Undecorated GFET for determinations of heavy metal ions aided by machine learning algorithms

Abstract

The Graphene field effect transistor (GFET) has become one of the most popular biochemical-sensing platforms. The sensitivity of traditional GFET sensor usually depends on functionalization of aptamers. Here, we reported a machine learning algorithms assisted heavy metal ions detecting method by using pristine GFET. In this work, GFETs are prepared by screen printing technology, which has the advantages of total solution processes and low costs. The graphene ink (G-ink) used in screen printing procedures of GFET is synthesized by combining liquid exfoliation with solvent exchange method. The FTIR and XPS show there are oxygen-containing groups on the graphene channel, which have different affinities for heavy metal ions and can be used to determine ions. Seven algorithms are used to establish and train prediction models by using the data of transfer curves as inputs and ions’ types and concentrations as labels, the results indicate that the accuracy of AdaBoost (0.91) is the highest, which demonstrate this machine learning assisted GFET sensing platform can be used to determine metal ions’ types and concentrations at the same time. Moreover, the complicated processes for bio-functionalizing GFET and data analysis are omitted, which is benefit to simplify the fabricating procedures, we believe it will be an attractive strategy to promote the development of GFET.

Keywords

• Graphene field effect transistor
• Machine learning algorithms
• Graphene ink
• Screen printing
• Heavy metal ions