1 Introduction

As a spatial exploration technology，remote sensing has gone through three stages：ground-based，airborne，and space-based. With increasing demands for detection distance and sensitivity in space remote sensing，infrared detectors with single-photon detection performance are urgently needed. Near-infrared InGaAs（P）/InP single-photon avalanche diodes（SPADs）have attracted wide attention in the field of space remote sensing due to small size，low power consumption，stable operation，and insensitivity to ultra-low operating temperatures^［1，2］. However，in the complex space environment，many high-energy particles such as electrons，protons，neutrons，and heavy ions^［3-5］ create a radiation environment in which photodetectors experience displacement，total-ionization-dose，and single-event effects. These will temporarily or even permanently damage the performance of the detector，leading to performance degradation or even failure^［6-9］. Therefore，in order to verify the reliability of the detector's operation in space，it is necessary to study how radiation affects device performance^［10］.

There have been various studies on the radiation resistance of InGaAs（P）/InP photodiodes. For example，Richard D. Harris et al. investigated how proton and gamma irradiation affected the performance of an InGaAs avalanche photodiode（APD）^{［11，12］}：under proton irradiation with an energy of 63 MeV and an incident flux of 2×10¹² P/cm²，the dark current of the detector increased from 5.6 nA to 1µA. Under gamma irradiation with a total dose of 269 krad（Si），the dark current of the detector increased from 0.1 nA to 1 nA. In addition，Zhang et al. gamma-irradiated an InGaAs p-i-n photodiode with a radiation dose rate of 16 rad（Si）/s and a total radiation dose of 30 krad（Si），and the results showed that the dark signal of the detector increased slightly after irradiation^［13］.

Previous studies on γ-irradiation of InGaAs（P）/InP photodetectors have mainly focused on p-i-n photodiodes or APDs，with little research on SPADs. In this paper，InGaAsP/InP SPADs were gamma-irradiated with 60Co at different irradiation doses and dose rates，and the dark current，the photon detection efficiency（PDE），the after pulse probability（APP），and the dark count rate（DCR）were compared to analyze how gamma radiation affected the performance of InGaAsP/InP SPADs.

1　Irradiated samples and experiments

1.1　Device structure

The APD chip is composed of a separate absorption，grading，charge，and multiplication（SAGCM）heterostructure，as shown in Fig. 1（a）. Epitaxial materials were sequentially grown on n-doped InP substrate：n^--doped InP buffer layer，intrinsic InGaAsP absorption layer，InGaAsP grading layer，n⁺-doped InP charge layer，and intrinsic InP cap layer. In the undoped InP cap layer，the deep and shallow diffusion of Zinc formed a double p-type diffusion，which controlled the photosensitive surface diameter of the chip to 80 μm. Here，a charge layer is added to balance the electric field between the absorption and the multiplication layers. The electric field in the absorption region is kept small to suppress the generation of tunneling dark current. While the electric field in the multiplication region is kept large to ensure sufficient impact ionization for the photoexcited carriers，which generates a macroscopic current and achieves single photon detection capability. After surface passivation，a p-contact electrode was selectively grown on the p- region，and gold was deposited on the back-side of the substrate as the n-electrode，as shown in Fig. 1（b）. The APD chip was packaged in a standard TO-66 metal tube shell，which integrated a three-stage thermoelectric cooler，as shown in Fig. 1（c）.

图 1. TO-66封装的InGaAsP/InP单光子雪崩探测器（SPAD）原理图: (a) InGaAsP/InP APD结构截面示意图; (b) APD 芯片; (c) 物理外观

Fig. 1. Schematic of a TO-66 packaged InGaAsP/InP single-photon avalanche diode (SPAD): (a) Cross-sectional schematic of the InGaAsP/InP APD structure; (b)APD chip; (c) Physical appearance

下载图片 查看所有图片

1.2　Irradiation conditions

The irradiation source was a 60Co point source with an intensity of 140 000 Ci，and all irradiations were performed at room temperature. According to the spatial application requirements of InGaAsP/InP SPADs，they should be able to work normally at an radiation dose rate of 5 krad（Si）/h and a total dose of 10 krad（Si）. Therefore，the detectors were exposed at a dose rates of up to 50 krad（Si）/h and a total dose of up to 20 krad（Si）.

Five devices with similar single-photon performance were selected from the same batch for irradiation experiments. The irradiation conditions were summarized in Table 1. Device1# and 2# were irradiated with a dose rate of 5 krad（Si）/h and a radiation dose of 10 krad（Si）. Device 3# was irradiated with a dose rate of 50 krad（Si）/h and the radiation doses were 1/7/10/20/50/70 krad（Si）. Device 4# was irradiated with a dose rate of 50 krad（Si）/h and a radiation dose of 10 krad（Si）. Device 5# was irradiated with a dose rate of 50 krad（Si）/h and a radiation dose of 20 krad（Si）. Among them，Device1# and 3# were biased at 0.95 V_br（breakdown voltage）. Device 3# was tested in situ，while the others were performed with shift testing.

1.3　Test parameters

We tested the SPADs’ dark currents in linear mode and the single-photon performances in Geiger-mode before and after irradiations，including PDEs，APPs，and DCRs. The dark current was tested using a Keithley 2635B programmable source meter. To test the single-photon performance，we built a pulse-gated single-photon test system as shown in Fig. 2. The signal generator（SG）generated two synchronizing signals：one was sent to the APD with a frequency of 1 MHz. The other was sent to the laser with a frequency of 200 kHz. The amplitude of the pulse voltage was 5 V with a pulse width of 4 ns. The light generated by the laser was calibrated to an average power of 0.1 photons per pulse after two-stage attenuation.

图 2. 门控模式单光子探测系统原理图

Fig. 2. Schematic diagram of gated-mode single-photon detection system

下载图片 查看所有图片

2　Experimental results and analysis

2.1　Influence of Gamma irradiation on dark current

Fig. 3 showed the dark current of Device 1# and 2# before and after irradiation. It can be seen that the dark current of the devices remained unchanged when the dose rate was 5 krad（Si）/h and the irradiation dose was 10 krad（Si）.

图 3. 辐照前后暗电流: (a) Device 1#; (b) Device 2#

Fig. 3. Dark current before and after irradiation: (a) Device 1#; (b) Device 2#

下载图片 查看所有图片

During irradiation，high-energy particles incident on the device lose energy due to ionization processes and generate electron-hole pairs in the material. If the rate of introducing electron-hole pairs is lower than the recombination rate，the performance of the device will tend to rapidly stabilize，which typically occurs within a few seconds to minutes after irradiation^［14］. To illustrate this possibility，an in-situ experiment was conducted on Device 3#. The irradiation dose rate remained at 5 krad（Si）/h. When the irradiation dose reached 1/7/10/20/50/70 krad（Si），the device was taken out and tested as soon as possible within 15 minutes. Fig. 4 shows the dark current of Device 3# during the in-situ test. It can be seen that when the irradiation dose rate was fixed at 5 krad（Si）/h，the APD's dark current did not change with the increase of irradiation dose. Therefore，for a low irradiation dose rate，the electron-hole pairs introduced by irradiation will tend to recombine in a very short time without causing too much impact on the device performance.

图 4. Device 3# 原位测试暗电流曲线

Fig. 4. Results of the dark current by in-situ test（Device 3#）

下载图片 查看所有图片

Becker et al. subjected InGaAs APDs to gamma irradiation with doses ranging from 1 krad to 200 krad（Si）：the maximum change in dark current after irradiation was 10 nA. However，Becker et al. did not specify the dose rate they used. We increased the dose rate from 5 to 50 krad（Si）/h and raised the total dose for Devices 4# and 5# to 10 krad（Si）and 20 krad（Si），respectively. The dark current of each device was tested before and after irradiation，and the results were shown in Fig. 5. As shown in Fig. 5（a），the dark current remained unchanged at a dose rate of 50 krad（Si）/h and an irradiation dose of 10 krad（Si）. However，as shown in Fig. 5（b），when the irradiation dose reached 20 krad（Si），the dark current of the Device 5# at 0.95V_br increased by 1.4 times after irradiation. In the subsequent room temperature annealing process，the dark current gradually recovered and returned to the pre-irradiation level after 30 hours. Fig. 6 showed the change of the dark current increment（ΔI）of Device 5# before and after irradiation. It can be seen that ΔI increased with the increase of the bias voltage，which was throughout the entire voltage range. For an APD，as the bias voltage increases，the depletion layer gradually expands from the multiplication layer to the absorption layer. The increment of the dark current caused by irradiation started from zero bias voltage，indicating that the ionization-induced electron-hole pairs were distributed throughout the device.

图 5. 辐照前后暗电流: (a) Device 4#; (b) Device 5#

Fig. 5. Dark current before and after irradiation: (a) Device 4; (b) Device 5

下载图片 查看所有图片

图 6. Device 5#辐照前后暗电流增量变化

Fig. 6. Change of the dark current increment before and after irradiation for Device 5#

下载图片 查看所有图片

As the electric field increased，the depletion layer gradually increased，collecting more and more electron-hole pairs caused by irradiation. As shown in Fig. 6，after 2 hours of irradiation，ΔI increased by 0.094 nA in the range of 0-20 V，0.106 nA in the range of 20-40 V，and 0.152 nA in the range of 40-60 V. The slow growth rate of ΔI represented a small multiplication.

Next，we will describe in detail the effects of irradiation on single-photon performance，such as PDE，DCR，and APP. It should be noted that these parameters after irradiation were obtained by shift testing 2 hours after irradiation.

2.2　Influence of Gamma irradiation on photon detection efficiency

For a SPAD，PDE is defined as the probability of detecting an incident single photon，which consists of three parts：quantum efficiency，i.e.，the photoelectric conversion efficiency of incident photons. This is mainly related to parameters such as the probability of optical coupling，the thickness of the absorption layer，and the absorption efficiency of the material. Another part is the probability of photo-excited carriers injecting into the multiplication layer，and the other is the probability of carriers injected into the multiplication layer to trigger avalanche breakdown which is determined by the electric field and the thickness of the multiplication layer.

As can be seen in Fig. 7，the PDEs were basically unchanged after irradiation. The PDEs in Fig. 7（a）and 7（c）fluctuated slightly and irregularly，with the maximum change of only 1.64%，which were caused by the inconsistent testing environments before and after irradiation and the testing errors. Photon absorption is an inter-band process. The collection and multiplication processes are controlled by the electric field and will not be affected by the electron-hole pairs generated by irradiation^［11］. Therefore，the PDEs will not change due to radiation，unless the radiation causes a strong recombination enhancement effect，which generally occurs in the displacement damage.

图 7. 辐照前后APD探测效率变化曲线: (a) Device 1#; (b) Device 2#; (c) Device 4#; (d) Device 5#

Fig. 7. PDEs of APDs before and after irradiation: (a) Device 1#; (b) Device 2#; (c) Device 4#; (d) Device 5#

下载图片 查看所有图片

2.3　Influence of Gamma irradiation on after pulse probability

The APP is also an important parameter for an APD，which represents the false avalanche probability caused by the release of captured carriers due to defects in the APD material in the absence of photon incidence. Fig. 8 shows the APPs before and after irradiation. As can be seen，the APPs did not change significantly after irradiation，only fluctuated slightly without obvious regularity. Under a dose rate of 50 krad（Si）/h and an irradiation dose of 20 krad（Si），the dark current of the APD increased by 1.4 times after irradiation，as shown in Fig. 5（b），while the PDE and APP，as shown in Fig. 7（d）and 8（d），remained unchanged. The irradiation may cause ionizing damage and displacement damage to APDs. The ionization process is the interaction between high-energy particles and atomic electrons，producing electron-hole pairs. It is a transient change and will not cause permanent damage to the material properties. In contrast，displacement damage creates vacancies and interstitials，that is，permanent lattice defects^［14］.The PDEs and APPs had no obvious changes，which meant that the device was free from irreversible damage such as defects to the material. Therefore，the main damage caused by gamma radiation to the SPADs was ionization damage，which corresponded to the dark current that recovered within a certain period of time.

图 8. 辐照前后APD后脉冲概率变化曲线: (a) Device 1#; (b) Device 2#; (c) Device 4#; (d) Device 5#

Fig. 8. PDEs of APPs before and after irradiation: (a) Device 1#; (b) Device 2#; (c) Device 4#; (d) Device 5#

下载图片 查看所有图片

3.4　Influence of Gamma irradiation on dark count rate

Fig. 9 shows DCRs of the APDs before and after irradiation. In order to better compare the changes in DCR，the PDE of each device was fixed at 40%，and the detailed changes in DCR were listed in Table 2. It can be seen that 2 hours after irradiation with a dose of 10 krad（Si）and a dose rate of 5 krad（Si）/h，the DCRs of Device 1# and 2# increased by 1.36 and 1.30 times，respectively，and then began to recover to the level of non-irradiation until 48 hours after irradiation. Meanwhile，under the irradiation doses of 10 k/ 20 krad（Si）and a dose rate of 50 krad（Si）/h，the DCRs of Device 4# and 5# increased by 1.30 and 1.58 times，respectively，and then began to recover to the level of non-irradiation until 120 hours after irradiation. After irradiation，the dark current of Device 1#，2#，and 4#，did not change，but the DCRs increased slightly. This was because the previously observed dark current was a parameter in the linear mode，where the multiplication factor was small，and the generation and recombination processes of electron-hole pairs can be offset. While in the Geiger-mode，due to a huge multiplication factor，the generated electron-hole pairs are greatly multiplied，which exceeds their recombination probability，and this is reflected in the DCR.

图 9. 辐照前后APD暗计数率变化曲线: (a) Device 1#; (b) Device 2#; (c) Device 4#; (d) Device 5#

Fig. 9. DCRs of APPs before and after irradiation: (a) Device 1#; (b) Device 2#; (c) Device 4#; (d) Device 5#

下载图片 查看所有图片

The difference in DCR and recovery time between Device 4#/5# and Device 1#/2# was mainly due to the difference in total radiation dose and the dose rate. As the radiation dose rate increased，the generation rate of electron-hole pairs exceeded the recombination rate，resulting in a multiplication process in the depletion layer and a longer time to recover to the non-irradiation level. In addition，as the total radiation dose increased，the number of generated electron-hole pairs increased，resulting in more obvious changes in DCR.

3　Conclusions

In this paper，InGaAsP/InP SPADs were gamma-irradiated at different doses and dose rates. At a radiation dose of 10 krad（Si），there were no changes in the dark current，PDE and APP，only a slight increase in DCR，which basically recovered to the non-irradiation level within a few days. When the radiation dose was increased to 20 krad（Si），the dark current and the DCR began to increase and gradually recovered after annealing at room temperature. The analysis indicated that the performance degradation of the device was mainly caused by ionization damage from gamma irradiation in the bulk material，resulting in many electron-hole pairs and a short-term degradation of device performance. During subsequent room temperature annealing，the device performance recovered to the level of non-irradiation due to the recombination of non-equilibrium carriers.