Powerful femtosecond lasers in the short-wave or mid-wave infrared (SW/MWIR) range are in high demand for many scientific applications, such as the generation of ultrafast incoherent/coherent X-rays [1,2], molecular spectroscopy [3], and particle acceleration [4]. They can also be used as the driving sources to generate long-wave infrared (LWIR) lasers [5] or THz radiation [6].

Optical parametric chirped pulse amplification (OPCPA) [7] pumped by near-infrared (NIR) lasers is the most commonly used method for obtaining SW/MWIR lasers. So far, benefiting from the commercial, robust NIR lasers, many high-repetition-rate (1 kHz or higher) SW/MWIR sources with GW-level peak power have been built. In the low-repetition-rate operation, sub-TW level outputs can be achieved [8–10]. In the single-shot mode, a 2.2-µm, 100-TW-level OPCPA system was also designed and reported [11,12].

Chirped pulse amplification (CPA) [13] is also a competitive choice for achieving powerful lasers in the SW/MWIR range owing to its high efficiency, high stability, and low requirements for pump lasers and time synchronization accuracy. In the past two decades, transition metal doped II-VI semiconductor polycrystals, such as Cr:ZnS/Cr:ZnSe (2.4 µm) and Fe:ZnS/Fe:ZnSe (4 µm), have attracted wide attention. These materials, with broad emission bands (${\gt}{800}\;{\rm nm}$, full width at half maximum) and high emission cross-sections (${\gt}{{10}^{- 18}}\;{{\rm cm}^2}$) [14,15], are very suitable for the generation of high-peak-power pulses in a CPA system. Besides, through the thermal diffusion doping (TDD) technique, a large-sized polycrystalline Cr:ZnS/ZnSe element can be conveniently achieved, which is necessary to support a high-energy amplified output. Since the first 2.4-µm CPA laser was reported [16], many CPA lasers based on Cr:ZnSe polycrystals have been constructed [17–21]. These lasers all operated at a high repetition rate (1 kHz) but with low-energy outputs and low peak power. So far, there has yet to be a report for high-peak-power lasers based on Cr:ZnS or Cr:ZnSe. In addition, using ${\rm Fe}^{2+}$ ion doping ZnSe or CdSe material, a 3.5-mJ, 150-fs, 10-Hz output at 4.4 µm [22] and a ${\gt}{1 {\rm -}\rm mJ}$ amplified output at 5.0 µm [23] were realized recently. In the LWIR range, several TW-level ${{\rm CO}_2}$ systems have also been built based on direct amplification [24] or the CPA scheme [25].

In this Letter, we demonstrate a TW-level system at 2.4 µm. In the amplifiers, we used combined Cr:ZnS elements to enhance the absorption of the pump energy. In addition, transverse parasitic lasing (TPL) inside the Cr:ZnS elements was suppressed using an ink-cladding technique. Through two stages of amplification, the pulse was amplified to over 200 mJ. After compression, we obtained an output of 147.1 mJ with a duration of 127.6 fs, corresponding to a peak power of 0.95 TW.

We used the Cr:ZnS material as the gain medium in the CPA system. Compared to ZnSe, ZnS has many advantages, such as lower nonlinear coefficients, higher thermal conductivity, and stronger mechanical characteristics [14,15]. Moreover, since the absorption band of the Cr:ZnS is near 1.6 µm, we can conveniently obtain a high-energy pump laser through the noncritical phase matching KTP-based optical parametric oscillator (OPO) pumped by the Nd:YAG laser. However, the diffusion coefficient of ${{\rm Cr}^{2 +}}$ ions in ZnS is considerably lower than that in ZnSe [26], which means that more time is required to prepare the Cr:ZnS samples, and it is difficult to obtain uniform longitudinal distribution inside the material. Therefore, we extend the diffusion time of the material to reduce the concentration gradient inside the material. In our system, all the Cr:ZnS elements were produced in a 960°C vacuum environment for two months by using the TDD method.

Figure 1 shows the scheme of the 2.4-µm CPA system. Our system began with a 2.4-µm optical parametric amplification (OPA) source. After passing through a grating stretcher, the pulse was amplified in a three-pass Cr:ZnS pre-amplifier and a two-pass Cr:ZnS main amplifier, and was compressed in a two-grating compressor. The pump lasers of the two amplifiers were provided by three home-built 1-Hz, 1.57-µm, KTP-based OPO sources (see Supplement 1, Section 2 for details).

 

Fig. 1. Schematic layout of the TW-level 2.4-µm laser system. PM, plate mirror; CM, concave mirror; RM1, RM2, roof mirrors; G0, G1, G2, gratings; BS, 1.57-µm beam splitter (${\sim}{50}:{50}$).

Download Full Size | PDF

In the OPA device, we used a 1-kHz, 790-nm, 35-fs Ti:sapphire amplifier as the pump laser. It could provide a 150-µJ, 2.4-µm output (see Supplement 1, Section 1 for details). After passing through a ${2} \times$ expander, the beam was sent into a grating-based stretcher. The stretcher included a gold-coated concave mirror with a radius of curvature of 1800 mm, a gold-coated plane mirror, a 600-lines/mm gold grating, and a roof mirror. The distance between the grating and its image was about 580 mm, and the incident angle was about 52°, giving a chirped ratio of 5.5 ps/nm.

An Nd:YAG-laser-pumped, 1.57-µm, KTP-based OPO was used as the pump of the pre-amplifier. Owing to the multimode feature of the Nd:YAG laser and the short-cavity structure of the OPO, the generated 1.57-µm beam was highly divergent, which not only degraded the beam homogeneity on the gain medium but also led to high energy loss during the beam propagation. Here, we used a set of image relay lenses to eliminate spatial inhomogeneities and ensure the flat-top beam profiles near the gain medium, thereby reducing the risk of damage to material surfaces. The image relay setup also helped to collect the pumping energy effectively. The pump laser was split into two beams to reduce the pump intensity on the gain medium surfaces, and the total pump energy was 102.9 mJ. The beam size on the gain medium of the two pump arms was set to about 5 mm in diameter. Considering that the output at the exit of the OPO was 118 mJ, the transmission efficiency of the pump laser was 87.8%, which was primarily limited by the size of the optical elements used for the beam propagation.

 

Fig. 2. (a) Single-pass gain measurements of the Cr:ZnS elements in the pre-amplifier. Blue upward triangle: Cr:ZnS A (89.7% doping); red downward triangle: Cr:ZnS B (97.2% doping); gray square: combined Cr:ZnS elements without cladding; black square: combined Cr:ZnS elements with cladding. (b) Output energy of the pre-amplifier. Inset: beam profile of the amplified pulse.

Download Full Size | PDF

Two Cr:ZnS elements (${12}\;{\rm mm} \times {12}\;{\rm mm} \times {7}\;{\rm mm}$ in size) without cooling were used for amplification in the pre-amplifier. All the faces of the Cr:ZnS elements were anti-reflection (AR) coated with a reflection of about 2% at 1.57 µm and about 0.5% between 2250 nm and 2450 nm. The low-energy pumping energy absorptions of the Cr:ZnS elements A and B at 1.57 µm were 89.7% and 97.2%, respectively, whereas the absorptions at the full pump energy (102.9 mJ) were 68.7% and 83.7%, respectively. To enhance the absorption of the pump energy, we combined the two Cr:ZnS elements in the pre-amplifier. In this case, the absorption at the full pump energy reached 97.1% despite the absorption saturation effect under high pump intensity.

The 42-µJ stretched pulse was sent into the pre-amplifier for amplification after collimation by a beam expander. To effectively extract the energy stored in the Cr:ZnS elements, the beam of the seed from the stretcher was collimated to match the size of the pump beam. The single-pass energy loss of the seed in the combined Cr:ZnS elements was 5.0%.

First, the single-pass gain of the Cr:ZnS elements was measured. All the measurements were performed at full pump energy. Figure 2(a) shows the single-pass gain of the two Cr:ZnS elements and their combination. The maximum gain values of the Cr:ZnS elements A, B, and their combination are 3.5 (blue upward triangle), 7.4 (red downward triangle), and 14.7 (gray square), respectively. The gain is sensitive to the relative delay between the signal and pump pulse. As shown in Fig. 2(a), the gain declines sharply when the time of the pump pulse is adjusted 6 ns before the optimal timing for the highest gain, while the fluorescence lifetime of ${{\rm Cr}^{2 +}}$ ion in ZnS is about 5 µs [14,15]. It is an obvious signal of the surface TPL of the gain medium. TPL usually appears in high-energy amplification cases like in high-energy Ti:sapphire amplifiers [27]. Owing to the high emission cross-section (about ${1.0} \times {{10}^{- 18}}\;{{\rm cm}^2}$ at 2.3 µm), Cr:ZnS amplifiers can easily achieve a high single-pass gain in the amplification. However, this is not always advantageous for an amplifier since the high transverse parasitic gain is also inevitable. In particular, in the TDD-method-produced material, such as Cr:ZnS and Cr:ZnSe, the doping concentration decreases exponentially along the beam path, with the maximum concentration at the surfaces and the minimum concentration at the center [26]. This feature further increases the pump energy absorption of the surfaces, aggravating the surface TPL of this type of material. Moreover, the reflectivity of Cr:ZnS (15%) caused by of internal Fresnel reflection is also higher than that of NIR materials like Ti:sapphire (7.6%). For Fe:ZnSe material, graphite has been used in the free-running oscillator [28] and the mJ-level CPA amplifier [29] to suppress TPL. Here, we use black ink as the cladding of the Cr:ZnS elements, in which the carbon powder is the main component. The typical refractive index of the carbon black powder is 2.26 at 1064 nm [30], which can match the refractive index of the ZnS material. The carbon black powder is also an ideal absorber of 2.4-µm light. The ink was painted on the sides of the Cr:ZnS elements, and it bonded well with the Cr:ZnS elements when it was dry. As shown in Fig. 2(a) (black square), the gain curve of the combined ink-cladding Cr:ZnS elements slowly decays with the pump-signal delay. In addition, the optimal timing for the maximum gain with the cladding elements is about 6 ns ahead of the optimized time of the non-cladding Cr:ZnS elements, indicating that the energy is effectively stored in the gain medium. These results indicate that TPL is well suppressed in the cladding Cr:ZnS elements. The ink cladding led to an increase of the single-pass gain to 39.0, which was 2.65 times that of the non-cladding case.

Next, the cladding combined Cr:ZnS elements were used as the gain medium in the three-pass pre-amplifier. As shown in Fig. 2(b), the output energy increases linearly with the pump energy. The output reaches 34.9 mJ at the full pump energy. The outputs of the three passes were 1.41 mJ, 14.4 mJ, and 34.9 mJ, respectively. Considering a seed energy of 42 µJ, the gains of the three passes were 33.6, 10.2, and 2.4, respectively. The pump-signal energy conversion efficiency of the amplifier is 33.9%. The beam profile of the amplified pulse is shown in the inset of Fig. 2(b).

In the main amplifier, we combined three Cr:ZnS elements (${20}\;{\rm mm} \times {20}\;{\rm mm} \times {7}\;{\rm mm}$ in size) as the gain medium. The Cr:ZnS elements without cooling were all ink-cladding on the sides. The surfaces of the Cr:ZnS elements were coated with the same AR coating as the Cr:ZnS elements in the pre-amplifier. The single-pass low-energy pump absorptions of the three Cr:ZnS elements were 92.5%, 97.9%, and 86.5%, respectively. Three Cr:ZnS elements ensured the absorption of the pump energy, and the pump absorption of the combined Cr:ZnS elements remained greater than 99.5% at all pump levels in the experiment.

In the main amplifier, two 1-Hz, 1.57-µm, KTP-based OPO sources were used as the pump lasers (see Supplement 1, Section 2 for details). Through image-relay systems, the beam profiles of the two OPO outputs were imaged on the combined Cr:ZnS elements, and the beam sizes on the combined Cr:ZnS elements were adjusted to about 11 mm in diameter.

The center of the output beam from the pre-amplifier was selected and expanded to about 11 mm. The pulse was then sent into the two-pass main amplifier. The seed energy before the gain medium was 26.1 mJ, and the transmitted energy was 21.1 mJ after two passes of the combined Cr:ZnS elements. We varied the pump energy of the amplifier and measured the two-pass amplified output [Fig. 3(a)]. For each measurement, the pump-seed relative time delay was optimized to achieve the maximum output. The output increases linearly with the pump energy, and at a total pump energy of 470 mJ, the exacting energy efficiency reaches a maximum of 37.8%. Here, the extracting efficiency is still far from the theoretical quantum efficiency of the Cr:ZnS amplifier (65.4%), and is also lower than the efficiency in saturated Ti:sapphire amplifiers [27], which have a similar quantum efficiency (66.5%) to our Cr:ZnS amplifiers. The relatively low efficiency can be attributed to several energy loss factors in the Cr:ZnS amplifier, such as absorption, scattering, or surface reflection loss in the gain elements and harmonic generation usually found in femtosecond oscillators [14]. After a total pump energy of 470 mJ, the extracting energy roll-off occurs, which indicates that the TPL suppression of the cladding layer becomes invalid. At higher pump energies, we even observed the destruction of the cladding layer. Considering the output energy and efficiency, the total pump energy was set to 520 mJ. In this case, the first-pass output was 156.2 mJ, and the output energy of the two passes was 208.9 mJ, corresponding to a total extracting energy efficiency of 36.1%. The average fluence of ${0.27}\;{{\rm J/cm}^2}$ was far below the laser-induced damage threshold (LIDT) of the Cr:ZnS elements (the 1-on-1 LIDT is ${2.3}\;{{\rm J/cm}^2}$ for the 1-Hz, 600-ps, 2.4-µm pulses). Figure 3(b) shows the one-hour output energy stability measurement of the main amplifier. The output exhibits high stability, reaching an RMS value of 1.16%, which benefits from the amplification saturation and the high stability of the pump lasers. The beam profile of the output is shown in the inset of Fig. 3(b). No obvious local hot spot inside the output beam was observed. We also checked the amplified spontaneous emission (ASE) at the exit of the main amplifier by blocking the seed injection before the pre-amplifier. The ASE energy at the full-energy pump was below 20 µJ.

 

Fig. 3. (a) Extracting energy (black square) and extracting efficiency (blue square) of the main amplifier; (b) energy stability of the amplified pulse. Inset: beam profile of the amplified pulse.

Download Full Size | PDF

After passing through the main amplifier, the beam was expanded by ${1.6} \times$ and sent into the compressor. After optimizing the angle and distance of the grating pair, a second harmonic generation frequency-resolved optical gating (FROG) device was used to measure the duration of the compressed pulse. The FROG measurement is shown in Fig. 4. The FROG error is ${7} \times {{10}^{- 3}}$ on a ${512} \times {512}$ grid. The retrieved duration is 127.6 fs, whereas the Fourier-transform-limited duration of the compressed pulse is 116 fs. Accurate dispersion compensation is necessary to shorten the pulse duration [20]. The output energy of the compressor was 147.1 mJ, corresponding to a transmission efficiency of 70.4%. According to the temporal pulse shape, we obtain an output with a peak power of 0.95 TW. The ${ M}^2$ values of the compressed pulse were ${M}_{x}^{2} = {1.28}$ and ${M}_y^2 = {1.03}$ (Supplement 1, Section 3). By focusing part of the pulse energy in air, the filament-induced harmonics up to 11th could be observed, and a high-efficiency third harmonic (${\sim}{1}\%$) was also measured (see Supplement 1, Section 4 for details).

 

Fig. 4. FROG measurement of the compressed pulses. (a), (b) Measured and retrieved traces; (c) measured (red) and retrieved (black) spectra, and retrieved spectral phase (blue); (d) retrieved (black) and Fourier-transform-limited pulse (red).

Download Full Size | PDF

In conclusion, we have demonstrated a 1-TW, Cr:ZnS-based system with a repetition rate of 1 Hz. In the amplifiers, we used combined Cr:ZnS elements to obtain high absorption of the pump laser, which ensured high energy storage. TPL inside the Cr:ZnS elements was effectively suppressed by using the ink-cladding technique. Via two stages of amplification, the pulse energy was amplified to over 200 mJ. The pulse was compressed to 127.6 fs at an energy of 147.1 mJ, corresponding to a peak power output of 0.95 TW. To the best of our knowledge, this is the first solid TW-level system in the SW/MWIR range based on the CPA technique. By focusing the laser in air, up to the 11th harmonic could be observed, and a high-efficiency third harmonic (${\sim}{1}\%$) could be obtained. At present, a more powerful output is limited by the pulse amplification and the compressed duration. Since obtaining large-sized gain material is not a bottleneck, a 10-TW level or even a 100-TW level output can be expected by further improving the gain material quality, suppressing the TPL effect, and using large-sized gratings.