Open Access
Add to BibSonomy
Abstract
We report results from characterizing the HgCdTe avalanche photodiode (APD) arrays developed for lidar at infrared wavelengths by using the high density vertically integrated photodiodes (HDVIP®) technique. The results show >90% quantum efficiencies between 0.8 μm and the cut-off wavelength, >600 APD gain, near unity excess noise factor, 6-10 MHz electrical bandwidth and <0.5 fW/Hz1/2 noise equivalent power (NEP). The detectors provide linear analog output with a dynamic range of 2-3 orders of magnitude at a fixed APD gain without averaging, and over 5 orders of magnitude by adjusting the APD gain settings. They have been used successfully in airborne CO2 and CH4 integrated path differential absorption (IPDA) lidar as precursors for use in space lidar.
© 2017 Optical Society of America
1. Introduction
Lidar (light detection and ranging) is an important remote sensing technique for Earth and planetary science investigations and for environmental monitoring [1,2]. Detectors play a key role in space-based lidar, especially for those using direct detection. The sensitivity of the detector usually dictates the laser transmitter power and receiver telescope diameter, which are usually major factors in instrument size and cost. The availability of a suitable detector can be a deciding factor in the choice of lidar wavelengths. Most space lidar developed to date operate in visible or near infrared wavelengths where photomultiplier tubes and silicon avalanche photodiodes have high sensitivity. However, many science measurements need to be carried out in the near- or mid -infrared wavelength regions because of the spectral features of targeted surfaces or gas species and lower solar background illumination.
There has been a lack of sensitive detectors suitable for lidar that operate in the near to mid infrared wavelengths. Unlike imaging sensors that integrate photons from the target and read them out frame-by-frame, lidar detectors need fast response time to provide range-resolved measurements. Ideally lidar detectors should be operated in the linear mode and have negligible dark noise. For best receiver sensitivity they need internal photoelectron multiplication gain to override electronics noise from the preamplifier or the read-out integrated circuit (ROIC). For most lidar applications it is important to have no dead-time and no after-pulsing and the capability to operate continuously without gating.
The development of HgCdTe avalanche photodiodes (APDs) was a significant advance in detector technology for the short-wave infrared (SWIR) to mid-wave infrared (MWIR) wavelength region. HgCdTe photodiodes have high quantum efficiency and low dark current at low temperature. The optimal operating temperature depends on the band-gap or the cut-off wavelength of the HgCdTe material and it is 77-120 K for the devices with a 4.3-μm cut-off wavelength discussed in this paper. Recently, the photoelectron multiplication gain has reached 500 or higher, more than sufficient to overcome the electronic noise of the ROIC, while still maintaining low dark current and near unity excess noise factor. HgCdTe APDs operate below the breakdown voltage and provide linear analog output. They have a wide dynamic range and no dead-time and no afterpulsing. More information about the basic physics of the HgCdTe APD can be found in [3,4]. The first published papers describing the near unity and gain-independent excess noise factor properties of HgCdTe APDs were published in 2001 and followed up in 2006 [5,6]. Further contributions were made in the areas of gain and dark current properties [7], Monte Carlo modeling [8], bandwidth [9], signal and noise modeling [10]. HgCdTe APDs have been shown capable of linear mode photon counting operation in SWIR and MWIR [11–14]. A review of more recent HgCdTe APD development with applications can be found in [15].
One HgCdTe APD structure is the high-density vertically-integrated- photodiode (HDVIP®) developed by DRS Technologies [5,6]. NASA Goddard Space Flight Center (GSFC) has been collaborating with DRS Technologies to develop HgCdTe APD arrays for the candidate CO2 integrated path differential absorption (IPDA) lidar operating at either 1.57 or 2.05-μm wavelengths for the Active Sensing of CO2 Emission of Days, Nights, and Seasons (ASCENDS) mission planned by NASA [16,17]. Some of these HgCdTe APD’s have been packaged in closed-cycle cryo-coolers and have demonstrated nearly quantum-limited receiver sensitivity with several orders of magnitude dynamic range. One detector system has been successfully used in the airborne CO2 Sounder lidar for flights in 2014 and 2016 [18–20]. The new detectors also have enabled NASA GSFC to greatly improve the performance of an airborne methane (CH4) lidar operating at 1.65 μm for environmental monitoring [21] as well as to demonstrate a multiple-wavelength laser absorption spectrometer operating from 2.6 to 3.5 μm that may be a candidate to map volatiles on the surface of the Moon [22].
In this paper, we give a brief overview of the HDVIP® HgCdTe APD detectors and report the results from characterizing the 4x4 pixel APD array detector systems at NASA GSFC.
2. HDVIP® HgCdTe APD Sensor Chip Assembly
The HDVIP® HgCdTe APD is a front-side illuminated, cylindrical, p-around-n photodiode formed around a small via in the HgCdTe [3,5]. The cut-off wavelength of the APD devices has been tailored from 2.2 to 9.7 um [4–6]. Figure 1 shows a schematic of a HDVIP® HgCdTe APD. Figure 2 shows a typical spectral response of the 4.3-μm cut-off devices. The via also serves as the interconnect conduit between the n-side of the photodiode and the input to the ROIC. Electron and hole pairs excited by incoming photons in the p-region undergo photo-electron multiplication with electrons traveling through the p-n junction to initiate the avalanche process. Benefits of the HDVIP® structure are: (1) interdiffused CdTe passivation of both surfaces of the HgCdTe film; (2) thermal cycle reliability that is array size independent; (3) low defects due to diode junction orientation with respect to threading dislocations; and (4) front side illumination for high quantum efficiency from the visible to the infrared cutoff wavelength determined by the material.
Fig. 1 Schematic of the HDVIP® HgCdTe APD, left: side view; right; top view (Conceptual drawing only, the layer thickness and the via size not to scale).
DRS designed and fabricated a set of 4x4 pixel HgCdTe e-APD detector arrays for the CO2 IPDA lidar, in work supported by NASA’s Earth Science Technology Office (ESTO) Instrument Incubator Program (IIP) [16]. The detector’s pixel pitch was 80 μm with no gaps between pixels. There is also a guard ring around the 4x4 pixel area, which consists of a single row or column of reversely biased pixels. The purpose of the guard ring is to ensure the pixels along edges of the 4x4 array are surrounded by the same devices so that all 16 pixels have uniform response. Each pixel consists of 4 identical HgCdTe APD’s connected in parallel. The photo sensitive area is the entire diode surface except for the void created by the via. However, only photo-electrons created outside the p-n junction area are multiplied by the full avalanche gain while those created in the multiplication region do not. Thus the optical fill factor, defined as the ratio of the photosensitive area to pixel area, changes with bias voltage or APD gain. At near unity APD gain, the fill factor is about 96% given the physical size of the via (6-μm diameter) plus the p-n junction area around the via (1-2 μm thickness). The fill factor becomes about 75% near the maximum APD gain as the n-region becomes optically inactive. All 16-pixel outputs are fanned out to the side of APD array for connection with the ROIC, which consists of 16 linear amplifiers, one for each pixel. The HgCdTe APD and the ROIC are mounted next to each other on a 64-pin leadless chip carrier (LCC) and connected with bond wires to form a sensor chip assembly (SCA), as shown in Fig. 3.
Fig. 3 Schematic and photograph of a 4x4 pixel HgCdTe APD sensor chip assembly (SCA) developed for NASA GSFC’s CO2 Sounder lidar.
The ROIC was developed by A/DIC Inc. and was designed to operate at cryogenic temperatures. Figure 4 shows a block diagram of the circuit. It was fabricated using Jazz Semiconductor foundry service. Each ROIC die consists of 16 transimpedance amplifiers (TIA), also known as resistive TIA (RTIA), one for each pixel. There are three feedback resistors that can be switched in and out. The maximum transimpedance is 320,000 V/A. The RTIA gain can be reduced by a factor of 2, 3, 4, 5, 6, and 7, depending on the combination of the feedback resistors selected. The switching and settling time between different RTIA gains is about 80 ns. When all three feedback resistors are switched open, the ROIC functions as a charge amplifier, or a capacitive TIA (CTIA), in combination with a sample-and-hold circuit and a reset switch. The CTIA gain is determined by the feedback capacitance, which is 0.063 pF or 2.5 μV/electron based on the ROIC layout. Both the RTIA and the CTIA gains were verified to within the measurement uncertainty during the ROIC testing at A/DIC. The electrical bandwidth at maximum transimpedance was chosen to match the 50-ns rise and fall times of the lasers used in the NASA GSFC CO2 Sounder lidar [18–20]. The ROIC also contains a digital control circuit to select the RTIA gain and to switch between RTIA and CTIA modes. The total electrical power dissipation of the ROIC chip is about 60 mW.
For the receivers of GSFC’s airborne CO2 and CH4 IPDA lidar, the HgCdTe APD SCAs are housed in a closed-cycle Dewar-cooler assembly (SE-IR DWR-CC with a L3-Com B1500E6-L6 1.5 W cooler and controller). The detector temperature can be varied but is usually set to 77 K. A cold filter is used inside the Dewar to block the ambient light and the thermal emission. The maximum numerical aperture for the incident optical signal is f/1.1 and can be adjusted by the aperture size of the cold shield. A separate control electronics assembly is supplied by DRS Technologies to power the ROIC, generate the bias voltage for the APD, control the ROIC operation mode, and set the RTIA gain. There is a set of 16 buffer amplifiers immediately outside the Dewar to drive 50Ω coax cables.
For space applications, the HgCdTe APD SCA can be mounted on a smaller ceramic chip carrier in an integrated detector-cooler assembly (IDCA) with a micro cryo-cooler, such as the 1/5-watt linear drive Stirling cooler by DRS Technologies [23]. DRS Technologies has successfully integrated similar HgCdTe APD arrays in such a cooler for a CubeSat demonstration [24]. This IDCA has passed the vibration, thermal shock, and thermal cycle tests per NASA’s General Environmental Verification Standard GSFC-STD-7000.
3. Performance Measurements of the 4x4 Pixel HgCdTe APD SCA
Several 4x4 pixel HgCdTe APD detector systems have been delivered to NASA GSFC. The basic functions and performance of these devices were verified at DRS Technologies [16]. NASA GSFC characterized the performance of these devices prior to the integration with the airborne lidar and the results are reported here. Most of the tests were performed using an optical-fiber-coupled continuous wave (CW) 1.55 μm laser diode modulated by an acoustic optical modulator and attenuated by a calibrated optical fiber attenuator. All measurements were taken with the HgCdTe APDs operating at 77 K, except for one set of measurements where the APD gain and dark noise were measured as a function of the temperature. All the measurements results reported here were from a single pixel, mostly Pixel 16. All other pixels were also measured and comparisons of different pixel outputs are given as bar graphs for some of the key device parameters, such as the APD gain.
3.1 Quantum efficiency
The quantum efficiency (QE) of the HgCdTe APD assemblies was measured from the detector responsivity at unity APD gain under a known incident optical power. Due to the device geometry and material properties [3, Ch.7.3] the APD approaches unity gain and the responsivity becomes nearly constant as the APD bias decreases to below 1 V. We also measured the device responsivity and confirmed it stayed nearly unchanged from 0.0 to 0.7 V APD bias. We then took the responsivity at 0.5 V APD bias as the unity gain responsivity in our QE measurements. Figure 5 shows the measurement results. The CTIA mode of the ROIC was used which gave a direct measurement of the detector response in A/W. The laser test source was collimated to flood illuminate all the 4x4 pixels and the laser power density in W/cm2 was measured with an optical power meter behind a known size pinhole at the same distance from the source as the detector under test. The optical power on each pixel was determined by multiplying the laser power density by the pixel size (80x80μm), the fill factor, and the throughput of the attenuator. The optical transmission of the Dewar window and the cold filter were also considered in the signal calculations. The detector fill factor was taken to be 96% at this low APD bias. The measurement uncertainty is estimated to be within ± 5% due to uncertainties in the optical power calibration and other factors. The average measured QE of all 16 pixels was 92% at 1.55 μm wavelength and varied within ± 10% among all the pixels.
Fig. 5 Quantum efficiency of all 16 pixels of the HVDIP HgCdTe APD SCA (Serial Number A8052-16B) at 1.55 μm wavelength based on the responsivity measurements in CTIA mode at 0.5 V APD bias.
We also measured the QE at 1.064, 1.573, and 2.050 μm using laser diodes and a blackbody radiation source followed by a narrow bandpass filter. The results showed the QE was about 90% at these wavelengths. The measured QE is affected by the anti-reflection (AR) coating, which can be tailored to maximize the detector responsivity at a given wavelength.
3.2 RTIA Responsivity and APD gain
The responsivity of the detector assembly in RTIA mode was measured as the ratio of the average detector output pulse amplitude to the incident optical signal power at a given APD bias voltage. Figure 6 shows the measurement results for a few pixels from one device vs. the APD bias voltage at the maximum RTIA gain setting (320,000 V/A). For these tests the laser diode illuminating the detector was modulated with 1-μs wide rectangular pulses. The ROIC output offset voltage was subtracted out to obtain the net pulse amplitude.
Fig. 6 Responsivity of the HgCdTe APD SCA (A8052-16B) vs. bias voltage at 1.55 μm wavelength. The buffer amplifier gain was 10.7 V/V for this detector assembly.
The APD gain was calculated as the ratio of the detector assembly output at a given APD bias voltage to that at 0.5 V bias. Figure 7 shows the measurement results with the ROIC in both RTIA and CTIA modes (left) and the gain versus the pixel number at biases of 11 and 12 V (right). Above 6 V bias, the APD gain increases with the bias as an exponential function, roughly doubling for every 1 V increase in the APD bias, and reaching about 900 at an APD bias of 12 V. At high APD gains the n multiplication region is less sensitive since photo-carriers generated in that region do not achieve full gain. Thus the optical fill factor, which is determined by the thickness of the n region and the diameter of the via, is lower at high gains, as mentioned previously. With a reduced acceptance angle for incident light, it is possible to use a microlens array on top of the APD array to concentrate the incident light onto the center region of the pixel and achieve a 100% optical fill factor that is independent of APD gain [25]. DRS Technologies has successfully integrated a microlens array on a similar HgCdTe APD array [24].
Fig. 7 APD gain of the HVDIP HgCdTe APD SCA (A8052-16B). Left: APD gain of Pixel 16 vs. bias voltage with both CTIA and RTIA measurement; right: APD gain for all the pixels at 11 and 12 V APD bias voltage.
3.3 Surface scan of the HgCdTe APD active area
Figure 8 shows a surface map of the responsivity from one center pixel of an HgCdTe APD at 11 V bias. The test laser light in this case was focused onto the detector using a long working distance objective lens that produced a laser spot size of about 5 μm in diameter on the detector surface. A translation stage with a set of stepping motors was used to move laser assembly in both the horizontal and vertical position at 1-μm step size. The small horizon ridges and groves shown in Fig. 8 were caused by intermittent noise in the test equipment over the 3-4 hour measurement period. The response of the pixel spilled over to areas outside the 80x80 μm area, due to the laser spot size, spurious spatial modes due to imperfection in the optics setup, and inherent cross-talk between the pixels.
Fig. 8 Normalized HgCdTe APD response vs. light spot position across one center pixel (A8052-13E) at 11 V APD bias. The laser spot size was about 5 μm in diameter and it was moved at 1-μm step size in a raster scan pattern across the APD active area.
3.4 Dark noise
The sources of detector “dark noise” are the APD dark current, the preamplifier (ROIC) electronics noise as well as dark current from photons that leak in through the cold shield. The APD dark current consists of the surface leakage current, that is not multiplied by the APD gain, the HgCdTe tunnel current in the multiplication region that is only partially multiplied by the APD gain, and the bulk dark current that is multiplied by the APD gain. We measured the total APD dark current using the CTIA mode with the APD bias set to 0.5 V (unity APD gain) and the results are shown in Fig. 9. The average dark current was 0.35 pA, or 2.2x106 electron/s. Most of it is believed to come from the infrared light leaking around the cold shield and possibly from photons emitted by the ROIC.
Fig. 9 Dark current of the HgCdTe APD (A8052-16B) at 0.5 V bias voltage measured (unity APD gain) measured with the ROIC in the CTIA mode.
The total output dark noise was measured with a radio frequency spectrum analyzer with the RTIA gain set to 320,000 V/A and the buffer amplifier gain of 10.7 V/V. The noise equivalent power (NEP) was taken as the root-mean-square (rms) of the output noise voltage spectral density in V/Hz1/2 divided by the responsivity in (V/W). We chose to evaluate the NEP at 3 MHz, which was about the center of the detector electrical bandwidth. Figure 10 shows the NEP vs. APD bias voltage for Pixel 16 and NEP for all pixels at APD biases of 11 and 12 V, respectively. It shows the NEP reached the minimum at APD bias of 11.5 V for this particular device.
Fig. 10 Left: NEP of the HgCdTe APD (A8052-16B, Pixel 16); Right: NEP for all pixels at 11 and 12 V APD biases. The RTIA gain was set to 320 kV/A and the buffer amplifier gain was 10.7 V/V.
3.5 Dark noise vs. temperature
The APD bulk dark current was measured as a function of the device temperature from 77 to 130 K using the CTIA mode at an APD bias of 0.5 and 11 V. The APD gain at 11 V bias was measured separately at each temperature from the ratio of the responsivity at 11 V APD bias to that at 0.5 V bias, which varied at a rate of about −0.87% per K. Figure 11 shows the gain-normalized APD dark current as a function of device temperature. It shows the optimal device operating temperature was about 110 K.
Fig. 11 Gain-normalized dark current of the HgCdTe APD (A8052-16B, pixel 16) at 0.5 and 11 V APD bias voltages divided by the APD gain.
3.6 Electrical bandwidth and responses to laser pulses
The electrical bandwidth of the HgCdTe APD with the ROIC was measured from the output noise spectrum while illuminating the APD with a relatively strong CW light so that the noise power spectrum was well above the dark noise, as shown in Fig. 12. We then determined the electrical bandwidth as the frequency at which the noise spectrum fell by a factor of 2 (3-dB). The bandwidth was 7.0 MHz for this device at this RTIA gain setting. The bandwidth varied among devices between 6 and 7 MHz. The bandwidth of the ROIC alone was 8.6 MHz based on independent measurement at A/DIC. We also measured electrical bandwidth and NEP as a function of the RTIA gain setting, as shown in Fig. 13. The bandwidth measurements had a relatively large uncertainty since the noise spectra rolled off gradually and it was difficult to determine the 3-dB points in the presence of noise. A linear fit to the bandwidth was given in Fig. 13 to show the general trend. The NEP decreased as the RTIA gain increased, reaching its minimum value at the highest RTIA gain setting.
Fig. 12 Noise spectral density of the HgCdTe APD (A8052-16B, pixel 16) under 0 and 11 V APD biases at dark and under a relatively strong CW illumination to determine the electrical bandwidth.
Fig. 13 NEP and bandwidth of the HgCdTe APD (A8052-4E, pixel 16) vs. the RTIA gain at 11 V APD bias. Note this is from a different device with a narrower bandwidth than that shown in Fig. 12.
The impulse response of the detector at maximum RTIA gain was measured at various APD biases by illuminating the detector with a 70-ps pulse width laser. Figure 14 shows the normalized impulse response waveform at 11 V APD bias. The signal was DC coupled and there was little ripple (ringing) after the pulse. The maximum amplitude of the ringing was <1% 150 ns after the pulse and <0.1% 1 μs after the pulse.
Fig. 14 Normalized impulse response of the HgCdTe APD (A8052-16B, pixel 16) at 11 V APD bias and maximum RTIA gain.
3.7 APD excess noise factor
The APD excess noise factor is defined as the ratio of the square of the mean to the mean square of the APD gain. It gives a measure of receiver signal to noise ratio (SNR) degradation from the randomness of the APD gain. We estimated the APD excess noise from the SNR of the detector response as follows. The SNR of the integrated APD response (i.e. pulse area) to a laser pulse is given by [26],
where and are the mean and the standard deviation of the net signal, is the APD quantum efficiency, is the fill factor, is the average APD gain, is the average number of photons per laser pulse incident onto the APD, is the standard deviation of the baseline noise, and is the excess noise factor. The variables , and can be directly estimated from the pulse waveforms output from the APD. The excess noise factor can be solved from the above equation, asWe measured the excess noise factor by illuminating the detector with both a 70-ps pulse width laser and a 1-μs rectangular pulse laser and recording the pulse waveforms with a 16-bit 120-Ms/s analog to digital converter (ADC). Figure 15 shows the measurement results. The average offset was estimated from the recorded waveform data prior to the laser pulse and subtracted before the pulse area calculations. The QE of the detector was assumed to be 90% and the fill factor was assumed to be 75%.
Fig. 15 Measured HgCdTe APD (A8052-13E, pixel 16) excess noise factor vs. number of incident photons per pulse in response to a 70-ps and a 1-μs pulse width lasers.
There was a relatively large uncertainty in the excess noise factor measurement because it involved subtraction of two similar numbers at low signal level and the signal amplitude approaching saturation level of the ADC at high signal level. It was not clear to us why the measured excess noise factors tended to be higher at high incident laser photon flux (photons/pulse divided by the pulse width) for both incident laser pulse widths. We repeated the tests using an oscilloscope as the ADC and the results were similar. Nevertheless, the measured excess noise factor was in general close to unity for low to mid signal level where low excess noise is important for lidar applications.
3.8 Linear dynamic range
The HgCdTe APD dynamic range was measured by using a 1-μs wide rectangular laser pulses. The average output pulse amplitudes were recorded as a function of the incident photons/pulse at different APD biases and the results are shown in Fig. 16. The maximum linear output is about 1.1 V. The standard deviation of the detector output with dark noise only was measured to be 2.0 mV at 11 V APD bias, which was consistent with the calculated noise from the NEP, bandwidth, and the responsivity. If we consider the minimum detectable signal to be equal to the standard deviation of the dark noise, the linear dynamic range at 12 VAPD bias was 550. In applications that need to average the received signals to achieve the required SNR, the detector dark noise is further reduced by the square roof of the number of averages, which gives an even larger dynamic range. For example, we average hundreds of pulse waveforms before recording a measurement in our CO2 IPDA lidar. This reduces the standard deviation of the noise by at least a factor of 10, resulting in a dynamic range of >5,500 at a fixed APD and RTIA gain setting. The APD gain can be adjusted by a factor of about three hundred and the RTIA gain by a factor of 7 to give a combined linear dynamic range of over 5 orders of magnitude without waveform averaging.
Fig. 16 Dynamic range of the HgCdTe APD (A8052-16B). The APD output pulse amplitude vs. the incident number of photons per pulse at various APD bias settings (left); and the output pulse amplitude divided by the APD gain vs. the incident photons/pulse (right).
We performed a linear fit to the output pulse amplitude shown above excluding near saturated points with >1.2 V amplitudes. The coefficients of correlation were R2 = 0.99998 for the data at 12 V APD bias and R2 = 0.99971 for the data at 11 V APD bias. The test results from the ROIC alone was R2 = 0.99999 at A/DIC by injecting calibrated electrical signal.
3.9 Pulsed signal to noise ratio
We measured the SNR of the detector output in response to both a 70-ps laser pulses and 1-μs wide rectangular laser pulses at 1.55-μm wavelength. A 16-bit 120-Ms/s ADC was used to collect the pulse waveforms and the quantization noise of the ADC was negligible. We then calculated pulse area by integrating the pulse waveform, as we do in our IPDA CO2 lidar, and determine SNR as the ratio the mean to the standard deviation of thousands of repeated pulse area measurements. Figure 17 shows the SNR of the pulse area measurement as a function of the APD bias voltage. The results showed the SNR first increased with the APD bias as the signal rises relative to the fixed ROIC noise. Above a certain APD bias, the SNR became signal shot-noise limited and SNR reached a plateau slightly under the quantum limit. Unlike in silicon and InGaAs APDs, further increasing the APD gain did not cause additional excess noise and degradation of the SNR. The APD gain of the HgCdTe APD was sufficiently high to allow nearly quantum-limited detection for weak optical signals.
Fig. 17 SNR of integrated pulse area from the HgCdTe APD (A8052-16B) as a function of the APD biases at different incident signal levels in response to a 70-ps short pulse laser (left) and a 1-μs rectangular pulse laser (right). Note the signal calibration for the 70-ps laser pulses was difficult and consequently less accurate than that for the 1-μs pulse laser.
3.10 Applications of the HgCdTe APD in airborne lidar
The HgCdTe APD detectors have recently been used in our airborne CO2 and CH4 IPDA lidar [18–21]. The entire detector system, including the power supplies, was packaged into a standard 7-U rack-mounted chassis (48.3 x30.5 x50.8 cm). The optical signal from the receiver telescope was coupled via a multimode optical fiber with a 400-μm core diameter and 0.22 numerical aperture and a focusing lens assembly attached to the Dewar. The optical signal illuminated a 3x3 pixel area and the outputs from these 9 pixels were summed together before being amplified and recorded by a 16-bit ADC. The relatively large optical spot on the HgCdTe APD array averaged out the small voids created by the vias. The lidar performance during the airborne campaigns agreed well with the model prediction [20]. There was no measurable disturbance in the received signal from the small vibration of the cryo-cooler. The detector’s linear dynamic range was sufficient to accommodate return signals from a wide range of natural surfaces for aircraft altitudes from 5 to 12 km without having to adjust the APD or the RTIA gain.
4. Summary
A set of HgCdTe APD detector assemblies developed by DRS Technologies for space lidar at SWIR to MWIR wavelengths were characterized. The results showed >90% QE over a range of laser wavelengths, >600 APD gain, near unity excess noise factor, 6-7 MHz electrical bandwidth, and <0.5 fW/Hz1/2 NEP. The APD gain was sufficiently high to achieve shot-noise limited lidar receiver performance at low signal levels. The detectors had a linear analog output with a dynamic range of 550 without APD and RTIA gain adjustment or signal averaging. The linear dynamic range was over 5 orders of magnitude when the APD and RTIA gains were adjusted. The detector assemblies have been successfully used in the receivers of GSFC’s CO2 and CH4 IPDA lidars in airborne campaigns. These detectors have filled an important need for high sensitivity lidar detectors in SWIR and MWIR wavelengths.
Funding
NASA Earth Science Technology Office (ESTO) Instrument Incubator Program (IIP) 2010-2014; Advanced Component Technology (ACT) program 2011-2014; NASA GSFC Internal Research And Research (IRAD) program 2008, 2009, 2011, 2015, and 2016.
Acknowledgments
We thank C. Kamilar, J. McCurdy, and P. Benken of DRS Technologies for the electronics design, detector system testing and device fabrication. We also thank W. Hasselbrack, G. Allan, and W. Lu of NASA GSFC for assistance in the device packaging and characterizations.
References and links
1. X. Sun, J. B. Abshire, J. F. McGarry, G. A. Neumann, J. F. Cavanaugh, J. C. Smith, D. J. Harding, H. J. Zwally, D. E. Smith, and M. T. Zuber, “Space lidar developed at NASA Goddard Space Flight Center – The first 20 years,” IEEE J. Sel. Topics Appl. Earth Observ. In Remote Sens. 6(3), 1660–1675 (2013).
2. M. P. McCormick, “Airborne and spaceborne lidar,” in Lidar: Range Resolved Optical Remote Sensing of the Atmosphere, Ch. 13, C. Weitkamp ed. (Springer Science + Business Media, 2005).
3. M. A. Kinch, Fundamental of Infrared Detector Materials (SPIE, 2007).
4. M. A. Kinch, State-of-the-Art Infrared Detector Technology (SPIE, 2014).
5. J. D. Beck, C. F. Wan, M. A. Kinch, and J. E. Robinson, “MWIR HgCdTe avalanche photodiodes,” Proc. SPIE 4454, 188–197 (2001). [CrossRef]
6. J. Beck, C. Wan, M. Kinch, J. Robinson, P. Mitra, R. Scritchfield, F. Ma, and J. Campbell, “The HgCdTe electron avalanche photodiode,” J. Electron. Mater. 35(6), 1166–1173 (2006). [CrossRef]
7. G. Perrais, O. Gravrand, J. Baylet, G. Destefanis, and J. Rothman, “Gain and dark current characteristics of planar HgCdTe avalanche photodiodes,” J. Electron. Mater. 36(8), 963–970 (2007). [CrossRef]
8. S. Derelle, S. Bernhardt, R. Haïdar, J. Primot, J. Deschamps, and J. Rothman, “A Monte Carlo study of Hg0.7Cd0.3Te e-APD,” IEEE Trans. Electron Dev. 56(4), 569–577 (2009). [CrossRef]
9. G. Perrais, S. Derelle, L. Mollard, J.-P. Chamonal, G. Destefanis, G. Vincent, S. Bernhardt, and J. Rothman, “Study of the transit-time limitations of the impulse response in mid-wave infrared HgCdTe avalanche photodiodes,” J. Electron. Mater. 38(8), 1790–1799 (2009). [CrossRef]
10. J. Rothman, L. Mollard, S. Goût, L. Bonnefond, and J. Wlassow, “History dependent impact ionization theories applied to HgCdTe e-APDs,” J. Electron. Mater. 40(8), 1757–1768 (2011). [CrossRef]
11. J. D. Beck, R. Scritchfield, P. Mitra, W. W. Sullivan III, A. D. Gleckler, R. Strittmatter, and R. J. Martin, “Linear mode photon counting with the noiseless gain HgCdTe e-avalanche photodiode,” Opt. Eng. 53(8), 081905 (2014). [CrossRef]
12. J. D. Beck, M. Kinch, and X. Sun, “Update on linear mode photon counting with the HgCdTe linear mode avalanche photodiode,” Opt. Eng. 53(8), 081906 (2014). [CrossRef]
13. G. Vojetta, F. Guellec, L. Mathieu, K. Foubert, P. Feautrier, and J. Rothman, “Linear photon-counting with HgCdTe APDs,” Proc. SPIE 8375, 83750Y (2011). [CrossRef]
14. W. Sullivan, J. Beck, R. Scritchfield, M. Skokan, P. Mitra, X. Sun, J. Abshire, D. Carpenter, and B. Lane, “Linear-mode HgCdTe avalanche photodiodes for photon-counting applications,” J. Electron. Mater. 44(9), 3092–3101 (2015). [CrossRef]
15. I. Baker, C. Maxey, L. Hipwood, and K. Barnes, “Leonardo (formerly Selex ES) infrared sensors for astronomy – present and future,” Proc. SPIE 9915, 991505 (2016). [CrossRef]
16. J. Beck, T. Welch, P. Mitra, K. Reiff, X. Sun, and J. Abshire, “A highly sensitive multi-element HgCdTe e-APD Detector for IPDA lidar applications,” J. Electron. Mater. 43(8), 2970–2977 (2014). [CrossRef]
17. X. Sun, J. B. Abshire, and J. D. Beck, “HgCdTe e-APD detector arrays with single photon sensitivity for space lidar applications,” Proc. SPIE 9114, 91140K (2014). [CrossRef]
18. J. B. Abshire, A. Ramanathan, H. Riris, J. Mao, G. R. Allan, W. E. Hasselbrack, C. J. Weaver, and E. V. Browell, “Airborne measurements of CO2 column concentration and range using a pulsed direct-detection IPDA lidar,” Remote Sens. 6(1), 443–469 (2013). [CrossRef]
19. G. R. Allan, H. Riris, J. B. Abshire, M. A. Stephen, ramanathan, chen, W. Hasselbrack, X. Sun, K. Numata, and S. Wu, “CO2 Sounder lidar development at NASA-GSFC for the ASCENDS mission,” in Conference on Lasers and Electro-Optics (CLEO), OSA Technical Digest (Optical Society of America, 2016), paper STh1H.3. [CrossRef]
20. J. B. Abshire, NASA Goddard Space Flight Center, Greenbelt, MD 20771, is preparing a manuscript to be called “Airborne measurements of XCO2 made with a pulsed, multi-wavelength-locked laser and HgCdTe detector.”
21. H. Riris, NASA Goddard Space Flight Center, Greenbelt, MD 20771, is preparing a manuscript to be called “Methane optical density measurements with an integrated path differential absorption lidar from an airborne platform,” Appl. Remote Sens. (submitted to).
22. P. G. Lucey, X. Sun, J. B. Abshire, and G. A. Neumann, “An orbital lidar spectrometer for lunar polar compositions,” 45th Lunar and Planetary Science Conference (2014), paper 2335.
23. D. Rawlings and G. Averitt, “A linear drive cryocooler for ultra-small infrared sensor systems,” Proc. SPIE 9070, 90702R (2014). [CrossRef]
24. R. Fields, X. Sun, J. Abshire, J. Beck, R. M. Rawlings, and D. Hinkley, “A linear mode photon-counting (LMPC) detector array in a CubeSat to enable earth science LIDAR measurements,” in International Geoscience And Remote Sensing Symposium (IGARSS),5312–5315, Paper FR2.B1 (2015). [CrossRef]
25. W. H. Southwell, “Focal-plane pixel-energy redistribution and concentration by use of microlens arrays,” Appl. Opt. 33(16), 3460–3464 (1994). [CrossRef] [PubMed]
26. R. M. Gagliardi and S. Karp, Optical Communications, 2nd ed, (John Wiley and Sons, 1995), Ch. 4.
