Fig. 2(a) shows the dark current density of the devices fabricated on the wafers A and B which are measured at 80 K. We note that the identical fabrication process is employed for the devices on both wafers A and B. The devices on two different wafers exhibit approximately one order of magnitude difference in the dark current density even if the same device design and process are employed. The dark current at cryogenic temperature (80 K) could be originated from different parts of the device. For example, it can be either from the depletion region in the vicinity between the barrier and the absorber or from the surface region covered with the passivation film. Thus, to further investigate the devices, the dark current density was measured by increasing temperature to 200 K. As the temperature increases, the dark current density is dominated by the diffusion current and recombination current from the absorber layer, which is greatly affected by the crystal quality of absorber layer rather than the process condition.
Next, temperature-dependent dark current density was measured at a bias voltage of –0.3 V for the devices on two different wafers as shown in Fig. 2(b). Both devices demonstrate the diffusion-limited dark current characteristics but with slightly different activation energies. The dark current density of the devices fabricated on the wafer B is approximately one order smaller than those on the wafer A. This result shows that the minority carrier lifetime in the absorber layer of the wafer B is longer than that of wafer A. Different minority carrier lifetime can be attributed to overall strain on the InAs/GaSb T2SL layer. Unbalanced strain can lead to the formation of extended defects that act as SRH recombination lefts, reducing the minority carrier lifetime. Especially, the dark current density of the fabricated devices on the wafer B are less than 2 × 10–5 A/cm2 at 120 K, which is similar to or excels the value recently reported using the same T2SL structure[18, 19].
The overall strain on the T2SL affects not only the electrical properties as described above but also the optical properties[5, 8, 11-13]. Fig. 3 shows the spectral responses of the devices fabricated on the wafers A and B, which were characterized using Fourier transform infrared (FTIR) spectrometer. The devices exhibit 40 % cut-off wavelength of 5.25 μm for the wafer A and 5 μm for the wafer B at 83 K. This result indicates that the activation energy of the devices fabricated on wafer B is larger than that of the devices on wafer A, confirming different absorber layer properties of two wafers and is also consistent with the results in Fig. 2(b). The energy bandgap of the absorber layers of the T2SL can be significantly varied despite of the identical design, MBE protocol, and the device fabrication process. Therefore, the growth quality of the T2SL layer can strongly affect the performance of the devices.
To further unveil the origin of different electrical and optical properties, HRXRD pattern was measured to compare the overall strain of the wafers A and B. Fig. 4 qualitatively shows the difference in the peak between GaSb substrate and the first superlattice (SL). The SL (+1) peak of the wafer B is closer to GaSb substrate than that of the wafer A. As shown in Table 1. the average lattice mismatch and periodicity obtained from HRXRD pattern quantitatively show different T2SL quality of the wafers A and B. The thickness of one period InAs/GaSb T2SL (periodicity) is calculated to be 59.18 Å for wafer A and 59.06 Å for wafer B, respectively. Although the periodicity of both wafers A and B is close to the original layer design and the growth of InAs/GaSb T2SL layer has been performed as expected; however, the mismatch values are clearly different: the mismatch of wafers A and B are 788.6 and 67.37 ppm, respectively. We believed that randomly formed 'InSb-like' interfacial layer during the MBE process greatly reduced the mismatch. The results of the HRXRD pattern in Table 1 and the dark current characteristics in Fig. 2 clearly demonstrate that overall strain on InAs/GaSb T2SL layer affects the minority carrier lifetime[7, 10, 11, 13], where the minority carrier lifetime could be increased by one order of magnitude due to the balanced strain.