Water diffusion in hydroxyethyl methacrylate(HEMA)-based hydrogels formed by g-radiolysisDavid JT Hill,* McKenzie CH Lim and Andrew K WhittakerPolymer Materials and Radiation Group, University of Queensland, Brisbane, QLD 4072, Australia

Abstract: Polymer hydrogels based upon methacrylates are used extensively in the pharmaceutical

industry, particularly as controlled release drug delivery systems. These materials are generally

prepared by chemically initiated polymerization, but this can lead to the presence of unwanted

initiator fragments in the polymer matrix. In the present work, initiation of polymerization by g-

irradiation of hydroxyethyl methacrylate, with and without added crosslinkers, has been investigated,

and the diffusion coef®cients for water in the resulting polymers have been measured through mass

uptake by the polymers. The diffusion of water in poly(hydroxyethyl methacrylate) at 310K was found

to be Fickian, with a diffusion coef®cient of 1.96�0.1�1011m2sÿ1 and an equilibrium water content of

58%. NMR imaging analyses con®rmed the adherance to a Fickian model of the diffusion of water into

polymer cylinders. The incorporation of small amounts (0.2±0.5wt%) of added ethyleneglycol-

dimethacrylate-based crosslinkers was found to have only a small effect on the diffusion coef®cient and

the equilibrium water content for the copolymers.

# 1999 Society of Chemical Industry

Keywords: HEMA hydrogels; synthesis by g-radiolysis; diffusion coef®cients; equilibrium mass uptake; NMRimaging

INTRODUCTIONPolymer hydrogels have found extensive applications

in the pharmaceutical industry and in medicine

because of their ability to absorb large amounts of

water. The family of polymer hydrogels which has

found widest application in medicine is that based on

hydroxyethyl methacrylate, because of the well known

biocompatibility of methacrylate based polymers.

The water absorption properties of poly(hydroxy-

ethyl methacrylate) (HEMA) and its copolymers have

been extensively investigated and reviewed.1 Poly

(HEMA) has been found2±5 to follow a Fickian model

for water diffusion, with a diffusion coef®cient at

310K in the range 1.55�1011m2sÿ1 to

2.00�10ÿ11m2sÿ1, in the absence of any added

crosslinker. In the presence of added crosslinker, such

as an ethyleneglycol dimethacrylate (EGDMA), the

diffusion coef®cient has been reported to decrease,

with the numerical value of the diffusion coef®cient

being dependent on the extent of crosslinking of the

polymer.6,7 It has been reported6 that for copolymers

of HEMA and monoethyleneglycol dimethacrylate

(mEGDMA) the diffusion coef®cient falls rapidly over

the range 0±1.0% mEGDMA, but beyond this

composition the decrease is much less rapid. Chen6

has suggested that below about 0.4% mEGDMA a

pore mechanism for diffusion is dominant, but above

this composition the diffusion mechanism changes.

However, other workers8,9 have found that water

uptake by poly(HEMA) is relatively insensitive to low

degrees of crosslinking by mEGDMA, which suggests

that the effective level of crosslinking in the gels is not

essentially changed at low crosslinker contents.

In most previous studies of the water absorption

behaviour of HEMA hydrogels, the polymer samples

have been prepared by chemically initiated polymer-

ization. In such polymerizations, the fragments of the

initiator remain in the polymer, and will diffuse out as

water is absorbed by the polymer matrix, converting it

to a rubbery state. Such initiator fragments may be of

concern if the materials are to be used in medical

applications. Therefore, in the present study the

polymers have been prepared by radiation initiated

polymerization, which avoids the presence of remnants

from the initiator, and the results are compared with

those observed for chemical initiation. The behaviour

of poly(HEMA) and a series of crosslinked HEMA

copolymers is reported.

Polymer International Polym Int 48:1046±1052 (1999)

* Correspondence to: David JT Hill, Polymer Materials and Radiation Group, University of Queensland, Brisbane, QLD 4072, AustraliaContract/grant sponsor: Australian Research CouncilContract/grant sponsor: Australian Institute for Nuclear Science and Engineering(Received 15 December 1998; accepted 3 February 1999)

# 1999 Society of Chemical Industry. Polym Int 0959±8103/99/$17.50 1046

EXPERIMENTALMonomer purification and polymerizationThe HEMA monomer was obtained from Rocryl,

puri®ed by the method outlined by Ghi et al,3 and the

middle fraction of the monomer which distilled at

342K at 6.6�102Pa was collected for use. The

ethyleneglycol dimethacrylate monomers were ob-

tained from Aldrich and the stabilizer was removed

from these monomers by passage through an anhy-

drous alumina column.

Monomer mixtures of the desired composition, in

the range 0.2±0.5wt% of the EGDMAs, were pre-

pared and put in cylindrical glass or plastic moulds

with diameters of 0.65cm and 0.75cm, respectively,

and placed in capped glass tubes. Oxygen was

removed from the mixtures by bubbling with dry

nitrogen gas and then the containers were sealed under

nitrogen. The samples were transferred to a 60Co

AECL Gammacell 200 facility for polymerizations at

ambient 300K (dose rate 1.1kGy hÿ1) or to a 60Co

Nordian 220 facility ®tted with a controlled tempera-

ture heating block for polymerizations at 373K (dose

rate 9.8kGy hÿ1). The doses used to effect polymer-

ization of the monomers were in the range 17±19kGy.

Following polymerization, the polymer samples

were allowed to stand overnight to allow any free

radicals formed during polymerization to decay, and

the cylinders of polymer were then removed from their

moulds and transferred to a vacuum oven at 320K for

7 days to remove any residual unreacted monomer or

other volatiles. After the cylinders were removed from

the oven they were checked over their length for the

presence of any residual monomer by FT-NIR

spectroscopy. No evidence was found for the presence

of any unreacted double bonds, which have an

absorbance at 6070cmÿ1.

Water sorption studiesThe water sorption studies were conducted by placing

the cylinders in test tubes of distilled water which was

kept at a controlled temperature of 310K. The

cylinders were removed at time intervals, the excess

water was removed by wiping with Kimberley-Clark

tissue wipes and their mass measured on a Mettler

AC100 analytical balance. The mass uptake measure-

ments were continued until the mass of the cylinders

remained effectively constant, which required approxi-

mately 3 weeks.

In addition to the mass uptake measurements, the

dimensions of the cylinders were monitored by

measurement with Mitutoyo digital vernier calipers.

NMR imaging studyTo con®rm that the nature of the diffusion front was

consistent with that expected for Fickian diffusion,

images of some samples were taken using a Bruker

AMX 300 NMR spectrometer. The images across a

central region of the cylinders consisted of 128�128

pixels obtained using a 3D spin echo method with the

following parameters: read gradient strength 0.08T

mÿ1; 90° pulse of duration 26ms; echo time 3.3ms;

recovery time 500ms.

Polymer glass transition temperatureThe glass transition temperatures of the dry polymers

were determined using a Perkin Elmer DSC-7 which

was calibrated using indium and zinc melting points

and the heat capacity of indium. The measurements

were made on 2±5mg of vacuum-dried, powdered

samples under nitrogen ¯owing at 15ml minÿ1, using

a scanning rate of 20K minÿ1 over the temperature

range 313±473K.

RESULTS AND DISCUSSIONMass uptake measurementsThe results of the experiments to determine the mass

uptake of water for the poly(HEMA) samples which

were prepared at 300 and 373K are shown in Fig 1.

The ®gure shows that there is little difference in the

sorption behaviour of the polymers prepared at the two

temperatures, even though the higher temperature is

close to the glass transition temperature for the

polymer, where polymerization would be expected to

proceed readily to complete conversion. This suggests

that the properties of the polymers formed at the two

temperatures are very similar.

Uncrosslinked poly(HEMA) is water soluble, so it

does not behave as a `true' hydrogel. Therefore, during

polymerization by g-radiolysis, the polymerizing sam-

ple must form a network. Hill et al10 have studied the

g-radiolysis of a soluble poly(HEMA) and reported

that it forms a gel, which has been attributed to

crosslinking through radical formation on the methy-

lene units of the HEMA side-chains. This would

Figure 1. Relative mass uptake versus time1/2 for poly(HEMA) synthesizedat 300K (*) and 373K (^). Fitted curves based on eqn (1): at 300K ——;at 373K –..–.

Polym Int 48:1046±1052 (1999) 1047

Water diffusion in HEMA-based hydrogels

explain the observed gel formation during polymeriza-

tion initiated by g-radiation.

The data in Fig 1 show a close to linear dependence

on t0.5 in the early stages of the sorption process, so

they have been ®tted to a Fickian model for diffusion.

Ghi et al3 have shown that for an in®nite cylinder of

radius a, the mathematical relationship for Fickian

behaviour is

Mt

M1� 1ÿ

Xn�1!1

4

�n2

expÿD�n

2t

a2

� ��1�

where D is the diffusion coef®cient, t is the time for

which penetrant diffusion has occurred and bn are the

roots of the zero order Bessel function J(bn)=0.

The data have been ®tted to this model using the

linear regression procedure described by the previous

workers.3 The values obtained for the diffusion

coef®cients for the two polymers shown in Fig 1 if

the full range of data are analysed were

1.80�10ÿ11m2sÿ1 and 1.56�10ÿ11m2sÿ1 at 310K

for the polymers prepared at 300K and 373K,

respectively. These values are in very good agreement

with values reported previously in the literature for

water diffusion into poly(HEMA) at 310K, which fall

in the range 1.55±2.00 10ÿ11m2sÿ1, depending on the

polymerization protocol, the sample size and the mass

uptake range used in the data analysis.2,4,5

The uncertainty in the value of the diffusion

coef®cient which is obtained from a ®t to the

experimental data can be assessed from the hypersur-

face for the `goodness-of-®t' parameter, here the sum

of the squared deviations of the data points from the

®tted curve. A typical curve for the data for the

poly(HEMA) cylinder prepared by radiolysis at 300K

is shown in Fig 2. The experimental error in the

diffusion coef®cients based on this method of estima-

tion is approximately�0.1�10ÿ11m2sÿ1 for a level of

con®dence greater than 95%.

The data in Fig 1 for the polymer prepared at the

higher temperature do appear to show a small over-

shoot, and a greater deviation from the ®tted curve

above a relative mass uptake of approximately 0.6,

which corresponds to the point at which the glassy core

disappears during the sorption process. These features

have been described previously,2,4,5 and are believed

to be associated with the relaxation of the polymer

chains.

When water is absorbed by the polymer cylinders,

their dimensions change, as demonstrated in Fig 3 for

a cylinder prepared at 300K. In the early stages of the

water sorption, the length of the cylinder changes only

slightly, with the most noticeable change being in the

cylinder diameter. However, after the relative mass of

water taken up by the cylinder reaches about 0.6, the

length of the cylinder increases signi®cantly, because

at this point the glassy central reinforcing core in the

Figure 2. Fitted hypersurface for poly(HEMA) at 300K.

Figure 3. Length and volume changes for water sorption into poly(HEMA)as a function of sorption time.

1048 Polym Int 48:1046±1052 (1999)

DJT Hill, MCH Lim, AK Whittaker

cylinder has disappeared. This allows the polymer

chains to relax and the length of the cylinder to

suddenly increase. Beyond the mass uptake fraction of

0.6 the nature of the diffusion of the water changes in

the now completely rubbery polymer, and there is an

apparent increase in the diffusion coef®cient, which is

responsible for the observed discontinuities in the

mass uptake curves.

Because the dimensional data indicate a change at a

relative mass uptake at approximately 0.6, the sorption

data were reanalysed to calculate the best value for the

diffusion coef®cient over the range 0±0.6. This

reanalysis of the data yielded values of the diffusion

coef®cient of 1.96�10ÿ11m2sÿ1 for the cylinders

prepared at 300K and 1.44�10ÿ11m2sÿ1 for the

cylinders prepared at 373K. These values are also in

good agreement with those previously reported by

Gehrke et al2 and Hill and co-workers4,5 at the same

temperature.

The relative mass uptake data for a series of

copolymers of HEMA with ethyleneglycol dimetha-

crylate, poly(HEMA±mEGDMA), diethyleneglycol

dimethacrylate, poly(HEMA±dEGDMA), triethyle-

neglycol dimethacrylate, poly(HEMA±triEGDMA)

and tertraethyleneglycol dimethacrylate, poly(HE-

MA±tetraEGDMA) prepared at 300K are shown in

Fig 4. The mass uptake data for the various crosslinked

systems in the early stages of water sorption show very

little dependence on the crosslinker concentration.

However, there are some small differences between the

Figure 4. Mass uptake curves for HEMA copolymers with (a) mEGDMA, (b) dEGDMA, (c) triEGDMA, (d) tetraEGDMA: * 0.2wt%; &, 0.3wt%; ~, 0.4wt%;!, 0.5wt%;.

Polym Int 48:1046±1052 (1999) 1049

Water diffusion in HEMA-based hydrogels

curves in the high mass uptake regions. The divinyl

comonomers introduce new types of crosslinks, and it

is clear from the curves in Fig 4 that these enhance the

tendency for the mass uptake to overshoot the

equilibrium value, as has been reported by some

previous workers5,7 for HEMA copolymers. It is

believed5 that this overshoot arises because the

relaxations of the polymer chains in the swollen gels

are slow processes, and the rates of relaxation are

in¯uenced by the extent of crosslinking.

The data in Fig 4 have been ®tted to eqn (1) using

the same curve ®tting procedures as before. The

diffusion coef®cients obtained from these curve ®ts

have been summarized in Table 1. While there is a

systematic small decrease in the values of the diffusion

coef®cients over that for poly(HEMA), with the

possible exception of tetraEGDMA, the changes are

within experimental error. In the case of tetraEGDMA

the diffusion coef®cient appears to increase slightly at

the higher crosslinker concentrations, which may be

due to the larger spacer group in this comonomer. As

pointed out previously, other workers8,9 have also

found little effect of mEGDMA on the sorption

properties of poly(HEMA), at low added crosslinker

contents.

Equilibrium water content measurementsThe equilibrium water content or mass uptake at

equilibrium (EWC) was measured for all the polymer

cylinders as the percentage increase in mass at

equilibrium relative to the mass of the polymer

cylinder. The values have been summarized in Table

1. The value of EWC of 58% for poly(HEMA) is very

similar to values found previously of 67%2 and 58.8%4

for poly(HEMA) which had been prepared by

chemical initiation. The incorporation of the cross-

linkers lowered the values of S to about 53±54%, with

the values almost independent of the crosslinker

content over the narrow composition range studied.

These observations are consistent with the EWC value

of 54.6% reported by Franson and Peppas11 for a

copolymer with 0.8wt% mEGDMA. The lower values

for the crosslinked polymers are consistent with the

formation of somewhat more restricted networks at

high penetrant contents.

Imaging StudiesFigure 5 shows a typical NMR image of the central

section of a polymer cylinder. The cylinder shown in

the ®gure contained 0.3wt% mEGDMA and the water

diffusion was conducted at 310K. In Fig 6 cross-

sections are shown of images obtained for this cylinder

after diffusion times of 58 and 1179min at 310K. The

image in Fig 5 shows that the highest concentration of

water is at the outer edge of the cylinder which is in

direct contact with the water, and the concentration of

the water decreases towards the glassy core of the

cylinder, which is clearly visible in the centre of the

image. These features are also evident in the cross-

sections given in Fig 6 which show that the water

concentration pro®le across the diameter of the

cylinder is that expected for Fickian diffusion, which

is shown by the dashed curves.

A close examination of the pro®le of the image in Fig

5 shows that there is a concentration ridge near the

edge of the glassy core. This ridge is also visible in the

form of small peaks in the cross-section of the image

shown in Fig 6 for 1179min water sorption. A similar

ridge has been observed previously by Ghi et al3 for

polymer cylinders formed by chemical initiation, and

has been attributed to crack formation resulting from

induced stress arising between the glassy core and the

adjactent swollen rubbery region. It has been sug-

gested3 that water enters these cracks and that this

water is characterized by a different relaxation

behaviour from that of the neighbouring water, which

is interacting strongly with the polymer.

The increase in the diameter of the cylinder as it

absorbs water can also be seen by comparing the the

two cross-sections shown in Fig 6. In the early stages of

the water uptake the major change in the volume of the

cylinder occurs through an increase in the diameter,

but when the glassy core disappears, and the whole

cylinder becomes rubbery, the cylinder length in-

creases rapidly.

Glass transition temperaturesThe glass transition temperatures of the polymer

cylinders were measured. All the polymers showed

two glass transition temperatures, as demonstrated in

Fig 7. The major transition occurred at about 380K

(approx. 2.2�10ÿ1Jgÿ1 Kÿ1) with a second, smaller

transition at about 417K (approx. 3.9�10ÿ2Jgÿ1

Table 1. Equilibrium water contents (EWC) anddiffusion coefficients (D) for crosslinked HEMAcopolymers of various compositions (wt%)

D�1011b (m2sÿ1)

Polymer EWCa (%) 0.0c 0.2c 0.3c 0.4c 0.5c

HEMA 58 1.8 ± ± ± ±

mEGDMA 53 ± 1.8 1.7 1.7 1.6

dEGDMA 54 ± ± 1.6 1.7 1.7

triEGDMA 53 ± 1.8 1.5 1.7 1.7

tetraEGDMA 54 ± 1.7 2.0 2.1 ±

a EWC=(M? ÿM0)�100/M0.b Experimental error ��0.1�10ÿ11m2sÿ1.c Crosslinker concentration (wt%).

1050 Polym Int 48:1046±1052 (1999)

DJT Hill, MCH Lim, AK Whittaker

Kÿ1). The transition at 380K is characteristic of

extended sequences of HEMA units, and compares

favourably with previously reported values of 371K12

and 382K13. The origin of the transition at 417K is

unclear, but because it appears in all of the samples it is

possibly associated with the motions of HEMA units

adjacent to crosslinking sites which are introduced by

the radiolysis process.

The principal glass transition temperature is not

signi®cantly dependent on the presence of added

crosslinker at the levels used in the present study, but it

would appear to decrease slightly for the two cross-

linkers with the longer spacer length between the

double bonds.

CONCLUSIONSThe properties of the HEMA-based polymers formed

by g-radiolysis have been found to be similar to those

prepared by chemical initiation. The polymer cylin-

Figure 5. Typical image for water sorptioninto a polymer cylinder containing 0.3wt%mEGDMA.

Figure 6. Central cross sections for images of a polymer cylindercontaining 0.3wt% mEGDMA after water sorption at 310K for 58min and1179min, together with curves for a Fickian model.

Figure 7. DSC trace of poly(HEMA) prepared by g-radiolysis of themonomer at 300K.

Polym Int 48:1046±1052 (1999) 1051

Water diffusion in HEMA-based hydrogels

ders prepared by radiolysis at ambient temperature

showed similar characteristics to polymer cylinders

prepared by radiolysis at 373K. For small relative

mass uptakes (0±0.6), the value of the diffusion

coef®cient for water in poly(HEMA) at 310K for

cylinders prepared at 300K was 1.96�10ÿ11m2sÿ1

and that for cylinders prepared at 373K was

1.56�10ÿ11m2sÿ1.

Inclusion of crosslinking monomers had only a small

effect on the diffusion coef®cients, but the equilibrium

mass uptake at 310K decreased from 58% for

poly(HEMA) to about 53±54% for the copolymers

with crosslinker compositions in the range 0.3±

0.5wt%.

ACKNOWLEDGEMENTSThe authors wish to acknowledge the ®nancial support

of the Australian Research Council and the Australian

Institute for Nuclear Science and Engineering.

REFERENCES1 Peppas NA and Moynihan HJ, `Hydrogels in Medicine and

Pharmacy', Vol 2, Ed by Peppas, NA, CRC Press, Boca

Raton. p 49 (1987).

2 Gehrke SH, Biren D and Hopkins JJ, J Biomater Sci Polym Edn

6:375 (1994).

3 Ghi PY, Hill DJT, Maillet D and Whittaker AK, Polym Commun

38:3985 (1997).

4 Hill DJT, Moss NG, Pomery PJ and Whitaker AK, Polymer, in

press.

5 Ghi PY, Hill DJT and Whittaker AK, J Polym Sci, Part B,

submitted.

6 Chen RYS, Polym Prepr 15:387 (1974).

7 Allen PEM, Bennett DJ and Williams DRG, Eur Polym J 28:347

(1992).

8 Refojo MF and Yasuda H, J Appl Polym Sci 9:2425 (1965).

9 Janacek J and Hasa J, Collect Czech Chem Commun 31:2186

(1966).

10 Hill DJT, O'Donnell JH, Pomery PJ and Saadat G, Radiat Phys

Chem 48:605 (1996).

11 Franson NM and Peppas NA, J Appl Polym Sci 8:1299 (1983).

12 Fambri L, Gavazza C, Stol M and Migliares C, Polymer 34:528

(1993).

13 Hodge PM, Simon GP, Whittaker MR, Hill DJT and Whittaker

AK, J Polym Sci Part B Polym Phys 36:463 (1998).

1052 Polym Int 48:1046±1052 (1999)

DJT Hill, MCH Lim, AK Whittaker