GaN-based heterostructure field-effect transistors (HFETs) are considered to be suitable for high temperature, high frequency, and high power applications. In GaN-based HFETs, the parasitic source resistance (R_S) is an important parameter for the device performance.^[1,2] Here, R_S includes the gate-to-source access resistance (R_GS) and the Ohmic contact resistance (R_Ohmic), and it leads to some undesirable effects. For example, R_S reduces the effective transconductance of the device and the current gain cutoff frequency. It is necessary to analyze the origin and properties of R_S at different temperatures for enhancing the device performance in various environments. As previously known, R_S is related to the scattering mechanisms for the electrons in the gate–source channel. Both theoretical and experimental results reveal that the polarization Coulomb field (PCF) scattering is an important scattering mechanism in AlGaN/AlN/GaN HFETs.^[3–5] Recently, the effect of the PCF scattering on R_S was verified at room temperature.^[6] However, the influence of PCF scattering on R_S is unknown at elevated temperatures. Therefore, for the AlGaN/AlN/GaN HFETs, it is important to study the relationship between PCF scattering and R_S at elevated temperatures. In this study, we investigate the influence of PCF scattering on R_S in AlGaN/AlN/GaN HFETs in the temperature range 300–500 K using the measured temperature-dependent R_S and both capacitance–voltage (C–V) and current–voltage (I–V) characteristics of the prepared device.

The undoped Al_0.28Ga_0.72N/AlN/GaN heterostructure layers used in this study were grown on a (0001) sapphire substrate by molecular beam epitaxy (MBE). Moreover, the components of its active structure were exhibited in detail in Fig. 1. The Al_0.28Ga_0.72N/AlN/GaN HFET was manufactured according to the standard process.^[7] Its source and drain Ohmic metal stack comprised Ti/Al/Ni/Au. The source and drain regions were rectangular with a length and width of 50 μm and 100 μm, respectively. The drain-to-source distance was 100 μm. Ni/Au was deposited to form the gate Schottky contact. The size of the gate Schottky contact was 20 μm/100 μm (length/width), and the gate was 20 μm away from the source. The respective C–V measurements for the prepared Al_0.28Ga_0.72N/AlN/GaN HFET at 300, 350, 400, 450, and 500 K were performed using Agilent B1520A at 1 MHz. Moreover, the respective I–V and R_S measurements at the above different temperatures were performed using the Agilent B1500A semiconductor parameter analyzer.

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. (color online) Schematic cross section of fabricated AlGaN/AlN/GaN HFET.

Figure 2(a) shows the measured C–V curves of the prepared Al_0.28Ga_0.72N/AlN/GaN HFET at different temperatures. The two-dimensional electron gas (2DEG) electron density (n_2D) can be obtained by the C–V curve integration, and the values are shown in Fig. 2(b).^[8] Table 1 lists the values of n_2D corresponding to zero gate bias (n_2D0) at each testing temperature. Figure 3 displays the output characteristics at different temperatures for the prepared sample. The drain-current, I_DS, with a source–drain voltage of 125 mV at zero gate bias can be obtained from the I–V characteristics, and the values at different temperatures are shown in Table 1.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. (color online) (a) Measured C–V curves for Al0.28Ga0.72N/AlN/GaN HFET at different temperatures. (b) Obtained 2DEG electron sheet density n2D from C–V measurements for the sample at different temperatures.

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. Measured output characteristics of the sample at (a) 300 K, (b) 350 K, (c) 400 K, (d) 450 K, and (e) 500 K.

Table 1.

Table 1.

Table 1. Parameters of the prepared AlGaN/AlN/GaN HFET at different temperatures. VGS is the gate–source bias, VDS is the drain–source bias, IDS is the channel current, n2D0 is the 2DEG electron density at zero gate bias, μn0 is the 2DEG electron mobility with zero gate bias, and RS0 is the parasitic source resistance corresponding to zero gate bias, which is the sum of the source access resistance and the Ohmic contact resistance. . T/K VGS/V VDS/mV IDS/mA n2D0/1012 cm−2 μn0/cm2⋅V−1⋅s−1 RS0/Ω 300 0 125 0.23465 5.01 2340.44 132.63 350 0 125 0.17338 5.16 1679.6 173.4 400 0 125 0.1325 5.41 1225.71 219.4 450 0 125 0.10818 5.61 964.94 260.07 500 0 125 0.087138 5.56 783.01 315.65

Table 1. Parameters of the prepared AlGaN/AlN/GaN HFET at different temperatures. VGS is the gate–source bias, VDS is the drain–source bias, IDS is the channel current, n2D0 is the 2DEG electron density at zero gate bias, μn0 is the 2DEG electron mobility with zero gate bias, and RS0 is the parasitic source resistance corresponding to zero gate bias, which is the sum of the source access resistance and the Ohmic contact resistance. .

The 2DEG electron mobility corresponding to zero gate bias (μ_n0) at 300, 350, 400, 450, and 500 K for the prepared Al_0.28Ga_0.72N/AlN/GaN HFET can be obtained using the same method,^[4] and the calculated results are shown in Table 1. Moreover, according to Ref. [4], the gate-to-source access resistance corresponding to zero gate bias (R_GS0) can be calculated by

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. Plot of ROhmic versus T for the sample.

R_S can be measured using the gate probe method, as described in detail in Ref. 11. The gate probe method shows that an accurate R_S can be extracted from the plot of the gate–source bias (V_GS) versus I_DS under the conditions of I_GS ≪ I_DS, the relatively low drain-to-source bias (V_DS), and the constant forward gate–source current I_GS. Figure 5 is the test configuration for measuring R_S using the gate probe. As shown in Fig. 5, the gate was forward driven with a constant current I_GS and the drain was driven with a range of currents, while the source was grounded. As I_GS is fixed, the voltage drop across the gate Schottky barrier V_drop is constant. Therefore, V_GS is the sum of the channel voltage at the source edge of the gate (V_b) and V_drop. Owing to V_b = I_DS × (R_Ohmic + R_GS), V_b increases with the increase of I_DS when V_DS increases. Here, R_GS is the gate-to-source access resistance. As V_GS = V_b + V_drop = I_DS × (R_Ohmic + R_GS) + V_drop, the derivative of V_GS with respect to I_DS is the value of R_Ohmic + R_GS, that is, R_S.^[1,6,11]

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. Test configuration for measuring RS using gate probe, IGS = constant.

According to the gate probe method, the value of R_S is equal to the slope in the V_GS versus I_DS plot, and the prominent linear relationship of V_GS versus I_DS suggests that an accurate extraction of R_S can be obtained. In order to study the effect of PCF scattering on R_S in the temperature range 300–500 K, the measurements of R_S were conducted at the same value of I_GS under the above different temperatures. Figure 6 shows the measured curves, that is, the plot of V_GS versus I_DS at different temperatures with forward gate current of 25 μA (I_GS = 25 μA). Here, the drain-to-source bias is maintained in the range from 0 V to 3 V. From Fig. 6, the value of R_S corresponding to I_GS = 25 μA at each testing temperature, R_S25, can be extracted. The black trace in Fig. 7 corresponds to the extracted R_S25. Moreover, the red trace in Fig. 7 corresponds to R_S0 (the value of R_S0 at each testing temperature is also listed in Table 1).

Fig. 6.

	Figure Option ViewDownloadNew Window
	Fig. 6. (color online) Plot of VGS versus IDS at different temperatures with IGS = 25 μA for the sample.

Fig. 7.

	Figure Option ViewDownloadNew Window
	Fig. 7. (color online) Variations of RS25 (black trace) and RS0 (red trace) with temperature. RS25 represents the value of RS at IGS = 25 μA, and RS0 represents the value of RS at VGS = 0 V.

As observed from Fig. 7, R_S25 is larger than R_S0 at each corresponding testing temperature, and the resistance difference between R_S25 and R_S0 (ΔR_S = R_S25 − R_S0 decreases with increasing temperature. R_S is related to the scattering mechanisms and n_2D for the electrons in the gate–source channel. As n_2D in the gate–source channel is not modulated by the gate–source bias V_GS,^[4] the value of n_2D in the gate–source channel is assumed to remain unchanged as V_GS or I_GS varies. It indicates that the value of n_2D in the gate–source channel corresponding to I_GS = 25 μA is equal to that corresponding to I_GS = 0 μA at the same temperature. Therefore, R_S25 and R_S0 correspond to the same value of n_2D at the same temperature. Hence, the resistance difference, ΔR_S, does not result from n_2D of the gate–source channel.

In undoped AlGaN/AlN/GaN HFETs, the longitudinal optical (LO) phonon scattering, the interface roughness (IFR) scattering, and the PCF scattering are primarily the three types of important scattering mechanisms.^[3,4] For LO phonon scattering, it is primarily related to the average phonon number and n_2D. For IFR scattering, it is mostly determined by the average distance of the 2DEG electrons from the AlN/GaN interface. The average phonon number is primarily determined by temperature, and the average distance of the 2DEG electrons from the AlN/GaN interface is considerably impacted by n_2D. As mentioned earlier, n_2D in the gate–source channel does not vary with I_GS. Therefore, for both LO phonon and IFR scatterings, they are I_GS-independent at the same temperature. Hence, both LO phonon and IFR scatterings cannot lead to the difference between R_S25 and R_S0 at the same temperature.

For PCF scattering as one of the primary types of important scattering mechanisms in AlGaN/AlN/GaN HFETs as mentioned earlier, it is closely related to the distribution of the polarization charges along the AlGaN/AlN/GaN heterostructure interface. The detailed illustration is provided in the following. The distribution of the polarization charges at the AlGaN/AlN/GaN heterostructure interface using the device processing mechanism and both the gate–source and drain–source biases is not uniform (see Fig. 8(a)). Without the deposition of the contact metals, the distribution of the polarization charges at the AlGaN/AlN/GaN heterostructure material interface is uniform (see Fig. 8(b)). The difference between the nonuniform polarization (Fig. 8(a)) and the uniform polarization (Fig. 8(b)) is considered as the additional polarization charges. Figure 8(c) provides the distribution of the additional polarization charge density. In Fig. 8(c), Δσ₁ is the additional negative polarization charge density near the Ohmic contact metals, which is generated by the Ohmic-contact processing owing to the diffusion of the Ohmic contact metal atoms. l is the diffusion length of the Ohmic contact metal atoms.^[4] Δσ₁ and l are related to the Ohmic contact processing only and not modulated by the gate bias and temperature. Δσ₂ is the additional polarization charge density under the ungated region except the range of l. Neither the Ohmic-contact processing nor the bias voltage influence Δσ₂. Hence, the value of Δσ₂ is considered to be 0.^[3,4] Δσ₃ is the positive additional polarization charge density underneath the gate, which is induced by the forward gate–source bias owing to the converse piezoelectric effect.

Fig. 8.

	Figure Option ViewDownloadNew Window
	Fig. 8. (color online) Schematic of polarization charges distribution. (a) Distribution of polarization charges with the Ohmic and Schottky contact metals. (b) Distribution of polarization charges without the contact metals. (c) Distribution of the additional polarization charge density. +: Positive polarization charges; —: Negative polarization charges; Ө: Two-dimensional electron gas electrons.

The PCF scattering theory indicates that the additional polarization charges establish the elastic scattering potential that scatters 2DEG electrons, and the larger the PCF scattering potential is, the stronger the PCF scattering will be.^[3–5] Here, the PCF scattering potential, V(x, y, z), can be expressed as^[6]

The polarization charges along the AlGaN/AlN/GaN interface involve both spontaneous and piezoelectric polarization. The spontaneous polarization does not vary with gate bias,^[4] and seldom changes with temperature.^[12,13] Hence, the variation of ΔR_S with temperature is closely related to the AlGaN barrier layer piezoelectric polarization. The piezoelectric polarization varies with z-direction electric field in the AlGaN barrier layer owing to the converse piezoelectric effect, and it is expressed as^[14]

Fig. 9.

	Figure Option ViewDownloadNew Window
	Fig. 9. (color online) Measured forward I–V characteristics of Ni/Au/AlGaN/AlN/GaN Schottky diode at different temperatures.

Fig. 10.

	Figure Option ViewDownloadNew Window
	Fig. 10. (color online) Variations of electric field and piezoelectric polarization of AlGaN barrier layer corresponding to IGS = 25 μA at different temperatures.

In addition, figure 10 shows that the biased z-direction electric field (E_PE,AlGaN = 5.57 × 10⁵ V/cm) at 500 K is still strong. Figure 7 shows the resistance difference ΔR_S at 500 K is considerably small. The reason can be explained below.

As mentioned earlier, the Ohmic-contact processing can generate the additional negative polarization charge density near the Ohmic contact metals, that is, Δσ₁ in Fig. 8(c). The positive gate–source bias corresponding to the forward gate current of 25 μA can generate the piezoelectric polarization charge density under the gate, that is, the additional positive polarization charge density underneath the gate contact, that is, Δσ₃ in Fig. 8(c). The scattering potential of the PCF scattering corresponding to I_GS = 0 μA (V_GS = 0 V) is only constituted by Δσ₁. The scattering potential of the PCF scattering corresponding to I_GS = 25 μA is constituted by Δσ₁ and Δσ₃. From Eq. (2), the PCF scattering potential between Δσ₁ and Δσ₃ can be eliminated. As mentioned earlier, as Δσ₁ do not vary with both temperature and gate bias, and Δσ₃, which is the piezoelectric polarization charge density induced by V_drop, decreases with increasing temperature. It can be inferred that the difference in the absolute value of Δσ₁ and Δσ₃ decreases with increasing temperature, and the absolute value of the above difference is approximately equal to that of Δσ₁ at 500 K. As a result, the scattering potential of PCF scattering corresponding to I_GS = 25 μA is approximately equal to that corresponding to I_GS = 0 μA (V_GS = 0 V). Therefore, R_S25 and R_S0 at 500 K are approximately equal. Thus, although the z-direction biased electric field E_PE,AlGaN at 500 K is still strong, the resistance difference ΔR_S is considerably small. Hence, all the above analysis shows that the PCF scattering exhibits an important influence on the parasitic source resistance R_S in the temperature range 300–500 K. This indicates that PCF scattering should be considered at elevated temperatures. In addition, the interaction between the positive additional polarization charges underneath the gate contact and the negative additional polarization charges near the source contact, which is related to PCF scattering, was verified during the variable-temperature study of R_S in the temperature range 300–500 K. Moreover, the PCF scattering, as a type of Coulomb field scattering, is related to distance. The shorter the gate–source spacing of the AlGaN/AlN/GaN HFETs is, the closer is the distance between the additional polarization charges and the 2DEG in the gate–source channel, and the stronger is the PCF scattering at elevated temperatures. Further research should be performed for the dependence of R_S on temperature, which is based on PCF scattering being the dominant mechanism in the AlGaN/AlN/GaN HFETs with the shorter gate–source distance.

In this study, we analyzed and studied the influence of PCF scattering on the parasitic source resistance R_S using the measured C–V and I–V characteristics and R_S in the temperature range 300–500 K for the prepared Al_0.28Ga_0.72N/AlN/GaN HFET. The results indicate that the PCF scattering should be considered at elevated temperatures as it was found to exhibit an important influence on R_S in the investigated temperature range. In addition, the interaction between the positive additional polarization charges underneath the gate contact and the negative additional polarization charges near the source contact was verified to influence the intensity of the PCF scattering.