Surveillance is increasingly performed with unmanned aerial vehicles fitted with infrared imaging systems. These long-wavelength, high-performance detectors require substantial cooling and draw a lot of power from on-board batteries. But these demands could be reduced with Sofradir’s new generations of mercury cadmium telluride detectors that promise to operate at higher temperatures, thanks to improvements in passivation layers and device architectures. Richard Stevenson reports.

It is incredibly beneficial to slash the cooling demands of any device. One of the biggest gains that follows is that a smaller, cheaper and lower weight unit is then good enough to provide device cooling. On top of this, power consumption is reduced, because the new cooler either draws less power or consumes less cooling fluids.

One family of devices that it is essential to cool are solid-state infrared detectors – fail to do this and the dark current is far too high for the acquisition of goodquality images. Cooling to cryogenic temperatures is mandatory for the three most common types of highperformance, infrared detector, which are based on InSb, mercury cadmium telluride (MCT) and quantum well heterostructures formed from the GaAs family of materials.

Traditional InSb detectors are relatively straightforward to build, but they suffer from two major weaknesses: The focal plane of the detector has to be cooled to 77 K, and its response is limited to a narrow spectral range of  just 3-5 μm. It is possible to increase this detector’s operating temperature to 150 K by switching from bulk InSb to more complex structures that include InAlSb and are grown by MBE. But this technology is in its infancy, and it is currently very challenging to manufacture detectors with this approach.

QWIPs is based on a far more mature materials technology, GaAs-based heterostructures. However, the sensitivity of a QWIPs detector is not as high as that for MCT, which can operate over a far wider spectral range. These II-VI devices can be tuned to cover the visible, or to detect radiation classified as short wave (1-2.5 μm), medium wave (3-5 μm), long wave (7-10 μm) or very long wave (around 14 μm).

One of the biggest downsides of the MCT detector is its cooling requirement. Although this is not as severe as that for conventional InSb detectors, commercial MCT detectors do require cooling to 80 -130 K, with the exact figure depending on the operating wavelength – it is higher for shorter wavelengths. It is possible to reach these temperatures with Stirling coolers, which operate just like refrigerators and use helium gas. And if cooling is required for a single, very short time, a Joule- Thompson cooler can be used, which involves the release of a high-pressure gas from a vessel.

Military demands



The biggest market for the MCT detector is the military, where it is often employed for surveillance. When these detectors are fitted on tanks, carriers, destroyers, submarines, fighter planes and helicopters, trimming the size, weight and power consumption of the detector’s cooling system is not necessarily a big deal. But it can make a major difference to the range of unmanned aerial vehicles (UAVs), pilotless planes that can have a wingspan of just a few metres.

Today, the power drawn by the cooler from a set of batteries typically exceeds that required to run the detector, but it should be possible to start re-dressing this balance by making detectors that can operate at significantly higher temperatures. One company that is trying to do just this is the French firm Sofradir, which is headquartered close to Paris and has its development and production facilities in Veurey-Voroize, a town within France’s ‘infrared valley’. Spun off from CEA-Leti in 1986, Sofradir has a clear long-term plan for increasing the operating temperature of its wide range of commercial MCT detectors.



The company’s standard technology combines an optimised passivation process with high-quality substrates and epilayers to yield detectors with operating temperatures up to 120 K.

In future, the company plans to launch detectors with an inverted doping structure that can operate at up to 150 K, followed by more complex epitaxial designs that will drive down dark current. These more sophisticated structures, which will require a switch of growth  technology from liquid phase epitaxy to MBE, should lead to operating temperatures of 200 K. Efforts to improve the performance of MCT detectors are carried out in partnership with CEA-Leti. “We have had a long, historical relationship, and in 2003 we set up a joint lab with CEA-Leti to share our R&D,” explains David Billon-Lanfrey, Vice President of R&D, technology and products at Sofradir. “[CEA-Leti] gives us a lot of new ideas and a lot of expertise, in terms of technology. We bring the needs of the market and industry constraints. It’s one thing to do one demonstrator – it’s another thing for our infrared technologies to turn production at large quantities.”

A key requirement for any high-quality detector is that it has very few defective pixels. According to Sofradir, the proportion of defective pixels should be less than 0.5 percent. Unfortunately, as operating temperature rises, the ratio of defective pixels to good ones increases, due to various forms of noise. Engineers at Sofradir and CEA-Leti have reduced the proportion of defective pixels by driving down the density of defects and dislocations in the MCT layer through improvements in substrate quality and epitaxy.

“Passivation layers are also playing a big role,” adds Billon-Lanfrey, because improvements to the passivation layer suppress the current leakage out of the device.



Sofradir is keeping the details of this improved passivation process under wraps, but it is willing to disclose the benefits brought to device performance. In 2010, more than thirty MCT detectors were made using a range of processes. Through optimisation of the passivation  process and epitaxial growth, the maximum operating temperature for these detectors – which featured 384 by 288 pixels with a 15 μm pitch and had a cut-off wavelength of 5 μm at 80 K – increased from 90 K to 120 K.

More recently, Sofradir’s engineers have started to apply these improvements to the company’s Scorpio detector, which has 640 by 512 pixels and a 5 μm cutoff wavelength. This detector’s operating temperature has increased from 90 K to 120 K and the power required for cooling has halved.

“ We go from something like 3.5 W to 1.7 W, depending on the size of the components,” says Billon-Lanfrey. The company took this to the SPIE Defense Security and Sensing conference held in Orlando last April. At this gathering its main rival was a form of InSb-based detector featuring an InAlSb or InAsSb barrier layer and sporting an operating temperature of up to 150 K.

However, according to Billon-Lanfrey, the higher operating temperature came at the expense of an inferior cut-off wavelength, which suppressed the dark current. “In the temperature range we are looking at, the maximum emission is in the four-to-five micron band. With a cut-off at four microns, there aren’t enough photons to achieve good image quality in poor weather conditions.”

Inverted architecture



Another aim for the French outfit is to demonstrate a ‘pover- n’ MCT detector in 2012. “We expect to have a focal plane temperature of between 150 K and 170 K,” says Billon-Lanfrey, who adds that it will take roughly another two years before this technology is used in production.

Switching from the conventional n-over-p MCT detector to a p-over-n variant enables a substantial increase in operating temperature by cutting dark current by oneto- two decades. The fabrication of such a device requires far greater modification of the production process than that which occurred during the introduction of a superior passivation process. But the engineers at Sofradir don’t have to start from scratch. That’s because they and their colleagues at CEA-Leti started to develop a p-over-n technology for longwavelength detectors in 2003, which is now used on the Scorpio LW that has a 9.5 μm cut-off wavelength.

Sofradir is continuing to develop its p-over-n technology for long-wave detectors and apply it to medium-wave cousins. Billon-Lanfrey believes that this should enable the production of medium-wave detectors operating at up to 150 K. To reach even higher temperatures will require the introduction of more complex architectures that may incorporate a barrier layer. Today CEA-Leti is working on the development of this technology, which requires a shift from liquid phase epitaxy to MBE growth of the epilayers.

The plan is to demonstrate this form of detector in 2014- 2015 and start production two years’ later. If Sofradir can hit these goals and its near-term targets, it will deliver significant improvements in the performance of infrared detectors throughout the remainder of this decade. Commercial success appears destined to follow, giving those working in the infrared valley yet more achievements to be proud of.