Heterostructures improve IR detector performance
IR imagers are important for a variety of applications, including (but not limited to) noninvasive medical diagnostics and technological functions in security and defense. However, cost is a major limitation because of the cooling systems required to keep such devices at the required low operating temperatures. Thermal-imaging devices exist that operate at room temperature, but they lack speed and do not acquire any spectral information. Rather, they only detect differences in emissivity. Commercial photodetector technologies like mercury cadmium telluride (MCT) are available, but they are expensive because of low production yield. To overcome the requirement for low operating temperatures that restrict use of IR photodetectors, semiconductor-heterostructure designs have been developed.
We are working on III-V semiconductor-heterostructure designs to achieve improved IR device performance at increased operating temperatures.1–4 We are focusing on a heterostructure that incorporates a design known as ‘nBn,’ where the sequence of letters refers to different layers of the device.5–7 Some problems that remain with this structure are related to parameter optimization of the different layers for various material systems and, more fundamentally, lack of a basic understanding of what is happening at the interface of the different layers. Our group recently investigated indium arsenide antimonide (InAsSb)-based nBn detectors with different material compositions of the B layer.
The nBn detector architecture operates like a regular semiconductor p-i-n photodiode but with lower dark current (i.e., the current present when no radiation is incident on the aperture), which is a limiting factor for IR detectors. The reason for this decrease is the absence of a depletion region (the built-in potential of a standard photodiode), which contributes to a dark-current component known as generation-recombination (G-R) current. By removing this dark-current source, nBn detectors ideally operate with superior performance compared to standard photodiodes at the same operating temperature, or with similar performance at higher operating temperatures (see Figure 1).
We used InAsSb and aluminum arsenide antimonide (AlAsSb) for the n and B regions, respectively, while we varied parameters such as the material composition and impurity doping concentration of the B region. We had access to four devices (A, B, C, and D). Devices A and B have a B-region material composition of AlAs0.15Sb0.85 and AlAs0.10Sb0.90, respectively, with no doping. Devices C and D have the same respective material composition, but with a doping concentration of 1×1017cm−3 of tellurium. We determined device characteristics such as spectral response (wavelength range seen by the detector), responsivity (output-current response for a given radiation input), detectivity (signal-to-noise ratio), and dark current.
Table 1 presents a summary of the characteristics measured at an operating temperature of 200K and applied bias of 0.2V. The spectral response showed that the detectors had a cutoff wavelength (longest wavelength seen) of ~4μm. In terms of responsivity and detectivity, device C performed best, with values of 1.84A/W and 2.0×1011Jones (1Jones = 1cm Hz1/2/W), respectively. Device B had the lowest dark-current density (1.6×10−4A/cm2), although less than a factor of two lower than that of device C (2.4×10−4A/cm2). For reference, a commercial MCT detector8 has a responsivity of 2.2A/W, detectivity of 5.4×1010Jones, and dark-current density of 2.5×10−3A/cm2 at an operating temperature of 200K. We also measured the dark-current density as a function of temperature, which indicates whether the G-R current limits detector performance. For devices A and B, we found that at low temperatures G-R current was the dominating current component, but for devices C and D its contribution was significantly reduced. These results indicate that for InAsSb/AlAsSb/InAsSb-based nBn detectors, n-type impurity doping in the B region is required to remove the G-R current.
| Device | Responsivity (A/W) | Detectivity (Jones) | Dark-current density (A/cm2) | Quantum efficiency (%) |
| A | 0.45 | 2.9×1010 | 6.7×10−4 | 0.16 |
| B | 0.96 | 1.3×1011 | 1.6×10−4 | 0.33 |
| C | 1.84 | 2.0×1011 | 2.4×10−4 | 0.62 |
| D | 1.76 | 1.1×1011 | 3.3×10−4 | 0.58 |
In summary, nBn heterostructure design is an effective tool for reducing the limitation imposed by dark current. We have investigated this architecture by adjusting the material composition and doping concentration in the B layer to optimize device performance. Among the four devices tested, we found that a structure with a composition of AlAs0.15Sb0.85 and doping concentration of 1×1017cm−3 tellurium achieved the best device performance. Our study establishes a basis for understanding how doping and composition of the B region affect InAsSb/AlAsSb/InAsSb-based nBn detectors. This will help us in developing models for understanding fundamental properties of the nBn design.
Stephen Myers received his BS degree in engineering physics from Tarleton State University in Stephenville (Texas). He is currently pursuing a PhD in nanoscience and microsystems, researching IR detectors using III-V materials, including type-II indium arsenide/gallium antimonide superlattices (in which the conduction and valence subbands are staggered, so that electrons and holes are confined in different layers) and InAsSb.
