SiC Material Developments Power Progress In Barcelona
For the past few years, the SiC community has waited tensely for chips made from the material to move confidently on from their tentative early steps in the high-power commercial electronics market. But other than LED-maker Cree, which produces and consumes SiC substrates in its light emitters, developers of the semiconductor material are still waiting for the big pay-off.
As SiC technology creeps forward, development funding ebbs and flows. At this year s European Conference for Silicon Carbide and Related Materials (ECSCRM) in Barcelona, Spain, some SiC groups admitted that their contingents had shrunk from previous years. Murmurs of difficulties in obtaining money to support ongoing SiC studies emerged from university professors gathered in darker corners, and were echoed by researchers with links to the US military. However, notable advances ranging from fundamental material research to commercial applications were on hand to provide any apathetic paymasters with food for thought.
The collective relief from such concerns provided by the conference s first-ever dedicated session on graphene was palpable. The University of Manchester s Andre Geim and Kostya Novoselov gained the Europhysics prize for the isolation of graphene in 2004 just a week before ECSCRM. With the pair subsequently touted as an outside shot at the Nobel Prize for Physics, the excitement surrounding SiC s potential impact in this area could be the perfect wake-up call to funding bodies.
Graphene is a continuous sheet of hexagonal carbon rings that when stacked with other sheets is well known as graphite. It was first characterized by removing monolayers from graphite with adhesive tape, shortly followed by production from SiC – a potentially more high-throughput approach. The SiC-based method heats conventional substrates so that silicon evaporates and leaves graphene behind.
The graphene research presented at ECSCRM ranged from the fundamental – for example microscopy analysis of graphene formed on SiC – to more practical concerns like improved preparation methods. Pierre Mallet of the multidisciplinary Néel Institute in Grenoble, France, bridged both areas. To date, much material made from SiC has been few layer graphene (FLG), which is essentially a graphene monolayer above graphite layers. Mallet and his colleagues made graphene by repeated annealing of 6H-SiC at temperatures of up to 1100 °C under ultra-high vacuum. Using scanning-tunneling microscopy the team showed that it had made 10–40 nm islands of true monolayer graphene.
Linköping University drew notable crowds during the poster sessions with its researchers attention-grabbing graphene developments. They showed characterization details and images of 2 inch diameter graphene monolayers produced in a specially modified furnace. Annealing at 2000 °C yields a single domain of graphene above most of the underlying 6H-SiC wafer, but also surprisingly generates some regions with a different structure according to low-energy electron microscopy.
Graphene boasts many exceptional qualities, but the interest at ECSCRM largely revolved around its promise as a basis for nanoelectronics. At the beginning of the year a team including Geim found that the intrinsic electron mobility of graphene was around 200,000 cm2/Vs. This remarkable figure compares with 1500 cm2/Vs for silicon and 8500 cm2/Vs for GaAs. The problem for this fledgling field is how to use these properties in devices that are compatible with graphene s single-atomic-layer thickness. "If you look at where processor speeds are now, and look at Intel s roadmap, we ve got 15 years to figure it out," joked one attendee. However, Shriram Shivaraman of Cornell University produced some of the first basic Schottky devices by evaporating metal dots on graphene formed from n-doped epitaxial SiC substrates. The graphene devices operated at higher currents than Shivaraman s SiC controls for the same forward voltage.
Advances in producing the cubic 3C-polytype of SiC represent a more immediately significant, and under appreciated, development than graphene. The flat surface and high elastic modulus of 3C-SiC makes it a strong candidate for the development of harsh-environment micro-electro-mechanical systems. 3C-SiC s properties may also help in metal-oxide-semiconductor (MOS) device applications, because reduced defect density at the SiC MOS gate interface may help to improve carrier mobility.
Ever since the first international conference on SiC in 1959, bulk single-crystal 3C-SiC growth has been held back by a lack of good-quality seed crystals. Now the European MANSiC project, set up specifically to advance 3C-SiC, is changing this. The institutions involved in the project, which is funding 19 young researchers, were represented in all of the oral presentations given on 3C-SiC at ECSCRM. In terms of bulk growth, 1.7 mm/h millimeter-size ingot production by sublimation was the key breakthrough, presented by Didier Chaussende of the Grenoble Institute of Technology.
A host of posters demonstrated the wider interest in both 3C-epitaxy and bulk growth, with the volume of work from Japan rivaling that coming from the European collaboration. The status of 3C-SiC epitaxy on non-native substrates has recently advanced rapidly and MANSiC member NovaSiC is formally selling 3C epiwafers. The company s presentation detailed epilayer thicknesses up to 12 µm grown on 2 inch silicon substrates, which hold the potential for expanding to larger diameters in the future.
4H-SiC remains the workhorse for SiC electronics, which typically use substrates sliced at an 8° angle from the primary 4H axis. This provides a saw-tooth-like surface for epitaxy that helps epilayers retain the 4H polytype. Unfortunately, each tooth can be the source of defects known as basal-plane dislocations (BPDs), which have been shown to evolve into other faults under bias and limit the lifespan of SiC devices. Unsurprisingly, given its pre-eminence in the SiC arena, Cree s pioneering method for eliminating BPDs is strongly defended by patents.
However, at ICSCRM 2007 – the international version of ECSCRM – in Otsu, Japan, much interest surrounded a new anti-BPD approach developed by the US Naval Research Laboratory (NRL). In Barcelona, Brenda VanMil explained how the NRL approach interrupts SiC chemical vapor deposition by cutting silane gas and reducing propane gas flows. Initially at each interrupt the team was able to convert 34% of BPDs to threading etch dislocations that do not limit device performance. Optimizing the temperature of the interrupt formation step to 1580 °C brings the conversion rate to more than 50%. Performing two interrupts with at least 2.5 µm of epitaxial material in between increases the proportion of BPDs converted to 84%.
Another star turn from ICSCRM was the Japanese Central Research Institute of the Electrical Power Industry, which this year showed that more techniques for BPD reduction are possible. The team detailed how it was able to perform virtually BPD-free SiC growth by turning down the growth rate in its high-speed epitaxy reactor from 250 to 20 µm/h.
Although the potential device lifetime benefits of these improvements are hopefully in the pipeline, SiC component manufacturers can already claim important performance achievements. At ECSCRM, Muhammad Nawaz from Swedish SiC transistor producer TranSiC detailed how his company s BJTs could be operated from –86 to 500 °C. The breakdown voltage remained unchanged, although gain diminished with increasing temperature.
Driving SiC advances
Before the automotive industry displaces silicon with SiC in its electric vehicles, it wants further performance improvements. However, according to the Institute for Electrical Machines, Traction and Drives at the Technical University of Braunschweig, there are some circumstances where SiC already wins out. Immo Koch compared the power losses seen with a SiCED SiC JFET with those from an Infineon silicon IGBT in inverters for switching frequencies between 5 and 30 kHz. At high temperatures and current densities the extensive conducting losses from the JFET inverter limit its benefit to higher frequencies. However, at reduced current density SiCED s JFET is also superior for lower switching frequencies.
Infineon, which part-owns SiCED with joint venture partner Siemens, previewed some of the technology that will be used in its third-generation SiC Schottky diodes, from which it is currently releasing pilot products. The German giant claims to have boosted reliability with a more thermodynamically and mechanically stable soldering process, an otherwise little-discussed topic in SiC device production. Semisouth, based in Starkville, MS, is also looking to add to the range of commercial SiC devices available with an official move from epiwafer supplier to vertical JFET manufacturer in 2009.
However, it was arguably Japanese automotive manufacturers who made the largest impact in Barcelona. Progress in SiC devices, especially in conjunction with their semiconductor industry compatriot Rohm, appears to be at a critical point. Development projects involving Honda and Nissan featured in the chip maker s status update, and represent some of the most successful work to date on power modules for electric vehicles.
Like Infineon, Nissan has successfully resolved reliability issues surrounding the packaging of SiC. Using a tantalum/TaN barrier, Satoshi Tanimoto was able to suppress a destructive reaction between Ni2Si contacts and the aluminum interconnect that Nissan uses with SiC devices. Kenichi Nonaka described Honda s suppressed surface recombination structure BJTs (SSR-BJTs). These BJTs possess a lightly doped n-type layer and a highly resistive p-type region that reduces concentrations of electrons and holes at the device s SiC surface (figure 2). This improves current gain of the SSR-BJT even if there is a high density of recombination centers on the SiC surface. In this way, Honda delivered a BJT with a record gain of 114, 2.8 mΩ cm–2 on-resistance and 1150 V blocking voltage.
ECSCRM s welcome reception brought a reminder of how tough the economic environment beyond SiC research has become, clashing with the lavish surroundings of the Liceu opera house in which it was held. Carles Gómara, director of innovation in the Catalonian government s ministry of industry, lauded the environmental and industrial benefits that are promised by SiC. He emphasized the importance of technological development in driving economic growth at a time when recessions are taking root around the world. With its slow but steady progress towards end markets, the SiC community is already cautious about the future. However, as Gómara suggests, its impact on areas of rapidly growing interest like graphene and electric vehicles should provide a strong bolster to its confidence.
View pdf of article