Structural, electronic and photocatalytic properties of atomic defective BiI3 monolayers

Chemical Physics Letters 691 (2018) 341–346

Contents lists available at ScienceDirect

Chemical Physics Letters journal homepage: www.elsevier.com/locate/cplett

Research paper

Structural, electronic and photocatalytic properties of atomic defective BiI3 monolayers Huang Yan, Hu Ziyu ⇑, Gong Xu, Shao Xiaohong ⇑ College of Science, Beijing University of Chemical Technology, Beijing 100029, China

a r t i c l e

i n f o

Article history: Received 23 October 2017 In final form 20 November 2017 Available online 22 November 2017 Keywords: Defective structures BiI3 monolayer Electronic properties Photocatalytic properties

a b s t r a c t The structural, electronic and photocatalytic properties of five vacancy-containing 2D BiI3 monolayers are investigated by the first-principle calculations. The electronic structures show that the five structures are stable and have comparable binding energies to that of the pristine BiI3 monolayer, and the defects can tune the band gaps. Optical spectra indicate that the five structures retain high absorption capacity for visible light. The spin-orbit coupling (SOC) effect is found to play an important role in the band edge of defective structures, and the VBi and VBi-I3 defective BiI3 monolayers can make absolute band edges straddle water redox potentials more easily. Ó 2017 Published by Elsevier B.V.

1. Introduction Solar energy is one of the most important renewable energy resources for its wide availability and environmental friendliness. The development of novel lowcost and highly efficient solar energy conversion materials is the most important task for its real applications. The 2D materials such as graphene, boron nitride and transition metal dichalcogenides have considered as good candidates for their high power conversion efficiency and better stability [1–13]. Recently, the metal halide Bismuth tri-iodide (BiI3) compound has been investigated as gamma-ray detection primarily due to the high atomic numbers of the constituent elements and high mass density, and also has good optoelectronic properties [14–18]. BiI3 crystallizes in an ABC stacking order layered structure, in which the I–Bi–I trilayers stacked along the [0 0 1] direction in a hexagonal lattice with a large van der Waals gap [19,20]. The structural, electronic and optical properties of bulk BiI3 crystal have been reported both experimentally and theoretically [14,21,22]. Lehner et al. [16] have determined the absolute band positions of BiI3 and suggested its promising optoelectronic properties. Guided by the predictive discovery framework, Brandt et al. [17] also considered BiI3 as a candidate for thin-film photovoltaic (PV) absorbers. Very recently, Zhang et al. [23] have investigated BiI3 and its single layer theoretically, and found that the single-layer BiI3 possesses an electronic gap of about 1.63 eV and high absorption for visible light, which provide useful guidelines for the experimental synthe⇑ Corresponding authors. E-mail addresses: [email protected] (H. Ziyu), [email protected] (S. Xiaohong). https://doi.org/10.1016/j.cplett.2017.11.044 0009-2614/Ó 2017 Published by Elsevier B.V.

sis of BiI3 monolayer and facilitate their practical applications. However, defects are inevitable in synthesis or processing, even sometimes created intentionally, they can usually play an important role to modulate their properties such as tailoring various electronic and optical properties of two-dimensional materials and have been the subject of intense research over the last few decades [24–32]. The most common and energetically favorable types of impurities are vacancy defects (VDs) [27]. Despite the single-layer BiI3 has been considered as a promising candidate for future low-dimensional solar energy conversion applications, vacancy defects in BiI3 monolayer have been barely explored. In this work, five types of VDs: (i) Bi vacancy (VBi), (ii) a vacancy complex of Bi and nearby three monoiodine vacancy (VBi-I3), (iii) monoiodine vacancy (VI-1), (iv) two-monoiodine vacancy (VI-12) and (v) three-monoiodine vacancy (VI-123) in (SL) BiI3 are investigated using first-principles calculations. The structural, electronic and photocatalytic properties are performed and the results show that these five defective BiI3 monolayers can modulate the band gaps and retain good absorption for visible light. The BiI3 monolayers with Bi vacancy (VBi) and vacancy complex (VBi-I3) defects are predicted to upshift the conduction band edge and could make absolute band edges straddle water redox potentials more easily relative to the pristine monolayer BiI3.

2. Computational methods All the calculations are performed by the projector augmented plane-wave (PAW) method [33] based on density functional theory (DFT) in the Vienna Ab initio Simulation Package (VASP) [34,35].

342

H. Yan et al. / Chemical Physics Letters 691 (2018) 341–346

The exchange and correlation potential are described by generalized gradient approximation in the Perdew Burkee Ernzerhof (GGA-PBE) form [36]. To simulate the long-range van der Waals interaction, a dispersion correction of the total energy of Grimme’s pair-wise correction (DFT-D3) [37], the Tkatchenko and Scheffler van der Waals correction (vdW-TS) [38] methods and optimized Becke88 van der Waals (optB88-vdW) [39,40] functional are also employed for comparison. To evaluate the importance of the spin-orbit coupling (SOC) effect, we perform comparative calculations for BiI3 with and without SOC. The projector augmented wave (PAW) pseudopotential is treated as 6s26p3 and 5s25p5 for Bi and I, respectively. A plane-wave basis set with a cutoff energy of 500 eV is used in the calculation. All the structures are fully relaxed with a force tolerance of 0.01 eV/Å. K-Points with a mesh of 13  13  1 and 7  7  1 generated by the scheme of Monkhorst–Pack [41] are used for pristine and defective BiI3 monolayer geometry optimization and optical property calculations, respectively. To study 2D systems under the periodic boundary conditions, a vacuum layer with a thickness of at least 20 Å is inserted to avoid the interaction between periodic images, and a hexagonal supercell 6  6  1 is adopted and five types of vacancy defects are introduced. 3. Results and discussion 3.1. Structure of pristine and different types of vacancy defects (SL) BiI3 We firstly calculated the bulk BiI3 (Fig. 1) and a 6  6  1 supercell of pristine BiI3 monolayer (see Fig. 2a). The corresponding lattice constants of the fully relaxed structures are list in Table 1. As is seen, compared with the experimental results, the PBE functional significantly overestimates the lattice parameters a or b and c by about 4%, 11%, respectively, which are all consistent with the previous calculations for bulk BiI3 [23]. Due to BiI3 consists of I–Bi–I trilayers binding with weak vdW interaction, and we thus use optB88-vdW to investigate the structure of BiI3, the calculated lattice constants of the bulk BiI3 and pristine single-layer BiI3 are 7.531 Å and 7.548 Å, respectively, which are good agreement with previously calculated value [23]. In addition, the volumes and lattice parameters predicted by vdW-D3 or vdW-TS approach in PBE are closer to the experimental values [42,43]. Therefore, we can conclude that the vdW-D3 and vdW-TS approximations are suitable for BiI3. However, the calculated lattice constants of bulk BiI3 and pristine single-layer BiI3 with vdW-TS correlation is 7.628 Å and 7.635 Å, respectively, which also overestimates the lattice parameters slightly than that of the optB88-vdW method. Finally, the calculated lattice constants of the bulk BiI3 and pristine

single-layer BiI3 by DFT-D3 approach decrease from 7.838 Å to 7.518 Å and 7.812 to 7.545 Å, respectively, which are much closer to the experimental value for bulk BiI3. Subsequently, we evaluate the possible vacancies defects in BiI3 monolayers, and the optimized structures of VBi, VBi-I3, VI-1, VI-12 and VI-123 are shown in Fig. 2b–f, respectively. Fig. 2b and d are the VBi and VI-1 defective structures, which are constructed by removing one Bi atom or I atom from the pristine BiI3 monolayers, as a results, which have slight effect on the structure. Fig. 2c, e and f are the other three defective BiI3 monolayers, named VBi-I3, VI-12 and VI-123, resulting from the lack of one Bi atom with nearby three I atoms, two and three I atoms, respectively. As is seen, most of the atoms have a significant displacement, especially around the vacancies, resulting in local distorted hexagons. Furthermore, in the VI-1, VI-12 and VI-123 defective systems, the Bi–I bond length around vacancy gets longer gradually. To check the stability of these defects in BiI3 monolayers, the calculated binding energies (Eb) for all the typical vacancy defects are summarized in Table 2. Eb is defined as Eb = [nE(Bi) + mE(I) E(BiI3)]/(n + m), where E(Bi) and E(I) are the average energy per bismuth and iodine atom, respectively. n and m denote the number of bismuth atoms and iodine atoms in the pristine or defective (SL) BiI3, and E(BiI3) is the total energy of the supercell with or without defects. Since the strong SOC effect is expected in BiI3 [23,43], we also calculate the binding energies of pristine and defective (SL) BiI3 using the PBE and PBE + SOC. The energies calculated by PBE, PBE + SOC, DFT-D3 and vdW-TS methods are positive vary from 0.698 to 1.135 eV, which indicates that pristine and defective (SL) BiI3 are stable. In addition, the VBi defect is the most stable in these five defective structures. In the VI-1, VI-12 and VI-123 defective systems, with the increase of I vacancies, the binding energies decrease gradually, implies that the VI-123 defective structure is more stable than the other two monoiodine vacancy defect structures. 3.2. Electronic structures and optical properties The calculated bandgaps of the five defective structures calculated by various methods are listed in Table 3. For comparison, we first check the band gap of pristine BiI3 monolayer. The obtained indirect band gap of pristine BiI3 monolayer are 2.58, 1.58, 2.54 and 2.53 eV from PBE, PBE + SOC, DFT-D3 and vdW-TS methods, respectively. The calculated bandgap of 1.58 eV by PBE + SOC is rather closer to the experimental value of 1.67 eV [15] and the previous calculated value of 1.63 eV [23]. Due to the existence of strong spin orbit effect in BiI3 monolayer, and the SOC effect should be considered for the bandgap calculation. The results

Fig. 1. The crystal structure in hexagonal (a) side view and (b) top view of bulk BiI3.

343

H. Yan et al. / Chemical Physics Letters 691 (2018) 341–346

Fig. 2. The top side views of atomic structures of the pristine and vacancies considered in this paper, red, green and white ball represent the iodine, bismuth and vacancies atoms, respectively: (a) pristine (b) Bi vacancy (VBi), (c) vacancy complex (VBi-I3), (d) monoiodine vacancy (VI-1), (e) two-monoiodine vacancy (VI-12) and (f) three-monoiodine vacancy (VI-123). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 1 The structural parameters of bulk and single-layer BiI3, and the bulk BiI3 experimental results from Ref. [20] is also listed for comparison. Pristine

PBE

optB88-vdW

DFT-D3

vdW-TS

Exp (Ref. [20])

Bulk a or b (Å) c(Å)

7.838 23.149

7.531 20.791

7.518 20.665

7.628 21.380

7.519 20.721

Single-layer a or b (Å)

7.812

7.548

7.545

7.635

–

Table 2 The binding energies (eV) per atom of the pristine and vacancies atomic structures calculated using various methods.

Pristine Bi Bi-I3 I-1 I-12 I-123

PBE

PBE + SOC

DFT-D3

vdW-TS

0.878 0.825 0.850 0.863 0.856 0.838

1.135 1.088 1.103 1.123 1.117 1.106

0.746 0.698 0.719 0.738 0.726 0.707

0.811 0.751 0.773 0.789 0.776 0.758

Table 3 The energy gaps in the unit of eV of pristine and defective BiI3 monolayers, and the experimental results from Ref. [15] are also listed for comparison. eV

PBE

PBE + SOC

DFT-D3

vdW-TS

Exp.

Pristine Bi Bi-I3 I-1 I-12 I-123

2.58 2.55 2.30 2.18 1.11 1.55

1.58 1.57 1.50 0.65 0.66 1.12

2.54 2.49 2.29 2.14 0.89 1.55

2.53 2.37 2.21 2.53 0.83 1.57

1.67 – – – – –

show that the five defective structures from PBE + SOC are all semiconductors with band gaps of 1.57 eV (VBi), 1.50 eV (VBi-I3),0.65 eV (VI-1), 0.66 eV (VI-12) and 1.12 eV (VI-123), respectively, which are smaller than the values from the PBE, DFT-D3 and vdW-TS methods. In addition, the VBi and VBi-I3 defective structures have the bandgap of 1.57 eV and 1.50 eV, respectively, may possess an excellent performance in harvesting the visible light.

The total and the projected density of states of pristine and defective BiI3 monolayers calculated using PBE and PBE + SOC are shown in Fig. 3a and b, respectively. As is seen, all the VBM are mainly contributed by I-p states, and all the CBM are dominated by I-p and Bi-p states. When SOC is included in the calculations, all electronic states broaden a little, and the I-p states still dominate the valence band, while the conduction band states move towards to the low energy region and narrow the energy gap. For the VBi defective structure, the position of conduction band state remains unchanged and the VBM has a widening appearance near the Fermi level slightly, which indicate the VBi defect decreases the band gap energy. But for the VBi-I3 defective structure, there is a left shift of the I-p orbitals and Bi-p orbitals to the CBM, and the position of valence states remains unchanged, which also decreases the band gap energy. For VI-1, VI-12 and VI-123 defective monolayers, there are emerging obvious peaks near the Fermi level in PDOS (PBE + SOC) and were located at around 0.06 eV, 0.13 eV and 0.02 eV, respectively, which are mainly contributed by I atom vacancies . In addition, there are leftshift both of the I-p orbitals and Bi-p orbitals to the CBM and I-p states to the VBM, which narrow the band gap (PBE + SOC) from 1.58 eV to 0.65 eV, 0.66 eV and 1.12 eV, respectively. Fig. 4 shows the calculated optical absorption coefficient of pristine and defective monolayer BiI3 structures using PBE + SOC functional. As is shown, the five defective structures retain high absorption for visible light. In Fig. 4b, the isotropic optical properties is obvious between the x and y axis in the VBi defective structure, which is similar to the optical properties of pristine (Fig. 4a) BiI3 monolayer. In addition, the curve of the VBi defective structure also has a peak at the position of 0.4 eV along the x and y direc-

344

H. Yan et al. / Chemical Physics Letters 691 (2018) 341–346

Fig. 3. The total and the projected density of states of pristine and defective monolayer BiI3 structures calculated using PBE functional (a) without SOC, (b) with SOC methods.

H. Yan et al. / Chemical Physics Letters 691 (2018) 341–346

tions. Fig. 4c–f are the absorption curves of the VBi-I3, VI-1, VI-12 and VI-123 defective BiI3 monolayers, which reveal anisotropic optical properties between the x axis and y axis slightly and have a little differ from that of the pristine and VBi defective structures. And the most obvious anisotropic positions of VBi-I3, VI-1, VI-12 and VI123 structures are located at 2.5 eV, 0.9 eV, 2.7 eV and 0.5 eV. Importantly, we can determine the types of the defects in singlelayer BiI3 by the isotropic or anisotropic properties in the absorption spectra, which is useful to investigate the relationship between the synthesizing procedure and the corresponding properties. 3.3. Band edge and photocatalytic properties The schematic of the origin of CBM and VBM in pristine and defective structures depicted are shown in Fig. 5. The optical properties are dominated by the band structure, and the band edges of potential materials for photocatalytic splitting water must straddle water redox potentials. To facilitate the illustration of the schematic, the standard water redox potentials for the reduction of 4.44 eV and for the oxidation of 5.67 eV are marked with respect to the vacuum level (the black dash lines). The calculated CBMs for the VBi, VBi-I3, VI-1, VI-12 and VI-123 five defective structures

345

using the PBE functionals are 3.608 eV, 3.767 eV, 4.037 eV, 5.099 eV and 5.318 eV, respectively, and the calculated VBMs are 6.155 eV, 6.076 eV, 6.227 eV, 6.217 eV and 6.869 eV, respectively, as shown in Fig. 5a. It seems that VBi, VBi-I3 and VI-1 defective structures fulfill the previous requirements of photocatalysis. However, when the SOC is considered, the CBM position and VBM position of VBi,VBi-I3 and VI-1 obtained using PBE + SOC changed to 4.583 ( 4.654, 5.559) eV and 6.156 ( 6.155, 6.204) eV (Fig. 5b), respectively, which reveals that SOC effect plays an important role all in the band edge of pristine and vacancies structures. In addition, the VBM is energetically favorable for oxygen evolution and the CBM is insufficient to produce hydrogen. Compared with the pristine single-layer BiI3, the CBM energy ( 4.583 eV) of the VBi defective structure is higher than that of the value ( 4.660 eV) in pristine single-layer BiI3, and the CBM energy ( 4.654 eV) of the VBi-I3 defective structure is almost equal to that of the pristine’s ( 4.660 eV). It should be pointed out that pristine BiI3 monolayer is not an intrinsic photocatalyst, and the CBM energy ( 4.660 eV) is in good agreement with the previous calculated value ( 4.7 eV) [23], which is very close to the reduction potential of H+/H2 ( 4.44 eV). To introduce the VBi and VBi-I3 defects, only a very small bias potential of 0.143 eV and 0.214 eV is required to drive hydrogen evolution, which is much smaller

Fig. 4. Absorption coefficients of the pristine and defective structures calculated using PBE with SOC: (a) pristine, (b) VBi, (c) VBi-I3, (d) VI-1, (e) VI-12 and (f) VI-123, respectively.

346

H. Yan et al. / Chemical Physics Letters 691 (2018) 341–346

Fig. 5. The absolute band edge of the pristine and defective single-layer BiI3 calculated using PBE (a) and PBE + SOC (b) methods.

than the pristine systems and should be feasible in experiments. For example, a bias potential of 0.9 eV is needed for single-layer SnS2 to proceed water splitting [44–46], which is evidenced as an efficient photocatalyst for water splitting experimentally [46,47]. 4. Conclusions In the present paper, we propose five vacancy defects in the monolayer BiI3, including Bi vacancy (VBi), vacancy complex (VBiI3), monoiodine vacancy (VI-1), two-monoiodine vacancy (VI-12) and three-monoiodine vacancy (VI-123) defects by using firstprinciples calculations. The results demonstrate that all the defective structures are thermodynamically stable and the VBi defective structure is the most stable, and these five vacancy defects effectively modulate the band gap of BiI3 monolayer. Calculations of the light absorption coefficient show that isotropic optical properties in the VBi defective monolayer BiI3 and anisotropic optical properties in other four defective BiI3 monolayers along the x and y axis. Moreover, compared with the pristine structure, the VBi and VBi-I3 BiI3 defective monolayers are predicted to upshift the conduction band edge and likely to make absolute band edges straddle water redox potentials more easily. Acknowledgements

[8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30]

This work was supported by the National Natural Science Foundation of China (Grant No. 11604008), the Key Program for Higher Education of the Education Department Henan Province of China (Grant No. 17A140010), and by BUCT Fund for Disciplines Construction (Project No. XK1702). References [1] K.S. Novoselov, A.K. Geim, S.V. Morozov, D. Jiang, Y. Zhang, S.V. Dubonos, I.V. Grigorieva, A.A. Firsov, Science 306 (2004) 666. [2] K.S. Novoselov, A.K. Geim, S.V. Morozov, D. Jiang, M.I. Katsnelson, I.V. Grigorieva, S.V. Dubonos, A.A. Firsov, Nature 438 (2005) 197. [3] A.K. Geim, Science 324 (2009) 1530. [4] C.R. Dean, A.F. Young, I. Meric, C. Lee, L. Wang, S. Sorgenfrei, K. Watanabe, T. Taniguchi, P. Kim, Nat. Nanotechnol. 5 (2010) 722. [5] A.A. Balandin, Nat. Mater. 10 (2011) 569. [6] Q.H. Wang, K. Kalantarzadeh, A. Kis, J.N. Coleman, M.S. Strano, Nat. Nanotechnol. 7 (2012) 699–712. [7] D. Jariwala, V.K. Sangwan, L.J. Lauhon, T.J. Marks, M.C. Hersam, Acs Nano 8 (2014) 1102.

[31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47]

Y. Jing, J. Mater. Chem. A 2 (2014) 12104–12122. X. Tan, C.R. Cabrera, Z. Chen, J. Phys. Chem. C 118 (2014) 25836–25843. S. Zhang, Z. Yan, Y. Li, Z. Chen, H. Zeng, Angew. Chem. 54 (2015) 3112. L. Kou, Y. Ma, X. Tan, T. Frauenheim, A. Du, S. Smith, J. Phys. Chem. C 119 (2015), 150309052646007. M. Zhou, W. Duan, Y. Chen, A. Du, Nanoscale 7 (2015) 15168. L. Dou, A.B. Wong, Y. Yu, M. Lai, N. Kornienko, S.W. Eaton, A. Fu, C.G. Bischak, J. Ma, T. Ding, Science 349 (2015) 1518. H. Yorikawa, S. Muramatsu, J. Phys.: Condens. Matter 20 (2008). N.J. Podraza, W. Qiu, B.B. Hinojosa, H. Xu, M.A. Motyka, S.R. Phillpot, J.E. Baciak, S. Trolier-Mckinstry, J.C. Nino, J. Appl. Phys. 114 (2013) 2461–12127. N.J. Podraza, W. Qiu, B.B. Hinojosa, H. Xu, M.A. Motyka, S.R. Phillpot, J.E. Baciak, S. Trolier-Mckinstry, J.C. Nino, Appl. Phys. Lett. 107 (2015) 419. N.J. Podraza, W. Qiu, B.B. Hinojosa, H. Xu, M.A. Motyka, S.R. Phillpot, J.E. Baciak, S. Trolier-Mckinstry, J.C. Nino, J. Phys. Chem. Lett. 6 (2015) 4297–4302. H.S. Han, M. Hong, S.S. Gokhale, S.B. Sinnott, K. Jordan, J.E. Baciak, J.C. Nino, J. Phys. Chem. C 118 (2014) 3244–3250. M. Ruck, Z. Kristallogr. 210 (1995) 650–655. D. Nason, L. Keller, J. Cryst. Growth 156 (1995) 221–226. G.E. Jellison, J.O. Ramey, L.A. Boatner, Phys. Rev. B: Condens. Matter 59 (1999) 9718–9721. D.J. Singh, Phys. Rev. B: Condens. Matter 82 (2010) 2741–2754. W.B. Zhang, L.J. Xiang, H.B. Li, J. Mater. Chem. A 4 (2016) 19086–19094. Y. Liu, H. Xiao, Nano Lett. 16 (2016) 3335. H.P. Komsa, J. Kotakoski, S. Kurasch, O. Lehtinen, U. Kaiser, A.V. Krasheninnikov, Phys. Rev. Lett. 109 (2012) 035503. F. Banhart, J. Kotakoski, A.V. Krasheninnikov, Acs Nano 5 (2011) 26–41. J. Hong, Z. Hu, M. Probert, K. Li, D. Lv, X. Yang, L. Gu, N. Mao, Q. Feng, L. Xie, Nat. Commun. 6 (2015) 6293. Q. Fang, Y. Huang, Y. Miao, K. Xu, Y. Li, F. Ma, J. Phys. Chem. C 121 (2017). M.A. Khan, M. Erementchouk, J. Hendrickson, M.N. Leuenberger, Phys. Rev. B 95 (2017). M.A. Khan, M. Erementchouk, J. Hendrickson, M.N. Leuenberger, Appl. Phys. Lett 109 (2016) 338. M.A. Khan, M. Erementchouk, J. Hendrickson, M.N. Leuenberger, J. Mater. Chem. C 5 (2017). X. Li, L. Ma, D. Wang, X.C. Zeng, X. Wu, J. Yang, Nanoscale 8 (2016). B. Pe, Phys. Rev. B: Condens. Matter 50 (1994) 17953. G. Kresse, J. Furthmüller, Comp. Mater. Sci. 6 (1996) 15–50. G. Kresse, J. Furthmüller, Phys. Rev. B: Condens. Matter 54 (1996) 11169. J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett 77 (1996) 3865. S. Grimme, J. Comput. Chem. 27 (2006) 1787. T. Bucˇko, Phys. Rev. B: Condens. Matter 87 (2013) 1664–1667. J. Klimeš, D.R. Bowler, A. Michaelides, Phys. Rev. B 83 (2011) 772. J. Klimes, D. Bowler, A. Michaelides, J. Phys. Condens. Matter 22 (2010) 022201. H.J. Monkhorst, Phys. Rev. B: Condens. Matter 16 (2005) 1748–1749. H. Lu, J. Gao, Z. Hu, X. Shao, Rsc Adv. 6 (2016) 102724–102732. F. Ma, M. Zhou, Y. Jiao, G. Gao, Y. Gu, B. Ante, Z. Chen, A. Du, Scient. Rep. 5 (2015) 17558. H.L. Zhuang, R.G. Hennig, Phys. Rev. B 88 (2013) 3925–3938. L. Li, H. Li, J. Zhou, J. Lu, R. Qin, Z. Gao, W.N. Mei, J. Comput. Theor. Nanos 7 (2010) 2100–2103. R. Wei, T. Zhou, J. Hu, J. Li, Mater. Res. Exp. 1 (2014) 025018. Y. Sun, H. Cheng, S. Gao, Z. Sun, Q. Liu, Q. Liu, F. Lei, T. Yao, J. He, S. Wei, Angew. Chem. 51 (2012) 8727.