Dopability, Intrinsic Conductivity, and Nonstoichiometry of Transparent Conducting Oxides

Abstract

Existing defect models for In(2)O(3) and ZnO are inconclusive about the origin of conductivity, nonstoichiometry, and coloration. We apply systematic corrections to first-principles calculated formation energies Delta H, and validate our theoretical defect model against measured defect and carrier densities. We find that (i) intrinsic acceptors ("electron killers") have a high Delta H explaining high n-dopability, (ii) intrinsic donors ("electron producers") have either a high Delta H or deep levels, and do not cause equilibrium-stable conductivity, (iii) the O vacancy V(O) has a low Delta H leading to O deficiency, and (iv) V(O) has a metastable shallow state, explaining the paradoxical coexistence of coloration and conductivity.

Conductivity in transparent oxide semiconductors

“…The [0/2+] transition levels are again found to be rather deep, although the exact positions are still under debate. As for ZnO, there is significant variation in the magnitude of the calculated formation energies, leading some studies to suggest high concentrations of V O under equilibrium conditions [53], while others again find negligible concentrations [57]. Early experimental studies suggested high concentrations of oxygen vacancies at high temperature in polycrystalline In 2 O 3 [60,61] and SnO 2 [62], although this situation may not necessarily hold for high-quality single-crystalline material at room temperature.…”

“…In both ZnO [53,76] and In 2 O 3 [53], Zunger and colleagues have proposed that the oxygen vacancy leads to the population of the conduction band with free carriers via a persistent photoconductivity (PPC) mechanism. In this model, upon photoexcitation, two electrons are promoted from the deep non-conductive V 0 O level to a metastable conductive shallow state [77].…”

Intrinsic defects and dopability of zinc phosphide

“…Since defect levels are sensitive not to the band gap itself, but to their position relative to the host band states, this ambiguity can result in many different predictions for defect ground states. 5,6,25 Furthermore, corrections for SIE are especially important for charged defect calculations as reducing interaction error tends to increase the ionicity of the crystal 26 resulting in more ionic relaxation as a defect state is populated and hence greater energy benefit for a charged defect.…”

“…For example, recent work of various groups on ZnO has lead to predictions that the oxygen vacancy acts as both a shallow and deep defect. 5,6 Recently, more accurate techniques such as hybrid functionals 7 and GW excited-state calculations 8 yield greatly improved band gap prediction. However, these techniques are still too expensive for the large scales required of defect systems.…”

Recent Advances in Understanding the Structure and Properties of Amorphous Oxide Semiconductors

“…Despite great technological demand for TCOs [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] and extensive experimental efforts to improve the conductivity via impurity doping, [21,22] to tune the work function and carrier concentration via cation composition, [23][24][25][26][27][28] to achieve two-dimensional transport via heterointerfaces, [29] and to p-dope the oxides toward active layers of transparent electronics, [30][31][32] theoretical understanding of these fascinating materials has lagged behind significantly. The first electronic band structure of ITO was calculated in 2001; [33] the role of native defects in prototype TCOs was understood after 2002; [34][35][36][37] the properties of multi-cation TCOs were first considered in 2004 [37][38][39][40][41][42] followed by modeling of novel TCO hosts [43,44] and spin-dependent transport in transition-metal-doped TCOs; [45] the nature of the band gap in In 2 O 3 was clarified in 2008; [46] and a first highthroughput se...…”