- Email: [email protected]
Study of the effect of Cu heavy doping on band gap and absorption spectrum of ZnO
Chemical Physics Letters 614 (2014) 15–20
Contents lists available at ScienceDirect
Chemical Physics Letters journal homepage: www.elsevier.com/locate/cplett
Study of the effect of Cu heavy doping on band gap and absorption spectrum of ZnO Shaoqiang Guo a , Qingyu Hou a,∗ , Chunwang Zhao a , Yue Zhang b a b
College of Science, Inner Mongolia University of Technology, Hohhot 010051, China School of Material Science and Engineering, Beijing University of Aeronautics and Astronautics, Beijing 100083, China
a r t i c l e
i n f o
Article history: Received 13 June 2014 In final form 3 September 2014 Available online 10 September 2014
a b s t r a c t Contradictory experimental absorption spectra blue shift and red shift results have been reported in the literatures. To solve this problem, this study investigates the electronic structure and absorption spectra of Zn1−x Cux O (x = 0, 0.0313 and 0.0625, respectively) and Zn32 CuO32 supercells employing firstprinciples calculations with GGA+U method. By increasing the Cu substitutional doping concentration from 3.13% to 6.25%, the following results are obtained: increased magnetic properties, narrower band gaps, and a significant red shift in the absorption spectrum. These findings are in good agreement with the experimental results. The changes of band gap and absorption spectrum for interstitial doping and substitutional doping are opposite. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Zinc oxide is an important II–VI type semiconductor that possesses a direct wide band gap of 3.37 eV [1] and a high exciton binding energy of 60 meV [2] at room temperature. ZnO also has low dielectric constant, large electric-coupling ratio, high chemical stability, as well as excellent piezoelectric and photoelectric properties. This type of semiconductor has potential applications in short-wavelength optoelectronic devices, such as laser and light-emitting diodes. In experimental and theoretical studies, the structure of ZnO can be changed easily by effective alloying with a dopant, which involves ZnO material energy band engineering. In addition, by regulating the ZnO band gap, the performance of ZnObased optoelectronic devices can be improved. Therefore, doped ZnO has vast applications in the production of solar cells [3], liquid crystal displays [4], gas sensors [5], UV laser diodes [6], transparent conductive films [7], and diluted magnetic semiconductors [8]. Extensive works on photoelectrical properties of doped ZnO have been carried out. Kulyk et al. [9] prepared Cu-doped ZnO by radio-frequency magnetron sputtering; their results showed that the grain size and band gap decreased, and the absorption band edge shifted to low levels with increased Cu concentration. Wu et al. [10] prepared singly doped Co and Cu as well as co-doped
∗ Corresponding author. E-mail addresses: [email protected] (S. Guo), [email protected] (Q. Hou). http://dx.doi.org/10.1016/j.cplett.2014.09.005 0009-2614/© 2014 Elsevier B.V. All rights reserved.
ZnO thin films using the sol–gel process. They also investigated the influence of Co and Cu doping on the surface morphologies of ZnO films, and found blue double peaks at all ZnO thin film samples caused by transition of electrons from conduction band minimum (CBM) to zinc vacancy or from Zn interstitial to valence band maximum (VBM). However, the green peak is highly relevant to the oxygen slip formed by doping. Ma et al. [11] prepared Cudoped ZnO films by radio-frequency reactive magnetron sputtering and studied the microstructures and optical properties of doped systems. The results indicated that ZnO films exhibited a strong orientation toward the c-axis and a uniform grain size after Cudoping. Furthermore, the intensities of blue (∼485 nm) and green (∼527 nm) peaks showed significant increase after annealing. Fu et al. [12] investigated the photocatalytic performance of Cu-doped ZnO nanoparticles prepared using the sol–gel process. Theoretical calculations have also been performed to determine the photoelectrical properties of doped ZnO. For instance, Kong and Gong [13] calculated the band gap of Be substitutional and interstitial doping ZnO using the GGA method, and found that substitution of Zn atoms by Be could enhance the band gap of ZnBeO phases, while interstitial Be atoms would decrease the band gap. Ye et al. [14] calculated ferromagnetism of Cu-doped ZnO by the local density approximation (LDA) (Hedin-Lunquist form) and generalized gradient approximation (GGA) (Perdew–Burke–Ernzerhof form). Li et al. [15] investigated the magnetic and optical properties of doped ZnO nanosheets using the GGA+U method. Their results revealed that the replacement of two Cu atoms by two Zn atoms enhanced the magnetic properties of the doped
16
S. Guo et al. / Chemical Physics Letters 614 (2014) 15–20
Figure 1. The 2 × 2 × 4 supercells for (a) undoped ZnO, (b) Zn0.9687 Cu0.0313 O, (C) Zn0.9375 Cu0.0625 O, and (d) Zn32 CuO32 .
system. Moreover, the absorption spectrum exhibited a noticeable red shift with increased doping concentration. Kang et al. [16] analyzed the variations in the ferromagnetism of carrier doping in Zn1−x Cux O1−y Xy (X = N and F; x, y = 0.0277–0.0833) by the full-potential linear muffin-tin orbital (FP-LMTO) method. Their experimental results showed that Cu magnetic moments at low Cu concentrations increased and decreased by N and F doping, respectively. Meanwhile, Liu et al. [17] calculated the band gap variations and absorption spectrum of wurtzite Zn1−x Cox O using the GGA method, and found that the band gaps broadened and the optical absorption edge exhibited a significant blue shift given an atomic fraction of Co. Duan et al. [18] investigated the electronic structure and optical properties of ZnO doped with transition metals and N. Their calculations indicated that the co-doping of transition metals (Mn, Fe, Co, and Cu) and N favored the formation of p-type ZnO. Although good results on the effects of Cu doping on the optoelectronic properties of ZnO have been theoretically and experimentally achieved, speculations remain upon comparing the band gaps and absorption spectra of undoped and Cu-doped ZnO. Increased Cu substitutional doping from 3.13% to 6.25% widened the band gap and caused a blue shift on the absorption spectrum [19], which contradicted the findings of Ref. [20]. From crystal periodicity, we can obtain useful results by performing first-principles calculations of the band structure, band gap, and the absorption spectrum of Cu-doped ZnO, with a doping concentration range similar to that presented in previous works [19,20]. We believe that the results can contribute to better design and improved preparation of short-wavelength optoelectronic devices from Cu-doped ZnO.
2. Theoretical model and computational method 2.1. Theoretical model ZnO has a hexagonal wurtzite crystal structure that belongs to the space group P63 mc and C64v symmetry. The unit cell comprises two hexagonal close packed lattices along the c axis and sleeve; the lattice parameters are a = 0.3249 nm and c = 0.5205 nm. The 2 × 2 × 4 supercells for undoped ZnO, Zn0.9687 Cu0.0313 O, Zn0.9375 Cu0.0625 O and Zn32 CuO32 were applied to calculate the process of geometric optimization and energy. Three substitutional doping levels of 0%, 3.13%, and 6.25% with zero, one and two Zn atoms replaced by Cu atom, respectively, were examined.
Figure 2. The test of energy convergence as the cut-off varied for undoping ZnO.
Zn32 CuO32 is a system of one Cu atom interstitial doping. The atomic positions of Cu dopants are displayed in Figure 1. 2.2. Computational method Calculations were performed by CASTEP package [21–23] as implemented in the Materials Studio software. The software is based on the density functional theory (DFT) using the planewave ultrasoft pseudopotential method [24–26]. Perdew Burke Ernzerhof (PBE) scheme in the generalized gradient approximation (GGA) was used to treat the exchange-correlation energy. We first optimized the structure of undoped ZnO, Zn0.9687 Cu0.0313 O, Zn0.9375 Cu0.0625 O and Zn32 CuO32 and then calculate the electronic and optical properties. During optimization, Brillouin zone integrations were first performed with the special k-point method over a 4 × 4 × 1 Monkhorst-Pack mesh. Energy convergence during cut-off variation for undoping ZnO supercells was also tested (Figure 2). The total energy of the system begins to converge at the cutoff energy of 320 eV and presents good convergence at the cutoff energy of 380 eV. Therefore, cutoff energy at 380 eV is the most ideal for all calculation models. Second, the plane wave energy was set at 1.0 × 10−5 eV/atom, and the maximum tolerances of force, stress, and displacement were set at 0.3 eV/nm, 0.05 GPa and
S. Guo et al. / Chemical Physics Letters 614 (2014) 15–20
1.0 × 10−4 nm, respectively. The valence atomic configurations of Zn, Cu and O were 3d10 4s2 , 3d10 4s1 and 2s2 2p4 , respectively. The effects of spin polarization were considered in energy calculations. Because of the GGA failed to properly describe the position of ZnO band narrowing, which corresponded to high localization in contrast to s and p bands. Hence, an orbital-dependent potential was used, including an additional Coulomb interaction U (GGA+U) [27–31]. Using GGA+U and choosing a reasonable value for U are both very important, because U acts as a fitting parameter to reproduce the experimentally observed position including the p states and d states. In the calculations, Ud,Zn , Up,O , and Ud,Cu were set at 5.5, 8.0 and 6.0 eV, respectively, this will be explained in the band gap analysis.
3. Results and discussion
17
3.2. Mulliken bond population and bond length analysis Mulliken bond population and bond length of undoped and Cu substitutional doped ZnO were calculated to analyze the doping mechanism of the latter (Table 1). Bond population assesses the ionic or covalent bonds [39], and indicates the distribution of electron population around atoms. Positive and negative values of the bond population imply covalent and ionic bonding, respectively. Given the increased Cu substitutional doping concentration, the Mulliken bond populations and bond lengths of Cu O parallel and vertical to c-axis are lower than those of Zn O in undoped ZnO cells. Apart from the influence of bond population, the replacement of Cu2+ with Zn2+ decreases that bond lengths of Cu O in the doped system. Moreover, increased Cu-doping concentration weakens the covalent bond, strengthens the ionic bond, and destabilizes the substitutional doped system. The results are in agreement with those of the formation energies.
3.1. Crystal structure and stability analysis 3.3. Orbital average charge analysis The lattice parameters, volumes, equivalent total energies, and formation energies of Zn1−x Cux O (x = 0, 0.0313 and 0.0625, respectively) and Zn32 CuO32 were obtained following geometry optimization. The calculated results are shown in Table 1. The optimized geometry lattice parameters of undoped and Cu substitutional doped ZnO are close to experimental values [32,33]. The Mulliken method was used to calculate the orbital charge. Results show that at 3d10 4s1 valence atomic configuration for Cu in Zn0.9687 Cu0.0313 O and Zn0.9375 Cu0.0625 O, the charge transfer sum of the s and d states is close to 2. Therefore, the valence of Cu is +2 and Cu2+ ion exists in the doped system, which are consistent with the experimental results [34]. Equivalent volumes of the crystal cell decrease with increasing Cu concentration, which is due to the substitution of Cu2+ for Zn2+ . The radius of Zn2+ (0.074 nm [35]) is larger than that of Cu2+ (0.072 nm [35]). As the volume of the Cu atom interstitial doping increases, due to the extremely short distance between Cu and the neighboring Zn atom, their repulsive interaction also increases. Experimental studies reveal that the transmittance of the doping system is maintained at 60% when the Cu-doping concentration reaches 14.5% [36]. The crystal undergoes phase transformation for doping concentrations greater than 14.5%. In this study, Cu-doping concentrations ranged between 3.13% and 6.25%. This selection meets the crystal into hexagonal wurtzite structure for doping concentrations do not beyond 14.5%. In addition, the inconsistencies in the absorption spectra for various studies [19,20] are resolved. Impurity formation energy (Ef ) analyzes the relative degree of difficulty of atomic doping and illustrates the influence of Cu doping on the stability of wurtzite ZnO. Ef is given by Eq. (1) [37] as: Ef = EZnO:Cu − EZnO − ECu + EZn
(1)
where EZnO:Cu and EZnO are the total energies of Cu-doped ZnO and undoped ZnO, respectively for supercells similar sizes; and ECu and EZn are the respective energies of bulk Cu and Zn metals. Table 1 illustrates the calculated results for undoped and substitutional doping systems, indicating that the total energies and energies for impurity formation increase, and the doping becomes increasingly difficult with increased Cu content. Therefore, the substitutional doping system structure becomes unstable. Meanwhile, impurity formation energy of the Cu interstitial doping system is higher than that of Cu substitutional doping system with a similar doping concentration. Doping becomes more difficult, and the interstitial doping system structure becomes highly unstable. These results are in good agreement with previously published theoretical results [38].
The Mulliken atomic populations method was used to calculate the orbital average charges of O, Zn, and Cu atoms in undoped ZnO, Zn0.9687 Cu0.0313 O and Zn0.9375 Cu0.0625 O, respectively (Table 2). The calculation was performed to support the analysis of the mechanism of the partial density of states of Cu substitutional doped ZnO. The results show that the orbital charges of O-2p, Cu-3p and Zn-3p are reduced; the orbital charges of Zn-3d and Cu-4s remain unchanged; and the orbital charge of Zn-4s increases. 3.4. Analysis of electron density differences of undoped and Cu substitutional doped ZnO Paraguay et al. [40] prepared ZnO thin films by spray pyrolysis technique using an aqueous medium. The XRD results showed that the films were polycrystalline with preferred orientation along the (0 0 2) plane. Electron density differences at (0 0 2) for undoped and Cu substitutional doped ZnO were calculated to investigate the interactions and bonding among adjacent atoms in the Cu substitutional doped system (Figure 3a–c). The extent of overlapping among the neighboring electronic clouds of Zn O in undoped ZnO becomes noticeable as the electronic communication strengthens. By contrast, the extent of the overlapping neighboring electronic clouds of Cu O and Zn O in the Cu substitutional doped system weakens with increasing Cu-doping concentration. In other words, the covalent and ionic bonds weaken and strengthen, respectively. The results are in good agreement with those of the orbital average charges. 3.5. Band gaps analysis We set Ud,Zn = 5.5 eV and Up,O = 8.0 eV for all systems in our calculation using the GGA+U method. First, the band structure of undoped ZnO supercell was determined (Table 3). The band gap for undoped ZnO is 3.44 eV, which is in good agreement with the experimental values [1,20]. Hence, the values of Ud,Zn and Up,O in undoped ZnO set at 5.5 and 8.0 eV, respectively, were reasonable. We selected 5.00, 6.00, and 7.00 eV for Ud,Cu to obtain the theoretical band gaps of the substitutional doped system coinciding with the experiment. In addition, the band gaps for Zn0.9687 Cu0.0313 O and Zn0.9375 Cu0.0625 O were calculated. The results are shown in Table 3. The data show that the relative band gap values do not change much in Zn0.9687 Cu0.0313 O and Zn0.9375 Cu0.0625 O when the Ud,Zn , Up,O , and Ud,Cu are set at 5.50, 8.00 and 6.00 eV, respectively. Therefore, Ud,Cu at 6.00 eV is most ideal for use in the compromise method. In addition, the lattice constants of the a and c axes for the Cu substitutional doping system are reduced with increasing
18
S. Guo et al. / Chemical Physics Letters 614 (2014) 15–20
Table 1 The physical parameters of Zn1−x Cux O (x = 0, 0.0313, 0.0625) and Zn32 CuO32 . Model
ZnO Zn0.9687 Cu0.0313 O Zn0.9375 Cu0.0625 O Zn32 CuO32
a, b, c (nm)
V (nm3 )
E (eV)
Ef (eV) This work
This work
Experimental values
This work
This work
a = 0.3249 c = 0.5205 a = 0.3248 c = 0.5204 a = 0.3247 c = 0.5203 a = 0.3300 c = 0.5338
a = 0.3250[32] c = 0.5210[32] c = 0.5202[33]
0.04759
−4294.45
0.04754
−4279.88
−2.54
c = 0.5199 [33]
0.04751
−4265.15
−1.28
–
0.05034
−4381.61
4.29
Bond
Zn Zn Cu Cu Cu Cu –
O (||c) O (⊥c) O (||c) O (⊥c) O (||c) O (⊥c)
Population
Bond length (nm)
This work
This work
0.39 0.43 0.37 0.42 0.35 0.41 –
0.2004 0.2014 0.1992 0.1973 0.1991 0.1972 –
Table 2 Orbital charges of doped system. Model
O-2p/e
Zn-4s/e
Zn-3p/e
Zn-3d/e
Cu-4s/e
Cu-3p/e
Zn0.9687 Cu0.0313 O Zn0.9375 Cu0.0625 O
5.128 5.118
0.373 0.379
0.688 0.670
9.980 9.980
0.560 0.560
0.640 0.630
Figure 3. Electron density differences in (0 0 2) for (a) undoped ZnO, (b) Zn0.9687 Cu0.0313 O, and (C) Zn0.9375 Cu0.0625 O. Table 3 The band gap of before and after ZnO supercells.
ZnO Zn0.9687 Cu0.0313 O
Zn0.9375 Cu0.0625 O
Zn32 CuO32
[20]
ZnO Zn0.98 Cu0.02 O Zn0.94 Cu0.06 O
Ud,Zn (eV)
Up,O (eV)
Ud,Cu (eV)
Eg (eV)
Eg (eV)
5.50 5.50 5.50 5.50 5.50 5.50 5.50 5.50 5.50 5.50
8.00 8.00 8.00 8.00 8.00 8.00 8.00 8.00 8.00 8.00
– 5.00 6.00 7.00 5.00 6.00 7.00 5.00 6.00 7.00
– 0.051 0.052 0.047 0.072 0.075 0.075 0.133 0.114 0.112
3.44 3.26 3.25 3.25 3.14 3.18 3.23 3.49 3.50 3.48
– – –
– – –
– – –
– – –
3.47 3.45 3.43
Cu concentration (Table 1), which is due to the substitution of Cu2+ for Zn2+ . This finding is consistent with the change trend of experimental results [31,32]. As the lattice constants of the a and c axes for Cu interstitial doping system increase, because of the extremely short distance between Cu and the neighboring Zn atom, their repulsive interaction increases. Theoretical calculation [41] has proven that Pulay stress has an impact on the band gap of
ZnO. Compressive strain and tensile strain can enhance and reduce the band gap of ZnO, respectively. Under the condition of the same strain with undoped ZnO, we obtain the band gap variable Eg , which represents the strain that causes the band gap to increase and then decrease after Cu substitutional doping (Table 3). As can be seen from Table 3, given that the band gap variable for Cu substitutional doping that causes the band gap to decrease is
S. Guo et al. / Chemical Physics Letters 614 (2014) 15–20
19
considerably greater than the amount of strain that causes the band gap to increase, the band gaps of all substitutional doping models significantly decreased compared with the band gap of undoped ZnO. Moreover, a heavier Cu amount results in a narrower band gap. This result is consistent with the trend reported in the literature [20]. The narrow mechanisms of the band gap were analyzed based on orbital theory. While the changes of band gap for interstitial doping are opposite to that for substitutional doping. In addition, the band gaps of all substitutional doping models significantly decrease compared with the band gap of undoped ZnO; moreover, a heavier Cu amount results in a narrower band gap (Table 3). This result is consistent with the trend reported in the literature [20]. The narrow mechanisms of the band gap were analyzed based on orbital theory. 3.6. Analysis of partial density of states of undoped and Cu substitutional doped ZnO Figure 4a–c show the partial densities of states of undoped ZnO, Zn0.9687 Cu0.0313 O, and Zn0.9375 Cu0.0625 O supercells. Chemical orbital theory reveals that the band gap of undoped ZnO is formed by the respective antibonding-like and bonding-like states of s and p orbitals from the electronic interaction of Zn-4s and O-2p states (Figure 4a). CBM significantly shifts toward low energies (Figure 4b and c). The interactions of s–s states shift the conduction band (CB) toward high energies, whereas the repulsion effects of s–p states shift the CB toward low energies for Cu-doped ZnO. Given the increase in Cu content, the interactions of s-s state increase with the orbital charges of Zn-4s and the repulsion effects of s–p states decrease with the orbital charges of Zn-3p and Cu-3p. Hence, the CBM shifts to low energies. Meanwhile, the repulsion effects of p–d states shift the VB toward high energies, and the interactions of s–p states shift the VB shift toward low energies. Considering the increase in Cu-doping concentration, the interactions of p–p states decrease with the orbital charges of Cu-3p and O-2p, whereas the repulsion effects of p–d states decrease with the orbital charges of Zn-3p and Cu-3p. Therefore, the interaction size of p–p is similar to that of p–d, and the interaction direction of p–p is opposite to that of p–d. Hence, the changes in VB position are minimal. Given the increase in Cu-doping concentration, a few changes in the VB position are observed. The CB significantly shifts toward low energies; hence, the band gaps of the substitutional doped system show significant decrease, which is consistent with the result of the band structure. 3.7. Magnetic property analysis Figure 4b and c reveals the asymmetry of the total density of state, and that the number of up spins is greater than that of down spins. Zn0.9687 Cu0.0313 O and Zn0.9375 Cu0.0625 O supercells generate magnetic moments of 1.06 B and 2.23 B , respectively, indicating that increased Cu-doping concentration enhances the magnetic property of the supercells. This result agrees with that presented in a previous study [34]. Figure 4b and c imply that the spontaneous spin polarization originates from the p–d exchange between Cu-3d and O-2p orbitals of the Cu-doped ZnO, which is in good agreement with previously reported experimental result [34].
Figure 4. Densities of states of (a) undoped ZnO, (b) Zn0.9687 Cu0.0313 O, and (c) Zn0.9375 Cu0.0625 O.
calculated according to the dispersion relation of Kramers–Kronig. In this way, absorption coefficient ˛(ω), can be obtained. Omitted the derivation process [42], the concerned formulas are as following. ε2 (ω) =
As we know, the optical properties of the medium can be described by complex dielectric response function ε(ω) = ε1 (ω) + iε2 (ω) in leaner response range, in which ε1 = n2 − k2 and ε2 = 2nk. The real part ε1 (ω) and imaginary part ε2 (ω) can be
V,C
ε1 (ω) = 1 =
˛(ω) = 3.8. Absorption spectrum analysis
c ω2
√
2 0
2
BZ
(2)3
∞
0
|MCV (k)|2 · ı(ECk − EVk − ω)d3 k
ω ε2 (ω) ω 2 − ω2
dω
2[
ε21 (ω) + ε22 (ω) − ε1 (ω)]
(2)
(3) 1/2
(4)
where, subscript C and V represent conduction band and valence band, respectively, BZ is first Brillouin zone, k is reciprocal lattice vector, |MCV (k)|2 is momentum matrix element, c is constant, ω is angular frequency, ECk and EVk is intrinsic energy level. The above formulas provide a theoretical foundation for the analysis of the band
20
S. Guo et al. / Chemical Physics Letters 614 (2014) 15–20
experimental results [20]. The changes of band gap and absorption spectrum for interstitial doping and substitutional doping are opposite. These results can help improve methods by which to design and prepare short-wavelength optoelectronic devices from Cu-doped ZnO. Acknowledgments All the calculations were carried out at the Network Information and Computing Center of Beijing University of Aeronautics and Astronautics. This work was supported by the Natural Science Foundation of China (Grant nos. 61366008 and 51261017). This work was also supported by the Ministry of Education “Spring sunshine” plan funding, and the College Science Research Project of Inner Mongolia Autonomous Region (Grant no. NJZZ13099). References Figure 5. Optical absorption spectra of undoped and Cu-doped ZnO.
structure and optical properties of the crystals, which reflect the emitting mechanism of the spectrum generated by the electronic transition between energy level [43]. The absorption spectra of undoped and Cu-doped ZnO samples were calculated using GGA+U method (Figure 5). The absorption capacity of ZnO in the UV light area increases and the absorption spectrum exhibits a red shift with Cu atom substitutional doping. This red shift can be attributed to the lower energy of the electronic transition between O2p and Cu4s than that between O2p and Zn4s, with increasing amount of substitutional doping. It can be seen from band gaps analysis, the band gaps of undoped ZnO, Zn0.9687 Cu0.0313 O, and Zn0.9375 Cu0.0625 O were 3.44, 3.25 and 3.18 eV, respectively. When these values are substituted in the absorption long-wave limit formula, = (1.24/Eg (eV)) m, the calculated wavelengths are 360.5, 381.5, and 389.9 nm, respectively. For doping amounts ranging from 3.13% to 6.25%, the absorption spectra in long-UV light red shift become significant with increasing substitutional doping amount compared with undoped ZnO. The absorption band-edge shifts toward low energies. These results are consistent with the trend in experimental results [20]. While absorption spectrum of Cu atom interstitial doping exhibits a blue shift. These can help improve methods in designing and preparing long-UV light optoelectronic devices from Cu-doped ZnO. 4. Conclusion The band structures, density of states, magnetic properties, and absorption spectra of Zn1−x Cux O (x = 0, 0.0313 and 0.0625, respectively) and Zn32 CuO32 supercells are calculated using the first-principles plate-wave, ultrasoft pseudopotential method based on density functional theory. Calculation results show that the band gaps of undoped and Cu-doped ZnO coincide with the experimental values obtained through GGA+U method with a reasonable U. Increased Cu-doping concentrations from 3.13% to 6.25% reduce the volumes, increase the total and formation energies, and destabilize the substitutional doped systems. Moreover, the covalent bond weakens, the ionic bond strengthens, the magnetic property enhances, and doping becomes increasingly difficult. The band gaps become narrower and the absorption spectrum exhibits a significant red shift, which are in good agreement with
[1] V. Srikant, D.R. Clarke, J. Appl. Phys. 83 (1998) 5447. [2] Z.K. Tang, G.K.L. Wong, P. Yu, M. Kawasaki, A. Ohtomo, H. Koinuma, Y. Segawa, Appl. Phy. Lett. 72 (1988) 3270. [3] Q. Zhang, C.S. Dandeneau, X. Zhou, G. Cao, Adv. Mater. 21 (2009) 4087. [4] B.Y. Oh, M.C. Jeong, T.H. Moon, W. Lee, J.M. Myoung, J.Y. Hwang, D.S. Seo, J. Appl. Phys. 99 (2006) 124505. [5] H. Gong, J.Q. Hu, J.H. Wang, C.H. Ong, F.R. Zhu, Sens. Actuators B: Chem. 115 (2006) 247. [6] K.W. Liu, et al., Solid-State Electron. 51 (2007) 757. [7] X. Jiang, F.L. Wong, M.K. Fung, S.T. Lee, Appl. Phys. Lett. 83 (2003) 1875. [8] H.J. Lee, S.Y. Jeong, C.R. Cho, C.H. Park, Appl. Phys. Lett. 81 (2002) 4020. [9] B. Kulyk, V. Figà, V. Kapustianyk, M. Panasyuk, R. Serkiz, P. Demchenko, Acta Phys. Pol. A 123 (2013) 92. [10] D.C. Wu, et al., Acta Phys. Sin. 58 (2009) 7261. [11] L. Ma, S. Ma, H. Chen, X. Ai, X. Huang, Appl. Surf. Sci. 257 (2011) 10036. [12] M. Fu, Y. Li, S. Wu, J. Lin, F. Dong, Appl. Surf. Sci. 258 (2011) 1587. [13] F.T. Kong, H.R. Gong, Comput. Mater. Sci. 61 (2012) 127. [14] L.H. Ye, A.J. Freeman, B. Delly, Phys. Rev. B 73 (2006) 033203. [15] F. Li, C.W. Zhang, M.W. Zhao, Physica E 53 (2013) 101. [16] B.S. Kang, K.S. Kim, S.C. Yu, H. Chae, J. Solid State Chem. 198 (2013) 120. [17] Y. Liu, Q.Y. Hou, H.P. Xu, C.W. Zhao, Y. Zhang, Chem. Phys. Lett. 551 (2012) 72. [18] M.Y. Duan, M. Xu, H.P. Zhou, Y.B. Shen, Q.Y. Chen, Y.C. Ding, W.J. Zhu, Acta Phys. Sin. 56 (2007) 5359. [19] T.M. Hammad, J.K. Salem, R.G. Harrison, R. Hempelmann, N.K. Hejazy, J. Mater. Sci.: Mater. Electron. 24 (2013) 2846. [20] S. Muthukumaran, R. Gopalakrishnan, Opt. Mater. 34 (2012) 1946. [21] V. Milman, et al., J. Mol. Struc. Theochem 954 (2010) 22. [22] L.B. Shi, C.Y. Xu, H.K. Yuan, Physica B 406 (2011) 3187. [23] Y. Imai, M. Mukaida, T. Tsunoda, Thin Solid Films 381 (2001) 176. [24] D. Vanderbilt, Phys. Rev. B 41 (1990) 7892. [25] K.Z. Ska, J. Thin Solid Films 391 (2001) 229. [26] Y. Yanfa, M.M. AL-Jassim, Phys. Rev. B 69 (2004) 085204. [27] S. Amari, S. Méc¸abih, B. Abbar, N. Benosman, B. Bouhafs, Physica B 407 (2012) 3639. [28] X.G. Xu, D.L. Zhang, Y. Wu, X. Zhang, X.Q. Li, H.L. Yang, Y. Jiang, Rare Metals 31 (2012) 107. [29] N. Ghajari, A. Kompany, T. Movlarooy, F. Roozban, M. Majidiyan, J. Magn. Magn. Mater. 325 (2013) 42. [30] G. Chris, Van de Walle, A. Janotti, Phys. Status Solidi B 248 (2011) 1. [31] D.M. Song, T.H. Wang, J.C. Li, J. Mol. Model. 18 (2012) 5035. [32] D. Vogel, P. Krüger, J. Pollmann, Phys. Rev. B 52 (1995) 14316. [33] N.E. Sung, S.W. Kang, H.J. Shin, H.K. Lee, I.J. Lee, Thin Solid Films 547 (2012) 285. [34] Y. Tian, et al., Appl. Phys. Lett. 98 (2011) 162503. [35] G.L. Narendra, B. Sreedhar, J.L. Rao, S.V.J. Lakshman, J. Mater. Sci. 26 (1991) 5342. [36] X.F. Wang, Ms. D. Dissertation, Suzhou University, Suzhou, 2008, pp. p37. [37] X.Y. Cui, J.E. Medvedeva, B. Delley, A.J. Freeman, N. Newman, C. Stampfl, Phys. Rev. Lett. 95 (2005) 256404. [38] Y.F. Yan, M.M. Aljassim, S.H. Wei, Appl. Phys. Lett. 89 (2006) 181912. [39] R.S. Mulliken, J. Chem. Phys. 23 (2004) 1833. [40] F. Paraguay, D.W. Estrada, L.D.R. Acosta, N.E. Andrade, M. Miki-Yoshida, Thin Solid Films 350 (1999) 192. [41] J.E. Jaffe, J.A. Snyder, Z. Lin, A.C. Hess, Phys. Rev. B 62 (2000) 1660. [42] J.C. Garcia, et al., J. Appl. Phys. 100 (2006) 104103. [43] Y.C. Cheng, X.L. Wu, J. Zhu, L.L. Xu, S.H. Li, P.K. Chu, J. Appl. Phys. 100 (2008) 073707.
Contents lists available at ScienceDirect
Chemical Physics Letters journal homepage: www.elsevier.com/locate/cplett
Study of the effect of Cu heavy doping on band gap and absorption spectrum of ZnO Shaoqiang Guo a , Qingyu Hou a,∗ , Chunwang Zhao a , Yue Zhang b a b
College of Science, Inner Mongolia University of Technology, Hohhot 010051, China School of Material Science and Engineering, Beijing University of Aeronautics and Astronautics, Beijing 100083, China
a r t i c l e
i n f o
Article history: Received 13 June 2014 In final form 3 September 2014 Available online 10 September 2014
a b s t r a c t Contradictory experimental absorption spectra blue shift and red shift results have been reported in the literatures. To solve this problem, this study investigates the electronic structure and absorption spectra of Zn1−x Cux O (x = 0, 0.0313 and 0.0625, respectively) and Zn32 CuO32 supercells employing firstprinciples calculations with GGA+U method. By increasing the Cu substitutional doping concentration from 3.13% to 6.25%, the following results are obtained: increased magnetic properties, narrower band gaps, and a significant red shift in the absorption spectrum. These findings are in good agreement with the experimental results. The changes of band gap and absorption spectrum for interstitial doping and substitutional doping are opposite. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Zinc oxide is an important II–VI type semiconductor that possesses a direct wide band gap of 3.37 eV [1] and a high exciton binding energy of 60 meV [2] at room temperature. ZnO also has low dielectric constant, large electric-coupling ratio, high chemical stability, as well as excellent piezoelectric and photoelectric properties. This type of semiconductor has potential applications in short-wavelength optoelectronic devices, such as laser and light-emitting diodes. In experimental and theoretical studies, the structure of ZnO can be changed easily by effective alloying with a dopant, which involves ZnO material energy band engineering. In addition, by regulating the ZnO band gap, the performance of ZnObased optoelectronic devices can be improved. Therefore, doped ZnO has vast applications in the production of solar cells [3], liquid crystal displays [4], gas sensors [5], UV laser diodes [6], transparent conductive films [7], and diluted magnetic semiconductors [8]. Extensive works on photoelectrical properties of doped ZnO have been carried out. Kulyk et al. [9] prepared Cu-doped ZnO by radio-frequency magnetron sputtering; their results showed that the grain size and band gap decreased, and the absorption band edge shifted to low levels with increased Cu concentration. Wu et al. [10] prepared singly doped Co and Cu as well as co-doped
∗ Corresponding author. E-mail addresses: [email protected] (S. Guo), [email protected] (Q. Hou). http://dx.doi.org/10.1016/j.cplett.2014.09.005 0009-2614/© 2014 Elsevier B.V. All rights reserved.
ZnO thin films using the sol–gel process. They also investigated the influence of Co and Cu doping on the surface morphologies of ZnO films, and found blue double peaks at all ZnO thin film samples caused by transition of electrons from conduction band minimum (CBM) to zinc vacancy or from Zn interstitial to valence band maximum (VBM). However, the green peak is highly relevant to the oxygen slip formed by doping. Ma et al. [11] prepared Cudoped ZnO films by radio-frequency reactive magnetron sputtering and studied the microstructures and optical properties of doped systems. The results indicated that ZnO films exhibited a strong orientation toward the c-axis and a uniform grain size after Cudoping. Furthermore, the intensities of blue (∼485 nm) and green (∼527 nm) peaks showed significant increase after annealing. Fu et al. [12] investigated the photocatalytic performance of Cu-doped ZnO nanoparticles prepared using the sol–gel process. Theoretical calculations have also been performed to determine the photoelectrical properties of doped ZnO. For instance, Kong and Gong [13] calculated the band gap of Be substitutional and interstitial doping ZnO using the GGA method, and found that substitution of Zn atoms by Be could enhance the band gap of ZnBeO phases, while interstitial Be atoms would decrease the band gap. Ye et al. [14] calculated ferromagnetism of Cu-doped ZnO by the local density approximation (LDA) (Hedin-Lunquist form) and generalized gradient approximation (GGA) (Perdew–Burke–Ernzerhof form). Li et al. [15] investigated the magnetic and optical properties of doped ZnO nanosheets using the GGA+U method. Their results revealed that the replacement of two Cu atoms by two Zn atoms enhanced the magnetic properties of the doped
16
S. Guo et al. / Chemical Physics Letters 614 (2014) 15–20
Figure 1. The 2 × 2 × 4 supercells for (a) undoped ZnO, (b) Zn0.9687 Cu0.0313 O, (C) Zn0.9375 Cu0.0625 O, and (d) Zn32 CuO32 .
system. Moreover, the absorption spectrum exhibited a noticeable red shift with increased doping concentration. Kang et al. [16] analyzed the variations in the ferromagnetism of carrier doping in Zn1−x Cux O1−y Xy (X = N and F; x, y = 0.0277–0.0833) by the full-potential linear muffin-tin orbital (FP-LMTO) method. Their experimental results showed that Cu magnetic moments at low Cu concentrations increased and decreased by N and F doping, respectively. Meanwhile, Liu et al. [17] calculated the band gap variations and absorption spectrum of wurtzite Zn1−x Cox O using the GGA method, and found that the band gaps broadened and the optical absorption edge exhibited a significant blue shift given an atomic fraction of Co. Duan et al. [18] investigated the electronic structure and optical properties of ZnO doped with transition metals and N. Their calculations indicated that the co-doping of transition metals (Mn, Fe, Co, and Cu) and N favored the formation of p-type ZnO. Although good results on the effects of Cu doping on the optoelectronic properties of ZnO have been theoretically and experimentally achieved, speculations remain upon comparing the band gaps and absorption spectra of undoped and Cu-doped ZnO. Increased Cu substitutional doping from 3.13% to 6.25% widened the band gap and caused a blue shift on the absorption spectrum [19], which contradicted the findings of Ref. [20]. From crystal periodicity, we can obtain useful results by performing first-principles calculations of the band structure, band gap, and the absorption spectrum of Cu-doped ZnO, with a doping concentration range similar to that presented in previous works [19,20]. We believe that the results can contribute to better design and improved preparation of short-wavelength optoelectronic devices from Cu-doped ZnO.
2. Theoretical model and computational method 2.1. Theoretical model ZnO has a hexagonal wurtzite crystal structure that belongs to the space group P63 mc and C64v symmetry. The unit cell comprises two hexagonal close packed lattices along the c axis and sleeve; the lattice parameters are a = 0.3249 nm and c = 0.5205 nm. The 2 × 2 × 4 supercells for undoped ZnO, Zn0.9687 Cu0.0313 O, Zn0.9375 Cu0.0625 O and Zn32 CuO32 were applied to calculate the process of geometric optimization and energy. Three substitutional doping levels of 0%, 3.13%, and 6.25% with zero, one and two Zn atoms replaced by Cu atom, respectively, were examined.
Figure 2. The test of energy convergence as the cut-off varied for undoping ZnO.
Zn32 CuO32 is a system of one Cu atom interstitial doping. The atomic positions of Cu dopants are displayed in Figure 1. 2.2. Computational method Calculations were performed by CASTEP package [21–23] as implemented in the Materials Studio software. The software is based on the density functional theory (DFT) using the planewave ultrasoft pseudopotential method [24–26]. Perdew Burke Ernzerhof (PBE) scheme in the generalized gradient approximation (GGA) was used to treat the exchange-correlation energy. We first optimized the structure of undoped ZnO, Zn0.9687 Cu0.0313 O, Zn0.9375 Cu0.0625 O and Zn32 CuO32 and then calculate the electronic and optical properties. During optimization, Brillouin zone integrations were first performed with the special k-point method over a 4 × 4 × 1 Monkhorst-Pack mesh. Energy convergence during cut-off variation for undoping ZnO supercells was also tested (Figure 2). The total energy of the system begins to converge at the cutoff energy of 320 eV and presents good convergence at the cutoff energy of 380 eV. Therefore, cutoff energy at 380 eV is the most ideal for all calculation models. Second, the plane wave energy was set at 1.0 × 10−5 eV/atom, and the maximum tolerances of force, stress, and displacement were set at 0.3 eV/nm, 0.05 GPa and
S. Guo et al. / Chemical Physics Letters 614 (2014) 15–20
1.0 × 10−4 nm, respectively. The valence atomic configurations of Zn, Cu and O were 3d10 4s2 , 3d10 4s1 and 2s2 2p4 , respectively. The effects of spin polarization were considered in energy calculations. Because of the GGA failed to properly describe the position of ZnO band narrowing, which corresponded to high localization in contrast to s and p bands. Hence, an orbital-dependent potential was used, including an additional Coulomb interaction U (GGA+U) [27–31]. Using GGA+U and choosing a reasonable value for U are both very important, because U acts as a fitting parameter to reproduce the experimentally observed position including the p states and d states. In the calculations, Ud,Zn , Up,O , and Ud,Cu were set at 5.5, 8.0 and 6.0 eV, respectively, this will be explained in the band gap analysis.
3. Results and discussion
17
3.2. Mulliken bond population and bond length analysis Mulliken bond population and bond length of undoped and Cu substitutional doped ZnO were calculated to analyze the doping mechanism of the latter (Table 1). Bond population assesses the ionic or covalent bonds [39], and indicates the distribution of electron population around atoms. Positive and negative values of the bond population imply covalent and ionic bonding, respectively. Given the increased Cu substitutional doping concentration, the Mulliken bond populations and bond lengths of Cu O parallel and vertical to c-axis are lower than those of Zn O in undoped ZnO cells. Apart from the influence of bond population, the replacement of Cu2+ with Zn2+ decreases that bond lengths of Cu O in the doped system. Moreover, increased Cu-doping concentration weakens the covalent bond, strengthens the ionic bond, and destabilizes the substitutional doped system. The results are in agreement with those of the formation energies.
3.1. Crystal structure and stability analysis 3.3. Orbital average charge analysis The lattice parameters, volumes, equivalent total energies, and formation energies of Zn1−x Cux O (x = 0, 0.0313 and 0.0625, respectively) and Zn32 CuO32 were obtained following geometry optimization. The calculated results are shown in Table 1. The optimized geometry lattice parameters of undoped and Cu substitutional doped ZnO are close to experimental values [32,33]. The Mulliken method was used to calculate the orbital charge. Results show that at 3d10 4s1 valence atomic configuration for Cu in Zn0.9687 Cu0.0313 O and Zn0.9375 Cu0.0625 O, the charge transfer sum of the s and d states is close to 2. Therefore, the valence of Cu is +2 and Cu2+ ion exists in the doped system, which are consistent with the experimental results [34]. Equivalent volumes of the crystal cell decrease with increasing Cu concentration, which is due to the substitution of Cu2+ for Zn2+ . The radius of Zn2+ (0.074 nm [35]) is larger than that of Cu2+ (0.072 nm [35]). As the volume of the Cu atom interstitial doping increases, due to the extremely short distance between Cu and the neighboring Zn atom, their repulsive interaction also increases. Experimental studies reveal that the transmittance of the doping system is maintained at 60% when the Cu-doping concentration reaches 14.5% [36]. The crystal undergoes phase transformation for doping concentrations greater than 14.5%. In this study, Cu-doping concentrations ranged between 3.13% and 6.25%. This selection meets the crystal into hexagonal wurtzite structure for doping concentrations do not beyond 14.5%. In addition, the inconsistencies in the absorption spectra for various studies [19,20] are resolved. Impurity formation energy (Ef ) analyzes the relative degree of difficulty of atomic doping and illustrates the influence of Cu doping on the stability of wurtzite ZnO. Ef is given by Eq. (1) [37] as: Ef = EZnO:Cu − EZnO − ECu + EZn
(1)
where EZnO:Cu and EZnO are the total energies of Cu-doped ZnO and undoped ZnO, respectively for supercells similar sizes; and ECu and EZn are the respective energies of bulk Cu and Zn metals. Table 1 illustrates the calculated results for undoped and substitutional doping systems, indicating that the total energies and energies for impurity formation increase, and the doping becomes increasingly difficult with increased Cu content. Therefore, the substitutional doping system structure becomes unstable. Meanwhile, impurity formation energy of the Cu interstitial doping system is higher than that of Cu substitutional doping system with a similar doping concentration. Doping becomes more difficult, and the interstitial doping system structure becomes highly unstable. These results are in good agreement with previously published theoretical results [38].
The Mulliken atomic populations method was used to calculate the orbital average charges of O, Zn, and Cu atoms in undoped ZnO, Zn0.9687 Cu0.0313 O and Zn0.9375 Cu0.0625 O, respectively (Table 2). The calculation was performed to support the analysis of the mechanism of the partial density of states of Cu substitutional doped ZnO. The results show that the orbital charges of O-2p, Cu-3p and Zn-3p are reduced; the orbital charges of Zn-3d and Cu-4s remain unchanged; and the orbital charge of Zn-4s increases. 3.4. Analysis of electron density differences of undoped and Cu substitutional doped ZnO Paraguay et al. [40] prepared ZnO thin films by spray pyrolysis technique using an aqueous medium. The XRD results showed that the films were polycrystalline with preferred orientation along the (0 0 2) plane. Electron density differences at (0 0 2) for undoped and Cu substitutional doped ZnO were calculated to investigate the interactions and bonding among adjacent atoms in the Cu substitutional doped system (Figure 3a–c). The extent of overlapping among the neighboring electronic clouds of Zn O in undoped ZnO becomes noticeable as the electronic communication strengthens. By contrast, the extent of the overlapping neighboring electronic clouds of Cu O and Zn O in the Cu substitutional doped system weakens with increasing Cu-doping concentration. In other words, the covalent and ionic bonds weaken and strengthen, respectively. The results are in good agreement with those of the orbital average charges. 3.5. Band gaps analysis We set Ud,Zn = 5.5 eV and Up,O = 8.0 eV for all systems in our calculation using the GGA+U method. First, the band structure of undoped ZnO supercell was determined (Table 3). The band gap for undoped ZnO is 3.44 eV, which is in good agreement with the experimental values [1,20]. Hence, the values of Ud,Zn and Up,O in undoped ZnO set at 5.5 and 8.0 eV, respectively, were reasonable. We selected 5.00, 6.00, and 7.00 eV for Ud,Cu to obtain the theoretical band gaps of the substitutional doped system coinciding with the experiment. In addition, the band gaps for Zn0.9687 Cu0.0313 O and Zn0.9375 Cu0.0625 O were calculated. The results are shown in Table 3. The data show that the relative band gap values do not change much in Zn0.9687 Cu0.0313 O and Zn0.9375 Cu0.0625 O when the Ud,Zn , Up,O , and Ud,Cu are set at 5.50, 8.00 and 6.00 eV, respectively. Therefore, Ud,Cu at 6.00 eV is most ideal for use in the compromise method. In addition, the lattice constants of the a and c axes for the Cu substitutional doping system are reduced with increasing
18
S. Guo et al. / Chemical Physics Letters 614 (2014) 15–20
Table 1 The physical parameters of Zn1−x Cux O (x = 0, 0.0313, 0.0625) and Zn32 CuO32 . Model
ZnO Zn0.9687 Cu0.0313 O Zn0.9375 Cu0.0625 O Zn32 CuO32
a, b, c (nm)
V (nm3 )
E (eV)
Ef (eV) This work
This work
Experimental values
This work
This work
a = 0.3249 c = 0.5205 a = 0.3248 c = 0.5204 a = 0.3247 c = 0.5203 a = 0.3300 c = 0.5338
a = 0.3250[32] c = 0.5210[32] c = 0.5202[33]
0.04759
−4294.45
0.04754
−4279.88
−2.54
c = 0.5199 [33]
0.04751
−4265.15
−1.28
–
0.05034
−4381.61
4.29
Bond
Zn Zn Cu Cu Cu Cu –
O (||c) O (⊥c) O (||c) O (⊥c) O (||c) O (⊥c)
Population
Bond length (nm)
This work
This work
0.39 0.43 0.37 0.42 0.35 0.41 –
0.2004 0.2014 0.1992 0.1973 0.1991 0.1972 –
Table 2 Orbital charges of doped system. Model
O-2p/e
Zn-4s/e
Zn-3p/e
Zn-3d/e
Cu-4s/e
Cu-3p/e
Zn0.9687 Cu0.0313 O Zn0.9375 Cu0.0625 O
5.128 5.118
0.373 0.379
0.688 0.670
9.980 9.980
0.560 0.560
0.640 0.630
Figure 3. Electron density differences in (0 0 2) for (a) undoped ZnO, (b) Zn0.9687 Cu0.0313 O, and (C) Zn0.9375 Cu0.0625 O. Table 3 The band gap of before and after ZnO supercells.
ZnO Zn0.9687 Cu0.0313 O
Zn0.9375 Cu0.0625 O
Zn32 CuO32
[20]
ZnO Zn0.98 Cu0.02 O Zn0.94 Cu0.06 O
Ud,Zn (eV)
Up,O (eV)
Ud,Cu (eV)
Eg (eV)
Eg (eV)
5.50 5.50 5.50 5.50 5.50 5.50 5.50 5.50 5.50 5.50
8.00 8.00 8.00 8.00 8.00 8.00 8.00 8.00 8.00 8.00
– 5.00 6.00 7.00 5.00 6.00 7.00 5.00 6.00 7.00
– 0.051 0.052 0.047 0.072 0.075 0.075 0.133 0.114 0.112
3.44 3.26 3.25 3.25 3.14 3.18 3.23 3.49 3.50 3.48
– – –
– – –
– – –
– – –
3.47 3.45 3.43
Cu concentration (Table 1), which is due to the substitution of Cu2+ for Zn2+ . This finding is consistent with the change trend of experimental results [31,32]. As the lattice constants of the a and c axes for Cu interstitial doping system increase, because of the extremely short distance between Cu and the neighboring Zn atom, their repulsive interaction increases. Theoretical calculation [41] has proven that Pulay stress has an impact on the band gap of
ZnO. Compressive strain and tensile strain can enhance and reduce the band gap of ZnO, respectively. Under the condition of the same strain with undoped ZnO, we obtain the band gap variable Eg , which represents the strain that causes the band gap to increase and then decrease after Cu substitutional doping (Table 3). As can be seen from Table 3, given that the band gap variable for Cu substitutional doping that causes the band gap to decrease is
S. Guo et al. / Chemical Physics Letters 614 (2014) 15–20
19
considerably greater than the amount of strain that causes the band gap to increase, the band gaps of all substitutional doping models significantly decreased compared with the band gap of undoped ZnO. Moreover, a heavier Cu amount results in a narrower band gap. This result is consistent with the trend reported in the literature [20]. The narrow mechanisms of the band gap were analyzed based on orbital theory. While the changes of band gap for interstitial doping are opposite to that for substitutional doping. In addition, the band gaps of all substitutional doping models significantly decrease compared with the band gap of undoped ZnO; moreover, a heavier Cu amount results in a narrower band gap (Table 3). This result is consistent with the trend reported in the literature [20]. The narrow mechanisms of the band gap were analyzed based on orbital theory. 3.6. Analysis of partial density of states of undoped and Cu substitutional doped ZnO Figure 4a–c show the partial densities of states of undoped ZnO, Zn0.9687 Cu0.0313 O, and Zn0.9375 Cu0.0625 O supercells. Chemical orbital theory reveals that the band gap of undoped ZnO is formed by the respective antibonding-like and bonding-like states of s and p orbitals from the electronic interaction of Zn-4s and O-2p states (Figure 4a). CBM significantly shifts toward low energies (Figure 4b and c). The interactions of s–s states shift the conduction band (CB) toward high energies, whereas the repulsion effects of s–p states shift the CB toward low energies for Cu-doped ZnO. Given the increase in Cu content, the interactions of s-s state increase with the orbital charges of Zn-4s and the repulsion effects of s–p states decrease with the orbital charges of Zn-3p and Cu-3p. Hence, the CBM shifts to low energies. Meanwhile, the repulsion effects of p–d states shift the VB toward high energies, and the interactions of s–p states shift the VB shift toward low energies. Considering the increase in Cu-doping concentration, the interactions of p–p states decrease with the orbital charges of Cu-3p and O-2p, whereas the repulsion effects of p–d states decrease with the orbital charges of Zn-3p and Cu-3p. Therefore, the interaction size of p–p is similar to that of p–d, and the interaction direction of p–p is opposite to that of p–d. Hence, the changes in VB position are minimal. Given the increase in Cu-doping concentration, a few changes in the VB position are observed. The CB significantly shifts toward low energies; hence, the band gaps of the substitutional doped system show significant decrease, which is consistent with the result of the band structure. 3.7. Magnetic property analysis Figure 4b and c reveals the asymmetry of the total density of state, and that the number of up spins is greater than that of down spins. Zn0.9687 Cu0.0313 O and Zn0.9375 Cu0.0625 O supercells generate magnetic moments of 1.06 B and 2.23 B , respectively, indicating that increased Cu-doping concentration enhances the magnetic property of the supercells. This result agrees with that presented in a previous study [34]. Figure 4b and c imply that the spontaneous spin polarization originates from the p–d exchange between Cu-3d and O-2p orbitals of the Cu-doped ZnO, which is in good agreement with previously reported experimental result [34].
Figure 4. Densities of states of (a) undoped ZnO, (b) Zn0.9687 Cu0.0313 O, and (c) Zn0.9375 Cu0.0625 O.
calculated according to the dispersion relation of Kramers–Kronig. In this way, absorption coefficient ˛(ω), can be obtained. Omitted the derivation process [42], the concerned formulas are as following. ε2 (ω) =
As we know, the optical properties of the medium can be described by complex dielectric response function ε(ω) = ε1 (ω) + iε2 (ω) in leaner response range, in which ε1 = n2 − k2 and ε2 = 2nk. The real part ε1 (ω) and imaginary part ε2 (ω) can be
V,C
ε1 (ω) = 1 =
˛(ω) = 3.8. Absorption spectrum analysis
c ω2
√
2 0
2
BZ
(2)3
∞
0
|MCV (k)|2 · ı(ECk − EVk − ω)d3 k
ω ε2 (ω) ω 2 − ω2
dω
2[
ε21 (ω) + ε22 (ω) − ε1 (ω)]
(2)
(3) 1/2
(4)
where, subscript C and V represent conduction band and valence band, respectively, BZ is first Brillouin zone, k is reciprocal lattice vector, |MCV (k)|2 is momentum matrix element, c is constant, ω is angular frequency, ECk and EVk is intrinsic energy level. The above formulas provide a theoretical foundation for the analysis of the band
20
S. Guo et al. / Chemical Physics Letters 614 (2014) 15–20
experimental results [20]. The changes of band gap and absorption spectrum for interstitial doping and substitutional doping are opposite. These results can help improve methods by which to design and prepare short-wavelength optoelectronic devices from Cu-doped ZnO. Acknowledgments All the calculations were carried out at the Network Information and Computing Center of Beijing University of Aeronautics and Astronautics. This work was supported by the Natural Science Foundation of China (Grant nos. 61366008 and 51261017). This work was also supported by the Ministry of Education “Spring sunshine” plan funding, and the College Science Research Project of Inner Mongolia Autonomous Region (Grant no. NJZZ13099). References Figure 5. Optical absorption spectra of undoped and Cu-doped ZnO.
structure and optical properties of the crystals, which reflect the emitting mechanism of the spectrum generated by the electronic transition between energy level [43]. The absorption spectra of undoped and Cu-doped ZnO samples were calculated using GGA+U method (Figure 5). The absorption capacity of ZnO in the UV light area increases and the absorption spectrum exhibits a red shift with Cu atom substitutional doping. This red shift can be attributed to the lower energy of the electronic transition between O2p and Cu4s than that between O2p and Zn4s, with increasing amount of substitutional doping. It can be seen from band gaps analysis, the band gaps of undoped ZnO, Zn0.9687 Cu0.0313 O, and Zn0.9375 Cu0.0625 O were 3.44, 3.25 and 3.18 eV, respectively. When these values are substituted in the absorption long-wave limit formula, = (1.24/Eg (eV)) m, the calculated wavelengths are 360.5, 381.5, and 389.9 nm, respectively. For doping amounts ranging from 3.13% to 6.25%, the absorption spectra in long-UV light red shift become significant with increasing substitutional doping amount compared with undoped ZnO. The absorption band-edge shifts toward low energies. These results are consistent with the trend in experimental results [20]. While absorption spectrum of Cu atom interstitial doping exhibits a blue shift. These can help improve methods in designing and preparing long-UV light optoelectronic devices from Cu-doped ZnO. 4. Conclusion The band structures, density of states, magnetic properties, and absorption spectra of Zn1−x Cux O (x = 0, 0.0313 and 0.0625, respectively) and Zn32 CuO32 supercells are calculated using the first-principles plate-wave, ultrasoft pseudopotential method based on density functional theory. Calculation results show that the band gaps of undoped and Cu-doped ZnO coincide with the experimental values obtained through GGA+U method with a reasonable U. Increased Cu-doping concentrations from 3.13% to 6.25% reduce the volumes, increase the total and formation energies, and destabilize the substitutional doped systems. Moreover, the covalent bond weakens, the ionic bond strengthens, the magnetic property enhances, and doping becomes increasingly difficult. The band gaps become narrower and the absorption spectrum exhibits a significant red shift, which are in good agreement with
[1] V. Srikant, D.R. Clarke, J. Appl. Phys. 83 (1998) 5447. [2] Z.K. Tang, G.K.L. Wong, P. Yu, M. Kawasaki, A. Ohtomo, H. Koinuma, Y. Segawa, Appl. Phy. Lett. 72 (1988) 3270. [3] Q. Zhang, C.S. Dandeneau, X. Zhou, G. Cao, Adv. Mater. 21 (2009) 4087. [4] B.Y. Oh, M.C. Jeong, T.H. Moon, W. Lee, J.M. Myoung, J.Y. Hwang, D.S. Seo, J. Appl. Phys. 99 (2006) 124505. [5] H. Gong, J.Q. Hu, J.H. Wang, C.H. Ong, F.R. Zhu, Sens. Actuators B: Chem. 115 (2006) 247. [6] K.W. Liu, et al., Solid-State Electron. 51 (2007) 757. [7] X. Jiang, F.L. Wong, M.K. Fung, S.T. Lee, Appl. Phys. Lett. 83 (2003) 1875. [8] H.J. Lee, S.Y. Jeong, C.R. Cho, C.H. Park, Appl. Phys. Lett. 81 (2002) 4020. [9] B. Kulyk, V. Figà, V. Kapustianyk, M. Panasyuk, R. Serkiz, P. Demchenko, Acta Phys. Pol. A 123 (2013) 92. [10] D.C. Wu, et al., Acta Phys. Sin. 58 (2009) 7261. [11] L. Ma, S. Ma, H. Chen, X. Ai, X. Huang, Appl. Surf. Sci. 257 (2011) 10036. [12] M. Fu, Y. Li, S. Wu, J. Lin, F. Dong, Appl. Surf. Sci. 258 (2011) 1587. [13] F.T. Kong, H.R. Gong, Comput. Mater. Sci. 61 (2012) 127. [14] L.H. Ye, A.J. Freeman, B. Delly, Phys. Rev. B 73 (2006) 033203. [15] F. Li, C.W. Zhang, M.W. Zhao, Physica E 53 (2013) 101. [16] B.S. Kang, K.S. Kim, S.C. Yu, H. Chae, J. Solid State Chem. 198 (2013) 120. [17] Y. Liu, Q.Y. Hou, H.P. Xu, C.W. Zhao, Y. Zhang, Chem. Phys. Lett. 551 (2012) 72. [18] M.Y. Duan, M. Xu, H.P. Zhou, Y.B. Shen, Q.Y. Chen, Y.C. Ding, W.J. Zhu, Acta Phys. Sin. 56 (2007) 5359. [19] T.M. Hammad, J.K. Salem, R.G. Harrison, R. Hempelmann, N.K. Hejazy, J. Mater. Sci.: Mater. Electron. 24 (2013) 2846. [20] S. Muthukumaran, R. Gopalakrishnan, Opt. Mater. 34 (2012) 1946. [21] V. Milman, et al., J. Mol. Struc. Theochem 954 (2010) 22. [22] L.B. Shi, C.Y. Xu, H.K. Yuan, Physica B 406 (2011) 3187. [23] Y. Imai, M. Mukaida, T. Tsunoda, Thin Solid Films 381 (2001) 176. [24] D. Vanderbilt, Phys. Rev. B 41 (1990) 7892. [25] K.Z. Ska, J. Thin Solid Films 391 (2001) 229. [26] Y. Yanfa, M.M. AL-Jassim, Phys. Rev. B 69 (2004) 085204. [27] S. Amari, S. Méc¸abih, B. Abbar, N. Benosman, B. Bouhafs, Physica B 407 (2012) 3639. [28] X.G. Xu, D.L. Zhang, Y. Wu, X. Zhang, X.Q. Li, H.L. Yang, Y. Jiang, Rare Metals 31 (2012) 107. [29] N. Ghajari, A. Kompany, T. Movlarooy, F. Roozban, M. Majidiyan, J. Magn. Magn. Mater. 325 (2013) 42. [30] G. Chris, Van de Walle, A. Janotti, Phys. Status Solidi B 248 (2011) 1. [31] D.M. Song, T.H. Wang, J.C. Li, J. Mol. Model. 18 (2012) 5035. [32] D. Vogel, P. Krüger, J. Pollmann, Phys. Rev. B 52 (1995) 14316. [33] N.E. Sung, S.W. Kang, H.J. Shin, H.K. Lee, I.J. Lee, Thin Solid Films 547 (2012) 285. [34] Y. Tian, et al., Appl. Phys. Lett. 98 (2011) 162503. [35] G.L. Narendra, B. Sreedhar, J.L. Rao, S.V.J. Lakshman, J. Mater. Sci. 26 (1991) 5342. [36] X.F. Wang, Ms. D. Dissertation, Suzhou University, Suzhou, 2008, pp. p37. [37] X.Y. Cui, J.E. Medvedeva, B. Delley, A.J. Freeman, N. Newman, C. Stampfl, Phys. Rev. Lett. 95 (2005) 256404. [38] Y.F. Yan, M.M. Aljassim, S.H. Wei, Appl. Phys. Lett. 89 (2006) 181912. [39] R.S. Mulliken, J. Chem. Phys. 23 (2004) 1833. [40] F. Paraguay, D.W. Estrada, L.D.R. Acosta, N.E. Andrade, M. Miki-Yoshida, Thin Solid Films 350 (1999) 192. [41] J.E. Jaffe, J.A. Snyder, Z. Lin, A.C. Hess, Phys. Rev. B 62 (2000) 1660. [42] J.C. Garcia, et al., J. Appl. Phys. 100 (2006) 104103. [43] Y.C. Cheng, X.L. Wu, J. Zhu, L.L. Xu, S.H. Li, P.K. Chu, J. Appl. Phys. 100 (2008) 073707.