Artificial Intelligence Used in High-Performance Graphene Sensor Detects Bacteria and Toxins in Drinking Water

Half of the world's population still does not have access to safe drinking water, according to the World Health Organization (WHO). In 2019, unsafe drinking water accounted for at least 1.4 million deaths. Aside from bacterial contamination, toxic heavy metals also pose serious health risks. For this reason, there is a need to develop sensors that can accurately and quickly identify contaminants and help prevent waterborne diseases.

Challenges in Water Monitoring

In the U.S., the Food and Drug Administration regulates the lead concentration in bottled water, which must not be more than five parts per billion. Detecting parts-per-billion or parts-per-million concentrations of bacteria, heavy metals, and other toxins can only be done by analyzing laboratory water samples.

There are sensors designed to measure contaminants in water, but they have reliability issues and demonstrate an inability to detect faulty devices. The sensors are also expected to be extraordinarily sensitive in catching minute concentrations of toxins that can harm human health.

Screening for Water Contamination

Researchers from Argonne National Laboratory, the Pritzker School of Molecular Engineering at the University of Chicago, and the University of Wisconsin-Milwaukee developed sensors that detect lead, mercury, and E.coli in flowing tap water to address this challenge. Led by Professor Junhong Chen, the team aims to create an innovation that can help defend public health by providing early warning for water contamination.

The sensors contain a one-nanometer-thick layer of carbon and oxygen atoms in the form of graphene, which is used to coat a silicon substrate. The semiconducting graphene oxide sheet is the medium between the source and drain electrodes in a field-effect tran (FET). Afterward, the graphene sheets are deposited on a silicon wafer and imprinted with gold electrodes. This is followed by an insulating layer of aluminum oxide separating the gate electrode from the semiconducting channel.

Chemical and biological molecules were attached to the graphene surface and were allowed to bind with the desired targets: E.coli bacteria, lead, and mercury. When contaminants, even in their tiniest amount, get attached to the graphene surface, the material will change its conductivity with magnitude correlated to toxin concentration.

Each sensor is customized to detect lead, mercury, or E.coli. and measure them in parts-per-million concentrations in flowing water. The contaminants are differentiated from each other using machine-learning algorithms. Like any other FET, the reliable sensor responds very fast, so the results are immediately seen. It is also potentially low cost because it uses cost-effective and scalable technology already utilized in laptops, computers, and cell phones.

Creating sensors with reliable, consistent performance has been a major challenge to the experts because the insulating aluminum oxide layer can have flaws that trap charges and affect its performance. To address this concern, Chen and his colleagues used a non-intrusive process in detecting defective devices. While the sensors are immersed in water, they are analyzed by the researchers using impedance spectroscopy, and their current is measured through the devices. This allows the scientists to identify structural defects present in the aluminum oxide.

The research team has also devised a pathway to mass manufacture the sensors, which will cost only $1 each. This will allow people to test their drinking water at home for the presence of any toxins.