Biomaterials 23 (2002) 21912200

The glass ionomer cement: the sources of soluble uoride

J.A. Williamsa,*, R.W. Billingtonb, G.J. Pearsonb
a

Department of Biomaterials, Eastman Dental Institute for Oral Health Care Sciences, 256, Grays Inn Road, London, WC1X 8LD, UK
b
Department of Biomaterials in Relation to Dentistry, Queen Mary & Westfield College, Mile End Road, London, E1 4NS, UK
Received 16 March 2001; accepted 12 October 2001

Abstract
This study aimed to investigate certain processes of uoride production which enable glass ionomer cements to leach
uoride. Two uoroaluminosilicate glasses, G338 and LG26 were used. The free and total uoride which could be dissolved
from the glasses was measured, before and after acetic acid washing. Both glasses contained appreciable amounts of soluble uoride
prior to any acid treatment. The latter process reduced the amount to some 75% of the original levels. Replacing the customary
polymeric acid with propionic acid produced a cement which disintegrated in water allowing the amount of uoride generated by the
cement forming process to be measured. Cement production increased soluble uoride by a further 3%. Both glasses behaved
similarly when undergoing the various processes. G338 produced signicantly greater quantities of uoride, of the order of 10,
compared with LG26 although containing only three times the amount of uoride in the glass formula. A substantial proportion, over half, of the total uoride was complexed especially after contact with cement and when G338 was used. During the
period of the experiment, 21 days, total uoride release did not seem to depend on the square root of time. r 2002 Elsevier Science Ltd.
All rights reserved.
Keywords: Fluoride release; Glasses; Cements

1. Introduction
Fluoride (F) release from glass ionomer cements (gics)
and their predecessors, the silicate and silicophosphate
cements, has been investigated for several decades
following the observation of a reduced incidence of
secondary caries in association with these materials.
Leachable uoride had not been intentionally included
to improve the suitability of the cement for dental use
but occurred as a constituent of one of the several
chemicals used in the manufacture of the glass. Each gic
may use a specialised glass and differ also in the type
and amount of polymeric acid and tartaric acid
constituents. Silicate and silicophosphate cements combine a glass with an aqueous solution of, mainly,
phosphoric acid. Measurements of uoride release
found all three types of cement to release similar
quantities of uoride for a number of years with part
of the release prole showing a linear relationship with
*Corresponding author. Tel.: +44-20-7915-1133; fax: +44-20-79151133.
E-mail address: j.williams@eastman.ucl.ac.uk (J.A. Williams).

the square root of time of cement immersion [13].

Fluoride release from gics was found to be related to the
specimen surface area but not to specimen volume [4]
and a number of investigations have attempted to
characterise release by mathematical equations [5,6]. All
these studies have indicated that release of uoride from
the cement is controlled by a diffusion process.
However, such experiments have measured only the
nal stage in the process of uoride migration, namely
the movement of uoride leaching from the outer
surface of the cement into a surrounding aqueous
medium. It is not clear how, and from where, the
uoride ion arrives at the cement surface. All intrinsic
uoride, i.e. uoride which has not been taken up from
an outside source, must come originally from the glass in
the cement. A restorative cement typically contains
about 75% glass which in turn may contain 1325%
uoride ion. The cements may, therefore, contain
amounts of the order of 150 mgF/g. It has been said
that the leachable uoride is produced by acid attack
upon the glass during the cement forming process [79]
but the means by which uoride initially associated with
the glass becomes uoride ultimately released from the

0142-9612/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 2 - 9 6 1 2 ( 0 1 ) 0 0 3 5 2 - 0

2192

J.A. Williams et al. / Biomaterials 23 (2002) 21912200

outer surface of a cement is poorly understood and the

aim was to investigate aspects of this process.
Apart from work carried out by Craneld [10] none of
the investigations of uoride release have measured the
amount of water-soluble uoride present in the glass
itself prior to any cement formation. This then was the
initial objective. Secondly, it was proposed to measure
the amount of uoride released from the glass into water
acidied to the same degree as it would have been if the
polymeric and tartaric acids present in a specimen of
commercial gic were available. Thirdly, since glasses
may have some form of surface treatment involving acid
washing [9,11] prior to use as a cement, it was proposed
to measure soluble uoride after this process. Finally,
the intention was to measure the ultimate amount of
uoride which could be released from a cement made
from this glass. That is, to measure the amount of
uoride created by the cement forming process, rather
than measure the slow diffusion of F to the surface of
the cement and thence into water over a lengthy period.
In order to do this the cement needed to be disintegrated
and this was accomplished by replacing the polymeric
acid solution normally used for a gic with propionic acid
solution. The latter chemical is equivalent to a single
unpolymerised unit of polyacrylic acid. Cements made
in this way, although capable of setting, disperse in
water when shaken.

2. Materials and methods

2.1. MaterialsFglass
Two types were used, G338 and LG26. G338 glass,
formulated by the Laboratory of the Government
Chemist (Teddington, Middlesex, UK), forms hard
setting cements when blended with aqueous solutions
of polyacrylic and tartaric acids [12]. It had a uoride
content of ca. 18.6% [9].
LG26 (Limerick University, Eire) was a sodium-free
glass previously used in a study by Hadley et al. [13] and
contained 5.2% F. While formulated for surgical
cementation it could be used at a restorative consistency
when mixed with polymeric acid solution.
Portions of both glasses were treated with an aqueous
solution of 5% acetic acid (Analar grade. BDH/Merck,
Poole, UK) for 1 h before ltering through hardened
lter paper (S & S grade 595, Aldrich, Gillingham,
Dorset, UK) and washing with deionised water to
remove residual ions and drying. The absence of
cloudiness in the ltrate was taken as evidence of
retention of ne particles by the lter paper. After
treatment the glass was designated TRE.
Initial experiments showed that both glasses in the as
received state (RAW) gave cements of working time

Table 1
Formulation of cement

Glass
Acidic component

Water
a

Experimental

Cheml IIa

73.7
14.3
Propionic &
tartaric acids
12.2

72.9
14.3
Polyacrylic &
tartaric acids
12.8

From manufacturers data.

insufcient to allow handling. Therefore, cements were

made only from TRE glass.
2.2. MaterialsFcement
An aqueous solution of 52.6% propionic acid
(99.5+% grade, Aldrich, Gillingham, Dorset, UK)
and 1.7% tartaric acid (BDH/Merck, Poole, UK) was
made in distilled water (LIQ). When mixed with TRE
glass at a powder/liquid ratio of 2.8/1.0, the nal cement
composition was calculated as containing 73.7% glass,
14.3% acids and 12.0% water, similar percentages to
those given by manufacturers data for ChemFil II
cement (Table 1).
2.3. Method
Portions of RAW glass (0.1200.180 g) were
weighed and added to 20 ml portions of distilled
water (water) in plastic vials. These were immediately shaken and stored at 371C. After 24 h, a 2 ml
portion of liquid was removed for analysis and the vial
shaken. This was repeated at 7, 14 and 21 days. The
same process was repeated using TRE glass. This
process was then carried out using RAW and
TRE glasses where distilled water was replaced by
water containing 0.014 m propionic acid/tartaric acid
(AcW). This molarity was equivalent to the acidity
produced by dispersing 0.15 g cement in 20 ml water
assuming that no acid had been consumed in forming
the cement.
Finally, portions of TRE glass were weighed and
mixed with LIQ to give a cement. Sufcient amount was
made to ll four metal ring moulds of 10 mm diameter
and 1 mm height from one mix. Three moulds
were clamped between acetate sheet and metal plates
and placed in an environment of high humidity (>80%)
at 371C. The fourth was pressed between acetate
sheet and metal plates for 5 min before transferring
the friable disc from the mould into a weighed
vial containing 20 ml distilled water. The vial was
shaken to disperse the cement and re-weighed to nd
the weight of cement added. The remaining discs were
likewise treated after 60 min, 24 h and 7 days. After 7
days the disc was set sufciently hard to require

J.A. Williams et al. / Biomaterials 23 (2002) 21912200

2193

Fig. 1. Schematic diagram of experimental protocol.

dispersion using a glass rod. The vials were maintained

at 371C from which 2 ml portions of liquid were
removed after 1, 7, 14 and 21 days. The whole scheme
is summarised in Fig. 1.
Six specimens were prepared for each glass/liquid and
cement/water combination. Water without the addition
of glass or cement acted as blanks.
Fluoride ion concentration was measured using a
calibrated uoride ion electrode system (Orion 801
meter, 913500 pH and 9409SC F electrodes, Orion
Research, Cambridge, MA, USA ). Standards contained
0.011000 mg/l F in decade steps. Both free (FF) and
total uoride (TF) were measured, the former before
adding buffer and the latter following addition of an
equal volume of total ionic strength adjustment buffer
(TISAB) IV. Samples and standards were treated in the
same manner.
This gave a measure of the cumulative uoride
ion found in the liquid for contact periods of 121
days. Since the cement was disintegrated no surface area
could be accurately measured and furthermore the glass
was present in powder form. Fluoride measurements
were therefore calculated as mgF/g glass or cement,
after subtraction of blank values. Since the cement
contained only 73.7% glass the amount of uoride
released from the cement was adjusted to be equivalent
to 100%.

3. Results
3.1. Analysis of the acetic acid filtrate after treatment of
glass
This was also analysed for F content. The ltrate from
both glasses contained 5 mg/l FF. TF was higher with
G338 glass producing 19 mg/l and LG26 giving 28 mg/l.
The addition of an equal volume of TISAB IV was not
sufcient to raise the pH to the level of 5.05.5 required
and at least three aliquots of buffer were needed. This
may have introduced additional errors.
3.2. Analysis of sample solutions
Blank solutions contained 0.02 mgF/l or less with no
signicant difference between FF and TF.
All other solutions, water and AcW which had been in
contact with RAW and TRE glass and cement
contained both FF and TF with TF always greater
than FF.
3.3. G338 RAW glass (Table 2)
1.83.4 mg/g FF were found in water with no
signicant change (t-test, po0:05) resulting from increasing contact time. TF (see Fig. 2) was higher than

J.A. Williams et al. / Biomaterials 23 (2002) 21912200

2194

Table 2
Free and total uoride release from G338 as mg/g and (standard deviation)
Time (days)

Combination

Free or total (F/T)

RAW/Wa

F
T
F
T
F
T
F
T

RAW/AcWa
TRE/Wa
TRE/AcWa

Actual release from cement

CEM/5 min
F
T
CEM/60 min
F
T
CEM//1day
F
T
CEM/7day
F
T
Adjusted for weight in cement (100/73.7)
CEM/5 min
F
T
CEM/60 min
F
T
CEM/1day
F
T
CEM/7day
F
T
Pooled mean
F
T

14

21

1.80
3.67
2.11
7.08
1.15
1.87
4.84
10.05

(1.04)
(0.46)
(0.84)
(2.82)
(0.16)
(0.60)
(0.61)
(3.59)

1.62
5.49
1.88
7.78
2.97
3.19
6.13
6.77

(0.52)
(2.16)
(0.53)
(2.93)
(2.48)
(0.92)
(3.25)
(2.19)

3.53
5.67
1.31
6.76
1.27
3.59
5.40
6.48

(0.93)
(0.71)
(0.80)
(2.28)
(0.51)
(1.25)
(1.90)
(1.92)

2.13
5.58
0.36
10.71
1.23
4.18
1.45
7.75

(0.84)
(1.14)
(0.04)
(1.28)
(0.68)
(0.52)
(0.82)
(3.09)

0.35
3.90
0.27
3.34
0.43
7.96
0.60
3.01

(0.02)
(0.25)
(0.05)
(0.06)
(0.31)
(1.00)
(0.86)
(2.22)

0.20
4.42
0.21
4.10
0.68
5.38
0.59
5.50

(0.04)
(0.37)
(0.02)
(0.68)
(1.01)
(3.32)
(0.02)
(3.62)

0.25
4.70
0.38
6.41
0.33
6.02
0.44
5.05

(0.06)
(0.90)
(0.23)
(4.19)
(0.17)
(2.29)
(0.26)
(6.32)

0.29
6.41
0.24
6.97
0.23
6.64
0.34
5.97

(0.07)
(2.75)
(0.07)
(3.77)
(0.03)
(3.58)
(0.24)
(4.03)

0.48
5.20
0.36
4.45
0.57
10.61
0.80
4.01
0.55
6.07

FF at all time intervals, signicantly so (po0:05) after

21 days. TF increased as contact time increased and
reached 5.6 mg/g after 21days.
In AcW FF was 1.32.1 mg/g. There was no
signicant change in FF resulting from the increased
acidity present. TF was signicantly or highly signicantly (po0:01) greater than FF. However, only the
increase after 21 days from the 5.6 mg/g found in water
to 11.0 mg/g in AcW was signicant. The increase
between 1 and 14 days in AcW was signicant.
In neither liquid was TF release proportional to
(time)1/2.
3.4. G338 TRE glass (Table 2)
In water FF ranged from 1.2 to 3.0 mg/g, generally
slightly less than from RAW/water but not signicantly
so. Like the RAW/water combination there was little
increase between 1 and 21days contact time. There was
less TF, 75% compared to that found for the RAW/
water combination and this reduction was signicant or
highly signicant. The maximum was 4.2 mg/g after 21
days and only the increase between 1 and 21 days was
signicant. Although TF increased with contact time
(Fig. 3), there was no dependence on (time)1/2.

0.27
5.89
0.28
5.47
0.91
7.17
0.79
7.33
0.56
6.46

0.33
6.27
0.51
8.55
0.40
8.16
0.59
6.73
0.46
7.43

0.39
8.55
0.32
9.20
0.31
8.85
0.45
7.96
0.38
8.64

In general, when AcW was used in combination with

TRE both FF and TF were signicantly higher than
when water was used. FF reached a maximum of 6 mg/g
maximum at 7 days before decreasing between 7 and 21
days. TF did not signicantly change with contact time;
after an initial peak of 10.0 mg/g it remained relatively
steady and was 7.8 mg/g after 21 days.
3.5. G338 cement (Table 2)
As can be seen in Fig. 4 only 0.20.7 mg FF/g was
produced regardless of cement maturity or contact times
of 121 days. Even when adjusted to the same weight
basis, less FF was found to have been produced from
cement compared to RAW or TRE glass in contact with
water or AcW.
TF was some 20 times higher than FF with all four
cement maturation times producing similar amounts.
TF from the cements was higher than TF from TRE/
water, RAW/water and TRE/AcW but slightly less than
RAW/AcW. After 21 days TF was 89 mg/g.
Since the cement maturity was not a factor the results
at each sampling period for the four maturation times
(5 min to 7 days) were pooled and the mean calculated.
On this basis TF increased as contact time progressed

J.A. Williams et al. / Biomaterials 23 (2002) 21912200

Fig. 2. Total cumulative uoride released from RAW G338 and LG26
glasses.

although the increase between 1 and 21 days was

not signicant. TF was not related to (time)1/2.

2195

Fig. 3. Total cumulative uoride released from TRE G338 and LG26
glasses.

increase both FF and TF (po0:01) compared to the

glass in contact with water.

3.6. LG26 RAW glass (Table 3)

When shaken with water FF was very low, approximately 0.1 mg/g. This remained relatively constant
throughout the 21 day period. It was signicantly less
than FF from G338 RAW.
TF at o0.5 mg/g was also signicantly lower than
that for G338. TF increased steadily as contact time
progressed (Fig. 2) reaching a maximum of 0.45 mg/g
after 21 days, the increase between 14 and 21 days being
signicant at 95% level.
The substitution of AcW for water produced a highly
signicant increase in FF especially during the rst 24 h
when it was 0.36 mg/g. FF continued to rise to a
maximum of 0.72 mg/g after 7 day and then decreased,
until by 21 day the amount was 0.10 mg/g and not
signicantly different to RAW glass.
TF was also higher than when water was used (Fig. 2).
After 24 h 1.25 mg TF/g was released and 2.1 mg/g after
21 days. The general effect of AcW was to signicantly

3.7. LG26 TRE glass (Table 3)

In water FF levels were close to zero. TF (Fig. 3)
was higher than FF and was not signicantly different
from levels produced by the RAW LG26/water combination. The amount increased steadily with time to
reach 0.43 mg/g after 21 day. Only between 1 and 14
days was the increase signicant.
When AcW was used FF was not signicantly
different to when water was used, apart from a
high, and possibly spurious, result at 7 day. TF in
AcW (Fig. 3) showed a sharp increase during the rst
24 h of contact giving a level of 0.91 mg/g. Thereafter it
rose to reach 1.37 mg/g after 21 day. There was no
signicant difference between RAW/AcW and TRE/
AcW LG26.
Fluoride release from LG26 glass was always very
much less than that produced by G338 glass (Fig. 5).

2196

J.A. Williams et al. / Biomaterials 23 (2002) 21912200

comparable to other studies measuring uoride release

from monolithic gics, that is where uoride is released
via diffusion related processes. There are some variations between these studies. For example, amounts
equivalent to 0.11 g in 20 ml [1], 0.12 g in 20 ml [2] and
0.02 g in 20 ml [6] have been used (assuming the density
of cement to be 2.0 g/cm3). With glass and
cement present as particles in the present study,
not as blocks of material, little or no diffusion is likely
to occur and therefore build-up of uoride ion is
unlikely to supress uoride release (as can happen
in a diffusion related process). The limiting factor
here is probably the solubility of the uoride compounds. Nevertheless, the experiment was terminated at
21days where there was still a considerable volume of
water remaining. Another consideration was the difculty of excluding stray particles if the liquid level were
too low.
It was clear from the results that the glasses
themselves contained soluble uoride and that each
process undergone by them altered the level of soluble
uoride. In addition the process might also affect the
ratio of free to total uoride. The effect that each
process has had is discussed below under separate
headings.
4.1. The presence of fluoride before cement formation
occurs
Fig. 4. Total cumulative uoride released from G338 and LG26
cements.

3.8. LG26 cement (Table 3)

Within experimental variation all four cements
released similar levels of FF, with maturity of cement
prior to dissolution having no signicant effect. The
levels were not dissimilar to that of TRE/AcW. The
amount decreased as the contact time increased and for
two of the cements (60 min and 1 day maturity) this
decrease was signicant between 1 and 21day.
TF increased signicantly as the length of contact
with water increased to reach 12 mg/g after 21 days but
like G338 was not dependent on (time)1/2. Although
there were some signicant differences caused by the
maturity of the cement these were not uniform. Once
adjusted on a weight basis TF was higher than for
RAW/water and TRE/water and similar to RAW/AcW
and TRE/AcW.

4. Discussion
The amounts of glass and cement (approximately
0.15 g) and water in which they were dispersed were

Far from relying upon the cement formation to

produce the soluble uoride later released from cement
both G338 (in particular) and LG26 glass powders
contained soluble uoride before any treatment or
cement formation occurred. The origin of this soluble
uoride is as yet undetermined but one possibility as a
source is the milling process used to reduce the glass to a
particle size suitable for cement use. This process
normally involves crushing larger particles of glass with
a dense, refractory medium such as quartz pebbles,
steatite balls or zirconia cylinders under dry conditions.
As a result the surface may be left coated with a ne,
reactive, dust, some or all of which could dissolve in an
aqueous medium.
Although the G338 glass composition contained
about three times as much uoride than LG26, 18.6%
compared with 5.2% F, the amount found in water after
21 days contact with RAW glass was higher by a factor
of ten. It is not known whether this higher uoride
release into water is a consequence of physical or
chemical effects. G338 may contain more ne material
of the type suggested above or it may be attributable to
differences in chemical composition. Both glasses
contain aluminium and calcium but G338 glass contains
8.2% sodium oxide (Na2O) equivalent to 6% sodium
whereas LG26 is essentially sodiumfree. Sodium
uoride (NaF) is readily soluble in water up to a level

J.A. Williams et al. / Biomaterials 23 (2002) 21912200

2197

Table 3
Free and total uoride release from LG26 as mg/g and (standard deviation)
Time (days)

Combination

Free or total (F/T)

RAW/Wa

F
T
F
T
F
T
F
T

RAW/AcWa
TRE/Wa
TRE/AcWa

Actual release from cement

CEM/5 min
F
T
CEM/60 min
F
T
CEM/1 day
F
T
CEM/7 day
F
T
Adjusted for weight in cement (100/73.7)
CEM/5 min
F
T
CEM/60 min
F
T
CEM/1 day
F
T
CEM/7 day
F
T
Pooled mean
F
T

14

21

0.09
0.13
0.36
1.24
0.18
0.05
0.04
0.92

(0.07)
(0.24)
(0.04)
(0.24)
(0.09)
(0.05)
(0.04)
(0.15)

0.04
0.23
0.72
1.49
0.03
0.17
0.49
1.09

(0.15)
(0.08)
(0.26)
(0.46)
(0.02)
(0.02)
(0.22)
(0.81)

0.09
0.16
0.32
1.06
0.01
0.27
0.01
0.75

(0.02)
(0.04)
(0.02)
(0.07)
(0.01)
(0.12)
(0.05)
(0.27)

0.00
0.47
0.10
2.09
0.00
0.45
0.06
1.37

(0.04)
(0.07)
(0.22)
(3.01)
(0.03)
(0.46)
(0.06)
(0.45)

0.38
0.96
0.29
1.07
0.30
0.56
0.19
0.45

(0.05)
(0.11)
(0.12)
(0.15)
(0.30)
(0.07)
(0.02)
(0.15)

0.48
1.19
0.66
1.07
0.17
0.61
0.04
0.72

(0.06)
(0.07)
(0.12)
(0.63)
(0.04)
(0.18)
(0.04)
(0.48)

0.08
1.28
0.10
1.70
0.03
1.09
0.02
0.40

(0.02)
(0.24)
(0.04)
(0.21)
(0.03)
(0.84)
(0.02)
(0.41)

0.07
1.58
0.09
1.86
0.03
0.37
0.01
0.96

(0.02)
(0.22)
(0.05)
(0.36)
(0.03)
(0.37)
(0.01)
(0.14)

0.51
1.28
0.39
1.43
0.40
0.76
0.25
0.60
0.39
1.1

of 40 g/l and it may be that G338 contains uoride in a

more water-soluble form. The nding that G338 lost less
uoride during the acid treatment process than did
LG26 may possibly mean that more has been left for
subsequent dissolution, or possibly that the nature of
the uoride has been altered by the acid treatment
process.
Only a proportion of F appeared as FF and this
proportion remained essentially constant suggesting that
a complexation process is present. In the RAW/water
combination, species which could affect the free/total
ratio can only have come from the glass itself. With TF
increasing with time it seems unlikely that uoride was
precipitated as calcium uoride (CaF) from the calcium
ion which both glasses contained, G338 with 10.7%
CaO and LG26 with 29% CaO. Both glasses contained
phosphate ions. Analysis showed other species such
as MgO were present at very low levels of o0.05%.
G338 contained 31% Al2O3 and 28.1% SiO2 and
LG26 24% Al2O3 and 36% SiO2. Although there may
be interactions in the silicate matrix since G338
has the higher proportion of Al/Si and also a higher
ratio of FF/TF aluminium would be the suspected
complexant.

0.64
1.59
0.92
1.43
0.23
0.81
0.05
0.96
0.46
1.20

0.10
1.71
0.13
2.27
0.04
1.45
0.03
0.53
0.08
1.49

0.09
2.11
0.12
2.48
0.04
0.49
0.01
1.28
0.06
1.59

4.2. The susceptibility of raw glass to acidic water

(AcW)
Contact with acidic water (0.014 m) had no signicant
effect on FF from either glass but did increase TF
(Fig. 2). This indicates that uoride is capable of
dissolving to a greater extent in this particular mildly
acidic solution. If so, uoride might therefore be more
soluble in the oral environment than in laboratory
studies in water. LG26 was more affected than G338
glass with an increase of about 450% in TF when
changing from W to AcW, compared to an increase of
200% for G338. Why LG26 should apparently be more
affected by the acidic solution is not clear but this
apparently large increase may be a reection of the low
levels of uoride released from LG26 when in contact
with water. Whereas RAW G338 is increased from 5.6
to 10.7 mg/g by changing from water to AcW, RAW
LG26 is increased from only 0.5 mg/g to 2.1 mg/g.
The ratio of TF measured after the RAW glass had
been in contact with AcW for 21days compared to the
percentage originally present in the glasses was (1.1%/
18.6%)=5.9 for G338 and (0.2%/5.2%)=3.8 for LG26.
A study by Hatton & Brooks [14] showed that to

2198

J.A. Williams et al. / Biomaterials 23 (2002) 21912200

Fig. 5. Free and total uoride released after 21 days for glasses and
cement (pooled means).

produce elements other than uoride in the matrix

between 1.4% and 9.3% of the glass was consumed.
These values are not dissimilar to those found for TF in
this study.
The effects of increasing pH which might increase
solubility is discussed further in the next section. The
presence of polymeric acids did not seem to affect the
free/total F ratio and it is therefore assumed that neither
is involved in the complexation process.
4.3. The effect of acetic acid washing
This has been shown to increase the working time of
cements compared to those made from untreated glass.
How it does this is not clear since Wasson & Nicholson
[9] have shown that while the concentration of ions in
the acid solution used for washing the glass increased as
the acetic acid concentration increased from 2.5% to
10%, the working time did not continuously increase
but passed through a maximum. However, the study
indicated that ions from glasses are more soluble in
increasingly acidic solutions. The present study found
that treating both glasses with 5% (0.83 M) acetic acid
for 60 min removed uoride from the glass and added it
to the acid washing solution. Although LG26 glass
contained only a third of the uoride of G338, slightly

more uoride ion was removed, producing acidic ltrate

solutions containing approximately 1.4 mm (LG26) and
1.0 mm (G338) TF. With similar amounts of uoride lost
from the glass regardless of the amount of uoride
initially contained therein these gures are either a
measure of the solubility of uoride in the acid aqueous
medium or the amount of soluble glass present on the
surface of the glass particles. It is a similar effect to that
mentioned in a previous section and it appears that the
LG26 glass is more vulnerable to acid attack than is
G338. Using a similar acid concentration to that of the
present study, Wasson & Nicholson [9] found G338 lost
6.4 mm Al, 2.7 mm Ca, 4.2 mm Si, 4.7 mm Na and 0.3 mm
P, amounts similar in quantity to the uoride loss found
above. In that study soluble ions remaining in the glass
before and after the process were not measured.
After acid treatment less TF was produced from both
G338 and LG26 glass, regardless of whether they were
subsequently in contact with water or AcW. The least
affected combination (LG26/water) was also that which
released the least (0.5 mg/g TF). The ratio of TF
produced from TRE/RAW glasses was 75% (G338/
water), 95% (LG26/water), 72% (G338/AcW) and 65%
(LG26/AcW).
FF remained essentially unchanged regardless of acid
treatment.
The acetic acid treatment reduced the amount of
soluble uoride. However, the reduction was small
compared to that already available for dissolution. At
worst some 65% of uoride originally present was left in
or on the glass prior to cement formation.
4.4. The susceptibility of the TRE glass to acidic water
(AcW)
As with RAW glasses, contact with this liquid
increased TF but had less effect on FF when compared
to water. G338 glass produced more TF than LG26.
Therefore, although acid washing had reduced soluble
uoride from both glasses the effect of contact with
AcW was generally the same as with RAW glass. Some
compounds, such as calcium uoride, are more soluble
in dilute acids compared to water [15] and this may
account for the increased uoride in AcW.
4.5. Comparison of glasses and cements
Allowing for the fact that the cement is 75% glass
gives an adjusted value of TF from G338 cement of 8.0
9.3 mg/g after 21 days. This is higher, but not
signicantly so, than 7.8 mg/g TF from the TRE/AcWa
combination and lower than the 10.7 mg/g for RAW/
AcW. Both these latter values were signicantly higher
than the 4.2 mg/g found for TRE/water. LG26 cement
also adjusted for weight, produced a maximum TF of
2.5 mg/g glass compared with 1.4 mg/g from the TRE/

J.A. Williams et al. / Biomaterials 23 (2002) 21912200

AcWa combination.This also seems to show that LG26

is more vunerable to acid attack. This is more than the
1 mg/g TF released from discs of LG26 cement
formulated as a conventional gic [13] but in the study
by Hadley et al. the uoride release was measured for 97
days and may not have been complete.
The percentage increase in TF caused by the process
of cement formation is similar (2.42.8%) for both
glasses in spite of the differing levels of uoride
originally present in the bulk glass (Table 4). This
increase from both G338 and LG26 could be produced
by the dissolution of only 0.1 mm glass from particles of
diameter 10 mm (see Table 5).
Once adjusted for weight G338 cement generally
produced more TF and less FF than glass. This altered
the FF/TF ratio and may indicate that the cement
forming process has increased the degree of uoride
complexation. Although this did not occur with either
TRE glass with AcW it must be remembered that during
the cement forming process the glass is in contact, for
however brief a time, with 5 m acid, a much higher
concentration than that in AcW. A 5 m solution is
similar to a 35% polyacrylic acid solution often used for
gics. The glass may be attacked more severely or
preferentially compared to when the 0.014 m AcW is
used. More ions which can complex F may be produced.
Neither cement maturity prior to contact with water nor

Table 4
Calculation of total uoride produced by cement formation
mg TF present/g
G338
Present in RAW glass initially
186
Released from TRE glass into water (21day) 4.2
Released from cement into water (21day)
8.6
Inc=(Cement-TRE)a
4.4
% increase (Inc/RAW)
2.4
a

LG26
52
0.4
1.6
1.2
2.8

The extra amount of F produced from cement formation.

Table 5
Calculation of the amount of glass which needs to be dissolved to
produce the increase due to cement formation
Glass

G338

LG26

% uoride present in RAW glass

TF g/g initially present in RAW glass
TF g/g after acid treatment
For R0=5 mm, R=

18.6
0.186
0.175
4.90 mm

5.2
0.052
0.050
4.94 mm

For a glass containing N particles/g of radius R0 from which a uniform

layer is removed leaving the particles of radius R: Assume both glass
have the same particle size and density d, therefore initial weight W 0 =
V0d and nal weight W Vd:
V0 N4=3p(R0)3 and V N4=3pR3 and W0 =W V0 =V
R0 3 =R3
For R0 = 5 mm, W0/W = 125/(R)3.

2199

the presence of polymeric/tartaric acid solution affected

the extent of complexation.
The effect of cement formation was to increase
slightly the amount of TF produced compared to that
from TRE/water combination. Thus, it may be concluded that some acid attack has occurred during
cement formation even though this was only a minor
source of F.
4.6. The effect of time
The nding that cement maturity had little effect on
the amount of uoride produced indicates a lack of
continuing reaction between the acidic liquid and the
glass. This may also occur when a silicate cement is
made since this also invokes the attack by a nonpolymeric acidic molecule on the glass. It may not be the
case for polymeric acid-based gics where properties such
as compressive strength may change with time [16,17].
Since only four time intervals were taken denite
conclusions about the effects of time on uoride release
can be only tentative. Considering TF, the type
normally measured in uoride release studies, it appears
that where glass powder or cement disintegrates in liquid
no strong correlation exists between TF and (time)1/2.
The best t plots were given by Ln/Ln relationships with
both RAW/water and TRE/water having slopes of the
order of 0.30.4. In AcW the slopes were reduced to 0.14
and this was the value found for cement. The cement
plots appear to have a higher degree of correlation with
Ln time than do those where glass was used but further
work would be needed to ascertain this. Craneld &
Kuhn [18] also reported results where the relationship
was given by a loglog plot and concluded that the
(time)1/2 relationship was a special case.
No initial burst of uoride was seen with in the
present study, unlike that generally seen with a
polymeric acid-based gic. However, silicate cements
have also been shown [1] not to have an initial burst of
uoride so that different types of uoride release are
exhibited by different materials and this may relate to
the type of reaction used to form the cement [1,2].
Alternatively, since neither the glass nor the cement
remained as monolithic specimens it may be assumed
that the dependence on (time)1/2 seen previously for
cements is due to processes such as diffusion or
permeation rather than uoride release itself. Both
glasses when formulated as conventional gics have been
shown to release uoride at a rate dependent on (time)1/2
[2,13] Since no measurements were made at time
intervals of o1day it is not known whether the uoride
is released immediately on contact with liquid or
uniformly over 24 h.
Improvements in the method, such as continuous
stirring of the sample solutions, could be made for
future work, which would assist in accuracy particularly

2200

J.A. Williams et al. / Biomaterials 23 (2002) 21912200

where only low levels of uoride are to be measured.

However it has been seen that continuously stirred
solutions are slow to settle once stirring ceases and this
would prevent sampling unless the suspensions of glass
in water were centrifuged to prevent particulates
remaining in the liquid samples. Complete re-dispersion
after sampling would then be required. This procedure
would limit the number of sample solutions under
evaluation.

4. The cement forming process only slightly increased

soluble uoride; for both G338 and LG26 glass this
was about a 3% increase.
5. Fluoride was complexed to a greater extent when
released from cement than from glass.
6. During the duration of the experiment the relationship TF proportional to (time)1/2 did not describe the
process of uoride solubility for powder dispersed in
liquid either before or after cement formation.

4.7. Complexation of fluoride ion

While TF increased steadily with time FF remained
generally constant for both RAW and TRE glasses. One
explanation is that soluble F is accompanied by other
water soluble ions which complex F either as it is
released in the ionic state, or prior to release such that
the F is released as a complexed ion. The process of
cement formation can also produce ions capable of
complexing F so that these ions are likely to be those
soluble in acids such as aluminum, silica and phosphate
[7]. Some F may be removed from the system by
precipitation as calcium uoride (CaF) and this compound although sparingly soluble in water is more
soluble in acids [15]. If calcium were released more
slowly than uoride ion, precipitation could explain a
decrease of TF with time. This would also occur for
compounds such as AlF3 and AlPO4 which are barely
soluble in either water or acids. The high initial FF/TF
found for the G338 TRE/AcW combination might be
explained if complexing ions are slower than uoride
ions to move out of the TRE glass. There is a somewhat
similar effect seen with RAW G338/AcW.

5. Conclusions
It was found that
1. Fluoride containing glasses had considerable quantities of water soluble uoride present before any
treatment to alter surface properties is given and
before the cement forming process took place. As
much uoride was released from the glass alone as
from cements made from such glasses. Thus, acid
attack on glass during the cement forming process did
not account for all the uoride available
2. Acid treatment to modify properties such as the
working time of a cement removed ion species from
the surface and reduced soluble uoride for both
glasses examined.
3. The presence of mildly acidic water increased soluble
uoride.

References
[1] Swartz ML, Phillips RW, Clark HE. Long term F release from
glass ionomer cements. J Dent Res 1984;63:15860.
[2] Tay WM, Braden M. Fluoride ion diffusion from polyalkenoate
(glass ionomer) cements. Biomaterials 1988;9:4546.
[3] Forsten L. Short and long term uoride release from glass
ionomers and other uoride-containing materials in vitro. Scand J
Dent Res 1990;98:17985.
[4] Williams JA, Billington RW, Pearson GJ. The inuence of sample
dimensions on uoride ion release from a glass ionomer
restorative cement. Biomaterials 1999;20:132737.
[5] Causton BE. The physico-mechanical consequences of exposing
glass-ionomer cements to water during setting. Biomaterials
1981;2:1125.
[6] De Moor RJG, Verbeeck RMH, De Maeyer EAP. Fluoride
release proles of restorative glass ionomer formulations. Dent
Mater 1996;12:8895.
[7] Wilson AD, Kent BE. Dental silicate cements. IX. Decomposition
of the powder. J Dent Res 1970;49:713.
[8] Crisp S, Wilson AD. Reactions in glass ionomer cements. I.
Decomposition of the powder. J Dent Res 1974;53:140813.
[9] Wasson EA, Nicholson JW. A study of the relationship between
setting chemistry and properties of modied glasspoly(alkenoate) cements. Br Polym J 1990;23:17983.
[10] Craneld MJ. An in vitro analysis of uoride release from glass
ionomer cements. M.Sc. thesis, University of London, 1980.
[11] Schmidt W, Purman R, Jochum P, Grosser O. Eur Patent Appl
23,013, 1981.
[12] Akinmade AO, Nicholson JW. Poissons ratio of glass-polyalkenoate (glass-ionomer) cements determined by an ultrasonic
pulse method. J Mater Sci: Mater Med 1995;6:4835.
[13] Hadley PC, Billington RW, Pearson GJ. Effect of monovalent
ions in glass ionomer on their uptake and re-release. Biomaterials
1999;20:8917.
[14] Hatton PV, Brooks IM. Characterisation of the ultrastructure of
glass-ionomer (poly alkenoate) cement. Br Dent J 1992;73:75277.
[15] Partington JR. General, inorganic chemistry. UK: MacMillan,
1947. p. 422.
[16] Williams JA, Billington RW. Changes in compressive strength of
glass ionomer restorative materials with respect to time periods of
24 h4 months. J Oral Rehabilition 1991;18:1638.
[17] Williams JA, Billington RW, Pearson GJ. The effect of
maturation on in vitro erosion of glass ionomer and other dental
cements. Br Dent J 1992;173:3402.
[18] Craneld M, Kuhn AT, Winter GB. Factors relating to the rate of
uoride-ion release from glass-ionomer cement. J Dent
1982;4:33341.