Sie sind auf Seite 1von 10

Journal J. Am. Ceram. Soc.

, 82 [6] 1519–28 (1999)

Multiple-Quantum Magic-Angle Spinning 17O NMR Studies of Borate,


Borosilicate, and Boroaluminate Glasses
Shuanhu Wang and Jonathan F. Stebbins*,†
Department of Geological and Environmental Sciences, Stanford University, Stanford, California 94305

Multiple-quantum magic-angle spinning (MQMAS) 17O quantum magic-angle spinning (MQMAS), which achieves
NMR spectroscopy has been applied to study boron oxide, high resolution through the correlation of the multiple- and
sodium borate, sodium potassium borosilicate, and sodium single-quantum coherences for a quadrupolar spin system.
boroaluminate glasses. Up to eight distinct oxygen sites are MQMAS does not introduce technical complications beyond
identified, with the chemical shift and quadrupolar cou- the simple MAS NMR technique and has been successfully
pling parameters determined for each. Previous assignment used to study 27Al, 23Na, and 87Rb in a variety of model com-
of the B–O–B and Si–O–B resonances has been found to be pounds as well as complex materials.14–19
incorrect. In contrast to standard models of glass structure, For 17O, earlier studies have focused on aluminosilicate zeo-
three-coordinated boron mixes with the silicon oxide net- lites and glasses.20 –25 It has been demonstrated that MQMAS
work to a great extent. Sodium borosilicate glasses with is capable of separating resonances from Si–O–Si, Si–O–Al,
low-sodium content are likely to be phase-separated on the and nonbridging oxygens (Si–O–A, where A represents one or
nanoscale. Those with intermediate sodium content form more network-modifier cations). In the study presented here,
homogeneous glasses with boron atoms distributed evenly we further explore the utility of 17O MQMAS in a number of
in the SiO2 network. A boron avoidance rule analogous to borosilicate and boroaluminate glass samples.26 Through MQ-
the aluminum avoidance rule may apply in this region, MAS, we are able to identify up to eight oxygen sites in these
when BO4 groups are abundant. Adding excess sodium cat- systems and to determine the chemical shift and quadrupolar
ions to the system may again lead to compositional hetero- parameters for each of them. The results provide new insights
geneity. The existence of diborate group as the basic build- into the structure of boron-containing oxide glasses and chal-
ing unit in glasses with appropriate sodium content is not lenge the previous structural model proposed by Dell and
well supported. The network-forming cations in sodium Bray.27 Our results also clearly demonstrate the limitations of
boroaluminate glasses are well mixed, with no Al–O–Al MAS NMR and suggest that high-resolution techniques are
resonance observed by MQMAS. The effect of cation type needed to allow reliable assignment of the 17O spectra.
and thermal history on glass structures is also discussed. Borosilicates and boroaluminosilicates are among the most
widely used glass materials. The structure of these glasses has
been studied for about 40 years with different spectroscopic
I. Introduction techniques, including NMR28 and Raman spectroscopy.29,30
The most informative technique has been 11B NMR, which can
O XYGEN is the most abundant element in many technologi-
cally important materials, such as zeolites, ceramics, and
glasses. As a powerful tool to characterize these materials, 17O
quantify the concentrations of three- and four-coordinated bo-
ron. For sodium borosilicate glasses, the relative ratio between
solid-state NMR is sensitive to the number and type of cations the two boron environments is best predicted by a model from
that are bonded to oxygen, and it provides a unique method to Bray et al.27,31,32 The model assumes that the glasses contain
study cation distributions in multicomponent oxides.1 Accord- several larger structural units, which, in turn, are composed of
ing to earlier magic-angle spinning (MAS) studies, changing basic units, such as four-coordinated silicon, three- and four-
one or more first-neighbor cations for an oxygen site often coordinated boron. The model is successfully used to predict
results in discernible changes in the oxygen spectrum.1,2 The the physical and chemical properties of borosilicate glasses
effects can be blurred, however, by the low resolution one can (such as density and glass transition temperature).33 The suc-
achieve through conventional MAS NMR.3 As a quadrupolar cess of the model also allows it to be extended to other systems,
nucleus, 17O experiences a quadrupolar interaction that can such as boroaluminate34,35 and borophosphate glasses.36
only be partially averaged out by MAS. In the past 10 years, Dell and Bray’s model was based primarily on boron spe-
dynamic-angle spinning (DAS)4,5 and double rotation (DOR) ciation. For sodium borosilicate glasses, two molar ratios, R ⳱
NMR6 have been developed to overcome the resolution prob- Na2O/B2O3 and K ⳱ SiO2/B2O3, are introduced to characterize
lem. Both techniques have been applied to oxide systems, in- the system. The fraction of the BO4 group (N4) increases with
cluding glasses, and superior resolution has been achieved.7–11 sodium content up to R ⳱ 0.5 + K/16. N4 is almost constant in
The applicability of these techniques is reduced, however, by the region with 0.5 + K/16 < R < 0.5 + K/4 and then drops
the associated technical difficulties. Recently, Frydman and linearly as more sodium is added. The model suggests that
co-workers12,13 introduced the new approach of multiple- when R < 0.5, many compositions, especially those high in
SiO2, undergo visible phase separation into borate- and silicate-
rich regions. Even for glasses in this range that are optically
homogeneous, it is generally assumed that there is little mixing
of the borate and silicate components of the network, with most
R. K. Brow—contributing editor
or all Na+ entering the borate-rich regions. The structure is
often described in terms of “diborate units,” which are assumed
to be predominant at R ⳱ 0.5. These account for the initial
increase in N4. When 0.5 < R < 0.5 + K/16, a new structural unit
Manuscript No. 190176. Received May 22, 1998; approved October 10, 1998. of the composition NaBSi4O10 (reedmergnerite) is formed,
Supported by the National Science Foundation under Grant No. NSF DMR which contains only four-coordinated boron, so N4 keeps in-
9626735. creasing. In the next region where R < 0.5 + K/4, additional
*Member, American Ceramic Society.

Author to whom correspondence should be addressed. sodium cations are consumed by reedmergnerite groups to
1519
1520 Journal of the American Ceramic Society—Wang and Stebbins Vol. 82, No. 6
17
form nonbridging oxygen, so N4 is almost constant. As more O-enriched B2O3 and SiO2 were made by hydrolyzing sili-
Na2O is added, both the diborate and the reedmergnerite groups con tetrachloride or boron tetrachloride in 47% 17O-enriched
react to form pyroborate groups, which absorb more nonbridg- water in diethyl ether and CH2Cl2 solvents, respectively. The
ing oxygen and give more BO3 and fewer BO4 units. boric acid obtained was heated to 1273 K in argon flow and
Despite the success of this model, it is difficult to directly then quenched to room temperature to give the B2O3 glass. The
observe the larger structural units. 29Si NMR and Raman spec- sodium borate glass NB12 was made by mixing the 17O-
troscopy have begun to be important in this issue, and it has enriched B2O3 glass with Na2CO3, heating the mixture at 1273
been admitted that the physical reality of such crystallike units K under argon for 2 h, and removing the crucible from the
in glasses is open to some question. In particular, a detailed furnace. The NBS126 glass was prepared by heating the mix-
study using 29Si, 23Na, 17O NMR and Raman spectroscopy ture of NB12 glass and 17O-enriched SiO2 at 1473 K prior to
suggested the need for significant revision of some aspects of quenching. Other sodium and potassium borosilicate glasses
the model.30 However, even in this case, spectral resolution were made similarly, except that the starting materials were
was not good enough to uniquely constrain the structure. With Na2CO3 or K2CO3, H3BO3, and enriched SiO2, and that the
high-resolution 17O NMR, new insight into these structural heating temperature ranged from 1073 to 1473 K, depending
units is available, suggesting the need for significant modifi- on the melting temperature of the mixture. To study the fictive
cations of the standard model. temperature effect, some of these samples were quenched at 40
Much less is known about boroaluminate glasses. Again, 17O K/h (annealed samples), and some of the others were made by
MQMAS provides unique information on oxygen environ- heating the slow-quenched samples in a platinum tube on a
ments, particularly on the extent of boron and aluminum mix- Bunsen burner (T ≈ 1223 K) and subsequently hammering
ing and the presence (or absence) of Al–O–Al linkages. them against a large piece of steel block into thin sheets of
glass. The estimated cooling rate was ∼500–1000 K/s (hammer
quenched samples). Sodium boroaluminate glasses were pre-
II. MQMAS NMR Theory pared similarly from Na2CO3, Al2O3, and 17O-enriched B2O3
The theory of MQMAS has previously been presented in the glass. Weight loss during preparation was found to match well
literature and is only briefly discussed here.12–14 The theory is with nominal stoichiometry (the discrepancy was at most 1%–
similar to that of DAS,5 where a two-dimensional correlation 3%). Powder X-ray diffraction showed that all the samples
experiment is performed to achieve high resolution. In DAS, were amorphous, and all glasses were optically clear. Except
high resolution is achieved through sample spinning around for the B2O3 glass, all other samples were doped with 1000
two angles consecutively; in MQMAS, the multiple-quantum ppm CO3O4 to efficiently reduce the oxygen spin-lattice relax-
coherence is the degree of freedom that is exploited to give ation time (T1) to 1–1.5 s. It should be noted that sample
high resolution. In the pulse sequence used in MQMAS, a first NBS126 could be within the region where optically detectable
pulse generates (through phase cycling) the triple-quantum co- phase separation could occur on annealing,38 as were several
herence, which evolves under triple-quantum Hamiltonian for a samples in previous investigations.30 Sample BS46 was close
short time t1 before it is converted to single-quantum coherence to the two-phase region, but was not definitely known to un-
by a second pulse. If this coherence is allowed to evolve for kt1 dergo phase separation.
under the single-quantum Hamiltonian, an echo forms, and the
(2) NMR Spectroscopy
total evolution under the two Hamiltonians is devoid of aniso-
tropic second-order quadrupolar effect. In our experiment, NMR spectra were obtained with a modified Varian (Palo
however, a ␲ pulse is added as the last pulse to give 17O Alto, CA) VXR-400S spectrometer at a Larmor frequency of
MQMAS spectrum with better phasing. Such a scheme has 54.22 MHz for 17O and 128.25 MHz for 11B (9.4 T magnetic
been described extensively elsewhere.15,37 field). A fast-spinning 5 mm probe (Doty Scientific, Columbia,
SC) was used, and the spinning rate was 10 kHz for 11B,
and 13.5–14.5 kHz for 17O. For both nuclei, nonselective
III. Experimental Procedures 30° pulses were used to collect the MAS and static spectra. A
delay of 1 s was used between successive pulses for 11B,
(1) Sample Preparation while for 17O glasses, 3–5 s delays were used. For some of
Glass samples are labeled according to the molar ratios of the samples, 10 and 100 s delays were used, but the spectra
the different components (Table I). For example, NBS316 rep- showed no differences compared to those with shorter de-
resents a glass with 30 mol% Na2O, 10 mol% B2O3, and 60 lays. The frequency reference for 17O was external tap water;
mol% SiO2. Similarly, sodium boroaluminate glass with 40 for 11B, 0.1M aqueous H3BO3 was used and assigned to +19.6
mol% Na2O, 30 mol% B2O3, and 30 mol% Al2O3 is named ppm relative to the more common standard, boron trifluoride
NBA433. For binary borate glasses, NB12 means that the mo- etherate.
lar ratio between Na2O and B2O3 is 1:2. MQMAS spectra were collected using the shifted-echo se-
quence. A 3␲ pulse of 9.5 ␮s was used to excite the triple-
quantum coherence, and a shorter ␲ pulse of 3.5 ␮s was used
to convert the triple-quantum coherence into single-quantum
Table I. Nominal Compositions of Samples Used in coherence.39 The third pulse was a selective ␲ pulse of 15 ␮s,
This Study and the echo time was usually 70–80 ␮s.15 Typically, 720–
Composition (mol%) 1440 scans were averaged for each of the 72 t1 points with a
Label Na2O K2O B2O3 SiO2 Al2O3 3 s delay between successive scans. The t1 dimension was
zero-filled to 512 points prior to Fourier transformation. The
B2O3 100
BS46 40 60 MQMAS NMR experiment yielded a two-dimensional NMR
NB12 33 67 spectrum, with an isotropic dimension devoid of second-order
NBS126 11 22 67 quadrupolar broadening and an anisotropic dimension corre-
NBS116 13 13 75 sponding to a somewhat distorted MAS spectrum. Other details
NBS226 20 20 60 of data processing can be found in our earlier work on 27Al
NBS316 30 10 60 MQMAS.16 In all of the plots of two-dimensional spectra
KBS226 20 20 60 shown here, 20 contour lines are drawn at uniform intervals
KBS316 30 10 60 between 10% and 100% of maximum intensity.
NBA351 33 56 11 For a spin-5/2 nucleus, such as 17O, the observed resonance
NBA433 40 30 30
frequency in the isotropic dimension of MQMAS is given by14
June 1999 NMR Studies of Borate, Borosilicate, and Boroaluminate Glasses 1521

17 CS 10 2Q
␦MQMAS = − ␦ + ␦ (1)
31 iso 31 iso
For a single-quantum experiment, such as MAS and DAS, the
frequency is
␦MAS = ␦iso
CS
+ ␦iso
2Q
(2)
where

␦iso
2Q
=−
40 I 共2I − 1兲
2 2 冋
3 × 106 关I共I + 1兲 − 3 Ⲑ 4兴 CQ
2
共 1 + ␩2 Ⲑ 3 兲
␻20
册 (3)

is the second-order quadrupolar shift and ␦CS iso the isotropic


chemical shift. CQ and ␩Q are the quadrupolar coupling con-
stant and asymmetry parameter, respectively. From a MQMAS
spectrum, ␦MQMAS can be directly read out, while ␦MAS can be
estimated by the center of gravity of the MAS dimension.16
Knowing both shifts allows the determination of both the iso-
iso ) and the quadrupolar product (PQ ⳱ CQ[(1 +
tropic shift (␦CS
␩2)/3]1⁄2). In the spectra of glasses, in which NMR peaks in-
variably represent sites with statistical distributions of struc-
tural variables, such as bond distance and angle, such determi-
nations represent only the most probable values of NMR
parameters. Similarly, fits to one-dimensional peak shapes
broadened by such disorder (as has been done in many previous
NMR studies of glass structure) produce averages of some-
times quite broad ranges of parameters. Typical uncertainties in
MQMAS average peak positions are about ±2 ppm; this results
in uncertainties in calculated PQ values that are typically ±0.2
MHz and in ␦CS iso values of ±3 ppm.
Peak intensities in MQMAS spectra, typically measured by
total projections in the isotropic dimension (which total all data
at each value of isotropic shift and thus reflect peak volumes)
are known to be significantly affected by variations in CQ, in
part because excitation of spins in sites with large CQ values is Fig. 1. Representative 17O MAS spectra for boron-containing oxide
less efficient.16 However, when CQ does not vary too much glasses. 100 Hz Gaussian line broadening was added before Fourier
from site to site, at least semiquantitative results can be ob- transformation.
tained. For example, in a recent study of 17O MQMAS in
aluminosilicate glasses, intensity ratios for Si–O–Al (CQ ≈ 3.5
MHz) and for Si–O–Si (CQ ≈ 5.1 MHz) peaks are within 10% peak in B2O3 glass (Fig. 3(a)). This peak position is also con-
to 20% of ideal values.40 sistent with two of the resonances observed previously by DOR
NMR,42 both of which give predicted MQMAS peaks near to
−74 ppm. However, a third resonance predicted to occur at −60
IV. Results and Discussion ppm is not observed. This may be due to the difference in the
sample preparation procedures (e.g., H2O content), but it is
(1) MAS and One-Dimensional MQMAS Spectra more likely that the earlier assignment based on DOR is inac-
Typical MAS spectra for 17O in some borosilicate and curate, perhaps because sample spinning speed was low. It
boroaluminate glasses are shown in Fig. 1. All the spectra remains likely that the distinct types of oxygen sites enumer-
contain more than one oxygen environment, but, at 9.4 T, few ated previously for B2O3 glass42 (B–O–B within rings versus
distinct features are obtainable. B2O3, BS46, and BS226 between rings, etc.) are indeed present, but MQMAS in our
glasses have been investigated previously, and multiple oxygen experiments does not resolve their contributions to the total
sites were characterized through curve fitting of the MAS and NMR signal. For typical NMR parameters in these samples, a
static NMR spectra.30,41,42 The isotropic chemical shift and ±10% uncertainty in CQ (e.g., from 4.5 to 5.5 MHz) changes
quadrupolar parameters obtained are summarized in Table II. the predicted MQMAS isotropic shift by about ±4 ppm. Likely
Using Eqs. (1) and (3), the isotropic shifts in MQMAS spectra variations in ␩ produce much smaller effects of 1–2 ppm.
at 9.4 T can be predicted and are also included in Table II. As for the B2O3 glass, sample BS46 has a peak at about
The isotropic dimension total projections of the MQMAS −70 ppm, which is thus assigned to B–O–B. (Previous
spectra of B2O3, NB12, BS46, NBS126, NBS226, KBS316, one-dimensional 17O NMR studies of this composition sug-
NBA351, and NBA433 shown in Fig. 2 have much better gested that such a peak should be at about −63 ppm.30) This
resolution than the MAS spectra. The full widths of the peaks peak does not exist in alkali borosilicate glasses with low boron
range from 10 to 20 ppm, which is about an order of magnitude content (see below), also suggesting that the new assignment is
narrower than the MAS spectral width. Up to eight peaks at correct. A second peak at −56 ppm is also observed in BS46,
different frequency regions can be identified and are labeled. which must then be due to Si–O–B. Based on several previous
The observed values are compiled in Table III, together with studies of silicates,2,7,20,23,43 the third peak at −39 ppm is
their final assignment. clearly due to Si–O–Si, consistent with previous assignment in
borosilicates.30
(2) Glasses with R ≤ 0.5 From the widths in the MAS dimension, it is clear that
The assignment for the B–O–B, B–O–Si, and Si–O–Si reso- oxygen sites in Si–O–Si, B–O–B, and Si–O–B have similar
nances is achieved by comparing the two-dimensional MQ- quadrupolar coupling constants. As shown in Table III, PQ
MAS spectra of B2O3 and BS46 shown in Fig. 3. The −70 ppm values are calculated from Eqs. (1) and (2) as 4.9–5.6 MHz.
(isotropic dimension) peak is from B–O–B, as it is the only Values for Si–O–B (5.6 MHz) are large compared with previ-
1522 Journal of the American Ceramic Society—Wang and Stebbins Vol. 82, No. 6

Table II. Previously Estimated Site Occupancies (occ) and 17O NMR Parameters for Glass Samples Similar to Those Studied
in This Paper
␦CS
iso CQ PQ ␦MQMAS‡
Sample Ref. Site occ (ppm) (MHz) ␩Q (MHz) (ppm)
B2O3 42 B–O–B (ring) 0.5 104 0.9 5.0 −74
41 B–O–B (ring) 0.67 4.7 0.6 5.0
42 B–O–B (nonring) 0.3 113 0.9 4.3 −74
41 B–O–B (nonring) 0.33 5.8 0.4 5.9
42 B–O–B (ring–to–ring) 0.2 79 0.9 5.0 −60

BS46 30 B–O–B 0.35
Si–O–Si 0.60
Si–O–B 0.05
NBS136 30 B–O–B + Si–O–Na 0.35
Si–O–Si 0.60
Si–O–B 0.05
NBS226 30 B–O–B + Si–O–Na 0.20
Si–O–Si 0.55
Si–O–B 0.25
NBS316 30 B–O–B§ 72 5.2 1.0 6.0 −63
Si–O–Si 0.45 48 5.0 0.0 5.0 −43
Si–O–B 0.10 20 4.3¶ 0.2 4.3 −18
Si–O–Na§ 38 2.5 0.0 2.5 −25
§
For these borosilicates, the reference gives only typical values of fitted NMR parameters, as listed below for NBS316.

Predicted 3QMAS isotropic shift at 9.4 T. §Occupancy for Si–O–Na plus B–O–B is 0.45. ¶Estimate based on simulated peak in figure.

Table III. 17O NMR Parameters for Different Oxygen


Environments, Determined by MQMAS†
Label in ␦MAS ␦MQMAS ␦CS
iso
Fig. 2 Site (ppm) (ppm) (ppm) PQ
A Si–O–Si (R < 0.5) −23 −39 37 5.4
A Si–O–Si (NBS316) 2 −43 51 4.9
B B–O–B† 30 −70 92 5.5
C B–O–B§ 30 −62 82 5.1
D Si–O–B 0 −56 64 5.6
E Si[B]–O–Na 25 −22 35 2.2
F Si[B]–O–K 67 −45 76 2.1
G Al–O–B 25 −43 59 4.1
H Al[B]–O–Na 10 −10 16 1.7

Typical uncertainties in ␦MAS are ±2 ppm, in PQ ±0.2 MHz (see text).
Si[B] denotes either Si or B, Al[B] either Al or B. ‡in B2O3 and BS46. §in NB12,
NBS126, and NBA351.

CQ ⳱ −0.203I(%) + 14.78 (4)


Using this equation, the predicted CQ value for Si–O–B is ∼5
MHz, consistent with the experimental value. Ab initio calcu-
lations44 also confirm our assignment. The similarity of the CQ
values for Si–O–Si, Si–O–B, and B–O–B also means that the
intensity ratio among these sites in MQMAS at least qualita-
tively reflects the actual concentration ratio. Even so, a rela-
tively large error may exist, especially when B–O–B resonance
is involved. We, thus, are cautious not to draw a definite con-
clusion, unless we see an intensity discrepancy that calls for at
least a 100% error. On the other hand, if we do not observe a
specific peak (with CQ < 6 MHz), we assume that its intensity
is lower than ∼5%–10%. The assumption is reasonable, as
previous experiments have demonstrated a detection limit
Fig. 2. Representative isotropic dimension total projections of two- ⱖ2%–3%.22
dimensional 17O MQMAS spectra. For positions and assignments of The two-dimensional 17O MQMAS spectra of NB12 and
labeled peaks, see Table III. All of the other features in the spectra are NBS126 are presented in Fig. 4. Glass NB12 has a crystalline
spinning sidebands or are indistinguishable from noise. analog (sodium diborate) in which 37.5% of the boron atoms
are four-coordinated. The crystalline phase contains oxygen
atoms either connecting two four-coordinated, or two three-
ous results of one-dimensional spectra,30 and much larger than coordinated, or one three- and one four-coordinated boron at-
typical values for Si–O–Al of ∼3.5 MHz,2,20,43 suggesting a oms. The structure of the glass is commonly assumed to be
significantly different effect on NMR parameters for the two similar to that of the crystalline material.45 The 17O spectra
different trivalent cations. It has been proposed that the qua- show a single peak at −62 ppm, suggesting that MQMAS of
drupolar coupling constant for bridging oxygens correlates lin- these samples at this field cannot resolve oxygens connected to
early with the (percentage) ionic character (I ) of the two- borons with different coordination numbers. However, the ob-
cation–oxygen bond.1 served peak is relatively broad in the isotropic dimension. This,
June 1999 NMR Studies of Borate, Borosilicate, and Boroaluminate Glasses 1523

17
Fig. 4. O MQMAS spectra for NB12 and NBS126 glasses.
17
Fig. 3. O MQMAS spectra for B2O3 glass and BS46 glass.

glasses, again suggesting multiple B–O–B environments, re-


lated either to variation in boron coordination number or to
together with the shift in peak position from those in B2O3 and coordination of the oxygen by Na+. The lack of detectable
BS46 (all BO3 groups), is consistent with the presence of more Si–O–Na peak in NBS126 also supports the standard view that
than one oxygen environment. It is possible that the shoulder Na+ does not interact with the silicate component in this region.
near to −55 ppm in the isotropic projection (Fig. 2) for this The scale of the implied compositional heterogeneities in
glass could represent NMR signal from oxygens with one or these glasses cannot be determined directly from spectroscopic
two BO4 neighbors, but further study is required to conclude data without knowledge of the compositions of the proposed
this definitively. When SiO2 is added to NB12 (NBS126), two domains. Regardless of domain size, however, substantial mix-
new peaks appear that can be unambiguously attributed to Si– ing of boron and silicon is required by the high observed con-
O–Si (−39 ppm) and Si–O–B (−56 ppm). centrations of Si–O–B sites in BS46 and NBS126. If the bo-
Glass BS46 (R ⳱ 0) and NBS126 (R ⳱ 0.5) are close to the rate-rich regions contain little SiO2, and Si–O–B sites are
region where liquid–liquid immiscibility can occur on anneal- present primarily at the interfaces between regions, then very
ing.38 Although both samples are optically clear, phase sepa- small domains would be required, perhaps in the range of 2–3
ration at the submicrometer level can be present. Nonetheless, nm as suggested in previous work on B2O3.42 We note, how-
their spectra are very useful in developing peak assignments, as ever, that our nonobservation of two of the 17O peaks predicted
discussed above. They are also helpful in testing the standard by that study implies that its conclusions may require revision.
model of Bray et al.,27 which assumes that little interaction 11
B MAS NMR spectra (not shown) show that all the boron
occurs between borate and silicate components in the R ⱕ 0.5 in BS46 and 60% of the boron in NBS126 are three-
region and that most or all Na+ interacts with borate groups. coordinated. Combined with the 17O spectra, which show a
The similar positions of the B–O–B peaks in B2O3 and BS46 large fraction of Si–O–B, it is clear that three-coordinated bo-
glasses suggest, e.g., that this species is similar in both ron mixes well with four-coordinated silicon. Conventionally,
samples. In NB12 and NBS126, the B–O–B peaks are again this is assumed to be unlikely, because the mixing is thermo-
nearly identical but are distinct from those in the sodium-free dynamically disfavored, and no crystalline materials with this
1524 Journal of the American Ceramic Society—Wang and Stebbins Vol. 82, No. 6

type of mixing have been observed.46 Our results thus demon- has been observed in other oxide glasses and is from non-
strate that the structure of a glass may be significantly different bridging oxygen.2,23,43 However, it is not clear at present
from that of the corresponding crystalline phase. The two- whether the peak is due to Si–O–Na, B–O–Na, or both. The
component BS46 provides a particularly simple look at the Si–O–Si resonance shifts from −39 ppm in BS46 to −41 ppm
extent of B–Si order/disorder. In this composition, if full anti- in NBS116 and to −42 ppm in NBS226, as a result of the
ordering prevails (no Si–O–B), then 50% of the oxygens would progressive increase in alkali content. Such a shift shows that
be B–O–B and 50% would be Si–O–Si. In the case of fully at least some of the sodium cations are associated with silicate
random mixing, one-third each of B–O–B, Si–O–B, and Si– groups.
O–Si would be present. The latter is not too different than the Dell and Bray’s27 model describes NBS226 by the following
observed relative peak areas for the isotropic total projection reaction
(30%, 38%, 28% respectively, each with uncertainties of
roughly a few percent), suggesting a high degree of disorder.
(3) Glasses with 0.5 < R < 0.5 + K/4
2Na2O + 2B2O3 + 6 SiO2 =
8 6冉
6 11
Na2O⭈B2O3⭈8SiO2 冊
5
Figures 5(a) and (b) are the two-dimensional 17O MQMAS + Na2O⭈2B2O3
spectra of the NBS226 and NBS116 glasses. Both samples 8
are well outside the known two-liquid field.38 Compared to
BS46 and NBS126, a new peak centered at −20 ppm in the The reaction scheme leads to about 42% Si–O–Si, 30% Si–O–
isotropic dimension is observed for NBS226. Such a peak B, and 22% B–O–B. Our experiment, however, suggests that
B–O–B content must be very low (−62 ppm, only one contour
level). This, in turn, means that the diborate group, which
previously was assumed to be one of the predominant units in
sodium borosilicate glasses, does not exist in large quan-
tity.27,30 Instead, the glass is relatively homogeneous, with bo-
ron atoms evenly distributed in the SiO2 network, which leads
to a large fraction of Si–O–B.
For NBS116, a similar reaction can be worked out

Na2O + B2O3 + 6 SiO2 = 冉


6 7
Na O⭈B2O3⭈8SiO2
8 6 2 冊
1
+ Na2O⭈2B2O3
8

This model predicts that ∼87.5% of the boron is four-


coordinated, consistent with the 11B NMR spectrum of the
sample (not shown). The B–O–B intensity in this case is at the
lower edge of the detection limit, and thus may not be observ-
able. The Si–O–B intensity, however, should be comparable to
that of the Si–O–Si resonance (2:3). Our experiment reveals a
ratio of 1:3, which does not agree well with the reaction
scheme. Further experiments are needed to clarify this incon-
sistency. It is clear, however, that boron and silicon networks
mix well, and diborate groups may not exist, consistent with
results for NBS226.
In multicomponent oxide materials, cation ordering is an
important question to be answered. In aluminosilicate zeolites
and glasses, extensive experimental results have shown that the
network-forming cations do not mix randomly. Instead, direct
Al–O–Al connections are at least partially avoided when the
composition allows.24,47 A boron avoidance rule may exist as
well, at least in systems where BO4 groups are abundant, plac-
ing excess negative charge on B–O–B oxygens. The lack of
observed B–O–B peaks in NBS116 and NBS226 supports this
hypothesis. The rule is less applicable when the alkali content
is low and BO3 groups predominate, as B–O–B oxygens are, in
such cases, at least formally neutral. In some systems with
abundant BO4 groups (e.g., the mineral danburite, CaB2Si2O8),
B–O–B pairs are actually energetically favored,48 perhaps be-
cause of the greater ability of divalent cations to charge balance
the bridging oxygens. This suggests that, as in the case of
aluminum avoidance, boron avoidance is more likely in alkali
glasses, rather than alkaline-earth, borate, and borosilicate
glasses.

(4) Glasses with R > 0.5 + K/4


Glass NBS316 belongs to this region, and its two-
dimensional MQMAS spectrum is shown in Fig. 5(c). The
MAS dimension for Si–O–Si is narrower than those for the
other glasses, suggesting a smaller quadrupolar coupling con-
Fig. 5. 17O MQMAS spectra for NBS226, NBS116, and NBS316 stant (4.9 MHz). Such a trend was also observed previously.30
glasses. Si[B] denotes either Si or B. The standard reaction scheme for this composition is
June 1999 NMR Studies of Borate, Borosilicate, and Boroaluminate Glasses 1525

3Na2O + B2O3 + 6SiO2 = 冉


5 5
8 2 2
Na O⭈B2O3⭈8SiO2 冊
1
+ 共2Na2O⭈B2O3兲
6
5
+ 共Na2O⭈2B2O3兲
48
+ Na2O⭈SiO2

which predicts 5% B–O–B, 28% Si–O–B, 47% Si–O–Si, and


20% Si[B]–O–Na. The large intensity of the nonbridging oxy-
gen peak in the experimental spectrum is consistent with the
model prediction, but Si–O–B intensity is low relative to Si–
O–Si intensity. A possible explanation is that the silicon and
boron networks in this case are less well mixed than the model
predicts. If this is the case, one can have a larger B–O–B
intensity if the predominant boron-rich unit is diborate
(Na2O⭈2B2O3). However, if the sodium cations preferably en-
ter the boron-rich phase to form pyroborate group diborate
(2Na2⭈B2O3), we can have a larger (Si–O–Si):(Si–O–B) ratio,
and a low (thus unobservable) B–O–B concentration. This ex-
planation is consistent with the high-temperature 11B and 29Si
study on NBS414 glass, which reveals that, at high tempera-
ture, the exchange rate between two silicon sites is different
from that for the two boron sites.49 The difference in dynamics
can be well accounted for if the system involves intermediate-
scale compositional heterogeneity. This suggestion is also con-
sistent with earlier fictive temperature studies, which suggest
that samples (NBS414) with different thermal histories have
different nonbridging oxygen contents but identical silicate
speciation.19

(5) Potassium Borosilicate Glasses


Substituting potassium for sodium causes the chemical shift
of the nonbridging oxygen peak to increase to a frequency
higher than those of the Si–O–Si and Si–O–B peaks. As shown
in Fig. 1, this shift leads to a better-resolved MAS spectrum.
The substitution also moves the observed shift in the isotropic
dimension from −22 to −45 ppm. Even though the Si–O–Si
resonance appears around −44 ppm, the MAS dimension sepa-
ration is significant enough that the two resonances are well
differentiated in the two-dimensional spectrum (Fig. 6(a)). The
Si–O–Si and Si–O–B peak positions are less affected by cation
substitution and are hardly shifted from those observed in the
NBS glasses.
The KBS316 and NBS316 17O spectra agree well with each
other. Both spectra contain large intensities for nonbridging
oxygen and Si–O–Si but a relatively small peak for Si–O–B. Fig. 6. 17
O MQMAS spectra for KBS316 and KBS226 glasses.
Similar to NBS316, compositional heterogeneity can exist in
KBS316 as well. The KBS226 spectrum in Fig. 6(b) clearly
shows the Si–O–Si and Si–O–B resonances, with their intensity at different rates are compared. The change in boron speciation
ratio consistent with that for NBS226 glass. However, the non- is clear. Similar results are obtained for KBS316 (Fig. 7(b)),
bridging oxygen peak has a much lower intensity. At present, except that the change is more drastic.
it is hard to determine what causes the difference. Previous In Fig. 8(b), the two-dimensional MQMAS spectrum for
study of lithium borosilicate (LBS)50 glasses has revealed that NBS316 quenched at 40 K/h is shown. This spectrum is almost
boron speciation for LBS and NBS can disagree by as much as the same as the one in Fig. 5(c), showing data for the hammer-
20%. Similar differences can exist between NBS and KBS quenched sample. This means that MQMAS is not sensitive
glasses. enough to detect the structural changes in NBS glasses. For
KBS316, however, the difference is clear by comparing Figs.
(6) Temperature-Dependent Structural Changes 8(a) and 6(a). As expected, the nonbridging oxygen peak in-
Previous statistical thermodynamic modeling has suggested tensity in the slow-quenched sample is lower. In the two-
that boron speciation changes with temperature in sodium bo- dimensional plot, this change is seen most clearly by the pres-
rosilicate systems.51,52 Experimental evidence for such changes ence of one additional contour line in the Si–O–Si peak for the
comes mainly from 11B NMR of NBS glass samples quenched annealed glass. Because the maximum contour height is as-
at different rates.19,53,54 One of the major reactions in these signed to the highest (NBO) peak, this means that the propor-
systems is BO4 ↔ BO3 + NBO, which shifts to the right as tion of the latter species is lower in this sample than in the
temperature is increased. Associated with this reaction is a hammer-quenched glass. This change also affects the intensity
change of the amount of nonbridging oxygen, which may be ratio (NBO + Si–O–Si):(Si–O–B) as clearly seen in the isotro-
observable in high-resolution 17O NMR. Another question that pic total projection (Fig. 9).
is unclear is whether the NBOs created form Si–O–Na or B–O– 17
O spectra on fast- and slow-quenched NBS226 and
Na. In Fig. 7(a), the 11B spectra for NBS316 glasses quenched KBS226 glasses are also studied. Similar to the N/KBS316
1526 Journal of the American Ceramic Society—Wang and Stebbins Vol. 82, No. 6

Fig. 7. 11B MAS spectra for NBS316 and KBS316 glasses with
different thermal histories. For each composition, two spectra are plot-
ted, normalized to the highest peak. ((—) data for hammer-quenched
samples; (– – –) data for slow-cooled (40 K/h) glasses; BO4 marks the
narrow peaks for four-coordinated B; and BO3 marks the high-
frequency side of the doublets for three-coordinated B.)

case, the fictive temperature effect is more dramatic for


KBS226 than for NBS226. At present, we are not able to
determine what causes this difference. Experiments on samples
with a wider range of compositions would help answer this
question in the future.

(7) Sodium Boroaluminate Glasses


11
B and 27Al NMR have been very useful in elucidating the
structure of boroaluminate glasses,34,35,55 but 17O NMR of
these materials has not yet been exploited. One key question
that remains ambiguous is how the framework aluminum atoms
mix with boron atoms. Indirect evidence shows that aluminum
avoidance may apply. Such evidence includes boron specia- Fig. 8. 17O MQMAS spectra for NBS316 and KBS316, quenched at
tion, trends in glass transition temperatures, and trends in 11B 40 K/h. Si[B] denotes either Si or B.
and 27Al NMR data reported in a recent one-dimensional and
MQMAS NMR study. 55 Direct evidence from double- tween two AlO4 groups (Stebbins et al.‡). These results predict
resonance NMR was also reported recently.56,57 17O NMR, an MQMAS NMR peak at approximately the location shown
which is sensitive to the atoms directly bonded to oxygen at- by the box in Fig. 10(a). In addition, ab initio calculations
oms, is an ideal technique that potentially provides the most predict that such Al–O–Al sites have CQ values of ∼2.7 MHz,
straightforward evidence for the mixing problem.
iso ∼4 ppm lower than that for Si–O–Si sites (the latter
with ␦CS
Three resonances are observed for glass NBA351 and with a predicted CQ of 4.9 MHz).44 Given the data presented
NBA433 (Fig. 10). The −60 ppm peak in NBA351 is only 2 for Si–O–Si in NBS316 (Table III), this predicts an MQMAS
ppm away from the B–O–B resonance in the binary borate peak for Al–O–Al centered at about −30 ppm in the isotropic
glass and can be confidently assigned to B–O–B. As expected, dimension and 32 ppm in the MAS dimension, not too different
it disappears when a large fraction of aluminum is present from the value for NaAlO2. The lack of any such feature in the
(NBA433). The −45 ppm peak in both glasses has not been spectra for boroaluminate glasses suggests that aluminum
previously reported. Such a peak has CQ ⳱ 3.8 MHz and must avoidance prevails. In addition, the large intensity of Al–O–B
be due to Al–O–B. For Al–O–B, Eq. (4) predicts a CQ of 4 in NBA351 also seems to support the aluminum avoidance
MHz (I ⳱ 53%),1 consistent with the experimental value. The rule. This, in turn, confirms earlier 11B and 27Al NMR re-
−10 ppm peak in NBA433 can be assigned as a nonbridging sults,55 which suggest that aluminum atoms prefer to have
oxygen peak, which generally has relatively small quadrupolar boron atoms as next-nearest neighbors.
coupling constants (1.5–2.3 MHz).2 It is not clear at present In some aluminoborate glasses, 27Al NMR has revealed that
whether this peak is due to Al–O–Na, B–O–Na, or both. Future
experiments on borate–aluminate glasses with high boron:alu-
minum ratios may help resolve this ambiguity. ‡
Editors note in proof: J. F. Stebbins, S. K. Lee, and J. V. Oglesby, “Al–O–Al
Recently, high-resolution 17O NMR data have been collected Oxygen Sites in Crystalline Aluminates and Aluminosilicate Glasses: High-
on crystalline NaAlO2, in which all oxygens are bridging be- Resolution Oxygen-17 NMR Results,” Am. Mineral., in press.
June 1999 NMR Studies of Borate, Borosilicate, and Boroaluminate Glasses 1527

Fig. 9. Isotropic total projections of 17O MQMAS spectra for


KBS316 glass quenched at two different rates.

a significant fraction of the aluminum is present in AlO5 and


AlO 6 sites. 19,35,55 The presence of additional species
complicates the structure, probably increases overall disor-
der, and certainly introduces new types of oxygen sites,
e.g., bridging oxygens that connect two BO3, BO4, or AlO4
sites but are charge balanced by coordination with an AlO5
or AlO6 group.58 Our MQMAS spectra do not display extra
peaks for such sites, but their abundances are probably low—
recent 27Al NMR on similar compositions suggests that
high-coordinated aluminum is only 3%–8% of the total
aluminum.55

V. Conclusions and Future Work

Our 17O MQMAS NMR spectra for B2O3-containing glasses


have revealed a rich array of new structural information. We
reassigned the Si–O–B resonance and observed two types of
B–O–B resonances in B2O3 and a variety of borate and boro-
silicate compositions. Direct evidence of cation mixing be- Fig. 10. 17O MQMAS spectra for NBA351 and NBA433 sodium
tween three-coordinated boron and four-coordinated silicon is boroaluminate glasses. Al[B] denotes either Al or B. Box labeled
obtained. Dell and Bray’s model on sodium borosilicate glasses “Al–O–Al?” shows the likely position of the peak for this species,
needs to be extended to accommodate our new results. A boron based on recent data for crystalline NaAlO2 (see text).
avoidance rule is found to agree well with our experimental
results on sodium and potassium borosilicate and boroalumi-
nate glasses, when stoichiometry allows and when alkali con-
a greater range of samples could lead to a more conclusive
tent is moderate. In sodium boroaluminate glasses, aluminum
structural model. Fourth, combining high-resolution 17O NMR
atoms tend to be coordinated by boron atoms. Previous 11B
with data from other spectroscopic techniques and alternative
NMR fictive temperature studies are extended to include so-
NMR methods would potentially give a much better under-
dium borosilicate glasses with other compositions and potas-
standing of a variety of boron-containing glassy systems. Fifth,
sium borosilicate glasses. Temperature-dependent structural
and finally, and perhaps most importantly, is the issue of ac-
change in potassium borosilicate glasses is also observed
curate quantification of species abundances. MQMAS has in-
through 17O NMR.
herent limits in this regard, but NMR parameters derived from
Our study also points to several questions to be answered in
MQMAS may be very useful in helping to model less-resolved,
the future. First, molecular-level understanding of the NMR
but more quantitative, one-dimensional MAS spectra.
parameters observed here requires more careful quantum cal-
culations. Second, we have, so far, not been able to resolve
Acknowledgments: We would like to thank Dr. Zhi Xu for collecting
resonances from various nonbridging oxygen species, which X-ray diffraction data on all the samples and Sung Keun Lee for help with the
leaves some uncertainty in our assignment of some peaks. calculation of oxygen species abundance. Three anonymous reviewers provided
Studies of binary borate glasses with high alkali content and helpful comments and corrections.
glasses with Si:Al < 1.0 will help answer these questions. On
the other hand, multiple-field NMR experiments would make
the determination of NMR parameters more accurate and may References
1
S. Schramm and E. Oldfield, “High-Resolution Oxygen-17 NMR of Solids,”
resolve new oxygen environments. Third, our borosilicate glass J. Am. Chem. Soc., 106, 2502–506 (1984).
study has focused on samples with high-silicon content and, 2
J. F. Stebbins, “Nuclear Magnetic Resonance Spectroscopy of Silicates and
thus, have not covered the whole glass-forming region. Data on Oxides in Geochemistry and Geophysics”; pp. 303–32 in Handbook of Physical
1528 Journal of the American Ceramic Society—Wang and Stebbins Vol. 82, No. 6
31
Constants. Edited by T. J. Ahrens. American Geophysical Union, Washington, Y. H. Yun and P. J. Bray, “Nuclear Magnetic Resonance Studies of Glasses
DC, 1995. in the System Na2O–B2O3–SiO2,” J. Non-Cryst. Solids, 27, 363–71 (1978).
3 32
E. Oldfield, H. Kyung, C. Timken, B. Montez, and R. Ramachandran, S. Z. Xiao, “A Discussion about the Structural Model of ‘Nuclear Magnetic
“High-Resolution Solid-State NMR of Quadrupolar Nuclei,” Nature (London), Resonance Studies of Glasses in the System Na2O–B2O3–SiO2’,” J. Non-Cryst.
318, 163–65 (1985). Solids, 45, 29–31 (1981).
4 33
K. T. Mueller, B. Q. Sun, C. G. Chingas, J. W. Zwanziger, T. Terao, and A. K. Budhwani and S. Feller, “A Density Model for the Lithium, Sodium and
Pines, “Dynamic-Angle Spinning of Quadrupolar Nuclei,” J. Magn. Reson., 86, Potassium Borosilicate Glass Systems,” Phys. Chem. Glasses, 36, 183–88
470–76 (1990). (1995).
5 34
P. J. Grandinetti, “Dynamic-Angle Spinning and Applications”; pp. 1–13 in H. T. Kim, S. J. Chung, and M. J. Park, “Studies of the Structure of Li2O–
Encyclopedia of Nuclear Magnetic Resonance. Edited by D. M. Grant and R. K. B2O3–Al2O3 Glasses Using 11B NMR”; pp. 42–50 in Borate Glasses, Crystals
Harris. Wiley, New York, 1995. and Melts. Edited by A. C. Wright, S. A. Feller, and A. C. Hannon. Alden Press,
6
Y. Wu, B. Q. Sun, A. Pines, A. Samoson, and E. Lippmaa, “NMR Experi- Oxford, U.K., 1997.
35
ments with a New Double Rotor,” J. Magn. Reson., 89, 297–309 (1990). B. C. Bunker, R. J. Kirkpatrick, R. K. Brow, G. L. Turner, and C. Nelson,
7
K. T. Mueller, Y. Wu, B. F. Chmelka, J. Stebbins, and A. Pines, “High- “Local Structure of Alkaline-Earth Boroaluminate Crystals and Glasses: II, 11B
27
Resolution Oxygen-17 NMR of Solid Silicates,” J. Am. Chem. Soc., 113, 32–38 and Al MAS NMR Spectroscopy of Alkaline-Earth Boroaluminate Glasses,”
(1990). J. Am. Ceram. Soc., 74, 1430–38 (1991).
8 36
J. Baltisberger, P. Grandinetti, M. Eastman, A. Pines, I. Farnan, and J. F. Y. H. Yun, “Nuclear Magnetic Resonance Studies of the Glasses in the
Stebbins, “17O Dynamic Angle Spinning NMR Studies of Silicate Glasses,” System K2O–B2O3–P2O5,” J. Non-Cryst. Solids, 30, 45–52 (1978).
37
Eos, Trans. Am. Geophys. Union, 72, 571 (1991). P. J. Grandinetti, J. H. Baltisberger, A. Llor, Y. K. Lee, Y. Werner, M. A.
9
P. J. Grandinetti, J. H. Baltisberger, I. Farnan, J. F. Stebbins, U. Werner, and Eastman, and A. Pines, “Pure-Absorption-Mode Lineshapes and Sensitivity in
A. Pines, “Solid-State 17O Magic-Angle and Dynamic-Angle Spinning NMR Two-Dimensional Dynamic-Angle Spinning NMR,” J. Magn. Reson. A, 103,
Study of the SiO2 Polymorph Coesite,” J. Phys. Chem., 99, 12341–48 (1995). 72–81 (1993).
10 38
P. Florian, K. E. Vermillion, P. J. Grandinetti, I. Farnan, and J. F. Stebbins, W. Haller, D. H. Blackburn, F. E. Wagstaff, and R. J. Charles, “Metastable
“Cation Distribution in Mixed Alkali Disilicate Glasses,” J. Am. Chem. Soc., Immiscibility Surface in the System Na2O–B2O3–SiO2,” J. Am. Ceram. Soc.,
118, 3493–97 (1996). 53, 34–39 (1970).
11 39
I. Farnan, P. J. Grandinetti, J. H. Baltisberger, J. F. Stebbins, U. Werner, M. J. P. Amoureux, C. Fernandez, and L. Frydman, “Optimized Multiple-
Eastman, and A. Pines, “Quantification of the Disorder in Network Modified Quantum Magic-Angle Spinning NMR Experiments on Half-Integer Quadru-
Silicate Glasses,” Nature (London), 358, 31–35 (1992). poles,” Chem. Phys. Lett., 259, 347–55 (1996).
12 40
L. Frydman and J. S. Harwood, “Isotropic Spectra of Half-Integer Quadru- Z. Xu, H. Maekawa, J. V. Oglesby, and J. F. Stebbins, “Oxygen Speciation
polar Spins From Bidimensional Magic-Angle Spinning NMR.” J. Am. Chem. in Hydrous Silicate Glasses—An Oxygen-17 NMR Study,” J. Am. Chem. Soc.,
Soc., 117, 5367–68 (1995). 120, 9894–90 (1998).
13 41
A. Medek, J. S. Harwood, and L. Frydman, “Multiple-Quantum Magic- G. E. Jellison, L. W. Panek, P. J. Bray, and J. G. B. Rouse, “Determinations
Angle Spinning NMR: A New Method for the Study of Quadrupolar Nuclei in of Structure and Bonding in Vitreous B2O3 by Means of B10, B11, and O17
Solids,” J. Am. Chem. Soc., 117, 12779–84 (1995). NMR,” J. Chem. Phys., 66, 802–12 (1977).
14 42
S. H. Wang, Z. Xu, J. H. Baltisberger, L. M. Bull, J. F. Stebbins, and A. R. E. Youngman, S. T. Haubrich, J. W. Zwanziger, M. T. Janicke, and B. F.
Pines, “Multiple-Quantum Magic-Angle Spinning and Dynamic-Angle Spin- Chmelka, “Short- and Intermediate-Range Structural Ordering in Glassy Boron
ning NMR Spectroscopy for Quadrupolar Nuclei,” Solid State NMR, 8, 1–16 Oxide,” Science (Washington, DC ), 269, 1416–20 (1995).
43
(1996). R. J. Kirkpatrick, T. Dunn, S. Schramm, K. A. Smith, R. Oestrike, and G.
15
D. Massiot, B. Touzo, D. Trumeau, J. P. Coutures, J. Virlet, P. Florian, and Turner, “Magic-Angle Sample-Spinning Nuclear Magnetic Resonance Spec-
P. J. Grandinetti, “Two-Dimensional Magic-Angle Spinning Isotropic Recon- troscopy of Silicate Glasses: a Review”; pp. 302–27 in Structure and Bonding
struction Sequences for Quadrupolar Nuclei,” Solid State NMR, 6, 73–84 in Noncrystalline Solids. Edited by G. E. Walrafen and A. G. Revesz. Plenum
(1996). Press, New York, 1986.
16 44
J. H. Baltisberger, Z. Xu, J. F. Stebbins, S. Wang, and A. Pines, “Triple- J. A. Tossell, “A Theoretical Study of the Molecular Basis of the Al Avoid-
Quantum Two-Dimensional 27Al Magic-Angle Spinning Nuclear Magnetic ance Rule and of the Spectral Characteristics of Al–O–Al Linkages,” Am. Min-
Resonance Spectroscopic Study of Aluminosilicate and Aluminate Crystals and eral., 78, 911–19 (1993).
45
Glasses,” J. Am. Chem. Soc., 118, 7209–14 (1996). K. L. Geisinger, R. Oestrike, A. Navrotsky, G. L. Turner, and R. J. Kirk-
17
C. Fernandez and J. P. Amoureux, “2D Multiquantum MAS-NMR Spec- patrick, “Thermochemistry and Structure of Glasses along the Join NaAlSi3O8–
troscopy of 27Al in Aluminophosphate Molecular Sieves,” Chem. Phys. Lett., NaBSi3O8,” Geochim. Cosmochim. Acta, 52, 2405–14 (1988).
46
242, 449–54 (1995). F. C. Hawthorne, P. C. Burns, and J. D. Grice, “The Crystal Chemistry of
18
J. V. Hanna, M. E. Smith, and H. J. Whitfield, “Multiple Quantum Magic Boron”; pp. 41–116 in Boron: Mineralogy, Petrology and Geochemistry. Edited
Angle Spinning NMR Detection of Impurity Phases in Na2HfO3,” J. Am. Chem. by E. S. Grew and L. M. Anovitz. Mineralogy Society of America, Washington,
Soc., 118, 5772–78 (1996). DC, 1996.
19 47
S. Sen, Z. Xu, and J. F. Stebbins, “Temperature Dependent Structural W. Loewenstein, “The Distribution of Aluminum in the Tetrahedra of
Changes in Borate, Borosilicate and Boroaluminate Liquids: High-Resolution Silicates and Aluminates,” Am. Mineral., 39, 92–100 (1954).
11
B, 29Si, and 27Al NMR Studies,” J. Non-Cryst. Solids, 226, 29–40 (1998). 48
J. A. Tossell and G. Saghi-Szabo, “Aluminosilicate and Borosilicate Single
20
Z. Xu and J. F. Stebbins, “Oxygen Sites in the Zeolite, Stilbite: A Com- 4-Rings; Effects of Counterions and Water on Structure, Stability, and Spectra,”
parison of Static, MAS, VAS, DAS, and Triple Quantum MAS NMR Tech- Geochim. Cosmochim. Acta, 61, 1171–79 (1997).
49
niques,” Solid State NMR, 11, 243–51 (1997). J. F. Stebbins and S. Sen, “Microscopic Dynamics and Viscous Flow in a
21
Z. Xu and J. F. Stebbins, “Oxygen Site Exchange Kinetics Observed with Borosilicate Glass-Forming Liquid,” J. Non-Cryst. Solids, 224, 80–85 (1997).
50
Solid State NMR in a Natural Zeolite,” Geochim. Cosmochim. Acta, 62, 1803– J. Zhong, X. Wu, M. L. Liu, and P. J. Bray, “Structural Modeling of
809 (1997). Lithium Borosilicate Glasses via NMR Studies,” J. Non-Cryst. Solids, 107,
22
J. F. Stebbins and Z. Xu, “NMR Evidence for Excess Non-Bridging Oxy- 81–88 (1988).
51
gens in an Aluminosilicate Glass,” Nature (London), 390, 60–62 (1997). R. J. Araujo, “Temperature Dependence of Boron Coordination in Alkali
23
J. F. Stebbins, J. V. Oglesby, and Z. Xu, “Disorder among Network Modi- Borate Glasses as an Example of a Second Order Transition,” Phys. Chem.
fier Cations in Silicate Glasses: New Constraints from Triple-Quantum Oxygen- Glasses, 21, 193–96 (1980).
52
17 NMR,” Am. Mineral., 82, 1116–24 (1997). R. J. Araujo, “Statistical Mechanics of Chemical Disorder: Application to
24
P. J. Dirken, S. C. Kohn, M. E. Smith, and E. R. H. v. Eck, “Complete Alkali Borate Glasses,” J. Non-Cryst. Solids, 58, 201–208 (1983).
53
Resolution of Si–O–Si and Si–O–Al Fragments in an Aluminosilicate Glass by P. K. Gupta, M. L. Lui, and P. J. Bray, “Boron Coordination in Rapidly
17
O Multiple Quantum Magic Angle Spinning NMR Spectroscopy,” Chem. Cooled and in Annealed Aluminum Borosilicate Glass Fibers,” J. Am. Ceram.
Phys. Lett., 266, 568–74 (1997). Soc., 68, C-82 (1985).
25 54
J. P. Amoureux, F. Bauer, H. Ernst, C. Fernandez, D. Freude, D. Michel, J. F. Stebbins and S. E. Ellsworth, “Temperature Effects on Structure and
and U. T. Pingel, “17O Multiple-Quantum and 1H MAS NMR Studies of Zeolite Dynamics in Borate and Borosilicate Liquids: High-Resolution and High-
ZSM-5,” Chem. Phys. Lett., 285, 10–13 (1998). Temperature NMR Results,” J. Am. Ceram. Soc., 79, 2247–56 (1996).
26 55
S. H. Wang and J. F. Stebbins, “On the Structure of Borosilicate Glasses: A L. Züchner, J. C. C. Chan, W. M. Warmuth, and H. Eckert, “Short Range
Triple-Quantum Magic-Angle Spinning 17O NMR Study,” J. Non-Cryst. Solids, Order and Site Connectivity in Sodium Boroaluminate Glasses: I. Quantification
231, 286–90 (1998). of Local Environments by High-Resolution 11B, 23Na, and 27Al Solid State
27
W. J. Dell, P. J. Bray, and S. Z. Xiao, “11B NMR Studies and Structural NMR,” J. Phys. Chem., B102, 4495–501 (1998).
56
Modeling of Na2O–B2O3–SiO2 Glasses of High Soda Content,” J. Non-Cryst. L. van Wüllen, B. Gee, L. Zuchner, M. Bertmer, and H. Eckert, “Connec-
Solids, 58, 1–16 (1983). tivities and Cation Distributions in Oxide Glasses: New Results from Solid-
28
P. J. Bray, “Structural Models for Borate Glasses.” J. Non-Cryst. Solids, 75, State NMR,” Ber. Bunsen-Ges. Phys. Chem., 100, 1539–49 (1996).
57
29–36 (1985). L. van Wüllen, L. Züchner, W. Müller-Warmuth, and H. Eckert,
29
W. L. Konijnendijk and J. M. Stevels, “The Structure of Borosilicate “11B{27Al} and 27Al{11B} Double Resonance Experiments on a Glassy Sodium
Glasses Studied by Raman Spectroscopy,” J. Non-Cryst. Solids, 20, 193–201 Aluminoborate,” Solid State NMR, 6, 203–12 (1996).
58
(1976). B. C. Bunker, R. J. Kirkpatrick, and R. K. Brow, “Local Structure of Al-
30
B. C. Bunker, D. R. Tallant, R. J. Kirkpatrick, and G. L. Turner, “Nuclear kaline-Earth Boroaluminate Crystals and Glasses: I, Crystal Chemical Con-
Magnetic Resonance and Raman Investigation of Sodium Borosilicate Glass cepts-Structural Predictions and Comparisons to Known Crystal Structures,” J.
Structures,” Phys. Chem. Glasses, 31, 30–40 (1990). Am. Ceram. Soc., 74, 1425–29 (1991). 䊐

Das könnte Ihnen auch gefallen