A novel backside-illuminated mesa-structure dual-drifting layer (DDL) uni-traveling-carrier photodiode (UTC-PD) is reported to demonstrate high-power performance at sub-THz frequencies. The DDL structure consists of a velocity overshoot layer and a velocity saturation layer, formed by inserting a 20 nm p-type cliff layer into the thick depletion region. In the overshoot layer, photo-generated electrons drift at overshoot velocity under the carefully designed electric field profile, thus resulting in a short electron transit time. The saturation layer serves as a voltage sacrificing layer to enable high bias voltage operation, which leads to alleviated load voltage swing effect, as well as improved saturation performance. Our DDL UTC-PD exhibits a 3-dB bandwidth of 106 GHz with a responsivity of 0.17 A/W under a wide bias voltage range from 4 to 8 V. The photocurrent reaches up to 28 mA, corresponding to an output power of 7.3 dBm at 105 GHz.
© 2016 Optical Society of America
With the exponential increase in the required data rate in many microwave and millimeter wave (MMW) applications such as sensor networks, radar and broadband wireless-over-fiber systems , a radio-frequency (RF) bandwidth famine is anticipated, and researches exploring the MMW (30~300 GHz) have recently attracted great interest. This results in an increasing demand for high performance devices capable of high-power MMW signal generation and detection [2,3 ]. In an RoF communication system, high-speed photodiodes (PDs) usually determine the maximum allowable operating frequency and the dynamic range of the entire system . Simultaneous improvement in the bandwidth (>100 GHz) and output power (>1 mW) of the PD is crucial to implementing a high performance RoF system. Edge-illuminated waveguide photodetectors suffer from limited saturation current due to the high current density near the input facet of the waveguide. To achieve both high-frequency operation and high power performance, surface-illuminated photodetectors are preferable. The key to fabricating a high-speed PD with a high saturation photocurrent is to downscale the mesa area as well as the depletion region thickness [5,6 ]. A thinner depletion region means a shorter carrier transit time, but also results in a larger junction capacitance. It is thus necessary to reduce the device mesa area so as to maintain a low junction capacitance, which, however, increases the difficulty of fabrication process.
Recently, (modified) uni-traveling carrier PDs (UTC-PDs) [7–9 ] and near-ballistic UTC-PDs (NBUTC-PDs) [10,11 ] have been proposed to overcome the trade-off between the resistance-capacitance (RC) delay and the carrier transit time limited bandwidth, and excellent bandwidth and saturation performance have been demonstrated. However, as the output photocurrent increases, a large voltage swing induced by the alternating current delivered to the load will be superimposed onto the PD bias. Therefore, electric field in the depletion region will redistribute, causing performance degradation in such UTC-PDs. In order to suppress such load voltage swing effect, it is possible to adopt a load resistance less than 50 Ω , but it also reduces the output RF power. Another method is to increase the reverse bias voltage to achieve high saturation performance. For example, a thin p-doped layer was inserted beneath the thin depletion layer to achieve wide bandwidth performance under a bias voltage of about 5 V . However, further increase in the bias voltage would cause bandwidth degradation due to deviation from the velocity overshoot condition.
To achieve both high-speed and high saturation power performance, a novel dual-drifting layer (DDL) UTC-PD has been proposed and fabricated . In the DDL UTC-PD, a thin p-doped layer is inserted into the middle of a thick depletion layer (>600 nm), and by optimizing the electric field profile, significant electron velocity overshoot in the depletion layer can be maintained even under high reverse bias. In this paper, we report a DDL UTC-PD with a responsivity of 0.17 A/W and a high saturation photocurrent of 28 mA. The 3-dB bandwidth of the device reaches up to 106 GHz. Thanks to their high bias voltage capability, DDL UTC-PDs working in conjunction with a standard 50 Ω load will allow further improvement of the high power performance of UTC-PDs in sub-THz regime.
2. Device structure
The epitaxial layers of the DDL UTC-PD are shown in Fig. 1(a) , which include a 300-nm-thick graded doped p-InGaAs photoabsorption layer and a 630-nm-thick dual-drift depletion layer. The dual-drifting layer mainly consists of a 310-nm-thick velocity overshoot layer and a 300-nm-thickness InP velocity saturation layer, separated by a 20-nm-thick depleted p-type cliff layer in between. An additional 10 nm n-type cliff layer is placed next to the 20-nm-thick InGaAsP transition layer to elevate the electric field at the interface between the absorption layer and the transition layer. The thickness of the overshoot layer is optimized to take full advantage of the overshoot effect . Compared with the epitaxial structure depicted in , a thinner photoabsorption layer is adopted here to further reduce the electron transit time. The doping density of the 20-nm p-type cliff layer is also optimized to tailor the electric field profile in the dual-drifting layer. As a result, the new UTC-PD with DDL structure has the potential to work under high bias voltage, thus sustaining high saturation current without suffering from the load voltage swing effect, as well as obtaining a bandwidth reaching up to sub-THz regime.
Nextnano was adopted to simulate the electric field distribution in our structure. During the simulation, the photo-carriers and the voltage drop over the load resistance have to be taken into consideration. For incident light with 100% modulation depth, the induced photocurrent can be described as i(t) = i 0 + i 0 ejωt, which consists of a direct current (DC) component of i 0 and an alternating current (AC) component i 0 ejωt. The saturation performance is determined by the lowest electric field distribution within the depletion region during an AC period, corresponding to the peak current point. Furthermore, the photo-carrier density in the overshoot layer and saturation layer under different voltages is different, as a result of the different average electron velocity in the two regions. In our simulation, the average overshoot velocity is taken to be 4.0 × 107 cm/s , while the saturation velocity decreases along with the elevated voltage .
Figure 2 compares the simulated electric-field profile within an UTC-PD and a DDL UTC-PD under different reverse biases. As can be seen, the average electric field in the depletion region of the UTC-PD remains relatively high for all biases, increasing from ~50 kV/cm under 3 V to ~150 kV/cm under 9 V. Under such a high electric field, the overshoot velocity can only be sustained within a short distance, and electrons will drift at a saturation velocity in most part of the depletion region. On the other hand, electric field in the DDL UTC-PD is controlled by the 20-nm p-type cliff layer with a proper doping density to produce a segmented profile. The average electric field within the first drifting layer varies from 20 to 50 kV/cm, which is in the range of critical electric field for velocity overshoot [10,11 ]. As a result, the photogenerated electrons will first drift at overshoot velocity across the low electric field layer, and then travel at saturation velocity in the 300-nm high electric field layer. The DDL structure takes full advantage of the velocity overshoot effect, and a thick saturation layer is further used as the voltage suffering layer to realize high bias voltage operation. Compared with conventional UTC-PDs, where electrons drift at saturation velocity in almost the entire depletion region, the transit time in a DDL UTC-PD is significantly reduced.
3. Device fabrication and performance
Backside-illuminated PDs with 6-μm diameter are fabricated following a fabrication procedure similar to that described in . The double-mesa structure is defined by inductively coupled plasma (ICP) dry-etching processes, with the assistance of wet-etching. Magnetron sputtering and lift-off are adopted to form Ti(20nm)/Pt(20nm)/Au(200nm) p-electrode and Ni(20nm)/Au(100nm) n-electrode. An 800-nm-thick SiO2 layer is deposited with plasma enhanced chemical vapor deposition (PECVD) to reduce the surface leakage current and parasitic capacitance. Coplanar waveguide (CPW) electrode is then sputtered on top of the SiO2, followed by a 1.5-μm-thick Au by electroplating for high-speed measurement, so the p- and n-electrodes are covered by the thick CPW electrode. As a final step, the wafer is polished down to 120 μm, and a 215-nm-thick SiNx layer is deposited onto the backside to reduce reflection of the incident light. Schematic view of the device is shown in Fig. 3(a) . The incident light enters the PD from the backside and is reflected by the top p-electrode, thus the light goes through the absorption layer twice.
Figure 3(b) shows the current-voltage curve of the DDL UTC-PD without illumination. The dark current is less than 5 nA under 6 V reverse bias, and about 700 nA under 9 V reverse bias. The series-resistance is reduced to 8 Ω by optimizing the device structure and the rapid thermal annealing treatment of the p- and n-mesa electrodes.
The frequency responses and output RF power of DDL-UTC PD are measured with a two-laser heterodyne system. Three different MMW power sensor heads are employed to cover the frequency range from dc to 50 GHz, V-band (50-75 GHz) and W-band (75-110 GHz), respectively. The measured frequency responses under 2 and 6 V reverse bias voltages are plotted in Fig. 4(a) , with a fixed output photocurrent of 10 mA. The frequency dependent loss of the power sensor heads and probes used during measurement has been carefully calibrated. As can be seen, the 3-dB bandwidth of the DDL-UTC PD exceeds 106 GHz. The DDL-UTC PD exhibits a responsivity as high as 0.17 A/W, which is much higher than the reported UTC-PDs with sub-THz bandwidth . Figure 4(b) shows the measured bandwidth under different reverse biases at a photocurrent of 10 mA. The 3-dB bandwidth for 2 V reverse bias is relatively low, as the depletion region of the device is not completely depleted. Obvious improvement in bandwidth performance occurs when the reverse bias is increased to 4 V. No degradation in bandwidth is observed when the voltage is increased to 8 V, since the electric field in the overshoot layer is still within the range for velocity overshoot. The frequency response of the DDL UTC-PD is different from that of previously reported UTC-PDs [11,15 ], where the bandwidth degrades seriously under 5 V reverse bias voltage. The high bias voltage capability of DDL UTC-PDs allows improved high power performance, as it helps alleviate the voltage swing effect.
The output RF power versus photocurrent of the 6-μm diameter DDL UTC-PD at 105 GHz frequency under different reverse biases are plotted in Fig. 5 . The photocurrent at 1-dB compression point is 16, 23 and 28 mA under 2, 3 and 3.8 V reverse bias, respectively, corresponding to an output RF power of 0.4, 5.5 and 7.3 dBm. It is clearly seen that the output photocurrent reaches up to 28 mA at the frequency of 105 GHz. To our knowledge, this is the highest saturation photocurrent for devices exhibiting a bandwidth exceeding 100 GHz when connected to a standard 50 Ω load, without adopting temperature control or flip chip bonding technology. The 1-dB compression point under higher reverse bias has not been measured, as a result of thermal damage under high photocurrent. Thermal failure can be overcome by flip-chip bonding technique. For example, flip-chip bonding on AlN for better heat dissipation can be adopted to reduce the Joule heating under high reverse bias , and higher saturation photocurrent can be obtained for the high-speed PD .
4. Analysis and discussion
In order to investigate whether the bandwidth of the DDL UTC-PD is resistance-capacitance (RC) constant or carrier transit time limited, an equivalent-circuit model is adopted to study the frequency response, as shown in Fig. 6(a) . Figure 6(b) shows the measured and simulated S22 reflection coefficients of the device under 6 V reverse bias. Fairly good agreement between the simulated and measured results has been obtained in the frequency range from 50 MHz to 40 GHz, which is limited by the bandwidth of our vector network analyzer. During the fitting progress, the entire 630-nm depletion region is divided into the 20-nm InGaAsP depletion layer and the 610 nm InP depletion layer according to the different residual charge densities . Following the fitting process detailed in , we can extract the bulk material resistance R 1 = 3 Ω, and the p- and n-contact resistance R 2 = 5 Ω. Ru = 20 kΩ and Cu = 120 fF are the equivalent resistance and capacitance of the 20-nm InGaAsP depletion region, respectively. Rj = 354 Ω and Cj = 4 fF are the resistance and capacitance of the 610 nm InP depletion layer shown in Fig. 1(a), respectively. Cp = 16 fF is the parasitic capacitance of the p-electrode. The parasitic inductance and capacitance of the CPW electrodes are assumed to be Lc = 0.1 nH and Cc = 2 fF, respectively.
The alternating photocurrent flowing to the external circuit (Region 2) is controlled by the time delay RtCt circuit (Region 1), which models the transit time response of the photodetector. The fitting results shown in Fig. 4(a) correspond to RtCt = 500 fF·Ω. Accordingly, the carrier transit time limited bandwidth ft = 1/(2πRtCt)  is found to be 318 GHz. On the other hand, the RC-limited 3-dB bandwidths fRC is about 140 GHz, extracted from the S21 parameter of Region 2 in the equivalent circuit model. We can thus conclude that the measured bandwidth of the DDL UTC-PD is mainly limited by the RC constant of the device. In addition, the transit time 2πRtCt of electrons is 3.1 ps for bias voltage higher than 4 V, as shown in Fig. 7 . Considering the drifting transportation of electrons in the absorption layer , the average drifting velocity in the 630-nm thick depletion region is estimated to be about 3.0 × 107 cm/s. This high drifting velocity is attributed to the novel dual-drifting layer structure, and it also helps improve the saturation performance of the device due to the lower carrier density in the high velocity region.
From the above discussion, we conclude that electrons maintain a high drifting velocity in the depletion region, thus allowing the DDL UTC-PD to achieve high bandwidth performance in a wide bias voltage range from 3 to 8 V. The capability to operate under high bias voltage is also beneficial to the high output photocurrent, and helps overcome the bandwidth deterioration under elevated voltage in previously reported UTC-PDs.
In this paper, we have presented a novel DDL structure for high performance backside illuminated UTC-PD providing high output power at sub-THz frequencies. The thick depletion region of the DDL UTC-PD is divided into an overshoot layer and a saturation layer by introducing a cliff layer in between. In the overshoot layer the photo-generated carriers travel at overshoot velocity, thus enhancing both the transit time limited bandwidth and the output RF power of the PD. The fabricated 1.55 μm InGaAs/InP mesa-structure DDL UTC-PDs exhibits an output photocurrent of 28 mA, corresponding to a power level exceeding 7.3 dBm for a 50 Ω load at 105 GHz. In addition, wide bandwidth operation with a 3-dB bandwidth beyond 106 GHz and a responsivity of 0.17 A has been demonstrated over a wide bias voltage range from 4 to 8 V. We anticipate that this work will provide competitive photodetector with extremely wide bandwidth and high power performance for optical communications and MMW applications.
This work was supported in part by the National Basic Research Program of China (Grant Nos. 2012CB315605, and 2014CB340002), the National Natural Science Foundation of China (NSFC) (Grant Nos. 61210014, 61321004, 61307024, 61574082 and 51561165012), the High Technology Research and Development Program of China (Grant No. 2015AA017101), the independent research program of Tsinghua University (2013023Z09N), and the Open Fund of State Key Laboratory on Integrated Optoelectronics (Grant Nos. IOSKL2012KF08 and IOSKL2014KF09).
References and links
1. J. Yao, “Microwave photonics,” J. Lightwave Technol. 27(3), 314–335 (2009). [CrossRef]
2. J. Capmany and D. Novak, “Microwave photonics combines two worlds,” Nat. Photonics 1(6), 319–330 (2007). [CrossRef]
3. J.-W. Shi, C.-B. Huang, and C.-L. Pan, “Millimeter-wave photonic wireless links for very high data rate communication,” NPG Asia Mater. 3(4), 41–48 (2011). [CrossRef]
4. S. Koenig, D. Lopez-Diaz, J. Antes, F. Boes, R. Henneberger, A. Leuther, A. Tessmann, R. Schmogrow, D. Hillerkuss, R. Palmer, T. Zwick, C. Koos, W. Freude, O. Ambacher, J. Leuthold, and I. Kallfass, “Wireless sub-THz communication system with high data rate,” Nat. Photonics 7(12), 977–981 (2013). [CrossRef]
5. T. Nagatsuma, H. Ito, and T. Ishibashi, “High-power RF photodiodes and their applications,” Laser Photonics Rev. 3(1–2), 123–137 (2009). [CrossRef]
6. J. Li, B. Xiong, C.-Z. Sun, Y. Luo, J. Wang, Z.-B. Hao, Y.-J. Han, L. Wang, and H.-T. Li, “Bandwidth improvement of high power uni-traveling-carrier photodiodes by reducing the series resistance and capacitance,” Chin. Phys. B 24(7), 078503 (2015). [CrossRef]
7. X. Wang, N. Duan, H. Chen, and J. C. Campbell, “InGaAs/InP photodiodes with high responsivity and high saturation power,” IEEE Photonics Technol. Lett. 19(16), 1272–1274 (2007). [CrossRef]
8. T. Shi, B. Xiong, C.-Z. Sun, and Y. Luo, “Back-to-Back UTC-PDs with high responsivity, high saturation current and wide bandwidth,” IEEE Photonics Technol. Lett. 25(2), 136–139 (2013). [CrossRef]
9. A. Beling, A.-S. Cross, M. Piels, J. Peters, Q. Zhou, J.-E. Bowers, and J. C. Campbell, “InP-based waveguide photodiodes heterogeneously integrated on silicon-on-insulator for photonic microwave generation,” Opt. Express 21(22), 25901–25906 (2013). [CrossRef] [PubMed]
10. Y.-S. Wu and J.-W. Shi, “Dynamic analysis of high-power and high-speed near-ballistic uni-traveling carrier photodiodes at w-band,” IEEE Photonics Technol. Lett. 20(13), 1160–1162 (2008). [CrossRef]
11. J.-W. Shi, F.-M. Kuo, and J.-E. Bowers, “Design and Analysis of ultra-high-speed near-ballistic uni-traveling-carrier photodiodes under a 50Ω load for high-power performance,” IEEE Photonics Technol. Lett. 24(7), 533–535 (2012). [CrossRef]
12. J. Li, B. Xiong, C.-Z. Sun, and Y. Luo, “Dual-drifting-layer uni-traveling carrier photodiode for wide bandwidth and high power performance,” in Conference on Lasers & Electro-Optics (CLEO) (2015), 1–2.
13. T. J. Maloney and J. Frey, “Transient and steady-state electron transport properties of GaAs and InP,” J. Appl. Phys. 48(2), 781–787 (1977). [CrossRef]
14. T. H. Windhorn, L. W. Cook, M. A. Haase, and G. E. Stillman, “Electron transport in InP at high electric fields,” Appl. Phys. Lett. 42(8), 725–727 (1983). [CrossRef]
15. J. Li, B. Xiong, C. Sun, D. Miao, and Y. Luo, “Analysis of frequency response of high power MUTC photodiodes based on photocurrent-dependent equivalent circuit model,” Opt. Express 23(17), 21615–21623 (2015). [CrossRef] [PubMed]
16. Z. Li, Y. Fu, M. Piels, H. Pan, A. Beling, J.-E. Bowers, and J. C. Campbell, “High-power high-linearity flip-chip bonded modified uni-traveling carrier photodiode,” Opt. Express 19(26), B385–B390 (2011). [CrossRef] [PubMed]
17. J.-W. Shi, F.-M. Kuo, C.-J. Wu, C. L. Chang, Ch.-Y. Liu, Ch. Y. Chen, and J.-I. Chyi, “Extremely high saturation current-bandwidth product performance of a near-ballistic uni-traveling-carrier photodiode with a flip-chip bonding structure,” IEEE J. Quantum Electron. 46(1), 80–86 (2010). [CrossRef]