Giulia S. da Silva,a,b,1 Luiza P. de Oliveira,a,1 Gabriel F. Costa,a,b,1 Gabriela F. Giordano,a,b,1 Caroline Y. N. Nicoliche,a,1 Alexandre A. da Silva,a Latif U. Khan,a Gabriela H. da Silva,a,c Angelo L. Gobbi,a José V. Silveira,d Antonio G. Souza Filho,e Gabriel R. Schleder,a,f Adalberto Fazzio,a,f Diego S. T. Martinez,a,c and Renato S. Limaa,b,*
a Brazilian Nanotechnology National Laboratory, Brazilian Center for Research in Energy and Materials, Campinas, São Paulo 13083-970, Brazil
b Institute of Chemistry, University of Campinas, Campinas, São Paulo 13083-970, Brazil
c Center for Nuclear Energy in Agriculture, University of São Paulo, Piracicaba, São Paulo 13416-000, Brazil
d Department of Computer Science, Federal University of Ceará, Sobral, Ceará 62010-560, Brazil
e Department of Physics, Federal University of Ceará, Fortaleza, Ceará 60455-900, Brazil
f Center for Natural and Human Sciences, Federal University of ABC, Santo André, São Paulo 09210-580, Brazil
ABSTRACT: While multidimensional sensors are powerful platforms towards multitarget analyses, the successive synthesis/fabrication of multiple probes and measurements to each one of these units still damage the device miniaturization, scalability, cost, consumption of samples, operational simplicity, precision, and analysis time. Herein, we describe an electrochemical sensing array that affords the discrimination of metal ions from a single ready-to-use probe and experiment. The sensing probe consisted of commercial stainless-steel capillaries, which defined a microfluidic circuit and acted as electric double-layer parallel capacitors into devices prototyped by a fast, cleanroom-free, and green technique. The probes assured differential responses due to heterogeneous interactions with samples and multichannel capacitance outputs. In addition, we address an effective strategy to further improve the repeatability and recognition ability of the sensor by using oxidized multi-walled carbon nanotubes as a single bulk probe. The nanotubes provided differential electrostatic adsorptions of ions, then increasing the variance of the capacitance responses. The approach was successfully applied in the identification of samples of mineral water, lake, and petroleum according to the presence of metal ions. The first two applications are important to human health and environment, whereas the third one brings new possibilities to oil industry by allowing the optimum dosage of fouling inhibitors. Using supervised machine learning tasks, the sensor assured reproducible, sensitive, and accurate classification and quantification of dozens of lake samples spiked with multiple heavy metals in accordance with the safe limits for Ni2+, Al3+, and Cu2+. Remarkably, simultaneous quantification of the individual concentration of these metal ions added in the mixings was also possible from universal impedimetric assays by treating the capacitance data through multi-output regression. The sensor will be of paramount significance for advanced pattern discriminations from a single ordinary probe and measurement in a direct mode using scalable chips. Furthermore, the coupling of bulk nanoprobes with electrochemical sensors may be extended to other multitargets and platforms towards an enhanced sensing performance.
This article was accepted for publication in Sensors and Actuators B: Chemical (Elsevier, DOI: 10.1016/j.snb.2019.127482). This project has been conducted in collaboration with Federal University of Ceará (UFC).