Per- and polyfluoroalkyl substances (PFAS) are synthetic chemicals, which can exert a range of toxic effects, including a behaviour as 'endocrine disruptor chemicals' (EDCs). Despite the ban under the Stockholm Convention, considering their high persistence in the environment, detectable concentrations can be found in different matrices like water, food, etc. In addition, to replace long-chain perfluoroalkyl acids, ether organofluorinated substances (ether-PFAS) have been synthesised, including GEN-X (C6H4F11NO3) and ADONA (C7H2F12O4), with the aim of increasing the molecule's solubility and thus degradability by introducing the ether functional group. Indeed, toxicological studies have shown that these synthesised 'emerging PFAS' also have a non-negligible toxicity towards humans and animals, so their detection, quantification and regulation will be necessary in the future. To develop innovative and improved methods aimed at the detection and quantification of PFAS in different matrices, mass spectrometry and liquid chromatography parameters were optimized by employing Design of Experiments (DoE). The selected responses to be optimized were chromatographic peak area (related to sensitivity) and resolution (chromatographic separation), while variables were the following: on the chromatographic side, flow, gradient ramp and column temperature; on the mass spectrometry side, capillary voltage, sheath gas flow and fragmentor voltage. Considering the number of variables, the Placket-Burman (PB) design was chosen using a standard mix of 20 PFAS to identify the more influent variables, with the minimum number of experiments (16, including replicates). In order to reduce the computational effort, a Principal Component Analysis (PCA) was performed on the matrix containing the peak area response. The loading plot of the first two components shows different correlations between the responses, highlighting 4 different clusters of analytes. Thus, the computation of one model for each cluster was performed instead of 20 models for the individual analytes. Coefficient plots show that most variables are significant; in particular, the sheath gas flow was significant with a positive sign for all 4 clusters. The positive effect on the response may be due to either the easier generation of smaller droplets in the ionization process, or the improved focalization of the ions into the spectrometer inlet. In terms of resolution response, unresolved 'critical pairs' were identified, for a total of 9 pairs. Also in this case most variables were significant. Specifically, the flow showed a positive sign: higher mobile phase flow rates probably reduce the phenomenon of longitudinal diffusion, producing narrower peaks and thus higher resolution. The validation of the models was performed by the execution of the experiment at coded level 0. For both peak area and resolution responses, validation by Student's t-test was confirmed for some models, but not for most of them, suggesting non-linear models. Therefore, a response surface DoE will be set, considering the significant variables from PB results in order to optimize sensitivity and resolution. By working in this way, it will be possible to apply the method optimized for analysis of various matrices like water, food or packaging, upon development of an appropriate pre-treatment technique.
Optimization of LC-Q-TOF Mass Spectrometry and Chromatographic Parameters for the analysis of PFAS: a multivariate approach to maximize sensitivity and resolution
Daniel Bona;Marina Di Carro;Barbara Benedetti;Emanuele Magi
2025-01-01
Abstract
Per- and polyfluoroalkyl substances (PFAS) are synthetic chemicals, which can exert a range of toxic effects, including a behaviour as 'endocrine disruptor chemicals' (EDCs). Despite the ban under the Stockholm Convention, considering their high persistence in the environment, detectable concentrations can be found in different matrices like water, food, etc. In addition, to replace long-chain perfluoroalkyl acids, ether organofluorinated substances (ether-PFAS) have been synthesised, including GEN-X (C6H4F11NO3) and ADONA (C7H2F12O4), with the aim of increasing the molecule's solubility and thus degradability by introducing the ether functional group. Indeed, toxicological studies have shown that these synthesised 'emerging PFAS' also have a non-negligible toxicity towards humans and animals, so their detection, quantification and regulation will be necessary in the future. To develop innovative and improved methods aimed at the detection and quantification of PFAS in different matrices, mass spectrometry and liquid chromatography parameters were optimized by employing Design of Experiments (DoE). The selected responses to be optimized were chromatographic peak area (related to sensitivity) and resolution (chromatographic separation), while variables were the following: on the chromatographic side, flow, gradient ramp and column temperature; on the mass spectrometry side, capillary voltage, sheath gas flow and fragmentor voltage. Considering the number of variables, the Placket-Burman (PB) design was chosen using a standard mix of 20 PFAS to identify the more influent variables, with the minimum number of experiments (16, including replicates). In order to reduce the computational effort, a Principal Component Analysis (PCA) was performed on the matrix containing the peak area response. The loading plot of the first two components shows different correlations between the responses, highlighting 4 different clusters of analytes. Thus, the computation of one model for each cluster was performed instead of 20 models for the individual analytes. Coefficient plots show that most variables are significant; in particular, the sheath gas flow was significant with a positive sign for all 4 clusters. The positive effect on the response may be due to either the easier generation of smaller droplets in the ionization process, or the improved focalization of the ions into the spectrometer inlet. In terms of resolution response, unresolved 'critical pairs' were identified, for a total of 9 pairs. Also in this case most variables were significant. Specifically, the flow showed a positive sign: higher mobile phase flow rates probably reduce the phenomenon of longitudinal diffusion, producing narrower peaks and thus higher resolution. The validation of the models was performed by the execution of the experiment at coded level 0. For both peak area and resolution responses, validation by Student's t-test was confirmed for some models, but not for most of them, suggesting non-linear models. Therefore, a response surface DoE will be set, considering the significant variables from PB results in order to optimize sensitivity and resolution. By working in this way, it will be possible to apply the method optimized for analysis of various matrices like water, food or packaging, upon development of an appropriate pre-treatment technique.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
