Epitaxial growth of Pd nanoparticles on molybdenum disulfide by sonochemistry and its effects on electrocatalysis

Abstract

In this work, we have synthesized Pd/MoS₂ samples in which about 7 nm sized Pd NPs are formed on a MoS₂ surface with the Pd content varied from 14.4 wt% to 33.6 wt% and studied their electrocatalytic activity for oxygen reduction reaction (ORR) in an alkaline medium. The syntheses were achieved by single-step sonochemical reactions between palladium acetylacetonate and 2H-MoS₂ in ethylene glycol. Based on the structural characterization data, the Pd NPs are firstly grown epitaxially on the MoS₂ surface in the [1−10]_Pd||[100]_MoS2 and (111)_Pd||(001)_MoS2 relationship. As the Pd-content increases, Pd NPs are formed on top of the first layer of Pd NPs as well as on the available MoS₂ surface. The epitaxially grown Pd NPs experience a tensile strain and charge-transfer to MoS₂, which raises the d-band center of Pd, lowering the on-set potential in ORR. On the other hand, the enhanced adsorption of O₂ on MoS₂ can facilitate the ORR kinetics of Pd NPs. The observed ORR data on Pd/MoS₂ can be explained as the net result of these two opposing effects of the MoS₂ support.

---

1. Introduction

Transition metal dichalcogenides such as MoS₂, MoSe₂, WS₂ and WSe₂ have received a lot of attention for their unique electrical, optical, chemical and mechanical properties, which may find applications in various fields such as gas-sensing, water-splitting, and Li-ions batteries to name a few.^1–3 Among them, MoS₂ has been the most extensively studied. Recently, a number of papers have appeared on the use of MoS₂ as an electrocatalyst or a support material of electrocatalysts in fuel cells.^4–9 As for the application as electrocatalysts, most studies have focused on the electrocatalysis for the hydrogen evolution reaction (HER).^4–7 For instance, exfoliated 1T-MoS₂ nanosheets and nanostructured MoS₂ are reported to show excellent electrocatalytic activity for HER.^5,6

On the other hand, there have been only a few papers that report the use of MoS₂ as a support for electrocatalysts in fuel cells.^8–10 Although small in number, all of them indicate that the electrocatalytic performance of noble metal nanoparticles (NPs) can be enhanced by MoS₂ supports. For instance, Yuwen et al. showed that the methanol oxidation reaction on Pd NP electrocatalyst is enhanced by a MoS₂ support.⁸ Wang et al. reported a synergy effect between Au NPs and a MoS₂ support in ORR.⁹ Apparently, MoS₂ is a promising candidate that can replace the predominant porous carbon supports which suffer from corrosion during the cell operation.¹¹ These positive effects of MoS₂ supports also suggest that there are certain interactions between MoS₂ and noble metal NPs, for which the details are not yet fully understood.

In attempts to observe the effects of MoS₂ on the electronic structures of noble metal NPs on it and, hence, the electrochemical performance of the noble metal catalysts, noble metal NPs grown epitaxially on the MoS₂ basal plane have been investigated theoretically and experimentally.^10,12 Saidi investigated epitaxial M–MoS₂ contacts (M = Pd and Pt) by using the density functional theory (DFT) calculations.¹² They showed that the strain caused by the lattice mismatch between M and MoS₂ and the thickness of M affected the electronic structure of M. Huang et al. synthesized and characterized epitaxially grown Au, Pt and Pd NPs on the MoS₂ basal plane.¹⁰ The epitaxial growth is achieved mainly through sharing the MoS₂ (001) plane with the (111) planes of the face-centered cubic lattices of noble metals. The lattice mismatches are ∼13%. They also showed that their Pt/MoS₂ had excellent electrocatalytic activity for HER. However, understanding on the effects of MoS₂ as a support material is still far from completion.

The purpose of the present paper is to investigate the effects of a MoS₂ support on the electrocatalytic properties of Pd NPs. Towards this goal we have devised a new synthesis method to prepare Pd NPs epitaxially on delaminated MoS₂. In this method, Pd/MoS₂ samples with various Pd loadings from by 14.4 wt% to 33.6 wt% were synthesized by single step sonochemical reactions. We found that the electronic structure and lattice strain of Pd are strongly affected depending on the Pd loading. These changes were directly correlated with the ORR activity of the Pd/MoS₂ samples. The details of synthesis and characterization data will be discussed herein.

2. Experimental section

2.1 Materials

Palladium acetylacetonate (Pd(acac)₂, Pd(C₅H₇O₂)₂, 99%) and MoS₂ powder (99%, 2H-MoS₂ according to the supplier, size < 2 μm) were purchased from Sigma-Aldrich. Ethylene glycol (EG, 99.9%) and ethanol (99.9%) were purchased from Samchun Pure Chemical Co. Ltd.

2.2 Preparation of Pd/MoS₂ samples

The synthesis of Pd/MoS₂ samples was achieved in a single-step sonochemical reaction. 30 mL of EG was purged by Ar and then Pd(acac)₂ and MoS₂ were added to it. This dispersion was sonicated by using a high-intensity ultrasonic probe (Sonic and materials, VC-500, 30% amplitude, 20 kHz, with 13 mm solid probe) for 3 h under Ar. Three samples were prepared by varying the amount of Pd(acac)₂ from 0.05 mmol to 0.15 mmol while keeping the amount of MoS₂ at 0.03 g. After the sonochemical reaction, the dark slurry was filtered with a nylon membrane filter paper with 0.2 μm pore size (Whatman int. Ltd.). After washing with a copious amount of ethanol to ensure complete removal of EG, the samples were dried in a vacuum desiccator at room temperature for 12 h. Through elemental analyses, the Pd-contents of the three samples were determined to be 14.4, 25.6 and 33.6 wt%, respectively. Hereafter, these will be denoted as Pd(14.4)/MoS₂, Pd(25.6)/MoS₂, and Pd(33.6)/MoS₂, respectively. The yields are 95.8–98.0% based on Pd and 82.8–93.3% based on Mo. Therefore, the Pd contents in the final products are slightly higher than the loaded ones.

2.3 Electrochemical measurements

5 mg of a Pd/MoS₂ electrocatalyst was dispersed in triply de-ionized water and the dispersion was irradiated by ultrasound to make a homogeneous suspension. 5 μL of the suspension was dropped on a glassy carbon electrode (d = 3 mm, area = 0.0707 cm²) and dried in an ambient condition. The amounts of Pd in the electrodes are calculated to be in the range of 50.7–70.4 μg cm⁻². 5 μL of a Nafion diluted dispersion (0.05 wt%, DuPont™) was dropped on top of the electrode for mechanical stability. A Hg/HgO electrode (BAS Inc.) and a Pt wire were used as reference and counter electrodes, respectively. Potential cycling for 50 cycles from −0.6 V to 0.0 V was carried out to remove contamination or organic residue on the surface of working electrode before each measurement. Cyclic voltammograms (CVs) of samples were recorded from −0.8 V to 0.3 V in a N₂ saturated 0.1 mol L⁻¹ KOH aqueous solution with a scan rate of 50 mV s⁻¹. Linear sweep voltammograms (LSVs) for ORR were recorded from 0.2 V to −0.8 V in a O₂ saturated 0.1 mol L⁻¹ KOH aqueous solution with a scan rate of 10 mV s⁻¹.

2.4 Characterization

The structural analyses of Pd/MoS₂ samples were carried out by powder X-ray diffractometry (XRD, Rigaku, CuKα₁, λ = 1.54056 Å), transmission electron microscopy (TEM, JEOL JEM-3011 model, 200 kV) and scanning transmission electron microscopy (STEM, JEOL JEM-ARM200F, 200 kV). The metal contents of samples were determined by inductively coupled plasma-atomic emission spectroscopy (ICP-AES, OPTIMA 4300DV Perkin Elmer). X-ray photoelectron spectroscopy (XPS, SIGMA PROBE) measurements were performed with a monochromatic Al Kα source (1486.6 eV). All XPS spectra were calibrated with C 1s (284.6 eV).

3. Results and discussion

The MoS₂ we used has the 2H-type structure, the more abundant form between the two polytypes of MoS₂ (Fig. S1†). As a preliminary study to see if MoS₂ is indeed delaminated by sonication, we treated bulk MoS₂ dispersed in EG without Pd(acac)₂ by ultrasound. The XRD patterns before and after the sonication show that the 2H-type structure is not changed by the sonication but that the intensity ratio between (002) and (110) peaks is decreased from 7.0 to 4.7 (Fig. S1†). The TEM image on the sonicated MoS₂ sample shows very thin flakes which are absent in the bulk MoS₂ sample (Fig. S2†). With these, we can conclude that ultrasound can indeed induce delamination of MoS₂. However, full exfoliation appears to be very hard or impossible by sonication alone. In the literature on exfoliation by sonication, additional treatments such as fractionation by ultracentrifugation appear to be necessary to obtain a fully exfoliated MoS₂ sample.^13,14 The same procedure was followed for the synthesis of Pd/MoS₂ samples except that Pd(acac)₂ was added in the EG suspension containing MoS₂.

Fig. 1(a) shows the XRD patterns of Pd/MoS₂ samples and those of MoS₂ and Pd black as references. All of the Pd/MoS₂ patterns can be explained with 2H-MoS₂ (JCPDS no. 77-1716) and Pd (JCPDS no. 87-0645). As in the case of the sonicated MoS₂ explained above, the MoS₂ (002) peak at ∼14.5° is considerably reduced compared with the other peaks of MoS₂. The two broad peaks at ∼39.5° and ∼47.0° are the (111) and (200) diffractions of Pd. These two peaks increase in intensity as the Pd-content increases. Since the Pd (111) peak overlaps with the MoS₂ (103) peak, they are deconvoluted to extract the information on the Pd (111) peaks (Fig. 1(b)). In fact, analyses of the Pd (200) peaks would be better because they do not overlap with the MoS₂ peaks. However, unfortunately, this peak in Pd(14.4)/MoS₂ is too weak to be analyzed (Fig. S3†). The Pd (111) peaks are located at lower 2θ than that of Pd black indicating lattice expansion upon forming Pd/MoS₂. Pd(14.4)/MoS₂, the one with the lowest Pd-content, shows the largest peak shift and the magnitude of peak shift decreases slightly as the Pd-content increases. By using the position of this peak, the lattice constants of Pd in Pd/MoS₂ samples are calculated to be 0.393–0.396 nm, larger than that of Pd black (0.389 nm). It is well known that Pd can absorb hydrogen atoms.^15,16 The lattice constant of hydrogen-containing Pd is in the range of 0.396–0.399 nm depending on the H-content.¹⁶ Therefore, the expanded Pd lattices in Pd/MoS₂ samples can be understood as a result of hydrogen uptake. Alternatively, the epitaxial relationship between Pd on MoS₂ to be discussed below can induce the expansion of Pd lattice. Probably, these two effects occur simultaneously and the epitaxial relationship provides an additional stability to the hydrogen-absorbed Pd. The crystallite sizes, estimated by applying the Scherrer equation to the widths of these peaks, are 7.2 nm for Pd(14.4)/MoS₂ and 7.0 nm for Pd(25.6)/MoS₂ and Pd(33.6)/MoS₂, slightly smaller than that of Pd black (9.6 nm). The structural information on the samples is summarized in Table 1.

Fig. 1 (a) XRD patterns of samples and (b) magnified view to highlight the deconvolution between Pd (111) (red lines) and MoS₂ (103) peaks (blue lines) on Pd/MoS₂ samples.

Table 1 Structural and electrochemical characteristics of samples

Samples	Crystallite size^a (nm)	Lattice constant^a (Å)	Pd–Pd distance^a (Å)	ECSAs^b (m^2 per gmetal)
a Crystallite size, lattice constant and Pd–Pd distance of samples were calculated from the (111) peaks of XRD patterns of samples.b ECSAs of samples were determined by using the HUPD.
Pd(14.4)/MoS2	7.2	3.96	2.80	17
Pd(25.6)/MoS2	7.0	3.95	2.79	18
Pd(33.6)/MoS2	7.0	3.93	2.78	14
Pd black	9.6	3.89	2.75	11

---

The TEM images of the three Pd/MoS₂ samples are shown in Fig. 2. In these low magnification images, one can see that the MoS₂ surface is covered by Pd NPs of sizes smaller than 10 nm. The NPs are well dispersed over the entire MoS₂ surface. The MoS₂ parts in these images are light in darkness indicating that they are thin, an indication of delaminated or even exfoliated MoS₂. In Pd(14.4)/MoS₂ sample of the lowest Pd-content, some NPs are isolated from the other ones but the majority appears to be connected to one another in the horizontal direction. As the Pd-content increases, the surface coverage by Pd NPs increases and the degree of horizontal interconnection between NPs also increases as shown in insets of Fig. 2. In addition, there are NPs stacked on top of the other one, as judged by the darkness, whose population also increases as the Pd-content increases. The increasingly darker image with the Pd-content, therefore, can be understood as due to the increase of the electron density by the thickening of the Pd NP layers on both sides of MoS₂. On the other hand, the size of the primary Pd NPs does not appear to change significantly in agreement with the estimation based on the XRD peak widths.

Fig. 2 Low magnified transmission electron microscopy images of (a) Pd(14.4)/MoS₂, (b) Pd(25.6)/MoS₂ and (c) Pd(33.6)/MoS₂. Inset shows enlarged images of dot marked area.

Pd(33.6)/MoS₂ sample was characterized by HRTEM, HAADF, and EDS line profiling. Fig. 3(d) shows a HAADF-TEM image of Pd(33.6)/MoS₂. Since the brightness in HADDF images is proportional to the square of atomic number (Z²),¹⁷ the bright part can be identified as Pd NPs on MoS₂ support. The identity of Pd of this part is further proven by the STEM-EDS line profile overlapping with the HAADF image. There are also much bright places, which we believe to be the stacking of Pd NPs on the ones below them.

Fig. 3 TEM analysis of Pd(33.6)/MoS₂. (a) high resolution TEM image, (b and c) the fast Fourier transfer (FFT) generated SAED patterns on the yellow and red marked area, (d) STEM-EDS line profile and (e) Pd coverage on MoS₂ (in three different levels of no coverage, single crystalline Pd, and multiple layer of Pd) determined from FFT-ED patterns generated on 64 grids (see text).

The HRTEM image (Fig. 3(a)) shows regions with roughly three different levels of darkness. In the brightest regions as the one encircled by a yellow line, the lattice fringe pattern has d-spacing of 0.27 nm that can be identified as the MoS₂ (100) planes. The ED pattern generated by fast-Fourier transformation (FFT-ED patterns) on this lattice fringe image (Fig. 3(b)) shows that this region is single crystalline with the hexagonal symmetry. Evidently, the brightest regions show the pristine MoS₂ substrate with no Pd on it. In the regions of intermediate level of darkness such as the one encircled in red, the lattice fringe pattern has d-spacing of 0.23 nm that corresponds to Pd (111) planes.⁸ The FFT-ED pattern of this image (Fig. 3(c)) shows two sets of ED patterns, both in the hexagonal symmetry. One of them matches well with that of MoS₂ in Fig. 3(b) and the other can be explained as the 〈111〉 projection of the diffraction pattern of Pd. This FFT-ED also indicates that the Pd crystal is grown epitaxially on MoS₂ in the [1−10]_Pd||[100]_MoS2 and (111)_Pd||(001)_MoS2 relationship. The darkest regions in Fig. 3(a) show more complicated lattice fringe patterns most likely because of the stacking of Pd NPs on top of other NPs. The FFT-ED patterns generated from these regions show the characteristics of a multiple crystalline sample in agreement with this understanding (data not shown).

We estimated the surface coverage of Pd NPs on MoS₂ by inspecting the FFT-ED patterns generated on the 8 × 8 = 64 grids made on the HRTEM image as shown in Fig. S4–S6.† Each FFT-ED pattern can be identified as that of pure MoS₂, single crystalline Pd on MoS₂, or multi-crystalline Pd on MoS₂. The proportion of the latter two types of grids to the total of 64 grids will be a rough estimate for the surface coverage of Pd on MoS₂. The corresponding plot is shown in Fig. 3(e). By this estimation, the surface coverage of Pd(33.6)/MoS₂ is 95%. The surface coverage on the other two samples was estimated to be 55% and 75% for Pd(14.4)/MoS₂ and Pd(25.6)/MoS₂, respectively. With these, one can see that the surface coverage increases with the Pd-content but is not exactly proportional to the Pd-content, which suggests that the proportion of multiple stacking of Pd NPs increases as the Pd-content increases.

The electronic structures of the elements in Pd/MoS₂ are probed by XPS. In Fig. 4(a)–(c), the Pd 3d, S 2p and Mo 3d regions of XPS spectrum taken on Pd(33.6)/MoS₂ are shown. The doublet of Mo⁴⁺ band at 229.5 eV and 232.6 eV in the Mo 3d region of XPS agree well with that of MoS₂ in the literature.^18,19 Contrary to Mo, Pd and S species undergo changes in binding energy (BE) upon forming Pd/MoS₂ samples. The S 2p regions of the spectra of Pd/MoS₂ samples show one set of peaks of S 2p_1/2 and S 2p_3/2 (Fig. 4(b)) with the respective BEs of 162.2 and 163.4 eV. These can be assigned to S²⁻ in MoS₂, but are slightly red-shifted from those of pure MoS₂ (S 2p_3/2 = 162.4 eV) suggesting charge transfer from Pd to MoS₂.²⁰ The absence of other types of sulfur such as SO₄²⁻ indicates that the chemical identity of MoS₂ is intact under the UPS condition. The Pd 3d regions of the spectra are deconvoluted to three sets of peaks as shown in Fig. 4(a). The Pd 3d_5/2 peaks at 335.5 eV, 336.8 eV and 338.4 eV correspond to Pd⁰, Pd²⁺ and Pd⁴⁺, respectively.²¹ Compared with metallic Pd (3d_5/2 = 335.1 eV), the peak assigned to Pd⁰ is blue-shifted, implying charge transfer from Pd to MoS₂.²⁰ Interestingly, however, within the three Pd/MoS₂ samples, the Pd 3d peak is red-shifted as the Pd content decreases. Probably, this behavior of Pd BE is related with the variation of the thickness of Pd NPs on MoS₂. While the Pd NPs in direct contact with MoS₂ electronically interact with MoS₂, those on top of the first layer of Pd NPs do not experience such interaction. It is in agreement with the ab initio DFT calculations results reported by Saidi in which the d-band center of the top Pd layer is down-shifted as the number Pd atom layers on MoS₂ increases through.¹²

Fig. 4 X-ray photoelectron spectra. (a) Pd 3d region, (b) S 2p region and (c) Mo 3d region of Pd(33.6)/MoS₂ (d) Pd 3d region of samples. Vertical lines in (b) and (c) respectively indicate the S 2p_3/2 and Mo 3d_5/2 position of MoS₂.

In recent papers, MoS₂ has been used as a support for noble metal NP catalysts.^8,9 However, from the view point of the changes of electronic structure of the noble metal, epitaxial growth appears to give a number of drawbacks. The catalytic properties of noble metals in alloys or composites are explained by ligand effect and/or strain effect.^22,23 In case of Pd, a charge transfer to Pd or a strain effect that squeezes Pd–Pd distances can increase the catalytic properties of Pd.²⁴ According to this view, the observed changes of Pd by forming Pd/MoS₂ are in the opposite directions for enhancement of the catalytic properties of Pd. That is, there is charge transfer from Pd, not to Pd, and the Pd NPs experience tensile strain, not compressive strain. Therefore, Pd/MoS₂ samples are expected to have inferior catalytic properties to Pd black.

We have tested the catalytic properties of Pd/MoS₂ samples for ORR in an alkaline medium. Cyclic voltammograms (CVs) of samples were recorded in N₂-saturated 0.1 mol L⁻¹ KOH aqueous solution with scan rate of 50 mV s⁻¹ (Fig. 5(a)). All Pd-based samples showed the oxide formation peak at 0.05 V and reduction peak −0.2 V. In general, the electrochemical surface areas (ECSAs) of samples are calculated by using the integrating of hydrogen desorption peak area.²⁵ Unfortunately, Pd-based samples do not have hydrogen desorption peaks. Therefore, we have obtained the ECSAs of samples by integrating of reduction peak area of oxide species. The literature value of 420 μC cm⁻² for polycrystalline Pd was used as the reference.²⁶ The ECSAs of samples were summarized in Table 1.

Fig. 5 Electrochemical analyses of samples. (a) Cyclic voltammograms in N₂-saturated 0.1 mol L⁻¹ KOH (scan rate: 50 mV s⁻¹), (b) linear sweep voltammograms in O₂-saturated 0.1 mol L⁻¹ KOH (scan rate: 10 mV s⁻¹).

As shown in Fig. 5(b), the on-set potential of ORR on Pd(14.6)/MoS₂ is 0.05 V lower than that of Pd black, which can be understood as a consequence of the up-shift of the d-band center of Pd. As the Pd-content increases, the proportion of Pd NPs that are not in epitaxial relation with MoS₂ increases. This makes the on-set potential close to that of Pd black. A physical mixture between Pd black and MoS₂ (Pd + MoS₂) would make a situation in which Pd NP are in the pristine state, which would be better suited for catalysis. Indeed, the electrode made of Pd + MoS₂ shows a higher on-set potential than Pd/MoS₂ in agreement with the prediction in Fig. 5(b). This electrode also shows a higher on-set potential than Pd black, which cannot be explained by the electronic structure of Pd. It has been shown that MoS₂ can function as a co-catalyst. Wang et al. showed that a physical mixture between Au NPs and exfoliated MoS₂ shows a synergy effect between them with a high ORR on-set from Au NPs and a facilitated kinetics from MoS₂.⁹ However, they did not give any explanation on how MoS₂ facilitates the ORR kinetics. In the literature, MoS₂ is reported to be a good substrate to adsorb oxygen molecules at the edges.²⁷ Through this property, it seems that MoS₂ can produce high O₂ concentration environment for the Au NPs in the neighborhood. It seems that the same mechanism works for the Pd + MoS₂ electrode in the present work. In fact, all of Pd/MoS₂ electrodes show improved kinetics, which also seem to originate from the O₂ adsorbing property of MoS₂.

4. Conclusion

We demonstrate the synthesis of Pd/MoS₂ nanocomposites by sonochemical reaction technique. Delamination of MoS₂ layers, formation of Pd NPs, and deposition of Pd NPs on MoS₂ surface are achieved in a single-step reaction. The first layer of Pd NPs is composed of Pd NPs grown epitaxially on the MoS₂ surface; additional Pd NPs are deposited on the first layer Pd NPs and do not have any specific crystallographic relation with MoS₂. The electronic structure of the first layer Pd NPs are affected by the epitaxial relation. Both the charge transfer from Pd to S²⁻ in MoS₂ and the tensile stress cause the up-shift of the d-band center of Pd. These effects are detrimental for the electrocatalytic properties of Pd NPs. ORR on-set potentials of Pd/MoS₂ are lower than that of Pd black, reflecting the rise of the d-band center. On the other hand, a physical mixture between Pd black and MoS₂ shows enhanced ORR properties with increased on-set potential and improved kinetics, which can be understood as a consequence of the adsorption of O₂ molecules on the edges of MoS₂. Our results show that MoS₂ has a certain merits to be used as a co-catalyst for other catalyst. However, MoS₂ has a few inherent limitations because of the nature of its interactions with noble metals to be deposited on it.

Acknowledgements

This work was supported by the Agency for Defense Development through Chemical and Biological Defense Research Center (CBDRC-25) and by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (NRF-2009-0083540).

References

1. Q. H. Wang, K. Kalantar-Zadeh, A. Kis, J. N. Coleman and M. S. Strano, Nat. Nanotechnol., 2012, 7, 699–712.
2. M. Chhowalla, H. S. Shin, G. Eda, L.-J. Li, K. P. Loh and H. Zhang, Nat. Chem., 2013, 5, 263–275.
3. M. Pumera, Z. Sofer and A. Ambrosi, J. Mater. Chem. A, 2014, 2, 8981–8987.
4. J. Kibsgaard, Z. Chen, B. N. Reinecke and T. F. Jaramillo, Nat. Mater., 2012, 11, 963–969.
5. M. A. Lukowski, A. S. Daniel, F. Meng, A. Forticaux, L. Li and S. Jin, J. Am. Chem. Soc., 2013, 135, 10274–10277.
6. Y. Li, H. Wang, L. Xie, Y. Liang, G. Hong and H. Dai, J. Am. Chem. Soc., 2011, 133, 7296–7299.
7. D. Gopalakrishnan, D. Damien and M. M. Shaijumon, ACS Nano, 2014, 8, 5297–5303.
8. L. Yuwen, F. Xu, B. Xue, Z. Luo, Q. Zhang, B. Bao, S. Su, L. Weng, W. Huang and L. Wang, Nanoscale, 2014, 6, 5762–5769.
9. T. Wang, J. Zhuo, Y. Chen, K. Du, P. Papakonstantinou, Z. Zhu, Y. Shao and M. Li, ChemCatChem, 2014, 6, 1877–1881.
10. X. Huang, Z. Zeng, S. Bao, M. Wang, X. Qi, Z. Fan and H. Zhang, Nat. Commun., 2013, 4, 1444.
11. S. Dhanushkodi, M. Tam, S. Kundu, M. Fowler and M. Pritzker, J. Power Sources, 2013, 240, 114–121.
12. W. A. Saidi, J. Chem. Phys., 2014, 141, 094707.
13. S. M. Notley, Langmuir, 2012, 28, 14110–14113.
14. M. Lotya, Y. Hernandez, P. J. King, R. J. Smith, V. Nicolosi, L. S. Karlsson, F. M. Blighe, S. De, Z. Wang and I. McGovern, J. Am. Chem. Soc., 2009, 131, 3611–3620.
15. H. Kobayashi, M. Yamauchi, H. Kitagawa, Y. Kubota, K. Kato and M. Takata, J. Am. Chem. Soc., 2008, 130, 1818–1819.
16. J.-H. Jang, J. Kim, Y.-H. Lee, C. Pak and Y.-U. Kwon, Electrochim. Acta, 2009, 55, 485–490.
17. M. Takahashi, K. Higashimine, P. Mohan, D. M. Mott and S. Maenosono, CrystEngComm, 2015, 17, 6923–6929.
18. A. Saha, P. Bharmoria, A. Mondal, S. C. Ghosh, S. Mahanty and A. B. Panda, J. Mater. Chem. A, 2015, 3, 20297–20304.
19. M. Acerce, D. Voiry and M. Chhowalla, Nat. Nanotechnol., 2015, 10, 313–318.
20. H. Li, K. Yu, X. Lei, B. Guo, H. Fu and Z. Zhu, J. Phys. Chem. C, 2015, 119, 22681–22689.
21. A. Drelinkiewicz, M. Hasik and M. Kloc, Catal. Lett., 2000, 64, 41–47.
22. E. Lee and Y.-U. Kwon, Ultrason. Sonochem., 2016, 29, 401–412.
23. E. Lee, J.-H. Jang, M. A. Matin and Y.-U. Kwon, Ultrason. Sonochem., 2014, 21, 317–323.
24. M. A. Matin, J.-H. Jang and Y.-U. Kwon, J. Power Sources, 2014, 262, 356–363.
25. J.-H. Jang, E. Lee, J. Park, G. Kim, S. Hong and Y.-U. Kwon, Sci. Rep., 2013, 3, 2872.
26. Y. Liu and C. Xu, ChemSusChem, 2013, 6, 78–84.
27. Z. Hong and J. R. Regalbuto, J. Phys. Chem., 1995, 99, 9452–9457.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra07064g

---