This page was machine-translated and may differ from the original. View original

[Technical contribution] Infineon Technologies - Reliability verification based on understanding the characteristics and operation of wide bandgap devices

Google 우선 소스Published2022.11.25 12:05
Ensuring WBG semiconductor reliability requires specialized technology

High possibility of defects, process optimization
and worst-case scenario preparation required
Infineon Completes Test and Device Reliability Verification Beyond Design Limits

e4ds&utm_medium=referral&utm_campaign=202301_ap_kr_pss_pss.ppwbg_p&utm_content=e4ds+media-jan-mar&utm_term=tech+articles+series-wbg" target="_blank">WBG devices, new approach to life cycle modeling required

Gallium nitride (GaN) and silicon carbide (SiC) devices are increasingly replacing silicon in power electronics applications due to their unique properties.

ppwbg_p&utm_content=e4ds+media-jan-mar&utm_term=tech+articles+series-wbg" target="_blank">SiC devices enable unipolar devices with low switching losses even at high voltages, while GaN devices can switch at very high frequencies.

These characteristics enable smaller, more power-efficient designs.

However, since these technologies have only recently become mass-producible, little information is known about their long-term performance in the field.

Silicon devices have well-developed long-term lifetime modeling, which allows for accurate prediction of reliability, but wide-bandgap devices have different physical structures, requiring new approaches to device lifecycle modeling.

In this article, we will discuss reliability.s-wbg" target="_blank">Learn about the characteristics and operation of SiC and GaN devices.

Describes the processes and test procedures developed by Infineon to predict the long-term operation of wide bandgap devices and to achieve reliability over the entire product life cycle in the field.

■ Characteristics and defect mechanisms of SiC devices
○ Gate oxide film

SiC MOSFETs have a much higher extrinsic defect density, making them up to four times more prone to failure than silicon devices without any additional measures.

Extrinsic defects in the gate oxide film can be modeled through a localized thinning effect.

This phenomenon causes certain A stronger electric field can be applied at the gate bias voltage, which can cause the oxide film to be destroyed earlier (Figure 1).

▲Figure 1: Extrinsic defects in the amorphous layer can be mathematically modeled through the local thinning phenomenon of the gate oxide film of a SiC device.

Infineon uses innovative inspection techniques during testing to screen and isolate potentially vulnerable SiC devices.

All devices are tested by applying a stress pulse pattern designed to destroy devices with potential extrinsic faults, but to ensure that devices without faults operate normally.

Bias Temperature Instability also affects reliability. Thermal runaway can occur due to a strong electric field between the gate oxide, causing a change in the device threshold voltage and consequently the on-resistance.

Temperature instability caused by static voltage is well known, but device switching during transition can also act as a factor that changes the threshold voltage of SiC devices.

This amount of drift depends on the turn-on bias voltage and switching frequency.

This drift can be mitigated by ensuring that the device always stays within its safe operating area (SOA).

Infineon recently introduced a technologically improved gate oxide formation process that significantly reduces the impact of dynamic drift on threshold voltage instability.

○ Body diode

The body diode of a SiC MOSFET can experience bipolar degradation due to defects originating from the substrate material penetrating into the active layer (epilayer, drift region).

Then, stacking faults can grow inside the drift region, which leads to a reduction in the effective device area.

This again leads to a change in RDS(ON) over time and an increase in the VSD of the body diode.

Then, when the stacking fault reaches the wafer surface, growth stops (saturation).

The time taken here depends on the current passing through the PN junction and the junction temperature.

Fundamental parameters such as breakdown voltage, leakage current and switching behavior are not affected.

Infineon has optimized the process to suppress stacking fault formation.

As mentioned above In addition to technical measures, additional inspection tests are applied to finished devices to identify devices that are experiencing stacking defects.

This ensures that SiC devices can operate for long periods without violating datasheet limits.

○ Cosmic ray

In blocking mode, the power device operates under strong internal electric fields.

SiC has this field nearly ten times higher than silicon. Particles from cosmic rays are permanent in the Earth's atmosphere and can penetrate the active region of a power device, causing local carrier multiplication and ultimately destroying the device.

This is called Single Event Burnout. The robustness against SEB at a specific reverse bias voltage can be improved by increasing the thickness of the base layer and reducing the doping.

However, this has the hidden pitfall of increasing on-state losses instead of enabling higher avalanche breakdown voltages.

The rate of defects due to cosmic radiation increases exponentially with increasing altitude.

Therefore, calculations must be made based on the needs of a specific application, including the applied voltage, altitude, and required lifespan.

Infineon's experienced application engineers help calculate overall fault rates based on test data, application details, and mission profiles.

Additionally, Infineon designs its devices to meet fault rate requirements based on worst-case mission profiles for key target applications.

○ Robustness Verification

Infineon has developed a quality assurance program based on more than 25 years of experience in SiC-based power devices and its participation in numerous standards organizations such as JEDEC International Standards.

This approach describes the product life cycle using a bathtub-shaped curve (Figure 2). First, we select devices with a high probability of defects through intensive screening.

All standard tests were performed for a minimum of 1,000 hours, later increased to 5,000 hours.

This is well beyond the time required by the standard, allowing potential end-of-life (EOL) behavior to be captured.

Comprehensive testing to verify device robustness demonstrates that all CoolSiC™ MOSFETs are robust, with no system-level EOL mechanisms found.

▲Figure 2: The overall product life cycle is described using a ‘bathtub’ shaped curve.

nt=e4ds+media-jan-mar&utm_term=tech+articles+series-wbg" target="_blank">Operation and failure mechanisms of GaN devices

GaN devices based on the high electron mobility transistor (HEMT) structure behave very differently from silicon-based devices.

Because of these differences, GaN HEMT devices are more sensitive than silicon to various stresses in power conversion applications.

○ Test profile according to application

To qualify these devices, Infineon examines potential failure modes using application-specific methods.

For example, telecom rectifier (AC-DC) sockets must withstand harsh operating conditions, operating for long periods of time in a wide range of ambient temperatures, including high temperatures.

Since this application is a representative use case for GaN devices, Infineon decided to use this application to qualify GaN devices.

Table 1 summarizes the switching behavior of power transistors in the power factor correction (PFC) design of a 2.5 kW telecom rectifier system.
k">
▲Table 1: Device switching behavior in telecom rectifier applications (voltage and current data for one temperature)

○ Drain-to-source bias voltage sensitivity

GaN HEMT devices exhibit a defect rate that is highly correlated with drain-source bias voltage when subjected to accelerated testing under voltage and temperature conditions above their nominal rated values.

In other words, the higher the drain-source bias voltage, the faster the defect occurs.

For example, when device samples were tested at 800 V and 125 C, all devices failed within just 30 hours, but lowering the bias voltage and temperature increased the time until failure occurred.

This test data can be used to predict defects under real-world usage conditions.

At 480 V and 125 C, it can be predicted that there will be no failure for up to about 2 million hours (Figure 3).
px; height: 310px;" />
▲Figure 3: Cumulative failure probability according to failure time with different DC biases (five lines on the left) and device model (dotted line) when operating at 480 V 125 C.

Although GaN devices can withstand much higher voltage stresses than silicon devices, higher voltages decrease their lifetime.

This model shows that the lifespan is three times longer than the target requirement. On this graph, the green stars represent the probability of a defect that can be detected using verification techniques currently used in silicon (which are insufficient to predict the lifetime of GaN devices).

The gray star indicates a target life of 15 years at a defect rate of 100 ppm (parts per million).

○ Hard switching

SOA switching, also known as dynamic high temperature operating life (DHTOL), is another important mechanism that degrades GaN device performance.

Many GaN device companies are publishing long-term application switching data.

This shows that the device can operate reliably for 1,000 to 3,000 hours (18 to 60 weeks) in hard switching applications (measured at constant case temperature).

On the other hand, this amount of time is insufficient to draw any conclusions about the performance of a device that is expected to operate over a lifespan of several years.

Accordingly, Infineon developed a test platform to test devices at higher voltages and currents than originally designed to verify device reliability under long-term hard switching operation.

This accelerated stress test operates the device in a hard switching boost configuration, similar to that used for PFC.

Damage to the device under test conditions was well controlled to enable subsequent defect analysis.

This test platform uses bus voltages up to 700 V and currents exceeding the device rating (typical usage conditions are 420 V).

When voltage and current exceed certain thresholds, the devices begin to exhibit defects.

From this information, current, voltage, we derived a model that can predict the defect rate according to frequency.

Figure 4 shows the results of this model as a switching SOA curve.

Therefore, circuit designers can take note of this and operate the device to avoid current and voltage combinations that could lead to premature failures.

▲Figure 4: Hard switching SOA curve, which shows which section to operate in to avoid premature failures.

Accelerated tests were also conducted with a soft switching configuration similar to that used in DC-DC converters employing LLC architecture for CoolGaN™ HEMTs.

No failures were observed under (or beyond) the operating conditions that previously caused failures during accelerated hard switching.

This shows that the switching method plays an important role in device failures (Figure 5).
ndgap-semiconductors-sic-gan?utm_source=e4ds&utm_medium=referral&utm_campaign=202301_ap_kr_pss_pss.ppwbg_p&utm_content=e4ds+media-jan-mar&utm_term=tech+articles+series-wbg" target="_blank">
▲Figure 5: SOA faults can be prevented by using soft switching, but caution is required when using hard switching.

Infineon provides the latest information on WBG devices

Wide bandgap devices (SiC and GaN) are relatively Because mass production has only recently become possible, there is limited information on reliability over the product's lifespan.

Infineon has developed a test and validation process based on its deep understanding of the characteristics and behavior of CoolSiC™ and CoolGaN™ devices.

This ensures that reliability issues are not hastily raised when the device is put into use in the field.

Visit Infineon's wide bandgap technology website for more information about our portfolio of silicon carbide and gallium nitride-based devices for applications requiring high system performance.

※ author
Peter Friedrichs, Vice President SiC;
Tim McDonald, Senior Director, Consulting Advisor to the CoolGaN™ program,
Infineon Technologies


※ References
- Infineon Technologies, “How Infineon controls and assures the reliability of SiC based power semiconductors”, Whitepaper, August 2020.
- Infineon Technologies, “Reliability and qualification of CoolGaN™, Technology and devices”, Whitepaper, October 2018.
본 기사에 대한 정정·반론·추후보도 청구는 보도 청구 안내를, 그간 게재된 보도문은 정정·반론보도 모아보기를 참고해 주세요.
명세환 기자
명세환 기자