Bioelectronic tongue and multivariate analysis: A next step in BOD measurements

Seven biosensors based on different semi-specific and universal microorganisms were constructed for biochemical oxygen demand (BOD) measurements in various synthetic industrial wastewaters. All biosensors were calibrated using OECD synthetic wastewater and the resulting calibration curves were used in the calculations of the sensor-BOD values for all biosensors. In addition, the output signals of all biosensors were analyzed as a bioelectronic tongue and comprehensive multivariate data analysis was applied to extract qualitative and quantitative information from the samples. In the case of individual biosensor measurements, most accurate result was gained when semi-specific biosensor was applied to analyze sample specific to that biosensor. Universal biosensors or biosensors semi-specific to other samples underestimated the BOD7 of the sample 10―25%. PLS regression method was used for the multivariate calibration of the biosensor array. The calculated sensor-BOD values differed from BOD7 less than 5.6% in all types of samples. By applying PCA and using three first principal components, giving 99.66% of variation, it was possible to differentiate samples by their compositions.