Machine Learning Field Calibration of a Low-Cost Carbon Nanotube Sensor Array Utilizing Near-Road Ambient Ozone Measurements

Real-time monitoring of environmental pollutants, such as ozone and particulate matter, using low-cost sensors, has become increasingly important in various fields, including wearable devices, public health and occupational safety. Nanotechnology-based sensors offer significant advantages over commercial metal oxide semiconductor and electrochemical sensors, including low-temperature operation, low power consumption, enhanced sensitivity, and chemical functionalization of nanomaterials for selective gas detection. In this study, we integrate carbon nanotube (CNT) sensors based on thin films of three novel functionalized nanomaterials into a compact five-element sensor array and conduct ambient ozone measurements at a near-road location in downtown Atlanta, an area characterized by high vehicular emissions. Three distinct ozone-sensitive functionalized CNTs are investigated, namely, (a) Pt-CNT, comprising pristine CNTs decorated with platinum nanoparticles (~2 nm in diameter); (b) PBD-CNT, a composite of pristine CNTs and polybutadiene (PBD); and (c) ODA-CNT, comprising pristine CNTs covalently modified with octadecylamine (ODA) functional groups. The ambient measurements are performed during summer when the ozone levels are high due to abundant sunshine and high concentrations of precursor pollutants emitted by vehicular traffic. The field data comprising impedance magnitude and phase responses of the five sensors are analyzed using machine learning regression employing ensembles of gradient-boosted and random forest decision trees with blocked cross-validation. The sensor array is collocated with a precision ozone monitor, which provides the reference ambient ozone concentrations to calibrate the CNT sensor elements. Prediction on an unseen data block (not used in training and cross-validation) reveals that the array discriminates ambient ozone levels above 20 ppb with RMSE < 4 ppb, and R2 = 0.9.

Authors

Institutions

Publication Details

Journal
Sensors
Published
2026-11-19
DOI
https://doi.org/10.3390/s26227301
Primary Topic
Air Quality Monitoring and Forecasting
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Machine Learning Field Calibration of a Low-Cost Carbon Nanotube Sensor Array Utilizing Near-Road Ambient Ozone Measurements

Elena B. Bekyarova, Gautam Naishadham, Krishna Naishadham
Sensors
Air Quality Monitoring and Forecasting
article

Machine Learning Field Calibration of a Low-Cost Carbon Nanotube Sensor Array Utilizing Near-Road Ambient Ozone Measurements

Elena B. Bekyarova, Gautam Naishadham, Krishna Naishadham
article en

Abstract

Real-time monitoring of environmental pollutants, such as ozone and particulate matter, using low-cost sensors, has become increasingly important in various fields, including wearable devices, public health and occupational safety. Nanotechnology-based sensors offer significant advantages over commercial metal oxide semiconductor and electrochemical sensors, including low-temperature operation, low power consumption, enhanced sensitivity, and chemical functionalization of nanomaterials for selective gas detection. In this study, we integrate carbon nanotube (CNT) sensors based on thin films of three novel functionalized nanomaterials into a compact five-element sensor array and conduct ambient ozone measurements at a near-road location in downtown Atlanta, an area characterized by high vehicular emissions. Three distinct ozone-sensitive functionalized CNTs are investigated, namely, (a) Pt-CNT, comprising pristine CNTs decorated with platinum nanoparticles (~2 nm in diameter); (b) PBD-CNT, a composite of pristine CNTs and polybutadiene (PBD); and (c) ODA-CNT, comprising pristine CNTs covalently modified with octadecylamine (ODA) functional groups. The ambient measurements are performed during summer when the ozone levels are high due to abundant sunshine and high concentrations of precursor pollutants emitted by vehicular traffic. The field data comprising impedance magnitude and phase responses of the five sensors are analyzed using machine learning regression employing ensembles of gradient-boosted and random forest decision trees with blocked cross-validation. The sensor array is collocated with a precision ozone monitor, which provides the reference ambient ozone concentrations to calibrate the CNT sensor elements. Prediction on an unseen data block (not used in training and cross-validation) reveals that the array discriminates ambient ozone levels above 20 ppb with RMSE < 4 ppb, and R2 = 0.9.

SensorsVol. 26(22)
Carbon Solutions (United States) (US)
Air Quality Monitoring and Forecasting
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon