Comparison of robot-deployable sensing methods for autonomous in-field screening of total petroleum hydrocarbons

Advances in robotics and automation enhance workforce safety and increase the speed of environmental site characterization. Accordingly, real-time, robot-deployable sensing methods with sufficient analytical performance are needed to support site management decisions. We evaluated two autonomous approaches for soil total petroleum hydrocarbon (TPH) screening: (1) thermal desorption followed by gas chromatography (TD) and (2) direct sensing using a portable near-infrared (NIR) spectrometer. Soil samples from two crude-oil-impacted sites were analyzed and benchmarked against EPA Method 8015 C (solvent extraction and gas chromatography with FID detection). The TD method achieved high precision and agreed with the EPA reference within 3-27 % across 2270-18,980 ppm TPH but requires complex thermal and gas-handling hardware that would have to be engineered for robustness and to be intrinsically safe. The NIR method enabled rapid, reagent-free analysis, however, it also showed larger deviations (20-40 %) among samples. A prototype robotic platform integrating the NIR module was developed and tested in controlled outdoor conditions, demonstrating the feasibility of autonomous, real-time screening. The present study selected a few soil matrices and a realistic range of TPH concentrations to evaluate analytical trends between the two robot-deployable TPH sensors and compared their potential robotic deployment feasibility. Future work should further expand the number of soils and sites, refine the NIR calibration transferability, and engineer the TD-GC components to achieve fully integrated robotic deployment in field settings.