## Abstract

We present a closed-form analytical description of the early stages of spectral broadening of ultrashort laser pulses beyond the basic theory of self-phase modulation (SPM). In the limit of short propagation paths, approximate analytical expressions derived as a part of our treatment recover the canonical SPM-theory results for the nonlinear shift and spectral broadening. For longer propagation paths, these expressions shed light on how dispersion effects enter the scene, decelerating the spectral broadening in the regime of normal dispersion and giving rise to an explosion-like bandwidth growth in anomalous-dispersion high-soliton-number pulse evolution scenarios. Based on this formalism, we will provide an analytical derivation for the relation between the maximum soliton self-compression length and the soliton number, which has been previously treated as purely empirical.

© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement

## 1. Introduction

Spectral broadening of ultrashort pulses is one of the central effects in nonlinear optics [1]. While in linear optics, spectral broadening is forbidden by the second law of thermodynamics, nonlinear-optical systems and materials offer a unique resource for a spectral transformation of ultrashort optical pulses [2], enabling numerous pulse-compression schemes [3], supercontinuum-seeded optical parametric amplification [4], supercontinuum-assisted spectroscopy [5] and microscopy [6], as well as multioctave supercontinuum generation for ultrafast photonics [7, 8] and frequency-comb technologies [9].

Self-phase modulation (SPM) is known to dominate early stages of spectral broadening, seeding cascades of parametric processes, kicking off soliton dynamics, and providing a spectral content for subsequent pulse self-compression in the regime of anomalous dispersion [7, 8]. The basic theory of SPM [1, 2] neglects dispersion effects to derive an analytical result for the Kerr-effect-induced phase shift that is linear in the peak power, optical nonlinearity, and the propagation distance. This approximation provides a powerful tool for the analysis of both desirable and unwanted nonlinear phase shifts in a vast class of laser systems and parametric converters [3], helping to tailor the short-pulse output of such systems [10].

However, as the pulse propagates further on down an extended nonlinear medium, e.g., an optical fiber, dispersion effects inevitably come into the picture. Acting jointly with nonlinearity, these effects can give rise to a broad diversity of soliton phenomena [1, 11–13], enabling efficient supercontinuum generation [7, 8] and pulse self-compression [2], including compression to single-cycle and even subcycle pulse widths [14]. With dispersion completely out of picture in the canonical basic SPM theory, one faces an uncomfortable gap between the SPM and soliton theory accounts of pulse evolution. Indeed, in basic SPM theory, the nonlinear phase shift and, hence, the pulse bandwidth display an unbounded growth, increasing with the propagation path. In soliton dynamics, on the other hand, the spectral bandwidth of an optical pulse can become a nonmonotonous function of the propagation length, often undergoing signature, breather oscillations [2, 11], in which an explosion-like bandwidth growth is followed by a stage of spectral narrowing, translating into pulse-self-compression − pulse-stretching cycles in the time domain [15].

The main goal of this work is to close this gap by providing a closed-form analytical description of the early decisive stages of spectral broadening of an ultrashort laser pulse that would allow the basic SPM theory to be continuously extended to the soliton-number-based description of nonlinear pulse evolution. This analytical description will be shown to shed light on how dispersion effects enter the picture of pulse evolution, leading to a spatial-scale renormalization, decelerating the spectral broadening in the regime of normal dispersion, and giving rise to an explosion-like bandwidth growth in anomalous-dispersion high-soliton-number pulse evolution scenarios. Based on this formalism, we will provide an analytical derivation for the relation between the maximum soliton self-compression length and the soliton number, which has been previously treated as purely empirical.

## 2. Self-phase modulation: basic theory

As a starting point, we consider the nonlinear Schrödinger equation (NSE) [1, 2]

*A*is the pulse envelope,

*P*

_{0}is the input peak power,

*z*is the coordinate along the propagation path,

*η*=

*t*−

*z*/

*v*

_{g},

*t*is the time variable,

*v*

_{g}is the group velocity,

*β*

_{2}is the group-velocity dispersion coefficient, and

*γ*is the nonlinear coefficient.

With the dispersion term dropped, Eq. (1) is reduced to an elementary SPM-theory equation, with a solution given by [1, 2]

The *η*-dependent nonlinear phase shift [Eq. (3)] gives rise to a deviation of the instantaneous frequency from the central frequency of the input field, $\delta \omega \left(z,\eta \right)=\partial {\varphi}_{SPM}\left(z,\eta \right)/\partial \eta $, leading to a standard order-of-magnitude estimate for the SPM-induced spectral broadening [1, 2]

*τ*

_{0}being the input pulse width. Taking ${\left|u\left(0,\eta \right)\right|}^{2}\approx 1-{\left(\eta /{\tau}_{0}\right)}^{2}$ near the peak of the pulse, we find for the SPM-induced chirp

## 3. Soliton dynamics

The full NSE, with both nonlinearity and dispersion terms included in Eq. (1), allows an infinite number of soliton solutions [2, 11–13]. Within a vast parameter space and across a broad diversity of soliton phenomena, the soliton dynamics is controlled by the soliton number *N*, defined as *N* = (*l*_{d}/*l*_{nl}), where ${l}_{d}={\tau}_{0}^{2}/\left|{\beta}_{2}\right|$ is the dispersion length and *l*_{nl} = 1/(*γP*_{0}) is the nonlinear length. However, only a few of these solutions can be written in a closed analytical form. Among those, the fundamental, *N* = 1, soliton solution [1, 2, 11], $u\left(\xi ,\theta \right)=\mathrm{sec}\text{h}\left(\theta \right)\mathrm{exp}\left(i\xi /2\right)$, with *θ* = *t*/*τ*_{0} and *ξ* = *z*/*l*_{d}, plays a very special role, as such a soliton pulse propagates in a lossless medium without changing its envelope and dominates over other features in spectral and temporal dynamics as long as *N* < 2 [2, 11].

The *N* = 2 soliton is a rare example of a closed-form analytical solution for an *N* > 1 soliton (see, e.g., Refs. 2, 11). This solution exhibits a periodic behavior of |*u*|^{2} as a function of *ξ*. This oscillatory behavior, observed both in the spectral (Fig. 1) and temporal (Fig. 2) domain, is one of the key properties of higher order optical solitons. However, except for a few rare examples, the dynamics of *N* > 1 solitons has to be studied with numerical methods. As an example of such numerical analysis, Figs. 1(a) and 1(b) present the maps of temporal evolution of NSE solitons with *N* = 5 and *N* = 7, respectively. At the initial stage of their oscillatory dynamics, the solitons undergo pulse self-compression, reaching a minimum pulse width *τ*_{m} at *ξ* = *ξ*_{m}. This self-compression stage is followed by pulse stretching, revealing a multisoliton structure of the waveform [Figs. 1(a), 1(b)]. For ideal NSE solitons, such cycles recur with an ideal periodicity and a period *ξ*_{s} = *π*/2, or, with the physical units of length restored, *z*_{s} = *πl*_{d}/2. High-order dispersion leads to a fission of this bound high-*N* soliton solution, giving rise to efficient supercontinuum generation [7, 8, 16]. Individual soliton features emerging from this soliton-fission dynamics can still be coherently combined to yield a high-energy supercontinuum output [17, 18].

Numerical analysis of high-*N* soliton dynamics, supported by a wealth of experimental data, suggests purely empirical, yet practically useful design rules for devices implementing soliton pulse self-compression. As perhaps the most significant design rule, the optimal pulse compression distance *z*_{m} = *ξ*_{m}*l*_{d}, that is, the distance at which the soliton pulse width reaches its minimum *τ*_{m}, has been found to obey the following scaling with the soliton number *N* [2]:

We will show below that this important relation, viewed in the earlier work as purely empirical [2], can, in fact, be derived analytically through an appropriate extension of the basic SPM theory.

## 4. Dispersion and scale renormalization

To extend the analytical description of the early stages of spectral broadening of ultrashort laser pulses beyond the simplest equations of the SPM theory [Eqs. (2)–(5)], we first resort to the solution of Eq. (1) without the nonlinear term. For an input pulse with an initial chirp *α*_{0}, ${\left|u\left(0,\eta \right)\right|}^{2}=\mathrm{exp}\left[-\left({\tau}_{0}^{-2}+i{\alpha}_{0}\right){\eta}^{2}\right]$, the solution to this equation is written as

Seeking an approximate description of spectral broadening beyond the basic SPM theory in the high-*N* regime, we note that, since *N*^{2} >> 1 is equivalent to *l*_{nl} < *l*_{d}, the nonlinear phase shift in the case of high *N* is accumulated much faster than the dispersion-induced phase shift. We use this observation as a rationale, which in no way qualifies as a rigorous justification, for the separation of *l*_{nl} and *l*_{d} scales and plug an ansatz *α*_{0} = *α*_{SPM}, with *α*_{SPM} as defined by Eq. (5), into Eq. (9) to find

Equation (10) describes pulse self-compression in the case of anomalous dispersion, sgn(*β*_{2}) = −1, and pulse stretching when sgn(*β*_{2}) = + 1, i.e., dispersion is normal.

We restrict our analysis to propagation distances *z* < (*l*_{d}*l*_{nl})^{1/2} = *l*_{b}. Notably, this restriction does not prevent us from analyzing soliton dynamics up to the point of maximum pulse compression, as, in accordance with Eq. (6), *z*_{m} ≈*l*_{b}/2. Then, dropping the small ${\left(z/{l}_{d}\right)}^{2}$ and ${z}^{4}/{\left({l}_{d}{l}_{nl}\right)}^{2}$ terms in Eq. (10), we find from the resulting expression for *τ*(*z*) that the minimum pulse width (zero *τ* in our approximation) is achieved, in the case of anomalous dispersion, at

This is a rewarding result. Indeed, the empirical design rule of Eq. (6) can be rewritten, for a convenience of comparison, as *z*_{m} ≈1.0048*l*_{d}/(2*N*), which agrees with Eq. (11) within 0.5%.

We now modify Eq. (4) for the SPM-induced spectral broadening to reflect variations in the pulse width *τ* and, hence, the peak power *P* as functions of *z*. To this end, we replace *τ*_{0} in Eq. (4) by *τ*(*z*) as defined by Eq. (10), and express the *z*-dependent peak power as *P*(*z*) = *P*_{0}/[*τ*_{0}*ψ*(*z*)], with *ψ*(*z*) as defined by Eq. (8). Then, within the same approximation as in the analysis above, we arrive at

Integration in Eq. (12) leads to

*β*

_{2}) = −1, and

*β*

_{2}) = + 1.

Equations (13) and (14) provide an approximate description of spectral broadening beyond the basic SPM theory in the high-*N* regime. As one important finding, these equations reveal the significance of the *l*_{c} length as the key spatial scale of the problem not only in the regime of soliton self-compression, when *l*_{c} defines the length of maximum pulse compression (Figs. 1a, 1b), but also in the case of normal dispersion. Indeed, for propagation paths *z* << *l*_{c}, $\mathrm{arcsin}\left(z/{l}_{c}\right)\approx z/{l}_{c}$ and $\mathrm{arctan}\left(z/{l}_{c}\right)\approx z/{l}_{c}$. In this limit, both Eq. (13) and Eq. (14) recover the celebrated result of the basic SPM theory [Eq. (4)], $\Delta \omega \left(z\right)\approx \gamma {P}_{0}z/{\tau}_{0}=\Delta {\omega}_{SPM}\left(z\right)$.

## 5. Discussion

In Figs. 2(a)–2(e), we compare predictions of Eqs. (13) and (14) with the numerical solution of the NSE [Eq. (1)]. To this end, we plot the normalized spectral bandwidth $\left[\Delta {\omega}_{0}+\Delta \omega \left(z\right)\right]/\Delta {\omega}_{0}$, Δ*ω*_{0} being the spectral bandwidth of the input pulse, for two different values of *N* in the regime of normal [Figs. 2(a), 2(b)] and anomalous [Figs. 2(c)–2(f)] dispersion. For both positive and negative *β*_{2}, our approximation provides a reasonably accurate description of SPM-induced spectral broadening within the entire applicability range of Eqs. (13) and (14), i.e., all the way up to *z* ≈*l*_{c}. When the propagation path is no longer much smaller than *l*_{c}, Eqs. (13) and (14) are instructive in showing how dispersion effects enter the scene, leading to spatial-scale renormalization. Indeed, for *z* << *l*_{c}, spectral broadening for both positive and negative *β*_{2} is given by Eq. (4), with the nonlinear phase shift accumulating as *z*/*l*_{nl}, articulating the significance of the nonlinear length *l*_{nl} as the spatial scale of the process. For longer propagation paths, however, the scenario of spectral broadening changes. As *z* ceases to be small compared to *l*_{c}, *l*_{nl} no longer provides a spatial scale that fully defines spectral broadening. Instead, as dispersion effects come into play, *l*_{nl} has to share its role as the spatial scale with the dispersion length *l*_{d}, through a geometric mean (*l*_{nl}*l*_{nl})^{1/2}, defining a new spatial scale *l*_{c}. This *l*_{nl}-to-*l*_{c} scale renormalization is analytically described by $\mathrm{arcsin}\left(z/{l}_{c}\right)$ in the anomalous dispersion regime [Eq. (13)] and $\mathrm{arctan}\left(z/{l}_{c}\right)$ in the case of normal dispersion [Eq. (14)], both providing a smooth transition from the *z* << *l*_{c} regime, where the *l*_{c} scale is completely eliminated, to the *z* ~*l*_{c} regime, where *l*_{nl} is suppressed as an isolated factor in Eqs. (13) and (14), entering the problem only through the *l*_{c} scale instead.

Unlike the basic SPM theory, whose predictions for the nonlinear phase shift [Eq. (3)] and the related spectral broadening [Eq. (4)] are insensitive to the sign of dispersion, Eqs. (13) and (14) provide a closed-form analytical description of how the difference between spectral broadening in the normal and anomalous dispersion regimes builds up as the pulse propagates along a nonlinear, dispersive medium. As *z* ceases to be much smaller than *l*_{c}, the difference becomes striking. For positive *β*_{2}, normal dispersion results in pulse stretching, decelerating spectral broadening. This effect, which is grasped by Eq. (14) and is clearly seen in Figs. 2(a) and 2(b), manifests itself in a broad class of nonlinear fiber-optic experiments [7, 8], including recent studies on the spectral broadening of high-energy short laser pulses in normally-dispersive large-mode-area photonic-crystal fibers (PCFs) [19, 20].

In the case of negative *β*_{2}, anomalous dispersion, acting jointly with nonlinearity, gives rise to soliton pulse self-compression [Figs. 1(a), 1(b)]. In the frequency domain, this translates into enhanced spectral broadening [Figs. 2(c)–2(f)], giving rise to an explosion-like growth in Δ*ω* near the point of maximum pulse compression, *z* ~*l*_{c}. As can be seen in Figs. 2(c)–2(f), our approximate analytical result of Eq. (13) agrees reasonably well with numerical simulations, thus offering a helpful analytical tool for the design of fiber-optic pulse compressors [7, 8], including PCF-based sources of single-cycle and few-cycle field waveforms [14, 21].

## 6. Conclusion

To summarize, we presented a closed-form analytical description of the early stages of spectral broadening of ultrashort laser pulses beyond the basic SPM theory. This formalism closes the gap between the basic SPM theory and the soliton-number-based description of nonlinear pulse evolution. Based on this formalism, we have provided an analytical derivation for the relation between the maximum soliton self-compression length and *N*, which has been previously treated as purely empirical.

## Funding

Russian Foundation for Basic Research (16-02-00843, 17-52-53092); Russian Science Foundation (17-12-01533); Welch Foundation (A-1801-20180324), ONR (00014-16-1-2578), Government of Russian Federation (project no. 14.Z50.31.0040, Feb. 17, 2017).

## Acknowledgments

Useful discussions with Alexander Voronin are gratefully acknowledged.

## References and links

**1. **Y. R. Shen, *The Principles of Nonlinear Optics* (New York, Wiley-Interscience, 1984).

**2. **G. P. Agrawal, *Nonlinear Fiber Optics* (Academic, San Diego, 2001).

**3. **F. X. Kärtner, ed., *Few-Cycle Laser Pulse Generation and Its Applications* (Springer, Berlin, 2004).

**4. **C. Manzoni, O. D. Mücke, G. Cirmi, S. Fang, J. Moses, S. W. Huang, K. H. Hong, G. Cerullo, and F. X. Kärtner, “Coherent pulse synthesis: towards subcycle optical waveforms,” Laser Photonics Rev. **9**(2), 129–171 (2015). [CrossRef]

**5. **S. O. Konorov, D. A. Akimov, E. E. Serebryannikov, A. A. Ivanov, M. V. Alfimov, and A. M. Zheltikov, “Cross-correlation FROG CARS with frequency-converting PCFs,” Phys. Rev. E **70**, 057601 (2004). [CrossRef]

**6. **H. N. Paulsen, K. M. Hilligsøe, J. Thøgersen, S. R. Keiding, and J. J. Larsen, “Coherent anti-Stokes Raman scattering microscopy with a photonic crystal fiber based light source,” Opt. Lett. **28**(13), 1123–1125 (2003). [CrossRef] [PubMed]

**7. **J. M. Dudley, G. Genty, and S. Coen, “Supercontinuum generation in photonic crystal fiber,” Rev. Mod. Phys. **78**(4), 1135–1184 (2006). [CrossRef]

**8. **A. M. Zheltikov, “Let there be white light: supercontinuum generation by ultrashort laser pulses,” Phys.- Usp. **49**(6), 605–628 (2006). [CrossRef]

**9. **T. Udem, R. Holzwarth, and T. W. Hänsch, “Optical frequency metrology,” Nature **416**(6877), 233–237 (2002). [CrossRef] [PubMed]

**10. **A. M. Zheltikov, A. L’Huillier, and F. Krausz, “Nonlinear Optics,” in *Handbook of Lasers and Optics*, ed. by F. Träger, New York, Springer, 2007, pp. 157 – 248.

**11. **A. Hasegawa, *Optical Solitons in Fibers* (Heidelberg, Springer, 1990).

**12. **V. E. Zakharov and A. B. Shabat, “Exact theory of two-dimensional self-focusing and one-dimensional self-modulation of waves in nonlinear media,” Sov. Phys. JETP **34**, 62–65 (1972).

**13. **L. F. Mollenauer, R. H. Stolen, and J. P. Gordon, “Experimental observation of picosecond pulse narrowing and solitons in optical fibers,” Phys. Rev. Lett. **45**(13), 1095–1098 (1980). [CrossRef]

**14. **T. Balciunas, C. F. Dutin, G. Fan, T. Witting, A. A. Voronin, A. M. Zheltikov, G. Frédéric, G. G. Paulus, A. Baltuska, and F. Benabid, “Sub-cycle gigawatt peak power pulses self-compressed by optical shock waves,” Nat. Commun. **6**, 6117 (2015). [CrossRef] [PubMed]

**15. **A. A. Voronin and A. M. Zheltikov, “Subcycle solitonic breathers,” Phys. Rev. A **90**(4), 043807 (2014). [CrossRef]

**16. **J. Herrmann, U. Griebner, N. Zhavoronkov, A. Husakou, D. Nickel, J. C. Knight, W. J. Wadsworth, P. St. J. Russell, and G. Korn, “Experimental evidence for supercontinuum generation by fission of higher-order solitons in photonic fibers,” Phys. Rev. Lett. **88**(17), 173901 (2002). [CrossRef] [PubMed]

**17. **A. A. Voronin, I. V. Fedotov, A. B. Fedotov, and A. M. Zheltikov, “Spectral interference of frequency-shifted solitons in a photonic-crystal fiber,” Opt. Lett. **34**(5), 569–571 (2009). [CrossRef] [PubMed]

**18. **X. Fang, M. Hu, L. Huang, L. Chai, N. Dai, J. Li, A. Yu. Tashchilina, A. M. Zheltikov, and C. Wang, “Multiwatt octave-spanning supercontinuum generation in multicore PCF,” Opt. Lett. **37**, 2292–2294 (2012). [CrossRef] [PubMed]

**19. **T. Südmeyer, F. Brunner, E. Innerhofer, R. Paschotta, K. Furusawa, J. C. Baggett, T. M. Monro, D. J. Richardson, and U. Keller, “Nonlinear femtosecond pulse compression at high average power levels by use of a large-mode-area holey fiber,” Opt. Lett. **28**(20), 1951–1953 (2003). [CrossRef] [PubMed]

**20. **A. V. Mitrofanov, A. A. Ivanov, M. V. Alfimov, A. A. Podshivalov, and A. M. Zheltikov, “Microjoule supercontinuum generation by stretched megawatt femtosecond laser pulses in a large-mode-area photonic-crystal fiber,” Opt. Commun. **280**(2), 453–456 (2007). [CrossRef]

**21. **G. Krauss, S. Lohss, T. Hanke, A. Sell, S. Eggert, R. Huber, and A. Leitenstorfer, “Synthesis of a single cycle of light with compact erbium-doped fibre technology,” Nat. Photonics **4**(1), 33–36 (2010). [CrossRef]