Using controlled interdiffusion to make a two-color quantum dot IR photodetector
Quantum dot infrared photodetectors (QDIPs), based on self-assembled quantum dots (QDs), have attracted extensive interest in the past decade for mid- to long-wavelength (3–20μm) infrared photodetection. Due to three-dimensional carrier confinement, QDIPs outperform their quantum well (QW) counterparts in sensitivity to normal-incident IR radiation, responsivity, detectivity, and ability to operate at higher temperatures. Based on mature III-V compound semiconductor technology that offers high uniformity and low cost, QDIPs are also competitive with conventional HgCdTe detectors, especially for fabrication of two-dimensional focal-plane arrays.
Recent efforts have focused on making two-color or multi-color QDIPs for high-performance infrared systems in applications such as remote temperature sensing, chemical analysis, target identification and spectrometers. A straightforward approach to fabricating two-color QDIPs involves stacking two QD structures made from different materials (such as InGaAs/GaAs and InAs/GaAs) in order to generate distinct photoresponse peaks at different wavelengths.1
Another method uses the so-called dots-in-a-well (DWELL) structure.2 QDs embedded in a QW can detect two or more colors based on photocurrent contributions from transitions between QD and QW states, and between QD states and the continuum. However, due to the extremely sensitive self-organized process of QD formation, growing multiple QD stacks and DWELL structures is highly challenging, and the results are difficult to reproduce.
Using the well-known interdiffusion technique,3 we have developed a simple post-growth scheme for fabricating two-color QDIPs that avoids complicated growth processes and structure designs. Developed in early 1980s, interdiffusion—or atomic intermixing—tailors the optical properties (such as absorption coefficient and bandgap energy) of quantum-confined heterostructures by modifying their potential-energy profiles. In QD structures, simple thermal annealing causes a blueshift in the band-to-band luminescence, and a redshift in the intersubband photoresponse. The shifts arise from the thermally-driven compositional interdiffusion of the QD and barrier materials. On the other hand, the thermal stress effects of dielectric capping layers such as titanium dioxide (TiO2) suppresses interdiffusion in QDs.4 Thus, a two-color photodetector can made by applying thermal annealing to adjacent QDIPs on a QDIP structure, one capped by TiO2, the other not, and then removing the cap material.
The QDIP structure used in this work was an n-i-n structure5 grown on a semi-insulating GaAs (001) substrate by metal-organic chemical vapor deposition. Arsine (AsH3), trimethylgallium (TMGa) and trimethylindium (TMIn) were used as sources for arsenic, gallium, and indium, respectively. As shown in Figure 1, the structure contained 10 QD layers alternating with 50nm GaAs barrier layers, all sandwiched between highly Si-doped top (300nm) and bottom (1000nm) GaAs contact layers. Each QD layer comprised 5.7 monolayers of undoped InGaAs (nominally 50%). During operation, IR light shining on the detector excites QD charge carriers, which are swept away by an external electric field and collected as photocurrent.
To create the two-color detector, a 180nm TiO2 film was deposited by electron beam evaporation onto half of a suitably-masked wafer, followed by the rapid thermal annealing of the entire wafer at 750°C for 30s. The TiO2 layer was then removed by reactive ion etching using CHF3/Ar chemistry. Using standard photolithography, wet chemical etching, and metallization processes, the whole sample was fabricated into devices with 250μm × 250μm mesa structures. A 150μm-diameter round top contact and bottom contact were formed by the e-beam evaporation of Ge/Ni/Au, which was alloyed at 380°C for 60s. Figure 2(a) presents the schematic of the resulting two-color device structure.
Figure 2(b) and (c) show the normally-incident spectral photoresponse from the two-color QDIP. The uncapped, interdiffused QDIP's peak response is located at 6.7μm, which is 0.8μm higher than the capped region's response peak at 5.9μm. However, other important device characteristics of the two QDIPs, such as dark current, responsivity, and detectivity, are comparable.
Our results shows that proper mask design and optimized fabrication processes can yield a two-dimensional QDIP focal plane array with alternate two-color pixels. This may be useful for a variety of future applications such as thermal imaging.
The authors would like to thank M. Aggett, T. Halstead, and D. Gibson for their technical assistance. The financial support from Australian Research Council is also acknowledged.
