Technical Support
Feedback
[Introduction] Silicon carbide (SiC) is produced by high-temperature smelting in a resistance furnace using quartz sand, petroleum coke (or coal coke), and wood chips as raw materials. Silicon carbide also occurs in nature as a rare mineral, moissanite. Silicon carbide is also known as carborundum. Among modern non-oxide high-tech refractory raw materials such as C, N, and B, silicon carbide is the most widely used and most economical. It can be called emery or refractory sand.
I. The Past and Present of Silicon Carbide
Because of its stable chemical properties, high thermal conductivity, low thermal expansion coefficient, and good wear resistance, silicon carbide has many other uses besides abrasives. For example, applying silicon carbide powder to the inner wall of a water turbine impeller or cylinder block through a special process can improve wear resistance and extend service life by 1 to 2 times; advanced refractory materials made from it offer thermal shock resistance, small volume, light weight, high strength, and good energy-saving effects. Low-grade silicon carbide (containing about 85% SiC) is an excellent deoxidizer; it can accelerate steelmaking, facilitate control of chemical composition, and improve steel quality. In addition, silicon carbide is widely used to make electric heating elements—silicon carbide rods.
Silicon carbide is very hard, with a Mohs hardness of 9.5, second only to diamond (10), the hardest material in the world. It has excellent thermal conductivity, is a semiconductor, and can resist oxidation at high temperatures.
Silicon Carbide Timeline
1905: Silicon carbide was discovered in a meteorite for the first time.
1907: The first silicon carbide crystal light-emitting diode was created.
1955: A major theoretical and technical breakthrough occurred when LELY proposed the concept of growing high-quality silicon carbide, establishing SiC as an important electronic material.
1958: The first World Silicon Carbide Conference was held in Boston for academic exchange.
1978: In the 1960s and 1970s, silicon carbide research was mainly conducted in the former Soviet Union. By 1978, the "improved LELY technique" was first used for grain purification and growth.
1987–present: Silicon carbide production lines were established based on CREE's research results, and suppliers began providing commercial silicon carbide substrates.
2001: Germany's Infineon launched SiC diode products, and companies such as Cree and STMicroelectronics in the United States followed with SiC diodes. In Japan, Rohm, New Japan Radio, and Renesas Electronics began producing SiC diodes.
On September 29, 2013, the International Conference on Silicon Carbide and Related Materials (ICSCRM 2013) was held, with 136 organizations from 24 countries, including semiconductor companies and research institutions, and 794 attendees—the highest in recent years. Internationally renowned semiconductor device manufacturers such as Cree, Mitsubishi, Rohm, Infineon, and Fairchild presented their latest mass-produced silicon carbide devices at the conference.
Today, many manufacturers produce silicon carbide devices, such as Cree, Microsemi, Infineon, and Rohm.
II. Advantages and Characteristics of Silicon Carbide Devices
Silicon carbide (SiC) is currently the most mature wide-bandgap semiconductor material. Countries around the world attach great importance to SiC research and are investing substantial human and material resources in its development. The United States, Europe, and Japan have not only formulated corresponding research plans at the national level, but several international electronics giants have also invested heavily in developing silicon carbide semiconductor devices.
Compared with ordinary silicon, components made from silicon carbide have the following characteristics:
High-Voltage Characteristics
Silicon carbide devices withstand 10 times the voltage of equivalent silicon devices.
Silicon carbide Schottky diodes can withstand up to 2400 V.
Silicon carbide field-effect transistors can withstand tens of thousands of volts, and their on-state resistance is not very high.
High-Frequency Characteristics
High-Temperature Characteristics
Today, as Si materials approach their theoretical performance limits, SiC power devices have long been regarded as "ideal devices" and highly anticipated because of their high voltage withstand capability, low loss, and high efficiency. However, compared with conventional Si devices, the balance between performance and cost of SiC power devices, as well as their demanding process requirements, will be key to whether SiC power devices can truly achieve widespread adoption.
At present, low-power silicon carbide devices have moved from the laboratory to the production stage of practical devices. Currently, silicon carbide wafers are still relatively expensive and have many defects. Through continuous research and development, it was expected that around 2010 silicon carbide devices would dominate the power device market. In reality, however, this has not been the case.
III. The Most Closely Watched Silicon Carbide MOS
SiC Device Classification
SiC-MOSFET
SiC-MOSFET is the most closely watched device in silicon carbide power electronics research. The most notable results have come from Cree in the United States and ROHM in Japan.
Today, as Si materials approach their theoretical performance limits, SiC power devices have long been regarded as "ideal devices" and highly anticipated because of their high voltage withstand capability, low loss, and high efficiency. However, compared with conventional Si devices, the balance between performance and cost of SiC power devices, as well as their demanding process requirements, will be key to whether SiC power devices can truly achieve widespread adoption.
Structure of Silicon Carbide MOS
In silicon carbide MOSFETs (SiC MOSFETs), the N+ source region and P-well doping are both performed by ion implantation, followed by annealing activation at 1700°C. Another critical process is the formation of the silicon carbide MOS gate oxide. Because the silicon carbide material contains both Si and C atoms, a very special gate dielectric growth method is required. The advantages of the trench structure are as follows:
Planar vs. Trench
SiC-MOSFETs with a trench structure can maximize the characteristics of SiC.
Advantages of Silicon Carbide MOS
Silicon IGBTs generally can only operate at frequencies below 20 kHz. Due to material limitations, high-voltage, high-frequency silicon devices cannot be realized. Silicon carbide MOSFETs are not only suitable for a wide voltage range from 600 V to 10 kV, but also offer excellent switching performance as unipolar devices. Compared with silicon IGBTs, silicon carbide MOSFETs have no current tail in switching circuits, resulting in lower switching losses and higher operating frequencies.
The losses of a 20 kHz silicon carbide MOSFET module can be half those of a 3 kHz silicon IGBT module, and a 50 A silicon carbide module can replace a 150 A silicon module. This demonstrates the enormous advantages of silicon carbide MOSFETs in operating frequency and efficiency.
The parasitic body diode of a silicon carbide MOSFET has an extremely small reverse recovery time trr and reverse recovery charge Qrr. As shown in the figure, for devices with the same rated current of 900 V, the parasitic diode reverse charge of a silicon carbide MOSFET is only 5% that of a silicon-based MOSFET with the same voltage rating. For bridge circuits (especially when the LLC converter operates above the resonant frequency), this metric is critical because it can reduce dead time and the losses and noise caused by body diode reverse recovery, making it easier to increase the switching frequency.
Applications of Silicon Carbide MOS Transistors
Silicon carbide MOSFET modules offer enormous advantages in medium- and high-power power systems such as photovoltaics, wind power, electric vehicles, and rail transit. The high-voltage, high-frequency, and high-efficiency advantages of silicon carbide devices can overcome the limitations imposed by device performance in current electric vehicle motor designs, which is a key focus of electric vehicle motor R&D in China and abroad. For example, the power control unit (PCU) in hybrid electric vehicles (HEVs) and pure electric vehicles (EVs) jointly developed by Denso and Toyota uses silicon carbide MOSFET modules and achieves a volume reduction to one-fifth. The EV motor drive system developed by Mitsubishi uses SiC MOSFET modules, with the power drive module integrated into the motor to achieve integration and miniaturization. It is expected that from 2018 to 2020, silicon carbide MOSFET modules will be widely used in electric vehicles in China and abroad.
IV. Silicon Carbide Schottky Diodes
Silicon Carbide Schottky Diodes
Structure of Silicon Carbide Schottky Diodes
Silicon carbide Schottky diodes (SiC SBDs) use a junction barrier Schottky (JBS) diode structure, which can effectively reduce reverse leakage current and provide better high-voltage withstand capability.
Advantages of Silicon Carbide Schottky Diodes
Silicon carbide Schottky diodes are unipolar devices, so compared with conventional silicon fast-recovery diodes (Si FRDs), they have ideal reverse recovery characteristics. When the device switches from forward conduction to reverse blocking, there is almost no reverse recovery current (as shown in Figure 1.2a), and the reverse recovery time is less than 20 ns; even for a 600 V/10 A silicon carbide Schottky diode, the reverse recovery time is within 10 ns. Therefore, silicon carbide Schottky diodes can operate at higher frequencies and achieve higher efficiency at the same frequency. Another important feature is that silicon carbide Schottky diodes have a positive temperature coefficient: resistance gradually increases with temperature, which is the opposite of silicon FRDs. This makes silicon carbide Schottky diodes very suitable for parallel use, increasing system safety and reliability.
In summary, the main advantages of silicon carbide Schottky diodes are as follows:
1. Almost no switching loss
2. Higher switching frequency
3. Higher efficiency
4. Higher operating temperature
5. Positive temperature coefficient, suitable for parallel operation
6. Switching characteristics almost independent of temperature
Applications of Silicon Carbide Schottky Diodes
Silicon carbide Schottky diodes can be widely used in medium- and high-power fields such as switching power supplies, power factor correction (PFC) circuits, uninterruptible power supplies (UPS), and photovoltaic inverters, significantly reducing circuit losses and increasing circuit operating frequency. Replacing conventional silicon FRDs with silicon carbide SBDs in PFC circuits allows the circuit to operate above 300 kHz with efficiency essentially unchanged, whereas circuits using silicon FRDs show a sharp drop in efficiency above 100 kHz. As the operating frequency increases, the size of passive components such as inductors decreases accordingly, reducing the overall circuit board volume by more than 30%.
V. How Do People Evaluate Silicon Carbide?
Almost every article you can read introduces silicon carbide as follows:
Silicon carbide has a bandgap 2.8 times that of silicon (wide bandgap), reaching 3.09 electron volts. Its dielectric breakdown field strength is 5.3 times that of silicon, as high as 3.2 MV/cm. Its thermal conductivity is 3.3 times that of silicon, at 49 W/cm·K. Schottky diodes and MOS field-effect transistors made from silicon carbide have a drift resistance region an order of magnitude thinner than silicon devices of the same voltage rating. Their impurity concentration can be two orders of magnitude higher than that of silicon. As a result, the impedance per unit area of silicon carbide devices is only one-hundredth that of silicon devices. Its drift resistance is almost equal to the total resistance of the device. Therefore, silicon carbide devices generate very little heat. This helps reduce conduction and switching losses, and the operating frequency is generally more than 10 times higher than that of silicon devices. In addition, silicon carbide semiconductors have an inherent strong radiation resistance.
In recent years, power devices such as IGBTs (insulated gate bipolar transistors) made from silicon carbide materials have been able to use processes such as minority carrier injection, reducing their on-state impedance to one-tenth that of conventional silicon devices. Combined with the low heat generation of silicon carbide devices themselves, their thermal performance is excellent. Moreover, silicon carbide power devices can operate normally at high temperatures of 400°C. They can control very large currents using very small devices. Their operating voltage is also much higher.
VI. What Is the Current Status of Silicon Carbide Device Development?
1. Technical parameters: For example, Schottky diode voltage has increased from 250 V to over 1000 V, and chip area has decreased, but current is only tens of amperes. The operating temperature has increased to 180°C, far from the 600°C described in introductions. The voltage drop is even less satisfactory, with no difference from silicon materials; a high forward voltage drop reaches 2 V.
2. Market price: about 5 to 6 times that of silicon material devices.
VII. Where Are the Difficulties in the Development of Silicon Carbide (SiC) Devices?
Based on various reports, the difficulty lies not in the chip's principle design—especially since chip structure design is not difficult to solve well. The difficulty lies in the fabrication processes for realizing the chip structure.
Examples are as follows:
1. Micropipe defect density in silicon carbide wafers. Micropipes are macroscopic defects visible to the naked eye. Until silicon carbide crystal growth technology develops to the point of completely eliminating micropipe defects, high-power power electronic devices will be difficult to manufacture from silicon carbide. Although the micropipe density of high-quality wafers has reached no more than 15 cm⁻², device manufacturing requires silicon carbide crystals with diameters exceeding 100 mm and micropipe densities below 0.5 cm⁻².
2. Low epitaxial process efficiency. Vapor-phase homoepitaxy of silicon carbide generally must be carried out at high temperatures above 1500°C. Due to sublimation issues, the temperature cannot be too high—generally not exceeding 1800°C—so the growth rate is relatively low. Liquid-phase epitaxy has lower temperature and higher rate, but lower throughput.
3. Special requirements for doping processes. For example, if diffusion is used for doping, the diffusion temperature for silicon carbide is much higher than for silicon. At such temperatures, the SiO2 layer used as a mask loses its masking effect, and silicon carbide itself is unstable at such high temperatures. Therefore, diffusion doping is not suitable; ion implantation doping must be used. If aluminum is used as the p-type implanted impurity, because aluminum atoms are much larger than carbon atoms, implantation causes relatively severe lattice damage and leaves impurities in an inactive state. Implantation often must be performed at a relatively high substrate temperature and annealed at an even higher temperature. This brings about problems of silicon carbide decomposition and silicon atom sublimation on the wafer surface. At present, there are still many issues with p-type ion implantation, and a series of process parameters from impurity selection to annealing temperature still need optimization.
4. Fabrication of ohmic contacts. Ohmic contact is a very important process for leading out device electrodes. Manufacturing metal electrodes on silicon carbide wafers requires contact resistance below 10⁻⁵ Ω·cm². Ni and Al electrode materials can achieve this, but their thermal stability is poor above 100°C. Al/Ni/W/Au composite electrodes can improve thermal stability to 600°C for 100 h, but their specific contact resistance is as high as 10⁻³ Ω·cm². Therefore, forming good ohmic contacts on silicon carbide is relatively difficult.
5. Temperature resistance of supporting materials. Silicon carbide chips can operate at 600°C, but the materials used with them may not withstand such high temperatures. For example, electrode materials, solders, packages, and insulating materials all limit increases in operating temperature.
The above are only a few examples, not an exhaustive list. There are still many process issues without ideal solutions—such as silicon carbide semiconductor surface trenching processes, termination passivation processes, and the impact of gate oxide interface states on the long-term stability of silicon carbide MOSFET devices, on which the industry has not yet reached consensus. These greatly hinder the rapid development of silicon carbide power devices.
VIII. Why Are SiC Devices Still Not Widespread?
As early as the 1960s, the advantages of silicon carbide devices were already well known. The reason they have not yet been popularized is that many technical problems, including manufacturing issues, remain. Until now, the main industrial application of SiC materials has been as an abrasive (emery).
SiC does not melt within a controllable pressure range; instead, it transforms directly into a gas at a sublimation point of about 2500°C. Therefore, SiC single crystal growth can only start from the gas phase, a process much more complex than Si growth; Si melts at about 1400°C. The main obstacle preventing SiC technology from achieving commercial success has been the lack of a suitable substrate material for industrial production of power semiconductor devices. In the case of Si, the single-crystal substrate is often referred to as a silicon wafer, which is the prerequisite and guarantee for production. A method for growing large-area SiC substrates was successfully developed in the late 1970s. However, substrates grown by the improved method known as the Lely method suffered from micropipe defects.
As long as a single micropipe passes through a high-voltage PN junction, it destroys the junction's voltage-blocking capability. Over the past three years, this defect density has dropped from tens of thousands per square millimeter to tens per square millimeter. In addition to this improvement, when the maximum device size is limited to a few square millimeters, production yield may be greater than a few percent, giving each device a maximum rated current of a few amperes. Therefore, before SiC power devices can achieve commercial success, greater technical improvements in SiC substrate materials are needed.
Progress in micropipe density of SiC wafers in industrial production and of the best wafers
Micropipe density values for manufacturing devices with yields of 40% and 90%
As the figure shows, current SiC materials already meet the requirements for optoelectronic devices, which are no longer affected by material quality; industrial production yield, reliability, and other performance of devices also meet requirements. High-frequency devices mainly include unipolar devices such as MOSFETs and Schottky diodes. The micropipe defect density of SiC materials has basically met requirements, with only some impact on yield. For high-voltage, high-power devices, SiC materials will require about another two years to further improve material defect density. In short, regardless of current difficulties and how semiconductors develop, SiC is undoubtedly a promising material for the new century.
Source: Wide Bandgap Semiconductor Technology Innovation Alliance Compiled by: Changsha Ligao Jiechuang Instrument Co., Ltd.
