With the rise of new energy technologies and the global trend in energy conservation and carbon reduction, leading automobile manufacturers are all including the promotion and expansion of Electric Vehicle (EV) functions in their research and development blueprints. According to predictions made by Yole, the global average compound annual growth rate (CAGR) for all types of EV markets between 2021 and 2027 was 21%. Among electric vehicle components, power components such as DC-DC converters, car chargers and inverters have also seen an increase in applications. Although the IC components for these power components are still far from mature, its CAGR is expected to see double digit growth until 2027. Take the SiC module, for example, its revenue could reach 4.4 billion USD with a CAGR of 38% (Figure 1) by 2027. Therefore, power devices will be one of the centers of development for various semiconductor industry chains in the future.
 Figure 1. Revenue and CAGR of Various Power Devices from 2021 to 2027 [1] |
|
The growing demand for power in the field of automotive power devices is gradually drawing more and more attention to third generation semiconductors. |
In contrast to the traditional silicon-based power devices, third generation semiconductors use wide bandgap materials such as silicon carbide (SiC) and gallium nitride (GaN), which enable a higher operating range in terms of both frequency and power. Therefore, they can be applied in many emerging industries, such as self-driving cars, 5G/6G, space, AI, high-speed quantum computing, and power generation, etc.. In view of this, China announced its “14th Five-Year Plan” in 2020, which incorporated third generation semiconductors into its industrial development. It anticipates investing 10 trillion RMB between 2021 and 2025 into making its third generation semiconductor industry independent in order to avoid being constrained by the handful of top western manufacturers who currently lead the global market.
Many Taiwanese manufacturers of traditional power devices have announced that they will start production of SiC devices in the second half of 2022, injecting new energy into the entire third generation semiconductor market. Figure 2 illustrates the operating frequency and power range of power devices made of various materials as well as their fields of application.
  Figure 2. The Operating Range of Various Power Devices; Their Applications in Home Appliances, Electric Vehicles, Railways, Power Plants and More [1] |
Device types and voltage ranges vary depending on the application. Figure 3 is the division of devices for different voltages. The range of today’s mainstream automotive power devices is below 900V. These can be supported with traditional Si and GaN MOSFETs. However, those applications that demand more than 1200V, namely railway and power plant applications, require IGBT or SiC devices to operate.
 Figure 3. Operating Voltage of Power Devices of Various Types of Materials [2] |
The increase in the output of power devices naturally leads to an increase in the demand for analysis and testing. In the field of failure analysis, understanding the device structure and electrical measurement are basic steps for getting started. Although the structure of power devices is simpler than that of ICs, their materials and the layout of metal connections are important factors that affect sample preparation and defect observation. In terms of electrical measurement, their unique specifications make it impossible to use a general parameter analyzer to confirm their failure behaviors. Therefore, the analysis requires the use of high-powered measuring instruments. Based on the above considerations, the overall analysis process can be summarized in three steps:
1. Electrical Parameter Measurement
2. Fault Isolation
3. Defect Observation
|
Electrical Parameter Measurement |
IC measurements can be divided into static tests and dynamic tests. DC measurement is the former. Both open/short and leak Hi/Lo fall into this category. Third party analysis laboratories tend to use reliable parameter analyzers for verification, and the locations of defects are mainly determined using static tests. Dynamic testing is essentially function testing and requires ATE or bench tests to complete. Different types of ICs have different types of testing procedures, and, generally, third party analysis laboratories do not have all the necessary capabilities. Therefore, most cannot handle the entire function failure analysis process.
Due to the simple structures and fixed electrical parameters of power components, there is already an instrument on the market that is able to handle all the relevant measurements. The electrical parameters are clearly defined in the manual specifications. As long as all the steps in the manual are followed, the values of each item can be extracted one by one. First it is necessary to understand exactly which electrical parameters need to be measured. Take, for example, the 600V MOSFET. Its electrical parameters and their definitions are shown in Figure 4. Once you understand the definitions of the electrical parameters, it becomes possible to speculate which structures are at fault when a parameter is abnormal. This facilitates the subsequent formulation of physical failure analysis plans.
 Figure 4. 600V MOSFET Electrical Parameters and Definitions |
|
In preparation to welcome the arrival of the era of third generation semiconductors, MA-tek spared no efforts to acquire B1506A electrical parameter analyzers for its Taiwan and Shanghai laboratories (Figure 5). It can automatically capture all electrical parameters and measure most specified parameters, including those items essential to failure analysis. The B1506A also has the following features: (1) Suitable for use with all discrete power devices (2) As long as the settings are completed, all of the parameters to be measured can be captured automatically in one go. (3) With a maximum operating voltage of 3KV and a maximum current of 500A, it is compatible with most power devices on the market. (4) An external probe station can be connected for wafer-level and die-level measurements. ((5) Has the ability to quickly determine failure mechanisms for product development and customer-returned product analysis |
 Figure 5. The B1506A |
|
Table 1. B1506A Measurable Electrical Parameters |
Figure 6. IV Curve Measured via B1506A, such as the ICE-VCE of the Breakdown Voltage and BJT |
|
Figure 7. Capacitance to Voltage Graph |
Figure 8. Vgs Voltage to Gate Charge Graph |
|
Table 2. The results from the batch measurement of samples can be used to effectively identify problematic components.
|
|
Fault Isolation |
No matter what kind of semiconductor devices fault isolation is to be performed, at least one of these three tools will be used—the Photon Emission Microscope (PEM), the Optical Beam Induced Resistance Charge (OBIRCH) analyzer and the Thermal Emission Microscope (Thermal EMMI). The most appropriate localization instrument is selected according to the electrical failure behavior and sample structure. Whether the hot spot should be observed from the front or the back of the wafer is chosen based on the difficulty of sample preparation.
Although the structures of power devices is simple, the difficulty of its sample preparation is greater than that of ICs. This is because there is a thick layer of aluminum on the surface of the wafers that blocks the observation of hot spots. We prefer the Thermal EMMI for the initial identification of hot spots. The first positioning is carried out using its heat conduction characteristics. If a finer range is required after the positioning is completed, select another localization tool.
The Thermal EMMI system currently being used by MA-tek can reach a maximum voltage of 3KV. This makes it ideal for fault isolation of high-power devices. It can detect a microampere leakage even under high voltage operation. It is an essential tool for power component failure analysis.
|
Figure 9. At 500V, though the leakage current is only 1uA, it is still possible to locate the hot spot near the terminal area. On the left is the overlay of the hot spot and the optical image. On the right is the original photo of the hot spot. |
Defect Observation
|
Due to power devices’ simple structures, such as the MOSFET and the IGBT, many cells are arranged in parallel in an array. A clear and solid hot spot represents the location of a defect, and the path of a leak can be determined via the electrical behavior. A possible physical failure phenomenon can be deduced by comparing the structures. Therefore, generally speaking, it is standard operating procedure to observe the cross section immediately after the fault isolation is completed.
For power devices, there are two main ways to prepare cross section samples and observe the appearance of defects. One is Focused Ion Beam (FIB) microscopy. The other is Transmission Electron Microscopy (TEM). The difference between the two lies mainly in resolution. The FIB can observe comparatively obvious abnormalities such as melting, abnormal processes and foreign material, etc., whereas the TEM can be used to observe lattice defects. Both have their uses. In third generation semiconductor materials, there may be lattice defects, like dislocation. If no obvious abnormalities are found using FIB, you can turn to TEM observation.
If the leakage is caused by an ion implantation, the above two microscopes will be of little help. Instead, use Scanning Capacitance Microscopy (SCM) to observe the p-type and n-type doping. In addition to causing cell leakages, abnormal concentrations can lead to electric field distributions that cause breakdowns and high current problems.
In view of the above, with the appropriate analysis tools and established analysis steps that incorporate the right electrical characteristics and physical properties, the true causes of failures can be effectively uncovered. As the application of power devices becomes more and more widespread, we believe that this analysis process will help power device manufacturers accelerate development and improve mass production yields. |
Figure 10. GaN MOSFET Crack and Dislocation Observed via FIB and TEM
Figure 11. SiC MOSFET Dislocation Observed via TEM |
Please feel free to reach out to us if you have any questions : marketing@matek.com
Reference:
[1]Yole
[2]Power semiconductor roadshow hosted by UBS, London, 10-11 Nov. 2018
