Effects of Mg substitution on the structural, optical, and electrical properties of CuAlO2 thin films
H.F. Jiang,X.B. Zhu,H.C. Lei,G. Li,Z.R. Yang,W.H. Song,J.M. Dai,Y.P. Sun,Y.K. Fu
DOI: https://doi.org/10.1016/j.jallcom.2010.10.036
IF: 6.2
2011-01-01
Journal of Alloys and Compounds
Abstract:Mg-doped CuAlO 2 thin films are prepared by the chemical solution method. The XRD results show that the solid solubility of Mg species on Al sites in CuAlO 2 lattice is lower than 2 at.%. When less than 2 at.% of Mg is added to the CuAlO 2 film, the surface roughness of the films was reduced with Mg substitution. Moreover, the c -axis orientation of the films improves because the in-plane fusion between CuAlO 2 crystallites is hindered. Optical and electrical measurements show that substituting Al 3+ in the films with Mg 2+ increases both their transmittance in the visible region and their optical band gaps. As well, their electrical conductivity is enhanced. At 300 K, the conductivity of the 1 at.% Mg-doped sample is up to 5.2 × 10 −3 S/cm. Thus, Mg-doped CuAlO 2 films may have potential applications as transparent conductive oxides. Keywords CuAlO 2 Delafossite Optical property Electrical property Chemical solution deposition 1 Introduction Transparent conducting oxides (TCOs) have attracted much attention because of their functional combination of transparency and electrical conductivity. In general, most of the known TCOs are n-type materials. N-type materials, such as tin oxide, zinc oxide, indium oxide, and cadmium oxide, were popularly used in optoelectronics before 1997 [1–4] . For the use in transparent electronics technology, a p–n junction is required. Therefore, p-type TCO is highly sought after. p-Type CuAlO 2 thin films were first fabricated in 1997 [5] . Since then, many attempts have been made to grow CuAlO 2 thin films with high performance (high conductivity and high transparency). Various production methods, such as pulsed laser deposition (PLD) [6] , RF magnetron sputtering [7] , chemical vapor deposition [8,9] , spray pyrolysis [10,11] , and sol-gel processes [12–14] , among others, have been reported. Unfortunately, some difficulties, including high crystallization temperatures, secondary phase introductions, low transmittance, and low conductivity were encountered with these techniques. These problems have limited the applications of CuAlO 2 thin films in photovoltaic cells, gas sensors, light-emitting diodes, transistors, and so on. CuAlO 2 has an ABO 2 delafossite structure (A = Cu, Ag, Pd, Pt; B = Sc, Cr, Fe, Y, La, Al, Ni, Rh, Co). In this structure, monovalent cation A is coordinated with two oxygen atoms along the c -axis, forming an O–Cu–O dumbbell structure. Trivalent cation B is coordinated with six oxygen atoms, forming an edge-sharing BO 6 octahedron. The Cu + layers and B 3+ O 2 layers alternately stack along the c -axis. Recent studies presented the O–Cu–O dumbbell layers are important for hole transport and oxygen intercalation [15,16] . As such, the substitution of bivalent cations in B sites, such as that in CuSc 1− x Mg x O 2 [17] , CuCr 1− x Mg x O 2 [18] , CuRh 1− x Mg x O 2 [19] , CuY 1− x Ca x O 2 [20] , CuIn 1− x Ca x O 2 [21] , and CuFe 1− x Ni x O 2 [22] , is generally considered a direct and effective route to increasing p-type electrical conductivity because this substitution can induce the change of Cu valence state from +1 to +2 in the O–Cu–O dumbbell layers according to the valence equilibrium in defect chemistry and then realize the hole hopping from the Cu 2+ to Cu + . It can be expressed as C u + + A l 3 + → C u + + M g Al 2 + + h → C u 2 + + M g Al 2 + , where M g Al 2 + and h denote the Mg 2+ cation on the Al 3+ site and a positive hole, respectively. To date, a large number of experiments have been devoted to research on the effects of Mg substitution in isostructure CuCrO 2 semiconductors. For example, Sadik et al. reported the electrical transport of CuCr 1− x Mg x O 2 thin films grown by pulsed laser deposition, and found that the optimal atomic ratio of Mg to Cr was 3 at.% [23] . Nagarajan et al. investigated the p-type conductivity of Mg-doped CuCrO 2 oxide thin films deposited using RF sputtering; they found that the optimal value for doping was 5 at.% [24] . Through solid state reactions, Maignan et al. found that the solubility of Mg 2+ in CuCrO 2 compounds was not over 1 at.% [25] . These studies show that the doping of the same cation in CuCrO 2 materials could be influenced by the preparation methods of the material. To the best of our knowledge, only Dong et al. [26] reported on Mg substitution in CuAlO 2 thin films deposited using the RF magnetron sputtering technique. In their experiments, polycrystalline targets of CuAl 1− x Mg x O 2 ( x = 0, 0.01, 0.02 and 0.05) used for sputtering were synthesized by heating a stoichiometric mixture of Cu 2 O (99%), Al(OH) 3 (99.9%) and MgO (99.9%) at 1373 K for 10 h. They found that the solubility of Mg species in the CuAlO 2 lattice is up to 5 at.%, and the crystalline quality of the films declines with increasing Mg concentration. In this paper, we report the preparation of CuAl 1− x Mg x O 2 thin films by the chemical solution deposition (CSD) method, and investigate the effects of Mg substitution on the structural, optical and electrical properties of the CuAlO 2 thin films produced. 2 Experimental Four solutions of CuAl 1− x Mg x O 2 ( x = 0, 0.005, 0.01 and 0.02) used for spin-coating were prepared by dissolving mixtures of Cu(CH 3 COO) 2 ·H 2 O (99%), Al(NO 3 ) 3 ·9H 2 O (99%), and Mg(NO 3 ) 2 ·6H 2 O (99%) in CH 3 CH(OH)CH 3 (99.7%). The molar ratio of Mg to Cu in each solution was determined by the x value and then the molar ratio of Al species was equal to the (1 − x ) value. The mixtures were stirred for 4 h in order to obtain well-mixed solutions. The concentrations of these homogenous solutions were 0.2 M. Before deposition, sapphire substrates were cleaned sequentially with acetone, ethanol, and deionized water in an ultrasonic cleaner. Droplets of these solutions were spin-coated onto (0 0 1) sapphire substrates at a speed of 6000 rpm for 20 s. The deposited films were preheated for 20 min at 400 °C in a quartz tube furnace to evaporate the solvent and remove organic residues. To obtain the desired thickness, spin-coating and preheating were repeated six times. Finally, the deposited films were annealed for 6 h at 900 °C under ultra-pure nitrogen gas at a flow rate of 150 mL/min. According to Ref. [27] , CuAlO 2 is not thermodynamically stable at 900 °C in air. Thus, nitrogen or argon must be used as an annealing gas to obtain the pure delafossite phase. After annealing, the furnace was allowed to cool down naturally to room temperature. In this paper, the thickness of all the obtained films was about 200 nm ( Fig. 4 ). Film crystallinity was examined using a grazing incidence X-ray diffractometer (GIXRD, Philips designed X’Pert Pro type) (grazing angle α = 1°). Field emission-scanning electron microscope (FE-SEM) was used to determine surface patterns. Optical transmittance was recorded using a UV–vis-NIR spectrophotometer (CARY-5E) at a fixed incidence angle perpendicular to the film surface and within the range of 200–800 nm. Uncoated substrate was used as a reference for transmittance measurements. In the inset of Fig. 7 , the preparation of the CuAl 1− x Cr x O 2 thin films followed the above mentioned CSD route for comparison. DC resistivity was measured using the four-probe configuration. Silver paste was used to create good ohmic contacts. Hall measurements were carried out at room temperature using the van der Pauw method. 3 Results and discussion Fig. 1 shows the GIXRD patterns for the synthesized CuAl 1− x Mg x O 2 ( x = 0–2%) samples. For the sake of description, we define here the CuAl 1− x Mg x O 2 films with x = 0%, 0.5%, 1% and 2% as the sample Mg 0 , Mg 0.5 , Mg 1 and Mg 2 , respectively. When the amount of Mg substitution is below 2 at.%, the obtained films are phase-pure and the structure can be indexed to delafossite with a space group of R -3 m . This result for solid solubility differs from the result reported by Dong et al. [26] . However, the Mg 2+ ionic radius (0.66 Å) is almost equal to that of the Cr 3+ radius (0.63 Å), and the Mg content in CuCr 1− x Mg x O 2 thin films derived from various preparation methods are not over 5 at.% under the premise of a single delafossite phase. Furthermore, the CSD method is a novel route used to obtain thin films with compositional homogeneity. Further work is necessary to clarify such inconsistencies between studies. Each peak in Fig. 1 is indexed according to standard powder data (JCPDS, No. 35-1401). When x = 2%, the diffraction peaks corresponding to the delafossite structure are strongly suppressed, while a peak at 42.3° is strengthened and another peak at 61.3° appears. The peak at 42.3° is not believed to be a sign of (0 1 4) preferred orientation growth. Instead, a new phase of MgO is thought to have been produced, especially when the peak at 61.3° is taken into account. With increasing x, the c -axis orientation of the samples Mg 0 , Mg 0.5 and Mg 1 increases sequentially. This result also differs from that found by Dong et al. [26] . Here, a possible interpretation of the variation of the c -axis orientation is presented. The primary characteristic of the chemical solution deposition process is that CuAlO 2 crystal nuclei form dispersively, accompanying both the decomposition and evaporation of chemical organics. These crystal nuclei on the Al 2 O 3 (0 0 1) substrate were all expected to be of c -axis orientation because of the match of both the crystal symmetry (rhombohedral crystal system) and the lattice parameter (the O–O distances for CuAlO 2 and (0 0 1) Al 2 O 3 are 2.86 and 2.70 Å, respectively) between the CuAlO 2 crystallite and the Al 2 O 3 (0 0 1) substrate. However, during annealing, these crystal nuclei grew up and fused with each other in the in-plane direction, resulting in c -axis degeneration. This is why thicker films have worse c -axis orientation. Mg introduction causes difficulties in the in-plane fusion between crystallites. Thus, the growth of the crystallites tends to become out-of-plane ( c -axis). With increasing Mg content, the enhancement of c -axis orientation appears. As a layered semiconductor possessing anisotropic conductivity [28] , the texture growth of the three films favors increases in p-type electrical conductivity. The Mg 2+ ionic radius is 0.66 Å while that for Al 3+ is 0.51 Å. Due to the mismatch between the ionic radii, the substitution of Mg 2+ ions in Al 3+ sites induces lattice distortion. This is clearly seen in the magnified XRD patterns of the (0 0 6) peak, displayed in Fig. 2 (a) . An oblique solid line, which reflects the shift of the (0 0 6) peaks, is also shown. The shift to lower angles occurs with increasing x. This finding is highly consistent with the variation of lattice parameters shown in Fig. 2 (b). Using the Scherrer formula ( D = 0.9 λ /( β cos θ )) and the FWHM of the (0 0 6) peak, the crystallite sizes for the samples Mg 0 , Mg 0.5 and Mg 1 are calculated to be 29.4, 30.1, and 33.8 nm, respectively. Fig. 3 shows the morphology of the obtained CuAl 1− x Mg x O 2 thin films. A needle-like surface forms for the undoped Mg 0 film. Judging from the colors of the grains, the film seems to consist of several phases. However, the XRD patterns indicate that the film is composed of pure delafossite phases. Therefore, the difference in the colors could be due to the high surface roughness present in the sample. When the amount of Mg is increased to 0.5 at.%, the surface morphology of the Mg 0.5 film significantly improves. Compared to the undoped film, Mg 0.5 has a much smoother surface except for some holes. When the Mg content is further increased to 1 at.%, the Mg 1 film becomes smoothest and most compact. The morphological evolution of the three films ( x < 2 at.%) illustrates the improvement of surface roughness by Mg substitution. When x is equal to 2 at.%, bigger grains are extruded on the surface of the film, causing an inhomogeneous morphology. Initially, it was assumed that larger grains were associated with MgO impurity phases, and this hypothesis was expected to be verified by energy dispersive spectral measurements. However, due to issues regarding the thickness of the films (∼200 nm) and the penetration depth of the X-ray used (much higher than 200 nm), the obtained results (not shown here) were undesirable. Thus, the origin of the larger grains remains unclear. Fig. 5 illustrates the transmittance of CuAl 1− x Mg x O 2 thin films in the range of 200–800 nm. The transmittance of the samples Mg 0 , Mg 0.5 , Mg 1 and Mg 2 at 500 nm are 71.1%, 74.9%, 74.8%, and 48.7%, respectively. Such high values for the three films ( x < 2 at.%) produced by the CSD method are comparable to those obtained by physical methods, such as PLD and RF sputtering. In the visible region, the transmittance of the Mg-doped films is a little higher than that of the undoped film. Given the surface morphology of these films, the improvement in their optical transparency could be due to decreased surface scattering ( Fig. 3 ). Between 350 and450 nm, the plots of the films, except for the sample Mg 2 , become steeper with increasing x. This reflects improvements in the crystallinity of the samples with increasing Mg substitution [29] , which is highly consistent with the XRD results in Fig. 1 . Tauc's relation, expressed as ( αhγ ) 2 = A ( hγ − E g ) [30] , was used to calculate the optical band gaps of the CuAl 1− x Mg x O 2 thin films. The intersection point of each linear fit to hγ -coordinate corresponds to the optical band gap of each sample. In Fig. 6 , the band gaps of the samples Mg 0 , Mg 0.5 and Mg 1 are 3.32, 3.51, and 3.58 eV, respectively. The change in optical band gaps could not be interpreted using Burstein-Moss shift or quantum size effect according to the above calculations for crystallite sizes [31] . The optical anomaly in 13 group delafossites of Cu 1+ M 3+ O 2 , such as CuAlO 2 , CuGaO 2 , and CuInO 2 , has remained debatable [32] . If the simple consideration that direct optical band gaps increase with increasing M 3+ ionic radius, i.e., gap CuAlO 2 < gap CuGaO 2 < gap CuInO 2 , is assumed, then the increase in the optical band gaps of the CuAl 1− x Mg x O 2 thin films could be simply attributed to Mg substitution as an effect of increasing M 3+ ionic radius. The sample Mg 2 does not have an optical band gap, suggesting that this film loses the function of optical window which means selective transmission for incident light. The optical transmittance of the films in the ultraviolet region (175–325 nm) provides information regarding the structure of the valence band maximum (VBM). Studies involving this region have not yet been reported in relevant publications. Enlarged plots of this region are shown in Fig. 7 . The undoped sample Mg 0 has higher optical transmittance, while the transmittance of the sample Mg 0.5 is lowered integrally on the premise of keeping the curve unchanged. This implies the increase of electron occupation in the VBM. However, the curve of the sample Mg 1 differs significantly from the two other samples (Mg 0 and Mg 0.5 ). The experimental results are entirely different from the results obtained for Cr-doped CuAlO 2 thin films investigated earlier by our group. In such films, transmittance in the ultraviolet region presented monotonous decreases with Cr doping (shown in the inset of Fig. 7 ). Considering the contribution of Cr 3d states to the VBM [33] , enhanced hybridization is expected. In the present films, however, because Mg 2+ ions have no d electrons, new considerations are proposed here. First, Mg substitution leads to bigger M 3+ ionic radii, which in turn increases the cell parameter a . Increasing this parameter simultaneously induces the increase of O interstitials in triangular Cu layers [34] , and the decrease of overlapping between Cu 3d states in the ab planes. On one hand, the increase of O interstitials may cause hybridization strengthening because the VBM is composed of hybridized Cu 3d and O 2p anti-bonding states [32] , which can cause increases in the total density of states. On the other hand, the character of the VBM is mainly dominated by overlapping between Cu 3d states when x is greater than 0.5% because the off-stoichiometric oxygen is limited [35] . Fig. 8 shows the resistivity of the films as a profile of temperature in the range of 200–310 K. From the plots, the three samples all exhibit semiconductor-like electrical behavior (d ρ /d T < 0). As known, CuAlO 2 is a layered p-type semiconductor and Al ions in its structure are in the trivalent state. Bivalent cations, such as Mg 2+ , Ca 2+ , and Zn 2+ , doped onto Al 3+ sites can increase p-type conductivity mainly by contributing to the hole concentration [26,36,37] . This conclusion has been thoroughly verified by many experiments involving the delafossite structure. Enlarged plots of the films in the temperature range of 280–310 K are shown in the inset of Fig. 8 . At 300 K, the resistivity of the films with x = 0%, 0.5%, and 1% are 822, 342, and 193 Ω·cm, respectively. These values are slightly larger than those of films with the same composition but prepared by the PLD method, indicating that CSD technique needs to be further improved. From the analysis of the transmittance in the ultraviolet region, it appears that both Mg 2+ cations and O interstitials contribute to hole carriers. In addition to the carrier concentration, mobility is another important factor that influences conductivity ( σ = en μ ). From the XRD and SEM results, with increasing Mg substitution, both the enhancement of the c -axis orientation and improvements of morphology are expected to increase the mobility of holes. In order to confirm this deduction, Hall measurements were carried out on the three films at room temperature using the van der Pauw method. The positive Hall coefficient measured for these films verifies the p-type conductivity of the samples. The carrier concentrations are 2.29 × 10 16 , 4.82 × 10 16 , and 7.58 × 10 16 cm −3 for the samples Mg 0 , Mg 0.5 and Mg 1 , respectively. From these data, the calculated mobility for Mg 0 , Mg 0.5 and Mg 1 are 3.31 × 10 −1 , 3.79 × 10 −1 , and 4.27 × 10 −1 cm 2 V −1 s −1 , respectively. These values show that the increase in conductivity is a result of the increase of both the carrier concentration and mobility. The plots of ln σ vs. 10 3 / T are presented in Fig. 9 to better understand the conduction mechanism of the films. From the figure, the data obtained can fit the thermal activation model well, similar to the results reported by Tate et al. [38] . The activation energy is the energetic difference from an acceptor level to the valence band maximum (VBM) for p-type semiconductors, or from the conduction band minimum (CBM) to a donor level for n-type semiconductors. The activation energies are 331.6, 253.5, and 246.4 meV for the samples Mg 0 , Mg 0.5 and Mg 1 , respectively, indicating that Mg-doping changes the position of the acceptor level in the band gap. The variation trend is the same as that for resistivity, confirming once again the helpful effect of Mg substitution on the electrical properties of the films. 4 Conclusion We successfully prepared CuAl 1− x Mg x O 2 thin films on sapphire substrates by the CSD method. XRD and SEM results reveal that the thin films crystallize in 3R-delafossite structure when x is lower than 2 at.%, and that Mg substitution on Al sites improves the morphology of the films produced. The optical band gaps increase sequentially with increasing x (0 ≤ x ≤ 1%), due possibly to the size effect of the M 3+ ion in delafossite Cu 1+ M 3+ O 2 . Changes in transmittance in the ultraviolet region indicate changes in both the degree of hybridization and electron occupation in the VBM. Resistivity measurements show that Mg doping is helpful in increasing p-type conductivity. CuAl 1− x Mg x O 2 thin films synthesized by the CSD method have higher optical transparency, comparable even to films prepared by other physical methods despite their lower electrical conductivity. These results suggest that the CSD method is a novel and valuable route for the preparation of oxide films. Acknowledgements This work was supported by the National Key Basic Research, under Contract No. 2007CB925002, and the National Nature Science Foundation of China, under Contract Nos. 10774146 and 50802096. Grants were also provided by the Anhui Province NSF (No. 070414162 ) and Director's Fund of Hefei Institutes of Physical Science, Chinese Academy of Sciences . References [1] R.K. Gupta K. Ghosh R. Patel P.K. Kahol Applied Surface Science 255 2009 6252 [2] F. Yakuphanoglu Journal of Alloys and Compounds 470 2009 55 [3] H.F. Liu G.X. Hu H. Gong Journal of Crystal Growth 311 2009 268 [4] M. Sahal B. Hartiti A. Ridah M. Mollar B. Mar Microelectronics Journal 39 2008 1425 [5] H. Kawazoe M. Yasukawa H. Hyodo M. Kurita H. Yanagi H. Hosono Nature 389 1997 939 [6] M.N. Spallart S.P. Pai R. Pinto Thin Solid Films 515 2007 8641 [7] W. Lan M. Zhang G.B. Dong P.M. Dong Y.Y. Wang H. Yan Materials Science and Engineering B 139 2007 155 [8] J.L. Cai H. Gong Journal of Applied Physics 3 2005 033707 [9] Z. Lockman L.P. Lin C.K. Yew S.D. Hutagalung Solar Energy Materials and Solar Cells 93 2009 1383 [10] C. Bouzidi H. Bouzouita A. Timoumi B. Rezig Materials Science and Engineering B 118 2005 259 [11] A. Belafhaili M.A. Aouaj R. Diaz A. Belayachi F. Rueda M. Abd-Lefdil Sensor Letters 5 2009 672 [12] K. Tonookaa K. Shimokawa O. Nishimura Thin Solid Films 411 2002 129 [13] S. Gotzendorfer C. Polenzky S. Ulrich P. Lobmann Thin Solid Films 518 2009 1153 [14] H.F. Jiang H.C. Lei X.B. Zhu G. Li Z.R. Yang W.H. Song J.M. Dai Y.P. Sun Y.K. Fu Journal of Alloys and Compounds 487 2009 404 [15] D.O. Scanlon K.G. Godinho B.J. Morgan G.W. Watson The Journal of Chemical Physics 132 2010 024707 [16] J. Li A.F.T. Yokochi A.W. Sleight Solid State Sciences 6 2004 831 [17] Y. Kakehi K. Satoh T. Yoshimura A. Ashida N. Fujimura Vacuum 84 2009 618 [18] D. Li X.D. Fang Z.H. Deng W.W. Dong R.H. Ta S. Zhou J.M. Wang T. Wang Y.P. Zha X.B. Zhu Journal of Alloys and Compounds 486 2009 462 [19] A. Maignan V. Eyert C. Martin S. Kremer R. Fresard D. Pelloquin Physical Review B 80 2009 115103 [20] R. Manoj M. Nisha K.A. Vanaja M.K. Jayaraj Bulletin of Materials Science 31 2008 49 [21] H. Yanagi T. Hase S. Ibuki K. Ueda H. Hosono Applied Physics Letters 78 2001 1583 [22] K. Hayashi T. Nozaki T. Kajitani Japanese Journal of Applied Physics Part 1 46 2007 5226 [23] P.W. Sadik M. Ivill V. Craciun D.P. Norton Thin Solid Films 517 2009 3211 [24] R. Nagarajan A.D. Draeseke A.W. Sleight J. Tate Journal of Applied Physics 12 2001 8022 [25] A. Maignan C. Martin R. Fresard V. Eyert E. Guilmeau S. Hebert M. Poienar D. Pelloquin Solid State Communications 149 2009 964 [26] G.B. Dong M. Zhang W. Lan P.M. Dong H. Yan Vacuum 82 2008 1321 [27] Y. Kumekawa M. Hirai Y. Kobayashi S. Endoh E. Oikawa T. Hashimoto Journal of Thermal Analysis and Calorimetry 99 2010 57 [28] M.S. Lee T.Y. Kim D. Kim Applied Physics Letters 79 2001 13 [29] S.H. Lim S. Desu A.C. Rastogi Journal of Physics and Chemistry of Solids 69 2008 2047 [30] J. Tauc A. Menth Journal of Non-Crystalline Solids 8 1972 569 [31] C.H. Ong H. Gong Thin Solid Films 445 2003 299 [32] M.N. Huda Y.F. Yan A. Walsh S.H. Wei M.M. Al-Jassim Physical Review B 80 2009 035205 [33] D.O. Scanlon K.G. Godinho B.J. Morgan G.W. Watson Journal of Chemical Physics 132 2010 024707 [34] J. Li A.F.T. Yokochi A.W. Sleight Solid State Sciences 8 2004 831 [35] E. Mugnier A. Barnabe P. Tailhades Solid State Ionics 177 2006 607 [36] G.B. Dong M. Zhang T.X. Li H. Yan Journal of the Electrochemical Society 157 2010 H127 [37] P.M. Dong M. Zhang G.B. Dong X.P. Zhao H. Yan Journal of the Electrochemical Society 155 2008 H319 [38] J. Tate H.L. Ju J.C. Moon A. Zakutayev A.P. Richard J. Russell D.H. Mcintyre Physical Review B 80 2009 165206