It is reported an innovative methodology based on in situ MoKα1 laboratory X-ray powder diffraction (LXRPD)
and microtomography (μCT) avoiding any sample conditioning. The pastes are injected in 2.0 mm capillaries and
the extremes are just sealed. The measurements take place in the same region of the hydrating paste. Thick
capillaries are key to avoiding self-desiccation, which dictates the need of high-energy X-ray radiation for the
diffraction study. This approach has been tested with a PC 42.5 R paste having w/c = 0.50. μCT data were
collected at 12 h and 1, 3, 7 and 79 days. LXRPD data were acquired at 1, 3, 7 and 77 days. In this proof-ofprinciple
research, the same paste was also cured ex situ. Portlandite contents obtained by thermal analysis,
ex situ powder diffraction, in situ mass balance calculation and in situ powder diffraction were 13.8, 13.1, 13.1
and 12.5 wt%, respectively. From the μCT study, the grey value histogram evolution with time showed a crossing
point which allowed us to distinguish (appearing) hydrated products from (dissolving) unhydrated cement
particles. Segmentations were carried out by global thresholding and the random forest approach (one type of
supervised Machine Learning). The comparison of the segmented results for the unhydrated cement fraction and
the Rietveld quantitative phase analysis outputs gave an agreement of 2 %. The potential of this methodology to
deal with more complex binders is also presented.