1-s2.0-S0022309322005622-main

Journal of Non-Crystalline Solids 598 (2022) 121967
Contents lists available at ScienceDirect
Journal of Non-Crystalline Solids

journal homepage: www.elsevier.com/locate/jnoncrysol
Glass-forming ability and structural features of melt-quenched and

gel-derived SiO2-TiO2 glasses
Alessio Zandonà a, *, Erwan Chesneau a, Gundula Helsch b, Aurélien Canizarès a,
Joachim Deubener b, Valérie Montouillout a, Franck Fayon a, Mathieu Allix a
a
CNRS, CEMHTI UPR3079, Univ. Orléans, Orléans F-45071, France
b
Institute of Non-Metallic Materials, Clausthal University of Technology, Clausthal-Zellerfeld 38678, Germany
A R T I C L E I N F O A B S T R A C T
Keywords: SiO2-TiO2 glasses produced by aerodynamic levitation coupled to laser heating or by sol-gel spray-drying were
Glass-forming ability compared to highlight their structural differences. Glass formation was possible by melt-quenching up to 10 mol
Raman spectroscopy % TiO2, while higher contents led to devitrification. Raman spectroscopy and solid-state 17O and 29Si magic-
Solid-state NMR
angle-spinning nuclear magnetic resonance confirmed the clear emergence of Ti-O-Si bonds and a tetrahedral
Titania silica glasses
oxygen coordination of Ti4+ leading to full network connectivity, as also substantiated by the synthesis of TiO2-
Cristobalite
doped cristobalite. In gel-derived glasses, water content induced partial network depolymerization, thereby
enhancing the solubility of TiO2 in the hydrous silicate matrix. However, full dehydration during heating proved
challenging due to a competing tendency towards devitrification: the glass-forming range in the anhydrous bi
nary SiO2-TiO2 system does not therefore appear to be significantly enlarged by the sol-gel synthesis route.
1. Introduction [17], configurational changes are frequently associated with the pre
cipitation of TiO2-bearing crystals in a silicate melt [18,19]. Analo
The incorporation of Ti4+ in silicate melts and glasses has attracted gously, the glass forming ability of TiO2-containing melts has been
an unceasing research interest in both geoscience and materials science. previously evaluated in terms of the ratio between low-coordinated
TiO2 is indeed a frequent secondary component of terrestrial and lunar species and modifier-like sixfold coordinated Ti4+, such as in the
rocks and has been evaluated as a sensitive indicator of magma genesis SiO2-TiO2 compositional system [20].
and differentiation [1–4]. More recently, the precipitation of nanosized This latter system, extensively studied in the past due to the above-
Fe-Ti-oxides during decompression has been associated to the inter mentioned technological importance, appears as a promising candi
mittent explosive behavior of low-viscosity magmas [5]. Due to its date for elucidating the solubility and structural role of Ti4+ in a
strong tendency to segregate from the melt and crystallize, this oxide simplified amorphous silicate matrix. Nevertheless, literature sources
serves moreover as a widespread nucleating agent for the synthesis of spanning over at least five decades provided occasionally contrasting
glass-ceramics [6], while it represents a fundamental constituent of views on the structural features of these glasses (a short review is pro
SiO2-TiO2 glasses/glass-ceramics that are appreciated for their low posed in Section 1.1). This is partially related to their high melting
thermal expansion and catalytic activity [7–9]. temperature and strong tendency to crystallize [21–23], which make the
The Ti4+ ion possesses an amphoteric or intermediate character with classical melt-quench route unfavorable, leading to a preference for
respect to those of the classically defined network formers and modifiers other synthesis methods including flame hydrolysis and sol-gel
[10]. Its oxygen coordination in silicate melts exhibits a strong depen processing.
dence on temperature, pressure and especially composition; it is This work was therefore conceived as a fundamental clarification of
generally pictured as a combination of coexisting tetrahedral, analogies and dissimilarities between SiO2-TiO2 glasses obtained by
square-pyramidal and/or octahedral units [11–16]. Since melt-quenching (or equivalently by flame hydrolysis, involving a very
low-coordinated species are predominant in silicate glasses [12,15], high sintering temperature) and by sol-gel processing. Combining
whereas Ti4+ is typically six-fold coordinated in the crystalline state Raman spectroscopy, solid-state nuclear magnetic resonance (NMR) and
* Corresponding author.
E-mail address: [email protected] (A. Zandonà).
https://doi.org/10.1016/j.jnoncrysol.2022.121967
Received 18 May 2022; Received in revised form 1 October 2022; Accepted 7 October 2022
Available online 13 October 2022
0022-3093/© 2022 Elsevier B.V. All rights reserved.
A. Zandonà et al. Journal of Non-Crystalline Solids 598 (2022) 121967
molecular dynamics (MD) simulations, we succeeded in developing an coordination in phase-separated TiO2-rich clusters and crystals.
all-encompassing picture of the glass-forming region and structural
features of these materials. 2. Experimental
1.1. Review of relevant literature on SiO2-TiO2 amorphous materials 2.1. Sol-gel glasses
Studies based on conventional melting focused mostly on the Sol-gel glasses were produced as spray-dried nanobeads according to
determination of a phase diagram for the system (Fig. Sup1), involving a the method presented in a previous publication [9]. Two solutions were
eutectic point at ~8 mol% TiO2 and ~1550 ◦ C [21] and a prepared, homogenized separately and mixed together in the desired
high-temperature liquid immiscibility field between ~20 mol% and ratios shortly before being spray-dried. The first solution contained
~90 mol% [23–25]. Despite the high melting temperatures and the tetraethoxysilane (TEOS, 99.0% (GC), Fluka), isopropanol, deionized
strong devitrification tendency, some authors succeeded in character water as a hydrolysis agent (molar ratio TEOS/H2O = 0.25) and
izing SiO2-TiO2 glasses synthesized by electromelting, setting the limit concentrated nitric acid (69%, Fluka) to adjust the pH to 1. The second
of glass-forming ability at 10–12 mol% TiO2 [26,27] and invariably solution was obtained by mixing equimolar amounts of Ti-butoxide
assuming a tetrahedral or mixed tetrahedral-octahedral oxygen coordi (97.0%, Sigma-Aldrich) and ethyl acetoacetate (99.0%, Sigma-Aldrich)
nation for Ti4+ [26–29]. in isopropanol. Mixed solutions corresponding to samples with 5, 8,
Nevertheless, large-scale production of monolithic glass samples has 10 and 17 mol% TiO2, (respectively T5s, T8s, T10s and T17s in the
first been achieved by the flame hydrolysis method, in which a stoi following) were nebulized into a tube furnace set at 200 ◦ C using an
chiometric vapor mixture of suitable volatile precursors (such as SiCl4 aerosol atomizer (Atomizer, AGK 2000, Palas) operated with pressurized
and TiCl4) is oxidized in a CH4-O2 flame to obtain a fine glassy partic air (2.8 bar); the resulting nanobeads (diameter between 50 and 200
ulate that can be thereafter sintered at temperatures > 1700 ◦ C [7]. The nm) were collected from a particle filter at the other end of the furnace.
glass-forming region was macroscopically inferred to extend up to Higher TiO2 loadings (e.g. from 25 to 50 mol%) could be sprayed but
~12.5 mol% [7]; early Raman and IR spectroscopic characterizations proved unstable to the subsequent treatment at 600 ◦ C (SI section,
suggested again a tetrahedral or mixed tetrahedral-octahedral oxygen Fig. Sup2); they were therefore discarded from our further analyses. For
coordination of Ti4+ [30–32], in agreement with the reported solubility the production of a glass containing 8 mol% TiO2 and enriched in 17O,
of TiO2 in cristobalite up to ~8 mol% [33]. More direct investigations the same procedure was applied, using 17O-enriched water (enrichment
performed by X-ray absorption spectroscopy (XAS) at the Ti K-edge [34, degree: 40%, Cortecnet) for the hydrolysis of TEOS; the sample is noted
35] revealed Ti4+ to be almost solely 4-fold coordinated in these glasses as T8s(17O).
in the range 2–6 mol% TiO2, with an increasing amount of octahedral
species (still < 30% of the total) subsequently appearing up to 11.5 mol 2.2. Melt-quenched glasses
%, where the first crystalline TiO2 polymorphs were detected. At im
purity levels (TiO2 < 0.03 mol%), Ti4+ would instead exhibit Melt-quenched glasses were prepared by aerodynamic levitation
modifier-like 6-fold oxygen coordination. Later analyses by X-ray coupled to CO2 laser heating (ADL), using a setup described in a previous
photoelectron spectroscopy [36], Raman and IR spectroscopy [37,38] work [64]. SiO2 (Chempur, 99.9%) and TiO2 (Evonik, Aeroxide TiO2 P
and electron energy loss spectroscopy [39] did not substantially alter 25, 99.5%) powders were mixed thoroughly in an agate mortar and
these views. pressed into pellets of approximately 1 g. Small chunks of the pellets
The sol-gel route, typically based on the condensation of sols con weighing from 10 to 50 mg were then melted using O2 as a levitation
taining hydrolyzed alkoxide precursors, started to be evaluated exten gas, gradually increasing the laser power until reaching 1900–2100 ◦ C
sively in the 1980s as a possible alternative to the high melting (as recorded by two pyrometers). The levitated droplet was left at this
temperatures of SiO2-TiO2 mixtures [40–42]; the potential use of these temperature for a few seconds and then quenched by cutting the laser
materials as porous substrates for catalytic applications particularly power; the cooling rate is estimated to be in the order of 300 K s− 1 or
boosted scientific investigations [8]. Nevertheless, the choice of suitable more [64]. Using this method, several glassy beads containing 0, 0.5, 1,
synthesis parameters resulted crucial to avoid phase separation already 2, 4, 6, 8 and 10 mol% could be prepared; they were accordingly named
in the sols or in the gels [43–45,45,46]. The need for drying the gels at T0m, T0.5 m, T1m, T2m, T4m, T6m, T8m, T10m (-m for
relatively high temperatures to obtain fully interconnected glasses, melt-quenched). In the range T0m-T6m, macroscopically homogeneous
moreover, appeared in clear conflict with the low thermal stability of colorless beads could reproducibly be obtained, whereas TiO2-richer
these glasses [41,47,48]; notice that water could be still detected in compositions frequently exhibited opaque regions due to the crystalli
SiO2-TiO2 amorphous materials after annealing at 900 ◦ C [40] and a zation of anatase, as confirmed by Raman spectroscopy. This technique
similar H2O persistence was reported in gel-derived SiO2 glass [49–52]. was shown in previous works [9, 65–67] to be particularly effective in
All these factors made the structural investigation of gel-derived identifying phase separation and TiO2 crystalline polymorphs in silicate
SiO2-TiO2 glasses rather challenging. For instance, above ~6 mol% glasses, even in very small amounts (< 0.1 vol%). It was impossible to
TiO2, heterogeneity and TiO2 crystallization affected the samples reliably obtain glassy samples with TiO2 contents >10 mol%. All glassy
employed by Henderson and coworkers for their Raman and XAS studies beads were very rich in bubbles, due to the high viscosity of the mate
[15,41,53–55]. Despite this, they inferred a mixed 4-fold and 5-fold rials at the employed melting temperatures, which also resulted in oc
coordination for Ti4+ in the glass, possibly involving the formation of casionally poor chemical homogenization of the melts (confirmed by
oxygen triclusters; they associated 6-fold coordinated species to Ti-rich SEM). Nevertheless, lower viscosities could not be attained due to
clusters. Concurrently, other authors [56–63] adopted a multipronged physical limitations, since Si and Ti are increasingly volatile above
approach including solid-state NMR, X-ray and neutron diffraction and 2000 ◦ C [68]. An analytical strategy based on the analysis of a large
XAS to elucidate the structural evolution of SiO2-TiO2 gels during number of samples was therefore adopted to overcome the uncertainties
condensation and annealing. According to their ultimate results, four possibly arising from the occasional heterogeneity of the samples (see
possible environments are suggested for Ti4+ in these materials: )1) below).
distorted octahedral oxygen coordination, prevalently in gels treated In the attempt of synthesizing a melt-quenched glass enriched in 17O,
below 250 ◦ C and possibly corresponding to a tetrahedron with two the as-sprayed sol-gel sample T8s(17O) was placed in a tungsten crucible
further bonds to hydroxyl groups; more regular (2) tetrahedral and (3) and melted at 1800 ◦ C for 1 h in vacuum (to avoid exchanges with the
octahedral units, dispersed in the amorphous structure of gel-derived atmospheric 16O2), using a Setaram DTA/TGA SETSYS Evo 2400. The so-
glasses treated at relatively high temperature; (4) six-fold oxygen obtained material was partially crystallized; it is referred in the
2
following as T8m(17O). melt-quenched beads of each nominal composition were analyzed at

several locations, to avoid partially crystallized regions and average out
2.3. Heat treatments possible compositional inaccuracies deriving from occasionally poor
homogenization of the melts. In the following data treatment (Fig. 2), at
The as-prepared sol-gel materials required a further annealing stage least three spectra for each melt-quenched composition were analyzed,
to fully decompose the residues of their organic precursors, as also after a first selection among several candidates.
observed before [9]. A treatment at 450 ◦ C for 12 h in air resulted Data evaluation involved a simple baseline subtraction between 80
insufficient: a strong photoluminescence still prevented the acquisition and 1300 cm− 1 (Figs. 3 and 4 examine a much wider spectral range and
of Raman spectra with the selected laser line at 514 nm; heating at therefore report the raw spectra); the spectra were then normalized to
600 ◦ C for 12 h was found more suitable, although the glasses still their maximum intensity. To verify the appearance and growth of a
contained substantial amounts of water. To dehydrate them, several Raman band at ~685 cm− 1 (marked as T1 in the following), the spectra
drying protocols were then attempted in air using conventional labo shown in Fig. 1 were smoothed and differentiated in the range 620–770
ratory furnaces, as detailed in Fig. 3b: heating ramps were kept at 5 K cm− 1; the slope of the first derivative was used to mathematically infer
s− 1, testing maximum temperatures between 900 and 1200 ◦ C and the concavity variation in this spectral region. Interpretation of the
isothermal segments from 1 h to 30 h. Raman spectra of crystalline materials was facilitated by the online
In the case of sample T8s(17O), the pre-emptive heat treatment at consultation of the RRUFF database [69]: reference IDs are accordingly
600 ◦ C was performed in Ar using a Setaram DTA/TGA SETSYS Evo reported in the text.
2400, to avoid exchanges with 16O2 from the atmosphere; however, the
sample still exhibited photoluminescence, so that the as-sprayed powder 2.5. X-ray diffraction (XRD)
was subsequently treated also at 800 ◦ C for 12 h, to enable its charac
terization by Raman spectroscopy. Different decomposition kinetics of The samples were characterized by XRD using a D8 Advance Bruker
the organic precursors according to the employed atmosphere have been laboratory diffractometer (Bragg-Brentano geometry, Cu Kα1,2 incident
previously reported [40]. radiation, LynxEye XE line detector). The powders were placed on low-
TiO2-bearing cristobalite was synthesized according to the procedure background flat Si sample holders and dispersed with a few ethanol
described by previous authors [33]: sample T5s was placed in a hot droplets. Lattice parameters were computed by Le-Bail fits using the
furnace at 1450 ◦ C for 2 h and then quenched in air. For comparison, a software HighScore Plus (Panalytical).
TiO2-free cristobalite sample was similarly produced, treating our SiO2
raw material at 1600 ◦ C for 4 h. These samples are, respectively named 2.6. Transmission electron microscopy (TEM)
as T5crist and T0crist in the following treatment.
Sample T8m(17O) was examined in a Philips CM20 TEM operated at
2.4. Raman spectroscopy 200 kV, performing bright- and dark-field imaging, selected area elec
tron diffraction (SAED) and energy-dispersive X-ray spectroscopy (EDX)
Raman spectra were collected in parallel polarization using a measurements. The powder was dispersed in ethanol, after which a
Renishaw InVia Qontor spectrometer, mounting an Ar green laser (514 droplet was deposited and dried on a copper grid layered by an amor
nm) and a holographic grating of 1800 lines/mm. The nominal laser phous holey carbon film.
power was generally set at 50 mW, but samples containing high amounts
of TiO2 and/or crystalline TiO2 polymorphs required attenuation down 2.7. Molecular dynamics simulation and DFT GIPAW computation
to 1% of this power to avoid detector saturation. The measurements
were performed with 15 s integration time and 6 accumulations, Classical molecular dynamics (MD) model structures of T10m glass
focusing 30–40 μm below the surface of the melt-quenched beads or on were generated with rigid ion potentials using the DL_POLY 4 package,
the gel-derived nanosized powders with a 50x objective. Numerous whose details are reported elsewhere [70,71]. This composition has
Fig. 1. Raman spectra acquired from SiO2-TiO2 glasses obtained by different synthesis methods. (a) Melt-quenched samples; (b) gel-derived materials, treated at
600 ◦ C for 12 h in air, excluding T8s(17O) which was treated in Ar at 800 ◦ C for 12 h. The main characteristic bands for vitreous SiO2 are labelled as S1, S2, S3, S4, S5
and S6, while those related to the incorporation of Ti4+ are identified by T1, T2 and T3; vertical lines are kept at the same position in both figures to facilitate
the comparison.
3
been chosen to remain close to that of melt-quenched glasses while chemical shifts are referenced relative to water at 0 ppm. 17O-{1H}
increasing the number of statistically relevant titanium atoms in the cell. polarization transfer experiments allowing the selective observation of
Cubic boxes with 150 atoms were used and the edge lengths were T-O-H moieties were performed using a refocused-INEPT [81,82]
adjusted to match the experimental density [7]. We used a previously sequence with a SR421 recoupling block [83,84]. The RF field strengths
proposed force field [72] in which interactions between Si-O, Ti-O and for the π/2 and π pulses on the 17O and 1H channels were set to 20 and
O-O are formalized by a Morse potential. Initial random structures were 78 kHz, respectively, while a 1H RF field strength of 66.66 kHz corre
first equilibrated by a 100 ps NVT run at high temperature (3500 K), sponding to the 2ωR condition was used for the recoupling block.
followed by quenching at a rate of 2.5 K ps− 1 from 3500 K to 300 K. Six Recoupling delay was kept short (120 µs) in order to selectively probe
structures were created to increase the statistics. short-range O-H distances.
29
For all obtained structural models, the atomic positions were opti Si FID were processed through a homemade python code using
mized by DFT using the CASTEP package [73]. Computations of NMR nmrglue and numpy libraries [85,86]. Gaussian apodization centered at
chemical shielding and electric field gradient tensors were then per the top of the signal was applied prior to Fourier transform with line
formed using the Gauge Including Projector Augmented Waves (GIPAW) broadening of 25 Hz for 29Si and 17O. Simulations of 17O spectra were
[74] and the Projector Augmented Waves (PAW) [75] methods, done with Dmfit software [87] and 29Si ones were done using fNMR
respectively, as implemented in CASTEP. All computations were per package [88].
formed using the GGA PBE functional [76] and ultrasoft pseudopoten
tials (Materials Studio 7.0) generated “on-the-fly” [77]. Isotropic 3. Results
chemical shifts were calibrated based on series of computations per
formed on model crystalline systems of known structures, according to 3.1. Glass-forming ability and overall examination
the relationships [78]:
( ) ( ) The Raman spectra collected from melt-quenched materials (Fig. 1a)
δ 29 Si = − 0.921σ 29 Si + 289
confirmed the formation of glasses and the successful incorporation of
( 17 ) ( ) Ti4+ up to 10 mol% TiO2. Sample T0m exhibited the typical vibrational
δ O = − 0.946σ 17 O + 252
features of vitreous SiO2, labelled from S1 to S6. According to the
available literature, S1 and S4 (the latter one possibly involving a LO-TO
where δ is the isotropic chemical shift and σ the corresponding chemical
splitting) can be assigned to the vibrations of Si-O-Si linkages in the glass
shielding.
network [89–91]; S2 and S3 are also known as “defect lines”, respectively
NMR parameters were also computed for anatase and rutile TiO2
stemming from the breathing modes of four- and three-membered rings
models considering supercells of 384 and 576 atoms, respectively, and
of SiO4 tetrahedra [92,93]; S5 and S6 were previously related to the
using the same setting as for MD structures. Structural and GIPAW cal
asymmetric stretching of bridging oxygens linking SiO4 tetrahedra,
culations were analysed using a homemade python code. Atoms were
again with a LO-TO splitting [89–91] or possibly corresponding to two
clustered as a function of their speciation (i.e. coordination numbers,
different sets of stretching vibrations [94]. Upon TiO2 incorporation, the
atom-linked…), calculating property distributions for each speciation.
broad maximum S1 shifted to lower wavenumbers, while S2 and S3
Distributions were smoothed by applying a Gaussian broadening to each
gradually subsided. In parallel, three supplementary Raman features
point with broadness equal to the standard deviation of the dataset. The
became visible and increased in intensity, labelled as T1, T2 and T3. The
cut-off radius used to perform structural analyses was set to 2.2 Å for Si-
assignment of these bands is controversial in the available literature and
O and Ti-O, corresponding to the first minimum of the partial distribu
will be addressed in higher detail in the Discussion section. Neverthe
tion function. All structural analyses reported in the text below were
less, the position of the maximum of T2 (Fig. 2a), the concavity of the
performed on these DFT-optimized structures.
Raman spectra at T1 (i.e. the slope of their 1st derivative, Fig. 2b) and the
intensity ratio T2/T3 (Fig. 2c) exhibited an approximately linear corre
2.8. Solid-state nuclear magnetic resonance (NMR) lation with the TiO2 content of the melt-quenched glasses.
Gel-derived glasses treated at 600 ◦ C for 12 h in air (Fig. 1b)
Only reproducibly amorphous samples were characterized by solid- exhibited strong similarities with those of their melt-quenched coun
state NMR, crushing and grinding several beads; the resulting powder terparts; amorphous materials could be easily obtained up to 17 mol%
was filled into ZrO2 rotors. In addition, partially crystallized T8m(17O) TiO2. However, the computation of the above-introduced indices (Fig. 2)
was also measured, to support the interpretation of 17O Magic Angle highlighted a general shift of T2 to higher wavenumbers, a lower visi
Spinning (MAS) NMR spectra. 29Si MAS NMR spectra were acquired on a bility of T1 and a much lower intensity of T3 in comparison to melt-
Bruker Avance III spectrometer operating at a magnetic field of 7.0 T quenched glasses of similar composition; the defect bands S2 and S3
(29Si Larmor frequency of ν0=59.6 MHz) using a 4 mm probe. Spectra appeared moreover more pronounced (compare for instance T5s and
were recorded at a spinning rate of 10 kHz using a 30◦ flip angle (radio T6m). As for sample T8s(17O) (dried at 800 ◦ C for 12 in Ar), its spectrum
frequency field of 55 kHz). 29Si spectra were recorded for different matched well with that of T8s but exhibited a slight overall shift to lower
recycling times up to 10 min and we found that a recycling time of 60 s wavenumbers, in line with previous observations of the Raman spec
was long enough to avoid differential relaxation effects across the line trum of 18O-substituted SiO2 [90]. The noticed differences between
shape and ensure quantitative results. Between 1000 and 3000 tran melt-quenched and gel-derived glasses were ascribed to the
sients were collected for each spectrum. 29Si chemical shifts are refer non-negligible water content of these latter, as demonstrated by Raman
enced relative to tetramethylsilane (TMS) at 0 ppm. Because the T8m spectra collected over a wider range (Fig. 3). A broad band assignable to
(17O) sample was only available in limited amount, a Carr-Purcell- OH stretching vibrations was indeed clearly identifiable at ~3600 cm− 1
Meiboom-Gill (CPMG) echo-train acquisition [79,80] was used to [95,96], while it was absent in melt-quenched materials.
enhance the signal-to-noise ratio by summing 256 echoes equally spaced Samples T5s, T8s and T10s were subjected to several tentative heat
by a duration of 5.2 ms. treatments at temperatures between 900 ◦ C and 1200 ◦ C: interestingly,
17
O MAS NMR spectra were acquired on a Bruker Avance III spec partial dehydration of T5s (at 1000 ◦ C for 1 h) and of T8s (at 900 ◦ C for 2
trometer operating at a magnetic field of 17.6 T (17O Larmor frequency h) led their spectra to resemble more closely those of their melt-
of ν0=101.7 MHz) using a 2.5 mm probe. Quantitative spectra were quenched analogs, with a visible broadening of the defect bands S2
recorded at a spinning rate of 33.33 kHz using a small flip angle of 10◦ and S3 and an increase in the relative intensity of T3, mirrored by a
(corresponding to a pulse duration of 0.55 µs). Between 16,384 and variation in the indices plotted on Fig. 2. However, full dehydration
20,480 transients were co-added with a recycle delay of 0.5 s. 17O proved challenging to achieve before the onset of crystallization,
4
Fig. 2. Representative indices computed from the Raman spectra of melt-quenched and gel-derived glasses (T8s(17O) was disregarded due to the isotopic shift). (a)
Location of the maximum intensity for the Raman band labeled as T2; (b) slope of the 1st derivative of the Raman signal at the position of T1; (c) intensity ratio
between T2 and T3. Light blue circles represent the values obtained for partially dehydrated (and still amorphous) gel-derived samples, namely T5s treated at 1000 ◦ C
for 1 h and T8s treated at 900 ◦ C for 2 h (the spectra are reported in Fig. 3).
Fig. 3. Raman spectra collected over a wider range (80–4200 cm− 1) from some of the samples. (a) Melt-quenched glasses (as-prepared) and gel-derived glasses (pre-
treated at 600 ◦ C for 12 h), manifesting the presence of water in their structure; (b) gel-derived glasses treated at various temperatures for up to 30 h, invariably
exhibiting the onset of crystallization before full dehydration was completed.
marked by the appearance of sharper bands assignable to TiO2(B), [ICDD 01-072-1806]. The Raman spectrum revealed the appearance at
anatase [97], rutile and cristobalite (see references: [RRUFF-ID: low wavenumbers of intense vibrational features characteristic for
R050031] and [RRUFF-ID: X050046]). As for T17s, previous in Ti-bearing crystals, which however did not completely obliterate the
vestigations evidenced the emergence of TiO2(B) and anatase crystals spectrum of the glass, hinting at a very low crystalline volume fraction.
even only after 15 min at 800 ◦ C [9]. Since treatments below the melting Judging from its Raman spectrum, the residual glass was anhydrous and
point were invariably affected by partial crystallization, full dehydration still contained some TiO2 despite the partial crystallization, as
of gel-derived T8s(17O) was attempted by complete melting at 1800 ◦ C confirmed by the clear persistence of the bands labeled as T2 and T3 and
and subsequent quenching in vacuum (to avoid loss of 17O to the at the absence of vibrational features at ~3600 cm− 1.
mosphere). The characterization by XRD and TEM of the resulting T8m
(17O) revealed that it was still for the most part amorphous (Fig. 4); 3.2. Structural characterization
however, the crystallization of few highly Ti-enriched noduli (compo
sition qualitatively confirmed by EDX) could be detected, whose 3.2.1. Melt-quenched samples
diffraction pattern was indexable as an oxygen-interstitial solid solution Further investigations of the incorporation of Ti4+ in silica glasses
derived from metallic Ti [98], such as Ti6O [ICDD 01-072-1471] or Ti3O were conducted using 17O and 29Si solid-state NMR, which have been
5
Fig. 4. Experimental characterization of sample T8m(17O), obtained by melting gel-derived T8s(17O) at 1800 ◦ C in vacuum. (a) XRD pattern and Raman spectrum
(inset) of the sample, manifesting the formation of small amounts of Ti-suboxide solid solutions (indexable as Ti6O or Ti3O) in an otherwise fully anhydrous and
amorphous matrix (labels: t for Ti-suboxide solid solution, w for rests of the tungsten crucible). (b) Bright-field and (c) dark-field TEM micrographs of one of the Ti-
enriched crystalline noduli found in the sample, with an inset exemplifying the well-defined electron diffraction patterns obtainable from them.
shown to provide detailed information about chemical bonding and As shown in Fig. 5, the quantitative 17O MAS NMR spectrum of
local structures in a variety of glasses [99–104]. For this purpose, a sample T8m(17O) exhibits two well-resolved 17O central transition (CT)
melt-quenched 17O-enriched glass with nominal 8 mol% TiO2 was syn resonances and their associated satellite-transition (ST) spinning side
thesized as described above (Section 2.2). band manifolds, thereby revealing two distinct oxygen bonding
Fig. 5. 17O MAS NMR spectra of 17O-enriched glasses: (a,b) melt-quenched T8m(17O) and (c,d) gel-derived T8s(17O), recorded at 17.6 T with a spinning rate of 33
kHz. The 17O-1H MAS spectrum of T8s(17O) recorded using refocused-INEPT [81,82] polarization transfer with SR412 dipolar recoupling [83] is shown in black in (c,
d). Expansions of the − 100 to 150 and 150 to 1000 ppm frequency ranges are shown in (b,d). Experimental spectra and their simulations correspond to dark blue and
red lines, respectively. CT lines and ST spinning sidebands patterns of O_2Si, O_1Si-1Ti and O_1Si-1H contributions are shown as light blue, gray and black lines,
respectively.
6
environments in the glass structure. According to 17O NMR literature

data on SiO2-TiO2 materials [45,57,62], the intense line centered at
24.7 ppm and the broader one at 280 ppm can be assigned to bridging
O_2Si and O_1Ti-1Si environments, respectively. For the O_2Si moieties,
the 17O average isotropic chemical shift and quadrupolar coupling
constant (δiso = 41.3 ppm, CQ = 5.33 MHz) obtained from the CT line
shape simulation assuming Gaussian distributions of δiso and CQ values
are close to those reported for pure silica glass [105,106] (Table Sup1).
For the O_1Ti-1Si resonance showing both a featureless CT lineshape
and ST spinning sidebands of low intensities, the δiso and quadrupolar
parameter PQ were determined from the CT position measured at two
different fields of 17.6 and 9.4 T (δiso = 285 ppm, PQ = 2.97 MHz). One
can also notice that the <±3/2,±1/2> satellite transition sidebands are
much broader than those of the O_2Si resonance suggesting a larger
distribution of 17O isotropic chemical shift for the O-1Ti-1Si moieties.
Except for the ST spinning sidebands of O_2Si and O_1Ti-1Si, we note the
absence of signals in the 400–800 ppm chemical shift range associated to
oxygen atoms bonded to four, three or two titanium atoms [107]. The
relative proportions of O_2Si and O_1Ti-1Si environments obtained from Fig. 6. 17O isotropic chemical shift distributions obtained from GIPAW calcu
the fit of the spectrum (tacking into account all ST contributions) are lations for the MD structural models with 10 mol% TiO2. Average isotropic
93% and 7%, respectively. This amount of Si-O-Ti bridging bonds is chemical shifts (and standard deviations) are 47.1 (8.7), 312.1 (25.1) and 656.1
lower than expected (15%) from the nominal composition, assuming a (17.1) ppm for O_2Si, O_1Si_1Ti and O_2Ti environments in the MD models.
homogenous 17O-enrichment of all T-O-T linkages in a fully polymerized Results of GIPAW calculations for O_3Ti environments in anatase and rutile
random network of SiO4 and TiO4 tetrahedra. Since XRD and TEM TiO2 are also provided for comparison.
revealed the formation of pseudo-metallic Ti-rich noduli in the sample,
the observed lower proportion of O_1Ti-1Si contributions is likely due to Fig. 7b. Since 17O NMR results indicate the formation of Si-O-Ti bonds, it
a deviation from the starting composition of the glass, i.e. to a lower final shows that the 29Si isotropic chemical shift is almost unchanged by the
TiO2 content than the nominal. Nevertheless, this does not affect the substitution of Si4+ by Ti4+ in the second coordination sphere of Q4
overall interpretation of the 17O MAS spectra. units. This contrasts with the substitution of a silicon atom by a four-fold
To support the interpretation of the 17O NMR results, DFT GIPAW coordinated Al3+ or with the formation of a non-bridging oxygen atom,
computations of 17O chemical shifts were performed for structural which lead to chemical shift variations of 5 and 10 ppm, respectively
models with composition T10m. Six amorphous structures of 150 atoms [103]. The trend is confirmed by GIPAW computations of the MD
were generated by classical force field molecular dynamics (MD) sim amorphous models with 10 mol% of TiO2, which allow to benchmark
ulations and relaxed by DFT (at 0 K). Structural details of these models the 29Si chemical shift ranges of Si-(OSi)4 and Si-(OSi)3(OTi) moieties in
are given in Table Sup2 (see also .cif files and Fig. Sup4 in the supple a fully-polymerized network of SiO4 and TiO4 tetrahedra. Indeed, the
29
mentary materials). In all cases, MD simulations led to a fully poly Si chemical shift ranges calculated for these two environments are
merized tetrahedral network containing only SiO4 and TiO4 units, with found to overlap completely, with a difference of less than 0.7 ppm
no other higher-coordination polyhedral species (e.g. 5-fold or 6-fold between their mean chemical shift values (Fig. 7c) and in good agree
coordinated Ti4+, 5-fold coordinated Si4+). The proportion of Si-O-Ti ment with former computations of 29Si chemical shifts in Ti-bearing
and Ti-O-Ti linkages exhibits a slight variability but, on average over zeolites [124]. In line with 17O NMR results, the 29Si MAS spectra thus
the six structures, corresponds respectively to 18 and 1% of the oxygen suggest the formation of Si-O-Ti linkages with incorporation of titanium
content, as expected from composition and assuming a random distri atoms as TiO4 tetrahedral units in a fully polymerized network up to
bution of TiO4 units in the network. The average Ti-O bond length (1.83 ~8–10 mol% TiO2.
Å) is found to be larger than the Si-O one (1.63 Å), in agreement with
their ionic radii [108]; similarly, the average Si-O-Si bond angle 3.2.2. Sol-gel samples
(147.9◦ ) is smaller than the Si-O-Ti one (150.1◦ ). For the six amorphous As in the case of melt-quenched glasses, the nature of Ti4+ incorpo
structures, DFT GIPAW calculations lead to average 17O isotropic ration in gel-derived glasses was studied by 17O NMR. The 17O MAS
chemical shift values of about 47 ppm for O_2Si and 312 ppm for spectrum of the gel-derived glass T8s(17O) showed, in addition to the
O_1Si-1Ti environments (Fig. 6, Table Sup3), in agreement with the previously observed O_2Si and O_1Si-1Ti signals, a third contribution at
proposed assignment. Moreover, the computations predict an average lower chemical shift partially overlapping with the O_2Si CT peak
17
O chemical shift difference of 265 ppm between oxygen atoms (Fig. 5). This new resonance located at 4.0 ppm could be selectively
bridging two SiO4 and those bridging one SiO4 and one TiO4 tetrahedral highlighted using 1H-magnetization transfer experiments, revealing a
unit, with a larger distribution of 17O isotropic chemical shift for the short O-H distance and being therefore assigned to O_1Si-1H moieties in
latter. These results closely mimic the observed frequency difference agreement with previous works [109,110]. A small variation of the 17O
between the two 17O signals (255.3 ppm), as well as the observed larger isotropic chemical shift of bridging O_2Si (δiso = 43.1 ppm, CQ = 5.19
broadening of the O_1Si-1Ti <±3/2,±1/2> satellite transition side MHz) is also observed with respect to the melt-quenched glass. The
bands, strongly supporting the formation of a fully polymerized spectrum did not show any additional signal characteristic of Ti-O-Ti
silica-like network incorporating TiO4 tetrahedral units up to several bonding, suggesting a homogeneous distribution of Ti atoms. Quanti
mol% TiO2. tative analysis of the spectrum gives relative proportions of 89, 7 and 4%
The local structure of melt-quenched glasses up to 8 mol% TiO2 was for bridging O_2Si, O_1Si-1Ti and non-bridging O_1Si-1H oxygen atoms,
also investigated using 29Si solid-state NMR. As depicted in Fig. 7a, the respectively. The proportion of O_1Si-1Ti is lower than expected from
29
Si MAS spectra of melt-quenched samples were very similar, almost nominal composition, which could be associated to a preferential
17
independently of the titanium content. For the TiO2-free silica glass, the O-enrichment of Si-O-Si linkages since TEOS was hydrolyzed sepa
29
Si spectrum shows a broad resonance at – 110.0 ppm characteristic for rately with 17O-enriched water during synthesis.
Q4 units and both the position and the linewidth of the 29Si peak stay The 29Si MAS NMR spectra of the gel-derived samples calcined at
nearly constant when increasing the TiO2 content, as illustrated in 600 ◦ C (Fig. 8) showed obvious differences from those of melt-quenched
7
Fig. 7. (a) 29Si MAS NMR spectra of melt-quenched samples recorded at 7.0 T with a spinning rate of 10 kHz. Experimental spectra are shown in blue, while red
dashed lines correspond to the best fit of the T0m spectrum with a Gaussian lineshape. (b) Percentage of difference between the experimental spectra and the
simulated spectrum of T0m (red circle, left axis) and position of the center of gravity of each spectrum (blue square, right axis). (c) 29Si isotropic chemical shift
distributions of Si(OSi)4 and Si(OSi)3(OTi) environments obtained from GIPAW calculations for the MD models with 10 mol% TiO2. Average isotropic chemical shifts
(and standard deviations) are − 112.7 (6.9) and − 113.4 (5.6) ppm for Si(OSi)4 and Si(OSi)3(OTi), respectively.
glasses, with shoulders appearing at higher chemical shifts associated to synthesized TiO2-doped cristobalite (sample T5crist) according to the
depolymerized SiO4 units. Indeed, 1H-29Si cross-polarization MAS devitrification method reported by previous authors [33]. The lattice
spectra recorded at different contact times highlighted the presence of parameters obtained from Le-Bail refinements of XRD data (Fig. 9)
three distinct contributions located at − 109.2, − 100.5 and − 91.5 ppm confirmed an expansion of the unit cell of cristobalite upon incorpora
(see Fig. Sup3). According to 29Si chemical shift ranges, the peak at tion of Ti4+, due to its larger effective ionic radius (0.42 Å) compared to
− 109.2 ppm is assigned to both Si(OSi)4 and Si(OSi)3(OTi) Q4 units, Si4+ (0.26 Å) [108]; anatase and rutile were limited to trace levels in the
while the line at − 100.5 ppm is assigned to Si(OSi)3(OH) Q3 units and sample. The Raman spectrum of T5crist resembled closely that of
the later one of much weaker intensity to Si(OSi)2(OH)2 Q2 moieties. T0crist, excluding the bands assignable to an anatase impurity; never
Simulations of quantitative spectra were done using the parameters theless, two additional vibrational features could be discerned at high
extracted from CP MAS spectra and only the intensity of each component wavenumber, with maxima, respectively at 955 and 1115 cm− 1.
was varied. The position of the Q4 line is found very similar with that
observed for melt-quenched samples and the proportion of Q2 moieties 4. Discussion
bearing two hydroxyl groups does not exceed 1% for all compositions
heated at 600 ◦ C. A significant decrease of the relative proportion of Q3 The results of our study provide a self-consistent structural view on
units is observed as the titanium content increases. For the T8s(17O) the formation of glasses in the binary system SiO2-TiO2. Anhydrous
sample, the amount of non-bridging oxygen atoms determined from the glasses can be obtained by melt-quenching at least up to 10 mol% TiO2,
relative intensities of Qn resonances is about 10%. This corresponds to despite the higher difficulty in quenching a homogeneously amorphous
about twice the value obtained from the 17O spectrum (4%) suggesting a material as TiO2 content increases. This observation is consistent with
non-uniform 17O-enrichment of the different bonding environments in the available phase diagram [21,23], whose eutectic locates at ~8 mol
the T8s(17O) sample. Apart from manifesting OH-bearing structural %: beyond this point, the first phase predicted to form during cooling
units, the 29Si and 17O spectra of gel-derived glasses show strong simi through the liquidus is rutile, whose ability to precipitate homoge
larities with those of melt-quenched samples and remain consistent with neously in the volume can clearly affect glass-forming ability more than
the incorporation of four-fold coordinated Ti atoms in the silicate cristobalite, which is typically found at the surface. Moreover, the
network at low TiO2 content. threshold of 10 mol% TiO2 corresponds to the statistical emergence of
Ti-O-Ti bonds in a random network of tetrahedral units (the most
consistent structural description of SiO2-TiO2 glasses, see also below), as
3.3. TiO2-doped cristobalite
mirrored by the average values of our MD models in Table Sup2. In other
words, the statistical presence of Ti-O-Ti direct linkages in the melt
To support our discussion of SiO2-TiO2 glasses, we additionally
8
mol% TiO2, since samples with 25 mol% and 50 mol% already exhibited
crystalline features after the first calcination at 600 ◦ C (SI, Fig. Sup2).
Other authors reported the synthesis of gels of high TiO2 content and
similarly noted their high instability [40,41,44,48,56,57,59,111–113].
The predictable H2O content of our samples, which can be roughly
estimated as a few wt% based on our Raman [96] and 29Si NMR spectra,
persisted even after a pre-emptive drying stage at 600 ◦ C, so that these
materials should be consistently described as belonging to a ternary
compositional system, i.e. SiO2-TiO2-H2O. Obtaining dry glasses from
these hydrous gels proved then rather challenging, probably also due to
the lowering of viscosity and to the enhancement of nucleation usually
noticed in water-bearing silicate melts [114–116]. Our observations
therefore confirm the conclusions previously inferred from other
compositional systems [117]: the sol-gel route does not seem to signif
icantly enlarge the anhydrous glass-forming ability of reluctant
glass-formers, due to the inevitable competition between dehydration
and devitrification on the way to a fully densified glass. This scenario is
in line with the updated definition of the glassy state of matter and its
ultimate fate to crystallize [118], which disallow a spontaneous
glass-to-crystal transformation. As for the observed formation of Ti-rich
pseudo-metallic noduli in T8m(17O) during melting at 1800 ◦ C in vac
uum, it was clearly caused by the extreme reducing conditions, which
were previously shown to even possibly lead to volatilization of silicon
as gaseous SiO [119]. Notice that the non-detection of these
pseudo-metallic noduli on the NMR oxygen spectrum (expected in the
O_2Ti range) can be explained by their low volume fraction, their minor
oxygen content and also the short recycling delay used in the mea
surements, optimized for amorphous signals.
The computational prediction and experimental confirmation of the
absence of an observable and quantifiable effect of Ti-O-Si bonding on
the 29Si NMR spectra of amorphous materials qualifies again the char
Fig. 8. Experimental 29Si NMR spectra of sol-gel samples (blue) and their acterization of SiO2-TiO2 glasses as a challenging task. Yet, our data
simulations (dashed red). The individual Gaussian contributions associated to
obtained from melt-quenched glasses corroborate full polymerization of
Q4 ,Q3 and Q2 species are shown as dashed green, yellow and purple,
the silicate network (only Q4 species on the 29Si spectra) upon actual
respectively.
incorporation of Ti in their structure; Si-O-Ti bonds are moreover clearly
observable in the 17O spectrum of T8m(17O) (despite its partial crys
appears to be energetically unfavorable for glass formation (as sup
tallization). Relating these observations, Ti4+ must be isomorphously
ported by the energy differences between various relaxed MD models
integrated into the silicate network, as previously proposed by various
listed in Table Sup2), possibly representing the main structural driving
authors assuming a prevalent tetrahedral coordination for Ti4+ (at least
force for the onset of crystal nucleation.
at low TiO2 content) in anhydrous and homogeneous SiO2-TiO2 glasses
Concerning gel-derived glasses, we analyzed only materials up to 17
[27,28,34,35]. This conclusion is supported by the synthesis of
Fig. 9. Results of the analytical characterization of samples T0crist and T5crist, devitrified, respectively at 1600 ◦ C and 1450 ◦ C to form cristobalite. (a) XRD patterns
and related Le-Bail fits, detailing the lattice parameters obtained for cristobalite (filled areas: calculated intensity; broken line: background; gray line: fit residuals; a
and r label the most intense peak of anatase and rutile, respectively); (b) Raman spectra of the same samples (labels: a for anatase, * for Raman bands absent in the
reference spectrum of tetragonal cristobalite [RRUFF-ID: X050046], to which all unmarked features can otherwise be related).
9
Ti4+-containing cristobalite, in which the transition metal necessarily incorporate substantially higher amounts of TiO2, most likely due to the
substitutes Si4+ in the tetrahedral network sites. The Raman spectrum of depolymerizing effect of a non-negligible water content: this result
this phase also exhibits two supplementary bands at 955 cm− 1 and 1115 suggests that dehydration may be crucial in driving the crystallization of
cm− 1, whose position is close to that of T2 and T3 in glasses, clearly transition metal oxides in melts. All in all, the sol-gel route does not
supporting their assignment to the vibrational features of bridging ox appear a viable solution for enlarging the glass-forming range of anhy
ygens related to a 4-coordinated network-forming Ti4+; this assignment drous silicates.
agrees with the majority of literature sources [27,29,30,32,120].
Regarding the Raman band T1, its assignment is less straightforward: CRediT authorship contribution statement
given its vicinity to the position of bands arising in most of the known
TiO2 polymorphs in the range 600–670 cm− 1 [97,121,122], its inter Alessio Zandonà: Investigation, Conceptualization, Validation,
pretation as the signature for 6-coordinated Ti4+ species in the amor Investigation, Writing – original draft, Visualization, Funding acquisi
phous network is particularly tempting, as also claimed by previous tion. Erwan Chesneau: Investigation, Validation, Writing – original
authors [27]. This assignment would however imply the formation of draft, Visualization, Formal analysis. Gundula Helsch: Investigation,
6-fold Ti4+ already at very low TiO2 content (see the linear dependence Writing – review & editing. Aurélien Canizarès: Resources. Joachim
from composition observed in Fig. 2b), which is in clear antithesis with Deubener: Resources, Writing – review & editing. Valérie Montouill
our MD simulations, NMR data and previous XAS investigations exhib out: Validation, Writing – review & editing, Supervision. Franck Fayon:
iting no significant detection of 6-fold Ti4+ up to at least ~6 mol% TiO2 Investigation, Validation, Formal analysis, Writing – review & editing,
[34,35]. Consequently, the Raman band may indeed originate from a Resources, Supervision. Mathieu Allix: Resources, Writing – review &
6-fold coordinated Ti4+ content at impurity level, made detectable due editing, Supervision.
to the especially high Raman sensitivity of these species as compared to
other techniques (the high scattering power of TiO2 polymorphs is well Declaration of Competing Interest
known to Raman users); nevertheless, the vibrational feature could
equally stem from an additional mode related to tetrahedrally coordi The authors declare that they have no known competing financial
nated Ti4+, shifted from the typical positions observed in SiO2 (e.g. the interests or personal relationships that could have appeared to influence
band S4) or in GeO2 [89]. the work reported in this paper.
Based on our data, we also do not find significant reasons for
assigning Ti4+ to a 5-fold oxygen coordination, which was proposed to Data Availability
be the dominant configuration in gel-derived and partially crystallized
SiO2-TiO2 glasses investigated by other authors [55]. Indeed, MD sim Research Data associated with this article can be accessed on the
ulations produce structural models in which Ti4+ is invariably 4-fold repository “www.zenodo.org” using the following link: https://doi.org/
coordinated. In addition, the presence of TiO5 units is expected to lead 10.5281/zenodo.6513443.
to extra-coordinated (3-fold) oxygen atoms with higher chemical shift
values, which can be excluded from the analyses of the 17O NMR spectra.
For the same reason, pronounced clustering of octahedrally coordinated Acknowledgments
Ti4+ in the amorphous state seems also unlikely, since no signal is
detected around 600 ppm on 17O spectra. Moreover, pure TiO2 has A.Z. wishes to acknowledge the Deutsche Forschungsgemeinschaft
virtually no glass-forming ability and no other vibrational features are (DFG) for funding his research through the Walter Benjamin Program,
observed at lower wavenumber (100–300 cm− 1), where the most intense grant no. ZA 1188/1-1. Also J.D. and G.H. acknowledge the support of
bands of TiO2 crystals are typically found [97,121,122]. DFG, grant no. DE 598/33-1. The authors are additionally grateful to: (i)
NMR data obtained from H2O-bearing gel-derived glasses still Vinzent Olszok and Prof. Alfred P. Weber, for the assistance during the
revealed the presence of Q3 species (29Si spectra) and of Si-OH linkages synthesis of spray-dried nanobeads at the Institute of Particle Technol
(17O spectra), demonstrating an incomplete dehydration/densification ogy at TU Clausthal; (ii) the ICMN laboratory (Orléans, France) for TEM
of their network even after a treatment at 600 ◦ C for 12 h. The lower access and Dr. Cécile Genevois as an operator; (iii) the French Agency for
fraction of Q3 species noticed in TiO2-richer glasses could reflect the Research for its financial support through the Equipex Planex ANR-11-
expected lower viscosity of these materials, allowing higher structural EQPX-36; (iv) the TGIR-RMN-THC Fr3050 CNRS for its financial sup
mobility and faster H2O removal during the drying stages: a calorimetric port and (v) the CaSciModOT project (Calcul Scientifique et la Modéli
effect interpretable as a glass transition was observed in T17s at ~675 ◦ C sation à Orléans et Tours) for the access to high performance computing
[9], while the Tg of anhydrous T10m was previously estimated as facilities.
~1000 ◦ C [7] and that of anhydrous SiO2 as ~1200 ◦ C [123]. More
generally, water incorporation is likely to represent the crucial factor for Supplementary materials
the increased glass forming ability of hydrous SiO2-TiO2 glasses or,
differently formulated, for the higher solubility of Ti4+ in gel-derived Supplementary material associated with this article can be found, in
water-bearing silicate glasses: a partially depolymerized, less rigid the online version, at doi:10.1016/j.jnoncrysol.2022.121967.
network can probably more easily accommodate network distortions
due to the substitution of Si4+ for the substantially larger Ti4+, also References
avoiding energetically less favorable Ti-O-Ti direct linkages.
[1] F.J. Ryerson, E.B. Watson, Rutile saturation in magmas: implications for TiNbTa
depletion in island-arc basalts, Earth Planet. Sci. Lett. 86 (1987) 225–239,
5. Conclusion https://doi.org/10.1016/0012-821X(87)90223-8.
[2] G.A. Gaetani, P.D. Asimow, E.M. Stolper, A model for rutile saturation in silicate
We used a combination of Raman spectroscopy, molecular dynamics, melts with applications to eclogite partial melting in subduction zones and
mantle plumes, Earth Planet. Sci. Lett. 272 (2008) 720–729, https://doi.org/
GIPAW computations and solid-state 29Si and 17O NMR as a powerful
10.1016/j.epsl.2008.06.002.
multipronged approach for the structural characterization of glasses. [3] K. Kularatne, A. Audétat, Rutile solubility in hydrous rhyolite melts at 750–900◦ C
Taking into account all hypotheses proposed in the literature, we have and 2kbar, with application to titanium-in-quartz (TitaniQ) thermobarometry,
been able to confirm that titanium prevalently assumes a tetrahedral Geochim. Cosmochim. Acta 125 (2014) 196–209, https://doi.org/10.1016/j.
gca.2013.10.020.
oxygen coordination in anhydrous SiO2-TiO2 glasses and that the glass [4] F.P. Leitzke, R.O.C. Fonseca, J. Göttlicher, R. Steininger, S. Jahn, C. Prescher,
solubility of TiO2 is limited to ~10 mol%. Gel-derived glasses can M. Lagos, Ti K-edge XANES study on the coordination number and oxidation state
10
of Titanium in pyroxene, olivine, armalcolite, ilmenite, and silicate glass during [30] M.C. Tobin, T. Baak, Raman spectra of some low-expansion glasses, J. Opt. Soc.
mare basalt petrogenesis, Contrib. Mineral. Petrol. 173 (2018) 103, https://doi. Am. 58 (1968) 1459, https://doi.org/10.1364/JOSA.58.001459.
org/10.1007/s00410-018-1533-7. [31] C.F. Smith, R.A. Condrate, W.E. Votava, The difference infrared spectra of
[5] D. Di Genova, R.A. Brooker, H.M. Mader, J.W.E. Drewitt, A. Longo, J. Deubener, titanium-containing vitreous silica, Appl. Spectrosc. 29 (1975) 79–81, https://
D.R. Neuville, S. Fanara, O. Shebanova, S. Anzellini, F. Arzilli, E.C. Bamber, doi.org/10.1366/000370275774455437.
L. Hennet, G.La Spina, N. Miyajima, In situ observation of nanolite growth in [32] H.R. Chandrasekhar, M. Chandrasekhar, M.H. Manghnani, Phonons in TiO2–SiO2
volcanic melt: a driving force for explosive eruptions, Sci. Adv. 6 (2020), https:// glasses, J. Non Cryst. Solids 40 (1980) 567–575, https://doi.org/10.1016/0022-
doi.org/10.1126/sciadv.abb0413. 3093(80)90130-1.
[6] W. Höland, G.H. Beall, Principles of designing glass-ceramic formation. Glass- [33] D.L. Evans, Solid solution of TiO2 in SiO2, J. Am. Ceram. Soc. 53 (1970) 418,
Ceramic Technology, John Wiley & Sons, Ltd, 2019, pp. 1–66, https://doi.org/ https://doi.org/10.1111/j.1151-2916.1970.tb12146.x. –418.
10.1002/9781119423737.ch1. [34] D.R. Sandstrom, F.W. Lytle, P.S.P. Wei, R.B. Greegor, J. Wong, P. Schultz,
[7] P.C. Schultz, Binary titania-silica glasses containing 10 to 20 Wt% TiO2, J. Am. Coordination of Ti in TiO2–SiO2 glass by X-ray absorption spectroscopy, J. Non
Ceram. Soc. 59 (1976) 214–219, https://doi.org/10.1111/j.1151-2916.1976. Cryst. Solids 41 (1980) 201–207, https://doi.org/10.1016/0022-3093(80)90165-
tb10936.x. 9.
[8] A. Baiker, P. Dollenmeier, M. Glinski, A. Reller, Selective catalytic reduction of [35] R.B. Greegor, F.W. Lytle, D.R. Sandstrom, J. Wong, P. Schultz, Investigation of
nitric oxide with ammonia: II. monolayers of vanadia immobilized on TiO2-SiO2 glasses by X-ray absorption spectroscopy, J. Non Cryst. Solids 55
titania—silica mixed gels, Appl Catal 35 (1987) 365–380, https://doi.org/ (1983) 27–43.
10.1016/S0166-9834(00)82873-0. [36] S.M. Mukhopadhayay, S.H. Garofalini, Surface studies of TiO2–SiO2 glasses by X-
[9] A. Zandona, A. Martínez Arias, M. Gutbrod, G. Helsch, A.P. Weber, J. Deubener, ray photoelectron spectroscopy, J. Non Cryst. Solids 126 (1990) 202–208,
Spray-dried TiO2(B)-containing photocatalytic glass-ceramic nanobeads, Adv. https://doi.org/10.1016/0022-3093(90)90820-C.
Funct. Mater. (2020), 2007760, https://doi.org/10.1002/adfm.202007760. [37] A. Chmel, G.M. Eranosyan, A.A. Kharshak, Vibrational spectroscopic study of Ti-
[10] A. Dietzel, Die Kationenfeldstärken und ihre Beziehungen zu substituted SiO2, J. Non Cryst. Solids 146 (1992) 213–217, https://doi.org/
Entglasungsvorgängen, zur Verbindungsbildung und zu den Schmelzpunkten von 10.1016/S0022-3093(05)80493-4.
Silicaten, Z. Elektrochem. Angew. Phys. Chem. 48 (1942) 9–23, https://doi.org/ [38] S. Richter, D. Möncke, F. Zimmermann, E.I. Kamitsos, L. Wondraczek,
10.1002/bbpc.19420480104. A. Tünnermann, S. Nolte, Ultrashort pulse induced modifications in ULE - from
[11] D.B. Dingwell, E. Paris, F. Seifert, A. Mottana, C. Romano, X-ray absorption study nanograting formation to laser darkening, Opt. Mater. Express 5 (2015)
of Ti-bearing silicate glasses, Phys. Chem. Miner. 21 (1994), https://doi.org/ 1834–1850, https://doi.org/10.1364/OME.5.001834.
10.1007/BF00203924. [39] S.C. Cheng, Coordination and optical attenuation of TiO2–SiO2 glass by electron
[12] F. Farges, G.E. Brown, A. Navrotsky, H. Gan, J.J. Rehr, Coordination chemistry of energy loss spectroscopy, J. Non Cryst. Solids 354 (2008) 3735–3741, https://
Ti(IV) in silicate glasses and melts: II. Glasses at ambient temperature and doi.org/10.1016/j.jnoncrysol.2008.03.045.
pressure, Geochim. Cosmochim. Acta 60 (1996) 3039–3053, https://doi.org/ [40] C.J.R. Gonzalez-Oliver, P.F. James, H. Rawson, Silica and silica-titania glasses
10.1016/0016-7037(96)00145-7. prepared by the sol-gel process, J. Non Cryst. Solids 48 (1982) 129–152, https://
[13] F. Farges, G.E. Brown, A. Navrotsky, H. Gan, J.R. Rehr, Coordination chemistry of doi.org/10.1016/0022-3093(82)90251-4.
Ti(IV) in silicate glasses and melts: III. Glasses and melts from ambient to high [41] T. Hayashi, T. Yamada, H. Saito, Preparation of titania-silica glasses by the gel
temperatures, Geochim. Cosmochim. Acta 60 (1996) 3055–3065, https://doi. method, J. Mater. Sci. 18 (1983) 3137–3142, https://doi.org/10.1007/
org/10.1016/0016-7037(96)00146-9. BF00700798.
[14] C. Romano, E. Paris, B.T. Poe, G. Giuli, D.B. Dingwell, A. Mottana, Effect of [42] M. Emili, L. Incoccia, S. Mobilio, G. Fagherazzi, M. Guglielmi, Structural
aluminum on Ti-coordination in silicate glasses: a XANES study, Am. Mineral. 85 investigations of TiO2-SiO2 glassy and glass-ceramic materials prepared by the
(2000) 108–117, https://doi.org/10.2138/am-2000-0112. sol-gel method, J. Non Cryst. Solids 74 (1985) 129–146, https://doi.org/
[15] G.S. Henderson, X. Liu, M.E. Fleet, A Ti L-edge X-ray absorption study of Ti- 10.1016/0022-3093(85)90407-7.
silicate glasses, Phys. Chem. Miner. 29 (2002) 32–42, https://doi.org/10.1007/ [43] M. Schraml-Marth, K.L. Walther, A. Wokaun, B.E. Handy, A. Baiker, Porous silica
s002690100208. gels and TiO2/SiO2 mixed oxides prepared via the sol-gel process:
[16] E.T. Nienhuis, J. Marcial, T. Robine, C.Le Losq, D.R. Neuville, M.C. Stennett, N. characterization by spectroscopic techniques, J. Non Cryst. Solids 143 (1992)
C. Hyatt, J.S. McCloy, Effect of Ti4+ on the structure of nepheline (NaAlSiO4) 93–111, https://doi.org/10.1016/S0022-3093(05)80557-5.
glass, Geochim. Cosmochim. Acta 290 (2020) 333–351, https://doi.org/10.1016/ [44] M. Beghi, P. Chiurlo, L. Costa, M. Palladino, M.F. Pirini, Structural investigation
j.gca.2020.09.015. of the silica-titania gel/glass transition, J. Non Cryst. Solids 145 (1992) 175–179,
[17] F. Farges, G.E. Brown, J.J. Rehr, Coordination chemistry of Ti(IV) in silicate https://doi.org/10.1016/S0022-3093(05)80451-X.
glasses and melts: I. XAFS study of titanium coordination in oxide model [45] L. Delattre, F. Babonneau, 17O solution NMR characterization of the preparation
compounds, Geochim. Cosmochim. Acta 60 (1996) 3023–3038, https://doi.org/ of sol− gel derived SiO2/TiO2 and SiO2/ZrO2 Glasses, Chem. Mater. 9 (1997)
10.1016/0016-7037(96)00144-5. 2385–2394, https://doi.org/10.1021/cm970372f.
[18] F. Farges, G.E. Brown, Coordination chemistry of titanium (IV) in silicate glasses [46] C. Gervais, F. Babonneau, M.E. Smith, Detection, quantification, and magnetic
and melts: IV. XANES studies of synthetic and natural volcanic glasses and tektites field dependence of solid-state 17O NMR of X− O− Y (X,Y = Si,Ti) linkages:
at ambient temperature and pressure, Geochim. Cosmochim. Acta 61 (1997) implications for characterizing amorphous titania-silica− based materials, J. Phys.
1863–1870, https://doi.org/10.1016/S0016-7037(97)00050-1. Chem. B. 105 (2001) 1971–1977, https://doi.org/10.1021/jp003519q.
[19] L. Cormier, O. Dargaud, N. Menguy, G.S. Henderson, M. Guignard, N. Trcera, [47] D.S. Knight, C.G. Pantano, W.B. White, Raman spectra of gel-prepared titania-
B. Watts, Investigation of the role of nucleating agents in silica glasses, Mater. Lett. 8 (1989) 156–160, https://doi.org/10.1016/0167-
MgO− SiO2− Al2O3− SiO2− TiO2 glasses and glass-ceramics: a XANES study at the 577X(89)90182-1.
Ti K- and L2,3-edges, Cryst. Growth Des. 11 (2011) 311–319, https://doi.org/ [48] I.M.M. Salvado, J.M.F. Navarro, Phase separation in materials of the systems
10.1021/cg101318p. TiO2-SiO2 and ZrO2-SiO2 prepared by the alkoxide route, J. Mater. Sci. Lett. 9
[20] D.L. Evans, Glass structure: the bridge between the molten and crystalline states, (1990) 173–176, https://doi.org/10.1007/BF00727707.
J. Non Cryst. Solids 52 (1982) 115–128, https://doi.org/10.1016/0022-3093(82) [49] R. Puyané, P.F. James, H. Rawson, Preparation of silica and soda-silica glasses by
90285-X. the sol-gel process, J. Non Cryst. Solids 41 (1980) 105–115, https://doi.org/
[21] E.N. Bunting, Phase equilibria in the systems TiO2, TiO2-SiO2 and TiO2-Al2O3, 10.1016/0022-3093(80)90196-9.
Bur. Stan. J. Res. 11 (1933) 719, https://doi.org/10.6028/jres.011.049. [50] A. Bertoluzza, C. Fagnano, M. Antonietta Morelli, V. Gottardi, M. Guglielmi,
[22] R.W. Ricker, F.A. Hummel, Reactions in the system TiO2-SiO2; revision of the Raman and infrared spectra on silica gel evolving toward glass, J. Non Cryst.
phase diagram, J. Am. Ceram. Soc. 34 (1951) 271–279, https://doi.org/10.1111/ Solids 48 (1982) 117–128, https://doi.org/10.1016/0022-3093(82)90250-2.
j.1151-2916.1951.tb09129.x. [51] L.C. Klein, T.A. Gallo, G.J. Garvey, Densification of monolithic silica gels below
[23] R.C. DeVries, R. Roy, E.F. Osborn, Phase equilibria in the system CaO-TiO2–SiO2, 1000◦ C, J. Non Cryst. Solids 63 (1984) 23–33, https://doi.org/10.1016/0022-
J. Am. Ceram. Soc. 38 (1955) 158–171. 3093(84)90383-1.
[24] S.A. Kirillova, V.I. Almjashev, V.V. Gusarov, Phase relationships in the SiO2–TiO2 [52] T.A. Gallo, L.C. Klein, Apparent viscosity of sol-gel processed silica, J. Non Cryst.
system, Russ. J. Inorg. Chem. 56 (9) (2011) 1464–1471. Solids 82 (1986) 198–204, https://doi.org/10.1016/0022-3093(86)90131-6.
[25] C. Zhang, X. Ge, Q. Hu, F. Yang, P. Lai, C. Shi, W. Lu, J. Li, Atomic scale structural [53] G.S. Henderson, M.E. Fleet, The structure of Ti silicate glasses by micro-Raman
analysis of liquid immiscibility in binary silicate melt: a case of SiO2‒TiO2 spectroscopy, Can. Mineral. 33 (1995) 399–408.
system, J. Mater. Sci. Technol. 53 (2020) 53–60. [54] G.S. Henderson, M.E. Fleet, The structure of titanium silicate glasses investigated
[26] D.G. Ostrizhko, G.A. Pavlova, Structure of glasses in the silicon dioxide titanium by Si K-edge X-ray absorption spectroscopy, J. Non Cryst. Solids 211 (1997)
dioxide system, Izv. Akad. Nauk SSR, Neorg. Mat. 6 (1970) 74–77. 214–221, https://doi.org/10.1016/S0022-3093(96)00640-0.
[27] B.G. Varshal, V.N. Denisov, B.N. Mavrin, G.A. Parlova, V.B. Podobedov, Kh. [55] G.S. Henderson, X. Liu, M.E. Fleet, Titanium coordination in silicate glasses
E. Sterin, Spectra of Raman and hyper-Raman light scattering of the TiO2-SiO2 investigated using O K-edge X-ray absorption spectroscopy, Mineral. Mag. 67
glass system, Opt. Spektrosk. 47 (1979) 619–622. (2003) 597–607, https://doi.org/10.1180/0026461036740120.
[28] P.P. Bihuniak, R.A. Condrate, Structures, spectra and related properties of group [56] M.E. Smith, H.J. Whitfield, Application of oxygen-17 NMR spectroscopy to
IVB-doped vitreous silica, J. Non Cryst. Solids 44 (1981) 331–343, https://doi. detection of atomic scale phase separation in titania-silica gels, J. Chem. Soc.
org/10.1016/0022-3093(81)90036-3. Chem. Commun. 723 (1994), https://doi.org/10.1039/c39940000723.
[29] K. Kusabiraki, Infrared and raman spectra of vitreous silica and sodium silicates [57] P.J. Dirken, M.E. Smith, H.J. Whitfield, 17O and 29Si solid state NMR study of
containing titanium, J. Non Cryst. Solids 95–96 (1987) 411–417, https://doi.org/ atomic scale structure in sol-gel-prepared TiO2-SiO2 materials, J. Phys. Chem. 99
10.1016/S0022-3093(87)80138-2. (1995) 395–401, https://doi.org/10.1021/j100001a059.
11
[58] J.S. Rigden, J.K. Walters, P.J. Dirken, M.E. Smith, G. Bushnell-Wye, W.S. Howells, NMR, J. Magn. Reson. 186 (2007) 220–227, https://doi.org/10.1016/j.
R.J. Newport, The role of titanium in : mixed sol-gels: an x-ray and neutron jmr.2007.02.015.
diffraction study, J. Phys. Condens. Matter. 9 (1997) 4001–4016, https://doi.org/ [83] A. Brinkmann, A.P.M. Kentgens, Proton-selective 17O− H distance measurements
10.1088/0953-8984/9/20/001. in fast magic-angle-spinning solid-state NMR spectroscopy for the determination
[59] R. Anderson, G. Mountjoy, M.E. Smith, R.J. Newport, An EXAFS study of of hydrogen bond lengths, J. Am. Chem. Soc. 128 (2006) 14758–14759, https://
silica–titania sol–gels, J. Non Cryst. Solids 232–234 (1998) 72–79, https://doi. doi.org/10.1021/ja065415k.
org/10.1016/S0022-3093(98)00373-1. [84] R. Giovine, J. Trébosc, F. Pourpoint, O. Lafon, J.P. Amoureux, Magnetization
[60] G. Mountjoy, D.M. Pickup, G.W. Wallidge, R. Anderson, J.M. Cole, R.J. Newport, transfer from protons to quadrupolar nuclei in solid-state NMR using PRESTO or
M.E. Smith, XANES study of Ti coordination in heat-treated (TiO2)x(SiO2)1-x dipolar-mediated refocused INEPT methods, J. Magn. Res. 299 (2019) 109–123,
xerogels, Chem. Mater. 11 (1999) 1253–1258, https://doi.org/10.1021/ https://doi.org/10.1016/j.jmr.2018.12.016.
cm980644u. [85] J.J. Helmus, C.P. Jaroniec, Nmrglue: an open source Python package for the
[61] G. Mountjoy, D.M. Pickup, G.W. Wallidge, J.M. Cole, R.J. Newport, M.E. Smith, analysis of multidimensional NMR data, J. Biomol. NMR 55 (2013) 355–367,
In-situ high-temperature XANES observations of rapid and reversible changes in https://doi.org/10.1007/s10858-013-9718-x.
Ti coordination in titania–silica xerogels, Chem. Phys. Lett. 304 (1999) 150–154, [86] C.R. Harris, K.J. Millman, S.J. van der Walt, R. Gommers, P. Virtanen,
https://doi.org/10.1016/S0009-2614(99)00302-4. D. Cournapeau, E. Wieser, J. Taylor, S. Berg, N.J. Smith, R. Kern, M. Picus,
[62] D.M. Pickup, F.E. Sowrey, R.J. Newport, P.N. Gunawidjaja, K.O. Drake, M. S. Hoyer, M.H. van Kerkwijk, M. Brett, A. Haldane, J.F. del Río, M. Wiebe,
E. Smith, The structure of TiO2− SiO2 sol− gel glasses from neutron diffraction P. Peterson, P. Gérard-Marchant, K. Sheppard, T. Reddy, W. Weckesser,
with isotopic substitution of titanium and 17O and 49Ti solid-state NMR with H. Abbasi, C. Gohlke, T.E. Oliphant, Array programming with NumPy, Nature 585
isotopic enrichment, J. Phys. Chem. B 108 (2004) 10872–10880, https://doi.org/ (2020) 357–362, https://doi.org/10.1038/s41586-020-2649-2.
10.1021/jp049053j. [87] D. Massiot, F. Fayon, M. Capron, I. King, S.Le Calvé, B. Alonso, J.O. Durand,
[63] G.W. Wallidge, R. Anderson, G. Mountjoy, D.M. Pickup, P. Gunawidjaja, R. B. Bujoli, Z. Gan, G. Hoatson, Modelling one- and two-dimensional solid-state
J. Newport, M.E. Smith, Advanced physical characterisation of the structural NMR spectra, Magn. Res. Chem. 40 (2002) 70–76, https://doi.org/10.1002/
evolution of amorphous (TiO2)x(SiO2)1− xsol-gel materials, J. Mater. Sci. 39 mrc.984.
(2004) 6743–6755, https://doi.org/10.1023/B:JMSC.0000045604.02983.96. [88] T. Charpentier, C. Fermon, J. Virlet, Numerical and theoretical analysis of
[64] A. Zandonà, M. Moustrous, C. Genevois, E. Véron, A. Canizarès, M. Allix, Glass- multiquantum magic-angle spinning experiments, J. Chem. Phys. 109 (1998)
forming ability and ZrO2 saturation limits in the magnesium aluminosilicate 3116–3130, https://doi.org/10.1063/1.476903.
system, Ceram. Int. (2021), https://doi.org/10.1016/j.ceramint.2021.12.051. [89] F.L. Galeener, G. Lucovsky, Longitudinal optical vibrations in glasses: GeO2 and
[65] A. Zandona, C. Patzig, B. Rüdinger, O. Hochrein, J. Deubener, TiO2(B) SiO2, Phys. Rev. Lett. 37 (1976) 1474–1478, https://doi.org/10.1103/
nanocrystals in Ti-doped lithium aluminosilicate glasses, J. Non Cryst. Solids X 2 PhysRevLett.37.1474.
(2019), 100025, https://doi.org/10.1016/j.nocx.2019.100025. [90] F.L. Galeener, J.C. Mikkelsen, Vibrational dynamics in O18 -substituted vitreous
[66] D. Di Genova, A. Zandona, J. Deubener, Unravelling the effect of nano- SiO2, Phys. Rev. B 23 (1981) 5527–5530, https://doi.org/10.1103/
heterogeneity on the viscosity of silicate melts: implications for glass PhysRevB.23.5527.
manufacturing and volcanic eruptions, J. Non Cryst. Solids 545 (2020), 120248, [91] F.L. Galeener, A.E. Geissberger, Vibrational dynamics in 30Si-substituted vitreous
https://doi.org/10.1016/j.jnoncrysol.2020.120248. SiO2, Phys. Rev. B 27 (1983) 6199–6204.
[67] A. Zandonà, S. Ory, C. Genevois, E. Véron, A. Canizarès, M.J. Pitcher, M. Allix, [92] R.A. Barrio, F.L. Galeener, E. Martínez, R.J. Elliott, Regular ring dynamics in AX 2
Glass formation and devitrification behavior of alkali (Li, Na) aluminosilicate tetrahedral glasses, Phys. Rev. B 48 (1993) 15672–15689, https://doi.org/
melts containing TiO2, J. Non Cryst. Solids 582 (2022), 121448, https://doi.org/ 10.1103/PhysRevB.48.15672.
10.1016/j.jnoncrysol.2022.121448. [93] F.L. Galeener, Planar rings in glasses, Solid State Commun. 44 (1982) 1037–1040.
[68] Leo. Brewer, Thermodynamic properties of the oxides and their vaporization [94] P.F. Mcmillan, B.T. Poe, P. Gillet, B. Reynard, A study of SiO2 glass and
processes, Chem. Rev. 52 (1953) 1–75, https://doi.org/10.1021/cr60161a001. supercooled liquid to 1950K via high-temperature Raman spectroscopy, Geochim.
[69] B. Lafuente, R.T. Downs, H. Yang, N. Stone, 1. The power of databases: the RRUFF Cosmochim. Acta 58 (1994) 3653–3664.
project. Highlights in Mineralogical Crystallography, De Gruyter, 2015, pp. 1–30, [95] R.H. Stolen, G.E. Walrafen, Water and its relation to broken bond defects in fused
https://doi.org/10.1515/9783110417104-003. silica, J. Chem. Phys. 64 (1976) 2623–2631, https://doi.org/10.1063/1.432516.
[70] I.T. Todorov, W. Smith, K. Trachenko, M.T. Dove, DL_POLY_3: new dimensions in [96] H. Behrens, J. Roux, D.R. Neuville, M. Siemann, Quantification of dissolved H2O
molecular dynamics simulations via massive parallelism, J. Mater. Chem. 16 in silicate glasses using confocal microRaman spectroscopy, Chem. Geol. 229
(2006) 1911–1918, https://doi.org/10.1039/B517931A. (2006) 96–112, https://doi.org/10.1016/j.chemgeo.2006.01.014.
[71] K. Okhotnikov, B. Stevensson, M. Edén, New interatomic potential parameters for [97] T. Beuvier, M. Richard-Plouet, L. Brohan, Accurate methods for quantifying the
molecular dynamics simulations of rare-earth (RE = La, Y, Lu, Sc) aluminosilicate relative ratio of anatase and TiO 2 (B) nanoparticles, J. Phys. Chem. C 113 (2009)
glass structures: exploration of RE3+ field-strength effects, Phys. Chem. Chem. 13703–13706, https://doi.org/10.1021/jp903755p.
Phys. 15 (2013) 15041–15055, https://doi.org/10.1039/C3CP51726H. [98] I.I. Kornilov, V.V. Vavilova, L.E. Fykin, R.P. Ozerov, S.P. Solowiev, V.P. Smirnov,
[72] G. Malavasi, A. Pedone, M.C. Menziani, Towards a quantitative rationalization of Neutron diffraction investigation of ordered structures in the titanium-oxygen
multicomponent glass properties by means of molecular dynamics simulations, system, Metall. Trans. 1 (1970) 2569, https://doi.org/10.1007/BF03038386.
Null 32 (2006) 1045–1055, https://doi.org/10.1080/08927020600932793. [99] I. Farnan, P.J. Grandinetti, J.H. Baltisberger, J.F. Stebbins, U. Werner, M.
[73] S.J. Clark, M.D. Segall, C.J. Pickard, P.J. Hasnip, M.I.J. Probert, K. Refson, M. A. Eastman, A. Pines, Quantification of the disorder in network-modified silicate
C. Payne, First principles methods using CASTEP, Z. Krist. Cryst. Mater. 220 glasses, Nature 358 (1992) 31–35, https://doi.org/10.1038/358031a0.
(2005) 567–570, https://doi.org/10.1524/zkri.220.5.567.65075. [100] S. Wang, J.F. Stebbins, Multiple-quantum magic-angle spinning 17O NMR studies
[74] C.J. Pickard, F. Mauri, All-electron magnetic response with pseudopotentials: of borate, borosilicate, and boroaluminate glasses, J. Am. Ceram. Soc. 82 (1999)
NMR chemical shifts, Phys. Rev. B 63 (2001), 245101, https://doi.org/10.1103/ 1519–1528, https://doi.org/10.1111/j.1151-2916.1999.tb01950.x.
PhysRevB.63.245101. [101] F. Angeli, T. Charpentier, S. Gin, J.C. Petit, 17O 3Q-MAS NMR characterization of
[75] M. Profeta, F. Mauri, C.J. Pickard, Accurate first principles prediction of 17O a sodium aluminoborosilicate glass and its alteration gel, Chem. Phys. Lett. 341
NMR parameters in SiO2: assignment of the zeolite ferrierite spectrum, J. Am. (2001) 23–28, https://doi.org/10.1016/S0009-2614(01)00423-7.
Chem. Soc. 125 (2003) 541–548, https://doi.org/10.1021/ja027124r. [102] F. Angeli, T. Charpentier, M. Gaillard, P. Jollivet, Influence of zirconium on the
[76] J.P. Perdew, K. Burke, M. Ernzerhof, Generalized gradient approximation made structure of pristine and leached soda-lime borosilicate glasses: towards a
simple, Phys. Rev. Lett. 77 (1996) 3865–3868, https://doi.org/10.1103/ quantitative approach by 17O MQMAS NMR, J. Non Cryst. Solids 354 (2008)
PhysRevLett.77.3865. 3713–3722, https://doi.org/10.1016/j.jnoncrysol.2008.03.046.
[77] J.R. Yates, C.J. Pickard, F. Mauri, Calculation of NMR chemical shifts for [103] K.J.D. MacKenzie, M.E. Smith, Multinuclear Solid-State NMR of Inorganic
extended systems using ultrasoft pseudopotentials, Phys. Rev. B 76 (2007), Materials, eds., Pergamon, Amsterdam, 2002.
024401, https://doi.org/10.1103/PhysRevB.76.024401. [104] S.K. Lee, J.F. Stebbins, Nature of cation mixing and ordering in Na-Ca silicate
[78] S. Cadars, R. Guégan, M.N. Garaga, X. Bourrat, L.Le Forestier, F. Fayon, T. glasses and melts, J. Phys. Chem. B 107 (2003) 3141–3148, https://doi.org/
V. Huynh, T. Allier, Z. Nour, D. Massiot, New insights into the molecular 10.1021/jp027489y.
structures, compositions, and cation distributions in synthetic and natural [105] T.M. Clark, P.J. Grandinetti, P. Florian, J.F. Stebbins, Correlated structural
montmorillonite clays, Chem. Mater. 24 (2012) 4376–4389, https://doi.org/ distributions in silica glass, Phys. Rev. B 70 (2004), 064202, https://doi.org/
10.1021/cm302549k. 10.1103/PhysRevB.70.064202.
[79] J.S. Waugh, Sensitivity in fourier transform NMR spectroscopy of slowly relaxing [106] N.M. Trease, T.M. Clark, P.J. Grandinetti, J.F. Stebbins, S. Sen, Bond length-bond
systems, J. Mol. Spectrosc. 35 (1970) 298–305, https://doi.org/10.1016/0022- angle correlation in densified silica—results from 17O NMR spectroscopy,
2852(70)90205-5. J. Chem. Phys. 146 (2017), 184505, https://doi.org/10.1063/1.4983041.
[80] F.H. Larsen, I. Farnan, 29Si and 17O (Q)CPMG-MAS solid-state NMR experiments [107] E. Scolan, C. Magnenet, D. Massiot, C. Sanchez, Surface and bulk characterisation
as an optimum approach for half-integer nuclei having long T1 relaxation times, of titanium–oxo clusters and nanosized titania particles through 17O solid state
Chem. Phys. Lett. 357 (2002) 403–408, https://doi.org/10.1016/S0009-2614 NMR, J. Mater. Chem. 9 (1999) 2467–2474, https://doi.org/10.1039/A903714D.
(02)00520-1. [108] R.D. Shannon, Revised effective ionic radii and systematic studies of interatomic
[81] C.A. Fyfe, K.T. Mueller, H. Grondey, K.C. Wong-Moon, Dipolar dephasing distances in halides and chalcogenides, Acta Cryst. 32 (1976) 751–767, https://
between quadrupolar and spin-12 nuclei. REDOR and TEDOR NMR experiments doi.org/10.1107/S0567739476001551. Section A.
on VPI-5, Chem. Phys. Lett. 199 (1992) 198–204, https://doi.org/10.1016/0009- [109] X. Cong, R.J. Kirkpatrick, 17O MAS NMR investigation of the structure of calcium
2614(92)80069-N. silicate hydrate gel, J. Am. Ceram. Soc. 79 (1996) 1585–1592, https://doi.org/
[82] J. Trebosc, B. Hu, J.P. Amoureux, Z. Gan, Through-space R3-HETCOR 10.1111/j.1151-2916.1996.tb08768.x.
experiments between spin-1/2 and half-integer quadrupolar nuclei in solid-state
12
[110] X. Cong, R.J. Kirkpatrick, 29Si and 17O NMR investigation of the structure of [117] E.D. Zanotto, The formation of unusual glasses by sol-gel processing, J. Non Cryst.
some crystalline calcium silicate hydrates, Adv. Cem. Based Mater. 3 (1996) Solids 147–148 (1992) 820–823, https://doi.org/10.1016/S0022-3093(05)
133–143, https://doi.org/10.1016/S1065-7355(96)90045-0. 80723-9.
[111] M.F. Best, R.A. Condrate, A raman study of TiO2-SiO2 glasses prepared by sol-gel [118] E.D. Zanotto, J.C. Mauro, The glassy state of matter: its definition and ultimate
processes, J. Mater. Sci. Lett. 4 (1985) 994–998, https://doi.org/10.1007/ fate, J. Non Cryst. Solids 471 (2017) 490–495, https://doi.org/10.1016/j.
BF00721102. jnoncrysol.2017.05.019.
[112] K.L. Walther, A. Wokaun, B.E. Handy, A. Baiker, TiO2/SiO2 mixed oxide catalysts [119] J.F. White, J. Lee, O. Hessling, B. Glaser, Reactions between liquid CaO-SiO2 slags
prepared by sol—Gel techniques. Characterization by solid state CP/MAS and graphite substrates, Metall. Mater. Trans. B 48 (2017) 506–515, https://doi.
spectroscopy, J. Non Cryst. Solids 134 (1991) 47–57, https://doi.org/10.1016/ org/10.1007/s11663-016-0788-5.
0022-3093(91)90010-4. [120] V.O. Sokolov, V.G. Plotnichenko, E.M. Dianov, Quantum-chemical modeling of
[113] J.P. Rainho, J. Rocha, L.D. Carlos, R.M. Almeida, 29 Si nuclear-magnetic- titanium centers in titanosilicate glass, Inorg. Mater. 42 (2006) 1273–1288,
resonance and vibrational spectroscopy studies of SiO2 –TiO2 powders prepared https://doi.org/10.1134/S0020168506110173.
by the sol-gel process, J. Mater. Res. 16 (2001) 2369–2376, https://doi.org/ [121] G.A. Tompsett, G.A. Bowmaker, R.P. Cooney, J.B. Metson, K.A. Rodgers, J.
10.1557/JMR.2001.0325. M. Seakins, The Raman spectrum of brookite, TiO2 (Pbca, Z = 8), J. Raman
[114] M.J. Davis, P.D. Ihinger, A.C. Lasaga, Influence of water on nucleation kinetics in Spectrosc. 26 (1995) 57–62, https://doi.org/10.1002/jrs.1250260110.
silicate melt, J. Non Cryst. Solids 219 (1997) 62–69, https://doi.org/10.1016/ [122] H.L. Ma, J.Y. Yang, Y. Dai, Y.B. Zhang, B. Lu, G.H. Ma, Raman study of phase
S0022-3093(97)00252-4. transformation of TiO2 rutile single crystal irradiated by infrared femtosecond
[115] J. Deubener, R. Müller, H. Behrens, G. Heide, Water and the glass transition laser, Appl. Surf. Sci. (2007) 4.
temperature of silicate melts, J. Non Cryst. Solids 330 (2003) 268–273, https:// [123] G. Urbain, Y. Bottinga, P. Richet, Viscosity of liquid silica, silicates and alumino-
doi.org/10.1016/S0022-3093(03)00472-1. silicates, Geochim. Cosmochim. Acta 46 (1982) 1061–1072, https://doi.org/
[116] P. Del Gaudio, H. Behrens, J. Deubener, Viscosity and glass transition temperature 10.1016/0016-7037(82)90059-X.
of hydrous float glass, J. Non Cryst. Solids 353 (2007) 223–236, https://doi.org/ [124] G. Ricchiardi, J. Sauer, Influence of Ti substitution on the 29Si NMR spectra of
10.1016/j.jnoncrysol.2006.11.009. silicalite. A computational study, Zeitschrift für Physikalische Chemie 209 (1999)
21–32, https://doi.org/10.1524/zpch.1999.209.Part_1.021.
13

1-s2.0-S0022309322005622-main

Uploaded by

Copyright:

Available Formats

1-s2.0-S0022309322005622-main

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1-s2.0-S0022309322005622-main

Uploaded by

Copyright:

Available Formats

Journal of Non-Crystalline Solids 598 (2022) 121967

Contents lists available at ScienceDirect

Journal of Non-Crystalline Solids

Glass-forming ability and structural features of melt-quenched and

following as T8m(17O). melt-quenched beads of each nominal composition were analyzed at

environments in the glass structure. According to 17O NMR literature

You might also like