support that observed M@C2n-positive ion distributions reflect neutral distributions.

Figure 2: Group 1–3 containing M@C2n from evaporated metal-doped graphite.
figure 2

Positive ions are representative of neutral distributions. M@C2n and C2n distributions are shown graphically for each mass spectrum. As more negative charge is progressively transferred to the cage, as shown for (a) Cs+@C2n, (b) Ca2+@C2n2− and (c) Y3+@C2n3−, non-IPR structures with particular pentagon motifs become highly stabilized and less reactive toward C2 insertion. The proposed carbon cage for (d) M@C44 contains four sets of TSFPs, (e) M@C50 exhibits five sets of DFPs and all pentagons are isolated for (f) M@C60.

Figure 3: Lanthanide-containing mono-EMFs from evaporated metal-doped graphite.
figure 3

Two distinct formation subsets are observed for lanthanides that involve transfer of two or three electrons to carbon cages, such as (a) Sm@C2n and (b) Nd@C2n.

Striking changes in formation distributions are observed for M@C2n across Groups 1, 2 and 3. Elements that formally donate a single electron to carbon cages, namely Na33, K, Rb and Cs (Fig. 2a and Supplementary Figs 8–10) exhibit M@C60 and M@C70 as dominant species, similar to empty cages. By contrast, a considerable shift toward smaller EMFs occurs for elements that donate two electrons, as for Ca, Sr and Ba (Fig. 2b and Supplementary Figs 11–13), in which M@C50 becomes nearly as abundant as M@C60. Three-electron transfer to cages, as for Group 3 containing EMFs in Fig. 2c and Supplementary Figs 14 and 15, affords a distribution that is more sharply skewed to smaller species, for example, highly abundant M@C44 (for M=Sc, Y). All M@C2n synthesized from evaporated metal/graphite show outstanding agreement between calculated and experiment isotope distributions. M@C82 is clearly observed for all elements that form endohedral fullerenes from graphite. Relative abundance of that compound can be enhanced by adjustment of formation conditions18,34 (namely, by increasing laser fluence to create more available carbon vapour); however, small EMFs still dominate the products of formation. Therefore, the observed M@C2n trends are primarily equilibrium distributions under the present conditions, because most of the available carbon vapour formed should be able to react, that is, exhibit carbon insertion reactions with smaller M@C2n. It follows that there must be some point during the reaction that no ‘new’ small EMFs can be generated, but already formed M@C2n may still continue to grow bottom-up into larger species.

Two distinct formation categories for the lanthanide series correspond to divalent and trivalent elements (Supplementary Figs 16–29). Sm@C2n is representative of divalent lanthanide EMFs, M2+@C2n2−. As shown in Fig. 3a, Sm@C50 is formed in conspicuous abundance that rivals Sm@C60, a distribution pattern that mimics Group 2 EMFs. Nd@C2n, a representative element for trivalent lanthanide EMFs (Ln3+@C2n3−), is also shown in Fig. 3b. Nd@C44 exhibits unusual abundance and an overall formation distribution that is nearly identical to Group 3 EMFs. We recently reported that Group 4 elements exhibit strongly altered formation distributions with the appearance of M@C28 (ref. 35). Figure 3 demonstrates the ultra-high resolution of the present FT-ICR mass spectrometer that enables in situ investigation of many of these metal–carbon systems for the first time.

Gas-phase isolation of M@C2n EMFs (C2n=44, 50, 60, 70, 80, 90) is carried out by the use of a stored waveform inverse Fourier transform event36. Subsequent structural investigation by sustained off-resonant irradiation collision-induced dissociation37 confirms the endohedral location of elements inside the carbon cages (Supplementary Note 1). Stored waveform inverse Fourier transform and sustained off-resonant irradiation collision-induced dissociation product ion mass spectra for M@C60 with all EMF-forming elements are shown in Supplementary Figs 35–39. Highly thermally excited M@C60 does not produce the smaller unconventional ‘missing metallofullerenes’ M@C28, M@C36, M@C44 or M@C50 in significant abundance under our conditions, providing additional support for bottom-up formation of these species from graphite. Further support is gained through theoretical investigations probing bottom-up formation of the Ca@C2n family, as shown in Fig. 4.

Figure 4: Computational modelling of bottom-up formation for Ca@C2n.
figure 4

(a) Energy per atom for C2n2− dianions (2n=42–62) and (b) reaction energies for Ca@C2n+C2→Ca@C2n+2 (2n=42–60). The computed values in a are fitted to an exponential function (blue line). The energies for D5h-C502−(271) and Ih-C602−(1,812) isomers, shown as red dots, are found to be 15.2 and 21.6 kcal mol−1 deviated from the general trend, indicating an enhanced stability. In b, computed reaction energies (red line) are compared with experimental relative abundances (green line). Ca@D5h-C50(271) and Ca@Ih-C60(1812) are the least reactive species to C2 insertion, in agreement with experimental observations.

Discussion

Figure 1 experimentally establishes transformation of medium-sized metallofullerene cages into large and giant species by C2 insertion reactions, without formation of smaller EMFs. We propose that the smallest EMF compounds for a given element are the first to form by synthesis from evaporated metal-doped graphite. Ion mobility and other cluster source studies suggest that an evaporated metal ion nucleates carbon to initially form such small EMFs, in agreement with our observations35,38. Ensuing C2 insertions into those small EMFs result in formation of medium-sized compounds, such as solvent-extractable M@C82, which is clearly the origin of large and giant EMFs formed through the bottom-up mechanism shown in Fig. 1. Therefore, conversion of smaller EMFs into larger cages is crucial for synthesis of known endohedrals39,40. Figures 2 and 3, and Supplementary Figs 8–33 show those important formation steps from evaporated graphite. Elements that form mono-metallofullerenes are displayed in Supplementary Fig. 5 and corresponding formation distributions are shown in Fig. 5. We have recently reported Group 4 mono-metallofullerenes35. All Group 1–4 metals, lanthanides and actinides (Supplementary Figs 30 and 31) form M@C2n from evaporated metal-doped graphite, except for Li, Be and Mg under the present conditions. We note, however, that a Li+@C60 material has been recently synthesized by Li+ implantation into pre-existing C60 (ref. 41). All other transition metals do not form EMFs from graphite in detectable abundance under the present conditions. Surprisingly, Ga and In are observed to encapsulate carbon cages (Supplementary Figs 32 and 33).

Figure 5: M@C2n formation distributions with elements across the periodic table.
figure 5

Metallofullerene formation trends for (a) Group 1, (b) Group 2, (c) Group 3 metals, (d) trivalent lanthanides, (e) divalent lanthanides, (f) actinides, and (g) Ga and In from evaporated element-doped graphite. The smallest observed cages are shown for each M@C2n. The largest cages typically extend to ~M@C90 or larger, but are not shown for simplicity.

Negative charge is preferentially located at pentagons or [5,6] bonds in the caged network of metallofullerenes, and to a greater extent when those pentagons are fused to result in [5,5] bonds and [5,5,5] junctions42. Consequently, charge transfer stabilizes highly pyramidalized C atoms induced by pentagon adjacency and cages that do not conform to the isolated pentagon rule (IPR) can become quite favourable systems43. All smaller fullerenes (<C60) must possess various types of important fused pentagon motifs (Fig. 2d–f). Therefore, extent of charge transfer to double fused pentagons (DFPs), triple sequentially fused pentagons (TSFPs) and triple directly fused pentagons (TDFPs), and effect on C2 insertion can be elucidated by in situ study of these small systems with various elements entrapped (Figs 2, 3 and 5). Particular significance is given toward tracking bottom-up transformation of M@C28→M@C36→M@C44→M@C50→M@C60 by C2 insertion reactions. The cage structure of Td-C28 comprises exclusively TDFPs35,44, D6h-C36 is the smallest cage that can exist without TDFPs43,45, D2-C44 consists entirely of TSFPs motifs as described below, D5h-C50 exhibits DFPs without TSFPs or TDFPs46, and Ih-C60 is the smallest in which all pentagons are isolated47. We have recently theoretically predicted that bottom-up construction of these smaller EMFs take place by exergonic C2 insertions with free energy barriers that are attainable at the high temperature of synthesis48, providing strong support for the bottom-up mechanistic route.

All possible isomers are computed for dianions, C2n2− and Ca@C2n EMFs from C42 to C60 (Fig. 4a and Supplementary Table 1), as recently done for tetravalent systems48, Ti@C2n. Carbon cages that encapsulate such divalent cations are advantageous for computational analysis because of the closed shell nature of their ground states. The computed values are fitted to an exponential function, as previously devised for hexanion EMFs49, to discern which Ca@C2n cages exhibit enhanced stability. We find that the relative energies of empty cage dianions and Ca@C2n follow the same trends and, therefore, the ionic model is applicable to Ca-encapsulated EMFs. Further, the most stable Ca@C2n and C2n2− match the least-strained, lowest-energy empty cages; thus, although smaller EMF systems are described by charge transfer from the encapsulated metal to cage (for example, Ca2+@C2n2−), strain is playing a major role in establishing the relative stability of isomers for a particular cage size.

Many of the smaller EMF cages exhibit identical isomers to empty fullerene cages, permitting comparison of specific non-IPR isomer reaction paths48 for EMFs with different metals encapsulated. Reaction energies for bottom-up, closed network growth18 of Ca@C2n+C2→Ca@C2n+2 (Fig. 4b) show that specific isomers (also shown in Fig. 2d–f) for Ca@C50 and Ca@C60 are the least reactive species to C2 insertion, in agreement with experimental observations. In addition, we computed encapsulation energies (Supplementary Table 2), which provide estimations of the interaction between the metal atom and carbon cage in EMFs, and determine that metal atoms with high oxidation states are favoured to encapsulate smaller cages. Encapsulation energies decrease when fewer electrons are formally donated to the cage and cannot compensate for the higher strain exhibited in small cages.

Bottom-up growth of the medium-sized EMF, Pr@C82 (Fig. 1), exhibits considerably different reactivity towards C2 insertions than do similar-size solvent-extractable empty cages, as known for C84 and C78 (ref. 18). The empty cages appear to grow much more rapidly than negatively charged cages (such as Pr3+@C823−). In essence, C2 insertion becomes more difficult with increasing oxidation state of the encapsulated element and thus EMFs with carbon cages electronically described as C2n3− ‘grow slower’ than empty fullerenes, in agreement with the changes in formation distributions with increasing negative charge on the cage for transformation of smaller M@C2n into medium-sized EMFs (Figs 2 and 3) and theoretical modelling (Fig. 4). Thus, charge transfer from metal to carbon cage and structural considerations can account for the fundamental M@C2n formation trends observed across the periodic table based on a bottom-up mechanism (Fig. 5).

Under synthesis conditions (Figs 2, 3 and 5) that give C60 as the most abundant empty cage for C2n, medium-sized endohedrals, such as M@C82, are present in low relative abundance for mono-metallofullerenes, M@C2n. We interpret such ‘trapping’ of EMFs as non-extractable, smaller cages (for M@C2n, C2n<C60) when sufficient charge is transferred by an encapsulated metal in the proposed bottom-up mechanism to be a major contributing factor to the low general yields of popular, solvent-extractable medium-sized and larger EMFs from evaporated graphite. On the other hand, that fundamental EMF formation property suggests that the abundant formation of smaller EMFs can be avoided by encapsulation of multiple elements or clusters of atoms, such as the M3N unit, to increase yield of solvent-extractable medium-sized or larger EMFs4. The metallofullerene formation ‘road map’ (Fig. 5 and Supplementary Fig. 5) produced in this work should permit additional insight into the many unknown details of bottom-up EMF formation. For example, Ga and In are new elements discovered to incorporate in cages. Based on observed formation distributions, they are predicted to be uncommon Ga+ and In+ when encapsulated in carbon, as corroborated by DFT calculations (Supplementary Fig. 34), and thus oxidation state of an encapsulated element may be identified by formation distributions. Such chemical information may be used to predict possible isomeric structures based on encapsulated metal.

MIII,IV@C44 (MIII,IV=trivalent or tetravalent metals) is found in unusual abundance when encapsulated by elements that donate three or more electrons to the cage (Fig. 5). The prominence of C28, C36, C50, C60 and C70 resulting from evaporated graphite was essential for rationalizing the existence of empty cage fullerenes47. We interpret the emergence of the EMF C44 cage (Figs 2, 3, 4, 5), to represent a previously unrecognized species that may be used to qualitatively understand the disparate behaviour of encapsulated metals in cages. The IPR rule was developed exclusively for pristine fullerenes but did not directly consider the TSFP motif (Fig. 2d)47. The unit of three sequentially fused pentagons is a linear triquinane that contains 11 carbon atoms50. Carbon cages require exactly 12 pentagons for closure and thus four sets of TSFPs bonded to each other would fulfill that fundamental constraint to yield a C44 cage size. The resulting structure of C44, comprising exclusively TSFPs, should exhibit a special stability and would exist as a magic-numbered chemical system if properly stabilized, by analogy to cages D5h-C70, Ih-C60, D5h-C50, D6h-C36 and Td-C28, because neighbouring cage sizes (C2n≤C44) must contain more severely pyramidalized pentagon fusions. M@C44 is the first EMF formed from smaller species in which all TSFP motifs are completely isolated from each other. The structure should be particularly stabilized to nucleophilic attack by C2 and bond rearrangements in a bottom-up formation process when sufficient charge is acquired from an entrapped metal and plausibly explains the observed formation trends in Fig. 5. Thus, the present results strongly suggest that M@C44 exhibits cage isomer D2-C44 (89) (Fig. 2d), which is predicted by DFT calculations48. Therefore, the linear triquinane (TSFP unit) should be found in other metallofullerenes, in addition to the known pentalene (DFP) motif51. Indeed, a particular recently isolated and characterized species of Sc2@C66 displays two TSFP motifs52. The robust stabilization of various pentagon fusions in smaller metallofullerenes in the bottom-up formation scheme is in distinct contrast to empty cage formation and that mechanism can rationalize the structural origin of larger non-IPR EMF cages, as well as aid in prediction of EMF cage isomers for specific endohedral systems.

Curious formation behaviour is observed for actinide-containing EMFs (Fig. 5 and Supplementary Figs 30 and 31). Uranium is known to form the U@Td-C28 species, which is stabilized by transfer of four electrons to the cage35,44. Accordingly, it was predicted that thorium should exhibit similar behaviour, because its primary oxidation state is Th4+. Interestingly, Th@C28 is not observed; however, Th@C36 seems to have been essentially substituted (Fig. 5) in the bottom-up formation path. This observation is likely due to the larger ionic radius of Th4+ relative to U4+. The smallest cage that the Th4+ ion can effectively nucleate must be greater than 28 carbon atoms, and in this case is Th@C36. The distribution of Th@C2n mimics U@C2n formation beyond Th@C36, which can now be explained by a similar extent of charge transfer during the bottom-up C2 insertion process. M@C28 is not observed for any rare earth encapsulated EMFs; however, Group 4 elements have been observed to encapsulate the C28 cage35, in addition to U. Indeed, comparison of M@C2n with different groups is accurate in this respect, because, for example, very similar ionic radii exist for encapsulated elements such as Na+ (1.02 Å), Ca2+ (1.00 Å) and Ce3+ (1.01 Å). Numerous other instances (Fig. 5) of M@C2n with similar ionic radii but different oxidation states additionally support charge transfer as a determinant factor in bottom-up EMF formation. As shown in Fig. 5, the smallest detectable cage(s) formed in abundance under the present conditions is M@C50 for Group 1 elements (K, Rb, Cs), M@C42 or M@C44 for Group 2 elements (Ca, Sr, Ba), M@C36 for trivalent rare earth metals (Sc, Y, La, Ce, Pr, Nd, Gd, Tb, Dy Ho, Er, Lu), although Sc@C34 can be observed in very low abundance, M@C44 or M@C42 for divalent lanthanides (Sm, Eu, Yb, except Tm), and M@C50 for Ga and In. These results also suggest that ionic radii may be a possible factor that alters EMF formation.

In conclusion, we experimentally report that small, medium, and giant endohedral mono-metallofullerenes are formed in the gas phase from metal-doped graphite through a bottom-up mechanism under EMF synthesis conditions. The acquisition of negative charge from the encapsulated metal to cage, the hallmark of EMFs which give them spectacular properties, occurs at the inception of metallofullerene formation. Ionization energies of the elements appear to determine oxidation state after graphite/metal evaporation and thus extent of charge transferred to carbon cages, Mn+@C2nn−. Although strain energy is a trend-setting factor, particular smaller, unconventional EMFs become significantly stabilized by charge transfer, thereby impeding C2 insertion reactions and thus inhibiting overall bottom-up formation of the small EMF compounds into solvent-extractable, larger metallofullerenes. The chemical manifestation also provides a natural explanation as to how non-IPR cage isomers may be formed for larger metallofullerenes. The understanding of EMF formation may bring into reality the ability to increase yield and drastically improve synthesis of particular molecular targets. There are still many unsolved aspects of metallofullerene formation that will require additional experimental evidence to elucidate, and further studies are underway in our laboratories. We note it is possible, and perhaps in some cases probable, that a C2 loss event can occur in a bottom-up reaction path and thus should be necessary for a comprehensive explanation of fullerene formation.

Methods

Molecular behaviour of M@C82 in evaporated graphite

M@C82 (M=Pr and so on) was synthesized by the arc discharge technique. Pr@C82 was isolated and purified by HPLC to yield the isomers Pr@C82 (I) and Pr@C82 (II). An isomerically purified sample of Pr@C82 was then uniformly applied to the surface a pure graphite (99.9999%, 2–15 μm) target rod for carbon plasma exposure studies by use of an Nd:YAG laser (532 nm, 15 mJ per pulse) cluster source.

Preparation and vapourization of metal-doped graphite

Metal-incorporated graphite material (1–2 atomic % metal) is produced by combination of graphite (99.9999%, 2–15 μm) and, in most cases, metal oxide (≥99.9%). Metallofullerenes, M@C2n, are formed by evaporation of that metal-doped graphite material, after being moulded into a rod, by the use of an Nd:YAG laser (532 nm, 5 mj per pulse) in a pulsed cluster source.

Cluster source and 9.4 T FT-ICR mass spectrometry

Nanocarbon mechanism experiments are analysed with a custom-built FT-ICR mass spectrometer based on a 9.4-T superconducting magnet and performed with positive ions produced by a pulsed laser cluster source18,33,35. Evaporation of a translating and rotating target rod (12.7 mm diameter) is achieved by a single laser shot fired from an Nd:YAG (532 nm, 3–5 ns pulse width, ~1.5 mm beam diameter, 5–15 mj per pulse) in conjunction with the opening of a pulsed valve (800 μs duration) to admit He flow over the sample. Carbon vapour produced then enters a channel 4 mm in diameter and ~8.5 mm in length18. The laser is fired ~2 ms after opening of the pulsed valve for evaporation of metal-doped graphite samples and ~5 ms for M@C82-coated graphite samples. Ions accumulated by ten individual laser and helium pulse events are transported to an open cylindrical ion trap (70 mm diameter, 212 mm long, aspect ratio ~2). The ions are accelerated to a detectable cyclotron radius by a broadband frequency sweep excitation (260 Vp-p, 150 Hz μ−1, 3.6 down to 0.071 MHz) and subsequently detected as the differential current induced between two opposed electrodes of the ICR cell. Each of the acquisitions is Hanning-apodized and zero-filled once before fast Fourier transform and magnitude calculation53. Ten time-domain acquisitions are averaged. The experimental event sequence is controlled by a modular ICR data acquisition system54. Ions are further probed by collision-induced dissociation.

Computational details

Amsterdam Density Functional code (ADF2011) was used55,56. The electronic density was provided by the local density approximation by use of Becke’s gradient-corrected exchange functional and Vosko–Wilk–Nusair parameterization for correlation, corrected with Perdew’s functional (BP86). Electrons for carbon and metals were described with Slater-type basis functions of triple-ζ polarization quality. We have included scalar relativistic corrections by means of the zeroth-order regular approximation formalism.

Additional information

How to cite this article: Dunk, P. W. et al. Bottom-up formation of endohedral mono-metallofullerenes is directed by charge transfer. Nat. Commun. 5:5844 doi: 10.1038/ncomms6844 (2014).