1. Introduction

Since the first experimental demonstration of permanent and photo-induced modifications in glasses through the ultrafast laser-matter interaction by Davis et al. [1], the development of three-dimensional photonic components has emerged in many areas including optical data storage [2,3], astrophotonics [4], sensing [5–7], optofluidic [8], optical communication [9,10], quantum photonics [11] and fiber laser fabrication [12,13]. The glass photonics domain is largely dominated by the silica-based technologies and thus restricted to applications at wavelength below 2.5 µm due to the intrinsic high phonon energy of the silicon oxide glass matrix (i.e., 1100 cm^-1). Consequently, alternative oxide glass families such as tellurite, bismuthate or lead-germanate glasses, are currently employed to cover the mid-infrared (Mid-IR) domain. However, their reduced chemical, thermal and mechanical properties as well as their low glass transition temperatures (Tg) of these so-called soft glasses make them less adapted for several applications such as high-power lasers.

The barium-gallium-germanium (BGG) oxide glass is a relevant alternative for the development of Mid-IR photonic devices. Indeed, they have demonstrated remarkable chemical, thermal, mechanical and optical properties, as they transmit up to 6.0 µm, have Knoop micro-hardness up to 5.4 GPa and a Tg extending above 750 °C [14], allowing them, for instance, to be considered as an alternative material to zinc sulfide windows in high-energy laser applications [15]. Moreover, a lot of effort has been put into the development of BGG fibers to enable the fabrication of robust Mid-IR fiber lasers as well as fiber-based photonic components [16–20]. In this context, several studies have already started to investigate the direct laser writing (DLW) functionalization of BGG glasses through the fabrication of waveguides [21,22], directional couplers [23] and subwavelength periodic structures [24,25]. By tailoring glass composition along with laser parameters, laser-induced modifications inside the glass can be optimised as well. However, fundamental studies exploring the impact of each basic chemical element of the BGG composition, i.e., Gallium, Germanium and Barium, on the DLW capabilities are currently lacking.

In this work, we thoroughly investigated the role of each chemical element on the glass physicochemical properties and DLW process, in particular through two glass series: one via a Ga³⁺/Ge⁴⁺ ratio fixed to 1 and a barium content varying from 25 to 40 cationic percent, the other via a 2Ba²⁺/Ga³⁺ ratio fixed to 1 and a Germanium content varying from 20 to 80 cationic percent. Among the glass compositions studied, it appeared that the glass structure and both the bonding nature and barium ions content inside the glass network play a significant role in the BGG glass properties. In the meantime, the glass photosensitivity under direct laser writing was explored, revealing a clear dependence on the glass composition. Finally, in a gallium-rich glass which was not dehydrated, we demonstrated the fabrication of an 8.2 cm-long waveguide with low losses down to 0.3 dB.cm^-1 at 1550 nm.

2. Materials and methods

2.1 Glass preparation

Two series of glasses have been synthesized within the ternary diagram GeO₂ – Ga₂O₃ – BaO. The first series, labelled Ba²⁺ series, corresponds to the nominal composition in cat% (100-)[50GeO₂ + 50GaO_3/2] + $x$BaO with $x$ = 25, 30, 35, 40 cat%. The second series, labelled Ge⁴⁺ series, corresponds to the nominal composition in cat% (100-$y$)[33.3BaO + 66.6GaO_3/2] + $y$GeO₂ with $y$ = 20, 30, 60, 80 cat%. The GeO₂ – Ga₂O₃ – BaO ternary diagram including both series is depicted in Fig. 1.

 

Fig. 1. Ternary diagram of GeO₂ – Ga₂O₃ – BaO with investigated compositions for both Ba²⁺ and Ge⁴⁺ series, respectively red dots and blue triangles. Grey region represents the BaO-GeO₂-Ga₂O₃ vitreous domain extracted from [33].

Download Full Size | PDF

The corresponding nominal and experimental compositions, with the 2Ba²⁺/Ga³⁺ ratio and Ga³⁺/Ge⁴⁺ ratios, are reported in Table 1. No significant discrepancy is revealed, even though the germanium concentration is overestimated and the gallium is underestimated, in general.

Table 1. Investigated glasses in both Ba²⁺ and Ge⁴⁺ series: glass labels, nominal and experimental cat%, nominal mol% and nominal ratios (2Ba²⁺/Ga³⁺ and Ga³⁺/Ge⁴⁺)

View Table | View all tables in this article

Glasses were synthesized from precursors, at least 99.99% of purity; homogeneously mixed and melted at high temperatures in a platinum crucible from 1400 °C to 1550 °C, depending on the glass compositions. Indeed, even though the germanium-rich compositions melt at a lower temperature, the casting temperature is increased due to the higher viscosity than the germanium-poor compositions. After casting, all samples were annealed 30 °C below the glass transition temperature, cut and optically polished on two parallel faces.

2.2 Glass characterization

Raman spectra were recorded at room temperature from 200 to 1100 cm^-1 using a Renishaw inVia Raman microscope and a 50X microscope objective. A continuous wave laser operating at 633 nm was used for excitation. The refractive indices were measured at five different wavelengths (532 nm, 632.8 nm, 972.4 nm, 1308.2 nm and 1537.7 nm) with a prism coupler refractometer (Metricon, 2010/M). The UV–visible–near-IR transmission spectra from 200 to 1100 nm were recorded on a Cary 60 UV-Vis (Agilent) spectrometer by steps of 1 nm, while the near-IR-MIR transmission spectra were obtained from 1 to 7 µm using a Fourier-transform infrared spectrometer with an average of 50 scans and a resolution of 4 cm^-1. The density, ρ, was determined by the Archimedes method using ethanol as the immersion liquid at room temperature with a Mettler Toledo XSE204 densimeter. Differential scanning calorimetry (DSC) measurements were performed at a heating rate of 10 °C.min^-1 and a temperature accuracy of ± 3 °C with a DSC 404 F3 Pegasus calorimeter. Thanks to DSC measurements, the glass transition temperatures were extracted. Chemical analyses were conducted by electron-probed microanalysis (EPMA) on a CAMECA-SX100 apparatus. Wavelength Dispersive Spectroscopy (WDS) was acquired to measure the cationic elements, with an average value based on 8 acquisitions.

2.3 DLW setup

After being cut and polished, the glass samples were irradiated simultaneously by a frequency-doubled femtosecond laser (Clark-MXR Impulse) at 515 nm with a pulse duration of 300 fs. A λ/4 waveplate was inserted in the beam path to induce circular polarization. The samples were mounted on a computer-controlled 3D motorized stage (Aerotech PlanarDL-200 and ANT130) so that they could be translated with respect to the position of the laser beam. The pulses were focused approximately 200 µm below the surface of the samples using a 100×/0.8 NA microscope objective (Nikon ELWD), yielding a spot size of approximately 800 nm. The irradiations were performed at repetition rates varying from 200 kHz up to 5 MHz.

2.4 DLW structure characterization

The laser tracks were analyzed using a bright-field microscope mounted with a wavefront sensor (Phasics SID4Bio) to measure the phase difference through the different laser tracks. This was computed along with the cross-sectional geometry to infer the refractive index change along the laser modifications.

The propagation losses in the waveguides were measured by butt-coupling the BGG sample to a single-mode fiber delivering 1550 nm light emitted from a laser diode. Light exiting the waveguide was then collected using a multi-mode fiber connected to a power-meter (Newport 818-IS). The power was recorded and compared to a reference made by removing the glass sample and directly butt-coupling the injection and collection fibers. The losses arising from Fresnel reflections at both interfaces of the BGG waveguides were then calculated and deducted from the total insertion losses.

3. Results and discussion

3.1 Glass properties

UV-to-mid-IR optical absorption and Raman spectra have been recorded on both the Ba²⁺ and Ge⁴⁺ series. In Fig. 2(a), raw Raman spectra are presented from 200 to 1050 cm^-1 normalized at 520 cm^-1 for both series, and a gradual spectrum evolution is shown. Each Raman spectrum can be separated into three regions. The highest spectral domain (650-1000 cm^-1) can be attributed to the antisymmetric and symmetric stretching modes of Germanium and Gallium tetrahedral units [TO₄] [26,27]. The intermediate spectral range (400-650 cm^-1) can be assigned to several vibrational contributions of T–O–T bending with in-the-plane T–O–T oxygen motions [27,28]. Lastly, the lowest spectral domain (200-400 cm^-1) can be assigned to either acoustic modes involving large atom clusters [28] or out-of-plane oxygen motions in bent T–O–T bridge (T = Ga or Ge in fourfold coordination) [27].

 

Fig. 2. (a) Raw data Raman spectrum normalized at 520 cm^-1 excited at 633 nm ; (b) Linear absorption coefficient in the UV-visible-to-Mid-IR wavelength range for both Ba²⁺ and Ge⁴⁺ series.

Download Full Size | PDF

For the Ge⁴⁺ series, at 60 and 80 cat% of Germanium, the main contribution is located around 470 cm^-1 with a shoulder at 520 cm^-1, and minor contributions are observed at 800 and 890 cm^-1. The 470 cm^-1 contribution is assigned to Ge-O-Ge bridges, while the shoulder at 520 cm^-1 and both 800 and 890 cm^-1 contributions are attributed to T-O-T bridges with in- and out-of-plane oxygen motions. At 20 and 30 cat% of Germanium, the main contribution appears at 520 cm^-1 with a shoulder peaking at 465 cm^-1, while minor contributions are now located at around 750 and 820 cm^-1. Latter minor bands are attributed to GeO₄ tetrahedral units with respectively two and one non-bridging oxygens, i.e., GeØ₂O₂^2- and GeØ₃O^- [29,30].

In the BGG glass, Gallium tetrahedral units have a negative charge [GaO₄]^- compensated by one of the two positive charges of a single Ba²⁺ ion. Hence, a single Ba²⁺ ion can stabilize up to two [GaO₄]^- tetrahedra. In the Ge⁴⁺ series, the positive/negative charge ratio of 2Ba²⁺/Ga³⁺ is kept constant at 1, indicating that there are right enough barium ions involved in the [GaO₄]^- compensation mechanism to stabilize all Gallium ions in the [GaO₄]^- glass former unit [28,31]. From 80 to 20 cat% of Ge⁴⁺, both Ba²⁺ and Ga³⁺ quantities increase equally at the expense of Germanium ions. As shown in the Raman spectra, the Ge-O-Ge bridges are progressively replaced by Ga-O-Ge linkages. Hence, it is expected that the glass network skeleton evolves progressively from a germanate glass network (80 Ge⁴⁺ cat%) to a germano-gallate glass one (20 Ge⁴⁺ cat%), with a changeover of Gallium and Germanium tetrahedral units following the same pattern. However, the appearance of GeØ₂O₂^2- and GeØ₃O^- units below 30 Ge⁴⁺ cat% would indicate that a very small portion of [GaO₄]^- tetrahedral units cannot be formed due to the lack of charge compensator Ba²⁺ ions involved within Germanium non-bridging oxygens. Consequently, one cannot evict the formation in a small quantity of some higher coordinated Gallium units, such as five- and/or six-fold coordinated sites, which are neutral and thus do not require a charge compensation ion. Further measurements in the NMR experiment may help to better understand the glass structure in gallium-rich BGG glasses.

In the Ba²⁺ series glasses, the main contribution is located at 520 cm^-1 with a shoulder at 465 cm^-1. Over the Ba²⁺ increase, the second main contribution peaking at 820 cm^-1 grows significantly up to equal the 520 cm^-1 band intensity for the glass with 45 cat% of Ba²⁺. Concomitantly, two shoulders peaking at 750 and 670 cm^-1 are rising within the 820 cm^-1 contribution increase. All previous contributions have already been assigned except the 670 cm^-1 band, attributed to [GaO₄]^- tetrahedral units possessing non-bridging oxygen [GaØ₃O]^2- [30,32].

In the Ba²⁺ series, Raman spectra highlight the formation of a germano-gallate/gallo-germanate glass network, which is consistent considering the Ga³⁺/Ge⁴⁺ ratio remaining equals to 1. While increasing the content of Ba²⁺ ions, acting as a glass network modifier, the presence of various types of non-bridging oxygens is observed. These non-bridging oxygens contribute to depolymerize the germano-gallate glass network skeleton.

Raman band assignments and their corresponding peak positions are reported in Table 2.

Table 2. Raman band assignments in BGG glasses. T refers either to Ge or Ga tetrahedron

View Table | View all tables in this article

Figure 2(b) depicts the linear absorption coefficient evolution from 200 to 7000 nm for both glass series. All glasses from the Ba²⁺ and Ge⁴⁺ series present a wide transparency window from 300 nm up to 6000 nm. While the UV cut-off at 10 cm^-1 shifts a little from 290 to 320 nm depending on the glass composition, the mid-IR cut-off at 10 cm^-1 undergoes a severe blue-shifting when the concentration of Germanium ions increases in both series. In the Ge⁴⁺ series, the Mid-IR cut-off blueshifts from 6030 nm to 5820 nm for Ge-20 and Ge-80, respectively, while in the Ba²⁺ series, it blueshifts from 6030 nm to 5840 nm for Ba-40 and Ba-25, respectively. One possible explanation of this constant MIR narrowing of the optical window in germanate rich-BGG composition would be the presence of one or several overtones of a characteristic germanate structural band observed at around 890 cm^-1 in the Raman spectra [30].

Since no special care was taken to dehydrate the glass precursors prior to and during the glass synthesis, a consequent quantity of hydroxyl groups is present in all studied glasses from 2700 to 5000 nm. Jewell and Aggarwal have already studied the broad domain of fundamental OH absorption in BGG glasses with compositions close to Ge⁴⁺ and Ba²⁺ series glasses. In their study, they deconvoluted the OH absorption bands in four components (arranged in ascending wavelengths): “free” hydroxyls linked to germanate Ge-OH or gallate Ga-OH structural units, strong T-OH—O and very strong T-OH–O Hydrogen bonds (with T = Ge or Ga) [33]. More precisely, the first three contributions overlap, while the latter related to very strong T-OH–O is observable on a standalone basis at about 4300 nm. Hence, depending on the BGG glass composition, the OH absorption bands are expected to vary.

In both series, as the concentration of Germanium decreases, the amount of Ge-OH decreases leading consequently to a redshift of the OH absorption peak. This redshift appearing in Germanium-poor BGG glasses is notably interesting since it highlights their improved transparency at the 2.7 µm Erbium emission for instance. In the Ba²⁺ series, the OH absorption range broadens as the formation of non-bridging oxygens is a source of Oxygen ions forming hydrogen bonds. To ease the reader’s observations in the mid-IR region, a zoom between 2500 and 6000 nm is reported in Fig. S2.

Glass properties for the Ba²⁺ series have also been determined. Measurements include density, glass transition temperature, refraction index dispersion and concentration of Ba²⁺ ions (Fig. 3). The raw thermograms for DSC measurements are depicted in Fig. 1(S)(b).

 

Fig. 3. Properties of Ba²⁺ series as function of BaO content x (a) Density (b) Glass transition temperature (c) Refractive index (d) Concentration of Ba²⁺ ions. Lines are guides for the eyes.

Download Full Size | PDF

After consideration of the error bars, it appears that Tg follows a non-linear growth, which is often representative of a glass structure change. As introduced with Raman spectroscopy (Fig. 2(a)), the formation of various types of non-bridging oxygens, i.e., [GaØ₃O]^2-, GeØ₂O₂^2- and GeØ₃O^-, is denoted which is in favor of a significative structural change. In the meantime, even though one cannot rule out the possibility of the appearance of higher coordinated structural sites for both Germanium and Gallium ions at high content of barium ions, it is more likely to solely observe the formation of higher germanium coordination sites [31]. Indeed, as the amount of Barium ions is more than enough to compensate each Gallium tetrahedron and considering the fact that, even though higher Gallium coordination sites can exist, they are less stable in a glassy network, Germanium’s higher coordination sites are chemically favored. Additionally, the 2Ba²⁺/Ga³⁺ ratio is more than 2 at the Tg plateau (between 30 and 35 Ba²⁺ cat% (Fig. 3(b)). Hence, rather than compensate two Gallium tetrahedra, each Barium ion must compensate only one Gallium tetrahedron, leading to a certain change of Barium ion role in the BGG glass network. This change in the Barium ion role and the possible appearance of higher coordinated structure sites would be responsible for the Tg non-linear growth.

Finally, refractive index, density and Ba²⁺ ions concentration evolve linearly, as shown in Fig. 3(a, c and d). Already mentioned for BGG containing Potassium ions [34], Barium ion has twice the molar mass of the second heaviest element, i.e., Germanium, and is almost one order of magnitude more polarizable than the second most polarizable element, i.e., gallium [35]. Consequently, it appears that both intrinsic Barium ion characteristics (e.g., molar mass and polarizability) and glass density affect predominantly those both properties, whereas the structural glass modifications reported do not seem to play a significant role in these property evolutions.

Glass properties for the Ge⁴⁺ series have been determined as well (Fig. 4). Measurements comprise density, glass transition temperature, refraction index dispersion and concentration of Ba²⁺ ions. The raw thermograms for DSC measurements are depicted in Fig. 1(S)(a).

 

Fig. 4. Properties of Ge⁴⁺ series as function of Ge⁴⁺ content y (a) Density (b) Glass transition temperature (c) Refractive index (d) Concentration of Ba²⁺ ions. Lines are guides for the eyes. Dark points in c) have been plotted and extracted from [36]. For (b), the error bars are hidden by the data points.

Download Full Size | PDF

As the Ge⁴⁺ ion content decreases, the concentration of Ba²⁺ ions, the density and the Tg decrease linearly considering the error bars. As for the Ba²⁺ series case, the concentration of Ba²⁺ in BGG glasses is largely affected by the Ba²⁺ molar mass. For both Tg and density, the glass structure rearrangement from a germanate to a germano-gallate network at high and low Ge⁴⁺ content, respectively, does not play an important role. Rather, it seems to follow mostly the Ba²⁺ concentration. As the 2Ba²⁺/Ga³⁺ ratio is constant, it can be interpreted that the replacement of one Germanium tetrahedral unit for one Gallium tetrahedral unit has only a minor effect on those properties. Conversely, the stronger bond strength of Ga-O over the Germanium one appears to have a more significant impact on both Tg and density, as also reported elsewhere [31].

Finally, the refractive index decreases while the concentration of Ge⁴⁺ decreases as well. A strong drop appears between 60 and 80 cat% of Ge⁴⁺. To emphasize or extrapolate this refractive index drop for Ge-80. We have added to the plot the refractive index for the fused germania glass, named Ge-100., extracted from [36]. As highlighted, the trend change in the refractive index drop seems to occur slightly beyond 50 cat% of Ge⁴⁺ and extends up to 100 cat%. Conjointly, it is assumed that the polarizability of Ba²⁺ ions involved in the [GaO₄]^- charge compensation mechanism is more than those linked to Oxygen. Also, the appearance of non-bridging oxygens (Fig. 2(a)) has been observed from 30 Ge⁴⁺ cat% and below. Hence, it is considered that the change in Ba²⁺ electronegativity due to its type of bonding would be at the origin of the refractive index’s decrease trend.

3.2 Laser direct writing

In the preliminary phase of our investigation, we have proceeded to an extensive study of the laser-induced modifications over a wide range of parameters for the eight glasses belonging to the Ba²⁺ and Ge⁴⁺ series: scanning speed, fluence, number of passes, wavelength, repetition rate and pulse duration were all investigated. From this thorough investigation, a regular trend was observed along each glass series. To simplify the results, we restricted ourselves to present here only the irradiation performed on both series extrema, i.e., the four glass compositions: Ba-25, Ba-40, Ge-20 and Ge-80, at a wavelength of 515 nm and a pulse duration of 300 fs. The repetition rates used in this study were 0.2, 0.4, 0.6, 0.8, 1.2, 2.3 and 5 MHz. This range of repetition rates allows for laser-induced modifications in the athermal, thermal and the so-called transitional regimes, defined hereafter.

To allow a better comparison between the various repetition rates for a fixed irradiation fluence, we fixed the pulse spacing to 2 nm by adjusting the scan speed accordingly. This indicates DLW speeds of 0.4 (at 0.2 MHz), 0.8, 1.2, 1.6, 2.4, 4.6 and 10 mm/s (at 5 MHz), respectively. Finally, the four glass composition samples were glued alongside on a flat substrate and then optically polished, allowing to proceed to laser-induced modifications at the same mechanical depth with the same DLW parameters.

The laser tracks inscribed in the BGG samples were observed and categorized to map the different waveguide writing regimes across the different glass compositions. The main differences stem from the morphology of the tracks and what constitutes the driver of the structural modifications (laser filament and/or heat accumulation) in conjunction with the usual type I, II and III classifications which are ubiquitous in direct laser writing. Type I refers to tracks displaying isotropic, smooth and homogeneous refractive index change. Type II refers to the formation of anisotropic nano-gratings. Lastly, type III is characterized by disorderly damage around the focal region constituting an arrangement of cracks and voids. Note however that no type II modifications (i.e., nano-gratings) were observed in our experiment. Moreover, Raman spectroscopy was conducted on the laser tracks and revealed no trace of crystallization.

The laser tracks were split into three categories describing the thermal effects within the focal region, as shown in Fig. 5.

 

Fig. 5. Microphotographs of the laser track transverse profiles in the three regimes (thermal (5 MHz), athermal (200 kHz) and transitional (600 kHz)) for the Ba-40 and Ge-20 glass compositions. For each microphotograph, the laser track inscribed at the highest pulse energy is reported on the left side and the lowest one on the right side.

Download Full Size | PDF

In the athermal writing regime (right column), tracks have a narrow and elongated morphology caused by spherical aberration and nonlinear effects, such as self-phase modulation and filamentation, of the tightly focused beam. The resulting modification tracks have thus dimensions approximately limited to the filament size. Slight linear increase of the track length and width arises as pulse energy is increased due to the filamentation process.

The thermal diffusion time for our glasses is typically of the order of ∼1 µs [37]. This means that for heat accumulation to occur, the delay between consecutive pulses must be below this value, yielding a repetition rate of approximately 1 MHz. In the thermal regime, isotropic and significant broadening of the track takes place as thermal diffusion extends the laser-heated region beyond the focal volume, leading to rapid expansion of track width as pulse energy is increased (left column). This is typical for the high repetition rate tracks due to the lower temporal delay between consecutive pulses.

Finally, a transition regime between thermal and athermal has been identified, where aspects of both regimes can be distinguished. These tracks feature the same morphology as athermal tracks but also presents localized isotropic broadening beyond the filament caused by heat accumulation. Pulses have enough energy and sufficient overlap to drive some heat accumulation around the focal spot, but not enough to cover the entirety of the laser track (center column). This marks the onset of thermal effects in the glass upon laser irradiation.

In Fig. 6 are presented top-views of the laser tracks corresponding to two pulse energies in the thermal (5 MHz), athermal (200 kHz) and transitional (600 kHz) regimes for the four glass compositions.

 

Fig. 6. Microphotographs of the laser tracks in the three regimes (thermal (5 MHz), athermal (200 kHz) and transitional (600 kHz)) for the four glass compositions. For each microphotograph, the laser track inscribed at the highest pulse energy is reported on the top and the lowest one on the bottom.

Download Full Size | PDF

In the athermal regime, the onset of the laser-induced type I modifications is barely visible, especially at 50 nJ pulse energy, for the Ba-40 sample, whereas they are clearly visible for the Ba-25 and Ge-20 samples. As for the Ge-80, the onset of irregularities along the laser track at both low and high pulse energy seems to indicate a poor DLW capability for this glass. In the transitional regime, only the Ba-40 sample presents smooth laser tracks of Type I at both high and low pulse energy. The onset of Type III is present for both Ba-25 and Ge-20 samples at high pulse energy. In the thermal regime, only the Ba-40 and Ge-20 samples exhibit broad laser tracks of Type I modifications, whereas the Ba-25 and Ge-80 samples are showing the formation of self-organized bubbles at the laser track center, although with less regularity in the Ge-80 sample. The overall picture of the laser tracks modifications for all four glasses is presented in Fig. 7.

 

Fig. 7. Mapping of direct laser writing regimes across the 4 compositions. The crosses indicate type III modifications whereas the full dots indicate type I modifications. The area between the athermal and thermal regions indicates a transitional regime between the two structural modification regimes.

Download Full Size | PDF

At this point it is interesting to draw a parallel between the behavior of the four glasses with respect to DLW and their glass transition temperatures.

Now, the Tg was shown to increase from 640 °C to 710 °C for Ba-25 and Ba-40 respectively, while it decreases from 725 °C to 575 °C for Ge-20 and Ge-80, respectively (Fig. 3 and Fig. 4). It is thus noteworthy to observe that the progressive evolution from no laser-induced modification to Type I and then Type III modifications seem to follow the same trend as that of the glass transition temperature. In fact, Ge-80 glass composition presents the lowest Tg and direct transition from no laser-induced modification to Type III modification, i.e., no type I modification, whereas Ba-40 and Ge-20 have the highest Tg and the widest domain for Type I, as shown in Fig. 7.

As reported for calcium aluminosilicate glasses [38], a crossover of the glass viscosity evolution appears between an aluminosilicate rich and poor in silica. In other words, even though the glass transition temperature is higher for an aluminosilicate poor in silica than for a rich one, below the crossover point of the viscosity evolution, the drop in viscosity is more important for the aluminosilicate poor in silica due to its glass viscosity fragility. The term of glass viscosity fragility refers to the degree to which the temperature dependence of the viscosity deviates from Arrhenius behavior. Hence, glasses that strongly deviate are called fragile and those that do not are called strong. A similar behavior is expected here for the barium gallo-germanate/germano-gallate glasses. As a matter of fact, it was observed for the eight investigated glasses that the richer a glass composition is in germanium oxide, the higher is its casting temperature due to the significant difference in viscosity. Thus, it is suspected that Ba-40 and Ge-20 are more fragile than their counterpart, Ba-25 and Ge-80, respectively, which are stronger in terms of viscosity [39]. In Fig. 7, Ge-80 and Ba-25 samples are the glass compositions presenting the smallest Type I domain and the possible lowest fragility, while Ge-20 and Ba-40 are the ones with the largest Type I domain and the possible highest fragility. Additionally, Bellouard and Hongler have proposed a phenomenological model where the rate of spontaneous bubble nucleation during laser irradiation (thermal regime) in glass increases proportionally to the increase of surface tension [40], e.g., directly related to the liquid viscosity. As exposed in both investigated glass series for the same temperature, the melt viscosity is higher when the germanium oxide content is more important, which could provide further insight into the apparition of self-organized bubbles in both Ba-25 and Ge-80 glass compositions at 5 MHz, and the difference of spontaneous bubble rates between them.

Recalling the laser track transverse profiles of the two glasses presenting the largest domain of Type I modification, i.e., Ba-40 and Ge-20 (cf. Figure 5), it is interesting to point out that these two glasses present more or less the same thermal behavior as a function of the repetition rate except for the thermal broadening which appears to begin sooner in the Ba-40 sample. In order to highlight the observed trends for track broadening for these two glasses, their laser track widths are plotted against the pulse energies for the three regimes in Fig. 8.

 

Fig. 8. Laser track width vs pulse energy in the three regimes (thermal (5 MHz), athermal (200 kHz) and transitional (600 kHz)) for the Ba-40 and Ge-20 glass compositions.

Download Full Size | PDF

At first glance, for Ge-20 and Ba-40 glasses, the laser track width increases with both the repetition rate and the pulse energy. Additionally, beyond the threshold for heat accumulation, the higher the repetition rate, the faster the increase in the laser track width and the lower the energy threshold for heat accumulation. Comparing both glass compositions, as soon as the heat accumulation occurs, the spatial broadening of the laser track is always more important in Ba-40 than in Ge-20. At 600 kHz, it can be seen that the thermal regime threshold arises at lower pulse energy in Ba-40 than in Ge-20. Hence, one can conclude that the laser-induced thermal effect is more predominant in Ba-40 than in Ge-20.

Laser-induced refractive index changes (Δn) were also investigated. The refractive index change in the DLW structures is characterized by quadri-wave lateral shearing interferometry. An incident wavefront passing through the sample is distorted due to the localized change in refractive index. The light is collected by a microscope objective and four replicas of the wavefront are then created by a bi-dimensional grating. The interferogram formed by these replicas is subsequently recorded by a CCD camera, from which the optical path difference is calculated [41]. The Δn were inferred from the optical path difference ($OPD$) for a given laser track thickness ($e$) using the following equation:

This equation yields exact results for homogeneous media. In real situations where the $\mathrm{\Delta }n$ distribution is inhomogeneous, it is the average Δn that is inferred. In this work, the average $\mathrm{\Delta }n$ values were calculated from the center of the tracks, which provides a lower bound on the maximum refractive index change of the laser tracks. Indeed, as the measurement corresponds to an average over the track and knowing that a lower Δn can be found at the very center of the laser track, the value corresponds to a lower bound on the absolute Δn [21]. These lower bound values are used as Δn for the remainder of the paper. In Fig. 9 are plotted the Δn at each repetition rate for both Ba-40 and Ge-20 glass compositions.

 

Fig. 9. Tracks presenting the highest refractive index change for every repetition rate in both Ba-40 and Ge-20 glass compositions.

Download Full Size | PDF

It is observed that below 300 kHz (athermal regime), the refractive index change is near 1.5 × 10⁻³ for both glasses. As the thermal regime is reached though, one clearly observes a significant increase of the Δn in the Ge-20 glass whereas it remains more or less the same for Ba-40. In order to interpret this result, let us recall that even though both glasses present a broad type I DLW window, the evolution of their laser track widths and refractive index changes highlights a strong composition-dependent factor. Many glass parameters, most of them intrinsically interdependent, can play a significant role in the difference of DLW modifications from the atomic to the macroscopic scale. Hence, one can cite the proportion of glass network (i.e., Ge and Si), intermediate (i.e., Ga and Al) and modifier ions (i.e., Ba and Ca) present in the glass composition [42], affecting the glass structure and the arrangement of these specific ions (i.e., mobile ions in charge compensation site or linked to non-bridging oxygen, the coordination site of intermediate ions, etc.) at the atomic scale [22,43], but also the ability of a glass to evacuate the accumulated heat during the laser irradiation, expresses as the thermal diffusion coefficient at the macroscopic scale.

In both Ge-20 and Ba-40 glass compositions, the strong proportion difference of modifier and intermediate ions, i.e., Ga³⁺ and Ba²⁺, has a direct incidence on the final properties of the DLW modification. By isolating the evolution of the Germanium (Ge series) and Barium (Ba series) contents, it appears that the BGG glass composition with the most promising future in terms of high refractive index change and wide DLW window would be the Ge-20 sample. This BGG composition has a similar content of Barium and Gallium and a low content of Germanium, which is in perfect agreement with comparable BGG-like glasses and aluminosilicate counterparts [22,42]. In the thermal regime, it was observed that higher pulse energy contributes to creating broader tracks rather than significantly changing the Δn.

To illustrate the potential of the Ge-20 glass composition, waveguides were inscribed using the same irradiation parameters as for the previous analysis, but in an 82 mm-long sample to assess its optical performances (Fig. 10). The refractive index change profile is shown, displaying a smooth profile. At a wavelength of 1550 nm, the lowest propagation losses of 0.3 dB.cm^-1 were recorded in waveguides written at a pulse energy of 150 nJ and a repetition rate of 5 MHz. This value represents an upper bound on losses since it includes losses due to fiber coupling such as mode mismatch, fiber misalignment, and scattering from the imperfectly polished surface at the end-faces of the waveguide. Strong heat accumulation during the writing of these laser tracks creates smooth and homogeneous structural modifications. This contributes positively towards achieving low propagation losses by minimizing the scattering of light propagating in the waveguide caused by irregularities along the track. The performance of the waveguides in Ge-20 can likely be improved further by optimizing the speed at which the tracks were written to increase the $\mathrm{\Delta }n$ achieved. Additionally, many beam-engineering techniques could be employed to produce waveguides with more circular cross-sections, such as using a spatial light modulator [44], an objective with a correction collar [45], a dual-beam technique [46], a simultaneous spatial and temporal focusing technique [47], or using an astigmatic shaped beam [48] formed by a pair of cylindrical lenses [49] or a slit [50] placed just before the beam focusing objective.

 

Fig. 10. (a) 82 mm-long Ge-20 sample in which the waveguides were inscribed, with a ruler for scale (b) top view of the waveguide (c) cross-section of the waveguide and (d) Normalized refractive index change of the waveguide measured by quantitative phase microscopy.

Download Full Size | PDF

4. Conclusions

A thorough investigation focused on the dependence on the BGG glass composition and DLW photosensitivity was carried out in two series by varying solely either the barium or germanium ions. Through several glass property analyses, a preponderant dependence on both barium concentration and structural stabilization sites was reported. Under direct laser writing, BGG glasses rich in barium and gallium ions presented the wider DLW capability window of Type I modification, but only glasses with a high concentration of gallium highlighted the higher refractive index change. In the gallium-rich BGG glass, an 82 mm-long waveguide was inscribed with low propagation losses of < 0.3 dB.cm^-1 at 1550 nm.