|
1.INTRODUCTIONHgCdTe APDs have opened a new horizon in photon starved applications due to their exceptional performance in terms of high linear gain, low excess noise and high quantum efficiency. Both focal plane arrays (FPAs) and large area single element using HgCdTe APDs have been developed to detect without signal degradation the spatial and/or temporal information contained in photon fluxes with a low number of photon in each spatio-temporal bin. The enhancement in performance that can be achieved with HgCdTe has subsequently been demonstrated in a wide scope of applications such as astronomical observations, active imaging, deep space telecommunications, atmospheric LIDAR and mid-IR (MIR) time resolved photoluminescence measurements using detectors manufactured at CEA/Leti [1]-[6], DRS Leonardo [7]-[10] and Leonardo [11], [12]. The present communication focus on the present development and tests of detector modules at CEA/Leti for applications that only require temporal information and can be addressed with a single or a few APDs, such as atmospheric LIDAR and free-space optical communication (FSO). The detectors are made of components and sub-assemblies, including the HgCdTe APDs, which are developed and/or adapted for each application in order to comply with their specific requirements. These sub-assemblies are described in the next section together with the status and objectives for our current detector module development for FSO and atmospheric LIDAR. In the thereafter following section 3, we report characterization results on a detector developed for high depth resolution atmospheric LIDAR, in collaboration with Airbus DS in the scope of an R&T CNES project. Section 4 present a summary and the perspectives for time resolved HgCdTe APDs detector modules at CEA/Leti. 2.HGCDTE APD DETECTOR DEVELOPMENT AT CEA LETI2.1Detector modulesThe detector modules that are developed at CEA/Leti are constituted of sub-assemblies that are designed and/or adapted to meet the requirements for each applications in terms of operating temperature, bandwidth, active area, optical coupling, electrical interface and cooling system:
2.2Detectors for FSOFSO is a typically photon starved application in which the number of detected photons per bit is limited by the combination of low emitted laser power, beam divergence and possible atmospheric absorption. A first HgCdTe APDs detector module was developed at CEA/Leti for deep space telecommunications. Characterized by a bandwidth of 80 MHz and a NEP=65fW/√Hz (BW= 80 MHz), it was successfully used to demonstrate 80 Mbps rate from moon to earth during the LLCD experiment, in collaboration with ESA and NASA [1],[2]. This demonstration has spurred the interest for HgCdTe APDs for this application and we are presently developing two new detector modules for high sensitivity deep space FSA, with ESA, and for high 10 Gbps communication closer to earth in collaboration with Mynaric Lasercom GmbH. The ESA project objective is to develop a 4 quadrant detector for deep space FSO applications, with single photon sensitivity and a bandwidth of 300 MHz on each detector. To reach single photon resolution, the detector is made using APD-ROIC hybrid integrated into a LN2 cryostat. An optical active area of 200 μm will be achieved using 4 quadrant micro lenses fabricated into the CZT substrate. At APD gains in excess of 100, the detector module should be capable of linear mode single photon detection. The bandwidth of the detector module is limited to 300 MHz by the Si ROIC. A hetero-structure APD architecture have been designed and is grown using MBE to optimize the quantum efficiency while minimizing the timing jitter at single photon level to a value below 100 ps. The expected performance of the detector is resumed in table 1. This sensitivity, high bandwidth and low jitter will not only be an asset for deep space FSO, but also for quantum optical applications such a Quantum Key Distribution (QKD) and we intend to develop a fiber coupled version of the detector. By distributing the signal on the 4 detectors, we expect to be able to detect single photons at GHz rates. Table 1.Expected characteristics of the HgCdTe APD detector modules under development for FSO at CEA/Leti.
The detector for fast 10 Gbps FSO is developed in collaboration with Mynaric Lasercom Gmbh. The focus of this development is the optimization of the response time of the APDs to reach a bandwidth in the order of 10 GHz. A specific MBE grown hetero-structure –have been designed in this purpose, which should also limit the dark current generation to achieve room temperature operation. The high bandwidth and operating temperature requirement do also imply that the APD active area needs to be limited. Again, micro-lenses made in the CZT substrate will be used to reduce the active APD diameter while complying with the application requirements in terms of collection efficiency. The APD will be connected to a deported commercial TIA. No specific packaging will be used for this detector as the APD will be operated close to room temperature. The expected performance for this detector is resumed in table 1. 2.3Detectors for atmospheric LIDARA first APD detector module was developed to respond the needs of Differential Absorption Lidar (DIAL) specified by Laboratoire de Méterologie Dynamique (LMD) within the scope of a CNES R&T project. The detector uses a deported RTIA and was cooled in a vacuum package using a 4-stage TEC. The characteristics of the detector and first CO2 DIAL measurements at LMD have been reported in references [3]-[4]. The first DIAL measurement showed that the use of HgCdTe APDs for direct detection LIDAR was in favor of increased range and higher temporal resolution, but also that a further increased sensitivity is needed to increase the range. In addition, a long term remanence was observed when the light was strongly absorbed by the atmospheric CO2, which impeded a correct extraction of the optical thickness at larger distances [4]. Large area APDs from the same batch were also submitted to proton irradiation and endurances tests that showed on a stable performance and motivated the pursuit of the development of large area HgCdTe APDs for applications in space [3]. Lidar detectors for atmospheric LIDAR are currently developed within the scope of R&T CNES projects and a H2020 project termed HOLDON. The aim of these projects is to optimize characteristics of the HgCdTe APD detector modules to comply the specific challenges imposed by atmospheric LIDAR measurements in terms of dynamic range, response time, temporal lag (remanence) and background noise. A new detector module has been developed within the scope of a CNES R&T project in collaboration with Airbus DS, aiming to meet specifications atmospheric and bathymetric profiling with high spatial depth resolution. This specification requires an increased bandwidth of the APD and the RTIA. The APD bandwidth has been increased by reducing the collection time of the photo-generated carriers by forming a macro-diode on specifically developed ceramic IC using HgCdTe with 15 μm pitch to form the large area APD. APD-IC hybrids are mounted onto a specifically designed cold PCB, illustrated in figure 5, and connected by wire bonding to two different deported RTIA. The first amplifier is a ROIC that was developed at CEA/Leti with an expected bandwidth of 100 MHz and a noise below 1 pA/√Hz, termed CL MEATS [16]. This amplifier was designed to be used with a direct hybridization with a 5x5 array at a pitch of 30 μm. This pitch do not allow to meet the BW specification with the available design of APDs. For this reason, the amplifier was used in a deported mode in which the performance of the ROIC could be degraded due to the additional capacitance at the input of the amplifier. A second commercial RTIA (C-RTIA), with a nominal bandwidth of 200 MHz and an expected noise of 3 pA/√Hz was therefore implanted as a back-up on the cold PCB. The cold PCB with amplifier and APD-IC hybrid, mounted on the cold post of the LN2 cryostat, is illustrated in figure 5. Two detector modules have been developed and delivered to Airbus DS for characterization on a specifically LIDAR simulation test bench at λ=1.06 μm. First characterization results obtained from measurements at CEA/Leti and Airbus are reported in section 3. The aim of the HOLDON project, which started in beginning of 2018, is to develop a versatile detection chain that shall meet most of the application requirements for atmospheric LIDAR measurements. The main tasks and partners of the project are:
The project objective and logic is illustrated in figure 6. LIDAR application requirements in fields varying from greenhouse gas detection (CO2 DIAL at LMD, CH4 IPDA at DLR), and cloud and aerosol profiling (at Airbus DS, using a LIDAR simulator from uv to NIR developed by ALTER) have been used to define the detector specifications. The detection chain is constituted of an HgCdTe APD-ROIC (developed at CEA/Leti) that will be integrated into low operating temperature packages (LN2 cryostat and sterling cooler). Specific proximity electronics will be developed by IDQ to control the functionalities of the detector and to sample the data. The main technical objective is to obtain a dynamic range, of 6 order of magnitude, while achieving photon shot noise limited detection for signals down to 1 photon per laser shot and observation time. The HOLDON approach to meet these extreme specifications is to use a Si-CMOS ROIC in which the gain and temporal resolution is adapted during duration of the return signal with the goal to enable a photon noise limited detection over the signals full dynamic range. This functionality is expected be met by implementing a dual mode of operation: a real-time trans-impedance mode will be used to sample the signal outside the cryostat on a proximity electronics card. This mode is adapted to the higher end of the dynamic range. The other mode is what is termed on-chip sampling mode (OCS), in which the signal will be sampled in a number of analog temporal bins in the ROIC and that will be read out at a lower speed. The OCS mode will in particular be used to address the lower end of the dynamic range. In addition to the development of the dedicated ROIC, dedicated HgCdTe APDs will be processed during HOLDON in order to optimize the quantum efficiency at the expected laser wavelengths, maximize the APD gain, reduce the excess noise factor and reduce the temporal remanence. Micro-lenses will also be explored to reduce the APD area in favor of lower dark currents and diode capacitance. The objective in performance of the HOLDON detector is compared to the other CEA/Leti LIDAR detection modules in table 2, illustrating the expected increase in performance on all parameters. Table 2.Expected characteristics of the HgCdTe APD detector modules under development for atmospheric LIDAR at CEA/Leti.
3.HIGH BANDWIDTH LIDAR DETECTOR MODULE CHARACTERIZATIONDetection modules for high depth resolution atmospheric profiling have been developed in collaboration with Airbus DS in the frame of an R&T CNES project, as described in section 2.2. The experimental set-up used to characterize the detector modules is illustrated in figure 7. The detector responsivity, noise equivalent power (NEP), and quantum efficiency to excess noise ratio measured by focusing a CW laser on the HgCdTe APD. The incident optical power, Popt, is estimated using a fiber coupled Variable Optical Attenuator (VOA) and a variable number of fixed 10 dB fiber coupled attenuators. The external responsivity of the detector is directly estimated from ratio of the voltage difference between induced by the laser ΔVlaser and Popt: The noise equivalent input power density, NEP, and noise equivalent number of photons, NEPh, are estimated from the noise standard deviation without incident light: Where TC is the characteristic time over which the photo current is integrated at given bandwidth, BW, limited by the detector and/or by applying a low pass filter on the sampled signal. In the figure 8 we report the measured responsivities and NEP values at the output of the C-RTIA and CL MEATS detection chains. The bandwidth was limited by numerical filtering on the sampled signals in order to control the BW over which the noise is integrated and to limit the integration of amplifier noise at higher frequencies where the spectral noise density tends to peak. The increase in gain in the HgCdTe APDs results in a strong increase in responsivity and a reduced NEP up to a reverse bias in the order of -12 V. The corresponding APD gain at -12 V reverse bias was 300 and 170 for the C-RTIA and CL MEATS outputs, respectively. The difference in APD gain is due to a difference in bias on the input nodes of the TIA which can be estimated to be larger than 1 V. At the highest reverse bias we observe a NEP in the order of 10-15 fW/√Hz on both detectors, which corresponds to NEPh values of about 5 photons at 80 MHz, for the C-RTIA output, and 15 Photons at 8 MHz, for the CL MEATS output. In terms of NEP, these values are at least a factor 3 lower than what was observed for the TEC cooled detectors that were used for deep space FSO (see section 2.2) and DIAL (see section 2.3). The information conservation in the detector due to the quantum efficiency (QE) and excess noise factor (F) of the APD, measured by the ratio QEFR=QE/F, can be directly estimated from the degradation of the signal to noise ratio with respect to the photon noise limited CW signal at the input with a power Popt: The value of QEFR was found to decrease when higher BW where used to integrate the noise. This dependency indicates a variation in gain between high and low frequencies. An intermediate bandwidth of 80 MHz where used to estimate the QEFR on the C-RTIA detection chain yielding an external fiber to detector QEFR=51 %. This value was obtained without AR coating and could also be reduced be an optical coupling losses due to multiple optical interface and the use of small field of view in the cold screen which could cause vignetting in case the optical axe is not well aligned with the APD active area. The unfiltered bandwidth of each output was estimated from Fourier analysis of the impulse response of the detection module. The typical impulse response of the C-RTIA detection chain is illustrated in figure 9 for an average signal of about 200 photons, at which the photon shot noise gives a dominating contribution to the signal amplitude. The width at half maximum is FWHM=2.1 ns and the corresponding total bandwidth was estimated to 180 MHz, which comply with the Airbus specifications for atmospheric profiling. Similar measurements were performed on the CL MEATS detection chain and showed on a total bandwidth of about 20 MHz (FWHM of 33 ns). This is below the expected bandwidth of this amplifier and the reduction in bandwidth is probably due an excess input capacitance when using the ROIC as a deported amplifier. The limited bandwidth do not comply with the Airbus DS specification which is why this detection chain is more adapted for DIAL or IPDA measurements with reduced depth resolution. The linearity of the detector modules was estimated using pulsed measurements and by attenuating the signal using the VOA. The use of a variable attenuator is associated with an uncertainty on the value of attenuation of a few percent and is not accurate enough to verify linearity down to 10-3 level which tend to be required for the most demanding lidar applications. A first linearity measurement with the C-RTIA detection chain is reported in figure 10. The detector is found to linear over close to 3 order of magnitudes but tend to saturate when the output signal approaches 100 mV. This saturation can be caused by the use of the C-RTIA at low temperatures and/or the saturation of a pre-amplifier which was not used in ideal conditions during this measurement. We are currently developing a new measurement technique based on the ratio of two consecutive pulses with constant power ratio to be able to probe the linearity with high precision over more than 6 orders of magnitudes. To reduce uncertainty, a complementary measurement is also being set-up at Airbus DS, with an integrated sphere and a calibrated radiometer. Remanence is a trailing signal that is observed on time scales longer than the linear response time of the detector that can interfere with atmospheric LIDAR signals, which can evolve over 5-6 orders of magnitudes. Such interference was for example observed during the CO2 DIAL measurements on a time scale of 10 μs after the detection of long duration high intensity peak. In atmospheric or bathymetric LIDAR that is developed at AIRBUS DS, the required depth resolution is one meter, which corresponds a time scale of 10 ns during which the detector should ideally relax. Remanence measurement was carried out at CEA/Leti using the set-up illustrated on the right hand side of figure 7. The goal of the set-up is to use a Mach-Zehnder fiber-coupled light modulator to increase the extinction ratio of an initially pulsed laser source to guarantee that no laser light is incident on the APD after the detection of the laser pulse. The polarization of the pulsed laser light was aligned with an in-line fiber optic polarizer that blocks orthogonal polarized light, which is not subject to modulation by the Mach-Zenhder modulator, by at least 23 dB. The axe of the light from the output of the polarizer is aligned with the axe of the Mach-Zehnder modulator in order to generate an extinction ratio of at least 30 dB. Given an initial extinction ratio of at least 30 dB of the pulsed laser, we expect the extinction ratio of the light on the detection to be higher than 50 dB (10-5) at 5 ns from the detection of the pulse. Hence, the measured remanence is dominated by long term relaxation that originates from the HgCdTe APD, RTIA amplifier and/or sampling electronics. Figure 11 reports the measured impulse response on the C-RTIA output signal, which was pre-amplified using a differential instrumental amplifier and sampled with an 8 bit Lecroy oscilloscope using 3 bit ERES filtering and waveform averaging to increase the dynamics of the sampled signal to about 40 dB. The shape of the response in figure 11 shows on three contributions to the response time. The fastest contribution (I), observed over the first 10 ns, corresponds to the linear response time of the APD. A slower contribution (II) is consistent with a contribution from the collection of residual carriers through diffusion and is characterized by an exponential decay with a time constant equal to the expected life time of the minority carriers in the HgCdTe substrate of about 15 ns. This contribution can be strongly reduced by concentrating the light within the active area of the APD using, for example, micro-lenses. A slower trailing edge (III) is observed after 50 ns of the pulse detection and reaches a relative intensity below 10-4, 200 ns after the main peak. At longer time scales, the signal is hidden in the residual noise. This contribution could origin from trapping in the APD and/or -relaxation phenomena in the amplifier or sampling electronics. Complementary measurement with different amplifier and sampling electronics needs to be made to separate the different contributions. Such measurements will be performed at Airbus DS using the 1.06 μm LIDAR test bench and a high resolution 16 bit ADC. The level of remanence observed in figure 11 is compared with an expected bathymetric LIDAR signal relative to the strong surface reflection peak for a standard and a more turbid scenario. It can be seen that the present detector is below the standard scenario but is likely to interfere with the more turbid case. Part of the remanence is due to the contribution (II) to the response and is likely to be reduced by concentrating the optical absorption within the active APD area. This optimization will be explored within the HOLDON project. Two detector modules have been delivered to Airbus DS in view of extensive characterization on their λ=1.06 μm LIDAR simulation test bench. Initial characterization have been performed on the CL MEATS output and the first results are consistent with the results obtained at CEA/Leti. A complementary measurement at Airbus DS of the optically active area of the HgCdTe APD hybrid is reported in figure 12. The response is found to be homogeneous over the full area of the detector, with a small modulation between the APDs that are interconnected to from a large area APD of few percent. The variation in sensitivity is due to a reduced collection efficiency between the APDs. The observation of a small variation is consistent with the observation of a high QEFR value, but it can also be limited by the optical spot-size. Further characterization of both detection chains on this detector module are currently made at Airbus DS and will the subject of a future communication. The results of the first characterizations are compared with the objectives in table 3. All characterized parameters are found to comply with the specifications except the QEFR value, which is reduced due to the absence of AR coating and a possible loss due to vignetting at the cold diaphragm, and remanence, which is a factor two above the specification at 100 ns. The remanence is partly limited by a slow response of carriers generated between the APDs in the macro pixel which is why a strong reduction can be expected by concentrating the light to the active APD area. This approach will be explored in the frame of the H2020 HOLDON project. Table 3.Comparison between the nominal and obtained characteristics of the high depth resolution LIDAR HgCdTe APD detector module.
4.SUMMARY AND PERSPECTIVESHgCdTe APD detector modules are developed at CEA/Leti for time resolved applications such as atmospheric LIDAR and FSO. After first demonstration with detector modules based on TEC cooled large area the development is presently dedicated to the development of detectors with higher sensitivities and/or faster response time through the optimization of the HgCdTe APDs, using MBE grown hetero-structures, of the optical coupling, using micro-lenses patterned into the CZT substrate, of the electrical pre-amplifier and/or of the cryogenic packaging. For FSO, detector are currently developed for deep space application for ESA, a 4 quadrant detector with down to single photon sensitivity and a bandwidth 300 MHz, and a detector for higher data rates, up to 10 Gbps, and that will be operated at room temperature is developed in collaboration with Mynaric Lasercom GmbH. Two detectors modules are also developed for atmospheric LIDAR applications. A dual channel detector with two 150 μm diameter HgCdTe detectors have been developed for Airbus DS within the scope of an R&T CNES project. The two channels are using two different deported amplifiers with different characteristics in terms of gain, input noise, bandwidth and remanence. Sensitivities in the range of 10-15 fW/√Hz were observed for the two detectors, representing a gain in sensitivity by a factor 3 compared to the TEC cooled detectors. A bandwidth of 180 MHz was measured on the channel that used a commercial RTIA and did comply with the requirements for high depth resolution LIDAR. In addition, remanence measurement showed that this detector had a trailing relative signal of about 10-4, 200 ns after the detection of an impulse, and could be compatible with the demanding requirements for bathymetric profiling. Further characterizations of this detector module are currently made at Airbus DS. The second detector module is developed within the H2020 project HOLDON with the objective to make a versatile full detection chain that will be capable of responding to needs of most LIDAR applications. The optimization of the HgCdTe APDs, ROIC and optical coupling should enable to reach a 60 dB photon noise limited dynamic range with down to single photon detection capability for observation times ranging from 10 ns to 10 μs. ACKNOWLEDGEMENTSThe authors which to acknowledge CNES, ESA, Mynaric LaserCom AG and the European Commission’s Horizon2020 program for supporting the development and optimization of time resolved HgCdTe APD detector modules. The HOLDON project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 776390. You can visit our website http://holdon-h2020.eu/ and join our community on linkedIn “HOLDON project” REFERENCESI. Zayer, et al.,
in SpaceOps 2014 Conference,
(2014). Google Scholar
J. Rothman et. al.,
in Proc. SPIE,
(2015). Google Scholar
J. Rothman, G. Lasfargues, B. Delacourt, A. Dumas, F. Gibert, A. Bardoux, M. Boutillier, CEAS Space J, 9
(507),
(2017). Google Scholar
A. Dumas, J. Rothman, F. Gibert, D. Édouart, G. Lasfargues, C. Cénac, F. Le Mounier, J. Pellegrino, J.-P. Zanatta, A. Bardoux, F. Tinto, and P. Flamant, Applied Optics, 56
(7577),
(2017). Google Scholar
B. Delacourt, P. Ballet, F. Boulard, A. Ferron, L. Bonnefond, T. Pellerin, A. Kerlain, V. Destefanis and J. Rothman, J. Electron. Mater, 46
(6817),
(2017). Google Scholar
A. Gassenq, L. Milord, J. Aubin, K. Guilloy, S. Tardif, N. Pauc, J. Rothman, A. Chelnokov, J. M. Hartmann, V. Reboud, and V. Calvo, Appl. Phys. Lett., 109
(242107),
(2016). Google Scholar
X. Sun, J. B. Abshire, J. D. Beck,
in Proc. SPIE,
(2014). Google Scholar
W. SullivanIII, J. D. Beck R. Scritchfield, M. Skokan, P. Mitra, X. Sun, J. Abshire, D. Carpenter, B. Lane, J. Electron. Mater, 44
(3092),
(2015). Google Scholar
P. F. McManamon, P. Banks, J. Beck, A. S. Huntington, E. A. Watson,
in Proc. SPIE,
(2016). Google Scholar
X. Sun, J. B. Abshire, J. D. Beck, P. Mitra, K. Reiff, and G. Yang, Optics Express, 25
(16589),
(2017). Google Scholar
Dani E. Atkinson, Donald N. B. Hall; Ian M. Baker; Sean B. Goebel; Shane M. Jacobson; Charles Lockhart; Eric A. Warmbier,
in Proc. SPIE,
(2016). Google Scholar
G. Finger, I. Baker, M. Downing, D. Alvarez, D. Ives, L. Mehrgan, M. Meyer, J. Stegmeier, H. J. Weller,
in Proc. SPIE,
(2017). Google Scholar
J. Rothman, J. Electron. Mater,
(2018) https://doi.org/10.1007/s1166 Google Scholar
G. Vojetta, F. Guellec; L. Mathieu; K. Foubert; P. Feautrier; J. Rothman,
in Proc. SPIE,
(2012). Google Scholar
J. Asbrock, S. Bailey, D Baley, J. Boisvert, G. Chapman, G. Crawford, T. De Lyon, B. Drafahl, J. Edwards, E. Herrin, C. Hoyt, M. Jack, R. Kvaas, K. Liu, W. McKeag, R. Rajavel, V. Randall, S. Rengarajan and J. Riker,
in Proc. SPIE,
(2008). Google Scholar
H. Amhaz, K. Foubert, F. Guellec and J. Rothman,
“A novel 0.5GHz real time asynchronous photon detection and counting technique: ROIC design for cooled SWIR HgCdTe infrared detector,”
in IEEE 11th International New Circuits and Systems Conference (NEWCAS),
1
–4
(2013). Google Scholar
|