Applying grid nanoindentation and maximum likelihood estimation for N-A-S-H gel in geopolymer paste: Investigation and discussion

Static nanoindentation and Maximum Likelihood Estimation (MLE) were applied for the nano/micromechanical properties investigation of alkali-activated fly ash (AAFA) in this study. Some critical issues of statistical nanoindentation were fully discussed, including properties of pure gel phase, influence of bin size when using least-square estimation (LSE), and suitable number of components for deconvolution. Results indicate that the model estimated by MLE method can effectively reflect the micromechanical distribution of AAFA. The number of components needed to separate sodium aluminosilicate hydrate (N-A-S-H) gels is sometimes more than the normally used 3 or 4, depending on the sample and testing factors. The gel phase does not always display as a prominent peak in the histogram and is easy to be mixed with other adjacent peaks even if the bin size is small, indicating the challenges of employing the LSE method to investigate the gel phase in highly heterogeneous materials, such as geopolymer. •When using MLE method, separating process of stable N-A-S-H gel from mixed phases is observed.•Bayesian Information Criterion is not necessary for the optimum value when just focusing on N-A-S-H phase.•The number of components to obtain pure N-A-S-H phase is more than the typically used value of 4.•GMM model for AAFA obtained by MLE can match histogram well when bin size is small (0.4 GPa for hardness).•Average modulus and average hardness of N-A-S-H with various alkali concentration change in small ranges.