Zero photoelastic and water durable ZnO–SnO–P₂O₅–B₂O₃ glasses

Transmission of electric power over long distances is utilized through electric current in high voltage conductors. For monitoring the electric current, an optical fiber current sensor, which has some advantages, such as passive measurement, free from saturation effect, high speed response, small size and light weight, and easy butting with the optical communication fiber, is employed on the basis of the Faraday effect.¹ The Faraday rotation angle of linearly polarized light corresponds to the magnetic field along the path followed by the fiber. The signal is linearly correlated to the rotation angle but the sensitivity is drastically reduced by a stress induced birefringence, so called photoelastic effect, which may occur in the situation of the number of turns of fiber which is wounded around the conductor. Therefore, very small birefringent glass fiber having zero photoelastic constant (PEC) is desirable for maintaining its sensitivity to measure the current.² To date, high PbO containing SiO₂ glass fiber similar to SF57³ is actually realized for this purpose,¹ but other glasses made by lead-free composition are seeking from a point view of environmental friendly materials. The divalent lead ion is essential for the special fiber of current sensor due to high polarization ability.⁴ However, the flint fiber glasses¹ will be subjected to the European Union directive Restriction of Hazardous Substances which restricts the use of hazardous element such as lead. The minimization of PEC is one of the breakthroughs to realize above the fiber current sensor device¹ using lead free composition.

Despite its widespread interest, microscopic modeling and quantitative examples of photoelasticity in phosphate glass are still lacking.^5–9 Recently, Zwanziger’s group has successfully explained the reason of the zero stress optics (zero photoelastic constant) in phosphate and other oxide glasses by a simple empirical model that predicts the existence of the composition showing zero stress optics property.^6,7,9 The empirical model is on the basis of the parameter d/Nc, where d is cation-oxygen bond length and N_c is cation coordination number. Here, glasses with average d/N_C values greater than 0.5 Å have negative PEC, whereas glasses with d/N_C < 0.5 Å have positive one.⁶ 

Our group has reported the series of values of PEC in xZnO–(67–x)SnO–33P₂O₅ glasses.¹⁰ When x = 18.5 mol. %, the glass has minimum absolute PEC values of |0.04 × 10⁻¹²| Pa⁻¹ with an accuracy of ±0.02 × 10⁻¹² Pa⁻¹.¹⁰ The composition having zero PEC (<|0.05 × 10⁻¹²| Pa⁻¹) can be successfully explained by applying the empirical model.^6,10 Although some compositions of ZnO–SnO–P₂O₅ glasses have low/zero PEC values, improvement of water durability remained to be solved for practical use of the fiber current sensors. Accordingly, we substituted P₂O₅ by B₂O₃ in composition, which can enhance water durability that retains the optical and thermal properties in SnO–P₂O₅–B₂O₃ glasses.^11,12 For B₂O₃ containing glasses, the anion forms of borate unit¹³ need to be revealed to utilize the empirical model. Since proportion of BO₃ and BO₄ units (X_i, subscription i termed BO₃ and BO₄), which deduce cation (B³⁺)-oxygen bond length (d_i) and coordination numbers of oxygen (N_c,i)⁶ can be obtained by the information of the borate form elucidated by spectroscopic methods. Particularly, the Raman and NMR spectroscopies are powerful tools to determine the ratio of borate proportions in the zero/low PEC glasses.

In this letter, we report a new series of xZnO–(67–x)SnO–(33–y)P₂O₅–y B₂O₃ glasses providing with both properties of zero PEC and good water durability. The structures are revealed by Raman and magic angle spinning (MAS)-NMR spectroscopies. The validity of the Zwanziger’s empirical model for several compositions possessing the low PEC is also mentioned.

Glass samples were prepared by a conventional melt quenching method. A series of glass compositions was prepared from reagent grades SnO (99.5%), ZnO (99.99%), P₂O₅ (99.99%), and B₂O₃ (99.99%). Sample compositions were xZnO–(67–x)SnO–(33–y)P₂O₅–y B₂O₃, where x = 0–30, y = 0, 3, and 10 in mol.%. Batches that yielded 12–20 g were mixed in an alumina mortar and then melted for 60 min at 950 °C in a vitreous carbon crucible in a silica tube furnace under flowing argon gas in order to avoid oxidation of Sn²⁺ to Sn⁴⁺. The melting process can minimize the oxidization being about 3%.¹⁴ The melts were quenched on a carbon plate in air and annealed to remove the residual stress between 280 and 340 °C in air, depending on their glass compositions. The obtained glasses were cut and polished to disk shapes with ∼10 mm or ∼20 mm in diameter and ∼10 mm in length for PEC measurement. The disk samples with larger diameter were needed for examining the glass having low PEC (<0.5 × 10⁻¹² Pa⁻¹) in order to apply relatively high uniaxial loads up to ∼20 kg.¹⁰ A frequency stabilized transverse Zeeman He-Ne laser¹⁵ operating at 632.8 nm was used as a light source. The reproducibility of PEC measurements for each glass was within ±0.002 × 10⁻¹² Pa⁻¹ due to sample quality variations in our laboratory. Optical transmittance spectra and refractive index were measured using a 1 mm-thick plate and a 3.5 × 3.5 × 10 mm rectangular bar, respectively. Standard Material Characterization Center-1 static leaching experiment¹² with glass sample plates (10 mm × 10 mm × 1 mm) was employed for water durability test. The moderate water durability was characterized by the weight loss per specific area less than ∼10⁻⁸ kg/mm² and pH change after immersion tests at 40 °C for 24 h in distilled water. Raman spectra for polished glasses were obtained using a micro-Raman spectrometer (inVia, RENISHAW) with a 532 nm YAG laser as the excitation source. ³¹P and ¹¹B MAS-NMR were performed at on a 18.8 T spectrometer (Avance III, Bruker BioSpin) using a prototype B-free stator 3.2-mm boron phosphorus probe operating at a spinning frequency of 20 kHz.

Figure 1 shows the variations of average PEC values with ZnO concentration in the glasses. At least three samples were measured for averaging on each composition. The PEC increases with increasing ZnO concentration in the range of −3.1– + 2.2 × 10⁻¹² Pa⁻¹. In a composition of x = 18.5 and y = 0, the minimum PEC value was +0.04 × 10⁻¹² Pa⁻¹.¹⁰ Whereas, in x = 19, y = 3 and x = 20.5, y = 10, the minimum values were −0.002 × 10⁻¹² Pa⁻¹ and +0.006 × 10⁻¹² Pa⁻¹, respectively.

Figure 2 shows results of water durability of the glasses after immersion test. It is obvious that there is composition dependence of water durability in relation with ZnO and B₂O₃ concentrations. Less changes of pH were observed for y = 3 and y = 10 glasses compared to for y = 0 glasses, which mean the less weight loss of phosphate component into the distilled water. In particular, concentration dependence of B₂O₃ (y) is more prominent at nearly same ZnO compositions (zero photoelastic compositions of x = 17.5–22). At the weight loss per unit area in x = 19, y = 3 and x = 20.5, y = 10, samples are at least in 2 and 3 orders lower than that in series of y = 0, respectively. These durable properties by B₂O₃ substitution have been reported in SnO–P₂O₅–B₂O₃ glasses,^11,12 which were explained by making a protection coating film against water molecular attack.¹² And the normalized weight loss on the SF57 glass is more than 1.5 × 10⁻¹⁰ kg/mm² per hour in 50 °C water,¹⁶ which is about 4 times higher than the zero photoelastic glasses.

Figure 3 shows optical transmittance of the zero photoelastic glasses. The absorption edges on those glasses depend on amount of SnO and B₂O₃. The optical bandgap is primarily governed by 5s²-5p transition in Sn²⁺ ions.¹⁷ Typical wavelengths of absorption edge are 320-340 nm on these glasses, while absorption edge of SF57 glass is around 340 nm.³ In near infrared region up to 2 μm, present zero photoelastic glasses are transparent. Refractive index of current zero photoelastic glasses for d-line (n_d) is 1.77-1.80, which is less than that of SF57 (n_d = 1.85).

The structures in the low/zero PEC compositions were investigated by micro-Raman technique as shown in Fig. 4. Three representative compositions of Raman spectra having very low PEC are shown in Fig. 3. The marked points are that peak of constituents of phosphate are shifted from Q¹ units in 1050 cm⁻¹ to Q⁰ units in 980 cm⁻¹ accompanying decrease of peak at 745 cm⁻¹ with increasing of B₂O₃ concentration (here, Qⁿ represents the number of bridging oxygen (n) per PO₄ tetrahedron²³). The increase of Q⁰ units might cause water durable glass since non-polar environment around Q⁰ units is hard to attract a polar water molecule. Regarding borate structures, little information is derived from the Raman spectra. Anionic ring in BO₄ units in 580 cm⁻¹ and B–O–Sn linkages in 200 cm⁻¹ is observed. However, the BO₃ units in 1200-1400 cm⁻¹ (Ref. 24) cannot be observed in the x = 19, y = 3 and x = 20.5, y = 10 compositions.

The ³¹P MAS-NMR spectra (Fig. 5(a)) confirm the amorphous character of the materials through the presence of broad peaks, signature of distributed and disordered structure. Presence of boron leads to broader signal (Fig. 5(b) (ii) and (iii)) due to the presence of several mixed units with close chemical shifts.^14,25 Deshielded chemical shifts suggest the presence of Q¹ and Q⁰ sites and support the conclusions derived from the Raman experiments. The ¹¹B MAS-NMR spectra show the presence of BO₃ (∼15 ppm) and BO₄ (0/−5 ppm) units in the x = 20.5 and y = 10 samples and reveal the presence in very low amount of BO₃ in the x = 19 and y = 3 glasses. Signal integration allows determining the BO₃/BO₄ ratio in both samples. It turns out that the x = 19, y = 3 and x = 20.5, y = 10 samples contain 2.0% ± 1.5% and 52.5% ± 0.5% of BO₃, respectively. These proportions were used as input parameters in the Zwanziger’s empirical equation.⁶ 

Figure 6 shows plots of the relationship between observed PEC and d/N_c. Here, the data of d and N_c are adapted from Refs. 6 and 9. We here assume that BO₃/BO₄ ratios elucidated are 2%/98% for y = 3 and 52.5%/47.5% for y = 10, respectively. The three lines are crossing at around ∼0.49 for in the zero PEC compositions. The estimation value of d/N_c is in good agreement in the criterion of the change of PEC sign.⁶ 

The Sn⁴⁺ ratio for binary 67SnO-33P₂O₅ glasses has estimated with ³¹P NMR technique (¹¹⁹Sn not showing at present).¹⁴ The spectrum could be decomposed by 94% of Q¹ and 6% of Q⁰ components. It is supposed that the formation of Q⁰, which have non-bridging oxygens, was derived from the oxidation of Sn²⁺ during melting. Then, totally 3% of Sn²⁺ was oxidized in the binary composition. Previous study on ZnO-SnO-P₂O₅¹⁹ and SnO-P₂O₅-B₂O₃¹⁴ glasses revealed that content ZnO and B₂O₃, which are substituted with SnO and P₂O₅ in the 67SnO-33P₂O₅ composition, resulted in creation of Q⁰. So, it is difficult to distinguish between the formations of Q⁰ by Sn⁴⁺ or ZnO/B₂O₃ substitution solely judged from the ³¹P NMR spectra for the nominal molar composition of the zero photoelastic glasses. Further Mössbauer and/or titration analysis should be employed. We estimate that based on the empirical model,^6,9 a 3% of Sn⁴⁺ content affects cross points of the zero PEC which are shifted to ∼0.48 from ∼0.49.

In conclusion, we have demonstrated the compositions of the lead free ZnO–SnO–P₂O₅–B₂O₃ glasses which have very low PEC values and water durable property compared to other SnO–P₂O₅ glass systems. From the structural inspection by Raman and ³¹P and ¹¹B MAS-NMR spectroscopies, the very low photoelastic constant glasses have existence of dominant Q⁰ units in phosphate network and BO₃ and BO₄ units in borate network. At least in terms of our samples, the compositions having zero photoelastic constant are explained by the empirical model based on bond lengths and coordination numbers around constituent cations.

A.S. and H.T. thanks graduate students S. Anan and Y. Oba for experimental assistance. A.S. also appreciates Professor H. Toyota for use of micro-Raman scattering apparatus. This work was supported by JSPS KAKENHI Grant No. 25820333. Financial support from the TGIR-RMN-THC Fr3050 CNRS for conducting the research is gratefully acknowledged.