IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 58, NO. 3, MARCH

IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 58, NO. 3, MARCH 2010 665 Quadratic Electrooptic Effect for Frequency Down-Conversion Yifei Li, Member, IEEE, Renyuan Wang, Member, IEEE, Jonathan S. Klamkin, Member, IEEE, Leif A. Johansson, Member, IEEE, Peter R. Herczfeld, Life Fellow, IEEE, and John E. Bowers, Fellow, IEEE Abstract A novel optical frequency down-conversion approach employing the quadratic electrooptic effect is proposed, analyzed, and verified. This down-conversion method can seamlessly integrate with an optical phase-locked-loop linear phase demodulator. It has the potential to achieve a large dynamic range and low conversion loss. Comprehensive theoretical studies are performed to investigate the noise, efficiency, and linearity of this down-conversion approach. Experiments are then conducted. It is found that the unwanted fourth-order electrooptic effect of the InP modulator device limits the dynamic range. Index Terms Frequency conversion, optical phase-locked loop (OPLL), quadratic electrooptic effect. I. INTRODUCTION I N RADAR frontend applications, RF/photonic links can provide the required connectivity between antennas and the signal processing unit. They are attractive due to their high bandwidth, low attenuation, and immunity to electromagnetic interference (EMI). However, due to the nonlinear distortion incurred in the optical modulation process, existing microwave fiber-optic links that employ optical intensity modulation have inadequate spurious-free dynamic range (SFDR), which can be barely pushed over 70 db for 500-MHz instantaneous bandwidth [1] [3]. One potential solution for this problem is a true linear phase modulated (PM) fiber-optic link employing an optical phase-locked loop (OPLL) [4], [5] linear phase demodulator that linearly demodulates the optical phase by tight phase tracking. This new optical link aims to achieve a dynamic range over 140 db Hz, which is equivalent to 83 db over 500-MHz Manuscript received April 15, 2009; revised September 06, 2009. First published February 18, 2010; current version published March 12, 2010. This work was supported in part by the Defense Advanced Research Projects Agency (DARPA)/Microsystems Technology Office (MTO) Phorfront under Grant S3920000007850 and by DARPA under the Young Faculty Award Program. Y. Li and R. Wang are with the Department of Electrical Engineering, University of Massachussetus at Dartmouth, Dartmouth, MA 02747 USA (e-mail: yli2@umassd.edu; ryan.cywang@gmail.com). J. S. Klamkin, L. A. Johansson, and J. E. Bowers are with the Department of Electrical and Computer Engineering and the Department of Materials, University of California at Santa Barbara, Santa Barbara, CA 93116 USA (e-mail: jklamkin@gmail.com; leif@ece.ucsb.edu; bowers@ece.ucsb.edu). P. R. Herczfeld is with the Department of Electrical and Computer Engineering, Drexel University, Philadelphia, PA 19104 USA (e-mail: herczfeld@ece.drexel.edu). 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.2010.2040355 bandwidth. Different from the conventional OPLLs [6] that are for stabilizing the optical phase differences in laser heterodyne applications, the OPLL phase demodulator has a local optical phase modulator instead of a tunable local laser oscillator. It requires a very high open loop gain ( 20 db) over a very large bandwidth ( 1 GHz, to cover the entire RF information band) in order to perform linear phase demodulation. The gain and bandwidth requirements necessitate a short loop delay that is only achievable when implemented as a photonic integrated circuit (PIC). At present, the OPLL phase demodulator PIC is under intense development. On the other hand, a radar frontend needs to convert the antenna signal from RF to IF or baseband, where it can be digitized and processed by a radar signal-processing backend. The conventional approach for frequency down-conversion employs electronic mixers. However, the electronic mixers contain a large conversion loss, and more importantly, high spurious distortion, resulting in a restricted dynamic range [7], [8]. Thus, they cannot preserve the dynamic range of the new PM optical link. Several optical techniques for down-converting the RF signals have been investigated. These include frequency down-conversion using a Mach Zehnder (MZ) modulator [9], optical heterodyning [10], and optical sampling phase-locked loop [11]. However, to date, there are still many open questions on how to integrate them seamlessly with the OPLL without significant performance downgrade. We have proposed a different optical down-conversion technique based on quadratic optical phase modulators [12]. In this approach, we exploit the second-order nonlinearity of the quadratic optical phase modulator. As shown in Fig. 1, the frequency down-conversion is achieved by applying the RF and local oscillator (LO) signals simultaneously to a quadratic optical phase modulator. Due to the quadratic nonlinearity of the phase modulator, the mixing product between the RF and LO is encoded in the optical phase. The optical phase is then recovered through an OPLL linear phase demodulator. The integration of the OPLL and the quadratic optical phase modulator down-converter is seamless. In addition, the quadratic optical phase modulators can be conveniently implemented with an indium phosphide (InP) multiple quantum well (MQW) modulator device. In this paper, we extend our initial discussions in [12]. We provide a detailed analysis of the frequency down-conversion with quadratic optical phase modulators. We derive the phase down-conversion efficiency, nonlinear distortion, and dynamicrange limitations by assuming shot noise limit. We will also present updated and more comprehensive experimental results. 0018-9480/$26.00 2010 IEEE

666 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 58, NO. 3, MARCH 2010 Fig. 1. Frequency down-converter diagram. Quadratic phase modulator (QPM). Linear phase modulator (LPM), Optical phase-locked loop (OPLL). Uni-traveling-carrier photodiode (UTC PD). II. THEORETICAL ANALYSIS In this section, we discuss the mechanism of the frequency mixing process, its nonlinear distortion, and dynamic-range limitation. A. Frequency Mixing Using Quadratic Phase Modulators In the proposed frequency down-conversion scheme, we achieve frequency mixing by exploiting the second-order nonlinearity of quadratic optical phase modulators. In Fig. 1, the output optical phases of two ideal quadratic optical phase modulators (in Fig. 1) are given by (1a) (1b) where is the quadratic phase modulation sensitivity per unit length, is the modulator length, is the RF input voltage, and is the LO voltage. Thus, the difference between the two phases contains the mixing product between and, which is where and are the amplitudes of the LO signal and RF signal, and and are the angular frequency and the phase of the RF signal, respectively. and are the angular frequencies of the down-converted signal and LO signal, respectively. This process is identical to that of frequency mixing using single balanced electronic mixers. Similarly, a single optical quadratic optical phase modulator is equivalent to an unbalanced electronic mixer. The first term in (2) is the down-converted signal, and the second term is the up-converted signal. The up-converted signal can be filtered out. If necessary, it can be further diminished by an image-rejection scheme [12] shown in Fig. 2. We define the frequency down-conversion gain of the quadratic optical phase modulator as the ratio between the magnitude of the down-converted optical phase [in (2)] and the magnitude of the RF input. It is given by (2) (3) Fig. 2. Quadratic optical phase modulator with image rejection. Quadratic phase modulator (QPM). It should be noted that the conversion gain defined here has a unit of rad/v. The conversion gain can be improved by enhancing the quadratic optical modulator sensitivity and the electric LO voltage. When the down-converted optical phase is demodulated by an OPLL phase demodulator, the magnitude of the output IF voltage is given by where is the phase voltage of the local phase modulator inside the OPLL (Fig. 1). In deriving (4), it is assumed that the OPLL has sufficient open loop gain, which is a prerequisite for linear optical phase demodulation. In (4), can be treated as the phase demodulator responsivity. In order to get a large IF output, the ratio between and the demodulation responsivity needs to be large. B. Nonlinear Distortion, Noise, and Dynamic Range 1) Nonlinear Distortion: If the quadratic optical phase modulator is ideal, it should add no nonlinear distortion. However, a practical quadratic optical phase modulator, such as a semiconductor MQW phase modulator, contains higher order ( second order) nonlinearities that add spurious distortion to the down-converted optical phase. Including these nonlinearities, the outputs of the quadratic phase modulators in Fig. 1 should be expressed as (4) (5a) (5b) where and are the voltages applied to the modulators, respectively: and ; is the th-order phase modulation sensitivity per unit length. In (5), the odd-order terms cause out-of-band distortion that can be subsequently filtered out. However, the even-order terms (i.e., the fourth, sixth, ) cause in-band distortion. To quantify the linearity of the quadratic phase modulator as a frequency mixing

LI et al.: QUADRATIC ELECTROOPTIC EFFECT FOR FREQUENCY DOWN-CONVERSION 667 element, we define its down-conversion phase third-order intercept point (IP3), where the desired down-converted IF signal and the in-band spurious distortion have identical strength. To determine the down-conversion phase IP3 point, we assume the RF input contains two tones Upon substituting (6) into (5) and ignoring the higher order ( fourth order) nonlinear terms, we obtain the down-converted signal and the spurious distortion inside the optical phase, which are given by (6) (7a) (7b) At the intercept point, where the spurious distortion and downconverted signal have equal magnitude, the magnitude of the RF input must satisfy (8) Thus, in terms of the root mean square (rms) average of the optical phase, the down-conversion phase IP3 is given by The down-conversion phase IP3 has a unit of rms radian. It quantifies the down-conversion linearity of the quadratic phase modulator as a frequency-mixing element. 2) Noise Analysis: In general, the minimum discernable optical phase is determined by several noise sources [5]. Here we focus on the laser phase noise, relative intensity noise (RIN), and photodetector shot noise, as other noise sources are generally much smaller and negligible. In this scenario, the minimum discernable optical phase per hertz is given by (10) where is the optical phase noise power spectral density (PSD) of the laser source and is the delay difference between the two optical paths in Fig. 1. RIN is the laser RIN. is the RIN cancellation of the balanced photodetector. is the charge of an electron, and is the photocurrent of each photodetector. As shown in (10), the contribution from the laser phase noise would be eliminated if the optical path delays are matched, i.e.,. It is instructive to compare the minimum discernable optical phase that is achievable with two distinctive laser sources: a state-of-the-art solid-state laser and a low-cost narrow linewidth laser diode (Redfern PLANEX RIO009x). As shown in Fig. 3, the solid-state laser source offers much better noise performance. It can achieve shot noise limited performance with reasonable delay mismatch. In comparison, the system with the laser diode as its source is RIN (9) Fig. 3. Minimum discernable optical phase simulation: (a) with a high-quality solid-state laser (RIN: 0170 db/hz, optical phase noise: 0110 dbrad/hz @ 1-MHz offset with Lorenzian shape and (b) with a narrow linewidth laser diode (RIN: 0145 db/hz, optical phase noise: 0100 dbrad/hz @ 1-MHz offset with Lorenzian shape). The other parameters are I =100mA, 0 =20dB, and f =1GHz. noise. Thus, we conclude that for best noise performance, the solid-state laser should be used. 3) SFDR: If we assume the shot noise limit that is achievable with solid-state laser source, the SFDR limitation due to the quadratic optical phase modulator is db Hz (11) Using (10) and (11), we have analyzed the required downconversion phase IP3 for achieving a certain SFDR. As shown in Fig. 4, larger phase IP3 values are demanded for a higher SFDR. However, with a larger photocurrent, this demand is relieved. Specifically, if an SFDR of 140 db Hz is desired, the down-conversion phase IP3 should be at least with 100-mA photocurrent, which is typical for the OPLL currently under development. The IP3 requirement of is achievable, but involves significant challenges, and it is the current focus of our research effort. Although the discussions in this section focus on a pair of balanced quadratic phase modulators, as shown in Fig. 1, they can be applied to the case of a single quadratic phase modulator [12]. The only difference is that the balanced modulator pair doubles the conversion gain and the phase IP3.

668 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 58, NO. 3, MARCH 2010 Fig. 5. PIC for phase modulator linearity measurement. Quadratic phase modulator (QPM). Multimode interference coupler (MMI). where and are the phase modulations on the two phase modulators, respectively, Fig. 4. Required down-conversion phase IP3 versus photocurrent. Spuriousfree dynamic range (SFDR). In terms of bandwidth limitation of this approach, RF and LO frequency range of this down-conversion scheme is ultimately limited by the bandwidth of the quadratic optical phase modulator. The IF frequency range is limited by the bandwidth of the OPLL phase demodulator ( 1 GHz). Next, we report our experiments on frequency down-conversion using existing InP MQW optical phase modulators. III. DOWN-CONVERSION BY InP PHASE MODULATOR A. Measurement Approach We use a InP MQW phase modulator to implement the quadratic optical phase modulator. The InP MQW modulator consists of 28 periods of well (5-nm wide) and barrier (7-nm thick). The photoluminance (PL) peak of the quantum well is measured to be 1425 nm. The output phase of the quadratic optical phase modulator should be detected linearly by an OPLL, as suggested in Fig. 1. However, the photonic integrated OPLL device with high linearity is still under development. Therefore, in order to eliminate the spurious distortion due to an OPLL phase demodulator, an MZ interferometer, shown in Fig. 4, is first used for phase demodulation as a substitute. The core of the system is a PIC, which contains four independent phase modulators (QPM 1 4), two 3-dB multimode interference couplers, and two waveguide uni-traveling-carrier photodiodes. The UTC photodiode can achieve over 50-mA saturation photocurrent at 1 GHz [15]. The MQW phase modulators and the multimode interference couplers form an MZ interferometer. In each arm of the MZ interferometer, there are two phase modulators with lengths of 561 m and 786 m, respectively. Except for the length difference, the phase modulators have identical device structure. The nonlinearity of the MZ interferometer is carefully mitigated using the following approach. Let us consider the two phase modulators in the upper arm of the MZ interferometer (Fig. 5). The total phase perturbation of the arm is (12) (13a) (13b) where and are the voltages applied to the two phase modulators, respectively, and are the lengths of the two modulators, respectively. To isolate the nonlinearity of the MZ interferometer from the down-conversion process, and are adjusted to cancel the fundamental response in (12). This requires (14) Thus, the MZ interferometer operates in small phase condition, where its nonlinearity can be safely ignored. The MZ output should be linearly proportional to the combined optical phase (15) The quadratic optic phase modulator can thereby be determined. In deriving (15), we assume both modulators have identical. However, if the difference between of the two modulators exists, (15) should be modified as (16) where is the difference between the values of the two modulator devices. As long as is small compared with, the difference between (16) and (15) should be small. In the experiment, the two modulators have identical epitaxial structure and are integrated in the same PIC. Although nonuniformity in device fabrication may drift the values, the difference is small and we did not notice any significant perturbation to the measurement results. B. Down-Conversion Performance of InP Phase Modulator First, the output of the MZ interferometer was measured as a function of the bias voltage, from which the phase modulator response can be estimated. In this and the following measurements, light was launched into the modulator waveguide by

LI et al.: QUADRATIC ELECTROOPTIC EFFECT FOR FREQUENCY DOWN-CONVERSION 669 Fig. 7. Frequency spectrum captured at the PD output. (a) Fundamental tone due to the first phase modulator. (b) Down-converted tone. Fig. 6. PD output voltage versus the phase modulator bias voltage. Modulators 1 and 3 are biased simultaneously. The PD is terminated by a 1-k resistor. a tapered fiber, which was mounted on a submicrometer manipulation stage. Due to optical mode mismatch between the fiber and modulator optical waveguide, the coupling loss is large ( 5 db). The coupling efficiency of this configuration is also very sensitive to environmental temperature fluctuations. Over a long period ( 1 h), we observed up to 50% fluctuation in coupled optical power, which leads to 6-dB change in the photodiode output power. In order to eliminate its impact, a voltage ramp signal (6 V) with 5-ms duration was applied to the modulator. Both the input voltage ramp and photodiode output voltage were captured by a high-speed digital sampling scope (LeCroy DDA120). The measured MZ interferometer responses for different optical wavelengths are depicted in Fig. 6. The output deviates from sinusoidal function of conventional MZ modulator devices, as the phase modulation of the MQW modulator device is nonlinear. Unwanted electroabsorption is also evident with high bias voltage. At a short optical wavelength, the phase modulation is larger. However, it is at the expense of the enhanced electroabsorption. Thus, in order to minimize the electroabsorption, we chose 1570-nm optical wavelength for subsequent measurements. A curve fitting has been performed for the captured MZ response for 1570-nm optical wavelength. In the curve fitting, both the optical phase and the unwanted electroabsorption were modeled using a third-order polynomial. For this study, we estimated that quadratic phase modulation sensitivity per unit length is approximately equal to 0.3 rad/v mm. Next, we demonstrate the frequency down-conversion by the InP phase modulator. An RF tone (from a HP E4400B signal generator) and an LO tone (from a Rohde & Schwarz 5265 signal generator) were combined (Mini-circuits ZXC-2-4). Through an RF balun (Mini-circuits ADTL1-12), the combined RF/LO signal was then split into two paths ( and ) to feed the two MQW phase modulators separately. The RF balun provided an initial phase difference between two signal paths necessary for canceling the linear phase response. In addition, a voltage-controlled attenuator (Mini-circuits ZX73-2500 ) was employed to adjust the amplitude of, and a microwave line-stretcher phase shifter was applied to fine tune the phase of. During this and the subsequent measurements, the MZ interferometer was biased at quadrature. The RF bandwidth was limited to below 1 GHz because of the RC time of the lumped Fig. 8. Down-conversion versus input RF power. The phase modulator has a dc bias of 2 V. electrode of the InP phase modulator. With a traveling-wave electrode and velocity matching of the optical and RF/LO field, the RF bandwidth can be extended above tens of gigahertz. Fig. 7 shows a sample of the measured fundamental tone at 437.7 MHz and the down-converted tone at 76.6 MHz. Here, the MZ interferometer was biased at 2 V, and the 361.1-MHz LO signal was set at 10 dbm. The fundamental tone was captured at the PD output when the LO signal was turned off. The down-converted signal was captured with two channel canceling in presence. Thus, the down-converted signal only originated from the quadratic optical phase modulator, but not from the MZ nonlinearity. It should be mentioned that, during the measurement, there was strong EMI from the modulator electrode to the photodiodes. However, the EMI problem was mitigated using a carrier suppression technique reported in [16]. Fig. 8 shows the down-converted signal power of modulator 1 when the input RF power was swept from 4 to 3 dbm and the LO power was varied from 5 to 11 dbm. In this plot, (15) is applied to extract the down-converted signal power. The power of the down-converted signal is proportional to the input RF power. Increasing the LO power also enhances the down-converted signal. The down-converted optical phase corresponding to the output power is also determined and is marked in decibel scale on the right -axis of Fig. 8. The power loss from the RF input to the down-converted output from the photodetector exceeds 50 db, which is due to low optical power ( 1 mw) impinging into the photodiode.

670 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 58, NO. 3, MARCH 2010 Fig. 10. Spurious distortion in the down-converted signal. Fig. 9. Down-converted optical phase of phase modulator 1 (561-m long) versus input RF and LO power. G is the conversion gain. The phase down-conversion efficiency [defined in (3)] of the quadratic phase modulator (QPM1, 561- m long) is plotted against the LO power in Fig. 9. The phase conversion efficiency of this single device reaches 0.2 rad/v with an LO power of 11 dbm. This is already sufficient. In the high dynamic-range frequency down-conversion system, as suggested in Fig. 1, the length of the quadratic phase modulator should be a least a few millimeters long in order to obtain a large down-conversion phase IP3. The 0.2-rad/V conversion gain of a single 561-mm device will translate to larger than 1-rad/V conversion gain for a pair of the long device. The OPLL phase demodulator that is currently under development has a around 3 V. Thus, according to (4), the IF output will have a similar or even larger power than the RF input. Next, we evaluate the down-conversion linearity of the MQW quadratic phase modulator. A two-tone RF signal and a single LO signal was used as input. In addition, in order to minimize the impact of the MZ nonlinearity on the down-converted spurious level, small LO power was employed so that the down-converted optical phase was small enough for the MZ interferometer to still remain in the small angle condition. To compensate the resulting power loss at the photodiode output, a high saturation ( 20 dbm) amplifier (Mini-circuits ZX60-33LN ) was inserted behind the photodiode. A sample of the down-converted signal spectrum containing the spurious distortion is shown in Fig. 10. Here, the two-tone input centers at 437 MHz and separates by 0.04 MHz, and the phase modulator was biased at 2.5 V. To determine the intercept point, the down-converted signal level and the nonlinear distortion level are plotted as a function of the RF power, as shown in Fig. 11. The down-converted signal shows a 1:1 slope and the spurious distortion has a 3:1 slope. They intercept at 26-dBm RF input power. From the intercept point, we determine [using (8)] that the ratio between the second- and fourth-order coefficients (i.e., ) is 120 V. The two-tone distortion measurement has also been performed with other bias voltages. The results are summarized in Fig. 12. The ratio is large at low reverse-bias voltages. It decreases as the bias voltage increases. It is expected Fig. 11. Spurious distortion versus input RF power with the phase modulator bias voltage set at 2.5 V. Fig. 12. = versus modulator bias voltage. for the modulator device employing a conventional square quantum-well structure. Thus, low bias voltage is preferred for the quadratic optical phase modulator. If we choose a 10-dBm LO power, the aforementioned measurement results of the ratio suggest that the phase modulator device used here will have a down-conversion phase IP3 (per unit length) of 0.85 mm. This is already suitable for achieving a dynamic range in the range of 125 db Hz (see Fig. 4). However, this is inadequate for achieving a larger dynamic range such as 140 db Hz.

LI et al.: QUADRATIC ELECTROOPTIC EFFECT FOR FREQUENCY DOWN-CONVERSION 671 Fig. 13. Coherent PM optical link with down-conversion. C. PM Optical Link With Down-Converter Next, we present the preliminary results of a coherent PM optical link with a quadratic optical phase modulator for frequency down-conversion (see Fig. 13). The purpose of this experiment is to identify the conversion loss and feasibility of the proposed conversion approach when an OPLL is used for phase demodulation. Due to equipment limitation, only a single InP optical phase modulator is employed. The modulator is identical to QPM 1 (0.561-mm long), which was discussed in Section III-B. In addition, since the photonic integrated OPLL is still under development, a discrete OPLL was used instead. The details of the OPLL were discussed in [5]. It has a limited bandwidth ( 100 MHz). It contains an attenuation counter propagating (ACP) local phase modulator with equal to 2 V, which results in a phase demodulation sensitivity of 0.64 V/rad. In this link setup, the output of a single-mode fiber laser (Orbits Ethernal 2800A-30-PM) was amplified and split into the TX and LO optical paths. The TX path is fed into the InP phase modulator by a tapered fiber. Due to poor coupling efficiency, there is a large fiber-to-fiber loss ( 20 db). Therefore, we use another optical amplifier to boost up the optical power after the InP phase modulator. The optical power entering OPLL was 150 mw for each path. Three polarization controllers were employed to assure the correct polarization states for the light entering the InP phase modulator and OPLL. In addition, in order to correct the slow optical phase fluctuation induced by the environment, the low-frequency component of the OPLL output spectrum is fed back to a piezoelectric fiber line stretcher. This enables the optical link remain stably locked during the measurement process. An RF input was then applied to the InP phase modulator. Fig. 14 shows a sample of the output IF spectrum near 50 MHz when the (two-tone) RF input and the electrical LO frequency are 1 GHz and 950 MHz, respectively. The RF input power is 0 dbm per tone and the measured IF signal power is 16 dbm per tone. The observed conversion loss is 16 db. The conversion loss can be greatly reduced with a longer InP phase modulator and a improved OPLL phase demodulator. The 50-MHz IF output spectrum contains nonlinear spurious distortion. Fig. 15 shows the IF power and the spurious distortion level versus the RF input power. It is found that the input third-order intercept point (IIP3) is 20 dbm, which is lower than the measured value (26 dbm) for the 0.561-mmlong InP phase modulator (Section III-B). The enlarged nonlinearity is due to the phase demodulation process. With only one InP phase modulator, the down-converter is similar to an Fig. 14. IF output spectrum at 50 MHz. Input RF power is 0 dbm per tone and LO power is 10 dbm. The InP modulator was biased at 2.5 V. Fig. 15. IF power and spurious distortion versus RF input power. TABLE I COMPARISON OF DOWN-CONVERSION PERFORMANCE unbalanced mixer. Therefore, excessive fundamental tones, frequency doubled tones, and up-converted tones in the optical phase are contained. Those tones are beyond the OPLL bandwidth. They cause a large optical phase swing, resulting in an enlarged nonlinear distortion in the phase demodulation process. The image rejection configuration (as shown in Fig. 2) should mitigate this problem. The IF output also contains some noise spurs due to the relaxation oscillation of the fiber laser. However, if we assume shot noise limit, the dynamic range of the frequency down-conversion is 116 db Hz. The shot noise limit is achievable by

672 IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES, VOL. 58, NO. 3, MARCH 2010 Fig. 16. Optimized MQW phase modulator design. using a better laser source and by improving the fiber-to-fiber loss of the InP modulator to reduce the amplified spontaneous emission (ASE) noise from the optical amplifiers. IV. CONCLUSIONS AND CURRENT EFFORTS In this paper, we reported a new optical approach for down-converting RF signals by exploiting the nonlinearity of quadratic optical phase modulators. This approach can integrate with a linear PM optical links employing an OPLL phase demodulator. It has the potential to achieve a large dynamic range for radar frontend. It should also be applicable to any other applications that require high dynamic-range frequency down-conversion. Preliminary experimental studies have been performed to verify this concept using some existing InP modulator devices. We found that the down-conversion process is efficient. A 0.561-mm-long device achieved a phase down-conversion gain of 0.2 rad/v. However, the fourth-order nonlinearity of the InP phase modulator adds unwanted nonlinear distortion to the down-converted signal. Due to this, the current quadratic phase modulator device has a limited down-conversion phase IP3, which sets a dynamic-range limit of 125 db Hz with 10-dBm electric LO power. We then insert the InP phase modulator into a coherent PM optical link with a discrete OPLL linear phase demodulator. We achieved an electric RF to electric IF conversion loss of 16 db, which agrees with our measurement for the InP phase modulator. However, in the PM optical link, we also observed enlarged nonlinearity distortion. This is due to unwanted tones in the mixing process that interfere with the OPLL phase demodulation. Thus, we conclude that an image rejection scheme (Fig. 2) is required in order to preserve the OPLL linearity. Table I compares the achieved performance of the proposed approach and other down-conversion approaches [7] [11]. Even with a single 0.561-mm InP modulator device and a discrete OPLL phase demodulator, the proposed approach has already achieved better performance than other optical frequency downconversion methods [9] [11]. With help of an image rejection scheme and an improved InP modulator device, we expect the proposed approach surpass the state-of-the-art electronic downconverters [7], [8]. Our current research efforts focus on improving the InP phase modulator. As discussed in Section III, in order to achieve a dynamic range in the range of 140 db Hz, the InP modulator down-conversion phase IP3 must be improved. The phase IP3 can be improved by: 1) elongating the phase modulator and 2) enlarging the second-order coefficient and ratio. The first approach is straightforward. However, for the current device, this would necessitate an exceedingly long device length ( 14 mm), resulting in an unacceptable optical loss. Thus, an improved modulator design with minimized optical loss and a maximized ratio is required. One major source of the optical absorption comes from the p-doped cap of the modulator device. The absorption due to the p cap can be significantly mitigated by inserting an intrinsic InP layer between the P cap and the quantum well region, as shown in Fig. 16. Although this might reduce the field strength inside the quantum well region, it does not change the down-conversion phase IP3. Another source of the optical absorption comes from the absorption engendered by the quantum wells. To minimize this absorption, the quantum-well PL peak need to be detuned further away from the desired operating wavelength. This also reduces the unwanted electroabsorption. For enlarging the ratio, the quantum well used in this experimental study is not optimum. Numeric simulations have shown that by optimizing the quantum well width and the barrier height, the down-conversion phase IP3 can be doubled. The new generation InP optical phase modulator devices that includes the aforementioned optimizations is currently under development. ACKNOWLEDGMENT The authors wishes to thank B. Krantz, Boos Allen Harmilton, Fairfax, VA, Dr. G. Evans, South Methodist University, Dallas, TX, Dr. A. Rosen, Drexel University, Philadelphia, PA, and Dr. G. Gopalakrishnan, Defense Advanced Research Projects Agency (DARPA), Fairfax, VA, for some very helpful discussions. REFERENCES [1] E. I. Ackerman, Broad-band linearization of a Mach Zehnder electrooptic modulators, IEEE Trans. Microw. Theory Tech., vol. 47, no. 12, pp. 2271 2279, Dec. 1999. [2] J. H. Schaffner and W. B. Bridges, Inter-modulation distortion in high dynamic range microwave fiber-optic links with linearized modulators, J. Lightw. Technol., vol. 11, no. 1, pp. 3 6, Jan. 1993. [3] Y. Chiu, B. Jalali, S. Garner, and W. Steier, Broad-band electronic linearizer for externally modulated analog fiber-optic links, IEEE Photon. Technol. Lett., vol. 11, no. 1, pp. 48 50, Jan. 1999. [4] J. E. Bowers, A. Ramaswamy, L. A. Johansson, J. Klamkin, M. N. Sysak, D. Zibar, L. A. Coldren, M. J. Rodwell, L. Lembo, R. Yoshimitsu, D. Scott, R. Davis, and P. Ly, Linear coherent receiver based on a broadband and sampling optical phase-locked loop, presented at the Microw. Photon. 07, Victoria, BC, Canada, Oct. 2007. [5] Y. Li and P. Herczfeld, Coherent PM optical link employing ACP- PPLL, J. Lightw. Technol., vol. 27, no. 9, pp. 1086 1094, May 2009. [6] L. N. Langley, M. D. Elkin, C. Edge, M. J. Wale, U. Gliese, X. Huang, and A. J. Seeds, 7 14 GHz optical phase-locked loop (OPLL) source module for use in coherent optical beamforming applications, Electron. Lett., vol. 34, no. 24, pp. 2342 2343, Nov. 1998. [7] C. Fazi and P. G. Neudeck, Wide dynamic range RF mixers using wide-bandgap semiconductors, in IEEE MTT-S Int. Microw. Symp. Dig., Jun. 8 13, 1997, vol. 1, pp. 49 51. [8] J. P. Comeau and J. D. Cressler, A 28-GHz SiGe up-conversion mixer using a series-connected triplet for higher dynamic range and improved IF port return loss, IEEE J. Solid-State Circuits, vol. 41, no. 3, pp. 560 565, Mar. 2006. [9] G. K. Gopalakrishnan, W. K. Burns, and C. H. Bulmer, Microwaveoptical mixing in LiNbO3 modulators, J. Lightw. Technol., vol. 12, no. 12, pp. 2383 2381, Dec. 1993. [10] Y. Li and P. Herczfeld et al., Optical domain down-conversion of microwave signals for high dynamic range microwave fiber optics links, presented at the Microw. Photon. 06, Grenoble, France, Oct. 2006.

LI et al.: QUADRATIC ELECTROOPTIC EFFECT FOR FREQUENCY DOWN-CONVERSION 673 [11] A. Ramaswamy, L. A. Johansson, J. Klamkin, D. Zibar, L. A. Coldren, M. J. Rodwell, and J. E. Bowers, Optical phase demodulation of a 10 GHz RF signal using optical sampling, presented at the Coherent Opt. Technol. Appl. Conf., Boston, MA, Jul. 2008, Paper CTuC3. [12] Y. Li, R. Wang, J. Klamkin, L. Johansson, P. Herzfeld, and J. Bowers, RF frequency down-conversion using quadratic electro-optic effect, presented at the IEEE MTT-S Int. Microw. Symp., 2009, Paper 1685. [13] D. Pozar, Microwave Engineering, 2nd ed. New York: Wiley, 1998. [14] A. Yariv, Introduction to Optical Electronics, 2nd ed. New York: Holt, Renehart, Winston, c1976. [15] J. Klamkin, Y. C. Chang A. Ramaswamy, L. A. Johansson, J. E. Bowers, S. P. DenBaars, and L. A. Coldren, Output saturation and linearity of waveguide unitraveling-carrier photodiodes, IEEE J. Quantum Electron., vol. 44, no. 4, pp. 354 359, Apr. 2008. [16] Y. Li, R. Wang, G. Ding, J. Klamkin, L. Johansson, P. Herzfeld, and J. Bowers, Novel phase modulator linearity measurement, IEEE Photonic Technol. Lett, vol. 21, no. 10, pp. 1405 1407, Oct. 2009. InP-based PICs with emphasis on high-power photodiodes and novel coherent integrated receivers for highly linear microwave photonic links. In 2008, he joined the Electrooptical Materials and Devices Group, Massachusetts Institute of Technology (MIT) Lincoln Laboratory, where he is a member of the Technical Staff. His current research interests include directly modulated frequency stabilized slab-coupled optical waveguide lasers, GaN-based optical modulators, microwave photonic subsystems, high power photodiode arrays, and quantum-well intermixing techniques for novel PICs and devices. Leif A. Johansson (M 97) received the Ph.D. degree in engineering from University College London, London, U.K., in 2002. In 2002, he joined the University of California at Santa Barbara in a Postdoctoral position. His current research interests include design and characterization of integrated photonic devices for analog and digital applications. Yifei Li (M 03) received the B.Eng. degree in optoelectronics from the Huazhong University of Science and Technology, Luoyang, China, in 1996, and the M.S. and Ph.D. degrees in electrical engineering from Drexel University, Philadelphia, PA, in 2001 and 2003, respectively. From 2003 to 2007, he was a Research Faculty Member with the Center for Microwave/Lightwave Engineering, Drexel University. He is currently with the University of Massachusetts at Dartmouth. His research interests include high dynamic-range RF/photonic links, tunable microchip lasers, hybrid lidar/radar, fiber radio systems, coherent optical communications, and laser nonlinear dynamics. Peter R. Herczfeld (S 66 M 67 SM 89 F 91 LF 08) was born in Budapest, Hungary, in 1936. He received the M.S. degree in physics and Ph.D. degree in electrical engineering from the University of Minnesota at Minneapolis St. Paul, in 1963 and 1967, respectively. Since 1967, he has been with Drexel University, Philadelphia, PA, where he is currently the Lester Kraus Professor of Electrical and Computer Engineering. He is the Director of the Center for Microwave-Lightwave Engineering, Drexel University. He has also served as a consultant to numerous large and small corporations. He has authored or coauthored over 400 papers in solid-state electronics, microwaves, photonics, solar energy, and biomedical engineering. Renyuan Wang (M 08) received the B.Eng in electronic science and technology from the Harbin Institute of Technology, Harbin, China, in 2007, and is currently working toward the M.S. degree in electrical and computer engineering at the University of Massachusetts at Dartmouth. In 2007, he joined the RF/Photonics Group, University of Massachusetts at Dartmouth, where he was involved with development of high dynamic photonic RF radar front-ends. His research interests include quantum electronics, and opto-electronic, photonic, and high-speed electronic devices. Jonathan S. Klamkin (M 00) received the B.S. degree in electrical and computer engineering from Cornell University, Ithaca, NY, in 2002, and the M.S. degree in electrical and computer engineering and Ph.D. degree in electronic materials from the University of California at Santa Barbara (UCSB), in 2004 and 2008, respectively. While with UCSB, he was involved with the design, growth, fabrication, and characterization of widely tunable semiconductor lasers, photodetectors, optical intensity and phase modulators, and semiconductor optical amplifiers for John E. Bowers (F 04) received the M.S. and Ph.D. degrees from Stanford University, Stanford, CA. He is a Professor with the Department of Electrical Engineering and with the Technology Management Program, University of California at Santa Barbara (UCSB). He is also CTO and cofounder of Calient Networks. He is cofounder of the Center for Entrepreneurship and Engineering Management, and founder of Terabit Technology. He was with AT&T Bell Laboratories and Honeywell prior to joining UCSB. He has authored or coauthored eight book chapters, 400 journal papers, and 600 conference papers. He holds 49 patents. His research interests are primarily concerned with silicon photonics, opto-electronic devices, optical switching, and transparent optical networks. Dr. Bowers is a Fellow of the Optical Society of America (OSA) and the American Physical Society. He was an elected member of the IEEE Lasers and Electro-Optics Society (IEEE LEOS) Board of Governors, an IEEE LEOS Distinguished Lecturer, and vice president for Conferences for IEEE LEOS. He is a member of the National Academy of Engineering. He was a recipient of the IEEE LEOS William Streifer Award and the South Coast Business and Technology Entrepreneur of the Year Award. He was also the recipient of the 2007 ACE Award for Most Promising Technology for the hybrid silicon laser.