Zinc oxide (ZnO) has shown great promise as a wide-bandgap semiconductor with a range of optical, electronic, and mechanical applications. The presence of compensating donors, however, is a major roadblock to achieving p-type conductivity. Recent first-principles calculations and experimental studies have shown that hydrogen acts as a shallow donor in ZnO, in contrast to hydrogen's usual role as a passivating impurity. Given the omnipresence of hydrogen during growth and processing, it is important to determine the structure and stability of hydrogen donors in ZnO.
To address these issues, we performed vibrational spectroscopy on bulk, single-crystal ZnO samples annealed in hydrogen (H2) or deuterium (D2) gas. Using infrared (IR) spectroscopy, we observed O-H and O-D stretch modes at 3326.3 cm-1 and 2470.3 cm-1 respectively, at a sample temperature of 10 K. These frequencies indicate that hydrogen forms a bond with a host oxygen atom, consistent with either an antibonding or bond-centered model. In the antibonding configuration, hydrogen attaches to a host oxygen and points away from the Zn-O bond. In the bond-centered configuration, hydrogen sits between the Zn and O. To discriminate between these two models, we measured the shift of the stretch-mode frequency as a function of hydrostatic pressure. By comparing with first-principles calculations, we conclude that the antibonding model is the correct one.
Surprisingly, we found that the O-H complex is unstable at room temperature. After a few weeks, the peak intensity decreases substantially. It is possible that the hydrogen forms H2 molecules, which have essentially no IR signature. Electrical measurements show a corresponding decrease in electron concentration, which is consistent with the formation of neutral H2 molecules. The correlation between the electrical and spectroscopic measurements, however, is not perfect. We therefore speculate that there may be a second “hidden” hydrogen donor. One candidate for such a donor is a hydrogen-decorated oxygen vacancy.