We present a two-element back-illuminated symmetric-connected uni-traveling-carrier photodiode array (SC-PDA) integrated with sub-wavelength-gratings based beam-splitter (SWGs-BS) for high power optoelectronic applications. The SC-PDA with SWGs-BS, a top-illuminated SC-PDA and a single PD were fabricated and tested. The proposed SC-PDA with SWGs-BS demonstrates a 3dB bandwidth of 23.8GHz@60mA, a saturation current of 87.9mA@12GHz and a maximum output RF power of ~16dBm@12GHz. Compared to the top-illuminated SC-PDA and the single PD, the proposed SC-PDA with SWGs-BS achieved high-power handling capability at low coupling complexity and requires no phase-matching techniques in the system.
© 2017 Optical Society of America
Applications such as high-speed receivers, optically-fed phased array antennas, high-power antennas in RoF systems and photonic oscillators have been creating increasing demands on high performance photodiodes (PDs) [1,2]. In the past few decades, many efforts have been made to improve the bandwidth, saturation current and output power of photodiodes by designing new epitaxial structures. Several solutions, such as the partially-depleted-absorber PD , the dual-absorption PD , the dual-depletion region PD  and the uni-traveling-carrier (UTC) PD , has been brought up to optimize the PD’s high-speed performance under high-power illumination light. Since the InP based UTC-PDs could achieve high bandwidths by eliminating the slow transportation of holes that dominates the carrier transit times in traditional PIN-PDs, it has become an important structure for high-speed high-power photodiodes. Several modifications such as adding a highly doped cliff layer  and slightly dope the collection layer [8,9] has further improved the saturation power of UTC-PDs.
Meanwhile, the trade-offs between the figures of merit of a single PD has always been limiting its overall performances. For the single PDs, the mesa was often designed with a small size of area and a thin thickness to obtain a low capacitance and a small carrier transit time. But reducing the size of area of the mesa leads to a high photocurrent density and result in a low saturation current. Distributed photodiode arrays (PDAs) with large bandwidths and high saturation photocurrents have been proposed to overcome the trade-off [2,10-11]. By evenly splitting the high-power illumination light and feeding into several discrete photodiodes, the photocurrents are summed while keeping the total capacitance equivalent to the capacitance of a single photodiode . In this configuration, signals produced by each PD element have different delays and losses while propagating along the transmission lines. The difference in propagation lengths requires additional compensation technologies to solve the phase mismatch problem.
Two-element symmetric-connected photodiode array (SC-PDA) has been proposed in our previous work. Comparison between the symmetric-connected photodiode array and traditional traveling-wave photodiode array (TW-PDA) has been made  (referred to as parallel-connected PDA and serial-connected PDA, respectively). The PD elements in SC-PDA share the same signal propagation length and have no phase-mismatches, as long as the phases of the optical signals were matched. On the other hand, the SC-PDA suffers from the accumulated capacitance which affects its high-speed performance, and a single PD with a larger size of active area could achieve the same bandwidth and saturation current as the SC-PDA. However, the SC-PDA has a larger surface-to-volume ratio that is in benefit of the heat dissipation. The heat issue is critical since the SC-PDA is supposed to work in high-power applications. Still, the PDA requires several optical fibers or an optical fiber array for the coupling of light and thus increases the coupling complexity of the device.
In this paper, for the first time, we present a novel two-element normal-incidence back-illuminated symmetric-connected UTC photodiode array integrated with sub-wavelength gratings based beam-splitter (SWGs-BS). The SOI-based SWGs-BS splits the incident light into two beams and focus them at the PD elements. This design simplifies the coupling complexity and automatically matches the phase of the optical signals.
2. Device design and fabrication
The schematic diagram of the back-illuminated SC-PDA integrated with SWGs-BS is shown in Fig. 1. The device consists of a back-illuminated SC-PDA with an InP based UTC-type epitaxial structure, a SOI based SWGs-BS and a layer of BCB as bonding agent. The detailed design and fabrication process are described as follows.
2.1 Photodiode array
The epitaxial structure of the PD elements is grown on a semi-insulating InP substrate, as is shown in Fig. 2(a). The absorption layer thickness was chosen as 600nm to ensure a high responsivity for high RF power output. Graded doping was adopted in the absorption layer to build an electric field in the absorption layer and accelerate the photo-generated electrons. A 20nm highly doped InP cliff layer was adopted to balance the trade-off between the performance of the device and the fabrication difficulty. The InP collecting layer was slightly n-doped to provide charge compensation . In order to make performance comparisons, a top-illuminated SC-PDA (shown in Fig. 2(b)), a back-illuminated SC-PDA (shown in Fig. 2(c)) and a single UTC PD were fabricated. The diameter of the PD elements in the SC-PDAs and the single PD are 40μm. A coplanar-waveguide-like (CPW-like) electrode is used to connect the PD elements in the SC-PDAs. The width of the center conductor and the gap between the center conductor and ground pad are 20μm and 80μm, respectively. The spacing between the adjacent PD elements is 250μm.
2.2 Sub-wavelength gratings based beam-splitter
In our previous works, detailed design process of 2D non-periodic SWGs-BS with focusing ability has been reported . The previous SWGs-BS was designed with a grating layer thickness of 1.2μm and a period of 700nm. Considering the fabrication process, the SWGs-BS used in this work was redesigned with a grating layer thickness of 650nm. The blocks of the SWGs-BS have a same period of 600nm and different duty cycles. Simulation results of the redesigned SWGs-BS with an area of ~100μm2 is shown in Fig. 3. The transmittance of the SWGs-BS was calculated to be 82.236%. Then the size of area of the SWGs-BS was designed as 250μm × 250μm to maximum the coupling efficiency. The designed focal length and the distance between the focal spots are 200μm and 250μm, respectively. In the calculating and simulating process, the grating blocks were surrounded with materials with a refractive index of 1.5, which is approximately the refractive index of silicon dioxide and BCB. Considering the refractive index of the InP substrate, the focal length after bonding was calculated as 480μm.
2.3 Fabrication process
The mesa structures of the PD elements were fabricated with standard semiconductor fabrication processes including photo-lithography, wet chemical etching and magnetron sputtering of the contact electrodes. After the photodiode mesas and contact electrodes were formed, a layer of polyimide was spin coated on the wafer as a passive layer. Top electrodes were then evaporated on top of the passive layer to connect the PD elements. After the fabrication process, the InP substrate was mechanically thinned to 480μm and polished. Finally, the wafer with the back-illuminated SC-PDA was diced into 1mm x 1mm chip.
The SWGs-BS was fabricated by electron-beam lithography and inductively coupled plasma etching using C4F8 and SF6 on a SOI wafer. The thicknesses of top silicon and buried silicon dioxide are 650nm and 500nm, respectively. A 1mm x 1mm frame surrounding the SWGs-BS was also etched for alignment in the bonding process. Then the SOI chip was mechanically thinned to 400μm and polished. The fabricated SWGs-BS is shown in Fig. 4(a). The beam-splitting performance of the SWGs-BS was tested before bonding. Figure 4(b) shows the measurement results. The measurement results were obtained by moving an optical fiber along the line that connects the two focal spots and measuring the received light power while the SWGs-BS was back-illuminated. The distance between the tip of the optical fiber and the SWGs-BS was 200μm. The measured distance between the focal spots is approximately 200μm, which is slightly different from the designed one. This was probably caused by the change of refractive index around the SWGs-BS (from 1.5 to 1 of air). The measured power distribution at the two focal spots is approximately 0.9:1. The inset of Fig. 4(b) shows the intensity distribution of the transmitted light, which was measured by a slit scanner at a relatively longer distance. The captured image indicates two transmitted beam split by the SWGs-BS.
After the tests, a BCB layer was spin coated on the SWGs-BS chip. The 1mm x 1mm back-illuminated SC-PDA chip was aligned to the 1mm x 1mm frame around the SWGs-BS. In this case, the center of the SWGs-BS would be aligned with the middle of the two PD elements. The BCB was then cured at 250°C for two hours to finish the bonding process. Neither the InP substrate nor the SOI substrate was coated with anti-reflection coating.
3. Results and discussion
The fabricated SC-PDAs and single PD were tested on wafer. The back-illuminated SC-PDA with SWGs-BS was measured by aligning a fiber collimator to the center bottom of the chip. The top-illuminated SC-PDA was measured by aligning two lensed fiber probes to the top windows of the PD elements. An optical fiber splitter was used to evenly split the power of modulated/heterodyned source lights and fed them into the two fiber probes. A GSG150 microwave probe was used to extract the photo-generated current. The phases of the two optical signals were matched by aligning both fibers to a large area single photodiode and adjust one lensed fiber vertically until the coherent subtraction measured at the output pad disappeared. Then the output power of the two fibers were balanced by an optical tunable attenuator in the other channel. In the RF measurements, the reverse bias voltages were applied through a bias-tee inserted in the RF link.
3.1 DC characteristics
The dark currents of the SC-PDAs and the single PD were measured as ~20nA at −3V bias. Figure 5 shows the measurement results of the SC-PDAs at −3V bias with 1550nm incident light. The measured responsivities of the top-illuminated SC-PDA and the back-illuminated SC-PDA with SWGs-BS were 0.438A/W and 0.179A/W, respectively. The results indicates that a 4.55dB loss was introduced by the proposed coupling technique. The loss was caused by two major effects. On one hand, the interfaces between different materials along the light propagation route brought in many reflection losses. On the other hand, the size of the focal spot of the transmitted lights is larger than the size of the PD mesas. The oblique incidence of the transmitted light and the reflection at the top contact metal lengthened the absorption length and compensated some of the losses.
3.2 Frequency response
The frequency responses of the fabricated devices were carried out by a high-speed modulation system. The laser light generated by a 1550nm CW laser source was amplified and modulated by a 40GHz modulator. The modulated light was then coupled into a tunable optical attenuator for light power tuning. The electric modulation signals were generated by a 40GHz network analyzer and the RF response of the PD was coupled into the network analyzer for S21 measurement.
The relative frequency responses of the fabricated devices measured at −3V bias and photocurrents of 1mA and 60mA are shown in Fig. 6 and Fig. 7, respectively. The 3dB-bandwidths of top-illuminated SC-PDA, the back-illuminated SC-PDA with SWGs-BS and the single PD measured at 1mA are 15.2GHz, 15.4GHz and 19.6GHz, respectively. It’s obvious that the accumulated capacitances of the PD elements limited the high-speed performance of the SC-PDAs.
When we increase the light power to get a 60mA output photocurrent, as is shown in Fig. 7, the response of the single PD saturated and the 3dB-bandwidth dropped to 2.6GHz. Benefiting from the cliff-layered epitaxial structure of the PD elements, the response of the SC-PDAs, on contrary, went up to 23.2GHz and 23.8GHz, respectively. It should be noted that in order to get the same output photocurrents, the back-illuminated SC-PDA with SWGs-BS requires larger incident light power than the top-illuminated SC-PDA and the single PD.
Figure 8 shows the relative frequency responses measured when the phases of the two optical feeds were detuned for the top-illuminated SC-PDA. A periodical pattern could be observed on the frequency response curve of the top-illuminated SC-PDA. This was caused by the phase-mismatches between the signals generated by the two PD elements. For the back-illuminated SC-PDA with SWGs-BS, the light propagation lengths were always the same as long as the source light was aligned to the center of the SWGs. So the proposed SC-PDA with SWGs-BS requires no additional phase matching techniques.
3.3 Saturation characteristics
The saturation properties of the fabricated devices were measured with an optical heterodyne setup. The beat frequency and bias voltage were set at 12GHz and −3V, respectively. As is shown in Fig. 9, the SC-PDAs demonstrates similar saturation behaviors. The saturation currents measured for the back-illuminated SC-PDA with SWGs-BS and the top-illuminated SC-PDA are 86.8mA and 87.9mA, respectively, while the single PD has a saturation current of 46.7mA. The highest output RF power of the SC-PDAs were ~16dBm before failure, which is ~6dB bigger than the saturation power of the single PD. However, if the beat frequency was set at the dip frequencies in Fig. 8, the top-illuminated SC-PDA would produce much lesser RF power than the back-illuminated SC-PDA with SWGs-BS. Since the devices were measured on wafer with no cooling mechanisms, the measured saturation current and maximum output RF power should not be at the best values that the devices could produce.
The fabricated SC-PDA with SWGs-BS achieved a higher saturation current compared to the single UTC-PD, and it requires no additional phase matching techniques. Still, there are some drawbacks in the proposed device. First of all, a 4.5dB coupling loss was introduced by the proposed coupling technique and the SWGs-BS, which would increase the optical power budget of the system. The insertion loss was introduced mainly by the interface reflections and the large size of the separated light beams. The interface reflections could be reduced by a finer polishing of the InP and SOI substrates and anti-reflection coatings. The spot size of the separated beams could be reduced by an optimized design of the SWGs-BS. Meanwhile, the light scattering in the substrates might also take part in the loss. This can be reduced by designing a SWGs-BS with a smaller focal length and further thinning the InP and SOI substrates. So the coupling efficiency is further improvable. Furthermore, although the saturation and heat dissipation of single photodiodes are limited and discrete photodiode array is currently an efficient way to greatly increase the saturation photocurrent of a photodetector, additional costs and fabrication complexity was introduced by the design of the SC-PDA with SWGs-BS.
In this paper, a two-element SC-PDA with SWGs-BS is proposed for the first time for high power handling and simple coupling. A back-illuminated SC-PDA with SWGs-BS, a top-illuminated SC-PDA and a single PD with a same UTC epitaxial structure were fabricated and tested. The proposed device demonstrates higher power handling capability than the single PD and lower coupling complexity than the top-illuminated SC-PDA. For further improvements on high-power high-speed performance, the photodiode array could be designed with multiple PD elements symmetrically arranged around the output pad and a sub-wavelength gratings based beam-splitter that split the incident light into multiple transient light beams. By optimizing the coupling efficiency, the proposed design of heterodyne integrated photodiode arrays has a great potential in high-power high-speed applications on silicon platform.
National Nature and Science Foundation of China (NSFC) (61574019, 61674018, 61674020); Fund of State Key Laboratory of Information Photonics and Optical Communications; Specialized Research Fund for the Doctoral Program of Higher Education of China (20130005130001).
This work was supported, in part, by the Joint Laboratory of Quantum Optoelectronics and the Theory of Bivergentum and Beijing International Scientific and Technological Cooperation Base of Information Optoelectronics and Nano-heterogeneous Structure.
References and links
1. A. J. Seeds and K. J. Williams, “Microwave Photonics,” J. Lightwave Technol. 24(12), 4628–4641 (2006). [CrossRef]
3. X. Li, N. Li, X. Zheng, S. Demiguel, J. C. Campbell, D. Tulchinsky, and K. Williams, “High-Speed High-Saturation-Current InP/In0.53Ga0.47As Photodiode with Partially Depleted Absorber,” in Optical Fiber Communication Conference, Technical Digest (Optical Society of America, 2003), paper WF3.
4. R. Sankaralingam and P. Fay, “Drift-enhanced dual-absorption PIN photodiodes,” IEEE Photonics Technol. Lett. 17(7), 1513–1515 (2005). [CrossRef]
5. F. Jeffenberger and A. Joshi, “Ultrafast, dual-depletion region, InGaAs/InP p-i-n detector,” J. Lightwave Technol. 14(8), 1859–1864 (1996). [CrossRef]
6. T. Ishibashi, N. Shimizu, S. Kodama, H. Ito, T. Nagatsuma, and T. Furuta, “Uni-Traveling-Carrier Photodiodes,” in Ultrafast Electronics and Optoelectronics, M. Nuss and J. Bowers, eds., Vol. 13 of OSA Trends in Optics and Photonics Series (Optical Society of America, 1997), paper UC3.
7. X. Wang, N. Duan, H. Chen, and J. Campbell, “InGaAs-InP photodiodes with high responsivity and high saturation power,” IEEE Photonics Technol. Lett. 19(16), 1272–1274 (2007). [CrossRef]
8. Z. Li, H. Pan, H. Chen, A. Beling, and J. C. Campbell, “High-saturation-current modified uni-traveling-carrier photodiode with cliff layer,” IEEE J. Quantum Electron. 46(5), 626–632 (2010). [CrossRef]
9. J. Li, B. Xiong, Y. Luo, C. Sun, J. Wang, Z. Hao, Y. Han, L. Wang, and H. Li, “Ultrafast dual-drifting layer uni-traveling carrier photodiode with high saturation current,” Opt. Express 24(8), 8420–8428 (2016). [CrossRef] [PubMed]
10. A. Beling, H. Chen, H. Pan, and J. C. Campbell, “High-Power Monolithically Integrated Traveling Wave Photodiode Array,” IEEE Photonics Technol. Lett. 21(24), 1813–1815 (2009). [CrossRef]
11. A. Beling, H. Bach, G. G. Mekonnen, R. Kunkel, and D. Schmidt, “High-Speed Miniaturized Photodiode and Parallel-Fed Traveling-Wave Photodetectors Based on InP,” IEEE J. Sel. Top. Quantum Electron. 13(1), 15–21 (2007). [CrossRef]
12. C. Goldsmith, G. Magel, and R. Baca, “Principles and performance of traveling-wave photodetector arrays,” IEEE Trans. Microw. Theory Tech. 45(8), 1342–1350 (1997). [CrossRef]
13. J. Fei, Y. Huang, T. Liu, X. Ma, X. Duan, K. Liu, and X. Ren, “Performance Comparison between Serial-Connected and Parallel-Connected Photodiode Array,” in Conference on Lasers and Electro-Optics, OSA Technical Digest (online) (Optical Society of America, 2017), paper JW2A.129. [CrossRef]
14. K. J. Williams and R. D. Esman, “Design Considerations for High-Current Photodetectors,” J. Lightwave Technol. 17(8), 1443–1454 (1999). [CrossRef]
15. Y. Wang and Y. Huang, “Novel Beam Splitter Based on 2D Sub-wavelength High-Contrast Gratings,” in Asia Communications and Photonics Conference 2016, OSA Technical Digest (online) (Optical Society of America, 2016), paper AF2A.58. [CrossRef]