ZhangYu123
ZhouSongmin1
LiXun1
WangXi1
ZhuLiqi123
LinChun13*
-
(Key Laboratory of Infrared Imaging Materials and Detectors, Shanghai Institute of
Technical Physics, Chinese Academy of Sciences, Shanghai 200083, China)
-
(University of Chinese Academy of Sciences, Beijing 100049, China)
-
(School of Information Science and Technology, ShanghaiTech University, Shanghai 201210,
China)
Copyright © The Institute of Electronics and Information Engineers(IEIE)
Index Terms
HgCdTe, detector, unstable pixels, surface passivation, dark current
I. INTRODUCTION
At present, MCT infrared detector has played an essential role in many application
fields, such as night imaging, intelligent appliances, space, and biological detection
[1-7]. Many important physical parameters are presented to evaluate the performance of
infrared FPA, such as detection efficiency, noise, resolution, effective pixels rate,
etc. In a space between effective and non-effective pixels, the unstable pixels usually
appear as effective, turning into non-effective pixels when the conditions change.
With the improvements of system imaging resolution and stability requirements, it
is necessary to explore the characteristics of unstable pixels [8]. This paper describes unstable pixels in HgCdTe linear arrays of LWIR based on an
n-on-p structure. According to the different fluctuation characteristics, unstable
pixels are classified into four types: trend-clear type (including the rising and
declining pixels), fluctuating type, comb type, and telegraph type. The types and
numbers of unstable pixels under different bias voltages are counted, and their low-frequency
characteristics are illuminated. In addition, the dark currents of unstable pixels
and normal pixels are also compared.
II. EXPERIMENTS
The HgCdTe materials were grown on CdZnTe substrate by LPE (Liquid Phase Epitaxy)
technology. The P-type layers of the devices are all doped with Hg vacancies. The
n-type layers were formed using B+ ion implantation, generating abundant Hg interstitials
[9]. The schematic diagrams of pixel structures are shown in Fig. 1. Sample 1 used ZnS/CdTe double-layer as passivation (without annealing process).
The cutoff wavelength (${\lambda}$c) was 10.2 $\mu m$ at the temperature of 77~K.
Sample 2 was processed with interdiffusion annealing to form a graded CdTe/HgCdTe
heterojunction at CdTe/HgCdTe interface [10]. The cutoff wavelength of Sample 2 was 9.5 $\mu m$ at 77 K. Each linear array had
512 pixels. The length and width of the pixels studied in this paper were 20*56 $\mu
m^{2}$. The chips were mounted into a liquid nitrogen Dewar. The focal plane test
setup was composed of a host computer and a Field-Programmable Gate Array (FPGA) acquisition
circuit [11,12]. The schematic diagram of the test setup is shown in Fig. 2. The pixel test should be based on a low background, so a filter should be added
to control the number of photons passed. The test conditions are displayed in μTable 1.
Fig. 1. Schematic diagram of the n-on-p MCT infrared detector pixels with different passivation (a) ZnS/CdTe double-layer passivation, (b) processed with interdiffusion annealing.
Fig. 2. Schematic diagram of the test setup used in this study.
Table 1 Test conditions
|
Test conditions
|
Sample 1
|
Sample 2
|
|
Integration
capacitance
|
2 pF
|
1 pF
|
|
Integration time
|
100 μs
|
100 μs
|
|
Bias voltages
|
2 mV-10 mV
（Division value: 2 ）
|
10 mV-100 mV
（Division value: 10）
|
|
Blackbody temperature
|
293 K
|
|
Filter band
|
8925-9275 nm
|
|
Flux
|
$4.35 \cdot 10^{16}\left(\text { photon } \cdot \mathrm{s}^{-1} \cdot \mathrm{cm}^{-2}\right)$
|
|
Acquisition frame number
|
50000
|
III. RESULTS AND DISCUSSION
1. Unstable Pixels Classification
For an MCT infrared FPA, the MCT detector array is coupled with a read-out integrated
circuit (ROIC) by indium bumps. The photocurrents of MCT photodiodes are injected
into the integration capacitors of the ROIC. By recording the voltage across the capacitors
within a specific integration time, the photo-response of each pixel can be obtained.
The voltage across the capacitor recording integration time is stable under the specific
luminous flux for a normal pixel, as shown in Fig. 3(a). However, the voltage will fluctuate when it comes to the unstable pixels, as shown
in Fig. 3(b)-(f), where the fluctuation characteristics of five unstable representative pixels are
displayed. It can be seen that the fluctuation characteristics of these unstable pixels
are different. According to the different characteristics, the unstable pixels are
classified into four types: the trend-clear type (including the rising and declining
pixels), the fluctuating type, the comb type, and the telegraph type. The voltage
of the trend-clear type pixel shows an increasing or a decreasing trend shown in Fig. 3(b) and (c). The voltage of the fluctuating type pixel displays slow volatility, as shown in
Fig. 3(d). The voltage of the comb type pixel shows rapid, irregular jitters, see Fig. 3(e). The voltage of the telegraph type pixel exhibits a random telegraph noise, as shown
in Fig. 3(f).
Fig. 3. Four types of unstable pixels (a) normal pixel, (b) rising type, (c) declining type, (d) fluctuating type, (e) comb type, (f) telegraph type.
2. Spectrum Characteristics
A very effective mathematical algorithm performs Fast Fourier transform on various
unstable pixels in the data processing stage. A given signal can determine the relationship
between its frequency and energy, i.e., noise power spectrum. The noise power spectrum
can be calculated by converting the (voltage-number of frames) via fast Fourier transform.
It can exhibit the relationship between the size and frequency of each sinusoidal
signal. The following formula can define the fast Fourier transform:
The noise power spectrum curves of the above four types of unstable pixels are shown
in Fig. 4 [13]. It can be found that the low-frequency noise of the unstable pixels is higher than
that of standard pixels, especially in the noise part of low frequency [14-17]. Thus, it is crucial to understand the nature of low-frequency noise to optimize
the performance of MCT infrared detectors at low frequencies and reduce the number
of unstable pixels.
Fig. 4. Power spectrum of different types of unstable pixels (a) rising type/declining type, (b) fluctuating type, (c) comb type, (d) telegraph type.
3. Influence of Passivation
3.1 Number and Types of Unstable Pixels
The two different passivated samples’ unstable pixels were counted under different
bias voltages. Each sample was tested twice to ensure accuracy. The numbers of unstable
pixels for the two samples under different bias voltages are shown in Fig. 5. Comparing the number of unstable pixels of the samples, it can be found that the
interdiffusion annealing effectively reduces unstable pixels, which indicates that
the unstable pixels in the MCT infrared detector are related to the device’s passivation.
The number of unstable pixels of each type under different bias voltages is counted
for Sample 1 and 2, as shown in Fig. 6. It can be found that there are various types of unstable pixels in sample 1, and
the number of fluctuating type unstable pixels increases steadily with the bias voltage
increasing. As a comparison, there is only one type of unstable pixels in sample 2,
namely, fluctuating type. Comparing the unstable pixels type of the two samples, it
can be found that the unstable pixels of rising/declining type, comb type, and telegraph
type are sensitive to surface passivation. The interdiffusion annealing of the CdTe/HgCdTe
structure can effectively reduce defects and the interface traps. Therefore, the number
of unstable pixels is reduced, and the device can operate at a higher bias voltage
[18].
Fig. 5. The number of unstable pixels under different biases of the two samples (a) sample 1, (b) sample 2.
Fig. 6. The number of unstable pixels of each type under different bias voltage (a) sample 1, (b) sample 2.
3.2 Bias Characteristics
It can be easily seen in Fig. 5 that the number of unstable pixels increases drastically under a large bias voltage.
For Sample 1 that was passivated using ZnS/CdTe double-layer film, the number of unstable
pixels suddenly increased from 0 to ~10 suddenly when the bias voltage increased from
4 to 6 mV. As a comparison, the number of unstable pixels of Sample 2 increased from
2 to ~6, with the bias voltage increasing from 70 to 80 mV.
In both Sample 1 and 2, a steep increase of unstable pixels can be observed when the
bias voltage is large enough. It can be inferred that the cause of unstable pixels
may be related to the defects in the depletion region of the PN junction [19]. As the bias voltage increases, the width of the depletion region gradually increases,
and more defects are contained in the depletion layer, which leads to the instability
of pixels. Fig. 7(a) and (c) show the width of the depletion region of the two samples under different
bias voltages, and the number of unstable pixels under different depletion region
widths is shown in Fig. 7(b) and (d). It can be seen that the number of unstable pixels shows an upward trend as the width
of the depletion region increases.
Fig. 7. The width of depletion region versus bias voltage and the number of unstable pixels varies with the depletion region width of (a)-(b) sample 1, (c)-(d) sample 2.
Because the test bias voltage range in sample 2 is more comprehensive (the test conditions
are shown in μTable 1). It is easier to observe the characteristics, so the unstable pixels that appear
often are extracted, as shown in Fig. 8. Fig. 8 shows the voltage level of the integration capacitor of an unstable pixel records
at different bias voltages. It can be clearly seen in Fig. 8(b) that the voltage level of the pixel under high bias voltage is unstable, but it is
very stable under low bias voltages. It is again confirmed that the unstable characteristics
of the pixels are easier to appear under a large bias. The charge fluctuation theory
proposed by McWhorter points out that 1/f noise originates from the change of carrier
concentration caused by the exchange of trap electrons and bulk electrons [20]. From the experimental results, it can be inferred that the increase of the applied
bias voltage will affect the number of traps in the depletion region, thereby affecting
the 1/f noise and finally manifesting in unstable pixels.
Fig. 8. (a) The relationship between the voltage across the integration capacitor and the number of frames of an unstable pixel under different biases, (b) The detailed curves under large bias (the test conditions are shown in μTable 1).
4. Dark Current
Dark current is an essential parameter for infrared detectors. A dark current can
be measured using an integration capacitor for an infrared detector array coupled
with a readout circuit [11,12]. Eq. (2) displays the relationship between capacitor charge, voltage, and capacitance. Eq.
(3) is the definition of the current. According to the principle of infrared FPA imaging,
the pixel's current is collected by the integration capacitor within the integration
time. Thus, the dark current can be obtained by using Eq. (4).
The two samples' dark currents are calculated using Eqs. (2)-(4) [21,22], and the results are shown in Fig. 9. For each sample, four unstable pixels are selected as representative. As a comparison,
the average value of the FPA, which is the mean value of all pixels except the dead
ones and the unstable ones, is also shown in Fig. 9. By comparing the dark current of the unstable pixels and the standard pixels, it
can be found that the dark current of the unstable pixels are not significantly different
from that of the standard pixels. Therefore, we can infer that pixel instability factors
do not directly affect dark current.
Fig. 9. Dark current of the device with (a) ZnS/CdTe double-layer passivation, (b) interdiffusion annealing process.
IV. CONCLUSION
This article studies the unstable pixels of LWIR MCT linear arrays with different
passivation. Although the manifestations of the unstable pixels are various, it can
be found that they are all derived from low-frequency noise according to their power
spectral density. The instability characteristics are more likely to appear under
large bias voltage than the small ones, which indicates that the instability of the
unstable pixels may be related to the defect in the depletion region of the PN junction.
By comparing the characteristics of the unstable pixels of two linear arrays with
different passivation, it is found that the linear array passivated by ZnS/CdTe double-layer
has larger numbers of unstable pixels and more types-various unstable pixels, indicating
that the unstable pixels are sensitive to the surface passivation. Besides, there
is no apparent difference between the dark current of the unstable pixels and the
standard pixels, which means the origin of the unstable pixels is different from that
of the dark current.
ACKNOWLEDGMENTS
This work was supported by the Shanghai Sailing Program (20YF1456000) and Innovation
Special Fund from the Shanghai Institute of Technical Physics (CX-336). This research
was done at the Shanghai Institute of Technical Physics (SITP) and ShanghaiTech University.
My colleagues provided a lot of help in completing this work. For this, the author
expresses his sincere gratitude.
References
Martyniuk P., Antoszewski J., Martyniuk M., Faraone L., Rogalski A., 2014, New concepts
in infrared photodetector designs, Applied Physics Reviews, Vol. 1, No. 4
Norton P., 2006, Third-generation sensors for night vision, Opto-Electronics Review,
Vol. 14, No. 1, pp. 1-10
Hansen M. P., Vavilov V. P., Malchow D. S., Burleigh D. D., 2008, Overview of SWIR
detectors, cameras, applications, presented at the Thermosense XXX
Gravrand et al. O., 2016, HgCdTe Detectors for Space and Science Imaging: General
Issues and Latest Achievements, Journal of Electronic Materials, Vol. 45, No. 9, pp.
4532-4541
X. D. J. E. D. Wang , 2012, The Study of Self-Heating and Hot-Electron Effects for
AlGaN/GaN Double-Channel HEMTs, IEEE Transactions on, Vol. 59, No. 5, pp. 1393-1401
H. F.-Z. LI Xiong-Jun , LI Dong-Sheng , LI Li-Hua , HU Yan-Bo , KONG Jin-Cheng , ZHAO
Jun , ZHU Ying-Feng , ZHUANG Ji-Sheng , JI Rong-Bin , 2017, HgCdTe surface effect
based on gate-controlled diode device, Journal of Infrared and Millimeter Waves, Vol.
36, No. 3, pp. 295
Cui A., Sun C., Wang F., Ye Z. J. I. P., 2021., Electrical Properties of plasma-free
ultra-low-temperature ALD ZnS passivation on p-Type HgCdTe, Technology, pp. 103667
P. J. s. t. w. i. s. r. Mccullough , 2009, Radioactivity in HgCdTe devices: potential
source of
Bubulac L. O., Tennant W. E., 1987, Role of Hg in junction formation in ion‐implanted
HgCdTe, Applied Physics Letters, Vol. 51, No. 5, pp. 355-357
Mollard L., et al. , 2009, Planar p-on-n HgCdTe FPAs by Arsenic Ion Implantation,
Journal of Electronic Materials, Vol. 38, No. 8, pp. 1805-1813
G. L. J. S. S. Destefanis , 1991, HgCdTe infrared diode arrays, Vol. 6, No. 12c, pp.
c88
Rogalski A., 2020, HgCdTe photodetectors, in Mid-infrared Optoelectronics, pp. 235-335
Krapf D., Marinari E., Metzler R., Oshanin G., Xu X., Squarcini A., 2018, Power spectral
density of a single Brownian trajectory: what one can and cannot learn from it, New
Journal of Physics, Vol. 20, No. 2
Zhu L., Deng Z., Huang J., Guo H., Chen B. J. O. E., 2020., Low Frequency Noise-dark
Current Correlations in HgCdTe Infrared Photodetectors, Vol. 28, No. 16
Ciura Ł., Kopytko M., Martyniuk P., 2020., Low-frequency noise limitations of InAsSb-,
HgCdTe-based infrared detectors, Sensors and Actuators A: Physical, Vol. 305
Chen Y., et al. , 2020., Study on the noise characteristics of GaAs-based blocked-impurity-band
(BIB) detectors, Optical and Quantum Electronics, Vol. 52, No. 6
Bae S. H., Lee S. J., Kim Y. H., Lee H. C., Kim C. K., 2000., Analysis of 1/f noise
in LWIR HgCdTe photodiodes %J Journal of Electronic Materials, Vol. 29, No. 6
K. Zanio , T. J. J. o. E. M. Massopust , 1986, Interdiffusion in HgCdTe/CdTe Structures,
Vol. 15, No. 2, pp. 103-109
Yuan J. S., Liou J. J., 1998, Semiconductor device physics and simulation.
Philadelphia , R. H. J. U. o. P. P. Kingston , Semiconductor surface physics.
Baker R. J., 1997, CMOS Circuit Design, Layout, Simulation.
Chang B., et al. , 2013, ROIC for HgCdTe e-APD FPA, presented at the International
Symposium on Photoelectronic Detection and Imaging 2013: Low-Light-Level Technology
and Applications
Yu Zhang is currently studying for a master's degree in electronics science and
technology at ShanghaiTech University, under the tutelage of Chun Lin from the Shanghai
Institute of Technology, Chinese Academy of Sciences. His current research focuses
on the formation mechanism of unstable pixels of the long-wave mercury cadmium telluride
infrared detector.
Songmin Zhou received a master's degree in condensed matter physics from the Department
of Physics, School of Science, Shanghai University, in 2010. In 2010, he entered the
Institute of Technology and engaged in the research of mercury cadmium telluride infrared
focal plane device technology. He has been published in many journals.
Xun Li received the B.S. degree in the School of Materials Science and Engineering
from Nanchang Univer-sity, Nanchang, China, in 2014, and the Ph.D. degree in Hefei
National Laboratory for Physical Sciences at the Microscale, University of Science
and Technology of China, Hefei, China, in 2019. In 2019, he joined the Shanghai Institute
of Technical Physics, the Chinese Academy of Sciences. His current research interest
focuses on high-performance HgCdTe infrared detectors.
Xi Wang was born in 1988. He obtained his Ph.D. from the University of Chinese
Academy of Sciences, mainly engaged in mercury cadmium telluride infrared focal plane
devices, and has in-depth research on the dark current suppression of mercury cadmium
telluride infrared focal plane devices.
Liqi Zhu received a B.S. degree in microelectronics science and engineering from
Qingdao University, China, in 2018. He is working toward a Ph.D. degree in microelectronics
and solid state electronics with ShanghaiTech University, China and Shanghai Institute
of Technical Physics, Chinese Academy of Sciences. His research interest focuses on
the simulation and characterization of low noise infrared avalanche detectors and
high bandwidth infrared photodioes.
Chun Lin, born in February 1973, holds a doctorate. In 2001, he entered the Walter
Schottky Institute of Technical University of Munich, Germany, to engage in optoelec-tronics
research. In 2005, he joined the Shanghai Institute of Technical Physics, the Chinese
Academy of Sciences. The current researcher and doctoral supervisor of the institute.
He mainly researches semiconductor infrared optoelectronic device preparation and
testing technology. His research work includes mercury cadmium telluride long-wave
infrared focal plane detector, quantum well focal plane detector, mid-infrared antimonide
quantum well laser, and quantum cascade laser. More than 20 papers have been published
in international academic journals.