Multiplexed quantification of soil nutrients using an AI-enhanced and low-cost impedimetric sensor

Soil management technologies are key to tackling the UN development goals and ensuring the 70% increase in agricultural production needed by 2050. In particular, the determination of soil nutrients is crucial to optimising plant growth and maximising crop yields. However, the direct determination of ion availability is challenging due to environmental changes (e.g., soil moisture, temperature), and it requires costly and bulky equipment. While indirect methods, such as soil electric conductivity and hyperspectral imaging, offer advantages in terms of costs, their measurement accuracy is low, since they are influenced by a high number of parameters, including soil texture and composition. As such, there is a shortage of low-cost and accurate sensors to determine nutrient bioavailability in soil.

This work describes, for the first time, an impedimetric low-cost device for the simultaneous and direct determination of environmental parameters (i.e., temperature and humidity), as well as of the bioavailability of key ions in soil. The device could be incorporated into an Arduino-based setting, reducing manufacturing and operational costs and enabling flexible implementation across a wide range of settings. The impedimetric device was tested in vivo by implanting it into tomato plants and recording impedimetric signals over time. An AI algorithm was also trained to enable an accurate determination of ion concentrations. The final system could determine dynamic changes in Na⁺ and K⁺ with high accuracy (R²=0.98 in the case of sodium and R²=0.99 for potassium), and it could be used for the continuous monitoring of soil parameters.