Quantifying BTEX in aqueous solutions with potentially interfering hydrocarbons using a partially selective sensor array

Partially selective gold nanoparticle sensors have the sensitivity and selectivity to discriminate and quantify benzene, toluene, ethylbenzene, p-xylene and naphthalene (BTEXN) at concentrations relevant to the US Environmental Protection Agency. In this paper we demonstrate that gold nanoparticle chemiresistors can do so in the presence of 16 other hydrocarbons and that they did not reduce the discriminating power of the array. A two-level full factorial designed experiment was performed on unary, binary, ternary, quaternary, quinary combinations of BTEXN analytes with and without the possibly interfering hydrocarbons. The nominal component concentration of the mixtures was 100 μg L−1, equivalent to approximately 100 parts per billion (ppb). Concentrations predicted with the random forests method had an average root mean square error of 10–20% of the component concentrations. This level of accuracy was achieved regardless of whether or not the 16 possibly interfering hydrocarbons were present. This work shows that the sensitivity and selectivity of gold nanoparticles chemiresistor sensors towards BTEXN analytes are not unduly affected by the other hydrocarbons that are expected to be present at a petroleum remediation site.