Site Overlay

Research Izzati

Comparative Study Of Electrochemical Transducer Fabrication Methods With Glucose Oxidase As Recognition Layer

Non-invasive glucose monitoring approach for diabetes patients is a goal under ongoing process. The extremely small range of glucose in bodily fluids other than blood remains as a barrier in the commercialization of an accurate and reliable non-invasive biosensor. However, advancements in nanotechnology and artificial intelligence have revolutionized the scope of research in which could draw us closer to realize the non-invasive glucose sensing. In this study, combination of reduced graphene oxide (rGO) with a conductive polymer, poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate)(PEDOT:PSS) were modified on screen-printed carbon electrode (SPCE) as transducer. The rGO-PEDOT:PSS modified electrodes were fabricated with four different methods where the fabrication parameters such as reduction cycles, sequence and glucose oxidase immobilization techniques were varied. The aim is to elucidate how these fabrication parameters can affect the kinetics rate constant (k⁰) and effective surface area (Aeff) of rGO-PEDOT:PSS materials in which can be determined from cyclic voltammetry (CV). This study also apply machine learning to model the input and output data of the rGO-PEDOT:PSS biosensor based on support vector machine (SVM). The electrochemical reversibility and mass transport properties was analyzed via CV in 0.1 M potassium ferricyanide (K3(Fe[CN]6)). The results show rGO-PEDOT:PSS fabricated with method A, B, C and D are quasi-reversible. In terms of k⁰, each fabrication methods generated different trend of k⁰ value with increasing reduction cycles. Overall, the range of k⁰ for rGO-PEDOT:PSS-modified SPCE are from 0.52 x10-5 to 4 x10-5 cm/s. In terms of mass transport properties, the results show that the electron transfer process on the rGO-PEDOT:PSS-modified electrode following the diffusion-controlled process. In addition, the highest Aeffvalue with respect to fabrication method were obtained from different reduction cycle. Fabrication method A gave the highest Aeff when the composite was reduced for 5 reduction cycles (16.41 cm2). For methods B and D, the highest Aeff obtained was for 30 reduction cycles (17.48 cm2 for method B and 6.43 cm2 for method D) while for method C, the highest Aeffvalue was obtained for 15 reduction cycles (21.41 cm2). For SVM analysis, data from CV were used to construct a prediction model to predict the different fabrication methods. Linear and non-linear kernels were compared, and the best performance was showed by radial basis function (RBF) kernel with a perfect accuracy of 100%. However, the interaction of input and output variables showed inconsistency between runs. In general, this study provides new insights on kinetics and mass transport behavior and the use of machine learning on CV data.

Who is Izzati? Click here.

Congratulations #TeamDrAmani for WINNING the MIT SOLVE Community Award.