Influence of Proton Irradiation Energy on Gate–Channel Low-Field Electron Mobility in AlGaN/GaN HEMTs

Abstract

AlGaN/GaN high-electron-mobility transistors (HEMTs) with two different gate–drain distances (30 μm and 10 μm) were exposed to 1 MeV, 0.6 MeV, and 0.4 MeV protons at a fluence of 2.16 × 10¹² cm⁻². The gate–channel electron density and low-field mobility were obtained by measuring the capacitance–voltage characteristics and current–voltage characteristics. After proton irradiation, the gate–channel low-field electron mobility of the AlGaN/GaN HEMT with a 30 μm gate–drain distance increases and that with a 10 μm gate–drain distance decreases. It is studied and found that the mobility behavior is related to the polarization Coulomb field scattering, and the proton irradiation influences the intensity of the polarization Coulomb field scattering by changing the polarization/strain distribution in the barrier layer. The different gate–drain distances correspond to different variation trends of scattering intensity. The effect of 1 MeV protons on the barrier layer is smaller compared with 0.6 MeV and 0.4 MeV protons, so the mobility variation is smaller.

1. Introduction

AlGaN/GaN high-electron-mobility transistors (HEMTs) have the advantages of high breakdown voltage, high electron mobility, and high power density. They can work stably under harsh environments such as high temperature and high radiation due to the large band gap. Therefore, AlGaN/GaN HEMTs have broad application prospects in important fields such as radar communication, aerospace, and nuclear reactors [1,2,3,4,5,6].

Compared with Si- and GaAs-based devices, the better radiation resistance of AlGaN/GaN HEMTs has been proven theoretically [7,8]. However, during the growth process of the epitaxial layers, the lattice mismatch and thermal mismatch between the epitaxial layers induce a high density of defects [9,10]. It is a considerable challenge to the radiation resistance of AlGaN/GaN HEMTs in practical applications. In addition, as one of the main particles causing the degradation of the device performance in the radiation environment, the study of the proton irradiation of AlGaN/GaN HEMTs has received much attention. Hu et al. reported that the output characteristics and the transfer characteristics of the AlGaN/GaN HEMTs degraded slightly at the 15, 40, and 105 MeV proton irradiations, but the drain current and transconductance decreased by 10% and 6.1% for a 1.8 MeV fluence of 10¹² cm⁻², respectively [11]. The degradation of AlGaN/GaN HEMTs at different proton energies is consistent with the calculation of the non-ionizing energy loss. Kalavagunta et al. reported that the gate-lag of AlGaN/GaN HEMTs increases with a fluence increasing (from 3 × 10¹² cm⁻² to 3 × 10¹⁴ cm⁻²) at a 1.8 MeV proton irradiation [12]. Lv et al. studied the influence of 3 MeV protons on AlGaN/GaN HEMTs [13]. In addition, they observed that the drain saturation currents and maximum transconductance decreased by 20% and 5%, respectively. The degree of the threshold voltage shift increased with the proton fluence increasing. The density of vacancies at different proton fluences were calculated by using SRIM and adding the vacancies model into a Slivaco simulator. It is found that the result of the simulation is consistent with the experimental data and the Ga vacancies may be the main reason for the degradation of the AlGaN/GaN HEMTs’ performance. Ives et al. studied the effects of displacement damage induced by protons on AlGaN/GaN HEMTs [14]. The threshold voltage has a negative shift at a fluence of 10¹³ cm⁻² and a positive shift at a high fluence of 10¹⁴ cm⁻². It is attributed to the proton-induced donor defects (N vacancies) dominating at a low fluence and acceptor defects (Ga vacancies) dominating at a high fluence. Kim et al. studied the DC characteristic degradation of AlGaN/GaN HEMTs at the 0.5 MeV, 5 MeV, and 60 MeV proton irradiations, and the transfer characteristics at a 0.5 MeV proton irradiation has the most serious degradation. Moreover, the threshold voltage at the 0.5 MeV and 5 MeV proton irradiations has a positive shift, while the threshold voltage at a 60 MeV proton irradiation has a negative shift [15]. According to different scattering mechanisms, Tang et al. developed a model to simulate two-dimensional electron mobility at different proton energies and fluences at a low temperature [16]. It is found that the proton-induced ionized impurity scattering is the dominant factor in the degradation of electron mobility at a very low temperature (<5 K). The theoretical model of the proton irradiation effects on the two-dimensional electron density of AlGaN/GaN HEMTs were developed in the subsequent work [17]. The decrease of the electron density is a result of two mechanisms: the Fermi level is affected by the Ga vacancies in the GaN cap layer and AlGaN barrier layer, and the Ga vacancies in the GaN cap layer traps two-dimensional electrons in the potential well. These studies all consider that the displacement effect caused by proton irradiation is the main reason for the change of the AlGaN/GaN HEMTs’ performance. However, little attention has been paid to the proton influence on piezoelectric polarization.

For AlGaN/GaN HEMTs, the existence of two-dimensional electron gas (2DEG) in the channel depends on the spontaneous and piezoelectric polarization [18]. A change of the lattice structure, which is induced by the displacement effect, will inevitably affect the polarization charge distribution by changing the strain distribution. It has been proven that the polarization Coulomb field (PCF) scattering is a special and important scattering mechanism in AlGaN/GaN HEMTs, and the PCF scattering is directly related to the device size and polarization charge distribution at the AlGaN/GaN interface [19,20,21,22,23]. Therefore, proton irradiation will alter the electron mobility of AlGaN/GaN HEMTs and device performance through PCF scattering. Therefore, particular attention should be paid to the proton influence on PCF scattering.

In this paper, AlGaN/GaN HEMTs with two different gate–drain distances (30 μm and 10 μm) were prepared, which were irradiated with different proton energies (1 MeV, 0.6 MeV, and 0.4 MeV) at a fluence of 2.16 × 10¹² cm⁻². The capacitance–voltage (C–V) characteristics and current–voltage characteristics were measured before and after irradiation. The low-field electron mobility is extracted, the low-field electron mobility of the device with a 30 μm gate–drain distance gets increased, and that of the device with a 10 μm gate–drain distance gets decreased after proton irradiation. The degree of the low-field electron mobility variation decreases with the increase of proton energy. The proton irradiation effect on the two-dimensional electron transport properties is analyzed by using the theoretical model of PCF scattering related to the device size and polarization/strain distribution at the AlGaN/GaN interface.

2. Materials and Methods

Figure 1 shows the structure of AlGaN/GaN HEMTs in this study. The metal–organic chemical vapor deposition (MOCVD) was used to fabricate the AlGaN/GaN heterostructure material on Si substrate. The epitaxial layers are formed by 3.9 μm GaN buffer layer, 0.26 μm i-GaN layer, 27.5 nm Al_0.25Ga_0.75N barrier layer, and 3.3 nm GaN cap layer.

Source and drain are both Ohmic contacts and gate is Schottky contact. Gate–source distance L_GS is 4 μm. Gate length L_G is 4 μm, and gate width W_G is 104 μm. The gate–drain distance L_GD are 30 μm and 10 μm, which are marked as Device 1 and Device 2, respectively.

Proton irradiation was performed in a proton accelerator with the device temperature controlled at room temperature during irradiation. Devices were irradiated with 1 MeV, 0.4 MeV, and 0.6 MeV protons at a fluence rate of 10⁸ cm⁻²/s. The proton irradiation lasted for 6 h, meaning a fluence of 2.16 × 10¹² cm⁻². The C–V characteristics and current–voltage characteristics were measured before and after proton irradiation.

3. Results and Discussion

Figure 2 shows the C–V curves of the AlGaN/GaN HEMTs measured at 1 MHz before and after proton irradiation. The C–V curve at 1 MHz reveals the information of 2DEG at the AlGaN/GaN interface. The gate–channel 2DEG electron density n_2D can be obtained by the following equation [24]:

$$n_{2\mathrm{D}} = {\displaystyle {\int }_{V_{\mathrm{T}}}^{V_{\mathrm{GS}}}\raisebox{1ex}{$CdV$}\!\left/ \!\raisebox{-1ex}{$(Me)$}\right.}$$

(1)

where C is the gate–source capacitance, V_T is the threshold voltage, V_GS is the gate–source voltage, M is the area of the gate Schottky contact, and e is the electron charge. V_T is obtained by differentiating the C–V curve. Before irradiation, the V_T values of Device 1 and Device 2 are both −4.95 V. After irradiation, the V_T values of Device 1 are −4.95 V (1 MeV proton energy), −4.9 V (0.6 MeV proton energy), and −4.85 V (0.4 MeV proton energy). The V_T values of Device 2 after irradiation are −4.95 V (1 MeV proton energy), −4.95 V (0.6 MeV proton energy), and −4.85 V (0.4 MeV proton energy). The shift of V_T is related to the electron density [25]. The obtained n_2D is shown in Figure 3. It is found that the n_2D of Device 1 and Device 2 decrease by different degrees after irradiation. Protons cause displacement damage at the AlGaN/GaN interface, which is the main reason for the decrease of the electron density [13]. The decrease in n_2D is the most obvious after a 0.4 MeV proton irradiation. With proton energy increasing, the degree of decrease in electron density gradually gets smaller. The decrease after a 1 MeV proton irradiation is the least. It explains that the degree of threshold voltage shift increases as proton energy decreases.

Figure 4 and Figure 5 are the I_DS–V_DS curves of Device 1 and Device 2 before and after proton irradiation. I_DS and V_DS are the drain-source current and drain-source voltage, respectively. It is found that the I_DS of Device 1 increases and I_DS of Device 2 decreases after 0.4 MeV irradiation. With proton energy increasing, the difference in the I_DS between Device 1 and Device 2 is gradually disappearing. I_DS is dominated by both electron density and electron mobility. The electron density of Device 1 and Device 2 both decrease after irradiation (Figure 3), and it cannot explain the increase of I_DS in Device 1 (Figure 4c). Thus, the difference in the I_DS between Device 1 and Device 2 should be induced by electron mobility. According to the I_DS–V_DS curves, the gate–channel 2DEG electron mobility μ_n is obtained as follows [24,26,27,28]:

$${\mu }_{\mathrm{n}} = \frac{I_{\mathrm{DS}}{L}_{\mathrm{G}}}{e{n}_{2\mathrm{D}}{W}_{\mathrm{G}}[V_{\mathrm{DS}}-I_{\mathrm{DS}}(R_{\mathrm{d}}+R_{\mathrm{s}}+2R_{\mathrm{c}})]}$$

(2)

$$R_{\mathrm{d}} = \frac{L_{\mathrm{GD}}}{e{n}_{2\mathrm{D}0}{\mu }_{\mathrm{n}0}{W}_{\mathrm{G}}}$$

(3)

$$R_{\mathrm{s}} = \frac{L_{\mathrm{GS}}}{e{n}_{2\mathrm{D}0}{\mu }_{\mathrm{n}0}{W}_{\mathrm{G}}}$$

(4)

where R_d is the gate–drain access resistance, R_s is the gate–source access resistance, R_c = 3.54 Ω∙mm is the Ohmic contact resistance, which is obtained by the transmission line method (TLM), and n_2D0 and μ_n0 are the gate–channel 2DEG electron density and mobility corresponding to V_GS = 0 V. The values of I_DS at V_DS = 0.1 V are adopted corresponding to the low-field region [27,28,29].

Figure 6 and Figure 7 show the μ_n of Device 1 and Device 2 before and after proton irradiation. In order to provide a clear comparison, the average change rates of μ_n for Device 1 and Device 2 under the 1 MeV, 0.6 MeV, and 0.4 MeV proton irradiations are given in

Table 1. Average change rate of μ_n for Device 1 and Device 2 with 1 MeV, 0.6 MeV, and 0.4 MeV protons.

Proton Energy (MeV)	Average Change Rate of μn for Device 1 (%)	Average Change Rate of μn for Device 2 (%)
1	+5.4	−5.4
0.6	+10.2	−9.0
0.4	+14.3	−15

. After a 0.4 MeV proton irradiation, the μ_n of Device 1 increases by 14.3% (Figure 6c) and that of Device 2 decreases by 15.0% (Figure 7c). As the proton energy increases, the magnitude of the μ_n variation for Device 1 and Device 2 gradually decrease. After a 1 MeV proton irradiation, there is a slight increase (+5.4%) in the μ_n of Device 1 (Figure 6a) and a slight decrease (−5.4%) in the μ_n of Device 2 (Figure 7a). It is found that the electron mobility varies differently between the devices with two different gate–drain distances and the magnitude of the variation decreases with the proton energy increasing. This μ_n behavior is explained as follows.

In AlGaN/GaN HEMTs, electron mobility is mainly determined by the polar optical phonon (POP) scattering, acoustic phonon (AP) scattering, interface roughness (IFR) scattering, dislocation (DIS) scattering, and PCF scattering [29,30,31]. The first four scattering mechanisms are determined by the inherent properties of the AlGaN/GaN material and are not affected by the device size [20,31,32]. The devices in this study are fabricated on the same heterostructure material, so the first four scattering mechanisms of Device 1 are consistent with that of Device 2. Since the protons are uniformly irradiated over the devices, the variations of the inherent properties are the same for Device 1 and Device 2 under the same irradiation environment. For devices in the same proton irradiation environment, the variation of μ_n should be consistent under the influence of the first four scattering mechanisms after proton irradiation, which is inconsistent with the experimental results (Figure 6 and Figure 7). Previous studies have reported that PCF scattering is closely related to device size [19,20,21,22,23,28,29]. Therefore, although μ_n is dominated by POP scattering at room temperature, the differences in μ_n behavior corresponding to the different L_GD (Figure 6 and Figure 7) must be attributed to PCF scattering.

Before device fabrication, the polarization charges in the heterostructure material are uniformly distributed. The device fabrication processing (rapid thermal annealing) and gate bias cause the nonuniform distribution of polarization charges at the AlGaN/GaN interface [24,29]. The PCF scattering of the AlGaN/GaN HEMTs originates from the difference between the nonuniform polarization charges and the uniform polarization charges, which is defined as the additional polarization charges ∆σ. Figure 8 is the ∆σ distribution diagram. The influence of the additional polarization charges near the Ohmic contact is negligible, considering that L_GS and L_GD are large enough and μ_n before irradiation is almost the same (Figure 6 and Figure 7) [29]. Therefore, the distribution of ∆σ is mainly divided into two parts, as shown in Figure 8. In region I, which is influenced by the converse piezoelectric effect, the negative gate bias compresses the lattices and the polarization charges decrease, generating the negative additional polarization charges ∆σ₁. Due to the lattice continuity, the compression of the lattices in region I causes the tensile strain of the adjacent lattices in region II and the polarization charges increase, generating the positive additional polarization charges ∆σ₂ [18,28]. The tensile effect weakens with the distance increasing, so the additional polarization charges in region II are linearly distributed. In addition, although L_GS is different from L_GD, the interaction effect of region I is the same for both the gate–source region and gate–drain region, and the effect of protons on the gate–source region and gate–drain region is also the same. For a clearer presentation and ease of understanding, the additional polarization charges of the gate–source region are treated the same as that of the gate–drain region. With this distribution of the additional polarization charge, the PCF scattering potential V(x,y,z) generates the following [29]:

$$\begin{array}{l}V(x,y,z)=-\frac{e}{4\pi {\epsilon }_s}{\displaystyle {\int }_{-L_{GS}-\frac{L_G}{2}}^{-\frac{L_G}{2}}dx’}{\displaystyle {\int }_0^{W_G}\frac{\Delta {\sigma }_2\cdot (-x’-L_{GS}-L_G/2)}{L_{GS}\cdot \sqrt{{(x-x’)}^2+{(y-y’)}^2+z^2}}dy’}\\ -\frac{e}{4\pi {\epsilon }_s}{\displaystyle {\int }_{-\frac{L_G}{2}}^{\frac{L_G}{2}}dx’}{\displaystyle {\int }_0^{W_G}\frac{\Delta {\sigma }_1}{\sqrt{{(x-x’)}^2+{(y-y’)}^2+z^2}}dy’}\\ -\frac{e}{4\pi {\epsilon }_s}{\displaystyle {\int }_{\frac{L_G}{2}}^{L_{GD}+\frac{L_G}{2}}dx’}{\displaystyle {\int }_0^{W_G}\frac{\Delta {\sigma }_2\cdot (x’-L_{GD}-L_G/2)}{L_{GD}\cdot \sqrt{{(x-x’)}^2+{(y-y’)}^2+z^2}}dy’}\end{array}$$

(5)

where ε_s is the GaN dielectric constant. It is found from Equation (5) that the PCF scattering intensity is closely related to the device size and ∆σ distribution.

Based on Figure 3, the level of the n_2D variation ∆n_2D after proton irradiation at V_GS = 0 V as a function of proton energy is obtained and is shown in Figure 9. Since the 2DEG of the AlGaN/GaN HEMTs is determined by the strain and polarization, the n_2D variation after proton irradiation indirectly reflects the lattice strain and polarization of region I, and the increase/decrease of the electron density corresponds to the enhanced/weakened tensile strain [18,33,34]. It is found from Figure 9 that the n_2D of both Device 1 and Device 2 decrease, and the ∆n_2D of Device 1 is larger than that of Device 2. This means that the tensile strain in region I is weakened after proton irradiation. In addition, the weakening amplitude of Device 1 is greater than that of Device 2. A previous study has reported that the effects of protons on the AlGaN/GaN heterostructure material are the ionization effect and displacement effect [35]. The ionization effect is the proton interaction with the lattice electrons, and the displacement effect is protons forcing atoms to leave the original lattice position [35]. The ionization effect will not change the lattice structure, so the change of the strain and polarization is caused by the displacement effect. Due to the same structure and size of the Schottky gates, at the same proton energy, the proton effect on region I should be the same for Device 1 and Device 2. Device 1 and Device 2 differ only in the gate–drain distance, so the variation of the tensile strain in region I must be attributed to the indirect effect of the adjacent region II. Since region I is covered by the gate metal, which has a blocking effect on protons, the proton irradiation effect of region I is different from that of region II. This makes the proton irradiation effect on the strain/polarization distribution of region I and region II different. Moreover, the PCF scattering is directly related to the nonuniform polarization charge distribution. Therefore, the main focus of this study is on the difference of the polarization distribution brought by proton irradiation for region I and region II. The enhancement of the tensile strain in region II can be inferred from the weakened tensile strain in region I. Compared with Device 2, the enhancement of the tensile strain in region II of Device 1 is greater, considering the greater weakening of the tensile strain in region I. The positive additional polarization charges ∆σ_2’ is added to ∆σ₂ due to the enhanced tensile strain in region II (Figure 8). In addition, since ∆σ is the additional polarization charges per unit area, and protons are uniformly irradiated on the devices, ∆σ_2’ is the same in Device 1 and Device 2. The enhanced tensile strain in region II weakens the tensile strain in region I because of the lattice continuity, generating the extra negative additional polarization charges ∆σ_1’ (Figure 8). According to Equation (5), the additional polarization charges in region I and region II have an offsetting effect, which affects the PCF scattering potential. A larger L_GD will correspond to a stronger offsetting effect. For Device 1, the offset of ∆σ₂ + ∆σ_′ to ∆σ₁ + ∆σ_1′ is stronger due to the larger L_GD. According to Equation (5), the PCF scattering potential becomes weaker and μ_n gets larger after proton irradiation (Figure 6). For Device 2, L_GD is smaller, which corresponds to a smaller integration interval of the last term in Equation (5). In addition, the offset of ∆σ₂ + ∆σ_2′ to ∆σ₁ + ∆σ_1′ is weaker. Therefore, ∆σ₁ + ∆σ_1′ plays a dominant role at this point, enhancing the PCF scattering potential and causing a decrease in the μ_n after proton irradiation (Figure 7).

The action distance of higher energy protons is farther from the barrier layer, and with the proton energy increasing, the non-ionizing energy transferred from the proton to the atom decreases, which causes the degree of displacement damage to decrease [35,36]. This is the reason that the action intensity of higher energy protons on the barrier layer is lower in a certain energy range. It is found from Figure 9 that the amplitude of the n_2D variation decreases with increasing proton energy. This indicates that the weakening degree of the tensile strain in region I gradually decreases as the proton energy increases. In other words, the enhancement degree of the tensile strain in region II gradually decreases. Therefore, for devices with the same L_GD, the absolute value of ∆σ_2′ decreases with the increasing proton energy, and the absolute value of ∆σ_1′ decreases simultaneously. For Device 1 at a 1 MeV proton irradiation, the influence of ∆σ₂ + ∆σ_2′ is weaker than that at a 0.4 MeV proton irradiation. Following Equation (5), the decrease degree of the PCF scattering potential gets lower with the proton energy increasing. As a result, compared with Figure 6c, the increase of μ_n is smaller in Figure 6a. For Device 2 at a 1 MeV proton irradiation, ∆σ₁ + ∆σ_1′ plays a dominant role. With the proton energy increasing, the absolute value of ∆σ_1′ decreases, and the PCF scattering potential gets lower. This makes the decrease degree of μ_n smaller in Figure 7a, compared with Figure 7c.

4. Conclusions

Based on the PCF scattering theory, the proton irradiation effect on the transport characteristics of AlGaN/GaN HEMTs is analyzed in terms of the gate–drain distance and proton energy. It is found that the displacement damage caused by protons changes the polarization/strain distribution at the AlGaN/GaN interface, which affects the 2DEG electron density and intensity of the PCF scattering. Due to the displacement damage, the electron density decreases after proton irradiation. In addition, the decrease degree of the device with a large gate–drain distance is larger. The electron mobility of the device with a large gate–drain distance increases after irradiation due to the weakening of the PCF scattering. For the device with a small gate–drain distance, the electron mobility decreases after irradiation due to the enhancement of the PCF scattering. The change of the electron mobility is the main reason for the difference in the output current. The impact of proton irradiation on the electrical performance gradually decreases with the proton energy increasing. Although the electron density decreases after a 0.4 MeV proton irradiation, the current and low-field electron mobility of the device with a larger gate–drain distance increase. Thus, the proton effect and gate–drain distance can be adjusted appropriately to improve the device performance. Moreover, based on a theoretical model of the PCF scattering, the variation of the strain at the AlGaN barrier layer is obtained by an analysis of the electrical properties. It is helpful to understand the mechanism of the proton irradiation influence on AlGaN/GaN HEMTs at the microscopic level. This is beneficial for improving the irradiation resistance of AlGaN/GaN HEMTs in the field of aerospace.