The Structure of Simple Phosphate Glasses

Journal of Non-Crystalline Solids 263&264 (2000) 1±28
www.elsevier.com/locate/jnoncrysol
Section 1. Structure
Review: the structure of simple phosphate glasses

Richard K. Brow *
Department of Ceramic Engineering, University of Missouri-Rolla, Rolla, MO 65409-0330, USA
Abstract
Recent developments of phosphate glasses for a variety of technological applications, from rare-earth ion hosts for
solid state lasers to low temperature sealing glasses, have led to renewed interest in understanding the structures of these
unusual materials. In this review, spectroscopic and diraction studies of simple phosphate glasses, including v-P2 O5
and binary phosphate compositions, are described. Special attention is given to the structures of anhydrous ultra-
phosphate glasses, which have received close attention from the glass community only in the past six years. Ó 2000
Published by Elsevier Science B.V. All rights reserved.
1. Introduction By studying such materials, Van Wazer [2] estab-

lished the foundations for much of our present
Phosphate glasses with low dispersion and rel- understanding of the nature of phosphate glasses.
atively high refractive indices (compared with sil- About the same time, Kordes and co-workers [3,4]
icate-based optical glasses) were developed for re-examined the alkaline earth phosphate glasses,
achromatic optical elements about 100 yr ago by including UV-transmitting compositions, and
Schott and co-workers. Subsequent interest in al- noted some ÔanomalousÕ trends in properties which
kaline earth phosphate glasses stemmed from their they suggested showed a compositional-depen-
high transparency for ultraviolet (UV) light, again dence for the coordination number of metal ca-
when compared with silicate glasses. However, the tions like Zn2 .
poor chemical durability of these early optical The advent of solid state lasers in the 1960s
glasses limited their applications and (temporarily) heralded a new era of phosphate glass research.
discouraged their further development. (See [1] for Certain compositions have large rare-earth stim-
a description of these early studies and citations to ulated emission cross-sections and low thermo-
them.) optical coecients (compared with silicate glasses)
In the 1950s, interest in amorphous alkali and are the materials of choice, particularly for
phosphates was stimulated by their use in a variety high power laser applications [5]. (Campbell and
of industrial applications, including sequestering Suratwala review the development of Nd-doped
agents for hard water treatments and dispersants phosphate laser glasses elsewhere in this volume
for clay processing and pigment manufacturing [2]. [6].)
More recently, phosphate glasses have been
developed for a variety of specialty applications.
*
Tel.: +1-573 341 6812; fax: +1-573 341 6934. Alkali aluminophosphate compositions have glass
E-mail address: brow@umr.edu (R.K. Brow). transition temperatures under 400°C and thermal
0022-3093/00/$ - see front matter Ó 2000 Published by Elsevier Science B.V. All rights reserved.
PII: S 0 0 2 2 - 3 0 9 3 ( 9 9 ) 0 0 6 2 0 - 1
2 R.K. Brow / Journal of Non-Crystalline Solids 263&264 (2000) 1±28

expansion coecients greater than 150 10ÿ7 = C electrons (see, e.g., [15,16]). These tetrahedra link
and so are used for specialty hermetic seals [7]. through covalent bridging oxygens to form various
Zinc phosphate compositions are chemically du- phosphate anions. The tetrahedra are classi®ed
rable, have processing temperatures under 400°C using the Qi terminology [17], where ÔiÕ represents
and can be co-formed with high temperature the number of bridging oxygens per tetrahedron
polymers to produce unusual organic/inorganic (shown schematically in Fig. 1). The networks of
composites [8]. The chemical durability and low phosphate glasses can be classi®ed by the oxygen-
processing temperature of iron phosphate glasses to-phosphorus ratio, which sets the number of
have led to their development as nuclear waste tetrahedral linkages, through bridging oxygens,
hosts [9]. Biocompatible phosphate glasses and between neighboring P-tetrahedra [2].
glass-ceramics have medical applications [10] and Zachariasen [18] identi®ed vitreous P2 O5 as one
amorphous lithium phosphate [11] and phospho- of the prototypical Ôrandom networkÕ glass form-
rus oxynitride glasses [12] have fast ion conduc- ers. Indeed, recent diraction studies of v-P2 O5 ,
tivity that make them useful as solid state described below, are consistent with the Zacha-
electrolytes. riasen description of an open, distorted network of
The properties that make phosphate glasses Q3 tetrahedra. Hagg [19], however, challenged
candidates for so many dierent applications are ZachariasenÕs idea that glass forming tendency
related to their molecular-level structures. There relied upon the development of a random, three-
have been many excellent reviews of structural dimensional network, using the metaphosphate
studies on phosphate glasses, including those of composition as an example of glass formation in
Van Wazer [2], Abe [13] and Martin [14]. Since which a large one-dimensional ÔmolecularÕ group
MartinÕs 1991 review, a great deal of new struc- (based on Q2 tetrahedra) inhibits crystallization.
tural information about phosphate glasses has The chromatography studies of Van Wazer [2] and
been reported. For the ®rst time, the structures of others since have reinforced the utility of the de-
anhydrous ultraphosphate compositions have been scription of phosphate glass structures in terms of
determined, as have the structures of polyphos- distributions of molecular species such as those
phate compositions with very low ( < 40 mol%) proposed by Hagg.
P2 O5 contents. New structural probes, including These two descriptions of glass structure are not
solid state nuclear magnetic resonance (NMR) incompatible; both predict local bonding similari-
spectroscopy, X-ray absorption spectroscopy ties in glasses and crystals of identical composi-
(XAS), and neutron diraction analyses, have tion. (That is, a metaphosphate glass and crystal
provided unprecedented detail about the bonding will both have structures based on linkages of Q2
arrangements in glasses, from simple alkali and tetrahedra.) More importantly for the present re-
alkaline earth phosphates, through more complex view, these two models also illustrate the need to
and technologically useful, multicomponent describe Ôglass structureÕ at dierent length scales.
glasses. These more recent studies, particularly Having information about the local bond ar-
those published since 1991, will be emphasized in rangements of the glass forming (or modifying)
this review. polyhedra will not necessarily complete the de-
scription of a glass structure. That short-range
information (generally de®ned by the immediate
2. Background
The basic building blocks of crystalline and

amorphous phosphates are the P-tetrahedra that
result from the formation of sp3 hybrid orbitals by
the P outer electrons (3s2 3p3 ). The ®fth electron is
promoted to a 3d orbital where strong p-bonding Fig. 1. Phosphate tetrahedral sites that can exist in phosphate
molecular orbitals are formed with oxygen 2p glasses.
R.K. Brow / Journal of Non-Crystalline Solids 263&264 (2000) 1±28 3
coordination environment of an ion; e.g., see [20]) sublimation) that is then typically melted in sealed
must also be coupled with longer range informa- silica ampoules (see [25] for details).
tion, including how those polyhedra are linked to Recent neutron diraction studies [26,27] con-
form larger structures, whether ordered or disor- ®rm that Q3 P-tetrahedra are the basic building
dered, before a proper structural description is blocks of v-P2 O5 . Fig. 2 (from Ref. [26]) shows the
available. The studies reviewed below describe real-space correlation function, T(r), of v-P2 O5 .
both the short-range and longer-range structures There are two P±O contributions to T(r) near 1.5
of phosphate glasses. and the second is three times more intense than
A
Phosphate glasses can be made with a range of the ®rst, consistent with their assignment to P±TO
structures, from a cross-linked network of Q3 tet- and P±BO bonds, respectively [26]. Table 1 com-
rahedra (vitreous P2 O5 ) to polymer-like meta- pares bond lengths determined from neutron dif-
phosphate chains of Q2 tetrahedra to ÔinvertÕ fraction studies of v-P2 O5 with those from
glasses based on small pyro- (Q1 ) and orthophos- diraction studies of the three crystalline poly-
phate (Q0 ) anions, depending on the [O]/[P] ratio morphs. In general, the P±TO bond lengths are
as set by glass composition. This review is likewise shorter and the P±BO bond lengths are longer in
organized in this order. the glass than in the dierent crystalline
polymorphs.
Longer-range information about how the Q3
3. Vitreous P2 O5 ([O]/[P] 2.5) tetrahedra are linked can also be obtained from
diraction studies. The neutron scattering factor,
There are a number of polymorphs of crystal- S(Q), from v-P2 O5 and several xNa2 O(1)x)P2 O5
line P2 O5 , including a hexagonal form (I) based on glasses are shown in Fig. 3 [27]. Of interest are the
discrete P4 O10 molecules [21], and two ortho- two peaks at smaller S(Q) from the v-P2 O5 struc-
rhombic forms (II [22] and III [23]) based on he- ture factor, identi®ed by the arrows at Q 1.30
lices of [PO4 ] tetrahedra, each sharing three and 2.12 A ÿ1 . These peaks represent intermediate
corners. The tetrahedra in each polymorph are range order in the P±O network corresponding to
linked to three other tetrahedra through bridging periodicities of 4.8 and 3.0 A, respectively, similar
(P±O±P) oxygens. The fourth bond is to a terminal to the intermolecular distances in crystalline P4 O10
oxygen P¸O and is signi®cantly shorter than the (H±P2 O5 (I)). Suzuya et al. [27] contend that ori-
others, a consequence of its enhanced p-bond entation correlations of P4 O10 molecular units in
character, as con®rmed by recent ab initio calcu- the glass account for these features in the scatter-
lations [24]. Table 1 lists the average lengths for ing factor. Hoppe et al. note that the distribution
bridging (BO) and terminal (TO) P±O bonds for of O±O distances in v-P2 O5 (centered at 2.5 A,
the dierent P2 O5 polymorphs. Fig. 2) cannot be explained exclusively by the
Until recently, little detailed information about presence of P4 O10 molecular units. On the other
the structure of vitreous P2 O5 was available. P2 O5 hand, based on a similar double maximum in the
is hygroscopic and volatile. Anhydrous glasses X-ray scattering factor from Mg-ultraphos-
must be prepared using dry-box handling proce- phate glasses, Walter et al. [28] proposed that
dures from crystalline P2 O5 (puri®ed by vacuum- the structure of O0 ±P2 O5 (III) was a better model
Table 1
Average bond lengths for P2 O5 polymorphs (adapted from Refs. [26,27])
Phase Density (g/cm3 )
P±BO (A)
P±TO (A) Reference
H±P2 O5 (I) 2.28 1.590 1.433 [21]
O±P2 O5 (II) 2.705 1.576 1.445 [22]
O0 ±P2 O5 (III) 2.928 1.570 1.445 [23]
v-P2 O5 2.445 1.581 1.432 [26]
v-P2 O5 2.385 1.579 1.420 [27]
Fig. 2. Real-space correlation function from a neutron dif-

fraction study of v-P2 O5 . Dots represent the experimental curve
and the solid line (upper curve) marks a ®tted model. The lower
plot reproduces the Gaussian peaks used to obtain P±O
(Table 1) and O±O distances. (The ®gure is from Ref. [26].)
for v-P2 O5 . They propose that the glass is made up

of corrugated sheets of six-member rings of Q3
tetrahedra. The two peaks in the structure factor
are then ascribed to distances between (eectively)
aligned P¸O bonds within the rings (shorter) and Fig. 3. Neutron structure factors from v-P2 O5 (top) and a series
between the sheets (longer). Hoppe et al. discuss of xNa2 O(1)x)P2 O5 glasses. The curves were displaced upward
the diraction studies of v-P2 O5 and other phos- by 0.5 for clarity. (The ®gure is from Ref. [27].)
phate glasses elsewhere in this volume [29].
Some of the details about the Q3 tetrahedra in v-P2 O5 and from O0 ±P2 O5 (III). The Raman
v-P2 O5 can also be obtained by various spectro- spectra are dominated by bands near 1380 cmÿ1 ,
scopic techniques. Vibrational spectroscopy has due to P±TO stretching modes, and at 630 cmÿ1 ,
long been used to investigate the short-range due to symmetric P±O±P stretching modes. The
structures of phosphate glass. Galeener and intense IR absorption band, centered near 970
Mikkelsen [30] reported the ®rst Raman spectra cmÿ1 is assigned to asymmetric P±O±P stretching
from anhydrous v-P2 O5 . By comparing those modes. This IR spectrum is similar to that re-
spectra with ones collected from various gas ported by Hudgens and Martin [25] for anhydrous
phases, they were able to con®rm the tetrahedral v-P2 O5 . Meyer notes that the shift to greater
structure of the phosphate building blocks and in wavenumbers for the P±TO band is due to a
so doing, associated the 1390 cmÿ1 Raman peak shorter P±TO bond length in v-P2 O5 compared to
(discussed below) with the short P±TO bond. O0 ±P2 O5 . This dierence agrees with the diraction
Galeener's assignments were con®rmed by the in- results shown in Table 1, and led Meyer to the
frared and Raman spectra calculated from an ab conclusion that the crystalline structure of O0 ±
initio study of phosphate clusters [31] and in more P2 O5 is not similar to the extended range structure
recent Raman studies of v-P2 O5 and series of alkali of v-P2 O5 . Meyer rules out H±P2 O5 (I) as a proper
[32,33] and Zn-ultraphosphate glasses [34]. From structural model because it is the low temperature
the latter study [34], Fig. 4 shows the Raman polymorph, and so considered unlikely because of
(top) and infrared (bottom) spectra collected from the high temperature for glass processing.
ordering. That the density of v-P2 O5 is intermedi-

ate to those of the H- and O-forms of crystalline
P2 O5 (Table 1) indicates that the glass likely con-
tains longer-range structural similarities to both
forms.
4. Binary phosphate glasses
The addition of a modifying oxide to v-P2 O5

results in the creation of non-bridging oxygens at
the expense of bridging oxygens. The resulting
depolymerization of the phosphate network with
the addition, e.g., of alkali oxide, R2 O, can be
described by the pseudo-reaction [2]
2Qn R2 O ! 2Qnÿ1 : 1
Fig. 4. Raman (top set) and infrared absorption spectra (bot-
tom set) collected from anhydrous samples of v-P2 O5 (top) and For binary xR2 O (or R0 O) (1)x)P2 O5 glasses,
crystalline O0 ±P2 O5 (bottom). (The ®gure is from Ref. [34].) the concentrations of Qi tetrahedra (Fig. 1) gen-
erally follow simple predictions based on the glass
Solid state 31 P nuclear magnetic resonance composition. In the ultraphosphate region
(NMR) spectroscopy also provides information 0 6 x 6 0:5, the fraction of Q2 and Q3 tetrahedra
about the local structure of phosphate glasses [35]. are given by [2]
Grimmer and Wolf [36] collected 31 P NMR spectra ÿ x
from the three crystalline P2 O5 polymorphs and f Q2 ; 2a
1ÿx
from v-P2 O5 . For each sample, they found axial
symmetric tetrahedra with large shielding tensors ÿ 1 ÿ 2x
(Dd) due primarily to very shielded ()240 to )260 f Q3 : 2b
1ÿx
ppm) r33 tensor elements. For axial symmetry, r33
lies along or near the P±TO bond, consistent with Anhydrous ultraphosphate glasses must also be
the expected greater shielding associated with the prepared in sealed ampoules because of the hyg-
localized electron density. Hudgens [37] examined roscopicity and volatility of P2 O5 . Because of
a drier v-P2 O5 sample and reports a 31 P isotropic preparation diculties, detailed spectroscopic in-
chemical shift, relative to H3 PO4 , of )54.2 ppm, formation about ultraphosphate structures has
vs. )38.1 ppm reported by Grimmer and Wolf ± only recently been obtained, as described below.
and drew similar conclusions, despite a greater The reactivity of Q3 sites to water limits the tech-
degree of deviation from axial symmetry. nological usefulness of ultraphosphate glasses even
In summary, the basic building block for the though they possess UV edges at wavelengths
structure of P2 O5 glass is the Q3 tetrahedron, shorter than found for silicate glasses [38].
possessing three covalent bridging oxygen bonds Metaphosphate glasses x 0:50 have net-
to neighboring tetrahedra, and one short (P±TO) works based entirely on Q2 tetrahedra that form
bond to a terminal oxygen. These tetrahedra are chains and rings. The chains and rings are linked
linked to form a three-dimensional network by more ionic bonds between various metal ca-
that appears to possess some similarities to the tions and the non-bridging oxygens. Because of the
structure of P4 O10 molecules found in crystalline diculty in obtaining an exact stoichiometry, these
H±P2 O5 (phase I), although there remains uncer- compositions are usually more accurately
tainties as to the details of such extended range described as long-chained polyphosphates (see
below). The term ÔmetaphosphateÕ is reserved for techniques [42]. These latter glasses are analogous
cyclic anions of the exact POÿ 3 n composition [39]. to the silicate ìnvert glasses', stable compositions
However, the convention is to label compositions with [O]/[Si] > 3 [43]. Such glasses do not have a
with x 0:50 Ômetaphosphate glassÕ and that continuous random network of glass-forming
convention will be followed here. polyhedra, at least as described by Zachariasen
Polyphosphate glasses (x > 0:50) have networks [18]. Instead, the glass structure consists of isolated
based on Q2 chains terminated by Q1 tetrahedra tetrahedra and small molecular fragments linked
(one bridging and three non-bridging oxygens per through their non-bridging oxygens by larger,
tetrahedron). The average chain length becomes smaller valence (`non-glass forming') cations.
progressively shorter as the [O]/[P] ratio increases. For binary glasses, the bridging-to-terminal
At O=P 3:5 (the pyrophosphate stoichiometry, oxygen ratio, for 0 6 x 6 0:75, depends on com-
x 0:67), the network structure is dominated by position according to
phosphate dimers, two Q1 tetrahedra linked by a BO=TO 0:5 3 ÿ 4x: 3
common bridging oxygen. Glasses for which [O]/
[P] > 3.5 contain isolated Q0 (orthophosphate units) As will be shown below, the predictions repre-
tetrahedra. Between the metaphosphate x 0:50 sented by Eqs. (1)±(3) can be tested using probes
and pyrophosphate x 0:67 boundaries, the such as solid state 31 P magic angle spinning nu-
fraction of Q1 and Q2 tetrahedra are given by [2] clear magnetic resonance (MAS NMR) spectros-
ÿ 2x ÿ 1 copy, X-ray photoelectron spectroscopy (XPS),
f Q1 ; 2c and chromatography, to provide quantitative
1ÿx
descriptions of the phosphate glass network
ÿ 2 ÿ 3x structures.
f Q2 : 2d
1ÿx
4.1. Ultraphosphate glasses (2.56[O]/[P] < 3.0)
Between the pyrophosphate x 0:67 and or-
thophosphate (x 0:75) boundaries, the fraction The hygroscopic nature of ultraphosphate
of Q0 and Q1 tetrahedra are given by glasses must be taken into account when analyzing
ÿ 3x ÿ 2 structure and property trends. Binary sodium ul-
f Q0 ; 2e traphosphate glasses prepared in open crucibles
1ÿx
will absorb water; e.g., over 20 mol% H2 O is pre-
ÿ 3 ÿ 4x sent as P±OH bonds in a glass with the nominal
f Q1 : 2f composition 25 mol% Na2 O and 75 mol% P2 O5
1ÿx
[44]. Water is an additional structural modi®er
Much research has been done on the structure that depolymerizes the ultraphosphate network
and properties of metaphosphate and polyphos- (Eq. (1)) beyond that expected from the nominal
phate glasses and details can be found in a number alkali-to-phosphorus ratio. Fig. 5 compares the
of review papers (e.g., [13,14,40]). Glass formation densities of Na-ultraphosphate glasses prepared by
in binary xR2 O (or R0 O) (1)x)P2 O5 systems conventional open crucible techniques [45,46] with
(where R2 O is an alkali oxide and R0 is an alkaline those prepared using anhydrous raw materials
earth oxide) is typically limited to x < 0.55±0.60 melted in sealed ampoules [37]. Although some of
(mole fraction) for conventional melt processing. the density dierences between the series might be
Rapid quenching techniques have yielded Li- ascribed to ®ctive temperature eects due to the
phosphate glasses with x as large as 0.70 ([O]/ dierent preparation conditions, the density trends
[P] 3.67), having structures based on Q1 and Q0 for the two series dier markedly. Anhydrous Na-
tetrahedra [41]. Pyrophosphate glasses based on a ultraphosphate glasses have a minimum in density,
variety of low-coordinated (coordination number near 20 mol% Na2 O that does not occur when
6 6) metal oxides, including SnO, ZnO, CdO, PbO glasses are prepared in open crucibles. (Bartenev
and Fe2 O3 , can be made by conventional melt [47] reported a minimum in the speed of
Fig. 5. Densities of Na-ultraphosphate glasses prepared in

open crucibles [45,46] and in sealed ampoules [37]. The former
will possess signi®cant amounts of structural water that further
depolymerize the ultraphosphate network. Lines are drawn as
guides for the eye.
ultrasound in Na-ultraphosphate glasses near 25

mol% Na2 O and related this to an analogous
minimum in density.) The densities (and glass Fig. 6. 31 P NMR spectra from anhydrous Na phosphate
glasses. Isotropic peaks that are due to dierent P-tetrahedra
transition temperatures) of anhydrous Li-ultra-
are labeled with the appropriate Qi notation; the remaining
phosphate glasses have similar compositional de- peaks are spinning sidebands. The inset shows the range of
pendencies [25]. chemical shifts reported for sodium phosphate glasses and
Care must also be taken when analyzing anhy- crystals. (The ®gure is from Ref. [35].)
drous ultraphosphate glasses prepared using
Q1 (0 ppm) tetrahedra are evident in the various
sealed ampoule techniques. Meyer et al. [48] have
spectra. The decrease in shielding (less negative
shown that 30 s of exposure to ambient air is en-
chemical shift) with decreasing polymerization
ough to degrade the IR re¯ectance spectra from
(fewer bridging oxygens per tetrahedron) is due to
(originally) anhydrous v-P2 O5 . The hydration of
the increased ionicity of the non-bridging oxygen
alkaline earth ultraphosphates is slower, but still
bonds [44]. The chemical shifts of the individual
evident.
peaks are also sensitive to the glass composition.
Each peak becomes less shielded with increasing
4.1.1. Phosphate site distributions in ultraphosphate Na2 O content. Similar changes in isotropic chem-
glasses ical shifts have been noted in other NMR studies
The evolution of the cross-linked Q3 network of of binary phosphate glasses [35] and indicate a
P2 O5 to the chain-like Q2 network of a meta- redistribution of electron density among tetrahe-
phosphate glass can be readily seen in the 31 P dra with changing bridging to non-bridging oxy-
NMR spectra collected from the series of Na- gen ratios. One-dimensional 31 P NMR studies
phosphate glasses shown in Fig. 6 [35]. The dif- have been reported for a variety of anhydrous bi-
ferent Qi tetrahedra have isotropic 31 P NMR nary ultraphosphate glass systems [32,37,49,50]
chemical shift peaks that are separated by ca. 20 and for ultraphosphate glasses that contain sig-
ppm (inset to Fig. 6). Peaks due to Q3 (isotropic ni®cant (>5 mol%) amounts of structural water
chemical shift at )51 ppm), Q2 ()22 ppm), and [44,51].
Such spectra can be used to test the site-distri- without hydroxyl bonds and that identi®es both
bution models outlined above. Fig. 7 shows the intra-and intermolecular hydrogen bonding. This
tetrahedral site distributions from a number of 31 P `structurally complex' incorporation of water
NMR studies of dierent ultraphosphate glasses. supports the conclusions drawn from earlier IR
The solid lines are the predicted distributions from and 1 H NMR analyses [53]. Eqs. (2a) and (2b)
Eqs. (2a) and (2b). are consistent with the measured Q-distributions
Also shown in Fig. 7 are the Q3 site fractions when ÔxÕ is de®ned as R2 O H2 O [44]. In this
(shown as asterisks) determined from Na-ultra- way, P±OH bonds will aect ultraphosphate glass
phosphate glasses melted in open crucibles [44], properties, such as the density data in Fig. 5. The
plotted against their analyzed Na2 O content. presence of P±OH bonds likely explains the
These glasses contain large amounts of water (5± smaller than expected Q3 /Q2 ratios reported in a
31
20 mol%) which also depolymerize the phosphate P NMR study of Zn- and Pb-ultraphosphate
network by forming P±OH bonds and so reduc- glasses [54]. Those glasses were prepared in open
ing the Q3 fraction to much less than that ex- crucibles with water-producing raw materials
pected from the Na2 O-content alone. That is, such as NaH2 PO4 á H2 O.
bridging oxygens associated with Q3 tetrahedra The spectra in Fig. 6 indicate that only Q2 and
react with H2 O to form Q2 tetrahedra with P± 3
Q tetrahedra are present in the ultraphosphate
OH bonds. Mercier et al. [51] show that 1 H±31 P range; this is typical for ultraphosphate glasses and
cross polarization NMR spectroscopy can dis- indicates that no site disproportionation of the
tinguish between Q2 tetrahedra with and without type
P±OH bonds in a series of `wet' Zn-ultraphos-
phate glasses. Wenslow and Mueller [52] report a 2Q2 $ Q1 Q3 4
similar study of water-containing (near) meta-
phosphate glass that reveals Q1 sites with and occurs in these glasses. The absence of Q1 sites
implies that the pseudo-reaction indicated by
Eq. (4) would have an equilibrium constant
kQ2 0, in agreement with Van Wazer's predictions
for ultraphosphate glasses [2]. (There has been one
report for Q2 disproportionation kQ2 > 0 in a Ca-
metaphosphate composition [55].) Site dispropor-
tionation is important for the structures of poly-
phosphate glasses, as described below.
Neutron and X-ray diraction studies of Ca-
and Zn-ultraphosphate glasses indicate that the
average P±TO and P±BO bond lengths increase
with increasing metal oxide (MeO) contents
[26,66]. The mean P±O bond length does not
change within experimental errors in the same se-
ries because the fraction of shorter P±TO bonds
increases (Fig. 8 [26]). These studies could not re-
solve dierent P±TO bond lengths from Q2 and Q3
tetrahedra coexisting in ultraphosphate glasses.
This latter observation is not surprising since
Fig. 7. Q2 (closed symbols) and Q3 (open symbols) site distri- HoppeÕs survey of diraction studies of ultra-
butions from 31 P NMR studies of anhydrous Na (s) [32], Zn
phosphate crystals indicates that such a dierence
(h) [34], Ca (n), Sr (Ñ) and Ba (e) [50] ultraphosphate glasses. [56], too small to
The solid lines are the predicted distributions (Eqs. (2a) and would be on the order of 0.01 A
(2b)). The asterisks are Q3 site distributions from water- be evident in the broad diraction peaks from
containing Na-ultraphosphate glasses [44]. glasses.
There has been some controversy about the bridging' oxygens [14]. For example, a Q2 tetra-
dierent oxygen species in phosphate glasses, hedron is often represented as possessing one
particularly as it relates to the relevance of P¸O and one P±Oÿ R terminal oxygen bond, the
distinguishing Ôdouble bondedÕ oxygens from `non- implication being that the double bond is shorter
because of the localization of the ®fth valence
electron from P5 . This description was supported
by an early X-ray photoelectron study [57] in
which the O1s spectra from various phosphate
glasses were decomposed into three components,
representing distinct P¸O; P±Oÿ R and P±O±P
species.
Subsequent diraction studies of metaphos-
phate glasses [58,59] challenge this assumption,
revealing instead that the two P±TO bonds on a Q2
tetrahedron are indistinguishable. Ab initio mo-
lecular orbital calculations indicate that the nega-
tive charge associated with the ®fth valence
electron of P5 is almost equally delocalized over
each non-bridging oxygen on the Q2 tetrahedron
[24,60]. Given these recent results, the proper way
to decompose the O1s spectra from simple phos-
phate glasses then seems to be the two-peak ®t,
®rst proposed by Gresch et al. [61].
Fig. 8. Comparison of the P±TO and P±BO bond lengths of
v-P2 O5 (h) with those of Ca- (s) and Zn-ultraphosphates (n). Examples of the O1s spectra collected from
(Adapted from Ref. [26].) The asterisks are predicted bond anhydrous sodium ultraphosphate glasses [32] are
lengths from ab initio calculations [24]. shown in Fig. 9 (left). The greater binding energy
Fig. 9. O1s photoelectron spectra collected from anhydrous Na-ultraphosphate glasses (left) and the bridging-to-terminal oxygen ratio
as a function of Na2 O content (right), determined from the O1s spectra as well as Qi -distributions from 31 P NMR spectra. The dashed
line is the expected ratio, from Eq. (3). (Adapted from [32].)
peak is due to bridging oxygens (P±BO) and the

other binding energy peak is due to terminal
oxygens (P±TO) on the phosphate tetrahedra.
Note that the separation between the two binding
energies decreases, from 2.56 eV for v-P2 O5 , to
1.90 eV for Na-metaphosphate glass. This trend is
consistent with the ab initio molecular orbital
predictions of Uchino and Ogata [24]. Fig. 9
(right) shows the bridging-to-terminal oxygen ra-
tio, determined by decomposing the O1s spectra,
in comparison to those calculated from the Qi -
distributions (Fig. 7) from 31 P NMR spectra and
to the predicted ratio from Eq. (3). Similar results
were obtained in other XPS studies of binary
phosphate glasses [61±64].
The progression from a cross-linked Q3 net-
work, to chain-like Q2 structures, to depolymer-
ized Q1 /Q0 glasses with increasing [O]/[P] ratio can
be monitored by a variety of other spectroscopic
probes. For example, Fig. 10 shows Raman spec-
tra collected from several Li-ultraphosphate [32]
and polyphosphate [65] glasses. The anhydrous
ultraphosphate glasses were prepared using sealed
ampoule techniques and the polyphosphate glasses
were quenched to avoid crystallization. Bands in
the 1000±1400 cmÿ1 range are due to the sym- Fig. 10. Raman spectra collected from Li-ultraphosphate [32]
metric and asymmetric stretching modes of P±non- and polyphosphate [65] glasses.
bridging (or terminal) bonds, whereas bands in the
620±820 cmÿ1 range are due to the symmetric and
asymmetric stretching modes of bridging oxygens. ultraphosphate glasses [34]. Infrared spectroscopy
Note that the frequency of the dominant `non- can also provide useful, complementary informa-
bridging' stretching mode decreases from 1390 tion about the network structures of anhydrous
cmÿ1 (x 0, symmetric P¸O stretch on Q3 tet- ultraphosphate glasses, as reported for series of
rahedra), to 1180 cmÿ1 (x 0:50, symmetric PO2 binary Li- [25] and Zn-phosphate [34] composi-
stretch on Q2 tetrahedra), to 1050 cmÿ1 (x 0:67, tions.
symmetric PO3 stretch on Q1 tetrahedra) and 950 Comparisons of Raman (and infrared) spectra
cmÿ1 (x 0:70, symmetric PO4 stretch on Q0 tet- from ultraphosphate glasses prepared by dierent
rahedra) with increasing [O]/[P] ratio. This de- techniques illustrate the depolymerizing eects of
crease in frequency shows the progressive water on the phosphate network. Fig. 11 shows
depolymerization of the phosphate network, as Raman spectra from glasses with the same nomi-
well as the increase in the average bond length of nal composition (mol%) of 25Na2 O á 75P2 O5 . The
P±non-bridging oxygens. This latter observation is top spectrum is from an anhydrous glass prepared
consistent with neutron and X-ray diraction in a sealed ampoule [32] and the bottom spectrum
studies of ultraphosphate glasses (e.g., [66]) dis- is from a glass prepared and handled in a dry box
cussed above and can be explained by a decrease in [68]. The relative intensity of the m P¸Osym peak
the average p-bond character of the P±TO bond (near 1370 cmÿ1 ) from the anhydrous glass is
[24]. Raman spectra have also been reported for greater than from the ÔwetÕ glass, indicating that
series of anhydrous Na- [32,33], Ca- [67] and Zn- the former structure is more cross-linked; i.e., has
Na-ultraphosphate glasses with 30±40 mol% Na2 O

[32,33] but not in the spectrum of a `wet' 40 mol%
ultraphosphate glass [68]. The band amplitude is
temperature dependent, disappearing when the
glass is heated to a liquid, and reappearing when
the liquid is cooled to a temperature near Tg
[32,33]. These observations indicate that the band
is associated with strained Q2 sites that apparently
react in the presence of water and relax when the
glass is heated above Tg .
The frequency of a number of vibrational
modes changes with composition. Fig. 12 shows
that the position of the m P¸O band in the Ra-
man spectra of Li- and Na-ultraphosphate glasses
(Fig. 10) decreases in frequency with increasing
alkali content [33]. (Note in Fig. 10 that the
m P¸O peak becomes indistinguishable from the
m(PO2 )asym band when Li2 O > 40 mol%.) This de-
Fig. 11. Raman spectra collected from 25Na2 Oá75P2 O5 glasses. crease is particularly evident when R2 O exceeds
The top spectrum (from Ref. [32]) is from an anhydrous glass about 20 mol%. The decreasing frequency indi-
prepared in a silica ampoule whereas the bottom spectrum
(from Ref. [68]) was collected from a glass that appears to have
cates that the P¸O bond on Q3 tetrahedra be-
a signi®cant water content (see discussion). comes longer, particularly when R2 O > 20 mol%
[33], because of p-bond delocalization. Similar
decreases at 20 mol% Li2 O occur to the frequency
a greater Q3 -content. A comparison of the relative of the P¸O IR absorption band [25] and there are
intensities of these bands with those reported for a changes in the shielding of the 31 P chemical shift
series of anhydrous xNa2 O(1)x)P2 O5 [32] indi- tensor elements from several Li-ultraphosphate
cates that the spectrum of the ÔwetÕ
25Na2 Oá75P2 O5 glass in Fig. 11 is more like that
from a glass with 35±40 mol% modi®er content. It
is likely, therefore, that this glass possesses
20 mol% water, as has been measured elsewhere
for this composition [44,46]), despite the reported
precautions taken in preparation and handling of
the glass.
Similar precautions are warranted when ana-
lyzing the spectra of other ultraphosphate glasses
prepared by open crucible techniques, including
recent infrared studies of Na- [69] and Pb-ultra-
phosphate glasses [70] and an early Raman study
of a K-ultraphosphate glass [71].
4.1.2. Network linkages in ultraphosphate glasses

An arrow in Fig. 10 indicates the presence of an
additional band on the high frequency shoulder of
the m(PO2 )sym band in the Raman spectrum of the Fig. 12. Frequency shift of the m P¸O Raman band from Li
anhydrous 30Li2 O á 70P2 O5 glass. Similar bands (d) and Na (s) ultraphosphate glasses (adapted from Ref.
can be seen in the Raman spectra of anhydrous [33]).
glasses [49] that are consistent with this explana- are, however, several dierent ways in which these
tion. tetrahedra could be linked to form the phosphate
The change in frequency of the m P¸O band network, including: (a) randomly distributed link-
indicates a possible change in the longer-range ages; (b) chemically ordered linkages; and (c) phase
(beyond a local P-tetrahedron) ultraphosphate separated structures. Q3 ±Q2 linkages are preferred
network structure at 20 mol% R2 O. Such struc- in the `chemically ordered' network; the fraction of
tural modi®cations are indicated in the neutron Q3 ±Q2 linkages will equal one at 37.5 mol% R2 O
scattering factors from the Na-ultraphosphate (or R0 O) [66]. There are virtually no Q3 ±Q2 linkages
glasses shown in Fig. 3. Note that the two inter- in a phase-separated network. For the `random
mediate-range order peaks of v-P2 O5 at Q equal to distribution' network, the number of Q3 ±Q2 link-
1.30 and 2.12 A ÿ1 (indicated by arrows in Fig. 3) ages can be determined from a simple binomial
are replaced by a single peak Q 1:60 A ÿ1 ) for expansion to calculate probabilities [49].
the 20 mol% Na2 O glass, corresponding to a pe- There are a number of alkaline earth ultra-
[27]. From 20 to 50 mol% Na2 O,
riodicity of 3.7 A phosphate crystals with the stoichiometry
the intermediate-range order peak shifts to larger R0 O á 2P2 O5 that possess chemically ordered struc-
Q, and a new peak arises Q 1.16 A ÿ1 for the tures dominated by Q3 ±Q2 linkages. The similari-
50 mol% Na2 O sample, implying order on an ex- ties between the vibrational spectra of these crystals
presumably associated
tended length scale (5.8 A) and the respective glasses led Meyer and co-work-
with metaphosphate chains. Walter and co-workers to conclude that the structures of ultraphos-
ers [28,72] have reported similar compositionally phate glasses were best described as a combination
dependent changes in the X-ray scattering factors of chemically ordered substructures similar to the
from alkaline earth ultraphosphate glasses. They appropriate, equilibrium crystalline phases
propose that ultraphosphate glasses with up to [34,67,74,75]. In this model, a Zn-ultraphosphate
20 mol% modifying oxides retain the `quasi-layer' glass with a composition 0 6 ZnO 6 33:3 mol% will
structure of v-P2 O5; with isolated modi®er ions have a structure based on a combination of chem-
occupying positions between the corrugated sheets ically ordered substructures similar to crystalline
of Q3 rings. Above about 20 mol% modifying P2 O5 and ZnP4 O11 [34]. A Zn-ultraphosphate glass
oxide, these Q3 layers are consumed as Q3 tetra- with a composition 33:3 6 ZnO 6 50 mol% will
hedra are converted to Q2 tetrahedra. have a structure based on a combination of chem-
Recent inelastic neutron scattering studies of ically ordered substructures similar to crystalline
Na-ultraphosphate glasses indicate changes in the ZnP4 O11 and Zn(PO3 )2 . Likewise, the glasses in the
phonon density of states (PDOS) that are consis- Ca-ultraphosphate series should have phosphate
tent with the description of the evolution of a 3D network structures based on the combinations of
P2 O5 network to corner-shared PO4 chains with the substructures of appropriate intermediate
increasing Na2 O content [73]. The breadth of the phases [67].
PDOS was greatest at 20 mol% Na2 O, indicating Two-dimensional (2D) solid state NMR exper-
larger distributions of P±O±P and O±Na±O coor- iments can provide direct detection of the linkages
dination environments. A narrower PDOS at the between neighboring P-tetrahedra in phosphate
metaphosphate composition was related to the glasses. Two types of experiments have been re-
appearance of extended-range order in the static cently described. The ®rst probes P±P proximity
structure factor (viz., Fig. 3 [27]). by magnetization exchange mediated by dipole±
There has been speculation that the way that the dipole coupling, using a radio frequency dipolar
Q2 and Q3 tetrahedra are arranged may be re- recoupling (RFDR) scheme [49,76,77]. The second
sponsible for these indicated changes in èxtended- is a double quantum (DQ) experiment that is
range' order [25,33,49,56]. As discussed earlier, the sensitive only to directly coupled pairs of 31 P spins
concentrations of dierent tetrahedra depend on [78,79]. Both techniques rely on the sensitivity of
composition according to the simple depolymer- the dipolar coupling constant to the distance
ization model described by Eqs. (1)±(2b). There between nuclei of interest for connectivity infor-
mation. An advantage of the DQ technique over ultraphosphate glasses (e.g., Fig. 5) may be related
the RFDR type of experiment is the poorer reso- to the variations in the distributions of these rings.
lution of the latter due in part to magnetization The conclusion by Feike et al. that Q2 ±Q2 and
transfer from a third (or more) spins [78]. Q ±Q3 linkages are present in this 35Na2 O á 65P2 O5
3
Fig. 13 (from Ref. [79]) shows a 31 P 2D DQ glass diers from that drawn in earlier 2D 31 P
spectrum from a 35Na2 O á 65P2 O5 (mol%) glass. RFDR NMR. Those earlier studies found only
0
The connectivities between Qn and Qn tetrahedra heteronuclear linkages in a similar glass and led to
are indicated by n±n0 . Signal intensity along the the conclusion that the Ôchemically orderedÕ model
indicated diagonal represents homonuclear (2±2 best described the linkages that make up an alkali
and 3±3) connections and o-diagonal intensity ultraphosphate glass network [76,77]. Alam and
represents heteronuclear (2±3 and 3±2) connec- Brow [49,82] used 2D 31 P RFDR NMR to deter-
0
tions. Feike et al. suggest that the )23 ppm mine the relative concentrations of Qn ±Qn linkages
chemical shift for the 2±2 auto-peak indicates that for a series of Li-ultraphosphate glasses and those
Q2 ±Q2 linkages are present in this glass, either in results are summarized in Fig. 14. The solid lines are
the form of 3-membered rings or in chains with at the expected populations of linkages based on a
least four Q2 tetrahedra. Trimetaphosphate rings random distribution model and are in good agree-
were presumed to be present in alkali ultraphos- ment with the measured connectivity concentra-
phate glasses, based on the hydrolytic degradation tions. The `chemically ordered' model predicts
of Q3 sites [80]. Similar moieties have been pro- much greater numbers of Q3 ±Q2 and Q2 ±Q3 link-
posed [33] to explain the high frequency m(PO2 )sym ages and the `phase separation' model predicts
Raman band (Fig. 10) discussed above. Elsewhere fewer Q3 ±Q2 and Q2 ±Q3 linkages than are found
in this volume, Liang et al. [81] describe molecular experimentally.
dynamics simulations of ultraphosphate glass
structures which possess rings of varying sizes and 4.1.3. Modi®er bonding in ultraphosphate glasses
they suggest that the property minima in alkali There seems to be little connection between the
ratio of P-tetrahedra (Fig. 7) or their distribution
(Fig. 14), and the signi®cant property minima that
have been reported near 20 mol% alkali oxide
(e.g., Fig. 5). The local bonding arrangements of the
tetrahedra do change near 20 mol% however, as
indicated by the m P¸O frequency shifts shown in
Fig. 12.
Hoppe [56] presented a structural model that is
based on the coordination environments of the
modi®er cations to explain packing densities in
alkaline earth and zinc ultraphosphate glasses.
(That model ± and supporting diraction experi-
ments ± is also discussed in Hoppe's article else-
where in this volume [29].) Structural studies of
crystalline ultraphosphates (e.g., [83,84]) indicate
that the local coordination environment of the
modifying cations includes terminal oxygens from
both Q2 and Q3 tetrahedra. Hoppe recognized that
similar bonding arrangements were likely present
in the ultraphosphate glasses and proposed that
the structure (and properties) of ultraphosphate
Fig. 13. 31 P double-quantum NMR spectrum from a glasses would be dependent on the number of
35Na2 O á 65P2 O5 glass. (The ®gure is from Ref. [79].) terminal oxygens available to coordinate the
(region I) and when MTO < CNMe (region II). For

region I, sucient individual terminal oxygens are
available for coordination with each Mev ion and
so these ions can exist as isolated coordination
polyhedra within the phosphate network (Fig. 15,
left). In region II, there are not enough individual
terminal oxygens to satisfy the coordination envi-
ronments of every Mev ion, and so these ions
must share the available TOs. As a result, in region
II Mev coordination polyhedra share corners and
edges and act as bridges between neighboring Q2
polyhedra (Fig. 15, right).
Consider ®rst the single valent ions. Diraction
studies indicate that lithium ions in metaphosphate
crystals [85] and glasses [86] are tetrahedrally co-
ordinated. A recent 6 Li NMR study of Li-ultra-
phosphate glasses reached a similar conclusion,
that the average CN for Li is in the range of 4±5
[87]. Diraction studies of NaPO3 glass indicate
that the coordination number of Na is about 5
[88,89]. Using these coordination numbers for
CNMe , from Eq. (5) the transition from region I to
region II is calculated to occur at 20±25 mol% for
Li and Na ultraphosphate glasses. This analysis
then provides an explanation for the minima in
properties (viz., Fig. 5) and for the spectral
changes such as those shown in Fig. 12. In glasses
with less than 20 mol% Li2 O or Na2 O, isolated
alkali polyhedra such as those shown in Fig. 15
(left) form. Because there is an excess of Q3 ter-
minal oxygens, these alkali ions likely are bonded
to the terminal oxygens on a single Q2 -tetrahe-
dron, and less-strongly bonded to neighboring Q3 -
Fig. 14. Connectivity distributions obtained from 31 P RFDR
analyses of a series of xLi2 O(1)x)P2 O5 glasses. The solid lines
tetrahedra through the terminal P¸O bonds.
are the expected connections based on a random distribution Density decreases (Fig. 5) as the Q3 network ex-
model [49]. pands to accommodate the alkali polyhedra. Tg
decreases [25,33] because the fraction of bridging
oxygens is decreasing (Eq. (3)) as Q2 tetrahedra
modi®er ions (Mev , where v is the ion valence). replace Q3 tetrahedra. Above 20 mol% Me2 O,
Such cations will have coordination numbers MTO < CNMe and so neighboring alkali polyhedra
(CNMe ) similar to those found in the appropriate must share the terminal oxygens on Q2 tetrahedra
crystalline phases. For glasses with the stoichi- (Fig. 15, right). Density increases as more Me-
ometry x(Me2=v O) (1)x)P2 O5 , the number of ter- polyhedra share corners and edges. In addition,
minal oxygens per modifying ion is the bonds between alkali ions and neighboring Q3
tetrahedra are strengthened, as indicated by the
MTO v 1=x: 5
lengthening of the Q3 P¸O bond (Fig. 12). Thus,
Hoppe noted that there will be two structurally the alkali polyhedra act as bridges between
dierent compositional ranges: when MTO > CNMe neighboring Q2 and Q3 tetrahedra, strengthening
Fig. 15. Schematic representations of alkali bonding in ultraphosphate glasses when MTO > CNMe (left) and MTO < CNMe (right).
the network (and so increasing Tg ) even as P±O±P transition from region I to region II will depend on
bonds are replaced by P±O±Me bonds as the alkali the coordination number of the Me2 ion and this
concentration increases. eect in fact has been reported. Minima in ultra-
In this volume, Prabakar et al. [90] describe a phosphate glass packing densities, in the transition
23
Na ® 31 P cross-polarization (CP) NMR experi- from region I to region II, occur at 50 mol%
ment that is consistent with this modi®er structural MgO (CN 4), 33 mol% CaO (CN 6), and
model. The CP time constant (Tcp ) for magneti- 25 mol% BaO (CN 8) [56,91]. There are cor-
zation transfer between neighboring Na and Q2 -P responding changes in the X-ray diraction
nuclei is longer for a 20Na2 Oá80P2 O5 glass than structure factors that are consistent with the for-
for either a 44Na2 Oá56P2 O5 or a 50Na2 Oá50P2 O5 mation of more compact structures of Me-poly-
glass. This observation indicates that the number hedra sharing corners and edges [72].
of sodium ions associated with an average Q2 site There are dierences in the vibrational spectra
in the 20Na2 O glass is less than that found for the of Ca- and Zn-ultraphosphate glasses that are re-
glasses with greater Na2 O contents, as would be lated to the expected coordination eects of the
expected if TOs in the latter glasses were coordi- dierent modi®ers on the phosphate network.
nated to more than one sodium ion. In the same Fig. 16 compares the Raman spectra collected from
study, Prabakar et al. report a correlation between binary ultraphosphate glasses with 30 mol% Li2 O,
Na nuclei and Q3 -P nuclei in the 20Na2 Oá80P2 O5 Na2 O (both from [33]), CaO [67] and ZnO [34]. Of
glass, indicating that the P¸O bond likely par- particular interest is the presence of separate bands
ticipates in the alkali coordination environment, as due to the m P¸O and m(PO2 )asym modes in the
indicated in Fig. 15. spectrum from the Zn-ultraphosphate (and, to a
When divalent or trivalent ions are incorpo- lesser extent, the Li-ultraphosphate), whereas these
rated into an ultraphosphate network, their charge two modes are indistinguishable in the spectra from
requirements dictate that at least two Q2 tetrahe- the Ca- and the Na-ultraphosphate glasses. The
dra are present in their immediate coordination smaller Zn2 ion apparently does not form bonds
sphere. As a result, `restructuring' through Q2 ± to terminal oxygens P¸O on the remaining Q3
Me2 ±Q2 bridges will occur in region I, even tetrahedra which are as strong as those of larger
though the Me2 -polyhedra can remain isolated as Ca2 and Na ions, and so does not aect the fre-
long as MTO > CNMe [56]. When MTO < CNMe , quency of the P¸O band to nearly the degree
the Me2 -polyhedra again begin sharing corners found for the other modifying ions.
and edges, with resulting changes in property The frequencies of the m(PO2 )sym and m(PO2 )asym
trends. One prediction of Eq. (5) is that the peaks in Fig. 16 are cation dependent, increasing
range is consistent with the larger coordination

numbers. In¯ections in packing density trends
have been correlated with the changes in coordi-
nation number [56].
The modi®er substructure that forms in region
II is a hallmark of the `modi®ed random network'
(MRN) theory for glass structure, introduced by
Greaves to describe the structure of modi®ed sili-
cate glasses [95]. The apparent segregation of
modi®ers has obvious consequences for the prop-
erties of oxide glasses [96]. For phosphate glasses,
the èxtra' terminal oxygen associated with each P-
tetrahedron makes it possible for simple phosphate
compositions to be model systems for illustrating
the consequences of the development of `MRN'
structures, such as at the transition from region I
to region II compositions (Fig. 15).
In summary, anhydrous ultraphosphate glasses
have network structures based on combinations of
Q2 and Q3 tetrahedra, the concentrations of which
can be predicted from Van Wazer's chemically
simple depolymerization model. These tetrahedra
Fig. 16. Raman spectra reported for binary ultraphosphate appear to be randomly linked, at least in alkali
glasses with 30 mol% Li2 O, Na2 O (both from [33]), CaO [67] ultraphosphate glasses. In glasses with large P2 O5
and ZnO [34]. The spectra are shown with normalized m(PO2 )sym contents (greater than about 75 mol%), the phos-
peak intensities. phate network most resembles that found for
v-P2 O5 . With increasing modi®er content, there
in frequency as the ®eld strength of the modifying is a loss of extended-range order associated with
cation increases (in the order Na < Li < v-P2 O5 , as the fraction of Q2 tetrahedra increases.
Ca2 < Zn2 ), consistent with previous studies of The composition at which this transition occurs
metaphosphate glasses [92]. The full-width at half depends on the coordination number and valence
maximum (FWHM) of the m(PO2 )sym band also of the modifying cation. Water has a depolymer-
increases with increasing cation ®eld strength izing eect on the phosphate network and account
(from about 22 cmÿ1 for the 30Na2 O glass to must be made for P±OH bonds when describing
about 52 cmÿ1 for the 30ZnO glass), indicating an the structure and properties of these glasses.
increase in the degree of disorder that may be due
to changes in the distributions of the Q2 P±O bond 4.2. Metaphosphate glasses ([O]/[P] 3.0)
lengths and bond angles. Again, these observa-
tions are consistent with previous spectroscopic Metaphosphate and polyphosphate glasses have
studies of metaphosphate glasses [93]. structures that are based on phosphate anions. The
Finally, there are neutron and X-ray diraction average chain-length (nav ) of an anion in a poly-
studies which show that the coordination numbers phosphate glass is given by Van Wazer [2]
of divalent ions in ultraphosphate glasses are nav 2 1 ÿ x= 2x ÿ 1: 6
larger in region I (when MTO > CNMe ) than in
region II [56,66,91,94]. For example, the coordi- Note that when x 0:5 (the metaphosphate stoi-
nation number of Zn changes from approximately chiometry), n is 1. (This conclusion assumes that
six at 33 mol% ZnO to four at 50 mol% ZnO [66]). there are no cyclic anions, the presence of which
The availability of èxcess' terminal oxygens in this will reduce the average chain length. In general,
the percentage of cyclic metaphosphate anions, element) [17]. The chain constitution in a meta-
including (P3 O9 )3ÿ and (P4 O12 )4ÿ is greatest when phosphate glass is also aected by the modifying
x 0.5 and decreases with increasing x [97].) Av- cation, as indicated by a variety of spectroscopic
erage chain lengths can be estimated by end group studies. For example, the full width at half maxi-
titration techniques (e.g., Ref. [98]) and typical mum (FWHM) of the 31 P NMR peaks from series
`metaphosphate' glasses have chains with nav of metaphosphate glasses increases with the ®eld
ranging from 40 to 100 P-tetrahedra and termi- strength of the modifying cation, indicating an
nated by hydroxyl groups [99]. By removing water increase in the distribution of P±O bond lengths
from alkaline earth metaphosphate liquids, glasses and bond angles [35,115]. A similar trend was re-
with chain lengths approaching 1000 tetrahedra ported in the breadth of P±O bond lengths from a
have been reported [13]. Abe [13] reviews the ef- neutron diraction study of several metaphos-
fects of water on the properties of metaphosphate phate glasses [125]. Luminescence studies of Eu-
glasses. Grith reports that glasses with average doped metaphosphate glasses con®rm the increase
chain lengths no greater than 400 tetrahedra are in the degree of disorder with increasing modi®er
possible by quenching alkali polyphosphate liq- ®eld strength [116].
uids, in contrast to crystallized sodium Kurrol's The phosphate chains are linked through bonds
salts with chain lengths over 10 000 tetrahedra between terminal oxygens and the modifying ca-
[100]. tions, as shown schematically in Fig. 15 (right).
The transition from cross-linked ultraphos- Raman band broadening, frequency shifts, and a
phate structures to polymer-like metaphosphate decrease in the ratio of the amplitudes of the
structures aects the temperature dependence of m(PO2 )sym (1180 cmÿ1 ) and m(POP) (690 cmÿ1 )
viscosities of alkali phosphate liquids; in general, Raman bands (Fig. 17), as well as an increase in
metaphosphate liquids are more fragile (as de®ned the amplitude ratio of depolarized vs. polarized
by Angell, e.g., Ref. [101]) than ultraphosphate spectra, indicate that the modi®er-TO bond be-
liquids [102]. The rheological properties of meta- comes increasingly covalent as the ®eld strength of
phosphate liquids are more like those of a polymer the modi®er increases [117±120,130]. The resulting
liquid than a conventional cross-linked oxide glass increase in the rigidity of the metaphosphate net-
[103±105]. The motion of intact phosphate chains work is consistent with a concomitant increase in
determines macroscopic phenomena such as vis- the glass transition temperature [121]. Thus, the
cous ¯ow near Tg [106]. Chain alignment in properties of metaphosphate glasses are less de-
metaphosphate ®bers can create anisotropic opti- pendent on the nature of the P±O±P bonds that
cal and mechanical properties [107±109]. (A recent form phosphate chains than on the P±O±Me inter-
2D rotor-synchronized NMR experiment identi- chain bonding.
®ed partial alignment of phosphate chains along A narrow (5±6 cmÿ1 ) band is present on the low
the extrusion direction [110]. J ager et al. [111] de- frequency side (630 cmÿ1 ) of the m(POP) band in
scribe these, and newer, experiments elsewhere in the Raman spectra of Cs-, Rb- and K-metaphos-
this volume.) Consistent with this eect are recent phate glasses and melts [122] that is not present in
X-ray diraction studies of NaPO3 which indicate the spectra of the metaphosphate glasses shown in
that upon heating above Tg the alkali polyhedral Fig. 17. This narrow band increases in intensity as
environment expands and the phosphate chains the size of the modifying cation increases and is
remain intact and are more ordered in the liquid assigned to a stretching mode of ordered phos-
than in the glass [112±114]. phate chain segments with P±O±P bond angles of
Diraction studies of crystalline (linear) meta- 180° [122].
phosphates indicate that the P-tetrahedral chains Uchino and Yoko, in this volume [123], de-
have a variety of conformations. In general, larger scribe alkali metaphosphate clusters simulated by
®eld strength cations can be correlated with chains ab initio calculations and they note the impor-
with greater conformational periodicities (de®ned tant role played by the alkali ions in cross-link-
by the number of tetrahedra per repeating chain ing and distorting neighboring metaphosphate
Fig. 17. Raman spectra from several metaphosphate glasses. (The ®gure is from Ref. [120].)
clusters. From simulations of the vibrational measured by X-ray scattering [126] and by neutron
spectra of their clusters, they correlate low fre- diraction [27]; prominent peaks in the structure
quency (<100 cmÿ1 ) bands with rotational mo- factor were related to quasi-periodic superstruc-
tions of Q2 -tetrahedra and suggest that it is the tures, the dimensions of which depend on the
thermal excitation of such modes that is re- modifying cation. Scattering from such structures
sponsible for the structural rearrangements at the may involve the electron density dierences be-
glass±liquid transition. Larger ®eld strength ca- tween the metaphosphate chains and the inter-
tions increase structural rigidity, increasing the vening metal oxide polyhedral con®gurations
temperature required to activate the rotational [127]. (See [29] for further discussion of these
motions, and so increasing Tg . (The low fre- scattering studies.)
quency rotational modes modeled by Uchino The structural environments of modifying ca-
and Yoko may be related to slower, sub-Tg re- tions in metaphosphate glasses have been deter-
laxation processes, decoupled from faster viscous mined in a variety of diraction and X-ray
modes, in mixed alkali metaphosphate glasses absorption spectroscopic measurements. Table 2
described by Rue et al. [124].) summarizes some results. Of interest are the zinc
Neutron diraction studies indicate a decrease metaphosphate glasses, in which the Zn[IV] ions
in the mean P±O bond length with increasing combine with the Q2 P sites to form a tetrahedral
modi®er ®eld strength [125], consistent with the network which has low temperature vibrational
interpretations of the Raman spectral trends de- anomalies [128] and an unusually large photo-
scribed above. Variations in the intermediate range elastic coecient [129], reminiscent of three-di-
order of metaphosphate glasses have been mensional network glasses such as silica.
Table 2
Cation coordination numbers (CN), average metal±oxygen and P±O bond distances reported for metaphosphate glassesa
Cation CN
rMe±O (A)
Avg. rP±O (A) Technique Reference
Li 4.0 2.02 1.52 XRD [86]
Li 4.0 (0.4) 2.03 (0.04) ND [130]
Na 5.0 (0.4) 2.38 1.57 XRD [94]
Na 5.7 (0.4) 2.43 (4) ± XRD [88]
K 6.7 (5) 2.77 (2) 1.48 (1) TO 1.625 (10) BO XRD, ND [59]
Rb 5.0 (0.6) 2.9 (0.1) ND [130]
Ag 3.4 (3) 2.362 (5) 1.590 (3) XRD [132]
Mg2 4.0 2.06 1.53 XRD [132]
Ca2 4.9 2.33 1.52 XRD [132]
Ca2 7.0 (0.4) 2.39 ) XRD [88]
Sr2 5.1 (5) 2.55 (3) 1.54 (1) EXAFS [133]
Ba 8.0 2.79 1.56 XRD [94]
Zn2 3.7 (3) 1.94 (1) ) EXAFS [134]
Zn2 4.0 (2) 1.95 (1) 1.53 (1) XRD [134]
Zn2 4.7 (4) 1.996 (5) 1.558 (3) XRD [131]
Zn2 3.9 (4) 1.939 (3) 1.50 (1) TO 1.60 (1) BO XRD [54]
Pb2 5.0 (4) 2.484 (5) 1.593 (3) XRD [131]
Pb2 5.2 (6) 2.47 (2) 1.49 (1) TO 1.59 (1) BO ND [125]
Pb2 8 (1) 2.47 ) EXAFS [135]
La3 7.1 (5) 2.458 (15) 1.484 (5) TO 1.604 (5) BO ND,XRD [136]
La3 7.3 (1.2) 2.42 (0.03) ± EXAFS [137]
Nd3 10.4 (2.7) 2.35 (0.010) ) EXAFS [138]
Sm3 6.9 (0.3) 2.31 (0.01) ) EXAFS [137]
Eu3 7.6 (0.2) 2.28 (0.002) ) EXAFS [138]
Eu3 6.2 (0.3) 2.31 (0.01) ) EXAFS [137]
Gd3 5.7 (0.3) 2.27 (0.002) ) EXAFS [138]
Gd3 5.0 (0.4) 2.28 (0.01) ) EXAFS [137]
Tb3 8.0 (0.2) 2.25 (0.002) ) EXAFS [138]
Ho3 7.0 (0.1) 2.20 (0.001) ) EXAFS [138]
a
The P±O distances are averages, unless otherwise indicated. Limit errors, from original references, are in parentheses.
Kordes identi®ed a number of phosphate by tetrametaphosphate rings to one based on

glass-forming systems, including BeO±P2 O5 , pyrophosphate dimers, although there appears
MgO±P2 O5 and ZnO±P2 O5 , as ànomalous' be- not to be supporting spectroscopic evidence for
cause they have unusual discontinuities in com- glass structures dominated by tetrametaphos-
position-property dependence (i.e., refractive phate rings.
index and molar volume) at the metaphosphate The Ômodi®er coordination modelÕ described in
composition [3]. This dependence was originally Section 4.1.3 provides a likely explanation for the
explained by a coordination change of the mod- properties of the anomalous glasses identi®ed by
ifying cation [3,4], but subsequent spectroscopic Kordes. Consider that the ÔanomalousÕ divalent
and diraction studies generally failed to identify cations (e.g., Zn2 and Mg2 ) have coordination
such a change. (For example, see Ref. [64] for numbers of about 4 (Table 2). From Eq. (5), the
references to studies of the ZnO±P2 O5 glasses transition from isolated divalent tetrahedra to
and Ref. [139] for a review of the MgO±P2 O5 corner-sharing tetrahedra will occur at 50 mol%
glasses.) Okura et al. [140] proposed that the RO; viz., the anomalous composition. The com-
anomaly in the properties of the MgO phosphate positions at which the coordination requirements
glass forming system was related to the transition for monovalent and larger divalent cations (e.g.,
from a metaphosphate glass structure dominated Ca2 ) exceed the number of available terminal
oxygens are in the ultraphosphate range and and these reactions will have equilibrium constants
hence those metaphosphate glasses do not have de®ned by
the anomalous properties. Hoppe et al. [136] h ih i
used a similar analysis to explain the correlation of kn Pn1 O3n4 ÿ n3 Pnÿ1 O3nÿ2 ÿ n1
La±O±La bonds in a diraction study of a La- h i2
metaphosphate glass. La3 has a coordination = Pn O3n1 ÿ n2 : 7b
number of about 7, which exceeds the number of
available terminal oxygens in that composition The distribution of chain lengths, at constant [O]/
(Eq. (5)), necessitating the corner sharing by [P], depends on the type of modifying cation and
neighboring La-polyhedra. such analyses can be related to various glass
properties. Fig. 18 (left) shows chromatographs
4.3. Polyphosphate glasses ([O]/[P] > 3.0) obtained from three dierent polyphosphate
glasses [151]. The Na-glass has the narrowest dis-
A variety of spectroscopic techniques detect tribution of P-anions and the Al-glass has the
changes in the average chain length (or, by ex- broadest distribution. As seen in Fig. 18 (right),
tension, to the [O]/[P] ratio) for polyphosphate the latter liquids have larger equilibrium constants
glasses. These include 31 P NMR chemical shifts for the disproportionation reactions (Eqs. (7a) and
[44,141], P2p/P2s electron binding energies from (7b)); these reactions appear to shift to the right as
XPS experiments [61], and P L-edge and K-edge the ®eld strength of the modifying cation increases.
energies from X-ray absorption spectroscopy The more complicated phosphate structures that
[142]. Such studies indicate a decrease in the av- result from the larger anion distributions can be
erage p-bond character of a P±O bond with in- correlated with ease of glass formation [147,151].
creasing modi®er content [24] and can account for A similar cation sensitivity for `structural com-
variations in optical basicity as measured by ¯u- plexity' was noted above for the metaphosphate
orescence and absorption spectroscopy of phos- glasses. (Sales reviews chromatographic studies of
phate glasses doped with optically active ions polyphosphate glasses elsewhere in this volume
[143±146]. The evolution of extended range order, [152].)
from that associated with a 3D cross-linked ul- Anionic distributions in polyphosphate glasses
traphosphate network to metaphosphate chains to can also be described using the tetrahedral-sensi-
pyrophosphate dimers with increasing modi®er tive spectroscopic techniques discussed above. The
content has been measured by neutron diraction advantage is that information can be obtained
and inelastic scattering [73]. about the phosphate anions without ®rst dissolv-
The distribution of chain lengths in polyphos- ing the glass, as must be done for chromatographic
phate glasses is best determined by chromato- analyses. For example, Fig. 19 (left) shows the 31 P
graphic analyses of solutions of dissolved glasses NMR spectrum collected from a 67ZnOá33P2 O5
e.g., [147,148]. Such techniques can quantitatively glass [64]. Based on stoichiometry, only Q1 sites
detect phosphate anions with chain lengths up to should be present, but the spectrum indicates de-
30 tetrahedra [149]. Anion distributions were tectable concentrations of Q2 and Q0 sites that
quantitatively explained by Van Wazer [2], using result from Q1 site disproportionation reactions
an approach originally developed by Flory [150] to that occur during the reorganization of the glass
describe the distributions of polymer molecular liquid
weights. Reorganization of phosphate anions in a 2Q1 $ Q2 Q0 : 8
liquid depends on disproportionation reactions of
the type Fig. 19 (right) shows the quantitative distributions
of tetrahedra from a series of xZnO(1)x)P2 O5
2 Pn O3n1
ÿ n2
$ Pn1 O3n4
ÿ n3 glasses. The open symbols are tetrahedral distri-
ÿ n1
butions determined from high-performance liquid
Pnÿ1 O3nÿ2 7a chromatography [153] and the closed symbols are
Fig. 18. (Left) High-performance liquid chromatographs from Na- (top), Zn- (middle) and Al-polyphosphate glasses (bottom), with
average chain lengths as indicated (right). Equilibrium constants calculated from Eqs. (7a) and (7b) using anionic information based on
the chromatographic analyses. (Figures are from Ref. [151].)
the results from a 31 P NMR study [64]. By treating Montagne et al. [154] used NMR to observe
the disproportionation reaction (8) as an equilib- similar site disproportionation reactions in poly-
rium reaction and assuming constant activity co- phosphate glasses when ZnO is added to NaPO3 .
ecients for each site, the NMR results indicate Tatsumisago et al. [65] used Raman spectroscopy
that k1 0:028 0:018. One consequence of this to study the same in Li-polyphosphate glasses.
analysis is that kn depends not only on the ®eld E®mov [155] used infrared spectroscopy to study
strength of the modifying cation (Fig. 19, right) the structures of alkali zinc pyrophosphate glasses
but also on the liquid temperature [147], and so it and concluded that the Q2 tetrahedra in Eq. (8) are
is possible to modify the constitution of the incorporated into (P3 O9 )3ÿ metaphosphate rings,
phosphate network by varying the melt conditions. particularly in the zinc-rich compositions.
Fig. 20. Raman spectra collected from xLi2 O(67)x)ZnO

33P2 O5 glasses. (Adapted from Ref. [156].)
Fig. 19. 31 P MAS NMR spectrum from a 0.67ZnOá0.33P2 O5

glass (top) indicating the presence of Q0 , Q1 and Q2 tetrahedra viz., replacing ZnO with an alkali oxide shifts the
and the distributions of tetrahedra (bottom) determined from disproportionation reaction (8) to the left to yield
the NMR spectra from a series of Zn-phosphate glasses
a more durable glass. This change in structure is
(adapted from [64]). The solid lines in the distribution ®gure
represent predicted tetrahedral distributions based on Eqs. seen in the Raman spectra [156] shown in Fig. 20
(2c)±(2f), whereas the dashed lines account for Q1 dispropor- where the replacement of ZnO by Li2 O decreases
tionation, according to Eqs. (7a) and (7b). Open symbols the width of the spectral envelope due to various
are tetrahedral concentrations reported in a high-performance P±Oÿ stretching modes. (Note that the bottom
liquid chromatography study of similar glasses [153]. spectrum corresponds to the Zn-pyrophosphate
glass. The broad spectral envelope centered near
An interesting feature of Zn-pyrophosphate 1050 cmÿ1 is a consequence of the distribution of
glass is that the addition of an alkali oxide (re- P-anions as noted above.) Similar improvements
placing ZnO) increases its aqueous corrosion re- are noted in the durabilities of iron phosphate
sistance [156]. This improvement in durability has glasses. An Fe-pyrophosphate glass is over 1000
been correlated with a concomitant decrease in the times more durable than an Fe-metaphosphate
fraction of more easily hydrolyzed Q2 sites [156]; glass [157], with the improved durability correlated
with the elimination of bridging P±O±P bonds LiPO3 leaves the metaphosphate chains unchanged
associated with Q2 tetrahedra [158]. but further separated by the intervening LiCl
Distributions of tetrahedral linkages in poly- phase [86], analogous to the eects of AgI addi-
phosphate glasses have been measured using the tions to AgPO3 , described above.
multi-dimensional NMR techniques described Two-dimensional heteronuclear correlation
above. For example, the 31 P double-quantum NMR techniques are now being used to describe
NMR spectrum of 58Na2 Oá42P2 O5 glass reported modi®er bonding in polyphosphate glasses.
by Feike et al. [79] resolves Q2 ±Q2 , Q2 ±Q1 and Q1 ± Mueller and co-workers [90,164] have resolved Na
Q1 linkages, although the latter might be due to ions near Q2 tetrahedra from ones near Q1 tetra-
weak dipolar couplings between neighboring Q1 hedra in single and mixed alkali polyphosphate
end-groups. In this volume, J ager et al. [111] re- glasses. Such techniques will become increasingly
port procedures by which the DQ experiments important for the determination of structural or-
provide quantitative information about poly- ganization beyond a local polyhedron.
phosphate network connectivity. They found that There are other interesting properties associated
Q2 tetrahedra in trimer anions (P3 O10 )5ÿ that are with ìnvert' poly- and pyrophosphate glasses.
bonded to two Q1 tetrahedra (labeled Q2;11 ) can be Glasses in the alkali-modi®ed zinc phosphate sys-
resolved from Q2 tetrahedra connected to one Q1 tem can be prepared with transformation temper-
and one Q2 (Q2;12 ) or from Q2 tetrahedra con- atures of 275±375°C, low enough to allow them to
nected to two other Q2 (Q2;22 ). Similar information be blended with polymers and molded using con-
is provided for the various Q1;jk tetrahedra. By ventional plastic forming equipment [8]. A related
determining the relative concentrations of the dif- family of glasses is based on the (SnO + ZnO)-py-
ferent Qi;jk sites, it will be possible to obtain chain- rophosphate system, under development for low
length distribution information for phosphate temperature (<200°C), PbO- and alkali-free seal-
glasses similar to that obtained by chromatogra- ing applications [165]. Properties in this system can
phy but without dissolving the sample [159]. J ager be tailored by varying the SnO/ZnO ratio.
et al. [111] also raise the possibility that, by de- The depolymerized Sn-phosphate system has
termining the relative orientations of the chemical yielded another unusual family of glasses. Tick
shift tensors, bonding angle data might be ob- [166] showed that durable compositions with glass
tained in future DQ NMR experiments. transformation temperatures as low as 100°C
Olsen et al. [160] examined Ag-metaphosphate could be formed from mixtures of SnO, SnF2 , and
and Ag-polyphosphate glasses, with and without P2 O5 particularly when ([O] + [F]/2)/[P] 4. The
AgI dopants, using both radio frequency dipole resulting glass networks consist of low coordinated
recoupling (RFDR) and 31 P double-quantum (probably 3 or 4) Sn(O, F) polyhedra linked to
experiments. They showed for the former glass isolated or dimer ¯uorophosphate units in the
that the Q2 metaphosphate structure was retained most durable compositions [167]. The very high
after AgI additions and they could resolve pyro- thermal expansion coecients and (too) low glass
phosphate tetrahedra from Q1 sites terminating transformation temperatures probably limit seal-
chains in the latter, polyphosphate system. The ing applications for these materials. However, the
addition of AgI to AgPO3 appears to òpen up' the glasses have been used as chemically stable hosts
metaphosphate network [161] without breaking for low-temperature optically active organic mol-
the metaphosphate chains. This eect diers from ecules [168].
the eects of the addition of alkali ¯uorides to Orthophosphate compositions from the Li2 O±
alkali metaphosphate glasses, where ¯uorine re- Al2 O3 ±TiO2 ±P2 O5 system have been used to pre-
places bridging oxygens [162], shortening the av- pare fast-ion conducting glass-ceramics [11] and
erage polyphosphate chain-length. When LiF is porous glass-ceramics with membrane, ion ex-
added to LiPO3 , ¯uorine is partitioned between change and catalytic applications [169]. Such
¯uorophosphate terminal groups and Li±F frag- compositions are from the NASICON (Na-su-
ments [163]. In contrast, the addition of LiCl to perionic conductors) family of orthophosphate
glass forming systems (see [170] for a review of glasses will also provide critical information to test
some of these compositions). Other depolymer- structural models. X-ray photoelectron spectros-
ized titanophosphate glasses have been developed copy has provided some quantitative information
for non-linear optical applications. The hyper- about the concentrations of dierent (bridging and
polarizability of Ti-polyhedra increases the non- non-bridging) oxygen sites, as well as some limited,
linear refractive index coecient (n2 ) to make ti- qualitative information about bond order. Solid
tania-containing glasses useful for photonic de- state 17 O NMR experiments has provided detailed
vices [171]. Analogous to the Fe-phosphate information about oxygen bonding, including
glasses [172], the titanophosphate glasses possess bond angles and local coordination environments
ìnvert' networks based on linkages between small in silicate-based glasses (e.g., see [175,176]) and
(Q0 , Q1 , Q2 ) phosphate anions and distorted oc- similar quantitative information is expected for the
tahedral titanate species, charge balanced by al- phosphate glasses. Among the questions that
kali or alkaline cations [173] and, when might be answered by such experiments: Are there
crystallized, can form bioactive ortho- and pyro- dierences between the non-bridging oxygens on
phosphate phases [174]. individual Q2 and Q1 tetrahedra? Does the Q3
double-bonded oxygen P¸O lengthen at x 20
mol% alkali oxide, as indicated in the Raman
5. Concluding remarks spectra? Do bridging oxygens participate in the
local coordination environment of modifying ca-
New technological applications have spurred tions, in con¯ict with the predictions of the
much of the recent interest in the structures of `modi®er coordination' structural model? How do
simple phosphate glasses. The ultraphosphates are the P±O±P bond angles change at the transition
among the simplest compositions but their struc- from isolated to linked alkali polyhedra? What
tures and properties were not properly understood relative level of order exists for bridging and non-
before the recent recognition that special handling bridging oxygens?
procedures are necessary to fabricate anhydrous Critical experiments that describe in detail the
samples. The anomalous compositional depen- longer-range phosphate network structures will
dencies of glass properties (e.g., Fig. 5) cannot also be important for establishing quantitative
(yet) be adequately explained by changes in the structure-property relationships for phosphate
details of the phosphate network structure. There glasses. Chromatography experiments provide in-
are no P-coordination changes, as found in simple formation about P-anion distributions in poly-
borate and germanate systems, nor do there ap- phosphate compositions, but critical information
pear to be any particular relationships between the about chain conformation (including P±O±P bond
distributions of tetrahedral linkages and proper- lengths and angles) that is implied in broad vib-
ties. Instead, it appears that the `phosphate rational and NMR spectra is lost in the prepara-
anomaly' depends on the modi®er sub-structure, tion of liquid chromatography samples, as is
with breaks in physical properties associated with network bonding information in ultraphosphate
compositions that represent a transition from compositions. New solid state multi-dimensional,
isolated modi®er polyhedra to polyhedra linked multiple quantum NMR experiments hold prom-
through non-bridging oxygens on Q2 tetrahedra. ise for providing that information, particularly for
Future spectroscopic and modeling studies that the ultraphosphate compositions that cannot be
provide details about the modi®er bonding envi- analyzed by liquid chromatography. These new
ronments will provide critical tests for this spectroscopic techniques will also be important for
structural description. New heteronuclear, multi- determining the relative importance of rings and
dimensional NMR experiments will be among the chains to the properties of phosphate glasses. For
more important of these analyses. example, does the formation of small, energetic
A quantitative understanding of the local envi- phosphate rings contribute to the `phosphate
ronments of oxygen ions in simple phosphate anomaly'? Is the anomaly a consequence of the
entanglement of long phosphate chains with in- [2] J.R. Van Wazer, Phosphorus and its Compounds, vol. 1,
creasing alkali content? Interscience, New York, 1958.
[3] E. Kordes, W. Vogel, R. Feterowsky, Z. Elektrochem. 57
These and other spectroscopic techniques will (1953) 282.
also be applied to studies of glass liquids to de- [4] E. Kordes, R. Nieder, Glastechn. Ber. 41 (1968) 41.
termine structural changes upon heating through [5] M.J. Weber, J. Non-Cryst. Solids 123 (1990) 208.
the glass transition. Among the questions to be [6] J.H. Campbell, T.I. Suratwala, these Proceedings, p. 318.
answered are those related to temperature depen- [7] R.K. Brow, L. Kovacic, R.E. Loehman, Ceram. Trans. 70
(1996) 177.
dencies of ring and chain formation and the rela- [8] C.J. Quinn, G.H. Bell, J.E. Dickinson, in: Proceedings of
tive role of changes in the modi®er coordination the XVIth International Congress on Glass, vol. 4, 1992,
environment in determining the properties of p. 79.
phosphate liquids. A separate question is whether [9] D.E. Day, Z. Wu, C.S. Ray, P. Hrma, J. Non-Cryst. Solids
simple anhydrous ultraphosphate melts have liq- 241 (1998) 1.
[10] J. Vogel, P. Wange, P. Hartmann, Glastech. Ber. 70 (1997)
uid±liquid immiscibility. I am unaware of reports 23.
on this topic. [11] J. Fu, J. Mater. Sci. 33 (6) (1998) 1549.
Finally, future spectroscopic studies of the [12] J.B. Bates, N.J. Dudney, G.R. Gruzalski, R.A. Zuhr, A.
structures of phosphate glasses will bene®t from Choudhury, C.F. Luck, J.D. Robertson, Solid State Ionics
improved structural simulations based on ab initio 53±56 (1992) 647.
[13] Y. Abe, in: M. Grayson, E.J. Grith (Ed.), Topics in
and molecular dynamics (MD) calculations. Im- Phosphorus Chemistry, vol. 11, Wiley, New York, 1983,
proved molecular orbital descriptions will help p. 19.
decipher subtle spectroscopic information associ- [14] S.W. Martin, Eur. J. Solid State Chem. 28 (1991) 163.
ated with structural changes due to variations in [15] D.W.J. Cruickshank, J. Chem. Soc. (1961) 5486.
O/P ratios and the nature of modifying cations. [16] K.A.R. Mitchell, Chem. Rev. 69 (2) (1969) 157.
[17] F. Liebau, in: M. O'Keefe, A. Novrotsky (Eds.), Structure
Such calculations will also lead to more accurate and Bonding in Crystals II. Academic Press, New York,
potentials for P5 ions that will then be used in 1981, p. 197.
future MD simulations that should provide im- [18] W.H. Zachariasen, J. Am. Chem. Soc. 54 (1932) 3841.
portant long-range structural information. [19] G. H agg, J. Chem. Phys. 3 (1935) 42.
[20] A.C. Wright, J. Non-Cryst. Solids 106 (1988) 1.
[21] M. Jansen, B. L uer, Z. Kristallogr. 177 (1986) 149.
[22] El.H. Arbib, B. Eloudi, J.P. Chaminade, J. Darriet, Solid
Acknowledgements State Chem. 127 (1996) 350.
[23] D. Stachel, I. Svoboda, H. Fuess, Acta Crystallogr. C 51
The author thanks the many colleagues with (1995) 1049.
whom he has collaborated on many of the studies [24] T. Uchino, Y. Ogata, J. Non-Cryst. Solids 181 (1995) 175.
[25] J.J. Hudgens, S.W. Martin, J. Am. Ceram. Soc. 76 (1993)
reviewed here, including David Tallant, Todd 1691.
Alam, Jim Kirkpatrick, Jim Hudgens and Gary [26] U. Hoppe, G. Walter, A. Barz, D. Stachel, A.C. Hannon,
Turner. He also thanks Uwe Hoppe, Steve Martin J. Phys.: Condens. Matter 10 (1998) 261.
and Chandra Ray for critically reviewing this [27] K. Suzuya, D.L. Price, C.K. Loong, S.W. Martin, J. Non-
manuscript. This work was supported by Sandia Cryst. Solids 232±234 (1998) 650.
[28] G. Walter, U. Hoppe, R. Kranold, D. Stachel, Phys.
National Labs. Sandia is a multiprogram labora- Chem. Glasses 35 (6) (1994) 245.
tory operated by Sandia Corporation, a Lockheed [29] U. Hoppe, G. Walter, R. Kranold, D. Stachel, these
Martin Company, for the United States Depart- Proceedings, p. 29.
ment of Energy under Contract DE-AC04- [30] F.L. Galeener, J.C. Mikkelsen, Solid State Commun. 30
94AL85000. (1979) 505.
[31] W.B. Chang, Z.Z. Jin, X.W. Zou, Phys. Lett. A 159 (1991)
361.
[32] R.K. Brow, D.R. Tallant, J.J. Hudgens, S.W. Martin, A.D.
References Irwin, J. Non-Cryst. Solids 177 (1994) 221.
[33] J.J. Hudgens, R.K. Brow, D.R. Tallant, S.W. Martin,
[1] N.J. Kreidl, W.A. Weyl, J. Am. Ceram. Soc. 24 (11) (1941) J. Non-Cryst. Solids 223 (1998) 21.
372. [34] K. Meyer, J. Non-Cryst. Solids 209 (1997) 227.
[35] R.J. Kirkpatrick, R.K. Brow, Solid State Nucl. Magn. [65] M. Tatsumisago, Y. Kowada, T. Minami, Phys. Chem.
Reson. 5 (1995) 9. Glasses 29 (1988) 63.
[36] A.-R. Grimmer, G.U. Wolf, Eur. J. Solid State Inorg. [66] U. Hoppe, G. Walter, D. Stachel, A. Barz, A.C. Hannon,
Chem. 21 (1991) 221. Z. Naturforsch. A 52 (1997) 259.
[37] J.J. Hudgens, PhD thesis, Iowa State University, 1994. [67] K. Meyer, Phys. Chem. Glasses 39 (2) (1998) 108.
[38] W. Vogel, in: D.R. Uhlmann, N.J. Kreidl (Eds.), Optical [68] C. Nelson, D.R. Tallant, Phys. Chem. Glasses 26 (1985)
Properties of Glass. American Ceramic Society, 1991, p. 1. 119.
[39] D.R. Gard, in: Kirk-Othmer Encyclopedia of Chemical [69] Y.M. Moustafa, K.El. Egili, J. Non-Cryst. Solids 240
Technology, vol. 18, 4th Ed., 1991, p. 669. (1998) 144.
[40] A.E.R. Westman, in: J.D. Mackenzie (Eds.), Modern [70] C. Dayanand, G. Bhikshamaiah, V. Jaya Tyagaraju, M.
Aspects of the Vitreous State, Butterworths, London, Salagram, A.S.R. Kishna Murty, J. Mater. Sci. 31 (1996)
1960, p. 63 (Chapter 4). 1945.
[41] M. Tatsumisago, Y. Kowada, T. Minami, Phys. Chem. [71] N.H. Ray, Glass Technol. 16 (5) (1975) 107.
Glasses 29 (1988) 63. [72] G. Walter, U. Hoppe, T. Baade, R. Kranold, D. Stachel,
[42] W.A Weyl, E.C. Marboe, The Constitution of Glasses, J. Non-Cryst. Solids 217 (1997) 299±307.
Interscience, New York, 1964, p. 576. [73] C.K. Loong, K. Suzuya, D.L. Price, B.C. Sales, L.A.
[43] H.J.L. Trapp, J.M. Stevels, Phys. Chem. Glasses 1 (1960) Boatner, Physica B 241±243 (1998) 890.
107. [74] K. Meyer, H. Hobert, A. Barz, D. Stachel, Vibrat.
[44] R.K. Brow, R.J. Kirkpatrick, G.L. Turner, J. Non-Cryst. Spectrosc. 6 (1994) 323.
Solids 116 (1990) 39. [75] A. Barz, K. Meyer, D. Stachel, Glastech. Ber. 68 (C1)
[45] E. Kordes, H. Becker, Z. Anorg. Allgem. Chemie 260 (1995) 79.
(1949) 185. [76] C. Jager, M. Feike, R. Born, H.W. Spiess, J. Non-Cryst.
[46] P.E. Gray, L.C. Klein, Glass Technol. 24 (1983) 202. Solids 180 (1994) 91.
[47] G.M. Bartenev, in: The Structure and Mechanical Prop- [77] R. Born, M. Feike, C. J ager, H.W. Spiess, Z. Naturforsch
erties of Inorganic Glasses, Wolters-Noordho, Gronin- A 50 (1995) 169.
gen, The Netherlands, 1970, p. 166 (F.F. Jaray, Trans.). [78] M. Feike, R. Graf, I. Schnell, C. Jager, H.W. Spiess, J. Am.
[48] K. Meyer, A. Barz, D. Stachel, J. Non-Cryst. Solids 191 Chem. Soc. 118 (1996) 9631.
(1995) 71. [79] M. Feike, C. J ager, H.W. Spiess, J. Non-Cryst. Solids 223
[49] T.M. Alam, R.K. Brow, J. Non-Cryst. Solids 223 (1,2) (1998) 200.
(1998) 1. [80] J.B. Gill, S.A. Riaz, J. Chem. Soc. A (1969) 843.
[50] P. Losso, B. Schnabel, C. Jager, U. Sternberg, D. Stachel, [81] J.J. Liang, R.T. Cygan, T.M. Alam, these Proceedings,
D.O. Smith, J. Non-Cryst. Solids 143 (1992) 265. p. 167.
[51] C. Mercier, L. Montagne, H. S®hi, G. Palavit, J.C. Boivin, [82] T.M. Alam, R.K. Brow, in: Proceedings of the 18th
A.P. Legrand, J. Non-Cryst. Solids 224 (1998) 163. International Congress on Glass, San Francisco, CA, July
[52] R.M. Wenslow, K.T. Mueller, J. Phys. Chem. B 102 (1998) 1998, p. 5.
9033. [83] P.I. Tordjman, M. Bagieu-Beucher, R. Zilber, Z. Kristal-
[53] H. Hosono, Y. Abe, K. Deguchi, J. Non-Cryst. Solids 142 logr. 140 (1974) 145.
(1992) 103. [84] C. Baz-D olle, D. Stachel, Z. Kristallogr. 203 (1993) 282.
[54] A. Lai, A. Musinu, G. Piccaluga, S. Puligheddu, Phys. [85] P.J.C. Guitel, I. Tordjman, Acta Crystallogr. B 32 (1978)
Chem. Glasses 38 (4) (1997) 173. 2960.
[55] J.P. Fletcher, R.J. Kirkpatrick, D. Howell, S.H. Risbud, [86] K. Muruganandam, M. Seshasayee, S. Patnaik, Solid State
J. Chem. Soc. Faraday Trans. 89 (17) (1993) 3297. Ionics 89 (1996) 313.
[56] U. Hoppe, J. Non-Cryst. Solids 195 (1996) 138. [87] T.M. Alam, S. Conzone, R.K. Brow, T.J. Boyle, J. Non-
[57] R. Bruckner, H.U. Chun, H. Goretzki, M. Sammet, J. Non- Cryst. Solids 258 (1999) 140.
Cryst. Solids 42 (1980) 49. [88] U. Hoppe, J. Non-Cryst. Solids 183 (1995) 85.
[58] K. Suzuki, M. Ueno, J. Phys. Paris 46 (1985) C8-261. [89] U. Hoppe, G. Herms, W. Gerike, J. Sakowski, J. Phys.:
[59] U. Hoppe, G. Walter, D. Stachel, A.C. Hannon, Z. Natur- Condens. Matter 8 (1996) 8077.
forsch. A 51 (1996) 179. [90] S. Prabakar, R.M. Wenslow, K.T. Mueller, these Proceed-
[60] A.F. Zatsepin, V.S. Kortov, Yu.V. Shchapova, Phys. Solid ings, p. 82.
State 39 (8) (1997) 1212. [91] U. Hoppe, G. Walter, R. Kranold, D. Stachel, A. Barz,
[61] R. Gresch, W. M uller Warmuth, H. Dutz, J. Non-Cryst. J. Non-Cryst. Solids 192&193 (1995) 28.
Solids 34 (1979) 127. [92] B.N. Nelson, G.J. Exarhos, J. Chem. Phys. 71 (1979) 2739.
[62] Y. Kaneko, Yogyo-Kyokai-Shi 88 (1980) 497. [93] R.K. Brow, C.C. Phifer, G.L. Turner, R.J. Kirkpatrick,
[63] Yu.F. Zhuravlev, V.K. Slepukhin, Sov. J. Glass Phys. J. Am. Ceram. Soc. 74 (6) (1991) 1287.
Chem. (Engl. Trans.) 12 (4) (1986) 210. [94] U. Hoppe, D. Stachel, D. Beyer, Phys. Scr. T 57 (1995)
[64] R.K. Brow, D.R. Tallant, S.T. Myers, C.C. Phifer, J. Non- 122.
Cryst. Solids 191 (1995) 45. [95] G.N. Greaves, J. Non-Cryst. Solids 71 (1985) 203.
[96] G.N. Greaves, W. Smith, E. Giulotto, E. Pantos, J. Non- [126] R. Kranold, G. Walter, U. Lembke, T. Rieker, D. Stachel,
Cryst. Solids 222 (1997) 13. J. Non-Cryst. Solids 232±234 (1998) 502.
[97] A.E.R. Westman, P.A. Gartaganis, J. Am. Ceram. Soc. 40
[127] L. Cervinka, J. Bergerov
a, M. Trojan, J. Non-Cryst. Solids
(9) (1957) 293. 192&193 (1995) 121.
[98] J.R. Van Wazer, J. Am. Chem. Soc. 72 (1950) 647. [128] J.T. Krause, C.R. Kurkjian, J. Am. Ceram. Soc. 51 (4)
[99] B.C. Bunker, G.W. Arnold, J.A. Wilder, J. Non-Cryst. (1968) 226.
Solids 64 (1984) 291. [129] K. Matusita, C. Ihara, T. Komatsu, R. Yokota, J. Am.
[100] E.J. Grith, in: E.N. Walsh, E.J. Grith et al. (Eds.), Ceram. Soc. 68 (7) (1985) 389.
Phosphorus Chemistry, ACS Symposium Series 486, [130] J. Swenson, A. Matic, A. Brodin, L. B orjesson, W.S.
American Chemical Society, Washington DC, 1992, Howells, Phys. Rev. B 58 (17) (1998) 11331.
p. 86. [131] A. Musinu, G. Paschina, G. Piccaluga, G. Pinna, J. Non-
[101] C.A. Angell, J. Non-Cryst. Solids 73 (1985) 1. Cryst. Solids 177 (1994) 97.
[102] R. Wasche, R. Br uckner, Phys. Chem. Glasses 27 (1986) [132] E. Matsubara, T. Kawazoe, Y. Waseda, M. Ashizuka, E.
80. Ishida, J. Mater. Sci. 23 (1988) 547.
[103] A. Eisenberg, T. Sasada, J. Polym. Sci. C 16 (1968) 3473. [133] A. Balerna, M. Bionducci, A. Falqui, G. Licheri, C.
[104] R.L. Sammler, J.U. Otaigbe, M.L. Lapham, N.L. Bradley, Meneghini, G. Navarra, M. Bettinelli, J. Non-Cryst. Solids
B.C. Monahan, C.J. Quinn, J. Rheol. 40 (2) (1996) 285. 232±234 (1998) 607.
[105] J.U. Otaigbe, G.H. Beall, TRIP 5 (11) (1997) 369. [134] M. Bionducci, G. Licheri, A. Musinu, G. Navarra, G.
[106] C.F. Callis, J.R. Van Wazer, J.S. Metcalf, J. Am. Chem. Piccaluga, G. Pinna, Z. Naturforsch. A 51 (1996) 1209.
Soc. 77 (1955) 1471. [135] G.N. Greaves, S.J. Gurman, L.F. Gladden, C.A. Spence,
[107] M. Goldstein, T.H. Davies, J. Am. Ceram. Soc. 38 (7) P. Cox, B.C. Sales, L.A. Boatner, R.N. Jenkins, Philos.
(1955) 223. Mag. B 58 (3) (1988) 271.
[108] P.J. Miller, G.J. Exarhos, W.M. Risen, J. Chem. Phys. 59 [136] U. Hoppe, R. Kranold, D. Stachel, A. Barz, A.C. Hannon,
(6) (1973) 2796. J. Non-Cryst. Solids 232±234 (1998) 44.
[109] R. Bruckner, J. Murach, H. Shen, J. Non-Cryst. Solids 208 [137] R. Anderson, T. Brennan, G. Mountjoy, R.J. Newport,
(1996) 228. G.A. Saunders, J. Non-Cryst. Solids 232±234 (1998) 286.
[110] M. Braun, Y. Yue, C. R ussel, C. J
ager, J. Non-Cryst. [138] D.T. Bowron, G.A. Saunders, R.J. Newport, B.D. Rain-
Solids. 241 (1998) 204. ford, H.B. Senin, Phys. Rev. B 53 (9) (1996) 5268.
[111] C. Jager, P. Hartmann, R. Witter, M. Braun, these [139] T. Kanazawa, J. Non-Cryst. Solids 52 (1982) 187.
Proceedings, p. 4554. [140] T. Okura, Y. Yamashita, T. Kanazawa, Phys. Chem.
[112] U. Hoppe, G. Herms, W. Gerike, J. Sakowski, J. Phys.: Glasses 29 (1988) 13.
Condens. Matter 8 (1996) 8077. [141] S. Hayashi, K. Hayamizu, J. Solid State Chem. 80 (1989)
[113] G. Herms, W. Gerike, J. Sakowski, Ber. Bunsenges. Phys. 195.
Chem. 100 (1996) 1625. [142] Z. Yin, M. Kasrai, G.M. Bancroft, K.H. Tan, X. Feng,
[114] G. Herms, J. Sakowski, W. Gerike, U. Hoppe, D. Stachel, Phys. Rev. B 51 (2) (1995) 742.
J. Non-Cryst. Solids 232±234 (1998) 427. [143] J.A. Duy, R.G. Grant, Phys. Chem. Glasses 16 (6) (1975)
[115] R.K. Brow, C.C. Phifer, G.L. Turner, R.J. Kirkpatrick, 124.
J. Am. Ceram. Soc. 74 (6) (1991) 1287. [144] P. Gray, L.C. Klein, J. Non-Cryst. Solids 68 (1) (1984) 75.
[116] I.A. Zhmyreva, V.P. Kolobkov, S.G. Lunter, I.N. Moroz- [145] A.A. Dymnikov, A.K. Przhevuskii, Y.K. Fedorov, V.A.
ova, Y.K. Federov, A.N. Chikovskii, Fiz. Khim. Stekla Chernova, Fiz. Khim. Stekla (Engl. Trans.) 17 (6) (1991) 514.
(Engl. Trans.) 13 (6) (1988) 464. [146] J.A. Duy, G.O. Kyd, Phys. Chem. Glasses 36 (3) (1995)
[117] G.B. Rouse, P.J. Miller, W.M. Risen, J. Non-Cryst. Solids 101.
28 (1) (1978) 192. [147] T.R. Meadowcraft, F.D. Richardson, Trans. Far. Soc. 61
[118] B.N. Nelson, G.J. Exarhos, J. Chem. Phys. 71 (7) (1979) (1965) 54.
2739. [148] B.C. Sales, R.S. Ramsey, J.B. Bates, L.A. Boatner, J. Non-
[119] V.A. Sinyaev, L.V. Levchenko, Fiz. Khim. Stekla (Engl. Cryst. Solids 87 (1986) 137.
Trans.) 18 (2) (1992) 178. [149] F.A. Stover, J. Chromatogr. A 769 (1997) 349.
[120] A. Matic, L. B orjesson, Philos. Mag. B 77 (1998) 357. [150] J. Flory, Am. Chem. Soc. 64 (1942) 2205.
[121] A. Eisenberg, H. Farb, L.G. Cool, J. Polym. Sci. Part A-2 [151] B.C. Sales, L.A. Boatner, J.O. Ramey, J. Non-Cryst. Solids
4 (1966) 855. 232±234 (1998) 107.
[122] Y.K. Voronko, A.V. Gorbachev, A.A. Sobol, Neorgan. [152] B.C. Sales, these Proceedings, p. 155.
Mater. (Engl. Trans.) 28 (3) (1992) 576. [153] B.C. Sales, J.U. Otaigbe, G.H. Beall, L.A. Boatner, J.O.
[123] T. Uchino, T. Yoko, these Proceedings, p. 180. Ramey, J. Non-Cryst. Solids 226 (1998) 287.
[124] B. Rue, S. Beau®ls, Y. Delugeard, G. Coddens et al., [154] L. Montagne, G. Palavit, R. Delaval, J. Non-Cryst. Solids
Mater. Res. Soc. Symp., vol. 407, 1996, p. 155. 215 (1997) 1.
[125] U. Hoppe, G. Walter, D. Stachel, A.C. Hannon, Z. [155] A.M. E®mov, J. Non-Cryst. Solids 209 (1997) 209.
Naturforsch. A 50 (1995) 684. [156] J.E. Dickinson, Chim. Chron. New Series 23 (1994) 355.
[157] X. Yu, D.E. Day, G.J. Long, R.K. Brow, J. Non-Cryst. [166] P.A. Tick, P.A. Phys. Chem. Glasses 25 (1984) 149.
Solids 215 (1997) 21. [167] R.K. Brow, C.C. Phifer, X.J. Xu, D.E. Day Day, Phys.
[158] R.K. Brow, C.M. Arens, X. Yu, D.E. Day, Phys. Chem. Chem. Glasses 33 (2) (1992) 33.
Glasses 35 (3) (1994) 132. [168] P.A. Tick, D.W. Hall, Di. Defect Data 53&54 (1987) 179.
[159] R. Witter, P. Hartmann, J. Vogel, C. J ager, Sol. State [169] H. Hosono, Y. Abe, J. Non-Cryst. Solids 190 (1995) 185.
Nucl. Magn. Reson. 13 (3) (1998) 189. [170] K.C. Sobha, K.J. Rao, J. Non-Cryst. Solids 201 (1996) 52.
[160] K.K. Olsen, J.W. Zwanziger, P. Hartmann, C. J ager, [171] E.M. Vogel, M.J. Weber, D.M. Krol, Phys. Chem. Glasses
J. Non-Cryst. Solids 222 (1997) 199. 32 (1991) 231.
[161] J.D. Wicks, L. Borjesson, G. Bushnell-Wye, W.S. Howells, [172] G.K. Marasinghe, M. Karabulut, C.S. Ray, D.E. Day,
R.L. McGreevy, Phys. Rev. Lett. 74 (1995) 726. M.G. Shumsky, W.B. Yelon, C.H. Booth, P.G. Allen,
[162] R.K. Brow, Z.A. Osborne, Surf. Interf. Anal. 24 (1996) 91. D.K. Shuh, J. Non-Cryst. Solids 222 (1997) 144.
[163] S.V. Dvinskikh, I.V. Murin, A.F. Privalov, A.A. Pronkin, [173] R.K. Brow, D.R. Tallant, W.L. Warren, A. McIntyre,
E. Rossler, H.M. Vieth, J. Non-Cryst. Solids 240 (1998) 79. D.E. Day, Phys. Chem. Glasses 38 (6) (1997) 300.
[164] R.M. Wenslow, K.T. Mueller, J. Non-Cryst. Solids 231 [174] T. Kasuga, Y. Abe, J. Non-Cryst. Solids 243 (1999) 70.
(1998) 78. [175] K.E. Vermillion, P. Florian, P.J. Grandinetti, J. Chem.
[165] B.G. Aitken, D.C. Bookbinder, M.E. Greene, R.M. Phys. 108 (17) (1998) 7274.
Morena, US Patent No. 5,246,890, 21 September 1993; [176] S.H. Wang, J.F. Stebbins, J. Non-Cryst. Solids 231 (1998)
R.M. Morena, these Proceedings, p. 382. 286.

The Structure of Simple Phosphate Glasses

Uploaded by

Copyright:

Available Formats

The Structure of Simple Phosphate Glasses

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Structure of Simple Phosphate Glasses

Uploaded by

Copyright:

Available Formats

Journal of Non-Crystalline Solids 263&264 (2000) 1±28

Review: the structure of simple phosphate glasses

1. Introduction By studying such materials, Van Wazer [2] estab-

The basic building blocks of crystalline and

Fig. 2. Real-space correlation function from a neutron dif-

for v-P2 O5 . They propose that the glass is made up

ordering. That the density of v-P2 O5 is intermedi-

4. Binary phosphate glasses

The addition of a modifying oxide to v-P2 O5

Fig. 5. Densities of Na-ultraphosphate glasses prepared in

ultrasound in Na-ultraphosphate glasses near 25

peak is due to bridging oxygens (P±BO) and the

Na-ultraphosphate glasses with 30±40 mol% Na2 O

4.1.2. Network linkages in ultraphosphate glasses

(region I) and when MTO < CNMe (region II). For

range is consistent with the larger coordination

Kordes identi®ed a number of phosphate by tetrametaphosphate rings to one based on

Fig. 20. Raman spectra collected from xLi2 O(67)x)ZnO

Fig. 19. 31 P MAS NMR spectrum from a 0.67ZnOá0.33P2 O5

You might also like