Dissolution Kinetics of Hot Compressed Oxide Glasses - The Journal

Sep 1, 2017 - III, we have discussed the influence of NBOs and atomic packing density on Dr, but we note that these two factors cannot alone explain t...
0 downloads 6 Views 2MB Size
Article pubs.acs.org/JPCB

Dissolution Kinetics of Hot Compressed Oxide Glasses Nerea Mascaraque,† Mathieu Bauchy,‡ José Luis G. Fierro,§ Sylwester J. Rzoska,∥ Michal Bockowski,∥ and Morten M. Smedskjaer*,† †

Department of Chemistry and Bioscience, Aalborg University, 9220 Aalborg, Denmark Department of Civil and Environmental Engineering, University of California, Los Angeles, California 90095, United States § Instituto de Catálisis y Petroleoquímica (CSIC), Cantoblanco, 28049 Madrid, Spain ∥ Institute of High-Pressure Physics, Polish Academy of Sciences, 01-142 Warsaw, Poland ‡

S Supporting Information *

ABSTRACT: The chemical durability of oxide glasses is an important property for a wide range of applications and can in some cases be tuned through composition optimization. However, these possibilities are relatively limited because around 3/5 of the atoms in most oxide glasses are oxygens. An alternative approach involves post-treatment of the glass. In this work, we focus on the effect of hot compression on dissolution kinetics because it is known to improve, for example, elastic moduli and hardness, whereas its effect on chemical durability is poorly understood. Specifically, we study the bulk glass dissolution rate of phosphate, silicophosphate, borophosphate, borosilicate, and aluminoborosilicate glasses, which have been compressed at 0.5, 1.0, and 2.0 GPa at the glass transition temperature (Tg). We perform weight loss and supplementary modifier leaching measurements of bulk samples immersed in acid (pH 2) and neutral (pH 7) solutions. Compression generally improves the chemical durability as measured from weight loss, but the effect is highly composition- and pressure-dependent. As such, we show that the dissolution mechanisms depend on the topological changes induced by permanent densification, which in turn are a function of the changes in the number of nonbridging oxygens and the network cross-linking. We also demonstrate a direct relationship between the chemical durability and the number of chemical topological constraints per atom (nc) acting within the molecular network.

I. INTRODUCTION

occurs at a significantly lower pressure, thus enabling preparation of bulk samples. The effect of compression on chemical durability remains relatively poorly understood. Zhang et al. showed that the densification of silicate and borate glasses at a pressure up to 5 GPa at Tg improves the resistance to corrosion.19,20 We note that densification can involve various mechanisms, which can alter the dissolution in different ways. (i) Densification is associated with a lower amount of free volume within the atomic network, which is likely to slow down the migration of hydration species through the glass. (ii) Densification can also result in the appearance of some structural defects within the network, for example, Si−O−Si angular distortions. Such distortions are likely to result in the formation of some internal stress, thereby rendering the network less stable and more prone to dissolve. (iii) An increase in pressure can lead to the formation of over-coordinated species (e.g., through the conversion of 3-fold into 4-fold coordinated B atoms). The resulting increase in the glass connectivity (or degree of

Various approaches for tuning the dissolution rate of glasses are currently being investigated, such as intrinsic composition optimization involving growth of silver particles,1 rare-earth introduction,2 and substitution of elements.3−5 However, around 3/5 of the atoms in most oxide glasses are oxygens and the possibilities for property tuning are thus relatively limited. Many efforts are therefore also focusing on various post-treatment methods, such as cold and hot solution treatment,6 fire polishing,7 gaseous-reagent treatment,8 coating with polymers as grafting-from method,9 coating with metal,10 and SO2 surface treatment.11 As an alternative route to enhance or tune the corrosion rate, we here investigate hot isostatic compression, which is a method that does not induce a compositional modification of the glass and could also improve other properties, such as elastic moduli and hardness.12,13 The permanent densification of oxide glasses can be achieved by cold compression at pressures above 8−10 GPa,14,15 but most characterization of such densified samples would be prohibited due to their typical small sample volumes. By increasing the temperature during compression, for example, to the glass transition temperature (Tg),16−18 permanent densification © 2017 American Chemical Society

Received: May 11, 2017 Revised: August 30, 2017 Published: September 1, 2017 9063

DOI: 10.1021/acs.jpcb.7b04535 J. Phys. Chem. B 2017, 121, 9063−9072

Article

The Journal of Physical Chemistry B

compositions from a previous study,25 including phosphate, silicophosphate, borophosphate, borosilicate, and aluminoborosilicate systems. The selected compositions were 36Na2O· 64P 2 O 5 , 30Na 2 O·20SiO 2 ·50P 2 O 5 , 50CaO·5B 2 O 3 ·45P 2 O 5 , 15Na2O·10CaO·15B2O3·60SiO2, and 15Na2O·17Al2O3·5B2O3· 63SiO2 (in mol %) and are referred to as P, SiP, BP, BSi, and AlBSi, respectively (Table 1). The five glasses were prepared

polymerization) is likely to increase the resistance to dissolution. As such, the effect of the pressure on dissolution kinetics is nontrivial and might result from a balance of several competing mechanisms. In this work, we study the dissolution behavior of phosphateand silicate-based glasses subjected to permanent densification under pressures up to 2 GPa at Tg. When phosphate- or silicatebased glasses come in contact with an aqueous solution, two main processes occur individually or in combination:21,22 (i) ion exchange involving diffusion between M(+, 2+) (e.g., Na+, Ca2+) close to nonbridging oxygens (NBOs) and protonated species and (ii) hydrolysis involving dissolution of phosphate or silicate chains as a consequence of the breaking of P−O−P and Si−O−P bonds. The densification of silicate and borate glasses19,20 induces a decrease of the average bond angle and void volume, producing an improvement of chemical durability. However, water corrosion depends not only on the shortening of the bond length or the decrease of void volumes but also on the changes in the coordination number of the network-formers (e.g., P, Si, B, or Al) and the concentration of NBOs. For example, many dissolution studies21−23 have focused on the effect of the number of NBOs, that is, the degree of network connectivity, on the resistance to corrosion. In recent studies of oxide glasses,24,25 a direct relation has been demonstrated between the dissolution rate and the number of chemical topological constraints per atom (nc) acting within the molecular network. Topological constraint theory26−28 has been used to establish the dependence of various glass properties on the balance between the number of atomic degrees of freedom and the number of interatomic chemical constraints. For a three-dimensional system, each atom has three translational degrees of freedom, which are removed through the presence of rigid bond constraints, that is, twobody bond stretching (BS) and three-body bond bending constraints. Here, we apply such a topological approach to understand the chemical durability of densified oxide glasses by studying the dissolution behavior in acidic and neutral solutions of phosphate, silicophosphate, borophosphate, borosilicate, and aluminoborosilicate glasses previously compressed at 0.5, 1.0, and 2.0 GPa at their respective Tg. The activation energies needed to break a unit atomic constraint during dissolution are calculated for the densified samples at pHs 2 and 7. We discuss these results based on structural characterization using X-ray photoelectron spectroscopy (XPS) and supplementary measurements of the concentration of leached calcium and sodium modifier cations using atomic absorption spectroscopy (AAS). We show a direct relationship between the pressure-induced changes in nc and the dissolution rates, suggesting that with hot compression and the use of existing topological models, the design of oxide glasses with a specific chemical durability for a determined pH could become possible. As such, the aim of the present study is not to provide additional understanding of the details of the glass dissolution mechanism but rather to enable prediction of the overall practical glass dissolution rate as calculated from the weight loss measurements. For most industrial applications of glass, weight loss experiments provide valuable data and mimic industrial application conditions more accurately than controlled dissolution leaching tests.

Table 1. Nominal Compositions (in mol %), Hot Compression Treatment, and Calculated Number of Constraints Per Atom (nc) of the Investigated Oxide Glassesa composition (in mol %) 36Na2O·64P2O5 P glass

30Na2O·20SiO2·50P2O5 SiP glass

50CaO·5B2O3·45P2O5 BP glass

15Na2O·10CaO·15B2O3·60SiO2 BSi glass

15Na2O·17Al2O3·5B2O3·63SiO2 AlBSi glass

P (GPa)

nc (−)

ambient 0.5 1.0 2.0 ambient 0.5 1.0 2.0 ambient 0.5 1.0 2.0 ambient 0.5 1.0 2.0 ambient 0.5 1.0 2.0

2.76 2.84 2.86 2.84 2.90 2.96 2.98 2.90 3.01 3.06 3.09 3.13 3.35 3.50 3.53 3.54 3.63 3.73 3.76 3.92

reference 32 this this this 32 this this this 33 this this this 34 this this this 35 this this this

work work work work work work work work work work work work work work work

a

The values of nc of the as-prepared glasses are taken from refs32−35, whereas the calculation of nc for the densified glasses is explained in the text.

using the melt-quenching method by mixing batches of the following reagent-grade materials: Na2CO3 (99.999% Suprapur, Merck), CaCO3 (Reag. Ph Eur, Merck), NH4H2PO4 (≥98% ACS Reagent, Sigma-Aldrich), v-B2O3 (B2O3 previously melted at 850 °C using H3BO3 as raw material), SiO2 (Sigma-Aldrich, purum p.a.), and Al2O3 (Sigma-Aldrich, ≥99%). The meltquenching procedure is described in detail elsewhere.25 Samples of each oxide glass were cut to rectangular shape (approximately 0.7 × 0.6 × 0.3 cm3), and their faces were polished using a 4000-grit SiC paper at the final step. These samples were isostatically compressed at 0.5, 1.0, and 2.0 GPa at their respective glass-transition temperature in a nitrogen gas pressure reactor using a vertically positioned chamber for pressures of 0.5 and 1 GPa and horizontally positioned for a pressure of 2.0 GPa. The samples were heated at a constant rate of 600 °C/h to their Tg values, kept under N2 pressure for 30 min, and subsequently quenched with a cooling rate of 60 °C/ min and a decompression rate of 30 MPa/min at room temperature. Additional details about this high-pressure setup can be found in ref29. X-ray photoelectron spectra (XPS) of as-prepared and densified powder samples were obtained with a VG Escalab 200R spectrometer equipped with a hemispherical electron analyzer and a Mg Kα (hν = 1253.6 eV) X-ray source, powered at 100 W. The kinetic energies of photoelectrons were

II. EXPERIMENTAL SECTION To study the effect of compression on chemical durability in various oxide glass systems, we have selected five different 9064

DOI: 10.1021/acs.jpcb.7b04535 J. Phys. Chem. B 2017, 121, 9063−9072

Article

The Journal of Physical Chemistry B measured using a hemispherical electron analyzer working in the constant-pass energy mode (pass energy of 20 eV). The background pressure in the analysis chamber was kept below 8 × 10−9 mbar during data acquisition. The XPS data signals were taken in increments of 0.07 eV and with dwell times of 50 ms. Binding energies were calibrated relative to the C 1s peak at 284.8 eV. High-resolution spectra envelopes were obtained by curve fitting synthetic peak components using software XPS peak. The raw data were used with no preliminary smoothing. Symmetric Gaussian−Lorentzian functions were used to approximate the line shapes of the fitting components. Atomic ratios were computed from experimental intensity ratios and normalized by atomic sensitivity factors.30 To evaluate the effect of the high-pressure treatment on the network densification, we determined the density (ρ) and Vickers hardness (HV) of the as-prepared and densified glasses. Density was measured at room temperature using the Archimedes method. The weight of each glass was measured in ethanol using a balance measuring to ±0.1 mg and repeated 10 times in each glass to determine the standard deviation. Hardness was determined using Vickers microindentation with a Duramin 5 microindenter (Struers A/S) equipped with a Vickers geometry diamond tip. Prior to indentation, the samples were polished using water-free diamond suspensions to avoid surface hydration. Each sample was indented at least 30 times in air at room temperature using a load of 0.49 N and a dwell time of 15 s. Prior to the measurements of durability, the densified samples were cleaned with distilled water under ultrasonication for 5 min and then with acetone under ultrasonication for 5 min and finally dried in an oven at 105 °C for 1 h. Afterward, the weight and surface area were measured. Aqueous solutions (50 mL) of 2 mM HCl and distilled water were used to obtain pH values of 2 and 7, respectively. For each experiment, the solid-to-liquid ratio was selected to be low enough to avoid any saturation effect. The bulk dissolution rate of each glass was determined by measuring the weight loss after immersion of the bulk sample in acidic (pH 2) and neutral (pH 7) aqueous solutions at room temperature as a function of time (t in hour) and the pressure (P in GPa) used during the hot compression treatment. The experiment was repeated twice for each glass composition in each solution. The pH of the media was measured as a function of dissolution time using a pH meter (M220-X9541) and maintained constant throughout the experiments by adding basic or acid solutions. Note that the duration of the dissolution test was adjusted for each glass (from 4 h to nearly 3 months) to achieve substantial values of weight loss. In addition to studying the bulk dissolution rates of the samples, the leaching of calcium and sodium ions was also investigated after the end of bulk dissolution testing, that is, after the longest immersion time. The concentrations of Na+ and Ca2+ (in ppm) were determined using atomic absorption spectrometry (AAnalyst 100, PerkinElmer).

Figure 1. Deconvolution of O 1s core-level spectra of as-prepared and compressed SiP glasses. The dashed and solid lines represent the experimental and fitted data, respectively.

The spectra of the four other glass series are shown in Figures S1−S4 in the Supporting Information. According to the model of Brückner,31 the O 1s spectra of phosphate-based glasses (glasses SiP, P, and BP in Figures 1, S1, and S2, respectively) should be deconvoluted in three Gaussian−Lorentzian mixed components (Voigt profile) attributed to two nonbridging oxygens, PO and M(+ or 2+)···−O−P, at 531.1 and 532.0 eV, respectively, and bridging oxygens, P−O−P, B−O−P, and/or Si−O−Si, at 533.1 eV. For the silicate-based glasses (glasses BSi and AlBSi in Figures S3 and S4, respectively, in the Supporting Information), the O 1s peak is deconvoluted in two Gaussian− Lorentzian mixed components associated with NBOs (Si−O−) at 530.4 eV and BO (Al−O−Si, B−O−Si, and/or Si−O−Si) at

III. RESULTS AND DISCUSSION III.I. X-ray Photoelectron Spectroscopy. To establish the impact of the number of nonbridging oxygens (NBOs) on dissolution rates, we have used XPS to determine the number of bridging oxygens (BOs) and NBOs from the deconvolution of the O 1s core-level peak. Because of the large number of XPS spectra (as-prepared, 0.5, 1, and 2 GPa samples for the five compositions), we here show only the XPS spectra of the O 1s peak for the as-prepared and compressed SiP glasses (Figure 1). 9065

DOI: 10.1021/acs.jpcb.7b04535 J. Phys. Chem. B 2017, 121, 9063−9072

Article

The Journal of Physical Chemistry B

Table 2. Fractions of Bridging and Nonbridging Oxygens (BOs and NBOs) of the As-Prepared and Compressed Glassesa deconvolution of O 1s XPS peak as-prepared

a

compressed at 0.5 GPa

compressed at 1.0 GPa

compressed at 2.0 GPa

ID

BO (%)

NBO (%)

BO (%)

NBO (%)

BO (%)

NBO (%)

BO (%)

NBO (%)

P SiP BP BSi AlBSi

35.3 36.0 28.4 92.4 94.5

64.7 64.0 71.6 7.6 5.5

39.8 38.4 29.9 87.4 96.5

60.2 61.6 70.1 12.6 3.5

42.3 37.9 29.2 86.9 96.1

57.7 62.1 70.8 13.1 3.9

41.6 37.1 28.5 89.0 98.3

58.4 62.9 71.5 11.0 1.7

The error is smaller than ±0.5%.

Figure 2. (a) Density (ρ in g cm−3) and (b) Vickers hardness (HV in GPa) of the five glass compositions as a function of the applied pressure (P in GPa) at the glass-transition temperature. The errors associated with ρ and HV are smaller than ±0.002 g cm−3 and ±0.18 GPa, respectively.

constraints for the as-prepared glasses. We thus calculate nc of the compressed samples based on the variation in hardness between the compressed and the as-prepared glass by assuming a linear scaling between the pressure-induced increase in hardness and nc. The calculated nc values for the compressed glasses are given in Table 1. Both the as-prepared and compressed P and SiP glasses are flexible, whereas the BSi and AlBSi glasses are stressed-rigid. The as-prepared BP glass network is isostatic, becoming stressed-rigid upon compression. III.III. Dissolution Kinetics of Densified Glasses. We test the chemical durability of the compressed glasses in acidic (pH 2) and neutral solutions (pH 7). The dissolution rates of the asprepared glasses in these solutions have already been determined in our recent study.25 Because of the large number of durability experiments (three densified glasses for each glass composition in two solutions), we here show only the weight loss normalized by the initial surface area as a function of time for the densified SiP glasses (Figure 3). The results for the remaining four glass compositions are shown in Figures S6−S9 in the Supporting Information. Dissolution in aqueous solutions can occur through various processes, comprising hydrolysis, hydration, or ion exchange. Processes involving diffusion or leaching of cations are associated with a square root time dependence, whereas processes such as hydrolysis exhibit a linear time dependence.21 In Figure 3 and Figures S6−S9 in the Supporting Information, we observe a linear increase of weight loss with dissolution time, with some initial nonlinearity at short time, which suggests that surface hydrolysis is ratelimiting. Over the immersion time, each solution remains far from saturation and no plateau in the weight loss versus time curve is observed, that is, the kinetics of the dissolution is controlled by only the structural and topological changes with glass composition and pressure.

531.9 eV. The fractions of the as-prepared and densified glasses are summarized in Table 2. In general, the densification leads to an increase in the fraction of bridging oxygens, that is, an increase in the degree of connectivity of the phosphate or silicate network. However, each pressure treatment induces different structural changes depending on the glass composition. We discuss the pressure dependence of the number of bridging and nonbridging oxygens for each glass composition and its relationship with the chemical dissolution behavior in Section III.III. III.II. Density, Hardness, and Topological Constraints. The effects of hot compression on density and hardness are illustrated in Figure 2a,b, respectively. Data are also summarized in Table S1 in the Supporting Information. Figure 2a shows an increase in density with increasing applied pressure at Tg, as observed previously in refs12, 13, 29. A similar increase with the applied pressure is also observed in hardness (Figure 2b). The magnitude of the increase in hardness is compositiondependent as a result of the different pressure-induced topological changes. Following Maxwell’s stability criterion,36 an atomic network can be classified as flexible if nc (number of atomic constrains per atom) 3, or isostatic for nc = 3. In our recent study,25 we have enumerated the value of nc for the present glasses based on detailed structural information to explain the trend in chemical durability. However, such information is not available for the compressed glasses in this work and we therefore apply an alternative approach to obtain nc of the compressed glasses based on the measured hardness data. Hardness has been shown to scale linearly with nc in various oxide glass systems,32−35 and Figure S5 in the Supporting Information indeed shows a direct relationship between the measured hardness and the calculated number of 9066

DOI: 10.1021/acs.jpcb.7b04535 J. Phys. Chem. B 2017, 121, 9063−9072

Article

The Journal of Physical Chemistry B

Figure 3. Weight loss (in mg dm−2) of the densified SiP glass as a function of time (t in h) under different conditions: (a) pH 2 for P = 0.5 GPa, (b) pH 7 for P = 0.5 GPa, (c) pH 2 for P = 1.0 GPa, (d) pH 7 for P = 1.0 GPa, (e) pH 2 for P = 2.0 GPa, and (f) pH 7 for P = 2.0 GPa. Two repeated experiments are performed for each combination of glass, solution pH, and pressure. The dashed lines represent linear fits to the data.

P − O − B + H 2O → P − OH + B − OH

Figures 4 and 5 show the pressure and pH dependence of the logarithmic dissolution rate, Dr, for all of the compressed glasses. Dr is calculated from the slope of linear fits to the weight loss curves in Figure 3 and Figures S6−S9 in the Supporting Information. The dissolution mechanism of the phosphate-based glasses could be explained through the reactions of ion exchange (1) and hydrolysis (2).25 In neutral solution, there is an equilibrium between the concentrations of H+ and OH−, that is, both protonation and hydrolysis are strongly coupled, and the possible reactions are P − O− − M+ + H 2O → P − OH + M+ + OH−

In acidic solutions (pH 2), reaction 1a,b is favored as the ionexchange step is a process of protonation of phosphate chains, and a higher concentration of H+ is presented in the aqueous solutions

P − (O−)2 − M2 + + 2H 2O (1b)

P − O − P + H 2O → 2(P − OH)

(2a)

P − O− − M+ + H+ → P − OH + M+

(3a)

P − (O−)2 − M2 + + 2H+ → P − (OH)2 + M2 +

(3b)

As for phosphate network glasses, the dissolution mechanism of silicate-based glasses could be explained through the reactions of ion exchange (4a,b and 5a,b) and hydrolysis (6) for neutral (pH 7) and acidic (pH 2) solutions25

(1a)

→ P − (OH)2 + M2 + + 2OH−

(2b)

(Si, B, or Al) − O− − M+ + H 2O → (Si, B, or Al) − OH + M+ + OH− 9067

(4a)

DOI: 10.1021/acs.jpcb.7b04535 J. Phys. Chem. B 2017, 121, 9063−9072

Article

The Journal of Physical Chemistry B

Figure 4. Pressure dependence of the logarithmic dissolution rate (Dr, in mg dm−2 h−1) for different flexible and isostatic samples: (a) P glass at pH 2, (b) P glass at pH 7, (c) SiP glass at pH 2, (d) SiP glass at pH 7, (e) BP glass at pH 2, and (f) BP glass at pH 7. Dissolution data for the as-prepared glasses are also included and taken from ref24.

(M+ or M2+, respectively) close to NBOs and the protons (H+) is preferred in acidic media (reactions 3a,b and 5a,b). However, nucleophilic attack (reactions 2a,b and 6) or the breaking of P− O−P, Si−O−Si, B−O−P, Al−O−Si, or B−O−Si bonds is associated with the network cross-linking, that is, with a higher degree of polymerization network, and the nucleophilic attack or the diffusion of water to break bonds is more impeded. Figure 4 shows a general enhancement of the corrosion resistance at both pHs 2 and 7 (i.e., Dr decreases) of the phosphate-based P, SiP, and BP glasses following permanent densification. Figure 4a,b shows the same trend of corrosion behavior for the P samples in acidic and neutral media. The compression up to 1 GPa leads to an improvement of chemical durability due to an increase of atomic packing density, as a consequence of the decrease of the average bond angle and

(Si, B, or Al) − (O−)2 − M2 + + 2H 2O → (Si, B, or Al) − (OH)2 + M2 + + 2OH−

(4b)

(Si, B, or Al) − O− − M+ + H+ → (Si, B, or Al) − OH + M+

(5a)

(Si, B, or Al) − (O−)2 − M2 + + 2H+ → (Si, B, or Al) − (OH)2 + M2 +

(5b)

Si − O − (Si, B, or Al) + H 2O → 2(Si, B, or Al) − OH (6)

The protonation reactions (1a,b and 4a,b) or the ion exchange between the network modifier alkali or alkaline earth cations 9068

DOI: 10.1021/acs.jpcb.7b04535 J. Phys. Chem. B 2017, 121, 9063−9072

Article

The Journal of Physical Chemistry B

Figure 5. Pressure dependence of the logarithmic dissolution rate (Dr, in mg dm−2 h−1) for different stressed-rigid samples: (a) BSi glass at pH 2 (a), (b) BSi glass at pH 7, (c) AlBSi glass at pH 2, and (d) AlBSi glass at pH 7. Dissolution data for the as-prepared glasses are also included and taken from ref24.

void volume,19,20 and a decrease in the number of NBOs (see Table 2), that is, a higher phosphate network connectivity. However, when the P glass is compressed at 2.0 GPa, the number of NBOs is higher than that at 1.0 GPa, leading to a smaller increase in chemical durability from 1.0 to 2.0 GPa relative to that from 0 to 1.0 GPa. The Dr and NBO values of the P glasses exhibit the same trend, with the 1.0 GPa sample exhibiting the lowest Dr and NBO value. This result points to the importance of the ion-exchange process, that is, the impact of the NBOs on corrosion behavior, which is also observed in Figure S10a in the Supporting Information. On the other hand, the faster protonation reaction at pH 2 is verified through the results of the leaching study (Table S2 and Figure S11 in the Supporting Information). A higher concentration of sodium is measured in the pH 2 solution compared to that in pH 7 solution, implying that the rate of the diffusion process 1 is faster for lower pH. Moreover, the concentrations of Na+ measured by AAS and those calculated by assuming congruent dissolution are fairly similar (see Table S2 in the Supporting Information), indicating that the P glasses exhibit congruent dissolution of modifiers. The densified SiP glasses (Figure 4c,d) also exhibit lower dissolution rate values, which align with the pressure dependence of fraction of NBOs and density. The densified BP samples (Figure 4e,f) with stressed-rigid networks (nc > 3) show more impeded water attack due to the higher network connectivity, and the kinetics of dissolution is thus controlled by nucleophilic attack (reaction 4b). Therefore, the Dr is almost constant for all densified BP samples under neutral conditions

(Figure 4f), and the high influence of the number of NBOs is clear for acidic solutions (Figure 4e). We note that in Figure 4f the change in Dr with pressure is significantly smaller than that for P (Figure 4b) and SiP glasses (Figure 4d). This could be because the modifier cation in BP glasses is calcium, which diffuses more slowly compared with sodium.3 The larger influence of nucleophilic reaction 2b relative to that of ionexchange reaction 1a,b is confirmed by the leaching of calcium cations (see Table S2 in the Supporting Information). That is, the concentration of leached Ca2+ is lower than that calculated assuming congruent modifier dissolution, indicating that the modifier dissolution is not congruent and diffusion process 1a is slower than hydrolysis process 2b. On the other hand, the small changes in Dr of BP glass after densification could also be related to the changes in the boron species. In B2O3-containing glasses,12 it is observed that the densification is accompanied by the partial transformation of B3 to B4. In borate glasses,20 it has been shown that the lowest Dr is associated with the highest fractions of tetrahedral boron. For the densified silicate-based BSi and AlBSi glasses (Figure 5), the stressed-rigid networks limit the diffusion of water across the glass. The nucleophilic attack in pH 7 media (reaction 6) becomes predominant in samples with a higher degree of network cross-linking, whereas ion-exchange process 5a is preferred for acid media and in turn influenced by the number of NBOs. Figure 5a reveals an increase in dissolution rate at pH 2 upon compression at 0.5 and 1.0 GPa of the BSi glass as a consequence of the increase of the fraction of NBOs (see Table 2). On the other hand, Figure 5c shows a 9069

DOI: 10.1021/acs.jpcb.7b04535 J. Phys. Chem. B 2017, 121, 9063−9072

Article

The Journal of Physical Chemistry B

Figure 6. Logarithmic dissolution rate (Dr, in mg dm−2 h−1) of the densified oxide glass compressed at (a) 0.5 GPa, (b) 1.0 GPa, and (c) 2.0 GPa as a function of the calculated number of constraints per atom (nc) in aqueous solutions with pHs 2 and 7. The dashed lines are drawn as guides for the eyes. Insets: Dr on a linear scale as a function of nc and pH. Error bars are of the size of the symbols.

numbers that contribute to larger nc values, because the nc values are extracted from the measured hardness data. The strong correlation between network topology and the resistance of the glasses to water attack is illustrated in the universal plot of ln Dr versus nc in Figure 7, which highlights that the network topology controls the majority of the variation in dissolution rate, independent of whether nc is changed by compositional design or hot compression. The hot-compression process in

continuous improvement of dissolution resistance at pH 2 with increasing pressure for the AlBSi glass. This is due to a lower fraction of NBOs upon compression, combined with the increase of density. The results of the leaching study confirm that the protonation reaction is preferred for pH 2 (Table S2 in the Supporting Information). In neutral solutions (Figure 5b,d), Dr decreases from as-prepared to densified glass due to the higher density and network cross-linking for both BSi (Figure 5b) and AlBSi (Figure 5d) glasses. This could be also due to the pressure-induced changes in the aluminum and boron speciation through the transformations of B3 to B4 or Al4 to Al5/Al6.12,13 III.IV. Correlating Network Topology and Dissolution Kinetics. We have previously shown that the dissolution rate scales with nc for the as-prepared glasses in this study.25 Now, we test whether this scale also holds for the compressed glasses. To this end, log Dr is plotted as a function of nc for the glasses compressed at 0.5, 1.0, and 2.0 GPa in Figure 6. A direct relationship between the dissolution rate and nc is observed for all densified glasses in both acidic and neutral pH media at the various pressures. The insets in Figure 6 show the nc dependence of Dr on a linear scale, clearly highlighting the difference in Dr among the flexible and stressed-rigid networks. In Section III.III, we have discussed the influence of NBOs and atomic packing density on Dr, but we note that these two factors cannot alone explain the pressure dependence of Dr, as shown in Figures S10 and S12 in the Supporting Information, respectively. On the other hand, the network topology captures the full details of the atomic structure following densification, including the pressure-induced increases in coordination

Figure 7. Dependence of the natural logarithm of the dissolution rate (Dr, in mg dm−2 h−1) in different aqueous solutions (pHs 2 and 7) on the number of atomic constraints per atom (nc) of the as-prepared and densified oxide glasses (0−2.0 GPa). The data for the as-prepared glasses are taken from refs32−35. The two dashed lines indicate the master curves for dissolution at pHs 2 and 7, respectively. 9070

DOI: 10.1021/acs.jpcb.7b04535 J. Phys. Chem. B 2017, 121, 9063−9072

Article

The Journal of Physical Chemistry B oxide glasses leads to general improvement of chemical properties (Figures 4 and 5), as well as mechanical properties (Figures 2). The reported nc−Dr relations in Figure 7 could open up for the atomistic design of new oxide glasses with a specific hardness bulk dissolution rate for a determined pH value. As discussed in Section III.III, the dissolution in the different types of oxide glasses involves different reactions, which could occur individually or in combination, depending on pH. To understand the origin of the nc−Dr relations in Figure 7, we next consider an energy landscape point of view. Each topological constraint creates an energy minimum inside the energy landscape, whereas internal degrees of freedom (3 − nc) induce the formation of channels between these energy minima,37 allowing atomic reorganizations with low energy barriers. To determine this energy barrier, we consider an Arrhenius-like relation of the form24

⎛ E eff ⎞ Dr = Dr,0 exp⎜⎜ − a ⎟⎟ ⎝ RT ⎠

Figure 8. Energy barrier needed to break a unit constraint (E0) as a function of solution pH and pressure. E0 is determined by fitting the data in Figure S13 in the Supporting Information to eqs 7 and 8.

(7)

mechanism controlling the kinetics of dissolution with the reactions of hydrolysis and ion exchange occurring more in combination when the pressure reaches 2.0 GPa, as the difference between E0 in acidic and neutral solutions is significantly smaller at 2.0 GPa pressure than under ambient conditions.

where R is the gas constant, T is the temperature, and Dr,0 is a rate constant that depends on the solution phase chemistry, yielding the barrierless dissolution rate of completely depolymerized glass with nc = 0. Eeff a is the effective activation energy given as24 Eaeff = ncE0

IV. CONCLUSIONS The dissolution kinetics of different phosphate- and silicatebased densified glasses has been studied by measuring the weight loss and modifier leaching concentrations of bulk samples immersed in acid (pH 2) and neutral (pH 7) media. Supplementary density, hardness, and XPS measurements have also been performed. The hot compression treatment leads to a general improvement of the chemical durability, which is found to be directly correlated to the increase in network rigidity. The activation energy to break a unit constraint during dissolution, which is an indicator of the dissolution mechanism, is always found to be lower at pH 2 than at pH 7 independent of the applied pressure, highlighting the universal topological control of dissolution. With this research and existing topological models, it is thus possible to design new oxide glasses with a specific chemical durability based on a combination of compositional tuning and hot compression post-treatment.

(8)

where E0 is the energy barrier that needs to be overcome to break a unit atomic constraint. Here, we are considering isothermal dissolution and E0 is therefore determined from the slope of linear evolution of the logarithmic dissolution rate with respect to nc (see Figure S13 in the Supporting Information). The values of Eeff a are summarized in Table 3, and the fitted Table 3. Effective Activation Energy (Eeff a ) for Dissolution of All of the Compressed Samples at pHs 2 and 7 −1 effective activation energy, Eeff a (kJ mol )

compressed at 0.5 GPa

compressed at 1.0 GPa

compressed at 2.0 GPa

nc (−)

pH 2

pH 7

nc (−)

pH 2

pH 7

nc (−)

pH 2

pH 7

2.84 2.96 3.06 3.50 3.73

27.1 28.2 29.2 33.4 35.6

49.3 51.4 53.1 60.8 64.8

2.86 2.98 3.09 3.53 3.76

24.5 25.5 26.4 30.2 32.2

49.5 51.6 53.5 61.2 65.1

2.84 2.90 3.13 3.54 3.92

29.7 30.3 32.7 37.0 41.0

45.3 46.3 50.0 56.5 62.6



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.jpcb.7b04535. O 1s core-level XPS spectra of as-prepared and compressed samples at 0.5, 1.0, and 2.0 GPa; hardness of as-prepared samples as a function of the number of atomic constraints per atom; weight loss curves for the compressed P, BP, BSi, and AlBSi glasses in different pH solutions; dependence of the dissolution rate on the pressure-induced changes in density and fraction of NBOs; atomic constraint dependence on dissolution rate; concentrations of leached sodium and calcium ions; density and hardness values of densified glasses (PDF)

values of E0 are plotted against the applied pressure in Figure 8 for pHs 2 and 7. In general, the activation energy is lower in acidic solution compared to that in neutral pH solution as the ion-exchange process involves local deformation of the network, which should be associated with lower activation energy. Hydrolysis involves the breaking of bonds and thus higher activation energy. As expected, this infers that the dissolution by hydrolysis is predominant for pH 7, whereas ion exchange is predominant at pH 2. Considering the error range of the data, E0 appears to remain constant with pressure (at least up to 1.5 GPa) at each pH value, confirming the universal topological control of dissolution discussed above. However, we note that E0 slightly decreases and increases when the applied pressure reaches 2.0 GPa in pH 2 and pH 7 solutions, respectively (Figure 8). This could indicate a change in the



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. 9071

DOI: 10.1021/acs.jpcb.7b04535 J. Phys. Chem. B 2017, 121, 9063−9072

Article

The Journal of Physical Chemistry B ORCID

(16) Uhlmann, D. R. Densification of Alkali Silicate Glasses at High Pressure. J. Non-Cryst. Solids 1973, 13, 89−99. (17) Yamada, A.; Gaudio, S. J.; Lesher, C. E. Densification of MgSiO3 Glass with Pressure and Temperature. J. Phys.: Conf. Ser. 2010, 215, No. 012085. (18) Guerette, M.; Ackerson, M. R.; Thomas, J.; Yuan, F.; Watson, E. B.; Walker, D.; Huang, L. Structure and Properties of Silica Glass Densified in Cold Compression and Hot Compression. Sci. Rep. 2015, 5, No. 15343. (19) Zhang, Z.; Soga, N.; Hirao, K. Water Corrosion Behavior of Densified Glass. I. Silicate glass. J. Non-Cryst. Solids 1991, 135, 55−61. (20) Zhang, Z.; Hirao, K.; Soga, N. Water Corrosion Behavior of Densified Glass. II. Borate glass. J. Non-Cryst. Solids 1991, 135, 62−66. (21) Bunker, B. C.; Arnold, G. W.; Wilder, J. A. Phosphate Glass Dissolution in Aqueous Solutions. J. Non-Cryst. Solids 1984, 64, 291− 316. (22) Bunker, B. C. Molecular Mechanisms for Corrosion of Silica and Silicate Glasses. J. Non-Cryst. Solids 1994, 179, 300−308. (23) Hamilton, J. P.; Brantley, S. L.; Pantano, C. G.; Criscenti, L. J.; Kubicki, J. D. Dissolution of Nepheline, Jadeite and Albite Glasses: Toward Better Models for Aluminosilicate Dissolution. Geochim. Cosmochim. Acta 2001, 65, 3683−3702. (24) Pignatelli, I.; Kumar, A.; Bauchy, M.; Sant, G. Topological Control on Silicates’ Dissolution Kinetics. Langmuir 2016, 32, 4434− 4439. (25) Mascaraque, N.; Bauchy, M.; Smedskjaer, M. M. Correlating the Network Topology of Oxide Glasses with their Chemical Durability. J. Phys. Chem. B 2017, 121, 1139−1147. (26) Phillips, J. C. Topology of covalent non-crystalline solids I: Short-Range Order in Chalcogenide Alloys. J. Non-Cryst. Solids 1979, 34, 153−181. (27) Phillips, J. C.; Thorpe, M. F. Constraint Theory, Vector Percolation and Glass Formation. Solid State Commun. 1985, 53, 699− 702. (28) He, H.; Thorpe, M. F. Elastic Properties of Glasses. Phys. Rev. Lett. 1985, 54, 2107−2110. (29) Smedskjaer, M. M.; Rzoska, S. J.; Bockowski, M.; Mauro, J. C. Mixed Alkaline Earth Effect in the Compressibility of Aluminosilicate Glasses. J. Chem. Phys. 2014, 140, No. 054511. (30) Wagner, C. D.; Davis, L. E.; Zeller, M. V.; Taylor, J. A.; Raymond, R. H.; Gale, L. H. Empirical Atomic Sensitivity Factors for Quantitative Analysis by Electron Spectroscopy for Chemical Analysis. Surf. Interface Anal. 1981, 3, 211. (31) Brückner, R.; Chun, H.-U.; Goretzki, H.; Sammet, M. XPS Measurements and Structural Aspect of Silicate and Phosphate Glasses. J. Non-Cryst. Solids 1980, 42, 49−60. (32) Hermansen, C.; Guo, X. J.; Youngman, R. E.; Mauro, J. C.; Smedskjaer, M. M.; Yue, Y. Z. Structure-Topology-Property Correlations of Sodium Phosphosilicate Glasses. J. Chem. Phys. 2015, 143, No. 064510. (33) Hermansen, C.; Youngman, R. E.; Wang, J.; Yue, Y. Z. Structural and Topological Aspects of Borophosphate Glasses and their Relation to Physical Properties. J. Chem. Phys. 2015, 142, No. 184503. (34) Smedskjaer, M. M.; Mauro, J. C.; Youngman, R. E.; Hogue, C. L.; Potuzak, M.; Yue, Y. Z. Topological Principles of Borosilicate Glass Chemistry. J. Phys. Chem. B 2011, 115, 12930−12946. (35) Smedskjaer, M. M. Topological Model for Boroaluminosilicate Glass Hardness. Front. Mater. 2014, 1, No. 23. (36) Maxwell, J. C. L. On the Calculation of the Equilibrium and Stiffness of Frames. Philos. Mag. 1864, 27, 294−299. (37) Naumis, G. G. Energy Landscape and Rigidity. Phys. Rev. E: Stat., Nonlinear, Soft Matter Phys. 2005, 71, No. 026114.

Morten M. Smedskjaer: 0000-0003-0476-2021 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS N.M. and M.M.S. acknowledge financial support from VILLUM Fonden (Postdoctoral Block Fellowship Program). M.B. acknowledges support from the National Science Foundation (Grant No. 1562066). S.J.R. acknowledges the support from the National Science Center of Poland under Grant No. UMO-2016/21/B/ST3/02203.



REFERENCES

(1) Meincke, T.; Pacheco, V. M.; Hoffmann, D.; Boccaccini, A. R.; Klupp Taylor, R. N. Engineering the Surface Functionality of 45S5 Bioactive Glass-Based Scaffolds by the Heterogeneous Nucleation and Growth of Silver Particles. J. Mater. Sci. 2017, 52, 9082−9090. (2) Nicoleau, E.; Angeli, F.; Schuller, S.; Charpentier, T.; Jollivet, P.; Moskura, M. Rare-earth Silicate Crystallization in Borosilicate Glasses: Effect on Structural and Chemical Durability Properties. J. Non-Cryst. Solids 2016, 438, 37−48. (3) Hopf, J.; Kerisit, S. N.; Angeli, F.; Charpentier, T.; Icenhower, J. P.; McGrail, B. P.; Windisch, C. F.; Burton, S. D.; Pierce, E. M. Glass− Water Interaction: Effect of High-valence Cations on Glass Structure and Chemical Durability. Geochim. Cosmochim. Acta 2016, 181, 54−71. (4) Zhang, H.; Corkhill, C. L.; Heath, P. G.; Hand, R. J.; Stennett, M. C.; Hyatt, N. C. Effect of Zn- and Ca-Oxides on the Structure and Chemical Durability of Simulant Alkali Borosilicate Glasses For Immobilisation of UK High Level Wastes. J. Nuclear Mater. 2015, 462, 321−328. (5) Le Sauze, A.; Marchand, R. Chemically Durable Nitrided Phosphate Glasses Resulting from Nitrogen/Oxygen Substitution within PO4 Tetrahedra. J. Non-Cryst. Solids 2000, 263−264, 285−292. (6) Maschinen, W. Fire polishing adds quality to all glass container types. Glass Int. 2009, 32, 49. (7) Tichane, R. M.; Carrier, G. B. Microstructure of a Soda-Lime Glass Surface. J. Am. Ceram. Soc. 1961, 44, 606−610. (8) Yashchishin, I. N.; Kozii, O. I.; Bedzhanova, O. F.; Zelinski, T. B. Gaseous-Reagent Treatment of Glass. Glass Ceram. 1987, 44, 230− 232. (9) Sangermano, M.; Periolatto, M.; Castellino, M.; Wang, J.; Dietliker, K.; Grützmacher, J. L.; Grützmacher, H. A Simple Preparation of Photoactive Glass Surfaces Allowing Coatings via the “Grafting-from” Method. ACS Appl. Mater. Interfaces 2016, 8, 19764− 19771. (10) Chen, H.; Tai, Y.; Xu, C. Fabrication of Copper-Coated Glass Fabric Composites through Electroless Plating Process. J. Mater. Sci.: Mater. Electron. 2017, 28, 798−802. (11) Douglas, R. W. The Action of Sulphur Dioxide and of Water on Glass Surfaces. J. Soc. Glas. Technol. 1949, 33, 289−335. (12) Svenson, M. N.; Guerette, M.; Huang, L.; Lönnroth, N.; Mauro, J. C.; Rzoska, S. J.; Bockowski, M.; Smedskjaer, M. M. Universal Behavior of Changes in Elastic Moduli of Hot Compressed Oxide Glasses. Chem. Phys. Lett. 2016, 651, 88−91. (13) Bechgaard, T. K.; Goel, A.; Youngman, R. E.; Mauro, J. C.; Rzoska, S. J.; Bockowski, M.; Jensen, L. R.; Smedskjaer, M. M. Structure and Mechanical Properties of Compressed Sodium Aluminosilicate Glasses: Role of Non-Bridging Oxygens. J. NonCryst. Solids 2016, 441, 49−57. (14) Hemley, R. J.; Mao, H. K.; Bell, P. M.; Mysen, B. O. Raman Spectroscopy of SiO2 Glass at High Pressure. Phys. Rev. Lett. 1986, 57, 747−750. (15) Salmon, P. S.; Zeidler, A. Networks under Pressure: The Development of In Situ High-Pressure Neutron Diffraction for Glassy and Liquid Materials. J. Phys.: Condens. Matter 2015, 27, No. 133201. 9072

DOI: 10.1021/acs.jpcb.7b04535 J. Phys. Chem. B 2017, 121, 9063−9072