Carrier compensation by deep levels in Zn1−xMgxO/sapphire

Abstract

A systematic analysis of the deep level spectrum in the lower half of the bandgap of Au–Zn1−xMgxO (0.056<x<0.18) Schottky diodes is presented. Two deep levels are observed at Ev+580 and Ev+280 meV regardless of the bandgap energy with trap concentrations linearly increasing with the Mg content. The Ev+280 meV trap concentration becomes as high as 1.01×1018 cm−3 at 18% Mg, partially compensating the films and causing a decrease from 8.02×1016 to 1.27×1016 cm−3 in the net electron concentration and an increase by three orders of magnitude in the diode series resistance due to electron trapping.

Deep levels in a-plane, high Mg-content MgxZn1−xO epitaxial layers grown by molecular beam epitaxy

“…In addition, the R s values of the Schottky diodes increase gradually with Mg content (except for the highest Mg content) for the same device geometry, similar to what has been reported previously in the literature. 14,20 The high series resistance of the diodes is attributed to the low net carrier concentration observed in these unintentionally doped layers (see C-V results, below) are consistent with the the coplanar device structure and the large distance between Ohmic and Schottky contact metals.…”

“…18 While the SSPC approach provides accurate trap concentrations in the dilute concentration limit (for N t < 0.1N d ), this method underestimates the trap concentration for very high trap concentrations and LCV was found necessary for many of the traps detected in the MgZnO alloys here. 14,18 …”

“…More details about DLOS measurements can be found in previous studies, including its application to Mg x Zn 1Àx O with lower Mg content than explored here. 14,17 In order to accurately determine the concentrations of the traps observed in the DLOS measurements for cases where the trap concentration is an appreciable fraction (or exceeds), the background doping (i.e., the dilute concentration approximation is violated), lighted C-V (LCV) measurements were conducted with the Agilent LCR meter at 1 kHz frequency. 18 While the SSPC approach provides accurate trap concentrations in the dilute concentration limit (for N t < 0.1N d ), this method underestimates the trap concentration for very high trap concentrations and LCV was found necessary for many of the traps detected in the MgZnO alloys here.…”

“…As a result of the dispersion at high frequency, we chose a frequency of 1 kHz using a lock-in amplifier and/or LCR meter for subsequent capacitance-based measurements. 14,20 and/or it could be due to an increase in the donor activation energy for the Mg x Zn 1Àx O thin films due to the shift in the conduction band minimum as the Mg content increases. [22][23][24] In order to explore this further, DLOS and LCV measurements were performed and are discussed in Sec.…”

“…14 The work presented here focuses on experimental characterization of deep level defects within Mg x Zn 1Àx O grown by plasma assisted molecular beam epitaxy (MBE) covering a much wider range of alloys compositions from x ¼ 0 to x ¼ 0.56, and bandgap energies that range from 3.36 eV for ZnO to 4.56 eV for Mg x Zn 1Àx O for x ¼ 0.56.…”