The bandwidth and saturation power of germanium photodetectors are two crucial parameters for implementing analog and microwave photonics circuits. In conventional schemes, it is hard to optimize these two parameters simultaneously, due to different requirements for the size of absorption region. We report the design and demonstration of a high-power and high-speed germanium photodetector with distributed absorption regions. In this distributed-absorption photodetector (DAPD), the junction is formed by a multiple absorption region (n-cell) on a mutual substrate, and the input light is split and fed into the n cells. A comprehensive theoretical model is developed, and the device bandwidth and power loss in aspect of the number of cells is discussed. Experimentally, 2-, 4- and 8-cell DAPDs are investigated, and the 2-cell scheme shows the superior performance with the radio-frequency saturation photocurrent as high as 16.1 mA and the 3 dB bandwidth as high as 50 GHz. Without changing the standard process in the silicon photonic foundry, the DAPD can be seamlessly integrated with other photonics devices, and it is very attractive to applications such as integrated microwave photonics systems.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
High-speed and high-power photodetectors (PD) are greatly needed in analog and microwave photonics . During the past decade, various kinds of high-speed III-V PDs with high-power handling capability have been demonstrated, such as the uni-traveling carrier (UTC) PD  and the modified UTC (MUTC) PD . On the other hand, silicon-based device has attracted great attentions and seems to be a possible solution for high-speed and high-power applications [4,5] due to its low cost, high thermal conductivity and easy compatibility with complementary metal oxide semiconductor (CMOS) technology. Among various silicon photonics (SiPs) components, the waveguide coupled germanium PD (Ge PD) obtained extensive research interests [6–8]. While most of the researches focus on the improvement of dark current , responsivity  and bandwidth , it is significant to improve both the bandwidth and saturation power of the Ge PD simultaneously.
Nowadays, the bandwidth of Ge PD in literatures can achieve over 70 GHz [9,10], however these schemes cannot be operated in high-power condition simultaneously. The saturation power of the Ge PD is mainly restricted to the thermal failure and space-charge effect [11–14]. The thermal failure can be avoided by using material with high thermal conductivity. Compared with III-V devices, Ge PD features with higher thermal failure threshold due to the higher thermal conductivity . On the other hand, the space-charge effect can be alleviated by three different methods, including light field manipulation [16,17], junction structure design [18–21] and using PIN arrays with a traveling wave electrode configuration [22–24]. In the previous work, we proposed and experimentally demonstrated a kind of high-power Ge PD with new junction structure, based on two distributed absorption regions (cells) that formed on a mutual doping substrate . The measured results validated this new structure has the improved power handling capability compared with conventional one. However, the optimal cell number for the best performance is not investigated and concluded. More importantly, enlarging the effective absorption volume of the photodetectors has been proposed as the most direct way to increase the saturation photocurrents . According to this method, an assumption that the saturation power of DAPD can be improved by increasing the cell number should be judged experimentally.
In this paper, to elaborately investigate the inherent mechanism of the distributed-absorption PD (DAPD), the optimal cell number is discussed in aspect of high-speed and high-power characters, through the conventional theoretical model. Three kinds of DAPDs, including 2-, 4- and 8-cell schemes, are designed, fabricated and tested. It turns out the 2-cell DAPD has a high radio-frequency (RF) saturation photocurrent of 16.1 mA and a large 3 dB bandwidth of 49.3 GHz. Comparing with conventional PD, the 2-cell DAPD has the larger effective absorption volume. On the other hand, it has the lowest current loss among a series of n-cell DAPDs. Therefore, the 2-cell DAPD has the large-bandwidth and high-power performance. The DAPD shows a great potential for complex high-power microwave applications, due to its zero-change fabrication process in the silicon photonic foundry. These results provide a guidance for the design of high-power and high-speed Ge PD.
2. Theoretical model of DAPD
Here, an expanded PIN junction is designed using a multiple intrinsic absorption region on a mutual P-doped substrate, and the DAPD can be implemented with lumped structure. Following an assumption that the saturation power of DAPD can be improved by increasing the cell number, we give a comprehensive bandwidth analysis of the proposed DAPD.
Figure 1 shows the cross-section view of conventional PD and 2-cell DAPD, and the corresponding equivalent circuits are also provided. In the circuit, the current source is introduced to represent the photoelectric conversion in the absorption region. The parameters of the equivalent circuit include the junction capacitance Cj, the series resistance Rp+ (mainly comes from the P+ doping region) and the external load RL. The junction resistance and parasitic parameters from electrodes are ignored for simplification. In the circuit of conventional PD, the current response can be easily calculated and its 1.5 dB bandwidth fA is given in Eq. (1). This result equals to the 3 dB bandwidth of power response. In the circuit of 2-cell DAPD, the current response can be calculated by using the superposition principle, and the 1.5 dB bandwidth fB is given in Eq. (2). Obviously, even though the absorption region is different, the bandwidths of the two PDs are equal.
When the absorption region number n increases, the calculation of n-cell DAPD’s response becomes very difficult, as all Ge regions share a common P+ doped substrate. Figure 2(a) shows the cross-section view of n-cell DAPD. For clarification, the circuit between points A and B is reconstructed and shown in Fig. 2(b). The equivalent circuit of n-cell DAPD consists of a cell array with series resistance Rp+, junction capacitor Cj, and current source. The current source i’ph is related to the photocurrent from individual absorption region iph through the Norton equivalent-circuit analysis , as shown in Eq. (3). The circuit analysis of the n-cell DAPD is formulated using the transmission matrix method. When the sequence number of nodes m≥1, the recursion formula between nodes m and m+1 can be written as Eq. (4).
Furthermore, the current response IL is the sum of Im and I0. If the values of Rp+, RL and Cj are set to be 100 ohm, 50 ohm and 20 fF, respectively, the current response of DAPD can be calculated. The current responses of 2-, 3-, 4-, 6- and 8-cell schemes are shown in Fig. 3. The simulated frequency range is from 10 MHz to 100 GHz. The device bandwidth is decreased from 32.96 GHz (2-cell) to 4.1 GHz (8-cell) with the increasing number of absorption regions. Meanwhile, the current loss is also increased. The current loss comes from the offset caused by bidirectional currents at each series resistance Rp+, and the value increases with the cell number. The detailed results are shown in Table 1.
According to this bandwidth analysis, the bandwidth degradation from this junction structure is revealed, and it is a main issue for the high-speed application of DAPD. Alternatively, the gain peaking technology can be used to alleviate such a problem . The principle of gain peaking is that the inductor is introduced into the PD electrode to compensate part of the frequency response roll-off caused by junction capacitance. Here, to prove the effectiveness of the gain peaking technology, an inductor (540 pH) is introduced into the equivalent circuit model of DAPD, and the simulated results of bandwidth are listed in Table 1. The simulated power losses of DAPDs are not include, as they are independent to the inductor.
3. Fabrication and experimental results
The proposed DAPD with different cells are fabricated on the 220 nm thick silicon on insulator (SOI) wafer with 2 µm buried oxide (BOX) using standard silicon photonics platform. The length, width and height of each Ge cell are 10, 5, and 0.5 µm, respectively. The evanescently coupling and PIN doping structures are implemented. For the effective light absorption, dual-illumination is also implemented for each cell. The input light propagates along the 0.5 µm wide single-mode channel waveguide and it is split by the cascaded 1×2 multimode interference (MMI) couplers. The lumped ground-signal-ground (GSG) electrodes are designed to collect the photocurrent from each cell. Optical path for each cell is designed to be identical. In practice, an n-cell DAPD could be constructed by n Ge cells and 2n-1 1×2 MMI couplers.
We design and fabricate three DAPDs, namely 2-, 4-, and 8-cell ones. The microscope images are shown in Fig. 4. Considering the bandwidth degradation with the increasing cell numbers, an on-chip inductor of 540 pH is used in the GSG electrodes to boost the bandwidth. The grating couplers used for light coupling between fiber and chip have ∼4.5 dB coupling loss for the TE polarization, and the working wavelength is 1550 nm. The waveguide loss is 2 dB/cm, and the insertion loss of MMI coupler is about 0.3 dB. The routing waveguide lengths of 2-, 4- and 8-cell DAPDs are 1026.3, 3639.3 and 19154.45 µm, respectively.
3.1. Direct-current performance
The static performances of DAPDs, including the dark current, responsivity and direct-current (DC) power handling capability, are first characterized. The current-voltage results under dark-illuminated condition are shown in Fig. 5(a). Under the -3 V bias voltage, the dark currents of 2-, 4-, and 8-cell DAPDs are 0.46, 0.86 and 3.46 µA, respectively. The highest dark current can be measured for the 8-cell DAPD, as it has the largest active area.
The photocurrent at 1550 nm is further measured in Fig. 5(b), and the DC saturation of 2- and 4-cell DAPDs can be observed. Under low-power condition (from 0.1 to 10 mW), 2-cell DAPD has a responsivity of 1 A/W, while the 4- and 8-cell DAPDs have lower responsivity of 0.85 and 0.77 A/W. Under high-power condition, 2- and 4-cell DAPDs both have ∼37 mA maximum currents which correspond to the optical power of 56 and 112 mW. The maximum current of the 8-cell DAPD cannot be easily determined in our measurements, limiting by the available maximum optical power. In conclusion, while implementing the n-cell DAPD, the optical loss from the splitters and waveguide has an inevitable influence on the responsivity and DC saturation performance. The maximum optical input power of DAPD is increased with the cell number but the DC maximum current is not, and it is the comprehensive result from the optical loss from the splitter/waveguide and the current loss from the junction structure.
3.2. Small-signal RF response
The small-signal RF measurements are also carried out to assess the opto-electrical bandwidth. The conventional RF-test setup is implemented with a vector network analyzer (VNA: Anritsu MS4647B), as shown in Fig. 6(a). The frequency response of the Mach-Zehnder modulator (MZM) has been calibrated in advance. The Impedance Standard Substrate (Cascade, 101-190C) is used to calibrate the influence of the bias-tee, cables and microprobe (Cascade, I67-GSG-150). S21 traces of the three DAPDs are measured under the -3 V bias voltage and 0.5 mA photocurrent, as shown in Fig. 6(b). With the help of the gain-peaking effect introduced by the on-chip inductor, the bandwidths of 2-, 4- and 8-cell DAPDs are boosted to 49.3, 30.84 and 5.35 GHz, respectively. The difference between the simulated and measured bandwidth results is mainly due to the impact of the parasitic parameters from the fabrication process.
3.3. High-power RF performance
The high-power RF performance contains the bandwidth degradation and RF-power saturation. The same setup in Fig. 6(a) is used. The S21 responses of the three DAPDs at different photocurrents are recorded, as shown in Fig. 7. While the photocurrent increases from 1 to 20 mA, the 2-cell DAPD suffers a rapid bandwidth degradation from 49.3 to 7 GHz. On the other hand, the bandwidth of 4-cell DAPD decreases from 30.8 to 5.8 GHz, and the degradation of 8-cell DAPD is from 5.3 to 1.1 GHz. On the other hand, the maximum responses at 1 GHz are 22.7, 19.3 and 16.3 dB corresponding to 2-, 4- and 8-cell DAPDs, and the difference of them is caused by the power loss from the distributed absorption structure.
The saturation characteristics are measured by extracting the RF output power at a fixed frequency. The experimental setup is shown in Fig. 8(a). The 10 GHz sinusoidal signal is generated by the arbitrary waveform generator (AWG: Keysight M8195A). The bias voltage of the MZM and RF signal amplitude are optimized to achieve 30% modulation depth and the cable loss has been subtracted. Figure 8(b) shows the RF output power of the three DAPDs under -3 V bias voltages. The saturation power of 2-cell DAPD is 1.2 dBm at an average photocurrent of 16.1 mA. The values of 4- and 8-cell schemes are -2.3 dBm at an average photocurrent of 11.4 mA and -11.1 dBm at an average photocurrent of 10 mA, respectively. To be noted, the RF output powers of three DAPDs are decreased with the cell number, due to the power losses from the distributed absorption structure. Especially for the 8-cell DAPD, the saturation power under 10 GHz signal frequency is also limited by its 3dB bandwidth.
A summary of the device performances is shown in Table 2. The responsivity decreases with the increasing cell number, due to the increasing optical loss of the optical splitter, which is 1×2 MMI in the DAPDs. Obviously, the 2-cell DAPD has the best bandwidth and responsivity.
The saturation characteristics can be divided into two situations, i.e. the DC saturation and RF saturation. In the DC saturation results, the optical saturation power is doubled while the cell number increases from 2 to 4. However, 2- and 4-cell DAPDs have the same saturation photocurrent, due to the different current loss. According to the simulation results at Section 2, if the current loss of 4-cell DAPD is equal to the loss of 2-cell scheme, the saturation photocurrent should be 46.58 mA. Even though 8-cell DAPD can handle quite large optical power, the corresponding photocurrent is very low due to the large current loss (3.8 dB from Section 2). On the other hand, in the RF saturation results, the 2-cell DAPD has the largest RF saturation power due to its lowest current loss. Furthermore, the bandwidth degradation of the three DAPDs is shown in Fig. 9, as a function of photocurrents. The 2-cell DAPD has the largest bandwidth in the whole range of photocurrents.
An expanded PIN junction structure is implemented in the DAPD. This junction structure introduces the inevitable current offset at the Si substrate, and this turns into the current loss in aspect of output photocurrent. While the cell number n is 1, the loss can be canceled and the DAPD becomes the conventional one, which cannot operation in high-power condition . While n≥2, according to the theoretical and experimental results, we can conclude that the optimal cell number is 2, and the 2-cell DAPD is demonstrated to have the high-speed and high-power performance. Furthermore, this junction can be seamlessly implemented as a single state of PD array with a traveling wave electrode configuration for higher RF saturation performance.
We expand the concept of conventional PIN structure to enhance the power handling capability of Ge PD. The proposed junction structure can be designed as a multiple intrinsic absorption region on a mutual P-doped substrate. The n-cell DAPD is implemented with the expanded junction structure and lumped electrodes for high-speed and high-power applications. Based on this design, the optimal cell number has been discussed theoretically and the current loss from the junction structure has been revealed. The 2-, 4- and 8-cell DAPDs have been fabricated and measured, and the 2-cell one has the optimal performance with high RF saturation photocurrent (16.1 mA) and 3 dB bandwidth (49.3 GHz). Finally, Table 3 gives a comparison of the high-power waveguide-coupled Ge PDs in literatures and our work. The current density is introduced to be the figure of merit (FOM) while characterizing the DC saturation performance. On the other hand, the 1 dB compression current of PDs is usually used to characterize the RF saturation performance and can be accurately measured without considering the RF loss of system. Thus, we introduce a bandwidth-current product as another FOM. The 2-cell DAPD has the best bandwidth and current density among all PDs. Meanwhile, the DAPD can be easily implemented as a single state of PD array with a traveling wave electrode configuration for improving the bandwidth-current product, and it is very attractive to applications such as integrated microwave photonics systems, thanks to the low cost and zero-change fabrication process in the standard silicon photonics platform.
National Key Research and Development Program of China (2019YFB1803801, 2019YFB2203502); National Natural Science Foundation of China (61775073, 61922034); Huazhong University of Science and Technology (2018QYTD08).
The authors declare no conflicts of interest.
1. A. J. Seeds, “Microwave photonics,” IEEE Trans. Microwave Theory Tech. 50(3), 877–887 (2002). [CrossRef]
2. H. Ito, S. Kodama, Y. Muramoto, T. Furuta, T. Nagatsuma, and T. Ishibashi, “High-speed and high-output InP-InGaAs unitraveling-carrier photodiodes,” IEEE J. Sel. Top. Quantum Electron. 10(4), 709–727 (2004). [CrossRef]
3. Q. Li, K. Sun, K. Li, Q. Yu, P. Runge, W. Ebert, A. Beling, and J. C. Campbell, “High-Power Evanescently Coupled Waveguide MUTC Photodiode With >105-GHz Bandwidth,” J. Lightwave Technol. 35(21), 4752–4757 (2017). [CrossRef]
4. D. Zhou, Y. Yu, Y. Yu, and X. Zhang, “Parallel radio-frequency signal-processing unit based on mode multiplexed photonic integrated circuit,” Opt. Express 26(16), 20544–20549 (2018). [CrossRef]
5. D. Zhou, C. Sun, Y. Lai, Y. Yu, and X. Zhang, “Integrated silicon multifunctional mode-division multiplexing system,” Opt. Express 27(8), 10798–10805 (2019). [CrossRef]
6. C. T. DeRose, D. C. Trotter, W. A. Zortman, A. L. Starbuck, M. Fisher, M. R. Watts, and P. S. Davids, “Ultra compact 45 GHz CMOS compatible Germanium waveguide photodiode with low dark current,” Opt. Express 19(25), 24897–24904 (2011). [CrossRef]
7. M. M. P. Fard, G. Cowan, and O. Liboiron-Ladouceur, “Responsivity optimization of a high-speed germanium-on-silicon photodetector,” Opt. Express 24(24), 27738–27752 (2016). [CrossRef]
8. G. Chen, Y. Yu, S. Deng, L. Liu, and X. Zhang, “Bandwidth improvement for germanium photodetector using wire bonding technology,” Opt. Express 23(20), 25700–25706 (2015). [CrossRef]
9. H. Chen, P. Verheyen, P. De Heyn, G. Lepage, J. De Coster, S. Balakrishnan, P. Absil, W. Yao, L. Shen, G. Roelkens, and J. Van Campenhout, “-1 V bias 67 GHz bandwidth Si-contacted germanium waveguide p-i-n photodetector for optical links at 56 Gbps and beyond,” Opt. Express 24(5), 4622–4631 (2016). [CrossRef]
10. A. Novack, M. Gould, Y. Yang, Z. Xuan, M. Streshinsky, Y. Liu, G. Capellini, A. E.-J. Lim, G.-Q. Lo, T. Baehr-Jones, and M. Hochberg, “Germanium photodetector with 60 GHz bandwidth using inductive gain peaking,” Opt. Express 21(23), 28387–28393 (2013). [CrossRef]
11. E. Rouvalis, F. N. Baynes, X. Xiaojun, L. Kejia, Z. Qiugui, F. Quinlan, T. M. Fortier, S. A. Diddams, A. G. Steffan, A. Beling, and J. C. Campbell, “High-Power and High-Linearity Photodetector Modules for Microwave Photonic Applications,” J. Lightwave Technol. 32(20), 3810–3816 (2014). [CrossRef]
12. A. Beling, X. Xie, and J. C. Campbell, “High-power, high-linearity photodiodes,” Optica 3(3), 328–338 (2016). [CrossRef]
13. K. J. Williams and R. D. Esman, “Design Considerations for High-Current Photodetectors,” J. Lightwave Technol. 17(8), 1443–1454 (1999). [CrossRef]
14. M. Dentan and B. d. Cremoux, “Numerical simulation of the nonlinear response of a p-i-n photodiode under high illumination,” J. Lightwave Technol. 8(8), 1137–1144 (1990). [CrossRef]
15. A. Ramaswamy, M. Piels, N. Nunoya, Y. Tao, and J. E. Bowers, “High Power Silicon-Germanium Photodiodes for Microwave Photonic Applications,” IEEE Trans. Microwave Theory Tech. 58(11), 3336–3343 (2010). [CrossRef]
16. M. J. Byrd, E. Timurdogan, Z. Su, C. V. Poulton, N. M. Fahrenkopf, G. Leake, D. D. Coolbaugh, and M. R. Watts, “Mode-evolution-based coupler for high saturation power Ge-on-Si photodetectors,” Opt. Lett. 42(4), 851–854 (2017). [CrossRef]
17. Y. Zuo, Y. Yu, Y. Zhang, D. Zhou, and X. Zhang, “Integrated high-power germanium photodetectors assisted by light field manipulation,” Opt. Lett. 44(13), 3338–3341 (2019). [CrossRef]
18. M. Piels and J. E. Bowers, “40 GHz Si/Ge Uni-Traveling Carrier Waveguide Photodiode,” J. Lightwave Technol. 32(20), 3502–3508 (2014). [CrossRef]
19. C. Li, C. Xue, Z. Liu, H. Cong, B. Cheng, Z. Hu, X. Guo, and W. Liu, “High-responsivity vertical-illumination Si/Ge uni-traveling-carrier photodiodes based on silicon-on-insulator substrate,” Sci. Rep. 6(1), 27743 (2016). [CrossRef]
20. D. Zhou, Y. Yu, Y. Zuo, and X. Zhang, “Germanium Photodetector with Carrier Acceleration,” in Conference on Lasers and Electro-Optics, (Optical Society of America, 2018), SM2I.2.
21. D. Zhou, Y. Yu, N. Yang, and X. Zhang, “Germanium Photodetector With Alleviated Space-Charge Effect,” IEEE Photonics Technol. Lett. 32(9), 538–541 (2020). [CrossRef]
22. X. Luo, J. Song, X. Tu, Q. Fang, L. Jia, Y. Huang, T.-Y. Liow, M. Yu, and G.-Q. Lo, “Silicon-based traveling-wave photodetector array (Si-TWPDA) with parallel optical feeding,” Opt. Express 22(17), 20020–20026 (2014). [CrossRef]
23. L. Bogaert, K. Van Gasse, T. Spuesens, G. Torfs, J. Bauwelinck, and G. Roelkens, “Silicon photonics traveling wave photodiode with integrated star coupler for high-linearity mm-wave applications,” Opt. Express 26(26), 34763–34775 (2018). [CrossRef]
24. C.-M. Chang, J. H. Sinsky, P. Dong, G. de Valicourt, and Y.-K. Chen, “High-power dual-fed traveling wave photodetector circuits in silicon photonics,” Opt. Express 23(17), 22857–22866 (2015). [CrossRef]
25. G. Chen, Y. Yu, X. Xiao, and X. Zhang, “High speed and high power polarization insensitive germanium photodetector with lumped structure,” Opt. Express 24(9), 10030–10039 (2016). [CrossRef]
26. L. Y. Lin, M. C. Wu, T. Itoh, T. A. Vang, R. E. Muller, D. L. Sivco, and A. Y. Cho, “High-power high-speed photodetectors-design, analysis, and experimental demonstration,” IEEE Trans. Microwave Theory Tech. 45(8), 1320–1331 (1997). [CrossRef]
27. J. M. Lee, S. H. Cho, and W. Y. Choi, “An Equivalent Circuit Model for a Ge Waveguide Photodetector on Si,” IEEE Photonics Technol. Lett. 28(21), 2435–2438 (2016). [CrossRef]
28. M. Gould, T. Baehr-Jones, R. Ding, and M. Hochberg, “Bandwidth enhancement of waveguide-coupled photodetectors with inductive gain peaking,” Opt. Express 20(7), 7101–7111 (2012). [CrossRef]
29. T. Tzu, K. Sun, R. W. Costanzo, D. Ayoub, S. Bowers, and A. Beling, “Foundry-Enabled High-Power Photodetectors for Microwave Photonics,” IEEE J. Sel. Top. Quantum Electron. 25(5), 1–11 (2019). [CrossRef]