Abstract
Gasfilled hollowcore fiber (HCF) pulse postcompressors generating few to singlecycle pulses are a key enabling tool for attosecond science and ultrafast spectroscopy. Achieving optimum performance in this regime can be extremely challenging due to the ultrabroad bandwidth of the pulses and the need of an adequate temporal diagnostic. These difficulties have hindered the full exploitation of HCF postcompressors, namely the generation of stable and highquality nearFouriertransformlimited pulses. Here we show that, independently of conditions such as the type of gas or the laser system used, there is a universal route to obtain the shortest stable output pulse down to the singlecycle regime. Numerical simulations and experimental measurements performed with the dispersionscan technique reveal that, in quite general conditions, postcompressed pulses exhibit a residual thirdorder dispersion intrinsic to optimum nonlinear propagation within the fiber, in agreement with measurements independently performed in several laboratories around the world. The understanding of this effect and its adequate correction, e.g. using simple transparent optical media, enables achieving highquality postcompressed pulses with only minor changes in existing setups. These optimized sources have impact in many fields of science and technology and should enable new and exciting applications in the few to singlecycle pulse regime.
Introduction
The chirped pulse amplification (CPA) technique applied to Titanium Sapphire lasers has made intense nearinfrared (NIR) ultrashort pulses in the 20−100 fs range widely available for scientific, biomedical and industrial applications. Special efforts have been devoted to generate even shorter pulses, in the few and singlecycle regime, due to a number of interesting applications. In particular, such pulses have paved the way for attosecond physics and metrology^{1,2,3,4,5,6}, via the extreme ultraviolet (XUV) attosecond pulse trains and isolated attosecond pulses^{7} that can be obtained by highharmonic generation (HHG). The use of fewcycle optical pulses with durations close to or shorter than 10 fs in the nearinfrared, visible and nearultraviolet spectral regions has been extended in recent years to a wide range of spectroscopic techniques such as impulsive vibrational spectroscopy^{8,9,10}, timeresolved stimulated Raman spectroscopy^{11,12,13,14}, and ultrafast pumpprobe absorption spectroscopy^{15,16}. Fewcycle optical pulses have also become an interesting tool for transient absorption microscopy^{17}, nearfield imaging techniques^{18} and for generating ultrashort terahertz radiation^{19}.
While it is possible to obtain sub10 fs pulses from CPA^{20} or from optical parametric amplification^{21} systems, the former is not easy to accomplish, and the latter is not commonplace. Therefore, postcompression techniques are usually employed for the generation of intense few and even singlecycle pulses in the near and midinfrared spectral regions^{22,23}. In order to postcompress ultrafast pulses down to the fewcycle regime, two steps are usually needed. First, nonlinear processes broaden the pulse spectrum, thus decreasing the Fourierlimited pulse duration. In a second step, the spectral phase resulting from the previous stage is compensated, typically using chirped mirrors, gratings, prisms, or other dispersive systems, resulting in a temporally compressed pulse. This scheme was first proposed in the context of optical fibers in the 1980s^{24}, and enabled achieving 6 fs pulses when compensating simultaneously the outcoming group delay dispersion (GDD) and thirdorder dispersion (TOD)^{25}. The scheme was successfully expanded in 1996 by Nisoli and coworkers to the ultraintense laser pulse regime (mJlevel pulses) by using hollowcore fibers (HCF) filled with gases^{26}. Using the latter scheme, together with chirped mirrors as the compression system, fewcycle pulses in the hundreds of μJ energy range with 0.1 TW peak power were obtained^{27}. Later, similar results were obtained using the light filamentation process in the spectral broadening stage^{28}.
In spite of requiring a finer input beam alignment than filamentationbased compressors, HCF compressors are today the most widely used highenergy postcompressed sources, in part due to their intrinsic spatial filtering properties which result in very high quality beam profiles. Furthermore, their longterm stability can be greatly improved by using, e.g., piezodriven mirror mounts to ensure stable and constant spatial coupling of the input laser pulses into the HCF via a simple feedback loop. Hollowcore fiber postcompressed pulses have shown a great potential in a wide range of applications, such as pumpprobe experiments in conjunction with attosecond pulses^{29}, ultrafast measurement of electrical and optical properties of solids^{30,31}, timeresolved studies of Coulomb explosion dynamics^{32}, ultrafast spectroscopy techniques^{33,34,35,36,37,38,39}, and very recently a new generation of compact kHz laserplasma accelerators based on singlecycle pulses^{40}.
To access the Fourier Limit of a pulse after a nonlinear propagation process one has to deal with the complex phase that the pulse acquires due to the interplay between different linear and nonlinear effects. In general, researchers optimize their HCF compressors by empirically adjusting several key parameters, such as gas type and pressure, input pulse characteristics and coupling conditions, with the final result usually involving a delicate compromise between output efficiency, amount of spectral broadening and achievable degree of compression (pulse duration and quality) for their particular system and chirped mirror set. The ability to measure and quantify the achieved degree of compression is therefore paramount to identify the main characteristics of the output pulse and to further optimize its compression. Both spectral phase oscillations and the overall spectral phase of a pulse can be visualized in a very straightforward way using the dispersionscan (dscan)^{41} technique, which has been extensively used in the last years to characterize many stateoftheart fewcycle pulse sources around the world^{6,38,39,40,42}.
Dscan is a recent approach for the simultaneous measurement and compression of femtosecond laser pulses. Its experimental setup is fully inline, without any beam splitting, recombination or temporal delaying of the pulses. Experimentally it involves the measurement of the spectrum of a nonlinear signal, such as secondharmonic generation (SHG), as a function of dispersion applied to the pulse. This can be performed with pulse compression setups, such as a chirped mirror (CM) and glass wedge compressor, where the amount of glass traversed by the pulse is an independent variable controlled by insertion of one of the wedges: while the CMs impart negative dispersion, the variable positive dispersion introduced by the wedges will vary the total dispersion experienced by the pulse to be measured. In secondharmonic generation dscan (SHG dscan), measuring the SHG signal after the compressor provides a twodimensional trace of the SHG spectrum vs. insertion. An optimization algorithm is then used to retrieve the spectral phase of the pulse from the measured dscan trace and calibrated linear spectrum^{41}. A recent approach to dscan retrieval can also be used to obtain the pulse amplitude and phase from the measured trace^{43}, but in this case, the trace itself must be calibrated.
Dscan has been successfully demonstrated with fewcycle pulses since its inception^{22,41}, and over octavespanning singlecycle pulses have been measured directly with SHG dscan^{6,40,43,44,45}. Apart from its robustness and performance, another important advantage of dscan is the fact that it directly results in very intuitive traces that provide useful guiding information on the quality of the achieved pulse compression, which motivates our present use of the dscan trace as a diagnostic tool. For instance, a flat and thin trace is indicative of excellent compression, since this means that for a particular position of the compressor, all spectral components are equally compressed and hence their SHG signal is maximized. If the trace has a tilt, this means that different parts of the spectrum are being compressed for different applied dispersions. In other words, we have a frequencydependent chirp in the pulses, i.e., the pulses have a residual thirdorder phase due to uncompensated thirdorder dispersion. A curved, paraboliclike trace would be indicative of fourthorder dispersion, and so on. In the case of phase oscillations, these will produce spectral modulations and a wavy appearance in the resulting dscan trace (for a more detailed description of dscan traces and their interpretation, the reader is referred to refs^{22,41}).
In this paper we look for a route to obtain the optimum stable postcompressed pulse from a HCF using the dscan technique as the compression device and the dscan trace to univocally identify this route. Numerical simulations and experimental measurements reveal that, in quite general conditions, the best performance results in the postcompressed pulses typically exhibiting a residual thirdorder phase of nonlinear origin, i.e., a signature tilt (residual TOD) intrinsic to optimum nonlinear propagation within the fiber, in agreement with experimental observations made in several laboratories around the world. Overdriving the HCF above this optimum regime invariably results in an increasingly complex nonlinear spectral phase which renders compression very hard to optimize. These results and behavior have been obtained for different gases and different setup parameters (including HCF length, gas pressure and pulse energy) showing the universality of the existence of the optimum regime. We have also probed the spatial quality of the optimum output pulse, showing a high spatiospectral homogeneity. Moreover, we have been able to develop a simple theoretical model that explains the main nonlinear effects underlying the optimum regime, which is very useful for finding out the proper parameters to achieve the desired few to singlecycle pulses. The understanding of the optimum propagation regime, of its intrinsic TOD and its subsequent correction using, e.g., unusual transparent optical media with adequate ratio between second and thirdorder dispersion, enables achieving optimized highquality postcompressed pulses with only minor changes in existing setups.
Results and Discussion
Regimes of nonlinear propagation in the hollowcore fiber. Identification of the optimum dscan trace
The identification of the optimum parameters of a nonlinear process to obtain a desired output pulse is not an easy task. Fortunately, in the context of nonlinear propagation of ultrashort laser pulses, we have helpful theoretical models to guide us. We used a nonlinear spatiotemporal model (see the Methods section) to simulate the nonlinear propagation of laser pulses in a static HCF filled with different gases (argon, neon and air) and in different conditions (gas pressure; input pulse energy and duration), and their subsequent compression with a dscan system based on chirped mirrors and glass wedges. All cases studied are representative of real experimental situations and present qualitatively similar dynamics.
In Fig. 1 we present the theoretical (top row) and experimental (bottom row) dscan traces obtained in a pressurescan experiment. The left, middle and right column represent the low pressure cases (weak interaction regime), the optimum pressure cases (optimum interaction regime) and the high pressure cases (strong interaction regime), respectively (see the figure caption for the detailed set of parameters). The theoretical dscan traces are calculated from the onaxis field obtained at the end of the HCF. The dscan compressor is composed of chirped mirrors followed by a pair of BK7 glass wedges, as in the dscan setup commonly used for fewcycle pulse compression^{41}, but taking only into account their group delay dispersion effect and neglecting higher order terms. We assume ultrabroadband chirped mirrors, which introduce −120 fs^{2} of pure GDD, with the BK7 wedges introducing 46.6 fs^{2}/mm at the central wavelength 780 nm. As we will see in the comparison between simulations and experimental results, neglecting higher order terms in the numerical dscan stage is a very good approximation because the highorder contributions of CMs and wedges cancel out. This simplification helps us identify the origin of the higher order dispersion terms that may appear in the pulse phase, which must come from the propagation in the HCF because the compression/measurement stage does not introduce them. Therefore, and based on the explanation given in the Introduction section, we will be able to identify any high order phase terms of the output pulse directly from the dscan trace structure.
The observed similarity between the theoretical and experimental pressure scans indicates that the theoretical model, including the calculation of dscan traces, takes into account the main effects occurring in the experiment. The results presented in Fig. 1 also help us identify the optimum dscan traces (those in the middle column). It is clear that the weak interaction regime traces (left column) represent situations where the pulses did not experience enough nonlinear interaction to broaden their spectra, showing narrow dscan traces in the spectral coordinate. In contrast, the intricate modulations of the strong interaction regime traces (right column) indicate that they correspond to situations where the pulses had such intense nonlinear interaction that, although the spectral broadening was large, the complex spectral phase makes the output pulse useless for most applications. Moreover, this regime is easily identified in the experiments due to a poor output stability. We can therefore define the optimum interaction regime as the one that generates dscan traces similar to those of the middle column, which represent situations where the pulses are spectrally broadened but in a regime in which the acquired spectral phase can be adequately compensated for. The retrieved pulses obtained from the optimal dscan traces presented in Fig. 1 have durations of ∼4.1 fs FWHM for the theoretical case (top row) and ∼4.0 fs FWHM for the experimental case (bottom row). These optimum dscan traces already correspond to subtwocycle pulses, and additional compensation of their intrinsic residual TOD^{44} allows obtaining highquality pulses in the singlecycle regime^{6,43,44}.
The exact limits of the three identified propagation regimes are somewhat flexible but, as shown in Fig. 1, they can be easily identified using the dscan as the guiding tool. There is a large number of parameters that one can use to reach the optimum regime in the laboratory: input pulse energy, input pulse temporal duration (chirp), coupling conditions (focusing, mode and numerical aperture matching), gas type and pressure. Using some of them one can get into the optimum regime to achieve the optimal postcompressed output pulse, which will not be the one with the broadest spectrum, but the shortest among the less temporallystructured obtained pulses, which we have been able to univocally identify through the resulting dscan trace.
Optimum dscan trace properties: Universality, Spatial and Spectral structure, and Physical origin
In the previous results, we have been able to identify the optimum dscan trace when using argon. We now apply the theoretical model to verify that similar optimum dscan traces appear using different gases and experimental parameters. Figure 2 shows the theoretical optimum dscan traces obtained at the end of a HCF using argon (left), neon (middle) and air (right). The retrieved output pulse duration for the three cases is ∼4.1 fs, ∼4.0 fs and ∼4.7 fs FWHM, respectively. All these traces were identified after simulating pressure scans as the one presented in the top row of Fig. 1. When using argon or neon we obtained almost identical optimum traces, while when using air there are some variations due to the presence of the Raman effect, which is not present when using atomic gases. Surprisingly, regardless of the different conditions (fiber length, pulse energy, gas pressure and gas type), we have always been able to identify the optimum dscan trace at the end of the HCF. This universal behavior, together with the particular structure of the optimum dscan trace as a univocal fingerprint, is a very useful tool both for optimizing existing systems and to promote the spreading of fewcycle pulse systems for many more applications that can benefit from a stable and reproducible source.
For completeness we present a movie (see Supplementary Material) showing the evolution of the dscan trace at the end of the HCF during a pressure scan with argon, together with the evolution of the wedgeinsertiondependent postcompressed pulse. The left plot in the movie shows the particular features of the dscan trace along all three stages. The right plot shows the final compressed pulse that can be obtained, displaying the continuous evolution of the structure of the output pulse.
The structure of the optimum dscan trace gives us important information about the output pulse obtained from the postcompression system. The first distinctive feature of the optimum dscan trace is a slightly negative slope which corresponds to a negative TOD. This is the remaining residual TOD accumulated during the nonlinear propagation of the pulse inside the HCF. According to the simulations, this TOD mostly comes from the selfsteepening effect, similarly to what occurs in postcompression setups for pulses in the midinfrared spectral region^{46}. This tilt of the dscan trace is very helpful to find experimentally the optimum regime to obtain singlecycle postcompressed pulses. To our knowledge, this remaining TOD is clearly present in experimental HCF postcompressed pulses obtained in different laboratories around the world when optimizing for the output pulses (e.g., United Kingdom^{45}, Portugal^{44,45,47}, France^{48}, Sweden^{42,49}, Germany^{50}, Canada^{51}, China^{52}, Austria^{46} and United States of America^{6,39}, as some examples). Those experiments were done using HCFs with different characteristics (length and/or core diameter), filled with different gases (Ar, Ne or He) and with lasers in the NIR and midIR regions. All these observations support that the route to obtain an optimum postcompressed pulse described here is universal.
Another very important feature of the output pulse related to the optimum dscan trace is spatial homogeneity. All the theoretical dscan traces presented until now have been calculated from the onaxis field at the end of the HCF. One would expect that, as the optimum regime corresponds to a moderate nonlinear interaction regime, the resulting output beam should have a quite good spatial homogeneity, as it has in fact been proved experimentally in similar conditions^{47}. Figure 3 shows the theoretical farfield dscan calculated from a spatial selection of the field obtained at the end of the HCF. The parameters here are the same used to obtain the optimum dscan trace in argon (leftmost case of Fig. 2). The farfield distribution is calculated by doing the Hankel transform integrating spatially over the selected part of the beam. Then the dscan trace is finally calculated using the farfield corresponding to k_{⊥} = 0, i.e. the zero divergence farfield. Figure 3a shows the dscan trace of the farfield when taking only into account the onaxis near field, which is basically the same presented using the field onaxis (left picture of Fig. 2). Figure 3b shows the dscan trace of the farfield when integrating the nearfield up to 62.5 μm (half of the HCF core radius, which contains 94% of the output energy). As it can be observed, the dscan trace essentially retains the same shape as before, being slightly narrower in the spectral direction and more stretched along the dispersion axis. Figure 3c shows the dscan of the farfield obtained by integrating up to 100 μm (which accounts for 99% of the output energy). In this case the spectral narrowing of the dscan is more pronounced, as we are taking into account the most external part of the beam that sees less nonlinearity. Note that even this last dscan trace preserves the TOD signature related to the optimum regime, although the optimal pulse duration changes from ∼4.1 fs in the first case to ∼4.7 fs and ∼5.3 fs in the second and third cases, respectively. Finally Fig. 3d shows the beam fluence at the end of the HCF with circles indicating the integrated area used to calculate the farfield dscan for the three cases shown, which helps to visualize the good homogeneity of the beam obtained under the optimum parameters. The smooth output spatial profile also shows that the postcompressed pulse is suitable for applications which require good beam quality.
The identification of the optimum regime and the spatial homogeneity of the resulting output beam indicate that this regime is achieved for moderate nonlinear interaction conditions (not presenting, for instance, selffocusing dynamics, strong gas ionization or similar phenomena related to intense nonlinear interactions). To unveil the main nonlinear effects underlying this regime, we can use a theoretical model based on propagation equations (this model is discussed in Chapter 4 of ref.^{53}). As we will show below, this model is useful for estimating the parameters of the optimum propagation regime and, consequently, for obtaining the desired few or singlecycle postcompressed pulses. The model assumes that selfphase modulation (SPM) and linear losses are the only phenomena affecting the pulse during propagation. With these assumptions we are able to anticipate the maximum induced spectral broadening after propagating a distance L in the fiber, which in the case of a Gaussian pulse turns out to be Δω_{ max } = 0.86Δω_{0}B_{ max }, where Δω_{0} and B_{ max } represent the initial spectral width and the maximum nonlinear phase shift, respectively. Assuming now that the input and output pulses are Fourierlimited, which is valid if we are able to compensate properly the spectral phase of the output pulse, their spectral widths are directly related to their respective temporal widths by \({\rm{\Delta }}{\omega }_{0}\propto {\rm{\Delta }}{T}_{i}^{1}\) and \({\rm{\Delta }}{\omega }_{max}\propto {\rm{\Delta }}{T}_{f}^{1}\). Thus we can obtain an estimate of the required maximum nonlinear phase shift, i.e. the maximum Bintegral, for any given initial and final (compressed) pulse durations:
This basic model indicates that to broaden the spectrum of Fourierlimited input pulses of similar duration towards the singlecycle regime, we will always need to accumulate roughly the same maximum amount of the Bintegral. This way, the Bintegral becomes a very useful tool for predicting adequate values of some of the parameters.
We now proceed to finding the accumulated Bintegral of a pulse with energy E_{ in }, temporal duration t_{ p }, and spectrum centered at λ_{0}, propagating in a HCF of length L_{ F } and with a core of radius r_{ F }. We assume that the beam couples perfectly to the fundamental mode of the HCF, with wellknown losses denoted by α^{54}. Inside the HCF we have a gas at constant pressure, with nonlinear index (1 − x_{ R })n_{2}, where x_{ R } indicates the amount of Raman effect (x_{ R } = 0 for atomic gases and \({x}_{R}\sim 0.6\) for molecular gases like air). The maximum accumulated Bintegral then reads:
where \({I}_{0}={E}_{in}/({t}_{p}\pi {r}_{F}^{2})\) is the maximum input intensity. We should point out another argument that further confirms the suitability of the Bintegral as a good parameter for determining the optimal compression regime, based on the scaling properties of nonlinear wave equations for pulse propagation in gases. From Eq. 2, in the case of weak linear absorption (\(\alpha {L}_{F}\ll 1\)), we have \(B\simeq 2\pi {\lambda }_{0}^{1}{I}_{0}\mathrm{(1}{x}_{R}){n}_{2}{L}_{F}\). Following Heyl et al.^{49,55}, the propagation equations remain invariant if we transform the input parameters with a scaling dimensionless parameter η according to certain rules. In our case, the scaling relations E_{ in } → η^{2}E_{ in }, r_{ F } → ηr_{ F }, n_{2} → η^{−2}n_{2} and L_{ F } → η^{2}L_{ F } make B invariant under this transformation.
Applying Eq. 2 to all the simulations presented in Fig. 1 we obtain similar values for the Bintegral, \({B}^{sim}\gtrsim 4\) (4.2, 4.4 and 5.6 for the simulations corresponding to argon, neon and air, respectively), as predicted by the basic model described above. In fact, Eq. 1 gives a constant value \({B}_{max}\gtrsim 7\), when using ΔT_{ i } = 25 fs and ΔT_{ f } corresponding to a 1.5cycle, 3.9 fs pulse, proving that this simple model includes the main effects underlying this regime: SPM and linear losses. The optimum Bintegral, defined as the lower limit value of B^{sim}, \({B}_{opt}^{sim}=4\), provides a useful hint for finding optimum propagation parameters for input pulses of this duration. Taking into account that the nonlinearity depends linearly on the gas pressure, we have used Eq. 2 to calculate, for example, the pressure needed to obtain the optimum dscan trace at the end of a HCF of fixed length L_{ F } and core radius r_{ F }, filled with some gas with nonlinear refractive index \(\mathrm{(1}{x}_{R}){n}_{2}^{\ast }\) (being \({n}_{2}^{\ast }\) the nonlinear parameter at 1 bar pressure), with an input pulse of energy E_{ in } and a temporal duration t_{ p }:
We have checked that the parameters obtained from Eq. 3 bring us to a situation very close to the optimal dscan trace for argon, neon and air, over a wide range of parameters, and whenever the propagation occurs in the moderate nonlinear regime. These expressions demonstrate the tunability of the optimal trace with respect to key parameters such as pressure, type of gas and input energy or pulse duration, setting this procedure as a universal route for obtaining fewcycle pulses in the visibleNIR spectral region.
We have also estimated the maximum Bintegral for some experimental cases^{6,39,44,47}, besides the experiments presented in this work, obtaining a value \({B}^{exp}\gtrsim 9\) (range of values from 9 to 17 with an average of 14). To calculate these values we have estimated that the energy coupled into the HCF is 40% of the available input energy and we have assumed that the input pulse is Fourierlimited. Even with all these assumptions, which are not always fulfilled in experiments, we have obtained similar values of the experimental maximum Bintegral, which demonstrates that this parameter is also useful for bringing an experiment close to the optimum regime. We can see that the estimated values of the Bintegral in experiments are consistently larger than in numerical simulations, i.e., B^{exp} > B^{sim}. This is a consequence of the less effective nonlinear interaction present in real experiments compared to the numerical models, due to factors such as the imperfect spatial intensity distribution of the laser beam, as reported in the literature since the early studies of nonlinear ultrashort pulse propagation (see ref.^{56}).
TOD compensation of the optimal dscan trace
The residual TOD which appears in the optimal propagation regime can be further compensated for, not only to reduce the output pulse duration but also to increase the amount of energy in the pulse. This can be done, for example, by using an adequate transparent medium such as water^{42,44,45}, zcut KDP^{43} or zcut ADP^{6}, which directly enables achieving highquality pulses in a more strict singlecycle regime. In spite of the relatively small magnitude of this TOD (e.g., −40 fs^{3} reported in ref.^{44}), its effect on a few and singlecycle pulse can be dramatic: its compensation enabled going from a 3.8 fs pulse (with satellite pulses due to TOD) to a neartransformlimited 3.2 fs 1.3cycle pulse, which was accompanied by a significant improvement in pulse contrast and peak intensity^{44}.
In the center plot of Fig. 4 we show the dscan trace obtained after compensating the residual TOD by adding +20 fs^{3} (at the central wavelength of 780 nm) in the measurement/compression stage, compared with the noncompensated case (left plot). We see that TOD compensation effectively tilts the dscan trace so that its average negative slope practically disappears. To illustrate the improvement in intensity throughput after TOD compensation we show on the right plot of Fig. 4 the direct output pulse and the TODcompensated pulse, together with the Fourierlimited pulse. The used glass insertion in this case is the one which results in maximum intensity in the dscan trace. Clearly, the pulse corresponding to the TODcorrected trace has better features: it presents an increase in peak intensity of the order of 33%, a reduction of the temporal FWHM pulse duration from ∼4.1 fs to ∼2.9 fs, which is in very good agreement with experimental results, and smaller secondary temporal structures than the noncorrected case, as also confirmed by experiments.
Conclusion
In this work we have presented and analyzed a route to obtain optimum hollowcore fiber postcompressed pulses using the dscan setup as the compression and diagnostic device. We have demonstrated that by changing the gas pressure and/or input pulse energy, one is able to identify the optimal postcompression parameters to obtain the shortest, less structured and stable output pulses. This optimal setup can be univocally identified by the dscan trace of the output pulses, which shows a marked TOD feature, whereas overdriving the HCF above this optimum regime invariably results in an increasingly complex nonlinear spectral phase, which renders compression very hard to optimize. We have verified that the optimum dscan trace can be found for different gases and setup conditions, highlighting the universality of this phenomenon. Moreover, we have also proved the good spatiospectral characteristics of the output pulse obtained under these conditions. For the optimum propagation regime, the remaining TOD can be further corrected, for example by simple linear propagation in an adequate transparent medium, which improves the temporal shape and the peak power of the output pulse. The identified propagation regime and approach enable the generation of stable and highquality few to singlecycle pulses, which has direct implications in the performance of current and new HCF pulse postcompression systems and will help improve and extend the applications of these extreme light sources in many fields of science and technology.
Methods
Simulations
To study the nonlinear propagation of a laser pulse in a HCF we have implemented the standard nonlinear envelope propagation equation^{57}. The model includes the spatial and temporal pulse dynamics. We use a local frame moving with the pulse T = t − z/v_{ g }, being v_{ g } the group velocity of the pulse, and assume cylindrical symmetry (r radial coordinate, z axial coordinate). The propagation equation for the temporal envelope of the pulse, A(r, z, T), is
The first part of the propagation equation, \(\hat{L}\), represents the linear propagation effects: diffraction, dispersion and linear losses. The other part, \(\hat{N}\), represents the nonlinear propagation effects, which include selfphase modulation, Raman scattering, ionization, losses due to the ionization process and plasma absorption, and selfsteepening.
To solve Eq. 4 we use a splitstep method dividing each propagation step into two substeps^{53}. The first substep consists in applying only the linear effects by decomposing the input pulse into the EH_{1m} modes of the hollowcore fiber^{54,58}
where \(\tilde{A}(r,z,\omega )\) represents the Fourier Transform of A(r, z, T). The coefficients of the decomposition can be calculated by doing the inverse Hankel Transform of the spatial beam distribution in the core of the HCF
being a the core radius of the HCF and α_{ m } the m^{th}zero of J_{0}(x), where J_{ ν } is the Bessel function of the first kind of order ν. We solve Eq. 6 by using the discrete Hankel Transform scheme proposed in^{59}. The complete linear propagation in the HCF is simulated by using the complex propagation coefficient of each mode, β_{ m }(ω)^{54}, as shown in Eq. 7. The real and imaginary parts of β_{ m }(ω) take into account all the dispersion and losses of the m^{th}mode of the HCF, respectively
The second substep of the method consists in applying the nonlinear effects. Separating\(\hat{N}[A(r,z,T)]=\) \({N}_{SPM}(A)+{N}_{ioniz}(A)+{N}_{abs}(A)\), a mathematical expression for each term can be obtained^{57}, as given below. For the selfphase modulation, Raman scattering and selfsteepening, we have
In Eq. 8, k_{0} = n_{0}2π/λ_{0}, with λ_{0} the central wavelength of the pulse. n_{2} the nonlinear refractive index, ω_{0} = 2πc/λ_{0} and K(T − t) representing the SPM together with the Raman scattering, which has the following form (1 − x)δ(T − t) + x/τ_{ K }exp(−(T − t)/τ_{ K }), where x is the ratio between the SPM and the Raman effect, and τ_{ K } fs the characteristic time for the Raman response. For the ionization, we have
where σ is the cross section for the inverse Bremsstrahlung that depends on the collision time (τ_{ C }), the critical density of the medium and the central frequency of the laser pulse^{57}; ρ represents the ionized electron density, whose evolution is governed by ∂ρ/∂t = W(A^{2})(ρ_{ at } − ρ), where W(A^{2}) is the ionization rate calculated using the PPT model^{60} and ρ_{ at } is the atomic density of the medium. Finally, the absorption term is
where U_{ i } represents the ionization potential of the gas.
Experiments
The experiments were performed by employing 23–25 fs Fouriertransformlimited laser pulses centered at 780 nm. These pulses were generated with a 1 kHz Ti:Sapphire CPA laser system (Femtolasers FemtoPower Compact PRO HE CEP) which is part of the CLPU facility. The maximum pulse energy available was 2.5 mJ and we adjusted it to 1 mJ in our experiments. The laser pulses were focused by a spherical mirror (1 m focal length) into a hollowcore fiber (HCF) with an inner diameter of 250 micron and 1 meter length. The HCF was filled with argon gas at different pressures. The output pulses were compressed and measured using a dscan system from Sphere Ultrafast Photonics. The scanning SHG dscan setup includes a doubleangle chirped mirror set (Ultrafast Innovations GmbH) and motorized BK7 glass wedges to induce the dispersion scan.
Data availability
The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.
References
 1.
Hentschel, M. et al. Attosecond metrology. Nature 414, 509–513 (2001).
 2.
Kling, M. F. & Vrakking, M. J. J. Attosecond electron dynamics. Annu. Rev. Phys. Chem. 59, 463–492 (2008).
 3.
Krausz, F. & Ivanov, M. Attosecond physics. Rev. Mod. Phys. 81, 163–264 (2009).
 4.
Gallmann, L., Cirelli, C. & Keller, U. Attosecond science: recent highlights and future trends. Annu. Rev. Phys. Chem. 63, 447–469 (2012).
 5.
Krausz, F. & Stockman, M. I. Attosecond metrology: from electron capture to future signal processing. Nat. Photonics 8, 205–213 (2014).
 6.
Timmers, H. et al. Generating highcontrast, near singlecycle waveforms with thirdorder dispersion compensation. Opt. Lett. 42, 811–814 (2017).
 7.
Sansone, G. et al. Isolated singlecycle attosecond pulses. Science 314, 443–446 (2006).
 8.
Polli, D. et al. Broadband pumpprobe spectroscopy with sub10fs resolution for probing ultrafast internal conversion and coherent phonons in carotenoids. Chem. Phys. 350, 45–55 (2008).
 9.
Liebel, M., Schnedermann, C., Wende, T. & Kukura, P. Principles and applications of broadband impulsive vibrational spectroscopy. J. Chem. Phys. A 119, 9506–9517 (2015).
 10.
Du, J. et al. Surfaceenhanced impulsive coherent vibrational spectroscopy. Sci. Rep. 6, 36471 (2016).
 11.
Kukura, P., McCamant, D. W. & Mathies, R. A. Femtosecond stimulated Raman spectroscopy. Annu. Rev. Phys. Chem. 58, 461–488 (2007).
 12.
Fujisawa, T., Kuramochi, H., Hosoi, H., Takeuchi, S. & Tahara, T. Role of coherent lowfrequency motion in excitedstate proton transfer of green fluorescent protein studied by timeresolved impulsive stimulated Raman spectroscopy. J. Am. Chem. Soc. 138, 3942–3945 (2016).
 13.
Dietze, D. R. & Mathies, R. A. Femtosecond stimulated Raman spectroscopy. Chem Phys Chem 17, 1224–1251 (2016).
 14.
Kuramochi, H., Takeuchi, S. & Tahara, T. Femtosecond timeresolved impulsive stimulated Raman spectroscopy using sub7fs pulses: Apparatus and applications. Rev. Sci. Instrum. 87, 043107 (2016).
 15.
Wang, Y.T. et al. Use of ultrafast timeresolved spectroscopy to demonstrate the effect of annealing on the performance of P3HT:PCBM solar cells. ACS Appl. Mater. Interfaces 7, 4457–4462 (2015).
 16.
Luo, C.W., Wang, Y.T., Yabushita, A. & Kobayashi, T. Ultrabroadband timeresolved spectroscopy in novel types of condensed matter. Optica 3, 82–92 (2016).
 17.
Schnedermann, C. et al. Sub10 fs timeresolved vibronic optical microscopy. J. Phys. Chem. Lett. 7, 4854–4859 (2016).
 18.
Nishiyama, Y., Imura, K. & Okamoto, H. Observation of plasmon wave packet motions via femtosecond timeresolved nearfield imaging techniques. Nano Lett. 15, 7657–7665 (2015).
 19.
Darmo, J. et al. Fewcycle terahertz generation and spectroscopy of nanostructures. Phil. Trans. R. Soc. Lond. A 362, 251–262 (2004).
 20.
Seres, J. et al. 10fs terawattscale Ti: sapphire laser system. Opt. Lett. 28, 1832–1834 (2003).
 21.
Cerullo, G., Nisoli, M., Stagira, S. & De Silvestri, S. Sub8fs pulses from an ultrabroadband optical parametric amplifier in the visible. Opt. Lett. 23, 1283–1285 (1998).
 22.
Miranda, M. et al. Characterization of broadband fewcycle laser pulses with the dscan technique. Opt. Express 20, 18732 (2012).
 23.
Cardin, V. et al. 0.42 TW 2cyle pulses at 1.8μm via hollowcore fiber compression. Appl. Phys. Lett. 107, 181101 (2015).
 24.
Tomlinson, W. J., Stolen, R. H. & Shank, C. V. Compression of optical pulses chirped by selfphase modulation in fibers. J. Opt. Soc. Am. B 1, 139–149 (1984).
 25.
Fork, R., Cruz, C., Becker, P. & Shank, C. Compression of optical pulses to 6 femtoseconds by using cubic phase compensation. Opt. Lett. 12, 483–485 (1987).
 26.
Nisoli, M., De Silvestri, S. & Svelto, O. Generation of high energy 10 fs pulses by a new pulse compression technique. Appl. Phys. Lett. 68, 2793–2795 (1996).
 27.
Sartania, S. et al. Generation of 0.1TW 5fs optical pulses at a 1kHz repetition rate. Opt. Lett. 22, 1562–1564 (1997).
 28.
Hauri, C. P. et al. Generation of intense, carrierenvelope phaselocked fewcycle laser pulses through filamentation. Appl. Phys. BLasers and Optics 79, 673–677 (2004).
 29.
Blättermann, A. et al. In situ characterization of fewcycle laser pulses in transient absorption spectroscopy. Opt. Lett. 40, 3464–3467 (2015).
 30.
Schiffrin, A. et al. Opticalfieldinduced current in dielectrics. Nature 493, 70–74 (2013).
 31.
Schultze, M. et al. Controlling dielectrics with the electric field of light. Nature 493, 75–78 (2013).
 32.
Bocharova, I. A. et al. Timeresolved Coulombexplosion imaging of nuclear wavepacket dynamics induced in diatomic molecules by intense fewcycle laser pulses. Phys. Rev. A 83, 013417 (2011).
 33.
Liu, J., Okamura, K., Kida, Y., Teramoto, T. & Kobayashi, T. Clean sub8fs pulses at 400 nm generated by a hollow fiber compressor for ultraviolet ultrafast pumpprobe spectroscopy. Opt. Express 18, 20645–20650 (2010).
 34.
Liu, J. et al. Ultrafast timeresolved pumpprobe spectroscopy of PYP by a sub8 fs pulse laser at 400 nm. J. Phys. Chem. B 117, 4818–4826 (2013).
 35.
Kobayashi, T. & Kidaw, Y. Ultrafast spectroscopy with sub10 fs deepultraviolet pulses. Phys. Chem. Chem. Phys. 14, 6200–6210 (2012).
 36.
Gueye, M., Nillon, J., Cregut, O. & Léonard, J. Broadband UVVis vibrational coherence spectrometer based on a hollow fiber compressor,. Rev. Sci. Instrum. 87, 093109 (2016).
 37.
Paolino, M. et al. Design, synthesis, and dynamics of a green fluorescent protein fluorophore mimic with an ultrafast switching function. J. Am. Chem. Soc. 138, 9807–9825 (2016).
 38.
GonÃ§alves, C. S. et al. A dualcolour architecture for pumpprobe spectroscopy of ultrafast magnetization dynamics in the sub10femtosecond range. Sci. Rep. 6, 22872 (2016).
 39.
Chang, H.T. et al. Simultaneous generation of sub5femtosecond 400 nm and 800 nm pulses for attosecond extreme ultraviolet pumpprobe spectroscopy. Opt. Lett. 41, 5365–5368 (2016).
 40.
Guénot, D. et al. Relativistic electron beams driven by kHz singlecycle light pulses. Nat. Photonics 11, 293–296 (2017).
 41.
Miranda, M., Fordell, T., Arnold, C., L’Huillier, A. & Crespo, H. Simultaneous compression and characterization of ultrashort laser pulses using chirped mirrors and glass wedges. Opt. Express 20, 688–697 (2012).
 42.
Louisy, M. et al. Gating attosecond pulses in a noncollinear geometry. Optica 2, 563–566 (2015). (Supplementary Material).
 43.
Miranda, M. et al. Fast iterative retrieval algorithm for ultrashort pulse characterization using dispersion scan. J. Opt. Soc. Am. B 34, 1190–197 (2017).
 44.
Silva, F. et al. Simultaneous compression, characterization and phase stabilization of GWlevel 1.4 cycle VISNIR femtosecond pulses using a single dispersionscan setup. Opt. Express 22, 10181–10190 (2014).
 45.
Fabris, D. et al. Singleshot implementation of dispersionscan for the characterization of ultrashort laser pulses. Opt. Express 23, 32803–32808 (2015).
 46.
Fan, G. et al. Hollowcorewaveguide compression of multimillijoule CEPstable 3.2μm pulses. Optica 3, 1308–1311 (2016).
 47.
Alonso, B. et al. Characterization of subtwocycle pulses from a hollowcore fiber compressor in the spatiotemporal and spatiospectral domains. Appl. Phys. B 112, 105–114 (2013).
 48.
Bohle, F. et al. Compression of CEPstable multimJ laser pulses down to 4 fs in long hollow fibers. Las. Phys. Lett. 11, 095401 (2014).
 49.
Heyl, C. M. et al. Scaleinvariant nonlinear optics in gases. Optica 3, 75–81 (2016).
 50.
Tajalli, A. et al. Fewcycle optical pulse characterization via crosspolarized wave generation dispersion scan technique. Opt. Lett. 41, 5246–5249 (2016).
 51.
Schmidt, B. E. et al. Compression of 1.8μm laser pulses to sub two optical cycles with bulk material. Appl. Phys. Lett. 96, 121109 (2010).
 52.
Huang, Z. et al. Design of intense 1.5cycle pulses generation at 3.6 μm through a pressure gradient hollowcore fiber. Opt. Express 24, 9280–9287 (2016).
 53.
Agrawal, G. P. Nonlinear Fiber Optics 3rd ed. (Academic Press, San Diego, 2001).
 54.
Marcatili, E. & Schmeltzer, R. Hollow metallic and dielectric waveguides for long distance optical transmission and lasers. Bell Syst. Tech. J. 43, 1783–1809 (1964).
 55.
Heyl, C. M., Arnold, C. L., Couairon, A. & L’Huillier, A. Introduction to macroscopic power scaling principles for highorder harmonic generation. J. Phys. B: At. Mol. Opt. Phys. 50, 013001 (2017).
 56.
Lange, H. R. et al. Anomalous longrange propagation of femtosecond laser pulses through air: moving focus or pulse selfguiding? Opt. Lett 23, 120–122 (1996).
 57.
Couairon, A. & Mysyrowicz, A. Femtosecond filamentation in transparent media. Phys. Rep. 441, 47–189 (2007).
 58.
Granados, E., Chen, L., Lai, C., Hong, K. & Kartner, F. Wavelength scaling of optimal hollowcore fiber compressors in the singlecycle limit. Opt. Express 20, 9099–9108 (2012).
 59.
GuizarSicairos, M. & GutierrezVega, J. C. Computation of quasidiscrete Hankel transforms of integer order for propagating optical wave fields. J. Opt. Soc. Am. A 21, 53 (2004).
 60.
Peremolov, A. M., Popov, V. S. & Terentev, M. V. Ionization of atoms in an alternating electric field. Sov. Phys. JETP 23, 924 (1966).
Acknowledgements
We acknowledge funding from the following institutions: Junta de Castilla y León (Projects SA116U13, SA046U16); MINECO (FIS201344174P, FIS201571933REDT, FIS201675652P, FIS201787970R); Fundação para a Ciência e a Tecnologia (FCT), Portugal, (Grants ‘UltraGraf’ MERANET2/0002/2016, UID/NAN/50024/2013, NORTE070124FEDER000070, NORTE010145FEDER022096 and the Network of Extreme Conditions Laboratories  NECL); Consejo Nacional de Ciencia y Tecnología, México (CONACYT México for Postdoctoral Research Fellowships). CLPU is acknowledged for granting access to its facilities.
Author information
Affiliations
Contributions
H.C. noticed the universality of the observation. E.C.J. and J.S.R. performed the simulations. I.J.S., F.S. R.R., W.H. and M.A.G. conducted the experiments. E.C.J., J.S.R., B.A., I.J.S. and H.C. analyzed the results. All authors reviewed the manuscript.
Corresponding author
Ethics declarations
Competing Interests
The authors declare that they have no competing interests.
Additional information
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
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
Conejero Jarque, E., San Roman, J., Silva, F. et al. Universal route to optimal few to singlecycle pulse generation in hollowcore fiber compressors. Sci Rep 8, 2256 (2018). https://doi.org/10.1038/s41598018205801
Received:
Accepted:
Published:
Further reading

Relativisticintensity nearsinglecycle light waveforms at kHz repetition rate
Light: Science & Applications (2020)

Detection and elimination of pulse train instabilities in broadband fibre lasers using dispersion scan
Scientific Reports (2020)

Postcompression of picosecond pulses into the fewcycle regime
Optics Letters (2020)

Influence of the spatial confinement on the selffocusing of ultrashort pulses in hollowcore fibers
Scientific Reports (2019)

Hollowcore fiber compression of a commercial Yb:KGW laser amplifier
Journal of the Optical Society of America B (2019)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.