A method for characterizing the trap position distribution of GaN HEMT devices using amplitude response

Abstract

A method for characterizing the trap position distribution of GaN HEMT devices using amplitude response relates to the field of semiconductor device reliability. When the GaN HEMT device is under the bias of gate-source voltage and drain-source voltage, the traps inside the device are filled with electrons under the action of a constant stress. After the stress ends, a constant current is applied across the drain and source of the device, and the change of the drain-source voltage with time is monitored, which can characterize the trap release process. By measuring the transient voltage change during this trap release process and analyzing and processing the trap response peak based on the time constant spectrum, the absolute amplitude of the traps inside the device can be obtained. The trap filling at different positions inside the device is related to the electric field distribution. When the electric field strength increases, the trap amplitude corresponding to the same position increases. The present invention realizes the effective characterization of the trap position distribution by obtaining the precise amplitude change of the traps under different bias voltages and analyzing the influence of the electric field on trap filling under different voltages.

Description

Technical Field:

[0001] The present invention relates to the field of semiconductor device reliability, and is mainly applied to the measurement of internal trap parameters and the characterization of the distribution of GaN HEMT (Gallium Nitride Based High Electron Mobility Transistor) devices. Background Art:

[0002] GaN HEMT devices are a new generation of wide-bandgap semiconductor devices, which have the characteristics of high operating voltage, high frequency, high temperature resistance, etc. They have become key core devices in the fields of aerospace, power electronics, radar communication, etc., and have extremely broad application prospects. However, in actual application scenarios, the problem that the performance parameters of the device degrade over time due to internal traps has become an important issue affecting its application reliability.

[0003] At present, the main method for testing traps in GaN HEMT devices is deep level transient spectroscopy (DLTS) testing. This method can realize the test and characterization of traps and defects in GaN materials. However, the capacitance of GaN HEMT devices is small, so it is difficult to obtain the precise change of the capacitance signal by varying the temperature. In addition, trap-related parameters can be calculated through deep level transient spectroscopy testing, but it is difficult to determine the distribution position of the traps. In the actual application process, determining the position distribution of internal traps in GaN HEMT devices is of great significance for further improving device fabrication technology. At present, there is still a lack of effective characterization techniques to achieve the localization of internal traps in the device.

[0004] The technology of the present invention proposes a method for localizing the distribution of internal traps in a device by collecting the changes in trap amplitudes under different filling voltages. In traditional trap testing, the ordinate of the peak corresponding to the trap in the time constant spectrum is usually used as the trap amplitude. This value only summarizes the relative peak corresponding to the peak height and does not include the influence of the peak distribution. Therefore, on the basis of the time constant spectrum, the amplitude corresponding to the trap peak distribution is taken into consideration as the absolute trap amplitude, so as to obtain the precise change of the trap amplitude under different filling voltages to localize the distribution of internal traps in the device. This technology can be applied to the parameter extraction and position analysis of internal traps in GaN HEMT devices, further confirm the trap distribution affecting the device performance parameters, and provide an effective characterization means for the performance degradation analysis of GaN HEMT devices under voltage bias. This method can achieve non-destructive and in-situ testing, and the operation process is simple and convenient, which is applicable to the field of reliability analysis of wide-bandgap semiconductor devices. Summary of the Invention:

[0005] When the GaN HEMT device is under the bias of gate-source voltage and drain-source voltage, the traps inside the device are filled with electrons under the action of a constant stress. After the stress ends, a constant current is applied across the drain and source of the device, and the change of the drain-source voltage with time is monitored, then the trap release process can be obtained. By measuring and analyzing the transient voltage change in this trap release process and processing the corresponding peak of the trap based on the time constant spectrum, the accurate amplitude of the trap inside the device can be obtained. By measuring the transient voltage change during the release process under different bias voltages, the amplitude change of the trap corresponding to different gate-source voltages and drain-source voltages can be obtained, and thus the position distribution of the trap inside the device can be analyzed.

[0006] A method for characterizing the trap position distribution of a GaN HEMT device by amplitude response, characterized in that:

[0007] 1. Place the device under test on a constant temperature platform at a temperature of T1; during the trap filling stage, apply a gate-source voltage VG and a drain-source voltage VD to the device, and the filling time is t1. Where the drain-source voltage VD ranges from 0V to 20V, the gate-source voltage VG ranges from -15V to 0V, and the filling time t1 ranges from 1ms to 100s. After the t1 time ends, test the trap recovery process. Apply a 200mA drain-source current ID under a gate-source measurement voltage VGS of 0V to ensure that the device channel is in the open state and the gate-source measurement voltage VGS has no influence on the trap recovery process; the captured electrons are released from the trap and cause the drain-source voltage to change; monitor the curve of the drain-source voltage of the device under test changing with time until the transient drain-source voltage becomes stable and unchanged, then the transient voltage response curve VDS of the trap recovery process can be obtained. The trap recovery time is denoted as t2, where t2 ranges from 1ms to 300s.

[0008] 2. Perform Bayesian iteration on the obtained transient voltage response curve to obtain the corresponding time constant spectrum. The n peaks in the time constant spectrum correspond to the n traps contained in the transient voltage curve. Among them, the parameter corresponding to the peak of the time constant spectrum on the x-axis is the characteristic time constant τ of the trap, and the parameter on the y-axis is the relative amplitude action intensity corresponding to the trap. Accumulate the y values of the time constant spectrum, that is, let Then exchange the horizontal and vertical coordinates and perform differential processing on the abscissa at this time to obtain the differential amplitude spectrum DAS; the difference between the abscissas corresponding to adjacent peaks of the differential amplitude spectrum DAS is the absolute amplitude Amp of the trap, and the sum of the absolute amplitudes of the n traps is the total change amount of the transient voltage curve caused by trap filling.

[0009] 3. Keep the gate-source voltage VG constant and repeat the measurement process of Step 1 at M drain-source voltages (VD1, VD2, ……, VDM); keep the drain-source voltage VD constant and repeat the measurement process of Step 1 at N gate-source voltages (VG1, VG2, ……, VGN). Repeat the processing process of Step 2 for the obtained M and N transient voltage curves respectively to obtain the corresponding time constant spectra and differential amplitude spectra, so as to extract the time constants and absolute amplitudes corresponding to the traps at M drain-source voltages and N gate-source voltages respectively.

[0010] 4. Plot the dotted line relationship diagram of the absolute amplitude of the trap versus the filled gate-source voltage and drain-source voltage. Since the electric field distribution affects the trap filling, the trap filling in the AlGaN layer is affected by the lateral and longitudinal electric fields brought by both the gate-source voltage VG and the drain-source voltage VD, and the trap filling in the GaN layer is determined by the lateral electric field brought by the drain-source voltage VD. Analyze the absolute amplitudes of the traps at different gate-source voltages and drain-source voltages obtained in Step 3, and define the change in the trap amplitude with respect to the gate-source voltage VG as a, and the change in the trap amplitude with respect to the drain-source voltage VD as b. If 0.5 ≤ a / b, then the trap is affected by both the gate-source voltage VG and the drain-source voltage VD, and it can be obtained that the trap is in the AlGaN layer; if a / b < 0.5, then the trap is affected much more by the drain-source voltage VD than by the gate-source voltage VG, and it can be obtained that the trap is in the GaN layer.

[0011] The present invention first proposes a method for locating the trap distribution inside the device by collecting the changes in the trap amplitude under different filling voltages. This method can obtain the trap amplitude sizes under different filling voltages, realize the trap location inside the GaN HEMT device, and thus effectively characterize the trap distribution inside the device. Description of the Drawings:

[0012] Figure 1 The transient voltage response test conditions corresponding to Step 1 and Step 2, the test result diagrams, the time constant spectra and the differential amplitude spectra;

[0013] Figure 2 The transient voltage response test result diagrams, the time constant spectra and the differential amplitude spectra under different filling voltages corresponding to Step 3;

[0014] Figure 3 The dotted line relationship diagram of the absolute amplitude of the trap versus the filling voltage corresponding to Step 4. Detailed Embodiments:

[0015] The following gives an example of analyzing the trap position distribution inside the GaN HEMT device based on the changes in the trap amplitude under different filling voltages, which is only used to explain the present invention and should not be construed as a limitation to the present invention.

[0016] 1. Place the GaN HEMT device under test on a constant temperature platform at 298K. During the trap filling stage, apply a gate-source voltage VG of -7V and a drain-source voltage VD of 10V to the device, and the filling time t1 is 30s. After 30s of filling, test the trap recovery process. Apply a drain-source current ID of 200mA under a gate-source measurement voltage VGS of 0V, and simultaneously monitor the curve of the drain-source voltage changing with time. The obtained transient curve is the transient voltage response VDS of the trap recovery process, and the trap recovery time t2 is 120s. The test conditions and test result diagrams of the transient voltage response are respectively as shown in Figure 1 (a)(b).

[0017] 2. Perform Bayesian iteration on the obtained transient voltage response curve to obtain the corresponding time constant spectrum as shown in Figure 1 (c). There are two peaks in the time constant spectrum, which are the two traps contained in the transient voltage curve. The abscissa parameter corresponding to the peak is the time constant of the trap. Among them, the trap with a smaller time constant is named DP1, and the trap with a larger time constant is named DP2. Accumulate the y values of the time constant spectrum, that is, let Subsequently, exchange the horizontal and vertical coordinates and perform differential processing on the abscissa at this time to obtain the differential amplitude spectrum DAS as shown in Figure 1 (d). Under this filling voltage condition, the absolute amplitude of DP1 is 3.34%, and the absolute amplitude of DP2 is 12.72%. The sum of the absolute amplitudes of the two traps is the overall change amount of the transient voltage curve, which is 16.06%.

[0018] 3. Keep the drain-source voltage VD at 5V and repeat the measurement process in step 1 under different gate-source filling voltages VG (-7V, -8V, -9V, -10V). The obtained transient voltage response curves are as shown in Figure 2 (a); Keep the gate-source voltage VG at -7V and repeat the measurement process in step 1 under different drain-source filling voltages VD (5V, 7V, 9V, 10V, 12V, 15V). The obtained transient voltage response curves are respectively as shown in Figure 2 (b). Repeat the processing process in step 2 to obtain the time constant spectra under different gate-source filling voltages VG and different drain-source filling voltages VD and read the corresponding time constants, as shown in Figure 2 (c)(d); Obtain the differential amplitude spectra under different gate-source filling voltages VG and different drain-source filling voltages VD and read the corresponding absolute amplitudes, as shown in Figure 2 (e)(f).

[0019] 4. Plot the absolute amplitude changes of the traps under different gate-source voltage and drain-source voltage filling conditions as dot-line diagrams, as shown in Figure 3(As shown in (a) and (b), the position distribution of traps can be further analyzed by analyzing the variation of the absolute amplitude of traps with the gate-source voltage and the drain-source voltage. Among them, the trap filling in the AlGaN layer is affected by both the lateral electric field and the longitudinal electric field under the bias of the gate-source voltage and the drain-source voltage, and the trap filling in the GaN layer is determined by the lateral electric field under the bias of the drain-source voltage. From Figure 3 (a) and (b), it can be seen that the variation of the amplitude of the DP1 trap with the drain-source voltage VG is 0.012, and the variation with the drain-source voltage VD is 0.028. a / b = 0.42 < 0.5. The DP1 trap is affected much more by the drain-source voltage VD than by the gate-source voltage VG. Therefore, DP1 is located in the GaN layer. The variation of the amplitude of the DP2 trap with the drain-source voltage VG is 0.06, and the variation with the drain-source voltage VD is 0.09. a / b = 0.66 > 0.5. The DP1 trap is affected by both the drain-source voltage VD and the gate-source voltage VG. Therefore, the DP2 trap is located in the AlGaN layer.

Claims

1. A method for characterizing the trap position distribution of a GaN HEMT device using amplitude response, characterized in that: 1) Place the device under test on a constant temperature platform at temperature T1; during the trap filling stage, apply a gate-source voltage VG and a drain-source voltage VD to the device, and the filling time is t1; after the end of t1, test the trap recovery process. Apply a drain-source current ID of 200 mA at a gate-source measurement voltage VGS of 0 V. The trapped electrons are released from the traps and cause a change in the drain-source voltage. Monitor the curve of the drain-source voltage of the device under test changing with time until the transient drain-source voltage becomes stable and unchanged, and the transient voltage response curve VDS of the trap recovery process can be obtained. The trap recovery time is denoted as t2; 2) Perform Bayesian iteration on the transient voltage response curve obtained in step 1) to obtain the corresponding time constant spectrum. The n peaks in the time constant spectrum correspond to the n traps contained in the transient voltage curve. Among them, the parameter corresponding to the peak of the time constant spectrum on the x-axis is the characteristic time constant τ of the trap, and the parameter on the y-axis is the relative amplitude action intensity corresponding to the trap. Accumulate the y values of the time constant spectrum, that is, let Then, after exchanging the horizontal and vertical coordinates, perform differential processing on the abscissa at this time to obtain the differential amplitude spectrum DAS. The difference between the abscissas corresponding to adjacent peaks of the differential amplitude spectrum DAS is the absolute amplitude Amp of the trap, and the sum of the absolute amplitudes of the n traps is the total change in the transient voltage curve caused by trap filling; 3) Keep the gate-source voltage VG unchanged, and repeat the measurement process in step 1) at M drain-source voltages VD1, VD2, ……, VDM; keep the drain-source voltage VD unchanged, and repeat the measurement process in step 1) at N gate-source voltages VG1, VG2, ……, VGN; repeat the processing process in step 2) for the obtained M and N transient voltage curves respectively to obtain the corresponding time constant spectra and differential amplitude spectra, so as to extract the time constants and absolute amplitudes corresponding to the traps at M drain-source voltages and N gate-source voltages respectively; 4) Plot the dot-line relationship diagram of the absolute amplitude of the trap versus the filled gate-source voltage and drain-source voltage; since the electric field distribution affects trap filling, the trap filling in the AlGaN layer is affected by both the lateral and longitudinal electric fields brought by the gate-source voltage VG and the drain-source voltage VD, and the trap filling in the GaN layer is determined by the lateral electric field brought by the drain-source voltage VD. Analyze the absolute amplitudes of the traps at different gate-source voltages and drain-source voltages obtained in step 3), and define the change amount of the trap amplitude with respect to the gate-source voltage VG as a, and the change amount of the trap amplitude with respect to the drain-source voltage VD as b; if 0.5 ≤ a / b, then this trap is affected by both the gate-source voltage VG and the drain-source voltage VD, and it can be obtained that this trap is in the AlGaN layer; if a / b < 0.5, then this trap is affected far more by the drain-source voltage VD than by the gate-source voltage VG, and it can be obtained that this trap is in the GaN layer.

2. A method for characterizing the trap position distribution of a GaN HEMT device using amplitude response according to claim 1, characterized in that: During the trap filling stage, apply a gate-source voltage VG and a drain-source voltage VD to fill the inside of the device, where the drain-source voltage VD ranges from 0 V to 20 V, the gate-source voltage VG ranges from -15 V to 0 V, and the filling time t1 ranges from 1 ms to 100 s.

3. A method for characterizing the trap position distribution of a GaN HEMT device using amplitude response according to claim 1, characterized in that: During the trap recovery process, the gate-source measurement voltage VGS is 0 V to ensure that the device channel is in the open state and the gate-source voltage VGS has no influence on the trap recovery process; the drain-source current ID is 200 mA to ensure that the device is in the linear region and the influence of self-heating temperature rise on trap recovery can be ignored, i.e., △T < 1 °C; the filling and recovery time t2 ranges from 1 ms to 300 s.

Images