Graphite induced periodical self-actuation of liquid metal

Abstract

Self-actuation phenomena of liquid metal spheres in NaOH solution, including spreading, oscillating and stretching, induced by graphite alone were demonstrated for the first time. A liquid metal sphere could spontaneously spread on the surface of the graphite once immersed in the NaOH solution. The surface tension gradient on the sphere induced by the graphite/liquid metal galvanic cell was responsible for this deformation. When a liquid metal sphere was leaned against the side of a piece of graphite, it could oscillate periodically. As the sphere contacted the graphite, it rapidly collapsed, while the curvature radius of the sphere at the contact point decreased. Also, as the capillary force imposed on the sphere was larger than the friction force, the sphere recovered its original spherical shape. The surface tension of the liquid metal sphere acted as the restoring force of the oscillatory movement. Further, a phenomenon of resonance could be observed when two spheres were laid respectively on the top and the side of the graphite. The vibration of the top sphere was induced by the vibration of the side sphere. This finding provides a novel enhancement for the fabrication of future liquid metal beating heart systems and graphite/liquid metal-based batteries or machines.

---

1 Introduction

Investigation of the characteristics of liquid metal, including its thermal, electrical and rheological properties, has gained momentum in recent years.¹ Among these, the rheological characteristics of liquid metal under an electric field^2–4 have received special interest. A sheet of liquid metal film immersed in an electrolyte can transform itself from the flattened shape into a sphere as it contacts a cathode electrode,² and a liquid metal sphere will spread into a disc-shaped or petal-shaped object as it contacts an anode electrode.^3,5 When a liquid metal sphere is located between the electrified electrodes, it will move from the cathode to the anode while maintaining internal self-rotation.^2,4 The deformation of the liquid metal is generated by the electrocapillary force due to the electrowetting effect^6–8 as an external electric field is applied. In addition, other experiments such as mercury beating heart (MBH) systems, which are based on chemical redox reactions, have been investigated extensively. MBH is a classical electrochemical oscillator that was first reported by Lippmann as early as 1873.⁹ The experimental setup consists of a glass tube containing mercury covered by an aqueous solution of sulfuric acid and an oxidizing agent. As a piece of ion or aluminum is brought near the mercury drop periphery, a sustained pulsating motion of the drop is induced.^10,11 The electrocapillary effect and the surface tension gradient are generally considered to be responsible for the oscillations of the mercury drop.¹² Keizer et al.^13,14 associated the oscillations in sulfuric acid solution of strong oxidants with the formation and removal of a film of Hg₂SO₄ on the mercury drop. Subsequently, the MBH system was studied under different experimental conditions, including the presence and the absence of an applied voltage.^15–17 Smolin et al.¹⁸ observed standing waves in a linear geometry with a number of nodes depending on the potential of the used metal tip. Castillo-Rojas et al.¹⁹ investigated the dynamics of the MBH system in the presence of γ-radiation and demonstrated different topological configurations of the mercury drop. However, few studies exist regarding liquid metal deformation induced by graphite, which is otherwise the focus of many research. As a non-metallic functional material, graphite possesses superior physical properties, including electrical and thermal conductivities, which enable its wide application, especially in the fabrication of nanocomposites.^20–23 Also, its excellent chemical stability makes it a good candidate as an electrode material in the fabrication of electrochemical devices, including fuel cells, batteries and supercapacitors.²⁴ In this study, a fundamental, novel phenomenon of spreading, oscillating and stretching behaviors of liquid metal immersed in NaOH electrolyte solution and laid against a piece of graphite is disclosed, and the mechanisms are interpreted.

2 Results

2.1. Experimental materials

Galinstan (68.5% gallium, 21.5% indium and 10% tin by weight) with a melting point of about 11 °C is taken as the research object. It offers remarkable properties, such as high electrical conductivity, high surface tension, and negligible toxicity compared to mercury.²⁵ The preparation process for making this metal fluid was as follows: three metals, gallium, indium and tin (with high purity of 99.99%) are weighed according to the ratio of 68.5 : 21.5 : 10. These pure metals are placed in a beaker for 2 hours at 150 °C in an electric vacuum drying oven. Then, the mixture is stirred in the beaker, which is placed in a water bath at 50 to 60 °C for 30 min to ensure a well-mixed alloy solution. Ultra-pure graphite (99.999%) is utilized through all the experiments. NaOH solution with a concentration of 0.5 mol L⁻¹ is selected as the electrolyte.

2.2. Spreading phenomenon of liquid metal on the graphite surface

A piece of cuboid graphite with a liquid metal sphere with a mass of 0.6 g on the surface was placed in a petri dish, and NaOH electrolyte solution with a concentration of 0.5 mol L⁻¹ was gradually added to the dish. The variation of the voltage between the liquid metal and graphite with time was measured and is shown in Fig. 1(A). As the liquid metal sphere proceeded from unsubmerged (see snapshot 1 in the inset) to submerged (see snapshots 2 and 3 in the inset) in the solution, it rapidly collapsed and spread on the graphite surface (see Movie S1†). Meanwhile, the voltage value changed from positive to negative. This spreading deformation of the liquid metal and the voltage variation can be interpreted as follows: before the liquid metal is submerged, it is in direct contact with the graphite surface. The contact potential difference V_lg between the liquid metal and the graphite can be expressed as:^26,27where e is the charge of an electron, V_l and Φ_l represent the potential and the work function of the liquid metal, respectively, and V_g and Φ_g represent the potential and the work function of the graphite, respectively. As the work function of the graphite is equal to 4.7 eV (ref. 28) and those of each component (Ga, In and Sn) of the liquid metal are lower than 4.4 eV,²⁹ V_lg is a positive value according to eqn (1). This means that the potential of the liquid metal is larger than that of the graphite. An electric double layer (EDL) is formed at the graphite/liquid metal interface when the electrons come to a dynamic balance. Because electrons are prone to transfer from the liquid metal, which has a smaller work function, the residual electrons are concentrated on the graphite side. This is verified in the curve, where the voltage between the liquid metal and the graphite is a small positive value (about 0.031 mV) before the liquid metal is submerged in the NaOH solution. The value of the voltage (about 0.031 mV) is far lower than the theoretical calculation result (larger than 0.3 V) obtained from eqn (1). This can be attributed to the rough surfaces of the graphite and the liquid metal.²⁷

Fig. 1 Spreading deformation of a liquid metal sphere on the graphite substrate immersed in the NaOH solution. (A) Variation of the voltage between liquid metal and graphite before and after their immersion in the NaOH solution. The insets are images of the liquid metal at time points 1, 2 and 3, respectively. Scale bar is 3 mm in length. (B) Schematic of the charge distribution on the surface of the liquid metal in the initial state. The enlarged image (marked by the dashed box) shows the charge distributions and the chemical reactions on the graphite/electrolyte and the liquid metal/electrolyte EDLs. (C) Schematic of the deformation of the liquid metal in the end state. The red dashed-line circle represents the outline of the liquid metal before deformation.

As the liquid metal is immersed in the NaOH solution, its surface oxide (main component: Ga₂O₃ (ref. 30)) is removed by the electrolyte, and the chemical formula can be expressed as:³¹

Ga₂O₃ + NaOH → NaGaO₂ + H₂O (2)

Pure gallium will be exposed to the solution, and gallates such as [Ga(OH)₄]⁻ are produced and confer a negative charge on the liquid metal. Then, the positive ions in the solution are electrostatically attracted by these anions, and an EDL is generated at the liquid metal/electrolyte interface.^1,3 Meanwhile, electrons concentrate at the graphite/electrolyte EDL from the liquid metal through the NaOH solution, and the graphite side of the EDL is positively charged. A galvanic cell system is thus formed with graphite, liquid metal and NaOH solution as the anode, cathode and electrolyte, respectively. The chemical reactions taking place at the anode and the cathode of this graphite/liquid metal cell can be respectively expressed as:

Cathode:³²

Ga + 4OH⁻ − 3e⁻ → H₂GaO₃⁻ + H₂O (3)

Anode:

2H⁺ + 2e⁻ → H₂↑ (4)

It can be seen that hydrogen gas is generated from the graphite anode. This is verified in our experiment, as shown in Fig. 2; when several liquid metal spheres remain on the graphite for an hour, the surface will be covered with bubbles, compared with the clean surface without liquid metal (see Movie S2†). This graphite/liquid metal cell provides a novel method for producing hydrogen in a convenient way.

Fig. 2 Comparison between graphite with (A) and without (B) liquid metal droplets on the surface. Scale bars are all 3 mm in length.

The liquid metal can be regarded as an equipotential body for its high electric conductivity.⁴ Also, there will be a uniform charge distribution along the liquid metal/electrolyte EDL without an external electric field. However, in the presence of the graphite/liquid metal galvanic cell, there will be a potential gradient with downward direction on the surface of the sphere via the electrolyte, whose resistance can be seen as R. The surface tension of the sphere is thus altered due to the potential gradient. According to Lippmann's equation:⁶

where γ is the surface tension of the liquid metal, c is the capacitance of the EDL per unit area, V is the potential difference across the EDL and γ₀ is the maximum value when V = 0. Taking the derivation of both sides of eqn (5) along the z direction as shown in Fig. 1(B), we have:

Assuming the direction of V is from the electrolyte to the liquid metal, then V is a positive value while is negative in eqn (6). Therefore is positive, which means that the surface tension on the upper hemisphere of the liquid metal sphere is lower than that on the other half. As a result, the sphere spreads on the graphite; the schematic diagram is presented in Fig. 1(C).

As a liquid metal sphere is laid on the surface of copper in NaOH solution, it will rapidly spread, which indicates the formation of a copper/liquid metal galvanic cell (as shown in Fig. 3). However, the sphere will remain intact without spreading when the substrate is Teflon, which is a type insulator. This shows that the deformation of liquid metal cannot be induced by insulators because no galvanic effect is generated. The deformation characteristics of liquid metal gradually pressed by graphite, copper and Teflon in NaOH solution were also comparatively investigated (see Movie S3 and S4†). A liquid metal sphere can be pressed into a thin, round cake shape by graphite and copper. Also, the voltage between these substrates and the liquid metal is changed from about 1.2 V to about 0 V, which is presented in Fig. 4. As the liquid metal sphere is pressed by Teflon, it will be squeezed out without being flattened, and the voltage shows no obvious change. These experiments indicate the similar electrochemical properties of graphite and copper, which can both constitute galvanic cells with liquid metal.

Fig. 3 (A) The lateral spread deformation of a liquid metal sphere on copper in NaOH solution. (B) The liquid metal sphere on Teflon in NaOH solution. Scale bars are all 3 mm in length.

Fig. 4 (A) Voltage–distance relationships between graphite, copper, Teflon and liquid metal when it is pressed. The curves on the left side of the green dotted line represent the approaching process, while those on the right side represent the departure process. The inset is the schematic of the experimental setup. (B) Snapshot of liquid metal just contacting the graphite. (C) Snapshot of liquid metal pressed flat by the graphite. (D) Image of liquid metal just contacting the copper. (E) Image of liquid metal pressed flat by the copper. (F) Image of liquid metal just contacting the Teflon. (G) Image of liquid metal pressed flat by the Teflon. Scale bars are all 8 mm in length.

2.3. Self-oscillation phenomenon of liquid metal on the side of the graphite

When a liquid metal sphere is laid on the side of the graphite, periodic oscillation behavior of the sphere can be observed. The substrate was tilted slightly to ensure that the liquid metal sphere could roll toward the graphite spontaneously. The top view snapshots of a liquid metal sphere of 1.2 g (see Movie S5 and S6†) are shown in Fig. 5(A), while the side view snapshots of a sphere with 0.9 g (see Movie S7†) are shown in Fig. 5(B). Relative variations of the frontal projected area of the 1.2 g sphere and the height of the 0.9 g sphere during the oscillation process are shown in Fig. 5(C). The green dashed lines represent the departure of the spheres from the graphite. It can be seen that at the beginning of the oscillation, the liquid metal spheres are in the shape of an approximate spheroid with a minimum frontal projected area A₀ and a maximum height H₀. As the spheres are in contact with graphite, the area rapidly increases and the height decreases simultaneously. This is due to the reduced surface tension of the spheres caused by the graphite/liquid metal galvanic cell. Then, the area decreases gradually and the height increases simultaneously. When the spheres leave the graphite, they recover their original shapes rapidly, which is attributed to their own surface tension. Therefore, the surface tension acts as the restoring force of the oscillatory movement of a liquid metal sphere.

Fig. 5 Oscillation movement of liquid metal on a graphite substrate immersed in NaOH solution. (A) Top view snapshots of liquid metal sphere 1 with a mass of 1.2 g. (B) Side view snapshots of liquid metal sphere 2 with a mass of 0.9 g. (C) Relative variations of the frontal projected area of sphere 1 and the height of sphere 2 during the oscillation process. Symbol A denotes the frontal projected area of the liquid metal viewed from the top and A₀ denotes the area when t = 0. Symbol H denotes the height of the sphere viewed from the side, and H₀ denotes the height when t = 0. The horizontal coordinate unit T represents the oscillation period of the liquid metal sphere. Scale bars are all 3 mm in length.

Fig. 6(A) presents the deformation schematic diagrams (images 1, 3 and 5) and the corresponding snapshots (images 2, 4 and 6) of liquid metal spheres in contact with graphite in NaOH solution, viewed from the top. The horizontal force diagram of the sphere is shown in image 1. As the sphere contacts the graphite, it is subjected to the capillary force F_s directed rightward and the friction force F_f directed toward the graphite. The magnitude of F_s which is caused by the surface tension gradient of the liquid metal can be expressed as:³³

where E_gl is the electric field strength induced by the graphite/liquid metal galvanic cell, V is the potential difference across the liquid metal/electrolyte EDL, and ε and λ_D are the electric permittivity and the Debye screening length of the electrolyte, respectively. The friction force F_f includes the viscous force between the sphere and the NaOH solution as well as the frictional force between the sphere and the substrate. F_s > F_f is a necessary condition for pushing the sphere off the graphite and oscillating (see image 1 and 3). r represents the curvature radius on the contact point between graphite and liquid metal, which is denoted by P. As the surface tension on the left hemisphere is larger than that on the right side, the liquid metal will flow rightward inside the sphere.⁴ As a result, r becomes smaller because the pressure originates from the surface of the sphere surrounding the point P. The relative variation of the curvature radii at the contact point P of the 0.7 g and 1.2 g spheres are presented in Fig. 6(B). The green dashed lines represent the departure of the spheres from the graphite. It can be seen that whatever the mass of the sphere, it detaches away from the graphite when the curvature radius reaches the minimum value. The ratio of the maximum value to the minimum value is 1.30 for the 0.7 g sphere (images 1 and 2), compared to 2.36 for the 1.2 g sphere (images 3 and 4). This indicates that the sphere with a larger mass has a larger deformation degree, and it shows a melon seed shape at the maximum deformation. When F_s < F_f, the capillary force imposed is too small to push the sphere away from the graphite. Also, the 1.2 g sphere will be elongated to a wormlike cylinder (see Movie S8†) without oscillation, as shown in images 5 and 6.

Fig. 6 (A) Deformation schematic diagrams and snapshots of liquid metal in contact with graphite in NaOH solution, viewed from the top. The red dashed-line circles represent the outline of the liquid metal before deformation. (B) Relative variation of the curvature radius at the contact point P between the liquid metal spheres and the graphite. Symbol R denotes the curvature radius of the contact point viewed from the top and R₀ denotes the minimum value. The horizontal coordinate unit T represents the oscillation period of the liquid metal spheres. Scale bars are all 3 mm in length.

The variation of the oscillation period–time relationship of the liquid metal sphere is investigated, and the result is shown in Fig. 7(A). It can be seen that the oscillation period becomes increasingly larger with time and presents the form of a ladder, as shown by the blue dashed lines. In the interval of each step, longer time periods occasionally appear, which indicates a slowing trend of the oscillation movement. The sphere ultimately stops oscillating, with slight but continuous deformation, after about ten minutes. However, it oscillates again after the surface layer of the graphite is scraped off. To find the reason, a comparison was made between the surface of a piece of graphite after the oscillation experiment and that of a pure one that was not used in the experiment. The scanning electron microscope (SEM) images obtained under the same conditions are presented in Fig. 7(B) and (C), respectively. It can be seen that the experimented graphite shines brightly and unevenly, which indicates the poor conductivity of the surface. Energy dispersive X-ray (EDX) studies were performed to analyze the surface element composition of the graphite after the oscillation experiment, and the result is shown in Fig. 7(D). Na element with 3.18% by weight is found on the surface, which indicates that some Na⁺ ions in the NaOH solution are adsorbed on the graphite surface. These absorbed ions may destroy the graphite/liquid metal EDL and further weaken the induced electric field strength E_gl. The driving force F_s imposed on the sphere is thus reduced to a value lower than the friction force F_f and ultimately results in termination of the oscillation.

Fig. 7 (A) Variation of the oscillation period of the liquid metal spheres with the number of times. (B) SEM image of the graphite after the liquid metal sphere oscillation experiment. (C) SEM image of the pure graphite. (D) EDX elemental analysis of the graphite after the liquid metal sphere oscillation experiment. Scale bars are all 2 μm in length.

2.4. Resonance phenomenon of two liquid metal spheres

As two liquid metal spheres are respectively laid on the top and the side of the graphite, an interesting phenomenon of resonance can be observed (see Movie S9†). The vibration of the side sphere, denoted as LM2 and having a mass of 1.0 g, can induce the vibration of the top sphere, which is denoted as LM1 and has a mass of 0.2 g. The schematic diagram and snapshot of the liquid metal double-sphere resonance oscillation experiment are respectively shown in Fig. 8(A) and (B). As the side sphere contacts the graphite, its height decreases with an increase of the top sphere. As the side sphere leaves the graphite, however, its height increases with a decrease of the top sphere. This can be seen from the height–time relationship curves of the two spheres, presented in Fig. 8(C). The curve crest of LM1 and the curve trough of LM2 correspond to each other and are marked by the green dashed lines. Also, LM2 detaches from the graphite at the time corresponding to these lines. V_gl1 and V_gl2 represent the potential differences across the graphite/LM1 and the graphite/LM2 galvanic cells, respectively. As LM2 is not in contact with the graphite, a loop circuit CL1 is formed which comprises LM1, the graphite and the electrolyte, whose equivalent resistance is R₁. The voltage imposed on LM1 is V_gl1 directed downward, which causes LM1 to spread. As LM2 contacts the graphite, a loop circuit CL2 is formed which comprises LM1, LM2, the graphite and the electrolyte, whose equivalent resistance is R₂. The voltage imposed on LM2 is V_gl1 + V_gl2 directed leftward, which actuates LM2 to spread. However, the voltage imposed on LM1 is V_gl1 − V_gl2, which is smaller than V_gl1, and LM1 regains its spherical shape.

Fig. 8 (A) Schematic and (B) image of the liquid metal double-sphere resonance oscillation experiment. Scale bar is 3 mm in length. (C) Variation of the heights of the top sphere LM1 and the side sphere LM2 with time in the oscillation state, viewed from the side.

As a non-metal conductor, graphite possesses more stable chemical properties than common metals such as copper. It cannot fuse or react with the liquid metal, thus ensuring the maintenance of the oscillation movement. In the case of copper, however, a filament will be formed, as shown in Fig. 9, when the liquid metal is detached from it (see Movie S10†). The self-actuation of liquid metal demonstrated in this paper is driven by the surface tension gradient, which is induced by the graphite/liquid metal galvanic cell. This mechanism is different from that of the traditional mercury beating heart system induced by the oxidation–reduction chemical reaction, which has been abundantly researched. Based on the actuation of the graphite/liquid metal system, oscillators with complex structures, including series–parallel connections, can be developed. In addition, graphite/liquid metal-based batteries can also be conceived with vast application prospects.

Fig. 9 The adhesion of the liquid metal to copper. Scale bar is 10 mm in length.

3 Conclusions

In summary, the deformation of liquid metal in 0.5 mol L⁻¹ NaOH solution was investigated systematically without an external electromagnetic field. When a liquid metal sphere was placed on the surface of graphite, it would spread and collapse rapidly. This is because a graphite/liquid metal galvanic cell was formed and provided the sphere with a downward directed potential, and the sphere spread because of its reduced surface tension. When a liquid metal sphere was placed against the side of the graphite, it oscillated periodically. Once the sphere was in contact with the graphite, an electric capillary force F_s was induced by the graphite/liquid metal galvanic effect and imposed on the sphere. As F_s was larger than the friction force F_f, the sphere detached from the graphite and the oscillation movement was triggered. Also, as F_s was lower than F_f, the sphere would be elongated to a wormlike cylinder without oscillating. When two liquid metal spheres were respectively laid on the top and the side of the graphite, a phenomenon of resonance was observed. The spheres oscillated in opposite directions, and the vibration of the top sphere was induced by the vibration of the side sphere. The mechanism of this self-actuation of liquid metal is quite different from that of the traditional mercury beating heat system, which is ascribed to an oxidation–reduction chemical reaction. The oscillating method proposed in this paper provides a novel idea for investigating graphite/liquid metal-based oscillators and batteries, as well as hydrogen production strategies.

Acknowledgements

The authors appreciate Dr Zhizhu He for beneficial discussion on the deformation mechanism of the liquid metal. This work is partially supported by the Dean's Research Funding of the Chinese Academy of Sciences.

References

1. Q. Zhang, Y. Zheng and J. Liu, Front. Energy, 2012, 6, 311.
2. L. Sheng, J. Zhang and J. Liu, Adv. Mater., 2014, 26, 6036.
3. J. Zhang, L. Sheng and J. Liu, Sci. Rep., 2014, 4, 7116.
4. S. Y. Tang, V. Sivan, P. Petersen, W. Zhang, P. D. Morrison, K. Kalantar-zadeh, A. Mitchell and K. Khoshmanesh, Adv. Funct. Mater., 2014, 24, 5851.
5. J. T. H. Tsai, C. M. Ho, F. C. Wang and C. T. Liang, Appl. Phys. Lett., 2009, 95, 251110.
6. G. Beni, S. Hackwood and J. L. Jackel, Appl. Phys. Lett., 1982, 40, 912.
7. A. R. Wheeler, Science, 2008, 322, 539.
8. Z. L. Wan, H. J. Zeng and A. Feinerman, Appl. Phys. Lett., 2006, 89, 201107.
9. G. Lippmann, Ann. Phys., 1873, 149, 546.
10. C. W. Kim, I. H. Yeo and W. K. Paik, Electrochim. Acta, 1996, 41, 2829.
11. M. Najdoski, V. Mirceski, V. M. Petrusevski and S. Demiri, J. Chem. Educ., 2007, 84, 1292.
12. J. L. Ocampo-Espindola, E. Ramirez-Alvarez, F. Montoya, P. Parmananda and M. Rivera, J. Solid State Electrochem., 2015, 19, 3297.
13. J. Keizer, P. A. Rock and S. W. Lin, J. Am. Chem. Soc., 1979, 101, 5637.
14. S. W. Lin, J. Keizer, P. A. Rock and H. Stenschk, Proc. Natl. Acad. Sci. U. S. A., 1974, 71, 4477.
15. E. Ramirez-Alvarez, J. L. Ocampo-Espindola, F. Montoya, F. Yousif, F. Vazquez and M. Rivera, J. Phys. Chem. A, 2014, 118, 10673.
16. D. K. Verma, H. Singh, A. Q. Contractor and P. Parmananda, J. Phys. Chem. A, 2014, 118, 4647.
17. J. Olson, C. Ursenbach, V. I. Birss and W. G. Laidlaw, J. Phys. Chem., 1989, 93, 8258.
18. S. Smolin and R. Imbihl, J. Phys. Chem., 1996, 100, 19055.
19. S. Castillo-Rojas, J. L. Gonzalez-Chaves, L. Vicente and G. Burillo, J. Phys. Chem. A, 2001, 105, 8038.
20. C. Martin, M. Alias, F. Christien, O. Crosnier, D. Belanger and T. Brousse, Adv. Mater., 2009, 21, 4735.
21. J. W. Gu, X. T. Yang, Z. Y. Lv, N. Li, C. B. Liang and Q. Y. Zhang, Int. J. Heat Mass Transfer, 2016, 92, 15.
22. J. W. Gu, N. Li, L. D. Tian, Z. Y. Lv and Q. Y. Zhang, RSC Adv., 2015, 5, 36334.
23. J. W. Gu, Z. Y. Lv, X. T. Yang, G. E. Wang and Q. Y. Zhang, Sci. Adv. Mater., 2016, 8, 972.
24. R. L. McCreery, Chem. Rev., 2008, 108, 2646.
25. T. Y. Liu, P. Sen and C. J. C. J. Kim, J. Microelectromech. Syst., 2012, 21, 443.
26. J. F. Bresse, Microchim. Acta, 2000, 132, 449.
27. M. S. Xue, W. F. Wang, F. J. Wang, J. F. Ou, C. Q. Li and W. Li, J. Alloys Compd., 2013, 577, 1.
28. S. Y. Davydov, Tech. Phys. Lett., 2009, 35, 847.
29. K. I. Ibragimov and V. A. Korol'kov, Inorg. Mater., 2001, 37, 567.
30. F. Scharmann, G. Cherkashinin, V. Breternitz, C. Knedlik, G. Hartung, T. Weber and J. A. Schaefer, Surf. Interface Anal., 2004, 36, 981.
31. Y. X. Gao and J. Liu, Appl. Phys. A: Mater. Sci. Process., 2012, 107, 701.
32. D. Jahn and H. G. Plust, Nature, 1963, 199, 806.
33. A. J. Pascall and T. M. Squires, J. Fluid Mech., 2011, 684, 163.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra12177b

---