1. Introduction

Polycrystalline infrared (PIR) fibers are used for numerous applications like mid-IR spectroscopy, flexible IR pyrometry [1], flexible IR-imaging systems [2], modal wavefront filtering [3], power delivery for quantum cascade lasers [4], CO– and $\mathrm {CO_2-}$lasers [4]. AgClBr is one of the compounds that can be used for polycrystalline fiber production. Its optimal combination of parameters ensures the successful use of AgClBr PIR-fibers for a broad range of applications. Some of those characteristics are nontoxicity, flexibility [5], low optical losses (0.2–0.3 dB/m in 9–13 µm range) without absorption peaks over the 9–13 µm range, and decent values of single-surface transmittance (only about 25% of Fresnel reflection losses). For mid-IR radiation delivery above 3 µm, other types of fibers could be used. Hollow-core photonic bandgap fibers provide only small regions of good transmission with zero transmission bandgap zones [6,7]. Another type of hollow-core fibers is antiresonant fibers which provide a wide transmission zone with almost zero losses but above 4.5 µm they start to suffer losses due to material absorption [6,7]. It is also possible to use hollow-core fibers with AgI coating on the inner side but drawbacks, however, include an additional loss on bending, small NA, and relatively short available lengths [8,9]. Based on these disadvantages, AgClBr fibers seem to be the most optimal option for radiation delivery in the mid-IR range.

However, to improve the efficiency of some devices that use PIR-fibers, it is necessary to increase the fiber tip surface’s transmittance. For instance, power delivery for $\mathrm {CO_2-}$ lasers requires high transmittance values of the optical fibers to utilize the maximum amount of available power [10]. Therefore, it is important to achieve the fiber tip’s utmost transmittance for a 10.6 µm wavelength.

The conventional method for increasing the surface transmittance is to use single-layer or multi-layer antireflection coatings (ARCs). Although ARCs are deposited at the fiber facets, the coefficient of thermal expansion plays a role in the transmission of high-power laser radiation. In the case of a one-component material, it is possible to select a coating material with an approximately equal coefficient of thermal expansion. But when it comes to fiber, then three materials are present at once – the core, the cladding, and the ARC itself. The difference in the coefficients of thermal expansion (CTE) of these materials leads to an additional problem - it can result in cracking of the ARC. Therefore it is difficult or often impossible to select the appropriate coating material for that purpose [11]. In the case of AgClBr ARCs cannot be applied to the surface of a fiber tip due to its unevenness. Moreover, the plasticity of material prevents perfect chipping or cleavage. Thus a slicing technique should be used to prepare the fiber, which produces a slice that is non-perpendicular to the optical axis of a fiber. The surface cannot be polished because of the AgClBr softness, and even with the most precise cut, perfect flatness cannot be achieved. The polycrystalline fiber core diameter typically ranges from one hundred to several hundred microns [3] which makes it difficult to use conventional methods for increasing fiber tip transmittance due to its small size. Another challenge is that inferior substrate surface quality can hamper the coating’s adhesion to the substrate which is quite common with materials such as AgClBr compounds [12].

An alternative approach is to fabricate antireflection microstructures (ARMs) on the surface of a fiber tip [13–16]. An ARM can be described as a system of micro-sized cavities or protrusions spaced on the sample surface. Physical phenomena that underlie the operation of ARMs can be described using the effective medium theory [17]. For wavelengths greater than $\lambda _{diff}~=~n~\cdot ~p$ (where $n$ is the refractive index and $p$ is the period of a microstructure), ARM acts like a layer with a refractive index gradient, which leads to an increase of transmittance [18].

Direct imprinting is one of the most straightforward ARM fabrication methods. The desired microstructure is imprinted onto the sample surface using the master mold, which usually is a stamp with a negative fabricated form of the microstructure on its surface. To eliminate collateral mechanical damage, the surface of a sample is preheated (the melting point of 410$^{\circ }$C for AgClBr has to be considered for). The master mold is then applied with a sufficient force to imprint the microstructure relief on the sample surface. This method can produce structures of high-fidelity profiles with periods ranging from 0.1 to 5 µm and larger [19–23].

There are, however, several significant disadvantages to this method. The first one has been mentioned earlier - the force applied to the material can result in cracking, chipping off and other mechanical damage. For instance, fiber tip deformation ("bulging") mentioned in [20] can drastically decrease the performance of the ARM. However, applying less force doesn’t provide acceptable results: the depth of the microstructures wouldn’t be sufficient, and the profile of the master mold wouldn’t be perfectly imprinted on the fiber tip. Another drawback is that the master mold needs to be fabricated using a different fabrication technique, like lithography, etching, etc. This can slow down the production rate, becoming a significant disadvantage, especially if a different master mold has to be fabricated for different geometries of the ARM.

In the case of AgClBr fibers, their non-flatness is an important issue - only the protruding parts of the tip will contact the shim, therefore the master structure won’t imprint perfectly on the whole surface. A hot-stamping technique for fabricating a diffractive microrelief on an end face of a silver-halide PIR- fiber proposed in [24,25] involves fabrication of a grating by photolithography, followed by the proper stamping of the microrelief. The process has several disadvantages related, on the one hand, to chemical etching and, on the other hand, to thermal and mechanical deformations of the mask due to the stamping of the surface.

Another highly promissory method of ARM fabrication is the single-pulse direct femtosecond laser ablation. The technique requires fewer steps in comparison to others [13–15] and provides both the good quality of ARM and high throughput [26]. Besides, this method has already been tested on chalcogenide glasses used for optical fiber production, such as $\mathrm {As_2S_3}$ [27], and bulk materials such as ZnSe [28] and CdSSe crystals [26]. The main advantage of this method is the ability to apply it even on uneven surfaces such as GaSe with a layered structure [29] and soft fiber materials such as AgClBr.

Here we demonstrate for the first time the AgClBr PIR fiber transmittance increase by applying ARM to a fiber tip surface using the single-pulse direct femtosecond laser ablation method.

2. Experimental setup

The proposed method consists of the formation of periodic holes on the working surface through local material removal due to the high energy density of a single fs pulse. This process is described by the two-temperature model. At the start, the process of multiphoton ionization is initialized due to the high intensity of the incident fs pulse. After that, produced electron gas accelerates in the field of the very same pulse (free-carrier absorption also known as inverse bremsstrahlung [30]), and more ionization events happen, during collisional interactions of already energized electrons and electrons in a valence band. These processes happen on a timescale comparable with pulse duration, forming non-thermal electron plasma, where temperatures of electron gas and lattice are in great non-equilibrium. Then, on a timescale of several picoseconds, energy from overheated (tens of thousands K) electron gas to lattice, and, therefore, the lattice receives energy later than electron gas. After that, the fragmentation of material bulk occurs (with partial transfer to gaseous or micro-dispersed state), then thermal plasma (where ions and electrons are in thermal equilibrium) forms, material ejection and plume formation happen [31,32]. As a result, local material removal occurs in the zone where a laser beam intensity exceeds the material’s ablation threshold.

Microstructures can be fabricated using single or multiple pulses to form a single cavity. Moreover, there are two possible ways to utilize the multi-pulse laser ablation technique. First, several high-energy pulses form a single ablation crater. The method should provide a larger depth. Second, the microstructure can be fabricated via precise carving it with several thousand pulses [33]. However, for both cases, without high-scanning speed, the fabrication time will be unsatisfactory. Therefore, whereas multi-pulse laser ablation can provide microstructures with higher aspect ratio, for efficient fabrication of large-area microstructure it requires having a laser system with both high repetition rate (at least several MHz) and large pulse energy (depending on the material, so in the particular case of AgClBr ablation it should be comparable with values achieved in this article). However, such laser systems are still not available off the shelf and it is challenging to design a custom one.

Therefore, in our method, to accelerate the production rate and utilize a commercially available laser system, a single pulse per microstructure feature was used to form the ARM. This is the simplest and the fastest way to fabricate ARMs using ablation. Other methods, like using a single pulse to form several craters or, vice versa, using several pulses to form a single crater, are considered to be more complex as they require higher levels of precision which cannot be afforded with our setup.

To achieve the highest transmittance values several aspects must be taken into consideration while fabricating the ARM. Firstly, the ARM has to be persistent and periodical, with the period calculated according to the desired wavelength for maximum transmittance. Secondly, the aspect ratio (depth to diameter ratio) of every cavity should be higher than 1. Thirdly, the transmittance depends on the profile and the fill factor of the ARM. Therefore these parameters must be chosen reasonably [26,34,35].

Pharos Yb:KGW laser (Light Conversion, Lithuania) was used as the femtosecond pulse source for ARM fabrication. Both second and third harmonic generation system is implemented, therefore providing a choice of three operating wavelengths: 1026 nm, 513 nm, and 342 nm. The laser system can produce 200 to 1000 fs pulses at repetition rates from 1 kHz to 200 kHz. In this work, a 1026 nm wavelength was used, and the maximum available average power was 450 mW. The 1026 nm wavelength was chosen because of the specific properties of AgClBr. Illumination by visible light with a wavelength shorter than approximately 650 nm results in photo-induced silver formation at the fiber surface, which in turn drastically decreases the transmittance.

For sample positioning, Aerotech ANT-90 (Aerotech Incorporated, USA) three-axis nanopositioners were used to provide in-position stability of 2 nm and repeatability of 75 nm at velocities of up to 200 mm/s. A 100^× objective lens (Mitutoyo Corporation, Japan) was used to focus the laser beam slightly underneath the fiber tip’s surface. The exact shift from the surface was determined experimentally. The ablation craters of test samples were inspected visually with a microscope to provide prompt response and find a suboptimal regime, while the fine-tuning of the z-shift parameter was chosen based on the transmittance measurements.

To remove residual ablative material and other debris from the surface of the sample, the constant supply of compressed air directed to the ablation zone was used. The air pressure was experimentally adjusted so that it would not disturb the ablation process.

The ARM transmittance was measured with two different setups. The first measurement was performed with a FTIR spectrometer Bruker Vertex 70v. For fiber measurement, an additional hand-made unit with two adjustable parabolic mirrors and two SMA connectors was used. To provide the measurement of 25 cm piece of AgClBr fiber (core diameter 900 µm, cladding diameter 1000 µm) due to the bend radius limitation of 150 mm, reference fiber was used. To avoid the appearance of losses at the connection of two fibers, a 400 µm core diameter of the reference fiber was chosen. When assembling the fiber and gluing it to the polymer sheath, we always slightly recessed it in the connectors so that the measured and reference fibers do not touch each other during butt coupling. Therefore, the measurement was first carried out with the ARM, and then the fibers fixed to each other were removed from the SMA connectors and the structure was cut off. Then the fibers were reinstalled and transmission was measured. This procedure made it possible to avoid the appearance of the reference effect since the same distance between the measured and the reference fiber took place in both measurements and thus this effect was reduced when calculating the ARM transmission. The spectral resolution was 4 cm$^{-1}$ and the internal aperture was 6 mm. A schematic of the described setup is presented in Fig. 1(a).

 

Fig. 1. Schematic of the measurement setup: FTIR spectrometer measurement (a); direct transmittance measurement with $\mathrm {CO_2}$ laser (b)

Download Full Size | PDF

To confirm the data from the spectrometer measurement, direct transmittance measurement with a $\mathrm {CO_2}$ laser at 10.6 µm was also carried out. The experimental setup is shown in Fig. 1(b). It includes the laser with a maximum 10 W output power (Coherent Inc., USA), two power meters for reference and main power measurements (Thorlabs Inc., USA), ZnSe beamsplitter and focusing lens (Thorlabs Inc., USA), and optomechanics for alignment of the beam (Thorlabs Inc., USA).

3. Experimental results

Several ARM samples were fabricated on the fiber tips of test fibers. The core of the used fibers is made of $\mathrm {AgCl_{0.3}Br_{0.7}}$ and the cladding is made of $\mathrm {AgCl_{0.5}Br_{0.5}}$. A lamp with a special light filter was used for illumination during the experiments to prevent the photo-induced silver formation at the fiber tip surface. Transmittance spectra for all the fabricated samples were recorded and analyzed.

The ARM sample with the highest transmittance was fabricated with the following parameters: the wavelength of 1026 nm, the pulse duration of 200 fs, the repetition rate of 200 kHz, and the average laser power of 450 mW. An additional pulse picker of approximately 1.25 kHz was used. The value of the pulse picker is calculated automatically based on the velocity of the sample, the repetition rate, and the period between craters. In this case, the speed during the fabrication was 5 mm/s. The peak power was 11.2 MW, the pulse energy was 2.25 µJ. The estimated spot radius in the focal plane was 0.46 µm. Due to the aforementioned z-shift, the spot radius on the surface was around 1.77 µm, therefore providing the fluence of 49.56 J/cm$^{2}$. Figure 2(a) demonstrates the profile of the fabricated ARM achieved with the laser confocal optical profilometer VK-X1100 from KEYENCE with 100^× objective (NA = 0.8, the working distance of 2 mm, the nominal resolution of 12 nm in height and 40 nm laterally).

 

Fig. 2. A cross-section profile (a), an overview image (b), magnified top-down view (c) and a close-up angled view (d) of the fabricated ARM

Download Full Size | PDF

Figure 2(a) demonstrates a cross-section profile of one area of the ARM. The diameter of the cavity is 2.4 µm (calculated as the difference between the dimensions labeled as 1 and 7) and the period is approximately 3.4 µm (labeled as 1). The average values of the period, cavity diameter, and depth, however, need to be calculated from several ARM areas. The average period calculated from 50 measurements on Fig. 2(c) is 3.8 µm, the average diameter of the cavity is 2.85 µm, therefore the average fill factor is 0.75. The average diameter of the formed craters correlates well with the estimated spot size. In addition, a close-up angled view of the ARM is demonstrated in Fig. 2(d).

Figure 2(b) demonstrates an image of the fabricated ARM. The average ARM cavity depth of approximately 1.7 µm and the average ARM cavity diameter of 2.85 µm provide the aspect ratio (depth to diameter ratio) of 0.6. Furthermore, it can be noticed that the microstructure is persistent and periodical.

Figure 2(b) also demonstrates larger periods of changes in depth that form a crosshatch pattern of red and blue. They appear on most ARM samples. This is an unintentional appearance, and the exact origin of these larger periods is yet undetermined. We presume that this can happen because of small vibrations occurring during the acceleration and deceleration of the sample. Although the absolute values of crater depths change, the difference between the highest and the lowest points in different areas remains constant and close to the values mentioned in the paper.

The total transmittance of untreated and one side treated fiber measured by $\mathrm {CO_2}$ laser at 10.6 µm is 75% and 83%, respectively. It needs to be considered that the error of the $\mathrm {CO_2}$ laser measurement is 3% for the total fiber transmittance due to the measurement error of the power meters. The transmittance spectrum of the fabricated ARM sample from two measurements (FTIR spectrometer measurement and direct $\mathrm {CO_2}$ laser measurement) along with ARM simulations with different parameters is shown in Fig. 3. The average ARM single-surface transmittance is 91.8% in the 7–14 µm range compared to the 87.4% transmittance of the untreated AgClBr fiber tip. The single-surface transmittance at 10.6 µm is 92.8%, which is 5.4% higher than the before-treatment transmittance.

 

Fig. 3. Comparison of single-surface transmittance of untreated AgClBr fiber and AgClBr fiber with ARM along with several ARM simulations

Download Full Size | PDF

The second Y-axis in Fig. 3 corresponds to the pink and purple graphs that represent the transmittance spectrum of the whole measurement setup, including the reference fiber (as described in section 2. The purple graph corresponds to the case when an untreated fiber was measured, and the pink one - to the ARM-treated fiber measurement. All the values are given in arbitrary units, the way the FTIR spectrometer records them. All calculations were based on this raw data.

Transmittance simulations of microcavities with different morphology were performed. For that, the commercially available finite-element modeling package was used [26]. The modeling setup was a three-dimensional domain with an emitter port, receiver port, and with single microcavity placed in between. Therefore, the dimensions of the domain in a plane perpendicular to light propagation direction equals microstructure period. We used a plane-wave light source because the field-microcavity interaction happens on a scale comparable to a wavelength of light. Domain size along light propagation direction was 300 µm, and ports were placed at a sufficient distance from microcavity, to avoid near field influencing them. The perfectly matched layers were used to effectively cancel the field before the emitter port (catching reflected wave) and after the receiver port (catching transmitted wave). Floquet periodical boundary conditions were used on sides of the domain to simulate an infinite array of microcavities, interacting with the large-scale Gaussian spot.

Figure 3 also demonstrates simulations for the ARMs with different parameters: "ff" stands for the fill factor, "p" - for the period, and "d" - for the depth. Other parameters correspond to the ones of the fabricated ARM and the fifth order of curvature was chosen for each of the simulation. The depth and fill factor were varied, while the period was constant. The shape of micro-cavities was approximated with y-axis revolution of $\mathrm {y=|x|^5}$ function for x ranged from 0 to 1. Such a shape has a close match with the actually observed morphology of fabricated samples. The resulting shape then was scaled to the desired period and depth. Therefore, any possible non-uniformity of microcavity shape was ignored in the simulation, and the results demonstrate the expected performance of perfectly fabricated microstructure.

It can be noticed that the transmittance spectrum of the fabricated ARM correlates well with the simulation results. Furthermore, to achieve the maximum amount of transmittance, it’s not enough to increase only fill factor or only ARM depth. According to our previous work [26], both fill factor and depth need to be increased because each of them influences the effective refractive index gradient smoothness and therefore maximum achievable transmittance. However, the period and the cavity diameter need to remain the same. This cannot be achieved with a Gaussian beam, therefore appropriate laser beam shaping is required. In our future work, we plan to focus on this aspect.

Furthermore, in Fig. 3 the transmittance increase is observed only at wavelengths larger than 6.5 µm because the diffraction starts on the ARM at 7 µm [26] due to the chosen ARM period and material refractive index.

4. Discussion

To achieve the demonstrated result, some of the fabrication parameters were varied during a series of experiments. The microstructure period, the laser frequency, the pulse picker settings, and the scanning speed remained unchanged after the initial choice of the values calculated to achieve the desired spectral transmittance characteristics of the microstructure. The laser average power and focal shift have been adjusted during a series of experiments until the formation of persistent microstructure was observed and then the transmittance sample fabricated with the set of parameters was measured. After some iterations, reliable fabrication of samples with the desired transmittance values was achieved.

To achieve a higher transmittance, the larger aspect ratio of microstructure should be maintained. However, with our current setup, the demonstrated result is considered as the limit. The limitation of the current setup is the shape of the volumetric "ablation zone" formed by a Gaussian beam. The ablation zone is the volume where the incident radiation is intensive enough to cause the substrate material ablation. With Gaussian beam three-dimensional shape of the ablation zone is determined by characteristics of the laser beam (intensity, M2, divergence) and properties of the material (refractive index and absorption coefficient). This is the simplest case that does not take into account any nonlinear effects and pulse interaction with plasma plume. Therefore, the aspect ratio is nearly constant because it depends on parameters, which cannot be adjusted. The only free parameter is the intensity, however, maximizing the intensity always leads to an increase in ablation crater diameter, and, consequently, the period of the structure should be increased to fit the new diameter which, in turn, affects the transmittance spectral characteristic. In detail, these problems and possible solutions were described in our original paper [26].

5. Conclusion

A straightforward method for ARM fabrication was proposed to increase the transmittance of AgClBr fiber used for $\mathrm {CO_2}$ lasers power delivery. The fabricated ARM transmittance was measured using two different setups: a FTIR spectrometer and a $\mathrm {CO_2}$ laser. The results show a significant increase in transmittance: the best-fabricated ARM demonstrates a 92.8% single-surface transmittance at the operation wavelength of the $\mathrm {CO_2}$ laser and at least 91.8% single-surface transmittance in the 7–14 µm range compared to that of 87.4% for untreated AgClBr.

In our further work, we will focus on optimizing the ARM fabrication process and increasing the maximum transmittance values in the 3–14 µm range to enable the AgClBr fiber usage with sources operating in the wide wavelength range. Therefore it will be possible to create an efficient FTIR fiber-based spectrometer with a wide transmission range.

Authors contribution

Andrey A. Bushunov, Andrei A. Teslenko and Mikhail K. Tarabrin designed fabrication method and experimental setup, fabricated ARM samples on AgClBr fibers, measured transmittance using a spectrometer and mostly contributed to the manuscript. Mikhail K. Tarabrin and Vladimir A. Lazarev supervised the research project. Tatiana Sakharova, Iskander Usenov and Viacheslav Artyushenko created AgClBr fiber samples, measured transmittance using a $\mathrm {CO_2}$ laser. Jonas Hinkel, Torsten Döhler and Ute Geißler provided the microstructure profile measurement.