Hazardous aromatic VOC quantification through spectroscopic analysis and intelligent modeling to assess drinking water quality

• Accurate non-linear model to estimate concentrations of VOCs in mixtures
• Inexpensive analytical approach for reliable data gathering in the micromole range
• The compounds analyzed have been characterized as hazardous water pollutants.
• Neural networks have proven to be a suitable tool for assessing water safety.
• Wide applicability span, from industries to water treatment facilities

Concentrations within the micromole range have been accurately determined for mixtures of two common aromatic volatile organic compounds (VOCs) that are hazardous for both the environment and human health: styrene and 1,2,4-trimethylbenzene. This task has been developed by combining 165 UV/Vis absorption measurements and intelligent modeling based on artificial neural networks (ANNs), and is especially relevant as these compounds have been located in sources of drinking water within the concentration range covered. The performance of the designed non-linear models proves the existence of a relation between spectroscopic information and compound concentration, as well as the ability to obtain reliable quantifying chemometric tools for environmental control applications. The average mean absolute errors for three internal validations were near 0.5 μM and 1.5 μM for styrene and 1,2,4-trimethylbenzene concentration estimation, respectively.