Development of Numerical Methods for Signal Smoothing and Noise Modeling in Single Wire-Based Electrochemical Biosensors

作者：Goncalves W D; Lanfredi A J C; Crespilho F N^*

来源：Journal of Physical Chemistry C, 2011, 115(32): 16172-16179.

DOI：10.1021/jp204180e

Recently, many efforts have been made to build biosensors and bioelectronic devices on the nanometer scale, where individual 1D structures are applied as the working electrode to obtain a low current signal, typically on the order of femtoamperes (fA) and picoamperes (pA). However, considering the most modem equipment for current measurements (potentiostats and multimeters) and the scale of the magnitude of the measures, several types of noise affect and compromise the interpretation of the results. Thus, the objective is to identify types of noise and propose different methods for smoothing signals obtained from bioelectrochemistry nanostructured devices. As a case study, the sign of a biochip comprising an indium tin oxide nanowire (ITO-NW) modified with a glucose oxidase (GOx) enzyme was evaluated. The biochip was used as a nanobiosensor for glucose, which revealed currents on the order of pA from the biocatalytic process. New computer software based on numerical methods was developed for this purpose. This program was able to identify types of noise present in measurements and propose several methods of smoothing the signals, such as the Moving-Average (MA), Savitzky-Golay (SG) and Fast Fourier Transform (FFT). The statistical evaluation of the signals was calculated by the power spectral density and the distribution of the probability density function. As a result, the biodevice presented picoampere currents but with noise of the same order of magnitude. The noise was identified as being mostly composed of the effect of thermal noise, with magnitudes of the same order as the acquired signal and also contributions from shot noise, but this contribution was much less significant. Having identified the source of the noise, numerical methods were used to accomplish the smoothing of the data. MA and SG attenuated signals with higher frequencies, and the FFT method completely eliminated the high-frequency signals. Thus, the results showed that the signals of interest were concentrated in the low-frequency range, and through statistical analysis, the removal of white noise and high-frequency noises was validated.

• 出版日期2011-8-18