Investigation on concentration detection of turbid solution based on hemisphere sample cell and multidimensional spectroscopy

The study developed a hemisphere sample cell and constructed a multidimensional spectroscopy acquisition system to enhance the accuracy of detecting turbid solution with scattering properties. The system simultaneously captures absorption and scattering information from the tested samples. Monte Carlo simulation was employed to model the transmission light intensity distribution data from sample cells of various shapes. It was found that the hemisphere sample cell increases the dimensionality of transmission light intensity information and enables the acquisition of more scattering-related data. Furthermore, 46 samples of intralipid-20% solution with varying concentrations were examined. Models were constructed using partial least squares (PLS) regression with one-dimensional and two-dimensional light intensity distribution data. The results indicate that the model using two-dimensional light intensity distribution data significantly outperforms the model using one-dimensional data, reducing the root mean square error by 39.96% and increasing the correlation coefficient by 0.332%. Experimental results demonstrate that the multidimensional spectroscopic modeling method employing the hemisphere sample cell can significantly enhance the accuracy and speed of detecting chemical composition concentrations in turbid solution.