1. Introduction

Although pump-induced photodarkening (PD) and related effects in Yb-doped fibers have been investigated intensively for many years [1–11], the understanding of the processes on a microscopic scale is obsolete and characterized by conflicting assumptions. The charge transfer absorption of Yb³⁺ ions involving nearby oxygen ligands was proposed to initiate the procedure [3]; subsequent trapping of charge carriers could result in the formation of Yb²⁺ ions and hole-related color centers. This model was taken up by other research groups (e.g., [6,10]). In [6], a PD-induced reduction of Yb³⁺ absorption only around 976 nm was claimed and interpreted as a partial change of valence of Yb³⁺ ions to Yb²⁺. However, core transmission measurements on very short Yb fiber pieces are sophisticated and not reliable around this absorption peak because the remaining probe signal can be obscured by the noise floor of the spectrometer or by a small amount of probe light guided in the fiber cladding. Actually, we did not find any reduction of Yb³⁺ absorption in our fibers after pump irradiation, which could point to a partial transformation to Yb²⁺. Moreover, no indication of Yb²⁺ absorption or emission was found in the ultraviolet/visible (UV/VIS) spectra determined on common Yb fibers (from preforms collapsed in an oxidizing atmosphere) after pump-induced PD. In Yb fibers with oxygen deficiency caused by the preparation technology (preform collapsing in a He atmosphere), the distinct Yb²⁺ signatures were rather reduced than enhanced by the PD process [8].

Owing to lack of evidence of Yb²⁺ creation, other mechanisms are to be considered to explain color center generation including Yb³⁺ ions that are doubtlessly involved in the photodarkening process. In [12], electron transfer between nearest-neighbor non-binding oxygen atoms caused by doping silica with Yb³⁺ and Al³⁺ is proposed as origin of the PD process that takes place by the action of two pump photons in collaboration with phonons produced by excited Yb³⁺ ions.

A direct proof of charge carrier movement between atomic defects acting as precursors or traps is out of the scope of this paper; however, modifications of the environment should have consequences on the Yb³⁺ ions themselves and can be investigated by the analysis of fluorescence properties. Despite unchanged absorption, a distinct reduction of Yb³⁺ fluorescence and lifetime after PD is observed. These effects are quantitatively evaluated in this paper for fibers with various Yb concentrations, for different photodarkening states below the equilibrium at room temperature, and also for degradation by different Yb inversions. The results are discussed in correlation to the measured PD loss values.

2. Experimental

Preform samples were prepared by MCVD (Modified Chemical Vapor Deposition) and solution doping and collapsed in an O₂/Cl₂ atmosphere. For characterization purposes, fiber samples with a cladding diameter of 125 µm and a core diameter of approximately 10 µm were drawn. All fibers were doped with nearly constant amounts of Al₂O₃ (≈4.5 mol%) and P₂O₅ (≈0.5 mol%). The concentration of Yb₂O₃ varied between 500 and 6800 mol-ppm and was restricted to the range without concentration quenching effects [13].

To analyze PD and to prepare samples for the subsequent fluorescence investigations, the fibers were cladding-pumped at room temperature with up to 16 W at 915 nm (Yb inversion up to 0.7) in the setup described in [5]. A tapered fiber bundle (TFB) coupler with signal feed-through was used to couple the pump power into the cladding and the probe light (wavelength 633 nm, power < 1 nW) into the core of short fiber pieces (1 to 2 cm) under test to observe the temporal evolution of PD loss. The steady state values of PD loss achieved after pump switch-off are later on discussed in relation to the observed changes of fluorescence properties.

Three series of fiber samples were prepared in this photodarkening setup and kept for subsequent fluorescence analysis in comparison to the pristine samples:

By this procedure, fiber samples with final PD loss varying in a wide range were prepared. A linear dependence of PD loss on Yb inversion and a slightly nonlinear increase with Yb concentration (for constant inversion) are known from earlier investigations [5,13].

The characterization of fluorescence intensity and decay was done with continuous-wave (CW) or chopped (pulses of 0.5 ms, repetition rate 10 Hz) excitation, respectively, using a MOPA system emitting at 976 nm. The laser beam is focused perpendicularly to the fiber core to excite a very small volume (fiber core with diameter 10 µm) thus avoiding ASE (amplified spontaneous emission) and reabsorption effects. A collecting fiber (diameter 400 µm), which guides the fluorescence light to the spectrometer or to the detector, is oriented perpendicularly to both, the exciting beam and the fiber axis. This twice-perpendicular measurement method is described more detailed in [14]. Yb inversion was kept low by using the same low excitation power with an estimated value of ≈1 mW corresponding to an Yb inversion of ≈0.1 for all CW measurements, whereas the inversion is even lower in chopped operation. Moreover, the measurement time for averaging the spectrum or chopped signal was kept below 10 s. Thus, PD effects in pristine samples or modifications of PD loss in already photodarkened samples are avoided. A quantitative comparison of the Yb³⁺ fluorescence properties between pristine and photodarkened fibers and between the samples of the three series introduced above is enabled by the advanced adjustment technique, which guarantees good reproducibility of the measurements by the use of a calibration fiber [14]. The fluorescence decay after the end of exciting 976 nm pulses is characterized in the near-infrared (NIR) region (λ > 1 µm). The discussed fluorescence intensity values in NIR (1075 nm) and VIS (500 nm) are picked up from the measured spectra of Yb³⁺ NIR emission and cooperative luminescence.

3. Results and discussion

The evolution of PD loss during pumping with 16 W (Yb inversion 0.7) is depicted in Fig. 1(a) for the fibers of series I with Yb content of 2000 mol-ppm or higher. (No measurable PD loss was observed in fibers with Yb content < 2000 mol-ppm for pump duration up to 900 min.) The pump-induced PD process was applied to prepare the fibers for subsequent fluorescence analysis, also in case of the fibers of series II and III.

 

Fig. 1 (a) PD evolution in the fibers of series I with Yb content ≥ 2000 mol-ppm, (b) NIR fluorescence decay measured on the fibers of series II after preparation by the PD process up to different PD loss values (pumping of pristine fiber pieces was interrupted after different times), compared to the decay curve of a pristine fiber.

Download Full Size | PDF

In Fig. 1(b), the measured decay of normalized NIR fluorescence intensity I_f (t)/I_f (0) after the end of 976 nm excitation is shown for the fibers of series II, serving as examples. Similar curves were recorded for all pristine and photodarkened samples of series I and III, too. In all cases, the measured fluorescence decay could be fitted by the single-exponential decay function of Eq. (1) with good accuracy, thus delivering one meaningful time constant τ for each measurement.

Figure 2(a) shows the results for series I. The NIR fluorescence lifetime of the pristine fibers τ₀ varies slightly between 813 and 842 µs (estimated measurement error ± 5 µs) with an already known dependence on the Yb content [7]. For the samples with Yb content ≥ 2000 mol-ppm, a reduced lifetime τ_PD was detected after PD. Moreover, a reduction of the fluorescence intensity was observed for these samples and the fibers of series II and III, too. The NIR fluorescence intensity after PD normalized by the values of pristine samples is depicted in Fig. 2(b) for series I as a function of Yb content, and in Fig. 3 for the series II and III as a function of PD loss.

 

Fig. 2 Fibers with variation of Yb content (series I): (a) measured NIR Yb³⁺ fluorescence lifetime before and after PD, (b) NIR fluorescence intensity after PD related to the value before PD for each fiber sample, measured (open squares) and calculated from measured lifetimes (full squares).

Download Full Size | PDF

 

Fig. 3 (a) Measured fluorescence properties (normalized to values before PD) vs. PD loss for series II, (b) measured and calculated NIR fluorescence intensity (normalized to values before PD) vs. PD loss for series III. The solid lines represent linear fits to the lifetime values.

Download Full Size | PDF

The VIS fluorescence intensity is also influenced by PD, even to a stronger extent because of the Yb³⁺ pair effect of cooperative luminescence as shown in Fig. 3(a) for series II. However, the absolute values of measured VIS intensity are more than 3 orders of magnitude below the NIR ones and can thus be neglected in the following discussion.

PD effects on NIR fluorescence lifetime and intensity are not independent from each other. The accelerated but still single-exponential fluorescence decay after PD points to a certain probability of a non-radiative energy transfer concerning all excited Yb³⁺ ions (homogeneous deactivation) with a rate R_nr. Thereby, the measured lifetime τ_PD is reduced in comparison to the pristine value τ₀ according to

To give an idea of the numbers: the rate of spontaneous fluorescence τ₀⁻¹ in the pristine samples is about 1200 s⁻¹ for the fiber types discussed here, and the maximum value of R_nr derived from the minimum of lifetime observed after PD is about 230 s⁻¹. This non-radiative loss of excitation energy also reduces the mean Yb inversion in the photodarkened fibers. The NIR fluorescence intensity after PD I_PD related to the value of pristine fiber I₀ can be calculated from the ratio of Yb inversions (upper level populations n₂) in both cases; see Eq. (3). The Yb inversions n₂(0) before and n₂(PD) after PD were calculated from the fiber and pump parameters as described in [15], taking into account the measured NIR fluorescence lifetimes τ₀ and τ_PD. Because no reduction of Yb³⁺ absorption was found after PD, modifications of Yb³⁺ absorption and emission cross sections σ_abs and σ_em were excluded. Also, the exciting power P_exc (photon energy hν, core area A) for fluorescence analysis was kept constant.

The results of I_PD /I₀ calculated from Eq. (3) with P_exc = 1 mW are included in Figs. 2(b) and 3(b) for the samples of series I and III, respectively. Similar results were found for series II. It is to be observed that the courses of calculated values show lower fluctuations than the measured ones because of the higher accuracy of temporal analysis compared to intensity measurements. Therefore, only the experimental results of lifetime analysis are used in the following discussion.

Generally, a strong correlation of fluorescence decrease with increasing PD loss, which is supposed to be proportional to the density of generated color centers, is indicated from Fig. 3. This is also valid for the fiber series with increasing Yb content (series I). From these observations we conclude that the non-radiative energy transfer discussed above could take place from Yb³⁺ ions (donors) excited during fluorescence analysis to the color centers generated by the pump-induced PD process before. The absorption of color centers spectrally overlaps with Yb³⁺ emission; they can thus be hypothesized as acceptors of transferred energy if they remain in close vicinity to the Yb³⁺ ions after the PD process. Moreover, the complete recovery of fluorescence intensity and lifetime after thermal bleaching of PD loss (temperature 600°C) points to the color centers as the real acceptors of energy in the photodarkened fibers.

An appropriate measure to quantitatively evaluate this effect is the transfer efficiency η [16], which can be calculated in our case from the NIR fluorescence lifetimes measured before and after PD,

The results are depicted in Fig. 4 for all three series of fiber samples investigated in this work. An almost linear dependence of transfer efficiency on PD loss (proportional to density of color centers) is observed. All samples of series I and III were pumped up to PD loss near the equilibrium value. The differences in transfer efficiency for PD loss below 800 dB/m can be attributed to the higher Yb content (6800 mol-ppm) in the samples of series III; a stronger energy migration between Yb³⁺ ions (donors) can enhance the non-radiative transfer to the acceptors [16]. The results for series II (Yb content of 5900 mol-ppm) are found between both other curves. An influence of the fiber states below PD equilibrium (PD loss < 700 dB/m), which could be characterized by particular distributions of distances between Yb³⁺ ions and color centers, cannot be proved by these results.

 

Fig. 4 Efficiency of non-radiative energy transfer vs. measured PD loss.

Download Full Size | PDF

4. Conclusions

In [1], an inhomogeneous deactivation of Yb ions during pumping is reported for fiber types with low Yb₂O₃ concentrations (< 1600 mol-ppm) but strong loss of pump power probably caused by photodarkening. In the fiber types with detectable PD effects investigated in our work, the reduction of NIR fluorescence intensity as well as lifetime was examined after PD. The rather homogeneous deactivation of excited Yb³⁺ ions, as concluded from the measured still single-exponential decay of fluorescence intensity after PD, may be promoted by rapid energy migration between Yb ions for the higher Yb concentrations.

The atomic defects generated during photodarkening in Yb fibers not only operate as absorbing color centers for photons in the wavelength range from UV up to NIR. They are suspected also to take over energy from excited Yb³⁺ ions directly, because a strong, nearly linear correlation of PD loss (proportional to density of color centers) and reduction of fluorescence was found. From the distinct efficiency of this non-radiative energy transfer we conclude that color centers remain in close vicinity of the Yb³⁺ ions where they were generated during the pump-induced PD process.

This energy transfer can cause additional loss of laser efficiency by the deactivation of Yb³⁺ ions that absorb pump power without contributing to the laser process. Further investigations including modeling of Yb fiber lasers will study the impact of both, color center absorption and Yb³⁺ deactivation, on the performance particularly of high power fiber lasers and amplifiers.

Photodarkening can also occur during fluorescence analysis of Yb fibers, especially in case of high Yb content. As our investigations show, the results will then not correspond to the fluorescence properties of pristine fibers. To avoid this inaccuracy, the exciting intensity or the measurement time should be reduced until no further increase of fluorescence intensity and lifetime are observed. The upper limits of reasonable intensity and measurement time depend on the Yb concentration as well as on the host glass composition.