Accuracy in Cement Hydration Investigations: Combined X-ray Microtomography and Powder Diffraction Analyses

Cement hydration is a very complex set of processes. The evolution of the crystalline phases during hydration can be accurately followed by X-ray powder diffraction data evaluated by the Rietveld method. However, accurate measurements of some microstructural features, including porosity and amorphous content developments, are more challenging. Here, we combine laboratory X-ray powder diffraction and computed microtomography (μCT) to better understand the results of the μCT analyses. Two pastes with different water–cement ratios, 0.45 and 0.65, filled within capillaries of two sizes, ϕ = 0.5 and 1.0 mm, were analysed at 50 days of hydration. It was shown that within the spatial resolution of the measured μCTs, ~2 μm, the water capillary porosity was segmented within the hydrated component fraction. The unhydrated part could be accurately quantified within 2 vol% error. This work is a first step to accurately determining selected hydration features like the hydration degree of amorphous phases of supplementary cementitious materials within cement blends.


Introduction
Cements are multi-phase materials and they can be understood well when studied by a multi-technique approach [1], mainly: (i) X-ray fluorescence for elemental analysis; (ii) X-ray powder diffraction, using the Rietveld method, for quantitative phase analysis; and (iii) a set of techniques for textural characterization (mainly Blaine air permeability, laser diffraction and nitrogen sorption isotherm). The understanding of cement hydration features is less developed. Hydration can take place in pastes, mortars or concretes. Here, the discussion is restricted to the hydration of cement pastes, as mortars and concretes have an additional layer of complexity due to the aggregates.
The hydration of cement pastes is itself very complex due to the interplay of several parameters. Firstly, the starting materials are multiphase with a wide range of particle sizes. Secondly, there is an increasing trend of using supplementary cementitious materials (SCMs) [2] and organic and inorganic admixtures [3]. Thirdly, the hydration conditions can vary: water-to-cement (w/c) mass ratio, temperature and pressure. And fourthly, after some time, hydration conditions can change and important processes for durability could take place such as carbonation or chloride ingress [4].
It is acknowledged that many characterization techniques are necessary to fully understand the features of cement hydration. Interested readers are referred to an excellent book where many are thoroughly treated (i.e., calorimetry, shrinkage measurements, thermal analysis, nuclear magnetic resonance (NMR), mercury intrusion porosimetry, electron microscopies, etc.) [1]. Our work is framed within a current trend of developing procedures that are reliable, relevant and rapid [5]. Powder diffraction and µCT do not require sample preparation once the capillaries are filled. They also have the potential for automation Table 1. Rietveld quantitative phase analysis (RQPA) results (wt%), including amorphous and crystalline non-quantified (ACn) and free water (FW) contents, for the w/c = 0.45 samples hydrated for 50 days at 20 • C and referred to 100 g of paste. Mass density of each phase is also given.  a This binder also contains at t 0 : 0.6 wt% of gypsum and 0.3 wt% of quartz. b These capillaries also contain 0.5 wt% of quartz.

Paste Preparation
Cement pastes were prepared with deionized water (w/c = 0.45 and 0.65) by mechanical stirring at a speed of 800 rpm for 90 s. After waiting for 30 s, another identical stirring step was undertaken. The pastes were injected into glass capillaries of diameters of 0.5 and 1.0 mm with a syringe and sealed with UV hardening resin. In addition, and to prepare reference samples that can also be characterised by thermal analysis, the same pastes were poured into polytetrafluoroethylene (PTFE) cylinder moulds, 35 mm long and 10 mm in diameter, that were hermetically closed and rotated at 16 rpm, at 20 ± 1 • C. The samples were kept in the moulds for 24 or 48 h for the pastes with w/c = 0.45 and 0.65, respectively. Then, the cylinders were demoulded and placed into a bottle of water at 20 ± 1 • C to be used at the selected hydration age. The bottle was tightly closed to minimise carbonation and the amount of water was the minimum to minimise calcium leaching. For a fraction of the cylinders, the hydration of the samples was stopped before their characterisation by manually grinding and washing twice with isopropanol and finally with diethyl ether. The two hydration-arrested powder samples were gently ground with the internal standard, 20 wt% of quartz, for RQPA.

Thermogravimetric Analysis
A SDT-Q600 analyser from TA instruments (New Castle, DE, USA) was used to collect the thermogravimetric (TGA) traces for the four samples, i.e., two hydration-arrested and two neat pastes. The temperature was varied (10 • C/min) from RT to 40 • C, held at 40 • C for 30 min and then raised up to 1000 • C at the same rate. Measurements were made in open platinum crucibles under synthetic air flow. The weight loss from 45 to 550 • C was assigned to chemically bounded water [37,38] and the loss from 550 to 1000 • C was considered as CO 2 . The free water (FW, wt%) was calculated from the TGA traces of the hydration-arrested pastes by assuming that the chemically combined water was responsible for the weight loss from 45 to 550 • C. The FW was obtained by subtracting this value from the total added water (nominal value).

Laboratory X-ray Powder Diffraction (LXRPD) and Data Analysis
The pastes within the capillaries were measured on a D8 ADVANCE (Bruker AXS) diffractometer, located at SCAI at UMA (Malaga, Spain), using monochromatic Mo-Kα 1 radiation (λ = 0.7093 Å). The optics configuration was a primary monochromator, a focusing mirror, 1 or 2 mm anti-scatter slits (for 0.5 and 1 mm of capillary diameters, respectively), and 2.5 • Soller slits for the incident and transmitted beams. The X-ray detector was an EIGER (from DECTRIS, Baden, Switzerland) specially designed and optimized for Mo anodes, which was used with an aperture of 7 × 21 degrees, working in VDO mode. Data were collected from 3 to 35 • (2θ) during approximately 4 h. The tube worked at 50 kV and 50 mA. The pastes from the cylinders were measured in transmission geometry placing the powders between two Kapton foils. Measurements were carried out in the same diffractometer with the same conditions and very similar optics (2 mm incident anti-scatter slit).
All the LXRPD patterns were analysed by the Rietveld method using a GSAS software package [39] by using a pseudo-Voigt peak shape function with the asymmetry correction included [40,41]. The refined overall parameters were: background coefficients, zero-shift error, phase scale factors, unit cell parameters, peak shape parameters and preferred orientation coefficient if needed. The employed crystal structures are given elsewhere [42,43]. The determination of amorphous and crystalline non-quantified (ACn) content was performed by the internal standard methodology [17] mixing the powders with~20 wt% of quartz.

Laboratory X-ray Computed Microtomography (µCT) and Data Processing and Analysis
The µCT images were acquired on a SKYSCAN 2214 (Bruker) scanner at SCAI at UMA. Images were obtained with an X-ray tube using a LaB 6 source (operated at 55 kV and 130 µA) and employing a 0.25 mm Al foil to reduce beam hardening. The capillaries were rotated 360 • during data acquisition. Images were taken every 0.2 • , i.e., 1801 frames, with an exposure time of 2.0 s resulting in an overall recording time of 4.72 h per CT. The geometrical settings were a source to object distance of 14.480 mm and an object to detector distance of 315.449 mm. The detector system was the CCD3 which has a physical pixel size of 17.4 µm (binning 1 × 1), which together with the geometric settings resulted in a voxel size of 0.80 µm.
NRecon (version 2.1.0.1, Bruker) was used for the reconstruction of the CTs. Dataviewer (Bruker) software was used for 3D visualization. The analysed volumes were cylinders with a height of~1215 µm, and diameters of~450 and 900 µm, for φ = 0.5 and 1.0 mm glass capillaries, respectively. Many filtering methods have been developed to reduce the background noise of the images [44,45]. Here, a 3 × 3 × 3 median 3D filter was applied to the four µCTs with Fiji program [46]. This was to reduce image noise and to even out contrast variations between adjacent slices. This procedure slightly sharpens the greyscale value histograms. Fiji was also used for data analysis. Image segmentation was performed by applying a global threshold approach with boundaries determined as described in the results section. Rendered volumes displaying the outcomes of the segmentations were pictured using CTVox software (Bruker).

Thermal Analysis
Thermal analysis data were collected for the powder samples cast as cylinders with w/c = 0.45 and 0.65, i.e., with and without arresting the hydration (four traces). Figure 1. shows the TGA profiles for the neat pastes, i.e., without arresting the hydration, which allows extracting three relevant values: (i) total weight loss which corresponds to all the water plus the contributions from limestone and any hydrated phase within the pristine cement; (ii) the portlandite -CH or Ca(OH) 2 -; and (iii) calcite-CC or CaCO 3 -contents, which are referenced to the neat pastes. This allows a direct comparison with the RQPA results, see below. The percentages are calculated in a two-step process, (a) the weight loss is obtained by the tangential method [47], and (b) the value is converted to weight percentage using the 4.11 or 2.27 factors, for portlandite or calcite, respectively. The FW percentages, given in Tables 1 and 2, were determined from the TGA curves of the hydration-arrested pastes as detailed in the experimental section.
Materials 2021, 14, x FOR PEER REVIEW 5 of 18 mm glass capillaries, respectively. Many filtering methods have been developed to reduce the background noise of the images [44,45]. Here, a 3 × 3 × 3 median 3D filter was applied to the four μCTs with Fiji program [46]. This was to reduce image noise and to even out contrast variations between adjacent slices. This procedure slightly sharpens the greyscale value histograms. Fiji was also used for data analysis. Image segmentation was performed by applying a global threshold approach with boundaries determined as described in the results section. Rendered volumes displaying the outcomes of the segmentations were pictured using CTVox software (Bruker).

Thermal Analysis
Thermal analysis data were collected for the powder samples cast as cylinders with w/c = 0.45 and 0.65, i.e., with and without arresting the hydration (four traces). Figure 1. shows the TGA profiles for the neat pastes, i.e., without arresting the hydration, which allows extracting three relevant values: (i) total weight loss which corresponds to all the water plus the contributions from limestone and any hydrated phase within the pristine cement; (ii) the portlandite -CH or Ca(OH)2-; and (iii) calcite-C ̅ or CaCO3-contents, which are referenced to the neat pastes. This allows a direct comparison with the RQPA results, see below. The percentages are calculated in a two-step process, (a) the weight loss is obtained by the tangential method [47], and (b) the value is converted to weight percentage using the 4.11 or 2.27 factors, for portlandite or calcite, respectively. The FW percentages, given in Tables 1 and 2, were determined from the TGA curves of the hydrationarrested pastes as detailed in the experimental section.

Laboratory X-ray Powder Diffraction Study
For a given paste, three LXRPD patterns were collected: two for the pastes within the capillaries (φ = 0.5 and 1.0 mm) and a third one for the paste casted as cylinder. This last sample underwent hydration stoppage and it was gently milled with the internal standard. The Rietveld fits for the three patterns, w/c = 0.45 pastes, are displayed in Figure 2 and the results are reported in Table 1. The results for the pastes hydrated with w/c = 0.65 at 20 °C for 50 days are given in Table 2 and the corresponding Rietveld plots are displayed in Figure 3. The same calculations/assumptions, described above for the w/c = 0.45 pastes, were followed here. As the RQPA results for the capillaries can be referenced to 100 g of paste, the overall hydration For reporting the data shown in Table 1, some additional calculations/assumptions were made. Firstly, this table contains the FW determined from the TGA as detailed above. Secondly, and as an internal standard was used, an overall 'amorphous' amount was determined. The ACn values reported in Table 1 are the subtraction of the FW to the ACn contents. The value for ACn is mainly, but not only, C-S-H gel. Thirdly, as the capillaries cannot contain an internal standard (to avoid the filler effect and to carry out µCT of the pastes), here it is assumed that they contain the same fractions of FW and ACn as the analogous cylinder samples. Furthermore, the external standard method cannot be used for capillary samples as it is not possible to control the packing factor.
The results for the pastes hydrated with w/c = 0.65 at 20 • C for 50 days are given in Table 2 and the corresponding Rietveld plots are displayed in Figure 3. The same calculations/assumptions, described above for the w/c = 0.45 pastes, were followed here. As the RQPA results for the capillaries can be referenced to 100 g of paste, the overall hydration degree of the pastes can be calculated. From data in Tables 1 and 2, the overall hydration degrees of the φ = 1.0 mm pastes for w/c = 0.45 and w/c = 0.65 are 78.3 and 82.3 %, respectively.  Figure 4a,c. The determined free water contents were 25 and 14 wt%, respectively. Shrinkages are evident in the µCTs as there is air between the glass capillary walls and the binders. However, the study of this shrinkage and its evolution with hydration age was out of the scope of the present work.

Laboratory X-ray Computed Microtomographic Study
To illustrate the spatial resolution of the reconstructed tomographic data, Figure 5 displays two enlarged views. The plot profiles, shown in Figure 5b,d, allow us to visualize the three main components: (i) porosity (water and air) with grey-scale values ranging~0-55; (ii) hydrated components,~55-100/120; and (iii) unhydrated cement particles, >100/120. It can also be seen that relatively small pores, size~4 µm, are readily measured which highlights the good spatial resolution of the acquired data. From several measurements, the spatial resolution is estimated as~2 µm. It is noted that although the voxel size was 0.80 µm, the spatial resolution can be/is poorer.
A first data analysis can be carried out by inspecting the greyscale histograms (GSHs). The linear attenuation coefficient µ is mapped in the reconstructed tomograms, where the higher is the µ value, the whiter is the corresponding greyscale value. Figure 6 displays the GSHs for the four studied capillaries. For three out of the four capillaries, there are clear valleys between the hydration products and the unhydrated cement products. For selecting the greyscale thresholds in the absence of valleys between the contributions, the tangent-slope approach was used [48]. The grayscale threshold separating two components was defined as the point where a change of tangent-slope for both contributions occurred. The procedure, and the derived values, are given in Figure 6.
It is evident in Figure 6 that for a given capillary size, the increase in added water displaces the maximum of the GSH for the HP to less absorbing regions. This observation plus the absence of a large contribution of water in the pore regions, greyscale values lower than~60, indicates that most free water contribution are within the hydration product peak.   have intermediate absorption and hence, middle grey-scale values. A first qualitative analysis indicates that the pastes with w/c = 0.65, see Figure 4b,d have more porosity, as expected, than those with w/c = 0.45, Figure 4a,c. The determined free water contents were 25 and 14 wt%, respectively. Shrinkages are evident in the μCTs as there is air between the glass capillary walls and the binders. However, the study of this shrinkage and its evolution with hydration age was out of the scope of the present work. To illustrate the spatial resolution of the reconstructed tomographic data, Figure 5 displays two enlarged views. The plot profiles, shown in Figure 5b,d, allow us to visualize the three main components: (i) porosity (water and air) with grey-scale values ranging ~0-55; (ii) hydrated components, ~55-100/120; and (iii) unhydrated cement particles, >100/120. It can also be seen that relatively small pores, size ~4 μm, are readily measured The four recorded µCTs were segmented by global thresholding using Fiji software. The thresholds for the component assignment were derived as detailed above and it is acknowledged that the manual threshold choice may slightly affect the accuracy of the results of the segmentation. The errors related to this approach have been previously estimated as~2 vol% [49]. The segmentation results are listed in Table 3. Figure 7 displays the rendered volume of the segmented tomograms. As the porosity contribution was mainly air, Table 3 also reports the segmented values but renormalized just taking into account the hydrated products and the unhydrated cement fraction labelled as rHP and rUHP, respectively. As expected, the overall degree of hydration is larger for the pastes hydrated with larger w/c ratio, i.e., w/c = 0.65. Furthermore, for the same paste (i.e., the same w/c ratio) the results from the two capillary sizes are the same within the variability of the measurements, the differences were lower than 1.5 vol%. Porosity, see Figure 7, is mainly observed close to the glass walls due to chemical shrinkage. The segmented porosity within the interior of the capillaries, because of its tiny dimensions, can only be observed in the w/c = 0.65 pastes.
Materials 2021, 14, x FOR PEER REVIEW 1 which highlights the good spatial resolution of the acquired data. From several mea ments, the spatial resolution is estimated as ~2 μm. It is noted that although the voxe was 0.80 μm, the spatial resolution can be/is poorer. A first data analysis can be carried out by inspecting the greyscale histog (GSHs). The linear attenuation coefficient μ is mapped in the reconstructed tomog where the higher is the μ value, the whiter is the corresponding greyscale value. Fig  displays the GSHs for the four studied capillaries. For three out of the four capill there are clear valleys between the hydration products and the unhydrated cement ucts. For selecting the greyscale thresholds in the absence of valleys between the c butions, the tangent-slope approach was used [48]. The grayscale threshold separ two components was defined as the point where a change of tangent-slope for both tributions occurred. The procedure, and the derived values, are given in Figure 6.
It is evident in Figure 6 that for a given capillary size, the increase in added w displaces the maximum of the GSH for the HP to less absorbing regions. This observ plus the absence of a large contribution of water in the pore regions, greyscale v hydrated with larger w/c ratio, i.e., w/c = 0.65. Furthermore, for the same paste (i.e., the same w/c ratio) the results from the two capillary sizes are the same within the variability of the measurements, the differences were lower than 1.5 vol%. Porosity, see Figure 7, is mainly observed close to the glass walls due to chemical shrinkage. The segmented porosity within the interior of the capillaries, because of its tiny dimensions, can only be observed in the w/c = 0.65 pastes.

Discussion
In this work, accuracy in cement hydration is studied in three stages. Firstly, thermal analysis results were compared to those obtained from RQPA of the pastes casted within cylinders. Secondly, the RQPA results from the three powder samples hydrated with the same w/c ratio were compared. Finally, the results from the analyses of the μCT were compared to those obtained from the capillaries by LXRPD. All laboratory X-ray data sets (diffraction and microtomography) were taken in approximately the same regions of the capillaries. Note that this study was carried out at 50 days of hydration. This choice was

Discussion
In this work, accuracy in cement hydration is studied in three stages. Firstly, thermal analysis results were compared to those obtained from RQPA of the pastes casted within cylinders. Secondly, the RQPA results from the three powder samples hydrated with the same w/c ratio were compared. Finally, the results from the analyses of the µCT were compared to those obtained from the capillaries by LXRPD. All laboratory X-ray data sets (diffraction and microtomography) were taken in approximately the same regions of the capillaries. Note that this study was carried out at 50 days of hydration. This choice was larger than the standard maturing time, i.e., 28 days, for two reasons: (a) to ensure that longer hydration times do not lead to significant carbonation; and (b) that low unhydrated cement product contents could be measured to an acceptable accuracy level.

Comparison Involving the Thermal Analyses
TA was chosen because it is a technique readily available, widely used for analysing cement hydration, which allows measuring at least three relevant parameters: (i) portlandite content, (ii) chemically bound water, and (iii) calcium carbonate content [47]. It had already been used in 2004 to verify the findings from Rietveld quantitative phase analysis of pastes by comparing the portlandite contents determined from the two methods and also by backscattered electron images [28]. TA, in combination with RQPA [50], has been used for characterizing the hydration of many types of cements, see for instance: (i) limestone-Portland blends [51]; (ii) limestone-fly ash-Portland ternary blends [37]; and (iii) limestone-fly ash-blast furnace slag-Portland quaternary blends [52]. Furthermore, the amount of measured portlandite depends upon several factors including the mineralogy of the employed cement, the hydration temperature and the use of additions. For comparisons, the content should be referenced to the same basis, for instance 100 g of paste. In this context, this cement was employed in an independent study by the authors and the portlandite contents determined at 28 days for a paste hydrated at 20 • C with w/c = 0.50 by TA and RQPA were 12.2 and 13.0 wt%, respectively [49]. These data compare very well with the portlandite percentages given in Tables 1 and 2 and Figure 1 (ranging 10.7-12.4 wt%), taking into account that a small consumption of portlandite could take place at longer hydration ages to give AFm-type phases.

Thermal Analysis and Nominal Compositions
On the one hand and for the non-arrested samples, the total weight loss expected for the pastes corresponds to the added water plus the losses from the decomposition of calcite and bassanite. For the w/c = 0.45 paste, the measured value was 29.5%, see Figure 1, which agrees relatively well with the expected overall weight loss, 32.5%. For the w/c = 0.65 paste, the measured value, 39.2%, also agrees well with the expected one, 41.0%. The gentle grinding before the thermal analysis could be the responsible for the small difference,~2 wt%, which is very likely due to water loss at that stage. On the other hand, the TGA determined CC content,~5 wt% for the w/c = 0.65 paste, is larger than the expected value,~2.2 wt%. This minor carbonation is not negligible but it does not importantly influence the comparisons carried out next.

Thermal Analysis and Rietveld Quantitative Phase Analysis (RQPA) of the Pastes from the Cylinders
The portlandite contents allow comparing the results from these two analytical approaches. For the w/c = 0.45 paste, the CH content from TGA was 10.7 wt%, see Figure 1. The corresponding value for the paste within the cylinder, from RQPA, was 10.8 wt%. Furthermore, the agreement for the w/c = 0.65 pastes is also good, 10.9 and 10.0 from TGA and RQPA, respectively. The Rietlved value is slightly lower very likely due to the small carbonation of this sample. In fact, the CC content from RQPA was 5.5 wt%, which is higher than the value coming from the pristine cement,~2.2 wt% after dilution.
The successful comparison between the values from TGA and RQPA for the cylindercasted pastes indicates that the RQPA for these samples were accurate and they can be used as reference "ground truth" for the next comparison: RQPA results from pastes hydrated within tiny capillaries and the reference samples. It must be noted that the final goal is to combine data taken in the same regions of the same capillaries by LXRPD and µCT. The optimal size of the capillary is a trade-off between two opposite effects. On the one hand, the tiniest capillary results in lower overlapping in the powder patterns due to the narrower diffraction peak widths and better resolution in the µCT as the field of view is smaller. However, very narrow capillaries result in poor particle statistics and also it is more difficult to ensure the correct (nominal) w/c ratio [53,54] in the probed volume as microbleeding could develop. On the other hand, wider capillaries have better particle statistics theoretically enabling better accuracy, but if the capillaries are very wide, the LXRPD patterns as well as the µCTs will have poorer resolution, which could lead to larger errors. Because of these opposing effects, capillaries of two different sizes, φ = 0.5 and 1.0 mm were filled with exactly the same pastes.

Comparison of the RQPA Results or the Same Pastes in Different Capillary Sizes
Firstly, the results for the three patterns containing the w/c = 0.45 pastes will be discussed. The data reported in Table 1 show a very good agreement between the cylinder and the φ = 1.0 mm capillary analyses. For instance, the C 3 S contents are 4.3 and 4.2, for the cylinder and φ = 1.0 mm samples, respectively. This agreement is also evident in the formed phases, i.e., the CH contents are 10.8 and 11.6 wt%, respectively. Unfortunately, the agreement with the φ = 0.5 mm capillary results is not that good. The C 3 S and CH contents are 7.5 and 8.3 wt%, respectively. It seems clear that the degree of hydration for the unhydrated cement phases is lower for the w/c = 0.45 paste within the narrow capillary. We justify this because ensuring the nominal w/c ratio within tiny capillaries is very challenging, as previously reported [53,54]. Finally, it is worth noting that ettringite contents for the sample within the φ = 1.0 mm capillary and the hydration arrested paste were 11.2 and 10.8 wt%, respectively. This means that the hydration arresting step has not significantly decreased the crystalline ettringite content.
Secondly, the results for the three w/c = 0.65 patterns will be considered. Table 2 shows a very good agreement for the contents derived for the three pastes. For instance, the ettringite fractions were 10.2, 11.1 and 9.8 wt% for the pastes φ = 0.5 mm, φ = 1.0 mm and cylinder, respectively. In this case, with a relatively large water/cement ratio, the RQPA results for the sample within the φ = 0.5 mm is accurate as microbleeding/segregation seems less important.
It is noted that carbonation has not taken place in any of the two wide (φ = 1.0 mm) studied capillaries, see Tables 1 and 2. Altogether, and because low water/cement ratios are very common in field applications, it is concluded here that relatively thick capillaries are needed to have RQPA results as accurate as possible.

Comparison of the RQPA and µCT Results for the Pastes within the Same Capillaries
The mass densities of the crystalline phases are known and are given in Table 1. For the amorphous phases, we have used the approximation of the density being the same than that of C-S-H gel, which is variable at the microscale, i.e.,~2.0 g/cm 3 [55], and the latest value 2.11 g/cm 3 was used here [56]. Under this approximation, the results of the RQPA, given in Tables 1 and 2, can be transformed from wt% to vol%. The different percentages were grouped as unreacted cement products and hydrated products, including the free water, and they are given in Table 4. This table also reports the results from the segmentation of the µCT after renormalization in order not to take into account the porosity. As can be seen in Table 4, the agreement between the values obtained from diffraction and tomography, for a given w/c ratio, is relatively good. Firstly, and as expected, both approaches yield a larger degree of reaction to the pastes with a larger w/c ratio, i.e., lower UHP content. Secondly, the agreement between the results for the same paste within capillaries of two sizes are also good. However, it is noted that any experimental variation within the narrow capillaries, i.e., capillaries with a nominal diameter of 0.5 mm, will result in larger deviations. Thirdly, the relative large disagreement for the w/c = 0.65 paste within a 1.0 mm capillary is likely due to the small fraction of UHP, which slightly increase the errors. Finally, and from the results shown in Table 4, it is estimated that the segmented values in the µCT are accurate within~2 vol%.