Shallow donor and deep DX-like center in InAlN layers nearly lattice-matched to GaN

Nonintentionally doped 200-nm-thick ${\mathrm{In}}_{0.16}{\mathrm{Al}}_{0.84}\mathrm{N}/n^{+}\text{−}\mathrm{GaN}$ samples were grown by metal-organic vapor phase epitaxy and used for the electrical characterization of InAlN. In the temperature range 180–400 K, the forward current of Schottky diodes is dominated by a tunneling mechanism below 1.2 V. Capacitance and conductance-temperature characteristics were measured at 1 MHz in the 90–400 K range and at various voltages. The conductance vs temperature reveals two peaks ${\mathrm{D}}_1$ and ${\mathrm{D}}_2$, which are attributed to bulk states in InAlN. Their characterization by admittance spectroscopy gives thermal activation energies of $\approx 68$ meV and 290 meV, and thermal capture cross section of $9.7\times {10}^{-17}$ ${\mathrm{cm}}^2$ and $\approx 6.2\times {10}^{-15}$ ${\mathrm{cm}}^2$, respectively. The same levels are also revealed by extracting the temperature dependence of the carrier density in the neutral region of InAlN from $I\text{−}V\text{−}T$ characteristics on the Schottky diode. A partial carrier freeze out is demonstrated and discussed in the framework of an existing theory for DX centers. The use of this approach is supported by the evidence of persistent photoconductivity effects, which strongly indicate the presence of DX centers in our material. It results that each donor in InAlN would exist in two distinct lattice configurations, a substitutional one (${\mathrm{D}}_1$, hydrogenic state) and a lattice-distorted one (${\mathrm{D}}_2$, DX state). From secondary ion mass spectrometry data, theoretical grounds, and previous experimental evidence in the $\mathrm{Al}{}_x\mathrm{Ga}{}_{1-x}\mathrm{N}$ system, oxygen is the most probable candidate for such an unintentional dopant.

Figure 1

Experimental current density-voltage characteristics at various temperatures between 200 and 400 K. J increases with rising temperature. Actual $\Delta T$ steps were 10 K. Area of the Schottky contact $S=3.14\times {10}^{-4}$ ${\mathrm{cm}}^2$. The energy parameter $E_0$ [Eq. (1)] vs temperature is shown in the inset. Evaluation of the series resistance $R_s\left(T\right)$ was extracted from the linear dependence at high positive biases.

Figure 2

(a) Temperature dependence of the carrier concentration (arbitrary units) as deduced from the thermionic-field emission theory [see text and Eq. (4) for details]. (b) Reciprocal series resistance $R_s$ versus inverse temperature. The activation energies refer to $n_0\left(T\right)$, assuming various temperature dependences of the mobility (see text).

Figure 3

(a) Measured conductance $G$ and (b) associated capacitance $C$ at 1 MHz vs temperature at fixed applied voltages between $-5$ V and 0 V. Heating rate: 4 K/min, $S=3.14\times {10}^{-4}$ ${\mathrm{cm}}^2$. The two peaks in conductance and corresponding inflexion points in capacitance are attributed to a shallow donor ${\mathrm{D}}_1$ and a deep level ${\mathrm{D}}_2$. They appear when their emission rate $e_n\left(T\right)$ is comparable (see text) to $\omega$. (c) Simulated emission rate $e_n$ vs temperature from the parameters extracted from Fig. 6. This plot helps to define the various frequency regimes of the capacitance. The experimental points $e_n\left(T\right)$ are also displayed to visualize the regions of extrapolations.

Figure 4

Measured $G/\omega$ vs. temperature for a few frequencies, at a fixed applied voltage of $-3$ V. The $G/\omega$ peaks shift towards higher temperatures with increasing frequency. The Arrhenius plot for $D_1$ was built from such thermal admittance experiments. Heating rate: 4 K/min, $S=3.14\times {10}^{-4}$ ${\mathrm{cm}}^2$.

Figure 5

$G_{\mathrm{corr}}/\omega$ versus frequency at fixed temperatures and fixed applied voltage $V_a$. $G_{\mathrm{corr}}$ is the corrected conductance for the effect of the series resistance $R_s\left(T\right)$ in a parallel mode of measurement. The Arrhenius plot for $D_2$ was built from such frequency-dependent admittance experiments. $S=3.14\times {10}^{-4}$ ${\mathrm{cm}}^2$.

Figure 6

Arrhenius plots for the two levels with their signatures, deduced from the admittance spectroscopy data taken at $V_a=-3$ V. The values of capture cross sections assume $m^{*}=0.28m_0$ and $g=2$. ${\mathrm{D}}_1$ points are obtained from thermal admittance spectroscopy, ${\mathrm{D}}_2$ points from frequency-dependent admittance spectroscopy.

Figure 7

Simulated frequency dispersion of the capacitance in the intermediate-frequency regime for the $D_2$ level. The experimental linear temperature dependence, likely due to $C_{\mathrm{HF}}$, was included in Eq. (9) and $e_n\left(T\right)$ was calculated as in Fig. 3. $S=3.14\times {10}^{-4}$ ${\mathrm{cm}}^2$.

Figure 8

Series resistance $R_s$ of an $\mathrm{InAlN}/n^{+}-\mathrm{GaN}$ Schottky diode (sample B) in the dark. The data are taken before (black points) and after (red point) illumination at 100 K by a subband gap (2 eV) monochromatic light. A PPC effect is clearly seen at low temperature. $S=1.26\times {10}^{-3}$ ${\mathrm{cm}}^2$.

Figure 9

2DEG sheet carrier concentration $n_s$ at 15 K versus time measured on a thin barrier InAlN/AlN/GaN heterostructure. The time sequence for illumination was: dark (OFF), illumination (ON), dark again (OFF, PPC state). Three experiments were performed at various photon energies. A PPC effect is clearly seen after illumination at 2 or 2.6 eV. The variations of $n_s$ in the dark depend on the history of the sample.

Figure 10

2DEG sheet carrier concentration $n_s$ and Hall mobility $\mu$ in the dark versus temperature following sufficient photoexcitation at 2 eV to reach saturation of $n_s$ at 15 K. The sample used was the same as in Fig. 9.