1. Introduction

Ytterbium-doped silica optical fiber is now an essential element of many fiber laser systems but is known to suffer from photodarkening [1,2], a photodegradation effect that manifests itself as a gradual reduction in output laser power over time. Although several mechanisms and species potentially responsible for photodarkening, such as charge transfer mechanisms [3,4] and intrinsic defects [5], as well as mitigation approaches [6,7], have been proposed, a definitive explanation for the photodarkening process remains a subject of debate. Furthermore, while photodarkening has generally been characterized in drawn fiber [1], investigations of its precursors have focused on the fiber preforms and not the optical fiber itself.

It has long been understood that the photosensitivity of germanosilicate glasses in the ultraviolet (UV), upon which the fabrication of fiber Bragg gratings is based, is attributable to the existence of intrinsic point defects in the silica matrix such as the oxygen deficiency centers (ODCs) [8,9]. Specifically, Amossov and Rybaltovsky [8] note that: “As is generally known, the absorption band in the spectral range of 242-248 nm is connected with one of the most widespread defects in the silica glass network, namely, ODCs formed either on the basis of silicon itself or germanium…” It has also been demonstrated previously [10] that the process of drawing fibers is capable of introducing defects such as the ODCs, two of which appear to be dominant [8,9,11].

The first of these, known as ODC (I), consists of an oxygen vacancy lying between two bonded silicon atoms and is represented chemically as ≡Si-Si≡ (also known as the ‘relaxed’ vacancy). The other prominent oxygen vacancy defect, known as ODC (II) (B₂ band), has two plausible configurations [9]. One is the silylene-type defect (represented as: Si = ) in which two oxygen vacancies reside at a single silicon atom, whereas the second is represented as an oxygen vacancy on two unbounded Si sites (≡Si…Si≡, the ‘unrelaxed’ vacancy). ODCs are produced when one or more oxygen bonds in the glass are cleaved, resulting in defects that are able to profoundly impact the optical properties of the fiber. In this work, we focus on the ODC (II) variant since it has been suggested as a precursor to the E’- center and the formation of the non-bridging oxygen hole center (NBOHC) [9,11,12], both of which are participants in a UV-induced defect interconversion process that provides one potential avenue for photodarkening.

To date, few studies characterizing the optical and kinetic properties of optically-active, intrinsic defects in rare earth-doped silica fibers have appeared in the literature. To that end, the experiments reported here were designed and conducted in an effort to explore a potential linkage between Yb ions and ODCs in doped silica fibers that was proposed in Ref. 13. The subjects of the present study are silica fibers co-doped with Al and Yb (at Yb number densities [Yb] ≤ 5 × 10¹⁹ cm⁻³) and produced by an industry standard solution-doping method, and Yb:YAG-derived [14], heavily-doped ([Yb] < 2.4 × 10²⁰ cm⁻³) silica fibers.

Experiments are described in which the characteristic absorption band of ODC (II) in silica, exhibiting maximum absorption at ~246 nm [8] and a spectral width of 15-18 nm [8,9], is probed by laser excitation spectroscopy. Spectra acquired by photoexciting Yb-doped silica fibers at discrete wavelengths in the 225-265 nm interval while monitoring fluorescence at 978 (Yb³⁺) or 282 nm (ODC (II)) are found to be identical to within experimental uncertainty, demonstrating unambiguously that the photoexcited ODC (II) defect is a precursor to Yb³⁺ emission. This result confirms the recent proposal [13] of ODC → Yb³⁺ excitation transfer in Yb-doped silica fibers irradiated at λ ~250 nm, and provides further support for the premise that Yb³⁺ sites and ODC defects are coordinated. Measurements of the ODC (II) absorption spectrum in a set of fibers, in which the Yb concentration is varied, demonstrate that the ODC (II) number density is linearly proportional to the Yb concentration ([Yb], expressed in units of cm-3). Specifically, the ODC (II) defect density is ~1% of [Yb]. Furthermore, fluorescence lifetime measurements detect no precipitate-like (fast-decay clustering) regions in the most heavily-doped fibers examined to date. Finally, experiments show that two very different performs (solution-doped, modified chemical vapor deposition glass, and crystalline ytrrium aluminum garnet (YAG)) both yield glass fibers having similar absorption and emission spectra when photoexcited near 250 nm in the UV.

II. Experimental arrangement and data acquisition

Eight Yb- or Er-doped silica fibers were investigated in this study and Table 1 provides both the physical dimensions and dopant concentrations for each fiber. Fibers A, B, and C are Al-Yb co-doped and were manufactured by Nufern (East Granby, CT) with a solution doping process. Shaped claddings were provided for both Fibers B and C, but as described in Sect. III, this geometry was found to have no impact on the spectroscopic measurements. Fibers D and E are Yb:YAG-derived fibers, fabricated at Clemson University. Details regarding the fabrication and general characteristics of these fibers can be found in [14]. However, it should be noted that while the precursors to these fibers were crystalline, the resulting drawn fibers are fully vitrified. Fibers F and H are Er:YAG-derived fibers, while Fiber G is an Er-doped fiber also produced by a solution doping process. Fiber I is OFS coreless termination fiber (produced from silica) and Fiber J is an undoped YAG-derived fiber. Fibers I and J serve as references in the analysis to follow for the background absorption occurring in pure silica claddings. Microanalysis finds the composition of Fiber J to include 16.4 wt% of alumina and 23.0 wt% of yttria.

Table 1. Physical parameters for the Er or Yb-doped silica fibers investigated in these experiments. The measured peak wavelength (λ_p) and spectral breadth (FWHM) for the ODC (II) absorption band are also indicated for each fiber. The fluorescence decay constant τ is discussed in in Sect. IIIC.

View Table

Measurements of the fiber cladding absorption coefficient in the deep-UV (230 ≲ λ ≲ 265 nm) were conducted with the experimental arrangement shown schematically in panel (a) of Fig. 1 . A light-emitting diode (LED), radiating in the UV and producing a ~12.5 nm FWHM (Δλ≃24 nm at the e⁻² points of the spectral profile) continuum peaking at ~243 nm, serves as the optical source. The recent development of efficient, GaN-based LEDs in the UV now provides intense sources of broadband radiation from the visible to wavelengths below 250 nm. The spectral brightness (mW-sr⁻¹-nm⁻¹) of these new sources surpasses that of conventional UV lamps (such as deuterium) by several orders of magnitude. Optical fluence is of particular value when the number density of the absorber under study is high (i.e., N > 2/σL where N, σ, and L are the absorber number density, cross-section, and fiber length, respectively). Consequently, absorption (and emission) experiments in fibers can now be conducted readily throughout much of the UV with one or a series of UV LEDs. For the present studies, the diode described above was chosen from a set of five, each emitting at a different peak wavelength below 350 nm. A monolithic ball lens, pre-mounted onto the LED and having a focal length of 15-20 mm, produces a minimum spot size of 1.5-2.0 mm in diameter. In order to obtain cladding absorption measurements, each of the fibers tested was first treated in an acetone bath so as to remove completely the coating material. One end of the fiber was subsequently inserted into a fiber positioner and aligned with the ball lens on the UV LED (no additional lenses were necessary). The far end of the fiber was connected to a spectrometer/diode array (Ocean Optics HR 4000) by means of an SMA fitting. Owing to the exceptionally large absorption coefficients for the Yb:YAG-derived fibers, it was necessary to restrict the fiber lengths to ~7 cm. For the sake of convenience in working with such short lengths, the Yb:YAG-derived fibers (and only these fibers) were spliced to reference OFS termination fibers (such as Fiber I) with a Vytran GPX glass processing system. The reference fiber has a measured loss of 22 dB/m at 243 nm (see Table 1).

 

Fig. 1 Schematic diagrams of experimental arrangements for the following measurements: (a) Fiber absorption in the ~230-265 nm region, obtained with a UV LED (λ_peak ~243 nm); (b) Photoluminescence or excitation spectra acquired with an ultrafast Ti:sapphire laser and optical parametric amplifier (OPA) system in which β-BaB₂O₄ (BBO) frequency-doubles blue pulses into the ultraviolet. Several experiments were also conducted with a KrF excimer laser (248 nm) as the excitation source; (c) Apparatus for Yb³⁺ lifetime measurements.

Download Full Size | PDF

If the degree of mode mixing during the propagation of an optical wave in a fiber is sufficient, then absorption in the fiber core can be determined by simply multiplying the measured cladding absorption by the ratio of the cladding cross-sectional area to that for the core. Inadequate mode mixing can result in severely under-estimating the core absorption, and measurements of cladding absorption coefficients that are dependent upon sample lengths. Fibers with circular claddings often suffer from this difficulty [15] whereas non-circular claddings generally exhibit mode mixing sufficient to yield length-independent loss measurements. Having confidence, therefore, in the cladding absorption measurements described above demands that fiber length-dependent experiments be conducted and the results are presented in Fig. 2 for Fibers D and E. Despite varying the lengths of both fibers by more than an order of magnitude, the measured absorption coefficient remains constant to within experimental error and the largest fluctuations in the data are observed with the shortest fibers. These results, as well as those of Table 1 and data to be presented later, demonstrate that shaped claddings (such as those of Fibers B and C) are not necessary for obtaining reliable absorption and emission data.

 

Fig. 2 Dependence of the cladding absorption coefficients (dB/m) on fiber length for Fibers D and E (cf. Table 1). For clarity, data for Fiber D were intentionally reduced uniformly by 100 dB/m. All measurements were recorded for a probe wavelength of nominally 243 nm.

Download Full Size | PDF

Photoluminescence data and laser excitation spectra were obtained with the experimental arrangement of Fig. 1(b). An optical parametric amplifier (OPA), pumped by a Ti: sapphire ultrafast laser system (~2.7 mJ, ~100 fs pulses, ~1 kHz pulse repetition frequency), produces ~120 fs pulses tunable throughout the visible and near-infrared. The output of the OPA is frequency-doubled in β – BaB₂O₄ (BBO, 10 mm x 10 mm x 0.5 mm, cut at 39° with respect to the crystal’s optical axis). This system generates pulses tunable in wavelength from ~225 nm to 265 nm. After this radiation was launched into the fiber under test with a UV-grade lens having a focal length of 15 cm, the fluorescence produced within the doped fiber was recorded with a fiber-coupled spectrometer. Several experiments were also conducted in which the ultrafast laser system and doubling crystal were replaced by a KrF excimer laser (248.4 nm, ħω ≈5 eV, ~25 ns pulses). To our knowledge, laser excitation of optically-active defects in fibers has not been reported prior to Ref. 13 but the narrow bandwidth, temporal resolution, and brightness afforded by these coherent sources are of considerable value in clarifying the identity of the defect, dopant, or impurity under study.

Panel (c) of Fig. 1 is a diagram of the apparatus with which the temporal decays of Yb³⁺ fluorescence in the near-infrared and green (owing to Yb³⁺- Yb³⁺ interactions) were recorded for several fibers. A pulse-modulated pump laser diode (JDSU 27-2601), operating at 975 nm, photoexcited short segments of Yb-doped fiber spliced between two passive fibers. The fiber length and pump power levels were chosen such that the observed amplified spontaneous emission (ASE) was negligible. A monochromator or a bandpass (interference) filter selected the wavelength region of interest and the Yb³⁺ fluorescence waveforms were detected and recorded with a Si photon-counting avalanche photodiode (APD) and a scaler/averager.

III. Experimental results and conclusions

A. Absorption Spectra and Coefficients

Measurements of the cladding absorption spectra in the ~228 – 270 nm wavelength region for two fibers (C and E, Table 1) are presented in Fig. 3 . Acquired with the UV LED described in Sect. II, these spectra have been normalized to the LED emission profile and are representative of those obtained for all of the fibers studied. As indicated by the dashed black curve of Fig. 3, fitting a Gaussian to the experimental Fiber E spectrum yields an absorption peak and breadth of 245 nm and 19 nm, respectively. Both characteristics are consistent with the known parameters of the deep-UV absorption of the ODC (II) defect in silica [8,9]. Table 1 summarizes the results of cladding absorption measurements for all of the fibers examined in these experiments. As discussed previously, the absorption coefficient in the core for each fiber (α_core) is determined by scaling the cladding coefficient (α_clad) by the ratio of the cladding-to-core cross-sectional areas. Furthermore, the entries of Table 1 assume that the ODC (II) background absorption of 22 dB/m for the coreless silica fiber (Fiber I, Table 1) is identical to that for the silica claddings of the doped fibers [10]. Consequently, 22 dB/m was subtracted from α_clad for each fiber prior to calculating α_core. The measurement of cladding absorption for the undoped, YAG-derived fiber (Fiber J, Table 1) demonstrates that the presence of Y and/or Al contributes to an insignificant degree to the optical loss. Also, the difference of 2 dB between α_clad for Fibers I and J lies within the estimated measurement uncertainty of ± 10%. Figure 4 illustrates the absorption measurements of Table 1 for both Yb- and Er-doped fibers. Data showing the dependence of α_core on [Yb] and [Er] are represented in the figure by the solid circles (●) and open circles (○), respectively and, to within experimental error, α_core varies linearly with the Yb concentration in the fiber. Although the data of Fig. 4 are limited to five fibers and further measurements at additional [Yb] values are warranted, these results suggest that the ODC (II) number density is linearly proportional to the Yb doping concentration.

 

Fig. 3 Measured cladding absorption spectra for Fiber C (green) and Fiber E (red). The least-squares fit of a Gaussian to the Fiber E spectrum yields the dashed black curve having a peak wavelength and spectral breadth of λ₀ = 245 nm and Δλ = 19 nm, respectively.

Download Full Size | PDF

 

Fig. 4 Dependence of fiber core absorption on the rare earth number density. All of the data were recorded with the LED source operating with a peak emission wavelength of ~243 nm [see text]. Linear least-squares fits to both the Yb and Er data of Table 1 are shown, as are estimated uncertainties for several of the measurements.

Download Full Size | PDF

We note in Table 1 an apparent, slight blueshift of ~3 nm in the peak absorption as the Yb concentration is increased from [Yb] ~9 × 10¹⁸cm⁻³ to 2.4 × 10²⁰cm⁻³. Such a trend, if confirmed by further studies, is statistically significant and is attributable to the rising influence of other absorbers such as the peroxy radical having an absorption band peaking at ħω ≈5.4 eV (λ ~230 nm) [9,16–18]. Other potential contributors to absorption in this spectral region include E’ centers and NBOHC defects having absorption maxima lying at ~215 nm and ~258 nm, respectively. However, because the absorption spectra associated with E’ centers and NBOHC defects are considerably broader than that of ODC (II), we conclude that the contribution of E’ and NBOHC defects to the data of Fig. 3 can be neglected.

If the peak absorption cross-section for the ODC (II) continuum of Fig. 3 is assumed to be 2 × 10⁻¹⁷cm² [11], then the ODC number density for Fiber E (for example) is estimated (assuming Beer-Lambert absorption) to be ~2.6 × 10¹⁸ cm⁻³, or approximately 1% of [Yb]. Furthermore, the ratio of the ODC number density to that for Er³⁺ in fibers such as G and H (Table 1) is an order of magnitude lower (i.e., ~0.1%) than the corresponding ratio for the Yb-doped fibers. On this basis, it is reasonable to assume that if the ODC (II) defect is found to be the primary cause of (or contributor to) photodarkening, Er-doped fibers can be expected to be considerably less susceptible than their Yb-doped counterparts to photodegradation effects.

Before leaving this section, the potential influence of absorbers other than ODC (II) on the results of Figs. 3 and 4 and Table 1 should be addressed. Specifically, Engholm and Norin [3,4,6] observed in 2007 strong absorption in Yb-doped aluminosilicate fiber preforms in the deep-UV. Peaking at ~230 nm and having a FWHM of almost 40 nm, the reported continuum was attributed [4] to a charge transfer (CT) absorption mechanism in which the charge state of a Yb ion in the glass is reduced from + 3 to + 2 by an electron donated from the surrounding matrix. Because of the emission bandwidth of the GaN diodes employed for the absorption measurements in this study, it is not possible to be dogmatic as to the presence or absence of CT absorption in the fibers reported here. Nevertheless, the absorption spectra of Fig. 3 are virtually identical to those measured previously for the ODC (II) defect in silica. Furthermore, because of the breadth and strength of the CT absorption reported in Refs. 3, 4, and 6, it is expected that the CT continuum would underlie the comparatively narrow ODC (II) spectrum and significant absorption beyond 265 nm would be observed. Neither of these is evident and it can be concluded that, although the possible presence of CT absorption cannot be dismissed outright, the absorption spectra in the deep-UV for all of the fibers examined are fully explained solely by ODC (II) absorption. Finally, we note that no luminescence resulting from CT absorption has been observed [4]. The present experiments are in agreement with this report because, as described in detail in the next section, photoexcitation of the ODC (II) defect throughout the absorption spectrum of Fig. 3 (with fixed wavelength or tunable sources) yields only emission attributable to the ODC (II) or NBOHC defects in silica [19], or to Yb³⁺. Because the experiments of Refs. 3, 4, and 6 were conducted with preforms whereas the present study employed fiber exclusively, one interpretation of the apparent absence of (or minor role played by) CT absorption in the present experiments focuses on the process of drawing fiber. It is possible that the procedure for producing fiber from a preform is responsible for a transformation in the optical properties of rare earth-doped silica, particularly insofar as the optically-active defects are concerned. Such a conjecture is consistent with [10].

B. Photoluminescence and Laser Excitation Spectra

The direct connection between the populations of ODC (II) defects and Yb³⁺ ions in doped fibers, implied by the data of Fig. 4, is corroborated by emission spectra recorded in the UV, visible, and near-infrared regions. Figure 5 is representative of the panoramic spectra acquired over the 200-1100 nm wavelength interval when Yb-doped fibers are photoexcited with UV pulses from the KrF excimer laser (λ = 248.4 nm, ħω ≈5 eV). The specific spectrum shown in the figure is that for Fiber C of Table 1 but it is obvious that pumping the absorption band of Fig. 3, associated with the ODC (II) defect, does indeed result in intense fluorescence that is known to originate from the ODC. Two prominent continua, peaking at ~280 nm (ħω = 4.4 eV) and ~480 nm (ħω = 2.6 eV), are observed in Fig. 5 and both have been studied extensively previously [8,9,13]. Weak emission from the NBOHC defect in silica [13,19] is also evident. Of greatest interest, however, is the presence of Yb³⁺ emission lying between ~900 nm and 1050 nm. Similar Yb³⁺ emission profiles are observed when the Yb-doped fibers are illuminated with the UV LED at λ ~243 nm. These results demonstrate that a fraction of the energy absorbed by ODC (II) defects via the ~19 nm FWHM continuum of Fig. 3 appears as Yb³⁺ luminescence, thus indicating the existence of ODC (II)→ Yb³⁺ excitation transfer as proposed in Ref. 13.

 

Fig. 5 Emission spectrum recorded over the 200-1100 nm region when a Yb-doped fiber (Fiber C, cf. Table 1) is pumped by a KrF excimer laser (λ_L = 248.4 nm, ħω ≈5 eV). Representative of the emission observed when any of the Yb-doped silica fibers of this study are photoexcited at 248 nm, this spectrum comprises fluorescence generated by optically-active defects in silica (such as ODC (II)) and Yb³⁺. Faint emission from NBOHC defects is detected, and all of the prominent features in the spectrum are identified. Note that the λ > 380 nm portion of the spectrum has been magnified in intensity by a factor of two.

Download Full Size | PDF

Despite the results described above, it is laser excitation spectroscopy that provides an unambiguous link between ODC (II) absorption and Yb³⁺ photoexcitation. Specifically, a series of experiments were conducted in which relative emission intensity spectra, recorded over the 200-1100 nm wavelength region, were obtained while tuning the wavelength (λ_L) of the laser excitation over the ODC (II) absorption spectrum of Fig. 3. In these studies, the output of the Ti: sapphire/OPA/BBO, ~120 fs laser source of Fig. 1(b) was scanned in discrete increments over the ~225-265 nm wavelength interval and representative spectra, recorded for four specific values of λ_L, are presented in Fig. 6 . For λ_L = 231 nm, only faint emission is observed throughout the UV, visible, and near-infrared spectral regions because absorption by the ODC (II) defect (Fig. 3) is weak. However, as the laser system is tuned closer to the ODC absorption maximum, intense emission from both the 4.4 eV and 2.6 eV bands [8,9] of the ODC (II) defect, weak NBOHC fluorescence, and Yb³⁺ radiation (λ ~1 μm) are all observed. Weak features due to OPA (fundamental) radiation transmitted by the BBO crystal (labeled “2 × λ_L” in Fig. 6) and leakage of the Ti:sapphire laser pump (λ ~780 nm) through the OPA/BBO system are also evident in Fig. 6. In short, the excitation spectra of Fig. 6, obtained with a tunable laser source, and the KrF fixed wavelength laser (248.4 nm) are consistent in showing emission features attributable to Yb³⁺ or an optically-active, intrinsic defect in silica.

 

Fig. 6 Spectra representative of those recorded when the wavelength of the Ti:sapphire/OPA laser system is scanned from 225 nm to 265 nm. Key features in the spectra (including artifacts such as the weak pump signal in second order and Ti:sapphire laser leakage at ~780 nm) are identified. All of the spectra shown were acquired with Fiber C (cf. Table 1).

Download Full Size | PDF

Information regarding the dynamics of energy transfer in the UV-photoexcited Yb:silica matrix can be gleaned from spectra similar to those of Fig. 6. Specifically, the dependence of the relative ODC (II) and Yb³⁺ yields on the laser wavelength (λ_L) are readily determined and the results are summarized in Fig. 7 . Experimental measurements in both panels are represented by the open circles and, for convenience, all data are normalized to the peak value of the relative intensity at λ_L ~248 nm. Assuming the ODC (II) absorption band profile of Fig. 3 to be represented by a Gaussian, then the functional form of the excitation spectra can similarly be expressed in terms of the Gaussian

 

Fig. 7 Laser excitation data acquired in the ~225 - 265 nm wavelength range by monitoring the relative ODC (II) (a) or Yb³⁺ (panel (b)) fluorescence intensity at ~282 nm and 978 nm, respectively, as the UV laser source (Ti:sapphire/OPA/BBO) wavelength was scanned. The solid curves represent best fits of Eq. (1) to the data and both spectra are normalized to the peak intensity (at ~248 nm).

Download Full Size | PDF

C. Yb³⁺ Fluorescence Decay

Further insight into the origin of the spectra of Figs. 5 and 6, and the local environment in which the emitters reside, is provided by measurements of the temporal decay of Yb³⁺ fluorescence. When Yb³⁺ is photoexcited at 975 nm, spontaneous emission in the near-infrared and visible is generated, as illustrated in Fig. 8 . Panel (a) of the figure shows the characteristic continuum of Yb³⁺ peaking near 1.03 µm and the blue-green fluorescence of Fig. 8(b) is the result of a cooperative (ion-ion), upconversion process [20]. With the apparatus of Fig. 1(c), the temporal history of Yb³⁺ fluorescence, produced in response to 1 µs laser pulses at 975 nm, was recorded for several fibers and representative results are presented in a semilog format in Fig. 9 for Fibers A, D, and E. These data were chosen to illustrate the monotonically increasing decay rate that is observed as the Yb³⁺ number density is raised. All of the red curves were acquired by monitoring the temporal history of the Yb ion fluorescence in the long-wavelength tail of the 1.03 µm continuum (Fig. 8) at ~1.06 µm. It is evident that the near-infrared fluorescence decays exponentially over a range in intensity of at least two orders of magnitude. This behavior indicates that the ODCs are not formed at devitrification sites, nor are they the result of precipitate-like clustering [21]. This conclusion is reinforced by similar data obtained for the blue-green spectrum of Fig. 8(b). Recording the temporal dependence of the Yb³⁺ - Yb³⁺ cooperative emission (monitored near 530 nm) yields the green curve of Fig. 9 for Fiber E. Although the declining fluorescence is best fit with a double exponential, the initial (fast) decay constant is 294 ± 10 µs or almost precisely 50% of the decay constant for the ~1.03 µm waveforms for Fiber E (τ = 584 ± 5 µs; cf. Table 1).

 

Fig. 8 Spontaneous emission generated in a Yb-doped fiber (Fiber C, Table 1): (a) the near-infrared (~950 – 1100 nm), and (b) the blue-green region of the spectrum (460-560 nm) by photoexcitation of a Yb-doped fiber (Fiber E, Table 1) at 975 nm. Both spectra were recorded at 90° to the fiber axis and, as a reference, the spectrum of the 975 nm pump for these experiments is shown in red in panel (a).

Download Full Size | PDF

 

Fig. 9 Temporal decay of Yb³⁺ fluorescence following the photoexcitation of three fibers at 975 nm (cf. Figure 1 (c)). Measurements of the Yb³⁺ spontaneous emission at ~1.06 µm are indicated by the red profiles for Fibers A, D, and E of Table 1. The green curve reflects the temporal history of emission produced near 530 nm by cooperative ion processes (upconversion) in Fiber E. Note that the ordinate in logarithmic.

Download Full Size | PDF

It is also instructive to consider the dependence of the near-infrared exponential decay constant (τ) on the Yb³⁺ number density, [Yb³⁺]. Measurements for the five Yb-doped fibers examined in these experiments are shown in Fig. 10 in which the Al/Yb co-doped fibers (Table 1, Fibers A-C) are represented by solid circles (●). The two open circles (○) in the figure denote the measured lifetimes for the Yb:YAG-derived fibers of Table 1. The curve in Fig. 10 is the best fit to the data of the relation [22]

 

Fig. 10 Measured dependence of the Yb³⁺ radiative lifetime τ on the Yb number density of Yb-doped silica fibers. The solid curve represents the best fit of Eq. (2) to the data, which yields the quenching number density, [Yb]_q, of 4.1 × 10²⁰ cm⁻³. Solid circles (●) represent data obtained for the Al/Yb co-doped fibers fabricated by a solution doping process (Fibers A-C, Table 1) whereas the two open circles (○) denote measurements for the Yb:YAG-derived fibers (Fibers D and E, Table 1).

Download Full Size | PDF

IV. Summary and conclusions

Absorption measurements and laser excitation spectroscopy of Yb-doped fibers, solution-doped or YAG-derived, have been described. Absorption spectra in the deep-UV (~228 – 280 nm) show consistently the ODC (II) absorption continuum peaking near 248 nm and measurements demonstrate that the ODC (II) absorption coefficient is directly proportional to the number density of the Yb dopant for both solution-doped and YAG-derived fibers. Not only do these data provide evidence directly linking the ODC (II) number density to that for Yb but they also suggest that the appearance of the ODC defect is associated with the introduction of Yb to the fiber and, possibly, the process by which the fiber itself is formed [10]. Furthermore, the ODC (II) number densities in Er-doped fibers are an order of magnitude lower than those observed for their Yb-doped counterparts.

Photoexcitation of the ODC (II) absorption spectrum with a KrF (248 nm) laser indicates the existence of ODC (II) → Yb³⁺ excitation transfer. Unambiguous confirmation of this energy transfer mechanism is provided by laser excitation spectroscopy experiments in which the ODC (II) absorption band was pumped at several wavelengths in the ~225 – 265 nm region by a tunable Ti:sapphire/OPA/BBO laser system. Monitoring ODC (II) and Yb³⁺ fluorescence at 282 nm and 978 nm, respectively, reveals a variation with the UV laser wavelength that is virtually identical for both emitters. Furthermore, both the Yb³⁺ and ODC (II) excitation spectra in this UV region match precisely the ODC (II) absorption spectrum, thereby reinforcing the conclusion that electronically-excited Yb³⁺ is formed directly in these experiments by ODC (II) → Yb³⁺ excitation transfer.