Ultrasonication-switched formation of dice- and cubic-shaped fullerene crystals and their applications as catalyst supports for methanol oxidation

Abstract

Three-dimensional (3D) nanostructures of an endohedral fullerene, Sc₃N@C₈₀ (I_h), including both cubic-shaped and unprecedented dice-shaped crystals, have been successfully prepared for the first time via a modified liquid–liquid interfacial precipitation (LLIP) method integrating ultrasonication. By simply switching ultrasonication on/off upon mixing the good/poor (mesitylene/isopropanol) solvents, the dice- and cubic-shaped Sc₃N@C₈₀ crystals have been selectively prepared, and their size can be readily controlled by varying the concentration of the starting Sc₃N@C₈₀ solution in mesitylene. The growth mechanism of the cubic- and dice-shaped Sc₃N@C₈₀ crystals has been proposed, highlighting the formation of local mesitylene cavities which lead to the nucleation and growth of Sc₃N@C₈₀ molecules, while the formation of the dice structure. i.e., a hole in each facet of the cube, was primarily due to the hindered diffusion of Sc₃N@C₈₀ solutes by ultrasonication-generated gas microbubbles. The crystal structure of the Sc₃N@C₈₀ cubes has been determined as a simple cubic unit cell with a lattice constant of a = 10.8 Å. Finally the dice-shaped Sc₃N@C₈₀ crystals were applied as a Pt catalyst support for the methanol oxidation reaction (MOR), revealing the improved MOR activity compared to the cubic-shaped Sc₃N@C₈₀ crystals due to its larger surface area resulting from the dice (hole) structure.

---

1. Introduction

Since the first success in the macroscopic production of fullerenes, along with the facile formation of different forms of C₆₀ crystals,¹ the study of fullerene crystals with controllable morphology, size and dimensions, which show potential applications in catalysis, field effect transistors and photoelectric devices etc., has been attracting considerable attention.^2–5 So far, vapor-driven and solution-driven methods are the most common for the preparation of fullerene crystals.^6–11 In particular, as a facile method affording predetermined morphology and improved uniformity of the crystals, the solution-based liquid–liquid interfacial precipitation (LLIP) method has been widely used for the growth of different types of single-crystalline C₆₀ with tunable morphologies ranging from one-dimensional (1D) rods or tubes to 2D plates.^12–16 However, the preparation of 3D fullerene nanostructures, specifically cubic-shaped C₆₀ crystals, has been rarely reported. Instead, single-crystalline C₇₀ cubes have been successfully prepared in two reports. In 2010, Choi et al. used the LLIP method and succeeded in fabricating the first cubic-shaped C₇₀ crystals, which exhibited enhanced photoluminescence over that of C₇₀ powders.¹⁷ Recently Wågberg et al. also accomplished the preparation of C₇₀ microcubes by using a fast solvent-assisted evaporation method with aromatic solvents and halogen radicals as a controller.¹⁸ Thus, the formation of 3D fullerene crystals, particularly cubes, seems to be dependent on the nature of the fullerene, and an intriguing question to be addressed is whether cubic- or other shaped 3D crystals could be formed for other types of fullerenes, including endohedral fullerenes which represent a special type of fullerenes with atoms, ions or clusters entrapped in carbon cages.^19–23

During the process of crystal nucleation and growth, any negligible perturbation may result in a considerable change to the final crystal in terms of morphology and size. In particular, ultrasonication has recently been widely implemented in the crystal growth of different molecules including fullerenes such as C₆₀ because it can readily induce gas microbubbles which function as physical matrices for nanocrystal nucleation and growth, and consequently affect the supersaturation level required for nucleation and the final particle size distribution.^4,24–27 However, to our knowledge, to date, ultrasonication perturbation has not been applied in the crystal growth of endohedral fullerenes.^28–32

Herein, we report the first preparation of 3D nanostructures of an endohedral fullerene, Sc₃N@C₈₀ (I_h), via a modified LLIP method integrating ultrasonication, affording both cubic-shaped and unprecedented dice-shaped crystals, which can be tuned by simply switching ultrasonication on/off upon mixing the binary solvents. The growth mechanism of the morphology-tunable Sc₃N@C₈₀ crystals was proposed on the basis of a systematic study on the effects of ultrasonication, good/poor solvents and their mixing ratio on the morphology of the Sc₃N@C₈₀ crystals. Finally the dice- and cubic-shaped Sc₃N@C₈₀ crystals were applied as Pt catalyst supports for the methanol oxidation reaction (MOR), and the correlation between the morphology and the MOR activity of the Sc₃N@C₈₀ crystals was unveiled.

2. Experimental

2.1. Materials

Sc₃N@C₈₀ (I_h) was synthesized by a modified Krätschmer–Huffman DC-arc discharging method and isolated with a high purity (>99%), as reported previously.³³ Other reagents, including the solvents (mesitylene and isopropanol), were commercially available, and used without further purification.

2.2. Preparation

The preparation procedure of the dice-shaped Sc₃N@C₈₀ crystals using a modified LLIP method is shown in Scheme 1 (route (I)). The detailed procedure was as follows: 3 mL isopropanol was quickly added into 1 mL Sc₃N@C₈₀ solution in mesitylene (0.1 mg mL⁻¹) with ultrasonication (KQ5200B ultrasonicator, 200 W, 40 KHz, Kunshan Ultrasonic Instruments Co., Ltd, China) switched on in a 5 mL glass vessel, and then ultrasonication was performed for 5 min. A large amount of microbubbles emerged and Sc₃N@C₈₀ molecules nucleated and crystallized during the ultrasonication process. After ultrasonication, the mixture solution became light grey. Finally the mixture was kept at room temperature for 12 h for crystal growth.

Scheme 1 A schematic illustration of the procedure to prepare the dice- and cubic-shaped Sc₃N@C₈₀ crystals via a modified LLIP method integrating ultrasonication.

The preparation procedure of cubic-shaped Sc₃N@C₈₀ crystals using the traditional LLIP method was as follows: 3 mL isopropanol was slowly added into 1 mL Sc₃N@C₈₀ solution in mesitylene (0.1 mg mL⁻¹) in a 5 mL glass vessel by keeping a placid interface between mesitylene and isopropanol (see route (II) of Scheme 1). The sample vessel was then shaken by hand for 1 min, resulting in a homogeneous yellowish mixture. The mixture solution was then ultrasonicated for 5 min and many small microbubbles formed, in which Sc₃N@C₈₀ molecules nucleated and crystallized. After ultrasonication, the sample solution became light grey. Finally the mixture was kept at room temperature for 12 h for crystal growth.

2.3. Preparation procedure of Pt/Sc₃N@C₈₀ crystals/ITO electrodes

Sc₃N@C₈₀ crystals were dispersed in acetone and then transferred to a small cell in which two ITO electrodes were kept at a distance of 5 mm by a Teflon spacer. A DC voltage of 120 V was applied between the two electrodes using a DC power supply. The electrode surface connected to the positive terminal became dark grey after the deposition of Sc₃N@C₈₀ crystals, and then the film was dried in air. Two different types of modified ITO electrodes, namely Sc₃N@C₈₀ dice/ITO, Sc₃N@C₈₀ cubes/ITO as well as blank ITO electrodes, were exposed to an aqueous solution of K₂PtCl₆ (1 mM) and LiClO₄ (1 M) for reduction, and consequently platinum (Pt) nanoparticles were formed and deposited onto the Sc₃N@C₈₀ crystal film by applying a potential at 315 mV vs. the Ag/AgCl electrode. The amount of Pt deposited as determined from coulometry was ∼16 μg.

The electrochemical active surface area measurement was carried out in 0.5 M H₂SO₄ solution. The scan rate was 100 mV s⁻¹. Sc₃N@C₈₀ crystals/ITO electrodes were used as the working electrode. A platinum plate and an Ag/AgCl electrode were used as the counter and reference electrode, respectively. The electrochemical active surface area of the catalyst was calculated by measuring the charge for hydrogen desorption, Q_H, according to formula (1):³⁴

where v represents the scanning speed, I_d represents the current modifying value of the double layer, and (a, b) is the voltage range of hydrogen desorption. Then Q_H was used to calculate the active platinum area of the electrodes according to the following formula:³⁴where M_Pt represents the platinum loading in the electrode and Q_Href represents the charge required to oxidize a monolayer of H₂ on flat Pt (210 μC cm⁻²).³⁴

2.4. Measurements

SEM imaging was carried out on a JSM-7401F instrument (JEOL, Japan). High resolution transmission electron microscopy (HR-TEM) and selected area electron diffraction (SAED) measurements were made by applying a drop of the sample to a copper grid using a JEOL-2010 TEM instrument (JEOL, Japan). X-ray diffraction (XRD) data were collected with a TTR-III instrument (RIGAKU, Japan) using graphite monochromator filtered Cu Kα radiation. The samples for XRD analysis were prepared by dropping samples onto a silicon substrate in air. Thermal gravimetric analysis (TGA) was carried out under a 100 sccm N₂ gas flow on a DTG-60H instrument (Shimadzu, Japan). Methanol oxidation reaction (MOR) measurements were performed in 2.0 M methanol and 1.0 M H₂SO₄ solution on a 660A electrochemical analyzer (CH instruments Inc., USA) at room temperature. Pt/Sc₃N@C₈₀ crystals/ITO electrodes were used as the working electrode. A platinum plate and a Ag/AgCl electrode were used as the counter and reference electrodes, respectively. The scan rate was 20 mV s⁻¹. Chronoamperometric measurements were also performed on the 660A electrochemical analyzer at 0.81 V.

3. Results and discussion

3.1. Preparation of dice- and cubic-shaped Sc₃N@C₈₀ (I_h) crystals

Sc₃N@C₈₀ (I_h) crystals were prepared by the LLIP method as shown in Scheme 1 with mesitylene and isopropanol used as the “good” solvent for dissolving Sc₃N@C₈₀ (I_h) and the “poor” solvent for precipitation, respectively. High-purity (>99%) Sc₃N@C₈₀ (I_h, hereafter the label of isomer I_h is omitted for clarity), which was synthesized by a modified Krätschmer–Huffman DC-arc discharging method,³³ was dissolved in mesitylene with variable concentrations (0.05, 0.1 mg mL⁻¹), and then 3 mL isopropanol was slowly added into 1 mL Sc₃N@C₈₀ solution in mesitylene by two different means. In the first set of the study, the addition of isopropanol into Sc₃N@C₈₀ solution in mesitylene was made by keeping a placid interface between these two solvents (see route (II) of Scheme 1). After the complete mixing of isopropanol and mesitylene via shaking by hand followed by ultrasonication treatment, the Sc₃N@C₈₀ molecules nucleated and crystallized, and the SEM images of the prepared Sc₃N@C₈₀ crystals are shown in Fig. 1A, with the initial concentration of Sc₃N@C₈₀ solution in mesitylene set as 0.1 mg mL⁻¹. Clearly uniform cubic-shaped crystals with an average size of ∼0.8 μm were obtained (see Fig. S1A, ESI†), indicating the successful preparation of 3D microcubes of Sc₃N@C₈₀.

Fig. 1 SEM images of cubic- (A) and dice-shaped (B) Sc₃N@C₈₀ crystals prepared by the LLIP method with or without ultrasonication. The concentration of the Sc₃N@C₈₀ solution in mesitylene was 0.1 mg mL⁻¹.

Interestingly, when isopropanol was added rapidly into the Sc₃N@C₈₀ solution in mesitylene (0.1 mg mL⁻¹) and ultrasonication was switched on simultaneously (see route (I) of Scheme 1), while the following procedure was exactly the same as that used for the growth of cubic-shaped crystals, the resultant crystals change to uniform dice shapes according to the SEM image (Fig. 1B), and the average size of the dice is 0.9–1.0 μm (see Fig. S1B, ESI†). Since the dramatic changes of both the morphology (dice vs. cube) and size result from simply switching ultrasonication on/off upon mixing the binary solvents (isopropanol and mesitylene), this result indicates the strong influence of perturbation induced by ultrasonication on the crystal nucleation and growth of the Sc₃N@C₈₀ crystals.

For both cases of cubic- and dice-shaped Sc₃N@C₈₀ crystals, the size of the final crystals is found to be sensitively dependent on the initial concentration of Sc₃N@C₈₀ solution in mesitylene, while when the initial concentration of Sc₃N@C₈₀ solution in mesitylene decreases to 0.05 mg mL⁻¹, the average size of the Sc₃N@C₈₀ cubes (ultrasonication off) and dice (ultrasonication on) increases to ∼2.0 and 2.1 μm, respectively (see Fig. S2, ESI†). This phenomenon is in accordance with other fullerene nanostructures reported in the literature,^8–16 and can be interpreted by the formation of more nucleation sites leading to more crystal seeds and a larger concentration depletion of Sc₃N@C₈₀ molecules under a higher concentration of Sc₃N@C₈₀ solution. Therefore, the size of the dice- or cubic-shaped Sc₃N@C₈₀ crystals can be readily controlled by simply varying the concentration of the starting Sc₃N@C₈₀ solution in mesitylene.

3.2. Effects of ultrasonication, good/poor solvents and their mixing ratio on the morphology of Sc₃N@C₈₀ crystals

To unveil the role of ultrasonication applied during mixing the good/poor solvents (mesitylene/isopropanol) on the transformation of the cubic-shaped Sc₃N@C₈₀ crystals into dice-shaped crystals, we performed a comparison experiment by applying vigorous vibration, which may also induce perturbation on the mixing process. When a vigorous vibration from a mechanical shaker was applied during the rapid addition of isopropanol into the Sc₃N@C₈₀ solution in mesitylene, we found that mesitylene and isopropanol mixed only locally but not thoroughly in the vessel despite the vessel being shaken vigorously; as a result, non-uniform cubic-shaped Sc₃N@C₈₀ crystals with a much larger size distribution were obtained (see Fig. S3, ESI†). This is understandable because the Sc₃N@C₈₀ solutes distribute inhomogeneously around the local mixing of mesitylene and isopropanol. It is notable that, despite the perturbation by vigorous vibration, only Sc₃N@C₈₀ cubes were observed and no dice formed under the vigorous vibration conditions. This suggests that a perturbation induced by ultrasonication is necessary and more effective for dice formation because of the more homogeneous mixing of the binary solvents, and more importantly, the generation of gas microbubbles, which provide physical matrices for crystal nucleation and growth.^17,27

Given that the role of ultrasonication on the formation of dice-shaped Sc₃N@C₈₀ crystals was confirmed, we next carried out a series of control experiments by varying the combination of good/poor solvents and their volume ratio in order to investigate the roles of good/poor solvents (mesitylene/isopropanol) on the formation of dice-shaped Sc₃N@C₈₀ crystals under ultrasonication conditions. First we selectively used analogous alkyl-substituted benzenes such as p-xylene and m-xylene as an alternative good solvent to dissolve Sc₃N@C₈₀. With the identical preparation procedure to that using mesitylene and ultrasonication, 1D micron-sized rods and micron-sized needle-like rods were obtained for p-xylene and m-xylene respectively, as we reported very recently (see Fig. S4, ESI†).³² Previously, using toluene or CS₂ as the good solvent to dissolve Sc₃N@C₈₀, Wakahara et al. obtained either 1D Sc₃N@C₈₀ nanowhiskers when using toluene, or 1D nanowhiskers/2D tetragonal nanosheets by using CS₂ (ultrasonication was not applied in both cases).³⁰ Because the same solvent (isopropanol) was used as the poor solvent for interfacial precipitation in all cases, including our present study, the large diversity of the final Sc₃N@C₈₀ crystals must result from the nature of the good solvent. The dependence of the solvent nature (molecular geometry, solubility etc.) on the morphology of the nanostructures of fullerenes, especially C₆₀, has been extensively investigated in the literature.¹¹ It is commonly believed that the good solvent generally provides local cavities for crystal growth, guiding the self-crystallization of fullerene molecules into different morphologies.¹¹ In this sense, the solvent geometry and solubility of fullerenes are two major factors determining the final morphology of fullerene crystals. In our present study, mesitylene has a higher molecular symmetry than p-xylene and m-xylene and Sc₃N@C₈₀ has a lower solubility in mesitylene, and presumably this is beneficial for the formation of homogeneous local mesitylene cavities, leading to the cocrystallization of Sc₃N@C₈₀ molecules with mesitylene. Thus homogeneous crystal growth in all axial directions is fulfilled and consequently 3D cubic-/dice-shaped Sc₃N@C₈₀ crystals form.

The good solvent determines the geometry of the Sc₃N@C₈₀ crystals, but does not seem to account for the formation of holes in each facet (i.e., dice-shape) of the Sc₃N@C₈₀ cubes, for which ultrasonication plays the determining role as discussed above. The effect of the poor solvent on the morphology of the Sc₃N@C₈₀ crystals was then studied further by substituting isopropanol with n-hexane and acetone. When n-hexane with a lower polarity than isopropanol was used as the poor solvent for precipitation, random Sc₃N@C₈₀ crystals, including both irregular polyhedra and pseudo-cubes with obscure edges and quite small and shallow holes in each facet, were obtained. On the other hand, when acetone with higher polarity replaced isopropanol, inhomogeneous dice-shaped Sc₃N@C₈₀ crystals with much larger holes in each facet were observed (see Fig. S5, ESI†). This phenomenon is probably because, similar to the case of C₆₀,³⁵ the relatively lower solubility of Sc₃N@C₈₀ in acetone than in n-hexane led to hindered Sc₃N@C₈₀ solute diffusion when using acetone, and thus crystals grew along the preferred direction. These results reveal that the polarity of the poor solvent is responsible for the size of the holes in each facet of the Sc₃N@C₈₀ dice.

Since mesitylene and isopropanol was determined to be the most favorable good/poor solvent combination for the formation of dice-shaped Sc₃N@C₈₀ crystals, the influence of their mixing ratio was studied next to gain deeper insight into the growth mechanism of the Sc₃N@C₈₀ crystals. Under the conditions of mesitylene : isopropanol = 1 : 3 (v/v), perfect Sc₃N@C₈₀ dice were obtained (see Fig. 1B). When the mixing ratio of mesitylene to isopropanol increased to 1 : 1 (v/v), the relative concentration of Sc₃N@C₈₀ in the mesitylene and isopropanol mixed solvents increased, resulting in the formation of more crystal nuclei and a sufficient solute supply, and consequently smaller cubic-shaped crystals instead of dice were obtained. On the other hand, the decrease of the mixing ratio of mesitylene to isopropanol to 1 : 5 (v/v) led to the coexistence of both dice and cubes with sharp edges and comparable size (see Fig. S6, ESI†). Since in this case the relative concentration of Sc₃N@C₈₀ in the mesitylene and isopropanol mixed solvents decreases, the size of the dice-shaped Sc₃N@C₈₀ crystals is obviously much larger compared to that obtained under the optimum conditions discussed above (mesitylene : isopropanol = 1 : 3 (v/v)). Interestingly, for some Sc₃N@C₈₀ dice, it was found that the holes were filled or attached with many small particles or cubes, and this can be regarded as an intermediate structure between the perfect dice and cube.

3.3. Growth mechanism and crystal structures of cubic- and dice-shaped Sc₃N@C₈₀ crystals

Based on the above systematic studies, a plausible growth mechanism of the cubic- and dice-shaped Sc₃N@C₈₀ crystals is proposed in Scheme 2. For Sc₃N@C₈₀ cubes, in the first step, local mesitylene cavities form immediately upon the addition of isopropanol, in which Sc₃N@C₈₀ molecules nucleate. Then crystal seeds start to grow by the deposition of Sc₃N@C₈₀ molecules onto the crystal nucleus via π–π stacking with the guidance of mesitylene solvent molecules, leading to the formation of Sc₃N@C₈₀ cubes. A similar mechanism has been reported for C₇₀ cubes by Choi et al.¹⁷ When ultrasonication was switched on, upon mixing mesitylene and isopropanol, gas microbubbles are instantaneously generated, which on one hand promote the homogeneous formation of the local mesitylene cavities for Sc₃N@C₈₀ nucleation, but on the other hand hinder the diffusion of Sc₃N@C₈₀ solutes for crystal growth.^4,24–27 Due to the slower diffusion of Sc₃N@C₈₀ solutes from mesitylene to isopropanol, Sc₃N@C₈₀ molecules would stack along the preferable 〈100〉 and 〈111〉 directions, followed by the principle of crystal growth under a shortage of the solute, leading to hole formation in each facet of the cube, i.e., a dice-shaped structure. It should be noted that, for the preparation of C₇₀ cubes by Choi et al., ultrasonication along with rapidly mixing mesitylene and isopropanol was also applied, and the implementation of ultrasonication was found to be advantageous for improving the size homogeneity of the C₇₀ cubes, but, similar to the case without applying ultrasonication, only cubes were obtained without dice-shaped crystals.¹⁷ The dramatic difference between Sc₃N@C₈₀ and C₇₀ on the crystallization behavior under ultrasonication conditions suggests that the crystal growth of fullerenes with perturbation is sensitively dependent on the nature of the fullerenes, such as the cage size and charge transfer which takes place within endohedral fullerenes,^19–23 and such a statement is to be further verified by other (endohedral) fullerenes.

Scheme 2 A schematic illustration of the growth process of cubic- and dice-shaped Sc₃N@C₈₀ crystals. Small cubes form after nucleation and then grow by preferential deposition.

Given that C₆₀ and C₇₀ crystals prepared by the LLIP method always contain solvent molecules in the crystal lattices,^12–17,36 an open question arises: whether the dice-shaped Sc₃N@C₈₀ crystals contain the solvent molecules in the crystal lattices or not. Thermal gravimetric analysis (TGA) was then carried out under a 100 sccm N₂ gas flow to investigate the decomposition process of the dice-shaped Sc₃N@C₈₀ crystals upon heating to over 800 °C, and two weight loss processes were clearly observed. In the first weight loss step (160–340 °C), a ∼15% weight loss was observed, which can be assigned to the removal of mesitylene solvent molecules. Since only one weight loss process was found in this temperature region, the coexistence of isopropanol solvent molecules can be ruled out. In the second weight loss process (>800 °C), the weight loss is attributed to the decomposition of dice-shaped Sc₃N@C₈₀ crystals (see Fig. S7, ESI†). The TGA results reveal that the dice-shaped Sc₃N@C₈₀ crystals contain mesitylene solvent molecules with the molar ratio of mesitylene to Sc₃N@C₈₀ being ∼1.6 : 1. Besides, the existence of mesitylene solvent molecules in both the dice- and cubic-shaped Sc₃N@C₈₀ crystals is further confirmed by FTIR measurements (see Fig. S8, ESI†).

The crystal structures of the cubic- and dice-shaped Sc₃N@C₈₀ crystals were examined by powder X-ray diffraction (XRD, Fig. 2) in combination with the electron diffraction pattern (Fig. 3). In the XRD pattern of the Sc₃N@C₈₀ cubes (Fig. 2(I)), two intense diffraction peaks with d-spacings of 10.8 Å and 5.41 Å respectively can be indexed as the (100) and (200) planes, suggesting that the Sc₃N@C₈₀ cubes lie on the Si substrate through {100}-series planes. There are several other weak diffraction peaks with d-spacings of 6.24, 4.85, 4.43, 3.62, 3.42 and 3.27 Å corresponding to the (111), (210), (211), (300), (310) and (311) planes, respectively (see S9, ESI†). Such a diffraction pattern perfectly matches to the simple cubic structure (a = 10.8 Å). A high-resolution transmission electron microscopy (HR-TEM) image of a Sc₃N@C₈₀ cube (inset A of Fig. 3) shows a clear lattice image with the distance between two adjacent planes of 10.8 Å, which could be assigned to the d-spacing of the (100) crystalline plane as is clearly observed in the corresponding selected-area electron diffraction (SAED) pattern (inset B of Fig. 3). These data coincide with that obtained from the XRD characterization, confirming the simple cubic structure with a = 10.8 Å for the Sc₃N@C₈₀ cubes.

Fig. 2 XRD patterns of cubic- (I) and dice-shaped (II) Sc₃N@C₈₀ crystals. Inset: magnified XRD pattern in the region of 2θ = 10–30°.

Fig. 3 High-resolution TEM (HR-TEM) image of a cubic-shaped Sc₃N@C₈₀ crystal. Insets: (A) TEM image of a single Sc₃N@C₈₀ cube. The red ring marks the site of electron diffraction; (B) selected area electron diffraction (SAED) pattern.

For the dice-shaped Sc₃N@C₈₀ crystals, only one intense diffraction peak with a d-spacing of 10.8 Å was observed in the XRD pattern (Fig. 2(II)), which can be indexed as the (100) plane as well. Similar to the case of the Sc₃N@C₈₀ cubes, the dominating appearance of only (100) plane diffraction suggests that the dice-shaped Sc₃N@C₈₀ crystals contact with Si substrate through {100}-series planes as well, while the (200) plane diffraction was not detected. This might be ascribed to the weak crystallinity of the dice-shaped Sc₃N@C₈₀ crystals. In fact, the sample with weak crystallinity suffers amorphization after high-energy electron beam bombarding, thus an attempt to measure the HR-TEM image and SAED pattern of the dice-shaped Sc₃N@C₈₀ crystals failed.

3.4. Application of cubic- and dice-shaped Sc₃N@C₈₀ crystals as Pt catalyst supports for the methanol oxidation reaction (MOR)

To our knowledge, the catalytic properties of endohedral fullerenes have never been studied. The unprecedented dice shape of the Sc₃N@C₈₀ crystals with the hole structure stimulated us to investigate their application as a Pt catalyst support for the methanol oxidation reaction (MOR),^2–4,37,38 and its catalytic activity is compared with that of the cubic-shaped Sc₃N@C₈₀ crystals so as to understand the structure–property relationship of the Sc₃N@C₈₀ crystals. Sc₃N@C₈₀ dice were first deposited onto an indium tin oxide (ITO) film electrode by an electrophoretic deposition method,³² and then ∼16 μg platinum (Pt) was loaded onto the Sc₃N@C₈₀ crystal surface (see Experimental details).^3,4Fig. 4(I) shows the cyclic voltammogram (CV) of the Pt/Sc₃N@C₈₀ dice/ITO electrode in 0.5 M H₂SO₄ solution, which includes also those of the Pt/Sc₃N@C₈₀ cubes/ITO and Pt/ITO electrodes for comparison. All three CV curves show obvious redox peaks associated with hydrogen adsorption/desorption below 0.15 V (vs. Ag/AgCl). The electrochemical active surface area was calculated by measuring the charge (Q_H) associated with the adsorbed hydrogen monolayer in the potential region of −0.2 to 0.15 V, which is 5.93, 3.56 and 2.43 m² g⁻¹ for the Pt/Sc₃N@C₈₀ dice/ITO, Pt/Sc₃N@C₈₀ cubes/ITO and Pt/ITO electrodes, respectively (see Experimental details). It is noteworthy that the electrochemical active surface area of the Pt/Sc₃N@C₈₀ dice/ITO electrode is ca. 1.7 and 2.4 times larger than those of the Pt/Sc₃N@C₈₀ cubes/ITO and Pt/ITO electrodes, respectively, suggesting that the dispersion of the Pt catalyst nanoparticles is largely improved upon applying Sc₃N@C₈₀ crystals as catalyst supports.

Fig. 4 (I) Cyclic voltammograms of the Pt/Sc₃N@C₈₀ dice/ITO, Pt/Sc₃N@C₈₀ cubes/ITO and Pt/ITO electrodes recorded in 0.5 M H₂SO₄ with a scan rate of 100 mV s⁻¹; (II) normalized cyclic voltammograms of the Pt/Sc₃N@C₈₀ dice/ITO, Pt/Sc₃N@C₈₀ cubes/ITO and Pt/ITO electrodes in 2.0 M methanol + 1.0 M H₂SO₄ solutions with a scan rate of 20 mV s⁻¹. The curves are normalized to the electrochemical active surface area and Pt loading (∼16 μg).

The MOR curves of the three modified ITO electrodes normalized to the electrochemical active surface area and Pt loading, which were measured in 2.0 M methanol and 1.0 M H₂SO₄ solution at room temperature with a scan rate of 20 mV s⁻¹, are compared in Fig. 4(II). The Pt/Sc₃N@C₈₀ dice/ITO and Pt/Sc₃N@C₈₀ cubes/ITO electrodes show onset potentials of 0.15 V and 0.17 V, respectively, which exhibit a negative shift of about 0.09 and 0.07 V over that of the Pt/ITO electrode (0.24 V). The oxidation peak currents observed for the Pt/Sc₃N@C₈₀ dice/ITO and Pt/Sc₃N@C₈₀ cubes/ITO electrodes are both higher than that for the Pt/ITO electrode (see Fig. 4(II)). Besides, chronoamperometric measurements performed at 0.81 V confirmed the highest methanol oxidation current for the Pt/Sc₃N@C₈₀ dice/ITO electrode, and that the catalytic activity follows the order of Pt/Sc₃N@C₈₀ dice/ITO > Pt/Sc₃N@C₈₀ cubes/ITO > Pt/ITO (see Fig. S9, ESI†). A higher electrocatalytic stability of the Pt/Sc₃N@C₈₀ dice/ITO electrode over that of the Pt/Sc₃N@C₈₀ cubes/ITO electrode is observed despite the fast current decay due to the poisoning of the catalyst by the intermediate species formed during the methanol oxidation reaction.³⁸ The improved MOR activity of Pt/Sc₃N@C₈₀ dice/ITO compared to Pt/Sc₃N@C₈₀ cubes/ITO is presumably due to the larger surface area of the Sc₃N@C₈₀ dice resulting from the unique dice (hole) structure.⁴

4. Conclusions

In summary, for the first time, 3D nanostructures of Sc₃N@C₈₀, including both cubic-shaped and unprecedented dice-shaped crystals, have been successfully prepared via a modified LLIP method integrating ultrasonication. By simply switching ultrasonication on/off upon mixing the good/poor solvents (mesitylene/isopropanol), the dice- and cubic-shaped Sc₃N@C₈₀ crystals were selectively prepared, and their size can be readily controlled by varying the concentration of the starting Sc₃N@C₈₀ solution in mesitylene. Based on a systematic study on the effects of ultrasonication, good/poor solvents and their mixing ratio on the morphology of the Sc₃N@C₈₀ crystals, a plausible growth mechanism of the cubic- and dice-shaped Sc₃N@C₈₀ crystals was proposed, highlighting the formation of local mesitylene cavities which lead to the nucleation and growth of Sc₃N@C₈₀ molecules, while the formation of the dice structure. i.e., a hole in each facet of the cube, was primarily due to the hindered diffusion of Sc₃N@C₈₀ solutes by ultrasonication-generated gas microbubbles. The crystal structure of the Sc₃N@C₈₀ cubes was determined to feature a simple cubic unit cell with a lattice constant of a = 10.8 Å. Finally, the dice-shaped Sc₃N@C₈₀ crystals were applied as a Pt catalyst support for the MOR, revealing the improved MOR activity compared to cubic-shaped Sc₃N@C₈₀ crystals due to their larger surface area resulting from the dice (hole) structure. As the first success of the facile preparation of 3D nanostructures of endohedral fullerenes, our study provides new insight into the crystallization behavior of fullerenes as well as their potential application in direct methanol fuel cells.

Acknowledgements

We are grateful to valuable discussions with Professor Suhua Wang (Institute of Intelligent Machines, Chinese Academy of Sciences). This work was partially supported by the National Basic Research Program of China (2010CB923300, 2011CB921400) and the National Natural Science Foundation of China (no. 21132007, 21371164).

Notes and references

1. W. Krätschmer, L. D. Lamb, K. Fostiropoulos and D. R. Huffman, Nature, 1990, 347, 354.
2. B. S. Sherigara, W. Kutner and F. D'Souza, Electroanalysis, 2003, 15, 753.
3. K. Vinodgopal, M. Haria, D. Meisel and P. Kamat, Nano Lett., 2004, 4, 415.
4. Y. Zhang, L. Jiang, H. Li, L. Z. Fan, W. P. Hu, C. R. Wang, Y. F. Li and S. H. Yang, Chem. - Eur. J., 2011, 17, 4921.
5. Y. Zhang, W. Liu, L. Jiang, L. Z. Fan, C. Wang, W. P. Hu, H. Zhong, Y. F. Li and S. H. Yang, J. Mater. Chem., 2010, 20, 953.
6. L. Wang, B. B. Liu, D. Liu, M. G. Yao, Y. Y. Hou, S. D. Yu, T. Cui, D. M. Li, G. T. Zou, A. Iwasiewicz and B. Sundqvist, Adv. Mater., 2006, 18, 1883.
7. X. D. Shi, A. R. Kortan, J. M. Williams, A. M. Kini, B. M. Savall and P. M. Chaikin, Phys. Rev. Lett., 1992, 68, 827.
8. J. F. Geng, I. A. Solov'yov, W. Z. Zhou, A. V. Solov'yov and B. F. G. Johnson, J. Phys. Chem. C, 2009, 113, 6390.
9. M. G. Yao, B. M. Andersson, P. Stenmark, B. Sundqvist, B. B. Liu and T. Wågberg, Carbon, 2009, 47, 1181.
10. L. Wang, B. B. Liu, D. D. Liu, M. G. Yao, S. D. Yu, Y. Y. Hou, B. Zou, T. Cui, G. T. Zou, B. Sundqvist, Z. J. Luo, H. Li, Y. C. Li, J. Liu, S. J. Chen, G. R. Wang and Y. C. Liu, Appl. Phys. Lett., 2007, 91, 103.
11. C. Park, H. J. Song and H. C. Choi, Chem. Commun., 2009, 4803.
12. K. Miyazawa, Y. Kuwasaki, A. Obayashi and M. Kuwabara, J. Mater. Res., 2002, 17, 83.
13. Y. Jin, R. J. Curry, J. Sloan, R. A. Hatton, L. C. Chong, N. Blanchard, S. V. Tolojan, W. H. Kroto and S. R. P. Silva, J. Mater. Chem., 2006, 16, 3715.
14. K. Miyazawa, J. Minato, T. Yoshii, M. Fujino and T. Suga, J. Mater. Res., 2005, 20, 688.
15. M. Sathish and K. Miyazawa, J. Am. Chem. Soc., 2007, 129, 13816.
16. M. Sathish, K. Miyazawa, J. P. Hill and K. Ariga, J. Am. Chem. Soc., 2009, 131, 6372.
17. C. Park, E. Yoon, M. Kawano, T. Joo and H. C. Choi, Angew. Chem., Int. Ed., 2010, 49, 9670.
18. M. G. Yao, X. H. Fan, D. D. Liu, B. B. Liu and T. Wågberg, Carbon, 2012, 50, 209.
19. A. A. Popov, S. F. Yang and L. Dunsch, Chem. Rev., 2013, 113, 5989.
20. T. S. Wang and C. R. Wang, Acc. Chem. Res., 2014, 47, 450.
21. X. Lu, L. Feng, T. Akasaka and S. Nagase, Chem. Soc. Rev., 2012, 41, 7723.
22. S. F. Yang, Curr. Org. Chem., 2012, 16, 1079.
23. S. F. Yang, F. P. Liu, C. B. Chen, M. Z. Jiao and T. Wei, Chem. Commun., 2011, 47, 11822.
24. N. Lyczko, F. Espitalier, O. Louisnard and J. Schwartzentruber, Chem. Eng. J., 2002, 86, 233.
25. H. Li, J. K. Wang, Y. Bao, Z. C. Guo and M. Y. Zhang, J. Cryst. Growth, 2003, 247, 192.
26. Z. Guo, M. Zhang, H. Li, J. Wang and E. Kougoulos, J. Cryst. Growth, 2005, 273, 555.
27. B. W. Zeiger and K. S. Suslick, J. Am. Chem. Soc., 2011, 133, 14530.
28. C. J. Li, Y. G. Guo, B. S. Li, C. R. Wang, L. J. Wan and C. L. Bai, Adv. Mater., 2005, 17, 71.
29. T. Tsuchiya, R. Kumashiro, K. Tanigaki, Y. Matsunaga, M. O. Ishitsuka, T. Wakahara, Y. Maeda, Y. Takano, M. Aoyagi, T. Akasaka, M. T. Liu, T. Kato, K. Suenaga, J. S. Jeong, S. Iijima, F. Kimura, T. Kimura and S. Nagase, J. Am. Chem. Soc., 2008, 130, 450.
30. T. Wakahara, Y. Nemoto, M. Xu, K. Miyazawa and D. Fujita, Carbon, 2010, 48, 3359.
31. Y. Xu, C. C. He, F. P. Liu, M. Z. Jiao and S. F. Yang, J. Mater. Chem., 2011, 21, 13538.
32. Y. Xu, J. H. Guo, T. Wei, X. Chen, Q. Yang and S. F. Yang, Nanoscale, 2013, 5, 1993.
33. S. F. Yang, C. B. Chen, A. A. Popov, W. F. Zhang, F. P. Liu and L. Dunsch, Chem. Commun., 2009, 6391.
34. A. Pozio, M. D. Francesco, A. Cemmi, F. Cardellini and L. Giorgi, J. Power Sources, 2002, 105, 13.
35. R. S. Ruoff, D. S. Tse, R. Malhotra and D. C. Lorents, J. Phys. Chem., 1993, 97, 3379.
36. J. F. Geng, W. Z. Zhou, P. Skelton, W. B. Yue, I. A. Kinloch, A. H. Windle and B. F. G. Johnson, J. Am. Chem. Soc., 2008, 130, 2527.
37. H. R. Barzegar, G. Hu, C. Larsen, X. Jia, L. Edman and T. Wågberg, Carbon, 2014, 73, 34.
38. L. Gao, L. Ding and L. Fan, Electrochim. Acta, 2013, 106, 159.

---

Footnotes

† Electronic supplementary information (ESI) available: Size distributions of cubic- and dice-shaped Sc₃N@C₈₀ crystals, SEM images of Sc₃N@C₈₀ crystals obtained under other conditions, TGA curve of Sc₃N@C₈₀ dice, FTIR spectra of Sc₃N@C₈₀ dice and cubes, detailed indexing data of Sc₃N@C₈₀ cubes, chronoamperometric curves, etc. See DOI: 10.1039/c4mh00037d

‡ These authors contributed equally to this work.

---