Solid State Communications 142 (2007) 6–9 www.elsevier.com/locate/ssc Structural distortion of B1-structured MnO and FeO Wei-Bing Zhang a , Yong-He Deng a , Yu-Lin Hu a , Ke-Li Han b , Bi-Yu Tang a,∗ a Department of Physics, Xiangtan University, Key Laboratory of Low Dimensional Materials & Application Technology (Xiangtan University), Ministry of Education, Hunan Province 411105, China b Center for Computational Chemistry and State Laboratory of Molecular Reaction Dynamics, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023, China Received 7 November 2006; received in revised form 31 January 2007; accepted 31 January 2007 by J.R. Chelikowsky Available online 4 February 2007 Abstract The structural distortion of B1-structured MnO and FeO under high pressure is investigated using ab initio methods within GGA and GGA+U approximations. Present calculations indicated that the ordinary density-functional calculation overestimated the structural distortion of MnO and FeO under pressure. When the strong electronic correlations are included in the form of GGA + U, a reasonable description of the structural distortion under high pressure can be obtained. The overestimation of exchange interactions in ordinary DFT calculations may be the most important reason for the overestimation of structural distortion. c 2007 Elsevier Ltd. All rights reserved. PACS: 71.15.Mb; 64.30.+t; 61.82.Ms Keywords: A. Magnetically ordered materials; C. Crystal structure and symmetry; D. High pressure The high-pressure behavior of transition-metal monoxide (TMMO) with rocksalt (B1) structure such as MnO and FeO has a particular importance in condensed-matter physics and geophysics. Under high pressure, TMMO exhibits rich phenomena such as structural phase transition, insulator–metal transition and magnetic collapse. Despite their high-pressure properties having been studied extensively both experimentally and theoretically, many questions still haven’t been solved. Many high pressure experiments have indicated that FeO [1–5] and MnO [6–8] undergo a pressure-induced first-order phase transition around 70 and 90 GPa. Recently, using firstprinciples calculation and X-ray experiments, Fang et al. [9] investigated systematically the high-pressure phase stability and have confirmed that the high-pressure phase of MnO and FeO is B8 (NiAs) and inverse B8 (i B8) structure, respectively. Under relative low pressure, a type II antiferromagnetic (AFII), rocksalt phase with rhombohedral distortion along the [111] direction is observed experimentally when the temperature is below their Néel temperature. With increasing pressure, the ∗ Corresponding author. Tel.: +86 732 2371004; fax: +86 732 8292195. E-mail address: tangbiyu@xtu.edu.cn (B.-Y. Tang). c 2007 Elsevier Ltd. All rights reserved. 0038-1098/$ - see front matter doi:10.1016/j.ssc.2007.01.045 rhombohedral distortion is enhanced and the cell is more deviated from the ideal B1 structure [10,11]. Because of strong electronic correlations between 3d electrons, the theoretical studies of TMMO have long been a challenge to investigators. Using ultrasoft pseudopotential and generalized gradient approximation (UP-GGA), Fang et al. [9] have also investigated the rhombohedral distortion of B1-structured MnO and FeO under pressure. It is found that the rhombohedral distortions are overestimated under high pressure and there is a large discrepancy between GGA calculations and experiment. Other researches about the rhombohedral distortions at ambient pressure also show that the GGA overestimated the distortions [12,13]. The similar results also appeared in the case of NiO [14,15]. It seems that the overestimated rhombohedral distortion of ordinary DFT calculations is a common problem in TMMO with distorted B1 (r B1) structure, which have been proved to mainly result from the difficulty in describing the 3d electronic correlation in DFT. Our previous work [14,15] has shown that this discrepancy between theory and experiment can be corrected in the case of NiO when the strong electronic correlation is included in the form of GGA + U method [16,17]. In order to shed light on 7 W.-B. Zhang et al. / Solid State Communications 142 (2007) 6–9 experimental results and clarify the underlying mechanism, in the present study, we investigate the structural properties of B1structured FeO and MnO under high pressure based on density function theory within GGA and GGA + U methods. All calculations in this paper have been performed with Vienna Ab initio Simulation Package (VASP) [18–20], a generalized gradient approximation [21] was chosen for the exchange correlation functional, and the spin interpolation of Vosko et al. [22] was also used. The interaction between ions and valence electrons was described by the projector augmented-wave (PAW) method [23]. The Kohn–Sham equations were solved via iterative matrix diagonalization based on the minimization of the norm of the residual vector to each eigenstate and optimized charge- and spin-mixing routines [24–26]. To take into account the electronic correlation, a simple rotationally invariant DFT + U version proposed by Dudarev et al. [27–29] was used. In this method, parameters U and J represented on-site Coulomb interaction energy and exchange energy, respectively. The parameters U and J did not enter separately, only the difference U − J was meaningful. J was kept fixed to 1 eV in the present calculation. Because of increasing screening, the electronic correlation is expected to become weak and U parameter should be pressure dependent. Recently, a LDA + U approach has been implemented in the plane wave pseudopotential method [30], where U is calculated in an internally consistent way and the change of U induced by pressure and impurity can also be considered. This novel method has been very successful in describing FeO [30] and magnesiowüstite [31] under pressure. The results indicate that the high-pressure properties of strongly correlated systems depend crucially on the pressure dependence of U and the more accurate description should take into account the change of U . However, the change of U is usually very small for many systems under the considered pressure range and is expected to have a minor influence on the structural properties. So the variation of U can be neglected and U has been approximately set as a constant, as have many authors in other systems such as LaMnO3 [32] and GaN [33]. This is also a common practice until now, and a reasonable result can be given. In the present study, GGA + U method implemented in VASP code was used with U = 7 eV for MnO [27] and a U = 6 eV for FeO [34], which are taken from constrained LDA computations. The results reported in this paper were carried out on a rhombohedral antiferromagnetic supercell including two formula unit cells. A plane-wave basic set expanded in energy cutoff of 600 eV and k point sampling with a mesh of points 8× 8 × 8 generated by the scheme of Monkhorst and Pack [35] can ensure the convergence accuracy with total energies difference less than 3 meV/atom. The unit cell shape optimizations for each volume were performed using the conjugate gradient method and a Gaussian smearing with a width of 0.2 eV. Forces acting on atoms and stress tensors on unit cells were used in the optimization process. For total energy calculations, the integration over the Brillouin zone was performed using the linear tetrahedron method with Blöchl corrections [36– 38]. The calculated total energies as a function of volume were fitted to the Murnaghan equation of state (EOS) [39] to Table 1 Equation-of-state parameters and magnetic moments of MnO and FeO at ambient pressure ´ a (Å) B0 B00 Moment (µ B ) MnO GGA GGA + U Fang et al. Experiment 4.433 4.496 4.46 4.435a 152 150 157 151b ,162c 3.13 3.59 3.23 3.6,4.8 4.304 4.684 4.47 4.58d ,4.79e FeO GGA GGA + U Fang et al. Experiment 4.268 4.333 4.28 4.334f 169 173 180 142f ,180g 3.59 3.74 3.55 4.9 3.384 3.754 3.46 4.2h a b c d e f g h [40]. [41]. [42]. [43]. [44]. [45]. [46]. [47]. obtain equilibrium volume V0 , bulk modulus B0 and its pressure derivatives B00 . First, we give the equilibrium properties of FeO and MnO at ambient pressure. Recently, Cococcioni et al. [30] suggested a broken symmetry phase as ground state of FeO, which has a lower energy than the standard structure and also displays the correct pressure dependence of the rhombohedral angle. However, the validity of the new phase needs to be confirmed further both theoretically and experimentally. Thus, the present results including bulk properties and structural distortion of FeO are still based on B1 (r B1) structure. As shown in Table 1, using PAW-GGA, the calculated lattice parameter for MnO and ´ and 4.268 Å ´ which are in good agreement with FeO is 4.433 Å experimental result [40,45]. Whereas the GGA + U seems to give relatively larger results than GGA calculation, the obtained ´ and 4.333 Å, ´ respectively. One also lattice parameter is 4.496 Å can find from the table that the calculated bulk modulus and pressure derivative of bulk modulus using both methods are also quite reasonable. As shown above, GGA can successfully predict bulk properties, but the ordinary DFT calculation is known to be not powerful enough to describe the bandgaps of Mott insulators. For MnO, the GGA predicts that it is an insulator with bandgap about 1.3 eV, which is largely underestimated. For FeO, GGA even predicts that its ground state is metallic, in contrast with the experimental observation. When the strong electronic correlations are included, the correct insulator ground state can be obtained. A bandgap about 3.0 and 1.85 eV for MnO and FeO appears in GGA + U calculation, which is also similar to other calculations. The magnetic moments are also underestimated in GGA calculations. Present GGA calculations give 4.304µ B and 3.384µ B for MnO and FeO, which are much smaller than experiment. Using GGA + U, the obtained values are 4.684µ B and 3.76µ B , respectively, which also agree with other calculations [48,9]. It should be pointed out that the more proper description of the ground state needs a treatment of spin–orbit coupling. 8 W.-B. Zhang et al. / Solid State Communications 142 (2007) 6–9 The structural distortion is very important for the AFII r B1 phase of TMMO. As shown above, MnO and FeO are observed to be compressed and stretched respectively. At ambient pressure, present results show that GGA and GGA + U can predict the correct sign of the rhombohedral distortion, but GGA overestimates the rhombohedral distortion for both oxides. Using GGA, the obtained rhombohedral distortion c/a at experimental lattice parameters are 2.35244 and 2.83933 for MnO and FeO. The experimental rhombohedral distortion of MnO is 2.40984 (α = 60.72◦ ) [49]. For FeO, the value extrapolated from the experimental strain as a function of vacancy concentration suggests the value of c/a is 2.54037 (α = 58.40◦ for stoichiometric FeO) [50]. This overestimated structural distortions also appear in the Fang et al. [9] and Pask et al.’s calculations [13]. The present GGA + U calculations give the value of c/a as 2.41501 and 2.57079 respectively, which are in good agreement with experimental results. As shown above, although GGA overestimated the structural distortion of TMMO, it can give a reasonable lattice parameter and bulk modulus. The reason may be lattice parameter and bulk modulus are determined by the total energies through EOS, whereas structural distortion is decided by the magnetic exchange interactions. Energy is an integral quantity, whereas it is well known that a magnetism related quantity is not able to be described accurately by ordinary DFT calculation. Now we turn our attention to the structural distortion under high pressure. It should be noticed that MnO and FeO may undergo a pressure-induced structural phase transition around 70 GPa and 90 GPa, respectively. Above this transition pressure, the phase structure of MnO and FeO should be B8 and inverse B8 structure. In our study, the structure is supposed to keep in the B1 structure and other high pressure structure is beyond the scope of the present study. Fig. 1 shows the cell volume dependence of structural distortion of MnO. As shown in the figure, both calculations predict the correct sign of c/a under pressure and can also give the reasonable tendency of the change of distortions with decreasing volume. Consistent with experiment [10,11], the structural distortion predicted with GGA and GGA + U method increases with decreasing cell volumes. However, present PAW-GGA and Fang et al.’s UP-GGA calculations underestimated axial ratio c/a under the entire pressure region, which indicates the GGA has difficulties in describing the structural distortion. It’s very clearly seen that the GGA + U gives a more reasonable description for structural distortion under high pressure. It should be noticed that when the volume of cell is lowered to a smaller value (16.0 Å3 for GGA and 13.718 Å3 for GGA + U), the c/a shows a significant change in both GGA and GGA + U calculation which indicate there is a phase transition in MnO under high pressure. However, the continuous changes in the figure indicate that there is no phase transition occurred in the considered volume range. The calculated and experimental structural distortions of FeO under pressure are shown in Fig. 2. Similar to the case of MnO, UP-GGA and PAW-GGA cannot give a reasonable description for structural distortion. Only the GGA + U can reproduce the reasonable experimental results [49]. However, it is noticed that in the case of FeO, c/a is larger than the Fig. 1. The calculated rhombohedral distortion versus volume for AF r B1 structured MnO, and comparison with experimental results and Fang et al.’s GGA calculation (Ref. [9]). Exp. 1 and Exp. 2 come from √ Ref. [10] and Ref. [49]. The value of c/a shown in the figure is divided by 6. Fig. 2. The calculated rhombohedral distortion versus volume for AF r B1 structured FeO, and comparison with experimental results and Fang et al.’s GGA calculation. Exp. 1 and Exp. 2 come from √ Ref. [11] and Ref. [50]. The value of c/a shown in the figure is divided by 6. ideal B1 structure and increases with pressure under the entire pressure region, whereas in the case of MnO, the value of c/a is smaller than the ideal B1 structure and decreases with pressure. In fact, both MnO and FeO have a similar tendency. With increasing pressure, the c/a of both monoxides deviates from B1 structure more and more largely. Recently, Cococcioni et al. have investigated the structural distortion of FeO using GGA and LDA + U method and also found GGA overestimates the rhombohedral distortion and its pressure dependence. Based on the B1 structure, they did not give a correct distortion using their LDA + U method. And they suggested that the correct pressure dependence of the rhombohedral angle could be obtained based on a new broken symmetry phase [30]. However, using GGA + U methods, we reproduced well experimental results based on conventional B1 phase. It seems that different functional forms may play a very important role in calculations about structural distortion of FeO. The underlying mechanism of structural distortion below TN for TMMO has been discussed by many authors since the late 1950s [51,52]. It’s argued that the weak NN (nearest-neighbor) W.-B. Zhang et al. / Solid State Communications 142 (2007) 6–9 interaction plays a decisive role in the structural distortion. Base on distance dependent NN interactions, Rodbell and Owen [53] used a molecular-field approach to derive an expression for distortion, which shows the rhombohedral distortion explicitly depends on the NN interactions. Recently, Pask et al. [13] investigated the exchange constants of MnO and also find that the distortion can be understood with distance dependent NN interactions. However, their obtained exchange constant is significantly larger than experimental data. It’s also noticed that the overestimated exchange constant also appears in other DFT calculation (a detailed comparison of theoretical and experimental results about the exchange constant of MnO can be found in Ref. [48]). Our previous work about NiO [16, 17] also shows that the direct magnetic interactions are overestimated hugely in the case of ordinary DFT calculation, which leads to a large deviation of structure distortion from experimental results. Thus, the suggested mechanism for the overestimation of structural distortion is the overestimation of NN magnetic exchange interaction. When the strong electronic correlations were included in the form of GGA+U, much better description of structural distortion under the pressure range could be obtained. In summary, we have investigated the structural distortion of B1-structured MnO and FeO under high pressure using GGA and GGA + U methods. Although GGA gives a reasonable bulk property, it gives an incorrect description of the electronic structure and overestimates structural distortions of both oxides. Only when the strong electronic corrections are included, the reasonable ground state and structural distortion can be obtained. Under high pressure, the overestimated structural distortion in GGA calculation still exists. Using GGA + U, the structural distortions under high pressure are in good agreement with experiment due to the reasonable description of NN magnetic exchange interaction. These results suggest that the structural distortions are driven by magnetic exchange interaction and the strong electronic correlation plays a very important role in properties of TMMO even under high pressure. DFT + U can give a much better description of strongly correlated system even under high pressure. Acknowledgement This work is supported by the Open Project Program of Key Laboratory of Low Dimensional Materials & Application Technology (Xiangtan University), Ministry of Education (KF0504). References [1] R. Jeanloz, T.J. Ahrens, Geophys. J. R. Astron. Soc. 62 (1980) 505. [2] T. Yagi, K. Fukuoka, H. Takei, Y. Syono, Geophys. Res. Lett. 15 (1988) 816. [3] T. Yagi, T. Suzuki, S. Akimoto, J. Geophys. Res. 90 (1985) 8784. [4] H.K. Mao, et al., Phys. Earth Planet. Inter. 96 (1996) 135. 9 [5] Y.W. Fei, H.K. Mao, Science 266 (1994) 1678. [6] R. Jeanloz, A. Rudy, J. Geophys. Res. 92 (1987) 11433. [7] Y. Noguchi, K. Kusaba, K. Fukuoka, Y. Syono, Geophys. Res. Lett. 23 (1996) 1469. [8] T. Kondo, T. Yagi, Y. Syono, T. Kikegawa, O. Shimomura, Rev. High Pressure Sci. Technol. 7 (1998) 148. [9] Z. Fang, I.V. Solovyev, H. Sawada, K. Terakura, Phys. Rev. B 59 (1999) 762. [10] T. Kondo, T. Yagi, Y. Syono, T. Kikegawa, O. Shimomura, Rev. High Pressure Sci. Technol. 7 (1998) 148. [11] T. Yagi, T. Suzuki, S. Akimoto, J. Geophys. Res. 90 (1985) 8784. [12] D.G. Isaak, R.E. Cohen, M.J. Mehl, D.J. Singh, Phys. Rev. B 47 (1993) 7720. [13] J.E. Pask, D.J. Singh, I.I. Mazin, C.S. Hellberg, J. Kortus, Phys. Rev. B 64 (2001) 024403. [14] T. Eto, S. Endo, M. Imai, Y. Katayama, T. Kikegawa, Phys. Rev. B 61 (2000) 14984. [15] T. Sasaki, Phys. Rev. B 54 (1996) R9581. [16] W.-B. Zhang, Y.-L. Hu, K.-L. Han, B.-Y. Tang, Phys. Rev. B 74 (2006) 054421. [17] W.-B. Zhang, Y.-L. Hu, K.-L. Han, B.-Y. Tang, J. Phys.: Condens. Matter 18 (2006) 9691. [18] G. Kresse, J. Furthmüller, Phys. Rev. B 54 (1996) 11169. [19] G. Kresse, J. Furthmüller, Comput. Mater. Sci. 6 (1996) 15. [20] G. Kresse, D. Joubert, Phys. Rev. B 59 (1999) 1758. [21] J.P. Perdew, J.A. Chevary, S.H. Vosko, K.A. Jackson, M.R. Pederson, D.J. Singh, C. Fiolhais, Phys. Rev. B 46 (1992) 6671. [22] S.H. Vosko, L. Wilk, M. Nusair, Canad. J. Phys. 58 (1980) 1200. [23] P.E. Blöchl, Phys. Rev. B 50 (1994) 17953. [24] D.M. Wood, A. Zunger, J. Phys. A 18 (1985) 1343. [25] D.D. Johnson, Phys. Rev. B 38 (1988) 12807. [26] P. Pulay, Chem. Phys. Lett. 73 (1980) 393. [27] V.I. Anisimov, J. Zaanen, O.K. Anderson, Phys. Rev. B 44 (1991) 943. [28] S.L. Dudarev, G.A. Botton, S.Y. Savrasov, C.J. Humphreys, A.P. Sutton, Phys. Rev. B 57 (1998) 1505. [29] O. Bengone, M. Alouani, P. Blöchl, J. Hugel, Phys. Rev. B 62 (2000) 16392. [30] M. Cococcioni, S. de Gironcoli, Phys. Rev. B 71 (2005) 035105. [31] T. Tsuchiya, et al., Phys. Rev. Lett. 96 (2006) 198501. [32] G. Trimarchi, N. Binggeli, Phys. Rev. B 71 (2005) 035101. [33] C.-G. Duan, et al., Phys. Rev. Lett. 94 (2005) 237201. [34] W.E. Pickett, S.C. Erwin, E.C. Ethridge, Phys. Rev. B 58 (1998) 1201. [35] H.J. Monkhorst, J.D. Pack, Phys. Rev. B 13 (1976) 5188. [36] O. Jepsen, O.K. Anderson, Solid State Commun. 9 (1971) 1763. [37] M. Methfessel, A.T. Paxton, Phys. Rev. B 40 (1989) 3616. [38] P.E. Blöchl, O. Jepsen, O.K. Anderson, Phys. Rev. B 49 (1994) 16223. [39] F.D. Murnaghan, Proc. Natl. Acad. Sci. USA 30 (1944) 244. [40] L.F. Mattheiss, et al., Phys. Rev. B 5 (1972) 290. [41] Y. Noguchi, et al., Geophys. Res. Lett. 23 (1996) 1469. [42] R. Jeanloz, et al., J. Geophys. Res. 92 (1987) 11433. [43] A.K. Cheetham, D.A.O. Hope, Phys. Rev. B 27 (1983) 6964. [44] D.E.F. Fender, A.J. Jacobson, F.A. Wegwood, J. Chem. Phys. 48 (1968) 990. [45] C.A. McCammon, et al., Phys. Chem. Miner. 10 (1984) 106. [46] I. Jackson, et al., J. Geophys. Res. 95 (1990) 21671. [47] P.D. Battle, A.K. Cheetham, J. Phys. C 12 (1979) 337. [48] C. Franchini, V. Bayer, R. Podloucky, J. Paier, G. Kresse, Phys. Rev. B 72 (2005) 045132. [49] H. Shaked, J. Faber Jr., R.L. Hitterman, Phys. Rev. B 38 (1988) 11901. [50] B.T.M. Willis, H.P. Rooksby, Acta Cryst. 6 (1953) 827. [51] J. Kanamori, Progr. Theoret. Phys. 17 (1957) 197. [52] S. Greenwald, J.S. Smart, Nature 166 (1950) 523. [53] D.S. Rodbell, J. Owen, J. Appl. Phys. 35 (1964) 1002.