Microstructure Characterisation of Ordinary Portland Cement Composites for the Immobilisation of Nuclear Waste

UoM administered thesis: Master of Philosophy

  • Authors:
  • laura dodds


Steel-based container materials with cementitious backfill materials are proposed forimmobilising Intermediate Level radioactive Waste (ILW), and as such theirmicrostructure and properties are of great importance. Cementitious materials havebeen selected because they are durable, reliable, economical, and have long-termstability. The microstructure of cement is complex and contains several solid phases,with the distribution of each affecting the properties and performance of the material.This can lead to changes in its ability to, for example, inhibit corrosion of steelreinforcements, or may affect the permeability of liquids and gaseous phases throughcement microstructure.A variety of techniques have been employed to study the structure, composition andphase distribution of Ordinary Portland Cement (OPC), OPC/Blast Furnace Slag(OPC:BFS) composites of different concentrations and the Nirex Reference VaultBackfill (NRVB). The techniques used include scanning electron microscopy(SEM), energy dispersive x-ray (EDX) spectroscopy, x-ray computed tomography(XCT), and x-ray powder diffraction (XRD).SEM and EDX data confirmed the presence of different phase compositions in theOPC:BFS samples However, due to sample charging problems, attributed to surfacetopography, only general cement compositions of C-S-H, Ca(OH)2 and the BFSderivates could be distinguished. Low magnification (300x and 1000x) gave areasonable overview, but not much local information. Higher magnification (2000x)gave insight into local cement composition, but the large interaction volume was stillinterfering. Subsequent studies may benefit from using lower energy EDX and evenhigher magnification.Powder XRD results were in agreement with previous studies, highlighting thepresence of Calcium Hydroxide (CH), Gehlenite (G), and Monosulfate (AFm) inNRVB. All OPC:BFS spectra produced no identifiable peaks, which may be relatedto the presence of amorphous cement phases, such as BFS derivates, as well as theprolonged storage time of the sample. A XRD spectrum on BFS confirmed poorlycrystalline. Even with high levels of BFS we would not expect to see BFS spectra,only OPC. OPC may transform to C-S-H (poorly crystalline) and as the sampleswere stored for a long period it may have resulted in almost no detectable peak.Using X-ray computed tomography (XCT) four distinct phases were identified; 1)un-reacted cement grains / calcium hydroxide, 2) inner calcium silicate hydrate (C-S-H), 3) air or water filled porosity and 4) Blast Furnace Slag (BFS) derivates.Analysis of greyscale XCT images showed there was significant overlap of phaseboundary locations and intensities, giving rise to errors when estimating overallcompositions.When comparing greyscale distributions for the different OPC-based samples, theoverall results agreed reasonably well with sample compositions. The onediscrepancy between the OPC sample and the BFS-containing samples is the un-reacted cement peak, since all BFS samples had a distinctive fourth peak. Thisdifference may be attributed to different stages of the setting period when the datawas acquired (1 day for OPC; more than 2 weeks for BFS samples), as well as allBFS-samples contained a higher proportion of metallic compounds with highdensity/capture cross section. This would suggest, in fact, that there are actually onlyfour peaks that can reliably segmented in both cases.In order to assess homogeneity of cement samples, the BFS:OPC (7:3) XCT datawere characterised. No discernible difference in the distribution of any of the phaseswas observed in this sample. Although differences were expected to be only small,errors arising from estimating the volume and location of the cement phases mayhave obscured small trends. A higher resolution XCT assessment of BFS:OPC (7:3)gave differences in sample composition greater than could be attributed to randomerrors alone. The biggest change was a decrease in the average grain size of the innerC-S-H phase, which is by far the largest phase and accounts for around 90% of thesample as measured by volume. At the increased resolution the smaller particleswere able to be resolved giving a greater accuracy although this was as a trade ofwith scan area.Using XCT a comparison of the volume fraction of the phases in BFS:OPC 9:1,BFS:OPC 7:3 and NRVB was made. Again the overall comparison was in agreementwith what was expected from sample compositions. The NRVB sample showed alarger volume fraction of porosity.Overall, each of the techniques applied to the samples provided information aboutdifferent parts of the puzzle, allowing a fuller picture to be constructed. The XCTgave some very high quality images providing information about the size, shape anddistribution of the phases in the cement. It also allowed in-situ measurements to bemade of potential induced corrosion of steel encased in cement. Identification of thespecific phases was not a straight forward process, however, and greater clarity wasgained by supporting the XCT data with information from XRD and EDX. XRD wasparticularly useful for providing information about the crystal structure of the bulk ofthe cement. This, in turn, gave information about some of the processes that hadoccurred in the cement, for example, the conversion of OPC into C-S-H after aprolonged period. EDX and SEM gave a lot of information about the surface of thesamples which is particularly of interest as it is the surface of the cement that isgenerally the first place any reactions will occurGenerally, the study provided useful information about the chemical and structuralstates of the composite cements. The work also demonstrated the strengths of thetechniques for use in further studies and raised a lot of further questions which maybe tackled in future research.


Original languageEnglish
Awarding Institution
Award date1 Aug 2013