1. Introduction

Mid-infrared (MIR) fiber lasers based on rare earth (RE)-ion doped MIR-transmitting glass-hosts, are of interest for cutting/welding soft materials, like plastics, and also cutting of biological tissue - potentially offering new wavelengths for medical laser surgery. The MIR region encompasses the fundamental vibrational absorptions of molecular materials and so a MIR fiber laser tuned to a specific molecular vibrational absorption has the potential to act as a direct molecular sensor. The large extinction coefficients of fundamental vibrational absorption bands in the MIR, mean such molecular sensors should show good contrast, and thus good sensitivity, and be quantitative. MIR fiber lasers are required for pumping MIR fiber supercontinuum generation [1] for wideband, MIR molecular sensing with broad applicability across many sectors including environmental monitoring, agriculture, security (including food) and in medicine, for in vivo early cancer diagnosis [2, 3] and optical biopsies.

Although the development of mid-infrared fiber laser is a growing field of photonics, there is significant challenge in fabrication fiber lasers beyond 3 μm wavelength [4]. No MIR fiber laser has yet been demonstrated at >4 μm wavelength [4–6]. Chalcogenide glasses are attractive hosts for RE-ions, which provide the MIR emission [7–9]. The chalcogenide glasses have low phonon energies, giving transparency at both pump and emission MIR wavelengths, and show reasonable solubility for RE-ions. However, only Nd³⁺-ion doped Ga:La:S glass fiber lasing in the near-infrared (NIR) at 1.08 μm wavelength has been demonstrated (in 1996) [10], nevertheless, in the last 15 years, RE-ion doped chalcogenide glass fibers were suggested as one of the most promising material to underpin MIR emission as a fiber laser source [4, 7, 11, 12]. MIR photoluminescence (PL) in Pr³⁺-doped chalcogenide bulk glass has been comprehensively studied in the work of Shaw et al. [13]. Our prior modeling [14] has suggested that praseodymium (Pr³⁺) is a suitable RE-ion dopant for a selenide-based, chalcogenide-glass, step index fiber (SIF) MIR fiber laser operating at 4-5 μm wavelength.

Fabrication of RE-ion doped chalcogenide glass fiber, especially with a small core, is a demanding process because crystallization must be avoided during the heat treatments required to effect shaping [11, 15–17]. The fabrication of both unclad. and large-core step-index Pr³⁺-doped selenide chalcogenide fibers has been reported in the literature. For the former: (i) Charpentier et al. [18] have reported unclad Ga₅Ge₂₀Sb₁₀Se₆₅ at% (atomic %) fiber doped with 0.05 wt% and 0.1 wt% (i.e. 500 ppmw and 1000 ppmw (parts per million parts, by weight), respectively) Pr³⁺ ions and (ii) Park et al. [9] have fabricated 0.02 mol% (sic) Pr³⁺- doped Ge₃₀Ga₂Sb₈Se₆₀ fiber. For the latter: (i) Cole et al. [19] have prepared multimode (core/clad. diameters = 160/200 μm), selenide SIF where the fiber core glass was Ge-As-Ga-Se doped with 200 ppm (parts per million) Pr³⁺, and (ii) we have presented multimode (core/clad. diameters = ~280/~300 μm) selenide SIFs and the fiber core was ~500 ppmw Pr³⁺-doped, either Ge-As-Ga-Se glass or Ge-As-In-Se glass [20, 21]. MIR fiber PL was reported in all of the above work [9, 18–21].

A small core structure in a RE-ion doped chalcogenide glass fiber would theoretically increase the gain in the active material and thus could facilitate the realization of lasing. Further, such structure allows for the realization of fiber lasers that operate in a single transverse mode and hence yield a better beam quality than the lasers using multi-mode fibers. Obviously, a photonic crystal fiber (PCF) offers a larger cross-sectional area within the single mode operation regime. However, a small core structure of a classic core/clad. fiber is easier to fabricate than a PCF, in these novel glasses, and hence provides a more pragmatic path towards the realization of a fiber laser that operates in a single transverse mode.

Compared to unclad. and large-core step-index RE-ion doped chalcogenide glass fiber, there is much less work in the literature, on small-core step-index, RE-ion doped chalcogenide-glass fiber, and only few reports may be found. Kobelke et al. [22] have described a double-clad, small-core fiber with a core diameter of 2.5 μm (760 ppmw Pr³⁺ doped As_37.8Ge₁.₃Ga_0.5S_60.4 at%), inner clad of 20 μm diameter (As₃₆.₆Ge_3.2S_60.2 at%) and outer cladding of 125 μm diameter (As₄₀S₆₀ at%). Also, Chung et al. [23] have reported a 0.02 mol% (sic) Pr³⁺-doped Ge₃₀Ga₂Sb₈Se₆₀ core/Ge₃₀Ga₂Sb₈Se₅₅S₅ clad. fiber, with 110 μm fiber diameter and a small core (core diameter was not specified). In both of these reports, of Kobelke et al. [22] and Chung et al. [23], the Pr³⁺-doped small core fibers exhibited a numerical aperture (NA) of ~0.4 and were single mode at 1.3 μm wavelength. No MIR fiber PL was investigated in these small core fiber papers [22, 23] and fiber performance in the 1-1.3 μm wavelength NIR spectral region was presented rather than in the MIR (>3 μm wavelength) spectral region of current interest.

Here, we report fabrication and characterization of a small-core, single-clad Pr³⁺-doped Ge-As-Ga-Se SIF and the focused spectral region is MIR. Thus, 500 ppmw Pr³⁺-doped Ge-As-Ga-Se glass SIFs with a 10 μm or 15 μm diameter core, and a cladding glass of Ge-As-Ga-Se-S, were fabricated. Thermal properties of the Pr³⁺-doped Ge-As-Ga-Se/Ge-As-Ga-Se-S glass pair were studied using differential thermal analysis (DTA) and thermomechanical analysis (TMA). The small core fiber structure was confirmed by means of scanning electron microscopic (SEM) back-scattered electron (BSE) imaging and SEM energy dispersive X-ray spectroscopy (EDX) mapping. Due to the importance of avoiding crystallization in the fiber fabrication, this work, in comparison with previous work [22, 23] on the RE-ion doped chalcogenide glass small core fiber, provides first time evidence of the crystalline/amorphous nature of the small core and core/clad. interface of the fiber using high resolution transmission electron microscopy (HRTEM) accompanied by selected area electron diffraction (SAED), and assisted by the technique of X-ray diffraction (XRD). Optical properties, including Fourier transform infrared spectroscopy (FTIR) and refractive index (RI) dispersion of the core/clad. glass pair, spectral optical loss of part-processed unstructured fiber of the same glass formulation as the small core fiber and termed here the 'intermediate fiber', and near-field behavior of the small core fiber, are presented. For the first time, the MIR PL spectrum and MIR PL lifetime in a small core RE-ion (i.e. Pr³⁺ in this work) doped chalcogenide glass fiber are reported.

2. Experimental

2.1 Bulk glass preparation, extrusion and fiber-drawing

2.1.1 Preparation of Pr³⁺-doped core glass cane and intermediate fiber

Ge (5N purity, Cerac), As (7N5 purity, Furakawa Denshi; prior heat treated under vacuum (10⁻³ Pa)), Se (5N purity, Materion; prior heat treated under vacuum (10⁻³ Pa)) and 1000 ppmw TeCl₄ (3N purity, Alfa Aesar; used as a [H] getter) were melted for 12 hours at 850°C, quenched and annealed, in a silica glass ampoule under vacuum (10⁻³ Pa). Then, the as-annealed Ge-As-Se glass was taken from the ampoule and was batched with 500 ppmw Al (5N purity, Alfa Aesar; used as an [O] getter) into a silica glass distillation rig and the rig was sealed, followed by a distillation under vacuum (10⁻³ Pa) of the chalcogenide glass and getter charge. After distillation, the Ge-As-Se was remelted in situ for 7 hours at 800°C, followed by quenching and annealing to form glass. Then, this as-annealed Ge-As-Se glass, which had been purified by distillation, was remelted again with Ga (5N purity, Testbourne Ltd.) and 500 ppmw Pr foil (3N, Alfa Aesar; it should be noted that care was taken to add the Pr foil accurately to the batch) for 6 hours at 850°C, followed by quenching and annealing, to achieve the 500 ppmw Pr³⁺-doped Ge-As-Ga-Se glass rod preform (see Fig. 1(a)).This preform was drawn into both 230 μm diameter fiber (i.e. unclad fiber which is the “intermediate fiber” in this paper) for optical loss measurement, and 500 μm diameter cane for the fabrication of the small core fiber (see Fig. 1(b)). A radio frequency fiber-drawing furnace on a customized Heathway fiber-drawing tower was used in this work.

 

Fig. 1 Flow chart of fabrication of 500 ppmw Pr³⁺-doped Ge-As-Ga-Se small core fiber. (a) 500 ppmw Pr-doped Ge-As-Ga-Se glass. (b) 500 μm diameter cane (and 230 μm diameter intermediate fiber) produced by the first drawing. (c) Ge-As-Ga-Se-S glass. (d) 1.5 mm ID jacket tube produced by extrusion. (e) Assembled preform before the final fiber drawing. (f) A selection of the small core fibers produced by the second fiber drawing.

Download Full Size | PDF

2.1.2 Cladding glass, extrusion and small core fiber drawing

Ge, As, Ga, Se (purity and prior treatment were as in section 2.1.1), and S (5N purity, Materion; prior heat treated under vacuum (10⁻³ Pa)) were melted for 12 hours at 850°C, quenched and annealed, to achieve a Ge-As-Ga-Se-S glass boule (see Fig. 1(c)). The Ge-As-Ga-Se-S glass boule was extruded to form an inner diameter (ID)/outer diameter (OD) = 1.5 mm/11.5 mm jacket tube (see Fig. 1(d)) using an in-house extruder. The final step was to place the Pr³⁺-doped Ge-As-Ga-Se core glass cane (described in section 2.1.1) inside the Ge-As-Ga-Se-S jacket tube (see Fig. 1(e)) and co-draw these into small core Pr³⁺-doped Se-based SIFs (see Fig. 1(f)), using the rod-in-tube method. Vacuum was applied during the core/clad. fiber drawing. Small core fibers of 10 μm diameter core (fiber OD = ~230 μm) and 15 μm diameter core (fiber OD = ~330 μm) were drawn.

2.2 Bulk glass and fiber characterization

2.2.1 Bulk glass characterization: DTA, viscosity-temperature, XRD, FTIR and RI dispersion

For DTA, 100 ± 1 mg of as-prepared chalcogenide bulk glass samples (in chunks) was sealed inside a small silica glass ampoule (ID/OD = 3 mm/4.6 mm, QB Glass) under vacuum (10⁻³ Pa) and placed inside a Perkin Elmer DTA7 differential thermal analyzer. DTA curves were collected from room temperature to the desired temperature typically >600°C at 10°C/min. From the DTA curves, the glass transition temperatures (T_g) was obtained constructing the intersection of the extrapolated onset of the maximum gradient of endothermic peak with the pre-T_g baseline, to an accuracy of ± 2°C.

A parallel-plate method [24] was applied for viscosity-temperature measurement which was carried out in a Perkin Elmer TMA7 thermomechanical analyzer at 10°C/min up to ~100-200°C above T_g. The TMA viscosity disc samples, cut from a glass remelt, were 4 mm in diameter and 1.6 mm or 4.1 mm in height. A constant load of either 70 mN (used for 1.6 mm high samples) or 430 mN (used for 4.1 mm high samples) was applied in the viscosity-temperature measurements. The accuracy was ± 2°C on the temperature values and 10 ^{± 0.05} Pa.s on the viscosity values.

XRD of as-prepared chalcogenide bulk glasses, which had been powdered, were measured in a Siemens Krystalloflex 810 X-ray diffractometer, with CuK_α radiation from 10 to 70 °2θ at 40 seconds per step size 0.05 °2θ, i.e. 13 hours running time. FTIR spectra of the bulk glasses were collected using a Fourier transform infrared spectrometer (Bruker IFS 66/S). FTIR disk samples had an optical pathlength of 2.6-2.7 mm and were polished to a 1 micron finish. Finally, the RI dispersion of the bulk glasses was determined using IR-VASE and VUV-VASE ellipsometers (J.A. Woollam; as in [25]).

2.2.2 Fiber characterization: XRD, fiber loss measurement, SEM, near-field imaging, HRTEM and photoluminescence

XRD samples of the 500 ppmw Pr³⁺-doped Ge-As-Ga-Se unstructured intermediate fiber and unclad. cane were prepared into powder form. The fiber/cane XRD patterns were obtained with the same equipment and method as described in section 2.2.1. The optical loss spectrum of the intermediate fiber was collected using the cut-back method on the Fourier transform infrared spectrometer (Bruker IFS 66/S). The focused spot size of the Globar/tungsten source in the FTIR was >1 mm diameter. The error of the fiber loss was ± 0.2 dB/m (more details of fiber loss measurement can be found in [20]).

For the 500 ppmw Pr³⁺-doped Ge-As-Ga-Se small core fibers, SEM-BSE images and SEM-EDX analysis were achieved by means of a FEI XL30 SEM system and an Oxford Instruments INCA x-sight Si(Li) detector with ATW2 window. In order to obtain samples for HRTEM imaging, the Pr³⁺-doped Ge-As-Ga-Se small core fiber was crushed into powder samples, which were placed on a carbon/Cu grid. A JEOL 2100F Field Emission gun (FEG)-TEM with a Gatan Orius camera was used to obtain HRTEM phase contrast images and SAED. In the collection of TEM-EDX spectra, an Oxford Instruments INCA TEM 250 EDX system was applied. In near-field imaging, a 1.319 μm wavelength laser (Thorlabs, FPL1053) was launched into the 10 μm diameter core of the 500 ppmw Pr³⁺-doped Ge-As-Ga-Se fiber (length = 90.5 mm) and near-field images of the light emitted from the other end of the fiber were obtained.

For evaluation of the PL intensity spectra of the Pr³⁺-doped Ge-As-Ga-Se intermediate and small core fibers, the pump was a 1.55 µm wavelength fiber-coupled single-mode, CW laser diode (Thorlabs FPL 1009S) focused through a microscope objective (Thorlabs) and a collimator (Thorlabs) and the resultant beam size was expected to be ~18 μm diameter. Thus, in the PL measurement, we anticipate that < 1/3 of applied pump power was launched into the 10 μm diameter core of the Pr³⁺-doped SIF. The exit-end fiber PL signal was collected by means of an ambient MCT detector (Vigo System PVI-6), in a motorized monochromator (Spex MiniMate). A digital 250 MHz PC oscilloscope (Picoscope5204 PicoTechnology) was used for the lifetime measurements of the Pr³⁺-doped small core fiber. The 1.55 μm pump laser diode was directly modulated (8-10 Hz) during the lifetime measurements. The sample lengths of the 10 μm diameter small core fiber and the intermediate fiber were 117 mm and 115 mm, respectively. PL spectra were corrected for the system response [20].

3. Results and discussion

3.1 Thermal properties of core/clad. glass pair

Figure 2 presents the viscosity-temperature curves of the 500 ppmw Pr³⁺-doped Ge-As-Ga-Se core glass and the Ge-As-Ga-Se-S clad. glass which were used in the fabrication of the small core fiber. At the extrusion viscosity of ca. 10⁸ Pa.s, the core supercooled glass melt temperature was 275°C and the clad. supercooled glass melt temperature was 284°C, i.e. a 9°C temperature difference. Within the 275-284°C region, the viscosity difference of the core and clad. glasses was 10^0.4 Pa.s. Moreover, a smaller temperature gap of 6°C was found for the core/clad. glass pair, for the ca. 10^4.5 Pa.s fiber drawing viscosity; for this the viscosity mismatch was 10^0.15 Pa.s in the 369-375°C region (see inset of Fig. 2).

 

Fig. 2 Viscosity-temperature curves of the core and clad. glasses used in the 500 ppmw Pr³⁺-doped Ge-As-Ga-Se, step-index, small core fiber. Inset expands these curves around the fiber drawing temperature. Solid curves are for clad. glass; dotted curves are for core glass.

Download Full Size | PDF

Figure 3 shows the DTA curves of the 500 ppmw Pr³⁺-doped Ge-As-Ga-Se core glass and the Ge-As-Ga-Se-S clad. glass, together with the approximate fiber drawing and extrusion temperatures found from the viscosity-temperature curves (see Fig. 2). The T_gs of the core and clad. glasses were close: T_g for the core = 226°C, T_g for the clad. = 227°C. For both core and clad. glasses, there was no distinct crystallization peak observed at the temperatures used for the extrusion (core glass: 275°C, clad. glass: 284°C) and fiber drawing (core glass: 369°C, clad. glass: 375°C). This appears to indicate sufficient glass stability of the core/clad. glass pair. Section 3.3 will describe further studies of the crystalline/amorphous nature to investigate the glass stability of the core/clad. pair.

 

Fig. 3 DTA curves of the as-annealed core glass 500 ppmw Pr³⁺-doped Ge-As-Ga-Se, and clad. glass Ge-As-Ga-Se-S from which the step-index, small core fiber was fabricated. The extrusion and fiber drawing temperatures have been taken from the viscosity-temperature curves of Fig. 2. The liquidus temperatures (T_liq) were calculated using equation shown inside the Fig [26].

Download Full Size | PDF

3.2 Compositional structure of the small core fiber

In the SEM-BSE images of both the 10 μm (see Fig. 4(a)) and the 15 μm (see Fig. 4(b)) core diameter fibers, the cores were sitting centrally in each fiber, but each core was slightly elliptical. The 10 μm diameter core was ca. 9 μm × 11 μm diameter (see Fig. 4(c)) and the 15 μm diameter core was ca. 15 μm × 16 μm diameter (see Fig. 4(d)). Figure 4(e) shows that the core cane contacted one side of jacket tube at the very beginning of the 15 μm core fiber drawing to leave a transitory gap which swiftly healed during the fiber-drawing. In the core region, the ID of the jacket tube (ca. 1.5 mm) was three times that of the cane OD (ca. 0.5 mm) for the rod-in-tube fiber drawing, and this is suggested to be the main reason for the elliptical core (i.e. unwanted distortion from circular, which happened in the collapse down during fiber drawing). In future work, a tighter rod-in-tube dimension match would be used in the fabrication. The 10^0.15 Pa.s viscosity mismatch at the fiber drawing temperature (see inset of Fig. 2) is another possible reason for the elliptical core.

 

Fig. 4 SEM-BSE images of the 500 ppmw Pr³⁺-doped Ge-As-Ga-Se, step-index, (a) and (c) 10 μm diameter core and (b) and (d) 15 μm diameter core fibers, and (e) transitory gap in 15 μm core fiber.

Download Full Size | PDF

The SEM-EDX mapping in Fig. 5 shows the S, As and Ga elemental differences between the Pr³⁺-doped Ge-As-Ga-Se core and the Ge-As-Ga-Se-S cladding glasses. In Fig. 5, the sulfur mapping shows the best distinction of the core/clad. geometry for both the 10 μm diameter and the 15 μm diameter core fibers, based on the lack of sulfur in the core. In summary, the designed core/clad. glass pair composition gave good contrast in both SEM-BSE images and SEM-EDX mapping.

 

Fig. 5 SEM-EDX-mapping (S, As, Ga) of the 500 ppmw Pr³⁺-doped (a) 10 μm diameter core and (b) 15 μm diameter core SIFs.

Download Full Size | PDF

3.3 Crystalline/amorphous nature study

It is important to avoid crystallization during processing RE-ion doped chalcogenide glass fiber because crystallization can increase the optical scattering loss [7, 15]. Our previous work [15, 16, 27] has shown that rare earth associated compounds (e.g. RE₂O₃, RE₂Cl₃) can act as nucleation agents for crystal growth in bulk glasses. Also, one of our previously reported early prototype 500 ppmw Dy³⁺-doped Ge-As-Ga-Se fibers was found to exhibit crystallization peaks in its XRD pattern [28]. In Fig. 6(a), the XRD pattern of the as-annealed Ge-As-Ga-Se-S bulk cladding glass in the present work does not show any crystallization peaks. Moreover, no distinct crystallization peak was found in the XRD pattern of the 500 ppmw Pr³⁺-doped intermediate unclad. fiber (see Fig. 6(b)). Also, the 500 ppmw Pr³⁺-doped cane used in the rod-in-tube method to make the small-core fiber was XRD amorphous (see Fig. 6(c)) before drawing the small core fiber.

 

Fig. 6 Powder XRD patterns of: (a) the as-annealed Ge-As-Ga-Se-S bulk clad. glass; (b) the 500 ppmw Pr³⁺-doped Ge-As-Ga-Se intermediate unclad. fiber and (c) the 500 ppmw Pr³⁺ doped Ge-As-Ga-Se unclad. cane. All of the XRD patterns were K_α2 stripped.

Download Full Size | PDF

After the final process step involving a heat-treatment, i.e. the small core fiber drawing of the Pr³⁺-doped glass, the XRD technique was not sensitive enough to check the crystalline/amorphous structure of the Pr³⁺-doped glass because: (1) the volume of the small core fiber was only 0.2% of the entire fiber and (2) XRD has a sensitivity limitation of ~2-5 volume% for crystals [16]. Thus, in the following, the HRTEM technique was applied to explore the amorphicity of the Pr³⁺-doped small core SIF.

It should be noted that, in addition to the core, the core/clad. interface is a highly likely region for crystal growth because this interfacial region is vulnerable to trapping defects and any defect in the interfacial region can act as an heterogeneous nucleation agent for crystal growth. Because of the 0.2% volume of the core, most of the glass material on the sample grid was the cladding and it was very difficult to find both core material and the interfacial region but this was successfully achieved using TEM-EDX. Figure 7(a) shows a Scanning Transmission Electron Microscopy (STEM) image of a sample coming from the Pr³⁺-doped small core fiber and Fig. 7(b) shows TEM-EDX spectra of the three regions marked in Fig. 7(a). In all the spectra of Fig. 7(b), the EDX-band of Au is attributed to a known stray signal from the TEM sample holder. Spectrum 1 of Fig. 7(b) shows no EDX-band for S, which indicates that this spectrum must have been from the 500 ppmw Pr³⁺-doped Ge-As-Ga-Se core material. In Fig. 7(b), spectrum 2 has an increased S band and its quantification result is close to the nominal batched S level in the cladding. Spectrum 2 was taken from a region of the fiber located adjacent to the region from which spectrum 1 (see Fig. 7(a)) was taken. Thus, it is believed that the spectrum 2 region of Fig. 7(a) includes the core/clad. interface and maybe some clad. material. Please note that in Fig. 7(a), no compositional difference could be found between the core and clad., which is in contrast to the SEM image of Fig. 4 which clearly shows a well-defined core/clad. interface. Finally, spectrum 3 of Fig. 7(b) shows a clear EDX S-band. Therefore, the spectrum 3 region is suggested to be from the Ge-As-Ga-Se-S clad. material.

 

Fig. 7 (a) Scanning Transmission Electron Microscopy (STEM) image of core region (spectrum 1), core/clad. interfacial region (spectrum 2) and clad. region (spectrum 3) of the 500 ppmw Pr³⁺-doped Ge-As-Ga-Se, step-index, 10 μm diameter core fiber. (b) TEM-EDX spectra of the three regions arrowed in Fig. 7(a).

Download Full Size | PDF

Figures 8(a), (b) and (c) are the HRTEM-SAED patterns for the core region (spectrum 1 in Fig. 7(a)), the core/clad. interfacial region (spectrum 2 in Fig. 7(a)) and the clad. region (spectrum 3 in Fig. 7(a)), respectively. The associated concentric circular SAED patterns are diffuse, indicating that all of these glassy materials are crystal free. Also, the phase contrast images of Fig. 8(d), (e) and (f) are taken from the core region, the interfacial region and the clad. region, respectively, and they give further evidence that the different regions of the small core fiber are amorphous. During the HRTEM measurement, many more phase contrast images were monitored carefully over the core, interface and clad. regions but no crystalline pattern was ever found. Thus, the HRTEM results are self-supporting and suggest no detectable crystals were present in the Pr³⁺-doped small core fiber, after the two step heat-treatment processing to fabricate fiber. In summary, all of the XRD and HRTEM results of section 3.3, together with the DTA curves results discussed in section 3.2 indicate that the Pr³⁺-doped Ge-As-Ga-Se core glass, the Ge-As-Ga-Se-S clad. glass were thermally stable and that the fabrication process of the small core fiber avoided crystallization.

 

Fig. 8 HRTEM-SAED patterns and HRTEM phase contrast images of the core material region (spectrum 1 of Fig. 7(a)), the core/clad. interfacial region (spectrum 2 of Fig. 7(a)) and the clad. material region (spectrum 3 of Fig. 7(c)) indicating all fiber regions were amorphous.

Download Full Size | PDF

3.4 Optical properties

3.4.1 Optical loss

Figure 9 shows the FTIR spectra of the as-prepared 500 ppmw Pr³⁺-doped Ge-As-Ga-Se core bulk glass and the Ge-As-Ga-Se-S clad. bulk glass used in the fabrication of the Pr³⁺-doped small core fiber. Firstly, the shape of the shorter wavelength baseline of each of the two FTIR spectra overlap each other well, indicating no extra scattering loss of the Pr³⁺-doped glass. Secondly, the FTIR spectrum of the core glass manifests the electronic absorption bands of Pr³⁺ appearing at wavelengths of 1.5 μm (³H₄→³F₄), 1.6 μm (³H₄→³F₃), 2.0 μm (³H₄→³F₂ + ³H₆) and 4.5 μm (³H₄→³H₅). We suggested that the Pr³⁺ absorption band at 4.5 μm wavelength overlaps with the Se-H contamination band of the glass matrix [20]. A simplified energy level diagram of the Pr³⁺-ion can be found in the inset to Fig. 9 [21]. Thirdly, in the clad. glass spectrum, contamination absorption bands are found at wavelengths of ~2.9 μm (due to vibrational absorption of O-H in hydroxide and water) [29], 4.0 μm (S-H) and 4.5 μm (Se-H) [30, 31]. For both the core and clad. glasses, the FTIR spectra show Si-O vibrational absorption at 9-10 μm wavelength [31].

 

Fig. 9 FTIR spectra of the 500 ppmw Pr³⁺ doped Ge-As-Ga-Se core bulk glass and the Ge-As-Ga-Se-S clad. bulk glass used in the fabrication of the Pr³⁺-doped, step-index, small core fiber. Inset is taken from [21] and it shows simplified energy level diagram of Pr³⁺-ion with notional inter-level values.

Download Full Size | PDF

The optical loss spectrum of the 500 ppmw Pr³⁺-doped Ge-As-Ga-Se unclad, intermediate fiber is shown in Fig. 10 and it does not have distinct extra scattering loss at NIR wavelengths. This is consistent with the XRD result also indicating the amorphous nature of the glass forming the intermediate fiber, as shown in Fig. 6. In the spectrum of Fig. 10, the bands at wavelengths of 1.5 μm (Pr³⁺), 1.6 μm (Pr³⁺), 2.0 μm (Pr³⁺) and 4.5 μm (Pr³⁺ + Se-H) are found to be overloaded due to too high absorption in the loss measurement. At 3.5 μm wavelength [30], loss due to the Se-H band was 1.5 dB/m. The contamination O-H band [29] exhibited a loss < 0.15 dB/m at 2.9 μm wavelength and no distinct H₂O band can be found at 6.3 μm wavelength [30]. The Ge-O/As-O band was 0.6 dB/m at 7.8 μm [32]. As found for the FTIR of the bulk core glass (see Fig. 9), the Si-O band at > 9 μm was observed [31]. The baseline loss was 2.4 dB/m in the 2.8-3.2 μm wavelength region and 2.0 dB/m (lowest loss) in the 6.5-7.1 μm wavelength region. The loss of the intermediate fiber is comparable to other rare earth ion doped chalcogenide glass fiber (either unclad. or large-core step-index) results in the literature [21, 28, 33–35]: generally, the lowest loss reported has been of the order 1-2 dB/m.

 

Fig. 10 Optical loss spectrum of the 500 ppmw Pr³⁺-doped Ge-As-Ga-Se unclad, intermediate fiber. Inset shows optical micrographs of the two best fiber cleaves used in the loss calculation.

Download Full Size | PDF

The glass distillation of Ge-As-Se (section 2.1.1) should have helped to decrease the fiber baseline loss and impurity bands (e.g. O-H vibrational absorption band). Ga and Pr were not further purified, and might have brought impurities into the fiber, contributing to higher optical losses (e.g. increasing the Se-H vibrational absorption band). Compared to the intermediate fiber, the final small core Pr³⁺-doped fiber is generally anticipated to have greater loss because of the additional processing; moreover the extra core/clad. fiber drawing could have led to further defects at the core/clad. interface. Yet no defects were found in the core/clad. interface of the small core fiber by means of SEM and HRTEM.

3.4.2 Light guiding

The refractive index dispersion curve of the 500 ppmw Pr³⁺-doped Ge-As-Ga-Se core glass can be found in Fig. 11 The refractive index dispersion curve of the 500 ppmw Pr³⁺-doped Ge-As-Ga-Se core glass can be found in Fig. 11 and was 2.5661-2.5586 at 3.5-6 μm wavelengths, respectively (measured using IR-VASE and VUV-VASE ellipsometers, J.A. Woollam). The refractive index dispersion data of a Ge-As-Ga-Se host glass (i.e. not rare earth doped) was measured using the same ellipsometers (J.A. Woollam). The design of the Ge-As-Ga-Se-S cladding glass in this work was based on this Ge-As-Ga-Se host glass, substituting 3 at% of Se by S to lower the refractive index. The refractive index behavior when Se is substituted for S in Ge-As-Se glasses is known from the literature [36] and is assumed to be upheld for the small 3 at% Ga addition in the Ge-As-Ga-Se-S glass formulations used here. Thus, the refractive index dispersion curve of the Ge-As-Ga-Se-S clad. glass was estimated based on the measured RI dispersion of the Ge-As-Ga-Se host glass, and the assumed refractive index change upon substituting S for Se (~0.016 decrease for 3 at% substitution [36]. This refractive index change was assumed to be wavelength independent away from the optical bandgap and multiphonon edge), and is presented in Fig. 11. According to the refractive indices of the core and clad. glasses, the calculated NA is given in Fig. 11 and is 0.3906-0.4046 at 3.5-6 μm wavelength. Near-field images of the 500 ppmw Pr³⁺-doped 10 μm diameter core SIF are presented in Fig. 12.The fiber was multimode at the 1.319 μm NIR test wavelength. According to the NA curve (see Fig. 11) and standard fiber theory [37], single mode propagation is expected for wavelengths ≥5.2 μm in the 500 ppmw Pr³⁺-doped 10 μm diameter core SIF.

 

Fig. 11 Refractive index (RI) dispersion of the core/clad. glass pair and numerical aperture (NA) of the 500 ppmw Pr³⁺ doped Ge-As-Ga-Se, step-index, small core fiber.

Download Full Size | PDF

 

Fig. 12 (a), (b) Near-field images of the 10 μm diameter guiding core of the 500 ppmw Pr³⁺-doped SIF (length = 90.5 mm) at 1.319 μm wavelength for different input excitation, showing the fiber to be multi-mode at this wavelength.

Download Full Size | PDF

3.4.3 PL spectrum and lifetime

The PL spectrum of the 500 ppmw Pr³⁺-doped 10 μm diameter core fiber pumped at 1550 nm is presented in Fig. 13.This is the first time to our knowledge that a MIR PL spectrum from a RE-ion doped chalcogenide glass small core fiber has been reported. The two insets inside Fig. 13 show the launch end cleave and emitting end cleave of the fiber sample used in the PL measurement of this work. In Fig. 13, the PL emission band spans in a broad range of 3500-6000 nm and is suggested possibly to be a combination of two (or more) transitions of ³H₅→³H₄ and ³H₆→³H₅ [18] (please refer to the simplified Pr³⁺ ion energy level transition shown in the inset to Fig. 9). The dip at 4.2 μm wavelength is suggested to have been due to CO₂ in the optical path of the PL measurement set-up [18–20].

 

Fig. 13 PL spectrum of the 500 ppmw Pr³⁺ doped Ge-As-Ga-Se, step-index, 10 μm diameter small core fiber (length = 117 mm). Pump wavelength 1550 nm and pump power 50.2 mW. It is estimated that < 1/3 of the applied pump power was launched into the 10 μm core. Relevant absorption band positions are shown.

Download Full Size | PDF

Based on the critical guiding angles at the core/clad. and clad./air interfaces (the refractive indices of the core/clad. glasses can be found in Fig. 11), it is estimated that for the PL emission which was guided in the small-core fiber, only around 2% of PL emission propagated within the core glass. In other words, 98% of the guided PL emission was expected to be carried in the Pr³⁺-ion-free clad. material. Thus it is suggested that the dip at 4.5 μm wavelength can be associated with the vibrational absorption of Se-H contamination in both the clad. and core. Due to the presence of S in the clad. glass, there was a S-H absorption band at 4.0 μm wavelength in the FTIR spectrum of the bulk clad. glass (absorption coefficient = 0.066 cm⁻¹, Fig. 9); corresponding to this, a dip in the small core fiber PL spectrum was observed at 4.0 μm wavelength (see Fig. 13).

Figure 14 shows the normalized PL spectra of the 500 ppmw Pr³⁺-doped 10 μm diameter core fiber, the 500 ppmw Pr³⁺-doped 230 μm diameter unclad. (intermediate) fiber and the 500 ppmw Pr³⁺-doped bulk glass (the bulk datum was taken from [20] for comparison). All of these glasses had nominally the same Ge-As-Ga-Se host batch composition. At 4.2-5 μm wavelengths of the FTIR spectra (see Fig. 9), the absorption of the clad. material (no Pr³⁺; the absorption is from the tail of S-H band and the small Se-H band; coefficient = 0.072 cm⁻¹ at 4.5 μm wavelength) was much smaller than the absorption of the core glass (the absorption here is from both Se-H and Pr³⁺; absorption coefficient = 0.072 cm⁻¹ at 4.5 μm wavelength). Also, as noted, 98% of the guided PL light is expected to have transmitted through the clad. material in the small core fiber. Hence, the absorption was distinctly less at 4.2-5 μm wavelengths in the small core fiber than in the unstructured fiber and the bulk glass. This probably led to the relatively higher PL intensity band at wavelengths in the <4.4 μm range in Fig. 14 for the small core fiber (the highest peak PL intensity at ~4.7 μm wavelength was normalized to 1.0).

 

Fig. 14 PL spectra of the 500 ppmw Pr³⁺-doped 10 μm diameter core fiber (continuous solid curve), the 500 ppmw Pr³⁺-doped 230 μm diameter unclad./intermediate fiber (continuous solid curve) and the 500 ppmw Pr³⁺-doped bulk glass (dashed curve; the bulk datum was taken from [20]). 50.2 mW pump power at 1550 nm wavelength was applied in the collection of these spectra.

Download Full Size | PDF

In the unstructured fiber (of length 115 mm), the emitted PL must have traveled through the Pr³⁺ doped material for a greater pathlength than the pathlength in the PL measurement of the Pr³⁺-doped bulk glass (≤2 mm) [20]. In the fiber core glass material, the Pr³⁺ electronic absorption and the Se-H vibrational absorption were predominantly in the wavelength range 4-5 μm (see Fig. 9). Hence in the unstructured fiber, longer wavelengths of >5 μm, were less attenuated than the shorter wavelengths, which will have led to an apparent increase in the long wavelength PL intensity in the unstructured fiber after normalization. Moreover, although the emitted PL essentially traveled through a 117 mm path of the small core fiber, we surmise that only 2% of the guided PL emission would have been guided in the core material. Thus, from Fig. 14, at longer wavelengths, of >5 μm, the intensity of the small core fiber PL spectrum is apparently lower than that of the unstructured fiber PL spectrum and is close to the bulk glass PL spectrum.

Figure 15 presents the PL decay characteristic and lifetime of the 500 ppmw Pr³⁺ doped Ge-As-Ga-Se, step-index, 10 μm diameter core fiber, at 4700 nm wavelength. The decay lifetime is defined as the time taken for the PL intensity to decrease to a faction of 1/e of its initial level and it is 7.8 ms for the Pr³⁺-doped small core fiber in Fig. 15. Our previous work has shown that the lifetime was 7.8 ms in a 500 ppmw Pr³⁺-doped Ge-As-Ga-Se bulk glass [20]. So, the lifetime in the Pr³⁺-doped bulk glass and the small core fiber were found to be the same, which indicates that the processing of the bulk glass into the small core of the small core fiber did not adversely affect the PL lifetime of the doped Pr³⁺ ions in the selenide chalcogenide host glass. To the best of our knowledge, this sustainment of lifetime in small core chalcogenide glass fiber has not previously been reported. More information (e.g. emission and absorption cross-section) of Pr³⁺-doped Ge-As-Ga-Se (bulk) glass can be found in our previous work [14].

 

Fig. 15 PL decay and lifetime of the 500 ppmw Pr³⁺ doped Ge-As-Ga-Se, step-index, 10 μm diameter core fiber, at wavelength of 4700 nm. 64.8 mW pump power at 1550 nm wavelength was used.

Download Full Size | PDF

4. Conclusions

The fabrication and characterization of 500 ppmw Pr³⁺-doped Se-based step-index 10 μm and 15 μm diameter core fibers have been presented in this paper. The thermal properties of the core/clad. glass pair have been shown to be sufficiently well matched for co-processing to fiber according to the viscosity-temperature curves, and the differential thermal analysis, and fiber was made. The Pr³⁺-doped Ge-As-Ga-Se core and Ge-As-Ga-Se-S cladding of the fabricated small core fiber gave good contrast for imaging of the fiber geometry using SEM-BSE and EDX-mapping. XRD and HRTEM were engaged to explore the crystalline/amorphous nature of the core and clad. glasses materials and no crystals were found, even down to the nano-level in the 10 μm diameter core fiber. The lowest background optical loss of the Pr³⁺-doped intermediate, unclad fiber was 2.0 dB/m in the 6.5-7.1 μm wavelength region. Refractive index dispersion and numerical aperture (~0.4 at the 4.7 μm wavelength of peak PL) of the fiber were presented. Near-field images of the Pr³⁺-doped SIF showed multimode propagation in the 10 μm diameter core at 1.319 μm wavelength and the fiber was estimated to have single mode propagation at ≥5.2 μm wavelengths. Finally, the MIR PL spectrum and lifetime of a rare earth doped chalcogenide glass small core fiber were presented for the first time: 3.5-6 μm wavelength emission in the Pr³⁺-doped 10 μm diameter core fiber was observed and studied; a PL decay lifetime of 7.8 ms was measured at wavelength of 4700 nm and was found to be the same as in our previous lifetime measurements but on Pr³⁺-doped bulk Ge-As-Ga-Se glass. The small core Pr³⁺-doped fiber achieved in this work is a useful step towards future work on the fabrication of a >4 μm wavelength MIR fiber laser. In the future, the PL behavior in Pr-ion doped chalcogenide glass small core fiber will be studied further in a laser cavity for testing. At the same time, purification of the glass fiber will be an important parallel work for achieving lasing in RE-ion doped chalcogenide glass fiber.