Abstract
Millimetrewave (mmWave) technology continues to draw great interest due to its broad applications in wireless communications, radar, and spectroscopy. Compared to pure electronic solutions, photonicbased mmWave generation provides wide bandwidth, low power dissipation, and remoting through lowloss fibres. However, at high frequencies, two major challenges exist for the photonic system: the power rolloff of the photodiode, and the large signal linewidth derived directly from the lasers. Here, we demonstrate a new photonic mmWave platform combining integrated microresonator solitons and highspeed photodiodes to address the challenges in both power and coherence. The solitons, being inherently modelocked, are measured to provide 5.8 dB additional gain through constructive interference among mmWave beatnotes, and the absolute mmWave power approaches the theoretical limit of conventional heterodyne detection at 100 GHz. In our freerunning system, the soliton is capable of reducing the mmWave linewidth by two orders of magnitude from that of the pump laser. Our work leverages microresonator solitons and highspeed modified unitraveling carrier photodiodes to provide a viable path to chipscale, highpower, lownoise, highfrequency sources for mmWave applications.
Introduction
Millimetre waves (mmWaves) provide key advantages in communication bandwidth, radar resolution, and spectroscopy thanks to their high carrier frequencies^{1,2,3}. Photonic oscillators operate at frequencies of hundreds of THz, and the frequency of the electrical signal produced by, e.g., the heterodyne detection of two lasers, is limited only by the photodiode bandwidth. However, at mmWave frequencies, the output power of the photonic system suffers from the power rolloff associated with the photodiode bandwidth. In terms of signal coherence, stabilizing the frequency difference of two lasers to a lowfrequency reference is challenging for mmWaves due to the high frequency.
The recent development of dissipative Kerr solitons in microresonators^{4,5,6,7,8,9} provides an integrated solution to address the challenges of photonicgenerated mmWaves in both power and coherence. These solitary wave packets achieve modelocking by leveraging Kerr nonlinearity to compensate for cavity loss and to balance chromatic dispersion^{4,10}. Microresonator solitons have been applied to metrology^{11}, optical communications^{12}, and spectroscopy^{13,14} in the form of microresonatorbased frequency combs (microcombs)^{15}. Due to the miniaturized dimension, the repetition rate of microresonator solitons ranges from a few GHz to THz^{16,17}. Direct detection of solitons with a fast photodiode produces mmWaves at the repetition frequency of the solitons. When compared with the conventional twolaser heterodyne detection method, soliton modelocking provides up to a 6 dB gain in mmWave output power due to the constructive interference among beatnotes created by different pairs of neighbouring comb lines^{18}. This additional gain is of great importance at high frequencies since it can relax the bandwidth requirements in the photodiode. In terms of signal coherence, recent studies have shown that the phase noise of the soliton repetition frequency at 10’s of GHz can be orders of magnitude smaller than that of its pump laser^{5,19,20,21}. When microresonator solitons are married with integrated lasers^{22,23}, amplifiers^{24}, and highspeed photodiodes^{25} through heterogeneous or hybrid integration, a fully integrated mmWave platform can be created with highpower, highcoherence performance, and the potential for largescale deployment through mass production (Fig. 1).
In this letter, we demonstrate highpower, highcoherence photonic mmWave generation at 100 GHz frequency through the combination of integrated microresonator solitons and a modified unitraveling carrier photodiode (MUTC PD). A 5.8 dB increase in mmWave power is obtained by using microresonator solitons when compared to the output power of conventional heterodyne detection. Importantly, the power level we achieve with microresonator solitons approaches the theoretical limit of heterodyne detection, which assumes an ideal photodiode with zero power rolloff in its frequency response. The system also achieves a maximum mmWave power of 7 dBm, one of the highest powers ever reported at 100 GHz^{26}. For our freerunning system, the 100 GHz signal has Lorentzian and Gaussian linewidths of 0.2 kHz and 4.0 kHz, respectively, which are two orders of magnitude smaller than that of the pump laser. The dependence of output power on the number of comb lines and chromatic dispersion is carefully studied both theoretically and experimentally. Our demonstration paves the way for a fully integrated photonic microwave system with soliton microcombs and highspeed photodiodes.
Results
In conventional heterodyne detection, mmWaves are generated when two laser lines mix with each other on a photodiode and create one beat note. However, when using an optical frequency comb, each comb line will beat with its two adjacent neighbouring lines to create beatnotes at the comb repetition frequency. For a comb that consists of N comb lines, (N−1) beatnotes will be created at the comb repetition frequency. Therefore, for the same average optical power, the comb can produce up to twice the number of beatnotes per laser line than heterodyne detection, and thus generate twice the AC photocurrent. The output power from the photodiode at the comb repetition frequency can be described as^{18,27}
where I_{DC} is the average photocurrent, R_{L} (50 Ω) is the load resistor, and N ≥ 2 is the number of comb lines. Γ is the measured relative mmWave power rolloff for the photodiode, and is ~5.5 dB for the 7 μm and ~6 dB for the 8 μm diameter PDs used in this work at 100 GHz. Clearly, the power at the limit of N → ∞ is 4 times (6 dB) higher than the power of heterodyne detection, where N = 2.
In practice, however, conventional frequency combs are not the best candidates to achieve the 6 dB gain for mmWave generation due to their low repetition frequencies. Previously, two attempts with electrooptics modulation frequency combs were reported, where linebyline amplitude and phase shaping were used to remove the unnecessary comb lines and increase the repetition rate from 20 GHz to 100 and 160 GHz^{18,28}. This postspectral filtering nonetheless increases the complexity and costs of the system. Conversely, microresonator solitons have comb repetition rates ranging from a few GHz to 1 THz, and can be directly applied to mmWave generation. MmWave generation with soliton microcombs in taperedcoupled microtoroid resonator^{29}, from dualcomb structure^{30}, and from a pair of comb lines^{31} has been shown, but there was no investigation into the output power.
The dissipative Kerr solitons used in this work are generated in an integrated, buswaveguide coupled Si_{3}N_{4} microring resonator with a freespectral range (FSR) of ∼100 GHz. The experimental setup is shown in Fig. 2. The single soliton state with a 35.4 fs pulse width is generated, and its squared hyperbolic secant spectral envelope is characterized by an optical spectrum analyser (Fig. 3a). The comb is then amplified by an erbiumdoped fibre amplifier (EDFA) and sent to the photodiode, and an optical programmable waveshaper (WS) is used to compensate for the group velocity dispersion and to suppress the amplified spontaneous emission (ASE) noise from the EDFA. The inset of Fig. 3a shows the optical spectrum after amplification and dispersion compensation. The photodiode used in this work is based on the chargecompensated modified unitraveling carrier photodiode (MUTC PD) structure. MUTC PDs operate under the principle of single carrier transit, and compared to traditional pin photodiodes, isolating electrons for this transit process eliminates the dependency on the slowertraveling holes, leading to higherspeed operation. To further enhance performance and limit thermal degradation, the PDs are then flipchip bonded to a ceramic substrate made of gold transmission lines grown on aluminium nitride submount (AlN)^{32}. Pictures of the microresonator and a PD die are shown in Fig. 3b and Fig. 3c, respectively. Details of microresonator solitons and photodiodes are described in the “Materials and methods” section.
To characterize the 6 dB power increase from the microresonator solitons, the PD output powers are measured for both microresonator soliton detection and heterodyne detection on four of our PDs with 7, 8, 10, and 11 μm diameters. The heterodyne measurements are performed using two continuouswave lasers with the same optical power and polarization. A variable optical attenuator is used to control the optical power illuminating on the PD. In the linear region of PD operation, the 100 GHz mmWave powers at different photocurrents are shown in Fig. 3d for the 7 µm device. The DC photocurrent is a direct measurement of the optical power illuminating on the PD. In the experiment, the coupling distance from the fibre to the PD is increased for uniform illumination, resulting in 1 mA photocurrent for 11 mW optical input power. The mmWave power generated from the microresonator solitons is measured to be 5.8 dB higher than that of heterodyne detection. This power increase approaches the 6 dB theoretical limit, and is verified on all four PDs with different diameters (shown in the inset of Fig. 3d). As a result of the 6 dB power increase, the mmWave power generated using microresonator solitons is within 1 dB of the theoretical power limit of heterodyne detection (solid black line in Fig. 3d), where the detector is assumed to be ideal and has no power rolloff at mmWave frequency. It shall be noted that no optical spectrum flattening is applied in our measurement. For a 5.8 dB power improvement, a 3 dB bandwidth of 7 comb lines is required for the Sech^{2} or Gaussian spectral envelope. As discussed in the “Materials and Methods” section, the shape of the spectral envelope has little effect on the mmWave power when the number of comb lines is large.
The electrical spectrum of the 100 GHz mmWave signal is measured and shown in Fig. 3e. Limited by the available bandwidth of our electrical spectrum analyser, we down convert the 100 GHz mmWave by sending it to an RF mixer to mix it with the fifth harmonic of a 20.2 GHz local oscillator. The mixer generates a difference frequency at Δf = 5f_{LO} − f_{r}. Δf is measured to be 1.2410 GHz, and we can derive the mmWave frequency as f_{r} = 99.7590 GHz. A lownoise, narrow signal is clearly observed at 3 kHz resolution bandwidth (RBW) in Fig. 3e (red trace). The signal is fitted with a Lorentzian, and the 3dB bandwidth is 0.2 kHz (zoomedin panel in Fig. 3e). Note that the soliton repetition rate is subject to fluctuations (laser frequency drift, temperature, etc.), and the central part of the signal is Gaussian with a 3dB linewidth of 4 kHz. This narrow linewidth at 100 GHz frequency is obtained for a freerunning microcavity soliton, which is driven by a pump laser with significantly broader linewidth (∼200 kHz, New Focus 6700 series specification). We note that there are a few bumps around 50 kHz offset frequency, which are likely to be derived from the technical noise of the pump laser. To compare the signal coherence between the conventional heterodyne method and the soliton method, the heterodyne signal of beating the pump laser and another 6700 series New Focus laser is also measured and shown in Fig. 3e (blue trace). At the same RBW, the heterodyne signal has poor coherence, and its frequency is drifting >5 MHz. Our measurements show that using freerunning microcavity solitons can reduce the linewidth of mmWave signals by two orders of magnitude, giving the microresonator soliton platform a key advantage over conventional heterodyne detection. No RF reference is used to stabilize the mmWave; in fact, the only controls used are the coarse temperature controls of the laser and the microresonator, to offset the change in environmental temperature. Further measurements of phase noise and Allan deviation will be introduced in later paragraphs.
Next, we verify the dependence of the mmWave power increase on the number of comb lines, which is described in Eq. (1). A linebyline waveshaping filter is used to select the number of comb lines that pass to the MUTC PD. We tested the number of comb lines from 2 to 22 at four different photocurrent levels (optical power), and the result is shown in Fig. 4a. Three representative optical spectra for 2, 12, and 22 comb lines are shown in Fig. 4b. The measured mmWave power follows the calculated curves. Interestingly, a 3 or 5 dB increase in power requires only 4 or 9 comb lines. This relatively small demand for comb lines relaxes the microresonator soliton requirement in terms of its optical bandwidth.
The increase in mmWave power occurs only when the beatnotes generated by different pairs of comb lines are in constructive interference. This is not always the case if there is dispersion between the microresonator and the PD. This effect is studied by applying programmable dispersion using a waveshaper. The measurement of mmWave power versus waveshaper dispersion is shown in Fig. 4c. The effect can be calculated analytically by adding the phase to each comb line, and will modify Eq. (1) to
where c is the speed of light, and d = d_{0} + d_{c} is the accumulated group velocity dispersion between the microresonator and PD. d_{0} denotes the offset dispersion in the system introduced by fibres and amplifiers, and d_{c} represents the dispersion compensation added by the waveshaper. The derivation of Eq. (2) is shown in the “Materials and Methods” section. The measurement and theory prediction agree very well when an offset dispersion of d_{0} = 1.95 ps/nm is included. The offset dispersion exists in our system because of the 70m fibre used to connect the microcomb lab and photodetector lab (contributing 1.26 ps/nm), with the rest of the dispersion coming from the fibres in the EDFA. N is used as a free parameter for fitting the experimental curve, and N = 15 is used for the dashed line in Fig. 4c. The fitted N should be interpreted as the effective number of comb lines to account for the spectral envelope shape. When the entire system is fully integrated, the overall length of waveguides will be well below a metre, and the dispersion will not impact the mmWave power.
We obtain a maximum output power of 7 dBm at 22.5 mA for the 8 μm device shown in Fig. 5a, due to the optimized light coupling from the size match of the 8 μm spotsize collimated fibre and diameter of the PD’s absorber. Using Eq. (1), we find that the ideal heterodyne response for this 8 μm device would need 26.7 mA to achieve 7 dBm, which means we can produce the same power at a lower average photocurrent using soliton excitation. The 7 dBm saturation power is recorded at −3.6 V bias. Increasing the reverse bias can improve the saturation power; however, ultimately this can cause PD thermal failure^{33} due to the rise in junction temperature from the dissipated power in the PD (reverse bias × average photocurrent). One advantage of using solitons is that they can generate the same RF output power at a lower photocurrent than the twolaser heterodyne method, and thus can reduce the dissipated power and allow the PD to be operated further below the point of thermal failure.
We further characterize the phase noise of the mmWaves generated from the freerunning microcavity solitons, and compare it to the phase noise from the heterodyne method. Similar to the linewidth measurement, the 100 GHz mmWave signal is downconverted in an RF mixer, where it is mixed with the fifth harmonic of a 20.2 GHz local oscillator. To minimize the effect of frequency drifting in the phase noise measurement, the frequency of the downconverted signal is further divided down electrically by a factor of 14 and 100 for the soliton and heterodyne, respectively. The phase noise is then measured in the electrical spectrum analyser with direct detection technique, and the result (at 100 GHz) is shown in Fig. 5b. Due to the large frequency drift, the heterodyne phase noise below 20 kHz offset frequency cannot be accurately characterized and thus is not presented. The soliton phase noise beyond 100 kHz is potentially limited by the measurement sensitivity, which is set by the noise floor of the spectrum analyser (dashed green), and the phase noise of the local oscillator (Keysight, PSG E8257D) used to down convert the mmWave (dashed black). The measurement shows that the freerunning solitons can reduce the mmWave phase noise by >25 dB compared to that of the heterodyne method. The reduction in phase noise from the pump laser frequency to the soliton repetition rate is a result of the noise transfer mechanism in microresonator solitons^{20}. Our observation is in agreement with previous reports of Xband and Kband microwave generation with microresonator solitons^{20,21,34}. The phase noise of solitonbased mmWaves can be further reduced in the future by using a pump laser with higher stability^{35}, tuning the soliton into quiet operation point^{20}, and implementing better temperature control of the entire system. For instance, a compact externalcavity diode laser has recently achieved a Lorentzian linewidth of 62 Hz^{36}. Using this laser to drive the soliton could further reduce the freerunning mmWave phase noise.
Finally, the Allan deviations of the mmWave generated from the soliton and the heterodyne detection are measured by counting the frequency of the downconverted signal on a zero deadtime counter (Fig. 5c). At 1 ms gate time, the Allan deviation of the solitonbased mmWave reaches the minimum of < 0.7 kHz, which is more than two orders of magnitude better than that of heterodyne detection. Above 1 ms gate time, the Allan deviation of the solitonbased mmWave increases due to the pump laser frequency drift and temperature fluctuation. Stabilizing the mmWave signal to a lowfrequency reference could provide longterm stability, which will increase the system complexity, but is possible through the electrooptics modulation method^{31} or dual microcavity soliton methods^{11,37}.
Discussion
In summary, we have demonstrated highpower, highcoherence mmWave generation at 100 GHz by using integrated microresonator solitons and MUTC PDs. Extending the frequency to several hundred GHz is possible. For microresonator solitons, the highest repetition rate reported is 1 THz^{17}, while demonstrated MUTC PDs have detection capabilities of at least 300 GHz^{38,39}. As the microresonator solitons consume very little pump power, and most of the pump transmits through the waveguide^{5}, it is possible to recycle the pump laser power to drive the next microresonator solitons (Fig. 1). Two tandem microresonator solitons driven by the same pump laser have been reported previously^{14}. The proposed platform has the potential to be fully integrated on a single chip, which can enable largescale mmWave arrays. The four critical components, laser, Kerr microresonator, amplifier, and ultrafast photodiode, have all been shown to be compatible with Si_{3}N_{4} photonic platforms through heterogeneous integration. Once all components are fully integrated, we expect that the platform can deliver a new paradigm regarding scalable, integrated photonics technologies for applications at very high frequencies, and thus provide a path to compact, lownoise, highfrequency sources for spectroscopy, ranging, and wireless communications.
Materials and methods
Microresonator soliton generation
The dissipative Kerr solitons used in this work are generated in an integrated, buswaveguide coupled Si_{3}N_{4} microring resonator. The resonator has an FSR of ∼100 GHz, an intrinsic quality factor of 2.6 × 10^{6} and a loaded quality factor of 2.2 × 10^{6}. The SiN resonator has a crosssection, width × height, of 1.55 × 0.8 µm^{2}, and is coupled to a buswaveguide of the same crosssection. The resonator radius is 0.24 mm, and the solitongeneration mode has anomalous dispersion of ∼1 MHz/FSR. A thermoelectric cooler (TEC) is placed beneath the microresonator to coarsely overcome environmental temperature fluctuations. To generate a single soliton state, a rapid pump laser frequency scanning method^{40} is applied to overcome the thermal complexity when accessing the reddetuned soliton existence regime. The detailed experimental setup is shown in Fig. 2. The pump laser is derived from the first phasemodulated sideband of a continuouswave laser, and the sideband frequency can be rapidly tuned by a voltagecontrolled oscillator (VCO). The pump laser scans its frequency at the speed of ∼20 GHz/µs, and the scan is stopped immediately once the pump laser frequency reaches the reddetuned regime of the resonator. The pump power in the waveguide is 1 Watt, which could be reduced in the future by 2 orders of magnitude by improving the quality factor and minimizing the thermal effect^{41}. The optical spectrum has a 3dB bandwidth of 5.4 THz, which contains a sufficient number of comb lines for photodetection. The solitons are coupled from the SiN onchip bus waveguide into a lensed fibre. Before reaching the MUTC PD, the soliton/heterodyne laser is amplified to >200 mW, and a variable optical attenuator is used to precisely control the illumination power. Finally, the solitons are coupled to the surface normal MUTC PD through an 8 μm collimated fibre.
Modified unitraveling carrier photodiode
The chargecompensated modified unitraveling carrier photodiode (MUTC PD) operates under the principle of single carrier transit, and compared to traditional pin photodiodes, isolating electrons for this transit process eliminates the dependency on the slowertraveling holes leading to higherspeed operation. To accomplish this, the photon absorption process which generates electronhole pairs in the PD absorber layer, occurs close to the pcontact layer, allowing the excess holes to be quickly collected in response to the ptype material dielectric relaxation time. To further enhance the speed of the PD response, by step grading the doping of the partially depleted absorber, an electric field is generated which accelerates the electrons through the absorber and towards the transparent and depleted drift layer. To prevent electric field collapse at the heterointerface of the absorber and drift layer, a fully depleted absorber layer and a moderately doped cliff layer help to maintain electric field strength and accelerate the electrons into the drift layer^{42,43}. Once in the drift layer, electron spacecharge effects are mitigated or chargecompensated by the light ntype doping in the drift layer^{44}. Fabrication flow of the PDs and similar PD epitaxial layering structures have been reported previously^{45}, and so has the AlN submount^{46}. The MUTC PDs used in this experiment have demonstrated dark currents as low as 200 pA at − 2 V, a 3dB bandwidth of up to 145 GHz (4 μm diameter PD), a responsivity of 0.2 A/W at 1550 nm, and a −2.6 dBm maximum output power at 160 GHz at − 3 V bias^{38}. They have also been investigated as viable receivers for soliton applications ranging from 50 to 500 GHz^{47,48}. The 3dB bandwidth at 5 mA and − 3 V bias for the 7, 8, 10, and 11 μm diameter PDs used in this experiment are 92 GHz, 90 GHz, 70 GHz, and 70 GHz, respectively. Note that in Fig. 3d, the ideal heterodyne power calculated using Eq. (1) where N = 2, assumes a 100% modulation depth; however, the measured modulation depth of the signal was 89%, leading to the observed mismatch in the measured and calculated heterodyne power.
MmWave linewidth reduction
Our observation of linewidth reduction is in agreement with previous reports of microresonator solitons at X and Kband repetition frequencies^{20,21}. The soliton repetition frequency equals the cavity FSR at the wavelength of the soliton spectral envelope centre. Both Raman selffrequency shift^{49} and dispersive wave recoils can affect the soliton envelope centre wavelength^{6,20}, and they are functions of lasercavity frequency detuning. This can be clearly seen in Fig. 3a, as our soliton’s envelope centre is to the red side of the pump laser. Because of the chromatic dispersion, the FSR at different wavelengths is different, and thus, the variation in the pump laser frequency, f_{p}, will alter the soliton spectral envelope centre, and change the soliton repetition rate, f_{r}. To the first order, the transfer of frequency variation from the pump (δf_{p}) to the repetition rate (δf_{r}) can be described as \(\delta f_{\rm{r}} = \frac{{\partial f_{\rm{r}}}}{{\partial f_{\rm{p}}}} \times \delta f_{\rm{p}}\), where δ denotes the variation. For both silica and silicon nitride resonators^{20,50}, this transfer coefficient \(\frac{{\partial f_{\rm{r}}}}{{\partial f_{\rm{p}}}}\) has been measured to be on the level of 10^{−2}, and thus, the soliton repetition rate linewidth is much smaller than that of the pump laser.
MmWave power versus dispersion
Optical pulses that propagate in an optical fibre will acquire additional phase due to group velocity dispersion in the fibre. Suppose the centre frequency of the pulse is ω_{p}; then, the component at frequency ω will acquire a relative phase after propagation of distance z^{51}:
where \(E(0,w)=E_{0}{\sqrt{(2N)}}\,{\exp}({}{iwt})\) is the electrical field of light at frequency ω and position z = 0, normalized to the photon number per unit time. Here, we have assumed a flat spectrum for the comb, and N is the total number of comb lines. D_{λ} is the group velocity dispersion parameter, and D_{λ} ≈ 18 ps/nm/km for the SMF28 fibre at 1550 nm. For soliton frequency combs, (ω − ω_{p})/2π = n × f_{r} for the nth comb line from the spectral envelope centre, where f_{r} is the comb repetition frequency. Therefore, the photocurrent generated in the photodiode is
where we have used \(\mathop {\sum }\nolimits_{k = m}^n ar^k = a(r^m  r^{n + 1})/(1  r)\) to derive the term cos(2πf_{r}t), and we have set 2N_{0} + 1 = N. Higher harmonics of the repetition frequency are neglected as they are beyond the detection limit of our photodiode. Considering I_{DC} as the average photocurrent flowing through the load resistor R_{L}, the detected mmWave power at frequency f_{r} is as follows:
where we have defined d = D_{λ} × z as accumulated dispersion, and Γ is the PD power rolloff at the repetition frequency. This equation is the same as Eq. (2) in the main text. When dispersion is very small (d → 0), the detected mmWave power is approximated by
which is Eq. (1) in the main text.
MmWave power versus optical spectral envelope
In this section, we calculate the impact of the optical spectral envelope on the mmWave power. For simplicity, we assume that the optical envelope is symmetric along the envelope centre, and we assume no accumulated dispersion. For the nth comb line, we have
where function f(n) is real and describes the spectral envelope. We focus on the case where the number of comb lines is large, so that we can assume the envelope is smooth, and f(n + 1) − f(n) ≪ f(n). The photocurrent is then expressed as
where we have neglected higher harmonics of the repetition frequency again. The sum can be simplified by using the symmetric envelope condition, f(−n) = f(n), and we can substitute f(n + 1) = f(n) + Δf(n + 1/2), where Δf(n + 1/2) is the difference between f(n + 1) and f(n), and Δf(x) is an odd function. Therefore, we have
where we have used f(n)Δf(n + 1/2), which is approximated to an odd function when the spectrum is broad, and thus f(n + 1) − f(n) ≪ f(n), and Δf(n + 1/2) ≈ Δf(n). It is clear that when N and N_{0} are very large, the sum is dominated by the total optical power, \(\mathop {\sum }\nolimits_{n =  N_0}^{N_0} f^2(n)\), and is almost irrelevant to the function of the envelope. The mmWave power can be expressed as
when N_{0} → ∞, \(f\left( {N_0} \right)f(N_0 + 1) \ll \mathop {\sum }\nolimits_{  N_0}^{N_0} f^2(n)\), and the power gain relative to heterodyne detection approaches 6 dB regardless of the spectral envelope f(n). It shall be noted that this result applies only to the case where the spectral envelope is symmetric and smooth; otherwise, the approximation used in Eq. (9) will fail.
Data availability
The data that support the plots within this paper and other findings of this study are available from the corresponding author upon reasonable request.
References
 1.
Cooper, K. B. et al. Penetrating 3D imaging at 4 and 25m range using a submillimeterwave radar. IEEE Trans. Microw. Theory Tech. 56, 2771–2778 (2008).
 2.
KleineOstmann, T. & Nagatsuma, T. A review on terahertz communications research. J. Infrared Millim. Terahertz Waves 32, 143–171 (2011).
 3.
Koenig, S. et al. Wireless subTHz communication system with high data rate. Nat. Photonics 7, 977–981 (2013).
 4.
Herr, T. et al. Temporal solitons in optical microresonators. Nat. Photonics 8, 145–152 (2014).
 5.
Yi, X. et al. Soliton frequency comb at microwave rates in a highQ silica microresonator. Optica 2, 1078–1085 (2015).
 6.
Brasch, V. et al. Photonic chip–based optical frequency comb using soliton Cherenkov radiation. Science 351, 357–360 (2016).
 7.
Gong, Z. et al. Highfidelity cavity soliton generation in crystalline AlN microring resonators. Opt. Lett. 43, 4366–4369 (2018).
 8.
Gaeta, A. L., Lipson, M. & Kippenberg, T. J. Photonicchipbased frequency combs. Nat. Photonics 13, 158–169 (2019).
 9.
He, Y. et al. Selfstarting bichromatic LiNbO_{3} soliton microcomb. Optica 6, 1138–1144 (2019).
 10.
Kippenberg, T. J. et al. Dissipative Kerr solitons in optical microresonators. Science 361, eaan8083 (2018).
 11.
Spencer, D. T. et al. An opticalfrequency synthesizer using integrated photonics. Nature 557, 81–85 (2018).
 12.
MarinPalomo, P. et al. Microresonatorbased solitons for massively parallel coherent optical communications. Nature 546, 274–279 (2017).
 13.
Suh, M. G. et al. Microresonator soliton dualcomb spectroscopy. Science 354, 600–603 (2016).
 14.
Dutt, A. et al. Onchip dualcomb source for spectroscopy. Sci. Adv. 4, e1701858 (2018).
 15.
Del’Haye, P. et al. Optical frequency comb generation from a monolithic microresonator. Nature 450, 1214–1217 (2007).
 16.
Suh, M. G. & Vahala, K. Gigahertzrepetitionrate soliton microcombs. Optica 5, 65–66 (2018).
 17.
Li, Q. et al. Stably accessing octavespanning microresonator frequency combs in the soliton regime. Optica 4, 193–203 (2017).
 18.
Kuo, F. M. et al. Spectral power enhancement in a 100 GHz photonic millimeterwave generator enabled by spectral linebyline pulse shaping. J. IEEE Photonics 2, 719–727 (2010).
 19.
Liang, W. et al. High spectral purity Kerr frequency comb radio frequency photonic oscillator. Nat. Commun. 6, 7957 (2015).
 20.
Yi, X. et al. Singlemode dispersive waves and soliton microcomb dynamics. Nat. Commun. 8, 14869 (2017).
 21.
Liu, J. Q. et al. Photonic microwave generation in the X and Kband using integrated soliton microcombs. Nat. Photonics 14, 486–491 (2020).
 22.
Stern, B. et al. Batteryoperated integrated frequency comb generator. Nature 562, 401–405 (2018).
 23.
Xiang, C. et al. Narrowlinewidth IIIV/Si/Si_{3}N_{4} laser using multilayer heterogeneous integration. Optica 7, 20–21 (2020).
 24.
de Beeck, C. O. et al. Heterogeneous IIIV on silicon nitride amplifiers and lasers via microtransfer printing. Optica 7, 386–393 (2020).
 25.
Yu, Q. H. et al. Heterogeneous photodiodes on silicon nitride waveguides. Opt. Express 28, 14824–14830 (2020).
 26.
Sun, K. Y. & Beling, A. Highspeed photodetectors for microwave photonics. Appl. Sci. 9, 623 (2019).
 27.
Urick, V. J. Jr., McKinney, J. D. & Williams, K. J. Fundamentals of Microwave Photonics (John Wiley & Sons, New York, 2015).
 28.
Wun, J. M. et al. Photonic highpower 160GHz signal generation by using ultrafast photodiode and a highrepetitionrate femtosecond optical pulse train generator. IEEE J. Sel. Top. Quantum Electron. 20, 3803507 (2014).
 29.
Zhang, S. Y. et al. Terahertz wave generation using a soliton microcomb. Opt. Express 27, 35257–35266 (2019).
 30.
Zang, J. Z. et al. Wideband millimeterwave synthesizer by integrated microcomb photomixing. In Conference on Lasers and ElectroOptics (CLEO), SF1O.1 2020 (Optical Society of America, 2020). https://doi.org/10.1364/CLEO_SI.2020.SF1O.1.
 31.
Tetsumoto, T. et al. 300 GHz wave generation based on a Kerr microresonator frequency comb stabilized to a low noise microwave reference. Opt. Lett. 45, 4377–4380 (2020).
 32.
Xie, X. J. et al. Improved power conversion efficiency in highperformance photodiodes by flipchip bonding on diamond. Optica 1, 429–435 (2014).
 33.
Xie, X. J. et al. Photonic generation of highpower pulsed microwave signals. J. Lightwave Technol. 33, 3808–3814 (2015).
 34.
Weng, W. L. et al. Frequency division using a solitoninjected semiconductor gainswitched frequency comb. Sci. Adv. 6, eaba2807 (2020).
 35.
Lucas, E. et al. Ultralownoise photonic microwave synthesis using a soliton microcombbased transfer oscillator. Nat. Commun. 11, 374 (2020).
 36.
Volet, N. et al. Microresonator soliton generated directly with a diode laser. Laser Photonics Rev. 12, 1700307 (2018).
 37.
Wang, B. C. et al. Vernier frequency division with dualmicroresonator solitons. Nat. Commun. 11, 3975 (2020).
 38.
Beling, A. et al. Highspeed integrated photodiodes. In Proc. 2019 24th OptoElectronics and Communications Conference (OECC) and 2019 International Conference on Photonics in Switching and Computing (PSC). Fukuoka: 2019, 1–3, https://doi.org/10.23919/PS.2019.8818022.
 39.
Dülme, S. et al. 300 GHz photonic selfmixing imaging system with vertical illuminated tripletransitregion photodiode Terahertz emitters. In 2019 International Topical Meeting on Microwave Photonics (MWP) 1–4 (IEEE, 2019). https://doi.org/10.1109/mwp.2019.8892098.
 40.
Stone, J. R. et al. Thermal and nonlinear dissipativesoliton dynamics in Kerrmicroresonator frequency combs. Phys. Rev. Lett. 121, 063902 (2018).
 41.
Liu, J. Q. et al. Ultralowpower chipbased soliton microcombs for photonic integration. Optica 5, 1347–1353 (2018).
 42.
Jun, D. H. et al. Improved efficiencybandwidth product of modified unitraveling carrier photodiode structures using an undoped photo absorption layer. Jpn. J. Appl. Phys. 45, 3475–3478 (2006).
 43.
Shimizu, N. et al. InPInGaAs unitravelingcarrier photodiode with improved 3dB bandwidth of over 150 GHz. IEEE Photonics Technol. Lett. 10, 412–414 (1998).
 44.
Li, N. et al. Highsaturationcurrent chargecompensated InGaAsInP unitravelingcarrier photodiode. IEEE Photonics Technol. Lett. 16, 864–866 (2004).
 45.
Li, Q. L. et al. Highpower flipchip bonded photodiode with 110 GHz bandwidth. J. Lightwave Technol. 34, 2139–2144 (2016).
 46.
Morgan, J. S. et al. Highpower flipchip bonded modified unitraveling carrier photodiodes with −2.6 dBm RF output power at 160 GHz. In Proc. 2018 IEEE Photonics Conference (IPC), 1–2. https://doi.org/10.1109/ipcon.2018.8527260 (2018).
 47.
Zang, J. Z. et al. Soliton microcombbased millimeterwave synthesizer. In: Proc. 2019 IEEE Avionics and Vehicle FiberOptics and Photonics Conference (AVFOP) 1–2 (2019).
 48.
Zang, J. Z. et al. Millimeterwave synthesizer based on microresonator soliton dualcomb photomixing. In OSA Advanced Photonics Congress (AP) 2019 (IPR, Networks, NOMA, SPPCom, PVLED). IT1A. 4 (Optical Society of America, 2019). https://www.osapublishing.org/abstract.cfm?uri=IPRSN2019IT1A.4#Abstract.
 49.
Karpov, M. et al. Raman selffrequency shift of dissipative Kerr solitons in an optical microresonator. Phys. Rev. Lett. 116, 103902 (2016).
 50.
Bao, C. Y. et al. Soliton repetition rate in a siliconnitride microresonator. Opt. Lett. 42, 759–762 (2017).
 51.
Agrawal, G. P. Nonlinear Fiber Optics 4th edn (Academic Press, Boston, 2007).
Acknowledgements
The authors thank Ligentec and VLC Photonics for resonator fabrication, S. Bowers at UVA for access to the spectrum analyser, and Q.F. Yang at Caltech for helpful comments during the preparation of this manuscript. The authors also gratefully acknowledge the support from the National Science Foundation and Defense Advanced Research Projects Agency (DARPA) under HR001115C0055 (DODOS). X.Y. is also supported by Virginia Space Grant Consortium.
Author information
Affiliations
Contributions
X.Y. and A.B conceived the concept. B.W. and J.S.M. performed the experiments, with assistance from K.S., M.J., and Z.Y. J.S.M., M.W., and S.E. fabricated the modified unitraveling carrier photodiode. B.W., J.S.M., A.B. and X.Y. analysed the data. All authors contributed to the writing of the manuscript.
Corresponding authors
Ethics declarations
Conflict of interest
The authors declare that they have no conflict of interest.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Wang, B., Morgan, J.S., Sun, K. et al. Towards highpower, highcoherence, integrated photonic mmWave platform with microcavity solitons. Light Sci Appl 10, 4 (2021). https://doi.org/10.1038/s4137702000445x
Received:
Revised:
Accepted:
Published:
Further reading

Visualising the heart of chaos
Light: Science & Applications (2021)

Special Issue on the 60th anniversary of the first laser—Series I: Microcavity Photonics—from fundamentals to applications
Light: Science & Applications (2021)

Direct observation of chaotic resonances in optical microcavities
Light: Science & Applications (2021)