Fiber lasers, especially those based on ytterbium are important for engraving, cutting, or marking applications [1]. However, current limitations arising from the photodarkening effect preclude such laser systems to be reliably employed for microfabrication or laser surgery. A common phenomenon in rare earth doped materials [2–4], photodarkening manifests itself as the formation of a large absorption band across the VIS and NIR spectral windows during laser operation [5], thus degrading laser efficiency and output power. Light-induced creation of colour centres is not only a problem for lasers [5] but also for other applications, such as non-linear optics [6,7]. Understanding origins of photodarkening is a key to overcome this issue. Here we show how thulium contamination, presents at the ppb level in the raw doping material, can induce photodarkening in ytterbium-doped continuous fiber laser through energy transfer from ytterbium ions (Yb³⁺) and explain how this problem could be alleviated. The photodarkening effect depends strongly on the optical power density [8,9] and Yb³⁺ concentration [9]. The development of large mode area (LMA) fibers [10] has allowed increasing output power levels by decreasing the optical power density. However, the applicability of the solution is limited [11] and it cannot adequately address the ever-increasing demand for output power.

Because of a lack of understanding of the physical process leading to photodarkening, we characterized the luminescence properties of a phospho-aluminosilicate fiber doped with 1.7w% ytterbium; 3.3 w% Aluminium and 1 w% phosphorus. Its mode spot size was ~9 µm, his core diameter was 5.41 µm and the numerical aperture was 0.16. We used the standard soaking method to incorporate the ytterbium ions, by means of with a high purity (99.998%) ytterbium chloride solution. As reported in literature [12], luminescence in the visible spectrum was observed when the fiber was pumped at 980 nm. The fluorescence spectrum of the ytterbium-doped fiber is shown in Fig. 1 (black dotted) for a 500mW pump centred at 976nm. Four distinct emission bands are observed across the spectral range; each of these bands was normalized to its maximum in Fig. 1. Almost all bands can be attributed to the presence of Tm³⁺ ions according to the spectroscopy database; the origin of the emission around 520nm will be discussed later. Despite the high purity of the raw doping material (N48), the presence of thulium contamination even in very small quantities (below 340ppbw- parts per billion weight) still provides significant light emission.

 

Fig. 1 UV and visible normalized emissions of the ytterbium doped sample as black doted lines and ytterbium-thulium codoped sample as red solid lines (200mW cw excitation at 976 nm)

Download Full Size | PDF

To confirm this observation, a second fiber with the same ytterbium and other co doping content was drawn. In the second fiber, Tm³⁺ was intentionally added and fixed to the level of 300ppmw wich correspond to at least 1,000 times higher than in the first sample. This thulium-enriched sample will be referred to as Yb-Tm. Its mode spot size was ~8.3 µm, his core diameter was 5.61 µm and the numerical aperture was 0.16. The corresponding luminescence spectrum is reported in Fig. 1 (red line). The comparison of the two spectra confirms the attribution of the 300nm, 360nm, 475nm and 650nm bands to Tm³⁺ . These emissions are assigned, in the literature, to the (¹I₆,³P₀)→³H₆; ¹D₂→³H₆; ¹G₄→³H₆ and ¹G₄→³F₄ transitions, respectively [13–16] One emission band (around 520nm) cannot be assigned to Tm³⁺ emission, but results from a cooperative process involving two Yb³⁺ ions [12]. Such an emission band could not be detected in the ytterbium-doped sample with high thulium content, because the 475nm Tm³⁺ emission band was few orders of magnitude higher than for the initial ytterbium doped fiber, resulting in a saturation of our detection system. The presence of thulium contamination is therefore clearly established in our ytterbium-doped sample and probably in most of the samples reported in the literature. This contamination is coming from the ytterbium raw material, which is known to be contaminated by other rare earths even for its highest available degree of purity. When pumped at 976nm, thulium ions are responsible for emission bands of higher energies due to the presence of Yb³⁺ ions.

Figure 2 depicts the corresponding multi-step process for these up-converted emissions. In the centre of this figure, the energy level diagram of Tm³⁺ is reported for silicate and fluoride glass samples, for which the spectroscopic characteristics of rare earth are similar to those of our samples [17,18]. Luminescence emissions depicted in Fig. 1 is represented by green arrows on the diagram. It is clear that the 976 nm laser pump is not resonant with Tm³⁺ levels for the step 1, nevertheless Yb³⁺ ions can act as a donor. On the left of the figure, the emission spectrum of Yb³⁺ is reported derived from [19]. The Yb³⁺ will absorb the 976nm pump and transfer their energy to thulium as the emission peak of the donor, namely Yb³⁺ (~10,000 cm⁻¹) is close to the ³H₅ level absorption (~8,500 cm⁻¹) of Tm³⁺. Tm³⁺ acts as an acceptor, resulting in quasi-resonant energy transfer between Yb³⁺ and Tm³⁺ [17,20]. After non-radiative relaxation downward the ³F₄ level (~6,000 cm⁻¹), excited state absorption (ESA [21]) of the 976nm pump or energy transfer up-conversion due to the presence of another Yb³⁺ ion populate the (³F₂-³F₃) and then the ³H₄ level (~13,000cm⁻¹) non-radiatively. Further up-conversion towards higher excited levels takes place due to the combined effect of the long decay lifetime of excited levels and the resonance with the pump and/or the Yb³⁺ emission wavelengths. As in Fig. 2, up-conversion can cascade up to the highest ³P₀, and ¹I₆ levels (35,000-40,000cm⁻¹), which are responsible of emission in the UV window.

 

Fig. 2 Schematic of up conversion processes in the ytterbium-thulium doped sample: The Yb emission [21] and Tm UV [19] and the visible - infrared [20] energy level diagram are depicted together with the luminescence transitions of Fig. 1 represented by green arrows.

Download Full Size | PDF

Such a cascaded process is fostered by the high number of neighbouring Yb³⁺ ions around a Tm³⁺ ion. For the reported concentration of 1.7w%, a 5nm-radius sphere centred at a Tm³⁺ ion comprises between 50 and 100 Yb³⁺ ions. All these Yb³⁺ are thus likely to be involved in the 4 or 5 successive energy transfers up to the UV energy range. Furthermore, corresponding emissions are far from being negligible, as for a 10cm-long fiber pumped with a 200mW source at 976nm, the 360nm emission band involves, order of magnitude, of 1011 photon.s-1 and the 300nm emission one 109 photon.s-1, as an estimation for the Yb doped sample.

Further fluorescence dynamics measurements could clarify the up-conversion processes involved, either Excited State Absorption (ESA) or energy transfer. Energy transfer between Tm³⁺ ions is very unlikely, owing to the low Tm³⁺ concentration. These mechanisms are comparable to the ones reported for up conversion lasers scheme [22]. Meanwhile the 520nm emission is readily attributed to cooperative energy transfer involving two Yb³⁺ ions. It is not yet clear if this emission plays a role in the process to feed the Tm³⁺ high energy levels.

The observations and the interpretations reported above allow understanding why UV levels of Tm³⁺ in both fibers provide up-converted emission when excited at 976nm.

Because UV light is known to induce colour centres in silica-glass [23], we strongly suspected UV emission to induce photodarkening in silica fibers. So, we performed photodarkening rate measurements in our two samples: ytterbium doped and ytterbium-thulium doped fibers. The experimental setup described by Koponen et al. [24] was employed without the cooling water bath because in our case according to [25] the increase of temperature is less than 1K. The light probe was a white lamp filtered at 440nm by a monochromator; the darkening laser was a single mode transverse laser diode (650mW maximum output power) emitting at 976nm. The laser power was set to have a power density of ~10.8W.m⁻² for our two 10cm-long fiber samples, which is sufficient to guarantee complete population inversion.

In Fig. 3 , we characterized the photodarkening by representing the measured Photo Induced Absorption (PIA) of the two samples as a function 976nm irradiation time. The photodarkening rates of the two samples differ by more than one order of magnitude. This measurement demonstrates that the presence of thulium in ytterbium-doped fiber stimulates photodarkening and supports our analysis that thulium contamination plays an important role in photodarkening process.

 

Fig. 3 Photo-induced absorption as function of time at 440 nm when excited with a 10W.m⁻² power density laser at 976 nm (●: ytterbium doped sample and ▲ for ytterbium-thulium codoped sample). Inset: detail of the absorption evolution from 0 to 50 minutes.

Download Full Size | PDF

On Fig. 4 , we see the overlap between the glass host absorption and the UV emissions measured here. We measured the absorption of the preform threw a pinole selecting only the core with a Lambda 900 (Perkin Elmer) absorption spectrometer. Thus we can imagine that the silicate matrix can absorb the energy present on these emission bands. Formation of colour centres is stimulated by the interactions of the glass host and Tm³⁺ ions excited at highly levels (UV energy), be they radiative (long distance) or non-radiative (short distance). The colour centres manifest as photodarkening.

 

Fig. 4 Comparison of the host absorption spectrum (grey-shaded area) and the (normalized) up-conversion emission bands of ytterbium doped sample (black line).

Download Full Size | PDF

We believe that the presence of thulium even at contamination levels in ytterbium doped fiber contributes extensively to photodarkening. Such a conclusion is consistent with several observations previously reported on the dependence of photodarkening on the operating power [8,9] and on the rare-earth concentration [9], as well as with the erbium effect on photodarkening [8] and the relationship between glass host absorption in UV and photodarkening [26].

In conclusion, it was experimentally demonstrated that photodarkening effects could result from the presence of Tm³⁺ions that are excited to UV-emitting levels owing to cascaded up-conversion processes or ESA.

This study confirms that under great optical power densities, even contamination at ppb levels are significant. A sustainable reduction of the photodarkening effect can be anticipated by improving firstly, the purity of the ytterbium precursor elements to decrease the thulium concentration, and secondly engineering the vicinity of the rare earths atoms within the host glass, for example using nanoparticle deposition [27]. Finally, we can reduce interaction between excited thulium and glass hosts by avoiding their respective emission and absorption spectra to overlap. This can be achieved by modifying the characteristics of the UV absorption with appropriate chemical glass composition. Improving both reliability and lifetime of ytterbium doped fiber laser-based devices is undoubtedly a substantial aspect to successfully address the fiber laser market estimated $260 million worth in 2007 [1] and ever growing since then. Our propositions for reducing photodarkening can be expected to contribute an important step for a competitive and reliable high-power laser device.