Power semiconductors usher in the SiC era

Silicon carbide production is increasing rapidly, driven by automotive end-market demand and price parity with silicon.

Thousands of power semiconductor modules are already used in electric vehicles for on-board charging, traction inverter and DC-to-DC conversion. Today, most of them are manufactured using silicon-based IGBTs. The move to silicon carbide-based MOSFETs doubles power density and increases switching speeds in smaller, lighter packages.

Electric vehicles and charging stations increasingly require high voltages and the ability to operate in hot, harsh environments, but silicon carbide (SiC) has taken some time to become available due to the high cost of manufacturing and packaging this wide-bandgap material. solid foundation. However, this is changing. Victor Veliadis, executive director and chief technology officer of PowerAmerica, said SiC power modules are now priced on par with silicon-based modules, which in turn is promoting supply partnerships and the construction of new SiC factories.

There is still much work to be done. SiC wafer technology needs to be upgraded. Manufacturing these devices requires 20% new process tools and 80% improved tools. The goal is to accelerate the turnaround of integrated and discrete power devices, which is why automakers are turning to direct fab-to-module collaboration.

Figure 1: Acquisition and partnership agreements ensuring supply and rapid technological advancement.

New wafer process tools include high-temperature epitaxial growth (>1,500°C), thermal ion implantation, rapid thermal processing (RTP) and faster pulsed atomic layer deposition. Significant improvements are occurring in wafer grinding, CMP, polishing pads and slurries for hard and brittle SiC materials. New materials, including strippers and cleaning chemicals, address equipment and sustainability needs.

From the packaging side, high-power printed circuit boards with discrete components are being replaced by integrated packages such as integrated circuits and chip-scale packages (CSP) to enable smaller and more reliable high-voltage operation. This allows electric vehicles to be equipped with smaller, lighter battery packs, which helps increase driving range. While the focus today is on SiC power and the expansion of Si power modules into hybrid and electric vehicles, SiC modules will dominate electric vehicles in the future. Additionally, GaN will find niche markets in electric vehicles, grid power, and smart energy.

By 2030, 39 million pure electric vehicles will be produced globally, equivalent to a compound annual growth rate of 22% from 2022 to 2030. This in turn is driving the power semiconductor market, which is expected to utilize approximately 50% silicon devices, 35% silicon carbide devices, and 12% gallium nitride devices by 2030. In electric vehicles, a traction inverter converts DC power from the battery pack into AC power to power the electric motors that drive the front and rear axles. SiC can also accelerate on-board and off-board charging, bringing electricity from the grid into electric vehicles.

Most importantly, SiC modules form the cornerstone of the transition from 400V to 800V batteries. Consumers will adopt electric vehicles more quickly when vehicles charge faster, have sufficient range and batteries cost less than $10,000 per vehicle.

SiC modules are reaching a tipping point where they are comparable in price to silicon-based power solutions, while enabling more efficient and compact systems. This, coupled with the expansion of the 800V battery range from currently used 400V batteries (including 600V or 650V devices), is stimulating high-volume production of 1,200V SiC devices. However, supply chain changes such as the impact of wafer crystal defects on yield, losses in device packaging and module integration, and closer ties between automakers and power system manufacturers are still ongoing. From a practical perspective, it will take some time for new silicon carbide wafer and fab capacity to reach high volumes.

However, this has not dampened enthusiasm for the technology. Analysts continue to raise SiC market forecasts. Yole Group predicts that the power semiconductor market will reach US$6.3 billion by 2027, 70% of which will be used in automotive applications. Looking only at SiC wafer production (starting with SiC wafers), TECHCET predicts a compound annual growth rate of 14% from 2022 to 2027.

Figure 2: Automakers are moving toward more direct collaboration with module suppliers and ultimately chipmakers.

Leaders such as Wolfspeed, STMicroelectronics, ON Semiconductor, Rohm, Infineon, and Bosch are key players in chip manufacturing. The largest cost contributor to these devices, silicon carbide wafers, is beginning to migrate from 150mm to 200mm manufacturing, but the growth, slicing and preparation processes still rely on expensive, time-consuming manual operations.

All parties, especially IDMs and fabs, are working hard to reduce SiC lattice defect rates and develop SiC-specific tool platforms, such as high-temperature ion implantation, epitaxial deposition furnaces operating above 1,500°C, and improved CMP slurries , polishing pads and cleaning chemicals for materials that are nearly as hard as diamond.

As silicon, SiC and GaN based power circuits all compete in the 400V battery range, the technology transition is already complete. However, SiC power systems are capable of delivering much higher power levels than GaN (see Figure 3).

“I call it the 650V battleground because really all three technologies are competitive in that range,” said PowerAmerica’s Veliadis, who participated in the “Connecting the Automotive Ecosystem with SiC Manufacturing” forum at SEMICON West. GaN has higher electron mobility than SiC, but it is less mature and cannot match SiC's high power levels. Even so, GaN still has huge appeal for making high-frequency devices. Additionally, some GaN-on-silicon approaches currently being used by Intel, imec, and others look very promising.

Figure 3: The power density operating windows of SiC, Si and GaN (left) overlap at the low end for 650V devices (400V batteries), but 1,200V devices for 800V batteries are coming soon

Silicon carbide modules are considered critical to improving the efficiency of electric drivetrains in electric vehicles. The dramatic shift from silicon-based devices to silicon carbide devices will significantly help increase the power density of power systems while reducing the size, weight and, most importantly, cost of electric vehicles. This happens because silicon-based power semiconductors, while still being optimized, have reached their operating limits in terms of conduction and switching losses. Silicon carbide's wider bandgap (3.26eV versus 1.12eV for silicon) reduces such losses and provides superior high-temperature and high-frequency performance.

To date, many SiC chip manufacturers have converted 150mm silicon production lines to SiC manufacturing. “The model that has been very successful so far has been to process silicon carbide in mature, fully depreciated silicon plants, with a capital investment of about $30 million, and the return is certainly huge,” Veliadis said. For power modules, cost is most important. “With silicon carbide you would pay approximately three times as much for a semiconductor chip, but the final system cost would be less than a silicon power module, which is counterintuitive. But the answer is simple. The ability to operate efficiently at high frequencies is significantly reduced The size of the magnetics and passive components offsets the higher cost of chip manufacturing.”

However, the industry has no old plants left that could spend $30 million to renovate. New silicon carbide fabs are being built rapidly. At the same time, fabless companies are competing for capacity.

"We have two competing markets - the automotive market and the renewable energy market - both looking for capacity," said Ralf Bornefeld, senior vice president of power semiconductors and modules at Robert Bosch. "We know from the COVID-19 pandemic that one competitive market can close another market, so we need to take that into account." Bosch is currently producing third-generation SiC MOSFET modules with a breakdown voltage of 1,200V.

SiC devices are particularly suitable for automotive because they can provide high power density operating at higher temperatures in harsh environments. SiC power devices can achieve extremely low switching losses and ultra-low RDSon (resistance between source and drain during operation). Smaller RDSon is associated with lower power losses in the MOSFET.

Device capabilities start with SiC materials. "Crystal quality is the number one factor that key players have been addressing for the past 20 years, but there are still basal plane dislocations, stacking faults, etc. in the crystals that need to be engineered to make 20, 30 and 40 square millimeter devices," said Christophe Maleville, Senior Vice President of SOITEC Innovation. “Four years ago, when we entered the silicon carbide space, the first thing we noticed was that the feasibility of each ingot and each wafer was different, and engineers often needed to adjust and verify epitaxy. So, its role in manufacturing The implementation is not yet lean.”

Electrically, power devices can be sensitive to parasitic inductance, sparks, and other challenges. Unlike analog mixed-signal plants, parameters are the primary focus and power engineers have to deal with changes.

“In the past, [simulation] lacked shrinkage. But they had a mature process in terms of defects,” said Dieter Rathei, CEO of DR Yield. “As compound semiconductors such as silicon carbide, gallium nitride and gallium arsenide become more mainstream and grow faster, parametric yield issues will improve.”

Currently, most 100mm and 150mm sized wafers use single crystal silicon carbide with a hexagonal lattice structure (4H and 6H represent 4-inch and 6-inch hexagonal wafers). But the largest SiC device producers are well underway with the transition from 150mm to 200mm, and other producers are also taking advantage of the supply.

For example, Infineon obtains its wafers from multiple suppliers, according to Yole Group analysts. This includes STMicroelectronics’ acquisition of a majority stake in Sweden’s Nortel. Silicon power device supplier Renesas Electronics is strengthening production capacity and partnerships. In July, Renesas signed a 10-year agreement and paid a $2 billion deposit to Wolfspeed to supply 150mm bare and epitaxial SiC wafers. Renesas Electronics also reached an agreement with Mitsubishi, which will spend 260 billion yen on technology and expansion, including building a new SiC factory in Japan.

"[Renesas] was a latecomer in the traditional power semiconductor field, but now [our products] are valued for their high efficiency," company president Hidetoshi Shibata said in a recent press release. “SiC can do this, too.”

Meanwhile, SOITEC and STMicroelectronics are exploring polycrystalline SiC on monocrystalline methods, which divide a monocrystalline silicon wafer into multiple slices and reuse the donor wafer substrate to reduce waste. The advantage of a polysilicon substrate is its ability to conduct heat through the substrate to the metal connectors, allowing for faster switching and excellent heat dissipation.

In some ways, silicon carbide is following silicon's trajectory. But due to SiC's defect levels, some data sharing is required.

“We exchange equipment data with raw material data from silicon wafer suppliers,” Bosch’s Bornefeld said. "We also use advanced artificial intelligence-based systems to identify good correlations and share this so both companies can move forward."

Still, data sharing is not widespread. Unlike silicon, boule sizes from 150mm to 200mm do not bring high returns in the form of more wafers/boule. Additionally, 200mm requires larger seeds, which take longer to grow at 2,500°C. Today, productivity (wafer/bogot) improvements are probably in the 20% range. TECHCET analysts estimate that the cost contribution of ingot growth will decline relative to slicing, grinding, polishing and CMP (see Figure 4).

Figure 4: Due to the high material cost per millimeter of pillar height, maximizing the number of SiC wafers per pillar is critical.

“We are seeing demand for semiconductors in automotive applications growing faster than EV production,” said Lee Bell, marketing director for automotive smart power and discrete products at STMicroelectronics. “This is due to a number of factors. Advanced drivers Safety features, autonomous vehicle controls, advanced connectivity and convenience features are all driving semiconductor demand, but not in the same way as powertrain electrification,” he said. “By 2022, about two-thirds of electric vehicles will be hybrids, with about one-third of these being battery-powered. By 2030, this trend will reverse. This is due to increased market acceptance, charging infrastructure Facility availability increases, but perhaps most importantly, this is where automakers place their R&D and manufacturing budgets.” This change is a key driver for the use of SiC MOSFETs.

Bell noted that traction inverters tend to use larger chips. Charging systems in vehicles and DC-DC converters that reduce voltage from batteries to IoT systems are huge consumers of power semiconductors, he added. Neither exists in hybrid vehicle architecture.

He also emphasized that the primary focus is on efficiency (packaging devices and modules) because less power is lost in the system, the longer the range of cars and trucks. “We conducted a study comparing a 210kW inverter system (equivalent to approximately 280 horsepower) with SiC MOSFETs and silicon IGBTs (insulated gate bipolar transistors),” he said. "The silicon carbide method consistently delivers 98% operating efficiency, while the IGBT method is less efficient, especially in the low operating load range, where the vehicle's service life is about 95%."

Total power is on-state losses plus switching losses. “Silicon carbide reduces switching losses by a factor of four,” he said. ST is producing fourth-generation SiC products with a 30% improvement in RDSon performance.

Bosch's Bornefeld presented demand and capacity estimates through 2030, indicating considerable global wafer and fab capacity in Japan, South Korea, China, Malaysia, Germany, Austria and the United States. Indeed, the industry needs to be careful not to overbuild (see Figure 5). “The question is, ‘What is happening in China?’ China is already a leader in silicon carbide raw materials, and they are providing very high-quality, affordable wafers,” Bonefield said. "They're catching up quickly on the equipment side as well. So we really need to watch and track overall capacity."

Figure 5: Global distribution map of facilities that process SiC raw materials into wafers.

Finally, PowerAmerica's Veliadis discussed the workplace training required to become proficient in implementing wide-bandgap semiconductors such as SiC and GaN in fabs. “Engineers with extensive experience in SiC and GaN MOSFET manufacturing are in short supply, and there are significant differences between SiC fabs and silicon fabs.”

The transition to clean energy and electric vehicles will require alternative semiconductor materials such as SiC and GaN, and power devices are certain to be significantly optimized in the coming decades. The frenzy of technology improvements and capacity expansion may not last, but power devices will remain key to many companies' roadmaps.

"We know the semiconductor industry is heading toward a $1 trillion market, but everyone wants to know what happens beyond 2030," said David Britz, director of strategic marketing at ICAPS at Applied Materials. "I'm here to prove that the fifth era of semiconductors is indeed driven by transformations in energy production and transportation."

Managing growth in silicon carbide wafers, devices and modules is probably the most difficult aspect of the silicon carbide market to date, along with supply chain issues, filling technology gaps and geopolitical changes. Still, the semiconductor technology community seems to agree on many things, especially the need for next-generation power efficiency and performance.

*Disclaimer: This article is original by the author. The content of the article is the personal opinion of the author. The reprinting by Semiconductor Industry Watch is only to convey a different point of view. It does not mean that Semiconductor Industry Watch agrees or supports the view. If you have any objections, please contact Semiconductor Industry Watch.

Today is the 3504th issue of "Semiconductor Industry Observation" shared with you. Welcome to pay attention.