2.1.

Fibers and Preparation

The investigated fibers were fabricated in-house by modified-chemical vapor deposition technology and solution doping. For our experiments, we chose fibers with comparable ${\mathrm{Yb}}_2{\mathrm{O}}_3$ concentrations around 0.35 mol. %. Our reference fiber is co-doped with 1.85 mol. % ${\mathrm{Al}}_2{\mathrm{O}}_3$, whereas our PD-reduced fibers are co-doped with cerium (Ce),¹⁹ with P/Al at ratio 1:1 and with much higher ${\mathrm{P}}_2{\mathrm{O}}_5$ content, respectively.²⁰ Detailed compositions of all fibers are given in Table 1. In addition, hydrogen loading of the reference fiber, which is known to decrease near-infrared (NIR)-PD as well,²¹ was performed before FBG inscription. In contrast,²² where ${\mathrm{H}}_2$ loading was performed at high temperature that resulted in ${\mathrm{Yb}}^{2+}$ formation, our fiber was loaded at room temperature conditions. All the fibers employ a standard geometry with an 8-μm core and a 125-μm cladding diameter.

Table 1

Chemical fiber compositions.

2.2.

FBG Inscription and Characterization

A modified Talbot interferometer and UV fs pulses were used for FBG fabrication (for details see Ref. 12). The laser system consisting of a seed laser (Mantis, Coherent, Inc., Santa Clara, California), a regenerative amplifier (Legend Elite, Coherent), and a frequency tripler (HSG-T, Coherent) provides pulses of 266-nm wavelength, approximately 500-fs pulse duration, with a 1 kHz repetition rate. For FBG inscription, a power of approximately 300 mW was applied with a $1/e^2$-beam diameter of approximately 6.3 mm, resulting in a maximum intensity in the focal region of $320\mathrm{GW}/{\mathrm{cm}}^2$, where a cylindrical lens with a focal length of 390 mm was used to focus the laser pulse on the fiber.

For FBG fabrication, we chose a phase mask period of 1075 nm, resulting in a Bragg reflection wavelength near 1550 nm, to avoid any distortion of the Bragg reflected signal by pump light, amplified spontaneous emission (ASE) or PD-induced losses (NIR tail) of the Yb-doped fibers during temperature measurement. Furthermore, the fabricated FBGs show a medium reflectivity ($<85\%$) to guarantee a proper peak measurement in reflection, as it is necessary for the applied measurement technique.

The temperature sensitivity of the FBGs was determined earlier to $(10.9\pm 0.3)\mathrm{pm}/\mathrm{K}$.²³ Because the fibers to be studied vary only slightly in doping concentrations, this temperature sensitivity was assumed for all fiber samples.

After grating inscription, the FBG possesses high losses, ranging from UV to NIR, that can be described as UV-induced PD in Refs. 18 and 23. Those losses disturb the proper sensor application of the FBG, because pump light will be absorbed causing a pronounced temperature enhancement due to UV-PD, which is not focus of this study. UV-PD can be removed by a temperature treatment²³ or photobleaching.¹⁸ The main advantage of a temperature treatment is that the absorbing species are annealed completely within just a few minutes when tempering the sample at about 550°C. But the disadvantage of this technique is a decreasing grating reflectivity, because the treatment affects not only the disturbing absorbing defects but also stress and compaction inducing defects, on which the index modulations defining the FBGs in case of fs pulse fabrication are mainly based. Utilizing a photobleaching treatment instead, the grating strength could be maintained due to the fact that only the disturbing color centers are affected. In addition, this technique is easily accessible and can be performed right after FBG inscription in the same setup using the same light source such as for FBG fabrication, just with lower average power. A detrimental influence is given by a residual absorbing defect concentration, which remains according to the model of equilibrium states of defects reported for NIR exposure²⁴ as well as UV.¹⁸ Nevertheless, by either of these treatments the disturbing broadband loss induced by the FBG fabrication can be removed.

2.3.

Pump Experiments

To study the thermal behavior of Yb-doped fibers under pump load, core-pumping was performed. The core-pumping technique was preferred because the pump light is restricted to the fiber core and disturbances in the cladding area, i.e., through grating inscription and fiber fabrication, are negligible and do not affect the pump absorption or the temperature measurement. A sketch of the setup is given in Fig. 1. Pumping was performed using a commercial single-mode fiber coupled diode at 976 nm with 250 mW optical power. For analysis of the temperature behavior during pumping, only the part containing the FBG, approximately 1 cm fiber, was spliced between passive delivery fibers, ensuring a homogeneous inversion along this length. For the spatially resolved measurement, an FBG array of 10-cm length containing four FBGs was spliced between the delivery fibers. In that case, pump absorption changes over fiber length, thus an inhomogeneous temperature profile along the fiber arises. Experimental investigations on the spatially distributed temperature evolution were restricted to a Ce-doped PD-reduced and the Yb/Al-doped reference fiber. The fibers can be moved into a tube furnace for thermal annealing of PD-induced defects.

Fig. 1

Setup for spatially distributed temperature measurement via fiber Bragg grating (FBG) analysis, combined with near-infrared photodarkening (NIR-PD) measurement at 633 nm with a core-pumping at 976 nm. For proper analysis of thermal effects in PD-reduced fibers, only short fiber pieces containing one FBG were analyzed.

To determine the temperature within the test fiber, we use an interrogator (FBGscan 608, FOS&S, Geel, Belgium) delivering test light around 1550 nm for the FBG wavelength monitoring and at the same time analyzing the Bragg wavelengths in reflection. From the wavelength shift of the FBG peak reflection, the temperature change within the fiber core can be estimated using the temperature sensitivity given above.

During pumping, it is also possible to monitor the PD loss at 633 nm, which is induced by the pump light itself (NIR-PD), using a chopper-lock-in technique as described earlier in Ref. 15. This measurement setup gives the possibility to correlate NIR-PD effects to the measured temperature changes.

3.1.

Temperature and PD Measurements

Prior to the annealing of the reference fiber, the UV-induced PD causes a strong temperature increase of the FBG region during pumping. This is due to the UV-PD involving pump light absorbing defects. After removing the FBG inscription-induced losses by a temperature treatment as described before, the pumped FBG yields two distinguishable contributions to the temperature change as indicated in Fig. 2. An initial temperature step is assigned to an energy input through permanent loss of the fiber core, including the quantum defect (QD) that is the energy difference between pump and ASE light plus any additional absorption due to the background attenuation of the fiber. The second contribution is a dynamic loss, which increases during pumping. Comparing the temperature evolution to the NIR-PD measurement, we assign this effect to the growing absorption induced by pump light. This dynamic loss can be prevented, e.g., by appropriate co-doping of the Yb fibers (PD-reduced fibers, see Sec. 2.1).

Fig. 2

Evolution of fiber temperature change and PD loss during pumping at 976 nm (reference fiber in water bath). The temperature increase before thermal treatment of the FBG is caused by UV-PD. After annealing of the UV-induced PD defects, only the quantum defect (QD) and subsequently induced NIR-PD contribute to the heating of the fiber core.

3.2.

Temperature Evolution in Differently Doped Fibers

In this section, we want to compare the temperature characteristics of differently co-doped fibers, i.e., our reference fiber and similar PD-reduced fibers (see Sec. 2.1). For proper comparison, all fibers containing an FBG have been thermally treated to remove possible UV-PD before pumping. Temperature evolution of the pumped fibers is depicted in Fig. 3(a).

Fig. 3

(a) Temperature evolution during 976-nm pumping for differently doped, preannealed fibers measured utilizing an UV fs pulse inscribed FBG. (b) Absorption cross sections of differently co-doped Yb-doped fibers. Ce co-doping does not change the cross section significantly in comparison to Al co-doping.

Although the Yb/Al-doped reference fiber shows an explicit dynamic temperature increase due to the NIR-PD, this effect is not observed for hydrogen loading of the same fiber prior to grating inscription. But in this particular case, the initial temperature step is dramatically increased. The reason for this performance is still under investigation. Changes in the valence state of the active ion may play an important role. Measured absorption spectra indicate the formation of ${\mathrm{Yb}}^{2+}$ by the exposure to UV radiation during FBG inscription for the hydrogen-loaded reference fiber (Fig. 4). This ${\mathrm{Yb}}^{2+}$-formation through optical energy deposition is even reinforced by the following thermal treatment at 550°C.

Fig. 4

Spectrum of pristine and hydrogen-loaded reference fiber showing ${\mathrm{Yb}}^{2+}$ absorption after UV irradiation and further on by subsequent thermal treatment.

The remaining PD-reduced fibers with Ce and P co-doping show also a negligible dynamic temperature increase during pumping. The initial temperature step at the beginning differs for the investigated fibers, although their Yb content is comparable. One reason for this behavior could be the difference in the Yb absorption cross sections, which directly influence the energy deposition through the QD, causing lower temperature shifts for $\mathrm{P}/\mathrm{Al}=1/1$ and the highly P co-doped fiber [compare Fig. 3(b)]. The slightly larger temperature step of the Ce co-doped fiber cannot be explained by this effect and must have another origin, since the absorption cross section is not influenced by Ce co-doping.²⁵

3.3.

Spatially Distributed Temperature Measurement

To demonstrate the applicability of the FBG temperature measurement in the spatial domain, we first fabricated an array of four FBGs within the Ce co-doped fiber, which is free of NIR-PD. During pumping, we observe different but temporally constant changes of the Bragg wavelength, i.e., the fiber core temperature at different FBG positions in the fiber (Fig. 5). At the pump side (FBG 1), pump light is efficiently absorbed causing heating of the fiber due to permanent losses (QD). At the other side of the array, the available pump power decreases, resulting in a reduced absorbed power and inversion, and therefore in a lower temperature jump at the start of pumping. During pumping, only negligible changes in the dynamic loss, i.e., NIR-PD, occur [Fig. 6(a)].

Fig. 5

Pumping of 10 cm Ce/Yb/Al-doped fiber containing four FBGs for temperature sensing. FBG 1 is situated next to the pump coupling, FBG 4 most distant. Different shifts of Bragg wavelength reflection peaks indicate an inhomogeneous temperature distribution along the fiber.

Fig. 6

Temperature evolution at different positions in 10-cm-long Yb fibers. The temperature step at the start of pumping is caused by the Yb absorption: (a) PD-reduced Ce-doped fiber and (b) reference fiber. NIR-PD occurs during pumping and causes an additional temperature increase.

To demonstrate the possibility of spatially distributed NIR-PD measurement, we repeated the same experiment on a 10-cm sample of our reference fiber. During pumping, we clearly observe the influence of the NIR-PD [Fig. 6(b)]. After the abrupt temperature step at the beginning of pumping, the temperature subsequently increases. This dynamic increase is more pronounced for FBG 1 than the other gratings because NIR-PD depends on the inversion, which is highest at the pump coupling side.