Optimisation of a surface enhanced Raman scattering method using design of experiments and Bayesian design space modelling

Surface enhanced Raman scattering (SERS) is an alternative technique based on Raman spectroscopy, which has been increasingly applied to pharmaceutical analytical chemistry in the last decade. It consists in enhancing the Raman effect by performing analyses using metallic surfaces, such as silver and gold colloids, on which the target molecules are adsorbed to be detected. It has been observed that in this way, an enhancement factor of 103-106 times can be obtained and the lack of sensibility related to conventional Raman scattering overcome [1].
Nowadays, design of experiment (DoE) is widely employed for modelling phenomena in analytical method development and optimisation, especially in the context of separation techniques. It is a structured approach that allows correlating key responses to controllable variables. Ideally, a certain number of factors may affect the critical method attributes (CMAs) of an analytical process in a negative or positive way. These factors are named critical method parameters (CMPs). DoE is employed, as a chemometric tool, to individuate CMPs and then, deeply study how they affect the process under study. To do so, CMAs are linked to CMPs by a regression model built by means of multivariate linear or partial least squares regression. Generally, the designs can be classified in two categories: screening and optimization designs. The formers are generally implemented when a high number of parameters are supposed to influence the analytical process and no much prior information is available. They result in useful tools to study the effects of both continuous and discontinuous factors. Instead, the optimisation designs are principally used to study wisely selected continuous factors [2].
The design space (DS) is defined as a multidimensional area in which the specifications given to the CMAs are met with a defined level of probability. Obviously, the larger the DS is, the more robust the method is. Its computation is achieved by several approaches, such as Monte-Carlo simulations, Bayesian methods as well as bootstrapping techniques [3].
The aim of this project was to combine and apply two potent chemometric tools such as DoE and Bayesian DS to SERS method development and optimisation.
[1] Cailletaud, J., De Bleye, C., Dumont, E., Sacré, P.-Y., Netchacovitch, L., Gut, Y., Boiret, M., Ginot, Y.-M., Hubert, P., Ziemons, E., Critical review of surface-enhanced Raman spectroscopy applications in the pharmaceutical field. J. Pharm. Biomed. Anal. 147, 458-472, 2018.
[2] Sahu, P.K., Ramisetti, N.R., Cecchi, T., Swain, S., Patro, C.S., Panda, J., An overview of experimental design in HPLC method development and validation, J. Pharm. Biomed. Anal. 147, 590-611, 2018.
[3] Deidda, R., Orlandini, S., Hubert, P., Hubert, C., Risk-based approach for method development in pharmaceutical quality control context: A critical review. J. Pharm. Biomed. Anal. 161, 110-121, 2018.