Novel photodetector developments poised to revolutionise infra-red imaging

Citation

(2007), "Novel photodetector developments poised to revolutionise infra-red imaging", Sensor Review, Vol. 27 No. 1. https://doi.org/10.1108/sr.2007.08727aaf.005

Novel photodetector developments poised to revolutionise infra-red imaging

Novel photodetector developments poised to revolutionise infra-red imaging

A problem with today's high performance thermal imaging and night vision systems is that they rely on costly, epitaxially grown InGaAs (indium- gallium-arsenide) detectors. Now, two technologically different developments by university teams are poised to overcome this limitation. Research by the Department of Electrical and Computer Engineering at the University of Toronto involves the use of films of semiconductor nanoparticles, whilst a group from the Physics Department at America's Harvard University, working with colleagues from Texas and Virginia universities, have used a novel surface modification technique to extend the operating wavelength of silicon into the IR region.

The Canadian group fabricated their detectors from colloidal PbS (lead sulphide) quantum dot nanocrystals by first making a solution of the nanoparticles in ultra-pure oleic acid. A drop of the solution was then spin- coated into a film on a glass slide with a prefabricated planar gold electrode array. Placing the film in a solvent for 2h and then allowing it to evaporate resulted in an 800nm thick layer of light-sensitive quantum dots with nanometre dimensions (Figure 7).

Figure 7 A 4nm diameter PbS quantum dot showing the pattern of atoms that constitute the crystal (University of Toronto)

These films showed large photoconductive gains with responsivities greater than 10³ A W^¯1. The best devices exhibited a normalised room temperature detectivity (D^*) of 1.8 x 10¹³ jones (see Box) at a wavelength of 1.3mm. This is significant, as today's highest performance IR photodetectors are photovoltaic devices fabricated from epitaxially grown InGaAs that exhibit a peak D^* in the 10¹² jones range at room temperature, whereas the previous record for D^* from a photoconductive detector is around 10¹¹ jones. According to the research group, in addition to thermal imaging, the technology could find applications in medical imaging (transparent tissue windows exist at 800 and 1,100nm), fibre optic communications (wavelength range 1.3-1.6mm) and environmental monitoring. This is possible because this development overcomes the limited wavelength photosensitivity of silicon which deteriorates from 800nm and ends abruptly at 1,100nm, and also eliminates the need for costly, conventional IR-responsive compound semiconductors such as InGaAs. The team also claims that, in principle, the technique could be extended out to wavelengths of around 2mm using current materials, allowing uses in solar cells (800-2,000nm). The materials have also been incorporated into conjugated polymers, notably MEH-PPV – poly[2-methoxy-5-(2'- ethylhexyloxy-p-phenylenevinylene)]. In contrast to traditional semiconductors, conjugated polymers allow simple processing, low costs, physical flexibility and large area coverage.

Performance of photodetectors

For a photodetector, the figure of merit used to characterise performance is equal to the reciprocal of the noise equivalent power, normalized to unit area and unit bandwidth. The specific detectivity (D^*) is given by:

where A is the area of the photodetector and Df is the effective noise bandwidth. The unit of D^* is the jones, where 1 jones=1cm Hz^1/2W^¯1. It is named in honour to D. Clark Jones who first defined this term in a paper in 1959.

As noted above, silicon is transparent to wavelengths longer than about 1mm which makes it unsuitable for many near-IR applications. However, the group from Harvard has overcome this limitation by developing a novel microstructuring technique which modifies the material's band-gap, thus allowing it to absorb in the IR region. A Ti:sapphire laser was used to irradiate n-doped silicon wafers with a 1kHz train of 100fs (femtosecond) pulses in a sulphur-rich atmosphere to create a surface covered with 2-3mm-sized structures (Figure 8). According to the team, the laser causes ablation and melting of the silicon surface, which evolves and interacts with the gas before re-solidifying with an altered morphology. The resultant morphology depends strongly on the characteristics of the laser pulses: they must be ultra- short and very intense, and the nature and pressure of the gas surrounding the silicon during irradiation is critical.

Figure 8 Scanning electron micrograph showing the silicon photodetector's laser etched surface. The microstructures are 2-3mm tall and spaced at around 2-3mm intervals (Harvard University)

Depending on the type of gas the shape of the microstructures can vary from sharp and tall to rounded and short. During this process, the silicon's surface, which is normally grey and shiny, turns deep black and has been dubbed “black silicon”. In addition, to near-unity absorption in the visible region, the irradiated surfaces absorb over 80 percent of IR light with wavelengths as long as 2.5mm. The group is investigating the use this extended absorption range to fabricate high performance silicon solar cells which will convert more of the sun's energy into electricity. Photodiodes with high responsivity in both the visible and IR regions can be made using this process. The microstructured surface encourages multiple reflections which promote the absorption of light. However, this is only part of the picture: it is a combination of increased absorption in the IR and large gain that leads to the extension of the operating wavelength.

Figure 9 Response of the microstructured photodetectors compared to that of a conventional silicon PIN (Harvard University)

The incorporation of large amounts of sulphur during the laser irradiation process is responsible for significant absorption beyond 1,100nm. The photodetectors exhibited a responsivity of 92A W^¯1 at 850nm and 119A W^¯1 at 960nm (with a 3volt reverse bias) and continued to show a response at 1.31 and 1.55µm (Figure 9). The group is now aiming to commercialise this development and anticipates imaging applications in security and surveillance systems.