Abstract
Recently, microwave photonic techniques have emerged to address the challenges that microwave systems face under high-frequency or wideband conditions. To a large extent, the performance of microwave photonic systems depends on the performance of individual optoelectronics devices, such as high-power and high power conversion efficiency photodiodes. Here, we report on a flip chip bonded on a diamond InP/InGaAs modified unitraveling carrier (MUTC) photodiode with record RF output powers of 32.7 dBm (1.86 W), 29.6 dBm, 28.2 dBm, and 26.2 dBm at 10, 15, 20, and 25 GHz, respectively, without active cooling. The corresponding dissipated powers are 34 dBm (2.5 W), 32.3 dBm, 30.4 dBm, and 28.3 dBm, respectively. Compared with previously reported RF power, the device on the diamond submount achieves higher RF output power. Using the high-power and high-frequency MUTC photodiode on diamond submount, a record power conversion efficiency of 50.7%–60% at 6–10 GHz with RF output power has been achieved as compared to previously reported efficiencies in the range in the corresponding frequency band.
© 2014 Optical Society of America
1. INTRODUCTION
High output power and high power conversion efficiency (PCE) in a wide-frequency band can significantly improve the performance of microwave systems. However, the PCE of high-power electronic amplifiers drops with frequency, and typically ranges from 10% to 30% [1]. Microwave photonics have demonstrated the potential to significantly influence a wide range of applications, including analog photonic links [2], distributed antenna arrays [3], and photonic microwave generation [4] owing to large instantaneous bandwidth, the low attenuation of optical fibers, as well as size, weight, and power consumption benefits. High-power, high-PCE, and high-frequency photodiodes enable high link gain, high power, low noise figure, and high dynamic range links [2]. In addition, for the photonic generation of low-noise microwave signals, these photodiodes permit the replacement of demanding postdetection radio frequency (RF) amplifiers with low-noise optical amplifiers. Recently, several high-performance photodiodes have been reported [5–27]. A 53.5% PCE with 24.4 dBm output power at 300 MHz was reported in [26]. The high output power and high linearity achieved by charge-compensated modified unitraveling carrier (CC-MUTC) photodiodes [15] have resulted in the generation of high-power microwave generation with a record low phase noise floor at 10 GHz [4]. However, when the photodiode operates at high current and high bias voltage, Joule heating will ultimately result in thermal failure. Higher RF output power necessitates improved thermal management. Among the various techniques to improve thermal dissipation, flip chip bonding has achieved the best results [5,6,8,12–14,21]. In this work, in order to improve heat dissipation further, CC-MUTC photodiodes have been flip chip bonded to a chemical vapor deposition (CVD) diamond submount. Owing to high thermal conductivity of the CVD diamond, the photodiodes with diameters of 28, 34, 40, and 50 μm achieved record RF output powers of 26.2 dBm at 25 GHz, 28.2 dBm at 20 GHz, 29.6 dBm at 15 GHz, and 32.7 dBm at 10 GHz, respectively, without active cooling. Compared with a 50 μm diameter device on an aluminum nitride (AlN) submount [8], the device on a diamond submount achieves 80% greater maximum RF output power. Two-dimensional thermal profiles of the device without submount, with AlN submount, and diamond submount were simulated. As expected, the results show that devices on diamond submounts achieve higher RF power. By low-biasing the optoelectronic modulator, the modulation depth of the input optical signal of photodiodes can be enhanced, and the operation condition of the photodiode changes from class A to class AB (note that the definitions of class A and class AB can be found in ref. [1]). In combination with this modulation-depth-enhancement technique, we report high PCE in the frequency band of 6 to 10 GHz. The device achieved 60% PCE with 27.8 dBm RF output power and 100 mA photocurrent at 6 GHz, which approaches the class AB PCE limit based on optoelectronic modulator (63.5% at 100 mA).
2. DEVICE FABRICATION AND DESIGN
Table 1 shows the epitaxial structure of the wafer, which is same as the previously reported photodiodes [15]. The epitaxial layer structure was grown by metal–organic CVD on semi-insulating InP substrate. Figure 1(a) shows the schematic cross-sectional view. In the CC-MUTC photodiode, the -doped drift layer acts as a space-charge compensation layer where the electrical field is predistorted to achieve a flat electric field profile at high photocurrent. In addition, a cliff layer is used to enhance the electric field in the depleted portion of the absorption layer in order to assist electron transport across the heterojunction interface at high photocurrent. The built-in electric field in the graded doped absorption layer assists diffusion of the photogenerated electrons, which enhances the bandwidth. Figure 1(b) shows the band diagram and electrical field distribution of PIN, UTC, and CC-MUTC photodiodes. Compared with a typical PIN photodiode and conventional UTC photodiode, the CC-MUTC photodiode is designed to mitigate saturation, which occurs when the electrical field drops below a critical value due to space charge and thus reduces the carrier velocity below the saturation velocity [15,28]. These modifications to the conventional UTC structure have resulted in high saturation current and high RF output power [15]. The active area of the device was defined by a dry-etched double-mesa procedure. A Ti/Pt/Au metal stack was deposited as both - and -type contacts. In order to flip chip bond chips on the diamond submount shown in Fig. 1(c), Au bonding bumps with a diameter of 6 μm and height of 2 μm were plated on the and contacts. A 250 μm thick layer was deposited on the back of the wafer as an antireflective coating, and then the wafer was diced into chips. The diced chips were flip chip bonded onto a high-thermal-conductivity diamond submount with coplanar waveguide (CPW) pads using a FINEPLACER pico ma system. Figure 1(d) shows a scanning electron microscope (SEM) image of the flip-chip-bonded device. Through the Au bonding bumps and CPW pads, the Joule heat generated in the junction can be dissipated into the diamond submount.

Fig. 1. (a) Simplified schematic cross-sectional view of a photodiode flip chip bonded on diamond submount. (b) Band diagrams and electrical field distributions of PIN, UTC, and CC-MUTC photodiodes. (c) Photomicrograph of a diamond submount with CPW pads. (d) SEM image of the flip chip bonded device.

Table 1. Epitaxial Layer Structure of CC-MUTC Photodiodes
3. THERMAL SIMULATION
In order to illustrate how the submount reduces the device temperature, a 3-D thermal model based on the finite-element simulator COMSOL Multiphysics was created. In [14], the surface temperatures of the MUTC photodiodes were measured using a thermal-reflectance imaging method. Since the device structure in this work is the same as that in [14], the simulation parameters were based on the imaging results in [14]. The simulation results are shown in Fig. 2. In our experiment, the 50 μm device on diamond submount failed at dissipated power of 2.5 W, and the corresponding calculated failure temperature is 438 K. Figure 2(a) shows that, for devices without a submount or with an AlN submount at the same failure temperature, the dissipated power is estimated to be 0.7 and 1.6 W, respectively. Compared with a device without a submount and a device bonded on AlN, the device flip-chip-bonded to diamond exhibits a dissipated power enhancement of and , respectively, which agrees with experimental results. Figures 2(b)–2(d) show the simulated thermal profiles of the three kinds of devices at the same dissipated power of 1.5 W. It is clear that diamond effectively reduces the junction temperature.

Fig. 2. (a) Junction temperature of the device at different dissipated powers and at a fixed dissipated power of 1.5 W for (b) back-illuminated photodiode, (c) photodiode flip chip bonded on AlN submount, and (d) photodiode flip chip bonded on diamond submount.
4. DEVICE CHARACTERIZATION AND RESULTS
An optical heterodyne setup with a modulation depth close to 100% was used to measure responsivity, bandwidth, and saturation characteristics. Responsivity and dark current at 10 V were and 500 nA, respectively. In the saturation measurement, the lensed fiber that illuminated the devices was pulled back to the position where the photocurrent dropped to half the peak photocurrent in order to maintain spatially uniform illumination. All devices under test were placed on a copper heat sink with a temperature of . As shown in Fig. 3(a), the 28, 34, 40, and 50 μm diameter photodiodes exhibit bandwidths of 28, 19, 15, and 10 GHz, respectively, with saturation currents of 128, 167, 234, and 300 mA. In Fig. 3(b), we plot the output power versus photocurrent. The corresponding maximum RF output powers are 26.2, 28.2, 29.6, and 32. 7 dBm (1.86 W), respectively. The corresponding dissipated powers are 28.3, 30.4, 32.3, and 34 dBm (2.5 W), respectively. As the photocurrent increases, the output RF power approaches the ideal RF power. This is the well-known bandwidth enhancement that originates from the current-induced electric field in the graded absorption layer [15]. The summary of RF output power versus frequency is plotted in Fig. 4. The output power of a 50 μm diameter device on diamond submount at 10 GHz is 1.86 W (without active cooling), and that for a similar device on AlN at is 1.0 W [8]. For the devices with diameters of 40, 34, and 28 μm on diamond, the output RF powers improve 22.7% at 15 GHz, 78.4% at 20 GHz, and 35.5% at 25 GHz, respectively. It should be noted that 1.5 W RF output power at 8 GHz was achieved by balanced photodiodes and active cooling [21]. Besides, the flip chip bonded photodiodes own larger failure power density. The failure power density of a 50-μm photodiode on diamond submount, as defined by the product of the reverse bias and the DC current at thermal failure, increased by 300% and 50% compared with that of a photodiode without a submount and a photodiode flip chip bonded to AlN.

Fig. 3. (a) Frequency response for different PD diameters. (b) RF output power versus average photocurrent at different frequencies and reverse bias voltages.
5. MODULATION DEPTH ENHANCEMENT TECHNIQUE
Some of the design aspects of the CC-MUTC structure promote high PCE. The combination of drift layer, cliff layer, and partially depleted absorption layer help to maintain high electric field across the heterojunction interfaces, which ultimately allows more of the electric and input optical power to be converted to RF output power. The PCE of an ideal photodiode can be expressed as [26]
where is the average power of the fundamental RF signal at the load, is the electrical power delivered to the photodiode, is the input average optical power, is the modulation depth of the fundamental RF signal, is the load resistance, and , , and are the average photocurrent, bias voltage, and responsivity of the photodiode, respectively. Equation (1) illustrates the dependence of PCE on modulation depth. The maximum PCE equals when approaches infinity. For the optical heterodyne setup used in this work, the photodiode was driven by a sinusoidal optical intensity envelope with maximum modulation depth of 100%, which would indicate that PCE should approach 50% at high photocurrent [1,26]. This agrees with classic class A operation of electronic transistor amplifiers. In our experiment, the PCE for sinusoidal optical intensity envelope was determined from the saturation characteristics. Figures 5(a) and 5(b) show the PCE versus frequency for heterodyne modulation and an external modulator, respectively. The PCE is 42%, 37.7%, and 37% at 10, 20, and 25 GHz, respectively, which compares favorably with the published results at the corresponding frequencies. PCE versus average photocurrent is plotted in Fig. 5(c). Class A operation is denoted by the orange dashed line.
Fig. 5. (a) and (b) PCE of the photodiode driven by heterodyne and modulator setup, respectively. Insert of (b) is the relationship between modulation depth and bias point of modulator. (c) Summary of PCE versus average photocurrent.
However, modulation depth of optical signal can be enhanced by using an optical filter [29] and stimulated Brillouin scattering [30], where the modulated optical signal has a modulation depth capable of reaching [30]. In this work, modulation depth is enhanced based on a Mach–Zehnder modulator by adjusting the bias point of the modulator and the power of the input RF signal delivered to the modulator. Figure 6 shows a block diagram of the setup. The modulation depth of the optical signal can be expressed as
where , and , , and are the half-wave voltage, bias voltage, and RF signal voltage of the modulator, respectively. The DC and AC optical power decrease as the modulator is biased away from quadrature toward its null point (), but the DC optical power decreases faster initially and then slower than the AC optical power [31]. Therefore, the modulation depth will initially increase, peak at a specific bias point, and finally decrease rapidly. This is illustrated in the inset of Fig. 5(b). The maximum is 1.414 when (close to null bias point) and (small input RF signal). The maximum PCE for the modulation-depth-enhancement technique is . When the modulator is biased at the quadrature point, the photodiode works at Class A condition, i.e., the input RF signal to the modulator is impressed on the signal to the photodiode. When the modulator is biased toward its null point, a part of the optical envelope (which corresponds with the input RF signal to the modulator) will be cut off, or “clipped,” and the photodiode operates in the Class AB condition. When the operation of a transistor changes from Class A to Class AB, the PCE increases. The amplitude of fundamental signal will increase at first and then drop beyond a specific point, and the DC component will decrease [1]. In other words, the modulation depth is enhanced when the operation changes from Class A to Class AB. The blue dashed line in Fig. 5(c) illustrates the relationship between Class AB PCE and average photocurrent. By optimizing the bias voltage of the modulator () and the RF signal voltage (), 60%, 52%, and 51% PCE with 28.3, 27.7, and 27.6 dBm RF output power at 6, 8, and 10 GHz, respectively, have been achieved [Fig. 5(b)], which approaches the class AB PCE limit based on optoelectronic modulator (63.5% at 100 mA), which is plotted as the blue dashed line in Fig. 5(c).6. CONCLUSION
We have demonstrated 1.8 W output power at 10 GHz and 60% PCE at 6 GHz with a flip chip bonded on diamond photodiode chip without active cooling. These CC-MUTC photodiodes achieved RF output powers of 29.6, 28.2, and 26.2 dBm at 15, 20, and 25 GHz, respectively. The failure power density on the diamond submount was 300% and 50% higher than that of photodiodes without flip chip bonding and bonded to an AlN submount, respectively. Based on the high-power and high-frequency MUTC photodiode on diamond submount and modulation-depth-enhancement technique, power conversion efficiencies of 51%–60% at 6–10 GHz with RF output power were achieved.
FUNDING INFORMATION
Defense Advanced Research Projects Agency (DARPA) (PULSE program); Naval Research Laboratory.
REFERENCES
1. I. Bahl, Fundamentals of RF and Microwave Transistor Amplifiers (Wiley, 2009).
2. V. J. Urick, F. Bucholtz, J. D. McKinney, P. S. Devgan, A. L. Campillo, J. L. Dexter, and K. J. Williams, “Long-haul analog photonics,” J. Lightwave Technol. 29, 1182–1205 (2011). [CrossRef]
3. A. Wootten and A. R. Thompson, “The Atacama large millimeter/submillimeter array,” Proc. IEEE 97, 1463–1471 (2009). [CrossRef]
4. T. M. Fortier, F. Quinlan, A. Hati, C. Nelson, J. A. Taylor, Y. Fu, J. Campbell, and S. A. Diaddams, “Photonic microwave generation with high-power photodiodes,” Opt. Lett. 38, 1712–1714 (2013). [CrossRef]
5. A. Beling, A. S. Cross, Q. Zhou, Y. Fu, and J. C. Campbell, “High-power flip-chip balanced photodetector with >40 GHz bandwidth,” in Photonics Conference (IEEE, 2013), pp. 352–353.
6. A. S. Cross, Q. Zhou, A. Beling, Y. Fu, and J. C. Campbell, “High-power flip-chip mounted photodiode array,” Opt. Express 21, 9967–9973 (2013). [CrossRef]
7. K. Sakai, E. Ishimura, M. Nakaji, S. Itakura, Y. Hirano, and T. Aoyagi, “High-current back-illuminated partially depleted-absorber p-i-n photodiode with depleted nonabsorbing region,” IEEE Trans. Microwave Theor. Tech. 58, 3154–3160 (2010). [CrossRef]
8. Q. Zhou, A. S. Cross, F. Yang, A. Beling, and J. C. Campbell, “Development of narrowband modified uni-travelling-carrier photodiodes with high power efficiency,” in Avionics, Fiber-Optics and Photonics Conference (IEEE, 2013), pp. 65–66.
9. X. Li, S. Demiguel, N. Li, J. C. Campbell, D. L. Tulchinsky, and K. J. Williams, “Backside illuminated high saturation current partially depleted absorber photodetecters,” Electron. Lett. 39, 1466–1467 (2003). [CrossRef]
10. S. Itakura, K. Sakai, T. Nagatsuka, E. Ishimura, M. Nakaji, H. Otsuka, K. Mori, and Y. Hirano, “High-current backside-illuminated photodiode array module for optical analog links,” J. Lightwave Technol. 28, 965–971 (2010). [CrossRef]
11. D. A. Tulchinsky, J. B. Boos, D. Park, P. G. Goetz, W. S. Rabinovich, and K. J. Williams, “High-current photodetectors as efficient, linear, and high-power RF output stages,” J. Lightwave Technol. 26, 408–416 (2008). [CrossRef]
12. Q. Zhou, A. S. Cross, F. Yang, A. Beling, and J. C. Campbell, “High-power high-bandwidth flip-chip bonded modified uni-traveling carrier photodiodes,” in Photonics Conference (IEEE, 2012), pp. 306–307.
13. Q. Zhou, A. S. Cross, A. Beling, Y. Fu, Z. W. Lu, and J. C. Campbell, “High-power V-band InGaAs/InP photodiodes,” IEEE Photon. Technol. Lett. 25, 907–909 (2013). [CrossRef]
14. Z. Li, Y. Fu, M. Piels, H. Pan, A. Beling, J. E. Bowers, and J. C. Campbell, “High-power high-linearity flip-chip bonded modified uni-traveling carrier photodiode,” Opt. Express 19, B385–B390 (2011). [CrossRef]
15. Z. Li, H. P. Pan, H. Chen, A. Beling, and J. C. Campbell, “High-saturation-current modified uni-traveling-carrier photodiode with cliff layer,” IEEE J. Quantum Electron. 46, 626–632 (2010). [CrossRef]
16. M. Chtioui, A. Enard, D. Carpentier, S. Bernard, B. Rousseau, F. Lelarge, F. Pommereau, and M. Achouche, “High-power high-linearity uni-traveling-carrier photodiodes for analog photonic links,” IEEE Photon. Technol. Lett. 20, 202–204 (2008). [CrossRef]
17. T. Nagatsuma, N. Sahri, M. Yaita, T. Ishibashi, N. Shimizu, and K. Sato, “All optoelectronic generation and detection of millimeter-wave signals,” in International Topical Meeting on Microwave Photonics (IEEE, 1998), pp. 5–8.
18. N. Shimizu, Y. Miyamoto, A. Hirano, K. Sato, and T. Ishibashi, “RF saturation mechanism of InP/InGaAs uni-travelling-carrier photodiode,” Electron. Lett. 36, 750–751 (2000). [CrossRef]
19. N. Duan, X. Wang, N. Li, H. Liu, and J. C. Campbell, “Thermal analysis of high-power InGaAs–InP photodiodes,” IEEE J. Quantum Electron. 42, 1255–1258 (2006). [CrossRef]
20. N. Li, X. Li, S. Demiguel, X. Zheng, J. C. Campbell, D. A. Tulchinsky, K. J. Williams, T. D. Isshiki, G. S. Kinsey, and R. Sudharsansan, “High-saturation-current charge-compensated InGaAs-InP uni-traveling-carrier photodiode,” IEEE Photon. Technol. Lett. 16, 864–866 (2004). [CrossRef]
21. Q. Zhou, A. S. Cross, F. Yang, A. Beling, B. M. Foley, P. E. Hopkins, and J. C. Campbell, “Balanced InP/InGaAs photodiodes with 1.5-W output power,” IEEE Photon. J. 5, 6800307 (2013). [CrossRef]
22. D. A. Tulchinsky, X. Li, N. Li, S. Demiguel, J. C. Campbell, and K. J. Williams, “High-saturation current wide-bandwidth photodetectors,” IEEE J. Sel. Top. Quantum Electron. 10, 702–708 (2004). [CrossRef]
23. X. Wang, N. Duan, H. Chen, and J. C. Campbell, “InGaAs–InP photodiodes with high responsivity and high saturation power,” IEEE Photon. Technol. Lett. 19, 1272–1274 (2007). [CrossRef]
24. T. Nagatsuka, S. Itakura, K. Sakai, and Y. Hirano, “High-power microwave photodiode array for radio over fiber applications,” in Opto-Electronics and Communications Conference (IEEE, 2008), pp. 1–2.
25. M. Chtioui, A. Enard, D. Carpentier, F. Lelarge, B. Rousseau, M. Achouche, A. Marceaux, A. Renoult, C. Feuillet, M. Queguiner, and T. Merlet, “High power UTC photodiodes design and application for analog fiber optic links,” in International Topical Meeting on Microwave Photonics (IEEE, 2009), pp. 1–4.
26. U. Gliese, K. Colladay, A. S. Hastings, D. A. Tulchinsky, V. J. Urick, and K. J. Williams, “53.5% photodiode RF power conversion efficiency,” in National Fiber Optic Engineers Conference 2010 (Optical Society of America, 2010), paper PDPA7.
27. E. Rouvalis, F. N. Baynes, X. Xie, K. Li, Q. Zhou, F. Quinlan, T. M. Fortier, S. A. Diddams, A. G. Steffan, A. Beling, and J. C. Campbell, “High-power and high-linearity photodetector modules for microwave photonic applications,” J. Lightwave Technol. 32, 3810–3816 (2014). [CrossRef]
28. T. Ishibashi, N. Shimizu, S. Kodama, H. Ito, T. Nagatsuma, and T. Furuta, “Uni-traveling-carrier photodiodes,” in Ultrafast Electronics and Optoelectronics, M. Nuss and J. Bowers, eds., Vol. 13 of OSA Trends in Optics and Photonics Series (Optical Society of America, 1997), paper UC3.
29. R. Esman and K. Williams, “Wideband efficiency improvement of fiber optic systems by carrier subtraction,” IEEE Photon. Technol. Lett. 7, 218–220 (1995). [CrossRef]
30. S. Tonda-Goldstein, D. Dolfi, J. Huignard, G. Charlet, and J. Chazelas, “Stimulated Brillouin scattering for microwave signal modulation depth increase in optical links,” Electron. Lett. 36, 944–946 (2000). [CrossRef]
31. X. Xie, K. Li, Q. Zhou, A. Beling, and J. Campbell, “High-gain, low-noise-figure, and high-linearity analog photonic link based on a high-performance photodetector,” J. Lightwave Technol. 32, 3585–3590 (2014). [CrossRef]