MATHEMATICAL MODEL OF BIOSENSOR WITH MULTILAYER CHARGED MEMBRANES N.Rossokhata, V.Rossokhaty Kiev National University, P.O.Box 493, 253222 Kiev - 222, Ukraine, E-mail: nross@nv.kiev.ua Biosensors combining the basic principals of microelectronics, chemistry and biotechnology attract considerable attention in such fields as medicine, environmental monitoring, pharmaceutics and food quality control. In this work we develop a mathematical model of enzyme potentiometric biosensor for glucose determination with charged multilayer membrane. Biosensor consists of two parts: solid-state transducer linearly transforming the concentration of protons near membrane-transducer interface into electrical signal and 3-layer membrane, with each layer involving uniformly distributed both enzyme and charged encapsulated molecules of given concentrations with various combinations. Test solution is considered to be infinite source of five types of species, namely, glucose, oxygen, protons, buffer and its conjugate base. Mathematically the model is reduced to a transmission problem with two interface points for a system of five one-dimensional nonlinear parabolic partial differential equations for concentrations and Poisson equation for charge. In order to find the numerical solution, we use the techniques known from semiconductor modeling literature and receive an explicit difference scheme. Because of concentration profiles change most rapidly at the beginning of the intervals corresponding to the enzyme layer(s), we use irregular space grid, which nodes are chosen automatically in dependence on physical data of the problem. To find charge and concentration distributions, we get uncoupled systems of linear algebraic equations with threediagonal matrix with diagonal prevalence. To solve them, we use effective iterative algorithms. To carry out a numerical experiment, the developed model of the biosensor was adopted to the experimental multilayer membrane system with the inner layer being enzyme one and the outer layer containing the uniformly distributed negative charge. The results of computer simulation give opportunity to optimize performances of enzyme potentiometric biosensors by varying combinations of physical and biochemical parameters of the membrane layers.