Open Access
Add to BibSonomy
Abstract
Surface dark current often limits the performance of infrared photodetectors, especially as detectors become smaller. Bulk dark current mechanisms are well understood, but surface dark current mechanisms are not. Here, surface dark current mechanisms are identified, and examples of detector designs that can block these currents are given.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Dark current is a concern in detector design. Fluctuations in carrier concentration and associated dark currents produce noise. Additionally, dark current increases power consumption, cooling requirements, and read-out challenges [1]. Dark current increases with increasing cutoff wavelength, as smaller bandgaps produce larger carrier concentrations. Therefore, dark current is often the main performance limitation of MWIR and LWIR detectors. Dark current can flow through the bulk or along the surface sidewall of the detector. Bulk dark current, and its sources, have been thoroughly studied and are well understood. Surface dark current is less understood, although it can be larger than bulk dark current, and can limit performance, especially as detector size decreases. Smaller detectors have a larger surface to volume ratio, which enhances the significance of surface dark currents. Smaller pixel sizes are desirable in imaging arrays, as they decrease size, weight, and power consumption, while enabling higher resolution [1].
2. Bulk dark current mechanisms
Before discussing surface dark current mechanisms, bulk mechanisms are reviewed. Four major mechanisms are majority carrier drift current, tunneling (band-to-band and trap-assisted-tunneling (TAT)), generation-recombination (G-R) current, and diffusion current. Each mechanism has a unique combination of dependencies on temperature, Fermi energy (carrier concentration), and defect concentration. These dependencies can be experimentally determined, which enables identification of the dominant mechanism in a given device.
2.1 Majority carrier drift current
The largest of the dark current mechanisms is majority carrier drift current, which occurs when there are no majority carrier barriers in the structure. An example of such a barrier-less structure is a photoconductor detector with two ohmic contacts. The magnitude of drift current is proportional to majority carrier concentration [2]. The thermal activation energy varies with carrier concentration, ranging from near half-bandgap at low concentration to near zero with high concentration. More complicated device architectures can suppress bulk majority carrier drift current by incorporating barriers to block majority carrier flow [3].
2.2 Tunneling currents
Ideally, barriers inserted into detectors suppress majority carrier flow to undetectable levels. However, in some cases, tunneling through the barrier can produce significant dark current. There are two types of tunneling current mechanisms, band-to-band tunneling in which the current carrier tunnels through the barrier in a single jump, and trap-assisted-tunneling (TAT) in which the current carrier tunnels via defect (trap) states inside the barrier. Both types of tunneling mechanisms have very little temperature dependence and similar (but not identical) voltage dependence. TAT current is proportional to the defect density [4], one of three bulk dark current mechanisms that depend on the semiconductor material quality.
2.3 Generation-recombination (G-R) current
G-R current occurs by drift in a depletion region after Shockley-Read-Hall (SRH) generation through trap states in the bandgap. SRH generation is greatest when the Fermi level is at the intrinsic Fermi level [5]. G-R current is also proportional to defect density. Unipolar barrier detectors can limit this mechanism by eliminating depletion of the absorber [3].
2.4 Diffusion current
If the above mechanisms are suppressed sufficiently, bulk dark current becomes diffusion limited. Diffusion current originates from minority carrier generation in a neutral region. There are three generation mechanisms that can cause diffusion current: SRH, Auger, and radiative. These generation processes exhibit approximately full-bandgap activation energy and little voltage dependence (for voltages greater than a few times kT/e).
SRH is most commonly applied to depletion region processes, such as in pn junctions, where the Fermi level passes through mid-gap, producing G-R current. In such cases, SRH has a thermal activation energy equal to half of the bandgap. SRH is not limited to depletion regions; it can also occur neutral regions, i.e., Fermi level near a band edge. In neutral regions, the SRH activation energy approaches full-bandgap. SRH in a neutral region produces diffusion current. SRH-generated diffusion current is proportional to defect concentration. If defect concentrations are low enough, diffusion dark current can be dominated by Auger or radiative generation [6].
Each mechanism is uniquely dependent on device parameters, such as temperature, operating voltage, device dimensions, Fermi level position, and defect concentration. Identification of the dominant mechanism can be made by measuring the dependence of dark current on these parameters. Temperature is perhaps the most common parameter, used for identifying the dominant dark current mechanism; temperature dependences (thermal activation energies) are listed in Table 1.
Table 1. Thermal Activation Energy
3. Surface dark current mechanisms
The basic mechanisms that enable surface dark current are the same as those of bulk dark current, although characteristics of surface currents can be much different. As described above, the current mechanisms can be strongly affected by, among other factors, defect density and Fermi level, both of which are often significantly different on the surface than in the bulk. The surfaces of III-V materials are characterized by a high defect density and a pinned Fermi level. A unified defect model [7] states that these defects create a density of states on the surface much larger than any doping density, which sets the Fermi level to the defect energy regardless of doping level or type. Measurements of Schottky barrier heights give good information on the pinned surface Fermi level for all III-V binary materials of interest [8]. These measurements also show that the absolute energy of surface Fermi levels are quite similar for the common III-V materials used in infrared photodetectors. The pinned surface Fermi level has major effects on the dark current mechanisms described above: majority carrier drift, tunneling, diffusion, and G-R.
3.1 Surface majority carrier drift current
Surface majority carrier drift current has a small or zero thermal activation energy, and is produced by uninhibited flow of majority carriers along the surface [9]. The small activation energy limits the effectiveness of device cooling in reducing this dark current. In simple photoconductors, there is no barrier to block majority carriers, and the resulting current is typically the dominant dark current. More advanced devices are typically designed with barriers (pn or unipolar) to block bulk majority carriers; however, such barriers are often not effective in blocking surface current, as discussed below.
Simple photodiodes use pn-junctions to block bulk majority carriers. Along the surface, however, there is no pn-junction due to surface Fermi level pinning. The surface is a single conductivity type, regardless of doping, allowing surface majority carrier drift [10]. Figure 1 shows the measurement of dark current at various temperatures in InAs and GaSb pn-junction devices, enabling determination of the activation energy. At lower temperatures, the devices are surface-limited, exhibiting dark currents with low activation energies, which match known Schottky barrier heights for the two materials [8,10]. The low temperature dark currents were determined to be surface currents by examining the device size dependence, as described in section 3.3. InAs has a surface current with near-zero activation energy due to its degenerate surface. GaSb has a surface current with small, but non-zero, activation energy due to its non-degenerate surface.
Fig. 1 Arrhenius plot of zero-bias conductance of InAs and GaSb pn-junction devices. Both are limited by bulk currents at higher temperatures, and by surface majority carrier drift currents at low temperatures [14].
Unipolar barrier detector designs use band offsets at a heterojunction to block majority carriers, but pass minority carriers. These devices are usually designed to block bulk majority carriers. If the bulk and surface have the same conductivity type, the surface majority carriers will be the same type as the bulk majority carriers, and the unipolar barrier will block both. The first device to apply this concept to infrared detectors was the nBn [3], which used n-type absorber and cap layers, along with an AlAsSb-based electron barrier. Both the bulk and the surface of the absorber are n-type, enabling the unipolar barrier to block both bulk and surface majority carriers. The absence of bulk and surface majority carrier current is evidenced by the full bandgap activation energy of its dark current down to low temperatures, enabling a six-order-magnitude reduction in dark current over the photodiode [10].
The InAs nBn can achieve bulk-limited performance over a wide temperature range [3]. In order to improve bulk performance, strained-layer superlattice (SLS) absorbers are of interest, based on the predicted suppression of Auger and background-limited high operating temperature performance [11]. As a natural consequence of carrier confinement in SLS materials, mobility is reduced, especially for holes. This impedes minority carrier extraction for an nBn, therefore the pBp design is often preferred. The SLS pBp design requires the absorber to be doped p-type, creating p-type conductivity type in the bulk, but these SLS materials typically have n-type surfaces [10]. The hole barrier of such a device blocks the majority carrier drift current of the bulk, but the surface’s n-type majority carrier drift current is not blocked, a significant problem in this device.
Surface majority carrier drift current exists in unpassivated photoconductors or photodiodes. Unipolar barrier structures can block surface majority current, but only if the bulk and surface have the same majority carriers (conductivity type).
3.2 Surface trap-assisted-tunneling current
TAT is enhanced at the surface due to its dependence on defect density, which is large at the surface. Simple pn photodiodes are limited by majority carrier drift current, as described in the section 3.1, which masks TAT current. Unipolar barrier devices can exhibit surface TAT through the unipolar barrier, if the barrier sidewall is exposed by device processing. This can be seen in Fig. 2, which shows dark current at low temperature with minimal activation energy only for the nBn in which the barrier is exposed (deep-etched) [12]. Similar surface TAT current has also been reported for more complicated unipolar barrier structures [13].
Fig. 2 Arrhenius plot of InAs-based deep-etched, shallow-etched, and inverted unipolar barrier detectors (left), and Arrhenius plot of surface dark current for InAs-based deep-etched and inverted unipolar barrier detectors, created through variable-area diode analysis, explained in section 3.3 (right).
A solution to surface TAT would be to forgo etching through the unipolar barrier layer during device processing, avoiding exposure of the barrier sidewall. This will, in fact, eliminate surface TAT, but results in undesirable lateral diffusion current because the absorber layer is not etched through. In order to eliminate lateral diffusion, the absorber must be isolated. The standard nBn has its barrier on top of the absorber, therefore, the barrier must be exposed in order to isolate the absorber. This problem may be avoided with the use of an inverted design, for which the absorber is on top of the barrier. In such a device, the etch to isolate the absorber does not need to also etch through the barrier. Neither the conventional nBn with shallow etch (only the contact layer etched) nor inverted nBn devices show signs of surface TAT current, as expected [12].
3.3 Surface diffusion current originated by SRH
The SRH process is enhanced by high defect density, as shown in section 2.4. Due to the large defect density on the surface, diffusion current along the surface will be dominated by carriers generated through the SRH process. Both bulk and surface diffusion currents have full-bandgap activation energies. In order to separate the two, variable area diode analysis (VADA) can be performed. VADA takes advantage of the simple geometric relationship:
where A is the device area in cm2, and P is the device perimeter in cm. Dark current is measured for devices of various sizes, and thus various P/A ratios. A graph of I/A vs P/A produces a straight line; the slope determines the surface current and the y-intercept determines the bulk current. VADA can be performed at several temperatures, producing temperature dependence values of surface current, and surface current thermal activation energy can be found. Figure 2 shows such an analysis for two nBn designs. Both designs show surface SRH G-R current with full-bandgap activation energy, as expected [12].3.4 Surface G-R current originated by SRH in a sub-surface depletion region
A device may exhibit surface G-R current when the surface is depleted or inverted. The surface becomes inverted when the bulk is doped to a conductivity type that is opposite to the natural conductivity type of the surface, creating a sub-surface pn-junction. Although this condition exists in many designs, surface G-R current is often masked by the larger surface majority carrier drift current, e.g., the SLS pBp discussed in section 3.1. The masking surface majority carrier drift current in the pBp can be eliminated by adding an additional barrier to the design. A diagram of such a complimentary-barrier infrared detector (CBIRD) is shown in Fig. 3. The hole barrier blocks bulk majority carriers, while the electron barrier blocks surface majority carriers. Since both majority carrier drift currents are blocked, surface G-R current may now become observable.
Fig. 3 Schematic of a complementary barrier detector. Carriers are generated in the sub-surface depletion region near the absorber surface, and are not blocked by the barriers.
Surface G-R current is generated in the sub-surface depletion region. As shown in Fig. 3, the generated electrons are pushed to the surface by the built-in field of the sub-surface pn junction, flow along the surface, and exit through the hole barrier; the generated holes are pushed into the bulk by the sub-surface pn-junction and exit through the electron barrier. Such a dark current mechanism reveals itself as a surface dark current (established by VADA) with a half-bandgap activation energy. Indeed, a surface dark current with these characteristics has been reported in a complimentary barrier structure detector [14].
4. Summary
There are multiple surface dark current mechanisms that can limit the performance of infrared detectors. The mechanisms are the same as the bulk mechanisms; however, the magnitudes and other characteristics of the surface currents are usually different from those of the bulk currents due to differing Fermi levels and large defect densities at the surface. By examining the effects of temperature, device dimensions, and processing schemes on surface dark current, the current mechanisms can be identified. Table 2 shows several infrared detector designs and their limiting surface dark current mechanisms. The only device that exhibits no detectable surface dark current is the shallow-etched nBn, however this design has a different problem: lateral diffusion current. Devices that do not use a unipolar barrier are subject to surface majority carrier drift current. Devices that are etched through the barrier can exhibit surface TAT current. Devices that are etched through the absorber can exhibit surface diffusion current originated by surface SRH. Devices that have opposite conductivity types in the bulk and on the surface can exhibit surface G-R current.
Table 2. Limiting Surface Dark Current Sources for Infrared Photodetector Designs
Funding
NASA Space Technology Research Fellowship; U.S. Army Research Office.
References and links
1. A. Rogalski, P. Martyniuk, and M. Kopytko, “Challenges of small-pixel infrared detectors: a review,” Rep. Prog. Phys. 79(4), 046501 (2016). [CrossRef] [PubMed]
2. D. A. Neamen, Semiconductor physics and Devices (McGraw-Hill, 2012), Chap. 4.
3. S. Maimon and G. W. Wicks, “nBn detector, an infrared detector with reduced dark current and higher operating temperature,” Appl. Phys. Lett. 89(15), 151109 (2006). [CrossRef]
4. J. Y. Wong, “Effect of trap tunneling on the performance of long-wavelength Hg 1-x Cd x Te photodiodes,” IEEE Trans. Electron Dev. 27(1), 48–57 (1980). [CrossRef]
5. C. T. Sah, R. N. Noyce, and W. Shockley, “Carrier generation and recombination in pn junctions and pn junction characteristics,” Proc. IRE45(9), 1228–1243 (1957).
6. G. R. Savich, D. E. Sidor, X. Du, C. P. Morath, V. M. Cowan, and G. W. Wicks, “Diffusion current characteristics of defect-limited nBn mid-wave infrared detectors,” Appl. Phys. Lett. 106(17), 173505 (2015). [CrossRef]
7. W. E. Spicer, I. Lindau, P. Skeath, C. Y. Su, and P. Chye, “Unified mechanism for Schottky-barrier formation and III-V oxide interface states,” Phys. Rev. Lett. 44(6), 420–423 (1980). [CrossRef]
8. S. Tiwari and D. J. Frank, “Empirical fit to band discontinuities and barrier heights in III-V alloy systems,” Appl. Phys. Lett. 60(5), 630–632 (1992). [CrossRef]
9. R. Pal, R. K. Bhan, K. C. Chhabra, and O. P. Agnihotri, “Analysis of the effect of surface passivant charges on HgCdTe photoconductive detectors,” Semicond. Sci. Technol. 11(2), 231–237 (1996). [CrossRef]
10. D. E. Sidor, G. R. Savich, and G. W. Wicks, “Surface Leakage Mechanisms in III–V Infrared Barrier Detectors,” J. Electron. Mater. 45(9), 4663–4667 (2016). [CrossRef]
11. C. H. Grein, H. Cruz, M. E. Flatte, and H. Ehrenreich, “Theoretical performance of very long wavelength InAs/InxGa1-xSb superlattice based infrared detectors,” Appl. Phys. Lett. 65(20), 2530–2532 (1994). [CrossRef]
12. X. Du, G. R. Savich, B. T. Marozas, and G. W. Wicks, “Suppression of Lateral Diffusion and Surface Leakage Currents in nBn Photodetectors Using an Inverted Design,” J. Electron. Mater. 47(2), 1038–1044 (2018). [CrossRef]
13. A. Soibel, S. B. Rafol, A. Khoshakhlagh, J. Nguyen, L. Hoglund, A. M. Fisher, S. A. Keo, D. Z. Y. Ting, and S. D. Gunapala, “Proton radiation effect on performance of InAs/GaSb complementary barrier infrared detector,” Appl. Phys. Lett. 107(26), 261102 (2015). [CrossRef]
14. C. Asplund, R. M. von Würtemberg, D. Lantz, H. Malm, H. Martijn, E. Plis, N. Gautam, and S. Krishna, “Performance of mid-wave T2SL detectors with heterojunction barriers,” Infrared Phys. Technol. 59, 22–27 (2013). [CrossRef]
