Sample Preparation for Quantitative X-Ray Diffraction (XRD/Rietveld)

Theoretical background: see chapters 3 and 4 in Lieven Machiels’s PhD thesis (2010). Sample preparation method modified from Machiels et al. (2008).

Representative sampling

The mineralogy and glass content of slags depend largely on the mode of cooling of the slags, e.g. slow cooling in a slag pot can result in a large amount of crystalline phases, and fast granulation in water can result in high glass content. A mineralogical analysis will thus only represent the mineralogy of a slag system for a certain grade of cooling and strongly different results can be obtained when cooling conditions are different between the different batches.

Even in a single slag pot, mineralogy and glass content can vary strongly. Taking a sample representative for a slag pot can be done by performing the sampling after a first size reduction of the bulk material or, by mixing representative parts of the slag pot (sides of the pot, center, near cracks, in center of slags, etc.)


A grain size of <10 micron is required for quantitative X-ray diffraction (Rietveld). To avoid amorphisation of the sample during grinding, wet milling is done in a McCrone Micronizing mill. Before grinding in the McCrone Mill, the samples are crushed and passed through a 500 micron sieve.

Take representative amounts of sample (50-100 g). Crush the sample by hand in a porcelain mortar. Use shock impact for grinding, avoid shearing. A jaw crusher can be used, but automatic milling devices which could induce shear stress of amorphisation such as ball mills should be avoided!

McCrone Micronizing mill

Micronizing of the sample

– Weigh 2.7 g of sample; add 0.3 g (10%) of ZnO internal standard. Note down the exact weights, they will be used in the calculations

– Micronize the samples in a MeCrone Micronizing mill using 5 ml of ethanol (methanol) as grinding agent and a grinding time of 5 minutes for soft material (e. g. limestones) up to 10 minutes for hard materials (eg. quartzites, slags). Since methanol tends to react with some artificial minerals, and as it is toxic, ethanol is preferred. To ensure that samples are ground up to < 10 micron, the size required for X-ray quantification, the grain size is best checked by (wet) laser diffractometry.

– After micronizing, recuperate the sample in porcelain cups. Cover the cups with plastic foil, because recovery of powder from the porcelain when dried is difficult. Wash with methanol to recuperate as much as possible of the sample.

– Dry for one—two days under a fume hood (methanol is toxic).

Preparation for X-ray diffraction

– Dried samples are gently disaggregated in an agate mortar and passed through a 250 micron sieve, to ensure good mixing of sample and ZnO standard.

– +/- 0.5 g of sample is needed for the X-ray analysis. Side-loading with frosted glass is recommended to fill sample holders, to prevent preferential orientation of fibrous zeolites and clay minerals. Sample holders are gently tapped while filling, to ensure good packing of the grains. Alternatively, back loading is used.

The content is offered by Dr. Lieven Machiels

Blended cement

Microstructure development of limestone blended cement

Though the chemical reaction of the fine particles of limestone in cement paste is limited to a low extent due to its inert property, the addition of limestone particles has significant effect on the microstructure development. Research conducted by Sellevold et al. using mercury intrusion method showed that specimens containing 12% CaCO3, which have the main body of the particles smaller than 0.1 µm had finer pore structure and somewhat reduced total pore volume.

They attribute this observation to the nucleation effect: The introduction of a large number of nucleation sites could result in a more homogeneous distribution of calcium silicate hydrate and thus a less open pore structure. The effect appears to be in the capillary pore structure rather than in the gel pores. The size and distribution of CH are also influenced by the presence of limestone.

In a limestone-filled paste which used 25% limestone larger regions of calcium hydroxide were unevenly distributed throughout paste while in the case of Portland cement small regions calcium hydroxide were evenly distributed. Further investigation found that limestone added in amounts of 5% or 25% enhance the formation of hydration rims of calcium silicate hydrate surrounding C3S particles due to the accelerated hydration rate of C3S.

Results from the image analysis showed that the interface between limestone and hydrates in limestone-filled self-compacting concrete (SCC) is quite porous, while the interface around the unhydrated cement in high performance cement paste is much denser.
Figures show a comparison of microstructure using BSE between limestone-filled cement paste and Portland cement paste. Two possible reasons could be responsible for this distinct difference in mcrostruture.

On the one hand, as the hydration process of cement proceeds, cement particles expand and get connect with each other to form a network, which result in strength gain; However, limestone particles do not expand during the whole hydration process due to its low reactivity, thus making the porosity of the former becomes lower than the latter one.

On the other hand, the addition of limestone may play a negative role in the microstructure development process, because it could block the expansion of hydrating cement, which makes the microstructure of limestone-filled cement paste more porous as showed in figure.

Comparison of BSE images of limestone cement (up) and Portland cement (down) at age of 7 days

Research by Pipilikaki et al. showed that limestone used in blended cements changes hardened cements’ pore structure. The capillaries pore size is increased from 20 nm to 40 nm when the maximum amount (35%) of limestone that is allowed by EN 197-1. At the same time, the addition of limestone reduces the threshold diameter of the paste and presents a more uniform pore size distribution.

The author concluded that this behavior can be attributed to the better packing of particles due to the filling effect of the limestone powder, resulting in that chemical species enter the hardened cement more difficult but transport easier inside the cement paste due to the pores have approximately the same size. As gel pores are concerned, addition of limestone in cement produces smaller gel pores probably due to the accelerated hydration rates.

Blended cement

Low reactivity and reported reaction extent of limestone in blended cement

The extent of reaction of limestone with cement clinkers is relatively low than the corresponding values of cement clinkers, the probable cause for this as stated above arise from the low solubility and low reactivity of limestone particles, though other factors of limestone, including fineness, particle size distribution, etc., do have influence on its reaction degree in blended system.

In experiments of 5% or 15% calcium carbonate reaction with type II cement hydrated up to one year, Klemm and Adams observed that monocarbonate hydrate is formed slower than that of ettringite, and after 129 days hydration, the amount of unreacted CaCO3 is still up to 80-90%. Ramachandran investigated that in mixtures hydrated with 15% and 5% limestone substitution, the amounts of limestone reacted with the hydrating C3S were reacted within the first 3 days. A typical simulation performed by Bentz showed that for a 20% by mass fraction substitution of ground limestone for cement, only about 5% of the limestone present reacts during the first 180d of hydration.

Taylor summarized that the maximum quantity of CaCO3 that can react appears to be 2-3% with most cements, and value up to 5.8% have also been reported; however, he did not mention the substitution level of limestone in cement. The relatively low reaction degree of limestone in blended cement is probably because of the low reactivity of limestone which is also the reason why limestone is regarded as inert “filler” by many researchers.

It is worth noting that the conclusion limestone being regarded as inert filler always arise from the fact high levels of limestone applied. At low substitution level, e.g. using up to 5% calcite, much or perhaps all of the added calcite is reactive with cement.

In terms of the quantitative relationship between the reactive fractions of limestone and time, it is still a challenge, no fundamental result is available from literature, not to mention limestone is considered as inert filler by many researchers provided that large amount of limestone is applied.