Abstract
We study the photoresponse and photocurrents of long-wavelength infrared quantum cascade detectors (QCDs) based on AlGaAs/GaAs material system. The photocurrent spectra were measured at different temperatures from 20 K to 100 K with a low noise Fourier transforming infrared spectrometer. The main response peak appeared at 8.9 μm while four additional response peaks from 4.5 μm to 10.1 μm were observed as well. We confirmed that the photocurrent comes from phonon assisted tunneling and the multipeak behavior comes from the complicated optical transition in the quantum cascade structure. This work is valuable for the future design and optimization of QCD devices.
1. Introduction
Quantum cascade detectors (QCDs) are photovoltaic detectors. They can operate under zero bias applied and consequently do not suffer any dark current which is expected as a promising alternative of quantum well infrared photodetectors (QWIPs) [1, 2]. For focal plane array (FPA) application, the QCDs perform pure photoresponse current to reach a longer integration time [3] and lower noise equivalent temperature difference (NETD) [4]. Based on bound-to-bound intraband transition, the line width of QCDs is narrow, which means high selectivity to wavelength. Moreover, QCDs have been shown to work at higher temperature [5] and been reported in different materials and all infrared atmospheric windows [5–9].
To prevent false-alarm and improve detection accuracy, two color FPAs were studied and fabricated [10–12]. Conventionally, two-color detection is achieved by two single-color detectors fabricated together. So, secondary photoresponse peaks are undesired which would act as cross talk and noise in response spectra. QCDs perform excellent for the narrowband detection. But confined with their multisized quantum wells, the photoexcited transitions are complicated. As a result, several secondary peaks appear. To eliminate cross talk and further improve the detector’s performance, detailed study on the secondary peaks is of particularly importance.
2. Materials and Methods
The samples under study are GaAs/AlGaAs QCDs designed to operate at 8.9 μm [3, 13]. The wafer used to fabricate this QCD sample was grown by molecular beam epitaxy (MBE), and the layer sequence from substrate to top was 400 nm n-doped GaAs bottom contact (Si-doped to 1 × 1018 cm−3), 40 periods of 7 coupled Al0.33Ga0.67As/GaAs superlattice active region, and 350 nm n-doped GaAs top contact in which doping concentration is equal to the bottom contact. The first QW of each period absorbs incident light and is n-doped with the doping concentration 5 × 1017 cm−3. The thickness of different QWs and barriers is 68/56.5/20/39.55/23/31/28/31/34/31/38/31/48/22.6 (Å) in roman and bold, respectively. The mesa size is 200 × 200 μm2 and electrical contacts were formed by a standard lift-off deposition of AuGe/Ni/Au (100/20/400 nm). To theoretically describe the electronic properties of QCDs, we applied the standard eight-band model. As shown in Figure 1, the electrons which the energy could be excited from to . is designed to equal which could allow the exited electrons resonant tunneling to reduce the recombination possibility and contribute to effective transport. Then, excited electrons tended to transport in the direction by LO-phonon assisted tunneling due to the coupling of wavefunction of adjacent wells.
3. Results and Discussion
Considering the complicated interactions between the intersubbands in QCDs, the absorption and response spectrum is not so pure and sensitive to 8.9 μm only. With the diffusion and tunneling of electrons, more than one absorption process could be figured out theoretically as Figure 2(a) dashed line indicated. Besides, we measured response photocurrent spectra under zero bias voltage at different temperature. QCD sample is soldered on a copper heat sink and placed in a closed cycle helium cryostat. The measurements were performed by using a SR570 low noise current preamplifier and a Fourier transform infrared spectrometer (Nicolet 6700). Response photocurrent spectra at different temperatures are shown as solid lines in Figure 2(a).
(a)
(b)
(c)
(d)
Three clear photoresponse peaks were found immediately in Figure 2(a) which are , , and processes. Other two peaks could be seen after zooming in which are corresponding to and transitions. It is to be noticed that the valley at 1250 cm−1 is a background absorption so that the peak at 1270 cm−1 is actually a fake one.
To further improve the detector’s performance, detailed study on the formation of multipeak spectra is needed. As is known, photocurrent could be expressed as [14] where is the quantity of photogenerated carriers, is electron charge, and is the transport velocity. Moreover, for a certain wavelength , the photocurrent could be given by Equations 1 and 2 are true when the electron concentration is high enough to be photoexcited. Otherwise, the is also restricted to electron concentration. Besides, corresponding to different photoexcited transitions, the transport paths are not all the same in QCDs, which would lead to different . At low temperature condition, the impurities are ionized less and mainly distributed in subband. Thus the optical excited transitions primarily occur from to , , and , respectively. With temperature rises, more impurities will be ionized and electron concentration will get higher. Consequently, more quantum states of subband would be occupied. According to Pauli exclusion principle and Hund’s rules, electrons would tend to be distributed at higher energy levels such as by means of diffusion and tunneling. Then, optical transition from is enhanced as Figures 2(b) and 2(c) indicated.
The calculated electrons distribution at different temperatures is presented in Figure 3. It can be seen that electron density at subband is hard to be identified at 50 K, while the density at or even other subbands becomes larger with temperature increases. At temperature higher than 40 K, the electron density grows exponentially rapidly as the inset Figure 3 indicated, and corresponding to Figure 2(b), the response of transition becomes clear at 40 K. This indicated that our QCD can be operated around 80 K, but the electron concentration of grows to ten times as that at 40 K which would be a nonignorable cross talk.
In addition, band-gap and barrier height would be influenced by temperature. Energy levels would be rearranged by structure variation of quantum wells, especially the quasibound energy levels such as . As for photoconductive QWIPs working at nonzero applied bias, the bound-to-quasibound transition appears redshift with temperature rises [15]. But for QCDs, null bias working condition makes the transport mechanisms different. As is expected, transition appears blueshift with temperature getting higher as Figure 2(d) indicated. The experiment results and calculated energy gap between and are shown in Figure 4. A general agreement has been obtained.
From responsivity and dark current, we can deduce the Johnson noise limited detectivity given by [2] where is the peak responsivity, is the device resistance at null bias, is the mesa surface, and is the temperature of the sample. We obtained = 1.4 × 1010 Jones at 80 K as Figure 5 presented. Moreover, after the elimination of secondary response peaks, the factor in spectra would get higher, leading to a higher .
In our QCD detector samples, the and transitions are expected to be the primary optical transitions. The spectra noises which act as cross talk could be classified in two categories. The first category is the transitions from to non- subbands such as and the second is excited from non- (for example, ) energy levels. For the first category, thicker barriers would reduce the overlap of wavefunctions, leading to a lower transition rate. For the second category, an effective way is to reduce the electron occupation of ground states at non- subbands.
4. Conclusion
In conclusion, we have demonstrated a GaAs/AlGaAs multipeak quantum cascade detector. The main response peak appeared at 8.9 μm while four other response peaks from 4.5 μm to 10.1 μm were observed and analyzed. We confirmed that the multipeak performance comes from the complicated optical transition in the quantum cascade structure. This work is valuable for the design and optimization of QCD devices.
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
Acknowledgments
This work was supported by the National Nature Science Foundation of China (Grant no. 61106092) and the State Key Development Program for Basic Research of China (Grants no. 2013CB632802).