This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES 1 High Power Integrated Photonic W-Band Emitter Keye Sun , Jesse Moody , Qinglong Li, Steven M. Bowers, Member, IEEE, and Andreas Beling Abstract— High power W-band integrated photonic emitters were successfully demonstrated by the integration of modified unitraveling-carrier photodiodes (PDs) with high directional gain Vivaldi antennas. An impedance matching network was designed to achieve conjugate matching between the PD and the antenna for better power delivery. An effective isotropic radiation power as high as 5 dBm at 110 GHz was measured for the emitter with a 5-µm-diameter PD. The impedance matching was shown to significantly affect the effective isotropic radiation power. Index Terms— Integrated photonic emitter, modified unitraveling-carrier (UTC) photodiode (PD), Vivaldi antenna. I. I NTRODUCTION T HE everlasting demand for high-speed data transmission rates is getting greater. Since the low frequency band is too crowded for high-speed wireless link applications, it becomes inevitable to extend carrier frequencies to the millimeter-wave (MMW) band above 60 GHz (V-band) or even to 100 GHz (W-band) , . Realization of such high-speed system in electronics has intrinsic drawbacks, such as high propagation loss in coaxial cables and narrow bandwidth of electronic circuits, which is incompatible with system operation at both V and W frequency bands. Developing MMW wireless links using photonic techniques seems advantageous due to the low loss of the optical fiber and inherent large bandwidth of photonic microwave components. Recently, significant interest has been paid to microwave photonic wireless emitters that tightly integrate a high-speed photodiode (PD) and an antenna , . To this end, the unitraveling carrier (UTC) PD was proposed in  owing to its excellent high-speed capability. In this structure, only fast electrons are required to drift across the depletion region as the slower holes relax within their short dielectric relaxation time . Consequently, the space charge effect in the depletion region is reduced as compared with the conventional p-i-n PD, which suffers from the slow hole drift velocity . As a result, the bandwidth performance and the power handling capability of the UTC PD are superior to that of the p-i-n PD. To further enhance device bandwidth and power handling capability of the UTC PD, a depleted region and a quasi-electric field generated by graded doping are incorporated into the absorber layer to switch the carrier transport mechanism from slow diffusion to fast drift , . Manuscript received May 17, 2017; revised August 24, 2017; accepted September 10, 2017. (Corresponding author: Keye Sun.) The authors are with the Department of Electrical and Computer Engineering, University of Virginia, Charlottesville, VA 22904 USA (e-mail: firstname.lastname@example.org; email@example.com; firstname.lastname@example.org; smb9cz@ virginia.edu; email@example.com). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TMTT.2017.2755008 This improved version of the UTC PD is called modified UTC (MUTC) PD –. The other essential component of the photonic MMW emitter is the antenna, which converts the electrical current generated by the PD into an electromagnetic wave. In order to compensate for the attenuation of MMW radiation in the atmosphere, antennas with high directional gain are desired to concentrate the beam in a certain direction. Commercial horn antennas and Si lenses have been used for this purpose , . Although a high directional gain can be achieved, they are usually bulky in size and not compatible with large-scale integrated systems. Planar structures, including slot, log-periodic toothed, Vivaldi antenna, and bow-tie antenna, as well as traveling wave antenna structures are preferred for the purpose of system miniaturization, planar integration, and wide bandwidth , , . In order to achieve high performance as well as reliable operation, the photonic emitter needs to have high mechanical strength, good thermal dissipation, and carefully designed RF characteristics. Among them, the RF characteristics are the most critical. In previous reports, PDs were directly connected to the antenna without an impedance matching network. However, in order to reach maximum output power, a conjugate impedance match between the antenna and the PD should be achieved. The effect of impedance matching has recently been discussed in  and was shown to have a significant effect on the radiation power. In this paper, we successfully demonstrate the W-band photonic emitters by the integration of the planar Vivaldi antennas with flip-chip bonded highpower high-speed MUTC PDs. The high gain of the Vivaldi antenna and the usage of a superstrate eliminate the need for the horn antenna or Si lens. AlN was used as the substrate, which has high mechanical strength and high thermal conductivity. An impedance matching network was designed to achieve a conjugate impedance matching between the MUTC PD and the antenna. Owing to these improvements, a recordhigh effective isotropic radiated power (EIRP) of 5 dBm was achieved using the planar structure. II. MUTC PD D ESIGN AND FABRICATION The epitaxial layer structure and the band diagram of the MUTC PD are shown in Fig. 1 . A 1000-nm heavily doped n-type InP layer acts as the n-contact layer. A charge compensated InP drift layer was then grown on top of the InP contact layer. The cliff layer is used to maintain a high electric field in the depleted absorber. The following quaternary layers are used to smooth the band discontinuity at the heterojunction interface, and thus avoid charge accumulation at the barriers. The absorber is composed of depleted and undepleted InGaAs layers. Layers with step-graded doping levels are used to 0018-9480 © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information. This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 2 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES Fig. 1. (a) Epilayer structure and (b) band diagram schematic of the MUTC PD. generate a quasi-electric field in the undepleted absorber. As a result, electron transport is facilitated, which further increased the bandwidth of the PD. The top quaternary and InP layers were grown to facilitate the hole collection and block the electrons from entering the anode. Finally, a heavily doped InGaAs layer is deposited as the p-contact layer. The PDs were fabricated using a double mesa process flow. The first mesa was dry-etched to the InP n-contact layer. The larger n-mesa was then formed by dry-etching down to the semi-insulating InP substrate. SiO2 surface passivation was deposited by plasma-enhanced chemical vapor deposition (PECVD). AuGe/Ni/Au and Ti/Pt/Au metal stacks were used for n- and p-metal, respectively. The top p-contact was connected to the bonding pads on the InP substrate by an air bridge. The metal layers and the air bridge were Au-plated to a thickness of 1.5 μm. A 220-nm-thick SiO2 antireflection coating was deposited by PECVD on the backside of the PD to reduce the optical reflection at 1.55-μm wavelength. The PDs with the mesa diameters of 5, 6, 14, and 20 μm were fabricated. A schematic and an optical image of the PD are shown in Fig. 2. III. A NTENNA AND I MPEDANCE M ATCHING N ETWORK D ESIGN , FABRICATION , AND I NTEGRATION The Vivaldi antenna was chosen for its wide RF bandwidth, highly directive endfire beam, and planar profile, all of which are suitable for integration into large-scale phased arrays. The Vivaldi antenna is a tapered slot line antenna, which has similar impedance characteristics to continuously tapered slot line matching networks. It presents an approximately real impedance over a very broad range of frequencies . However, the planar Vivaldi antenna has a distorted radiation pattern due to the substrate on which it is placed. A simple method for distortion suppression is to place a superstrate on top of the antenna, which equalizes the electromagnetic wave propagation velocity in air and substrate . While this limits the antenna bandwidth, it enables higher directional gain. The simulated radiation patterns using the commercial software (ANSYS HFSS) are shown in Fig. 3. The radiation pattern with superstrate has a narrow main lobe pointing forward. For comparison, the pattern without superstrate has a Fig. 2. the PD. (a) Cross-sectional schematic and (b) top view optical image of Fig. 3. Simulated antenna radiation pattern on E-plane at various frequencies (a) with and (b) without superstrate. Insets: simulated 3-D radiation patterns of the antenna with and without superstrate. wider main lobe and shows significant distortion. The antenna design was detailed in . In order to maximize the radiation power from the antenna, a resonant impedance matching technique was employed where the dc bias lines for the PD were used in conjunction with a quarter wavelength RF short circuit to synthesize an inductance, which resonates out the PD junction capacitance. This resonance was optimized between 100 and 110 GHz for maximum extraction of RF power generated by the PD. AlN was used for the antenna substrate due to its high thermal conductivity, which significantly increases the dc power dissipation of the PD . The antenna and the impedance matching network pattern were fabricated on the AlN substrate using electron-beam metal deposition, photolithography, and dry etching processes. The PDs were then flip-chip bonded on the pattern using an Au–Au thermocompression bonding process at 340 °C. Fig. 4(a)–(c) shows a cross-sectional view, optical image, and 3-D schematic of the integrated photonic emitter, respectively. This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. SUN et al.: HIGH POWER INTEGRATED PHOTONIC W-BAND EMITTER 3 Fig. 4. (a) Cross-sectional schematic and (b) optical image of the photonic integrated emitter. (c) 3-D schematic of the integrated photonic emitter. Fig. 6. (a) Circuit model of flip-chip bonded PDs for S11 fitting. (b) Measured and fit S11 parameters of 5-, 6-, 10-, and 20-μm-diameter PDs. Fig. 5. Experimental setups for (a) antenna gain measurement and (b) integrated photonic emitter radiation measurement. two distributed feedback laser diodes (Mitsubishi, FU-68PDF-5) with wavelengths around 1.55 μm were heterodyned to generate an optical beat signal with nearly 100% modulation depth. The frequency of the beat signal was controlled by thermally tuning the wavelength of one laser diode. The wavelengths of the two laser diodes were monitored by a multiwavelength meter (Hewlett Packard, 86120C). An erbium-doped fiber amplifier (EDFA) (Keopsys) was used to amplify the optical signal. The output optical signal incident on the PD was adjusted by a variable optical attenuator (Agilent, 81577A) following the EDFA. A commercial horn antenna (QuinStar, QGH-WPRR00) with 23-dB gain was used as the receiving antenna. A W-band mixer (Pacific Millimeter, WM harmonic mixer) was used to convert the W-band RF signal down to an IF frequency around 2 GHz. The IF signal was fed into a spectrum analyzer (Agilent, E4440A) from which the power was read. B. PD Impedance Characterization IV. E XPERIMENTAL R ESULTS AND D ISCUSSION A. Experimental Setup The experimental setups for measuring the antenna gain and the radiation characteristics of the integrated photonic emitter are shown in Fig. 5(a) and (b), respectively. For the antenna gain measurement, a constant CW power was fed into the antenna from a PNA network analyzer (Keysight, N5227A) equipped with frequency extender module (Keysight, N5251AL20). For the photonic emitter radiation measurement, Knowledge of the PD impedance is critical to the design of the matching network. To this end, the S11 scattering parameters of the PDs flip-chip bonded onto the coplanar waveguide (CPW) were measured using a PNA up to 100 GHz. Parameter fitting was done in the Advanced Design System (ADS) software using the circuit model shown in Fig. 6(a). C j , Rs , and L s represent the junction capacitance, series resistance, and series inductance associated with the air bridge, respectively. The effect of the CPW structure was calculated by the electromagnetic simulator in ADS. This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 4 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES TABLE I E XTRACTED C IRCUIT E LEMENTS OF THE PD S Fig. 6(b) shows the measured and fit S11 results of PDs with different mesa sizes. The fit parameters were summarized in Table I. The parameters of the PD with 14-μm diameter were calculated from the impedance trend line of the PDs of 5-, 6-, 10-, and 20-μm diameters as discussed in . C. Antenna and Impedance Matching Network Measurement The schematic of the antenna connected to the impedance matching network is shown in Fig. 7(a). The RF short is realized by a quarter wavelength open stub. The transmission line of length L 1 + L 2 after the RF short is effectively an inductor to resonate with the PD junction capacitance. The detailed dimensions are listed in Table II. The S11 scattering parameters were measured by a PNA network analyzer from 95 to 110 GHz, and the results are shown in Fig. 7(b). Since the 5- and 6-μm-diameter PDs have capacitive impedances in this frequency range, the antenna with matching network should have an inductive input impedance to achieve conjugate matching. The Vivaldi antennas in designs 1, 2, 3, 4, and 7 are parallelly connected with longer matching lines (L 1 +L 2 ), and thus, they are more inductive and closer to the optimum conjugate impedance matching condition than designs 5 and 6 with shorter matching lines. The real part of the antenna impedance was estimated to be 50 at around 100 GHz for design 2. The directional gain of these antenna designs was measured at 100 GHz in the far-field at a distance of 57 cm. The gain was calculated from the path loss, mixer loss, and receiving horn antenna gain using the Friis equation. Both the simulated and measured results are shown in Table II. The measured gain varied from 1.7 to 5.5 dBi. There is some discrepancy between the simulated and measured results. Potential reasons for the discrepancy include additional parasitics and a slight misalignment between the transmitting Vivaldi antenna and the receiving horn antenna during the measurement. Design 2 has the highest measured gain of 5.5 dBi. The directional gain of this antenna was measured over the frequency band from 95 to 110 GHz, and the result is shown in Fig. 8. The HFSS simulation results are also shown in Fig. 8, and they reasonably agree with the measurement values. The antenna gain reaches 3 dBi or more over a bandwidth of 10 GHz. Finally, design 2 was chosen to be integrated with the PDs due to its highest measured antenna gain and optimum impedance matching with the PDs. It should be noted that the S11 of the 14-μm-diameter PD is inductive instead of capacitive. Fig. 7. (a) Schematic and dimension of the antenna plus matching network. (b) S11 scattering parameters for different antennas plus impedance matching networks and 5-, 6-, and 14-μm-diameter PDs. To conjugately match its S11, a capacitive matching circuit should be connected to the PD. The metal lines of a capacitive matching circuit (L 1 + L 2 ) would be very long, and the loss associated with it would be large at such high frequency. So the conjugate matching of the 14-μm-diameter PD was not pursued. But the same design was also integrated with the 14-μm-diameter PD to verify the importance of impedance mismatching on the EIRP. D. EIRP of the Integrated Photonic Emitter The radiation power at 100 GHz from the emitters with 5- and 14-μm-diameter PDs biased at −3 V and 10-mA photocurrent was measured at different distances from the integrated photonic emitter. As shown in Fig. 9, the received radiation This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. SUN et al.: HIGH POWER INTEGRATED PHOTONIC W-BAND EMITTER 5 TABLE II A NTENNA D IMENSIONS AND D IRECTIONAL G AIN OF D IFFERENT A NTENNAS AT 100 GHz Fig. 8. Measured and simulated directional gain of antenna design 2 between 95 and 110 GHz. Fig. 9. Received radiation power at 100 GHz from the emitters with 5- and 14-μm-diameter PD at 10-mA photocurrent and −3-V bias voltage at different distance. power decreases with increasing distance due to a larger path loss and enters far-field at 57 cm. The received power for the emitter with the 5-μm-diameter PD is 8 dB higher than Fig. 10. EIRP at 100 GHz from the emitters with 5-, 6-, and 14-μm PD biased at −2 V at different photocurrents. the one with the 14-μm-diameter PD at the same photocurrent, which can be explained by an improved impedance matching. The EIRP at different photocurrents was measured in the far-field. The results are shown in Fig. 10. The EIRP increases with increasing photocurrent as expected. The EIRP from the emitters with the 5- and 6-μm-diameter PDs is 7.5 and 6 dB higher than the power from the emitter with the 14-μm-diameter PD, respectively. The higher power is attributed to the better impedance matching between the 5- and 6-μm-diameter PDs and the antenna as shown on the Smith chart in Fig. 7(b). The input impedance of the antenna with the matching network appears inductive and is located at the conjugate loci of the impedance of the 5- and 6-μm-diameter PDs on the Smith chart. On the other hand, the impedance of the 14-μm-diameter PD appears capacitive, which is far away from the point of optimum conjugate matching. As a result, the EIRP of the emitter with 14-μm-diameter PD is around 6 dB lower than the emitter with 6-μm-diameter PD. The emitter with 5-μm-diameter PD has an EIRP about 1.5 dB higher than the emitter with 6-μm-diameter PD does. Based on the fact that the impedance This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 6 Fig. 11. EIRP of the integrated photonic emitters with 5-, 6-, and 14-μm-diameter PDs biased at 10-mA photocurrent from 95 to 110 GHz. IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES Fig. 12. E-plane radiation patterns of the integrated photonic emitters with 6- and 14-μm-diameter PDs, the antenna without PD, and the HFSS simulation results. Inset: experimental setup for radiation pattern measurement. matching condition for the 5- and 6-μm-diameter PDs is similar, the possible reason for this difference is that the 5-μm-diameter PD has a slightly higher −3-dB bandwidth than the 6-μm-diameter PD . The EIRP does not increase linearly at high photocurrent, which indicates the power compression due to the space charge effect in the PD at high photocurrent. E. Radiation Bandwidth of the Integrated Photonic Emitter The radiation bandwidths of the integrated photonic emitters were measured by tuning the temperature of one laser diode to sweep the beat signal frequency from 95 to 110 GHz. The results are shown in Fig. 11. The EIRP is around 3, −3, and −7 dBm for the photonic emitters with 5-, 6-, and 14-μm-diameter PDs, respectively. The difference of the EIRP from the emitters with different PDs is due to the impedance matching condition as shown in Fig. 7(b). Moreover, the EIRP varies by several dB for each photonic emitter across the frequency band. This variation might result from the varying input impedance of the antenna plus matching network. As can be seen in Fig. 7(b), the S11 of the antenna plus matching network circles near the optimum conjugate matching loci. Consequently, the impedance matching condition varies across the frequency band, which, in turn, causes the variation of the EIRP. The −6-dB power bandwidths for all three emitters are around 10 GHz. F. Radiation Pattern of the Integrated Photonic Emitter The E-plane radiation patterns of the integrated photonic emitters were measured by changing the receiving horn antenna position in the far-field region. The results are shown in Fig. 12. Also shown in Fig. 12 is the E-plane radiation pattern of the Vivaldi antenna only for comparison. All measurements revealed a half main lobe of about 15°. As expected, the antenna radiation pattern does not change when the PD is integrated. The HFSS simulation of the antenna radiation pattern on E-plane is included in Fig. 12 for comparison. The simulation and measurement results agree reasonably well. Fig. 13. (a) Equivalent circuit model for calculating the EIRP of the integrated photonic emitter. (b) Alternative equivalent circuit model for calculating the EIRP. V. C OMPARISON W ITH S IMULATION R ESULTS An equivalent circuit model was used to calculate the EIRP of the integrated photonic emitter in the ADS software. The circuit model is shown in Fig. 13(a). The PD model includes an ideal current source, junction capacitance C j , series resistance Rs , and series inductance L s associated with the air bridge. The carrier transit time effect in the PD was taken into account by the fact that the photocurrent is a function of the beat signal frequency ω, as shown in  I (ω) = I0 · 1 − exp( j ωtr ) j ωtr (1) where I0 is the average photocurrent, ω is the beat signal frequency, and tr is the electron transit time across the drift layer, which is calculated to be 3.5 ps . The antenna plus the impedance matching network were represented by the measured S11 parameters. Finally, the EIRP was calculated by the total power delivered to the load multiplied by the antenna gain (5.8 dB) as a function of frequency. This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. SUN et al.: HIGH POWER INTEGRATED PHOTONIC W-BAND EMITTER Fig. 14. Measurement and ADS calculation results of the EIRP of integrated photonic emitters with (a) 5-μm, (b) 6-μm, and (c) 14-μm-diameter PDs at 10-mA photocurrent. The ADS calculation and measurement results using the circuit model in Fig. 13(a) are shown in Fig. 14 for the integrated photonic emitters with 5-, 6-, and 14-μm-diameter PDs. For the emitters with 5- and 6-μm-diameter PDs, the ADS calculations were about 10 dB higher than the measurement results. The difference is 4 dB for the emitter with 14-μm-diameter PD. There are several possible reasons for this discrepancy. First, since the layout of the antenna substrate is different from the CPW substrate, different parasitics can be expected once the PDs flip-chip bonded on the two substrates. In other words, the parasitics for the PD bonded on the antenna substrate could be significantly larger than the value extracted from the S11 parameters of the PD bonded on the CPW substrate. In order to show the effect of parasitics on the calculated EIRP, 7 an alternative circuit model shown in Fig. 13(b) was used. The capacitance Cstray represents the contribution from parasitics. Other circuit elements have the same meaning and values as the circuit model in Fig. 13(a). A stray capacitance of 30 fF was assigned to Cstray in the circuit model in Fig. 13(b). As can be seen from the calculation results in Fig. 14, even a small parasitic capacitance of only 30 fF can have significant impact on the EIRP, which can be several decibels lower than without parasitics. Second, the S11 of the PD was measured under dark condition. It has been shown that the input impedance of the PD shows a significant change under strong light illumination due to the space charge effect . Like microwave amplifiers, the S-parameters based on small input signal conditions may not be adequate to characterize the PD. Other more sophisticated techniques, such as load-pull measurement, may help to characterize the optimum load for the device. Since the impedance matching network is designed based on the fitting parameters of our simple circuit model with small PNA stimulus signal under dark condition, the actual matching condition might be worse than the one shown in Fig. 7(b). Based on the fact that the impedance matching at such high frequency range is highly sensitive and depends on the knowledge of the accurate impedance of the PD under strong light illumination deembedded from the CPW structure, a nonlinear circuit model and a 3-D full-wave modeling will be developed in the future. In order to show the effect of illuminationinduced impedance change on the EIRP, the circuit model in Fig. 13(a) with a different junction capacitance C j was used for calculation. Under strong illumination, the junction capacitance tends to increase as a result of electric field screening caused by photogenerated carriers. To illustrate this effect, we used a light-induced change of junction capacitance of 3 fF and recalculate the EIRP. As can be seen from the results in Fig. 14, a difference around 1 dB can be induced by this 3-fF increase in junction capacitance. A better matching between the calculation and measurement results can be found by using the circuit model extracted from S11 parameters under illumination. Finally, the circuit model used to represent the MUTC-PD in this paper is oversimplified. The model includes only junction capacitance, series resistance, and air-bridge inductance. More sophisticated circuit models have been used to represent the MUTC-PDs. In , a serial connection of resistance and capacitance was used to model the carrier transit time effect, and two separate capacitances were used to represent the collector and electrical field suffer layer. In , two parallelconnected resistance and capacitance are used to represent the absorber and drift layers. In , parallel connections of resistance and capacitance were used to model the effect of quaternary grading layers in order to precisely match the magnitude and the phase of the S11 parameters. According to , a circuit model composed of a simple junction capacitance and a series resistance may lead to substantial disagreement between calculated and measured radiation power. In order to show the effect of variations in the circuit model on calculated EIRP, the circuit model shown in Fig. 13(b) was used. All the circuit element values were extracted from the This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 8 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES VII. C ONCLUSION A W-band photonic emitter was successfully demonstrated by the integration of high-power flip-chip bonded MUTC PD and Vivaldi antenna. An EIRP as high as 5 dBm has been achieved at 110 GHz. The impedance matching has a large effect on the EIRP. The emitter has a wide −6-dB radiation bandwidth of around 10 GHz. The main lobe of the radiation pattern is 30°. The compact integrated photonic emitter is fully compatible with planar integration. The EIRP is among the highest that have been reported around 100 GHz in the literature. Fig. 15. Radiation power of integrated photonic emitters reported in the literature. momentum fitting of S11 of the flip-chip bonded PDs as discussed in Section IV-B. The EIRP was then calculated using this circuit model with the best fitting parameters and is shown in Fig. 14. As can be seen, the calculated power is already around 1 dB less just by using this slightly different circuit model. As can be expected, the discrepancy between the calculation and measurement results could be further reduced if more sophisticated circuit models were used as the literature suggested. In summary, the EIRP is sensitive to all the effects, including light-induced impedance change, parasitics, and the circuit model representing the PD. We believe that a combination of these effects can explain the discrepancy between the calculation and measurement results. Other minor effects include the orientation and the polarization misalignment of the transmitting and receiving antenna in the far-field region. VI. C OMPARISON W ITH L ITERATURE R ESULTS Fig. 15 gives an overview of the radiation power in this paper and results that have been reported in the literature. At around 100 GHz, the EIRP from our photonic integrated emitter is among the highest. In , , and , the photonic emitters were connected to a horn antenna with high directional gain. In addition, either teflon or Si lens were used in , , , and  to achieve high directional gain by focusing the emitted beam. Up to 12-dB additional gain can be acquired by a Si lens as reported in . In contrast, the emitters in this paper use a planar Vivaldi antenna. As a result, the footprint is greatly reduced, and the device is fully compatible with planar integration and suitable for phased array applications. Other types of planar antenna structures were also used in the literature to avoid the Si lens and horn antenna, such as Yagi antenna , , log-periodic antenna , bow-tie antenna , , , slot antenna , , and traveling-wave ring radiator . Among all these reported photonic emitters, the EIRP in this paper is the highest owing to the optimized impedance matching network and excellent high-power-handling capability of the MUTC PDs. ACKNOWLEDGMENT The authors would like to thank Prof. J. C. Campbell for fruitful discussions and P. Yao for his help with the flip-chip bonding process. R EFERENCES  J. Federici and L. Moeller, “Review of terahertz and subterahertz wireless communications,” J. Appl. Phys., vol. 107, no. 11, 2010, Art. no. 111101.  C. Park and T. S. Rappaport, “Short-range wireless communications for next-generation networks: UWB, 60 GHz millimeter-wave WPAN, and ZigBee,” IEEE Wireless Commun., vol. 14, no. 4, pp. 70–78, Aug. 2007.  A. Stohr et al., “Millimeter-wave photonic components for broadband wireless systems,” IEEE Trans. Microw. Theory Techn., vol. 58, no. 11, pp. 3071–3082, Nov. 2010.  S. M. Bowers, B. Abiri, F. Aflatouni, and A. Hajimiri, “A compact optically driven travelling-wave radiating source,” in Proc. Opt. Fiber Commun. Conf. (OFC), San Francisco, CA, USA, 2014, paper Tu2A.3.  T. Ishibashi, N. Shimizu, S. Kodama, H. Ito, T. Nagatsuma, and T. Furuta, in Ultrafast Electron. Optoelectron. Tech. Dig., Washington, DC, USA, 1997, p. 166.  T. Ishibashi, S. Kodama, N. Shimizu, and T. Furuta, “High-speed response of uni-traveling-carrier photodiodes,” Jpn. J. Appl. Phys., vol. 36, no. 10, pp. 6263–6268, 1997.  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. Topics Quantum Electron., vol. 10, no. 4, pp. 702–708, Jul. 2004.  P. D. Yoder and E. J. Flynn, “Linear theory of the quasi-unipolar photodiode,” J. Lightw. Technol., vol. 24, no. 4, pp. 1937–1945, Apr. 2006.  Z. Li, H. Pan, H. Chen, A. Beling, and J. C. Campbell, “High-saturationcurrent modified uni-traveling-carrier photodiode with cliff layer,” IEEE J. Quantum Electron., vol. 46, no. 5, pp. 626–632, May 2010.  M. Anagnosti et al., “Optimized high speed UTC photodiode for 100 Gbit/s applications,” IEEE J. Sel. Topics Quantum Electron., vol. 20, no. 6, pp. 29–35, Nov. 2014.  X. Xie, Q. Zhou, K. Li, A. Beling, and J. C. Campbell, “1.8 watt RF power and 60% power conversion efficiency based on photodiode flip-chip-bonded on diamond,” in Post-Deadline Paper Dig., 2014, paper JTh5B.9.  J. W. Shi et al., “Extremely high saturation current-bandwidth product performance of a near-ballistic uni-traveling-carrier photodiode with a flip-chip bonding structure,” IEEE J. Quantum Electron., vol. 46, no. 1, pp. 80–86, Jan. 2010.  H. Ito, S. Kodama, Y. Muramoto, T. Furuta, T. Nagatsuma, and T. Ishibashi, “High-speed and high-output InP-InGaAs unitravelingcarrier photodiodes,” IEEE J. Sel. Topics Quantum Electron., vol. 10, no. 4, pp. 709–727, Jul. 2004.  Q. Li et al., “High-power flip-chip bonded photodiode with 110 GHz bandwidth,” J. Lightw. Technol., vol. 34, no. 9, pp. 2139–2144, May 1, 2016.  N.-W. Chen, H.-J. Tsai, F.-M. Kuo, and J.-W. Shi, “High-speed W-band integrated photonic transmitter for radio-over-fiber applications,” IEEE Trans. Microw. Theory Techn., vol. 59, no. 4, pp. 978–986, Apr. 2011.  K. Sun, J. Moody, Q. Li, S. M. Bowers, and A. Beling, “High-power integrated photodiode-antenna emitter for 100 GHz applications,” Adv. Photon. (IPR, NOMA, Sens., Netw., SPPCom, SOF), Vancouver, BC, Canada, 2016, paper IW2A.6. This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. SUN et al.: HIGH POWER INTEGRATED PHOTONIC W-BAND EMITTER  M. Natrella et al., “Modelling and measurement of the absolute level of power radiated by antenna integrated THz UTC photodiodes,” Opt. Exp., vol. 24, no. 11, pp. 11793–11807, 2016.  T.-L. Hwang, D. B. Rutledge, and S. E. Schwarz, “Planar sandwich antennas for submillimeter applications,” Appl. Phys. Lett., vol. 34, no. 1, pp. 9–11, 1979.  H. Oraizi and S. Jam, “Optimum design of tapered slot antenna profile,” IEEE Trans. Antennas Propag., vol. 51, no. 8, pp. 1987–1995, Aug. 2003.  J. Moody, K. Sun, Q. Li, A. Beling, and S. M. Bowers, “A Vivaldi antenna based W-band MUTC photodiode driven radiator,” in Proc. IEEE Int. Topical Meet. Microw. Photon. (MWP), Long Beach, CA, USA, 2016, pp. 217–220.  X. Xie et al., “Improved power conversion efficiency in highperformance photodiodes by flip-chip bonding on diamond,” Optica, vol. 1, no. 6, pp. 429–435, 2014.  S. M. Sze, and K. K. Ng, Physics of Semiconductor Devices, 3rd ed. Hoboken, NJ. USA: Wiley, 2006.  J. Li, B. Xiong, C. Sun, D. Miao, and Y. Luo, “Analysis of frequency response of high power MUTC photodiodes based on photocurrentdependent equivalent circuit model,” Opt. Exp., vol. 23, no. 17, pp. 21615–21623, 2015.  M. Natrella et al., “Accurate equivalent circuit model for millimetrewave UTC photodiodes,” Opt. Exp., vol. 24, no. 5, pp. 4698–4713, 2016.  Y.-S. Wu, N.-W. Chen, and J.-W. Shi, “A W -band photonic transmitter/mixer based on high-power near-ballistic uni-traveling-carrier photodiode (NBUTC-PD),” IEEE Photon. Technol. Lett., vol. 20, no. 21, pp. 1799–1801, Nov. 1, 2008.  J.-W. Shi et al., “A W -band photonic transmitter-mixer based on highpower near-ballistic uni-traveling-carrier photodiodes for BPSK and QPSK data transmission under bias modulation,” IEEE Photon. Technol. Lett., vol. 20, no. 15, pp. 1039–1041, Aug. 1, 2009.  A. Hirata, H. Ishii, and T. Nagatsuma, “Design and characterization of a 120-GHz millimeter-wave antenna for integrated photonic transmitters,” IEEE Trans. Microw. Theory Techn., vol. 49, no. 11, pp. 2157–2162, Nov. 2001.  P. G. Huggard et al., “Generation of millimetre and sub-millimetre waves by photomixing in 1.55 μm wavelength photodiode,” Electron. Lett., vol. 38, no. 7, pp. 327–328, 2002.  F. Nakajima, T. Furuta, and H. Ito, “High-power continuous-terahertzwave generation using resonant-antenna-integrated uni-travelling-carrier photodiode,” Electron. Lett., vol. 40, no. 20, pp. 1297–1298, Sep. 2004.  K. Li et al., “High-power photodiode integrated with coplanar patch antenna for 60-GHz applications,” IEEE Photon. Technol. Lett., vol. 27, no. 6, pp. 650–653, Mar. 15, 2015.  J.-W. Shi, Y.-S. Wu, and Y.-S. Lin, “Near-ballistic uni-travelingcarrier photodiode-based V-band optoelectronic mixers with internal up-conversion-gain, wide modulation bandwidth, and very high operation current performance,” IEEE Photon. Technol. Lett., vol. 20, no. 11, pp. 939–941, Jan. 1, 2008.  A. Hirata et al., “Output power measurement of photonic millimetrewave and sub-millimetre-wave emitter at 100–800 GHz,” Electron. Lett., vol. 38, no. 15, pp. 798–800, 2002.  N. Shimizu and T. Nagatsuma, “Photodiode-integrated microstrip antenna array for subterahertz radiation,” IEEE Photon. Technol. Lett., vol. 18, no. 6, pp. 743–745, Mar. 15, 2006.  H. Ito, T. Furuta, F. Nakajima, K. Yoshino, and T. Ishibashi, “Photonic generation of continuous THz wave using uni-travelingcarrier photodiode,” J. Lightw. Technol., vol. 23, no. 12, pp. 4016–4021, Dec. 2005.  H. Ito et al., “Photonic millimetre-wave emission at 300 GHz using an antenna-integrated uni-travelling-carrier photodiode,” Electron. Lett., vol. 38, no. 17, pp. 989–990, 2002.  H. Ito, F. Nakajima, T. Furuta, K. Yoshino, Y. Hirota, and T. Ishibashi, “Photonic terahertz-wave generation using antenna-integrated unitravelling-carrier photodiode,” Electron. Lett., vol. 39, no. 25, pp. 1828–1829, 2003.  E. Rouvalis, C. C. Renaud, D. G. Moodie, M. J. Robertson, and A. J. Seeds, “Traveling-wave uni-traveling carrier photodiodes for continuous wave THz generation,” Opt. Exp., vol. 18, no. 11, pp. 11105–11110, 2010.  A. Stohr, A. Malcoci, A. Sauerwald, I. C. Mayorga, R. Gusten, and D. S. Jager, “Ultra-wide-band traveling-wave photodetectors for photonic local oscillators,” J. Lightw. Technol., vol. 21, no. 12, pp. 3062–3070, Dec. 2003. 9  A. Hirata, T. Minotani, and T. Nagatsuma, “A 120-GHz microstrip antenna monolithically integrated with a photodiode on Si,” Jpn. J. Appl. Phys., vol. 41, pp. 1390–1394, Mar. 2002.  I. H. Henning et al., “Broadband antenna-integrated, edge-coupled photomixers for tuneable terahertz sources,” IEEE J. Quantum Electron., vol. 46, no. 10, pp. 1498–1505, Oct. 2010.  E. Rouvalis, C. C. Renaud, D. G. Moodie, M. J. Robertson, and A. J. Seeds, “Continuous wave terahertz generation from ultra-fast InP-based photodiodes,” IEEE Trans. Microw. Theory Techn., vol. 60, no. 3, pp. 509–517, Mar. 2012.  A. Hirata, M. Harada, and T. Nagatsuma, “120-GHz wireless link using photonic techniques for generation, modulation, and emission of millimeter-wave signals,” J. Lightw. Technol., vol. 21, no. 10, pp. 2145–2153, Oct. 2003. Keye Sun received the B.S. degree in physics from Shanghai Jiao Tong University, Shanghai, China, in 2007, and the Ph.D. degree in electrical and computer engineering from the University of Virginia, Charlottesville, VA, USA, in 2015. He is currently a Research Scientist with the Electrical and Computer Engineering Department, University of Virginia. His current research interests include high-power and high-speed photodetectors, wireless photonic transmitters, high-frequency antennas, microwave photonics and circuits, Si photonics, solar cells, Ge-based devices, and high-power laser-based material processing. Jesse Moody received the B.S. degree from the University of Southern Florida, Tampa, FL, USA, in 2014. He joined the Electrical and Computer Engineering Department, University of Virginia, Charlottesville, VA, USA, in 2015, as a Graduate Researcher. His current research interests include ultralow power wake-up receivers for IoT applications and photonically driven radiators. Qinglong Li was born in Tianjin, China, in 1985. He received the bachelor’s degree from Nankai University, Tianjin, in 2008, and the M.Sc. degree from the Rochester Institute of Technology, Rochester, NY, USA, in 2013. He is currently pursuing the Ph.D. degree in electrical and computer engineering at the University of Virginia, Charlottesville, VA, USA. He worked within the semiconductor manufacturing industry in China for one year and a half. His current research interests include high-power and high-speed photodetectors for optical communication applications. This article has been accepted for inclusion in a future issue of this journal. Content is final as presented, with the exception of pagination. 10 Steven M. Bowers (GS’08–M’12) received the B.S. degree in electrical engineering from the University of California at San Diego, La Jolla, CA, USA, and the M.S. and Ph.D. degrees in millimeter-wave circuits and systems from the California Institute of Technology, Pasadena, CA, USA. He joined the faculty of the Charles L. Brown Department of Electrical and Computer Engineering, University of Virginia, Charlottesville, VA, USA, in 2014, where he is currently an Assistant Professor. Dr. Bowers is a member of the HKN and TBP. He was the recipient of the Caltech Institute Fellowship in 2007 and the Analog Devices Outstanding Student Designer Award in 2009. He was a recipient of the IEEE RFIC Symposium Best Student Paper Award in 2012, the IEEE IMS Best Student Paper Award in 2013, and the 2015 IEEE MTT Microwave Prize. IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES Andreas Beling received the Dipl.-Phys. degree (M.S.) in physics from the University of Bonn, Bonn, Germany, in 2000, and the Dr.-Ing. degree (Ph.D.) in electrical engineering from the Technical University of Berlin, Berlin, Germany, in 2006. He was a Staff Scientist with the Photonics Division, Fraunhofer-Institut für Nachrichtentechnik, Heinrich-Hertz-Institut, HHI, Berlin, from 2001 to 2006, and a Research Associate with the Department of Electrical and Computer Engineering (ECE), University of Virginia (UVA), Charlottesville, VA, USA, from 2006 to 2008. He possesses two years of industry experience as a Project Manager involved in optoelectronic receivers for high-speed fiber optic communication systems. He became an Assistant Professor with the Department ECE, UVA, in 2013. He has authored or co-authored over 130 technical papers and 2 book chapters. He holds four patents. Dr. Beling is a Senior Member of the OSA. He has served on the Technical Program Committees of the Optical Fiber Communication (OFC) Conference from 2010 to 2012, the International Conference on Indium Phosphide and Related Materials in 2014, the Microwave Photonics Conference in 2015 and 2016, and the Integrated Photonics Research conference in 2016. He was a Technical Program Subcommittee Chair for Subcommittee 8 (Optoelectronic Devices) at OFC in 2013. Since 2014, he has been an Associate Editor of the IEEE/OSA J OURNAL OF L IGHTWAVE T ECHNOLOGY.