A novel chemical image sensor consisting of integrated microsensor array chips and pattern recognition

A novel chemical image sensor developed for liquid component analysis is proposed in this paper; using it, pH values ranging from 1 to 12 and six kinds of metal ion, namely Cu2+, Fe2+, Fe3+, Ca2+, Zn2+, and Mg2+, can be detected qualitatively and quantitatively. The sensor applies the principles of optical chemistry and microfabrication technology to detect the ion concentrations in the solution, and has the advantages of high sensitivity, reduced contamination, a lower sample volume required, and the capability of detecting several indices at one time. Moreover, three multivariate data analysis methods are suggested in the paper for treating the raw data acquired from the microbeads, and predicting the results. The study demonstrates that the principal component analysis is capable of classifying six kinds of cation with success. Both partial least-squares regression (PLS) and artificial neural networks (ANN) can be used to compute the pH values quantitatively; furthermore, the PLS method has the advantage of requiring fewer iteration steps than the ANN approach.