1. Introduction

Femtosecond (fs) laser sources operating in the middle-infrared spectral range (MIR, 2–20 µm, 500–5000 cm⁻¹) are essential for a number of important applications. On the one hand, MIR frequency combs with high average power, high coherence, and a broad spectral span are crucial for the applications related to spectroscopy, sensing, and imaging in the molecular fingerprint region. On the other hand, availability of few-optical-cycle MIR sources with high pulse energy is of particular importance for strong-field physics and attoscience, e.g. high harmonic generation, x-ray generation, particle acceleration, time-resolved studies in physics and chemistry. Real-life applications also require compactness, robustness, and cost efficiency of the laser sources. Standard approaches to generation of few-cycle MIR pulses include down-conversion of well-established near-IR fs lasers via three-wave mixing [1], and pulse energy scaling in MIR optical parametric amplifiers (OPAs, OPCPAs) [2–4].

MIR lasers based on transition-metal-doped II-VI semiconductors (TM:II-VI) provide direct access to the spectral range 1.8–6 µm and represent an appealing alternative to complex and inefficient down-conversion of near-IR lasers. TM:II-VI materials were introduced as a new class of gain media in the late 1990s [5]. Features of II-VI semiconductor hosts (wide band gap, low phonon cutoff, broad transmission range, and tetrahedral coordination) are very favorable for doping by TM ions. Chemically stable divalent TM dopant ions provide the ‘right’ multiplet structure for broadly tunable MIR lasers, including broad absorption and emission bands, high cross-sections, and the absence of excited state absorption.

ZnS and ZnSe doped with Cr²⁺ and Fe²⁺ are typical and the best known representatives of the large TM:II-VI family, as reviewed in [6]. Advantages of Cr:ZnS and Cr:ZnSe lasers include room-temperature (RT) operation with close to 100% quantum efficiency, very broad tuning in the range of 1.9 – 3.4 µm, and convenient pumping by reliable erbium and thulium fiber lasers (EDFLs, TDFLs). In many respects, Fe:ZnS and Fe:ZnSe lasers are complimentary to Cr-based sources. They are pumped in the range 2.5–3.3 µm (e.g., by Cr:ZnS and Cr:ZnSe lasers or by Er:YAG lasers at 2.94 µm) and tunable over 3.4–5 µm range; they operate at RT in the nanosecond (ns) pulsed regime and require cooling to about 150 K in the continuous wave (cw) regime. An important feature of the Fe:ZnSe gain medium is excellent energy storage capability at cryogenic temperatures (55 µs luminescence lifetime at 77 K), which enables cost efficient, high energy MIR amplifiers. Recent achievements in TM:II-VI laser technology include room temperature (RT) ns pulsed Fe:ZnSe laser at 4.1 µm wavelength with 1.2 J pulse energy [7], RT cw Cr:ZnSe laser at 2.4 µm wavelength with a power in excess of 140 W at 62% pump conversion efficiency [8], GW-level fs Fe:ZnSe amplifier [9], and a kHz Cr:ZnSe based femtosecond laser system with 1 mJ pulse energy and a peak power of 5 GW [10].

2. Laser and nonlinear properties of polycrystalline Cr:ZnS and Cr:ZnSe

Broad emission bands of Cr:ZnS and Cr:ZnSe are favorable for a generation of ultra-short optical pulses; these materials are often referred to as the “Ti:sapphire of the middle-infrared”. The materials are superior to Ti:sapphire in terms of higher peak emission cross-section (13 × 10⁻¹⁹ cm² vs. 4 × 10⁻¹⁹ cm²) and longer RT lifetime (6 µs vs. 3 µs). On the other hand, they have inferior thermal optical parameters: ZnS, e.g., has a smaller RT thermal conductivity (27 W⋅m⁻¹⋅K⁻¹ vs. 35 W⋅m⁻¹⋅K⁻¹) and significantly higher RT temperature derivative of refractive index (40 × 10⁻⁶ K⁻¹ vs. 12 × 10⁻⁶ K⁻¹).

Absorption and emission spectra of Cr:ZnS and Cr:ZnSe are illustrated in Fig. 1. Broad absorption bands of the materials are conveniently located for optical pumping by a variety of readily available cw and pulsed lasers, e.g. fiber lasers (EDFLs at 1.5 µm, TDFLs at 1.9 µm) and Q-switched solid state lasers (e.g. Er:YAG at 1.65 µm, Ho:YAG and Ho:YLF at 2 µm).

 

Fig. 1 RT absorption (black) and emission (red) cross-sections of ZnS and ZnSe doped with Cr²⁺ ions (dashed and solid curves respectively). Black vertical arrows show the standard schemes of optical pumping of Cr:ZnS and Cr:ZnSe lasers.

Download Full Size | PDF

Cr (and Fe) doped ZnS and ZnSe have very similar spectroscopic and laser parameters. The choice between ZnS and ZnSe hosts is usually defined by the specific requirements to the laser source. Both materials are available in single crystal and in polycrystalline forms. Single-crystals of high optical quality and sufficiently high dopant concentrations are difficult to grow. The technology of post-growth thermal diffusion doping of polycrystalline materials [11] has enabled the fabrication of laser gain elements with high dopant concentration, uniform dopant distribution, and low losses. This technology is scalable and quantitative, i.e. it allows for mass-production of large-size gain elements with pre-assigned parameters (see, e.g., Fig. 2c).

 

Fig. 2 (a) Microstructure of polycrystalline Cr:ZnS with ~30 µm average size of the grain (the sample is optimized for high SHG yield, the batch was annealed at 950 °C during 2 weeks); (b) microstructure of polycrystalline Cr:ZnSe with ~500 µm average size of the grain (the sample is designed of suppression of the up-conversion via three-wave mixing, the batch was annealed at 1000 °C during 3 weeks); (c) state of the art in Cr:ZnS and Cr:ZnSe fabrication by post-growth thermal diffusion doping: large-size, uniformly doped polycrystalline gain elements for high power spinning ring MIR lasers Ø50 × 6 mm Cr:ZnSe (top) and Ø 50 × 5 mm Cr:ZnS (bottom)

Download Full Size | PDF

Cr:ZnS and Cr:ZnSe uniquely combine superb, ultra-fast laser capabilities with high nonlinearity of zinc-blende semiconductors. Second-order nonlinear susceptibility of ZnSe (d_eff ≈16–19 pm/V at 2.4 µm) is higher than that of lithium niobate (15 pm/V) [12, 13]. Compared to ZnSe, ZnS has lower nonlinearity (about 4 pm/V). However, physical properties of this material, e.g. reduced thermal-optical effects and wider bandgap (3.9 eV vs. 2.7 eV), are advantageous in high average and peak power laser applications.

Polycrystalline ZnS and ZnSe consist of a multitude of microscopic single-crystal grains. The broad distribution of grain sizes and orientations allows for three-wave mixing via random quasi-phase-matching process (RQPM). The main distinctive features of RQPM — a linear dependence of the conversion yield with the medium length, an ultra-wide bandwidth — were predicted in [14, 15] and then confirmed in the experiment [16]. RQPM is less efficient than conventional QPM process in cw and ns pulsed laser regimes. However, the technique is well suited for frequency conversion of fs pulses with a high peak power and a broad spectrum.

An important prerequisite for efficient three-wave mixing via RQPM process is the optimized microstructure of the material: the average size of the grain in polycrystalline medium should be of the order of the coherence length of three-wave mixing [14, 15]. For instance, 2400 nm → 1200 nm second harmonic generation (SHG) would be most efficient at the average grain size of about 20 µm for ZnSe, and about 30 µm for ZnS. Fabrication of Cr:ZnS and Cr:ZnSe laser gain elements using post-growth thermal diffusion doping technology retains the polycrystalline zinc-blend structure of the material. Furthermore, the technology allows, to some extent, to manipulate the microstructure of the polycrystalline material including the average size of the grain. Doping of polycrystalline II-VI materials by transition metals occurs under a vacuum in sealed ampoules, which are annealed at about 1000 °C, as described in [11]. Comparisons of the materials’ microstructure before and after doping have shown that annealing results in an increase of the average size of the grain. Importantly, the diffusion of the dopant and the growth of the grains occur at different rates and have different temperature dependences. Thus, the parameters of the gain element can be tailored in favor of a certain type of three-wave mixing by a proper choice of starting material grain size, the annealing time and temperature.

The Cr:ZnS and Cr:ZnSe gain elements were prepared as follows: undoped samples have been gown by chemical vapor deposition (CVD). Several batches of samples were annealed at different temperatures (900–1000 °C) and during different time (from several days to several weeks). After annealing, a sample from each batch was chemically etched and its microstructure was evaluated using a microscope.

Examples of the microstructures of polycrystalline Cr:ZnS and Cr:ZnSe are shown in Fig. 2. Cr:ZnS gain elements with ~30 µm grain size (as in sample a) are used in fs oscillators with enhanced SHG output. Cr:ZnSe gain elements with ~500 µm grain size (as in sample b) are used in ultrafast amplifiers with suppressed tree-wave mixing. Part (c) of the figure shows large size polycrystalline ring-shaped gain elements for high power MIR lasers.

An effect of SHG via RQPM in polycrystalline Cr:ZnS and Cr:ZnSe was first observed in the gain elements of SESAM mode-locked lasers [17, 18]. Reported SHG signals were at sub-mW level (inside the resonator) or about 10⁻⁴ of the main signal. Kerr-lens mode-locking in polycrystalline Cr:ZnS and Cr:ZnSe, the use of optimized gain elements, and the improvements in the design of the laser resonators has resulted in sub-W SHG levels, as reviewed in [19]. Most current results on RQPM in polycrystalline Cr:ZnS will be presented in Section 3.

Nonlinear refractive indices of ZnS and ZnSe (n₂ ≈10⁻¹⁴ cm²/W at 2.4 µm wavelength) are two orders of magnitude higher than that of sapphire. The critical power for self-focusing in these materials is proportionally low (~400 kW at the wavelength 2.4 µm). The formation of filaments has been recently observed in bulk ZnSe and ZnS under pumping by MIR fs pulses. It was shown that the higher multiplicity of the multi-photon ionization is very important for the formation of filaments in ZnSe [20]. Filamentation occurred in three-photon ionization regime but was easily perturbed by external conditions. Filaments with greater spatial stability were obtained at higher orders of multi-photon ionization. The spectral broadening of about 450 nm was observed in ZnSe at 2 µm pump wavelength in five-photon regime. On the other hand, a 3-octave-spanning continuum has been obtained from filaments in ZnS pumped at 2.1 µm (seven-photon regime) by a kilohertz OPCPA at about 1000 times the critical power for self-focusing [21].

Currently reached levels of peak power inside the gain elements of polycrystalline Cr:ZnS and Cr:ZnSe fs oscillators and amplifiers are high enough for pronounced effects arising from χ⁽³⁾. Spectral broadening in a Kerr-lens mode-locked oscillator will be discussed in Section 3. Recent results on spectral broadening in full repetition rate Cr:ZnS amplifiers (0.1 µJ pulses at 80 MHz) and on an octave-spanning supercontinuum generation in a ns pulse pumped Cr:ZnSe amplifier (10 µJ pulses at 1 kHz) are presented in Section 4.

3. Few-optical-cycle Kerr-lens mode-locked Cr:ZnS and Cr:ZnSe oscillators

Availability of high power fiber lasers for optical pumping and convenience of RT operation have stimulated rapid progress of ultrafast Cr:ZnS and Cr:ZnSe lasers. Femtosecond lasers based on all major mode-locking techniques have been implemented over the past decade including SESAM [22], Kerr-lens [23], and graphene [24] mode-locked lasers, as reviewed in [25]. Recent demonstration of Kerr-lens mode-locked lasers based on polycrystalline Cr:ZnS and Cr:ZnSe [26] has led to significant improvements in the output parameters of ultrafast MIR oscillators in terms of average power, pulse energy, and pulse duration.

Generic scheme of ultrafast Cr:ZnS (Cr:ZnSe) oscillator is shown in Fig. 3. We rely on Kerr-lens mode-locking technique as it allows for shorter pulses, higher power in comparison with, e.g., SESAM mode-locked oscillators. We use conventional X-folded resonators with unconventional normal incidence mounting of the AR coated gain element. The polycrystalline gain element is 4–9 mm long, is optically pumped by off-the-shelf EDFL, and is cooled with room-temperature water. The gain elements are plane-parallel cut (non-parallelism <1°). Typical low-signal transmission of the gain elements at the pump wavelength is 5–20%.

 

Fig. 3 Generic design of Kerr-lens mode-locked polycrystalline Cr:ZnS (ZnSe) oscillator (not to scale). EDFL, pump laser at 1550–1567nm; L, pump focusing lens; Cr:ZnS(ZnSe), AR coated polycrystalline gain element at normal incidence; HR, high reflectors; OC, output coupler; HR*, optional folding mirrors. All optical coatings are dispersion-controlled. SHG signal is generated in the gain element via RQPM process and separated by an optional dichroic mirror DM.

Download Full Size | PDF

The resonator includes two curved high reflectors (HR) with radii r, plane end mirror and an output coupler for fundamental MIR radiation (OC). The resonator’s legs are unequal with a typical ratio 2:5 and the OC being installed in a longer leg. The angle of incidence at the curved mirrors is minimized to reduce the astigmatism of the resonator. An optional SHG output can be implemented via curved dichroic mirror (DM) with high transmission in SHG range, as shown in Fig. 3. Optional folding mirrors HR* can be introduced in the resonator in order to reduce its foot-print (e.g. an oscillator at 79 MHz repetition rate fits into a 125 × 375 mm² rectangle). We avoid the use of prisms or plates for dispersion control and instead tailor the chromatic dispersion of the optical coatings. We control second and compensate third order dispersion of the resonator within a third of an octave. Dispersion metrology in 2–3 µm range was not available. Dispersions of the gain elements and of the optical coatings were evaluated using the standard Sellmeier equations (for undoped ZnS and ZnSe) and the theoretical data provided by the coaters. Thus, our estimates of resonators’ net dispersion are approximate.

We first optimize the laser for maximum cw output power. The distance between the curved mirrors is then fine-adjusted in order to enable the Kerr-lens mode-locking (initiated by OC translation). We usually operate the oscillators at 2.3–2.4 µm central wavelength near the maximum of the tuning curve; standard techniques, e.g. a birefringent plate, can be employed for wavelength tuning of the fs oscillator (see, e.g., [6]).

Main distinctive feature of this design of Kerr-lens mode-locked laser is the normal incidence mounting of the gain element, which provides (i) better management of the thermal optical effects due to circularity of the pump and laser beams, (ii) a significant increase of the pump and laser intensity inside the gain element, (iii) greater convenience when using gain elements with a large length and, hence, high pump absorption. Our experiments show that uncompensated astigmatism of the resonator is not an impediment for Kerr-lens mode-locking of Cr:ZnS and Cr:ZnSe lasers (see, e.g., [28]). Circular (or close-to-circular) output beam profiles can be obtained in most cases. The resonator does not include polarizing components; the degeneracy of two orthogonal polarizations is lifted by the resonator’s astigmatism. Therefore, the output of Kerr-lens mode-locked laser can be either p- or s- polarized, depending on the alignment of the resonator. Furthermore, the oscillator can be fine-tuned for simultaneous generation of two orthogonally polarized pulse trains at the fundamental repetition rate.

The developed design of the oscillator is flexible and allows addressing a wide range of laser parameters. We obtained stable Kerr-lens mode locked laser oscillations at fundamental repetition rates ranging from 69 MHz to 1.2 GHz (resonator’s lengths from 2.2 m to 125 mm, radii of curved mirrors r = 15–150 mm). Optimizations of the resonator at particular pulse repetition rate (f_R) include (i) fine-tuning of the resonator’s net group delay dispersion (GDD), (ii) optimization of OC’s reflectivity (R_OC), (iii) fine-tuning of pump power and pump focusing in the gain element, as described in [27]. Parameters of several selected laser configurations are summarized in Table 1 and correspond to multi-hour operation in Kerr-lens mode locked regime at a fundamental repetition rate and with linear polarization. To the best of our knowledge, those sets of parameters are record-high for fs MIR oscillators in the spectral range of 2–3 µm.

Table 1. Parameters of fs polycrystalline Cr:ZnS oscillators near 2.4 µm central wavelength^a

View Table | View all tables in this article

As can be seen from the table, Kerr-lens mode-locked polycrystalline Cr:ZnS oscillators can be optimized for rather unusual regimes, e.g. for a peak power, which exceeds the critical power for self-focusing in the gain medium (line 2 in Table 1), and (or) for efficient SHG of fs pulses directly in the gain medium (line 3 in Table 1).

Effects of high peak power on the parameters of Kerr-lens mode-locked laser are illustrated in Fig. 4. This laser was equipped with 5-mm long polycrystalline Cr:ZnS gain element (11% low-signal pump transmission). Repetition rate of the resonator was set to f_R = 84 MHz. The laser was then optimized for shortest pulse duration following the routine described in [27]. The broadest spectra were obtained at estimated net GDD of about −125 ± 50 fs². The spectra of pulses significantly depend on the reflectivity of output coupler R_OC. A smooth and uniform spectrum with 290 nm (16.1 THz) FWHM bandwidth was obtained at R_OC = 40% (see dashed black curve in Fig. 4(a)); 21 nJ output pulse energy allows us to estimate 35 nJ inside the resonator, which corresponds to 1.5 MW intracavity peak power assuming time-bandwidth product 0.32. Thus, the ratio of peak power to the critical power κ = P_Pk/P_Crit ≈3.75. Replacement of the OC by another one with higher reflectivity R_OC = 90% has decreased output pulse energy to 6.3 nJ but increased intracavity energy to about 63 nJ. So we can estimate a proportional increase of intracavity peak power and of the parameter κ (3 MW, 7.3, respectively). A measured spectrum of pulses at R_OC = 90% is shown in Fig. 4(a) by solid black curve. As can be seen, a twofold increase of the parameter κ has resulted in a strong broadening of pulses’ spectrum to 510 nm (31 THz) FWHM. Significant optical signal at 2.6–2.8 µm suggests that even broader spectrum can be obtained by purging of the optical resonator.

 

Fig. 4 Spectra of pulses (a, top) and autocorrelations (b, c, bottom) of Cr:ZnS oscillator at f_R = 84 MHz repetition rate. Two different output couplers (R_OC = 90, 40%) provide different peak power levels inside the resonator (~2.5 MW and ~1.5 MW respectively). Measured spectrum of pulses includes: (f), fundamental MIR band; (2f), second harmonic (3f), third harmonic; (4f), fourth harmonic; (SFG), sum frequency generation between fs MIR pulses and cw pump radiation; (Pump), residual pump at 1567 nm. Near IR and visible optical signals (shown only for R_OC = 90%) are attenuated and distorted during the transmission through the resonator’s HR mirrors, 2f and 3f peaks are normalized to unity for convenience. Gray background shows transmission of 1 m standard air. Inserts show measured output beam profiles.

Download Full Size | PDF

Broad bandwidth of RQPM in the polycrystalline laser medium leads to a cascade of three-wave-mixings from fundamental MIR band to near-IR SHG (red line, 2f peak in Fig. 4a) to third and fourth optical harmonics (green line, 2f and 4f peaks). There is also a signal that corresponds to sum frequency mixing between fs MIR pulses and cw pump radiation (SFG peak).

MIR pulses as short as two optical cycles (Δτ^(S) ∼15–20 fs) can be estimated from the spectra. Such short pulses would experience significant temporal broadening during propagation through the substrates of the output couplers: 3.2 mm thick YAG (R_OC = 40%) and 3.2 mm thick ZnSe (R_OC = 90%) substrates introduce −400 fs² and + 700 fs² respectively and broaden a two-cycle pulse by 300–600% see, e.g., Eq. (2).1.20) in [29]. Output pulses were re-compressed outside the resonator, but only to some extent as will be discussed below in more detail. Autocorrelations of re-compressed pulses are illustrated in Fig. 4b and Fig. 4c.

In the next experiment we optimized Kerr-lens mode-locked laser for high SHG power. We have installed in the resonator a relatively long 9 mm polycrystalline Cr:ZnS gain element (4% initial pump transmission) with 30 µm average size of the grain (to match the coherence length of SHG). The resonator was equipped with curved dichroic mirror DM with high transmission at SHG wavelength, as shown in Fig. 3. The repetition rate of the resonator was decreased to 75 MHz. We optimized the laser for short pulse duration using the output coupler with R_OC = 60% on a 3.2 mm thick YAG substrate. The broadest spectra were obtained at estimated net GDD of about −200 ± 100 fs². A spectrum of pulses is shown in Fig. 5a. Fundamental MIR band of the spectrum features uniform peak with 380 nm (22 THz) bandwidth. The spectrum spans 1.8 – 2.8 µm at −30 dB level. Most likely, a broader spectrum can be obtained by purging of the optical resonator.

 

Fig. 5 Spectrum of pulses (a, top) and autocorrelations (b, c, bottom) of Cr:ZnS oscillator optimized for high SHG power at f_R = 75 MHz. Fundamental MIR band (f) and second harmonic (2f) are normalized to unity; gray background shows transmission of 1 m standard air. Numbers near spectra show measured power and bandwidth. Numbers near autocorrelations show GDD, TOD of the optical components outside the resonator and estimated pulse widths. Insert shows measured output beam profile.

Download Full Size | PDF

The measured power of Kerr-lens mode-locked laser was 1.1 W and 0.35 W in fundamental MIR and SHG bands respectively (1.8 W MIR power was measured in cw regime after disruption of the mode-locking). The transmission of the gain element’s AR coating at SHG wavelength is about 75%. Furthermore, SHG spectrum was truncated during separation from residual pump radiation (bandwidth of the dichroic separators was 1100 – 1350 nm). Thus, we estimate SHG power in excess of 0.46 W inside the gain element in one direction and 16% single-pass SHG conversion efficiency of MIR fs pulses in 9 mm long polycrystalline Cr:ZnS sample. SHG of fs pulses in RQPM medium provides a combination of good efficiency with broad bandwidth: the spectral spans of f and 2f peaks in Fig. 5a are the same (if counted in frequency units). SHG pulses are broadened to few hundred fs due to chromatic dispersion of the gain medium (~350 fs²/mm at 1.1 µm) and due to a large group velocity mismatch between MIR and SHG pulses (about 170 fs/mm) [19].

A uniform and smooth fundamental spectrum of the oscillator suggests low time-bandwidth product of pulses. Pulse duration Δτ^(S) ~15 fs (two optical cycles) can be estimated from the spectrum assuming Δτ⋅Δν = 0.32. Autocorrelation of output pulses after their propagation through the substrate of the OC reveals significant chirp and temporal broadening to about ~80 fs, as shown in Fig. 5b. This observation is in good agreement with an analytical formula (Eq. (2).1.20) in [29]). We de-chirped and re-compressed output pulses outside the resonator by introduction of 2 mm thick ZnSe plate with the opposite sign of GDD. The autocorrelation of re-compressed pulses is shown in Fig. 5c. Dispersion control by the plate has allowed for GDD compensation; however, the plate introduced a significant amount of third order dispersion (TOD). Therefore, output pulses were re-compressed to about 30 fs and the autocorrelation shows the residual chirp. Most likely, more sophisticated GDD and TOD control, e.g. by the use of dedicated dispersive mirrors, and purging of the laser setup will reveal sub-two-cycle MIR pulses.

To conclude this section, we demonstrate that Kerr-lens mode locked oscillators based on polycrystalline Cr:ZnS (and Cr:ZnSe) provide access to few-cycle MIR pulses with Watt-level power at very broad range of pulse repetition rates.

4. Power end energy scaling in ultrafast Cr:ZnS and Cr:ZnSe laser amplifiers

A great demand for ultrafast MIR sources with high power and pulse energy, broad spectrum, and an ultra-short few-cycle pulse width stimulates our efforts on the power scaling of MIR oscillators. Single-pass fs laser amplifiers based on polycrystalline Cr:ZnS and Cr:ZnSe, which were proposed in [26], are very appealing due to their simplicity, compactness and robustness. Recent improvements in the design of polycrystalline Cr:ZnS and Cr:ZnSe amplifiers has allowed us to implement MIR fs sources with a unique combination of parameters [30, 31]. The schematic of a single-pass amplifier is illustrated in Fig. 6. The amplifier is seeded at 2 – 3 µm by a fs master laser (MO), e.g. one of the oscillators described in Section 3. The amplifier is optically pumped at 1.5–2.1 µm by cw or pulsed laser (Er- or Tm-, doped fiber laser or a solid state laser). Depending on the application, the amplifier can be either operated at full repetition rate or include a pulse picker.

 

Fig. 6 Single-pass Cr:ZnS (Cr:ZnSe) ultrafast amplifier (not to scale): (Cr:ZnS/ZnSe), amplifier’s gain element; (Pump), cw or pulsed laser for optical pumping; (MO), fs master oscillator; (L, M, DM), combination of lenses and mirrors for input/output dispersion control beam shaping, combining, and separation. The system can include an optional pulse picker and/or an optical isolator (OI).

Download Full Size | PDF

Single-pass arrangement of the amplifier provides great robustness and, at the same time, flexibility. The system can be configured for several distinctly different regimes of operation, depending on a combination of the following parameters:

Parameter γ defines whether the amplifier is configured for high pump conversion with moderate gain (γ ≈ 1) or for high gain with low efficiency (γ >> 1). On the other hand, parameter κ gauges nonlinear interactions in the gain medium.

We evaluated the parameters of single-pass Cr:ZnS and Cr:ZnSe amplifiers in three regimes: (a) high pump conversion and low nonlinearity; (b) high pump conversion and high nonlinearity; (c) high gain and high nonlinearity. Configurations (a) and (b) were continuously pumped at 1567 nm by off-the-shelf EDFL with 20 W power; the amplifiers were equipped with 9 mm long Cr:ZnS gain elements with ~2% small signal pump transmission. We used a combination of dispersive mirrors to compensate for dispersion in the amplifier’s gain element. Other parameters of the optical setup were similar to the amplifier described in [30]. Configuration (a) was seeded by Kerr-lens mode locked Cr:ZnS oscillator at 888 MHz repetition rate, 2360 nm central wavelength with 76 fs pulses at 0.9 W average power (γ = 18, κ = 0.03); configuration (b) was seeded at 81 MHz, 2380 nm with 50 fs pulses at 1.75 W average power (γ = 13, κ = 1), see [28, 32], respectively, for more details about the seed lasers.

Output parameters of full repetition rate cw pumped fs Cr:ZnS amplifiers (a) and (b) are compared in Fig. 7. In both cases we obtained output power in excess of 7 W with close to linear dependences of output power on pump power. Thus, a very simple and robust single-pass arrangement of the amplifiers provides for high conversion efficiency of low-cost cw pump radiation at 1.57 µm to ultra-short mid-IR pulses at rather high gain (η = 37%, G = 8 and η = 27%, G = 4.6 in configurations (a) and (b), respectively).

 

Fig. 7 Parameters of full repetition rate cw pumped single-pass fs Cr:ZnS amplifiers. Configurations (a) and (b) correspond to significantly different peak powers of seed pulses. Top: Measured autocorrelations (ACs). Initial ACs (In, amplifier’s pump is off) are compared with final ACs (Out, full pump power). Numbers near ACs show estimated pulse durations. Bottom: Measured spectra of pulses. Initial spectra (blue lines) normalized to unity; final spectra (red lines) normalized to optical power; grey lines show intermediate spectra, obtained during the gradual increase of pump power (all normalized to optical power). Numbers near the spectra show output power measured without pumping (In) and at full pump power (Out).

Download Full Size | PDF

Configuration (a) with low peak power of seed pulses (κ = 0.03) provides amplification without a distortion of the spectrum of seed pulses. Autocorrelation of output pulses shows equally low distortion of the temporal parameters of the seed. On the other hand, configuration (b) with κ ≈1 enables simultaneous amplification of MIR fs pulses and their spectral broadening. Remarkably, the spectra of pulses became broader with an increase of cw pump power (e.g. an increase of the amplifier’s gain). Obtained autocorrelations show that amplification of MIR fs pulses is supplemented by a decrease in pulse duration from 50 fs to about 40 fs. Most likely, even shorter pulses can be obtained with better output dispersion control of the amplifier: one can estimate 30 fs width of amplified pulses taking into account their 250 nm (13 THz) broad spectrum and assuming, e.g., time-bandwidth product 0.4.

Obtained experimental data were processed to derive dependences of spectral broadening (B) on the amplifier’s gain (G), as shown in Fig. 8. The parameter B was defined as B(G) = 100% × (Δν(G)–Δν_In)/Δν_In, where Δν(G) is the spectral bandwidth of output pulses and Δν_In is the initial bandwidth of seed pulses (both at −20dB level). Thus, we measure the spectral broadening B in % and dB/dG derivative in % ⋅1⁻¹.

 

Fig. 8 Spectral broadening (B) in single pass fs Cr:ZnS amplifiers vs. amplifier’s gain (G) (see main text for definition of the parameter B). Configurations (a) and (b) correspond to significantly different peak powers of seed pulses.

Download Full Size | PDF

Unsurprisingly, seeding of the amplifier at low peak power (curve a, κ = 0.03) produced minuscule broadening with a steady slope dB/dG ≈0.6% ⋅1⁻¹. The increase in the peak power of seed pulses to about the critical power for self-focusing in ZnS (curve b, κ = 1) results in significantly different spectral parameters. The spectral broadening is modest at a low gain of the amplifier: dB/dG ≈1.3% ⋅1⁻¹ for G < 3. However, with further increase of the gain (i.e. with an increase of peak power inside the gain medium), the spectral broadening shoots up with the slope as high as dB/dG ≈32% ⋅1⁻¹. As a result, we obtained a 65% spectral broadening at G = 4.6 (vs. modest 10% at G = 3). Our experiment was limited by the available EDFA power (20W) and peak power of the fs master laser (400 kW). One can expect that full repetition rate Cr:ZnS and ZnSe amplifiers with an octave-spanning spectrum will be implemented soon using the now available MW master oscillators [28] and the spinning ring technology [8].

In the subsequent experiment we re-configured the amplifier for high gain and high nonlinearity. CW EDFA pump was replaced with a Q-switched Er:YAG laser at 1645 nm with 2 mJ pulse energy, 100 ns pulse duration at a 1 kHz repetition rate. We have selected a polycrystalline Cr:ZnSe material with high dopant concentration (5 × 10¹⁸ cm⁻³) and with the largest available size of the grain (500–1000 µm). The gain element was 30 mm long and Brewster-cut with <0.01% initial transmission at pump wavelength. The amplifier was seeded by the master laser from the previous experiment (f_R = 81 MHz, E_Seed = 21 nJ). The setup was configured for γ = 10⁵ and κ_In = 1. The amplifier was seeded at a full repetition rate without down picking to 1 kHz repetition rate of pump pulses. Therefore, each 100 ns long pump pulse has interacted with dozens of seed pulses from the input pulse train with 12.4 ns period of pulse repetition.

The energy of output pulses and the amplifier’s gain were evaluated using DC coupled photoelectromagnetic mid-IR detector with ns resolution, as illustrated in Fig. 9. We measured the amplitude of output pulses while the amplifier’s pump was off (waveform (a) in Fig. 9). We then estimated gain as a ratio between the amplitude of amplified pulses to the initial amplitude (waveforms (b, c) in Fig. 9). We did not take into account any effects due to saturation of the detector. We did not use an optical isolation, wherefore; the parameters of the amplifier were limited by an optical feedback between the master laser and the amplifier, which resulted in appearance of a secondary pulse train (shown in waveform (c) by arrows). Therefore, we have limited the energy of pump pulses to 2 mJ and conservatively estimate the single-pass gain as G = 500 and output pulse energy as E_Out = 10 µJ per individual pulse in the train.

 

Fig. 9 Pulse trains detected at the output of single-pass Cr:ZnSe amplifier pumped by 100 ns pulses at 1645 nm wavelength: (a), initial pulse train (amplifier’s pump is off); (b), E_Pump = 1.6 mJ; (c) E_Pump = 2.2 mJ. All waveforms were acquired using the same MIR detector; the waveforms (b) and (c) were then normalized to the amplitude of the initial signal (a). Red arrows in waveform (c) show a trace of a secondary pulse train, which appears due to an optical feedback between the fs master laser and the amplifier. Waveforms (a) and (b, c) are shown in different time-scales.

Download Full Size | PDF

Spectrum of output pulses was measured by a grating monochromator. Amplified pulses were detected and separated from the background signal at 81 MHz repetition rate by a gated integrator. The obtained spectrum of amplified pulses is shown in Fig. 10. As can be seen, the amplification of seed pulses well above the critical power for self-focusing in ZnSe results in generation of mid-IR supercontinuum (SCG) directly in the gain element of a single-pass amplifier. We did not observe significant fluctuations or degradation of the spectrum during several hours.

 

Fig. 10 Measured spectrum of pulses of single-pass fs Cr:ZnSe amplifier with ns pulsed pumping. (In), Initial spectrum of 20 nJ seed pulses (obtained then the amplifier’s pump was off); (Out) spectrum of amplified ~10 µJ pulses. The sketch on the right illustrates timing diagram of the measurement. We stitched several spectra, which were measured with different detectors and filters (see Section 6). Error bars correspond to averaging of three sets of spectra that were acquired during ~1 hour.

Download Full Size | PDF

To the best of our knowledge, this is the broadest SCG obtained to-date in bulk ZnSe. In our opinion, the use of polycrystalline Cr:ZnSe, which was specially tailored for the large size of the grain, was an important prerequisite for an octave-spanning MIR SCG. The large grains suppress SHG and SFG via RQPM in polycrystalline medium. Reduced levels of near-IR and visible emissions at second, third, and fourth harmonic wavelengths greatly reduce the probability of multi-photon ionization in ZnSe and, hence, provide for a broader continuum. For instance, significantly smaller spectral broadening (up to 450 nm) has been reported for (presumably standard CVD-grown) ZnSe pumped by an OPA with a 0.8–2.4 µm tuning range and with a 6–20 µJ pulse energy at 30 fs pulse width [20].

5. Conclusion

Polycrystalline Cr:ZnS and Cr:ZnSe offer us unique possibilities for generation and nonlinear frequency conversion of ultra-short optical pulses in the MIR range. Kerr-lens mode-locked Cr:ZnS and Cr:ZnSe oscillators now provide the shortest pulses (2–4 optical-cycles) with the highest average power (1–2 W) in the range 2–3 µm. The oscillators are conveniently pumped by off-the-shelf fiber lasers (with 20% optical-to-optical conversion efficiency) and operate in a broad range of repetition rates.

We demonstrate that random quasi phase matching in polycrystalline Cr:ZnS and Cr:ZnSe is well suited for efficient three-wave mixing of few-optical-cycle pulses directly in the gain elements of ultrafast lasers and amplifiers. The fabrication of polycrystalline Cr:ZnS and Cr:ZnSe gain elements by post-growth thermal diffusion doping allows control of the material microstructure and hence tailor its parameters in favor of a certain type of three-wave mixing.

The availability of MIR oscillators with high power auxiliary output at second harmonic wavelength (up to 0.4 W) provides simple access to the range 1–1.5 µm, which is of high interest for a number of applications. Three-wave mixing in polycrystalline medium provides a number of interesting opportunities for control and stabilization of generated MIR optical frequency combs. For instance, significant optical signals at second, third and fourth optical harmonics can be used for stabilization of the comb’s tooth spacing and its carrier-envelope offset frequency. Sum frequency mixing between MIR and pump radiations provides for referencing of generated MIR comb to the pump laser at 1.5–2 µm (e.g. stabilized cw or pulsed fiber laser). Furthermore, down-conversion via random quasi phase matching process enables MIR frequency comb generators with exceptionally broad spectral coverage. A possible approach to frequency combs in the range of 2–10 µm includes a sub-harmonic synchronously pumped OPOs based on polycrystalline ZnS and ZnSe.

We demonstrate that polycrystalline Cr:ZnS and Cr:ZnS enable ultrafast MIR amplifiers with unique output parameters. Simple and robust single-pass amplifiers allow for power scaling of few-optical-cycle pulses to multi-Watt level at repetition rates from MHz to GHz. The amplifiers feature high conversion efficiency of a low-cost fiber laser radiation 1.5–2 µm to ultra-short MIR pulses at 2–3 µm. Furthermore, the amplifier can be configured for simultaneous amplification, spectral broadening, and compression of input pulses. We expect that MIR fs sources with 100 W average power will be implemented in the near future using the spinning ring Cr:ZnS and Cr:ZnS laser technology.

We also demonstrate very compact and cost efficient MIR supercontinuum generator with 10 µJ output pulse energy and the spectrum spanning 1.8–4.5 µm. The device utilizes the same single-pass laser amplifier arrangement. Bulk polycrystalline Cr:ZnSe is optically pumped at 1.6 µm by mJ pulses of a Q-switched Er:YAG laser and is seeded by 20 nJ pulses from a few-cycle Cr:ZnS oscillator. There is little doubt that the current output characteristics of this supercontinuum generator can be further significantly improved.

Note Added in Proof: An optical parametric oscillator based on polycrystalline ZnSe has been reported during the peer review of this paper [33]. The parameters of polycrystalline ZnSe OPO (RQPM process) are in line with those of a conventional OP-GaAs OPO (QPM process) under similar pumping conditions. Thus, ZnSe, ZnS and similar materials offer a promising route to generation of few-cycle pulses and multi-octave frequency combs in the middle IR.

Disclosures

Dr. Mirov declares competing financial interests. Dr. Mirov would like to acknowledge funding support from the AF Office of Scientific Research (Award No. FA9550-13-1-0234) and DARPA contract W31P4Q-15-1-0008.

6 Appendix (methods)

The experiments were carried out in a standard lab environment at 40–60% relative air humidity. Presented results correspond to stable laser operation in Kerr-lens mode-locked regime at a fundamental repetition rate with linear polarization. Spectral parameters of the mode-locked lasers were characterized using Princeton Instruments grating monochromators, as summarized in Table 2. We did not post-process the spectra to account for spectral dependences of the diffraction efficiency of the gratings and sensitivity of the detectors. Temporal parameters of the lasers were evaluated using an interferometric autocorrelator (A•P•E GmbH). The autocorrelator was customized for measurable MIR pulse widths of 20 fs. We relied on the autocorrelator’s control software and used sech² fit of the autocorrelation functions for evaluations of the pulse duration. Output beam profiles were characterized by a bolometric camera (DataRay WinCamD-FIR2-16-HR). Femtosecond pulse trains at fundamental MIR wavelength and at second harmonic wavelength were detected by fast MCT and InGaAs photo detectors (VIGO PEM and Time-Base PD1800, respectively) and acquired by TDS7254 digital scope.

Table 2. Monochromators and detectors that were used for the acquisition of the spectra^a

View Table | View all tables in this article

Acknowledgments

The work reported here partially involves intellectual property developed at the University of Alabama at Birmingham (UAB). This intellectual property has been licensed to the IPG Photonics Corporation.

References and links

1. I. Pupeza, D. Sánchez, J. Zhang, N. Lilienfein, M. Seidel, N. Karpowicz, T. Paasch-Colberg, I. Znakovskaya, M. Pescher, W. Schweinberger, V. Pervak, E. Fill, O. Pronin, Z. Wei, F. Krausz, A. Apolonski, and J. Biegert, “High-power sub-two-cycle mid-infrared pulses at 100 MHz repetition rate,” Nat. Photonics 9(11), 721–724 (2015). [CrossRef]  

2. D. Sanchez, M. Hemmer, M. Baudisch, S. L. Cousin, K. Zawilski, P. Schunemann, O. Chalus, C. Simon-Boisson, and J. Biegert, “7 µm, ultrafast, sub-millijoule-level mid-infrared optical parametric chirped pulse amplifier pumped at 2 μm,” Optica 3(2), 147–150 (2016). [CrossRef]  

3. P. Malevich, T. Kanai, H. Hoogland, R. Holzwarth, A. Baltuška, and A. Pugžlys, “Broadband mid-infrared pulses from potassium titanyl arsenate/zinc germanium phosphate optical parametric amplifier pumped by Tm, Ho-fiber-seeded Ho:YAG chirped-pulse amplifier,” Opt. Lett. 41(5), 930–933 (2016). [CrossRef]   [PubMed]  

4. S. Wandel, M. Lin, Y. Yin, G. Xu, and I. Jovanovic, “Parametric generation and characterization of femtosecond mid-infrared pulses in ZnGeP₂,” Opt. Express 24(5), 5287–5299 (2016). [CrossRef]  

5. L. D. DeLoach, R. H. Page, G. D. Wilke, S. A. Payne, and W. F. Krupke, “Transition metal-doped zinc chalcogenides: spectroscopy and laser demonstration of a new class of gain media,” IEEE J. Quantum Electron. 32(6), 885–895 (1996). [CrossRef]  

6. S. Mirov, V. Fedorov, D. Martyshkin, I. Moskalev, M. Mirov, and S. Vasilyev, “Progress in mid-IR lasers based on Cr and Fe doped II-VI chalcogenides,” IEEE J. Sel. Top. Quantum Electron. 21(1), 1601719 (2015). [CrossRef]  

7. K. Firsov, M. Frolov, E. Gavrishchuk, S. Kazantsev, I. Kononov, Yu. Korostelin, A. Maneshkin, S. Velikanov, I. Yutkin, N. Zaretsky, and E. Zotov, “Laser on single-crystal ZnSe:Fe with high pulse radiation energy at room temperature,” Laser Phys. Lett. 13(1), 015002 (2016). [CrossRef]  

8. I. Moskalev, S. Mirov, M. Mirov, S. Vasilyev, V. Smolski, A. Zakrevskiy, and V. Gapontsev, “140 W Cr:ZnSe laser system,” Opt. Express 24(18), 21090–21104 (2016). [CrossRef]   [PubMed]  

9. F. V. Potemkin, E. A. Migal, A. V. Pushkin, A. A. Sirotkin, V. I. Kozlovsky, Yu. V. Korostelin, Yu. P. Podmar’kov, V. V. Firsov, M. P. Frolov, and V. M. Gordienko, “Mid-IR (4–5 µm) femtosecond multipass amplification of optical parametric seed pulse up to gigawatt level in Fe²⁺:ZnSe with optical pumping by a solid-state 3 µm laser,” Laser Phys. Lett. 13(12), 125403 (2016). [CrossRef]  

10. E. Slobodchikov, L. R. Chieffo, and K. F. Wall, “High peak power ultrafast Cr:ZnSe oscillator and power amplifier,” Proc. SPIE 9726, Solid State Lasers XXV: Technology and Devices, 972603 (March 16, 2016).

11. S. B. Mirov, V. V. Fedorov, I. S. Moskalev, D. Martyshkin, and C. Kim, “Progress in Cr²⁺ and Fe²⁺ doped mid-IR laser materials,” Laser Photonics Rev. 4(1), 21–41 (2010). [CrossRef]  

12. H. P. Wagner, M. Kühnelt, W. Langbein, and J. M. Hvam, “Dispersion of the second-order nonlinear susceptibility in ZnTe, ZnSe, and ZnS,” Phys. Rev. B 58(16), 10494–10501 (1998). [CrossRef]  

13. I. Shoji, T. Kondo, A. Kitamoto, M. Shirane, and R. Ito, “Absolute scale of second-order nonlinear-optical coefficients,” J. Opt. Soc. Am. B 14(9), 2268–2294 (1997). [CrossRef]  

14. E. Yu. Morozov, A. A. Kaminskii, A. S. Chirkin, and D. B. Yusupov, “Second optical harmonic generation in nonlinear crystals with a disordered domain structure,” JETP Lett. 73(12), 647–650 (2001). [CrossRef]  

15. E. Yu. Morozov and A. S. Chirkin, “Stochastic quasi-phase matching in nonlinear-optical crystals with an irregular domain structure,” Quantum Electron. 34(3), 227–232 (2004). [CrossRef]  

16. M. Baudrier-Raybaut, R. Haïdar, P. Kupecek, P. Lemasson, and E. Rosencher, “Random quasi-phase-matching in bulk polycrystalline isotropic nonlinear materials,” Nature 432(7015), 374–376 (2004). [CrossRef]   [PubMed]  

17. E. Sorokin and I. T. Sorokina, “Femtosecond Operation and Random Quasi-Phase-Matched Self-Doubling of Ceramic Cr:ZnSe Laser,” in Conference on Lasers and Electro-Optics 2010, OSA Technical Digest (CD) (Optical Society of America, 2010), paper CTuGG2. [CrossRef]  

18. E. Sorokin, N. Tolstik, and I. T. Sorokina, “Femtosecond operation and self-doubling of Cr:ZnS laser,” in Nonlinear Optics, OSA Technical Digest (CD) (Optical Society of America, 2011), paper NThC1. [CrossRef]  

19. S. Vasilyev, I. Moskalev, M. Mirov, V. Smolski, S. Mirov, and V. Gapontsev, “Mid-IR Kerr-lens mode-locked polycrystalline Cr:ZnS and Cr:ZnSe lasers with intracavity frequency conversion via random quasi-phase-matching,” Proc. SPIE 9731. Nonlinear Frequency Generation and Conversion: Materials, Devices, and Applications XV, 97310B (2016).

20. M. Durand, A. Houard, K. Lim, A. Durécu, O. Vasseur, and M. Richardson, “Study of filamentation threshold in zinc selenide,” Opt. Express 22(5), 5852–5858 (2014). [CrossRef]   [PubMed]  

21. H. Liang, P. Krogen, R. Grynko, O. Novak, C.-L. Chang, G. J. Stein, D. Weerawarne, B. Shim, F. X. Kärtner, and K.-H. Hong, “Three-octave-spanning supercontinuum generation and sub-two-cycle self-compression of mid-infrared filaments in dielectrics,” Opt. Lett. 40(6), 1069–1072 (2015). [CrossRef]   [PubMed]  

22. E. Sorokin, N. Tolstik, K. I. Schaffers, and I. T. Sorokina, “Femtosecond SESAM-modelocked Cr:ZnS laser,” Opt. Express 20(27), 28947–28952 (2012). [CrossRef]   [PubMed]  

23. M. N. Cizmeciyan, H. Cankaya, A. Kurt, and A. Sennaroglu, “Kerr-lens mode-locked femtosecond Cr(²⁺⁾:ZnSe laser at 2420 nm,” Opt. Lett. 34(20), 3056–3058 (2009). [CrossRef]   [PubMed]  

24. M. N. Cizmeciyan, J. W. Kim, S. Bae, B. H. Hong, F. Rotermund, and A. Sennaroglu, “Graphene mode-locked femtosecond Cr:ZnSe laser at 2500 nm,” Opt. Lett. 38(3), 341–343 (2013). [CrossRef]   [PubMed]  

25. I. T. Sorokina and E. Sorokin, “Femtosecond Cr²⁺-based lasers,” IEEE J. Sel. Top. Quantum Electron. 21(1), 1601519 (2015). [CrossRef]  

26. S. Vasilyev, M. Mirov, and V. Gapontsev, “Kerr-lens mode-locked femtosecond polycrystalline Cr²⁺:ZnS and Cr²⁺:ZnSe lasers,” Opt. Express 22(5), 5118–5123 (2014). [CrossRef]   [PubMed]  

27. S. Vasilyev, I. Moskalev, M. Mirov, S. Mirov, and V. Gapontsev, “Three optical cycle mid-IR Kerr-lens mode-locked polycrystalline Cr(²⁺⁾:ZnS laser,” Opt. Lett. 40(21), 5054–5057 (2015). [CrossRef]   [PubMed]  

28. S. Vasilyev, I. Moskalev, M. Mirov, V. Smolski, S. Mirov, and V. Gapontsev, “Kerr-Lens Mode-Locked Middle IR Polycrystalline Cr:ZnS Laser with a Repetition Rate 1.2 GHz,” in Lasers Congress 2016 (ASSL, LSC, LAC), OSA Technical Digest (online) (Optical Society of America, 2016), paper AW1A.2. [CrossRef]  

29. U. Keller, “2.1 Ultrafast solid-state lasers,” in Laser Physics and Applications, G. Herziger, H. Weber, R. Poprawe, ed. (Springer Materials, 2007).

30. S. Vasilyev, I. Moskalev, M. Mirov, S. Mirov, and V. Gapontsev, “Multi-Watt mid-IR femtosecond polycrystalline Cr(²⁺⁾:ZnS and Cr(²⁺⁾:ZnSe laser amplifiers with the spectrum spanning 2.0-2.6 µm,” Opt. Express 24(2), 1616–1623 (2016). [CrossRef]   [PubMed]  

31. S. Vasilyev, I. Moskalev, M. Mirov, S. Mirov, and V. Gapontsev, “7 W Few-Cycle Mid-Infrared Laser Source at 79 MHz Repetition Rate,” in High-Brightness Sources and Light-Driven Interactions, OSA technical Digest (online) (Optical Society of America, 2016), paper MT2C.4. [CrossRef]  

32. S. Vasilyev, M. Mirov, and V. Gapontsev, “Mid-IR Kerr-Lens Mode-Locked Polycrystalline Cr²⁺:ZnS Laser with 0.5 MW Peak Power,” in Advanced Solid State Lasers, OSA Technical Digest (online) (Optical Society of America, 2015), paper AW4A.3.

33. Q. Ru, N. Lee, X. Chen, K. Zhong, G. Tsoy, M. Mirov, S. Vasilyev, S. Mirov, and K. Vodopyanov, “Optical parametric oscillation in a random polycrystalline medium,” Optica 4(6), 617–618 (2017). [CrossRef]