Metallic Nitride Microfluidic e-Tongue: A Novel Selective Approach for the Detection of Macronutrients in Soil

The growing global demand for food requires optimizing agricultural practices and more rational use of natural resources without expanding cropping areas. Precision agriculture (PA) tools are essential for accurately applying fertilizers and herbicides, reducing costs, and avoiding environmental impacts. Standard macronutrient mapping methods are costly and time-consuming, limiting denser sampling collection in the field. Consequently, devices employing new materials with specific properties matching sensitivity to agricultural nutrients, robustness to face intensive climatic variations, and economical manufacturing viability are mandatory. In this sense, microfluidic impedimetric e-tongues have emerged as practical tools in PA due to their high sensitivity, adaptability, affordability, and ease of use. These sensors provide rapid qualitative and quantitative results in liquid media, with applications extending to food analysis, environmental monitoring, and biosensing. Here, metallic nitride thin films (CrN, BN, and TiN) deposited via physical vapor deposition (PVD) using the glancing angle deposition (GLAD) technique are applied as sensing units presenting high sensitivity, controlled micro- and nanostructures, durability, reproducibility, and mechanical robustness, essential characteristics for future on-site soil analyses. The GLAD technique allows precise control over the micro- and nanostructures deposited on gold interdigitated electrodes to create molecular sieves for a possible capture of target species (e.g., K⁺, Na⁺, Mg²⁺, Ca²⁺, PO₄^3-) from soil samples. We demonstrate the feasibility of using distinct nitrides as sensing units in a microfluidic e-tongue tested with soil samples with distinct compositions diluted in water without pretreatments. The sensor successfully differentiated all samples tested, showing higher K and Mg macronutrient resolution. These findings demonstrate the high potential to detect minute changes (<1 mmol·L^-1) in soil fertilization, with results compared by four prediction models, paving the way for future in situ analyses envisaging a controlled delivery of macronutrients during fertilization.