New method for preparing graphene by peeling graphite and facile fabrication of bulk Bi_0.45Sb_1.55Te_3.02/graphene composites with dense texture and high ZT

Abstract

We report a new method for peeling graphite to graphene, with which we develop a facile procedure for the fabrication of bulk Bi_0.45Sb_1.55Te_3.02/graphene by pushing thin graphite foils into pressed Bi_0.45Sb_1.55Te_3.02 powders and then foliated into graphene under pressure and high direct current (up to 1000 A). The pushing force results from the huge repulsive Coulomb force between the layers in the graphite foil. The Coulomb force arises from electron agglomeration as a result of the Lorentz force that the large direct-current-produced magnetic field applies on the moving electrons in the graphite foil. The incorporated graphene sheets act as growth templates for the Bi_0.45Sb_1.55Te_3.02 grains. Consequently, fast grain growth and a densely textured microstructure with laminates were observed in the Bi_0.45Sb_1.55Te_3.02/graphene bulk. By combining direct current and applied pressure, anisotropic textures were obtained, with the laminates oriented mostly along the axial direction. Since the laminates can filter low-energy carriers and scatter long-distance phonons, an increased Seebeck coefficient, decreased thermal conductivity, and a consequential 25% enhancement in the figure of merit, ZT (1.40 at 90.9 °C), were observed in the direction along which pressure was applied. This work suggests that graphene can be utilized as a template to rapidly grow single crystals of materials with similar crystal structures, as well as to adjust the textures of materials. This facile method is expected to be applicable in the fabrication of bulk semiconductor or graphene/alloy composites.

---

Introduction

The unique physical and mechanical properties of graphene (GN), such as its surprisingly high electrical mobility (10⁴ cm² V⁻¹ s⁻¹),^1–3 large theoretical specific area (2630 m² g⁻¹),⁴ high thermal conductivity (3000–5000 W m⁻¹ K⁻¹),⁵ excellent optical transparency,⁶ and high Young's modulus (1 T Pa)⁷ have excited researchers across a wide spectrum of scientific disciplines. Extensive experiments have indicated that incorporating GN can enhance the intrinsic properties of a host material and impart novel properties.^8–18 The preparation of GN-incorporated composites, especially bulk materials, is generally very complicated, expensive, and low-yielding.^8–18 In most cases, GN is independently prepared and then mixed with the host material. Consequently, the complicated preparation procedures for GN result in high costs for the GN-incorporated composites.¹⁹ Furthermore, the mixing treatment can easily lead to GN agglomeration and loss of its unique properties. An exception is the incorporation of few-layer graphene (FLG) into paraffin to prepare FLG/wax composites by the split–press–merge approach.²⁰ This method has provided a feasible route for the large-scale production of some organic/GN composites.

In this work, we report a new facile method for preparing GN at low cost by peeling graphite foil (GF), as well as a one-step approach for incorporating GN into the alloy Bi_0.45Sb_1.55Te_3.02 (BST), which has potential applications in cooling.²¹ The incorporation of GN into BST was found to have a remarkable effect on the grain growth rate of the as-prepared columnar bulk materials. The incorporated GN sheets act as growth templates for the BST grains, as previously observed in thin films.²² The number of nucleation sites was enhanced, while the energy of BST grain formation was decreased by the GN template. Consequently, the grain growth was very fast and a dense composite texture was formed in 8 min. Combined with the effect of applied pressure, anisotropic textures were produced, most of which have laminates along the axial direction. Since the laminates can filter low-energy carriers and scatter long-distance phonons, the Seebeck coefficient (S) increased and the thermal conductivity (k) decreased, affording a 25% enhancement in the figure of merit (ZT) (1.40 at 90.9 °C) in the axial direction. This work suggests that GN can be utilized to adjust the texture of materials. This facile method is also expected to be applicable for the ready fabrication of solid-state GN-incorporated composites on large scale, as well as for readily preparing GN sheets by peeling graphite sheets.

Experimental section

Materials

Bismuth (99.5%), antimony (99.5%), and tellurium (99.99%) were obtained from Alfa Aesar, and used without further purification. GF (0.5 mm thickness, 99%) was purchased from Zhuzhou Chenxin Induction Equipment Company, Hunan Province, China.

Preparation

BST/GN bulk composites were fabricated as follows. Bismuth, antimony, and tellurium were mixed and ball-milled with a high-energy ball mill machine (SPEX 8000M Mixer/Mill, Metuchen, NJ, USA) for 20 h at 900 rpm. The ball-milled powder was placed in the columnar chamber of the graphite die, sandwiched on either side by two stacks of GF, and enclosed by two graphite bars after two packs of GF were placed between the graphite bars and the ball-milled powder, as the GF sources. Pressure and direct current were applied through the graphite bars. Although the pressure was kept at 80 MPa, the applied current was varied. For the first 2 min, a 100 A direct current was applied, which was then homogeneously increased to 1000 A over 5 min. When the temperature reached 480 °C, the direct current was decreased to an appropriate value to hold this temperature for 1 min. Then, the applied direct current and pressure were withdrawn. The as-prepared sample was cooled to room temperature over ∼10 min by circulating cooling water. During the first 2 min of the pressure and current treatment, the GF was exfoliated into smaller, thinner pieces, and pushed into the pressed BST powder by the Coulomb forces between the 2D carbon layers. As the current increased, the foliated GF pieces were further transformed into GN. The 2D GN can be stabilized through chemically bonding to the BST system after sintering at elevated temperatures. Thus, a stable BST/GN composite was fabricated. This preparation method is called the “press-and-direct current” route. Pure bulk BST was prepared by pressing the aforementioned ball-milled powder using the same preparation parameters, but without inserting the two GF stacks.

Verification of GN and characterization

X-ray diffraction (XRD; X'Pert PRO MPD, PANalytical, The Netherlands) was employed to verify the presence of the Bi_0.45Sb_1.55Te₃ and GN phases in the as-prepared samples. High-resolution transmission electron microscopy (HRTEM, JEM-2100, JEOL, Japan) was performed on the as-prepared bulk samples to study their microstructures and the interplanar distance. In addition, the elemental compositions of the samples were examined by energy dispersive X-ray spectrometry (EDS) during HRTEM investigation. The HRTEM specimens were prepared by dicing, polishing, and ion-milling the bulk samples. Micro-Raman spectroscopy (Jobin Yvon LabRAM HR800, HORIBA, Jobin Yvon Company, France) measurements (including Raman mapping) were performed to confirm the presence of GN in the as-prepared samples after being polished by abrasive paper. Freshly fractured cross-sections were observed by scanning electron microscopy (SEM; Zeiss Ultra 55, Carl Zeiss, Germany). Disks (12.5 mm in diameter and 2 mm in thickness) and bars (about 2 mm × 2 mm × 12 mm) were cut along both the radial and axial directions of the columnar samples. The disks and bars were also polished for electrical conductivity (σ), S (ZEM-3, Ulvac-Riko, Japan), and k (LFA457, Netzsch, Germany) measurements. ZT values in the radial and axial directions (ZT_‖ and ZT_⊥) were then calculated using the values of the properties measured along the corresponding direction.

Results and discussion

New method for preparing GN by peeling GF

The new method for peeling GF into GN is schematically shown in Fig. 1. A graphite die with a 12.5 mm inner diameter is loaded with a stack of GF foils (with a height less than that of the die) sandwiched between two graphite bars (to clearly show the foil, the graphite bars are not drawn in Fig. 1). When a direct current of 100 A is applied to the GF (Fig. 1a and 2a) in the die through the graphite bars, a static magnetic field is produced, as shown in Fig. 1b. Here, “×” and “●” indicate the directions of the entering and exiting magnetic fields, respectively. With this magnetic field, the moving electrons (opposite to the current direction) in the GF are subjected to the Lorentz force and are transferred away from the axis, leaving the area around the axis of the die positively charged (“+” and “−” indicate the positive and negative charges, respectively). Therefore, identical charges become locally concentrated between the GF layers in the foil, and as a result of their close proximity, a very high repulsive Coulomb force develops. As a result, the foils are peeled into smaller and thinner pieces (Fig. 2b). When the GF stacks are placed in the columnar space of the die (Fig. 1a) in the “GF bar/GFs/pressed BST powder/GFs/GF bar” configuration, these pieces are pushed into the pressed BST powder. When the direct current is increased to about 300 A, the GFs begin to split into GN sheets. During this process (hereafter called the “press-and-direct current process”), the temperature is also increased, and the peeled GN sheets become chemically well combined with the BST. In this work, a bulk GN-incorporated BST composite was fabricated using the route described above with four GFs, two for each end of the pressed BST powder.

Fig. 1 Schematic illustrations of (a) GF stack inside the columnar space of the graphite die, and (b) the magnetic field produced by the direct current and the transferred electrons in the pressed GF. “×” and “●” indicate the entering and exiting magnetic fields, respectively, whereas “+” and “−” indicate the positive and negative charges, respectively.

Fig. 2 (a) GFs before being pressed together with the applied 100 A direct current, and (b) GF particles obtained from the GF die (foliated GF powder) after application of 80 MPa pressure and 100 A current.

Verification of GN in the as-prepared bulk hybrid material

The existence of GN in the as-prepared bulk material was first proven by XRD measurements. Fig. 3 shows the XRD patterns of the as-prepared bulk BST/GN, the direct-current-foliated GF powder (Fig. 2b), and the starting GF samples (Fig. 1a), from the bottom to the top panel, respectively. The XRD patterns indicate that the foliated GF powder is composed of almost 98% GF with approximately 2% GN, whereas in the as-prepared bulk sample, only GN and BST phases (no GF phase) are found. This suggests that the GFs pushed into the BST powder are totally exfoliated into GNs under the larger direct currents such as 300 A.

Fig. 3 XRD patterns of the as-prepared BST/GN bulk sample, foliated GF powder, and starting GF, from the bottom panel to the top, respectively.

The presence of GN in the as-prepared bulk sample was further directly verified by HRTEM and EDS (Fig. 4). Fig. 4a and b present the low-resolution TEM images of a GN nanosheet embedded with BST particles. Fig. 4a shows that GN exhibits a scrolled topography²³ which is obviously different from the BST system.²¹ This scrolling morphology arises from the thermodynamic instability of the 2D membrane. The HRTEM (Fig. 4c) and EDS (Fig. 4d) measurements were conducted in the field indicated by the green circle in Fig. 4b. A red curve circling field B in Fig. 4c indicates a GN sheet with a carbon sp² plane (C–sp², (002)) at the view plane. The distance between the hexagonal carbon cells (6 C cell) estimated from Fig. 4c is 2.25 Å, which is close to the value calculated using the most accepted C–C bond length of GN, 1.43 Å.^24,25 Fig. 4d presents the elemental composition measured by EDS. Excluding some impurities mainly brought during preparation, the sample is composed of 6.79% C in addition to Bi, Sb, and Te. The interplanar distance of field A is 0.235 nm, indicating that the view plane in field A is the face (101̲0) of BST. It should be noted that there are traces of parallel lines with identical distances under the GN sheets in field B. The lines have a distance of 0.942 nm, which is close to that of the face (003) of BST (PDF# 82-0358). In other words, the material under the GN in area B is probably BST. The lattice arrays of BST along the 〈003〉 direction share the lattice arrays on the C–sp² plane of GN. Thus, GN is the growth template for BST. In addition to the close lattice dimensions, the fact that the lattice structures of both GN and BST are identical (hexagonal) allows GN to be a growth template for BST.

Fig. 4 (a) HRTEM of GN in the as-prepared BST/GN slice, indicating the scrolling topography of GN. (b) HRTEM image of a GN sheet embedded with the bulk BST composite. (c) HRTEM observation and (d) EDS spectrum acquired from within the green circle shown in (b).

In the centre of field B, the colour is more intense, probably because there are more GN layers or graphite sheets in this area. However, from the Raman results, it can be seen that there are few graphite sheets in the as-prepared samples.

The third piece of evidence indicating the presence of GN in the as-prepared samples is the existence of GN vibrational modes (Fig. 5a). In addition to the Raman modes of BST (114.4 cm⁻¹, 134.9 cm⁻¹, 248.9 cm⁻¹, 442.5 cm⁻¹),^26–28 the Raman activated modes of GN (1351.3 cm⁻¹ (D-band), 1578.2 cm⁻¹ (G-band), and 2645.3 cm⁻¹ (2D-band)) were observed for the as-prepared samples (Fig. 5a).^29,30 The D-band is associated with sp³ defects in the sp² lattice, whereas the G-band is related to the pristine sp² GN lattice. The D-band, G-band, and 2D-band signals are very weak in single measurements. To obtain clear D-band and G-band signals, the measurements were repeated 10 times and the data was combined into a single curve (red curve in the inset of Fig. 5a). The green curve in the inset of Fig. 5a corresponds to the vibration modes of the starting graphite, which confirms that the modes at 1351.3 and 1578.2 cm⁻¹ should not be attributed to graphite, but rather, to GN, because the vibration modes of the starting graphite are at lower frequency (1344.0 and 1574.0 cm⁻¹) in Fig. 5a.

Fig. 5 (a) Raman spectrum of GN/BST bulk. The red curve in the inset is a cumulative (10 measurements) Raman spectrum of the as-prepared bulk sample. The green curve is the Raman spectrum of the starting graphite. (b) Regions A, B, C, and D with dimensions of 10 μm × 10 μm used for Raman mapping. (c) Raman mapping in terms of prominent Raman band intensity (G peak) and (d) resulting I_D/I_G ratios of graphene in regions A, B, C, and D.

To demonstrate the quality and uniformity of the graphene sheets, Raman mapping was performed (Fig. 5b–d). Fig. 5b displays the Raman mapping regions A, B, C, and D with dimensions of 10 μm × 10 μm. Fig. 5c and d present the Raman maps and resulting D-(in terms of prominent Raman band intensities (G peak, around 1578 cm⁻¹)) and G-band intensity ratios (I_D/I_G) for the graphene sheets in regions A–D (as shown in Fig. 5b). From Fig. 5c, it can be seen that the graphene in the as-prepared samples is not particularly uniform. Fig. 5d demonstrates that all the I_D/I_G ratios are between 0.94 and 1.12, suggesting that the layer numbers of the graphene in the A–D regions are generally greater than 1 and smaller than 4.³¹

The effects of incorporated GN on the growth of BST

The SEM images of the BST/GN show that the incorporation of GN results in a homogeneous texture and dense structure (Fig. 6a and b). In comparison, the BST sample prepared under the same processing conditions exhibits an inhomogeneous structure (Fig. 6c and d). Nearly all of the microstructure of the BST/GN composite is textured, whereas the pure BST has little texture. We think that the texture in the BST/GN is derived from the incorporated GN sheets. The lattices of both BST and GN are hexagonal. Consequently, GN can act as a growth template during BST crystallisation. The incorporation of GN increases the number of nucleation sites for BST crystal formation. The grains can grow ubiquitously, and stop growing when they meet. During the sintering period, the mean grain size of the BST/GN grew to ∼22 μm; over the same period, the bulk BST grains grew to ∼4 μm, except for several grains that were 22 μm in size. We believe that the impurities in BST assist grain growth, affording texture. It is important to note that the GN significantly accelerates the grain growth. This suggests that GN can be used as a growth template to shorten the growth period for BST single crystals or other materials that are structurally similar to GN. On the other hand, in both BST and BST/GN, the textures are anisotropic and most of the textures have laminates along the axial direction of the columnar samples (Fig. 6b). As reported by other researchers, the pressure applied during preparation tends to produce an anisotropic texture,^32,33 which, in turn, would produce anisotropic thermoelectric properties.

Fig. 6 (a) Low- and (b) high-magnification SEM images of the fractured section of the axial plane of the as-prepared columnar sample (bulk BST/GN), respectively. (c) Low- and (d) high-magnification SEM images of the fractured section of the axial plane of the as-prepared columnar sample (bulk BST), respectively.

The effects of incorporated GN on the thermoelectric properties of BST

A comparison of the thermoelectric properties of BST/GN and BST prepared by identical procedures is presented in Fig. 7, to clarify the effects of the incorporated GN and textures on the thermoelectric properties. The incorporated GN and textures have almost no obvious effect on the σ values for BST in both either the axial (hereafter indicated by the subscript “⊥”) and or radial (hereafter indicated by the subscript “‖”) directions of the as-prepared columnar samples (bulk BST and BST/GN). However, highly favourable effects are observed for the values of S and k. With the incorporation of GN and the dense texture, the S_‖and S_⊥ of BST were increased by 6.65 and 2.96%, respectively, whereas k_‖ and k_⊥ were favourably decreased by 8.09 and 2.78%, respectively. Consequently, ZT_‖ and ZT_⊥ were enhanced by 21.74 and 12.5%, respectively. The maximum ZT_‖ of BST/GN was increased from 1.15 to 1.40 at 90.9 °C. Fig. 7 also indicates that T_m, which is the temperature at which ZT reaches its peak value, shifts to lower temperatures. The peak ZT_‖of this micron-sized BST/GN has a comparable value to that obtained in a nano-structured BST system.²¹ Therefore, it can be concluded that the layered texture is preferred for high ZT.

Fig. 7 Temperature dependence of (a) k, (b) σ, (c) S, and (d) calculated ZT values along the radial direction of the columnar sample (indicated by the subscript “⊥”) and along the axial direction of the columnar sample (indicated by the subscript “‖”).

From the foregoing results, it is can be concluded that the incorporation of GN has a favourable effect on S and k and, consequently, ZT. The probable reason for the improvement of these parameters is the change in the texture of BST/GN that results from the incorporation of GN. The preparation processes for BST and BST/GN were identical. The only significant differences between the BST and BST/GN samples are the presence of GN and the texture induced by GN in the latter. BST/GN consists of layers whose interfaces are good low-energy carrier filters and long-distance phonon scatterers.^34,35 The difference in the magnitude of the thermoelectric properties between the radial and axial directions of the as-prepared samples is in good agreement with the anisotropy of the texture of BST/GN. In the axial direction, there are more layers to filter low-energy carriers and scatter long-distance phonons, resulting in higher S and lower k values. Therefore, the thermoelectric properties are more favourable in the axial direction.

The results presented in this study demonstrate a new facile route for preparing GN sheets by peeling graphite, as well as a simple method for fabricating solid-state GN-incorporated composites on large scale.

Conclusions

In summary, this paper presents a new method for preparing GN by peeling graphite. Using this method, we have developed a facile procedure for the fabrication of bulk BST/GN by pushing thin GF into pressed BST powders and then foliating the GF into GN under pressure and high direct current (up to 1000 A). The incorporated GN significantly accelerated the rate of grain growth and increased the number of nuclei during BST crystallisation. The fast grain growth resulted in a dense-textured microstructure with laminates. When pressure was applied along with direct current, the texture of the as-prepared columnar sample appeared to be anisotropic, with more laminates along the axial direction. Since the laminates can filter lower-energy carriers and scatter long-distance phonons, we observed an increased S, decreased k, and consequently, a 25% enhancement in ZT (1.40 at 90.9 °C) in the axial direction. We propose that GN can be used as a template to reduce the time required for single crystal growth in the BST system or other materials that have structures similar to GN. GN can also be used to adjust the textures of appropriate materials. In addition, the press-and-direct current method can be extended to prepare GN-incorporated composites.

Acknowledgements

This work is supported by the National Natural Science Foundation of China (51172078 and 51372092), and the Guangzhou Science and Technology Project of China (2013J4100045).

Notes and references

1. K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, Y. Zhang, S. V. Dubonos, I. V. Grigorieva and A. A. Firsov, Science, 2004, 306, 666.
2. L. Ponomarenko, F. Sehedin, M. Katsnelson, R. Yang, E. Hill and K. Novoselov, Science, 2008, 320, 356–358.
3. Y. B. Zhang, V. W. Brar, C. Girit, A. Zettl and M. F. Crommie, Nat. Phys., 2009, 5, 722.
4. M. D. Stoller, S. Park, Y. W. Zhu, J. An and R. S. Ruoff, Nano Lett., 2008, 8, 3498.
5. A. A. Balandin, S. Ghosh, W. Z. Bao, I. Calizo, D. Teweldebrhan, F. Miao and C. N. Lau, Nano Lett., 2008, 8, 902.
6. R. R. Nair, P. Blake, A. N. Grigorenko, K. S. Novoselov, T. J. Booth, T. Stauber, N. M. R. Peres and A. K. Geim, Science, 2008, 320, 1308.
7. C. Lee, X. D. Wei, J. W. Kysar and J. Hone, Science, 2008, 321, 385.
8. K. Chu and C. C. Jia, Phys. Status Solidi A, 2014, 211, 184.
9. O. R. Jankovsky, P. Simek, D. Sedmidubsky, S. Huber, M. Pumera and Z. Sofer, RSC Adv., 2014, 4, 7418.
10. K. W. Shu, C. Y. Wang, M. Wang, C. Zhao and G. G. Wallace, J. Mater. Chem. A, 2013, 2, 1325.
11. B. J. Wang, H. Y. Li, L. Z. Li, P. Chen, Z. B. Wang and Q. Gu, Compos. Sci. Technol., 2013, 89, 180.
12. T. L. Chen, D. S. Ghosh, V. Mkhitaryan and V. Pruneri, ACS Appl. Mater. Interfaces, 2013, 5, 11756.
13. S. E. Lee, H. J. Kim, H. Kim, J. H. Park and D. G. Choi, Nanoscale, 2013, 5, 8986.
14. L. Kvetková, A. Duszová, M. Kašiarová, F. Dorčáková, J. Dusza and C. Balázsi, J. Eur. Ceram. Soc., 2013, 33, 2299.
15. Y. Du, K. F. Cai and S. Z. Shen, Funct. Mater. Lett., 2013, 6, 1340002.
16. S. Mitra, A. Singha and D. Chakravorty, J. Phys. D: Appl. Phys., 2013, 46, 375306.
17. D. R. Wang, T. Zhou, J. W. Zha, J. Zhao, C. Y. Shi and Z. M. Dang, J. Mater. Chem. A, 2013, 1, 6162.
18. W. L. Song, M. S. Cao, M. M. Lu, J. Yang, H. F. Ju, Z. L. Hou, J. Liu, J. Yuan and L. Z. Fan, Nanotechnology, 2013, 24, 115708.
19. L. Yang, S. L. Zhang, Z. Chen, Y. L. Guo, J. S. Luan, Z. Geng and G. B. Wang, J. Mater. Sci., 2014, 49, 2372.
20. S. Wang, M. G. Bee, K. M. Kiran, Q. L. Bao, P. Yang and K. P. Loh, ACS Nano, 2010, 4, 6180–6182.
21. B. Poudel, Q. Hao, Y. Ma, Y. C. Lan, A. Minnich, B. Yu, X. Yan, D. Z. Wang, A. Muto, D. Vashaee, X. Y. Chen, J. M. Liu, M. S. Dresselhaus, G. Chen and Z. F. Ren, Science, 2008, 320, 634–638.
22. A. K. Geim and K. S. Novoselov, Nat. Mater., 2007, 6, 183.
23. X. Huang, X. Z. Zhou, S. X. Wu, Y. Y. Wei, X. Y. Qi, J. Zhang, F. Boey and H. Zhang, Small, 2010, 6, 513.
24. G. X. Wang, J. Yang, J. Park, X. L. Gou, B. Wang, H. Liu and J. Yao, J. Phys. Chem. C, 2008, 112, 8192.
25. J. Ma, D. Alfe, A. Michaelides and E. G. Wang, Phys. Rev. B: Condens. Matter Mater. Phys., 2009, 80, 033407.
26. K. M. F. Shahil, M. Z. Hossain, V. Goyal and A. Balandin, J. Appl. Phys., 2012, 111, 054305.
27. L. M. Goncalves, C. Couto, P. Alpuim, A. G. Rolo, F. Volklein and J. H. Correia, Thin Solid Films, 2010, 518, 2816.
28. P. Fan, Z. H. Zheng, G. X. Liang, X. M. Cai and D. P. Zhang, Chin. Phys. Lett., 2010, 27(8), 087201.
29. M. Sun, H. J. Liu, Y. Liu, J. H. Qu and J. H. Li, Nanoscale, 2014, 7, 1250–1269.
30. Z. Sofer, O. Jankovský, P. Simek, K. Klímova, A. Mackova and M. Pumera, ACS Nano, 2014, 8(7), 7106.
31. H. J. Park, J. Meyer, S. Roth and V. S. kalova, Carbon, 2010, 48, 1088.
32. P. Srivastava and K. Singh, J. Electron. Mater., 2013, 42, 2733–2738.
33. D. Vasilevskiy, M. S. Dawood, J.-P. Masse, S. Turenne and A. Masut, J. Electron. Mater., 2010, 39, 1890.
34. B. Zhong, Y. Zhang, W. Q. Li, Z. R. Chen, J. Y. Cui, W. Li, Y. D. Xie, Q. Hao and Q. Y. He, Appl. Phys. Lett., 2014, 105, 123902.
35. M. Bachmann, M. Czerner and C. Heiliger, Phys. Rev. B: Condens. Matter Mater. Phys., 2012, 86, 115320.

---