Short-range structure of invert glasses along the pseudo-binary join MgSiO3-Mg2SiO4: Results from29Si and25Mg MAS NMR spectroscopy

Article abstract of DOI:10.1021/jp9079603

The short-range structure of "invert" glasses along the pseudobinary join MgSiO₃-Mg₂SiO₄ has been studied using ²⁹Si and ²⁵Mg MAS NMR spectroscopy. The results indicate a progressive compositional evo

Full text of DOI:10.1021/jp9079603

J. Phys. Chem. B 2009, 113, 15243–15248
15243
Short-Range Structure of Invert Glasses along the Pseudo-Binary Join MgSiO -Mg SiO :
3
2
4
2
9
25
Results from Si and Mg MAS NMR Spectroscopy
,
†
‡
§
S. Sen,* H. Maekawa, and G. N. Papatheodorou
Department of Chemical Engineering & Materials Science, UniVersity of California at DaVis, DaVis, California
9
5616, Graduate School of Engineering, Tohoku UniVersity, Aoba-ku, Sendai, 980-8579, Japan, and Institute of
Chemical Engineering and High Temperature Chemical Processes FORTH, P.O. Box 1414,
GR-26504, Patras, Greece
ReceiVed: August 17, 2009; ReVised Manuscript ReceiVed: October 2, 2009
The short-range structure of “invert” glasses along the pseudobinary join MgSiO
3
-Mg
using Si and Mg MAS NMR spectroscopy. The results indicate a progressive compositional evolution in
Q speciation that approximately follows a statistical distribution. The Mg SiO glass shows an abrupt deviation
from this trend with the presence of nearly 40% of the Si atoms as (Si
2 4
SiO has been studied
2
9
25
2
4
6
-
1
2+
2
O
7
)
dimers, i.e., Q species. Mg
ions are present in predominantly octahedral coordination in all glasses. When taken together, these results
indicate that glasses with MgO contents between 50 and 60 mol % are characterized by a structure consisting
6
primarily of at least three types of Q species and MgO octahedra. On the other hand, the structure of glasses
0
1
with >60 mol % MgO appears to consist of Q and Q species with structural connectivity being primarily
provided by the MgO octahedra. The possible consequences of such compositional evolution of structure on
the ability of glass formation in this system are discussed.
6
0
I. Introduction
polyhedra and isolated SiO
4 2 4
units (Q species) near the Mg SiO
composition. The coordination number of Mg is suggested to
Knowledge of structure and structure-property relationships
in silicate glasses and melts is important in earth and materials
sciences and has important ramifications for processes as diverse
as manufacturing of glasses and glass-ceramics and volcanic
increase abruptly from ∼4.5 near the MgSiO
3
composition to
,7
∼
5.0 near the Mg
2
SiO
4
composition.6 On the other hand,
Raman spectroscopic results have indicated the presence of
bridging oxygens in all glasses and a more gradual evolution
of structure and Q speciation in these glasses as a function of
eruption.1
–5
Magnesium-rich silicates are one of the most
important constituents of the terrestrial and lunar mantles and
8
1
4+
composition. It may be noted that the structural and chemical
meteorites. On the other hand, melt-grown Cr -doped Mg
2 4
SiO
local order within the network in silicate glasses can be
characterized by the concentration of various types of Si
(
forsterite) crystals have shown significant potential for applica-
5
tion as laser materials. A number of spectroscopic and
diffraction studies of the structures of MgO-SiO binary glasses
have been reported in the literature over the past several
n
environments, typically denoted as Q species, where n varies
2
between 0 and 4 and denotes the number of bridging oxygens.
2
Further insight into the structure of invert MgO-SiO glasses
decades. The so-called “invert” glasses in the MgO-SiO
binary system with e50 mol % SiO have recently received
significant attention as the glass-forming region has been
extended down to the orthosilicate Mg SiO composition with
using containerless levitation techniques and
2
2
29
can be gained if one considers the results of recent Si and
2
2
5
Mg nuclear magnetic resonance (NMR) spectroscopic studies
1,13
of MgSiO
3
and Mg
2
SiO
4
glasses.1
Simulation of the static
2
4
29
Si NMR spectrum of Mg
2
SiO glass has shown the presence
4
3
CO
3.33 mol % SiO
2
6-
laser melting.6
–13
2 7
of nearly 40% of the Si atoms as (Si O ) dimers, consistent
2
Systematic structural studies of these
with the Raman spectroscopic results.11 The nominal composi-
invert glasses are particularly interesting, as they can provide
the clue to their glass-forming ability at such low concentrations
of the nominal network-former SiO . Unfortunately, the results
2
of the structural studies of these glasses have remained
contradictory and significantly controversial.
0
tion of this glass demands the presence of only Q species, and
hence, formation of dimers would release “free” oxygens that
are not bonded to any Si atom. It has been speculated that these
“free” oxygens will be coordinated to Mg, resulting in an abrupt
increase in Mg coordination number. This scenario is consistent
with the Mg coordination numbers obtained by Wilding and
co-workers on the basis of diffraction data, as mentioned above.
Detailed composition dependent structural evolution of invert
2
MgO-SiO glasses has been studied by neutron and X-ray
6,7
diffraction by Wilding and co-workers and more recently using
8
25
Raman spectroscopy by Papatheodorou and co-workers. The
On the other hand, Mg magic-angle-spinning (MAS) NMR
diffraction studies indicate that the structure of these glasses
and triple-quantum (3Q) MAS NMR spectra have been inter-
preted to be resulting from Mg being in octahedral coordination
evolves from a sparse network of bridged SiO
a mixture of MgO and MgO polyhedra near the MgSiO
composition to a percolation domain of MgO (x ) 4, 5, 6)
4
tetrahedra and
with oxygen in the MgSiO
3
glass with a range of distorted
geometries.1 A combined X-ray diffraction and reverse Monte
Carlo simulation study of the structure of Mg SiO glass has
estimated the relative fractions of MgO , MgO , and MgO
4
5
3
2,13
x
2
4
*
Corresponding author.
University of California at Davis.
Tohoku University.
4
5
6
†
polyhedra to be ∼30, 50, and 20%, respectively, resulting in
‡
§
10
an average Mg coordination number of ∼5. On the basis of
Institute of Chemical Engineering and High Temperature Chemical
Processes FORTH.
these results, Shimoda et al.12 have hypothesized that the Mg
1
0.1021/jp9079603 CCC: $40.75  2009 American Chemical Society
Published on Web 10/23/2009

1
5244 J. Phys. Chem. B, Vol. 113, No. 46, 2009
Sen et al.
coordination number actually decreases along the pseudobinary
MgSiO -Mg SiO join with increasing Mg concentration. Such
a decrease results from the requirement of the formation of
MgO tetrahedra as additional network formers besides the SiO
tetrahedra for the stabilization of the Mg SiO glass. It should
3
2
4
4
4
2
4
be noted that this structural hypothesis is in strong contrast to
the scenario proposed by Wilding and co-workers based on an
abrupt increase in the Mg coordination number near the Mg SiO
2 4
composition (Vide supra).
We report here the results of a systematic Si and Mg MAS
NMR spectroscopic study of nine glasses along the pseudobinary
29
25
3 2 4 2
MgSiO -Mg SiO join, prepared using levitation and CO laser
heating methods. The purpose of the present work is to
investigate the compositional evolution of the short-range
structure of these glasses to test the validity of the various
structural models proposed in the literature.
II. Experimental Section
Sample Synthesis. The starting chemicals for the preparation
of the glasses were reagent grade MgO (Alfa Aesar, 99.95%)
and SiO
were first dried by heating in a neutral atmosphere to up to ∼400
C and were then transferred into a dry glovebox. Powder
mixtures corresponding to the stoichiometries of MgSiO and
Mg SiO were first prepared in 5 g quantities. These mixtures
were thoroughly ground and pelletized and subsequently
partially sintered in a furnace at ∼1000 °C for a few days. These
pellets were then crushed again inside the glovebox and were
used as starting materials for preparing the powder mixtures of
intermediate compositions. Approximately 10-20 mg of the
powder mixtures were melted in a water-cooled copper hearth
2
(Alfa Aesar, 99.999% puratronic grade). The oxides
Figure 1. 29Si MAS NMR spectra of (MgO)
x 2
(SiO )100-x glasses.
Corresponding glass compositions are given alongside each spectrum
°
in terms of SiO content in mol %.
2
3
2
4
The 25Mg MAS NMR spectra of select glass samples were
acquired using a JNM-ECA930 (JEOL) spectrometer equipped
with a JEOL magnet (21.8 T) operating at a Larmor frequency
of 56.8 MHz for 25Mg. All MAS NMR spectra were acquired
using a single rf pulse of a duration of 1 µs (solids 45°) and a
3 4
recycle delay of 1 s. Samples were taken in Si N rotors and
were spun at 16 kHz in a 4 mm JEOL MQ/MAS probe.
Approximately 80 000 to 168 000 FIDs were averaged to obtain
2
using a 240 W continuous wave CO laser (Synrad, Evolution
Series) operating at a wavelength of 10.6 µm. Only a small
percentage (<15%) of the laser power was used to avoid
evaporation of the oxides and reaching temperatures far above
the liquidus. Polycrystalline, nearly spherical, white samples
were thus obtained after turning off the laser. These samples
were then levitated in Ar gas flow through a conical nozzle
25
each spectrum. Chemical shifts for all Mg spectra are
referenced in parts per million relative to 1 M MgCl
solution.
2
aqueous
III. Results
and heated and melted using the CO
2
laser. Finally, glasses were
The 29Si MAS NMR spectra for all glasses are shown in
obtained by rapid melt-quenching via sudden shut down of the
laser. Multiple spheres of ∼1 mm diameter were prepared this
way, amounting to a total of nearly 100 mg for each composi-
Figure 1. The fast sample spinning results in sidebands that are
far removed from the position of the main peak and negligible
sideband intensities (1% or less). These sidebands are not shown
in Figure 1 and have not been included in the line shape
3
tion, except the MgSiO composition for which nearly 300 mg
2
9
25
sample was made available, for Si and Mg MAS NMR
spectroscopy. The homogeneity and purity of such glasses
synthesized in a similar fashion had been confirmed in previous
works by others.6 The compositions of the glass spheres used
in this study were checked using atomic absorption spectroscopy
as well as by energy dispersive spectroscopy. The compositions
of all samples were found to be within (1 wt % of the nominal
composition.
29
simulation process. It is clear that the average Si MAS NMR
peak position in these spectra systematically shifts to higher
ppm values (i.e., gets deshielded) with increasing MgO con-
centration. This result is consistent with the progressive depo-
,9,10
2
lymerization of the silicate network and replacement of Q and
3
1
0
Q species with Q and Q species upon addition of MgO. The
29
Si MAS NMR line shape for Mg-silicate glasses is relatively
featureless and therefore difficult to simulate without a priori
knowledge of the peak positions and widths for the various Q
MAS NMR Spectroscopy. The 29Si MAS NMR spectra of
all glass samples were collected with a 4 mm Bruker MAS probe
and a Bruker Avance 500 spectrometer equipped with a wide
bore ultrashield magnet operating at a Larmor frequency of 99.3
MHz for 29Si. The samples in the form of spherical beads were
0
species. Fortunately, the isotropic chemical shifts δ for Q ,
iso
1
2
Q , and Q species in MgO-SiO binary glasses are well-known
2
from previous 29Si NMR studies of Mg-silicate glasses and
crystals and from ab initio chemical shift calculations to be
1
1,14–16
spun in ZrO
2
rotors at 10 kHz without further crushing. All
∼ -66, -75, and -82 ppm, respectively.
In this work,
one-pulse 29Si MAS spectra were collected using a radio
frequency (rf) pulse of duration 1.5 µs (corresponding to a 45°
tip angle) and with a recycle delay of 60 s. Approximately 3000
free induction decays (FIDs) were averaged to obtain each
29
Gaussian peaks have been used to represent the Si MAS NMR
line shapes of Q , Q , and Q species. The δ values for the
Q , Q , and Q species have been kept constrained to within
(1 ppm of the above-mentioned values, and the corresponding
peak widths ∆ (full width at half-maximum) were found to vary
within the narrow ranges of 7-9, 9-11, and 10-12 ppm,
0
1
2
iso
0
1
2
2
9
spectrum. Chemical shifts for all Si spectra are referenced in
parts per million relative to tetramethylsilane (TMS).

Short-Range Structure of Invert Glasses
J. Phys. Chem. B, Vol. 113, No. 46, 2009 15245
n
TABLE 1: Isotropic Shift (δiso) and Full Width at Half-Maximum (∆) for Different Q Species as Obtained from Simulation of
2
9
a
the Si MAS NMR Spectra of the (MgO)
x 2
(SiO )100-x Glasses
Q0
Q1
Q2
Q3
Q4
x
δiso (ppm)
∆ (ppm)
δ
iso (ppm)
∆ (ppm)
δ
iso (ppm)
∆ (ppm)
δ
iso (ppm)
∆ (ppm)
δ
iso (ppm)
∆ (ppm)
5
5
5
5
5
6
6
6
6
0
2
4
6
8
0
2
4
-74.9
-75.0
-75.8
-74.7
-74.5
-75.0
-75.0
-74.5
11.7
11.0
10.9
11.0
9.0
9.0
10.1
9.0
-83.9
-84.0
-84.1
-82.7
-82.0
-82.0
-82.5
12.6
11.7
11.1
10.0
10.1
10.0
10.0
-94.2
-94.2
-92.8
-92.2
13.2
13.0
10.0
13.0
-107.2
14.9
-65.5
-65.5
-65.5
-67.5
-67.5
-67.5
9.7
8.0
7.5
7.0
9.0
7.0
6.7
a
The error associated with both δiso and ∆ is (1 ppm.
TABLE 2: Relative Fraction (Percentages) of Different Qn
Species as Obtained from Simulation of the Si MAS NMR
respectively, for all glasses. The 29Si MAS NMR spectra of
29
glasses with e56% MgO required the fitting of an additional
Spectra of the (MgO)
x
(SiO
% Q1
2
)
100-x Glasses
3
peak centered at a δiso value of -92 ( 2 ppm for Q species
% Q0
% Q2
% Q3
% Q4
x
whose ∆ was found to vary between 12 and 14 ppm. This value
5
0
0.0
0.0
6.0
10.0
14.5
27.5
40.0
50.0
60.0
25.0
39.0
47.0
58.0
54.3
58.0
52.0
50.0
40.0
42.0
48.0
31.0
26.0
31.2
14.5
8.0
25.7
13.0
16.0
6.0
0.0
0.0
0.0
0.0
0.0
7.3
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
3
of δiso for Q species is in good agreement with those reported
52
54
56
58
6
6
6
6
3
in the literature for Q species in a wide range of crystalline
1
5,17
alkali and alkaline-earth silicates.
In addition, simulation of
2
9
the Si MAS NMR spectrum of the glass with 50% MgO
required consideration of the presence of a small fraction of Q4
species. The corresponding peak was found to be located at a
0
2
4
6.7
0.0
0.0
δ
iso value of -107.2 ppm, again consistent with similar values
4
reported in the literature for Q species in crystalline alkaline-
earth silicates.15 The simulation parameters for all 29Si MAS
NMR spectra are listed in Table 1, and some typical fits are
characterized by a broad, asymmetric line shape with a high-
field tail and a peak maximum located near ∼ -20 ppm. These
3
shown in Figure 2. The relatively high values of ∆ for the Q
25
Mg MAS NMR spectra are typical of a powder pattern for
2
0
1
and Q species compared to those for the Q and Q species
Table 1) are consistent with similar observations made by
central transition of a quadrupolar nuclide with second-order
quadrupolar broadening resulting from a distribution of qua-
drupolar parameters. The 25Mg MAS NMR spectrum of the
(
2
9
Larson et al. in a recent Si MAS NMR study of binary Li-
silicate glasses with high Li
O contents of up to 64.3 mol %.18
2
3
MgSiO glass in Figure 4 is consistent with that reported by
The relative fractions of various Q species have been obtained
by integrating the respective component line shapes and are
listed in Table 2 for all glass compositions, and their compo-
sitional variations are shown in Figure 3.
Shimoda et al. in earlier studies on isotopically enriched (99%
25
12,13
25
Mg) glasses.
The high-resolution Mg 3QMAS NMR
spectrum of this glass was interpreted by Shimoda et al. to
VI
represent Mg sites in 6-fold coordination with oxygen with at
The 25Mg MAS NMR spectra, collected for select glasses in
the series, are shown in Figure 4. These spectra are all
least four different types of such sites in the glass characterized
by δiso ranging between 1 and 24 ppm and quadrupolar product
Q
P ranging between 3.3 and 7.3 MHz, implying a wide range
VI
of site distortions. The δiso of Mg sites in silicates ranges
Figure 3. Compositional variation of relative fractions of different
Figure 2. Examples of typical fits of 29Si MAS NMR spectra of
Q species (symbols) as obtained from simulation of the Si MAS
NMR spectra of the (MgO) (SiO 100-x glasses. The black, pink, red,
blue, and green colored symbols correspond to Q , Q , Q , Q , and Q
n
29
x 2
(MgO) (SiO )100-x glasses. Corresponding glass compositions are given
x
2
)
4
3
2
1
0
alongside each spectrum in terms of SiO
2
content in mol %. Experi-
mental and simulated spectra are shown with solid and dotted lines,
respectively. Individual simulation components corresponding to dif-
ferent Q species are shown as dashed lines (Gaussian peaks).
species, respectively. Solid and dashed lines with corresponding colors
denote compositional variation of Q species according to the predic-
tions of binary and statistical distribution models, respectively.
n
n

1
5246 J. Phys. Chem. B, Vol. 113, No. 46, 2009
Sen et al.
n
of the Q Species
n
TABLE 3: Equilibrium Constants k
Disproportionation Reaction (See Text for Details) in
Different (MgO) (SiO )100-x Glasses
x 2
a
x
k
1
k
2
k
3
5
5
5
5
5
6
6
6
6
0
2
4
6
8
0
2
4
0.364
0.220
0.782
0.515
0.464
0.084
0.077
0.153
0.118
0.118
6.7
1
7
Figure 4. 25Mg MAS NMR spectra of (MgO)
CaSiO3
Li O-SiO
Na
0.105
0.156
0.106
x
2
(SiO )100-x glasses with
1
8,23
0.062*, 0.258* 0.30, 0.044*, 0.088* 0.08, 0.055*
5
0 (dotted line), 40 (dashed line), and 33.3 (solid line) mol % SiO
The topmost line shape shown with the dot-dashed line represents
simulation with a Gaussian distribution of C and η values. See text
2
.
2
2
2
3
2
O-SiO
O-SiO
2
0.14
0.02
0.06
0.01
0.02
0.01
2
3
K
2
2
Q
25
for simulation parameters. The line shape simulation program QuadFit
was used for this simulation.
a
Previously published values of
k
n
for other alkali and
alkaline-earth silicate glasses are included for comparison. Values of
with an asterisk are from ref 18.
k
n
between ∼5 and 15 ppm, while δiso is ∼50 ppm for 4-fold
IV
V
coordinated Mg sites. Although δiso for 5-fold coordinated Mg
n
n-1
+ Qn+1 (1 e n e 3). This disproportionation
2
Q ) Q
sites is not known, it is expected to be situated somewhere
between 15 and 50 ppm. Shimoda et al. has therefore suggested
reaction shifts to the right with increasing cooling rate of the
glass and with increasing field strength, i.e., the charge/radius
ratio, of the modifier cation. A more quantitative comparison
can be made if one considers the equilibrium constant of the
disproportionation reaction at the glass transition/fictive tem-
perature that can be written as k
, and k values for different Mg-silicate glass compositions
as determined in this study are listed in Table 3 and are
compared with the published values for alkali silicates, and
CaSiO
that one of the four Mg sites in MgSiO
3
glass, characterized by
22
V
12
the highest value of δiso ) 24 ppm, may in fact be a Mg site.
The 25Mg MAS NMR spectra shown in Figure 4 can be
simulated reasonably well with a second-order quadrupolar
powder pattern with a Gaussian distribution of quadrupolar
parameters. The best fit in all cases yields mean values of ∼17
ppm for δiso, ∼7.6 MHz for the quadrupolar coupling constant
n-1
n+1
n 2
n
1
) [Q ][Q ]/[Q ] . The k ,
k
2
3
C
Q
, and ∼0.35 for the asymmetry parameter η (Figure 4). The
17,18,23
3
glass.
It is clear that Mg-silicate glasses are
corresponding widths of the distributions for C and η are 3
Q
characterized by some of the highest values of these equilibrium
constants ever reported in the literature for binary silicate glasses.
These high k values and the significant deviation of the Q
n
MHz and 0.1, respectively. Although the δiso value of 17 ppm
VI
can be safely assigned to Mg sites in 6-fold coordination with
oxygen,19 as noted earlier by Shimoda et al., the possibility
12
species distribution in Mg-silicate glasses from that predicted
by a binary model is consistent with the high field strength of
V
of the presence of some fraction of Mg sites in these glasses
cannot be entirely discarded. It is also important to note the
2+
Mg and fast quench rates (i.e., high fictive temperatures)
2
5
strong similarity between the Mg MAS NMR line shapes in
Figure 4 that implies that the average coordination number of
Mg in these glasses remains unchanged irrespective of
composition.
employed in the syntheses of these glasses (Figure 3, Table 3).
Addition of each mole of MgO to SiO
mol of NBO according to the reaction MgO + Si-O-Si )
2 Si-Ø)· · ·Mg, where Ø represents an NBO. Therefore, the
2
is expected to create
2
(
total number of NBO atoms per Si (NBO/Si) is fixed by the
glass composition. The NBO/Si can also be calculated from
IV. Discussion
One extreme of the Q species distribution in silicate glasses
is defined by the binary distribution that assumes sequential
conversion of Q species: Q f Q with increasing concentra-
the Q speciation result as NBO/Si ) ∑x
n
n
(1 - n), where x is
n
the relative fraction of Q species. A comparison between the
expected NBO/Si from glass composition and that obtained from
the Q speciation measured by 29Si MAS NMR for these Mg-
silicate glasses is shown in Figure 5. The expected and measured
NBO/Si are in reasonably good agreement for all glass
n
n-1
tion of network modifiers. The binary distribution predicts the
n
presence of at the most two types of Q species, i.e., Q and
n-1
Q
, for any particular composition. The other extreme of Q
species distribution is based on a purely statistical model that
assumes a random modification of a fully connected Si-O
network via formation of nonbridging oxygens (NBOs) in the
2 4
compositions except for the Mg SiO glass where the expected
NBO/Si ) 4 is significantly larger compared to the experimental
value of 3.6 (Figure 5). Clearly, this deviation is a result of the
20–22
1
glass structure upon addition of network modifying cations.
presence of ∼40% of the Si as Q species in the end member
n
The relative concentrations of the Q species in glasses along
the pseudobinary join MgSiO -Mg SiO as determined in this
study (Figure 3) fall between these two extremes, in agreement
2 4
Mg SiO composition. The presence of such dimers in this
3
2
4
composition is consistent with previous reports of Q speciation
in this composition by us and others.8 The formation of Q
species in the Mg SiO glass implies that “free” oxygen ions
are released from disproportionation reactions that can be written
,11
1
2
9
with the results of previous Si MAS NMR studies of Q
speciation in a wide range of binary silicate glasses.1
However, the deviation of Q species distribution from a simple
binary model in these Mg-silicate glasses is unusually strong
and in fact glasses with <64% MgO show a closer cor-
respondence with the predictions of the statistical model (Figure
2
4
8,20,22
0
1
2-
as 2Q ) 2Q + O . These “free” oxygen ions are expected
to be charge balanced by and bonded to Mg2 ions in the
structure, as is indeed observed in the high-pressure polymorph
+
2
4
2 4
of the mineral forsterite (Mg SiO ), known as wadsleyite. In
3
). The deviation of Q speciation from the binary model can
the structure of wadsleyite characterized by tetrahedrally
coordinated Si present as Q species, the “free” oxygen ions
1
be expressed in terms of disproportionation reactions such as

Short-Range Structure of Invert Glasses
J. Phys. Chem. B, Vol. 113, No. 46, 2009 15247
V. Summary
29
The Si MAS NMR spectra of glasses along the pseudobinary
join MgSiO -Mg SiO indicate a smooth evolution of the Q
3
2
4
speciation with composition. The Q species distribution falls
between the predictions of binary and statistical models and
corresponds more closely with the statistical model. This result
is consistent with the high fictive temperatures of these glasses
and especially the high field strength of Mg2 cations. The
+
2 4
Mg SiO end member shows a large deviation in Q speciation
from the binary and statistical models and an anomalous
1
presence of a large fraction of Q dimer species. This result
implies the existence of “free oxygen” ions in the glass that are
bonded only to Mg2 ions, forming Mg-O-Mg bridges. Mg
+
25
2+
MAS NMR results indicate the presence of Mg ions pre-
dominantly in octahedral coordination in all glasses. The ability
of glass formation in this system is possibly dictated by the
presence of multiple Q species in glasses with <60 mol % MgO.
In contrast, the presence of a network of interconnected MgO
and SiO polyhedra allows glass formation in compositions with
higher MgO content. Furthermore, the “free” oxygens in the
Mg SiO glass provide connectivity to the MgO polyhedra via
Mg-O-Mg bridges that may also contribute to the glass
stability.
6
Figure 5. Comparison of the experimental (symbols) and calculated
4
(
from glass composition: solid line) number of nonbridging oxygens
per Si atom in (MgO) (SiO
octahedra. The 29Si and 25Mg MAS NMR
x
2
)
100-x glasses.
2
4
6
are shared by MgO
spectra presented here conclusively indicate that the structure
of Mg
6
0
1
2
SiO
4
glass is characterized by a mixture of Q and Q
octahedra, and therefore, the
Acknowledgment. The authors wish to thank Mr. N. K.
Nasikas (Dept. of Material Sciences, Univ. of Patras) for his
help with the CO melting/quenching experiments, Dr. V.
2
species (see Table 2) and MgO
6
nearest-neighbor bonding environments in this glass may look
locally similar to that of a mixture of wadsleyite and forsterite
phases.
When taken together, these results indicate a structural
scenario where the Mg-silicate glasses with g60 mol % MgO
can be stabilized by the formation of a network of interconnected
Drakopoulos (ICE/HT-FORTH) for the XRD and EDS mea-
surements, and Dr. M. Orkoula (Dept. of Pharmacy, Univ. of
Patras) for the chemical analysis by atomic absorption. The
authors also wish to thank Ms. Mariko Ando, Dr. Yasuto Noda,
and Mr. Itaru Oikawa (Graduate School of Engneering, Tohoku
University) for their help with the 25Mg NMR data acquisition.
We express our gratitude to M. Tansho and T. Shimizu of NIMS
for offering an opportunity to conduct the ultrahigh field NMR
experiments. The kind financial support to one of us (G.N.P.)
from the Executive Board of FORTH is acknowledged. S.S.
was supported by the National Science foundation grant
DMR0906070.
SiO
4
and MgO
6
polyhedra where the SiO
as Q and Q species. A simple bond-valence argument indicates
that in an interconnected network of SiO and MgO polyhedra
4
tetrahedra are present
0
1
4
6
each Si-NBO atom is expected to be shared by three Mg atoms,
as is indeed observed in the crystal structures of forsterite and
2
4
wadsleyite polymorphs of Mg
is different from that proposed by Kohara et al. for the
stabilization of the Mg SiO glass which was suggested to be
due to the formation of the percolation domain of interconnected
MgO (x ) 4, 5, 6) polyhedra where 4 and 5 coordinated Mg
2 4
SiO . This structural scenario
2
4
x
References and Notes
1
0
sites dominate the structure. On the other hand, Shimoda et
al. ascribed the glass formation ability of the invert glasses to
(
(
1) Helfrich, G. S.; Wood, B. J. Nature 2001, 412, 501.
2) Mysen, B.; Richet, P. Silicate Glasses and Melts: Properties and
the formation of a sparse tetrahedral (Si,Mg)O
4
network
Structure; Elsevier B.V.: Amsterdam, The Netherlands, 2005.
3) Rawson, H. Glasses and Their Applications; Royal Institute of
Metals: London, 1991.
4) Shelby, J. E. Introduction to Glass Science and Technology;
structure.1 However, the Mg MAS NMR results presented
here indicate that Mg is predominantly 6-fold coordinated in
all invert Mg-silicate glasses. Although the presence of some
fraction of 5-fold coordinated Mg sites in these glasses cannot
be ruled out, the possibility of the presence of any significant
fraction of tetrahedrally coordinated MgIV sites can be safely
2,13
25
(
(
Cambridge: RSC, 2005.
(5) Petricevic, V.; Seas, A.; Alfano, R. R. Opt. Lett. 1991, 16, 811.
(
6) Wilding, M. C.; Benmore, C. J.; Tangeman, J. A.; Sampath, S.
Chem. Geol. 2004, 213, 281.
7) Wilding, M. C.; Benmore, C. J.; Tangeman, J. A.; Sampath, S.
(
discarded. Therefore, the existence of a tetrahedral (Si,Mg)O
4
Europhys. Lett. 2004, 67, 212.
network is not tenable as a hypothesis for the structure of invert
Mg-silicate glasses. Finally, the strong similarity between the
Mg MAS NMR line shapes of invert Mg-silicate glasses over
the entire composition range studied here directly contradicts
the conclusion of Wilding and co-workers made on the basis
(8) Kalampounias, A. G.; Nasikas, N. K.; Papatheodorou, G. N.
J. Chem. Phys. 2009, 131, 114513.
(9) Tangemann, J. A.; Phillips, B. L.; Navrotsky, A.; Weber, J. K. R.;
Hixson, A. D.; Key, T. S. Geophys. Res. Lett. 2001, 28, 2517.
10) Kohara, S.; Suzuya, K.; Takeuchi, K.; Loong, C.-K.; Grimsditch,
M.; Weber, J. K. R.; Tangeman, J. A.; Key, T. S. Science 2004, 303, 1649.
11) Sen, S.; Tangeman, J. A. Am. Mineral. 2008, 93, 946.
2
5
(
(
of diffraction results that the average coordination number of
composition.6 Such
,7
(12) Shimoda, K.; Tobu, Y.; Hatakeyama, M.; Nemoto, T.; Saito, K.
Mg increases abruptly near the Mg
2
SiO
4
Am. Mineral. 2007, 92, 695.
discrepancy may result from the fact that these authors
determined the Mg-O coordination number by integrating to
the first minimum in the pair distribution function. Such a short
cutoff distance may result in incorrect coordination numbers in
the case of the presence of a mixture of short and long Mg-O
bonds with relative ratios that are dependent on composition.
(
13) Shimoda, K.; Nemoto, T.; Saito, K. J. Phys. Chem. B 2008, 112,
6947.
(
14) Gaudio, S. J.; Sen, S.; Lesher, C. E. Geochim. Cosmochim. Acta
008, 72, 1222.
15) Stebbins, J. F. In Mineral Physics and Crystallography: Handbook
of Physical Constants; Ahren, T. J., Ed.; AGU Reference Shelf: Washington,
2
(
DC, 1995; p 303.

1
5248 J. Phys. Chem. B, Vol. 113, No. 46, 2009
Sen et al.
(
16) Ashbrook, S. E.; Le Polle, L.; Pickard, C. J.; Berry, A. J.; Wimperis,
S.; Farnan, I. Phys. Chem. Chem. Phys. 2007, 9, 1587.
17) Zhang, P.; Grandinetti, P. J.; Stebbins, J. F. J. Phys. Chem. B 1997,
01, 4004.
18) Larson, C.; Doerr, J.; Affatigato, M.; Feller, S.; Holland, D.; Smith,
M. E. J. Phys.: Condens. Matter 2006, 18, 11323.
(21) Lacy, E. D. Phys. Chem. Glasses 1967, 8, 238.
(22) Greaves, G. N.; Sen, S. AdV. Phys. 2007, 56, 1.
(
(23) Maekawa, H.; Maekawa, T.; Kawamura, K.; Yokokawa, T. J. Non-
1
Cryst. Solids 1991, 127, 53–64.
(
(
(
24) Horiuchi, H.; Sawamoto, H. Am. Mineral. 1981, 66, 568.
25) Kemp, T. F.; Smith, M. E. Solid State NMR 2009, 35, 243.
(
(
19) Kroeker, S.; Stebbins, J. F. Am. Mineral. 2000, 85, 1459.
20) Eckert, H. Prog. Nucl. Magn. Reson. Spectrosc. 1992, 24, 159.
JP9079603