Microsemi puts SiC on the radar

The versatility of radar is on the up. The systems that saw widespread deployment in the Second World War were only capable of searching the sky for objects. By the 1980s they could identify ballistic missiles, and since the atrocities on September 11, 2001, there has been a desire to take radar capability to an even higher level. The current goal is for relatively affordable radar for defense and commercial markets that operates over distances of 500 miles or more, while delivering detailed information regarding the identity of aircraft and missiles.

If radar is to reach this level, it will require the introduction of a new technology for the generation of electromagnetic pulses with frequencies of 0.1–4 GHz. If the purpose of radar is just to search the sky, then short pulses of less than 1 µs at a low duty cycle of 1% are adequate, which can be produced by vacuum tubes. But these tubes cannot be used for identification – this requires far longer pulses, in combination with higher duty cycles. Silicon is up to this task, and can produce pulses that are 1000 times longer than those generated by tubes, in combination with duty cycles of at least 10% and some form of pulse compression. The strengths of SiC The drawback of using silicon is lower peak powers (typically 300 W at 300 µs) that reduce the radar s operating range. Combining elements can address this – the US Pave PAWS (phased array radar warning system), built at the height of the cold war to detect and track sea-launched intercontinental ballistic missiles, has a range of more than 3000 miles. However, this system has more than 2000 elements, a component count that makes long-range silicon-based radar too expensive for commercial markets.

To meet today s goal – affordable, long-range radar that delivers an unprecedented level of detail – requires a new solid-state technology that offers affordable systems with operating pulses of 1 ms or more, higher duty cycles and output powers that are far higher than those of silicon devices. At Microsemi, which is headquartered in Irvine, CA, we believe that we have the answer in SiC. This wide-bandgap semiconductor can be used to create transistors with very high levels of performance, including peak output powers of 1.4 kW from a single device. Pulse lengths of 1 ms or more should soon be realized from transistors delivering a peak output of 750 W.

SiC devices can deliver longer, more-powerful pulses at a higher duty factor than silicon equivalents because they can operate at higher voltages. Transistors built from this wide-bandgap material can also operate at higher temperatures, which allows higher power densities and smaller chip sizes, thanks to the greater thermal conductivity of SiC.

Other compound semiconductors also have superior intrinsic properties to silicon. GaAs, for example, has enjoyed great success in the handset market. However, devices made from GaAs operate at lower voltages than those built from SiC and this limits the peak output power. GaN, meanwhile, shares many of the attributes of SiC, but again it produces transistors with lower operating voltages.

Our devices are designed to cover radar bands that are allocated by the International Telecommunications Union (table 1). The lower frequency bands, such as 138–144 MHz and 420–450 MHz, are used for long-range search and air route surveillance. Several systems operating at these frequencies also incorporate the capability to penetrate ground foliage. Higher frequency radars are used primarily for short-range air traffic control.

We cover the VHF band, through to the L-band, with a SiC static induction transistor (SIT) (figure 1). This device employs a vertical geometry, which leads to superior breakdown voltages and power densities compared with lateral structures, such as LDMOS, MOSFET and MESFET transistors. The SIT is operated in triode mode, which enables high voltage gain at operating voltages of up to 125 V. This leads to a high power gain without large excursions in current.

S-band and higher frequencies are covered with a MESFET (figure 2). This type of transistor has a lateral structure and can operate at high frequencies and voltages thanks to the short channel length and low chip parasitics.

Epiwafer growth is outsourced, which leaves us to focus on chip architecture, wafer processing and the design and manufacture of the finished product. With this approach we can exploit the expertise of epitaxial growth specialists, who are constantly improving the material quality of this wide-bandgap semiconductor.

Our devices are grown on 3 inch SiC, but we will make the transition to 4 inch material when this size becomes more readily available. We are investigating the quality of SiC substrates produced by a variety of manufacturers and have found that although the variations are significant, they are narrowing all the time. We are also evaluating the quality of epiwafers produced by several companies and have found small to moderate differences across suppliers. By working with various epiwafer and substrate suppliers we will qualify multiple sources at both levels, and ultimately ensure quality and consistency in production volumes at reasonable prices.

Chip fabrication is carried out in-house, including the key process steps of pattern generation (with ASML 5X steppers), materials etching, high temperature implanting and annealing, passivation and metallization. Many of these process steps employ tools that are also used for silicon processing. However, specialized equipment that operates at high temperatures is needed for the implanting step and subsequent annealing. The equipment that is shared between both materials has been adapted to reflect the unique characteristics of SiC, while ensuring a high level of consistency of the finished product.

We have produced SIT chips for VHF and UHF radar applications ranging from 1 × 1 mm to 1 × 2 mm in size. This includes the first solid-state devices to operate at peak powers above 1 kW.

The full 1000 W transistor has seven chips wired into the package. Each chip has five active areas, known as cells, and the final transistor uses 32 cells, which are each capable of producing a peak output of 45 W and 36 W at VHF (156 MHz) and UHF (450 MHz) frequencies, respectively. Dividing the chip into a handful of active areas aids transistor performance because it enables low inductance connections and good thermal distribution, leading to improved operation at higher frequencies. As the frequency is increased, the active cell size gets smaller and more cells are employed on a chip.

Our transistors are routinely tested at a 10% duty cycle using medium and long pulses with widths of 300 µs and 1 ms (figure 3). However, we have also studied our SITs under extended pulse width and duty factor, and the results show the capability of SiC for higher voltage operation. One of our SITs, which is run at 100 V and adjusted to operate at 200 MHz, produces 800 W from 1.8 ms pulses and an 18% duty cycle. This output is more than three times higher than that of the best silicon products. Pump up the power The performance of VHF and UHF products produced from SIT epiwafers via several growth runs has recently been evaluated. These transistors deliver high peak output powers, power gain and drain efficiency, in combination with a low reduction in output power over the duration of the pulse, when operated over the 100–125 V range. These results have encouraged us to generate high-performance products for the frequency range 1.2–1.4 GHz, and for applications relating to avionics equipment, which operate in the 960–1215 MHz band. Other plans include the development of a higher peak power chip that produces more than 2 kW, which will require a larger package.

In addition, we are developing high-performance subassemblies that will enable our customers to quickly implement their latest system design. During the last 12 months we have introduced a series of power solution modules, which have terminal impedances of 50 Ω and use a pair of devices mounted on a carrier. This approach promises to enable the operation of SIT products at higher power levels within a short time frame.

Another target is a 2 kW output from a single SIT. This will require a complete review and analysis of the entire chip and package structure, including efforts to improve thermal management and device and circuit layout for a targeted operating frequency band. Each element will be modeled, fabricated and tested to identify key performance characteristics and trade-offs. Customer demands mean that we will require chips that are about 50% larger than our existing designs, in combination with larger packages. This device should be completed by early 2010.

We are also gearing up for the manufacture of SiC MESFETs that are able to serve high-power S-band applications operating between 2.7 and 3.5 GHz. The gain produced by these transistors is at least 2 dB higher than silicon equivalents, and they can also handle longer pulses and higher duty factors. These products are planned for release late this year and will complement our silicon BJT products serving the same frequency band. By releasing SiC products that span the VHF to S-band, we will enable our customers to benefit from the deployment of this new form of solid-state technology.   

View pdf of article