Electronic Noses Sniff Out Diseases, WildFires, and Pestered Plants

Electronic noses or "e-noses" powered by gas sensors play a critical role in various applications to decode chemical and environmental information. This technology has made significant progress in recent years, with advancements in both hardware and algorithms. As a result, they have a wide range of potential use cases, from environmental monitoring to medical diagnosis. However, despite many advances in this technology, the stability and reliability of the sensors remain an issue.

 

Perhaps the most famous example of a nose on a chip is Intel's Loihi neuromorphic test chip. Image courtesy of Intel

 

This article discusses recent breakthroughs in e-noses and shows how these feats innovatively tackle the challenges of gas sensing.

 

What Are Electronic Noses?

An electronic nose mimics the olfactory systems of humans and animals. Our sense of smell relies on olfactory receptor neurons (ORNs). Odorant molecules interact with specific receptors on the surface of ORNs, triggering an electrochemical potential and eventually leading to a sensation. E-noses consist of an array of sensors that detect complex odorant molecules.

Because odors are complex mixtures of different chemicals, each having a unique combination of compounds, gas sensors try to capture these compounds. And by analyzing the pattern of chemicals, e-noses can identify or classify the smell.

 

The operation of a biological olfactory system vs. an electronic nose in odor category identification. Image courtesy of Intelligent Computing

 

Recently, there have been many efforts to improve these sensors' sensitivity, selectivity, and stability. Currently, e-noses face limitations, such as sensor drift that leads to inaccurate readings over time, and limited selectivity, making it difficult to differentiate between similar odors. These sensors also struggle with chemicals of low volatility.

Artificial intelligence (AI) and machine learning (ML) can help improve the performance of these devices to some extent. Researchers have explored deep learning algorithms to analyze the pattern of sensor responses and identify specific compounds. However, some types of odors may be underrepresented in their learning dataset, which could lead to bias and reduce accuracy.

 

E-nose With Microbial Nanowires for Health Monitoring

Scientists at the University of Massachusetts Amherst recently created a nanowire that can be economically grown with bacteria and optimized to identify various chemicals. An array of thousands of nanowires can be layered on a compact wearable sensor, making it useful in diagnosing medical conditions such as asthma and kidney disease. These sensors are far more environmentally friendly than conventional carbon nanowires since the wires are organic and biodegradable.

The researchers began with a bacterium they used during their previous research, Geobacter Sulfurreducens, to create a biofilm capable of producing electricity from sweets. While these bacteria can grow electrically conductive nanowires, they require a specific environment to grow.

Scientists spliced the G. Sulfurreducens to remove the DNA of Escherichia Coli, a widespread bacterium. They then modified it to include DLESFL, an extremely ammonia-sensitive peptide. Ammonia is present in the breath of patients with kidney diseases. The team harvested the ammonia-sensitive nanowires and built a sensor to detect this chemical in patients. 

 

Nanowires that can capture smell are built into a sensor. Image courtesy of Biosensors and Bioelectronics

 

The team reported that the genetically modified nanowire is 100 times more responsive to ammonia than it previously was. Toshiyuki Ueki, the study's co-lead author and research professor in microbiology at UMass Amherst, suggests that it is possible to design unique peptides and incorporate hundreds of sensitive nanowires to capture many odors and identify different health conditions beyond kidney diseases.

 

Detecting Wildfires With E-noses

In addition to sniffing out diseases in humans, electronic noses can also detect wildfires within the first hour of smoldering, drastically increasing the chances of extinction. The California Department of Forestry and Fire Protection (CAL FIRE) plans to pilot the wildlife detection technology of environmental startup Dryad Networks, which aims to identify wildfires faster and reduce the destruction associated with such events. The new sensors can detect wildfires in minutes while it is still possible for firefighters to put them out.

Beyond wildfire detection, Dryad sensors can monitor microclimate, temperature, humidity, and air pressure. It includes a precise gas sensing mode for ultra-low power air quality sensing.

 

Use of Dryad sensors during the smoldering phase (within the first 60 minutes) of a wildfire. Image courtesy of Dryad Networks

 

Combined with AI, Dryad's Silvanet Wildfire Sensor can detect hydrogen, carbon monoxide, and carbon dioxide, allowing one to distinguish between fuels that ignite them. The device is solar-powered and uses supercapacitors instead of Li-ion batteries. In addition, LoRaWAN communication capability enables Dryad's sensors to run for 10–15 years without needing to replace their capacitors.

 

Sniffing Whitefly Infestations of Tomato Plants

Electronic noses are also being put to use in agriculture research. Researchers of the Agricultural Research Service (ARS) and universities developed an e-nose to sniff out whitefly infestations of tomato plants, which works by identifying volatile organic compounds (VOCs) that these plants release in the air when infested by whiteflies. 

 

Using the electronic nose system to sample VOCs. Image courtesy of ARS

 

The team built a prototype e-nose that can operate in the greenhouse. Consisting of gas sensors, data acquisition modules, and other components, the device effectively converts VOC samples to digital signals, which are then transmitted to an algorithm programmed to recognize specific types and concentrations. The system can distinguish between the "smell fingerprints" of infested tomato plants versus healthy ones.

The team believes that with additional testing and development, their system could be a useful monitoring tool for greenhouse owners to tackle whiteflies before they reach damaging levels.