High-selectivity phenol detection in cumene process wastewater via bromination and dynamic optical path

Phenol is a vital chemical precursor in modern industry; however, the dominant cumene process for its production generates wastewater laden with complex organic compounds, making the accurate detection of phenol critically important. Traditional spectroscopic methods struggle with this challenge due to severe spectral overlap among the components. To overcome this analytical bottleneck, we introduce a novel framework that integrates chemical modulation with a custom-built, multidimensional sensor array for the highly selective and sensitive monitoring of such complex industrial effluents. At the core of our system is a bespoke U-shaped cuvette that permits the optical path length to be dynamically adjusted from 1 to 10 cm, thereby introducing a critical physical dimension to the dataset. By capturing the high-dimensional optical response of the sample before and after a selective bromination reaction-a process that specifically eliminates the spectral signature of phenol-we constructed an information-rich differential dataset that provides distinctive features for resolving the phenol signal from a convoluted background. This dataset was modeled using various machine learning algorithms, among which the Random Forest (RF) model yielded the best performance, achieving a coefficient of determination (R²) of 0.99826. Critically, our dynamic path length strategy lowered the limit of detection (LOD) to 0.0709 mg/L, a greater than five-fold enhancement in sensitivity over conventional fixed-pathlength methods. The robustness of this approach was thoroughly validated in authentic river and marine water matrices, where spike-and-recovery experiments produced ideal results, with recovery rates consistently falling within the 95 %-105 % range.