Carbon Nanotube-Based Sensor array for water contaminants detection: Sensing performance enhancement using artificial intelligence strategies 

Background

The proposed postdoctoral position will take place within the French ANR project 4Water “For WATER Quality Monitoring” dedicated to the development of environmental sensors for aquaculture application and the European-Indian Project LOTUS dedicated to the fabrication of “LOw-cost innovative Technology for water quality monitoring and water resources management for Urban and rural water Systems in India”. 

India is facing a water and sanitation crisis. While a staggering 17% of the world population resides in India, the country’s share of world water resources is small — only 4% — which creates a significant gap between water demand and availability. Furthermore, given that 80% of the country’s water is used for irrigation, only 20% is available for drinking water and other industrial needs. Only 32% of households have access to a drinking tap water from a treated water source.

Digital water solutions, i.e., integrating ICT to water management, have the potential to improve water quality and availability for municipalities and consumers worldwide. Much needed water quality monitoring, which is traditionally slow, expensive and requires specialized personnel, can now be achieved on a real-time basis with no expertise involved. LOTUS and 4Water projects bring new ICT solutions for water and sanitation challenges in rural and urban area.

It is based on an innovative multi-parameter, carbon-nanotube-based chemical sensor array for real time, adaptable water quality monitoring of contaminants. This sensor array is developed at Ecole polytechnique, within NACRE team, a joint research team between Ecole polytechnique, CNRS and IFSTTAR. The device exploits carbon-nanotubes-based resistive chemical sensors with optimized, differentiated chemical functionalizations. Sensitivity to pH, chlorine, chloride, hardness and nitrates has been demonstrated so far.

Principle of operation

Each sensor in the array consists of a conducting network of multi-walled carbon nanotubes (MWCNT) directly ink-jet printed on Silicon between interdigitated electrodes. The MWCNT are functionalized by specific polymers selected for their capability to interact reversibly and selectively with the target analytes and to interact strongly with the MWCNT. When the device is exposed to water containing a mixture of various species, in the absence of functionalization, most species interact with the MWCNT-based devices without selectivity. In the presence of functionalized MWCNT on the contrary, the devices respond selectively to the targeted analyte thanks to the functionalization.

Proposed research

The characterization in water of such sensor array, as well as their deployment in testing scenario (such as in Sense-City facility), generates massive amount of data. This data contains all the information on the sensitivity, the selectivity and the ageing features of the sensor array. So far, it is exploited in a straightforward statistical approach with little automation to derive the average sensitivities of the sensors to selected analytes. This approach has already enabled to prove results at the state of the art in terms of carbon nanotube sensor array for water monitoring.

However, it is fairly well-known that electronic tongue algorithms[i] capable to exploit deeply cross-correlations between sensors (principal component analysis & regression, partial least squares, various types of neural networks) may enable a drastic increase in sensitivity and selectivity.  They may also provide information on factors perturbing measurements (such as cross-sensitivity to other chemicals, temperature…) as well as on sensor ageing.

The aim of the project is to evaluate the performances of such methods, and more generally of supervised and unsupervised machine learning, applied to carbon-nanotubes-based sensor array for chemical sensing in water.

The project includes not only the development and testing of e-tongue algorithms, but also in-depth interactions with the rest of the team (in charge of device fabrication and characterization) in view of developing experimental protocols and data management practices adapted to the use of automated data processing algorithms.

Team & organization

Position located in greater Paris area: Ecole polytechnique, Route de Saclay, 91128 Palaiseau, France.

Work within NACRE research team (IFSTTAR, Ecole polytechnique, CNRS), in Platine research platform (https://portail.polytechnique.edu/lms/fr/projet-platine) in LPICM laboratory (https://portail.polytechnique.edu/lpicm/en)

The team counts 3 researchers, 2 engineers, 1 technician, 8 to 10 young researchers (interns, PhDs, postdocs) and 2 startup projects. One of these startups (in creation) notably focuses on the commercialization of the sensor array technology for water monitoring.

The researcher will be in charge (typically ½ day per week, up to 1 day per week) of project management activity for LOTUS. 

Application and deadlines

The successful candidate will have a strong background in data sciences, for instance in data analytics, machine learning or artificial intelligence.

Fluent English (written and spoken), high proficiency in technical writing and presentations. Autonomy, spirit of initiative, decision-making skills, creativity, team working. Project management skills.

To be considered, applications must include detailed resume and motivation letter.

Starting date: Adjustable. Preferably February 2019.

Contacts

Bérengère Lebental  berengere.lebental(at)polytechnique.edu