Open Access
Add to BibSonomyAbstract
We present experimental results for a selective epitaxially grown Ge-on-Si separate absorption and charge multiplication (SACM) integrated waveguide coupled avalanche photodiode (APD) compatible with our silicon photonics platform. Epitaxially grown Ge-on-Si waveguide-coupled linear mode avalanche photodiodes with varying lateral multiplication regions and different charge implant dimensions are fabricated and their illuminated device characteristics and high-speed performance is measured. We report a record gain-bandwidth product of 432 GHz for our highest performing waveguide-coupled avalanche photodiode operating at 1510nm. Bit error rate measurements show operation with BER< 10−12, in the range from −18.3 dBm to −12 dBm received optical power into a 50 Ω load and open eye diagrams with 13 Gbps pseudo-random data at 1550 nm.
© 2016 Optical Society of America
1. Introduction
Significant progress has been made in the past several years toward the development of a complete CMOS compatible integrated silicon photonics platform [1, 2]. Increasing functionality has been realized with the development of refined building block devices working at telecommunication wavelengths that can be combined into new compact photonic integrated circuits [3–8]. Recently photonic integrated circuits designed within the platform have targeted new types of chip-scale optical interconnect applications. These include classical optical communications for optical interconnection in data centers [9] and for quantum communications, specifically quantum key distribution (QKD) [10, 11] which require new device functionality within the silicon photonics platform.
Photodetection plays a key role in optical sensing and communications applications. The integration of waveguide-coupled photodetectors into our silicon photonics platform has been a central focus and we have previously developed an integrated high speed Ge on Si p-i-n photodiode into our silicon photonics platform [6, 12]. Up to now, a key missing component in our integrated optical platform is an optical detector with gain. This would enable improvements in integrated optical receivers, but also potentially lead to new pulsed photon counting and single photon detection capability for laser imaging detection and ranging (LIDAR) and quantum communications. Waveguide-coupled Ge avalanche photodiodes have recently been reported and integrated within a photonics platform [13]. These devices used Ge as both an absorber and the avalanche multiplication region with reasonable performance but has led to higher receiver noise figures due to excess noise from the Ge avalanche characteristics. Recently, a new high performance separate absorption and multiplication APD with top-down illumination in the Ge on Si system has been demonstrated [14–16]. In this vertical Ge on Si APD device, the photon is absorbed in the Ge and the avalanche multiplication takes place in the Si layer. This one-dimensional linear mode APD has shown record gain-bandwidth product at 1310 nm, exceeding state of the art InGaAs/InP APD’s [14–17]. Here we present the first compact separate absorption and charge multiplication waveguide-coupled avalanche photodiode with performance comparable to top-down illuminated devices in the same Ge on Si material system. Currently waveguide-coupled SACM APD’s at 1550 nm in the Ge on Si material system have reported gain-bandwidth performance of 105 GHz [18, 19], where top-illuminated SACM devices in the same material system have gain-bandwidth performance exceeding 300 GHz [20, 21]. Our highest performing waveguide-coupled device has gain-bandwidth product exceeding 430 GHz and is based on our selective epitaxial Ge on Si growth process. Thus our waveguide-coupled SACM APD’s are completely compatible and co-integratable with Ge on Si p-i-n detectors within our silicon photonics process.
Selective epitaxial growth of Ge on Si has been introduced into modern CMOS processing for strain engineering of transistors and for high speed digital and analog components [22, 23]. Furthermore, the Ge on Si system has been extensively studied for p-i-n photodiodes for both orthogonal entry [24–27] and waveguide coupled devices [6]. Colace et al. first demonstrated photodetection in the Ge on Si system using ultrahigh vacuum chemical vapor deposition (UHV-CVD) with a two step growth. First a low temperature growth of 50nm buffer at 330° C, followed by a high temperature, 600° C growth to fill to the desired thickness [26] [25]. The growth of the low temperature Ge buffer layer was originally explored via molecular beam epitaxy (MBE) [28, 29]. Kang et al. [14, 30] has made significant progress in the development of avalanche photodiode (APD) mesa type orthogonal entry devices and the use of selective epitaxy has been used to explore waveguide coupled APD devices with separate absorber and charge multiplication (SACM) regions [31]. Our device is based on a two step Ge growth mechanism. The high temperature growth step is critical to the performance of our devices since this step gives rise to tensile strain in the Ge through the coefficient of thermal expansion mismatch between Si and Ge and improves the response of our detectors in the telecom wavelength region (1530–1565 nm).
Our objective in this paper is to describe the development of a waveguide-coupled avalanche photodiode (APD) that operates as a linear mode gain element for enhanced optical communication purposes but also may be operated in gated Geiger mode for use in chip-scale photon counting applications. The APD device is an extension of the processes used for our previously demonstrated p-i-n photodiode [6], and in this paper we present a first design study for integrating waveguide-coupled Ge on Si APDs into our silicon photonics platform. We report on device design, manufacture, and measured characteristics of these devices operating as an integrated photonic receiver. The focus is for operation of the APD in linear mode, and Geiger mode operation will be the subject of a future study.
1.1. Device fabrication & characteristics
The initial design considerations for an avalanche photodiode (APD) evolved from previously demonstrated p-i-n photodiodes [6] integrated within the Silicon photonics platform. This has the advantage that both Ge on Si p-i-n photodiodes and our new Ge on Si APD’s can be integrated simultaneously with other active optoelectronic components within the platform and allows for complex chip-scale optical circuits with both photodetection elements.
Figure 1 shows the cross-section of the SACM waveguide-coupled device. In this design, the Ge acts as the optical absorbing material into which light can be input from a Si end-fire waveguide. The multiplication occurs in the lateral intrinsic region of the Si of width Wmulti. The symmetry of the device is important from field configuration considerations and both Tungsten Si contacts are symmetrically displaced. The p-charge layer is implanted in the Si and engineered to stand off the high field region in the Si, while keeping the field in the Ge absorber region below breakdown. The p-charge implant layer density can be varied, however here we examine a single Boron dose for all tested APDs on the same wafer. Instead, the p-charge layer was designed to have no overlap, t = 0, and positive overlap, 100 nm, relative to the Ge sidewall. This, in effect, offers a nominally higher doped charge layer thickness utilizing the same wafer for all devices. The multiplication width is parametrically studied, as well, with incremented variations from 200 nm to 1000 nm.
Fig. 1 Schematic cross-section and angled SEM image of waveguide coupled linear mode APD (a) Schematic of device with p-charge layer overlap t relative to Ge. The multiplication width Wmulti, indicates width of intrinsic Si multiplication region. (b) Angled SEM image of Ge on Si device structure with oxide cladding removed. The device structure is similar to our Ge p-i-n [6] with different implants.
The linear mode waveguide-coupled APD is based on a separate absorber layer design which consists of selective epitaxy of Ge on a p-implanted charge layer in the underlying Si. Figure 1(a) shows schematically the device cross-section and (b) shows an angled SEM of our deprocessed device showing the Ge on Si device with metal interconnects and Tungsten vias as contacts to the Si on insulator pedestal. The lateral APD device shown in Fig. 1(b) has a 1 micron wide multiplication region in the Si with a p-charge layer with no overlap (t=0 nm). A new p-charge layer mask and process step has been added and a 110 keV Boron difluride (BF2) implant with dose splits in the range of 2 − 6 × 1012cm2 are used targeting active p-charge layer carrier densities of 1 − 2 × 1017 cm3. Boron diffusion is well calibrated at our CMOS facility and has been shown to be constant for devices across the wafer. Silvaco simulations corroborate a uniformly doped p-charge layer post annealing. The Ge contact is formed by Boron difluoride (BF2) implantation into Ge and activation to a concentration of 1×1019cm−3 to form the p-type layer and top contact of the diode. A Ti/TiN metal liner is used to make ohmic contact to the Ge. The Si contact is made to a heavily doped N+ layer that is formed prior to Ge growth. The record gain-bandwidth product of our waveguide-coupled lateral APD is attributed to more refined fabrication and a significantly smaller absorber volume. The lateral multiplication region consists of undoped Si and is the high field region where impact ionization in the Si occurs. Due to the favorable ratio of electron to hole impact ionization coefficients in Si, the separate absorber charge multiplication design leads to good noise performance of the APD as a receiver [32].
The measured photoresponse of the APD is given by the responsivity, Iphoto = RPopt, where Iphoto = Ilight − Idark is the photocurrent and Popt is the incident optical on the device, we then have
Here η is the quantum efficiency, and G is the multiplication gain of the APD. The evaluation of the multiplication gain requires an estimate of the unit gain G = 1 responsivity from which we can determine the bias for unit gain, VG=1. The multiplication gain is where Ilight (V) and Idark (V) are the measured illuminated and dark currents, respectively at the applied bias voltage V. In the extracted gain G, we consider a quantum efficiency (QE) of η = 0.66 which is the measured Ge p-i-n diode QE.The electric field in the region near the Ge/Si interface is a key determinant of the device characteristics. At the interface, misfit dislocations occur due to the lattice mismatch at the Ge/Si heterojunction. The p-charge layer at the interface is designed to keep the electric field in the Ge below breakdown but high field in this region can give rise to high dark current. In the misfit dislocation region, we expect that these threading dislocation defects form mid-gap states in the Ge and are responsible for the dark currents seen in the APD dark IV characteristics in Fig. 2(a) and 3(a). Conversely in the Ge p-i-n device, the dislocation network is placed within the “contact”, or quasi-neutral, region for the diode, and recombination centers at the interface will not adversely affect device performance. Another consequence of the lattice mismatch is that it results in 0.25% tensile strain in the Ge. This splits the hole bands lowering the direct band gap and is a direct result of the growth temperature and conditions. This strain is responsible for the improved direct absorption at 1550 nm and longer wavelengths and gives rise to the high responsivity of our devices with improved performance.
Fig. 2 Device characteristics for the no-overlap device t = 0 in Fig. 1 for various multiplication widths. Solid lines are illuminated and dashed lines are dark IVs. (a) LIV characteristics for the each APD. (b) Measured responsivity for each device. (c) Gain obtained from the LIV characteristics. (Black Wmulti = 1.0µm Vbr = 30.6V, Grey Wmulti = 0.5µm Vbr = 15.8V, Blue Wmulti = 0.4µm Vbr = 12.3V, Green Wmulti = 0.3µm Vbr = 7.8V) The illumination wavelength was λ = 1510 nm for all devices.
Fig. 3 Device characteristics for the overlap device t = 100 nm in Fig. 1 for various multiplication widths. Solid lines are illuminated and dashed lines are dark IVs. (a) LIV characteristics for the each APD. (b) Measured responsivity for each device. (c) Gain obtained from the LIV characteristics. (Dark Red Wmulti = 0.9µm Vbr = 30.6V, Purple Wmulti = 0.4µm Vbr = 16V, Gold Wmulti = 0.3µm Vbr = 12.3V, Bright red Wmulti = 0.2µm Vbr = 8.2V) The illumination wavelength was λ = 1510 nm for all devices.
Figure 2 and 3 show the measured APD characteristics under illumination for the no-overlap (t = 0 nm) and the overlap (t = 100 nm) devices for various multiplication widths where the solid (dashed) lines are for illuminated (dark) IV’s respectively. The LIV characteristics and the measured responsivity are shown in panels (a) and (b). The downward trend in current with increasing reverse bias seen in nearly all IV’s is due to probe capacitance and is not an artifact of the actual device. The optical power received at the APD is calibrated from measured through waveguide loss in a straight waveguide section, and includes the grating coupling efficiency and the intrinsic waveguide loss. The grating-waveguide coupling was optimized for each APD to ensure constant 1.34 µW incident optical power. The unit gain voltage VG=1 is obtained from the measured responsivity of the device. We have used the Ge p-i-n responsivity of 0.8 A/W for unit QE for determination of the multiplication gain in the APD. The breakdown voltage of the APD, Vbr, is limited by a current compliance condition of I = 0.2 mA on a Keithley source meter.
The observed differences in APD gain between no-overlap and the t = 100 nm overlap devices is attributed to the higher electric field in the Ge absorber for the no-overlap device and the added recombination due to the Shockley-Read-Hall recombination in the larger p-charge volume in the overlap device. Avalanche gain is estimated from the measured responsivity of a reference a p-i-n diode and is in good agreement with experimental gain estimates. We therefore use the measured 0.8 A/W responsivity as the unit gain for all gain calculations used throughout the paper. We measured the noise performance and gain as a function of bias for a second identical APD with a 1µm multiplication region and gain vs. bias characteristics nearly identical to the APD in Fig. 2. For a gain of 10, 20, and 40, the noise equivalent power (NEP) was found to be , and , and respectively, and the excess noise factor (F) was ~ 1.6, 3.5, and 11.1 respectively. The noise performance is consistent with other Ge on Si APD designs [30], however with our larger gain there is a higher field in the Ge which contributes additional noise. To improve the performance of the device, further optimization of the device configuration, implants and growth are needed.
1.2. Bandwidth and bit error rate testing
The dynamic response of the waveguide-coupled APD is characterized by the bandwidth measurement shown in Fig. 4. In this experiment, two lasers are mixed in a 50%–50% combiner, one laser is at fixed frequency ν and the other laser frequency is tuned ν + Δν. This combined optical output is a heterodyned optical RF beat at the difference frequency Δν which is mixed in the waveguide-coupled APD under bias where it is converted to an electrical RF signal. The RF electronic signal is passed through a bias tee and into an RF power meter and recorded. By detuning the tunable laser we can sweep frequency and measure the frequency response of our detector. To calibrate the tunable laser to the fixed wavelength laser source, the tunable laser was swept in wavelength and mixed in a commercial high speed detector and the RF beat was measured in a spectrum analyzer. Bandwidth resolution was limited by the 2 pm wavelength step size applied to the tunable laser, chosen to optimize the time measuring the frequency response of multiple devices across the wafer.
Fig. 4 Schematic of heterodyned bandwidth measurement. In the measurement, two lasers are used to generate an RF beat in the device under DC bias. The RF signal is then measured on an RF power meter and one of the laser is detuned from the center wavelength to vary the RF frequency of the beat.
In order to further quantify performance, we focused our experimental examination on the no-overlap devices with the widest multiplication widths and hence higher gain. Figure 5 shows the performance comparison of the 500 nm and 1000 nm multiplication width devices. Figure 5(a) shows the measured bandwidths of these devices at the bias voltage corresponding to the highest gain. The shorter device shows considerably higher bandwidth but with lower gain. The bandwidth versus gain plot shows comparable performance at lower gain but with increasing voltage the shorter multiplication width device has superior bandwidth. The bandwidth vs gain plot of Fig. 5(b) shows nearly constant bandwidth over all gains. This is effectively the transit RC time constant limit of the device. We expect that at higher gain we will see the bandwidth decrease due to the avalanche build up in the Si layer. This bandwidth roll-off is not seen in the measured data due to current compliance needed to protect the APD devices. Since the gain is estimated from the measured responsivity, in Fig. 5(c) we show the bandwidth plotted against the measured responsivity of the device. We see that the bandwidth begins to plateau as the responsivity of the device increases for our linear mode APD. The higher bandwidth is associated with the shorter drift times in the multiplication region.
Fig. 5 Performance characteristics of high gain no-overlap linear mode APDs. Grey (Black) line is for 500 (1000) nm wide multiplication region. (a) Measured frequency response at bias voltage corresponding to highest gain. (b) Bandwidth vs. gain for each device. (c) Bandwidth vs. measured device responsivity for both devices.
A key figure of merit for these linear mode APD’s is the gain bandwidth product. For the highest applied bias 31 (17) Volts, we measured for the 1000 nm (500 nm) wide device a gain of 69.3 (14.3) and a bandwidth of 6.24 (9.8) GHz leading to a gain bandwidth product of 432 (140) GHz, respectively. Given the previously stated optical power of 1.34 µW, a light current of ~ 1 × 10−4 A, and a dark current of ~2.7 × 10−5 A for the 1000 nm APD, the responsivity at this operating point is (1 × 10−4 − 2.7 × 10−5)/1.34µW = 54.5 A/W.
The final test of the APD is to characterize the transient performance as a integrated waveguide-coupled receiver. In order to measure its response, we create a pseudo-random bit sequence (PRBS) of word length of 210 − 1 that is transmitted to modulated light incident on the waveguide coupled APD and measure the electrical dynamic response, as a bit error rate (BER) as a function of the received optical power. In Fig. 6, we show the measured BER as a function of the received power for the two APD devices for comparison. The APD output was amplified by a pair of 50 Ω input pre-amplifiers and a post amplifier to boost the signal to the level required by the error detector. The insets show the 10 Gbps eye-diagrams for each of the no-overlap devices at a wavelength of 1550 nm. The devices show error free operation, BER below 10−12, for reciever optical power ranges of −17.2 to −10.9 dBm for 500 nm wide device, and −18.3 to −12 dBm for 1000 nm wide device. The sensitivity is worse than a similar APD trans-impedance amplifier (TIA) [33, 34] because we used an amplifier with a 50 Ω input impedance. As the optical power is increased, we see an increase in the error rates associated with saturation of the detector. Both APDs have an open eye up to 13 Gbps which is our equipment bandwidth limit. This bandwidth may be limited by charging of the SiO2 sidewalls which could be contributing to the inter-symbol interference in the error rate measurements.
Fig. 6 Bit Error Rate measurements for no-overlap linear mode APDs. (a) BER vs received optical power for 500 nm multiplication width device. Inset shows open 10 Gbps eye diagram. (b) BER vs. received optical power for 1000 nm multiplication width device. Inset shows measured 10 Gbps eye diagram. All APDs run under λ = 1550 nm illumination for 10 GHz PRBS data.
1.3. Conclusions
In this paper, we have described the design, fabrication, and characterization of an integrated waveguide-coupled Ge on Si linear mode APD within our Si photonics process. We have demonstrated record 432 GHz gain-bandwidth product for a waveguide coupled APD operating in the telecom band (1510–1550 nm). This represents the highest gain-bandwidth product for a waveguide-coupled device [18,19] with performance approaching top-illuminated devices [20,21] in the Ge on Si material system.
Our device is based on a 2D lateral separate absorber charge multiplication region design where impact ionization occurs in a lateral Si multiplication region displaced from the Ge absorber. The dynamic response of the devices is obtained using a heterodyned laser technique to create an RF beat in the device. This method has been previously used to evaluate 3dB response of our Ge on Si p-i-n diodes [6]. The measured responsivity of the APD’s is calibrated using a measured Ge on Si p-i-n device and straight through waveguides to estimate the grating coupling loss. The APD multiplication gain is estimated from the measured APD responsivity and the p-i-n responsivity to estimate the unit gain voltage. This procedure is necessary since the lateral APD design is intrinsically 2D and calibration structures have vastly different field profiles leading to very different behavior in comparison to 1D planar top-down structures. This complex field profile throughout the device requires that the electrode design be symmetric in order to channel photogenerated carriers from the Ge into the multiplication region in Si and to keep the electrons away from the oxide interface.
From the frequency response, we find a measured gain-bandwidth product in excess of 430 GHz for our largest multiplication width devices. The bandwidth of the APD is seen to vary with the multiplication width (12.6 to 6.2 GHz, for 0.3 to 1.0 micron width range) and to depend on the bias voltage. The 3dB frequency response is seen to saturate around 12 GHz for all variations of the devices. We note that the devices do not exhibit a constant gain-bandwidth product, but rather something close to constant bandwidth independent of gain. By running the APD as a receiver, we find an open 13 Gbps eye for pseudorandom bit sequence data for both the 1000 nm and 500 nm wide multiplication width devices. The BER is seen to be < 10−12 for input optical power ranges of −18.3 dBm to −12 dBm and −17.2 to −10.9 dBm, respectively. Our receiver sensitivity is limited by our choice of amplifier and can be improved with a transimpedance amplifier [33].
The performance of our waveguide-coupled APD’s at 1550 nm is quite good compared to similar devices [18, 19], however further optimization and design parameter studies should be undertaken to improve the dark current, responsivity and bandwidth of the APD. For other applications, Geiger mode operation will be required [36] and will be the subject of a future study. Furthermore, noise characterization and the effects of temperature are also of interest for photon counting and other applications.
Acknowledgments
We would like to thank Chris Long for useful comments and Kate Musick of Sandia for SEM pictures of our devices. Funding for this work was provided by Sandia’s Laboratory Directed Research and Development (LDRD) program and the Department of Defense. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy’s National Nuclear Security Administration under contract DE-AC04-94AL85000.
References and links
1. A. L. Lentine, C. T. Derose, P. S. Davids, J. D. Nicolas, W. A. Zortman, J. A. Cox, A. Jones, D. C. Trotter, A. T. Pomerene, A. L. Starbuck, D. J. Savignon, M. Wiwi, and P. B. Chu, “Silicon Photonics Platform for National Security Applications,” IEEE Aerospace Conference pp. 1–9 (2015).
2. D. J. Lockwood and L. Pavesi, “Silicon Fundamentals for Photonics Applications,” Silicon Photonics, Topics in Applied Physics 94, 1–52 (2004). [CrossRef]
3. M. R. Watts, J. Sun, C. DeRose, D. C. Trotter, R. W. Young, and G. N. Nielson, “Adiabatic thermo-optic Mach-Zehnder switch,” Opt. Lett. 38, 733–735 (2013). [CrossRef] [PubMed]
4. J. A. Cox, A. L. Lentine, D. J. Savignon, R. D. Miller, D. C. Trotter, and A. L. Starbuck, “Very Large Scale Integrated Optical Interconnects : Coherent Optical Control Systems with 3D Integration,” Integrated Photonics Reseach, Silicon and Nanophotonics pp. IM2A–1 (2014).
5. J. A. Cox, D. C. Trotter, and A. L. Starbuck, “Integrated control of silicon-photonic micro-resonator wavelength via balanced homodyne locking,” Opt. Express 22, 52–53 (2013).
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, 527–534 (2011). [CrossRef]
7. A. L. Lentine, J. A. Cox, W. A. Zortman, and D. J. Savignon, “Electronic interfaces to silicon photonics,” SPIE Photonics West 2014-OPTO: Optoelectronic Devices and Materials 8989, 89890F (2014).
8. A. L. Lentine and C. T. DeRose, “Challenges in the implementation of dense wavelength division multiplexed (DWDM) optical interconnects using resonant silicon photonics,” Proceedings of SPIE 9772, 977207 (2016). [CrossRef]
9. D. Miller, “Device requirements for optical interconnects to silicon chips,” Proceedings of the IEEE 97, 1166–1185 (2009). [CrossRef]
10. P. Sibson, C. Erven, M. Godfrey, S. Miki, T. Yamashita, M. Fujiwara, M. Sasaki, H. Terai, M. G. Tanner, C. M. Natarajan, R. Hadfield, J. O’Brien, and M. Thompson, “Chip-based quantum key distribution,” arXiv preprint arXiv:1509.00768 (2015).
11. J. L. O’Brien, A. Furusawa, and J. Vučković, “Photonic quantum technologies,” Nature Photon. 3, 687–695 (2009). [CrossRef]
12. G. Masini, L. Colace, G. Assanto, H. C. Luan, and L. C. Kimerling, “High-performance p-i-n Ge on Si photodetectors for the near infrared: From model to demonstration,” IEEE Transactions on Electron Devices 48, 1092–1096 (2001). [CrossRef]
13. S. Assefa, F. Xia, and Y. A. Vlasov, “Reinventing germanium avalanche photodetector for nanophotonic on-chip optical interconnects,” Nature 464, 80–84 (2010). [CrossRef] [PubMed]
14. Y. Kang, H. Liu, M. Morse, and M. Paniccia, “Monolithic germanium/silicon avalanche photodiodes with 340 GHz gain–bandwidth product,” Nature Photon. 3, 59–63 (2008). [CrossRef]
15. Y. Kang, M. Morse, M. J. Paniccia, M. Zadka, Y. Saad, G. Sarid, A. Pauchard, W. S. Zaoui, H.-W. Chen, D. Dai, J. E. Bowers, H.-D. Liu, D. C. Mcintosh, X. Zheng, and J. C. Campbell, “Monolithic Ge/Si avalanche photodiodes,” 2009 6th IEEE International Conference on Group IV Photonics6, 25–27 (2009).
16. J. E. Bowers, D. Dai, W. S. Zaoui, Y. Kang, and M. Morse, “Resonant Si/Ge avalanche photodiode with an ultrahigh gain bandwidth product,” 2010 IEEE Photonics Society Winter Topicals Meeting Series (WTM)2, 111–112 (2010).
17. Y. Kang, Y. Saado, M. Morse, M. J. Paniccia, J. C. Campbell, J. E. Bowers, and A. Pauchard, “Ge/Si waveguide avalanche photodiodes on SOI substrates for high speed commnunication,” ECS Transactions 33, 757–764 (2015).
18. K.-W. Ang, M.-B. Yu, G.-Q. Lo, and D.-l. Kwong, “Low-Voltage and High-Responsivity Germanium Bipolar Phototransistor for Optical Detections in the Near-Infrared Regime,” IEEE Electron Device Letters 29, 1124–1127 (2008). [CrossRef]
19. K.-W. Ang, J. W. Ng, A. E.-J. Lim, M.-B. Yu, G.-Q. Lo, and D.-L. Kwong, “Waveguide-integrated ge/si avalanche photodetector with 105ghz gain-bandwidth product,” in “Optical Fiber Communication Conference,” (Optical Society of America, 2010), p. JWA36.
20. N. Duan, T.-Y. Liow, A. E.-J. Lim, L. Ding, and G. Q. Lo, “310 ghz gain-bandwidth product ge/si avalanche photodetector for 1550 nm light detection,” Opt. Express 20, 11031–11036 (2012). [CrossRef] [PubMed]
21. N. Duan, T.-Y. Liow, A. E. Lim, L. Ding, and G. Lo, “High speed waveguide-integrated ge/si avalanche photodetector,” in “Optical Fiber Communication Conference/National Fiber Optic Engineers Conference 2013,” (Optical Society of America, 2013), p. OM3K.3.
22. M. L. Lee, E. A. Fitzgerald, M. T. Bulsara, M. T. Currie, and A. Lochtefeld, “Strained si, sige, and ge channels for high-mobility metal-oxide-semiconductor field-effect transistors,” Journal of Applied Physics 97, 011101 (2005). [CrossRef]
23. J. Kavalieros, B. Doyle, S. Datta, G. Dewey, M. Doczy, B. Jin, D. Lionberger, M. Metz, W. Rachmady, M. Radosavljevic, U. Shah, N. Zelick, and R. Chau, “Tri-gate transistor architecture with high-k gate dielectrics, metal gates and strain engineering,” in “VLSI Technology, 2006. Digest of Technical Papers. 2006 Symposium on,” (2006), pp. 50–51.
24. L. Colace, P. Ferrara, G. Assanto, S. Member, D. Fulgoni, and L. Nash, “Low Dark-Current Germanium-on-Silicon Near-Infrared Detectors,” Ieee Photonics Technology Letters 19 No 22, 1813 (2007). [CrossRef]
25. L. Colace, G. Masini, and G. Assanto, “Ge-on-Si Approaches to the Detection of Near Infared Light,” IEEE JOURNAL OF QUANTUM ELECTRONICS 35, 1843–1852 (1999). [CrossRef]
26. G. Masini, L. Colace, and F. Galluzzi, “Ge/Si (001) Photodetector for Near Infrared Light,” Solid State … 54, 55–58 (1997).
27. J. Michel, J. Liu, and L. C. Kimerling, “High-performance Ge-on-Si photodetectors,” Nature Photon. 4, 527–534 (2010). [CrossRef]
28. A. Sakai and T. Tatsumi, “Ge growth on si using atomic hydrogen as a surfactant,” Appl. Phys. Lett. 64, 52–54 (1994). [CrossRef]
29. D. Dentel, J. Bischoff, T. Angot, and L. Kubler, “The influence of hydrogen during the growth of ge films on si (001) by solid source molecular beam epitaxy,” Surface Science 402, 211–214 (1998). [CrossRef]
30. Y. Kang, H.-D. Liu, M. Morse, M. J. Paniccia, M. Zadka, S. Litski, G. Sarid, A. Pauchard, Y.-H. Kuo, H.-W. Chen, W. S. Zaoui, J. E. Bowers, A. Beling, D. C. McIntosh, X. Zheng, and J. C. Campbell, “Monolithic germanium/silicon avalanche photodiodes with 340 GHz gain–bandwidth product,” Nature Photonics 3, 59–63 (2009). [CrossRef]
31. S. Zhu, K. W. Ang, S. C. Rustagi, J. Wang, Y. Z. Xiong, G. Q. Lo, and D. L. Kwong, “Waveguided Ge/Si avalanche photodiode with separate vertical SEG-Ge absorption, lateral Si charge, and multiplication configuration,” IEEE Electron Device Letters 30, 934–936 (2009). [CrossRef]
32. M. C. Teich and B. E. Saleh, Fundamentals of Photonics, vol. 2 (Wiley, 1991).
33. H. Chen, J. Verbist, P. Verheyen, P. De Heyn, G. Lepage, J. De Coster, P. Absil, X. Yin, J. Bauwelinck, J. Van Campenhout, and G. Roelkens, “High sensitivity 10gb/s si photonic receiver based on a low-voltage waveguide-coupled ge avalanche photodetector,” Opt. Express 23, 815–822 (2015). [CrossRef] [PubMed]
34. H. Chen, J. Verbist, P. Verheyen, P. De Heyn, G. Lepage, J. De Coster, P. Absil, X. Yin, J. Bauwelinck, J. Van Campenhout, and G. Roelkens, “Low-voltage waveguide ge apd based high sensitivity 10gb/s si photonic receiver,” in “Optical Communication (ECOC), 2015 European Conference on,” (IEEE, 2015), pp. 1–3.
35. H.-C. Luan, D. R. Lim, K. K. Lee, K. M. Chen, and J. G. Sandland, “High-quality Ge epilayers on Si with low threading-dislocation densities,” Applied Physics Letters 75, 2909–2911 (1999). [CrossRef]
36. R. Warburton, G. Intermite, M. Myronov, P. Allred, D. Leadley, K. Gallacher, D. Paul, N. Pilgrim, L. Lever, Z. Ikonic, R. Kelsall, E. Huante-Ceron, A. Knights, and G. Buller, “Ge-on-si single-photon avalanche diode detectors: Design, modeling, fabrication, and characterization at wavelengths 1310 and 1550 nm,” Electron Devices, IEEE Transactions on 60, 3807–3813 (2013). [CrossRef]
