Acceleration of the setting of hemihydrate plaster with calcium sulfate dihydrate

1 Introduction
Calcium sulfate binders are required to have widely varying properties to meet technical and economic requirements. In particular, it has to be possible to vary the workability time and the time of the onset of stiffening of the binders over a period ranging from a few minutes to several hours. To cope with these requirements it is necessary to use additives to regulate the setting.
 
Acceleration of the setting is of crucial importance in the production of gypsum wallboard. Modern plants for producing gypsum wallboard can reach manufacturing speeds of up to
180 m per minute. The plant can operate at maximum capacity only by means of sharply accelerated setting of the binder used (hemihydrate plaster = predominantly beta-hemihydrate = CaSO₄ ∙ 0.5  H_­2O). The use of highly effective accelerators is essential in this process. Very finely ground gypsum (dihydrate = CaSO₄ ∙ 2  H_­2O) is used as the setting accelerator. However, the understanding of its mode of operation and of the factors affecting its production and application is based essentially on empirical investigations. It is crucially important for the continuous gypsum wallboard production process that the accelerating effect is extremely stable. Production experience indicates that there are serious problems in producing a setting accelerator that remains uniformly effective over a long period.
 
This situation gave rise to these investigations, which were aimed at improving the effectiveness of the additive and stabilizing it at a high level by establishing the main influencing variables during the production and application of dihydrate-based setting accelerators.
 
2    Current state of knowledge
2.1 The setting of hemihydrate plaster
The following three aspects were taken into consideration when describing the setting process of calcined gypsum:
–    hydration    –    physico-chemical aspect
–    crystallization    –    mineralogical aspect
–    solidification    –    technological aspect
 
It is well known that hydration generally involves a dissolving-precipitating process. A solid phase releases ions to the liquid phase to reach a state of equilibrium with it. In turn, the liquid phase tries to achieve equilibrium with a second solid phase that has a lower solubility and precipitates from the solution. At 20 °C, for example, the solubility of hemihydrate is 0.885 g
CaSO4/100 g H2O and that of dihydrate is 0.204 g CaSO4/
100 g H2O. If the solution is saturated with respect to the hemihydrate it is then supersaturated with respect to dihydrate and is therefore metastable. In the presence of a crystallization nucleus the excess dissolved substance that is beyond the saturation concentration is precipitated in accordance with the following equation:
 
CaSO4 ∙ 0.5 H2O + 1.5 H2O  g  CaSO4 ∙ 2 H­2O
 
The formation of gypsum crystals follows the law of nucleation and crystal growth. Immediately after supersaturation occurs the first precipitates of dihydrate can be observed “in the form of localized agglomerations of very small crystals with excess water” [2]. The degree of order of these “clusters” increases through further crystal growth, and elongated dihydrate crystals are formed.
The crystallization process results in the intergrowth and mechanical matting of the gypsum crystals with one another and with the remaining unhydrated constituents of the binder matrix. This causes increasing solidification of the binder slurry. An ever denser crystal structure is formed that exhibits its greatest strength after complete evaporation of surplus pore water that is not needed for the hydration.
 
2.2 Setting acceleration
The following factors play a part in the kinetic course of the setting process for hemihydrate plaster:

–    Factors originating from the binder itself
    ·    starting material (crystal morphology, impurities, ...)
    ·    calcining process (phase composition)
    ·    particle size and specific surface area
    ·    particle breakdown
    ·    ageing
 
–    External factors
    ·    water/binder ratio
    ·    temperature
    ·    mixing process
    ·    absorptivity of adjacent materials
(e.  g. plaster base, paperboard, ...)
    ·    additives
 
The rate of setting of hemihydrate plaster can be precisely ­controlled by the use of additives (e.  g. [2–6]). During the investigation of the effect of more than 100 set-regulating additives on the stiffening times of hemihydrate plaster Bertoldi [7] reached the general conclusion that the greatest “sensitivity of the hydration mechanism” lies in the nucleation phase. Additives that interfere with the nucleation – predominantly the retarders and the dihydrate-based accelerators – have a significantly stronger effect than substances that operate by affecting the solubility.
 
Calcium sulfate dihydrate is now widely used to accelerate hemihydrate plaster (e.  g. [8–10]), even though the understanding of its mode of action and of the factors affecting its production and application is based essentially on empirical investigations.
 
3    Investigative methods and materials
Various hemihydrate plasters based on FGD gypsum and natural gypsum that had been calcined in kettles on an industrial scale were used as the binders to be accelerated. The gypsum raw materials used for producing the setting accelerators were various, carefully selected, FGD gypsums and natural gypsums.
 
The very finely ground gypsums were characterized by determining the crystal water content, by measuring the BET specific surface area and by using laser granulometry to describe the particle size distribution. The hydration process of the hemihydrate plaster to be accelerated was analysed by using differential calorimetry (DCA), carrying out conductometric investigations and determining the degree of hydration by meas­uring the crystal water. The stiffening behaviour of the hemihydrate plaster was characterized with the aid of established penetration procedures and by ultrasound technology. Scanning electron microscopy (SEM as well as ESEM-FEG) and X-ray phase analysis were used.
4    Results
4.1 Mode of operation of the accelerator
The set-accelerating effectiveness of selected materials was compared in preliminary trials. Before the solids (gypsum, hemihydrate plaster, anhydrite II, limestone and quartz) were added to the hemihydrate plaster to be accelerated they were very finely comminuted in a laboratory vibratory disk mill (at equal stressing intensity). The results confirm that the setting of dihydrate plaster can be accelerated considerably more strongly by the presence of gypsum particles than by the addition of the other solids.
 
Electron microscopic examination of the hydration of hemihydrate plaster in the presence of fresh gypsum and quartz fracture surfaces provided a clear indication of the reason for this (Fig. 1). Using an environmental scanning electron microscope (ESEM-FEG) it was established that needle-like gypsum crystals grow directionally on a selenite surface (dihydrate), i. e. orientated on the existing selenite crystal structure, when in contact with calcium sulfate solution that is supersaturated with respect to dihydrate.
 
It was shown visually that immediate crystal growth (without prior nucleation) can take place on the surface of gypsum particles. This “seed crystal effect” starts within a few seconds after the hemihydrate plaster is sprinkled on water – as soon as the solution of calcium and sulfate ions is supersaturated with respect to dihydrate. As a consequence, the size of the surface area of the gypsum particles determines the rate of hydration of the hemihydrate plaster to be accelerated.
 
The accelerating action of gypsum in the form of seed crystals was made clear by differential calorimetric investigations (Fig. 2). The maximum in the heat release during the hydration of hemihydrate plaster is displaced towards earlier times in the presence of gypsum particles but the intensity of the hydration is not substantially affected. The shortening of the dormant period of hydration (resting stage before the accelerated progress of the reaction) is the main reason for the substantially earlier stiffening of the hemihydrate plaster on addition of very finely ground gypsum (Table 1). This table shows the change in the stiffening times of a setting hemihydrate plaster mix on addition of very finely ground gypsum rock that had previously been stored at different humidities (storage time 56 days, quantity added: 0.2  %, natural hemihydrate plaster, l/s = 0.6).
 
4.2 Production of a highly effective accelerator
The specific surface area of the gypsum is increased by mechanical stressing in grinding units. Laser granulometry investigations have shown that this produces an extremely wide particle size distribution. The particles in the micron range are dominant in terms of volume and mass but the particles in the nanometre range are predominant in terms of numbers and surface area. Because of their comparatively large ratio of surface area to volume the very fine nano particles are particularly relevant to the accelerating effect.
 
Gypsum crystals with a large number of particles in the range between 100 and 1000 nm were obtained in the laboratory trial by hydration in order to examine this relationship more closely. When used as a hemihydrate plaster additive these crystals had only a slight influence on the setting behaviour pattern. This confirmed that the extremely fine (< 100 nm) gypsum particles with defective lattices produced during grinding are required for the production of a highly effective setting accelerator.
 
The ultrafine particles have a significant tendency to agglomerate both in industrial scale plants and in laboratory mills with greatly increased input of grinding energy. This means that the success of grinding can only be quantified to a limited extent by the measurable increase in surface area. The increased accelerating action of the gypsum with respect to the setting of hemihydrate plaster caused by ­ultrafine grinding can be shown indirectly by the change in stiffening behaviour.
 
Analysis of the production process and laboratory investigations have shown that the changing ­engineering parameters of the grinding units and differences in the composition of the ­material ­being ground cause fluctuations in the quality of the accelerating additive. If, for example, partial ­dehydration of the dihydrate occurs during the grinding process as a result of heating of the material being ground then this leads to appreciable loss of the set-accelerating effectiveness. This is because of the phase conversion of the very finest particles of dihydrate into hemihydrate, with the result that they can no longer function as seed crystals.
 
The choice of mill feed material also has a crucial influence on the grinding result and therefore on the effectiveness of the setting accelerator. Natural gypsums that contain large quantities of harder and more brittle constituents (such as quartz or dolomite) are more effective as accelerators than natural gypsums with fewer impurities or than FGD gypsums. With the aid of laser granulometry and the Rietveld refinement of X-ray analy­sis it was shown that these foreign constituents in the natural gypsum act as grinding aids during the grinding process and lead to increased formation of extremely fine gypsum particles with a high level of defective lattices. The investigations also showed that the geological period in which the gypsum rock was formed is of secondary importance for the accelerating action of the ground product when compared with the above-mentioned effect of the impurities.
 
4.3 Influence of the storage conditions
on the effectiveness of the accelerator
Instability of the set-accelerating effectiveness is associated not only with the production process but also with the storage conditions. High levels of humidity cause a sharp reduction in the set-accelerating effectiveness of very finely ground dihydrate within a short time. This process, also known as ­
“ageing” [11], becomes apparent from the change in the stiffening times of setting hemihydrate plaster mixes on addition of very finely ground gypsums that have been stored under different conditions (Table 1).
 
This relationship is confirmed by the results of differential calo­rimetric investigations. When compared with hydration using dihydrate-based setting accelerators that have been stored under dry conditions the maximum in the rate of heat evolution is displaced towards that of the reference sample without accelerator. As expected, the total quantity of heat is not affected by the addition of accelerator (Fig. 2).
Electron microscope photomicrographs taken using ESEM show surface degradation caused by moist storage (Fig. 3). There is a reduction in the number of extremely fine particles. The edges and corners of the remaining fine particles become rounded and surface irregularities disappear. This can be explained by the fact that recrystallization processes occur through condensation in the interstitial spaces between the particles, during which some of the very small particles are completely dissolved and the surfaces of larger ones start to dissolve.
 
Once again the reduced set-accelerating effectiveness of the gypsums that have been “aged” in this way indicates the special role of extremely fine particles (Table 1).
 
As has been confirmed by other investigations, clayey impu­rities – which are often found in natural gypsums – can retard the “ageing process” of the dihydrate particles produced by very fine grinding. This is mainly because the inclusion of water in the structure of the clay minerals restricts the access of moisture to the extremely fine gypsum particles in clay-gypsum agglomerates.
 
The use of sugar as a grinding additive was also investigated. Enveloping the dihydrate particles in sugar or starch and the resulting retarded progress of ageing of the setting accelerator is frequently stated in the literature as the reason for the improved set-accelerating effectiveness (e. g. according to [8, 10, 12–15]). The term “coating” has also been applied in this context. During the course of the work it became apparent that the harder and more brittle (compared with gypsum) sugar crystals act like foreign mineral constituents (such as quartz or dolomite) in the mill feed material. Their presence during the gypsum grinding creates increased quantities of the extremely fine dihydrate particles with defective lattices that are relevant to the acceleration. However, the laboratory trial showed no indication that sugar had any influence as a grinding additive on the decrease in effectiveness during moist storage of the setting accelerator.
 
5    Conclusions
As a result of the work the known process of accelerating the setting of hemihydrate plaster with calcium sulfate dihydrate was analysed systematically in its entirety and new data in certain specific areas was provided. A model for the acceleration of the setting of hemihydrate plaster by calcium sulfate has been derived from the theoretical knowledge and the results of the experimental investigations.
 
The following relationship becomes clear from initial examination of the setting of hemihydrate plaster without any accelera­ting additive (Fig. 4). Some of the binder dissolves immediately after the hemihydrate plaster comes into contact with water (I, induction period) and a solution is formed rapidly that is saturated with respect to hemihydrate and supersaturated with respect to dihydrate. Because of the effect of the supersaturation on the crystallization kinetics the high level of supersaturation then results in preferential crystal nucleation (II, dormant period). This process takes place preferentially on the surfaces of the remaining solid particles (predominantly hemihydrate plaster) (= heterogeneous nucleation). These primary hydrates grow slowly, stabilize themselves and can take over the function of nucleation. A now greatly increased number of nuclei then grow into gypsum crystals (III, acceleration period). The super­saturation of the solution falls with the decrease in available material that can be hydrated, which in turn slows down the growth of the gypsum crystals and causes a change in morphology. Recrystallization phenomena eventually occur when there is only a small degree of supersaturation.
 
In contrast, the setting of hemihydrate plaster in the presence of seed crystals can be described as follows (Fig. 5). The basis of the process is again that part of the binder dissolves as a result of the contact of hemihydrate plaster with water and a solution that is supersaturated with respect to dihydrate is produced. At an even lower degree of supersaturation than is necessary for the nucleation process (Fig. 4 and Fig. 5, variant C) there is immediate crystal growth in the presence of seed crystals (variant A). The extremely fine gypsum particles with their surface defects in the gypsum crystal lattice that are produced during very fine grinding act as the centres for crystallization. Consequently, it is not the mass of dihydrate that is critical for the seed crystal effect of the dihydrate but the number of seed crystals, characterized indirectly by the particle surface area. The setting of the hemihydrate plaster is increasingly accelerated with rising number of extremely fine particles (e. g. by increasing the quantity of accelerator added, by increasing the grinding intensity or by the presence of harder and more brittle foreign constituents in the mill feed material).

On the other hand, a decrease in the set-accelerating effect can be expected if, before it is used as a setting accelerator, the finely ground gypsum comes into contact with high humidity (e. g. during storage in a moist environment) or if it is added too early to the mixing water (i. e. to a calcium sulfate solution that is not yet saturated) (variant B). The complete dissolving of some of the very small particles and the surface phenomena of partial dissolving – which is also particularly responsible for the reduction in the crystallographic lattice defects ­– are regarded as the reason for the decrease in effectiveness of dihydrate-based setting accelerators.