Electronic structure, elasticity, Debye temperature and anisotropy of cubic WO3 from first-principles calculation.

The electron structure, elastic constant, Debye temperature and anisotropy of elastic wave velocity for cubic WO3 are studied using CASTEP based on density functional theory. The optimized structure is consistent with previous work and the band gap is obtained by computing the electronic structure; the top of the valence band is not at the same point as the bottom of the conduction band, which is an indirect band-gap oxide. Electronic properties are studied from the calculation of band structure, densities of states and charge densities. The bulk and shear moduli, Young's modulus, hardness and Poisson's ratio for WO3 are studied by the elastic constants. We calculated acoustic wave velocities in different directions and estimated the Debye temperature from the acoustic velocity. The anisotropy of WO3 was analysed from the point of view of a pure wave and quasi wave.

Introduction

As a type of excellent semiconductor material, tungsten trioxide (WO3) has been widely used in multiphase catalysis, electroluminescence, photodegradation, high-temperature superconductivity and new energy fields [1-10]. The ideal cubic WO3 primitive cell is a kind of octahedral structure; W and O occupy the central and six corners of the octahedron (figure 1), respectively.

Figure 1.

Crystal structure of cubic WO3. There are four atoms in the WO3 primitive cell: three O atoms at (0, 0, 0), (0, a/2, a/2) and (a/2, 0, a/2) and one W (0, 0, a/2), where a is the lattice constant.

Tungsten trioxide experiences different structure transitions in the temperature range of −180 to 900°C; these phase transitions are not the recombination of tungsten and oxygen atoms, but the distortion of tungsten atoms in the original ideal crystal structure. The phase transition, from low to high temperature; is irreversible; a stable phase that forms at high temperature is also stable at low temperatures. As the temperature increases, the sequence of WO3 phase transitions is as follows: monoclinic (at low temperature) → triclinic → monocline (at room temperature) → orthorhombic phase → tetragonal phase. Of course, there is also hexagonal phase WO3. Although cubic WO3 has not been observed at high temperatures, in many works it is considered as a reference structure, with many reports on experimental and theoretical studies of WO3.

Since the 1970s, when Randin [11] found tungsten oxide with photochromic performance for the first time, many research works both at home and abroad have conducted a number of theoretical and experimental studies on photocatalysis, capacitance performance, photosensitivity and gas sensitivity of tungsten oxide [12-21]. Yan et al. [22] published an article on hydrogenation of WO3. They synthesized tungsten oxide single-crystal nanosheets via the exfoliation of layered tungstic acid to tungsten oxide nanosheets and subsequent introduction of oxygen vacancies (figure 2). In Taiwan, Hsieh et al. [23] studied the growth along the [001] direction of triclinic WO3 nanowires, and the optical absorption of nanowires by the red shift phenomenon. The results showed the influence of oxygen vacancy on the red shift, and the influence of crystallinity and particle size distribution of the nanowires on the forbidden bandwidth. Hao Lai et al. [24] fabricated mesoporous WO3 nanofibres and tube-like nanofibres by the gas-filled assistant sol–gel immersion method with porous anodic alumina membrane confinement, which has a porous structure (figure 3). Wang et al. [25] studied the structural and electronic properties of WO3 by various hybrid functionals combined with both plane wave and localized basis sets. Levy & Pagnier [26], through the establishment of sheet WO3 to simulate a nanoscale with the required stoichiometric ratio and oxygen-containing defects, computed the position of the atoms, electron density and the electron state density distribution, and pointed out that the WO3 surface could absorb oxygen atoms, which is due to the existence of the acceptor level at the bottom of the surface valence band. Chatten et al. [27] studied the WO3 electron structure of the oxygen vacancy model in different crystal systems and its forbidden band gap, and found that the polarizations of the bonding and anti-bonding states were not only related to the location of the crystal system but also to the position of oxygen vacancy.

Figure 2.

Schematic illustration describing the formation of tungsten oxide single crystal nanosheets [23].

Figure 3.

Gas-filled process of mesoporous tungsten oxide nanowire synthesis from triblock co-polymer within a porous anodic alumina membrane (AAM) [25].

By reviewing the previous work, it has been found that the study of tungsten oxide experimentally and theoretically has made great advancements, but systematic studies of its elastic properties, Debye temperature and anisotropy are rare. It is well known that elastic properties can be used to provide information about the potential of atoms and relate to a variety of basic solid-state phenomena. Therefore, this paper is based on density functional theory (DFT) to study the cubic WO3 electronic structure, elastic properties, Debye temperature and anisotropy in different directions.

Computational details

In recent years, the methods used for theoretical study of metal oxides have been mainly based on LDA, GGA, LDA + U and GGA + U [28-34]; Hubbard parameter U is known as the On-site Coulomb interaction energy [35]. However, a strong association effect is not considered in the calculation of LDA and GGA, that is, the unoccupied d and f orbits. Furthermore, because of the complexity of the electron cloud diffusion, the multibody effect of an orbit is difficult to accurately describe by LDA and GGA. In this system, the energy band structure of the metal is often given by LDA or GGA, and the bands across the Fermi level are often d or f orbits. In fact, the band structure of the system has semiconductor characteristics, there is a clear band gap between the bonding and the anti-bonding states, and the d or f orbit is tightly confined to the nucleus and does not show delocalization. To better describe the strong association system, it is necessary to surpass the traditional LDA or GGA approximation. In this respect, the more successful methods of improvement are LDA + U and GGA + U. Compared with the traditional DFT, the calculation of LDA(GGA) + U does not increase obviously, and the calculation results can be improved significantly with the proper parameters.

In this work, we use DFT to study electronic structure, elastic properties, Debye temperature and anisotropy of cubic WO3; the calculations have been performed by the CASTEP code [36,37]. By comparison of different approximation methods, LDA + U [38] is finally determined for structural optimization. W 5d46s2 electrons and O 2s22p4 electrons are explicitly treated as valence electrons. According to the results of the convergence (figure 4), the cut-off energy of the plane wave is taken as Ecut = 700 eV, the k-point integral in the Brillouin zone is set to 7 × 7 × 7 and the self-consistent convergence method is used to optimize the structure of cubic WO3.

Figure 4.

Convergences of the total energy of WO3 at different computational parameters. (a) Total energy versus the cut-off energy for the k-point of 4 × 4 × 4. (b) Total energy versus the k-point for the cut-off energy of 700 eV.

Results and discussion

We optimized at the LDA + U (U = 5 eV [35]) level the lattice constant of the cubic WO3 to be 3.82 Å, which agrees quite well with those of previous self-consistent calculations: a = 3.73–3.84 Å [39-41], as well as the experimental value of a = 3.71–3.75 Å [42]. Optimized results by GGA + U are larger than those of the above theoretical and experimental studies (figure 5), which indicates that LDA + U is more suitable for structural optimization than GGA + U in this work.

Figure 5.

Optimized lattice constant of WO3. U values are specified where applicable.

Electronic structure

The calculated band gap, Eg = 0.544 eV, is in line with the previous theoretical studies, Eg = 0.3–0.6 eV [43,44], which is much smaller than the experimental value of 2.62 eV (for the monoclinic phase) [45] due to the well-known DFT error. The results can be modified by using the scissors operator [46], and the band-gap width of WO3 is Eg = 2.70 eV, which does not affect our next step of calculation.

As shown in figure 6, the WO3 band gap can attain the maximum value at the R point (the top of the valence band) and the minimum value at the G point (bottom of the conduction band), so WO3 is an indirect band-gap semiconductor. The valence band consists of 12 levels, which can be divided into four bands: three bonding and one non-bonding bands of hybridized O-2s, -2p and W-5d, -6s states. Based on the crystalline field theory, it can be seen that the lowest three states form , which is mainly due to the hybridization of the O-2s orbitals with a few of W 5d(x) and 5d(z). Three states of are formed by the hybrids of W-5d(z) and 5d(x) with O-2p orbitals. The energy of is lower than that of the band, because the overlap of the d orbitals with O-2p orbitals is stronger than that of the t band, which is generated by the interaction between W-5d (d) and O-2p orbitals. Three levels near the top of the valence band form the non-bonding band; the formations are due to the interaction between the nearest O–O. In the conduction band, the energy of (anti-bonding band) is lower than that of (anti-bonding band). It is worth noting that the anti-bonding band is between and .

Figure 6.

Band structure of WO3. The red dashed line represents the Fermi energy level.

As can be seen in figure 7, the calculated density of states (DOS) of WO3 is divided into three groups. The low-valence band (i), that is, with a bandwidth of 2.93 eV, is mainly composed of O-2s, and a minority of W-5d. Group (ii), near the top of the valence band, with a bandwidth of 7.63 eV, is formed by hybridization between O-2p and W-5d, and the contribution of tiny amounts of W-5p, -6 s. Group (iii) is located in the conduction band with a bandwidth of 13.87 eV. In the bottom of the conduction band, we found that there is a mixture of dominant O-2p with W-5d orbitals, and in the upper area of group (iii), the main contribution is by hybridization between W-5p and -6s electrons. The electronegativity for W (1.7) and O (3.44) has a small difference; consequently, hybridized peaks are dispersive and intensities are not strong enough, which is also the cause of WO3 instability. Furthermore, the total charge densities of cubic WO3 is presented in figure 8; it shows that the bonds between W and O are covalent due to hybridization, which agrees well with our analysis of DOS.

Figure 7.

(a-c) The total and partial density of state for cubic WO3.

Figure 8.

Charge densities of cubic WO3.

Elastic and mechanical properties

According to Hooke's law, there are only three independent elastic constants C11, C12 and C44 for cubic crystal structures (C11 = C22 = C33, C12 = C13 = C23, C44 = C55 = C66). Here. C11 = 546, C12 = 35 and C44 = 71. Based on the Born's stability restrictions [47]: It is known that the elastic constants of the cubic WO3 are satisfied with the above stability conditions, so the cubic WO3 is stable in mechanics.

We further calculated the phonon dispersion of WO3, whose imaginary frequency can be found from figure 9, which indicates that the cubic WO3 structure is unstable, which is not a contradiction with the above equation (3.1). In cubic WO3, the obtained results by the Born's stability restrictions show that the stability is affected by the external force, that is, from the point of view of macroscopic mechanics. While the imaginary frequency of the phonon dispersion shows that the atomic arrangement is unstable at present, from the point of view of microscopic atomic structure, atoms in the WO3 will be rearranged by the lattice vibration to form a new and stable structure.

Figure 9.

Phonon dispersion diagram for WO3.

The bulk and shear modulus are two key parameters for the characterization of material hardness. The bulk modulus can be described as the ability of the material to resist the change of bulk, and is also understood as the mean value of the bond strength. The shear modulus can be described as the ability of the material to resist the shape change caused by shear force, and the ability to resist the change of the bond angle. The calculation of the bulk and shear modulus can be obtained by C, and there are two different methods to do so. One is the calculation of the strain continuity on grain boundaries by Reuss [48], and another is Voigt's [49] proposed stress continuity on the grain boundary. Hill [50] proves that the calculations of the Reuss and Voigt models are the lower and upper limit of the elastic constants, respectively, so the Hill model calculates the arithmetic mean of the results of the Reuss and the Voigt models: where

The Young's modulus E of cubic crystal material can be expressed as the corresponding coefficient: where x represents Reuss, Voigt and Hill.

The comparison of our calculated data and available theoretical studies [51-54] has been presented in table 1. It is obvious that our calculation constants are in accordance with previous work. According to the criterion of Pugh [55], BX/GX, which is defined as the shear modulus corresponding to the plastic deformation, and the bulk modulus are related to the fracture resistance. Pugh proposed that if BX/GX exceeds the critical value (1.75), the material will have ductility. In our calculation, the value is less than 1.75, which indicates that cubic WO3 has little ductility, which is also explained in the next study on Poisson's ratio.

Table 1.

According to equations (3.2)–(3.6), the bulk modulus, shear modulus and Young's modulus of WO3 are calculated.

bulk modulus			shear modulus
BV	BR	BH	GV	GR	GH
205	205	205	145	100	123
228(DFT), 225(ABOP) [51]; 224(DFT) [44]; 151 [52]; 254 [53]			123(DFT) [55]

Hence, we introduced the study of the Vickers hardness, which is an empirical formula for evaluating the hardness of the material. and where k is the Pugh ratio, which is the ratio of the shear G to the bulk B modulus, k = G/B.

Compared to other metal oxides in table 2, we can see that WO3 has greater hardness. This is due to the fact that the bulk modulus is not directly related to the hardness, and the shear modulus is more capable of characterizing the hardness of the material. Furthermore, with regard to the Poisson ratio, it reflects the strength of the covalent bond in the material. In general, the Poisson ratio 0.1 ∼ 0.28 represents the covalent property of the material, while 0.29 ∼ 0.33 represents the metal characteristics. The Poisson ratios for the above metal oxides, except for WO3 and MgO, are more than 0.29, which means that they have metal properties along with their mechanical properties, and their hardness is less than that of WO3.

Table 2.

Vickers’ hardness and Poisson's ratio for WO3 compared to the other calculated results.

	B	G	HV	υ
WO3	205	123	15.3	0.252
ZrO2 [56]	196	92	8.62	0.297
HfO2 [57]	293	103	5.85	0.342
TiO2 [58]	174	70	5.27	0.363
CaO [59]	144	88	12.4	0.266
MgO [60]	139	114	22.3	0.178

Debye temperature

Debye temperature is a very important thermodynamic parameter reflecting thermodynamic properties, which is related to many physical properties of solids, such as acoustic velocity, specific heat capacity and thermal expansion coefficient. According to the above bulk and shear modulus, we calculated wave velocity v, longitudinal wave velocity v and average wave velocity v, and further studied the Debye temperature of WO3 by equations (3.9) and (3.10). The relationship between the Debye temperature and the average wave velocity is as follows: and where h is the Planck constant, k is the Boltzmann constant, NA is the Avogadro constant, M is the molecular weight and ρ is the density (table 3).

Table 3.

Calculation of the elastic shear wave velocity vs, longitudinal wave velocity vp, average wave velocity vm and the Debye temperature of WO3.

vsX			vpX
X = V	X = R	X = H	X = V	X = R	X = H
4659	3869	4291	7722	7117	7432

The Debye temperature is calculated to be 582 K; unfortunately, there are no data available for comparison in the literature, as far as we know. To analyse the results, we only compared other functional materials, such as GaN (390), ZnO (303) and Al2O3(370) [61,62]; it can be found that WO3 has a higher Debye temperature, which shows that the interatomic binding force, melting point, hardness and thermal expansion coefficient of WO3 are larger than those of the above materials.

Anisotropy of elastic wave velocity

For cubic WO3, the propagation velocity of longitudinal and transverse waves along different crystal directions is related to the elastic constants C. In table 4, the equations relating velocities of propagation and elastic constants used for the calculations on WO3 are given.

Table 4.

Relation of velocity to elastic constants for various modes of propagation.

orientation of WO3 sample	mode of propagation	relation of velocity to elastic constants	calculated velocity
[100]	longitudinal	v_1=C_11/ρ	9040
[100]	transverse	v_2=C_44/ρ	3260
[110]	longitudinal	v_3=(C_11+C_12+2C_44)/2ρ	7356
[110]	transverse	v_4=C_44/ρ	3260
[110]	transverse	v_5=(C_11−C_12)/2ρ	6184
[111]	longitudinal	v_6=(C_11+2C_12+4C_44)/3ρ	6701
[111]	transverse	v_7=(C_11−C_12+C_44)/3ρ	5389

From table 4, we can draw a conclusion: for cubic WO3, v1 is the largest of longitudinal wave velocities; the wave is a pure longitudinal wave in the direction of [100], while in [110] and [111], waves of v3 and v6 are quasi-longitudinal waves. In a pure longitudinal wave, the direction of the particle vibration is perpendicular to the forward direction of the wave, whereas for the quasi-longitudinal wave it is not so, and the velocity component will be produced in other directions, and therefore v1> v3, v6. The velocities for v3, v4 and v5 are all different; transverse waves, which are formed by the degeneracy split into v4 and v5, have different velocities. Based on the above-mentioned reasons, we can conclude that it is anisotropy in this direction.

Conclusion

In this paper, we have performed first-principles calculations on cubic WO3, including structural parameters as well as the band structure, density of state, elastic constants, Debye temperature and the acoustic wave velocity in different directions. The calculated results show that the cubic WO3 is an indirect band-gap oxide; the valence band is mainly composed of O-2p, and the bottom of the conduction band is mainly contributed by W-5d and a few of O-2p. There is no doubt about the importance of elastic constants in this work; based on the elastic constants, the bulk, shear and Young's modulus are also further studied, which shows that the shear modulus is determined according to the stability of cubic WO3. Vickers' hardness and Poisson ratio are also investigated by obtaining the bulk and the shear modulus; there is no research on the hardness of WO3, and we can only compare it with other oxides.

The calculation of Debye temperature is further obtained by elastic constants, as 582 K, which is useful not only for the potential application of WO3 on thermoelectric and thermal resistance materials but also for the development of thermoelectric materials in future. Finally, we investigated the acoustic wave velocities in different directions. Based on the concept of pure and quasi wave, and the calculated result of wave velocity, it can be shown that cubic WO3 acoustic waves have anisotropy in the [110] direction.