SiC Diodes, SiC MOSFETs, IGBTs & Power Modules Manufacturer | Hypersemi
TechForum - application notes and selection guides

TECH FORUM

Leakage Doubled: Is the Transistor Faulty or Is Cooling Insufficient?

——Three Steps to Separate Thermal Runaway from Avalanche Damage in SiC SBDs: What to Check First, What to Check Next

Leakage Doubled: Is the Transistor Faulty or Is Cooling Insufficient?

When a SiC SBD exhibits increased leakage or reverse breakdown in a PV field application, the first fork an engineer faces is: thermal runaway, or avalanche damage? The former points to cooling and derating; the latter points to loop inductance and clamping. The failed parts may look indistinguishable, but the direction of the test data gives different answers. This article uses a three-step method to separate these two root causes at the test level, rather than jumping straight to "replace the

1. Why "High Leakage" Does Not Automatically Mean "The Device Is Damaged"

SiC SBD reverse leakage depends exponentially on Schottky barrier height. As temperature rises, thermionic emission current increases and leakage rises—this is part of normal operation, not necessarily damage. What needs to be distinguished are two qualitatively different abnormalities:

Thermal runaway is a positive-feedback process in which leakage generates heat, and heat further amplifies leakage. IEEE literature describes this mechanism as a "fatal feedback loop" between high leakage current and self-heating. The underlying problem is insufficient thermal path or derating margin, causing the device to lose thermal equilibrium near its operating temperature.

Avalanche damage occurs when reverse voltage exceeds VRRM, driving the device into avalanche breakdown and absorbing excess energy that damages the junction. The underlying problem is an unclamped turn-off voltage spike in the loop that exceeds the device's E_AS boundary.

The corrective directions are opposite. Thermal runaway calls for more cooling and adjusted derating; avalanche damage calls for better clamping and reduced loop inductance. Replacing the device with a higher-current part before separating the root cause may help one scenario and do nothing for the other.

2. Three Steps: Baseline First, Slope Second, Loop Retrospection Last

  • Step 1: Measure 25°C reverse leakage after power-down to establish a "damage baseline."

After the device cools to room temperature, measure reverse leakage at VR and compare it with the initial datasheet value. Reference criterion: an IR increase of more than one order of magnitude suggests the junction has experienced a non-negligible damage event (avalanche or over-temperature history). This step answers "has the device been damaged," not "why."

  • Step 2: Re-measure IR at 125°C/150°C—look at the slope, not the absolute value.

If room-temperature leakage has already deviated significantly, damage is present. If room-temperature leakage is still acceptable but high-temperature leakage rises super-exponentially, thermal runaway risk becomes dominant. IEEE research provides a reference logic: for SiC Schottky and PiN diodes, the temperature difference ΔT_d required to double leakage current is significantly higher than for Si PiN diodes, but this advantage must be reassessed at higher voltages or higher operating temperatures. In PV Boost or inverter stages with sustained high case temperatures, the high-temperature IR slope is closer to real lifetime risk than a single room-temperature point.

  • Step 3: Check whether the turn-off voltage spike exceeds VRRM without clamping.

If the first two steps indicate "damage present" and avalanche cannot be ruled out, the third step turns to the loop: measure or reconstruct the reverse voltage peak the diode sees at turn-off. When the peak exceeds VRRM and there is no effective clamping or snubber path, E_AS becomes the primary suspect. SiC SBD avalanche failure can occur in a very short time—recorded avalanche failure energy densities are on the order of several J/cm² with failure times below 10μs. This means a single unclamped lightning-induced or grid-disturbance event can complete the damage without repeated accumulation.

3. Hypersemi's Entry: Grounding the Three Steps in Verifiable Rth(j-c) and E_AS

The three-step method needs a carrier: a device whose Rth(j-c) and E_AS are explicitly specified in the datasheet. Hypersemi CR20120L2X100, 1200V / IF 20A / VF 1.4V / TO-247-2L, is used in Boost and PFC applications, and its Rth(j-c) and avalanche-related parameters can be extracted directly from the datasheet for thermal-runaway and avalanche boundary accounting. The same voltage tier includes CR15120L2X100 (1200V / IF 15A / VF 1.4V / TO-247-2L), covering lower-current auxiliary or branch positions.

Model

VRRM / IF / VF

Package

Role in the Three-Step Method

CR20120L2X100

1200V / 20A / 1.4V

TO-247-2L

Primary: Boost/inverter stage; verifiable Rth(j-c) and E_AS for thermal and avalanche boundary accounting

CR15120L2X100

1200V / 15A / 1.4V

TO-247-2L

Secondary: lower-current branch; comparison entry under the same diagnostic framework

For 650V/TO-220F-2L scenarios in junction boxes or auxiliary stages, "junction-to-ambient" thermal resistance is more easily underestimated by potting and mounting; the diagnostic logic is unchanged, but the thermal-runaway side carries more weight.

Measure room-temperature leakage first to establish a baseline; measure high-temperature leakage next to see whether the slope points to thermal runaway or to damage; finally, trace the turn-off spike against VRRM to confirm whether avalanche is in scope. Only after these three steps should the decision be made—fix the cooling or fix the clamping. Hypersemi CR20120L2X100 (1200V/20A/TO-247-2L) and CR15120L2X100 (1200V/15A/TO-247-2L) can be filtered by VRRM, IF, and package on the SiC diode product page (https://www.hypersemi.com.cn/products/sic-diodes), bringing the diagnosis from experience back to verifiable parameters.

Device parameters are subject to Hypersemi official datasheets. Root-cause determination must be confirmed at the system level in conjunction with the specific circuit, thermal design, and test conditions.


Back to top