Open Access
Add to BibSonomy
Abstract
Photonic microwave generation of high-power pulsed signals in the X-, Ku- and K-band using charge-compensated MUTC photodiodes is demonstrated. The impulse photoresponse without modulation showed a maximum peak voltage of 38.3 V and full-width at half-maximum of 30 ps. High power pulsed microwave signals at 10, 17 and 22 GHz with peak power up to 44.2 dBm (26.3 W), 41.6 dBm (14.5 W) and 40.6 dBm (11.5 W) were achieved, respectively.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Photonic generation of microwave signals has numerous advantages application areas, such as satellite communications, 5G communications, optical sensing and phased array antennas [1]. Previously, generation of continuous-wave microwave signals based on optical heterodyne [2,3], external modulation [4] and other approaches have been investigated [5]. Photonic generation of pulsed microwave signals has the potential to impact radar and wireless applications [6]. Among them, the X (8-12 GHz), Ku (12-18 GHz) and K (18-27 GHz) frequency bands are important and widely used for navigation, airport surveillance, and imaging [7]. Various techniques including spatial light modulator-based pulse shaping [8], optoelectronic oscillation-based pulse phase coding [9], and polarization modulator-based signal-phase coding [10] have been used to generate high-frequency pulsed radio frequency (RF) signals. Recently, Khaldoun et al. reported an optical frequency division system using a femtosecond laser frequency comb to generate microwave pulses with high spectral purity and stability [11]. However, for many of these applications, it is important to generate high-power pulsed RF signals. The photodiode is the key component to achieve this since it usually determines the peak power of the entire system. Previously, photonic generation of 1 GHz and 10 GHz RF signals with peak power of 41.5 dBm (14.2 W) and 40 dBm (10 W) using modified uni-traveling-carrier (CC-MUTC) photodiodes has been reported [12].
In this paper, we report an optimized MUTC structure that has achieved photonic generation of pulsed microwave signals at higher frequencies with higher peak power. The 50 µm-, 40 µm- and 34 µm-diameter photodiodes exhibit 3-dB bandwidth of 16, 20 and 27 GHz, respectively. The photodiodes show impulse photoresponse with 38.3 V peak unsaturated voltage and full width at half maximum of 30 ps. High power pulsed RF signals at X-, Ku- and K- bands with peak power of 44.2 dBm (26.3 W), 41.6 dBm (14.5 W) and 40.6 dBm (11.5 W have been measured, respectively. These are the highest pulsed RF power levels reported to date. The relationship between the maximum peak power and duty cycle was also studied.
2. Device fabrication and characterization
The back-illuminated MUTC photodiode structure was grown on semi-insulating InP substrate by metal organic chemical vapor deposition. The epitaxial layers of the InP/InGaAs wafer are shown in Fig. 1(a). The growth began with an n-type InP contact layer. An 800 nm lightly n-doped drift layer acts as a space charge compensation layer that pre-distorts the electric field in order to achieve a flat electric field profile at high photocurrent [13]. A cliff layer with a doping level of 1×1017 cm−3 and a thickness of 50 nm was used to adjust the electric field profile in the depleted absorber and drift layer [14]. The doping in the p-type undepleted absorber was graded in three steps in order to aid carrier transport. A 100 nm InP layer was used to block the electrons and a 50 nm p-type InGaAs was used as the contact layer.
Fig. 1. (a) Schematic cross section of InGaAs-InP MUTC photodiodes; (b) 3D view of flip-chip bonding process; (c) Photomicrograph of photodiodes.
The CC-MUTC photodiodes with diameters ranging from 34 µm to 50 µm were fabricated using the double-mesa structure reported in [15]. Ti/Pt/Au/Ti and AuGe/Ni/Au metal stacks were used as the p and n contacts, respectively. After polishing, a 194 nm-thick Si3N4 layer was deposited on the back of the wafer as an anti-reflection coating. The wafer was then diced into chips. In order to improve thermal dissipation, the diced chips were flip-chip bonded onto high thermal conductivity diamond submounts with coplanar waveguide (CPW) pads as shown in Fig. 1(b) [16]. Figure 1(c) is a photograph of a diamond submount with six sets of CPWs on the InP chip. The area with white dashed lines indicates the position of the InP chip after flip-chip bonding.
Since pulsed-illumination significantly suppresses the deleterious heating effects, photodiodes can be biased at higher voltage to achieve higher peak power compared to CW operation. The thickness of the absorber region and the drift layer were optimized using the Crosslight simulation tool to achieve high avalanche breakdown voltage as well as suppression of the space charge effect. The avalanche breakdown voltage corresponds to the electric field that generates a self-sustaining avalanche event. This is observed as an abrupt increase in the current. For InGaAs and InP the relationship between the ionization coefficients and the electric field can be found in Refs. [17–19]. It is worth mentioning that the breakdown electric field varies with the layer thickness. The 350 nm depleted absorber and 800 nm InP were found to enable a high junction breakdown voltage of -44 V as showed in Fig. 2(a), with the corresponding breakdown electric fields of 400 kV/cm in InGaAs and 560 kV/cm in InP. The electric field distribution under dark and intense illumination at -32 V and -44 V are shown in Fig. 2(b) and (c). From the field profiles, we see that impact ionization is most probable in the InGaAs layer under low illumination and in the InP layer under intense illumination. This is a result of the change in the electric field profile with current level. For depleted absorber thickness greater than 350 nm, the peak power decreases due to avalanche breakdown because the breakdown field decreases with increasing depletion width. On the other hand, for thickness less than 350 nm, the peak power saturates. In this region for high-level illumination, excess holes and electrons accumulate in the InGaAs intrinsic region and the electric field collapses below the value required to maintain saturation velocity of the carriers. This is the space-charge effect, which results in saturation [1]. The minimum required electric field in the depletion region to maintain electron saturation velocity sets a limit to the intensity of incident light and the maximum generated photocurrent. The simulated electric field collapses when the generated average photocurrent is 68 mA and the duty cycle of the gated signal is 5% at -44 V. At the optimized depleted absorber thickness of 350 nm, the maximum simulated peak power is 28.5 W for a 50 µm-diameter device at 10 GHz.
Fig. 2. (a) Simulated breakdown voltage and peak power versus absorber thickness; Simulated electric field distribution at (b) -32 V and (c) -44 V for 350 nm-thick depleted absorber.
Figure 3 shows the measured dark current versus bias voltage for a 50-µm diameter device. It is clear that tunneling dominates the dark current for bias greater than ∼ -20 V. Avalanche breakdown occurs when the number of impact ionization events and thus the gain M increases dramatically such that 1/M approaches zero [20]. Since the typical abrupt increase of current is not observed, we can only state that the avalanche breakdown does not occur for bias lower than -43 V.
An optical heterodyne setup with optical modulation depth close to 100% was used to characterize responsivity, bandwidth, and saturation current. The measured responsivity at 1550 nm is 0.69 A/W at room temperature. During the bandwidth measurements, the lensed fiber was pulled back until the maximum photocurrent dropped by half in order to achieve a uniform illumination. As shown in Fig. 4, the 50 µm-, 40 µm- and 34 µm-diameter photodiodes exhibit 3-dB bandwidths of 16, 20 and 27 GHz at 110 mA, 50 mA, and 50 mA photocurrent, respectively. A high impedance transmission line provides inductive peaking in order to boost the device bandwidth. The inductor results in a pole in the transfer function of the photodiode, which gives rise to peaking at the resonant frequency. Owing to the inductive peaking, all three devices show flat response up to 10 GHz. The bandwidth enhancement that originates from the self-induced electric field in the graded absorption layer is observed at high photocurrent [21].
Fig. 4. Frequency responses of the flip-chip-bond device with diameter of (a) 50 µm, (b) 40 µm and (c) 34 µm at different photocurrents.
Fig. 5. RF output power of 50 µm diameter device versus average photocurrent for different bias voltages at 10 GHz.
During the saturation power measurement, the photodiodes were mounted on a copper heat sink at room temperature. As shown in Fig. 5 A 50-μm device exhibited saturation current of 205, 265 and 290 mA at 1 dB compression point and corresponding RF output powers of 28.5, 31.2, and 32.3 dBm at −9, −12.5 and −14.7 V bias, respectively. The saturation phenomenon at - 12.5 V and – 9 V are due to the space charge effect, which results from the electric field collapse in the depleted region. Higher bias voltage can suppress the space charge effect and lead to a higher RF output power. The device failed at −14.7 V bias and 290 mA average photocurrent. The heat generated at high bias and high current leads to high junction temperature and thermal-related performance degradation [22,23].
It is well known that thermal failure is the primary performance limitation for MUTC photodiodes under CW illumination. In contrast, the photodiodes under pulsed-illumination can handle higher bias voltage and achieve higher output peak voltage pulses since the generated heat does not have enough time to build up [12]. Generated heat can be dissipated from the junction to the submount between the pulses.
3. Impulse response
A femtosecond Erbium-doped laser with pulse width < 50 fs and 100 MHz repetition rate was used to measure the pulse response. Figure 6 shows normalized output waveforms of the pulse photoresponse measured at -17 V and -36 V. The output peak voltage increases linearly with increasing optical power before saturation as shown in the inset of Fig. 7, and the waveform does not change significantly until it reaches space-charge saturation. The photodiodes exhibit a 22 ps fall time (80%-20%) [24] at average photocurrent of 1 mA and 25 ps at 2.1 mA at -17 V. The fall time remains the same at -36 V as input power increases. The fall time of the photoresponse does not significantly increase due to the fast response of electrons in the depletion layer [25]. When the device reaches saturation, the pulse width increases and the frequency response degrades gradually with increasing photocurrent. At −17 V, the photodiode starts to saturate and the pulse exhibits broadening after an average photocurrent of 1 mA. The pulse experiences larger shape distortion at 2.1 mA. The space-charge induced saturation and concomitant pulse shape distortion can be mitigated by higher bias voltage. At −36 V, the waveform remains unchanged at an average photocurrent of 2.1 mA.
Fig. 7. Maximum peak voltage versus bias voltage. Inset: Output peak voltage versus average optical power at −39.5 V
In order to measure the maximum peak voltage, the input optical power and corresponding average photocurrent were increased until the impulse response completely saturated. Figure 7 shows the maximum peak voltage versus bias voltage for a 50-μm device at room temperature. The slope is approximately 1. The inset plot is the output peak voltage versus optical power at −39.5 V bias voltage. The peak voltage reaches 38.3 V at average photocurrent of 4.6 mA, while the FWHM stays at 30 ps. The peak voltage (38.3 V) and corresponding peak power (29.3 W) are the highest reported for photodiodes with similar bandwidths at 1550 nm [25,26]. The peak voltage is 6 V higher than the previous reported MUTC photodiodes [12]. Even though pulsed operation suppresses thermal degradation, the maximum operating voltage is primarily thermal limited instead of avalanche breakdown.
4. Pulsed microwave signal generation
Figure 8 shows the pulsed microwave signal generation setup. A high power and narrow linewidth CW fiber laser working at 1550 nm was used as the source. The first Mach Zehnder modulator (MZM) was driven by an RF signal with power of 17 dBm. The modulator was biased at quadrature bias point. The gated RF signal was modulated by the second modulator. A bias controller was used to ensure that the second MZM was locked at its null point and the RF signal was confined within the pulse. The pulsed optical signal was amplified by an EDFA and the output optical intensity was controlled by a digital optical attenuator. The pulsed RF modulated optical signal was converted to a pulsed RF electrical signal by the MUTC photodiode. The output signal was then measured by a spectrum analyzer and an oscilloscope simultaneously after a bias tee. A microwave attenuator was used before the oscilloscope and spectrum analyzer due to the power limitations of the instruments. The waveform and spectrum of the pulsed RF signal are also shown in Fig. 8. The individual spectral components of the pulsed RF signal were measured with the spectrum analyzer. The peak power can be calculated using the equation [27]:
where τ is pulse width, T is pulse period, and Pin is the measured power at the central line of the main lobe in the spectrum. The peak power can also be calculated based on the peak voltage read from the oscilloscope. The RF signal was 100% modulated and was perfectly confined within the gate.The normalized waveforms of the measured pulsed signal with 4 GHz RF carrier signals are shown in Fig. 9. The corresponding input optical signals are shown in the inset. Three pulse repetition rates were used: 30 MHz, 20 MHz, and 10 MHz with a 5% fixed duty cycle. The corresponding pulse widths were 1.7 ns, 2.5 ns, and 5 ns, respectively. The waveforms were measured at reverse bias of -8 V and average photocurrent of 2 mA. Note that a bias point controller was used to remove the bias point shift and mitigate signal oscillation in the off state. The pulse distortion is primarily due to the pattern generator waveform distortion and EDFA saturation [28,29].
Figures 10(a), 10(b) and 10(c) show the peak powers of the pulsed 10, 17 and 22 GHz RF signals versus average photocurrent with different diameter devices. The experiments were conducted at room temperature at voltages in the range from -8 V to -32 V. The pulse repetition rate was fixed at 30 MHz and the duty cycle was 5%. The peak power increases quadratically with the photocurrent before reaching saturation, which is primarily due to the space-charge effect. For a 50-μm photodiode, the peak power begins to saturate at 11 mA average photocurrent as shown in Fig. 10(a) at -8 V. When the bias increases to −32 V, the maximum average photocurrent reaches 56 mA and the corresponding output peak power is 43.4 dBm. Larger bias voltage helps to mitigate the space-charge effect and thus enhances the peak power. The peak powers for gated modulation at 17 and 22 GHz are 41 dBm and 39.8 dBm for 40-μm and 34-μm photodiodes, respectively. The lower peak power at high frequencies is due to the higher current densities in smaller photodiodes. All the photodiodes failed before -32 V. Since avalanche breakdown for this device occurs at bias > -43 V as shown in the simulations, we concluded that the devices are limited by the thermal degradation rather than avalanche phenomena at 5% duty cycle.
Fig. 10. Peak power measured at (a) 10 GHz, (b) 17 GHz and (c) 22 GHz with 30 MHz repetition rate for different photocurrents at room temperature; (d) Maximum peak power and bias voltage at different duty cycles
Figure 10(d) shows the maximum peak power and bias voltage versus different duty cycles with 30 MHz repetition rate at room temperature. The maximum peak powers were obtained by increasing the photocurrent and the bias voltage until the devices failed. It can be observed that the bias voltage is higher at failure with the lower duty cycle. As duty cycle increases from 5% to 100%, the maximum peak power decreases by 11.3 dB and the maximum bias voltage decreases by 17.3 V. It can be concluded that thermal degradation is mitigated and the peak power is enhanced with decreasing duty cycle.
In order to verify the above assumption that the peak power is limited by thermal degradation, the pulsed RF power was measured at -10℃ using a thermoelectric cooler. Figures 11(a), (b) and (c) show the peak powers for different photocurrent at -10 °C. The photodiodes with 50 µm, 40 µm and 34 µm diameter achieved peak powers of 44.2 dBm (26.3 W), 41.6 dBm (14.5 W) and 40.6 dBm (11.5 W) at 10, 17 and 22 GHz, respectively. The peak power increases at low temperature. For duty cycle of 5% and repetition rate of 30 MHz, the thermal degradation of photodiodes is more significant than junction breakdown.
Fig. 11. Peak power measured at (a) 10 GHz, (b) 17 GHz and (c) 22 GHz with 30 MHz repetition rate for different photocurrents at -10 ℃; (d) Maximum peak power at different repetition rate.
The relationship between maximum peak power and pulse width/repletion rate was also studied. As shown in Fig. 11(d), the maximum peak power increases from 38 dBm to 42.6 dBm when the pulse repetition rate increases from 1 MHz to 30 MHz at -28 V and 5% duty cycle for the 50 µm-diameter device. The peak power performance is improved with increasing repetition rate and decreasing pulse width because the shorter pulses forestall thermal failure and experience less gain saturation in the EDFA [28]. The pulse width of the gate signal, the heat dissipation capacity and the saturation characteristics of the EDFA limit the maximum peak power.
5. Conclusion
We have demonstrated photonic microwave generation of high-power pulsed signals using flip-chip bonded MUTC photodiodes. The photodiodes were designed with avalanche breakdown voltage higher than 43 V. The pulse photoresponse of these photodiodes achieved peak voltage of 38.3 V and FWHM of 30 ps. The peak powers were 44.2 dBm (26.3 W), 41.6 dBm (14.5 W) and 40.6 dBm (11.5 W) at 10, 17 and 22 GHz, respectively.
Funding
Air Force Research Laboratory.
Disclosures
The authors declare that there are no conflicts of interest related to this article.
References
1. A. J. Seeds, “Microwave photonics,” IEEE Trans. Microwave Theory Tech. 50(3), 877–887 (2002). [CrossRef]
2. L. Goldberg, H. F. Taylor, J. F. Weller, and D. M. Bloom, “Microwave signal generation with injection locked laser diodes,” Electron. Lett. 19(13), 491–493 (1983). [CrossRef]
3. G. J. Simonis and K. G. Purchase, “Optical generation, distribution, and control of microwaves using laser heterodyne,” IEEE Trans. Microwave Theory Tech. 38(5), 667–669 (1990). [CrossRef]
4. G. Qi, J. Yao, J. Seregelyi, S. Paquet, and C. Belisle, “Optical generation and distribution of continuously tunable millimeter-wave signals using an optical phase modulator,” J. Lightw. Technol. 23(9), 2687–2695 (2005). [CrossRef]
5. T. Fortier, F. Quinlan, A. Hati, C. Nelson, J. Taylor, Y. Fu, J. Campbell, and S. Diddams, “Photonic microwave generation with high-power photodiodes,” Opt. Lett. 38(10), 1712–1714 (2013). [CrossRef]
6. H. Chi and J. Yao, “An approach to photonic generation of high-frequency phase-coded RF pulses,” IEEE Photonics Technol. Lett. 19(10), 768–770 (2007). [CrossRef]
7. M. I. Skolnik, “Radar handbook,” (1970).
8. J. D. McKinney, D. E. Leaird, and A. M. Weiner, “Millimeter-wave arbitrary waveform generation with a direct space-to-time pulse shaper,” Opt. Lett. 27(15), 1345–1347 (2002). [CrossRef]
9. S. Zhu, X. Fan, M. Li, N. Zhu, and W. Li, “Photonic generation and transmission of phase-modulated microwave signals,” In Semiconductor Lasers and Applications IX, 11182(111820B), (2019).
10. Y. Zhang and S. Pan, “Generation of phase-coded microwave signals using a polarization-modulator-based photonic microwave phase shifter,” Opt. Lett. 38(5), 766–768 (2013). [CrossRef]
11. K. Saleh, J. Millo, B. Marechal, B. Dubois, A. Bakir, A. Didier, C. Lacroûte, and Y. Kersalé, “Photonic Generation of High Power, Ultrastable Microwave Signals by Vernier Effect in a Femtosecond Laser Frequency Comb,” Sci. Rep. 8(1), 1997 (2018). [CrossRef]
12. X. Xie, K. Li, Y. Shen, Q. Li, J. Zang, A. Beling, and J. C. Campbell, “Photonic generation of high-power pulsed microwave signals,” J. Lightwave Technol. 33(18), 3808–3814 (2015). [CrossRef]
13. Z. Li, H. Chen, H. Pan, A. Beling, and J. C. Campbell, “High-power integrated balanced photodetector,” IEEE Photonics Technol. Lett. 21(24), 1858–1860 (2009). [CrossRef]
14. N. Shimizu, N. Watanabe, T. Furuta, and T. Ishibashi, “InP-InGaAs un-traveling-carrier photodiode with improved 3-dB bandwidth of over 150 GHz,” IEEE Photonics Technol. Lett. 10(3), 412–414 (1998). [CrossRef]
15. Y. Peng, J. Zang, K. Sun, Z. Yang, and J. C. Campbell, “High-Speed and High-Power MUTC Photodiode Working at 1064 nm,” IEEE Photonics Technol. Lett. 31(19), 1584–1587 (2019). [CrossRef]
16. N. Duan, X. Wang, N. Li, H. D. Liu, and J. C. Campbell, “Thermal analysis of high-power InGaAs–InP photodiodes,” IEEE J. Quantum Electron. 42(12), 1255–1258 (2006). [CrossRef]
17. J. S. Ng, C. H. Tan, J. P. R. David, G. Hill, and G. J. Rees, “Field dependence of impact ionization coefficients in In0.53Ga0.47As,” IEEE Trans. Electron Devices 50(4), 901–905 (2003). [CrossRef]
18. G. Meneghesso, A. Neviani, R. Parisotto, L. Vendrame, M. Hafizi, W. E. Stanchina, C. Canali, and E. Zanoni, “Measurement of the Electron Ionization Coefficient Temperature Dependence in InGaAs-based Heterojunction Bipolar Transistors,” In ESSDERC'96: Proceedings of the 26th European Solid State Device Research Conference, pp. 447–450. IEEE, 1996.
19. Kenko Taguchi, Toshitaka Torikai, Yoshimasa Sugimoto, Kikuo Makita, and Hisahiro Ishihara, “Temperature dependence of impact ionization coefficients in InP,” J. Appl. Phys. 59(2), 476–481 (1986). [CrossRef]
20. H. T. J. Meier, “Design, characterization and simulation of avalanche photodiodes,” PhD diss., ETH Zurich, 2011.
21. N. Shimizu, N. Watanabe, T. Furuta, and T. Ishibashi, “Improved response of uni-traveling-carrier photodiodes by carrier injection,” Jpn. J. Appl. Phys. 37(Part 1, No. 3B), 1424–1426 (1998). [CrossRef]
22. Y. Shen, J. Gaskins, X. Xie, B. M. Foley, R. Cheaito, P. E. Hopkins, and J. C. Campbell, “Thermal analysis of high-power flip-chip-bonded photodiodes,” J. Lightwave Technol. 35(19), 4242–4246 (2017). [CrossRef]
23. K. J. Williams and R. D. Ekman, “Design considerations for high-current photodetectors,” J. Lightwave Technol. 17(8), 1443–1454 (1999). [CrossRef]
24. Y. Liao, G. Lin, H. Kuo, K. Feng, and M. Feng, “10-gb/s operation of an In/sub 0.53/Ga/sub 0.47/As pin photodiode on metamorphic InGaP buffered semi-insulating GaAs substrate,” IEEE Photonics Technol. Lett. 18(17), 1822–1824 (2006). [CrossRef]
25. H. Ito, S. Kodama, Y. Muramoto, T. Furuta, T. Nagatsuma, and T. Ishibashi, “High-speed and high-output InP-InGaAs unitraveling-carrier photodiodes,” IEEE J. Sel. Top. Quantum Electron. 10(4), 709–727 (2004). [CrossRef]
26. J. Taylor, S. Datta, A. Hati, C. Nelson, F. Quinlan, A. Joshi, and S. Diddams, “Characterization of power-to-phase conversion in high-speed P-I-N photodiodes,” IEEE Photonics J. 3(1), 140–151 (2011). [CrossRef]
27. Agilent Application Note 150-2 (2012).
28. K. Sasahira, K. Kuroda, and Y. Yoshikuni, “Pulse amplification characteristics of a CW pumped erbium-doped fiber amplifier,” in Proc. Opto Electron. Commun. Conf., 346–347 (2010).
29. K. Kuroda, K. Sasahira, and Y. Yoshikuni, “Gain saturation of a CW-pumped erbium-doped fiber amplifier for nanosecond pulses,” Opt. Fiber Technol. 18(1), 44–46 (2012). [CrossRef]
