1. Introduction

Optical beam delivery in the infrared spectral region has historically been hindered by a lack of efficient light sources and the unavailability of low-loss materials suitable for fabricating optical fibers. Over the past two decades advances in commercial quantum cascade lasers (QCL) have dramatically changed the landscape of available high-power sources in the infrared [1–3]. However, the large divergence of the optical output from QCLs, as with all diode emitters, may be detrimental to their use in some free-space applications, which further highlights the need for appropriate infrared fiber delivery systems. Efficient coupling to optical fibers in this spectral region will help counter this limitation, thereby enabling infrared QCL-based applications in defense, sensing, industry and medicine.

There has been a recent resurgence in the development of infrared fibers spurred by advances in infrared materials and the emergence of new fibers designs and fabrication strategies [4]. To date, most attempts at optical-fiber delivery of QCL radiation have made use of either large-diameter hollow-core fibers or highly multimode chalcogenide or fluoride fibers [5–11]. Hollow-core fibers can deliver near-fundamental mode output, but are difficult to fabricate and can suffer high losses if the hollow core becomes contaminated [12]. In contrast, multimode chalcogenide/fluoride fibers have poor mode quality due to their large core diameters and corresponding large $V$-numbers.

The ideal optical fiber for transporting infrared radiation should be mechanically robust, possess low single-mode optical losses across the entire infrared spectrum, including the near-infrared (NIR, 0.8-2 $\mu$m), mid-infrared (MIR, 2-5 $\mu$m) and long-wave infrared (LWIR, 5-12 $\mu$m) wavelength bands, and provide a high source-to-fiber coupling efficiency. These requirements are satisfied by a recently developed versatile class of multi-material chalcogenide (ChG) fibers [13–16]. These fibers combine ChG glasses (that perform the optical functionality of optical guidance in a step-index structure) and thermo-mechanically compatible thermoplastic polymers constituting a built-in outer jacket (that provide the mechanical robustness to the fiber), which are – uniquely – co-drawn from the same structured preform. The particular multi-material ChG fiber structure we utilize here is described in [17]. These fibers have demonstrated power-handling capabilities $>\,10$ W at a wavelength of 2053 nm with excellent core-mode confinement, high coupling coefficient by virtue of a single-layer anti-reflection alumina coating deposited on the ChG-polymer fiber tip, as well as significant enhancement in the mechanical robustness of these fibers compared with traditional bare ChG fibers or fibers provided with a low-temperature polymer jacket post-drawing [17,18]. Coupled with the broad transmission window of ChG glasses in the spectral range $1-12$ $\mu$m [10] and their potential for truly single-mode beam quality, these fibers offer versatility not found in hollow core or multimode fibers.

In this paper we describe an approach for the efficient coupling of a high-numerical-aperture QCL at a wavelength of 4.55 $\mu$m to AR-coated multimaterial ChG optical fibers. A detailed design and analysis of the optical system used to couple light into the ChG fiber is presented. Finally, experimental results and a discussion of the fiber coupling scheme are reported. Our results are posed to contribute to further expansion of the applications of QCLs in remote sensing and infrared spectroscopy.

2. Overview

2.1 QCL specifications

The QCL used in this study was a Fabry-Pérot type commercial device (Thorlabs QF4550CM1) with emission centered at a wavelength of 4550 nm. For the purposes of this study, the QCL was operated with active cooling via a thermoelectric cooler at a maximum output power level of 200 mW. The total angular spread of the QCL output relative to the slow and fast axes were $\theta _{\mathrm {slow}}\,=\,60^{\circ }$ and $\theta _{\mathrm {fast}}\,=\,110^{\circ }$ as specified by the manufacturer, which correspond to numerical apertures (NA) of NA$_{\mathrm {slow}}\,=\,0.50$ and NA$_{\mathrm {fast}}\,=\,0.82$ along the slow and fast axes, respectively. The physical dimensions of the fundamental mode of the QCL along the $x$ and $y$ directions, $w_{0x}$ and $w_{0y}$, can be estimated from the NA of each dimension using the beam parameter product (BPP), given the operating wavelength and angular divergence [19]. Assuming single-mode operation such that $M^2\,\approx \,1$, the BPP is given by

2.2 Fiber specifications

The ChG fibers used in this study were thermally drawn using an in-house drawing tower from a multimaterial preform consisting of a commercially sourced step-index arsenic-sulfide glass cane (comprising the core and cladding) produced via the double-crucible technique [4] encapsulated within a polyetherimide (PEI) jacket for enhanced tensile and flexural strength. The polymer selection was dictated by the need to co-draw the materials together at the same temperature [4,14,20]. More details on the fiber fabrication and characterization of its mechanical properties can be found in [17]. The core material is As$_{39}$S$_{61}$ glass while the cladding is As$_{38.5}$S$_{61.5}$ glass. The refractive index difference between these two compositions gives a NA$\approx \,0.2$. For the purposes of transmission around $\lambda \,=\,4.55$ $\mu$m, the ChG fibers were drawn with a 25 $\mu$m core diameter, giving a $V$-number of $\sim \,3.45$. Thus, the fiber supports two core modes, LP$_{01}$ and LP$_{11}$. Using the modified version of the Petermann mode-field radius [21], the mode-field diameter (MFD) of the fundamental core mode is calculated to be MFD $\,=\,22.2$ $\mu$m. A microscope image of the fiber facet is shown in Fig. 1(a) along with an expanded inset detailing the glass core and cladding dimensions in Fig. 1(b).

 

Fig. 1. (a) Optical micrograph of a polished, uncoated facet of a 25-$\mu$m-diameter-core multimaterial ChG fiber. The outer built-in polymer jacket diameter is $\approx \,920$ $\mu$m. (b) Expanded image of the ChG core and cladding. (c) Calculated reflectivity curves for AR-coatings with the ideal (690 nm) and actual (640 nm) thicknesses. The reflectivity at $\lambda \,=\,4.55$ $\mu$m is denoted by a red dot on both curves.

Download Full Size | PDF

It has been found experimentally [22,23] that coupling QCL radiation into uncoated ChG fibers results in a large Fresnel reflection of $\sim \,17\%$ that is returned from the fiber facet into the QCL, which destabilizes both the transverse spatial mode and the output power, and leads to significant temporal fluctuations [24]. In the present study, this issue was resolved by applying anti-reflective (AR) coatings to both fiber facets. These were deposited on the fiber facets simultaneously to minimize any variation in the AR-coating thickness on the two facets. The AR-coatings comprise a single layer of alumina (Al$_2$O$_3$) evaporated onto the fiber facets using an electron beam evaporator (Temescal FC-2000). Using the Sellmeier equations for both Al$_2$O$_3$ and As$_2$S$_3$, a minimum reflectivity of $R\,=\,0.37\%$ is possible for a layer of thickness $d\,=\,690$ nm at $\lambda \,=\,4.55$ $\mu$m [25–27]. To determine the final AR-coating thickness, the coating was deposited on a reference wafer simultaneously and subsequently measured using spectroscopic ellipsometry. Experimentally, we obtained $d\,=\,640$ nm thick films on each fiber facet, corresponding to $R\,=\,0.6\%$, which is a substantial improvement over the Fresnel reflectivity of $\sim \,17\%$ for the uncoated fiber. Calculated reflectivity curves for both the 640 nm and 690 nm AR-coatings are shown in Fig. 1(c), where we highlight the reflectivity at $\lambda \,=\,4.55$ $\mu$m using red dots. Note that such a coating provides a reflectivity of $R\,<\,1\%$ over a broad spectral range.

2.3 Layout of the optical system

Due to the extremely high numerical aperture (NA$=\,0.82$) along the QCL fast axis, a high-NA lens is required to initially capture and collimate this optical field. For this purpose, a BD2 mounted aspheric lens (Thorlabs C037TME-E) of focal length $f_{1}\,=\,1.873$ mm was utilized as shown in Fig. 2. This lens provides NA$>\,0.8$ and high internal transmission $>\,99\%$ at 4.55 $\mu$m, which is made possible with AR-coatings optimized over $3-5$ $\mu$m (reflectivity of 0.25% per surface). A 3.6-mm-diameter collimated beam was obtained after placing the lens at the calculated working distance of 0.698 mm from the QCL chip facet. The beam then entered a $2\times$ magnifying AR-coated CaF$_2$ correction telescope, comprising a bi-concave lens with $f_{2}\,=\,-50$ mm and a plano-convex lens with $f_{3}\,=\,100$ mm, that serves to magnify the beam and permits optimal coupling into the ChG fiber. The final focusing objective with $f_{4}\,=\,40$ mm is an AR-coated CaF$_2$ meniscus lens. CaF$_2$ was chosen as the lens substrate for the final three lenses due to its excellent transmission in the MIR, low cost, low refractive index, and commercial availability.

 

Fig. 2. Layout of the 4-lens system used to couple QCL radiation into the ChG fiber. All the dimensions are to scale. Focal lengths of the lenses are $f_{1}\,=\,1.873$ mm, $f_{2}\,=\,-50$ mm, $f_{3}\,=\,100$ mm, and $f_{4}\,=\,40$ mm. The distances between the optical elements were optimized via Zemax and are provided in the figure.

Download Full Size | PDF

3. Results

3.1 Zemax results

To accurately simulate the propagation of the QCL radiation through this optical system, we employed the Zemax software package using the scalar diffraction integral via the physical optics propagation package [26]. This approach accounts for the curvature and phase introduced by each optical element, vignetting on finite apertures, as well as the dispersion of each lens substrate material. Additionally, Zemax calculates the overlap integral of the beam with the fiber fundamental mode after propagation, providing the total system transmission and coupling efficiency into the fiber. To seed these simulations, the beam waist at the QCL output facet and the fiber MFD specified in Sec. 2.1 and Sec. 2.2 were used as initial inputs with a flat phase front assumed at the QCL output facet.

The starting optic in the simulation was the aspheric lens that was used to initially collimate the QCL radiation when placed at its calculated working distance of 0.698 mm. This position was fixed in the simulations to ensure that the beam initially remained collimated, thereby facilitating optical alignment in the laboratory. The inter-lens spacings between the three CaF$_2$ lenses as well as the fiber input facet location were considered free parameters that were allowed to vary during system optimization. In particular, the system was optimized to maximize the total system transmission and the fiber coupling efficiency. This system was constrained, however, by the choice of commercially available lenses, leading to the selection of lenses shown here. The final optimized layout of the optical system to couple QCL radiation into the ChG fiber is outlined in Fig. 2.

The simulated beam is initially astigmatic as it exits the QCL, as seen by the ratio of its two transverse beam waists [Fig. 3(a)], with $w_{0x}/w_{0y}\sim 0.56$. At the output of the optical system, the mode field radii are $w_{x,\mathrm {out}}\,=\,13.20$ $\mu$m and $w_{y,\mathrm {out}}\,=\,14.81$ $\mu$m. The astigmatism in the fiber focal plane is significantly reduced with $w_{x,\mathrm {out}}/w_{y,\mathrm {out}}\sim 0.89$. The simulated focal plane distribution after the QCL radiation propagates through the optical system is shown in Fig. 3(b). Since no astigmatism-correcting optics were employed, this reduction is due solely to the diffractive effects of propagation through the optical system and accumulates gradually during propagation. This approach increases the coupling efficiency into the ChG fiber and eliminates the need for astigmatism-correcting optics.

 

Fig. 3. (a) Simulated intensity distribution of the beam profile at the output of QCL and (b) at transmitted focal plane after the optical system in Fig. 2.

Download Full Size | PDF

Both the spatial distribution of the beam intensity in the focal plane and the phase distribution across the profile are critical to the coupling into the fiber. We plot in Fig. 4 the simulated amplitude and phase distributions through the center of the beam along the $x$ and $y$ axes. The intensity distributions are Gaussian ($R^{2} > 0.999$) along both directions. Across the central portion of the beam, the phase front in both dimensions is nearly flat. Although a perfectly flat phase front is required theoretically to achieve $100\%$ coupling into a fiber [28], the phase fronts observed here nevertheless still allow for efficient fiber coupling. Given these focal-plane amplitude and phase distributions, the maximum simulated coupling efficiency into the fundamental mode of the ChG fiber is $\eta = 75.4\%$. This is significantly higher than the calculated butt coupling efficiencies [28] for the fast and slow axes of the QCL into the ChG fiber of 7.5% and 21.8%, clearly demonstrating the effectiveness of the optical system described here to improve the coupling efficiency between the QCL and the ChG fiber.

 

Fig. 4. Simulated intensity and phase distributions along the $x$ and $y$ directions in the focal plane of the optical system in Fig. 2.

Download Full Size | PDF

Finally, no off-axis aberrations such as coma or astigmatism are present in the fiber focal plane. However, the Zemax simulation reveals a large amount of spherical aberration on the order of 24 waves in the output focal plane. This is unavoidable with our choice of optics, and can be corrected only with custom aspheric optics. Because this is the only aberration present, we expect the coupling efficiency to improve considerably by reducing spherical aberration.

3.2 Experimental results

The full experimental layout for coupling the QCL output into the ChG fiber is shown in Fig. 5. The system consists of the four-lens system described above [Fig. 2] built to the specifications of the optimized Zemax file. A pellicle beam splitter is inserted into the system to monitor power fluctuations in real time that may arise from back reflections, and to allow for calibrated fiber transmission measurements. A 37.5-mm-focal-length ZnSe meniscus lens ($f_{5}$) was used after the ChG fiber to capture and re-image transmitted light onto a Pyrocam infrared camera (Spiricon Pyrocam III) for beam profiling. In this experiment, we tested a 20-cm-long AR-coated ChG fiber. The fiber was mounted straight during all these experiments, although no significant bend-induced effects were observed previously with this fiber upon bending down to 5-cm bend radii [17].

 

Fig. 5. Optical system used to couple into the ChG fiber. Lenses $f_1$ through $f_4$ are detailed in Fig. 2 while focal lengths of the lenses used in detecting the input and output fields are $f_{5} = 37.5$ mm and $f_{6} = 40$ mm.

Download Full Size | PDF

To verify the performance of this optical system and confirm the validity of the above Zemax simulations, the focal-plane distribution was re-imaged onto the infrared camera to measure the spatial profile using lens $f_{5}$, which is plotted in Fig. 6(a). Although there is noticeable aberration, the transverse intensity distributions in the $x$ and $y$ directions agree well with the simulated Gaussian focal plane distributions. The measured focal-plane distribution was also compared against the results from Zemax simulations, showing good agreement in both the $x$ and $y$ directions [Fig. 6(b) and 6(c)].

 

Fig. 6. (a) Measured focal plane distribution of the QCL radiation before the multimaterial ChG fiber. The fiber core/cladding boundary is overlaid as a dotted white circle. (b-c) Lineouts comparing numerical and experimental results in (b) the $x$ and (c) the $y$ directions.

Download Full Size | PDF

Predominantly single mode excitation in the core was achieved for all launched powers and transmitted powers of > 160 mW, limited only by the output of the QCL [Fig. 7(c)]. Based on our simulations and prior experimental work [29], we expect > 70% core confinement at the output of the ChG fiber. The transmitted beam profile is shown in Fig. 7(a) with an expanded image of the core mode in Fig. 7(b). The dashed white circle denotes the glass cladding/polymer interface while the dotted white line denotes the fiber core/cladding interface. The output of the ChG fiber is plotted in Fig. 7(c) for an input power up to $\sim \,180$ mW. The transmission remains linear with $88.9\%$ transmission throughout the entire tested range of the QCL output power, demonstrating in particular the effectiveness of the AR-coating of the fiber facets when compared with an uncoated fiber ($T\,\sim \,72\%$). At no point during the experiment was the polarization state of the QCL output or ChG fiber output measured since our fiber was not polarization-maintaining fiber.

 

Fig. 7. (a) Transmitted mode profile through the ChG fiber. Outer dashed white circle denotes the cladding/polymer interface, whereas the dotted white circle denotes the core/cladding interface. (b) Expanded view of the core mode. The dotted white circle denotes the core/cladding interface. (c) Total transmitted power versus the launched power through the ChG fiber.

Download Full Size | PDF

When measuring the transmitted power, the power meter (Gentec XLP12-3S-H2-DO) was placed at a distance $<\,5$ mm from the ChG fiber output facet to ensure any high-NA cladding light is captured by the detector. Therefore, the measured value of transmission is the total transmitted power including cladding light, rather than only core light. Assuming ideal launch conditions such that no light is launched into the polymer jacketing, we estimate a coupling efficiency into the fiber core/cladding of $98\pm 2\%$. Coupled with a measured background loss of 1.1 dB/m in the fiber at 4.6 $\mu$m, the AR-coating reflectivity was estimated to be $2.1\pm 0.4\%$ [17,18,30]. The observed discrepancy between the experimental value and simulated value of 0.6% for the AR-coating reflectivity is most likely due to cracking of the AR-coating [18] as well as minor variations in the propagation loss and coupling efficiency used to calculate this value. Future work focused on optimizing the quality of AR-coatings for this application will enable even higher transmission values and power-handling capabilities.

4. Discussion

The ChG fibers described in the previous sections represent a versatile platform for fiber delivery of high-power single-mode radiation throughout the infrared. By tailoring the ChG fiber core diameter, single-mode operation can be achieved at any wavelength within the transmission band of As$_2$S$_3$ glass [4]. Furthermore, although the coatings used in this study were anti-reflective, coatings with different properties can be applied directly on the fiber facet to enable seamless system integration of various optical components [31–33].

4.1 Potential for power scaling

These in-house drawn fibers have been previously shown to handle CW powers $>\, 10$ W (12 MW/cm$^2$ facet intensity) at a wavelength of 2053 nm and $>\,1$ W at 4110 nm [18]. They offer an excellent platform for power scaling in the infrared. Although the transmission losses of this particular class of ChG fibers have not been rigorously evaluated across the entire infrared (see [10], however, for Te-based multimaterial ChG fibers), low background losses in As$_2$S$_3$ fibers as low as 14 dB/km at 4.8 $\mu$m have been previously demonstrated [34,35]. This indicates that similar or better performance should be obtainable at 4550 nm compared to 2053 nm operation and indeed any wavelength not impacted by intrinsic absorption features of the As$_2$S$_3$ glass [36].

To ensure stability of the QCL and to prevent component failure, more robust anti-reflection coatings will be needed in the future. The AR-coatings used in this study suffered from cracking during deposition due to the mismatched thermal expansion coefficients between the polymer jacket, As$_2$S$_3$ glass, and Al$_2$O$_3$ coating [30]. Optimization of the coating material and deposition process will enable the fabrication of higher quality, lower reflectivity coatings. Other approaches to AR-coatings such as recently reported moth-eye structures provide additional possibilities for minimizing back reflections while avoiding the thermal issues observed with thin film coatings [31–33]. Advancements in these and other anti-reflective technologies will allow higher absolute transmission through the fiber as well as enabling higher power transmission without optical-feedback-induced destabilization of the QCL. Additionally, whereas we relied on off-the-shelf optics, a more compact design is possible using custom-made optics. Given the results obtained at 2053 nm, we expect these AR-coated ChG fibers to support the transport of $>\,10$ W at 4550 nm and longer wavelengths with no significant background or propagation loss as higher power QCL sources become commercially available.

5. Conclusions

We have demonstrated the efficient coupling of light from a high-NA QCL at a wavelength of 4.55 $\mu$m into a 25-$\mu$m-diameter AR-coated ChG fiber supporting two core modes. Power levels $>\,160$ mW were transmitted with 88.9% transmission. With the use of more efficient aspheric coupling optics and higher quality AR-coatings, this versatile platform has excellent potential for high-average-power handling across the infrared spectral region.