Open Access
Add to BibSonomy
Abstract
A longwave-infrared photodetector made of double layers of 100nm amorphous germanium (a-Ge) and 25nm amorphous silicon (a-Si) have been demonstrated. Under room temperature, the device shows the responsivity of 1.7 A/W, detectivity of 6×108 Jones, and noise equivalent power (NEP) of 5pW/√Hz under 5V bias and at 20kHz operation. Studies of frequency dependent characteristics and device modeling indicate that, above 100Hz or beyond the bandwidth of thermal response, the device operates as a quantum detector having the photoelectrons produced by optical excitation from the bandtail states to the mobile states of a-Ge. The superior device performance may be attributed to the combination of two amplification mechanisms: photoconductive gain in a-Ge and cycling excitation process (CEP) in a-Si, with the latter being the dominant factor. Besides its attractive performance, the device has a simple structure and is easy to fabricate at low cost, thus holding promise for night vision, sensing, autonomous driving, and many other applications.
© 2019 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Long-wave Infrared (LWIR) detection has been widely used in a large number of systems including night vision, thermal sensing, remote sensing, autonomous driving, robotics vision, machine vision, disease detection, and scientific research [1]. Over the past few decades, a large number of LWIR detectors have been demonstrated, including microbolometer [2,3], HgCdTe (MCT) [4], multi-quantum well (MQWIP) or quantum dots (QDIP) [5,6], Type-II superlattice structure (T2SL) [7,8], blocked impurity band (BIB) trap detectors [9,10], and graphene detectors [11]. According to the operation environment, LWIR detectors can be divided into two groups, uncooled detectors and cooled detectors. The former are mostly bolometers that detect IR induced temperature change and the output signal can be represented via different transduction mechanisms [12,13]. Even though microbolometers have demonstrated room temperature detectivity of 109 Jones, they generally have limited bandwidth less than 100 Hz [2,3]. As a quantum detector, Type-II superlattice (T2SL) structure has shown uncooled, broad IR spectrum detection ranging from short-wave infrared (SWIR) to very-long wavelength infrared (VLWIR) light [14,15]. The highest reported TIIS LWIR detectivity is about 6×108 Jones with an optical immersion design [7].
In contrast with uncooled LWIR detectors, almost all cooled LWIR detectors are quantum detectors with their photoresponse generated by photoexcited electrons or electron-hole pairs and having photocurrent as the output signal. At cryogenic temperature, cooled LWIR detectors show higher performance than uncooled detectors in general. For example, HgCdTe detectors have shown detectivity of 1011 Jones at lower than 100K [4]. Multi-quantum well (MQWIP) or quantum dots infrared photodetectors (QDIP) also achieved detectivity of 1011 Jones under ∼77K [6,16,17]. Lastly, BIB trap LWIR detectors using shallow impurity states in highly doped Si or Ge have achieved detectivity as high as 1014 Jones, but the devices have to be operated at extremely low temperature (∼10K) or the impurity states would be thermally excited to produce unacceptably high dark current [10].
For most applications, it is highly desirable to have room temperature LWIR detectors that are easy to fabricate at low cost, and can produce high detectivity, low noise equivalent power (NEP), and high frequency response much greater than 100Hz. However, few LWIR detectors reported to date can meet all these requirements. Most micro bolometers have limited frequency response, and most quantum detectors require cooling and complicated materials or processing. HgCdTe, being one of the prevailing choices for LWIR detectors, is expensive and difficult to grow and scale. Similar concerns also apply to MQWIP, quantum dots, T2SL, and BIB detectors.
In this paper, we demonstrate a unique design for uncooled LWIR detectors using a combined structure of amorphous Ge (a-Ge) and amorphous silicon (a-Si), both intentionally undoped. The a-Ge layer absorbs LWIR light and the a-Si layer produces carrier multiplication via cycling excitation process (CEP) we reported earlier [18–23]. The device has a very simple structure, is easy and low cost to fabricate, and can operate under room temperature, showing high responsivity (1.7 A/W), high detectivity (6×108 Jones), low noise equivalent power (5pW/√Hz), and high frequency response of greater than 20 kHz, limited by the modulation bandwidth of our CO2 laser source.
2. Fabrication and methods
2.1 Fabrication
A 25nm thick a-Si layer is deposited at 270°C by plasma-enhanced chemical vapor deposition (PECVD) on a highly n-doped silicon substrate, followed by room temperature sputtering of a 100nm thick a-Ge layer as the absorption medium. After the a-Si and a-Ge deposition, 30µm diameter device mesas are photolithographically defined and dry etched with SF6:C4F8. To passivate the mesa sidewall, Al2O3 (40 nm) and SiO2 (200nm) are deposited by atomic layer deposition at 200°C and by PECVD at 270°C, respectively. Finally, a 18nm Cr/Au layer is deposited to form the top electrode with 10% transmission at 10.6µm wavelength light, and a 220nm Ti/Au layer is further deposited to form the top and ground contact pads. The cross-section and top view of the device structure are shown in Fig. 1.
2.2 Methods
Dark current is measured with a Keysight B2900 source meter from 0V to −5V. Bias dependent photocurrent is measured by an RF spectrum analyzer, after being amplified by a transimpedance amplifier (TIA). The RF spectrum analyzer also gives rise to the noise spectrum for noise equivalent power (NEP) measurement. To measure the device response between 2 Hz and 6 kHz, we use a chopper to modulate a CO2 laser (10.6µm wavelength) operated at 100% duty cycle. To characterize the detector response at higher frequency than 6 kHz, we modulate the CO2 laser output at 5 kHz and characterize the output power at higher (up to the 4th) harmonic frequencies as the input optical power. Using this approach, we are able to extend the range of measurement to 20 kHz. The laser spot on the device surface is 5×5 mm2, producing an optical power density of 4mW/cm2. The incident light power is calibrated with a commercial detector (Thorlabs: S401C). All measurements are performed at room temperature.
From the measured photocurrent at 1Hz resolution bandwidth from an RF spectrum analyzer, we calculate frequency-dependent responsivity (R = Iphoto/Poptical), shot noise limited specific detectivity (D*=R/√(2eJdark)), and noise equivalent power (NEP=√A/D*) under different bias voltage. In these expressions, A is the device area, e is magnitude of electron charge, Poptical is the CO2 laser power incident on the device area, and Jdark is the device dark current density.
3. Results and discussion
We measure the device photo-response under 27nW CO2 laser power modulated at different frequencies (DC to 20 kHz) and under different bias voltage. The bias dependent DC dark current and responsivity at 5kHz are shown in Fig. 2. We plot the responsivity between 2.5V and 5V bias here because, lower than 2.5V bias, the responsivity is too low to be measured reliably. Both the dark current and photo responsivity increase with the bias voltage. The detailed analysis of the device behaviors will be presented later.
Fig. 2. Bias dependent (a) DC dark current and (b) Responsivity at 5kHz under 27nW CO2 laser illumination
Using the measured responsivity, we can calculate the specific detectivity (D*) because above 2.5V, the dark current and the RF spectrum of the device show the dominant noise of the device is shot noise. The results are shown in Fig. 3. We observe the highest specific detectivity of 6×108 Jones and the lowest noise equivalent power (NEP) of 5pW/√Hz at 5V reverse bias with a 30µm diameter device at 10.6 µm wavelength. Unlike bolometers or thermopiles that measure thermal effects and have limited (usually <100 Hz) frequency response, our device can produce high detectivity and low NEP at 20 kHz frequency, limited by the CO2 laser response.
Figure 4 shows the frequency dependence of device characteristics from 2 Hz to 20 kHz. We clearly observe that the device characteristics are divided into two regimes according to frequency. At 10 Hz or lower frequency, the device responsivity and detectivity reach 10 A/W and 7×109 Jones, which by themselves set a record value for uncooled LWIR detector, but the numbers drop rapidly above 10 Hz, showing the general property of thermal detectors. As a thermal detector, different materials of the device absorb the CO2 laser power, causing temperature rise which subsequently increase dark current that is revealed as “photoresponse”. However, what is more interesting and important is the higher frequency response of the detector. At higher than 100 Hz and up to 20 kHz, the maximum frequency we have measured, the device shows a photo responsivity of around 1.7A/W and detectivity of 6×108 Jones, and the values are nearly independent of frequency within measurement errors. The results indicate that above 100Hz, thermal effects vanish because of their slow response, and the device operates under a different mechanism which we will investigate in detail later.
Fig. 4. (a) Frequency dependent responsivity and detectivity of the device under 5V reverse bias (b) Photocurrent trace from RF Spectrum Analyzer at 5kHz laser modulation.
Table 1 shows a comparison of our detector with other LWIR detectors that operate under room temperature. In spite of its simple structure and fabrication process, our device has shown highly favorable performance in many key metrics. Compared with thermal detectors, our device has demonstrated much higher speed. Compared with quantum detectors, our detectors offer the best room temperature performance and are easiest to fabricate with low cost materials such as a-Ge and a-Si.
Table 1. Comparison of NEP (pW/√Hz) and specific detectivity, D* (100M Jones) for different uncooled LWIR detectors
The frequency dependent characteristics suggest that our detector operates as a quantum detector (i.e. signal produced by photoexcited electrons or electron-hole pairs by absorption of LWIR photons) rather than a thermal detector at higher than 100Hz. One key question to answer is what quantum transition(s) is responsible for the device behaviors at this wavelength. Another important question is what signal amplification mechanism(s) take place to give the device its superior performance. In the following we propose a physical model and use carrier transport equations to analyze the optical excitation pathway and signal amplification mechanism for the LWIR detector.
To obtain information to help develop the physical model, we have measured the photoresponse of two test structures on n-Si substrate, one with a 100 nm a-Ge layer and another with a 25 nm a-Si layer. The test structure with 100 nm a-Ge film produced responsivity of 0.01 A/W at 5kHz under 0.5 V bias. On the other hand, the test structure with a 25 nm a-Si layer showed very low LWIR responsivity (in the order of 10−4 A/W). This result indicates that it is the a-Ge layer that contributes to the primary photocurrent. The purpose of a-Si layer is to amplify the primary photocurrent via the CEP mechanism.
Based on the above observation, we hypothesize that the main operation mechanism under high (>100Hz) frequency is due to photoexcitation of electrons from the bandtail states to the mobile states of the conduction band of a-Ge. The excited electrons in the mobile states have much greater mobility than those in the bandtail states and can cross the a-Ge/a-Si heterointerface to enter the a-Si layer where a high electric field exists under voltage bias. The a-Si layer provides efficient carrier multiplication via the cycling excitation process (CEP) reported in our earlier publications [19,22,23] producing high responsivity.
Assuming that photoexcited electrons are generated from electron transition between the bandtail states and the mobile states of conduction band of a-Ge, we can write continuity Eqs. (1) and (2) for electron concentrations in the mobile states of conduction band (n) and in the conduction bandtail (nt).
To write Eqs. (1) and (2), we have assumed $\frac{d}{{dx}}{J_n} \approx \frac{d}{{dx}}({en{v_d}} )\approx e{v_d}\frac{{dn}}{{dx}} \approx 0$ where Jn represents the electron current density in a-Ge and vd is the drift velocity of electrons in the mobile states. Here we have ignored the current contribution by electron hopping over the localized states. We can express vd in terms of electron transit time ttr and a-Ge thickness daGe as ttr = daGe/vd.
After crossing the a-Ge/a-Si interface, electrons will experience bias-dependent CEP gain in a-Si. Since the energy barrier for a-Ge/a-Si interface is small for electrons, we assume 100% electron injection efficiency from the mobile states of conduction band of a-Ge to a-Si. Ignoring the drift velocity difference between electrons and holes in the mobile states of conduction and valence band for simplicity, we can represent the total particle current density, J, as
where J is the sum of the electron current density and the CEP generated hole current density, as illustrated in Fig. 5.Fig. 5. Band diagram of the device under reverse bias. Here, τ’−1(τ−1) represent generation recombination rate from conduction band (bandtail) to valence band, γn(γt) are electron capture (emission) rate to (from) bandtail state, σGp is electron generation rate via photoexcitation of bandtail states by incident LWIR photon flux.
By setting the d/dt terms to be zero for DC analysis and using the relation noγn = ntoγt, n can be derived from Eqs. (1) and (2) as
From Eqs. (3) - (5), we can find the dark current density (Jdark), photocurrent density (Jphoto), and responsivity $(R )$ as
In Eq. (8), the responsivity contains two signal amplification terms, the CEP gain produced in a-Si and the photoconductive gain in a-Ge defined by GPC = tlife/ttr with tlife being the effective electron lifetime represented by Eq. (5) and ttr the electron transit time across the a-Ge layer.
The above model shows that our device can operate as a “quantum detector” in which LWIR light excite electrons from the bandtail states to the mobile states of conduction band. Such excited electrons experience two gain mechanisms in series, the conventional photoconductive gain in a-Ge and the CEP gain in the high field a-Si region. The CEP gain is critical because the photoconductive gain is modest or in some case, negligible in our device. Due to the high density of bandtail states in a-Ge and high phonon emission rate for electrons to fall into the bandtail states from the mobile states, the effective lifetime for electrons to stay in the mobile states is in picoseconds. On the other hand, the low mobility and low E-field in a-Ge (most E-field drops in a-Si layer) produces a transit time of the order of 10ps. As a result, the photoconductive gain GPC = tlife/ttr in our device is modest. The device largely relies on the CEP gain to increase its photo responsivity and detectivity.
Figure 6 shows the bias-dependent CEP gain under visible (639 nm) light excitation from the same device structure without the a-Ge layer. Multiplying the responsivity of ∼10−2 A/W for the a-Ge test structure by the CEP gain of over 100 from Fig. 6, we achieve the net responsivity of > 1A/W for the a-Si/a-Ge heterostructure device under 5 V bias. This value agrees well with the measured responsivity (1.7 A/W) in Fig. 2(b). The result further confirms that CEP gain plays a key role in achieving high responsivity, high detectivity, and low NEP for our device.
Fig. 6. Photoresponse (at 639 nm wavelength) of a device consisted of an a-Si layer on an n-Si substrate (same structure but without the a-Ge LWIR absorption layer) for characterization of CEP gain.
4. Conclusion
We have demonstrated a novel design for room temperature LWIR detector. The device uses a-Ge for light absorption and a thin a-Si layer to produce CEP gain for photocurrent amplification. At 20kHz optical modulation, a 30µm diameter device has shown a detectivity value of 6×108 Jones, which is among the highest detectivity for room temperature LWIR detectors. The device shows a NEP value of 5pW/√Hz, the lowest value for all uncooled LWIR detectors. Frequency dependent measurements show the detector operates as a quantum detector above 100 Hz, and the CEP gain is key to its superior performance. The fast response, room temperature operation, high detectivity, low NEP, and simple and low-cost fabrication process make the device attractive to night vision, machine vision, autonomous driving and many other applications.
Funding
Office of Naval Research (N00014-15-1-2211); National Science Foundation (ECCS-1542148).
Acknowledgments
We thank the staff of the UCSD Nano3/SDNI facility and Dr. Yu-Hsin Liu for their technical support. SDNI is a site of NNCI.
Disclosures
The authors declare no conflicts of interest.
References
1. J. Caniou, Passive Infrared Detection: Theory and Applications (Springer US, 1999), Chap.1.
2. C. Chen, X. Yi, X. Zhao, and B. Xiong, “Characterizations of VO2-based uncooled microbolometer linear array,” Sens. Actuators, A 90(3), 212–214 (2001). [CrossRef]
3. A. K. R. Koppula, A. Abdullah, T. Liu, O. Alkorjia, C. Zhu, C. Warder, S. Wadle, P. Deloach, S. Lewis, E. Kinzel, and M. Almasri, “Material response of metasurface integrated uncooled silicon germanium oxide SixGeyO1-x-y infrared microbolometers,” Proc. SPIE 11002, 110021L (2019). [CrossRef]
4. VIGO, “VIGO Catalogue,” https://www.photonicsolutions.co.uk/upfiles/VIGOCatalogueLG06Feb18.pdf.
5. J.-W. Kim, J.-E. Oh, S.-C. Hong, C.-H. Park, and T.-K. Yoo, “Room temperature far infrared (8/spl sim/10 µm) photodetectors using self-assembled InAs quantum dots with high detectivity,” IEEE Electron Device Lett. 21(7), 329–331 (2000). [CrossRef]
6. S. Chakrabarti, A. D. Stiff-Roberts, X. H. Su, P. Bhattacharya, G. Ariyawansa, and A. G. U. Perera, “High-performance mid-infrared quantum dot infrared photodetectors,” J. Phys. D: Appl. Phys. 38(13), 2135–2141 (2005). [CrossRef]
7. K. Michalczewski, Ł Kubiszyn, P. Martyniuk, C. H. Wu, J. Jureńczyk, K. Grodecki, D. Benyahia, A. Rogalski, and J. Piotrowski, “Demonstration of HOT LWIR T2SLs InAs/InAsSb photodetectors grown on GaAs substrate,” Infrared Phys. Technol. 95, 222–226 (2018). [CrossRef]
8. H. Mohseni, J. Wojkowski, M. Razeghi, G. Brown, and W. Mitchel, “Uncooled InAs-GaSb type-II infrared detectors grown on GaAs substrates for the 8-12-/spl mu/m atmospheric window,” IEEE J. Quantum Electron. 35(7), 1041–1044 (1999). [CrossRef]
9. S. I. Woods, J. E. Proctor, T. M. Jung, A. C. Carter, J. Neira, and D. R. Defibaugh, “Wideband infrared trap detector based upon doped silicon photocurrent devices,” Appl. Opt. 57(18), D82–D89 (2018). [CrossRef]
10. F. Szmulowicz, F. L. Madarasz, and J. Diller, “Temperature dependence of the figures of merit for blocked impurity band detectors,” J. Appl. Phys. 63(11), 5583–5588 (1988). [CrossRef]
11. Y. Kawano, “Wide-band frequency-tunable terahertz and infrared detection with graphene,” Nanotechnology 24(21), 214004 (2013). [CrossRef]
12. Y. Zhao, M. Mao, R. Horowitz, A. Majumdar, J. Varesi, P. Norton, and J. Kitching, “Optomechanical uncooled infrared imaging system: design, microfabrication, and performance,” J. Microelectromech. Syst. 11(2), 136–146 (2002). [CrossRef]
13. F. Niklaus, C. Vieider, and H. Jakobsen, “MEMS-based uncooled infrared bolometer arrays: a review,” Proc. SPIE 6836, 68360D (2007). [CrossRef]
14. B.-M. Nguyen, M. Razeghi, V. Nathan, and G. J. Brown, “Type-II M structure photodiodes: an alternative material design for mid-wave to long wavelength infrared regimes,” Proc. SPIE 6479, 64790S (2007). [CrossRef]
15. M. Razeghi and B.-M. Nguyen, “Band gap tunability of Type II Antimonide-based superlattices,” Phys. Procedia 3(2), 1207–1212 (2010). [CrossRef]
16. S. Adhikary, Y. Aytac, S. Meesala, S. Wolde, A. G. Unil Perera, and S. Chakrabarti, “A multicolor, broadband (5–20 µm), quaternary-capped InAs/GaAs quantum dot infrared photodetector,” Appl. Phys. Lett. 101(26), 261114 (2012). [CrossRef]
17. S. D. Gunapala, S. V. Bandara, J. K. Liu, E. M. Luong, N. Stetson, C. A. Shott, J. J. Bock, S. B. Rafol, J. M. Mumolo, and M. J. McKelvey, “Long-wavelength 256/spl times/256 GaAs/AlGaAs quantum well infrared photodetector (QWIP) palm-size camera,” IEEE Trans. Electron Devices 47(2), 326–332 (2000). [CrossRef]
18. Y.-H. Liu, L. Yan, A. C. Zhang, D. Hall, I. A. Niaz, Y. Zhou, L. J. Sham, and Y.-H. Lo, “Cycling excitation process: An ultra efficient and quiet signal amplification mechanism in semiconductor,” Appl. Phys. Lett. 107(5), 053505 (2015). [CrossRef]
19. L. Yan, Y. Yu, A. C. Zhang, D. Hall, I. A. Niaz, M. A. Raihan Miah, Y.-H. Liu, and Y.-H. Lo, “An amorphous silicon photodiode with 2 THz gain-bandwidth product based on cycling excitation process,” Appl. Phys. Lett. 111(10), 101104 (2017). [CrossRef]
20. M. A. R. Miah, I. A. Niaz, Y.-H. Liu, D. Hall, and Y.-H. Lo, “A high-efficiency low-noise signal amplification mechanism for photodetectors,” Proc. SPIE 10108, 101080X (2017). [CrossRef]
21. L. Yan, M. A. R. Miah, Y.-H. Liu, and Y.-H. Lo, “Single photon detector with a mesoscopic cycling excitation design of dual gain sections and a transport barrier,” Opt. Lett. 44(7), 1746–1749 (2019). [CrossRef]
22. I. A. Niaz, M. A. R. Miah, L. Yan, Y. Yu, Z. He, Y. Zhang, A. C. Zhang, J. Zhou, Y. Zhang, and Y. Lo, “Modeling Gain Mechanisms in Amorphous Silicon due to efficient carrier multiplication and trap induced junction modulation,” J. Lightwave Technol. 37(19), 5056–5066 (2019). [CrossRef]
23. Y. Yu, Z. Xu, S. Li, A. C. Zhang, L. Yan, Z. Liu, and Y. Lo, “Plasmonically Enhanced Amorphous Silicon Photodetector With Internal Gain,” IEEE Photonics Technol. Lett. 31(12), 959–962 (2019). [CrossRef]
24. Hamamatsu, “InAsSb photovoltaic detectors,” https://www.hamamatsu.com/resources/pdf/ssd/p13894_series_kird1133e.pdf.
25. H.-H. Yang and G. M. Rebeiz, “Sub-10 pW/Hz0.5 room temperature Ni nano-bolometer,” Appl. Phys. Lett. 108(5), 053106 (2016). [CrossRef]
26. BostonElectronics, “Heimann Thermopiles,” https://www.boselec.com/wp-content/uploads/Linear/Heimann/HeimannLiterature/Heimann-Thermopiles-3-29-19.pdf
27. J. Tauc, A. Abrahám, R. Zallen, and M. Slade, “Infrared absorption in amorphous germanium,” J. Non-Cryst. Solids 4, 279–288 (1970). [CrossRef]
