The pristine glass structure of lead crystal glass composition (glass A) was carefully investigated in a previous work7. 17O multi-quantum magic-angle spinning (MQMAS) NMR was used to extract various oxygen environments. The presence of sites mixing Pb and K near non-bridging oxygen (NBO) was pointed out as well as the absence of Pb-O-Pb clustering. Pristine lead crystal glass (glass A) 29Si MAS NMR spectrum is given in Fig. 1 and the quantification of this spectra is available in the Supplementary Information of Angeli et al.7 which resulted in 59% of Q3 units and 41% of Q4 units, in good agreement with NBO calculation considering that Pb2+ leads to the formation of two NBOs in the silicate network. Thus, in this case, Pb appears to play the role of a network modifier element.Fig. 1: 29Si MAS NMR spectra.Pristine glass A (lead crystal glass) and glass altered 969 days are represented as well as 1H-29Si cross-polarization (cpMAS) spectra obtained with proton-to-silicon magnetization transfer time of 2 ms for the altered glass.The 27Al MAS spectra for glasses B and C in Fig. 2 show a single contribution corresponding to the signature of tetrahedral aluminum8 in accordance with glass chemical composition containing an excess of alkalis and alkaline earth elements in comparison with Al content. As a consequence, all [AlO4]− groups are compensated by a positive charge. The same features as in glass B are observed in the soda-lime glass C from 27Al MAS NMR, reflecting that all the aluminum atoms are in four-fold coordination. The aluminum charge compensation can fully be provided by the high Na content in this glass (13–15 mol% of Na2O for <2 mol% of Al2O3).Fig. 227Al MAS NMR spectra obtained from pristine glass of B, C and D samples.A comprehensive investigation of the structure of a Pyrex glass (glass D) was proposed by Tricot et al.9. Our glass composition (SiO2:82.1|B2O3:10.7|Al2O3:1.6|Na2O: 5.6) mol% however is slightly different than their own (SiO2:83.0|B2O3:11.6|Al2O3:1.4|Na2O: 4) mol%. Consequently the borate speciation is different due to the higher Na/B ratio in our glass resulting in an increase of tetrahedral boron (35% instead of 20%) as shown in Fig. 3a. This BO4 content is expected from the glass composition considering Dell and Bray model10 after removing Na acting as Al charge compensator. In these conditions, the glass network is fully polymerized without non-bridging oxygen. Regarding aluminum, the 27Al MAS NMR spectrum is presented in Fig. 2 and shows tetrahedral coordination of Al only. But contrarily to the spectra obtained on B and C glasses, a decrease of its chemical shift and a broader line reflecting an increased quadrupolar coupling constant are observed for glass D. Both features are explained by the more constraint environment of Al in glass D, where only sodium can compensate Al units and the network is fully polymerized unlike glasses B and C. The two components of BO3 units are depicted by thin red dashed lines in Fig. 3b with the contribution of non-ring BO3 corresponding to an 11B isotropic shift of 35 ppm and BO3 involved in rings showing up at 42 ppm, while the two components for the BO4 line in Fig. 3a account for the B atoms with 4 Si as second neighbors for the component centered at −2 ppm and B with 1 B surrounded by 3 Si as second neighbors for the one centered at −0.25 ppm.Fig. 3: 11B spectra obtained from pristine glass D (borosilicate).a 11B MAS NMR experimental and simulated spectra. b 11B multi-quantum magic-angle spinning (MQMAS) NMR spectra.During the 3 years of alteration, the liquid medium has been regularly sampled to gain access to the kinetics of alteration of the studied glasses in identical conditions. Each reactor contained glass under the form of powder and slab (a single slab per reactor) with 97% of the exposed surface of glass originating from the powder. Consequently, the release of the main constituting elements shown in Fig. 4 can be considered as indicative of the alteration of the powder with a very small contribution from the glass slab. A potassium contamination caused by the pH probe was noticed in the reactors, therefore the leaching behavior of potassium for B and C glasses were discarded. The level of contamination has been measured by quantifying K by ICP-AES in the samplings of glass D reactor which has no internal source of potassium. This pollution is negligible in the case of glass A since the value of potassium measured in solution is always higher by at least 1 order of magnitude to the value of potassium in glass D. Unfortunately, it is not the case for sample glasses B and C and data about potassium measured in solution had to be discarded. No saturation limit for any element has been reached in this experiment thanks to the highly diluted medium. The markers of calcium have intentionally faded in Fig. 4 as this element was very much subjected to lab contamination resulting in noisy analytical data and unusable rate calculations. At first sight the depletion of alkali elements (Na, K) of glass A sample (Fig. 4a) outpaced the release of Na found in the three other glasses (B, C and D) by at least a factor 5. Overall the alteration of the four glasses showed different rates and mechanisms as explained more in depth for each glass hereafter.Fig. 4: Evolution of the equivalent thicknesses (ETh) of the main constituting elements.Glasses A (a, b), B (c), C (d) and D (e) were leached over 3 years (1096 days). ETh calculations are based on ICP-AES analysis of the alteration solutions and Eqs. (1)–(3), (6), (7).For glass A, as expected the leaching behavior of alkalis (Na, K) is not linear with respect to time scale, but perfectly complies with the second Fick’s law, which has a dependency towards the square root of time, and is archetypal of a diffusion-controlled dissolution process for these species (a plot of the equivalent thicknesses of Na, ETh(Na), versus the square root of the duration is available in Fig. 6b). Si measurements exhibit an almost linear release, as observed on Fig. 4b, synonymous with the constant dissolution by hydrolysis of the silicate network over time. Lead, the most critical element towards toxicology, was firstly released in solution with a maximal diffusion coefficient after 7 days of alteration D7d(Pb) = 1.6 × 10−19 m2 s−1 calculated from Eq. (5). A detailed view of Pb release in function of the square root of the duration of alteration is given in Supplementary Fig. 1 available in the Supplementary Information. This initial leaching of Pb is then followed up by a sharp and continuous decrease of the diffusion coefficient that dropped to D28d(Pb) = 4.6 × 10−20 m2 s−1 after 28 days of alteration. This behavior is also mirrored by the evolution of equivalent thicknesses of glass (ETh) values that reached a plateau after 56 days of alteration that lasted for the remaining 2 years and 10 months of the experiment. This observation is rationalized by the formation of a passivating layer at the interface between the pristine glass and the solution with different retention properties towards the various elements of the glass. The remaining altered layer was observed after 3 years of alteration by SEM-EDX on a polished cross section, as shown in Fig. 5a. Alkalis, and especially potassium, are clearly depleted from the surface of the grains, as evidence by the chemical mapping of K (EDX chemical mapping of Na, very scarce in the glass composition, is given in Supplementary Fig. 2 available in the Supplementary Information). In the SEM micrographs, the range of alteration layer widths encountered is high as the grains are not perfectly perpendicular to the surface of the cross-section, with non-perpendicular grains exhibiting higher altered layer. Consequently the measurement of the width of alteration was carried out by scaling the thinnest altered layer and these results remained in accordance with the ETh plot (Fig. 4a). Unlike displayed in the ETh plot no difference between the depletion of Na and K was noticed in the SEM micrographs. It is likely that this contrast between both alkalis came from the noisy data regarding sodium and that there is no difference between both elements, as depicted by the ToF-SIMS profile performed after 231 days of alteration on the glass slab (Fig. 7a). Alternatively Pb appeared entirely retained in the altered layer at a microscopic scale, perfectly overlapping the signal of Si and confirming the retention of Pb in the altered layer.Fig. 5: SEM micrographs (backscattered electrons).Glass grains were altered for 3 years (1096 days) in acetic acid 4% (v/v) at pH = 2.4 and 70 °C,.prepared as polished cross-sections from glasses A (a), B (b), C (c) and D (d) and EDX mapping of the colored mentioned elements.Barium crystalline glass, glass B, designed as an alternative to lead crystal glass, intending to improve its chemical durability, showed a very different pattern of alteration in terms of magnitude and mechanisms. Differences between glass constituents leaching are faint and the maximal altered depth is held by sodium, with 1.14 ± 0.09 µm, more than 10 times less than glass A.The addition of alkaline earth elements to the composition of glass B, strengthening the glassy network, can explain this considerable difference11. Samplings are easily contaminated with minor Ca pollution in a lab environment explaining the greater scatter among the Ca data points, nonetheless, Ca exhibits a leaching trend similar to the one of alkaline and alkaline earth elements. After 200 days of alteration, a global decrease of alteration rates is noticeable for all elements in Fig. 4c, most probably driven by the formation of an altered layer, slowing down the elemental diffusion from the interdiffusion front to the solution. Unlike Pb in glass A altered layer, no specific retention of any element is observed but, all elements, including Si, seemed released through a diffusion mechanism, explaining the non-linearity of the evolution of ETh. Despite clear signs of the formation of an altered layer, this latter was not observed after 3 years of alteration via SEM. Figure 5b images the grains and a close up view of the edge of a grain where the chemical mapping of Si and Na show no difference. Indeed, the depleted width in Na would be of 0.16 µm from the ETh calculations which would be hardly detectable by SEM. Even at high magnification the distribution of Al, Ba, Ca and K remained homogeneous and perfectly superimposed to the one of Si and Na. Hydrolysis probably also played an important role in the dissolution of the altered layer throughout the experiment leaving no depleted area in any of the constituting elements. Compared to glass A, glass B sample shows greater durability regarding tracing elements like Na, comparable to the ones measured on glass D.Leaching of soda-lime glass, glass C, given in Fig. 4d, showed a distinct preferential release of Na. Both initial alteration rate and ETh(Na) after 3 years are the highest of this glass, in good agreement with classical leaching descriptions by ion exchange/interdiffusion from the literature on soda-lime silicate commercial glasses under acidic12 or neutral13 pH conditions. A smooth decrease of the alteration rates of all elements, is perceived after 300 days, likely driven by the properties of the alteration layer. As seen in Fig. 4c for glass B, Ca and Ba are not released linearly in function of time, suggesting that they are not leached out by hydrolysis. Although Ca data is noisier, ion-exchange still appeared favored based on the plateau observed for Ca release in Fig. 4d. Magnesium, which is also an alkaline earth element, only present in glass C, appeared closely linked to the leaching of barium, which is also an alkaline earth element. It was observed that exchange between alkalis and alkaline earth elements for AlO4 compensation are possible during glass leaching14. From the SEM images given in Fig. 5c, there is an observable remaining altered layer depleted in sodium on the outline of the altered grains observed in cross-section. The width of this layer is about 1 µm, in close agreement with the 0.88 ± 0.05 µm given by alteration solution analysis and deducted from the difference between ETh(Na) and ETh(Si) at 1096 days.Alteration of glass D sample was mainly congruent considering the uncertainties of our measurements and analyses. A stationary regime in which the rate of interdiffusion and the rate of hydrolysis are very similar is rapidly reached and maintained during most of the alteration duration. To be more specific, during the first 30 days of alteration Na leaching has the highest rate, quickly followed up by B, Al and Si, in accordance with scientific literature on the alteration of borosilicate compositions in acidic environment15. For a better visualization of the first 60 days of alteration a more detailed plot of this time lapse is given in Supplementary Fig. 3 in the Supplementary Information. After a few days all elements undergo a slight decrease of their alteration rate and B, Al and Si are leached at similar rates, their ETh being in the same range, considering the error bars associated to each value. These observations are in favor of a congruent dissolution of glass D after 200 days of alteration, which could lead, ultimately, to a total dissolution of the alteration layer. Nonetheless, from 322 days on, ETh(B) is consistently lower than the other elements. These last two notices are also shared by Stone et al. for similar B/(Al + Si) molar ratio at pH = 216 on shorter periods of alteration. Although reproducible these results are not fully explained yet. Regarding Na, its leaching rate seems slightly higher than the other elements, in favor of a small contribution of the ionic exchange/diffusion mechanism specific to network modifiers cations, while dissolution of the glassy network remains the dominant alteration process, all elements considered. As expected from the leaching mechanisms and calculated ETh, no residual alteration layer was observed, even with high resolution scanning and the distribution of elements sensed by SEM-EDX (Si, Al and Na) appeared homogeneous at the finest scale as shown in Fig. 5d.To follow-up with the comparison of alteration kinetics, two elements are common to the four studied glass compositions: silicon and sodium. Their leaching plots are displayed for better comparison in Fig. 6a, b respectively. On the one hand, ETh of Na, the most mobile species, are plotted in function of the square root of the duration of alteration to emphasis the compatibility of the mechanism of alteration of sodium with interdiffusion as defined in the second Fick’s law for all glasses. The measured rates of alteration of sodium after 56 days of alteration range from r(Na) = (98 ± 20) × 10−3 µm day−1 for glass A to r(Na) = (2 ± 0.6) × 10−3 µm day−1 for glass D, reflecting a great diversity of durability among the different glasses studied in function of their chemical composition and structure. In the composition of glass D, Na acts as charge compensator for [AlO4]− and [BO4]− units, leading to a better retention and thus a low rate of interdiffusion compared to glasses with an excess of alkali. In glasses with an excess of alkalis, these latter play the role of network modifiers causing the formation of NBO that are easily subjected to ion exchange with protons from the solution and then interdiffusion. On the other hand, the leaching of Si showed very similar rates of alteration for all the different glasses despite their varying silicon content (56.7 wt% for glass A and 79.7 wt% for glass D) and degrees of polymerization of the glassy network. Therefore it can be suggested that under acidic conditions, the rate of hydrolysis of these silicate glasses only depends, to first-order, on the experimental conditions: namely temperature and SA/V ratio as defined by Eq. (1) for a given pH of 2.4. Finally, it can be noticed that although the leaching behavior of this element continued to be governed by hydrolysis, the plots in Fig. 6a are not perfectly linear with respect to the duration of alteration.Fig. 6: Evolution of the equivalent thicknesses (ETh).Glasses A, B, C and D are represented for Si (a) and Na (b) in function of the square root of the duration (days) (read on the left Y-axis for glass A and on the right Y-axis for glasses B, C and D).A very minor contribution from other phenomena may be conceivable. The slight decrease of the leaching rate after 500 days of alteration could originate from the growing concentration of silicon in solution (about 50 ppm at the end of the 3 years of alteration) which could impact the hydrolysis rate in accordance with the thermodynamic affinity model17,18 even though the saturation of amorphous silica in solution (124 ppm under our experimental conditions) is not reached19. Alternatively, there could also be a very weak interdiffusion of silica units across the altered layer. Overall these hypothetical conjectures do not change the global rate of alteration imposed by hydrolysis reactions: r(Si) = (9.8 ± 0.5) × 10−4 µm day−1 at 70 °C and pH = 2.4. For glass D mainly and glasses B and C to a lesser extend, Si may appear as one of the most depleted elements, which may be considered against established models of glass alteration stating that Si is the least released element as the hydrolysis of the silicate network forces the release of the elements hosted in the so-called silicate network. The differences between Si and other elements are relatively thin in our case and the accuracy of the absolute values obtained from the data processing of the altered solution analyses as well as the sampling protocol could be accountable for these minimal incongruities.Looking at the structure of altered glass, for glasses B and D a very thin surface dealkalized layer is hydrolyzed over time. Therefore no remaining altered layer is found at the sample’s surface after 3 years of alteration, indicating that hydrolysis is the dominant mechanism of alteration for these glasses. Consequently, MAS NMR analyses conducted on the altered powder disclosed no significant difference with the pristine powders inasmuch as measurable by this technique. The results for 29Si and 23Na MAS NMR are given in Supplementary Fig. 4 in the Supplementary Information. Even in the case of glass C, for which a remaining altered layer depleted in Na was observed by SEM-EDX, overlapping MAS NMR spectra of pristine and altered glass were obtained for 23Na (Supplementary Fig. 4c) and 29Si (Supplementary Fig. 4d available in the Supplementary Information) in Supplementary Information) nuclei respectively. Glass A, adversely, featured a large remaining altered layer (about 15 µm) as well as local recondensation of the silicate network resulting from the alteration process. The MAS and 1H-29Si cross- polarization (cpMAS) NMR spectra collected on glass A powder altered for 969 days are given in Fig. 1, superimposed with the pristine glass 29Si MAS NMR spectrum. The cpMAS spectrum indicates interactions between 29Si and 1H detected in the altered glass, exhibiting the important proton exchange with the Q3 and Q2 units of the silicate network, which evidences the Si-OH species formation. Additionally, the comparison of MAS 29Si NMR spectra of the pristine and altered glass exhibits differing Q3/Q4 repartitions. In the pristine glass, Q3 are more abundant whereas in the altered glass Q4 are predominant, indicating that a repolymerization process of the silicate network occurring in the material during alteration. This repolymerization results from the condensation of silanols groups, creating bridging oxygen that are strengthening the silicate network as already observed by Angeli et al.7 and confirmed by Lecanuet et al.20 with a different study on the same lead crystal glass. In addition, to some extend repolymerization is favored by the acidic pH of the alteration solution21. In summary, for glasses C and A we clearly witnessed the formation of an alteration layer by interdiffusion depleted in alkaline elements of different widths depending on the chemical composition and glass structure with high retention properties towards earth-alkalis and lead. But only glass A demonstrated repolymerization evidences of the silicate network detected by 29Si MAS NMR on the altered powder, explaining the outstanding retention of Pb over very long periods of time. The repolymerization observed was made possible by the strong alkalis depletion in the altered layer for glass A in comparison with the other glasses as seen in Fig. 4. On the contrary, glass D alteration is dominated by hydrolysis, with no residual alteration layer due to the overall congruent dissolution of the glass. Glass B came out as intermediate exhibiting features of both mechanisms with a significant drop of alteration indicative of the formation of an alteration layer. Although, this layer was not observable by SEM-EDX after 3 years pointing out a weak rate of interdiffusion compared to glasses A and C, leading to the congruent dissolution of the altered layer over time. Furthermore the rate of hydrolysis of the silicate network appeared similar for all the studied glasses.Characterization of the altered glass slabsAfter 231 days of alteration, the glass slabs were removed from the reactors where they were altered alongside the powder that yielded the kinetics of alteration. ToF-SIMS and Spectroscopic Ellipsometry (SE) were performed on the smooth surface of these slabs to complement the data on the leached out elements from the solution’s analyses with data on the retained elements in the slab after alteration. ToF-SIMS profiles shown in Fig. 7 provided an instantaneous and detailed picture of the altered layers with a qualitative information on the elemental distribution in the materials and precise determination of the altered layers positions in a scanned area of 50 × 50 µm2. These local measurements were compared with the total alteration depth obtained by SE on a larger sample’s surface (about 0.5 cm2) and then put in balance with the solution analyses of each glass as presented in Table 2. At this stage, 231 days, glass A, was clearly in a stationary alteration rate regime characterized by the combination of a constant hydrolysis rate and a passivating altered layer. While, after 231 days of alteration, glasses B and C initiated their rate’s decrease, glass D is still leaching at a constant, but very low, rate dictated by the hydrolysis mechanism.Fig. 7: Tof-SIMS profiles obtained from the analysis of slabs’ smooth surfaces after 231 days of alteration at 70 °C and pH = 2.4.The ions displayed represent the species detected by the mass spectrometer after sputtering. The signals are normalized to the total intensity and to the pristine glass for B (b), C (c) and D (d) plots and to the profile of Si as well for glass A (a).Table 2 Comparison of the altered depths obtained on the glass slabs altered 231 days at 70 °C, pH 2.4, by ToF-SIMS, spectroscopic ellipsometry (SE), and ETh calculations based on ICP-AES analysis of the altered solutions for the same duration of alterationRegarding glass A, in Fig. 7a the penetration and integration of hydrogenous species into the glassy network are shown by the profiles of H+ and SiOH+ respectively, framing a total altered zone of 7110 ± 200 nm, as stated in Table 2. In this altered zone, two layers differing by their composition and thus properties are present. The most external layer and the thinnest one, about 200 ± 10 nm thick, visible on the close-up plot in the center of Fig. 7a is fully depleted in Pb and alkalis (Na, K), mainly formed of silicate species. This peculiar constitution is responsible for the passivating property of this alteration layer towards Pb as this element is then detected at its pristine glass level in the consecutive depths analyzed. The second layer corresponds to the rest of the alteration layer, which is constituted of almost 7 µm of dealkalinized hydrated glass and Pb is perfectly retained in this structure, forming Si–O–Pb units, as demonstrated from the 17O MAS NMR spectra of pristine and altered lead crystal glass samples with the same composition as glass A7. The SE measurement confirmed the altered depth obtained on a smaller area by ToF-SIMS with a very nice consistency. Regarding the agreement of solution analyses and solid characterizations, the ratio of ToF-SIMS and SE data for Na given in line 7 of Table 2 is moderately satisfying (61%) but better results are obtained considering K instead of Na. In fact, the low content of Na in this glass brings greater uncertainties on the analyses of Na and no difference between Na and K was noticed in the SEM-EDX images after 3 years of alteration nor in the depleted depths of both elements by ToF-SIMS suggesting that discrepancies between Na and K leaching are unfounded. Considering K, the following is obtained: ToFSIMS(K+)/(ETh(K) − ETh(Si)) = 0.89 which is more coherent and probably closer to the actual altered layer depth. All together, collected data were in the same vein describing the rapid and vast depletion of the alkali species (Na and K) and the formation of a passivating layer towards Pb through the repolymerization of the Si units observed by 29Si MAS NMR. Si repolymerization was also evidenced by Lecanuet et al.20, with tracing experiments using D218O and noticed a low content of D in the silica repolymerized subsurface region. This observation seems highly correlated with a distinct abundance of H+ denoting a local minimum in a ≈100 nm thick zone located 50 nm below the surface of the altered slab. This drop of H+ presumably results from the highly polymerized local network and indicates the position of the recondensation of silanols groups.The most external region of glass B profiles, between 0 and 35 nm, is mostly composed of network forming species Si and Al, also observed for the other glass samples of this study, which are mainly released by hydrolysis at a slower rate than the other elements, as verified on the powder with the leaching plot (Fig. 4c). Although essentially depleted from the first 50 nm, Ca and Ba are retained in the subsequent alteration layer to very similar extent. The most depleted element remained Na which is almost absent from the outer 160 nm of the altered glass. While very low on the ToF-SIMS profiles in the 0 to 35 nm zone, a minimal amount of Na has to remain in the Al rich region to compensate the [Al(O4)]− units22. High resolution 27Al MAS NMR spectrum from glass B after 969 days of alteration showed tetrahedral configuration of aluminum only, as presented in Fig. S5 in the Supplementary Information. Consequently, charge compensating elements, like Na, are necessary to maintain tetracoordinated Al units.Probably because of the low alteration depth and optical contrast, the SE and ToF-SIMS values for the total depth of alteration differ by about 120 nm as shown in line 4 of Table 2. But the width of the ToF-SIMS layer depleted in sodium nicely correlates with the solution data (89% correlation), giving credit for the ToF-SIMS measured depths. Although the analyses of the leaching solutions and SEM-EDX imaging disclosed the strong impact of the hydrolysis mechanism, ToF-SIMS data gives access to the finest traces of interdiffusion explaining the drop of the alteration rate by the formation of an external alkali and earth-alkali depleted layer.For glass C, ToF-SIMS recorded a total altered layer of 410 nm of which 390 nm is sodium depleted as seen in Fig. 7c and reported in Table 2. This observation validates the release of sodium by ion-exchange and interdiffusion through the already formed altered layer. alkaline earth elements (Mg, Ca and Ba) are well retained in the altered layer despite the hydration and depletion of sodium. Magnesium, which is only present in glass C, appeared depleted from the most external 25 nm, behaving in the same fashion as Ca and Ba, the other alkaline earth elements, although slightly more retained than these latter. In this case, SE and ToF-SIMS are in good agreement regarding the remaining altered layer depth after 231 days of alteration. Satisfying comparison with the solution data is obtained, nonetheless the solution analyses data depicts slightly higher depleted equivalent thicknesses of glass compared to ToF-SIMS and SE.With the lowest alteration kinetics, glass D ToF-SIMS profiles presented in Fig. 7d also displayed the smallest depths affected by alteration. The very weak depths of alteration and small depth differences between the most released (Na) and the most retained (Si) elements, about 20 nm at most, confirmed the congruent dissolution mechanism for glass D to the first- order. H+ and SiOH+ profiles are very noisy because of the settings chosen to analyze Na without saturation. Nonetheless, a highly hydrated zone corresponding to the depleted layer in sodium is noticeable between 0 and 40 nm followed by a poorly hydrated portion of glass until 120 nm depth. But in this case, a substantial difference between the leaching of Na and B is noticed on the ToF-SIMS profile as well as in the corresponding glass powder experiment shown in Fig. 4e. Although the difference expected from the solution analyses data was of 120 ± 40 nm has to be considered cautiously because of the experimental noise from the leaching data, this value remained quite far from the 30 nm recorded by ToF-SIMS. A 120 nm zone depleted in Na would not only match the leaching data but would also be consistent with the H+ profile showing a penetration of hydrogenous species in the same range. For this reason a second ToF-SIMS profile was conducted in a different area of the sample in case the first area chosen was not representative. The same results were obtained for both locations, as presented in Supplementary Fig. 6 in the Supplementary Information. This surprisingly low depletion of sodium from the slab might originate from the unwanted presence of K (K is not comprised in the nominal composition of glass D) in solution brought by the pH-meter probe, the concentrations of potassium measured in solution were around 1 ppm based on ICP-AES analyses. Detected by ToF-SIMS in the most external layer of 40 nm the existence of K at the surface of the slab might have played a role in hindering the release of Na. Under this assumption the ToF-SIMS altered layer thickness of H+ should be considered and compared to ETh(Si)-ETh(Na). In this case, a ratio of 0.75 is obtained which is in far better agreement than the ratio obtained for the ToF-SIMS profile of Na+ as measured and equivalent to what is obtained for C glass. Finally the SE measurement of the altered layer is relatively high, as mentioned in the last column of Table 2, but not consistent with other observations. Indeed the 2 local spots analyzed by ToF-SIMS (50 × 50 µm2 each) are in very good agreement with each other and there are no other indicators of a mili to micro-metric size heterogeneity on this glass sample.Regarding the altered layer open porosity, measurements were obtained by performing adsorption-desorption isotherms and the values obtained with the Effective Medium Approximation (EMA) remain relatively low, as displayed in the last line of Table 2 excluding local reaction pits taking place in the altered layer.
