An Accurate Measurement Method for Transient Threshold Voltage Changes of p-GaN HEMTs

Abstract

The present invention discloses a method for accurately measuring the transient threshold voltage change of a p-GaN HEMT. This method can be used to measure the transient change of the threshold voltage of a p-GaN HEMT device under electrical bias. Before applying the gate voltage bias and testing the transient threshold voltage, a test for the transient voltage change introduced by the sampling current is designed and added, and the test conditions are consistent with the test conditions of the threshold voltage transient curve. After the transient threshold voltage curve is measured, this additional factor is removed from the measured curve. This method can correct the effect of the sampling current on the transient threshold voltage change to obtain an accurate value of the threshold voltage change over time, thereby correctly evaluating the threshold voltage drift and trapping effect of the p-GaN HEMT device under electrical bias.

Description

Technical Field

[0001] The present invention relates to accurate measurement of transient threshold voltage changes of p-GaN HEMTs, and belongs to the field of electrical characteristic testing and reliability of power semiconductor devices. Background Art

[0002] Enhancement-mode GaN-based devices have a wide range of applications in the power switching field. Schottky-mode p-GaN HEMTs, in particular, offer low switching losses and minimal gate leakage current, enabling widespread commercialization. However, the gate structure of p-GaN HEMTs is more complex, and trap effects under gate voltage can cause device threshold voltage drift, potentially reducing switching efficiency and device stability, posing a serious risk to the reliability of p-GaN HEMT devices and related systems.

[0003] The threshold voltage drift of p-GaN HEMTs under positive gate voltage has been extensively studied, and various threshold voltage measurement techniques have been developed. The device threshold voltage can be obtained by using transfer characteristics to read the gate-source voltage at a constant drain-source current (e.g., 1 mA). When the gate-source voltage and drain-source voltage are equal, the device threshold voltage can also be obtained by applying a constant sampling current to the device's drain-source terminals. Both methods provide convenient and rapid access to the device's static threshold voltage, but the transient variation of the threshold voltage under electrical stress requires further characterization. Pulsed transfer characteristics can be used to measure the temporal variation of the p-GaN HEMT threshold voltage under electrical bias. This test method requires a specific pulse duty cycle, resulting in a scatter plot of the threshold voltage variation rather than a continuous curve. Alternatively, the transient threshold voltage variation can be measured by applying a constant sampling current while keeping the device gate and drain short-circuited. This method produces a continuous threshold voltage curve. However, in actual testing, the constant sampling current can cause trap filling and voltage variations. This additional transient variation is included in the measured transient voltage curve, introducing errors. Therefore, it is necessary to develop accurate and non-destructive p-GaN HEMT threshold voltage transient testing technology.

[0004] This technology proposes a correction method for measurement errors in transient threshold voltage testing of p-GaN HEMTs. This method evaluates the transient voltage variation caused by trap effects under applied sampling current conditions and removes this additional effect from the transient voltage curve of the p-GaN HEMT under gate voltage stress. This method provides a more accurate estimate of the transient threshold voltage variation of the p-GaN HEMT, providing a technical means for accurately analyzing the internal trap effects and trap-related parameters of the p-GaN HEMT. This method is applicable to transient threshold voltage testing of p-GaN HEMT devices from different manufacturers and has a promising application foundation. Summary of the Invention

[0005] The transient threshold voltage change of a p-GaN HEMT under gate voltage is typically characterized by the transient voltage change caused by the sampling current after the device gate and drain are short-circuited. However, during testing, the sampling current applied to the p-GaN HEMT also triggers internal trapping effects. The measured transient threshold voltage curve includes the effects of both the threshold voltage change under gate voltage and the additional voltage change introduced by the sampling current. Further analysis based on the measured curve reveals possible errors in the trap parameters within the p-GaN HEMT under gate voltage. To address this issue, we designed and added a test and evaluation of the transient voltage change under sampling current before the traditional gate voltage application and threshold voltage test. We then subtracted this test result from the final measured threshold voltage transient curve to obtain an accurate p-GaN HEMT transient threshold voltage change curve and analyze it to obtain accurate information related to the traps.

[0006] The technical solution adopted by the present invention is a method for accurately measuring the transient threshold voltage change of a p-GaN HEMT. The implementation process of the method is as follows:

[0007] Step 1). Place the p-GaN HEMT device under test in an incubator at 303 K. First, apply a gate-drain voltage (VGD) between the device's gate and drain, while simultaneously applying a test current (Isense) between the device's source and drain. VGD is 0 V, and the test current (Isense) ranges from 0.1 mA ≤ Isense ≤ 20 mA. Monitor the change in the device's drain-source voltage (VSD) over time, and record the transient curve as VSD1. The test duration is tm1. After the VSD1 test, which lasts for tm1, the device under test is left unpowered for 10 minutes. The test duration, tm1, ranges from 100 s ≤ tm1 ≤ 1000 s, and the transient curve (VSD1) ranges from 0.1 V ≤ VSD1 ≤ 10 V.

[0008] Step 2). After the 10-minute rest period, apply a fill voltage, VGD_F, between the device gate and drain, while keeping the source and drain shorted (VSD = 0 V). At this point, the gate-drain voltage and the source-drain voltage of the device under test are equal, so the gate-source fill voltage, VGS_F, equals VGD_F. The fill time is tf. The fill voltage, VGD_F, ranges from 1V ≤ VGD_F ≤ 10V, and the fill time, tf, ranges from 10ms ≤ tf ≤ 1000s. After the fill time, tf, apply a gate-drain test voltage, VGD_M, of 0V between the device gate and drain, while simultaneously applying a test current, Isense, between the device source and drain. Monitor the change in the drain-source voltage, VSD, of the device under test over time, and record the transient curve as VSD2. The test time is tm2. The range of VSD2 is 0.1 V ≤ VSD2 ≤ 10 V, and the range of tm2 is 100 s ≤ tm2 ≤ 1000 s. The change of VSD over time is the measured change of the threshold voltage of the device under test over time, ΔVTH_measured, where the range of ΔVTH_measured is 0 V ≤ ΔVTH_measured ≤ 5 V.

[0009] Step 3). After the test duration tm2 is complete, compare the transient curves VSD1 and VSD2 showing the drain-source voltage change over time. The transient curve VSD1, obtained during the test duration tm1, reflects the effect of 0 V VGD and 1 mA Isense applied to the device on VSD. These test conditions are consistent with the 0 V VGD_M and 1 mA Isense applied to the device during tm2. The voltage difference between VSD1 and VSD2 at the same sampling time is VSD1-VSD2. The sampling time is considered to be the same if the difference in sampling time between VSD1 and VSD2 is less than 2 ms. The transient curve VSD2, obtained during the test phase after the fill voltage is applied, includes the effect of the test conditions. Therefore, ΔVTH_actual = VSD2-VSD1 represents the actual transient curve change caused by the fill voltage, where ΔVTH_actual is in the range of 0 V ≤ ΔVTH_actual ≤ 9.9 V.

[0010] Step 4). Change the fill time tf and repeat steps 1) through 3) to obtain the corresponding transient threshold voltage change. Using Bayesian iteration techniques, we can further construct a time constant spectrum and a differential amplitude spectrum to extract the trap time constant and amplitude information contained in the threshold voltage change. The absolute amplitude of the trap represents the exact contribution of the N traps to the threshold voltage change, where 1 ≤ N ≤ 8. The sum of the absolute amplitudes of the N traps is the actual threshold voltage transient change, ΔVTH_actual.

[0011] This invention proposes and corrects the effect of applied sampling current on the measured transient voltage curve during the transient threshold voltage test of a p-GaN HEMT. This method accurately captures the transient threshold voltage curve of a p-GaN HEMT under electrical stress and accurately assesses the threshold voltage variation of the p-GaN HEMT under different voltage biases. The present invention provides a convenient and rapid testing technique with simple testing conditions. Building on existing p-GaN HEMT threshold voltage testing methods, this method provides an accurate transient threshold voltage testing technique, providing an effective technical means for testing the electrical characteristics and reliability assessment of p-GaN HEMTs. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 : The transient threshold voltage test timing diagram involved in the present invention.

[0013] Figure 2 : Measured threshold voltage transient curve and accurate threshold voltage transient curve after processing.

[0014] Figure 3 : Accurate threshold voltage transient change curve under different filling time and the threshold voltage change caused by each trap. DETAILED DESCRIPTION

[0015] The following describes in more detail the accurate measurement of transient threshold voltage changes of a p-GaN HEMT, using the accompanying drawings and specific embodiments. A p-GaN HEMT manufactured by EPC, model EPC2007C, is used as the device under test. Its maximum operating voltage is 100 V and its maximum operating current is 6 A. The method of the present invention includes the following steps:

[0016] Step 1: Place the p-GaN HEMT device under test in an incubator at 303 K. First, apply a gate-to-drain voltage (VGD) of 0 V between the device's gate and drain, while simultaneously applying a test current (Isense) of 1 mA between the device's source and drain. Monitor the change in the device's drain-to-source voltage (VSD) over time, recording this transient curve as VSD1. The test duration (tm1) is 500 s. After the 500-s VSD1 test, the device under test is left unpowered for 10 minutes.

[0017] Step 2: After 10 minutes of rest, apply a 5 V fill voltage VGD_F between the gate and drain of the device, while keeping the source and drain of the device short-circuited, that is, VSD=0 V. At this time, the voltage between the gate and drain of the device under test is equal to the voltage between the source and drain, so the gate-source fill voltage VGS_F=VGD_F=5 V, and the fill time tf is 30 s. After the 30 s fill time tf is completed, a 0 V gate-drain test voltage VGD_M is applied between the gate and drain of the device, and a 1 mA test current Isense is applied between the source and drain of the device; monitor the change of the drain-source voltage VSD of the device under test over time and record the transient curve as VSD2. The test time tm2 is 500 s. At this time, the change of VSD over time is the measured change of the threshold voltage of the device under test over time ΔVTH_measured. The test timing is as follows: Figure 1 shown.

[0018] Step 3: After completing the 500 s test time tm2, compare the transient curves VSD1 and VSD2 of the drain-source voltage changing with time. The transient curve VSD1 obtained during the tm1 test time reflects the impact of 0 V VGD and 1 mA Isense applied to the device on VSD. This test condition is consistent with the 0 V VGD_M and 1 mA Isense applied to the device during tm2. The voltage difference between VSD1 and VSD2 at the same sampling time point is VSD1-VSD2, where the difference in the corresponding sampling time of VSD1 and VSD2 is less than 2 ms, which is the same sampling time point. The transient curve VSD2 obtained in the test phase after removing the fill voltage includes the influence of the test conditions, so ΔVTH_actual=VSD2-VSD1 represents the actual transient curve change caused by the fill voltage. The results are as follows. Figure 2 Under the fill voltage condition of VGS_F = VGD_F = 5 V, the threshold voltage change corresponding to VSD2 obtained after the tm2 test time is ΔVTH_measured; the corresponding threshold voltage change after removing VSD1 introduced during the tm1 test time is ΔVTH_actual.

[0019] Step 4: Change the filling time tf to 50 ms, 100 ms, 200 ms, 1 s, 5 s, 30 s and 300 s in sequence, and repeat steps 1 to 3 to obtain the corresponding transient threshold voltage changes as shown in the figure. Figure 3The Bayesian iteration technique can be used to further construct a time constant spectrum and a differential amplitude spectrum, from which the trap time constant and amplitude information contained in the threshold voltage change can be extracted. The absolute amplitude of the trap represents the exact contribution of the three traps to the threshold voltage change, and the sum of the absolute amplitudes of the three traps is the actual threshold voltage transient change ΔVTH_actual.

Claims

1. A method for accurately measuring transient threshold voltage changes of a p-GaN HEMT, characterized by: Step 1) The p-GaN HEMT device under test is placed in an incubator at 303 K. A gate-to-drain voltage VGD of 0 V is applied between the gate and drain of the p-GaN HEMT device under test, while a test current Isense is applied between the source and drain of the device. The change in the drain-to-source voltage VSD of the p-GaN HEMT device under test over time is monitored and the corresponding transient curve is recorded as VSD1. The test time is tm1. After the VSD1 test is completed for time tm1, the p-GaN HEMT device under test is kept in an unpowered state and allowed to stand for 10 minutes. Step 2) After the 10-minute rest period, a fill voltage VGD_F is applied between the device gate and drain, while the source and drain of the p-GaN HEMT device under test are kept short-circuited, i.e., VSD = 0 V. At this point, the gate-drain voltage and the source-drain voltage of the p-GaN HEMT device under test are equal, so the gate-source fill voltage VGS_F = VGD_F, and the fill time is tf. After the fill time tf is complete, a gate-drain test voltage VGD_M of 0 V is applied between the device gate and drain, while a test current Isense is applied between the source and drain of the p-GaN HEMT device under test. The change in the drain-source voltage VSD of the p-GaN HEMT device under test over time is monitored, and the corresponding transient curve is recorded as VSD2. The test time is tm2. The change in VSD over time is the measured change in the threshold voltage of the p-GaN HEMT device under test over time, ΔVTH_measured. Step 3) After the test time tm2 is completed, compare the transient curves VSD1 and VSD2 of the drain-source voltage over time. The transient curve VSD1 obtained during the test time tm1 reflects the impact of 0 V VGD and 1 mA Isense applied to the p-GaN HEMT device under test on VSD. This test condition is consistent with the 0 V VGD_M and 1 mA Isense applied to the p-GaN HEMT device under test during the test time tm2. The voltage difference between VSD1 and VSD2 at the same sampling time point is VSD1-VSD2, where the difference in the corresponding sampling times of VSD1 and VSD2 is less than 2 ms, indicating the same sampling time point. The transient curve VSD2 obtained during the test phase after the fill voltage is applied includes the impact of the test conditions. Therefore, VSD2-VSD1 represents the actual transient curve change ΔVTH_actual caused by the fill voltage. Step 4) Change the fill time tf and repeat steps 1) to 3) to obtain the corresponding transient threshold voltage change; use Bayesian iteration to construct a time constant spectrum and a differential amplitude spectrum, from which the trap time constant and amplitude information contained in the threshold voltage change are extracted; the absolute amplitude of the trap represents the exact contribution of the N traps to the threshold voltage change; and the sum of the absolute amplitudes of the N traps is the actual threshold voltage transient change.

2. The method for accurately measuring transient threshold voltage changes of a p-GaN HEMT according to claim 1, characterized in that: In the VSD1 test phase described in step 1), the gate-drain voltage VGD of the p-GaN HEMT device under test is 0 V, the test current Isense ranges from 0.1 mA ≤ Isense ≤ 20 mA, the test time tm1 ranges from 100 s ≤ tm1 ≤ 1000 s, and the transient curve VSD1 ranges from 0.1 V ≤ VSD1 ≤ 10 V.

3. The method for accurately measuring transient threshold voltage changes of a p-GaN HEMT according to claim 1, characterized in that: In step 2), the fill voltage VGD_F is in the range of 1 V ≤ VGD_F ≤ 10 V. At the same time, the drain and source of the p-GaN HEMT device under test are short-circuited to ensure that VGD_F = VGS_F, which is the gate fill voltage applied to the device. The fill time tf is in the range of 10 ms ≤ tf ≤ 1000 s.

4. The method for accurately measuring transient threshold voltage changes of a p-GaN HEMT according to claim 1, characterized in that: In step 2), during the VSD2 test, the range of VSD2 is 0.1 V≤VSD2≤10 V, the range of tm2 is 100 s≤tm2≤1000 s, and the range of ΔVTH_measured is 0 V≤ΔVTH_measured≤5 V.

5. The method for accurately measuring transient threshold voltage changes of a p-GaN HEMT according to claim 1, characterized in that: The test current Isense applied during the VSD1 and VSD2 tests is the same, and the test time tm1 = tm2. The voltage difference at the same sampling time point is VSD1 - VSD2. If the difference in the corresponding sampling times of VSD1 and VSD2 is less than 2 ms, they are considered to be at the same sampling time point.

6. The method for accurately measuring transient threshold voltage changes of a p-GaN HEMT according to claim 1, characterized in that: ΔVTH_actual=VSD2-VSD1 is the actual transient curve change ΔVTH_actual caused by the fill voltage, where the range of ΔVTH_actual is 0 V≤ΔVTH_actual≤9.9 V.

7. The method for accurately measuring transient threshold voltage changes of a p-GaN HEMT according to claim 1, characterized in that: The exact contribution of the N traps to the threshold voltage change is the absolute amplitude of the traps obtained using the differential amplitude spectrum, where 1≤N≤8; the sum of the absolute amplitudes of the N traps is the actual transient change of the threshold voltage.

Images