1. Introduction

The interaction between glass materials and femtosecond (fs) laser pulses can result in a variety of effects, including changes in: (i) local composition, (ii) valence states of certain ions, (iii) optical properties, as well as (iv) nucleation of nanoparticles and crystallization [1–3]. One of the most technologically interesting applications is the fabrication of photonic structures such as waveguides, splitters or interferometers [4–8], by fs-laser driven modification of the refractive index of a glass medium. Understanding and controlling the processes governing such structural alterations in glasses is important for the fabrication of photonic structures. The most common substrate investigated frequently is pure amorphous silica, with focus on writing of optical waveguides [4–8]. A consistent picture of fs-laser induced alterations, the mechanisms linked to the creation of microstructural defects and of defect-assisted densification in different glass networks still remains elusive.

When a glass medium is additionally doped with luminescent rare-earths ions (REs), fs-laser processing can be used to fabricate high-gain optical devices, such as waveguide lasers and waveguide amplifiers [9–12]. For instance, Er³⁺- doped glasses can function as gain media in fiber lasers and amplifiers using the ∼1.5 µm emission, the standard telecommunications wavelength [13–14]. The majority of studies that report on monitoring of fs-laser created defects in glasses have focused only on un-doped glasses, mostly silicate or phosphate compositions [4–10]. In fact, when a glass is doped with Er³⁺, the fundamental study of optically active defects becomes very challenging. This is because the strong emission bands of Er³⁺, spanning from visible to infrared, dominate over the emission signatures (luminescence), which are used to monitor the formation of fs-laser created defects.

Optical amplification requires a host glass that is capable of accommodating high concentration of REs luminescent dopants, such as Er³⁺, to achieve a sufficient pump-light absorption and net gain. However, increasing the concentration of Er³⁺ is often counteracted by the detrimental effect of erbium clustering and the deleterious phenomenon of concentration quenching (i.e. the decrease of the Er³⁺ luminescence intensity with increasing Er³⁺ concentration). We have recently demonstrated [15] a few non-esoteric compositions of oxide glasses, capable to accommodate a high level of Er³⁺ doping, while avoiding the deleterious effects of concentration quenching.

In this work, we report on the role of multi-pulse fs-laser exposure on the structural changes in one of the above mentioned Er³⁺-doped and heavy-metal co-doped oxide glasses. We examine the relationships between, on one hand the exposure conditions, such as laser wavelength, pulse energy, irradiation time and on the other hand the microstructural modifications in the glass network. In the first section, we induce and subsequently probe, using Raman spectroscopy, the subtle differences in the glass network architecture, which are generated by different laser wavelengths. In the second section, using luminescence spectroscopy, we monitor the fs-laser driven formation of optically-active defects in the glass network. We demonstrate how to decouple the luminescence signals of laser-induced optically-active defects from the much stronger and usually dominating emission of Er³⁺.

2. Experimental details

A glass of composition 4 K : 8Ga : 24Si : 64O ratio in at. %, doped with 1 mol% of Er₂O₃, was prepared using analytical grade reagents of 99.999% purity and was synthesized by the high-temperature melting and quenching route. Details of synthesis have been described elsewhere [15]. For optical and microstructural studies bulk glass samples were cut into rectangular plates of 20 × 10 × 2 mm and their surfaces were polished to optical quality. Raman spectra were measured using a confocal spectrometer (LabRam HR Evolution) equipped with 1800 lines/mm grating and a 442 nm continuous wavelength laser. All spectra were measured in a backscattering configuration with the incoming and scattered light focused and collected using an Olympus microscope (BX41) with 100x (NA = 0.95) microscope objective. A CCD detector (1024 × 256 pixels) was used for recording spectra in the range between 200 cm⁻¹ and 1200 cm⁻¹ with a spectral resolution of ∼1 cm⁻¹. The same optical setup was used to collect emission spectra. A train of 50 fs pulses (FWHM of intensity) at 1 kHz repetition rate, 1.6 mJ energy each, centered at 795 nm (and referred to as 800 nm in the remainder of the text), is generated by a Ti:sapphire oscillator-regenerative amplifier. These near-infrared pulses are frequency doubled to 400 nm by a 0.5 mm thick Lithium-tri-Borate (LBO) crystal pulse, creating a train of 50 µJ pulses. For irradiations, the laser beam was directed onto the surface of a polished glass plate covered with a circular mask (∼5 mm in diameter) to define the area of exposure. All laser irradiation runs used a repetition rate of 1 kHz to avoid heat accumulation effects.

3. Results and discussion

3.1 fs-laser irradiation, network densification & Raman

The material investigated in this project belongs to the oxide-type glass family and to the category of classic network glasses. As such, its architecture is based on building blocks of well-ordered tetrahedra (SiO₄), linked together by bridging oxygens, to form a 3-dimensional, disordered network arranged into differently sized, tetrahedral ring structures. The majority of tetrahedra are connected to form 6-fold rings, but bigger and smaller rings are also present. Within this network architecture, and as demonstrated in our previous works [15,16], the heavy metal dopant, gallium, takes the role of a tetrahedral glass network former, while the erbium luminescent dopant acts as a network modifier and assumes an octahedral coordination with oxygens. Gallium forms gallium-oxygen tetrahedral units (GaO₄) which are charge-balanced by a network modifier - an alkali ion (K⁺). This allows the GaO₄ to interconnect nicely with the main glass network formers, which are silicon oxygen tetrahedra (SiO₄) [15,16].

Here we discuss the key features that can be extracted from Raman mode assignments in our glasses, pre- and post-irradiation (Fig. 1). It is useful to recall that Raman bands are associated with the vibrational density of states of glass forming units and their spectral widths reflect the degree of structural disorder, specific to a given glass composition (Fig. 1(a)).

 

Fig. 1. Raman spectra: example of quantitative analysis (Lorentzian and Gaussian deconvolutions) of ring statistics illustrating the diversity of tetrahedral ring structures, from 3-fold to >6-fold rings in: (a) pre-irradiated glass, and (b) post-irradiated glass (800 nm, 1 kHz, 50 fs); (c) sketch illustrating the change in relative ring contribution in pre- and post-irradiated glasses as established from quantitative analysis. Insets are schematics of tetrahedral ring structures.

Download Full Size | PDF

A Raman spectrum of the glass material studied in this project, measured before fs-laser irradiation, is shown in Fig. 1(a). The overall spectral shape is similar to that of pure SiO₂ glass and its detailed Raman mode assignments can be found in literature [15,17–21]. The spectrum is dominated by a broad, intense main glass band (M), which reflects symmetric stretching motions of bridging oxygens (Si-O-Si) of 6-fold SiO₄ tetrahedral rings. Overlapped with the main glass band is the D₁ ‘defect’ band, assigned to the stretching motions of bridging oxygens in 4-fold SiO₄ rings. The D₁ band appears as a sharp shoulder in pure SiO₂ glass, whereas in our glass it is merged with the M-band. The second glass ‘defect’ band, marked as D₂, is assigned to the stretching motions of bridging oxygens in 3-fold SiO₄ rings. The deconvolutions in Fig. 1(a) and Fig. 1(b) illustrate how the Raman bands can be used to quantify the modification in ring statistics going from pre- to post-irradiated glasses. A comparison of average ring statistics in the networks of pre- and post-irradiated glasses, shown in Fig. 1(c), illustrates the diversity of tetrahedral rings, from 3-fold to >6-fold rings, as well as shows how their relative contribution varies after irradiation. For both, the pre- and the post-irradiated, glasses it is apparent that 6-fold tetrahedral SiO₄ rings dominate. However, the area under the defect bands D₁ and D₂ increases measurably as a result of fs-pulse irradiation, indicating that in the irradiated glass there is a larger contribution of smaller, 3- and 4-fold rings, formed at the expense of 6-fold and larger rings (Fig. 1 (c)). Such change in ring statistics is evidence that the glass network, within the irradiated volume, experiences a fs-laser-driven densification, which leads to a positive refractive index change, according to literature [4–9].

 

Fig. 2. Raman spectra of pre- and post-irradiated glasses using different irradiation doses: (a) irradiation at 800 nm, 1 kHz, 50 fs and (b) irradiation at 400 nm, 1 kHz, 50 fs; (c) - (d) zoom of the Raman D₂ band area, (e) - (f) D₂ defect band normalized to the D₂⁰ area of the pre-irradiated glass as a function of irradiation dose and (g) - (h) D₂ defect band normalized to the D₂⁰ area of the pre-irradiated glass as a function of number of laser pulses. In the pre-irradiated glass, the D₂/D₂⁰ ratio is equal to 1. Dashed lines are guides for the eye.

Download Full Size | PDF

In the course of laser irradiation experiments there are few parameters that can be used to control the process: laser wavelength (λ), laser repetition rate, pulse energy (E), pulse duration (τ_p) and laser fluence (F = E/area). In our irradiation experiments the pulse energy, fluence, and wavelength are tied together due to the nature of the experimental setup used. Specifically, the 800 nm irradiation (repetition rate 1 kHz, pulse duration 50 fs) deposits in glass > 32 times more energy per pulse, compared to the 400 nm irradiation (repetition rate 1 kHz, pulse duration 50 fs). The Raman spectra measured before and after irradiation, for two wavelengths, at different irradiation doses, are shown on Fig. 2 (a) and Fig. 2(b). In our experiments, the irradiation dose is defined as the product of exposure time (or number of pulses) and the energy per pulse. The trend of glass network densification, seen earlier in Fig. 1 (b), can be followed more in details in Fig. 2 (c) and Fig. 2 (d), which display a zoom of the Raman D₂ band regions. To quantify the microstructural changes in glass network in terms of the amount of 3-fold tetrahedral rings, we use the D₂ defect band normalized to the D₂⁰ area of the pre-irradiated glass, as a function of dose (Fig. 2(e) and Fig. 2(f)) or as a function of number of pulses (Fig. 2(g) and Fig. 2(h)). The increasing D₂/D₂⁰ ratio of areas is a measure the fs-laser induced increase in the number of 3-fold rings and therefore a measure of network densification. We conclude that, for each irradiation wavelength, densification becomes more effective for higher irradiation dose and that for 800 nm irradiation this process exhibits a close to nonlinear behavior. Our observation is consistent with literature studies of irradiation of pure silica glasses, without any REs luminescent dopants, where a gradual growth of D₂ band area with laser irradiation doses was also reported and associated with densification and with an increase in the glass refractive index [22–25].

The results of Raman studies can be summarized into a few key findings. Femtosecond laser exposure of Er³⁺-doped and heavy-metal co-doped oxide glass, leads to various degrees of microstructural reorganization in the network. Specifically, in the short-to-medium range order, such exposure generates an increasing number of small, i.e. 3- and 4-fold tetrahedral rings in the backbone of the irradiated volume of the glass. The formation of such low-dimensional, strained rings occurs at the expense of 6-fold and larger rings, increases connectivity of the network and leads to the creation of a more compacted network, meaning glass densification.

3.2 fs-laser irradiation, optically active defects & luminescence

In the first section of this work we used Raman spectroscopy to follow the densification of the glass network as a result of fs-laser irradiation. In this section we turn to luminescence spectroscopy to follow the formation of optically active defects in the irradiated volume of the Er³⁺-doped glass. Specifically, we demonstrate how to decouple the weak luminescence signals of laser-induced, optically active defects from the much stronger and usually dominating emission bands of Er³⁺.

Irradiation can generate different types of defects in glass by breaking of bonds in the network. We focus here on the search for optically active defects that have a spectral signature in the visible range of emission (Fig. 3 and Fig. 4). One type of such defects are non-bridging oxygen hole centers (NBOHC) and are important for the fabrication of optical devices written in glass. NBOHCs, already reported in irradiated pure silica and phosphate glasses [26,27], result from the breaking of O-Si-O bonds in the 3-dimensional network (Fig. 3(d)). Those defect centers are represented chemically as [≡ Si − O•] where [≡] symbolizes bonding to three oxygens and [•] represents an unpaired electron, i.e. a dangling bond [26,27]. NBOHCs, have absorption near 2.0 eV and luminescence near 1.9 eV, and are called R-centers or red centers.

 

Fig. 3. Luminescence spectra of glasses before and after 800 nm fs-laser irradiation, at the highest dose: (a) the full spectral range of the pre-irradiated glass (dotted line) and of post-irradiated glass (solid line), both showing the typical, intense Er³⁺ emission bands; (b) zoom on the spectral region of interest, showing a significant irradiation-generated background (shaded area and dash-dot line) in the post-irradiated glass; (c) a Gaussian deconvolution of the irradiation-generated background, illustrating the contribution of two kinds of defect centers; (d) the process of creation of defects, a sketch of the fs-laser irradiation geometry and a white light image of a post-irradiated area of glass.

Download Full Size | PDF

 

Fig. 4. Luminescence spectra of glasses before and after 400 nm fs-laser irradiation, at the highest dose: (a) the full spectral range of the pre-irradiated glass (dotted line) and of post-irradiated glass (solid line), both showing the typical, intense Er³⁺ emission bands; (b) zoom on the spectral region of interest, showing a significant irradiation-generated background (shaded area and dash-dot line) in the post-irradiated glass; (c) a Gaussian deconvolution of the irradiation-generated background, illustrating the contribution of two kinds of defect centers; (d) the Er³⁺ energy diagram.

Download Full Size | PDF

The luminescence spectra of our glasses, before and after fs-laser irradiation using two laser wavelengths, 800 nm and 400 nm, are shown in Fig. 3 and Fig. 4. Figure 3(a) and Fig. 4(a) show the typical, intense Er³⁺ emission lines in the 450-1050 nm spectral range (for band assignment see the erbium energy diagram in Fig. 4(d)). For each irradiation wavelength, the spectra compare the pre-irradiated glass with the post-irradiated glass at the highest dose applied. Figure 3(b) and Fig. 4(b) show a zoom on the spectral region of interest, where optically active defect centers are expected. In both cases, the zoom highlights the almost flat background of the pre-irradiated glasses versus the significant, broad luminescence signal that appears in the post-irradiated glasses. The observed excess of luminescence originates from fs-laser-created optically active defects, which are formed by cleavage of O-Si-O bonds in our glass network (Fig. 3(d)), leading to network depolymerization.

The broad luminescence signal from defects can be separated into two bands based on Gaussian deconvolution. The bands in Fig. 3(c) and Fig. 4(c) can be assigned to the presence of at least 2 defect centers: the first defect band is centered at ∼535 nm or ∼545 nm (for 800 nm or 400 nm irradiation, respectively) and the second defect band is centered at ∼660 nm or ∼670 nm (for 800 nm or 400 nm irradiation, respectively).

To the best of our knowledge, no study has reported the detection of luminescence from defect centers in glasses, when an Er³⁺ dopant is also present. In fact, strong emission bands of typical Er³⁺ radiative transitions can overwhelm [11,28,29] the weak intensity luminescence signals from irradiation-created defect centers, when those signals occur in the same spectral range. In this work we are able to successfully decouple the very weak luminescence of laser-generated defects from the dominating luminescence of Er³⁺. According to its spectral position, the defect band observed in our glass at ∼660 or ∼670 nm (or ∼1.9 eV) indicates the formation of NBOHCs (Fig. 3(c) and Fig. 4(c). The origin of the second band observed in our glass at ∼535 or ∼545 nm (∼2.3 eV) is debatable and differing interpretations can be found in literature, for various glasses, but without any REs or Er³⁺ dopant.

Different optically active defects created by fs-laser exposure have been reported in silica, phosphate and borate glasses and glass fibers. In works by Reichman et al. [30] and Nishikawa et al. [31], a ∼540 nm emission band was reported in fused silica after low-dose irradiation with an 800 nm fs-laser and it was attributed to self-trapped exciton defect E^’_δ centers from very small silicon nanoclusters. Still for various silica and silica-based glasses, but without any RE dopants, a luminescence band centered at ∼630 nm or ∼650 nm was detected, after fs-laser irradiation, and was attributed to NBOHC defects [30–36]. For example, in fused silica irradiated at 800 nm, Mischik et al. [35] observed two luminescence bands at ∼650 nm (NBOHC) and at ∼530 nm, which was labelled as an undefined laser-induced defect center. For commercial Schott Borofloat-33 glass irradiated by an fs-laser at 1031 nm, Babu et al. [36] reported a strong, red, asymmetric emission centered at 680 nm (NBOHC) and a weak band ∼546 nm proposed to originate from silanol groups (Si–O–H). In phosphate glasses after fs-laser writing, Fletcher et al. [8] observed a band centered at ∼600 nm and attributed it to phosphorus oxygen hole center (POHC) defects. Finally, one unlikely candidate for the origin of our emission at ∼535-540 nm could be oxygen deficiency centers (ODCs), but they are reported to have an emission rather in the blue region at 2.6-2.7 eV or ∼460-470 nm in commercial Suprasil/Infrasil glass [37] or in Yb-doped silica glass fibers [38].

In summary, while NBOHC defects are well documented, there are contradictory opinions on the nature of the emission at ∼535-540 nm and, as for today, no study has identified the nature of those defects. Our studies provide a point in further discussion on the origin and exact nature of those defect centers.

4. Conclusions

The present work examines effects of fs-laser irradiation on the microstructure and optical properties of an Er³⁺-doped and heavy-metal co-doped oxide glass medium. Our studies tie together Raman spectroscopy, which provides supporting evidence of changes to the glass network topology, and luminescence spectroscopy, which shows evidence of atomic level changes in the network. Irradiation by fs-laser leads to bond breaking in the backbone of the glass network - as shown by Raman – and to the formation of two types of optically-active defects - as shown by luminescence. We demonstrate how to decouple the very weak luminescence signals of laser-generated optically-active defects from the dominating emission of the Er³⁺ dopant. Femtosecond laser exposure ultimately brings a gradual, defect-driven glass network densification within the laser-irradiated volume. This study extends our understanding of fs-laser irradiation-driven structural modifications and demonstrates defect monitoring in Er³⁺-doped oxide glasses.