Effects of silver ion exchange and subsequent treatments on the UV–VIS spectra of silicate glasses. I. Undoped, CeO₂-doped, and (CeO₂ + Sb₂O₃)-codoped photo-thermo-refractive matrix glasses

Highlights

• UV absorption spectra of silver ion-exchanged photo-thermo-refractive glasses

• Changes in the spectra under the effects of the UV irradiation and heat treatment

• Shifts of the redox equilibria between dopants due to temperature and ion-exchange

Abstract

The effects of the silver ion exchange and of subsequent UV irradiation and/or heat treatment on the optical density spectra of photo-thermo-refractive matrix glass samples both virgin and doped with cerium and/or antimony oxides were investigated. The incorporation of silver by ion exchange for ~ 15 min at 310 °C was found to cause, for all glasses except those with CeO₂ content around 0.1 mol% or more, a shift of the UV limit of strong absorption to the greater wavelengths by ~ 70–100 nm due to the occurrence of Ag⁺-related absorption. Complicated changes caused by the UV irradiation and/or heat treatment of ion-exchanged samples in the location of their UV limit of strong absorption and in the absorption throughout the 330–550 nm region were analyzed and shown to be greatly influenced by (i) the kind of dopant(s), (ii) chances for shifting the redox and other equilibria involving the dopants at temperatures of particular treatment processes, and (iii) NaBr presence in the glass batch composition. The effect of NaBr addition was explained tentatively in terms of formation, after the heat treatment, of AgBr clusters or nanocrystals.

Introduction

Photo-thermo-refractive (PTR) glasses are used widely now as effective photosensitive media for recording the volume phase holograms [1], [2], [3] and also are known [4] to be quite promising materials for the integrated optics. In particular, they can combine simultaneously the laser and waveguide properties, which opens up a way to developing, on their basis, the polyfunctional devices for the integrated optics [5]. For implementing such prospects, a detailed knowledge of their optical properties is a must.

There are many sources in literature (see, for example, a brief review in [6] and references therein) that discuss the assumed mechanisms of photo-induced processes involved into the phase hologram recording on PTR glasses. It is expedient to remind here that, according to these sources, all electronic processes occurring in PTR glasses at room temperatures start with the photoelectron generation that originates from the Ce^3 + ion present in the glass compositions as a dopant. The UV irradiation of a sample at wavelengths corresponding to the region of Ce^3 + ion absorption (λ_max ≈ 312 nm) is known (see, for example, [7], [8], [9]) to cause the photo-ionization of the ion, thus freeing the photoelectrons as follows:

$${\mathrm{Ce}}^{3+}+\mathrm{h}\nu {\left[{\mathrm{Ce}}^{3+}\right]}^{+}+{\mathrm{e}}^{-}.$$

These photo-electrons are then trapped by various trapping centers, the principal trapping channel being assumed to be the one involving the Sb(V) valence state:

$$\mathrm{Sb}\left(\mathrm{V}\right)+{\mathrm{e}}^{-}\to {\left[\mathrm{Sb}\left(\mathrm{V}\right)\right]}^{-}.$$

Certain fractions of photo-electrons can be trapped also by Ce(IV) valence states (the latter being thus transformed into the [Ce(IV)]⁻ electronic centers), impurity Fe^3 + ions, and Ag⁺ ions, the latter case resulting finally in the neutral silver atoms:

$${\mathrm{Ag}}^{+}+{\mathrm{e}}^{-}\to {\left[{\mathrm{Ag}}^{+}\right]}^{-}\to {\mathrm{Ag}}^0.$$

When subjecting the UV-irradiated PTR glass sample to heat treatment at temperatures below the glass transition one (T_g), the [Sb(V)]⁻ centers lose the trapped photo-electrons, thus becoming the electron donors. Electrons thus freed are then trapped by Ag⁺ ions, thus converting the latter, in accord with Eq. (3), into the Ag⁰ state. Subsequent heating a sample to temperatures close to or, the more so, above T_g is assumed to start two physicochemical processes. First, the atomary Ag⁰ states agglomerate into the silver clusters or molecular complexes such as Ag₂⁰, Ag₂⁺, Ag₂^2 +, Ag₃⁰, Ag₃⁺, … Ag_n^m+, the further aggregation of the complexes leading to the formation of colloidal metal silver particles. Second, these colloidal silver particles serve as the nucleation centers around which the nano-sized fluoride crystals (mostly NaF and also AgBr and NaBr) grow, thus providing a local decrease in the refractive index compared to the PTR glass matrix. Silver nanoparticles formed due to the above processes manifest themselves in additional absorption maxima in the 410 to 500 nm spectral region, which is due to the surface plasmon resonance in the Ag metal nanoparticles (see, for example, [10], [11] and references therein).

The standard synthesis of any Ag-doped silicate glasses by melting the batch materials results, when exceeding the batch molar percentage of Ag₂O around ~ 0.2 mol%, in a spontaneous reduction of Ag⁺ ions to metallic silver yet in the course of batch melting (see, for example, [12]). This hinders prospects to obtain glass-based materials with high concentrations of metal nanoparticles and governing the nanoparticle distribution, which, in turn, impedes the development some up-to-date types of integrated optics devices.

A reasonable way to solving the above problems seems to be the use of ion exchange (IE) technique (see, for example, [13], [14]). In particular, the Na⁺ → Ag⁺ ion exchange denoted further by silver IE can increase substantially the concentration of Ag⁺ ions in the surface layer of a sample, which, in prospect, can allow for combining the waveguiding, plasmonic, and photo-sensitive properties in a single device. The latter, in turn, might open up further potentialities of using PTR glasses for developing biosensors, plasmonic waveguides, and other integrated optics devices that function based on the effect of the plasmon resonance.

The properties of silicate glass samples subjected to the silver IE were studied extensively with the luminescence spectroscopy and, in substantially less detail, with the UV–VIS absorption spectroscopy (see, for example, [15], [16], [19], [20]). Data obtained in these studies revealed a strong dependence of the optical density in the wavelength region above ~ 350 nm on IE temperature and on subsequent heat treatment. Garcia et al. [17] found, for soda-lime silicate glass doped with 1 mol% CeO₂, that annealing for 30 min even at temperature as low as 350 °C is capable of increasing the absorption of ion-exchanged samples at wavelengths above ~ 350 nm. Further studies [18], [19], [20], in general, agreed with these trends and also indicated a substantial influence of other dopants on the formation of silver aggregates after IE (see below).

As to the wavelength region below ~ 350 nm, data available are much less detailed than those for the region above ~ 350 nm.

Ahmed and Allah [15] reported the occurrence of three bands in the UV–VIS absorption spectra of ion-exchanged soda-lime silicate glasses; the band locations were estimated with Gaussian analysis to be ~ 305, ~ 350, and ~ 420 nm. The authors assigned a band at ~ 305 nm to be related to the Ag⁺ ions, a band at ~ 350 nm to be due to the absorption by Ag⁰ single atoms, and a band at ~ 420 nm to be due to the absorption by Ag_n⁰ silver microcrystals.

Borsella et al. [16] investigated the optical density and luminescence spectra of ion-exchanged commercial soda-lime silicate glass in the 250 to 800 nm region and reported, for a sample ion-exchanged at 320 °C for 30 min in a molten salt bath containing 1 mol% Ag₂O, a small red shift of the near-UV edge due to emerging an additional absorption maximum. The location of the maximum was estimated, with the difference spectrum technique, to be around 268 nm. No assignment of the maximum was proposed.

Paje et al. [19], [20] investigated the effect of silver IE on the luminescence and optical density spectra in the 220 to 720 nm region for 16Na₂O·10CaO·74SiO₂ glasses containing one of the dopants such as Sb₂O₃ and CeO₂. Various IE temperatures (325 to 450 °C) and IE process durations (1 to 45 min) were tested.

In [22], the optical density spectrum of untreated soda-lime silicate glass sample 1 mm thick doped with 1 mol% Sb₂O₃ was found to contain a strong broad maximum with a flat top in the 260–280 nm region. For a sample ion-exchanged at 325 °C for 10 min, the shift of absorption edge formed by the wing of this maximum to the greater wavelengths and also the occurrence of a weak broad shoulder around ~ 400 nm ascribed tentatively to silver aggregates were reported. The authors tried to deconvolute crudely this spectrum into components and found (i) a component corresponding to the above shoulder to be located at ~ 402 nm and (ii) another quite weak component to occur at ~ 297 nm, its location corresponding to the excitation spectrum of antimony fluorescence. Further, based on a decrease in the intensities of luminescence from Sb(III) valence state and Ag⁺ ions after IE, the authors assumed a decrease in the Sb(III) concentration due to the occurrence of a redox reaction,

$$\mathrm{Sb}\left(\mathrm{III}\right)+{2\mathrm{Ag}}^{+}\to \mathrm{Sb}\left(\mathrm{V}\right)+{2\mathrm{Ag}}^0.$$

In [23], the optical density spectra were recorded before and after the silver IE for soda-lime silicate glass samples 1 mm thick doped with 1 mol% CeO₂. The result of CeO₂ doping was found to consist in a great increase in the optical densities of both untreated and ion-exchanged samples in the wavelength region above ~ 370 nm up to ~ 2.8–4.0 or even more. The authors deconvoluted crudely these spectra, the component locations and halfwidths before and after the IE being estimated to be practically similar. Three greater-wavelength components found to be centered at ~ 355, ~ 338, and 303 nm were assigned to transitions in the Ce^3 + ion. The only effect of silver IE at 325 °C for 10 min was found to be a certain (no more than ~ 0.3–0.4) increase in absorption throughout the entire wavelength region studied, no distinct spectral features that might be ascribed to the silver nanoparticles being observed. Further, the authors noted a decrease, compared to an untreated sample, in the intensity of Ce^3 + luminescence. This was assumed, similar to the case of [19], to result from a decrease in the Ce^3 + concentration due to the occurrence of a redox reaction, namely,

$${\mathrm{Ce}}^{3+}+{\mathrm{Ag}}^{+}\to \mathrm{Ce}\left(\mathrm{IV}\right)+{\mathrm{Ag}}^0.$$

With an increase in the IE temperature up to 425 °C and duration up to 1 h, the sample was found to become opaque both in the UV and most part of the visible, which was interpreted in terms of precipitating the silver nanoparticles exhibiting broad distributions over the sizes and shapes.

Our preliminary study [21] into the optical density spectra of ion-exchanged PTR-based glass samples has shown that, for the exchange process duration of 15 min at temperature of 310 °C, the silver IE caused a quite substantial long-wavelength shift of the UV limit of strong absorption of the samples. This shift was shown to obscure completely the region of Ce^3 + absorption, so that the ion-exchanged PTR glass samples have lost a sensitivity to the UV irradiation.

All sources available on the UV spectra of silicate glasses doped with silver through glass batch report the Ag⁺-related absorption to be located at wavelengths shorter than 250 nm, which is in contrast to the location of a feature reported by Borsella et al. [16], ~ 268 nm, and Ahmed and Allah [15], ~ 305 nm. In particular, Spierings [12] and Bach et al. [22] stated this absorption to comprise two features around 208–227 (shoulder) and 192–202 nm, respectively. In [23], the Ag⁺-related envelope centered around ~ 230 nm was found to occur at the spectra of PTR glass matrix sample doped with silver oxide. Recently, Efimov et al. [24], [25] studied in detail the absorptivity spectra of PTR glass matrix samples doped, through synthesis, with small amounts of Ag₂O and CeO₂ and implemented the deconvolution of the spectra with the dispersion analysis based on Convolution model for the dielectric function of glasses (see, for example, [26], [27], [28]). The Ag⁺-related absorption maximum around ~ 43,500 cm^− 1 (~ 230 nm) was found to be in fact an envelope of three individual bands and the occurrence of the fourth band outside the high-frequency limit of spectrum recording was confirmed. So, the total number of individual Ag⁺-related bands inherent in the spectra of silver-doped glasses was shown to be four, these bands being due, judging by analogy with Ag⁺-doped crystals, to the 4d¹⁰ → 4d⁹5s¹ transitions. For the bromine-free glass matrix, the inherent frequencies of these four bands were found to be ~ 40,740, ~ 43,850, ~ 47,920, and ~ 51,780 cm^− 1 (~ 245, ~ 228, ~ 209, and ~ 193 nm, respectively). For samples co-doped with CeO₂ and Ag₂O, the intensities of Ce(IV)-related bands were found to be less by factor of ~ 20 compared to those for samples doped with CeO₂ alone. Based on this fact, Efimov et al. [24], [25] concluded that the Ag⁰ valence state plays the role of a reducer with respect to the Ce(IV) valence state, which is opposite to an assumption by Garcia et al. [17] and Paje et al. [20]. Thus, according to Efimov et al. [24], [25], cerium in PTR matrix glasses co-doped with silver occurs predominately in the Ce^3 + valence state and, correspondingly, silver in the untreated samples should occur mostly in the Ag⁺ valence state.

In glass matrices, the isolated Ag⁰ atom was assumed [15], [29], [30], [31] to manifest itself in a band or envelope located at 350 to 365 nm. At the same time, other investigators took the Ag⁰ valence state in glasses to occur mostly in the form of Ag_n^m+ complex with n = 2 and m = 1, i.e., the charged diatomic Ag₂⁺ molecule, thus considering the amount of isolated Ag⁰ atoms to be negligible.

Because some of the above sources (such as [17], [19], [20]) touched upon the effect of silver IE on the spectra of samples doped with CeO₂ or Sb₂O₃, it is expedient to cite here sources available on CeO₂- and Sb₂O₃-doped glasses not subjected to IE. In particular, there is a lot of sources dated from sixties to now that are devoted to the absorption of Ce^3 +–Ce(IV) and Sb(III)–Sb(IV) valence states¹ in silicate and other glasses (see, for example, reviews in [32], [33]). The more recent of the sources reported the well-known Ce^3 +- and Ce(IV)-related absorption maxima in the 280–320 and 210–260 nm regions to be envelopes covering two or three (or even, for Ce^3 + in phosphate glasses, six) bands. For PTR matrix glasses, the Ce^3 +- and Ce(IV)-related absorption was analyzed recently in detail by Efimov et al. [32], [33] who confirmed, by processing the near UV spectra with the dispersion analysis, the Ce^3 +- and Ce(IV)-related envelopes to comprise three individual bands each. According to [33], the inherent frequencies of three Ce^3 +-related bands due to the 4f¹ (²F_5/2) → 5d¹ transitions are ~ 31,410, ~ 32,770, and ~ 34,690 cm^− 1 (~ 318, ~ 305, and ~ 288 nm, respectively), thus differing substantially from the band locations estimated by Paje et al. [20]. For Ce(IV)-related bands, their inherent frequencies were found in [33] to be ~ 38,680, ~ 44,010, and ~ 49,440 cm^− 1 (~ 259, ~ 227, and ~ 202 nm, respectively).

Sources reporting data on the antimony-related UV absorption in Sb₂O₃-doped glasses are scarce. Mostly, the antimony-related absorption in PTR glasses is merely indicated to contribute, together with other activators, to a broad envelope centered around ~ 240–250 nm that overlaps greatly with the intrinsic absorption tail (see, for example, [34], [35]). Two strong antimony-related bands were resolved preliminarily in [23]. The refined magnitudes of inherent frequencies of these bands in the UV spectra of PTR matrix glasses co-doped with CeO₂ and Sb₂O₃ were found [33] to be ~ 46,660 and ~ 51,900 cm^− 1 (~ 214 and ~ 193 nm, respectively), which differs greatly from antimony-related absorption reported in [19]. Based on the UV spectra of these glasses, Efimov et al. estimated the fractions of Ce(IV) and Ce^3 + species. The Ce(IV) fraction was shown to decrease from approx. 1/4 of the total cerium content in Sb-free glasses to ~ 3–5% in Sb-containing glasses. This effect has demonstrated the Sb(III) valence state to play the role of a reducer with respect to the Ce(IV) valence state due to the redox reaction,

$$\mathrm{Sb}\left(\mathrm{III}\right)+\mathrm{Ce}\left(\mathrm{IV}\right)\to \mathrm{Sb}\left(\mathrm{V}\right)+{\mathrm{Ce}}^{3+}.$$

As seen from all the above, there are substantial contradictions between data reported in particular sources on the Ag⁺-, Ce^3 +-, and antimony-related absorption of virgin and/or ion-exchanged glasses reveals, which can hardly be explained by differences in ways for silver incorporation into initial glass or in the dopant concentrations.

A research described in the given paper was aimed to attaining further insight into the processes of forming the optical properties of ion-exchanged PTR glasses, and, thus, extending the physical and physicochemical background for developing PTR glass-based polyfunctional devices for the integrated optics. Because the process of developing intended variations in the refractive index of commercial PTR glasses involves the UV irradiation and subsequent heat treatment, we investigated not only the UV–VIS absorption spectra of ion-exchanged samples as such but also the effects of the UV irradiation and/or heat treatment on these spectra.