Abstract
Solitons are selfreinforcing localized wave packets that manifest in the major areas of nonlinear science, from optics to biology and Bose–Einstein condensates. Recently, optically driven dissipative solitons have attracted great attention for the implementation of the chipscale frequency combs that are decisive for communications, spectroscopy, neural computing, and quantum information processing. In the current understanding, the generation of temporal solitons involves the chromatic dispersion as a key enabling physical effect, acting either globally or locally on the cavity dynamics in a decisive way. Here, we report on a novel class of solitons, both theoretically and experimentally, which builds up in spectrally confined optical cavities when dispersion is practically absent, both globally and locally. Precisely, the interplay between the Kerr nonlinearity and spectral filtering results in an infinite hierarchy of eigenfunctions which, combined with optical gain, allow for the generation of stable dispersionless dissipative solitons in a previously unexplored regime. When the filter order tends to infinity, we find an unexpected link between dissipative and conservative solitons, in the form of Nyquistpulselike solitons endowed with an ultraflat spectrum. In contrast to the conventional dispersionenabled nonlinear Schrödinger solitons, these dispersionless Nyquist solitons build on a fully confined spectrum and their energy scaling is not constrained by the pulse duration. Dispersionless soliton molecules and their deterministic transitioning to single solitons are also evidenced. These findings broaden the fundamental scope of the dissipative soliton paradigm and open new avenues for generating soliton pulses and frequency combs endowed with unprecedented temporal and spectral features.
Similar content being viewed by others
Introduction
Soliton phenomenon has amazed and stimulated scientists since its first observation by John Scott Russell in 1834^{1} and led to many remarkable applications, particularly in optics^{2,3}. The stationary pulsed solutions of the conservative nonlinear Schrödinger equation (NLSE) once appeared as a fascinating prospect for highcapacity optical communications^{4}. The research was soon extended to the investigation of dissipative solitons in optical cavities for developing novel modelocked laser architectures^{5} and Kerr frequency combs^{6}. Dissipative solitons are sustained by the composite balance between dispersion and nonlinearity on the one hand and gain and loss on the other hand. In contrast to the conservative NLSE solitons, dissipative solitons feature frequency chirping, which reflects a necessary energy redistribution within the pulse as it propagates. The latter phenomenon increases the soliton diversity well beyond that available in Hamiltonian conservative systems. For instance, the mode locking of frequencychirped bright laser pulses becomes accessible within the normal dispersion regime through an interplay involving spectral filtering^{7,8,9}. Since it represents an elementary source of frequency chirping, the chromatic dispersion is considered as a key enabling physical effect in the generation of ultrashort pulses, acting either globally or locally on the cavity dynamics in a decisive way. For instance, some laser architectures, while they operate at a nearzero pathaveraged dispersion, utilize a strong local dispersion to enhance the intracavity dynamics resulting, for instance, in a significant energy increase for the dispersionmanaged propagating stretched pulses^{10,11}. In other architectures, the local dispersion is combined with a lumped spectral filtering to access, within the laser cavity, a selfsimilar propagation evolution that becomes a robust attracting state^{12}. Therefore, even if the net dispersion may be close to zero in the two architectures mentioned above, the local dispersion plays a major role in establishing the pulse dynamics. On the other hand, when the local secondorder dispersion approaches zero, higherorder dispersion terms may become important and also enable soliton formation^{13,14,15,16,17}. Although the existence of solitons in the absence of any chromatic dispersion has been suggested in the context of modelocked lasers involving saturable absorbers^{18,19}, no clear supportive results have ever been reported so far. Exploring such possibility is not only important for a broader understanding of the fundamental nonlinear dynamics of complex systems, but it is likely to enable novel ultrafast sources for a wide range of applications.
In this article, we present our findings concerning a novel category of dissipative optical solitons, which we call dispersionless solitons as they arise from the interplay between dissipation and Kerr nonlinearity in the absence of chromatic dispersion. We base these findings on a wellknown physical system, suitable for both theoretical modeling and experimental testing: the optically driven Kerr cavity^{20,21,22,23}. Driving the Kerr cavity to new parameter regions, we uncover previously uncharted regimes in which ultraflat spectrum, transformlimited pulse generation and superior energy scaling could be simultaneously achieved. We show that these dispersionless solitons are closely related to the eigenfunctions of combined selfphase modulation (SPM) and spectral filtering effects, which asymptotically evolve toward an energyconserved pulsed solution with the increase of the filter order. Such remarkable trend thus reveals the existence of Kerronly conservative solitary waves as limiting cases of the dispersionless dissipative solitons when the spectral filtering becomes an ideal bandpass transmission. By construct, the asymptotic solitary wave shares the properties of an ideal Nyquist pulse, featuring a fully confined ultraflat spectrum and a transformlimited temporal waveform^{24,25,26}. Owing to their flat and compact spectral support, Nyquist pulses would be instrumental in enabling Kerr frequency combs with unprecedentedly high spectral efficiency, thus benefitting a wide range of applications such as highcapacity telecommunications^{27}, spectroscopy^{28}, lidar ranging^{29}, optical neural computing^{30,31}, and radiofrequency photonics^{32}.
Results
Principle to generate dispersionless cavity solitons
To search the dispersionless soliton solutions that form when the average group velocity dispersion becomes negligible, we use the following normalized distributed propagation model (see the Supplementary Section 1.1 for its derivation):
where \(U\)is the field envelope of the optical pulse; \(Z\) and \(T\) are the scaled spatial and temporal coordinates; \(\Delta\) is the pumpcavity phase detuning; \(C\) is the homogenous background of the pumped mode in the cavity; \(n\) is the filter order. Here, the normalized filter bandwidth out of which the spectral loss exceeds the uniform loss is 1. With the filter order \(n \to \infty\), the spectrum of \(U\) becomes fully confined in a unit frequency span. The three terms enclosed in square brackets represent the parametric mixing between the soliton and the background pump. The last term is an additional phase shift induced by crossphase modulation that is usually negligible in comparison to \(\Delta\). To gain more insight into the dispersionless soliton dynamics and how the balance can be achieved between dissipation and Kerr nonlinearity, the following eigenvalue problem is investigated:
where \(U_{{{\mathrm{e}}}}\) is the eigenfunction of combined SPM and spectral filtering; \(\xi\) and \(\lambda\) are the imaginary and real parts of the eigenvalue respectively. It turns out that \(\lambda\) is always negative: this reflects an overall amplitude loss induced by spectral filtering, whose impact also depends on SPM since the latter is conducive to the broadening of the optical spectrum. Apparently, these eigenfunction solutions are not stationary solutions of the propagation Eq. (1) but represent pulses damped with a decay rate contribution from spectral filtering and SPM that is given by \(\left \lambda \right\). However, when the pulses are coherently pumped, the energy loss can be compensated by the parametric gain, so that a composite balance can be achieved between all these propagation effects, resulting in the formation of cavity solitons. In the schematic representation of such composite balance by counteracting arrows in Fig. 1a, the vertical dimension represents purely dissipative effects and the horizontal one, purely dispersive contributions. Therefore, the tilt of the parametric gain arrow reflects its frequency dependence, which contributes to the balancing of SPM in absence of chromatic dispersion. More remarkably, we find that for a given value of \(\xi\), the energy loss induced by spectral filtering decreases and tends asymptotically to zero with the increase of filter order \(n \to \infty\) (see Supplementary Section 1.2). Therefore, in the limit when the filter order tends to infinity, spectral filtering and SPM can balance each other, leading to an eigenfunction that becomes a Kerronly solitary wave in an ideal bandpass system. In a realistic system, though, we should include linear losses, which can be precisely compensated by the parametric gain.
Figure 1b displays the numerically solved results of Eqs. (1) and (2) with \(\xi = \Delta = 20\), at different filter orders n. When the filter order increases, the coherently driven pulse described by Eq. (1) gets increasingly closer to the asymptotic soliton described by Eq. (2). Moreover, the soliton spectrum becomes more confined, and the pulse chirp decreases. When the filter order \(n \to \infty\), the soliton spectrum becomes fully confined within a unit bandwidth, featuring an intensity variation limited here to 4.2 dB. We also see that the asymptotic soliton adopts a uniform spectral phase, which is perfectly consistent with the fact that, as the losses vanish, the soliton behaves as a stationary conservative soliton. Correspondingly, the soliton temporal waveform becomes a transformlimited Nyquist pulse with characteristic oscillating tails. An approximate closedform solution can be derived for the Nyquist soliton with an ansatz method. A perturbation analysis of the governing averaged propagation model is performed in the Supplementary Section 1.3, showing that such eigenfunction is a stable attractor for coherently driven Kerr cavities with ideal bandpass transmission and negligible chromatic dispersion.
Observation of dispersionless cavity solitons
To study the soliton dynamics, a fiber ring cavity which share the same physical model as miniature microcavities is built. The experimental setup is shown in Fig. 2a. A spectral shaper is inserted for programmable dispersion and filter control^{17,33,34,35}, and a short length of erbiumdoped fiber (EDF) is used to compensate for the roundtrip loss as in ref. ^{36}. We note that the incorporation of active EDF here is not intended to build a fiber laser. The EDF length, its optical pumping power, and the pumping strength of the ring cavity are carefully tailored such that the gain provided by the optical amplifier keeps the ring cavity below laser threshold, and that the optical amplifier operates in the linear regime. Figure 2b shows the throughport transmission of the cavity. The total roundtrip length is 54.4 m and the power loss is 16.6%, corresponding to a free spectral range (FSR) of 3.73 MHz and a finesse of 36. Various spectral filtering functions can be implemented by programming the amplitude of the spectral shaper, while the fiber group velocity dispersion is carefully compensated by programming the phase. Figure 2c shows an example of a 250GHz gate bandpass filter. The spectral shaper has a limited optical resolution of 10 GHz, correspondingly, a maximum filter order of 30 can be achieved.
The roundtrip fiber length is stabilized with respect to the pump laser frequency by using a homebuilt feedback servo controller. The fiber ring is then synchronously pumped by a sequence of 87.4ps Gaussian pulses with a peak power of 1.5 W through a 90/10 coupler^{37}. The generated soliton is extracted from the 5% output of a 95/5 coupler for ultrafast characterization. To measure the soliton waveform, a highresolution intensity crosscorrelation setup is built, with the sampling pulses (0.2 ps) generated by spectrally broadening and compressing the pump pulses (more details presented in Supplementary Section 2). Figure 2d, e show the results of the dispersionless cavity solitons observed when the pump is tuned into the cavity resonance from the blue side (i.e., pump frequency higher than the resonant frequency). The spectral shaper is programmed to apply superGaussian filters with orders \(n = 2\), 4, 10 and 30, respectively. The filter bandwidth is 250 GHz. An excellent agreement between the experiment and the simulation is obtained. With the increase of the filter order, the soliton spectrum becomes rectangular, and the soliton temporal waveform gets closer to a Nyquist pulse. With the filter order \(n = 2\), the measured soliton pulse full width at half maximum (FWHM) is about 5.5 ps while the transformlimited pulse duration assuming a uniform spectral phase would be 3.3 ps, entailing a pulse chirp parameter close to 1. In comparison, the soliton for \(n = 30\) is nearly chirpfree with a FWHM of 3.7 ps and becomes close to a Nyquist pulse. For such Nyquistpulse soliton, about 99.6% of the total spectral power is confined within the 250GHz frequency window. The latter contains more than 66,000 comb lines within a 6dB intensity range excluding the pump, corresponding to an unprecedentedly high spectral quality for coherently driven Kerr frequency combs^{6}.
Nyquist soliton transition dynamics and its existence domain
Using the maximum filter order \(n = 30\), we find that the Nyquist soliton can be generated deterministically under pulsed pumping. Figure 3a shows the mean intracavity power trace when the pump laser scans across the resonance from the blue side. Steps corresponding to soliton state transitions can be clearly observed. In contrast to conventional dispersionenabled cavity solitons which are usually excited stochastically through chaotic modulational instability^{23}, the Nyquist soliton generation is rather deterministic with an absence of noisy region observed from the intracavity power trace. The evolution of the soliton waveform with the transition step is visualized in Fig. 3b, with some selected states and their spectra detailed in Fig. 3c (see Supplementary Section 2.3 for the full data). All the solitary structures can be stably maintained when the scanning laser stops at the corresponding steps. The solitons appear as square pulses with characteristic oscillating peaks and spectral sidelobes. The pulse duration is quantized and has a very good linear relation with the power step number (Fig. 3d). After each transition step, the pulse width decreases by about 8.1 ps. Numerical simulations reveal that these square pulses are associated with compact Nyquist soliton molecules (i.e., closely bound solitons^{5,38}) which can also be sustained in ideal bandpass systems by the balance between Kerr nonlinearity and spectral filtering alone (Supplementary Section 1.4). With the increase of pump frequency detuning, the soliton number in the molecule decreases one by one and a single Nyquist soliton is ultimately formed. It is found that the soliton transition process can be affected by the desynchronization between the pump pulse repetition rate and the cavity FSR (Supplementary Section 2.4). In some cases, the existence region of some intermediate states may be too narrow to be observed and the soliton number decreases by two or more in a single transition step.
To explore the existence region of the single Nyquist soliton, a parameter scanning is performed numerically based on the normalized continuously driven equation. The results are shown in Fig. 4a. The Nyquist soliton can be maintained in a relatively wide parameter space that lies within the homogenously bistable region. Figure 4b, c illustrate the evolution of the soliton pulse shape with the pump phase detuning and the pump power, respectively. The soliton spectra at some selected locations marked in Fig. 4a are plotted in Figs. 4d and 4e. Along the lower boundary of the pump power (the minimum pump power required to maintain the soliton), the soliton intensity increases linearly with the increase of pump detuning. Meanwhile, the pulse width keeps nearly unchanged since the spectral bandwidth remains fully confined by the gate filter in the frequency domain. Detailed plots of the pulse energy and width are shown in Fig. 4f, g. The energy growing slope is about 1.5 and the pulse width is 0.94. This indicates that the ideal Nyquist soliton is free from energywidth scaling, in sharp contrast to the conventional NLSE solitons enabled by the secondorder dispersion for which the energy scales with the inverse of the pulse duration. Therefore, the Nyquist soliton can, in theory, be promoted indefinitely while keeping almost the same pulse and spectral shape (more discussions about the scaling law of dispersionless solitons are presented in Supplementary Section 1.2.2). When the pump detuning is kept constant, the soliton energy also increases slowly with the pump power, with an average slope of 0.035. A plot of the overlaid soliton pulses under different pump power levels is shown in Supplementary Section 2.5. As the pump power increases, the main pulse width keeps nearly unchanged while the oscillating tails get more prominent. In the frequency domain, the spectral components close to the passband edge get more enhanced, giving rise to an increasingly flat spectrum. Along the dashdotted line in Fig. 4a for which the pump phase detuning is kept at 20, the best achievable spectral flatness is about 2.3 dB when the pump power varies between the lower and upper boundaries. The change of the soliton pulse shape under increasing pump power levels can be understood intuitively as follows. When the pump power is high, for a pulse shape primarily determined by the balance between spectral filtering and SPM (i.e., Kerrandfilter eigenfunctions following Eq. (2)), the parametric gain will be obviously larger than that required for compensating for the uniform loss. The soliton shape thus changes adaptively such that a new composite balance is achieved between the physical effects involved. A schematic illustration is shown in Supplementary Fig. S10b. The extended parametric gain arrow and its tilt indicate the difference in the balanced state compared to when the pump power is low. Due to the difficulty of expressing the distorted soliton pulse in closed form, the analytical description of this phenomenon is a challenging task that we leave for future work.
Relation between filterdriven and dispersiondriven solitons
When the net group velocity dispersion is not zero in the fiber loop, both the dispersion and the spectral filter will be responsible for soliton pulse shaping. Figure 5 shows the results when the spectral shaper is programmed to a 250GHz gate filter and different amounts of secondorder dispersion are applied. The soliton spectrum becomes slightly flatter when the net roundtrip dispersion is slightly normal (0.63 ps^{2}). However, the soliton existence region becomes narrower, and it is more difficult to get stable single Nyquist soliton when the dispersion is further increased. When the net dispersion is anomalous, the spectral intensity around the pump wavelength gets more pronounced. Two cases with net dispersion values of −1.06 and −4.22 ps^{2} are shown. With an increase of the net anomalous dispersion, the soliton tends to adopt hyperbolicsecant spectrum that is typical from conventional dispersiondriven solitons. The oscillating tails aside the main pulse, which are a signature of a Nyquist pulse, get diminished. A chaotic region with higher noise can also be observed from the intracavity power transition trace, which is attributed to the modulational instability in the anomalous dispersion region^{23}. For all the three cases shown in Fig. 5b, the solitons are very close to transformlimited pulses with no obvious chirp.
Discussion
In summary, this article demonstrates the existence of dispersionless cavity solitons sustained by the balance between dissipation and Kerr nonlinearity. In addition to deepening the dissipative soliton paradigm, our study reveals an unexpected connection between dissipative and conservative cavity solitons by virtue of Nyquist pulses. We have developed a theoretical model relating the solitons to the eigenfunctions of SPM and spectral filtering effects. When the filter order tends to infinity, a Nyquistpulselike soliton is demonstrated. Such Nyquist soliton possesses several outstanding merits, including a fully confined flat spectrum, a transformlimited pulse shape, deterministic generation under pulsed pumping, and immunity from energywidth scaling.
We note that the dispersionless solitons presented here are distinct from the “zerodispersion” solitons that have been recently demonstrated in driven passive cavities^{14,15}, photonic crystal waveguides^{16} and modelocked lasers^{17} when the average secondorder dispersion is vanishing. Indeed, the formation of solitons in refs. ^{14,15,16,17} is supported by the interaction involving the higherorder group velocity dispersion terms along with the Kerr nonlinearity (thus is still dispersiondriven). By “dispersionless” in this article, we emphasize that the soliton is primarily stabilized by the balance between SPM and spectral filtering. The participation of any order of dispersion is not a necessitate. Even when there is a weak dispersion, it will just act as a higherorder perturbation. Such a soliton formation mechanism has never been revealed in coherently driven optical cavities, to the best of our knowledge. Moreover, whereas our experimental fiber ring is constructed from discrete elements with alternating normal and anomalous dispersion, the soliton dynamics is fundamentally different from the stretchedpulse modelocking^{10,11}. Such statement is supported by a comparison between the soliton spectra measured before and after the spectral shaper in our setup, see the Supplementary Section 2.6, showing no significant difference. This is explained by the fact that the local dispersion length is always much larger than the cavity length for the pulse duration involved, justifying the prevalence of the pathaveraged cavity dynamics.
We have experimentally generated dispersionless solitons having a pulse width in the range of a few picoseconds, which is typical of coherently driven temporal solitons obtained within fiber resonators^{22,36}. For some applications, shorter pulses in the femtosecond range featuring a larger frequency comb bandwidth would be required. Although the spectral shaper we employ is capable of compensating for the fiber loop dispersion over a large bandwidth^{39}, stable femtosecond soliton generation becomes very challenging in our experiments due to the increased requirement of a precise synchronization between the pump pulse and the cavity soliton. Furthermore, when the local dispersion length of femtosecond solitons becomes comparable to or shorter than the fiber length of each section, the soliton evolution departs from the dispersionless case and eventually transition to the conventional stretchedpulse modelocking regime. It is therefore quite impractical to generate femtosecond dispersionless solitons with the present setup. The purpose of our work is to demonstrate the concept of dispersionless dissipative solitons while benefitting from the versatile fiber ring cavity architecture, which can be subsequently scaled to femtosecond broadband solitons by using other photonic platforms. For instance, a fiberbased viable solution can be envisaged by using miniature Fabry–Pérot fiber resonators fabricated with fiber end polishing and highreflection coatings^{37}. By choosing a proper fiber with closetozero dispersion at the pump wavelength and with bandpass coating, a spectrally confined fiber resonator with nearly vanishing dispersion can be obtained to support dispersionless solitons. Another very promising implementation for practical applications is to generate femtosecond dispersionless solitons within onchip integrated microresonators. Flexible dispersion engineering and filter control can be achieved by tailoring the waveguide dimension and employing photonic crystal structures^{40,41}. The microring and Fabry–Pérot conceptual designs are illustrated in the Supplementary Section 2.7. Numerical simulations are performed by considering the typical parameters of the silicon nitride platform: they show that femtosecond dispersionless solitons having a flat comb spectrum spanning over 100 nm are accessible. Flat microcombs are very attractive for various applications such as highspeed communications, lidar ranging and optical computing.
Finally, in our dispersionless soliton experiments, we do not observe the Gordon–Kelly resonant sidebands that cause instabilities of highenergy pulses in dispersiondriven soliton modelocked fiber lasers^{17,42}. Since the group velocity dispersion is absent, the formation mechanism of Gordon–Kelly sidebands, namely constructive interference between the soliton and the dispersive wave, is no longer applicable. This implies an improved robustness of the dispersionless solitons compared to the conventional dispersiondriven solitons. The discovery of dispersionless dissipative solitons will be instrumental in the development of novel ultrafast lasers and frequency combs with previously unattainable spectral and temporal features.
Materials and methods
Soliton generation and characterization
The fiber ring cavity is mainly composed of a 21m highly nonlinear fiber (YOFC NL1550NEG) and a ~30m standard singlemode fiber. A length of 0.6m erbiumdoped fiber (EDF, YOFC EDF13) is used to compensate for the loss. The EDF is pumped by a 1.8W 976nm laser, offering a gain of ~4.1 dB. The total cavity length is about 54.4 m which is typical for coherently driven fiber solitons^{22,36}. The long fiber cavity facilitates the experiments by boosting the Kerr nonlinearity per roundtrip and reducing the required pump power level.
A commercial spectral shaper (Finisar WaveShaper 1000A) is inserted for programmable dispersion and filter control. Its dispersion trimming capability is characterized by a dispersionbandwidth product of 80 ps which is sufficient for our experiments^{39}. To perform an accurate dispersion compensation, a transformlimited pulse train (FWHM: ~0.5 ps) from a modelocked fiber laser is sent through the fiber link when the loop is open. The output pulse shape is monitored through intensity autocorrelation. The phase of the spectral shaper is then adjusted iteratively to make the output pulse transformlimited again. The cavity roundtrip transfer function after dispersion compensation is further checked with a commercial optical vector network analyzer (Luna) and the results are shown in Fig. 2c. The pulsed pump is generated by modulating a continuouswave narrowlinewidth laser (OEwaves) with the signals from a pulse pattern generator (Anritsu), and is amplified to a peak power of 1.5 W for soliton sustainment. The amplified spontaneous emission noise is rejected by a narrowband filter. To generate solitons, the pump laser frequency is manually tuned into the cavity resonance from the higherfrequency side. In the meanwhile, the relative detuning is stabilized by a homebuilt feedback servo controller to mitigate environmental interference. The soliton spectrum is measured by a commercial spectrum analyzer (Yokogawa) with a resolution of 0.02 nm; and the waveform is measured by a homebuilt intensity cross correlator having a resolution of 0.2 ps.
Theoretical model and numerical simulations
The results in Fig. 1b are obtained by numerically finding the solutions of Eq. (1) (with the left side set to zero) and (2) with the Newton–Rapson method. The phase shifting parameters are \(\xi = \Delta = 20\). Note that Eq. (1) is identical to the following equation
where \(\psi \left( {Z,T} \right) = U\left( {Z,T} \right) + C\) the total intracavity field; and \(S\) is the pump field. In each case, the pump power \(\left S \right^2\)is adjusted near its minimum value that is required to maintain the stable soliton. For the filter order \(n = 2\), 4, 10 and \(\infty\), the pump power \(\left S \right^2 = 420\), 280, 120 and 25, respectively. The homogenous background \(C\) is calculated by finding the lowerbranch solution of Eq. (3) with the derivative terms set to zero.
When solving the eigenfunctions of Eq. (2), the real part of the eigenvalue (i.e., \(\lambda\)) can also be obtained at the same time as an unknown variable that depends on \(\xi\).
The parameter scanning results in Fig. 4 are obtained by numerically integrating Eq. (3) with the standard splitstep Fourier method until the intracavity field reaches a stable state. The cavity roundtrip time is 100. Note that the plots of Figs. 4d and 4e include both the soliton and the homogenous pump background, different from Fig. 1b which shows only the soliton.
The simulation results in Figs. 2, 3 and 5 for the fiber cavity are obtained with the splitstep Fourier method, based on a simplified model in which the effects of Kerr nonlinearity, dispersion and loss are considered separately by lumped terms in one roundtrip. The field at the end of the (m + 1)th round is related to that after the mth round by
where \(A\) is the field amplitude normalized such that \(P = \left A \right^2\) represents the field power; \(L\) is the roundtrip length; \(\alpha _L\) is the universal roundtrip power loss for all the modes; \(\delta _L\) is the pumpcavity phase detuning; \(\gamma\) is the average Kerr coefficient; \(\theta\) is the pumpcavity power coupling ratio; and \(A_{{{\mathrm{p}}}}\) is the pump field. The operator \(\widehat F\) accounts for the effects that can be easily applied in the frequency domain, including the filtering loss, the group velocity dispersion and the pumpcavity desynchronization (see Eq. (S28) in Supplementary Section 1.5 for more details). The simulation parameters are \(\theta = 10.4\%\), \(\alpha _L = 16.6\%\), \(\gamma = 4.6 \times 10^{  3}{{{\mathrm{ m}}}}^{  1}{{{\mathrm{W}}}}^{  1}\), and \(L = 54.6{{{\mathrm{ m}}}}\). The pump pulse is Gaussian with a FWHM of 87.4 ps and a peak power of 1.5 W. The power transfer function of the superGaussian filter implemented by the spectral shaper is given by
where \(\nu _0\), \(B\), and \(n\) represent the filter central frequency, bandwidth, and order, respectively.
References
Russell, S. J. Report on Waves: Made to the Meetings of the British Association in 1842–1843 (Forgotten Books, 1845).
Boardman, A. D. & Sukhorukov, A. P. SolitonDriven Photonics (Springer, 2001).
Akhmediev, N., Ankiewicz, A. & Akhmediev, N. Dissipative Solitons: From Optics to Biology and Medicine (Springer, 2008).
Haus, H. A. & Wong, W. S. Solitons in optical communications. Rev. Mod. Phys. 68, 423–444 (1996).
Grelu, P. & Akhmediev, N. Dissipative solitons for modelocked lasers. Nat. Photonics 6, 84–92 (2012).
Kippenberg, T. J. et al. Dissipative Kerr solitons in optical microresonators. Science 361, eaan8083 (2018).
Chong, A. et al. Allnormaldispersion femtosecond fiber laser. Opt. Express 14, 10095–10100 (2006).
Huang, S. W. et al. Modelocked ultrashort pulse generation from onchip normal dispersion microresonators. Phys. Rev. Lett. 114, 053901 (2015).
Spiess, C. et al. Chirped dissipative solitons in driven optical resonators. Optica 8, 861–869 (2021).
Haus, H. A. et al. Stretchedpulse additive pulse modelocking in fiber ring lasers: theory and experiment. IEEE J. Quantum Electron. 31, 591–598 (1995).
Dong, X. et al. Stretchedpulse soliton Kerr resonators. Phys. Rev. Lett. 125, 033902 (2020).
Renninger, W. H., Chong, A. & Wise, F. W. Amplifier similaritons in a dispersionmapped fiber laser. Opt. Express 19, 22496–22501 (2011).
ParraRivas, P., Gomila, D. & Gelens, L. Coexistence of stable dark and brightsoliton Kerr combs in normaldispersion resonators. Phys. Rev. A 95, 053863 (2017).
Li, Z. D. et al. Experimental observations of bright dissipative cavity solitons and their collapsed snaking in a Kerr resonator with normal dispersion driving. Optica 7, 1195–1203 (2020).
Anderson, M. H. et al. Zerodispersion Kerr solitons in optical microresonators. Nat. Commun. 13, 4764 (2020).
BlancoRedondo, A. et al. Purequartic solitons. Nat. Commun. 7, 10427 (2016).
Runge, A. F. J. et al. The purequartic soliton laser. Nat. Photonics 14, 492–497 (2020).
Turitsyn, S. K. et al. Dissipative solitons in fiber lasers. Phys. Uspekhi 59, 642–668 (2016).
Grelu, P. in Handbook of Laser Technology and Applications (Laser Design & Laser Systems) 2nd edn, Vol. II (eds Guo, C. L. & Singh, S. C.) Ch. 39 (Taylor & Francis, 2021).
Haelterman, M., Trillo, S. & Wabnitz, S. Dissipative modulation instability in a nonlinear dispersive ring cavity. Opt. Commun. 91, 401–407 (1992).
Wabnitz, S. Suppression of interactions in a phaselocked soliton optical memory. Opt. Lett. 18, 601–603 (1993).
Leo, F. et al. Temporal cavity solitons in onedimensional Kerr media as bits in an alloptical buffer. Nat. Photonics 4, 471–476 (2010).
Herr, T. et al. Temporal solitons in optical microresonators. Nat. Photonics 8, 145–152 (2014).
Nakazawa, M., Yoshida, M. & Hirooka, T. The Nyquist laser. Optica 1, 15–22 (2014).
Turitsyn, S. K., Bogdanov, S. & Redyuk, A. Solitonsinc optical pulses. Opt. Lett. 45, 5352–5355 (2020).
Nakazawa, M. et al. Ultrahighspeed ‘orthogonal’ TDM transmission with an optical Nyquist pulse train. Opt. Express 20, 1129–1140 (2012).
MarinPalomo, P. et al. Microresonatorbased solitons for massively parallel coherent optical communications. Nature 546, 274–279 (2017).
Suh, M. G. et al. Microresonator soliton dualcomb spectroscopy. Science 354, 600–603 (2016).
Riemensberger, J. et al. Massively parallel coherent laser ranging using a soliton microcomb. Nature 581, 164–170 (2020).
Feldmann, J. et al. Parallel convolutional processing using an integrated photonic tensor core. Nature 589, 52–58 (2021).
Xu, X. Y. et al. 11 TOPS photonic convolutional accelerator for optical neural networks. Nature 589, 44–51 (2021).
Wu, J. Y. et al. RF photonics: an optical microcombs’ perspective. IEEE J. Sel. Top. Quantum Electron. 24, 6101020 (2018).
Weiner, A. M. Femtosecond pulse shaping using spatial light modulators. Rev. Sci. Instrum. 71, 1929–1960 (2000).
Schröder, J. et al. Dark and bright pulse passive modelocked laser with incavity pulseshaper. Opt. Express 18, 22715–22721 (2010).
Peng, J. S. & Boscolo, S. Filterbased dispersionmanaged versatile ultrafast fibre laser. Sci. Rep. 6, 25995 (2016).
Englebert, N. et al. Temporal solitons in a coherently driven active resonator. Nat. Photonics 15, 536–541 (2021).
Obrzud, E., Lecomte, S. & Herr, T. Temporal solitons in microresonators driven by optical pulses. Nat. Photonics 11, 600–607 (2017).
Stratmann, M., Pagel, T. & Mitschke, F. Experimental observation of temporal soliton molecules. Phys. Rev. Lett. 95, 143902 (2005).
Application note: Dispersion trimming using the programming group delay capability of the Waveshaper family of programmable optical processors. https://iivi.com/dispersiontrimming (2022).
Yu, S. P. et al. Photoniccrystalreflector nanoresonators for Kerrfrequency combs. ACS Photonics 6, 2083–2089 (2019).
Lucas, E. et al. Inverse spectral design of Kerr microcomb pulses. In Proc. SPIE 11672, Laser Resonators, Microresonators, and Beam Control XXIII 1167205 (SPIE, 2021).
Kelly, S. M. J. Characteristic sideband instability of periodically amplified average soliton. Electron. Lett. 28, 806–807 (1992).
Acknowledgements
The authors thank A. M. Weiner for discussions. This work was supported by the National Key R&D Program of China under grant no. 2018YFA0701902, the National Natural Science Foundation of China under grant no. 61690192, and Zhejiang Lab under grant no. 2020LC0AD01. P.G. acknowledges support from the EiPhi Graduate School under grant no. ANR17EURE0004 and the French ISITEBFC programs under grant no. ANR15IDEX0003.
Author information
Authors and Affiliations
Contributions
X.X. conceived the idea and performed the experiments with help from B.Y., M.W., and S.L. X.X. and P.G. analyzed the results, performed the numerical simulations, and prepared the manuscript. All the authors were closely involved in discussions and revising the manuscript.
Corresponding authors
Ethics declarations
Conflict of interest
The authors declare no competing interests.
Supplementary information
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
Xue, X., Grelu, P., Yang, B. et al. Dispersionless Kerr solitons in spectrally confined optical cavities. Light Sci Appl 12, 19 (2023). https://doi.org/10.1038/s41377022010528
Received:
Revised:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41377022010528
This article is cited by

Ultrafast laser state active controlling based on anisotropic quasi1D material
Light: Science & Applications (2024)

Ultrasound sensing with optical microcavities
Light: Science & Applications (2024)