Ab Initio Prediction of Elastic and Thermal Properties of Cubic TiO2

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Computational Materials Science 50 (2011) 2142–2147

Contents lists available at ScienceDirect

Computational Materials Science


journal homepage: www.elsevier.com/locate/commatsci

Ab initio prediction of elastic and thermal properties of cubic TiO2


R. Miloua ⇑, Z. Kebbab, N. Benramdane, M. Khadraoui, F. Chiker
Laboratoire d’Elaboration et de Caractérisation des Matériaux Faculté des Sciences de l’Ingénieur, BP 89, Université Djillali Liabès 22000 Sidi Bel Abbès, Algeria

a r t i c l e i n f o a b s t r a c t

Article history: In this paper we focus on the novel solar material, namely cubic TiO2. The full potential linearized aug-
Received 30 March 2010 mented plane wave method in combination with the local density approximation (LDA) and the general-
Received in revised form 15 February 2011 ized gradient approximation (GGA) have been used. We calculated structural parameters, elastic
Accepted 18 February 2011
constants, wave velocities and thermal properties of the material assuming the fluorite structure. The
Available online 15 March 2011
obtained values were in good agreement with the available theoretical and experimental data. Moreover,
the pressure and temperature dependences of the bulk modulus, Debye temperature, Heat capacity and
Keywords:
linear expansion coefficient have been addressed for the first time.
Cubic TiO2
FPLAPW
Ó 2011 Elsevier B.V. All rights reserved.
Elastic constants
Thermal properties

1. Introduction atoms (dTi–O = 2.06 Å), while each oxygen is tetrahedrally sur-
rounded by Ti atoms. The space group is the Fm3 m (2 2 5) with
In the last few years, titanium dioxide has been the subject of titanium occupying the (0, 0, 0) position and the two oxygen atoms
intensive investigations in both the theoretical and experimental at ±(1=4 , 1=4 , 1=4 ), as illustrated in Fig. 1.
sides due to its unique properties like excellent photoactivity, The present paper deals with a systematic ab initio investiga-
long-term stability, low cost and non-toxicity. TiO2 found potential tion of the structural, elastic and thermodynamic properties of
applications in many products and processes such as pigments, f-TiO2 within the FP-LAPW method and the quasi-harmonic Debye
dye-sensitized solar cells, air purification, anti-fogging and self- theory. In the following section we describe the calculation ap-
cleaning films and hydrogen production by water splitting [1–3]. proach; the next section contains the obtained results and the
TiO2 exhibits a rich phase diagram including more than seven discussion. Section 4 gives the conclusion.
different polymorphs. The most abundant forms, namely rutile
and anatase are commonly used as antireflection coatings for solar
2. Method of calculation
cells. The cotunnite-TiO2 structure has been identified as an ultra-
hard phase, characterized by a large bulk modulus of 431 GPa [4].
We performed ground state calculations by means of the full
Using first-principles approach, Mattesini et al. [5] presented theo-
potential linearized augmented plane wave method (FP-LAPW)
retical arguments on the possibility to use cubic TiO2 (c-TiO2) as
[8,9]. We used both the local density approximation (LDA) [10]
light absorber, since the calculated dielectric function exhibits en-
and the generalized gradient approximation (GGA) [11] to the ex-
hanced absorption in the visible range.
change correlation potential. A certain care has been taken in order
Experimentally, the c-TiO2 was synthesized at pressure of
to converge all properties. The wave functions in the interstitial re-
48 GPa and temperature of 1900–2100 K by laser-heating anatase
gion were expanded in plane waves with a cut-off parameter of
in a diamond-anvil cell [6]. Unfortunately, it was difficult to iden-
Kmax = 9/RMT. RMT values are assumed to be 2.0 a.u. for Ti and 1.6
tify the exact cubic structure among the fluorite (CaF2) and pyrite
a.u. for O atoms. In order to calculate the thermal response of the
(FeS2) types from X-ray diffraction patterns. Recent ab initio pho-
material, we used the quasi-harmonic Debye approach as imple-
non calculations indicated that the fluorite TiO2 stabilizes under
mented in the GIBBS code [12].
pressure, whereas the pyrite TiO2 shows instability throughout
the whole pressure range of 0–114 GPa [7]. Hence, it is supported
that the fluorite structure is the exact experimentally observed cu- 3. Results and discussion
bic phase. In this structure each Ti atom is eight coordinated with O
3.1. Structural properties

⇑ Corresponding author. Tel.: +213 550 716 156. First, the lattice parameter, bulk modulus and its pressure
E-mail address: [email protected] (R. Miloua). derivative have been obtained by fitting the total energy versus cell

0927-0256/$ - see front matter Ó 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.commatsci.2011.02.020
R. Miloua et al. / Computational Materials Science 50 (2011) 2142–2147 2143

pect that this discrepancy is not caused solely by intrinsic calcula-


tion errors, but also by the experimental uncertainty. The extreme
difficulties in extracting accurate P-V data on the cubic phase from
other high-pressure coexisting phases explain the small values of
the bulk modulus B (202 GPa) and its pressure derivative B0 (1.3)
[6].
In Table 1 we also collected other theoretical works which used
different methods and approximations. A reasonable agreement is
observed with the reported theoretical data. Our calculations gave
the lowest deviation from experiment over the whole presented
values. Finally we should notice that bulk moduli presented by
Swamy and Muddle [17] are 60% larger than ours and the pressure
derivatives B0 (1.75, 2.06) seem to be anomalously low. This casts
some doubt over their accuracy.

3.2. Elastic moduli

The elastic constants have been calculated from fitting of the to-
tal energies of strained crystal to a fourth-order polynomial equa-
tion of the strains. Since we have three independent elastic
Fig. 1. The fluorite structure of titanium dioxide, yellow balls are Ti, dark balls are
constants for the cubic structure, three types of strain, i.e., the vol-
O. (For interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article.) ume change, and volume-conserved tetragonal and rhombohedral
shear strains were applied to the optimized structure. The elastic
constants, displayed in Table 1, allow the investigation of the
volume within the Murnaghan EOS [13]. The resulting values are mechanical stability of f-TiO2 by checking the following
compiled in Table 1, with other theoretical and experimental data. restrictions:
Since we used the local density approximation to the exchange–
correlation potential, it is expected that our calculation underesti- C 44 > 0; C 11 > jC 12 j; and C 11 þ 2C 12 > 0; ð1Þ
mates the lattice parameter and overestimates the bulk modulus.
This short-coming is due to the difficulty in modeling the correla- These conditions guarantee that the strain energy is positive.
tion interaction between electrons and hence, calculating the cor- Hence, from both LDA and GGA calculations we notice a good
rect electron density functional. The generalized gradient agreement with the available theoretical data. It is found that
correction considers the dependence of the total energy on both TiO2 in the fluorite structure is mechanically stable, in consistency
the electron density and its gradient. In comparison to the LDA, with the literature.
GGA usually improves the structural parameters and cohesive en- In order to gain a deeper insight into the mechanical properties
ergy of the material. In our case we found that the lattice parame- of the material, we have used both the Voigt–Reuss–Hill (VRH) and
ters compare well with both theoretical and experimental values the Hashing–Shtrikman (HS) averaging schemes [18] for the cubic
taken from Refs. [6,14–17], the GGA yields better value than LDA. polycrystalline form. The Young modulus and Poisson ratio have
In spite, we notice that both LDA and GGA underestimate the been estimated for both single-crystal and aggregate forms using
experimental bulk modulus by 43% and 22%, respectively. We ex- the following equations:

Table 1
Structural and elastic constants of cubic TiO2 using the single-crystal (SC), the Voight–Reuss–Hill (VRH) and the Hashing–Shtrikman (HS) schemes. B (bulk modulus), G (shear
modulus), Cij (elastic constants), E (Young modulus) and H (hardness) are in (GPa). B/G ratio, A (anisotropy factor) and m (Poisson ratio) are dimensionless parameters.

a B B0 C11 C12 C44 G B/G A E m H


This work
SC-LDA 4.742 288.7 4.30 665.5 105.7 59.4 280 1.03 0.21 636.6 0.14 17.3
SC-GGA 4.840 247.1 4.63 610.7 70.1 46.3 270 0.93 0.17 596.3 0.10 15.9
VRH-LDA 292.3 117 2.50 310.2 0.32
VRH-GGA 250.3 103 2.43 270.7 0.32
HS-LDA 292.3 170 1.72 426.6 0.26
HS-GGA 250.3 158 1.58 392.3 0.24
Ref. [14]
LDA 4.744 293.6 670.3 103.5 68.6 154.5b 1.90 0.24 394.1 0.27 17.2
GGA 4.841 249.3 590 79.6 48.5 131.2a 1.90 0.19 334.9 0.27 15.9
Ref. [15]
LDA 4.728 324 4.68 687 143 73 128 2.53 0.27 339 0.33 17.5
GGA 4.82 272 4.66 604 107 52 104 2.61 0.20 276 0.33
Ref. [16]
GGA 4.817 254 603 75 43 131b 1.91 0.16 336 0.28 17.1
Ref. [17]
a c
GGA 4.83 395 1.75 239 , 244 1.63 601.8 0.24 16.0
B3LYP 4.82 390 2.06 227a, 234c 1.69 577.7 0.25 16.2
Experiment [6] 4.87 202 1.3 15.5d
a
Calculated using VRH.
b
Calculated using the Voigt scheme.
c
Calculated using HS.
d
Estimated.
2144 R. Miloua et al. / Computational Materials Science 50 (2011) 2142–2147

9BG 3B  2G Moreover, the velocities of the longitudinal and transversal


E¼ v¼ : ð2Þ
3B þ G0 2ð3B þ GÞ waves along [1 0 0], [1 1 0] and [1 1 1] directions have been calcu-
lated using the averaged LDA-GGA elastic constants Cij, and follow-
The shear modulus is calculated by taking G ¼ ðC 11  C 12 Þ=2 for
ing the relations [22]:
the single-crystal form and
 1=2  1=2  1=2
C 11 C 44 C 11 þ C 12 þ 2C 44
1 vL ¼ ; vT ¼ ; vl ¼ ; v t1
GVRH ¼ ðGV þ GH Þ; q q 2q
2  1=2
C 11  C 12
( ¼ ; v t2 ¼ v T ; v l0
GV ¼ 15 ðC 11  C 12 þ 3C 44 Þ 2q
With 5ðC 11 C 12 ÞC 44 ð3Þ  1=2  1=2
GR ¼ ½4C ; C 11 þ 2C 12 þ 4C 44 C 11  C 12 þ C 44
44 þ3ðC 11 C 12 Þ ¼ ; v t0 ¼ ; ð5Þ
3q 3q
and
Thus, the values are displayed in Table 2. The density was obtained
1 by averaging LDA and GGA cell volumes, and taking the molecular
GHS ¼ ðGHS þ GHS Þ; mass of the crystal. The calculated density equals 4.79 g/cm3, which
2
is in agreement with q = 4.75 g/cm3 after Ma et al. [16].
8  
>
< GHSþ ¼ C 44 þ 2 GC5 þ 5C18ðBþ2C 44 Þ
44 44 ð3Bþ4C 44 Þ 3.3. Thermal properties
With   ; ð4Þ
>
: GHS ¼ G þ 3 C 5G þ 5C12ðBþ2GÞ
44 s ð3Bþ4GÞ In order to get new insights into the thermodynamic properties
of f-TiO2, the quasi-harmonic Debye model as implemented in the
for the aggregate one. From Table 1 we see that the obtained shear GIBBS code [12] is used. In this approach, the nonequilibrium Gibbs
moduli follow the relation GVRH < GHS < GSC for both LDA and GGA function G(V; P, T) takes the form of:
values. This behavior influences directly the other macroscopic
mechanical parameters of f-TiO2. When comparing with the other G ðV; P; TÞ ¼ EðVÞ þ PV þ Av ib ðHðVÞ; TÞ; ð6Þ
theoretical works, the calculated values exhibit a relative dispersion where E(V) the total energy per unit is cell and H(V) is the Debye
which may be explained by the sensitivity of elastic constants to the temperature. The vibrational Helmholtz free energy Avib can be
accuracy of the method. written as [23,24]:
According to the relations (3) and (4), we calculated the B/G ra-     
tio which is related to the brittleness of the material. Pugh [19] 9H H H
Av ib ðH; TÞ ¼ nkB T þ 3 ln 1  e T  D ð7Þ
suggests that the material has ductile behavior when B/G > 1.75 8 T T
and it has brittle character when B/G 6 1.75 [20]. Such behavior where n is the number of atoms per formula unit and D(y) is the De-
is related indirectly to the anisotropy factor. For single-crystal bye integral defined by:
form, we found that f-TiO2 is brittle. However, for the polycrystal- Z y
line form, the ratio B/G depends on whether we use the VRH or the 3 x3 hx
DðyÞ ¼ dx and x ¼ : ð8Þ
HS models. For both LDA and GGA calculations, the VRH model y3 0 ex  1 kB T
yields to B/G > 1.75 resulting in a ductile feature, which is in agree-
For an isotropic solid, H is expressed as [23]:
ment with the results from Refs. [14–16]. In contrast the HS model rffiffiffiffiffi
predicts a brittle character in accordance with the results from Ref. h B
[17].
H ¼ ½6pV n f ðrÞ S ;
1=2 1=3
ð9Þ
kB M
Most studies on mechanical properties of hard materials as-
where M is the molecular mass per formula unit, and BS the adia-
sume that the bulk modulus relates to hardness. However, hard-
batic bulk modulus which can be approximated by the static com-
ness depends on the dislocation mobility which in turn depends
pressibility. The Poisson ratio r is taken to be 0.25; the function f(r)
on the shear modulus G. A rough estimation of the Knoop microh-
is given in Refs. [25,26]. The isothermal bulk modulus BT, the heat
ardness can be obtained from G by the simple empirical relation
capacities CV, CP and the thermal expansion coefficient a are given
H = 0.139G. Hence, the calculated shear moduli yield to hardness
respectively by [26]:
values ranging from 14.3 GPa (VRH-GGA) to 38.9 GPa (SC-LDA).

!
The intrinsic hardness can be estimated with better accuracy with- @ 2G ðV; P; TÞ
in the semi-empirical model of Simunek [21]. This model relates BT ðP; TÞ ¼ V ; ð10Þ
@V 2
the hardness with the equilibrium cell volume (V0), the atomic ra- P;T

dii (rTi, rO) and the valence electron numbers (ZTi, ZO). We used the    
following parameters: rTi = 1.46 Å, rO = 0.89 Å, ZTi = 4, ZO = 6. The H 3H=T
C V ¼ 3nkB 4D  3H=T ; ð11Þ
obtained results are similar to those calculated by the other T e 1
authors. It can be concluded that the f-TiO2 cannot be a superhard
material since the predicted hardness is less than 40 GPa. C P ¼ C V ð1 þ acTÞ; ð12Þ
On the light of the calculated anisotropy values, we can notice
that the f-TiO2 is characterized by a relatively strong anisotropy. cC V
a¼ : ð13Þ
Further, a fair agreement is observed for the aggregate Young mod- BT V
uli in comparison to the other reported values. It appears clearly
Table 2
that they are strongly dependent on the choice of the aggregate
Acoustic velocities of the longitudinal and transversal waves along [1 0 0],
model. The Poisson ratio can relates to the measurement of the [1 1 0] and [1 1 1] directions (in km/s).
material to resist shear. It has been proven that m = 0.25 is the low-
[1 0 0] Direction [1 1 0] Direction [1 1 1] Direction
er limit for central-force solids and m = 0.5 is an upper limit which
corresponds to infinite elastic anisotropy. From Table 1, the Pois- vL = 11.541 vl = 9.317 vl
0 = 8.447
son ratio remains higher than the Cauchy 0.25 value, so the inter-
vT = 3.321 vt1 = 7.578 vt 0 = 6.478
vt2 = 3.321
atomic forces in the compound are central [16].
R. Miloua et al. / Computational Materials Science 50 (2011) 2142–2147 2145

c is the Grüneisen parameter. el is chosen because of its familiarity, analytic simplicity, ease of use
The temperature-dependent bulk moduli have been fitted to the and its excellent agreement with the fitted data since it was suc-
Einstein-oscillator model described by: cessfully applied to more than 20 materials by Varshni [28]. The
s resulting values of the fitting parameters are indicated in Table 3
BðTÞ ¼ BðT ¼ 0Þ ¼  ; ð14Þ and used to get the temperature derivatives of BT and Bs. It is found
et=T  1
that the temperature derivative of BT is lower than that of Bs by a
Here, s relates to the zero-point-vibration-energy contribution to factor of two. The whole obtained values are slightly affected with
the elastic stiffness and t is the Einstein temperature [27]. This mod- the pressure.
Fig. 2a depicts the normalized volume-pressure dependence at
Table 3 300 K and 1000 K. It is shown that augmenting temperature in-
Fitting parameters of the Einstein-oscillator model. creases the material’s compressibility. The same behavior is exhib-
BT BT BS BS ited in the bulk modulus as displayed in Fig. 2b. Since BS = BT
(at 0 GPa) (at 20 GPa) (at 0 GPa) (at 20 GPa) (1 + acT), it is obvious that BS and BT coincide at zero-T and diverge
s (GPa) 17.22 20.87 11.68 15.34 more and more when T increases. Their trend is nearly constant
t (K) 422.16 515.52 601.25 756.33 from 0 to 150 K and decreases linearly for T > 200 K. We estimated
@B s
@T ¼  t
4.08  102 4.05  102 1.94  102 2.03  102 the Grüneisen parameter to be c = 2.04.
(GPa/K) The Debye temperature as a function of pressure and tempera-
ture is depicted in Fig. 3. It is well known that the Debye

1400
300 K
1,00 300
1000 K
600
1300
1000
0,95
Debye Temperature [K]

1200
V/V0

0,90
1100

0,85
1000

0,80
(a) 900

0 20 40 60 80
800
Pressure [GPa] 0 20 40 60 80 100
Pressure [GPa]
380

1400
360
1350
Bulk modulii BT, Bs [GPa]

1300
340
1250 0 GPa
Debye Temperature [K]

40 GPa
BT at 0 GPa 1200
320 90 GPa
BT at 20 GPa
1150
Bs at 0 GPa
300 1100
Bs at 20 GPa
1050
280
1000

950
260
(b)
900
0 200 400 600 800 1000
850
Temperature [°K] 0 200 400 600 800 1000
Fig. 2. Normalized volume–pressure curves and bulk moduli at different temper- Temperature [K]
atures for c-TiO2, BT and BS are the isothermal and adiabatic bulk moduli,
respectively. Fig. 3. Debye temperature at different pressure and temperature values.
2146 R. Miloua et al. / Computational Materials Science 50 (2011) 2142–2147

80 4,0
0 GPa
70 3,5 40 GPa
0 GPa
90 GPa
40 GPa
60 3,0
90 GPa

50 2,5
C v [J/mole K]

α [10 -5 /K]
2,0
40

1,5
30

1,0
20
0,5
10
0,0
0 0 200 400 600 800 1000
0 200 400 600 800 1000
Temperature [°K]
Temperature [°K]

3,6 300 K
80 P=0 600 K
P = 40 3,2 1000 K
70 P = 90
2,8
60
α [10-6/K]

2,4
50
Cp [J/mol°K]

2,0
40
1,6
30
1,2
20
0,8
10
0 20 40 60 80 100
0
Pressure [GPa]
0 200 400 600 800 1000
Temperature [°K] Fig. 5. Variation of the thermal expansion coefficient of f-TiO2.

Fig. 4. Variation of the specific heats Cv and Cp with temperature and pressure.
The temperature and pressure dependences of the thermal
expansion coefficient of f-TiO2 are given in Fig. 5. We found that
temperature is related to the bulk. We remark an increasing of HD as pressure increases, thermal expansion increases rapidly, and al-
when the pressure increases, in contrast to the decreasing trend most exponentially. In fact, the effect of pressure on the thermal
which results from the temperature effect. This can be understood expansion coefficient is very small at low temperature and be-
if we consider the fact that increasing pressure is equivalent to comes strong as the temperature increases. The temperature
decreasing temperature at the level of the crystal volume. dependent evolution is that a increases as T3 at low temperature
Fig. 4 shows the calculated specific heats at constant volume and gradually turns to a linear increase at high temperature. The
(Cv) and constant pressure (Cp) of f-TiO2. Thus, it is clear that their following equations describe the pressure dependence of the De-
temperature dependence follows a T3 law for low temperature and bye temperature, the specific heats and the thermal expansion
tends to be linear for higher values of T, where Cv becomes very coefficient at 300 K, which were obtained by a least square fitting
close to the Dulong–Petit classical limit. to a fourth-order polynomials.

H
H0 ¼ 1 þ 0; 0071802P  3:568  105 P2 þ 1; 7638  107 P3  4; 5338  1010 P4 ;
CV
C V0
¼ 1  0; 00564P þ 3; 18905  105 P2  1; 38958  107 P3 þ 3; 05909  1010 P4 ;
ð15Þ
CP
C P0
¼ 1  0; 00593P þ 3; 72544  105 P2  1; 87842  107 P3 þ 4; 85623  1010 P 4 ;
a
a0 ¼ 1  0; 01987P þ 2; 68933  104 P2  2; 20194  106 P3 þ 7; 72801  109 P4 :
R. Miloua et al. / Computational Materials Science 50 (2011) 2142–2147 2147

Although the experimental values of these properties under [5] M. Mattesini, J.S. de Almeida, L. Dubrovinsky, N. Dubrovinskaia, B. Johansson, R.
Ahuja, Phys. Rev. B 70 (2004) 115101.
high pressure and temperature are not yet available for compari-
[6] M. Mattesini, J.S. de Almeida, L. Dubrovinsky, N. Dubrovinskaia, B. Johansson, R.
son. Hence our calculations cover the lack of data for f-TiO2. Ahuja, Phys. Rev. B 70 (2004) 212101.
[7] D.Y. Kim, J.S. de Almeida, L. Koči, R. Ahuja, Appl. Phys. Lett. 90 (2007) 171903.
[8] E. Wimmer, H. Krakauer, M. Weinert, A.J. Freeman, Phys. Rev. B 24 (1981) 864.
4. Conclusion [9] P. Blaha, K. Schwarz, G.K.H. Madsen, D. Kvasnicka, J. Luitz, WIEN2K, An
Augmented Plane Wave + Local Orbitals Program for Calculating Crystal
Properties, Karlheinz Schwarz, Techn. Universitat, Wien, Austria, 2001. ISBN:
In summary, the structural, elastic and thermal properties of 3-9501031-1-2.
TiO2 in the fluorite phase have been addressed. We used the full- [10] D.M. Ceperley, B.J. Alder, Phys. Rev. Lett. 45 (1980) 566.
[11] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865.
potential LAPW approach in combination with le LDA and GGA [12] M.A. Blanco, E. Francisco, V. Luana, Comput. Phys. Commun. 158 (2004) 57.
schemes. It is found that GGA yields better values of the lattice [13] F.D. Murnaghan, Proc. Natl. Acad. Sci. USA 50 (1944) 667.
constant and bulk modulus than LDA. The calculated elastic con- [14] W. Lu, H. Wang, Y. Hu, H. Huang, H. Gu, Physica B 404 (2009) 79.
[15] Y. Liang, B. Zhang, J. Zhao, Phys. Rev. B 77 (2008) 094126.
stants fulfilled the mechanical stability conditions, in accordance [16] X.G. Ma, P. Liang, L. Miao, S.W. Bie, C.K. Zhang, L. Xu, J.J. Jiang, Phys. Status
with the literature. Further, using two different aggregate schemes, Solidi B 246 (2009) 2132.
it was possible to calculate shear moduli (G), B/G, ratios, Young [17] V. Swamy, B.C. Muddle, Phys. Rev. Lett. 98 (2007) 035502.
[18] G. Simmons, H. Wang, Single Crystal Elastic Constants and Calculated
moduli and Poisson ratios. The calculated intrinsic hardness of
Aggregate Properties: A Handbook, MIT Press, Cambridge, 1971.
the material showed that it cannot be considered as a superhard [19] S.F. Pugh, Philos. Mag. 45 (1954) 823.
material. Finally, the Debye temperature, heat capacities and linear [20] P. Ravindran, L. Fast, P.A. Korzhavyi, B. Johansson, J. Wills, O. Eriksson, J. Appl.
expansion coefficients as functions of pressure and temperature Phys. 84 (1998) 4891.
[21] A. Simunek, Phys. Rev. B 75 (2007) 172108.
have been predicted for the first time. [22] S. Adachi, Properties of Group-IV, III–V and II–VI Semiconductors, John Wiley &
Sons Ltd., England, 2005.
[23] M.A. Blanco, A. Martín Pendás, E. Francisco, J.M. Recio, R. Franco, J. Mol. Struct.
References (Theochem.) 368 (1996) 245.
[24] M. Flórez, J.M. Recio, E. Francisco, M.A. Blanco, A.M. Pendás, Phys. Rev. B 66
[1] A. Fujishima, K. Honda, Nature 238 (1972) 37. (2002) 144112.
[2] M.R. Hoffman, S.T. Martin, W. Choi, D.W. Bahnemannt, Chem. Rev. 95 (1995) [25] E. Francisco, J.M. Recio, M.A. Blanco, A. Martín Pendás, J. Phys. Chem. 102
69. (1998) 1595.
[3] W. Luo, T. Yu, Y. Wang, Z. Li, J. Ye, Z. Zou, J. Phys. D: Appl. Phys. 40 (2007) 1091. [26] E. Francisco, M.A. Blanco, G. Sanjurjo, Phys. Rev. B 63 (2001) 094107.
[4] L.S. Dubrovinsky, N.A. Dubrovinskaia, V. Swamy, J. Muscat, N.M. Harrison, R. [27] H. Ledbetter, Phys. Status Solidi B 181 (1994) 81.
Ahuja, B. Holm, B. Johansson, Nature London 410 (2001) 653. [28] Y.P. Varshni, Phys. Rev. B 2 (1970) 3952.

You might also like