Open Access
Add to BibSonomy
Abstract
Based on the commercial silicon photonics (SiPh) process platform, a flat 3 dB bandwidth of 80 GHz germanium-silicon (Ge-Si) photodetector (PD) is experimentally demonstrated at a photocurrent of 0.8 mA. This outstanding bandwidth performance is achieved by using the gain peaking technique. It permits an 95% improvement in bandwidth without sacrificing responsivity and undesired effects. The peaked Ge-Si PD shows the external responsivity of 0.5 A/W and internal responsivity of 1.0 A/W at a wavelength of 1550 nm under -4 V bias voltage. The high-speed large signal reception capability of the peaked PD is comprehensively explored. Under the same transmitter state, the transmitter dispersion eye closure quaternary (TDECQ) penalties of the 60 and 90 Gbaud four-level pulse amplitude modulation (PAM-4) eye diagrams are about 2.33 and 2.76 dB, 1.68 and 2.45 dB for the un-peaked and peaked Ge-Si PD, respectively. When the reception speed increase to 100 and 120 Gbaud PAM-4, the TDECQ penalties are approximatively 2.53 and 3.99 dB. However, for the un-peaked PD, its TDECQ penalties cannot be calculated by oscilloscope. We also measure the bit error rate (BER) performances of the un-peaked and peaked Ge-Si PDs under different speed and optical power. For the peaked PD, the eye diagrams quality of 156 Gbit/s nonreturn-to-zero (NRZ), 145 Gbaud PAM-4, and 140 Gbaud eight-level pulse amplitude modulation (PAM-8) are as good as the 70 GHz Finisar PD. To the best of our knowledge, we report for the first-time a peaked Ge-Si PD operating at 420 Gbit/s per lane in an intensity modulation direct-detection (IM/DD) system. It might be also a potential solution to support the 800 G coherent optical receivers.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
In 1969, the concept of integrated photonics was firstly proposed by Stewart E. Miller [1]. Since then, the great successful applications of integrated photonics have been very wide, including optical interconnects (data centers and coherent transceivers), quantum communications, 5 G mobile communications, cloud computing, and neuromorphic computing [2–7]. There are mainly two platforms for integrated photonics: indium phosphide (InP) and silicon photonics (SiPh). For the InP platforms, both passive and active components (such as lasers, amplifiers, and photodetectors) can be enabled and integrated on the same substrate [7]. For the SiPh platforms, although it possesses some clear advantages over InP, such as its compatibility with advanced 300-mm complementary metal-oxide semiconductor (CMOS) technologies with low cost and high volume, one performance bottleneck is the bandwidth and responsivity of photodetector (PD) [8]. Ten years ago, the InP-based uni-traveling carrier (UTC) PD had already achieved an optoelectrical (OE) 3 dB bandwidth of 170 GHz and responsivity of 0.27 A/W [9]. More recently, the recorded 3 dB bandwidth of 330 GHz InP-based PD with a moderate responsivity of 0.11 A/W was also reported by using the type-II structure [10]. By comparison, the typical germanium-based PDs show the bandwidths in the range of 30-67 GHz with responsivity of >0.85 A/W [11–20]. Remarkably, by employing the novel biconcave germanium (Ge) fin shape, the recorded 3 dB bandwidth of up to 265 GHz and 240 GHz with internal responsivities of 0.3 A/W and 0.45 A/W has been reported by IHP [21], which is an important milestone. Its fabrication processes might be a little complex. Several hard-masks, lithography, dry etch and chemical mechanical polishing (CMP) processes are executed to form a lateral Si-Ge-Si diode and define the actual width of Ge region. Its rather high dark currents of 100-200 nA are also obtained, which should be further improved. Consequently, for the Ge-based PDs, there still exist a critical trade-off among responsivity (quantum efficiency, QE), dark current, bandwidth (speed) and fabrication process (yields and costs), which intrinsically limits its competition with InP-based PDs.
The gain peaking technique has been successfully utilized to enhance the bandwidths of CMOS amplifiers in the field of integrated circuits, such as drivers and transimpedance amplifiers [22,23]. In 2006 and 2012, this technique in SiPh process platform was proposed and introduced by these groups [24–26]. They demonstrated that the bandwidth of Ge-Si PD can be improved significantly without sacrificing responsivity and increasing dark current. Up to now, the experimentally measured 3 dB bandwidth of 75 GHz is the maximum value [27,28]. However, the key issue of inductive gain peaking technique is the dispersion, which is the phase delay dependent on different frequency. It would distort the high-speed large signals. For the conventionally used vertical Ge-Si PD, whose intrinsic 3 dB bandwidth is less than 35 GHz, the over-peaking effect is unavoidable in order to achieve large bandwidth. The extreme peak of bandwidth might limit the signal reception bit rate to be 128 Gbit/s [27]. In this scenario, to satisfy the requirements of actual applications, a laudable goal would be to explore ultrahigh-speed, high responsivity, and low dark current PDs with easy fabrication processes by employing inductive gain peaking technique with suitable inductance.
In this work, with the aid of intrinsic 42 GHz lateral Ge-Si PD and optimized on-chip spiral inductor as peaking element, a flat OE 3 dB bandwidth of 80 GHz peaked PD is demonstrated. The use of gain peaking technique permits an 95% improvement in bandwidth without sacrificing responsivity and undesired effects. Our peaked PD shows the external responsivity of 0.5 A/W and internal responsivity of 1.0 A/W at a wavelength of 1550 nm under -4 V bias voltage. The external responsivity vs. temperature for 1530, 1550, and 1565 nm is carried out. The eye diagrams comparison of un-peaked and peaked PD is presented for the speed of 60, 90, 100, and 120 Gbaud PAM-4. Under the same transmitter state, the transmitter dispersion eye closure quaternary (TDECQ) penalties of the 60 and 90 Gbaud four-level pulse amplitude modulation (PAM-4) eye diagrams are about 2.33 and 2.76 dB, 1.68 and 2.45 dB for the un-peaked and peaked PD, respectively. Furthermore, the 136, 145, and 156 Gbit/s nonreturn-to-zero (NRZ), and 128, 136, and 145 Gbaud four-level pulse amplitude modulation (PAM-4), and 128, 136, and 140 Gbaud eight-level pulse amplitude modulation (PAM-8) clear open electrical eye diagrams are also obtained without utilizing offline digital signal processing at the receiver side. Their large signal reception performances are as good as the commercial 70 GHz Finisar PD. These demonstrated ultra-high-speed eye diagram reception capability would pave the way to practical applications of gain peaking technique.
2. Design and fabrication of the peaked Ge-Si PD
The Ge-Si waveguide PDs commonly require P or N type heavily-doped germanium as well as a metal contact on the Ge area to form the p-i-n junction. This will result in a reduced QE due to the light absorption of metal contacts. In order to improve the QE, a lateral Ge-Si waveguide PD without doped Ge or Ge metal contacts is proposed [29,30]. Figure 1 (a) show the cross-sectional view of the proposed Ge-Si waveguide PD with Si-contacted junction. Here, the 220 nm Si region are doped with N + and P+, which is helpful to reduce resistance. Their concentrations are about $2.0 \times {10^{18}}c{m^{ - 3}}$. Besides, the bottom width of Ge film is designed to be 600 nm to balance the relationship among yield, quality, and bandwidth. The length of Ge active region is 35 μm. The gap size of 200 nm is designed for the p-i-n junction region, which is beneficial to decrease the carrier transit time. The designed Ge-Si waveguide PDs are fabricated in a SiPh platform with simple fabrication processes. The Ge film is epitaxially grown on the 220 nm thick Si through a low-pressure chemical-vapor deposition (LPCVD) process. Figure 1 (b) shows the simulated static electric field distribution in Ge region. The overall electric field intensity of Ge region is larger than $5.0 \times {10^4}V/cm$ at -4 V bias voltage, which benefits from the well-designed doping profiles and reduced width of Ge. The maximum field intensity of Ge is about $1.0 \times {10^5}V/cm$, which is conducive to realize high bandwidth in small bias voltage. The cross-sectional transmission electron microscopy (TEM) of the PD is shown in Fig. 1(c). The actual width of bottom Ge is about 580 nm with a thickness of 350 nm. The optical micrograph of the peaked PD chip with integrated inductor is shown in Fig. 1(d). For the PD responsivity test, the light was coupled in via Si-based suspended spot size convert (SSC) edge coupler with -3 dB/facet coupling loss at 1550 nm wavelength. The design procedure of peaked PD can be summarized as follows [27]. Firstly, the precisely circuit models of un-peaked and peaked Ge-Si PD are introduced to further simulate how the PD equivalent circuit respond to the added inductor. Secondly, the series resistance, p-i-n junction capacitance, and RF pad parasitic capacitance of un-peaked Ge-Si PD are extracted by fitting S11 scattering parameter data to the small signal RC-model. Then, the value of inductor resistance, inductor capacitance, and inductance are optimized to obtain large bandwidth of peaked PD. Finally, by using a commercial software, the full 3-dimensional electromagnetic simulation of the inductor is performed to design the physical parameters. The simulation model of our inductor has also considered the processes of commercial standard SiPh platform. The thickness of Cu metal 1 and 2 are about 1.1 μm. The designed value of inductance is about 170 pH.
Fig. 1. (a) Cross-sectional view of the peaked Ge-Si PD. (b) Simulated static electric field distribution in the Ge region. The bias voltage is -4 V. (c) The cross-sectional transmission electron microscopy (TEM) image of the peaked Ge-Si PD. (d) Optical micrograph of the peaked Ge-Si PD with integrated inductor.
3. Experimental results
3.1 Static measurements
A typical static current-voltage (I-V) characteristic of the peaked PD in the dark illuminated state is shown in Fig. 2(a). The device exhibits dark current as low as 3 nA and 6 nA at -1 V and -4 V bias. The ultra-low dark current might benefit from the lateral Si-contacted p-i-n junction structure. To evaluate the optical absorption efficiency of the peaked Ge-Si PD, the photocurrent was measured as a function of the input power at 1550 nm with -2 and -4 V bias voltage, as depicted in the Fig. 2(b). Here, the external responsivity is defined as the photocurrent/fiber input power. The reduction of optical power injection to the PD mainly results from the coupling losses between single mode fiber and Si-based SSC. Therefore, for the calculation of internal responsivity, the coupling losses should be excluded. In Fig. 2(b), the external and internal responsivities of the peaked Ge-Si PD are shown for 2 and 4 V reverse bias voltage, which are marked with “2 V-External, 4 V-External” and “2 V-Internal, 4V-Internal”. A flat external responsivity spectrum of >0.5 A/W is achieved up to 1565 nm. When the bias voltage increases from 2 V to 4 V, the external responsivity rises a little bit. The internal responsivity spectrums are also calculated and depicted in Fig. 2(b). It can reach to 1 A/W and approximately equal in the C optical communication band under -2 and -4 V bias.
Fig. 2. (a) Current-voltage (I-V) characteristics of peaked Ge-Si PD in the dark illuminated state. (b) Measured external and calculated internal responsivities of the peaked Ge-Si PD in the C band under 2 and 4 V reverse bias voltage.
3.2 Small-signal measurements
The small-signal radio frequency (RF) measurements were implemented to experimentally verify the S11 and S21 response of the un-peaked and peaked Ge-Si PD. The calibration of the high-speed RF trail was used to consider the contributions from GSG probes and coaxial cables. The small-signal measurement setup is depicted in [17]. The bandwidth test experiments were achieved by collecting the response of the S21 transmission parameter in the DC∼110 GHz Lightwave Component Analyzer (LCA) tool versus frequency. Figure 3(a) shows the measured magnitude (S11) performances of the un-peaked and peaked PD under -6 V bias voltage. Their tendency of reflection curves is very similar before 80 GHz. In the range of 70∼100 GHz, the reflection reducing might attribute to the introduced inductor. The measured phase shift characteristics of the un-peaked and peaked Ge-Si PD under -6 V bias voltage are shown in Fig. 3(b). The phase shift of peaked PD affects the pulse fidelity in broad-band systems. The series resistance and p-i-n junction capacitance of un-peaked Ge-Si PD are extracted by fitting S11 scattering parameter data to the small signal RC-model [17]. The series resistance and capacitance are 31 Ω and 21 fF, respectively. In this work, we carefully designed the inductance of PD to decrease the non-uniform time delay for different frequencies, which can be verified by the obtained clear open eye diagrams in Fig. 6 and 7.
Fig. 3. (a) Measured magnitude (S11) performances of the un-peaked and peaked PD under -6 V bias voltage. (b) Measured phase shift (S11) performances of the un-peaked and peaked PD under -6 V bias voltage.
Fig. 4. (a) Measured optic-electro (S21) performance of the un-peaked Ge-Si PD under 2, 4, and 6 V reverse bias voltage. (b) Measured optic-electro (S21) performance of the peaked Ge-Si PD under 2, 4, and 6 V reverse bias voltage.
Figure 4 shows the measured optic-electro (S21) performances of the un-peaked and peaked Ge-Si PD at a photocurrent of 0.8 mA under 2, 4, and 6 V reverse bias voltage. For the un-peaked PD, the 3 dB OE bandwidth is about 23 GHz at -2 V bias voltage, which is limited by the carrier transit time. Increasing the bias to -4 V, the bandwidth rises to 40 GHz. However, the bandwidth enhancement is not obviously under -6 V bias, which is approximately 42 GHz. With the aid of gain peaking technique, the bandwidth is boosted to 80 GHz, as shown in Fig. 4(b). It permits an 95% improvement in bandwidth without sacrificing responsivity and undesired effects. This is a cost-effective and high efficiency approach. More importantly, the optic-electro S21 response curve is very flat without over-peaking phenomenon, which is greatly beneficial to realize high-quality eye diagrams reception.
3.3 Eye diagram measurements
As shown in Fig. 5, the schematic setup of high-speed NRZ, PAM-4, and PAM-8 eye diagram measurement is presented. The high-speed RF signal with word length of 215-1 pseudo random bit sequence (PRBS) data streams is generated by a 256 GS/s arbitrary waveform generator (AWG). After amplified by SHF 66 GHz driver, the RF signal is sent to the thin film lithium niobite on insulator (TFLNOI) Mach-Zehnder modulator, which has a 3 dB bandwidth of 64 GHz after packaging. The modulated optical signal was input into the un-peaked and peaked Ge-Si PD via standard single mode fiber, then converted to current signals. The high-speed electrical signals are output by the GSG probe and RF coaxial-cable. A supply voltage of 6 V was applied to the PD by bias Tee. At transmitter side, the AWG output Vpp is set to 800 and 350 mV for NRZ and PAM-N eye diagrams test, respectively. The Keysight (DCA-X, N1000A) sampling oscilloscope was used to obtain the electrical eye diagrams. At receiver side, the electrical filter of oscilloscope is used to improve the eye diagram quality.
Fig. 5. Schematic of the experimental setup for the measurement of the eye diagrams. The black and red lines represent the optical and electrical connections, respectively. AWG, arbitrary waveform generator; EDFA, Erbium doped fiber amplifier; VOA, variable optical attenuator; PC, polarization controller; PD, photodetector; TFLN MZM, thin film lithium niobite Mach-Zehnder modulator.
From Fig. 4, it is obviously that the 3 dB bandwidth of Ge-Si PD can reach to 80 GHz with the gain peaking technique. Although its S21 response is very flat without over-peaking region, the high-speed large signal reception capability still needs to be explored. An eye diagram provides a freeze-frame display of repetitively sampled digital signals. It is a useful and convincing tool to examine the signal integrity in fiber optic transmission system. Especially, the TDECQ is a significantly method of calculating the penalty for transmitters that have unequal sub-eyes. The TDECQs also depend on the state of the receiver or PD. So, the whole system design is vital to evaluate the performances of transmitter or receiver. The marked TDECQ values in Fig. 6 is to present the Ge-Si PD signal reception capability under different 3 dB bandwidth with the same transmitter conditions. In this work, the TDECQs were measured at a soft-decision forward-error correction threshold (SD-FEC) threshold of symbol error rate (SER) < 4.8E-4 (IEEE 802.3cd). Figure 6 shows the measured 60, 90, 100, and 120 Gbaud PAM-4 eye diagrams of the un-peaked and peaked PD. The first- and second-line eye diagrams are obtained from the (I) un-peaked and (II) peaked PD with the same input power and bias voltage, respectively. Before the speed of 90 Gbaud, their eye diagrams quality is very high and similar. The TDECQ penalties of the 60 and 90 Gbaud eye diagrams are about 2.33 and 2.76 dB, 1.68 and 2.45 dB for the un-peaked and peaked Ge-Si PD, respectively. Further increasing the reception speed to 100 and 120 Gbaud, the eye diagrams quality of peaked PD is clearly superior to the un-peaked PD, which exhibits the powerful of gain peaking technique. Their TDECQ penalties are approximatively 2.53 and 3.99 dB. However, for the un-peaked PD, its TDECQ penalties cannot be calculated by the oscilloscope. It might be attributed to the limited 3 dB bandwidth of 42 GHz which the bit error rate (BER) increases to the threshold of 4.8E-4.
Fig. 6. Measured 60, 90, 100, and 120 Gbaud PAM-4 eye diagrams of the (I) un-peaked and (II) peaked PD. The time scale is 5 ps/div.
To further increasing the high-speed signal reception capability, the PAM-4 and PAM-8 modulation formats are also investigated. The PAM-8 is also an efficient way to obtain higher bit rate with limited bandwidth. Before executing the eye diagram test, the RF loss of link which included coaxial cables and GSG probe was compensated to improve the signal quality. Here, the pre-emphasis is also carried out in the AWG to compensate for the bandwidth limitation of the TFLNOI modulator. The root raised cosine (RRC) filter and roll-off factor of 0.7 are employed at transmitter (TX) side without utilizing any offline digital signal processing (DSP) at receiver (RX) side. Figure 7(a) shows the measured 136, 145, and 156 Gbit/s NRZ, and 128, 136, and 145 Gbaud PAM-4, and 128, 136, and 140 Gbaud PAM-8 clear opening eye diagrams of the peaked Ge-Si PD under -6 V bias voltage. To the best of our knowledge, we report for the first-time a Ge-Si PD operating at 420 Gbit/s per lane in an intensity modulation direct-detection (IM/DD) system. We also compare its large signal reception ability with the commercial 70 GHz Finisar (XPDV3320R) PD under the same test condition, as shown in Fig. 7(b). It is obviously that the eye diagrams of 156 Gbit/s NRZ, 145 Gbaud PAM-4, and 140 Gbaud PAM-8 are as good as the Finisar PD. With advanced packaging technology, the overall performances of Ge-Si PD might be comparable to the InP-based PDs. Here, the PD chip is integrated with on-chip spiral inductor as peaking element without considering the impedance match conditions. When we implement the PD chip electrical packaging without integrated with TIA, the bonding wire length and width is very important, which affects the overall bandwidth performances of PD device. So, with the suitable packaging technology and matched impedance, the PD device bandwidth might not decrease obviously and keep larger than 75 GHz. When the PD chip integrates with TIA by packaging, we can utilize the advanced flip-chip technology. The flip-chip technology can reduce the capacitive, resistive, and inductive parasitics.
Fig. 7. Measured 136, 145, and 156 Gbit/s NRZ, and 128, 136, and 145 Gbaud PAM-4, and 128, 136, and 140 Gbaud PAM-8 eye diagrams of the peaked PD. (b) Measured 156 Gbit/s NRZ, 145 Gbaud PAM-4, and 140 Gbaud PAM-8 eye diagrams of the commercial 70 GHz Finisar PD. The time scale is 3 ps/div.
3.4 Comparison with state of the art
An ideal Ge-Si PD should have a high responsivity, low dark current, and high bandwidth. More importantly, its fabrication process complexity also needs to be as low as possible, which determines the devices yields and costs. Unfortunately, there are several trade-offs that make such devices difficult to achieve simultaneously. Here, we define the figure of merit (FOM) of Ge-Si PD as $\frac{{Responsivity(A/W) \times Bandwidth(GHz)}}{{Dark \cdot current(nA)}}$ . Up to now, various high-speed Ge-Si photodetection structures have been extensively investigated and demonstrated [12]. The characteristics of the representative Ge-Si PDs are summarized in Table 1. The responsivities are calculated at a wavelength of 1550 nm. It is worth noting that a recorded 3 dB cutoff frequency up to 265 GHz has been reported [21]. However, its fabrication process complexity is a bit high, which will influence the yields and costs. The QE and dark current also need to be further improved. The large dark current which is a direct determinant of signal noise will lead to low sensitivity in optical communication system. So, reducing the dark current is one of the most effective ways to enhance the performance of Ge-Si PD. In this work, the bandwidth improvement is achieved by the optimized on-chip inductor without increasing the process procedure. A flat 3 dB bandwidth of 80 GHz is demonstrated with internal responsivity of 1.0 A/W at 1550 nm and dark current of 6 nA at -4 V bias. The FOM of the peaked Ge-Si PD is about 13.3. The fabrication process complexity is relatively low. Furthermore, based on this device, the speed up to 420 Gbit/s PAM-8 clear open eye diagram is also obtained for the first time. These experimentally obtained results thoroughly verify the practical applications of gain peaking technique. Overall, the peaked Ge-Si PD achieves a good trade-off among responsivity, dark current, bandwidth, and fabrication process complexity. In the near future, by utilizing the advanced photonic electronic co-design and flip-chip bonding technology, it is possible to realize high-speed 3D-integrated silicon photonics receiver with sub-pJ/bit power consumption [32–36].
Table 1. Comparison with State-of-the-Art Ge-Si PDsa
4. Discussions
For the design of on chip integrated inductor, the goal is the appropriate inductance to boost the bandwidth. By using a commercial 3-dimensional electromagnetic simulation software, we can design the physical parameters of the inductor. The key parameter of inductor is the width and total length of Cu metal 2. For this Ge-Si PD, the ideal value of inductance is about 170 pH, which is the optimization goal. However, the process variability of silicon photonics platform affects the highly-optimized inductance design. To reducing this influence, we design different structures with the change of inductance. From the test results, we find that the designed structure of inductor satisfies the maximal bandwidth requirement.
As shown in Fig. 2, the very flat photo response up to 1565 nm is tested under room temperature, which is about 20℃. The flat photo response benefits from the high Ge absorption coefficient even at longer wavelength. Moreover, we do the measurements based on the thermoelectric cooler (TEC) from 10 to 60 ℃. As shown in the Fig. 8, when the temperature of Ge-Si PD is reduced, the external responsivity increases a little bit for all the representative wavelengths. After that, the responsivity increases with the temperature. The responsivity of 1565 nm optical wavelength is more sensitive to the temperature variation.
Fig. 8. The external responsivity as a function of temperature for the wavelength of 1530, 1550, and 1565 nm.
For this peaked Ge-Si PD, there are two typical applications: integrated with/without TIA. For the silicon photonics high-speed receiver (Ge-Si PD integrated with TIA), the total noise is mainly dominated by the thermal noise generated in the TIA (dark current in the >μA range). So, the ultralow dark current (<10 nA) is not necessary. The industrial success of high-speed silicon photonics receivers proves it. For some applications, the Ge-Si PD might work independently without integrated with TIA. Then, the ultralow dark current is more helpful to improve the sensitivity.
To further investigate the high-speed performances of un-peaked and peaked Ge-Si PD, we also measure the BER under the speed of 56, 60, 64, 70, 80, 90, 100, 110, 112, and 120 Gbaud PAM-4, as shown in Fig. 9(a). In the measurement process, the two PDs have the same DC photocurrent of about 0.3 mA. For the BER assessment, the commercial super high frequency (SHF) driver the 3 dB bandwidth of 66 GHz is used as the TIA for high-speed RF signal amplification at Tx side. Post-compensation is carried out by off-line digital signal processing (DSP). In the DSP, adaptive 51-tap feedforward equalization (FFE) and adaptive maximum likelihood sequence estimation (MLSE) are utilized to realize channel estimation, inter-symbol-interference (ISI) elimination, and faster than Nyquist transmission. The BERs of un-peaked PD is a little larger than the peaked PD. To explore the sensitivity for our proposed peaked Ge-Si PD receiver, the BER assessments under 100 Gbaud were performed. The BERs vs. received optical power of the un-peaked and peaked PD for the NRZ and PAM-4 modulation formats at 100 Gbaud are shown in Fig. 9(b). The overall BER performances of peaked PD is superior to the un-peaked PD.
Fig. 9. (a) The bit error rate (BER) of un-peaked and peaked PD under the speed of 56, 60, 64, 70, 80, 90, 100, 110, 112, and 120 Gbaud PAM-4. (b) The BER vs. received optical power of un-peaked and peaked PD for 100 Gbaud NRZ and PAM-4 modulation formats.
5. Conclusions
By employing the advanced gain peaking technique, an intrinsic 3 dB bandwidth of 42 GHz Ge-Si PD is boosted up to 80 GHz. It yields an 95% improvement in bandwidth without sacrificing responsivity and undesired effects. Another attractive feature of this technique is that the bandwidth enhancement comes with no additional power consumption. The optic-electro S21 response curve is very flat without over-peaking phenomenon. The external responsivity of 0.5 A/W and internal responsivity of 1.0 A/W at a wavelength of 1550 nm under -4 V bias voltage are also achieved. The external responsivity as a function of temperature for the representative wavelength of 1530, 1550, and 1565 nm is also measured. The external responsivity of 1565 nm wavelength is more sensitive to the temperature variation. To further verify the acceptable non-uniform time delay effect for different frequencies of this peaked PD, the research on large signal reception capability is carried out. Firstly, the eye diagrams comparison of the un-peaked and peaked Ge-Si PD are presented for the speed of 60, 90, 100, and 120 Gbaud PAM-4. Under the same transmitter state, the TDECQ penalties of the 60 and 90 Gbaud PAM-4 eye diagrams are about 2.33 and 2.76 dB, 1.68 and 2.45 dB for the un-peaked and peaked Ge-Si PD, respectively. When the signal reception speeds up to 100 and 120 Gbaud, the eye diagrams quality of the peaked PD is clearly superior to the un-peaked PD, which exhibits the powerful of gain peaking technique. Secondly, based on this peaked PD, the 136, 145, and 156 Gbit/s NRZ, and 128, 136, and 145 Gbaud PAM-4, and 128, 136, and 140 Gbaud PAM-8 eye diagrams are also obtained. Finally, under the same test condition, we find that the eye diagrams of 156 Gbit/s NRZ, 145 Gbaud PAM-4, and 140 Gbaud PAM-8 are as good as the commercial 70 GHz Finisar PD. We also measure the BER performances of the un-peaked and peaked Ge-Si PDs under the speed of 56, 60, 64, 70, 80, 90, 100, 110, 112, and 120 Gbaud PAM-4 and different optical power at 100 Gbaud NRZ and PAM-4 modulation formats. To the best of our knowledge, we report for the first-time a Ge-Si PD operating at 420 Gbit/s per lane in an IM/DD system. These experimentally demonstrated ultra-high-speed eye diagram reception capability would pave the way to practical applications of gain peaking technique. With advanced packaging technology, the overall performances of Ge-Si PD might be comparable to the InP-based PDs in the future.
Funding
National Natural Science Foundation of China (62205255, U21A20454); Young Top-notch Talent Cultivation Program of Hubei Province.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. S. E. Miller, “Integrated optics: An introduction,” Bell Syst. Tech. J. 48(7), 2059–2069 (1969). [CrossRef]
2. Y. Zhao, L. Chen, R. Aroca, N. Zhu, D. Ton, D. Inglis, and C. R. Doerr, “Silicon photonic based stacked die assembly toward 4×200-Gbit/s short-reach transmission,” J. Lightwave Technol. 40(5), 1369–1374 (2022). [CrossRef]
3. C. Doerr and L. Chen, “Silicon photonics in optical coherent systems,” Proc. IEEE 106(12), 2291–2301 (2018). [CrossRef]
4. C. Sun, M. T. Wade, Y. Lee, et al., “Single-chip microprocessor that communicates directly using light,” Nature 528(7583), 534–538 (2015). [CrossRef]
5. B. J. Shastri, A. N. Tait, T. Ferreira de Lima, W. H. Pernice, H. Bhaskaran, C. D. Wright, and P. R. Prucnal, “Photonics for artificial intelligence and neuromorphic computing,” Nat. Photonics 15(2), 102–114 (2021). [CrossRef]
6. G. Zhang, J. Y. Haw, H. Cai, F. Xu, S. M. Assad, J. F. Fitzsimons, X. Zhou, Y. Zhang, S. Yu, J. Wu, W. Ser, L. C. Kwek, and A. Q. Liu, “An integrated silicon photonic chip platform for continuous-variable quantum key distribution,” Nat. Photonics 13(12), 839–842 (2019). [CrossRef]
7. M. Smit, K. Williams, and J. van der Tol, “Past, present, and future of InP-based photonic integration,” APL Photonics 4(5), 050901 (2019). [CrossRef]
8. J. Michel, J. Liu, and L. C. Kimerling, “High-performance Ge-on-Si photodetectors,” Nat. Photonics 4(8), 527–534 (2010). [CrossRef]
9. E. Rouvalis, M. Chtioui, F. van Dijk, F. Lelarge, M. J. Fice, C. C. Renaud, G. Carpintero, and A. J. Seeds, “170 GHz uni-traveling carrier photodiodes for InP-based photonic integrated circuits,” Opt. Express 20(18), 20090–20095 (2012). [CrossRef]
10. J. M. Wun, Y. W. Wang, and J. W. Shi, “Ultrafast uni-traveling carrier photodiodes with GaAs0.5Sb0.5/In0.53Ga0.47As type-II hybrid absorbers for high-power operation at THz frequencies,” IEEE J. Sel. Top. Quantum Electron. 24(2), 1–7 (2018). [CrossRef]
11. X. Hu, D. Wu, H. Zhang, W. Li, D. Chen, L. Wang, X. Xiao, and S. Yu, “High-speed lateral PIN germanium photodetector with 4-directional light input,” Opt. Express 28(25), 38343–38354 (2020). [CrossRef]
12. Y. Shi, Y. Zhang, Y. Wan, Y. Yu, Y. Zhang, X. Hu, X. Xiao, H. Xu, L. Zhang, and B. Pan, “Silicon photonics for high-capacity data communications,” Photonics Res. 10(9), A106–A134 (2022). [CrossRef]
13. D. Benedikovic, L. Virot, G. Aubin, F. Amar, B. Szelag, B. Karakus, J. M. Hartmann, C. Alonso-Ramos, X. L. Roux, P. Crozat, E. Cassan, D. Marris-Morini, C. Baudot, F. Boeuf, J. M. Fédéli, C. Kopp, and L. Vivien, “25 Gbps low-voltage hetero-structured silicon-germanium waveguide pin photodetectors for monolithic on-chip nanophotonic architectures,” Photonics Res. 7(4), 437–444 (2019). [CrossRef]
14. X. Hu, D. Wu, Y. Liu, Min Liu, J. Liu, H. Zhang, Y. Zhang, D. Chen, L. Wang, X. Xiao, and S. Yu, “Single lane beyond 400 Gbit/s optical direct detection based on a sidewall-doped Ge-Si photodetector,” European Conference on Optical Communication (ECOC), Optica Publishing Group, Th2E.5 (2022).
15. H. Chen, M. Galili, P. Verheyen, P. De Heyn, G. Lepage, J. De Coster, S. Balakrishnan, P. Absil, L. Oxenlowe, J. Van Campenhout, and G. Roelken, “100-Gbps RZ data reception in 67-GHz Si-contacted germanium waveguide p-i-n photodetectors,” J. Lightwave Technol. 35(4), 722–726 (2017). [CrossRef]
16. X. Hu, H. Zhang, D. Wu, D. Chen, L. Wang, X. Xiao, and S. Yu, “High-performance germanium avalanche photodetector for 100 Gbit/s photonics receivers,” Opt. Lett. 46(16), 3837–3840 (2021). [CrossRef]
17. X. Hu, D. Wu, H. Zhang, W. Li, D. Chen, L. Wang, X. Xiao, and S. Yu, “High-speed and high-power germanium photodetector with a lateral silicon nitride waveguide,” Photonics Res. 9(5), 749–756 (2021). [CrossRef]
18. X. Hu, D. Wu, D. Chen, L. Wang, X. Xiao, and S. Yu, “180 Gbit/s Si3N4-waveguide coupled germanium photodetector with improved quantum efficiency,” Opt. Lett. 46(24), 6019–6022 (2021). [CrossRef]
19. 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]
20. L. Vivien, A. Polzer, D. Marris-Morini, J. Osmond, J. M. Hartman, P. Crozat, E. Cassan, C. Kopp, H. Zimmermann, and J. M. Fédéli, “Zero-bias 40 Gbit/s germanium waveguide photodetector on silicon,” Opt. Express 20(2), 1096–1101 (2012). [CrossRef]
21. S. Lischke, A. Peczek, J. S. Morgan, K. Sun, D. Steckler, Y. Yamamoto, F. Korndörfer, C. Mai, S. Marschmeyer, M. Fraschke, and A. Krüger, “Ultra-fast germanium photodiode with 3 dB bandwidth of 265 GHz,” Nat. Photonics 15(12), 925–931 (2021). [CrossRef]
22. S. S. Mohan, M. D. M. Hershenson, S. P. Boyd, and T. H. Lee, “Bandwidth extension in CMOS with optimized on-chip inductors,” IEEE J. Solid-State Circuits 35(3), 346–355 (2000). [CrossRef]
23. S. Shekhar, J. S. Walling, and D. J. Allstot, “Bandwidth extension techniques for CMOS amplifiers,” IEEE J. Solid-State Circuits 41(11), 2424–2439 (2006). [CrossRef]
24. 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]
25. A. Novack, M. Gould, Y. Yang, Z. Xuan, M. Streshinsky, Y. Liu, G. Capellini, A. E. 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]
26. C. L. Schow, L. Schares, S. J. Koester, G. Dehlinger, R. John, and F. E. Doany, “A 15-Gb/s 2.4-V optical receiver using a Ge-on-SOI photodiode and a CMOS IC,” IEEE Photonics Technol. Lett. 18(19), 1981–1983 (2006). [CrossRef]
27. D. Y. Wu, X. Hu, W. Z. Li, D. G. Chen, L. Wang, and X. Xiao, “62 GHz germanium photodetector with inductive gain peaking electrode for photonic receiving beyond 100 Gbaud,” J. Semicond. 42(2), 020502 (2021). [CrossRef]
28. X. Li, Y. Zhu, Z. Liu, L. Peng, X. Liu, C. Niu, J. Zheng, Y. Zuo, and B. Cheng, “75 GHz germanium waveguide photodetector with 64 Gbps data rates utilizing inductive-gain-peaking technique,” J. Semicond. 44(1), 012301 (2023). [CrossRef]
29. S. Sahni, N. K. Hon, and G. Masini, “The dual-heterojunction Ge on Si photodetector,” ECS Trans. 64(6), 783–789 (2014). [CrossRef]
30. Y. Zhang, S. Yang, Y. Yang, M. Gould, N. Ophir, A. E. Lim, G. Q. Lo, P. Magill, K. Bergman, T. Baehr-Jones, and M. Hochberg, “A high-responsivity photodetector absent metal-germanium direct contact,” Opt. Express 22(9), 11367–11375 (2014). [CrossRef]
31. D. Chen, H. Zhang, M. Liu, X. Hu, Y. Zhang, D. Wu, P. Zhou, S. Chang, L. Wang, and X. Xiao, “67 GHz light-trapping-structure germanium photodetector supporting 240 Gb/s PAM-4 transmission,” Photonics Res. 10(9), 2165–2171 (2022). [CrossRef]
32. K. Li, S. Liu, X. Ruan, D. J. Thomson, Y. Hong, F. Yang, L. Zhang, C. Lacava, F. Meng, W. Zhang, and P. Petropoulos, “Co-design of a differential transimpedance amplifier and balanced photodetector for a sub-pJ/bit silicon photonics receiver,” Opt. Express 28(9), 14038–14054 (2020). [CrossRef]
33. S. Saeedi, S. Menezo, G. Pares, and A. Emami, “A 25 Gb/s 3D-integrated CMOS/silicon-photonic receiver for low-power high-sensitivity optical communication,” J. Lightwave Technol. 34(12), 2924–2933 (2016). [CrossRef]
34. D. Wu, D. Wang, D. Chen, J. Yan, Z. Dang, J. Feng, S. Chen, P. Feng, H. Zhang, Y. Fu, L. Wang, X. Hu, X. Xiao, and S. Yu, “Experimental demonstration of a 160 Gbit/s 3D-integrated silicon photonics receiver with 1.2-pJ/bit power consumption,” Opt. Express 31(3), 4129–4139 (2023). [CrossRef]
35. G. Dziallas, A. Fatemi, A. Malignaggi, and G. Kahmen, “A 97-GHz 66-dBΩ SiGe BiCMOS low-noise transimpedance amplifier for optical receivers,” IEEE Microw. Wireless Compon. Lett. 31(12), 1295–1298 (2021). [CrossRef]
36. I. G. López, A. Awny, P. Rito, M. Ko, A. C. Ulusoy, and D. Kissinger, “100 Gb/s Differential linear TIAs with less than 10 in 130-nm SiGe:C BiCMOS,” IEEE J. Solid-State Circuits 53(2), 458–469 (2018). [CrossRef]
