Effect of storage conditions on calcium sulphate hemihydrate-containing products

dental materials Dental Materials 17 (2001) 134±141 www.elsevier.com/locate/dental

Effect of storage conditions on calcium sulphate hemihydrate-containing products T.K.-C. Chan a, B.W. Darvell b,* a

b

Dental Technology, Prince Philip Dental Hospital, 34 Hospital Road, Hong Kong, People's Republic of China Dental Materials Science, Faculty of Dentistry, The University of Hong Kong, Prince Philip Dental Hospital, 34 Hospital Road, Hong Kong, People's Republic of China Received 18 October 1999; accepted 8 May 2000

Abstract Objective: Dental gypsum products are liable to deterioration on exposure to the air because of reaction between calcium sulphate hemihydrate and water vapour. The purpose of this investigation was to re®ne the knowledge of the conditions and rates at which this occurred for a representative selection of such products with a view to improved storage recommendations. Materials and methods: Absorption of water from controlled atmospheres (25±408C, 75±99% RH) was monitored gravimetrically for a dental arti®cial Stone, Plaster of Paris, an Impression plaster and a gypsum-bonded Investment. Results: An initial steady absorption stage was followed by a transition to a rapid absorption; this led to a completion stage as the presumed hydration reaction approached completion. The logarithm of time of the transition showed approximately linear dependence on the relative humidity (and not on water vapour pressure as might have been expected) and a slight but signi®cant …P , 0:005† negative dependence on temperature in the range 25±358C for Stone, Plaster of Paris and the Impression plaster, but not for the Investment …P . 0:5†: The behaviour for the ®rst three materials at 408C departed signi®cantly from the temperature-dependent trend …P , 0:05† while the Investment showed no such effect. The absorption by the Investment was substantially faster than for the other materials. Signi®cance: There is no evidence to suggest that there is a `safe' humidity for exposure to the atmosphere although for the Stone and Plasters at RH ,70% in excess of 1000 h is predicted to be necessary before rapid reaction commences. For the Investment, the equivalent time at 70% RH is 1 d. q 2001 Academy of Dental Materials. Published by Elsevier Science Ltd. All rights reserved. Keywords: Gypsum; Plaster; Investment; Humidity; Adsorption; Storage

1. Introduction Calcium sulphate hemihydrate is used in large quantities in dentistry in various forms for models, dies, casting investments and impressions, as well as in the building and related industries. In all these contexts, reaction with water is the basis of the setting or hardening of pastes made with powdered hemihydrate or powders containing this. As such, these products are susceptible to premature reaction with atmospheric moisture. The little work that had earlier been reported in this area has already been reviewed [1]. The experimental work [1] then covered in a more systematic manner a range of humidities at 238C and showed a clear relationship between the rate of reaction and humidity. However, there was evidence of either delayed reaction or sensitivity to high(er) humidity triggering events, after which reaction proceeded steadily even at 81% RH, a matter * Corresponding author. Tel.: 1852-2859-0303; fax: 1852-2548-9464. E-mail address: [email protected] (B.W. Darvell).

in need of resolution if a `safe' humidity was to be identi®ed. It was the purpose of the present investigation to extend that work [1] to a wider range of temperatures, partly with a view to resolving the above uncertainty and partly in an attempt to put the process on a ®rmer chemical footing. 2. Materials and methods Four hemihydrate products were studied, taken as representative of the principal product types used in dentistry: Plaster of Paris (Surgical, British Gypsum, Newark, UK), dental arti®cial Stone (Kaf®r D, British Gypsum), Impression plaster (Mirrotrue, WhipMix, Louisville, KY, USA) and gypsum-bonded casting Investment (Beauty Cast, WhipMix). The Impression plaster was supplied in airtight plastic bottles, the ®rst two in multi-layered paper sacks, and the Investment in a multi-layered plastic bag. A 10-kg portion of one batch of each product was

0109-5641/01/$20.00 + 0.00 q 2001 Academy of Dental Materials. Published by Elsevier Science Ltd. All rights reserved. PII: S 0109-564 1(00)00052-X

T.K.-C. Chan, B.W. Darvell / Dental Materials 17 (2001) 134±141

135

Fig. 1. Schematic diagram of the incubator used for the controlled environment.

obtained and stored, if necessary, in double-layered polyethylene bags, having ®rst ascertained that: (a) the sources were previously unused; (b) the products passed through a 1.5 mm mesh sieve with no residue and (c) mixing behaviour was as expected and that setting times were consistent with those stated by the manufacturers. Silica gel packets were used in the storage containers to minimize pre-exposure. A laboratory incubator was modi®ed for use as the controlled environment chamber. The internal dimensions were 66 cm high, 56 cm wide and 56 cm deep. The incubator was a double-door (outer, inner) design with a vent at the top centre. The incubator was originally designed to maintain the humidity at or over 98% by placing water in reservoirs located at the bottom of the incubator chamber. The incubator was modi®ed in several ways (Fig. 1). Six shelves (10 cm deep £ 60 cm long, 10 cm apart) were placed at the rear of the incubator for holding the specimens. An optoelectronic dewpoint meter probe (Protimeter, model DP989M, Protimeter plc, Marlow, Bucks, England) was placed inside the incubator for continuous monitoring of Table 1 Standard salt systems used, saturated solutions in water System A B C D E

Nominal RH at 258C (%) NaCl (NH4)2SO4 KBr NH4H2PO4 H2O

77 81 85 93 100

the temperature and humidity changes by the indicator outside. The data display was updated continuously every 0.5±2 min. To minimize the changes of temperature and humidity during the weighing of samples, it was arranged to do this in the incubator. The vent, 3 cm in diameter, was utilized by placing a single-pan analytical balance (resolution 0.1 mg) above it. The suspension for the weighing pan was extended with a stainless steel wire so that the pan was near the centre of the work space. The vent was sealed by a split rubber bung when the balance was not in use. The glass inner door was replaced by a clear acrylic sheet (5 mm thick). Near the bottom of this, two ports were cut to allow the hands and wrists to enter. These ports included `star cut' double rubber sheets to assist the maintenance of the internal environment and were closed off with close-®tting acrylic discs when not in use. Shallow plastic reservoirs, 23 cm £ 18 cm £ 3 cm deep were placed beside the fan mounted at the bottom of the incubator, beneath the perforated `false ¯oor', for holding saturated salt solutions (see below). The fan enabled circulation and rapid equilibration of the atmosphere over the full volume of the incubator. An electronic top-loading single-pan balance (resolution 1 mg) was used for the initial weighing of the Petri dishes and samples outside the incubator. Polystyrene Petri dishes (90 mm dia.) were used for holding samples of the test material in order to provide a large area of exposure inside the incubator. The sample was spread in a even layer approximately 5 mm thick.

136

T.K.-C. Chan, B.W. Darvell / Dental Materials 17 (2001) 134±141

Table 2 Summary of trials (system letters: see Table 1) Temperature (8C)

Number of trials (RH (%)) System

25 30 35 40

A

B

3 (77) 4 (80) 4 (82) 3 (75)

3 4 4 3

(85) (85) (85) (80)

C

D

E

3 (89) 4 (90) 3 (87) 4 (84)

4 (96) 4 (96) 4 (93) 4 (92)

4 (99) 4 (99) 4 (97) 4 (97)

The incubator temperature was monitored by the dewpoint meter, but a mercury-in-glass thermometer was placed inside the compartment as an independent check. A three-term (`PID') temperature-controller (model 125M, Eurotherm, West Sussex, England) was used to control the temperature aiming to achieve ^0.58C. Humidity was controlled by standard saturated salt solutions or water that provided an equilibrium ambient humidity close to the desired value (Table 1). The actual readings of the hygrometer were noted for all temperatures and solutions. An investment model-drying oven was used for preheating samples of powder to prevent condensation on entering the incubator. This was also modi®ed by replacing its temperature-controller with a PID controller (model 125M, Eurotherm) to obtain better control and accuracy. 2.1. Procedure The desired temperature for the incubator was ®rst set and checked to be stable for at least 4 h. Then, two shallow plastic trays ®lled with saturated salt solution (using deionized water) were placed on either side of the fan at the bottom of the incubator and, after equilibration, the temperature and humidity were checked to be stable for at

least 24 h. Two water re®lling bottles containing deionized water were placed on the false ¯oor of the incubator for later topping up of the salt solution reservoirs to ensure that they did not dry out. The water in the bottles were thereby preheated so that the temperature would not be affected when re®lling. During re®lling, care was taken not to create a large in¯uence on the saturation of the solution and thus on the humidity. Firstly, a large amount of saturated salt solution (about 100 g salt, 200 mL solution, as appropriate) was used at the beginning of the test. Secondly, if replenishment was needed, only a small amount of water (,10 mL) was added at regular intervals. Thirdly, the solution was well-stirred with a glass rod immediately after the addition of water. The preheating oven was set to 58C higher than the required test temperature and checked to be stable for at least 4 h. Each Petri dish with its cover was then weighed, after which about 20 g of the test powder was spread evenly in it and weighed again. All trials were in triplicate or quadruplicate (Table 2). The cover was in place to lessen the effects from the environment during the weighing of the other samples. These three or four dishes were then put into the preheating oven for at least 5 min. The warmed dishes were then put into the incubator through one of the ports after turning off the circulation fan to prevent disturbing the balance and minimizing external air exchange when the ports were in use. After a few minutes to allow cooling, the dishes were then reweighed using the internal weighing pan. The cover was then opened and placed under the lower half of the dish, which was then placed on a shelf. The dishes were thereafter weighed inside the incubator after 30 min, 1 h and at other intervals as considered appropriate and necessary. After each weighing series, the vent and the ports were closed up and the fan turned on. The data were plotted as they were collected so as to monitor progress closely and determine the next appropriate time for measurement. The percentage change in mass of the content of each dish was calculated and plotted using a graphing program (SigmaPlot for Windows, ver. 2.1, Jandel Scienti®c Software, San Rafael, CA, USA) on a personal computer. 2.2. Reversibility checks Similar duplicate specimens of the four products were tested by desiccation over diphosphorus pentoxide in vacuum (,0.2 mbar) under three conditions: (1) as-supplied, i.e. from the stock used for the above trials; after exposure for (2) 2 h and (3) 1 month at 358C and 98% RH. The exposure conditions were chosen to be severe to obtain a prompt result and to bracket the time of the transition to the rapid stage (see below). Weighing was in air at 238C on a conventional single-pan analytical balance to 0.1 mg resolution. 3. Results and analysis

Fig. 2. Typical plot of change of sample mass against time (log scale) showing the initial (I), rapid (R) and completion (C) stages and the construction to ®nd the transition time (TR).

The replicates within materials under given conditions were essentially indistinguishable and the data for these

T.K.-C. Chan, B.W. Darvell / Dental Materials 17 (2001) 134±141

137

Fig. 3. Typical sets of results on log±log plots for 258C, each plot combining the data from the several runs to show the consistency of behaviour. The ®tted lines are rational polynomials for trend indication only.

were pooled. All materials under all conditions gave curves of the kind illustrated in Fig. 2. A typical set of results for 258C are shown in Fig. 3. Broadly, each specimen showed an almost immediate increase in mass (i.e. at the ®rst weighing), which was attributed to simple adsorption of water vapour onto the powder surface. This was not studied further. Thereafter a relatively slow increase in mass was observed, which might continue for many hours; this was termed the initial (I) stage. Sooner or later, this was followed by an in¯ection into the rapid (R) stage of accelerating mass increase. After passing a second in¯ection, the rate of mass increase diminished steadily in what is referred to as the completion (C) stage. This last mentioned stage has no practical interest in the present context, as attention was

focused on the transition to the R stage, the time of which (TR) was taken to be the key feature of concern. However, as the transition itself was a gradual rather than abrupt event, and therefore not precisely determinable, a conventional approach to identifying it was taken, in the manner of the `proof stress'. A line of minimum slope was ®tted by eye to the I stage curve up to the in¯ection in graphs of mass gain vs. log(time) (Fig. 2). A parallel line corresponding to 0.2% further change in mass was then drawn, and the time corresponding to the intersection of this with the data curve (estimated smooth curve) was taken as TR. The means of these data (from the three or four runs as appropriate) were then plotted against the relative humidity for each trial. Plotting

138

T.K.-C. Chan, B.W. Darvell / Dental Materials 17 (2001) 134±141

Fig. 4. TR (log scale) vs. RH for all temperatures and materials.

log(TR) gave a good approach to a straight line in each case (Fig. 4). The temperature dependence was further investigated. The response surfaces for log(TR) vs. RH and temperature are nearly planar in each case, although a slight systematic deviation in the residuals prompted the following. For Stone, Plaster of Paris and Impression Plaster, a plane …z ˆ a 1 bx 1 cy† was ®tted (TableCurve 3D, v3, SPSS, IL, USA) to the data for 25±358C and the residuals extracted for that data and the corresponding values for the excluded 408C data. In the case of Investment, no such systematic variation was detectable, and the plane was ®tted to the full data set and the residuals plotted in a similar fashion. These plots are shown in Fig. 5, and the regression details are given in Table 3. The large and abrupt deviation of the

408C data from the trend established at lower temperatures is plain, the bulk of the points lying well above the two standard error band for the regression (i.e. P , ,0:05 for the distinction), with the exception of the Investment where there is no trace of such an effect. The results of the reversibility checks are shown in Fig. 6. As-supplied, the Impression plaster appeared to have about 1.5% desorbable moisture while the other three products had about 0.5% (Fig. 6a) (all mass changes here are referred to the original measured mass of powder). After 2 h (Fig. 6b), the gains were consistent with those seen in the main trials, and the overall losses returned the masses to similar levels as the as-supplied specimens, with the exception of the Investment which, while returning to the pre-exposure mass, did not lose the 0.5% seen for the as-supplied

T.K.-C. Chan, B.W. Darvell / Dental Materials 17 (2001) 134±141

139

material. After the 1-month exposure, all materials showed ,18±19% gain in mass, but while the Plaster of Paris and the Impression plaster lost about 1% on desiccation, and the Stone about 0.5%, the Investment lost some 12% in about 2 h, with a further much slower and slight loss after that. 4. Discussion The present results con®rm the sensitivity to high humidity, and the rapidity of reaction, previously observed, although in poorer resolution, with Plaster and Stone [1]. More importantly, they clarify the `induction period' observation made then: it is now seen to be simply related to the relative humidity for all four products tested, and evidently without any indication that this is merely a `risk' but rather an inevitability. The high degree of reproducibility between runs is an indication of this through the operation of some continuous process. There is thus expected to be an underlying straightforward chemical kinetic explanation. In the light of this fact, the relationship with relative humidity (RH) is a surprise. RH is a dimensionless quantity de®ned as the ratio of the actual water content of the air (g/ kg) to the quantity that would be present at saturation at the same temperature (g/kg). It therefore is not a chemically meaningful measure in itself. Ordinarily, some other measure such as vapour pressure (representing the activity of the water) would be expected to be involved. However, the corresponding plots against water vapour pressure, which therefore take into account the temperature, spread rather than unify the data (Fig. 7). On the other hand, the logarithmic dependence of TR on RH at a given temperature is in keeping with the usual form of the dependence of the chemical potential m of a species on its activity a [2], i.e.

m ˆ m0 1 RT ln a

Fig. 5. Residuals of plane regression of log(TR) vs. RH and temperature, plotted against temperature. In the case of Stone, Plaster and Impression Plaster the regression was limited to data for 25±358C, but residuals were calculated for 408C data against that ®tted plane (triangles). Open symbols identify highest RH value point; connecting lines indicate sequence. Dashed control line (`2s ') corresponds to twice ®tted regression standard error.

where m0 is a standard state reference value, R the gas constant and T the absolute temperature. It has not been possible to identify precisely why this situation (i.e. RH rather than vapour pressure) arises, although it would seem that the rise in temperature offsets the effect of increasing water vapour pressure by increasing the activity of an intermediate or transition state, which therefore opposes the forward reactionÐthere is an equilibrium of some kind which is nearly temperature-independent over the range 25±358C. Nevertheless, in terms of making recommendations about storage, there is considerable convenience in the near uniformity of response with respect to RH, especially since the observed and small deviation from the general trend at 408C is at a temperature ordinarily not encountered in working conditions, although it is conceivable in storage areas. RH values are commonly obtained from `domestic' instruments and reported in weather bulletins and seem to be well-enough understood in practical terms to make the relevance to store management plain.

140

T.K.-C. Chan, B.W. Darvell / Dental Materials 17 (2001) 134±141

Table 3 Regression analysis of log(TR) vs. RH and temperature. r 2: coef®cient of determination; F: F ratio statistic for regression; P: signi®cance probability; a: constant, b: slope for RH, c: slope for temperature; sd: standard deviation; n: number of points. TRp : predicted TR at 258C and 100% RH Material

r2

F

P

a (sd)

b (sd)

P

c (sd)

P

n

TRp (h)

Stone Plaster of Paris Impression Plaster Investment

0.9466 0.9925 0.9841 0.7710

115.25 857.77 403.47 30.30

5:4 £ 1029 ,1 £ 10213 2 £ 10 212 1:7 £ 1026

10.291 (0.590) 7.348 (0.145) 8.883 (0.283) 2.777 (0.342)

20.0834 (0.0055) 20.0554 (0.0014) 20.0752 (0.0027) 20.0240 (0.0032)

1:3 £ 1029 ,1 £ 10213 4.6 £ 10 213 6:9 £ 1027

20.0323 (0.0091) 20.0253 (0.0022) 20.0231 (0.0043) 0.0023 (0.0041)

0.00346 3:9 £ 1028 0.00014 0.57723

15 15 15 20

14 15 6 3

The behaviour of the Investment is of a different kind. The dependency on RH is much less, the slope being less than half that for the Plaster of Paris, such that less than 1 d exposure is expected to elapse before the rapid stage begins even at 70% RH. The corresponding extrapolated time for the other products is .1000 h. However, it may be easier to judge from the predicted TR at 100% RH (TRp ), taking 258C as a reference temperature, found from the intersection of the ®tted regression with the 100% RH axis (as shown in the last column of Table 3). Thus, the Investment is about ®ve times more sensitive under these conditions than the Stone and Plaster of Paris. This may be due to the presence of hygroscopic salts, which are present in the Investment as part of design to control its expansion on setting or ®ring. These may also change the crystal growth energetics and thus their kinetics. The effect of their presence can be seen in Fig. 6c, where a large proportion of the mass gain was readily reversible. This might account for the lower value of TRp compared with those for Stone and Plaster of Paris, which were otherwise nearly indistinguishable. The adsorption isotherms for water on hemihydrate (-containing) powders might be of interest in understanding the immediate increase in mass on exposure that was not

Fig. 6. Results of reversibility trials: (a) desiccation of as-supplied materials; (b) after 2 h at 358C and 98% RH; (c) after month at 358C and 98% RH.

Fig. 7. Typical set of data for TR plotted vs. calculated water vapour pressure.

T.K.-C. Chan, B.W. Darvell / Dental Materials 17 (2001) 134±141

pursued here. Although there is some evidence of an RHdependent variation in that data (Fig. 3), since the powders were not desiccated before use, and room conditions for the initial handling were not as tightly controlled (50±60 RH at 23 ^ 0:58C) as in the incubator, some spurious variation would be present and further analysis too imprecise. Even so, the Impression Plaster did seem to have a rather large `free moisture' content (Fig. 6a). That this is merely adsorbed and not reacted water is seen from Fig. 6b, where after 2-h exposure at 98% RH, the return was to masses corresponding to those seen on drying the as-supplied material (except for the Investment). This suggests that no harm is to be expected from this absorbed water. However, it is clear that the exact nature of the conditions in the powder causing the transition to the Rapid stage requires elucidation.

141

Acknowledgements This work was undertaken in partial ful®lment of the requirements for the degree of MPhil by T.K.C.C. We wish to thank Mr P.K.D. Lee of the Dental Materials Science Laboratory for electrical support.

References [1] Torrance A, Darvell BW. Effect of humidity on calcium sulphate hemihydrate. Austral Dent J 1990;35(3):230±5. [2] Pitzer KS, Brewer L. Thermodynamics. 2nd ed. New York: McGrawHill, 1961.