1. Introduction

Since 2010 there has been a surge of new low-phonon-energy commercial glasses such as tellurites, tellurides and chalcogenides from the glass manufacturers aiming for the burgeoning mid-infrared device market [1–3]. Mid-infrared wavelength based devices have application in several vital fields like medical, environmental sensing/monitoring, military surveillance and high sensitivity gas detection. Tellurite glasses have received special attention due to their transparency from 350 nm to 6 µm and their better mechanical robustness and thermal stability compared to other low phonon energy glasses. The wavelength range from 3 to 6 µm is one of the most important spectroscopic windows due to the high absorbance level of several trace gases compared to overtone absorption in the NIR window. Despite the tremendous success of tellurite optical fibers [4–7] there have been relatively few reports of channel waveguides in tellurite glass, which are needed for the fabrication of compact 3D devices and sensors [5, 7]. Though waveguide fabrication in tellurite glass showed early promise [8, 9], there were only two successful reports of net gain from an Er doped Tellurite waveguide amplifier, one formed by reactive ion exchange [10] and the other by femtosecond laser inscription [9]. The tellurite glass in the latter report was achieved by intentionally modifying the Te glass with phosphate to lower the refractive index (1.64 at 1535 nm), thereby overcoming filamentation and other nonlinear issues in laser inscription [9, 11].

Very recently we have reported ion migrations in phosphate glass, identifying their roles in determining the total refractive index change during femtosecond laser waveguide writing [12]. Here, we apply these methods to study the ions responsible for refractive index change in a tellurite glass, superior in its versatile functionality due to its broad band operation capability. Though our tellurite glass was modified with phosphate units, the basic matrix is predominantly tellurite and its unique spectroscopic features were found to be preserved after femtosecond laser modification, with waveguides demonstrating gain across the C + L communication bands [9]. We have studied the correlation of ion migrations and structural changes, directionality of migrating ions and types of migrating ions in the tellurite glass waveguides, giving further insight into the underlying physics during the femtosecond laser-glass interaction. Our rigorous study of ion migration in tellurite waveguides will allow one to more intelligently design waveguides in this glass, which is notoriously difficult to process [8]. This report will help the glass manufacturers and new mid-infrared glass industries to produce glasses that are compatible for waveguide fabrication by means of the versatile 3D femtosecond laser inscription technique, the only method capable of producing waveguides in such heavy metal oxide glasses.

2. Experimental

The preparation of phosphate modified tellurite glass (50TeO₂:20P₂O₅:20Na₂O:5ZnO:5ZnF₂ with 1wt% Er₂O₃, 1wt% CeO₂ and 2wt% Yb₂O₃) and the waveguide writing setup are described elsewhere [9]. Waveguides were inscribed using a Yb:KYW femtosecond laser with 1-MHz repetition rate (femtoREGEN, High Q), 1040-nm wavelength, 400-fs pulse duration using a 1.4 numerical aperture (NA) oil immersion microscope objective (Zeiss Plan-APOCHROMAT). The collimated beam diameter D before the lens was 5 mm, slightly overfilling the 4.5-mm clear aperture (CA), allowing us to exploit the full NA of the objective. The transmission of the lens was measured to be 30%, which is expected given the 40% transmission (underfilled) quoted by Zeiss and the ratio of the area of the aperture to that of the overfilling beam (CA/D)² = 0.81. The theoretical spot size 2w₀ was 0.66 μm [13] for our laser having a beam quality of M² = 1.4. The two waveguides used for this study were written with 1 MHz repetition rate and 130 nJ laser energy before the objective (the on target pulse energy, fluence and peak intensity were 39 nJ, 2.9 J/cm² and 0.7 W/cm², respectively). Waveguides were formed with polarization perpendicular to the writing direction, although it has been previously shown that polarization has a minor influence on the properties of waveguides written in glass within the heat accumulation regime [14]. The only writing parameter varied was the writing speed, with waveguides A and B written with 2 mm/s and 4 mm/s, respectively. Electron probe microanalysis (EPMA) by energy dispersive X-ray microscopy was carried out on a Leica S440 SEM equipped with a Bruker AXS Quantax micro-analysis system. Micro-Raman spectra were performed with an excitation wavelength of 442 nm using a Renishaw inVia Raman microscope with a 50 × (0.75NA) focusing optic, leading to a spatial resolution < 1 µm and a spectral resolution of <2 cm⁻¹.

3. Results and discussion

Differential interference contrast (DIC) microscope images in Figs. 1(a) and 1(b) show the cross-section of the optimized waveguides in the tellurite glass. The waveguide cross sections are nearly circular and are significantly larger than the laser focal volume, which is expected for waveguides written in the heat accumulation regime [14]. From the best writing condition window [9] we have chosen to study two waveguides having significantly different morphologies. From the DIC images, the bright contrast zone is the guiding zone which is a densified, positive index change region and the dark zones correspond to the less dense (negative index change) regions. Waveguide A, written with a lower writing speed, has a single positive and negative index change zones whereas waveguide B has a single positive index change region sandwiched between two negative index contrast zones. The refractive index change obtained for B was reported to be 2.5 × 10⁻³ [9] whereas for A it is 3.5 × 10⁻³, which is expected due to the increased energy deposited. Figures 1(a1) and 1(b1) show the propagated 980-nm wavelength modes, further underscoring the difference in waveguiding properties of A and B.

 

Fig. 1 DIC microscope image of (a) waveguide A, (b) waveguide B and (a1) & (b1) respective 980 nm propagated modes. All figures have the same scale shown in (a) and (b).

Download Full Size | PDF

Electron probe microanalysis is an ideal tool to survey the local composition at the epicenter of waveguide formation [12, 15]. Figures 2(a) and 2(b) show that Tellurium migrations are responsible for densification, forming distinct waveguiding regions in A and B waveguides. A clear evidence of sodium ions moving out to form a rarefied region is also associated with the waveguide formation mechanism. Figures 2(a1) and 2(b1) show the secondary electron images where the respective lines scans are marked and correspond to the x-axis of Figs. 2(a) and 2(b). This is the first time strong ion migration have been reported in a tellurite based glass to form optimized waveguides for optical amplification [16, 17].

 

Fig. 2 Line scans showing strong ion migration in (a) waveguide (A), (b) waveguide (B) along with respective (a1) & (b1) secondary electron images

Download Full Size | PDF

At the waveguiding region the Te content increases by a ~7.4% (Fig. 2(a)) and ~6.2% (Fig. 2(b)) with respect to the non-irradiated glass zone, for waveguides A and B, respectively. The large relative atomic weight increase of incoming Te compared to the outgoing Na is main reason for glass densification in the waveguide regions. To give further proof of this, we must also show that there is no associated volume expansion due to the increase of tellurium content due to the formation of high volume network structures. A plausible explanation for the observed ion migration along the beam propagation axis was given by Luo et al. [18]. They were able to invert the tear-drop shape of the waveguide cross-section, and thus the regions of refractive index increase and decrease, by inverting the on-axis intensity profile through altering the sign of the spherical aberration function. They proposed that the driving force for ion migration along the beam propagation axis is a non-symmetric energy deposition profile, causing that certain ions tend to diffuse towards the slowly decaying tail whereas others concentrate at the peak of the intensity distribution.

In the present work we show that ion diffusion mechanisms are much more complex, and that the directionality can be changed without modifying the intensity distribution, but simply by tuning the net laser fluence. This suggests that the glass composition itself plays a major role and that thermal accumulation effects are able to invert the migration direction. The findings are consistent with our recent work in phosphate glass [12]. Na migrating towards the rarefied zones could be explained with the theory of ionic diffusion in glasses where the monovalent ions easily diffuse towards the defect rich zones via vacancy/interstitial/defect transport as explained by several models put forward [19, 20]. Hence we assume that Na migration is a secondary effect or at least triggered by the migration of Te which can be explained on the basis of a structural model: 2[TeO₄]_bipyramid = 2[TeO₃]_pyramid + [2O^2-]_{non-bridging oxygen}. At these non-bridging oxygen sites, Na⁺ preferentially occupy sites [19, 20]. In fact, Te displacement from its initial position to the densified region leaves behind a defect rich region. This result supports previous findings that mono valence ions (Na⁺) move towards the negative index change zone whereas the multivalence ones (Te^{2+,4 + ,6+}) are found in the densified zone [12]. The term ‘valency’ is solely used to classify the ions based on their common valences. The spatial distribution of the other ions making up the entire glass structure viz. Zn, P, Er, Yb, Ce, F and O are relatively unchanged compared to the non-irradiated zone.

We further characterized these waveguides with micro-Raman spectroscopy to gain insight into the glass network and structural state due to ion migrations. We have used waveguide B for Raman spectroscopy with a line scan performed longitudinally (y-axis, parallel to laser beam axis) across the center of the waveguide. The scan distance is as marked in Fig. 3(a) (0 µm to 30 µm) and a representative Raman spectrum of the bulk glass is shown in Fig. 3(b). The corresponding changes of three different tellurite vibration band peak intensities (Fig. 3(c), 789 (blue line), 709 (green line) and 462 (red line) cm⁻¹), peak positions (Fig. 3(d)) and FWHM (Fig. 3(e)) along the waveguide are as shown in the graphs. The 1030 cm⁻¹ and 1112 cm⁻¹ bands corresponding to the phosphate network units did not show significant variation in any aspects of its vibration along the scan axis and hence are not included in the graphs. The x-axis of all the graphs correspond to the scale marked in Fig. 3(a) and the discussion will be mainly focussed on the guiding region, which is the densified zone. We have also carried out Raman measurements along the direction perpendicular to the propagation axis. Since the changes are essentially similar to what is expected from the longitudinal scans, we have only included the result of the measurements at the most relevant points marked in Figs. 3(a), 3(d) and 3(e). This information provides a complete picture of the densified zone. The similarity in the longitudinal and transverse direction measurements is given by the expected revolution symmetry with respect to be beam propagation axis of the energy deposition and subsequent heat diffusion processes [13, 14]. In order to analyze the structural modifications induced, we have used two distinct vibration bands of Te-O bonds, the first corresponding to the four coordination system for TeO₄ in which the neighbouring atoms are arranged at the four vertices of a trigonal bipyramid (tbp). This vibration is found at ~709 cm⁻¹ (shoulder on the left side of the maximum intensity peak of Fig. 3(b)) [21, 22]. TeO₄ units of a pure tellurite glass are broken down, upon addition of alkali ions, to TeO₃ trigonal pyramids (tp) to accommodate the network modifying alkali ions (this is the second vibration band used and is found at 789 cm⁻¹) and is the stretching mode of the TeO₃ tp containing terminal Te-O bonds such as Te = O and Te-O- with NBO). In the present scenario of waveguide writing, the case is unique since a highly localized modification happens inside a glassy matrix such that sodium and tellurium enriches two adjacent regions. At the tellurium rich zone one would intuitively expect an increase of TeO₄ unit vibration but such a situation would show signs of volume expansion due to the large size of tbp units. However, from Fig. 3(c) one observes that at the center of the waveguide, the TeO₄ vibration intensity decreases while that of TeO₃ increases (this is seen as a localized upward jump sandwiched between two localized depression in the graphs of TeO₃ and TeO_bend). In Fig. 3(c) for TeO₃ and TeO_bend graphs there is a slow decrease of intensity from 0 to 30 µm which could be due to normalization, but the information on the intensity profiles at the positive and negative index zones are still preserved. From Fig. 3(e) the distribution of vibrational frequencies of the TeO₄ units are reduced (FWHM lowered by 14 cm⁻¹) as TeO₄ trigonal bipyramids exist intrinsically at a highly strained condition. In contrast, the FWHM of the TeO₃ peak is increased [22, 23]. This conversion of tbp to tp is reported to be reversible upon solely by a heating and cooling cycle and is stated that it requires the rupture of two network links of each -Te-O-Te- ring [21,22].

 

Fig. 3 (a) DIC image of waveguide (B) with marked scan distance and five relevant transverse scan points (cross, square, star, circle, and diamond), (b) Raman spectrum of the unirradiated bulk glass and (c) peak intensity variation in vibration frequencies of TeO₃ (789cm⁻¹ peak, blue line), TeO₄ (709cm⁻¹ peak, green line) and Te-O bending (462cm⁻¹ peak, red line) cm⁻¹ units, (d) shift in corresponding peak vibration frequency (e) variation in FWHM of corresponding peak vibration band.

Download Full Size | PDF

TBP to a TP unit conversion requires a breakage of bridging oxygen, but due to the formation of double bonds in TeO₃, a second Te-O bond in the ring network has to be broken in pair. Since this conversion can also happen without strong dissipative molecular displacements it is deduced that the constituents of the network are capable of becoming mobile [22]. We believe a very similar scenario is being recreated at a micro-scale with heat accumulation/diffusion and ion migrations. The large contrast in FWHM variation of TeO₄ and TeO₃ can also be explained with the above model. Both these bands after ion migration relax to a slightly low frequency vibration mode, ~7 cm⁻¹ and 9 cm⁻¹ for TeO₄ and TeO₃, respectively (Fig. 3(d)). The bending mode of Te-O-Te or O-Te-O linkages at 462 cm⁻¹ showed an increase of vibration intensity but the variation magnitude of FWHM was reduced to ~6 cm⁻¹, which we attribute to the formation of more double bonds (T = O) within the tp units.

In Table 1, we compare the magnitude of changes found in Raman, EPMA and refractive index [9] (experimentally measured using refracted near field (RNF) profilometer (Rinck Eletronik)) in the guiding regions of waveguides A and B. From the table we conclude that the conversion to tp units that has shorter average bond length (including a short double bond) compared to the tbp [23] increasing the packing fraction and also a large relative increase of atomic mass of Te and Na (Na – 22.99, Te – 127.6) jointly contributing to the densification. This observation is important feedback for glass making which could allow the tailoring of higher quality waveguides in tellurite glasses. As the increase in refractive index change is mainly due to the relative increase of tellurium ions, in fact, increasing the amount of Na or substituting/introducing K ions which has a larger ionic radii make more room for incoming Te. The upper limit of increasing Te ions is dependent on several factors such as increase in surrounding stress due to the strained Te-O bonds, increase in non-bridging oxygens due to lack of network modifiers (draining of Na) to occupy the lone pair sites of tellurium pyramidal structures and crystallization due to Te-O segregation. Although the addition of phosphate aided the femtosecond laser inscription by reducing the refractive index of the glass and thus avoiding linear and nonlinear propagation effects as well as increasing the Er³⁺ upper laser level (⁴I_13/2) life time to 2-3 times of that in a pure tellurite glass [9], it plays no role in the densification process.

Table 1. Refractive index change, Te ion increase, and changes to Raman bands for waveguides A and B.

View Table

7. Conclusion

Femtosecond laser written waveguides in a modified tellurite glass showed the effects of ion migration and structural changes due to densification. Strong ion migration was identified with tellurite ions found to move towards the densified zone and sodium ions towards the rarefied region. We presented two waveguides inscribed with two different fluences yielding different morphologies. The waveguide (B) with a positive index contrast zone sandwiched between two negative index changes clearly shows that tellurium ion migrates to the densified region from both rarefied zones (for sodium the opposite occurs). Hence the net fluence can be used to control the direction of ion diffusion and the final refractive index distribution. The observation that monovalent ions enrich the rarefied zone with multivalent ions concentrating at the densified zone are consistent with our previous findings in phosphate glass. Raman spectroscopy carried out on the waveguides revealed that the increase in refractive index due to the local increase of Te is enabled by the conversion of its trigonal bipyramids to trigonal pyramids, increasing the packing fraction to aid densification. Both these results should help to improve the refractive index change obtained in tellurite glass waveguides by optimizing the Te-Na ratio within the glass matrix. Additionally this should also help achieve a faster waveguide writing as the optimization of ion migration should reduce the fluence needed to achieve the desired refractive index change. Our research expands the parameter space for optimizing femtosecond laser written waveguides for photonic devices.