Inference of The Phase-To-Mechanical Property Link Via Coupled X-Ray Spectrometry and Indentation Analysis: Application to Cement-Based Materials
Chemo-mechanical characterization of cement-based materials is presented, which combines the classical grid indentation technique with elemental mapping by scanning electron microscopy energy dispersive X-ray spectrometry (SEM-EDS). The method is able to successfully isolate the calcium-silica-hydrate gel at the indentation scale from its mixtures with other products of cement hydration and anhydrous phases; thus providing a convenient means to link mechanical response to the calcium-to-silicon ratio quanti?ed independently via X-ray wavelength dispersive spectroscopy.
The grid indentation technique allows access to the characterization of cementitious microstructures at different length and time scales.
Materials: The cement-based system selected for this experimental study is based on the American Petroleum Institute (API) oil-well cement Class G (Dyckerhoff, Germany), and silica ?our.
Sample preparation: Samples for instrumented indentation, SEM imaging and X-ray spectrometry were prepared with a semi-automatic polishing procedure.
X-ray powder diffraction: X-ray diffraction analysis was carried out on the representative powder samples of the investigated systems.
Wavelength dispersive X-ray spectrometry (WDS): A quantitative analysis of the chemical composition was carried out with the JEOL JXA-8200 Super probe electron probe micro-analyzer (EPMA). The polished specimens were analyzed by WDS under the following analytical conditions: accelerating voltage 15 keV, beam current 10 nA, beam diameter 1 ?m and 10 s counting time per element with background correction applied at each spot.
Elemental mapping with energy dispersive X-ray spectrometry (EDS): Electron micrographs and EDS elemental maps were acquired with the JEOL 5910 general purpose scanning electron microscope equipped with Bruker EDX system for elemental analysis. The backscattered images of the region of interest (ROI) were obtained at a beam voltage of 15 kV and working distance of 10 mm, while 12 kV was used for the elemental maps.
Instrumented grid indentation: Instrumented indentation tests were carried out on the CSM Instruments standard nanoindentation tester equipped with an environmental control enclosure (18 ± 1 °C, 21 ± 2%RH) and a damping system. A single indentation grid comprises of N = 624 measurement points minimum, arranged in the rectangular grid and spaced regularly around 10 ?m apart in each direction. The test operated to the selected maximum indentation depth produced the residual imprint of approximately 1.5 ?m. Such spacing ensures that successive measurements are independent of one another.
Statistical model: The design of these coupled experiments allows mechanical data obtained in the instrumented indentation to be directly associated with the local phase chemistry, assessed qualitatively or quantitatively at each grid point location from SEM-EDS elemental maps.
Image processing: automated detection of the indentation points: The proposed experimental approach combines experimental results, indentation properties and in-situ chemistry, which are obtained by two independent techniques and instruments run in a sequence on the selected ROI. This experimental sequence includes transfer and repositioning of the sample in between both stages of measurements. In order to assure one-to-one spatial correlation between indentation properties and chemistry of the indented volume and to accelerate the entire process, an image processing algorithm was developed.
Technique limitations and perspectives: The proposed experimental approach represents an attractive way to characterize and understand cement-based materials, with special focus on linking the indentation properties to the in-situ local phase chemistry. Such a method combines the standard grid indentation with the electron probe method (SEM-EDS), each of which has its own limitations. One of these limitations relates to the size of the material volume sensed with both techniques. In general, it is assumed that the characteristic length of the material volume sensed in the instrumented indentation is roughly 3-to-5 times the maximum penetration hmax. Therefore, for the given experimental conditions this translates to characteristic length l ? 1 ?m, which is generally, the resolution (de?ned as the characteristic size of the beam-solid interaction volume including 90% to 95% of emitted X-rays for all measured elements) of the SEM-EDS operating at standard conditions. While the penetration depth may be easily reduced to assess indentation properties for much smaller material volumes, this is not necessarily the case for the SEM-EDS. Therefore, the spatial resolution limit of the coupled IND-EDS approach is governed by the resolution of the latter technique.
The micromechanics scale-separability condition is another point of attention for indentation testing on cement-based materials. This condition requires the chosen indentation depth to be such that the interaction volume is representative of the material, and the other microstructural length scales and related components do not interfere signi?cantly with the indentation response. Under such conditions, the indentation data provide access to the intrinsic phase properties. Given the high complexity of the microstructure of this class of materials and the large number of grid points, it is very likely that some fraction of the measurements might not obey this condition, e.g. ? ne grained C4AFand grain boundaries. Therefore, the indentation hardness and modulus measured at these locations do not stand for representative phase properties, but rather represent the local ? actuations re?ecting particular microstructural order and composition at a given spatial location. Depending on their number as well as their nature, these events may contribute to the overall variance in indentation properties measured for a phase of interest, or establish an independent statistically signi?cant family described by its mean, variance and probabilistic weight.
Finally, the current approach works with the elemental maps representing the auto-scaled intensities of major elements, which are not corrected for background. In most of the practical situations where the element (or phase) spatial distribution is investigated, such compositional mapping is of primary choice for data presentation and analysis provided by the commercial EDS software. While the use of such auto-scaled maps may be expedient as demonstrated in this study, the loss of true intensities clearly eliminates any possibility of performing operations which require true counts, e.g. quantitative analysis. It is apparent, in the methods framework laid out in this manuscript, the chemo-mechanical characterization may be extended beyond qualitative compositional mapping. Such extensions may be implemented by including quantitative maps, or k-ratio maps. This advancement will provide a direct link from indentation properties to elemental concentration at a given location, without the need to apply other methods (e.g. WDS) in parallel.
Summary and conclusions: It is shown that the combination of the widely accepted grid indentation technique together with EDS elemental mapping provides a suitable base of observables to identify microstructural components by their unique chemo-mechanical footprint. The insight into the chemistry of the indentation volume, coined here as a “chemical eye”, appears to be central to the problem of separation of indentation records originating from hydration products with different stoichiometry, residual clinker, or mineral ?ller. In this regard, the indentations predominantly due to the nanocomposite C-S-H solid could be successfully isolated from the tests in?uenced by other phases, as demonstrated here using a selected engineering example. The current approach uses the WDS technique to quantify the stoichiometry of the C-S-H solid. However, it appears possible that a future extension to the proposed technique will be able to include quantitative EDS.