for ancient samples. Gyp: Gypsum, Cal: Calcite, ND: Non-determined.
The grey colour of samples from CON and HED may come from the addition of charcoal. The change of colour from grey to white and a peak at 430 °C seem to indicate the presence of charcoal (Shi et al, 2012).
Thin sections and polished surfaces (Fig. 1) of samples from THM, CON, HED, FStP, ART and StC were observed. Figure 1. THM, CON and ART samples present two different layers. We can observe the differences in grain size between the two layers but also between different samples taken at the same façade. THM sample has a red-colour superficial layer which is partially detached. CON sample is the only one that contains agglomerations of small gypsum crystals as well as bigger crystals. In the THM sample the two layers appear detached by a fissure. In ART there is also a fissure parallel to the surface but in the inner layer. HED sample shows a thin layer in the surface with recrystallization zones. Quartz sand grains in the Saint Claire sample are a little bit bigger than gypsum crystals in the other samples. In most of the samples charcoal grains can be observed, only ART and FStP do not show any charcoal. In TGA results this last sample has only 2.5 % of “undetermined” components, which is the lowest value of all the samples. Porosity and density results for historic samples are shown in Figure 2. A general direct relationship is observed for samples FStP, ART, HED and THM, while the CON sample has a higher porosity with respect to its density which can be explained by its higher content of calcite compared to the other samples. Density of calcite and gypsum are 2,700 kg/m3 and 2,300 kg/m3, respectively. FStP has an extraordinarily high porosity and a low density that can be explained by the small grain size and the high gypsum content. Figure 2 In general, good correlation was found between density and porosity with the exception of the lime-rich sample mentioned above, a fact that supports the similarity in composition between the respective samples.
Figure 1: Digital optical incident light microscopy. All images are at the same magnification.
Water vapour absorption/desorption results for samples from historic renders are presented in Figure 3. A first group of samples absorbs much less water vapour than the others at 90 % relative humidity, THM (0.15 %), FStP (0.5 %) and ART (0.77 %). Two other samples present intermediate values, StC (1.66 %) and Heudouville (2.5 %). The 69CON sample has the highest value, 4 %. Samples with low water vapour absorption have higher contents of gypsum. CON samples, with the highest absorption, have particles with very fine crystals which can explain this result due to a high specific surface. HED has the highest percentage of undetermined minerals in TGA which may correspond to the charcoal content (Fig. 2). The StC sample is quite different from the others, with lime binder containing almost 85 % of quartz with low water vapour absorption properties.
Figure 2: Porosity and density values of ancient samples.
Figure 3: Water vapour absorption/desorption curves for some old renders.
The CON sample absorbs the maximum of vapour and has the highest hysteresis of desorption. 20 % of the absorbed water remains in the sample after the test. THM sample absorption is the lowest one but it keeps 18 % of the water after desorption. StC sample keeps 10 % and the three others around 4 %. The differences in the sorption/desorption hysteresis cannot be explained.
Laboratory samples
TGA analysis of laboratory made plasters was performed shortly after their preparation, in which lime had not time to carbonate. The results obtained are not comparable to samples from historic renders.
Porosity of new plaster samples goes from 58 % in samples of rough render for exterior walls with lime or charcoal grains, to 44 % in fat or pure gypsum renders.
In new render samples, the amount of absorbed water vapour by unit of mass goes from 0.18 to 0.50 i. e. values similar to THM, FStP and StC samples. The highest values correspond to samples with the highest content of lime or charcoal. Lowest values correspond to pure lime samples.
Water vapour permeability was measured for new mortars. Permeance in kg/(m² × s × Pa) goes from 1E-13 to 2E-13. The highest values correspond to samples with 20 % of lime and the lowest ones to pure plaster renders. Tests have been done at three different ages, 10, 25 and 90 days, and the permeability values slightly decrease with time. Compression and flexural strength were measured for several samples of new plaster. Compressive strength varies from less than 1 MPa in some samples of pure gypsum with charcoal to more than 6 MPa in samples of pure gypsum with or without lime. At the tested ages the admixture of lime does not produce any increase in strength, as more time is necessary to complete carbonation of lime.
The amount of mixing water controls mechanical and hydric properties. With increasing amounts of mixing water, compression strength decreases, porosity and water vapour absorption increase and density decreases. The most important variation has been observed in water vapour permeability. In several studied renders, a diminution of 15 % in mixing water induces a decrease of water vapour permeability between 30 and 40 %.
Conclusions
In this paper we present preliminary results in order to understand the properties of gypsum renders as a function of their composition, binder and aggregates, but also of the samples preparation, especially the amount of mixing water, It is difficult to draw general conclusions, as the sampling process is not easy and obtained samples cannot be considered as representative of the whole population of renders in the Vexin Français area.
Another difficulty of the study is the age of the samples. This study has been done during a Master degree research project of 6 months. Samples from buildings are at least 100 years old but laboratory samples have less than 3 months. This difference is especially important for samples with lime. In new 70samples we find some portlandite, not found in old samples, not even in Saint Clair sur Epte sample, fabricated entirely with lime binder.
Nevertheless, some conclusions can be inferred from this work:
a) Historical studied renders are very different from one to the other, even if most of them are composed of more than 80 % of gypsum. Aggregates granulometry and composition, amount of charcoal, number of layers, colour, etc. vary from one sample to the other.
b) The amount of calcite or quartz aggregates is more important in renders of farms or enclosure walls than in façades. Decorative elements are composed almost exclusively of gypsum. The composition also depends on the location of a building in respect to the areas of plaster of Paris production.
c) Mixing water and the amount of lime control the properties of renders. In this way, the plasterer’s skills determine the quality of the products.
d) Porosity cannot be directly related to the water absorption/desorption of plasters. Sorption/ desorption test seems more adequate to characterise hydric behaviour of plasters than porosity measurements.
e) Permeability of plaster is directly related to the capacity of samples to desorb the absorbed water upon drying.
f) It is difficult to compare new and old plasters due to the alteration of their properties with time.
