On the spectral similarity of bridging and nonbridging oxygen in tellurites

Article abstract of DOI:10.1021/jp052405h

We show by high field ¹⁷O solid-state nuclear magnetic resonance (NMR) and by ab initio calculations of both the NMR and the oxygen Is photoelectron spectra that the oxygen sites in tellurite glasses show no spectroscopic distinction, even when comparing bridging and nonbridging sites. This is remarkable because two such sites differ formally by a full electronic charge, and they are readily distinguished by these same methods in silicates. We argue that this similarity arises from the symmetry breaking that occurs when the original TeO₂ crystal solid forms, due to the pseudo-Jahn-Teller distortion induced by the two additional valence electrons present in Te ^IV as compared to Si^IV.

Full text of DOI:10.1021/jp052405h

7636
J. Phys. Chem. A 2005, 109, 7636-7641
On the Spectral Similarity of Bridging and Nonbridging Oxygen in Tellurites
R. T. Hart,† J. W. Zwanziger,*,‡ U. Werner-Zwanziger,‡ and J. R. Yates§
Intense Pulsed Neutron Source, Argonne National Laboratory, Argonne, Illinois, Department of Chemistry and
Institute for Research in Materials, Dalhousie UniVersity, Halifax, NS, Canada, and CaVendish Laboratory,
UniVersity of Cambridge, Cambridge, U.K.
ReceiVed: May 9, 2005; In Final Form: June 24, 2005
We show by high field 17O solid-state nuclear magnetic resonance (NMR) and by ab initio calculations of
both the NMR and the oxygen 1s photoelectron spectra that the oxygen sites in tellurite glasses show no
spectroscopic distinction, even when comparing bridging and nonbridging sites. This is remarkable because
two such sites differ formally by a full electronic charge, and they are readily distinguished by these same
methods in silicates. We argue that this similarity arises from the symmetry breaking that occurs when the
original TeO2 crystal solid forms, due to the pseudo-Jahn-Teller distortion induced by the two additional
valence electrons present in TeIV as compared to SiIV.
1. Introduction
not resolved in XPS.1,9 As in the case of silicates, one would
expect that 17O solid-state NMR should also show resolution
of the BO and NBO in tellurites, and indeed we found previously
that in crystalline TeO2 and Na2TeO3, the oxygens have rather
different NMR parameters.10 These cases are extreme, however,
in that TeO2 has only bridging oxygen,11 while Na2TeO3 is an
ionic solid consisting of isolated TeO32- anions coordinated by
Na+.12
We show in this paper that even at a relatively high field
(16.44 T), the oxygen sites in glassy sodium tellurites are not
resolved. This negative result is interesting in that it demon-
strates along with the XPS data that the BO and NBO in
tellurites are remarkably similar in electronic structure. We then
provide detailed ab initio calculations of both the XPS data and
the NMR data, to derive an understanding of the origin of this
similarity. Finally, we contrast this system to the better-known
silicate case.
Simple oxide glasses, especially silicates, contain two char-
acteristic oxygen bonding geometries: bridging oxygen (BO),
in which the oxygen is covalently bonded to two main group
atoms, and nonbridging oxygen (NBO), in which the oxygen is
bonded to a single main group atom and participates in ionic
interactions with modifiers. An example of BO would be -Si-
O-Si-, while a typical NBO is -Si-O-Na+. Such modifica-
tion is of great importance in glass technology, as a route to
control the thermal and mechanical properties of the resultant
glass. Since modification results in the replacement of strong
covalent bonds with weaker ionic interactions, the softening
point, glass transition temperature, and so forth can be greatly
reduced.
Because of the formal charge difference between bridging
and nonbridging oxygen, these two species are usually easy to
distinguish by a number of different methods, including X-ray
photoelectron spectroscopy (XPS) and nuclear magnetic reso-
nance spectroscopy (NMR).1-3 Furthermore, the bond length
difference between BO and NBO (about 1.65 vs 1.59 Å in
silicates) makes them resolvable by neutron diffraction.4
Telluria (TeO2) also gives rise to oxide solids, but the bonding
geometries are more complex than in the silicates.5,6 In
particular, in tellurite crystals nonbridging oxygen with a single
formal negative charge are not usually observed; instead, two
nonbridging oxygens sharing a single negative charge, or three
nonbridging oxygens sharing two negative charges, are fre-
quently seen.7 Nevertheless, as in the silicates these structures
do result from addition of an alkali oxide like Na2O, with the
concomitant cleavage of covalent Te-O bonds, so that telluria
is modifiable in much the same way as silica. The mechanical
consequences, such as significant lowering of the glass transiiton
temperature, are similar to those seen in silicates.8 The bridging
and nonbridging oxygen bond lengths are resolved by neutron
diffraction,5,6 but unlike the silicates, the BO’s and NBO’s are
2. Methods
2.1. Experimental Methods. 2.1.1. Sample Preparation.
17O-enriched R-tellurite (R-Te17O2) was synthesized from tel-
lurium isopropoxide (85%, Alfa Aesar) and water (99.9999%,
40% 17O, Cambridge Isotope Labs).10 When 2 equiv of water
were added to a stirring ethanol solution of Te(OiPr)4, Te17O2
was obtained instantly as a white precipitate. The reaction was
followed by recrystallization in air, accomplished by melting
the resulting white powder at 750 °C in a nonwetting ZGS Pt/
Au crucible and cooling at 2 °C/min. As TeO2 has very low
volatility at this temperature, no loss in enrichment was observed
due to the recrystallization step. After recrystallization, R-Te17O2
was obtained quantitatively as off-white needles. The phase
purity was confirmed by solid state 125Te MAS NMR, powder
X-ray diffraction, and Raman spectroscopy.
Glasses were prepared by melting R-Te17O2 together with
Na2CO3 in a Pt crucible at 800 °C for 10 min and quenching in
ice-water. The resulting materials were amorphous by X-ray
diffraction and showed glass transition temperatures by dif-
ferential scanning calorimetry in accordance with literature
values.8
* To whom correspondence should be addressed. E-mail: jzwanzig@dal.ca.
† Argonne National Laboratory.
‡ Dalhousie University.
§ University of Cambridge.
10.1021/jp052405h CCC: $30.25 © 2005 American Chemical Society
Published on Web 07/28/2005

Bridging and Nonbridging Oxygen in Tellurites
J. Phys. Chem. A, Vol. 109, No. 33, 2005 7637
2.1.2. NMR Spectroscopy. Spectra at 9.4 T (54.5 MHz 17
O
space of about 0.05 Å-1). For Na2Si2O5 this approach amounted
to a 4 × 2 × 4 Monkhorst-Pack grid, and for Na2Te4O9, a 3 ×
2 × 3 Monkhorst-Pack grid.25 The other samples have more
complex primitive unit cells and so have more complex
descriptions. All grids used were offset from the origin by 0.5
units in reciprocal space. Such optimal grids are computed by
the ABINIT code. The energies were computed using the
experimentally determined crystal structures.
XPS results were modeled using a thermodynamic cycle
approach.26,27 In this approach, it is noted first that extraction
of a core electron results in a solid with a core-ionized atom at
a specific site. The energy required to extract the electron is
the quantity observed in the XPS measurement. On the other
hand, the cohesive energy of the resulting defective solid relative
to free atoms can be computed by the planewave pseudopotential
method. This is done by representing the defective atom with a
pseudopotential constructed from an atom with a core hole. In
a second calculation, the defect is placed at a chemically distinct
site. The free atom reference state is the same in both cases.
Furthermore, in both cases the original, defect-free solid as
probed by XPS is the same. Therefore, one has
resonance frequency were acquired on a Bruker DSX spec-
trometer, under magic angle spinning conditions (12.0 kHz). A
Hahn echo sequence was used, with 4 and 8 µs times for the
90 and 180° pulses, respectively. A 500 s recycle delay was
used, and each spectrum consists of 2048 scans.
17O NMR spectra at 16.4 T (17O Larmor frequency of 94.9
MHz) were acquired on a Bruker Avance NMR spectrometer,
using 2.5 mm rotors and magic angle spinning at 30.0 kHz.
We employed a 2-dimensional multiple rotor-assisted population
transfer (RAPT) pulse sequence with Hahn echo detection.13
The RAPT transfer was repeated 21 times followed by acquisi-
tion of the free induction decay. 64 blocks of these spectra were
acquired in rapid succession and saved as a 2-dimensional data
set. A total of 724 scans (10% and 30% Na2O) or 1272 scans
(20% Na2O), respectively, were accumulated with a repetition
delay of 350 s. The spectra shown below are center slices after
Fourier transformation in both dimensions.
2.2. Computational Methods. 2.2.1. Molecular Modeling.
Modeling of molecular TeF4 and SiF4 was carried out with the
Amsterdam density functional code (ADF).14-16 Because the
object of this part of the study was only to develop a qualitative
Walsh diagram relating molecular shape to orbital energies, a
small basis set, double-ú STO, was used, together with frozen
core orbitals. Generalized gradient corrections to the local
density approximation as proposed by Becke,17 for exchange,
and Perdew,18 for correlation, were employed.
EB + EC ) E′ + E′
(1)
B
C
where EB is the binding energy of the core electron at the first
site and EC the cohesive energy of the resulting defective solid,
and the primed quantities refer to the defect appearing at the
second site. The shift between the binding energies can then be
computed as
2.2.2. Modeling Solids, XPS Data, and NMR Spectra.
Results relating to simulation of the XPS data were obtained
through use of the ABINIT code, which is based on pseudo-
potentials and planewaves.19 This is a density functional theory
approach to finding the energies of periodic solids, in which
the valence electron wave functions are expanded in terms of
planewaves, and the effect of the core electrons are ap-
proximated through the use of pseudopotentials. Pseudopoten-
tials were generated with the FHI98PP code.20 For the cases
discussed below, the pseudopotentials used were of Troullier-
Martins type,21 in separable form. For all atomic species studied
(Te, O, Na, Si), we used a maximum angular momentum value
of 2 for generation of the pseudopotentials, and also l ) 2 for
the local potential. For all elements, only the outer s and p
electrons were treated as valence electrons; in addition, for Na,
nonlinear core corrections22 were used to account for the core-
valence overlap from the relatively diffuse 2p orbitals.
EB - E′ ) -(EC - E′ )
(2)
B
C
Thus, the chemical shifts of an XPS spectrum can be computed,
despite the fact that the absolute values of the binding energies
cannot.
Replacing normal oxygen by a core hole oxygen lowers the
symmetry of the crystal, and this is taken into account in
constructing the k-space grid by ABINIT. Moreover, the
interactions between defects (in neighboring unit cells) should
be considered. We found however that for the relatively large
unit cells studied here, and at the 0.5 eV resolution necessary
for distinguishing different XPS shifts, replacing the unit cell
by a larger supercell made no significant difference in the energy
shifts.
The NMR parameters were computed using the NMR
module28,29 of the CASTEP code.30 This code also uses a density
functional theory approach within the planewave/pseudopotential
formalism. Additionally it includes code to calculate the
magnetic field response and the electric field gradient tensors.
This method has been validated recently on a variety of oxide
and glassy materials.31 The calculations were performed using
Troullier-Martins21 pseudopotentials. Pseudopotentials for Na
and O were generated using parameters given in ref 31 and for
Te using parameters given in ref 32. All calculations used the
experimental crystal structures. The PBE23 exchange-correlation
functional was used and the wave functions expanded in
planewaves with a maximum energy of 50 Ha.
For both the pseudopotentials and the calculations on the
periodic solids, the generalized gradient approximation for the
exchange and correlation terms due to Perdew, Burke, and
Enzerhof was used.23 For oxygen, an additional pseudopotential
with valence configuration 1s2s22p5 was generated, for use in
simulating oxygen 1s XPS data. All pseudopotentials constructed
were checked for existence of “ghost” states24 through the tests
available in the FHI98PP package and examination of the
logarithmic derivatives. Transferability tests were performed by
checking that the eigenvalues of the pseudoatoms and the all-
electron atoms agreed for a variety of test configurations, in
which partial charges were transferred from lower lying to higher
lying valence states or ionized completely.
The output of an ab initio NMR calculation is the absolute
chemical shielding tensor, σjiso(r) defined as the ratio between
a uniform external magnetic field, B, and the induced magnetic
field Bin(r)
As discussed below, the ab initio computations used here are
of the total energy of the model crystals, with defects at different
locations. In all cases, a planewave cutoff energy of 30 Ha (816
eV) was used. The necessary integration over reciprocal space
was performed by using an optimum grid, in the sense of fewest
k-points, for which the shortest vector in real space not included
was about 18.5 Å (this corresponded to a spacing in reciprocal
Bin(r) ) -σjiso(r)B
(3)
The isotropic shielding σiso(r), is one-third of the trace of

7638 J. Phys. Chem. A, Vol. 109, No. 33, 2005
Hart et al.
TABLE 1: Computed Oxygen 1s XPS Shifts for Sites in
Crystalline Na2SiO3, Na2Si2O5, Na2Te4O9, and Na2Te2O5,
together with Hirshfeld Net Charge Analysis for Each, in
Units of the Magnitude of the Electron Chargea
crystal
site
shift/eV
net charge/|e-|
Na2SiO3
BO-1
NBO-1
BO-1
BO-2
NBO-1
BO-1
BO-3
BO-5
BO-6
BO-7
BO-8
BO-9
NBO-2
NBO-4
BO-4
0.00
-1.69
0.09
-0.09
-1.71
0.11
0.05
0.19
-0.30
0.08
-0.10
-0.03
-0.41
0.18
-0.18
0.18
-0.56
-0.24
-0.22
-0.22
-0.37
-0.22
-0.23
-0.39
-0.28
-0.27
-0.27
-0.25
-0.28
-0.31
-0.25
-0.29
-0.32
-0.26
-0.27
-0.33
-0.32
-0.30
Na2Si2O5
Na2Te4O9
Na2Te2O5
BO-5
NBO-1
NBO-2
NBO-3
a BO and NBO refer to bridging oxygen and nonbridging oxygen,
respectively. The shifts are referenced to the average of the BO shifts
in each case.
TABLE 2: Theoretical and Experimental10 Results for 17O
NMR Parameters in Crystalline Tellurites
theory
site type CQ (MHz) ηQ
experiment
iso (ppm) CQ (MHz) ηQ
TeO2
δ
δ
iso (ppm)
180
O-1 BO
7.78
0.48 199.53
7.48
0.43
Figure 1. 17O magic angle spinning NMR spectra of sodium tellurite
glasses, of the stated compositions. Top: 9.4 T field (54 MHz oxygen
resonance), Hahn echo sequence. Asterisks denote spinning sidebands.
Bottom: 16.4 T field (94.9 MHz oxygen resonance, multiple RAPT
sequence).
Na2Te4O9
0.56 251.81
0.47 241.37
0.43 233.26
0.17 218.31
0.88 254.80
0.46 232.07
0.53 226.46
0.21 187.32
0.71 219.20
O-1 NBO
O-2 BO
O-3 BO
O-4 NBO
O-5 BO
O-6 BO
O-7 BO
O-8 NBO
O-9 BO
6.74
6.55
6.21
7.20
6.12
7.14
6.82
7.45
7.14
σjiso(r). To compare directly with experimentally measured
isotropic chemical shifts we use
δiso(r) ) -[σiso(r) - σref]
(4)
Na2Te2O5
0.25 230.04
0.04 207.21
0.27 255.58
0.57 220.03
0.49 279.19
with σref ) 259.32 ppm taken from previous work on glassy
materials.31
The quadrupolar coupling constant, CQ and the asymmetry
parameter, η values were extracted from the diagonalized electric
field gradient tensor whose eigenvalues are labeled Vxx, Vyy, and
Vzz, such that |Vzz| > |Vyy| > |Vxx|:
O-1 BO
O-2 NBO
O-3 NBO
O-4 BO
O-5 BO
7.65
7.05
6.53
5.91
6.41
Na2TeO3
0.12 182.3
0.10 171.01
0.02 214.61
O-1 NBO
O-2 NBO
O-3 NBO
6.16
6.28
6.41
6.63
0.33
158
CQ ) eVzzQ0/h
where h is Planck’s constant and
ηQ ) (Vxx - Vyy)/Vzz
(5)
the 17O NMR parameters in tellurite crystals, including R-TeO2,
Na2Te4O9, Na2Te2O5, and Na2TeO3. These systems differ
significantly in the range and types of bridging and nonbridging
oxygen they have.
(6)
We use a value for the electric quadrupole moment of the
oxygen nucleus of QO ) 2.55 fm2.31
4. Discussion
3. Results
4.1. Bonding in Tellurium-Centered Molecules. Under-
standing the bond lengths in tellurites is critical to understanding
their response to modification, and tellurium-centered molecules
are the simplest starting point.
Figure 3 shows a simple molecular orbital diagram for SiF4,
in its experimental ground-state geometry, as derived from ab
initio calculations. Note that no antibonding orbitals are oc-
cupied. Orbitals 2a1 and 2t2 together make up the σ-bonding
17O NMR spectra of the tellurite glasses are shown in Figure
1 for two fields, 9.4 and 16.44 T (17O frequencies of 54 and 95
MHz, respectively). Table 1 and Figure 2 show results of the
simulated oxygen 1s XPS shifts, for several model crystals: Na2-
SiO3, Na2Si2O5, Na2Te4O9, and Na2Te2O5. In each case, the
shifts are referenced to the average of the bridging oxygen shifts.
Table 2 shows computational and experimental results for

Bridging and Nonbridging Oxygen in Tellurites
J. Phys. Chem. A, Vol. 109, No. 33, 2005 7639
Figure 3. (a) Simple molecular orbital energy level scheme for SiF4
and TeF4 in tetrahedral geometry. The difference between the two arises
from the addition pair of electrons from Te (circled), which would
occupy an antibonding orbital. The distortion this causes is depicted
in part b, where the tetrahedral shape (left) is contrasted with the C2V
distortion adopted by TeF4.
a lower energy, lower symmetry structure.33 One might guess
that the resulting shape (four bonds in a C2V geometry plus a
lone pair) would require the use of d-orbitals on the tellurium,
but it is well-established that main group molecules such as
this do not use d-orbitals to form hybridized sets for bonding.34,35
The bond that forms in addition to the original directed σ bonds,
is best described as a three-center, four-electron bond across
the two axial fluorine and the tellurium atom. This weak bond
retains antibonding character, and it is doubtless why these
bonds are longer than the bonds to the equatorial fluorine
ligands.
This bonding pattern persists into the solid state, with an
interesting alternating pattern.11 The crystal structure of TeO2
shows it to consist of TeO4 polyhedrons, with shapes very
similar to molecular TeF4, including local C2V symmetry, two
long axial bonds, and two short equatorial bonds. In addition,
the oxygens in the axial positions are bonded equatorially to
adjacent Te atoms, and the equatorial oxygens are bonded axially
to adjacent Te atoms. In other words, each oxygen participates
in TeO2 in both a long and a short bond, in contrast to the
oxygen in SiO2, in which all Si-O bonds are equivalent.
4.2. XPS and NMR Data: Comparison with Theory. Since
the short-range structures of amorphous and crystalline oxides
are similar,36 crystalline model compounds were used for
calculation of the XPS shifts. These included Na2Te4O9,
Na2Te2O5, Na2Si2O5, and Na2SiO3. All four of these compounds
include a mix of bridging and nonbridging oxygens. Na2SiO3
includes two equivalent bridging and two equivalent nonbridging
oxygens on each silicon.37 Na2Si2O5 has two types of bridging
oxygens and one type of nonbridging oxygen, in a 2:1:2 ratio.38
Na2Te4O9 has nine distinct oxygen sites, six or seven of which
are bridging (depending on what cutoff criterion is used).39
Finally, Na2Te2O5 has five distinct oxygen sites, two of which
are bridging.40
Figure 2. Model oxygen 1s XPS spectra computed for (a) Na2SiO3,
(b) Na2Si2O5, (c) Na2Te4O9, and (d) Na2Te2O5. Spectra computed from
ab initio calculations as outlined in the text, using a pseudopotential
with a 1s hole. Spectra have been artificially broadened with Gaussians
of width 0.5 eV. In each plot, bridging oxygen sites are plotted with
solid lines, nonbridging oxygen sites with dashed lines, and the total
sum with heavy solid lines.
orbitals, which in the nomenclature of hybridization gives rise
to sp3 bonding. The electrons in orbitals 1e, 1t1, and 3t2 arising
from the fluorine atoms are nonbonding lone pairs. The change
from SiF4 to TeF4 in this geometry leads to addition of the two
circled electrons into the 3a/1 antibonding orbital. This addition
is not sufficient to destabilize the molecule relative to neutral
atoms, but does nevertheless lead to a symmetry-breaking
distortion to a C2V structure. This distortion can be understood
as arising from a pseudo-Jahn-Teller mechanism, in which the
occupied 3a/1 orbital interacts through the shape distortion with
the lowest unoccupied orbital, here of symmetry t2, to produce
We computed the oxygen 1s XPS chemical shift between
the different sites in these crystals, following the methodology
outlined in eq 2 above, and the results are summarized in Figure
2 and Table 1. Table 1 shows that there is a significant shift
(about 1.7 eV) between the bridging and nonbridging sites in
the silicates, just as is observed in experiments1 while in the
tellurites there is almost no discernible correlation between shift
and site. Figure 2 shows in fact that the tellurite shifts lie
virtually on top of one another, leading to a total lack of

7640 J. Phys. Chem. A, Vol. 109, No. 33, 2005
Hart et al.
nonbridging sites.31 The lack of correlation in Figure 4 thus
provides further evidence that in the tellurites these sites have
quite similar electronic structure.
4.3. Origin of Differences to Silicates. It is apparent, based
on the XPS and NMR data and the ab initio calculations, that
bridging and nonbridging oxygens in tellurites have electronic
environments that are remarkably similar, in sharp contrast to
what is observed in silicates. We now suggest a reason for this
similarity. We noted in Figure 3 that the energetic cost of the
extra pair of antibonding electrons in a molecule like TeF4, as
compared to SiF4, is at least partially regained by distorting to
a lower symmetry structure. In this new structure, there are both
long bonds and short bonds, the long bonds including the
majority of the antibonding character of the additional pair of
electrons. When these units are linked in R-TeO2, each oxygen
participates in both a long and a short bond, connected to
neighboring tellurite atoms. In silica, by contrast, all Si-O bonds
are equivalent. Therefore, it appears that in telluria already for
bridging oxygen there is a broken symmetry. When the material
is modified by adding an alkali oxide, such as Na2O, and bonds
break to accommodate the additional charge of the sodium
cations and the additional oxygen atoms, it is reasonable to
assume that the longer (weaker) of the two bonds will break
preferentially. Then when we compare the spectral response of
these newly formed nonbridging oxygen with the response of
the bridging conformation, we must remember that we are
comparing an oxygen with a short Te-O bond to one that did
have two bonds but for which one was extended and had more
antibonding character. Morever, as we saw from Hirshfeld
charge analysis, the additional charge introduced into the oxide
network through alkali ion modification is spread more evenly
over the oxygen sites in the tellurites, presumably facilitated
by the three-center four-electron bond. Thus, in the final
analysis, it is perhaps not so surprising that the bridging and
nonbridging oxygen sites appear similar spectroscopically in
the tellurites; the distinction between bridging and nonbridging
has already been mostly lost in the unmodified starting material,
TeO2.
Figure 4. Correlation between calculated isotropic chemical shift and
quadrupole coupling strength for 17O in tellurite crystals (data plotted
from Table 2). In all cases open symbols represent bridging oxygen,
and filled symbols represent nonbridging oxygen. Squares: TeO2.
Circles: Na2Te4O9. Triangles (point up): Na2Te2O5. Triangles (point
down): Na2TeO3.
resolution of the bridging and nonbridging sites. To be sure, in
the Na2Te2O5 sample the nonbridging oxygen shifts are all at
slightly lower binding energies than the bridging sites, so it
would appear that as the ionic environment in tellurites increases,
the bonding follows increasing silicate-like patterns.
The difference between the two types of sites was examined
more quantitatively by using Hirshfeld charge analysis41 to
assess the electron density around each oxygen. In the silicate
we found that the nonbridging oxygen holds about 0.17-0.19
e- more net charge than the bridging site, while in the tellurite
the total spread over all sites, from most charged to least, is
only 0.07 e-. Again, there is no noticeable correlation between
the charge and the structural nature of the site. We note also
that the magnitudes of the charges on the oxygen in the silicates
span that in the tellurites, such that the silicate BO sites have
less charge, and the NBO sites have more charge, than the
tellurite sites, with the tellurite charge values falling roughly in
the middle. Hirshfeld charge analysis on SiO2 and TeO2 give
respectively -0.22|e-| and -0.27|e-| as the net charges on
oxygen, both of which agree well the values computed for the
bridging sites in the silicates and tellurites.
Like oxygen 1s XPS, 17O NMR typically shows a significant
distinction between bridging and nonbridging sites in oxide
glasses. However, we found that even at moderately high field
(16.44 T, 95 MHz oxygen resonance frequency) the Magic
Angle Spinning spectra do not show any clear indication of
multiple sites with differing shifts or quadrupole coupling
constants (Figure 1). In contrast, at this field strength a silicate
glass such as Na2Si2O5 would show clear resolution of the
bridging and nonbridging sites (based on published values of
the oxygen NMR parameters3). The observed resonances do
narrow on going from the lower to the higher field, by a factor
of roughly two in ppm units. In this case the field has increased
by a factor of 16.44/9.4, or 1.74; since second order quadrupole
broadening varies as the inverse field squared, again when using
ppm units for the shift, we would anticipate a narrowing by a
factor of about 3. The remaining width is presumably due to
the intrinsic disorder, since even at extremely high magnetic
fields a glass would not give sharp NMR resonances. However,
no hint of resolution of two oxygen environments is apparent.
The ab initio calculations of the 17O NMR parameters show
a similar overlap between the different types of sites in
crystalline tellurites (Table 2). Figure 4 shows the very minimal
correlation that exists between the isotropic chemical shift and
the quadrupole coupling strength for 17O in the various tellurite
crystals; one sees that there is no clear separation between the
two types of sites plotted in this way. This is in contrast to
silicates, for which it has been shown previously that this type
of plot gives a very clear distinction between bridging and
5. Conclusions
We have shown in this paper that both the oxygen 1s XPS
and 17O NMR spectra of tellurites do not resolve the distinction
between bridging and nonbridging oxygen. These structures are
clearly resolved in silicates, and they play a large role in
understanding the mechanical and thermal properties of glasses.
We showed also that high level ab initio calculations indicate
for both XPS and NMR that the bridging and nonbridging sites
respond in a very similar way. We conclude from this study
that the reason for this similarity lies in the broken symmetry
inherent in the original R-TeO2 structure, for which the
additional antibonding electrons of TeX4, as compared to SiX4,
induce an asymmetry between Te-X bonds. Both long (weaker)
and short (stronger) bonds are already present in R-TeO2; upon
modification, this dissimilarity of course persists, but upon
comparing the spectral responses of the two cases, there is
relatively little change.
Acknowledgment. J.W.Z. thanks Prof. Neil Burford and
Prof. Russ Boyd for many helpful conversations. Financial
support for this research came from the US National Science
Foundation, the Canada Foundation for Innovation, the Canada
Research Chairs program, and the Natural Sciences and
Engineering Research Council of Canada.

Bridging and Nonbridging Oxygen in Tellurites
J. Phys. Chem. A, Vol. 109, No. 33, 2005 7641
(20) Fuchs, M.; Scheffler, M. Comput. Phys. Commun. 1999, 119, 67.
(21) Troullier, N.; Martins, J. L. Phys. ReV. B 1991, 43, 1993-2006.
(22) Louie, S. G.; Froyen, S.; Cohen, M. L. Phys. ReV. B 1982, 26,
1738-1742.
References and Notes
(1) Himei, Y.; Miura, Y.; Nanba, T.; Osaka, A. J. Non-Cryst. Solids
1997, 211, 64-71.
(2) Florian, P.; Vermillion, K. E.; Grandinetti, P. J.; Farnan, I.; Stebbins,
J. F. J. Am. Chem. Soc. 1996, 118, 3493-3497.
(23) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. ReV. Lett. 1996, 77,
3865-3868.
(3) Maekawa, H.; Florian, P.; Massiot, D.; Kiyono, H.; Nakamura, M.
J. Phys. Chem. 1996, 100, 5525-5532.
(24) Gonze, X.; Stumpf, R.; Scheffler, M. Phys. ReV. B 1991, 44, 8503-
8513.
(4) Du, J.; Cormack, A. N. J. Non-Cryst. Solids 2004, 349, 66-79.
(5) McLaughlin, J. C.; Tagg, S. L.; Zwanziger, J. W. J. Phys. Chem.
B 2001, 105, 67-75.
(25) Monkhorst, H. J.; Pack, J. D. Phys. ReV. B 1976, 13, 5188-5192.
(26) Johansson, B.; Ma˚rtensson, N. Phys. ReV. B 1980, 21, 4427-4457.
(27) Ma¨der, K. A.; Baroni, S. Phys. ReV. B 1997, 55, 9649-9658.
(28) Pickard, C. J.; Mauri, F. Phys. ReV. B 2001, 63, DOI 245101.
(29) Profeta, M.; Mauri, F.; Pickard, C. J. J. Am. Chem. Soc. 2003, 125,
541.
(30) Segall, M. D.; Lindan, P. J. D.; Probert, M. J.; Pickard, C. J.; Hasnip,
P. J.; Clark, S. J.; Payne, M. C. J. Phys.: Cond. Matt. 2002, 14, 2717-
2743.
(6) Niida, H.; Uchino, T.; Jin, J. S.; Kim, S. H.; Fukunaga, T.; Yoko,
T. J. Chem. Phys. 2001, 114, 459-467.
(7) Becker, C. R.; Tagg, S. L.; Huffman, J. C.; Zwanziger, J. W. Inorg.
Chem. 1997, 36, 5559-5564.
(8) Heo, J.; Lam, D.; Sigel, G. H., Jr.; Mendoza, E. A.; Hensley, D.
A. J. Am. Ceram. Soc. 1992, 75, 277-281.
(9) Lim, J. W.; Jain, H.; Toulouse, J.; Marjanovic, S.; Sanghera, J. S.;
Miklos, R.; Aggarwal, I. D. J. Non-Cryst. Solids 2004, 349, 60-65.
(10) Hart, R. T., Jr.; Zwanziger, J. W. J. Am. Ceram. Soc. 2005, in press.
(11) Lindqvist, O. Acta Chem. Scand. 1968, 22, 977-982.
(12) Masse, R.; Guitel, J. C.; Tordjman, I. Mater. Res. Bull. 1980, 15,
431-436.
(31) Charpentier, T.; Ispas, S.; Profeta, M.; Mauri, F.; Pickard, C. J.
Phys. Chem. B 2004, 108, 4147-4161.
(32) Yates, J.; Pickard, C.; Mauri, F.; Payne, M. J. Chem. Phys. 2003,
118, 5746-5753.
(33) Atanasov, M.; Reinen, D. Inorg. Chem. 2004, 43, 1998-2012.
(34) Kutzelnigg, W. Angew. Chem., Int. Ed. Engl. 1984, 23, 272-295.
(35) Reed, A. E.; Weinhold, F. J. Am. Chem. Soc. 1986, 108, 3586-
3593.
(36) Zwanziger, J. W. Int. ReV. Phys. Chem. 1998, 17, 65-90.
(37) McDonald, W. S.; Cruickshank, D. W. J. Acta Crystallogr. 1967,
22, 37-43.
(38) Pant, A. K. Acta Crystallogr. B 1968, 24, 1077-1083.
(39) Tagg, S. L.; Huffman, J. C.; Zwanziger, J. W. Chem. Mater. 1994,
6, 1884-1889.
(40) Tagg, S. L.; Huffman, J. C.; Zwanziger, J. W. Acta Chem. Scand.
1997, 51, 118.
(13) Kwak, H.-T.; Prasad, S.; Clark, T.; Grandinetti, P. J. Solid State
NMR 2003, 24, 71-77.
(14) te Velde, G.; Bickelhaupt, F. M.; van Gisbergen, S. J. A.; Foncesca
Guerra, C.; Baerends, E. J.; Snijders, J. G.; Ziegler, T. J. Comput. Chem.
2001, 22, 931-967.
(15) Foncesca Guerra, C.; Snijders, J. G.; te Velde, G.; Baerends, E. J.
Theor. Chem. Acc. 1998, 99, 391-403.
(16) “ADF2002.03”, http://www.scm.com. Theoretical Chemistry, Vrije
Universiteit: Amsterdam, The Netherlands.
(17) Becke, A. D. Phys. ReV. A 1988, 38, 3098-3100.
(18) Perdew, J. P. Phys. ReV. B 1986, 33, 8822.
(19) Gonze, X.; et al. Comput. Mater. Sci. 2002, 25, 478-492.
(41) Hirshfeld, F. L. Theor. Chim. Acta 1977, 44, 29.