X-ray microanalysis – EDX – of cementitious materials

Energy-dispersive X-ray microanalysis (or EDX, EDS, EDAX) is the most common form of X-ray microanalysis. X-rays are generated by the interaction of electrons with the specimen and the resulting X-ray spectra are used for qualitative or quantitative analysis. Spectra can be obtained from individual crystals, or from an area of the specimen containing more than one crystal type.

In the X-ray microanalysis of cement, with a little practice the different cementitious compounds can be recognized from their X-ray spectra within a few seconds of starting the spectrum acquisition. Some examples follow.

X-ray Spectrum of calcium silicate hydrate

X-ray Spectrum
X-ray spectrum of calcium silicate hydrate; the strongest peaks are due to calcium and silicon, with weak peaks also present due to magnesium, aluminium, sulfur and iron.

X-ray Spectrum of ettringite

X-ray spectrum of ettringite, showing strong peaks due to calcium, sulfur and aluminium.

X-ray spectrum of alite, showing strong peaks due to calcium and silicon.

X-ray spectrum of alite, showing strong peaks due to calcium and silicon and very weak peaks due to magnesium, iron and aluminium.

X-ray spectrum of belite

X-ray spectrum of belite, showing strong peaks due to calcium and silicon, with weak peaks due to aluminium, sulfur, potassium and iron.

(Note that, although oxygen was present in the specimens, no oxygen peak is shown in these spectra as the X-ray detector used was not sensitive to light elements.)

Usually, it is easy to tell what phases are present just by doing a qualitative analysis. Often, though, we want to collect quantitative data. This may be, for example, to assess the proportions of Portland cement and slag in a mix, or to find out in more detail what hydration products are present in a cement paste, or to help determine whether a clinker is reduced.

In X-ray microanalysis, good quantitative analysis requires a flat surface – that is, we need to make a polished section of the material we wish to examine. Without a polished section the path lengths of X-rays passing through the specimen towards the detector will be highly variable, leading to errors in the analysis. Examples of analyses of two of the main Portland cement clinker minerals follow.

 

Table 1  Quantitative analyses (as oxide weight percent.) of the alite and belite spectra above.
Na2O MgO Al2O3 SiO2 SO3 K2O CaO TiO2 Mn2O3 Fe2O3
Alite 0.1 0.3 1.3 24.8 0.1 0.1 72.6 0.0 0.0 0.7
Belite 0.1 0.1 0.7 33.6 0.2 0.5 64.4 0.1 0.0 0.3

 

In Table 1, the data is shown as weight of oxides. This is standard procedure for X-ray microanalysis data of unhydrated materials and follows normal geological convention. Oxygen was calculated by stoichiometry.

For X-ray microanalysis of hydrated cement, data are normally presented in elemental format, not as oxides. This is because water is present in significant amounts but it cannot be quantified. Data for hydrated cement is often presented in graphical form as plots of atomic ratios. This enables a large quantity of data to be interpreted quickly, identifying the principle hydration products.

Plot of Si/Ca v Al/Ca

Atomic ratio plot of Si/Ca v Al/Ca for a cement paste, showing mixtures of calcium silicate hydrate (C-S-H) and AFm phase, and C-S-H and calcium hydroxide (CH).

In the above plot of Si/Ca v Al/Ca, there is a cluster of points near (0.5, 0.05) corresponding to calcium silicate hydrate (C-S-H). There is another cluster near (0.0, 0.5), corresponding to AFm phase. We can’t tell from this which AFm phase is present – it could be monosulfate, monocarbonate or several others. There are numerous points on the plot indicating mixtures, both of C-S-H and AFm, and of C-S-H and calcium hydroxide (CH). Data points corresponding to CH (or gypsum) will evidently plot at the origin, since they do not contain silicon or aluminium. The C-S-H cluster is centred at approximately (0.5, 0.05) because C-S-H contains a small proportion of aluminium. However, it is not necessarily clear whether this aluminium is part of the C-S-H structure or present as small inclusions of other phases such as AFm. To estimate the Si/Ca ratio of the C-S-H it is therefore common practice to extend the line representing mixtures of C-S-H and AFm (or in some mixes C-S-H and AFt) to the origin. In this example, the Si/Ca ratio of the C-S-H is approximately 0.60.

The next plot is of atomic ratios of Al/Ca v S/Ca. The purpose of this plot is to differentiate between different forms of hydration product, principally AFm phase, that could not be differentiated in the plot above of Si/Ca v Al/Ca.

Plot of Al/Ca v S/Ca

Atomic ratio plot of Al/Ca v S/Ca, showing points falling on a line between a cluster close to the origin (C-S-H) and the expected position of monosulfate phase, indicating a mixture of C-S-H and monosulfate phase.

The first point to note is that the second plot uses data from the same spectra as the first plot. When the spectra were quantified, elemental concentrations were calculated for all the elements observed in the spectra: Na, Mg, Al, Si, S, K, Ca, Ti, Fe. The two plots were produced simply by calculating ratios of different elements from each analysis. Each spectrum collected is therefore represented in each of the plots.

In the second plot, a cluster of points is visible near the origin. This corresponds to the cluster representing C-S-H in the plot of Si/Ca v Al/Ca. Pink squares represent the calculated composition of pure ettringite and pure monosulfate phase. A well-defined line links the C-S-H cluster near the origin to the composition of pure monosulfate. This line corresponds to the line linking C-S-H with AFm in the plot of Si/Ca v Al/Ca and shows clearly that the paste contains the monosulfate form of AFm phase.

If monocarbonate were present instead of monosulfate, the plot of Si/Ca v Al/Ca would appear the same, since neither the sulfate or carbonate anions that differentiate these forms of AFm will affect the data. However, the plot of Al/Ca v S/Ca would be quite different, with a cluster near (0.5, 0), since monocarbonate does not contain sulfate. Caution is need here since other pure phases, principally hydroxy-AFm, hemicarbonate and C4AF will plot at the same location.

The atomic ratios used above – Si/Ca v Al/Ca and Al/Ca v S/Ca are the most commonly used in SEM cementitious work. However, other ratios may also be useful, such as Mg/Si if slag is present. Additionally, substitutions can be allowed for, principally those of iron for aluminium in AFm and AFt phases, and magnesium for calcium. For example, instead of plotting Si/Ca v Al/Ca the ratios of Si/(Ca+Mg) v (Al+Fe)/Ca would be plotted. Often, though, the proportions of these minor substitutions make little difference to the overall shape of the plot.

It is a simple matter to set up a spreadsheet containing all the quantitative X-ray microanalysis data from a specimen and then to plot Si/Ca v Al/Ca and Si/(Ca+Mg) v (Al+Fe)/Ca – and maybe some others – then use this as a template for future analyses. Once the EDX data is collected, it can be imported into the spreadsheet and the different plots will be generated automatically.

Much more information on qualitative and quantitative X-ray microanalysis can be found in Chapters 5, 6, 7 and 8 of “Scanning Electron Microscopy of Cement and Concrete”.