Introduction. Laser spectrometers based on parametric light generation allows one to carry out a wide tuning of the laser radiation frequency [1]. One of the common applications is to generate narrowband high-energy mid-infrared (mid-IR) laser pulses spectrally positioned near the absorption peaks of organic molecules such as N${_2}$O, CO${_2}$, CO, C${_2}$H${_6}$, and others [2]. Detection of these molecules can be used for remote sensing [3], gas analysis [4], breath analysis [5], and early detection of diseases [6].

An important component of such a spectrometer is the nonlinear converter of laser radiation. Nonlinear crystals used for parametric generation should have a high second-order nonlinearity, provide phase matching for conversion, and be transparent in a desired spectral range. For high-power applications they must also have a high damage threshold. The LiGaSe${_2}$ crystal possesses all the necessary characteristics for efficient nonlinear parametric conversion in the mid-IR [7]. It crystallizes in wurtzite-type orthorhombic space group Pna${2_1}$ and is characterized by a wide bandgap (3.47 eV and 3.66 eV at 300 K and 80 K, respectively) and a high optical damage threshold: 80 MW/cm${^2}$ (5.6 ns, 20 Hz, 1.064 µm), 100 MW/cm${^2}$ (6 ns, 100 Hz, 1.064 µm) [8–10]. The high damage threshold is the main advantage of the material, while the values of its nonlinear coefficients are not very high (d${_{31}}=9.9$ pm/V, d${_{24}}=7.7$ pm/V at 2.3 µm) [8]. For comparison, a widely used silver-containing analog AgGaSe${_2}$ has the nonlinear coefficient d${_{36}}$ of 39.5 pm/V at 10.6 µm [11].

LiGaSe${_2}$ is transparent in the 0.37–13.2 µm spectral range. Its average refractive index of $n=2.25$ [7] in the 2–12 µm wavelength range results in significant losses due to Fresnel reflection. Thus, it is necessary to increase the crystal’s transmittance to improve the efficiency of nonlinear conversion by reducing reflection losses.

One of the most common approaches to increase the transmittance by reducing reflection losses is to apply single- or multi-layer dielectric antireflection coatings (ARCs) to the surface of the sample. However, the task of creating such coatings with a transmittance of more than 98% in the both near-IR and mid-IR range, which are simultaneously capable of operating at large angles of incidence, is complicated and expensive. Besides, the ARCs significantly reduce the damage threshold of the material [12] which is a major issue for high-power parametric generation applications.

An alternative approach is to create antireflection microstructures (ARMs) on the surface of the sample [13,14]. ARM can be described as a system of cavities or protrusions located on the sample surface at specified or random distances from each other. For the mid-IR application the size of these features is of the order of micrometers. For wavelengths greater than $\lambda = n \cdot p$ (here $n$ is the refractive index of the material and $p$ is the period of the microstructure) the ARM behaves like a layer with a gradient refractive index, and its operation principle can be described using the effective medium theory [15]. This way the radiation is smoothly coupled into the material which leads to reduced reflection [16] and therefore, an increase in transmittance. This technique is not interferometric, thus ARMs can be used to increase the transmittance both in a wide spectral range and in a wide range of incidence angles. Besides, it has been demonstrated that ARMs do not affect the damage threshold of the material as much as the ARCs [12,17,18].

Several different methods have been developed for ARM fabrication. One of the most common methods of ARM fabrication uses lithography followed by reactive ion etching or plasma etching [19]. This method allows one to fabricate antireflection microstructures with a wide range of profiles, depths, and periods. Such ARMs can achieve high transmittance values at required wavelengths [20]. However, several complex fabrication steps are required to produce an ARM with this method, making it both expensive and time-consuming [21]. The single-pulse direct femtosecond laser ablation method requires fewer steps in comparison with others and provides good quality of ARMs, preserving relatively high throughput [21]. Moreover, it does not require any consumables, which significantly reduces fabrication costs. This method has already been tested on GaSe [22], CdSSe [23], AgClBr [24], ZnSe [25,26], ZnS [27,28], and others [29–32]. In this work we demonstrate single-pulse direct laser ablation method for ARM fabrication on the surface of LiGaSe${_2}$ crystals that can be used for parametric conversion.

Materials. LiGaSe${_2}$ crystals were grown using a modified Bridgman–Stockbarger method in a two-zone vertical thermal unit with a controlled heat exchanger, that allows a significant reduction in temperature gradients without impairing the controllability of the growing process [33]. Elemental Li (99.9%), Se (99.999%), and Ga (99.9999%) taken in a stoichiometric ratio were used as starting materials for the synthesis of the compound. The temperatures in the hot and cold zones were 890°C and 500°C, respectively, the temperature gradient in the growth zone was 2°C/cm, and the crucible was moved from the hot zone to the cold one at a speed of 0.02 mm/h.

As a result, homogeneous transparent crystal ingots up to 40 mm long and 15 mm in diameter could be obtained. For the present experimental studies, the single crystals were shaped as plane-parallel plates of $8\times 8\times 2$ mm${^3}$ in size with a certain orientation ($\theta =90^\circ, \phi =39^\circ$) corresponding to type-II (e-oe) phase matching condition of downconversion (optical parametric oscillation) from 2.1 µm into 3.7 and 5 µm. In this work, the goal was to obtain the highest possible average one-side transmittance in the 2–8 µm spectral range.

Experimental setup. For ARM fabrication a single-pulse direct femtosecond laser ablation method was used. It is based on the local material removal due to the high energy density of a single femtosecond pulse. The physics of laser ablation are described by the two-temperature model [34,35]. The energy from the laser pulse is transferred to the super-heated electron plasma induced in the volume of the substrate and then to the atomic lattice of the material. These processes take place on a time scale comparable to the pulse duration, which is not sufficient for long-range heat diffusion in crystal lattice. Thus, a large amount of accumulated energy in the microscopic volume causes ablation, which happens in the area where the laser beam intensity exceeds the material’s ablation threshold.

To fabricate the ARM, the sample was irradiated with laser pulses and precisely moved in the plane perpendicular to the axis of the laser beam. The 513 nm-wavelength pulses with a duration of 200 fs and a repetition rate of 200 kHz were focused on the sample surface. The movement speed of the sample was chosen so that the laser pulses impinge the surface at regular intervals, forming an ARM with a certain period. It is important to note that only one pulse was used to form one cavity, hence the name of the method. A more detailed description of the setup can be found in our previous works [23,24].

A Pharos Yb:KGW laser (Light Conversion, Lithuania) was used as the femtosecond pulse source for ARM fabrication. The maximum amount of available power at the selected wavelength was 1.5 W.

For sample positioning, an Aerotech ANT-90 three-axis nanopositioner (Aerotech Incorporated, USA) was used to provide an in-position stability of 2 nm and repeatability of 75 nm at velocities of up to 200 mm/s. The movement velocity of the stages was adjusted to produce ARM with the period from 0.8 to 1.2 µm (the values of the period where chosen based on the formula presented in the Introduction). A $100^{x}$ objective lens (Mitutoyo Corporation, Japan) was used to focus the laser beam on the sample’s surface. The ablation craters of the test samples were inspected visually with a microscope to provide prompt response and find a suboptimal regime, while the fine-tuning of the aforementioned parameters (average power, ARM period, focus shift) was performed based on the transmittance measurements.

The spectral characteristics of the obtained ARMs were measured with a Bruker Lumos Fourier-transform spectrometer (Bruker Corporation, USA). Each transmittance measurement was averaged over 32 scans, the aperture size during a single measurement was $100\times 100$ µm, and the spectral resolution was set to 4 cm$^{-1}$. The device does not allow for chamber evacuation, thus features of air and water vapor absorption can be observed when analyzing the data.

With the current experimental setup the smallest achievable period of the ARM is 0.8 µm, however, it is hard to get consistent results, therefore the smallest ARM period during the fine-tuning was chosen to be 1 µm. This limits the shortest wavelength of the increased transmittance to approximately 2.4 µm, as described in the Introduction.

Results and discussion. Test samples with different parameters, each $200\times 200$ µm${^2}$, were fabricated on the surface of the first LiGaSe${_2}$ crystal embedded into an aluminum frame with an aperture of 6 mm for transportation and mounting purposes. Their transmittance was measured, and the fabrication parameters of the ARM with the highest transmittance were recorded. Then two larger $2.2\times 2.2$ mm${^2}$ ARM samples were fabricated on the surface of the second LiGaSe${_2}$ crystal, also embedded into an aluminum frame. Their transmittance was measured and recorded.

The pictures of the best ARM sample were obtained with a Versa 3D (FEI Company, USA) scanning electron microscope (SEM) and are shown in Fig. 1. The best $2.2\times 2.2$ mm${^2}$ ARM sample was fabricated with the average laser power of 5.5 mW and the pulse energy of 2.75 µJ. An additional pulse picker of 2 kHz was used to reduce the speed of the nanopositioner for better accuracy. These values provide the maximum energy density at the surface sufficient to ablate LiGaSe${_2}$ and form cavities with a diameter of approximately 1 µm and a period of 1 µm. The fill factor (the ratio of the ARM-treated area of the sample to the untreated area of the sample) is therefore close to 1.

 

Fig. 1. SEM image of the ARM sample: (a) top-down overview; (b) magnified; (c) cross section profile.

Download Full Size | PDF

Figure 1(c) shows the cross sectional profile of the ARM. The average depth of the ARM is approximately 0.7 µm which provides an aspect ratio (depth to diameter ratio) of 0.7. The transmittance of the ARM sample was measured with a $5\times 5$ array of points evenly distributed along the whole $2.2\times 2.2$ mm${^2}$ sample.

The transmittance of the ARM averaged over 25 points of measurement is shown in Fig. 2 along with the theoretical and measured values for an untreated surface of LiGaSe${_2}$. The theoretical transmittance of LiGaSe${_2}$ was calculated using Sellmeier equations given in Ref. [36] and correlates well with the measured value for the untreated surface. Figure 2 also shows the transmittance spectrum of a 200 nm-thick layer of liquid water. The absorption peaks on the ARM graph correspond to the absorption peaks of liquid water. This is due to the fact that LiGaSe${_2}$ is hygroscopic, therefore some water can be absorbed by the porous layer on the surface of the ARM cavities. This layer is highlighted in Fig. 1(c) with a white arrow in a magnified part of the image (the contrast was increased). To avoid the formation of liquid water during the ARM fabrication it is likely necessary to perform the experiments in a low-humidity environment. Annealing results in the degradation of LiGaSe${_2}$ optical properties [37], therefore in our future work we will investigate for other residual water removal methods.

 

Fig. 2. Comparison of the single-surface transmittance of untreated LiGaSe${_2}$, ARM-treated LiGaSe${_2}$, and ARM simulation.

Download Full Size | PDF

An average ARM single surface transmittance of 97.2% was calculated accounting for water absorption in the 2–8 µm range. The maximum single surface transmittance of 98.6% at 4.1 µm was achieved. The ARM provides high transmittance values in the required wavelength range. A decrease in transmittance at wavelengths greater than 9 µm is observed because longer wavelengths fail to interact with the structure due to its insufficient depth, and the spectral behavior of ARM tends toward untreated surface. The decrease in transmittance of the flat surface is due to the absorption of the material.

Figure 2 also demonstrates a transmittance simulation of the ARM with the morphological parameters of the experimental one. A more detailed description of the model can be found in our previous work [24]. The absorption was not taken into account during the simulation, therefore there is a good correlation between the simulation and the experiment. The decrease in transmittance in shorter wavelengths is due to diffraction.

The consistency of the ARM transmittance on the $2.2\times 2.2$ mm${^2}$ sample was analyzed. Figure 3 demonstrates the deviation from an average transmittance value of 98.6% at 4.1 µm. It can be noticed that the total transmittance deviation is less than ${\pm }$0.3%, thus the ARM is uniform. Some areas even provide the transmittance of around 99% which is 14% higher than the transmittance of an untreated LiGaSe${_2}$ surface.

 

Fig. 3. Transmittance deviation of a $2\times 2$ mm${^2}$ ARM sample fabricated on the LiGaSe${_2}$ crystal, for the 4.1 µm wavelength.

Download Full Size | PDF

To assess the effect of laser ablation on the LiGaSe${_2}$ surface, energy-dispersive X-ray spectroscopy (EDX) was performed using a Hitachi SU8020 (Hitachi, Ltd., Japan) SEM. The chemical composition of the surface areas of LiGaSe${_2}$ crystals, with and without ARM, was measured. Four of the measurement points are shown in Fig. 4. In the ARM-treated areas of the surface an excess of selenium was observed in the ratio of ${\rm Ga/Se} > 1/2.5$ (see Table 1). It may occur due to chemical reactions that arise as a result of the ultra-fast localized impact of high temperatures on the surface. This leads to the formation of oxygen-containing Li and Ga compounds and the release of Se with its further deposition near the ARM. A similar mechanism was described during the surface modification of GaSe crystals [38].

Conclusion. In this work the fabrication of ARMs on the surface of a LiGaSe${_2}$ crystal using a single-pulse femtosecond laser ablation method was demonstrated. The resulting ARM provides an average transmittance of 97.2% in the 2–8 µm spectral range and the maximum transmittance of 98.6% at 4.1 µm. It was also demonstrated that the ARM is consistent and uniform over the area of $2.2\times 2.2$ mm${^2}$, and the total deviation is smaller than 0.3%.

 

Fig. 4. Points of EDX measurements on the SEM images of LiGaSe${_2}$ surface: for (a) the ARM-treated sample and (b) the untreated sample.

Download Full Size | PDF

Table 1. Surface Chemical Composition of the Untreated and ARM-treated LiGaSe${_2}$, Corresponding to the Points in Fig. 4

View Table

In our future work, we will focus on optimizing the ARM depth in order to increase the average transmittance in the 2–8 µm range for both sides of the crystal, improving the quality of the ARM and analyzing the effect of feature surfaces roughness on transmittance and scattering losses.