Long-term in situ X-ray Diffraction Studies on Ordinary Portland Cement Hydration with Correlative X-Ray Mapping

Globally, Portland cement (PC) is one of the most consumed materials with annual global production of approximately 4.6 billion tonnes and global demand is predicted to increase to 6 billion tonnes by 2050 [1]. However, the production of PC has significant environmental impacts and is currently responsible for approximately 8% of global CO2 emissions [2]. For this reason, the development of greener means for producing PC, as well as alternative cements, have become a burgeoning field of research with considerable attention. Furthermore, the hydration reaction mechanisms in PC are still not fully understood and the curing reactions in many alternative cements also require further investigation. Cement hydration mechanisms are complex and involve several phase changes that can develop across brief to prolonged timeframes. For this reason, the use of high-energy X-ray facilities such as synchrotron may not be suitable for long-term in-situ studies due to the cost and demand of beamtime. In contrast, laboratory-based X-ray diffractometers (XRD) have become widespread throughout research institutes and industry, and have undergone many advancements in recent decades in regard to analysis speed, data quality and the widespread adoption of auto-samplers. Furthermore, the recent development of high-resolution energy range 1D position sensitive detectors (PSD) have allowed operators of Cu-source XRD units the ability to filter out fluorescence signals from samples containing Fe without the need for additional filters or monochromators. Data obtained from cement clinkers containing Fe is therefore free of scattering artefacts such as absorption edges from metal filters. Here we describe an approach to long-term ­in-situ XRD analysis of ordinary Portland cement (OPC) hydration. For XRD analysis, a sample of OPC was taken and mixed with potable water until a thick paste ensured. The paste was then mounted into a Bruker airtight sample holder and sealed with Kapton foil. Remaining OPC paste was transferred to an airtight vessel and left to cure for future X-ray Mapping work. All XRD work was carried out on a Bruker D8 Advance XRD in Bragg-Brentano geometry with a Cu X-ray source operating at 40 kV and 40 mA and a Lynxeye XE-T PSD with voltage discriminator settings optimised to reject Fe fluorescence from the sample. In-situ scans from 0-24 hrs were carried out across a 7-53° 2θ scan range with a 0.02° step size and acquisition time of 0.5 seconds per step totalling 20 min. A new scan carried out every 30 min. Scans beyond 24 hrs were carried out with similar parameters but with a wider 5-90° 2θ scan range and a 36 min total scan time. X-ray mapping (XRM) was carried out using a Jeol 7001F field-emission gun scanning electron microscope (FEGSEM) with an Amptek EDS detector and Moran Scientific microanalysis package. Post processing of the X-ray maps was performed using Moran Scientific chemical imaging software [3-5]. To represent the OPC starting material, anhydrous OPC was pressed into a pellet, placed under vacuum then carbon coated. A subsample from the initial hydration reaction was taken and lightly sanded with 2500 grit emery paper. The subsample was then placed under vacuum to dehydrate the material and a carbon coating was then applied. Both samples were analysed under high vacuum at 15 kV and x100 magnification. XRD of anhydrous OPC was initially characterised using a Bruker backloading sample holder. Major phases identified include alite (C3S), belite (C2S), brownmillerite (C4AF), tricalcium aluminate (C3A) plus several additional minor and trace phases including calcium sulfate (gypsum) and others. This sample was then rerun in a Bruker airtight holder fitted with a Kapton foil, allowing us to observe which peaks come from the foil, as the hydrated OPC sample needed to be run in the airtight holder. In order to visualise 7 days of XRD analysis, the hydration of OPC is shown as a 2D XRD plot (Figure 1a). Presenting XRD data this way helps to identify phase changes in large datasets. As shown in Figure 1b, the induction period of hydration where dissolution of gypsum takes place is initially observed, beginning rapidly from hydration until all available gypsum has been consumed at ∼9 hrs. While this occurred, we observed the beginning of the acceleration period which can be noted by the formation of ettringite (AFt) directly from the onset of hydration, and then after ∼3.5 hrs calcium hydroxide (portlandite) formation was observed. Ettringite formation reached peak maxima after ∼18 hrs, while the formation of portlandite occurred rapidly between 3.5 and 11 hrs and continued to form at a slower pace until a peak maxima was reached at ∼24hrs. Additionally, throughout the in-situ experiment, the peak intensities for phases C3S, C2S, C4AF and C3A were reduced and the appearance of calcium aluminate hydrate (AFm) was noted. The formation of calcium silicate hydrate (C-S-H) was notably absent; likely, the signal is concealed by amorphous regions contributed by Kapton foil and free water [6]. Correlative microanalysis conducted on both the anhydrous OPC pellet and the 7-day hydration mixture were in good agreement with the observed phase changes in XRD. A clear and simple indication of this is the dissolution of sulfur-rich material thought to be gypsum. Figure 2 shows the sulfur-rich material is localised in masses throughout the anhydrous OPC pellet, while in the 7-day hydration mix, sulfur is dispersed throughout the sample matrix. It may be possible that the sulfur redistribution is related to the formation of ettringite. In conclusion we have reported XRD over a 7 day time period detailing the hydration reaction of OPC and correlated this with XRM showing the dissolution of sulfur-rich regions. Ongoing studies are being carried out to further probe the intricacies of OPC hydration, and we hope that this methodology will become adaptable to other alternative cement mixtures and substitutes [7]. a) 2D representation of in-situ XRD showing 5-65° 2θ across 7 days of data collection. Phase labels are to the right of phase 100 peak, AFt = ettringite, Gyp = gypsum and CH = portlandite. b) Select scans of initial 24 hr hydration overlayed and shown as a 1D plot. C3S = alite, C2S = belite, C4AF = brownmillerite, C3A = tricalcium aluminate, AFt = ettringite, AFm = calcium aluminium hydrate, Gyp = gypsum and CH = portlandite. Pseudo colour XRM on anhydrous OPC (top left) and 7-day hydration product (top right). Colour channels assigned as Ca = red, S = green and Fe = blue, HWOF = 900 µm. Individual element distributions on anhydrous OPC (Ca = bottom left, S = bottom centre left) and 7-day hydration product (Ca = bottom centre right, S = bottom right).