Planar waveguide enzyme sensors coated with nanocomposite membranes for water pollution monitoring.

Attenuated total reflection (ATR) of light in a SiO[2]/Si[3]N[4] planar waveguide was successfully exploited with a view to developing a highly sensitive enzyme sensor for monitoring typically agricultural and industrial water pollutants. The Si[3]N[4] surface of the waveguide sensing window was coated with nanocomposite polyelectrolyte selfassembled (PESA) membranes containing cyclotetrachromotropylene (CTCT) indicator and enzymes. The reaction of three enzymes namely urease, acetylcholine esterase or butyrylcholine esterase was accompanied by changes in pH, varying the absorption coefficient of the CTCT indicator.An experimental set-up of a single channel enzyme sensor was constructed to investigate the capability of the planar waveguide to register small changes in the light absorption of the PESA membrane. It was found that due to the multiple reflections phenomena, the sensitivity of a planar waveguide was higher than that for traditional absorption spectroscopy and previously reported waveguiding structures by at least three orders of magnitude. The respective enzyme reactions as well as their inhibition with toxic agents such as cadmium and lead ions were studied by in-situ monitoring the changes in output intensity of the planar waveguide. The results were remarkable, since traces of the above pollutants were detected with concentrations as low as to 1 ppb.The work was also extended to design and implement a laboratory scale enzyme sensor array. A multi-channel reaction cell and light guiding system were designed for this purpose. The analysis of the experimental results demonstrated that the multichannel enzyme sensor was able to produce adequate responses to the presence of different pollutants of industrial (cadmium, lead and nickel) and agricultural (imidacloprid, paraoxon and DVDP) origin, in the concentration range from 1 ppb to 1000 ppb. The distinct pattern of sensor responses was analysed by the implementation of artificial neural network algorithm. Despite a rather small amount of experimental data, the trained neural networks were able to classify and quantify the pollutants with an acceptable average error of 6.24 %.