A method for determining ammonium ion concentration from complex aqueous samples is presented in this work. It does not need to eliminate chemical interferences, mainly sodium and potassium, because an array of potentiometric sensors with intrinsic responses is used. The measurements taken from the array are processed with a multicomponent data treatment. This approach is already known as electronic tongue. Multivariable calibration was implemented with an artificial neural network, trained under the rules of the Bayesian regularization. The developed system has been applied to water samples from rivers and wastewaters with ammonium content in the range 1 × 10 -4 -5 × 10 -2 mol L -1 . Results are similar to those obtained with other reference methods.
|Publication status||Published - 15 Dec 2003|
- Bayesian regularization
- Electronic tongue
- Ion selective electrode array
- Neural networks
- Sensor array