ARTICLE IN PRESS

Biomaterials 25 (2004) 39073913

Characterisation of commercial ionomer glasses using magic angle

nuclear magnetic resonance (MAS-NMR)
Artemis Stamboulisa, Robert V. Lawb, Robert G. Hilla,*
a

Department of Materials, Imperial College, Prince Consort Road, London SW7 2BP, UK
b
Department of Chemistry, Imperial College, London SW7 2BP, UK
Received 20 February 2003; accepted 10 October 2003

Abstract
Five commercial ionomer glasses (Fuji IX, Ketac Molar, G338, G2, and G2SR) used to produce glass (ionomer) polyalkenoate
dental cements were studied. 29Si, 27Al, 31P and 19F magic angle spinning nuclear magnetic resonance (MAS-NMR) Spectroscopy
was used to characterise the glasses and the resulting spectra compared with previous studies of model glasses. The 29Si NMR
spectra were consistent with Q4(3Al) and Q4(4Al) units being present and agreed with the low non-bridging oxygen contents
calculated from the elemental composition. The 27Al NMR spectra typically exhibited three distinct sites at 4560, 20 and 0 ppm
which have been attributed to Al(IV), Al(V) and Al(VI) coordinate aluminium. The presence of Al(V) and Al(VI) are consistent with
previous studies of model ionomer glasses. The 31P spectra all exhibited a chemical shift between 8 and 23 ppm with the
exception of the Ketac Molar glass, which exhibited a peak at 23 ppm consistent with orthophosphate. The chemical shift of 31P in
the range 8 to 23 ppm indicates a PO4 tetrahedra surrounded by 14 Al moieties.
The 19F NMR spectra indicated the presence of AlFCa(n) in the G2 and G338 glasses, AlFSr(n) in the G2SR and Fuji IX
glasses and crystalline CaF2, LaF3, AlFCa(n) in the Ketac Molar glass. The G338 glass with a high non-bridging oxygen content
showed the presence of a FCa(n) species. There was also present in all the glasses a peak corresponding to AlFNa(n). The
intensity of this peak was approximately proportional to the sodium content.
r 2003 Elsevier Ltd. All rights reserved.
Keywords: Glass ionomer; Dental cement; Glass composition; Glass analysis; MAS-NMR

1. Introduction
The composition and structure of uoro-aluminosilicate glasses used to form glass (ionomer) polyalkenoate cements is critical to the setting reaction. Glass
polyalkenoate cements are formed from reacting aqueous poly(acrylic acid) with a nely powdered uoroalumino-silicate glass [1]. The glass is attacked by the
acid and there is SiOAl bond hydrolysis leading to
the release of metal cations, which are then chelated
by the carboxylate groups and serve to ionically
crosslink the polycarboxylate chains. The number and
type of cations and anions released from the glass
determine the extent of ionic crosslinking of the polysalt
matrix and the properties of the cement. Whilst a great
*Corresponding author. Tel.: +44-20-7594-6783; fax: +44-20-75843194.
E-mail addresses: r.law@ic.ac.uk (R.V. Law), r.hill@ic.ac.uk
(R.G. Hill).
0142-9612/$ - see front matter r 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biomaterials.2003.10.074

deal is known about how glass composition inuences

glass polyalkenoate cement properties [27] there is very
little information on the structure of this type of glass.
The chemical structure of the glass will exert a strong
inuence on subsequent cement formation and the
properties of the resulting cements.
In a recent series of studies, Matsuya et al. [8] and Hill
et al. [911] have characterised simple model glasses
based on 2SiO2.Al2O3.CaO.CaF2, 4.5SiO2.3Al2O3.
1.5P2O5.(5 X)CaO.XCaF2,
4.5SiO2.3Al2O3.1.5P2O5.
3SrO.2SrF2, 1.5SiO2.Al2O3.YP2O5. (1 X Z)CaO.
XSrO.0.5CaF2.ZNa2O using 29Si, 31P, 27Al and 19F
MAS-NMR.
Matsuya et al. [8] have investigated simple SiO2
Al2O3CaOCaF2 glasses using 29Si, and 27Al MASNMR. The results have given valuable information on
the coordination states of aluminium and silicon and
their next nearest neighbours. The 29Si NMR spectra
give information about the average number of nonbridging oxygens (NBO) attached to silicon and the

ARTICLE IN PRESS
A. Stamboulis et al. / Biomaterials 25 (2004) 39073913

3908

number of next nearest neighbour aluminium atoms.

The 27Al NMR spectra are again inuenced by the next
nearest neighbours and the coordination state. In
phospho-uoro-alumino-silicate glasses there is evidence
for AlOP linkages as a result of local charge
compensation of Al3+ by P5+. 31P MAS-NMR shows
the presence of pyrophosphate species and this combined with the data from 27Al spectra indicates the
presence of AlOPO34 type species. 19F studies by Hill
et al. [911] have also been extremely valuable and
indicate the presence of F as FCa(n), AlFCa(n), in
calcium uoro-alumino-silicate glasses and as FSr(n),
AlFSr(n) in the equivalent strontium glasses [11].
Strontium may be substituted for calcium with virtually
no change in the glass structure. In sodium-containing
calcium/strontium uoro-alumino-silicate glasses, the
uorine is also present as an AlFNa(n) species [10].
The present paper describes an investigation of ve
glasses (Fuji IX, Ketac Molar, G338, G2, and G2SR) used
for forming commercially available glass polyalkenoate
cements using 29Si, 31P, 27Al and 19F MAS-NMR.

2. Experimental
2.1. Materials
The sources of the glasses used are given in Table 1,
whilst Table 2 gives a chemical analysis of the glasses
Table 1
Source of the glasses

obtained from X-ray uorescence measurements to give

the Si, Al, P, Na, Sr and La content. Pyrolysis of the
glasses in the presence of water to convert the uorine to
HF was carried out and the resulting uoride ion
concentration measured by ion-selective electrode to
determine the uorine content. The chemical analysis of
the glasses was carried out by CERAM Research
(Queens Road, Stoke-on Trent ST4 7LQ, UK).
2.2. X-ray powder diffraction
X-ray powder diffraction analysis was completed on
the glass powders for qualitative purposes. A Phillips
powder diffractometer (Phillips Xpert diffractometer,
Phillips Eindhoven NL) was used with Cu Ka X-rays.
2.3. MAS-NMR
MAS-NMR analyses were conducted on 29Si, 27Al
and 31P and 19F nuclei at resonance frequencies of 39.77,
52.15, 81.01 and 188.29 MHz, respectively, using an FTNMR spectrometer (DSX-200, Bruker, Germany).
Spinning rates of the samples at the magic angle were
5 kHz for the 29Si, 27Al, and 31P MAS-NMR and 15 kHz
for 19F. Recycle time was 2 s for 29Si and 27Al and 4 s for
31
P in a 7 mm zirconia rotor. For the 19F measurements
the glass powder was packed in a 4 mm zirconia rotor
and the recycle time was 120 s. Reference materials for
the chemical shift (in ppm) were tetramethylsilane
(TMS) for 29Si, yttrium aluminium garnite (YAG) for
27
Al, 85% H3PO4 for 31P and their chemical shift was
adjusted to 0 ppm. The spectra for 19F were referenced
to CaF2 taken as 108 ppm relative to the more
common standard of CFCl3.
In addition 27Al spectra were run on a high eld
(Advance-600, Bruker, Germany) at a spinning rate of
15 KHz with a recycle time of 1 s.

Glass

Supplier

Fuji IX

GC Corporation 22-23, Coopers Court, Newport

Pagnell, Bucks MK16 8JS, UK

G2

CDL St. Marks Ind. Estate Corby, Northants NN18

8AN, UK

G338

CDL St. Marks Ind. Estate Corby, Northants NN18

8AN, UK

3. Results and discussion

G2SR

CDL St. Marks Ind. Estate Corby, Northants NN18

8AN, UK

Ketac
Molar

3M-ESPE Dental AG ESPE Platz, D-822279 Seefeld,

Germany

Table 2 shows the chemical analysis of the glasses

studied in atomic percent. All the glasses were found to
be completely amorphous by X-ray diffraction (XRD),
with the exception of Ketac Molar glass. The XRD
pattern of this glass is shown in Fig. 1. This glass was

Table 2
Chemical analysis of the commercial glasses: G2SR, G2, Ketac Molar, Fuji IX and G338 (values in at%)
Glass

Al

Si

Ca

Sr

La

Na

Fuji IX
G2
G338
G2SR
Molar

0.129
0.140
0.131
0.168
0.104

0.115
0.158
0.098
0.106
0.104

0.017
0.011
0.030
0.027
0.015

0.000
0.038
0.041
0.031
0.060

0.547
0.528
0.536
0.567
0.515

0.126
0.192
0.143
0.061
0.163

0.056
0.000
0
0.061
0

0
0
0
0
0.043

0.010
0.058
0.055
0.007
0.021

ARTICLE IN PRESS
A. Stamboulis et al. / Biomaterials 25 (2004) 39073913

Fig. 1. XRD pattern of Ketac Molar glass showing the presence of

calcium uoride.

Table 3
Calculated NCs and surplus charge per phosphorus
Glass

NC

Surplus charge/P

Al:(Si+P)

Fuji IX
G2
G338
G2SR
Ketac Molar

3.48
3.36
3.52
3.63
2.41

0.59
0.45
1.20
1.85
3.47

0.98
0.83
1.02
1.26
0.87

partially crystalline and the diffraction lines present

corresponded to calcium uoride. The network connectivity (NC) of the glasses was calculated according to
Ray [12] assuming the aluminium is Al(IV), there is local
charge compensation of Al3+ and P5+ and that uorine
forms non-bridging uorines. The low calculated NC
value of the Ketac Molar glass would facilitate
amorphous phase separation (APS) and subsequent
crystallisation to uorite (CaF2). Table 3 also shows the
calculated surplus charge per phosphorus. In this
calculation, the surplus charge is dened as the number
of charges available after charge balancing the aluminium. It must be pointed out again that this calculation
assumes all the aluminium in the form of Al(IV), the
uorine forms non-bridging uorines and the glasses are
homogeneous.
Fig. 2 shows the 29Si MAS-NMR spectra of
commercial glasses. The chemical shift indicates the
environment around the Si atoms in the glass. The
chemical shift observed for Si in a four coordinate state
is between 60 and 120 ppm. When a structural unit
of the alumino-silicate glass network is expressed as
Si(OSi)m n(OAl )n(O )4 m, (Qm(nAl), 4XmXnX0), the
chemical shift increases with decreasing m or increasing
n [13]. The shift therefore depends on the number of
NBO and bridging oxygens (BO) attached to the silicon
and the number of next nearest neighbour Al. Increasing
the number of NBO moves the peak in a positive
direction, whilst increasing the number of next nearest

3909

Fig. 2. 29Si MAS-NMR of G2, G2SR, Ketac Molar G338 and Fuji IX
glasses.

neighbour aluminium atoms moves the peak in a

negative direction. All glasses show a broad peak
around 90 ppm with the exception of G2 and G338,
which exhibit a slightly more negative chemical shift of
about 95 and 99 ppm, respectively. The chemical
shifts observed are what would be expected for Q3(3Al)
and Q4(4Al). The low chemical shift for the G2 glass
could indicate a lower number of next nearest neighbour
Al or a lower number of NBO attached to the Si. The
elemental composition of G2 indicates a relatively low
number of NBO, so the latter is more likely in this case.
The low chemical shift of the G338 glass would not be
expected on the basis of its low NC and high NBO
content. This point will be discussed further after the 19F
spectra are discussed.

4.

27

Al MAS-NMR spectra

The 27Al MAS-NMR spectra of G2, G2SR, Ketac

Molar, Fuji IX and G338 commercial glasses run with
the 200 MHz Bruker instrument are shown in Fig. 3. All
glasses exhibit a peak at 4760 ppm, which is the
dominant species present except in Ketac Molar glass
and G338. This peak corresponds to Al(IV). In simple
model, glasses based on 2SiO2Al2O3CaOCaF2 Al(IV)
was found at 60 ppm [9] and in phosphate containing
glasses where the molar Al:P ratio was 2:1 at about
50 ppm [10]. Since all the commercial glasses have Al:P
ratios o2:1 it would be expected that the Al(IV) peak
would be found between 50 and 60 ppm, but closer to
60 ppm, which is found for all cases apart from G338.
This glass which has the lowest Al:P ratio has, as
expected, the lowest chemical shift of 47 ppm and
indicates a high incidence of AlOP bonds; however,
the chemical shift is slightly lower than that observed
previously for model glasses with lower Al:P ratios [10].

ARTICLE IN PRESS
3910

A. Stamboulis et al. / Biomaterials 25 (2004) 39073913

Fig. 3. Low-eld 27Al MAS-NMR spectra of G2, G2SR, Ketac Molar

and Fuji IX commercial glasses.

There are additional peaks in the 27Al spectra for

most of the glasses. Kohn et al. [14] have proposed the
presence of Al(IV), Al(V) and A(VI) in a uoroalumino-silicate glass with AlF5 and AlF6 species
present with chemical shifts of 60, 20 and 0 ppm,
respectively. Stebbins et al. [15] have also found evidence
of Al(V) and Al(VI) in uoro-alumino-silicate glasses. It
must be pointed out, however, that there is insufcient
uorine present in these glasses for the observed peaks
at 20 and 0 ppm to correspond to AlF5 and AlF6. The
strong peak at 0 ppm in the Ketac Molar glass
apparently indicates that there is more Al(VI) than
Al(IV) present. G2SR shows a peak that corresponds to
the Al(VI) at around 5 ppm a little bit more negative
compared to Ketac Molar glass (ca. 0 ppm). It is worth
noticing that G2 as well as Fuji IX glass exhibit a quite
unsymmetrical and broad peak that suggests the
presence of small amounts of Al(V) and Al(VI) in these
glasses. These glass compositions with the exception of
G2SR and G338 meet the Lowensteins two conditions
for maintaining Al in a four-fold coordination state [16].
Namely, an Al:(Si+P) ratio p1.0 and sufcient P and
network balancing cations to charge balance the chargedecient AlO4 tetrahedron, the presence of Al(V) and
A(VI) is surprising. However, studies on model glasses
[911] with high uorine contents that also meet
Lowensteins conditions also show Al in apparent
coordination states greater than four. Furthermore,
the higher coordination states are associated with the
formation of aluminium uorine species of the type Al
FCa/Sr/Na(n) species. At this point of time, it is
unclear why uorine being in close proximity to
aluminium should force it into higher coordination
states. The high proportion of Al(VI) in the Ketac
Molar glass may occur as a result of APS and
crystallisation of the glass to uorite, during quenching,

which would effectively remove charge balancing

calcium cations from the glass and force Al into higher
coordination states. Given that much of the uorine in
the Ketac Molar glass is present as crystalline uorite
the Al(VI) in Ketac Molar is unlikely to be present as
AlF5/AlF6-type species. Whilst previous studies [9] have
attributed chemical shifts at 20 and 0 ppm to Al(V) and
Al(VI), it cannot be ruled out at this stage that these
peaks are not due to asymmetric four coordinate
AlO3F-type units, strongly dominated by a secondorder quadrupolar effect due the low magnetic eld.
In order to investigate this effect, samples were also
run using a 600 MHz spectrometer. The 27Al spectra
obtained are shown in Fig. 4. The peak half-width
reduced as a result of the higher eld, reducing the
quadrupolar line broadening arising from electric eld
gradients at the nucleus. The peaks corresponding to
Al(V) and Al(VI) are also reduced in amplitude for all
the glasses and are much more clearly resolved
particularly the Al(VI) peak at about 5 ppm. This
suggests that a large part of the signal at about 20 and
5 ppm attributed to Al(V) and Al(VI) previously is due
to AlO4F- and AlO4F2-type species combined with a
strong second-order quadrupolar effect. This phenomenon is most marked with the Ketac Molar spectrum
that shows a markedly reduced proportion of Al(V) and
Al(VI). Three distinct resonances are now observed at
5060 ppm, at about 20 ppm and 0 to 5 ppm corresponding to Al(IV), Al(V) and Al(VI).
Whilst all the glasses have Al(V) and Al(VI) present
the amount is generally small. However, the G338 and
G2SR glasses contain signicant amounts of Al(VI).
The Al: (Si+P) ratios of the glasses are given in Table 3.
The G2SR and G338 glasses break Lowensteins rules in
that the Al:(Si+P) is greater than 1. In the G2SR glass,
this ratio is 1.26 and for G338 the ratio is 1.02; it might
be expected, therefore, that the Al would be present as

Fig. 4. High eld

27

Al MAS-NMR spectra.

ARTICLE IN PRESS
A. Stamboulis et al. / Biomaterials 25 (2004) 39073913

P Chemical Shift (ppm)

Plot of 31P Chemical Shift Against Calculated Surplus Charge per P

31

an Al(VI) species in both these glasses and the G2SR

glass has a signicant proportion of Al(VI). However,
the G338 glass is only just outside Lowensteins rules,
but has the highest proportion of Al(VI) present. The
explanation for this will become apparent when the 19F
spectra are discussed.

5.

31

P MAS-NMR spectra

Fig. 5 shows the 31P MAS-NMR spectra of the

commercial glasses. Fuji IX glass shows a strong broad
peak at around 18 ppm. This peak could be attributed
to phosphorus in a pyrophosphate environment; however, the shift is somewhat too negative and more
negative than that observed in previous studies of model
glasses [10,11]. Looking at the composition of Fuji IX
glass, it is possible that this glass has been designed so
that all Sr and phosphorus charge balance the aluminium so there are virtually no NBO in the glass
structure. Consequently, it is not possible to form Al
PO34 species, since there are insufcient surplus cations
to charge balance the PO34 ion. However the chemical
shift is too low for an AlPO4 (ca. 29 ppm) [17] type
species in which PO4 is surrounded by four Al. It is
likely, therefore, that the PO4 is present in a combination of states between these two species, but probably
surrounded by three or more Als. The calculated surplus
charge per phosphorus is shown in Fig. 6 for all the
glasses except Ketac Molar. This glass was not included,
since it is both phase separated and crystalline. The
calculation is only strictly valid if the chemical shifts
observed for 27Al NMR at 20 and 0 ppm are not
attributable to Al(V) and Al(VI) and all the uorine
forms non-bridging uorines, rather than complexing
calcium, strontium or sodium. The 31P chemical shift
would be expected to decrease in the order P(0Al),
P(1Al), P(2Al), P(3Al), P(4Al) where P(0Al) refers to an

Fig. 5. 31P MAS-NMR spectra of G2, G2SR, Ketac Molar and Fuji
IX commercial glasses.

3911

5
0
-5 0
0.5
-10
-15
Fujii IX
-20
-25 G2
-30 AlPO4
-35

Ca3(PO4)2
1

1.5

2
2.5
G2SR

3.5

4.5

G338 Calculation not valid

since F-Ca(n) present
Surplus Charge/P

Fig. 6. 31P Chemical Shift plotted against the calculated surplus

charge/P atom and compared to the theoretical expected chemical shift
based on Ca3(PO4)2 and AlPO4 Note the increase in the 31P chemical
shift with decreasing surplus charge/P for the G2 Fuji IX and G2SR
glasses.

orthophosphate ion and P(4Al) refers to AlPO4, i.e. a

PO4 tetrahedron surrounded by four Al. The chemical
shifts for AlPO4 and Ca3(PO4)2 are plotted for
comparison. It would be expected that the counter ion
for charge balancing NBO on the P(3Al), P(2Al), P(1Al)
and P(0Al) would inuence the observed chemical shift.
However, since Ca2+ and Sr2+ have similar charge-tosize ratios it would not be expected that substitution of
Sr for Ca would inuence the observed 31P chemical
shift. The observed 31P chemical shift of the orthophosphate ion in calcium uorapatite is almost identical to
that in strontium uorapatite. In contrast, whether
sodium or calcium charge balances the NBO on the
phosphorus would be likely to have a marked inuence.
Sodium would be expected to increase the observed 31P
chemical shift based on studies of substituting Na2O for
CaO in bioactives glasses [18,19]. However, it must be
pointed out that there is very little sodium present in
these glasses and the 19F data discussed later indicates
that most of the sodium forms AlFNa(n)-type species.
So consequently, most of the sodium would be unavailable for charge balancing of any phosphate species.
The chemical shift ( 20) for the G2 glass is similar to
that for the Fuji IX glass at about 18 ppm. It has a
similar surplus charge to phosphorus ratio so this is
expected. The G2SR glass has a higher surplus charge to
phosphate ratio and consequently there are more cations
available. Consequently, we would expect each PO4
tetrahedron to be surrounded by fewer aluminium
atoms and this explains the increased chemical shift of
8 ppm, indicating the presence of phosphorus in an
AlPO34 pyrophosphate-type environment similar to
that found in our previous studies on model/experimental glasses [10,11].
The G338 glass has a higher surplus charge to
phosphorus ratio, consequently we would expect an
even higher chemical shift; however, the chemical shift
observed is even more negative than the Fuji IX and
G2SR glasses at 23 ppm. This may be explained (as
will be seen later) by the fact that uorine complexes

ARTICLE IN PRESS
A. Stamboulis et al. / Biomaterials 25 (2004) 39073913

3912

calcium thereby reducing the availability of calcium for

forming NBO and resulting in less cations available for
charge balancing phosphate groups. Previous studies of
the G338 glass have suggested that this glass has
undergone APS during quenching from the melt.
Wasson and Nicholson [20] concluded from ion release
data that this glass had separated into two cocontinuous phases, one phosphorus and calcium rich
and the other aluminium and silicon rich. Wood and
Hill [21] showed this glass to undergo APS and
crystallisation to uorapatite on slow cooling. The
MAS-NMR data indicate that no signicant phase
separation had occurred with this batch of the G338
glass.
In contrast, Ketac Molar glass exhibits a peak at
around 2 ppm, which can be attributed to phosphorus in
orthophosphate environment. This again may be
indicative of this glass having undergone APS and
crystallisation. It is also worth noticing that the
calculated surplus charge per phosphorus in this glass
is very high at over 12; as a consequence, there are
plenty of cations present for charge balancing the PO34
group.

6.

19

F MAS-NMR spectra

19

F MAS-NMR spectra of the glasses are shown in

Fig. 7. The spectrum of Fuji IX glass has only one broad
peak at around 150 ppm. The peak at 150 ppm
probably corresponds to AlFSr(n) species, similar to
the AlFCa(n) species, which strongly agrees with the
literature value ( 150 ppm) of Zeng and Stebbins [22].
Model strontium uoro-alumino-silicate glasses also
exhibit a similar peak at 150 ppm [10,11]. G2SR that
also contains Sr shows a similar peak, which is more

Fig. 7. 19F MAS-NMR spectra of G2, G2SR, Ketac Molar and Fuji
IX and G338 commercial glasses.

symmetrical compared to Fuji IX glass, but still very

broad and it again can be attributed to uorine in AlF
Sr(n) environment. G2 has a different spectrum. As
expected the rst peak is around
150 ppm and
corresponds to F in an AlFCa(n) environment. The
second peak, however, has a chemical shift at about
180 ppm. According to Stebbins and Zeng [23] a peak
at about 220 ppm corresponds to FNa(n) sites in
silicate glasses. In previous model glasses containing Na,
we have observed a peak at about 185 ppm which
Stebbins [23] attributes to AlFNa(n). It is likely then
that this peak is shifted slightly by sitting on the tail of
the 150 ppm peak and corresponds to AlFNa(n).
According to Stebbins [23], there is a competition
between Ca and Na for F and it is almost certain that
in mixed glasses there is a preference for FCa over
FNa. However, in G2 there is no indication of F in an
FCa environment since the peak at 80 to 110 ppm,
which would correspond to FCa(n) is absent. Ketac
Molar glass has a completely different behaviour. The
spectrum exhibits the three typical broad peaks at
around 70 ppm that correspond to FCa(n) sites, a
relatively sharp 108 ppm peak that corresponds to a
CaF2 environment and 150 ppm that is attributed to
AlFCa(n) sites, respectively. A very weak peak at
about 25 ppm may also be present that possibly may
correspond to very small amounts of LaF3. Youngman
and Dejneka [24] found LaF3 to exhibit two peaks at 24
and 19 ppm. The previous XRD analysis of this glass
detected peaks that approximately match to crystalline
CaF2. There is also evidence for APS and the presence
of cubic crystals from TEM studies of this glass and a
related glass [25,26]. These facts together indicate that
the peak at 108 ppm is probably due to crystalline
CaF2. There is also a weak peak at about 180 ppm that
again probably corresponds to AlFNa(n) type species.
The absence of peaks corresponding to FSr(n) and
FCa(n) below 108 ppm in G2, G2SR and Fuji IX
glass is expected on the basis of their chemical
composition, since in these three glasses there are just
sufcient metal cations to charge balance the Al and
maintain it in Al(IV) state. The G338 glass, however,
has a higher NBO content and exhibits a peak at
between 40 and 80 ppm due to FCa(n) as well as a
peak at 148 ppm due to AlFCa(n). There is also a
strong additional peak at 185 ppm due to AlFNa(n)
which is very pronounced and is presumably due to the
high sodium content of this glass.
The presence of FCa(n) in the G338 glass indicates
that calcium is complexing uorine and species such as
CaF+ rather than Ca2+ may be present, and as a
consequence the calculated NC values will be too low
and the calculated surplus charge per phosphorus gures
will be too high. This may partly explain the low
observed chemical shifts observed in the 31P and 29Si
spectra for this glass. In addition, there may be

ARTICLE IN PRESS
A. Stamboulis et al. / Biomaterials 25 (2004) 39073913

insufcient charge balancing cations to maintain the

Al(IV) state. Assuming 50% of the Ca is present as F
Ca(n) and n=1 the calculated charge available to charge
balance Al per Al is 0.8, thus there may be insufcient
cations to maintain the Al(IV) state and this may
explain the high proportion Al(VI) present in the G338
glass.

7. Conclusions
MAS-NMR gives very useful information on the
structure of ionomer glasses. The type of species present
is strongly dependent on the chemical composition. The
commercial glasses have silicon structural units that
correspond to Q4(4Al) and Q4(3Al). There is strong
evidence for AlOP bonds and phosphorus and
aluminium locally charge balancing one another in the
glass network. In glasses with high NC, the PO4
tetrahedron is probably linked to three or more
aluminium atoms. In the glasses with a lower NC and
higher NBO content, there is evidence for FCa/Sr(n)
species, whilst in the glasses with higher NCs only AlF
Ca/Sr(n) species are present. The aluminium is present
as a Al(IV), but may also be present Al(V) and Al(VI)
states.

Acknowledgements
The authors would like to thank the EU DGXII for
funding ULTRASET No:G5RD-CT-2001-00475.

References
[1] Hill RG, Wilson AD. Some structural aspects of glasses used in
ionomer cements. Glass Technol 1988;29:1507.
[2] Wilson AD, Prosser HJ, Crisp S, Lewis BG, Merson SA.
Aluminosilicate glasses for polyelectrolyte cements. Ind Eng
Chem Prod Res Dev 1980;19:2638.
[3] Grifn S, Hill RG. Inuence of glass composition on the
properties of glass polyalkenoate cements: part I inuence of
aluminium to silicon ratio. Biomaterials 1999;20:157986.
[4] Grifn S, Hill RG. Inuence of glass composition on the
properties of glass polyalkenoate cements: part II inuence of
phosphate content. Biomaterials 2000;21:399403.
[5] De Barra E, Hill RG. Inuence of glass composition on the
properties of glass polyalkenoate cements part III. Inuence of
uorite content. Biomaterials 2000;21:5639.

3913

[6] Grifn SG, Hill RG. Inuence of glass composition on the

properties of glass polyalkenoate cements part IV. Inuence of
uorine content. Biomaterials 2000;21:6938.
[7] de Barra E, Hill R. Inuence of alkali metal ions on the fracture
properties of glass polyalkenoate cements. Biomaterials
1998;30:495502.
[8] Maeda T, Matsuya S, Ohta M. Effects of CaF2 addition on the
structure of CaOAl2O3SiO2 glasses. Dent Mater J 1998;17:
10414.
[9] Hill R, Stamboulis A, Law RV. Characterisation of the structure
of calcium alumino-silicate and calcium uoro-alumino-silicate
glasses by magic angle nuclear magnetic resonance (MAS-NMR).
J Non-Cryst Solids, in press.
[10] Stamboulis A, Hill R, Law R, Characterisation of uorine content
glasses by 19F, 27Al, 29Si and 31P MAS-NMR spectroscopy. in
preparation.
[11] Hill R, Stamboulis A, Law RV. The inuence of strontium
substitution in uorapatite glasses and glass-ceramics. J NonCryst Solids, submitted for publication.
[12] Ray NH. Inorganic polymers. London: Academic Press; 1978.
[13] Engelhardt G, Noftz M, Forkel K, Wishmann FG, Magi M,
Samson A, Lippma E. Structural studies of calcium aluminosilicate glasses by high resolution solid state 29Si and 27Al magic
angle spinning nuclear magnetic resonance. Phys Chem Glasses
1985;26:15765.
[14] Kohn SC, Dupree R, Mortuza MG, Henderson CMB. Evidence
for ve and six coordinated aluminium uoride complexes in F
bearing alumino-silicate glasses. Am Mineral 1991;76:312.
[15] Stebbins JF, Kroecker SJ, Lee SK, Kiczenski TJ. Quantication
of ve and six coordinated aluminium ions in aluminosilicate and
uoride containing glasses by high eld, high resolution 27Al
NMR. J Non-Cryst Solids 2000;275:16.
[16] Lowenstein W. The distribution of aluminium in the tetrahedra of
silicates and aluminates. Am Mineral 1954;39:924.
[17] Dollase WA, Merwin LH, Sebald A. Structure of Na3 3xAlxPO4,
x=0 to 0.5. J Solid State Chem 1989;83:1409.
[18] Dupree R, Holland D, Mortuza MG, Collins JA, Lockyer MWG.
Magic angle spinning NMR of alkali phospho-alumino-silicate
glasses. J Non-Cryst Solids 1989;112:1119.
[19] El Gayar I, Hill RG. Boccaccini A, Abiliev A. Structural analysis
of bioactive glasses. J Non-Cryst Solids, submitted for publication.
[20] Wasson EA, Nicholson JW. A study of the relationship between
setting chemistry and properties of modied glass polyalkenoate
cements. Br Polym J 1990;23:17983.
[21] Wood D, Hill RG. Structure property relationships in ionomer
glasses. Clin Mater 1991;7:30112.
[22] Zeng Q, Stebbins JF. Fluoride sites in aluminosilicate glasses:
high resolution 19F NMR results. Am Mineral 2000;85:86386.
[23] Stebbins JF, Zeng Q. Cation ordering at uoride sites in silicate
glasses: a high resolution 19F NMR study. J Non-Cryst Solids
2000;262:15.
[24] Youngman RE, Dejneka MJ. NMR studies of uorine in
alumino-silicate glasses and glass-ceramics. J Am Ceram Soc
2002;85:107782.
[25] Ngo H. Personal communication.
[26] Hatton PV, Brook IM. Characterisation of the ultrastructure of
glass-ionomer (polyalkenoate) cement. Brit Dent J 1992;173:2757.