Optimization of SPME and chromatographic conditions
Several polymers with different thicknesses and polarity are commercially available and currently used as SPME coatings according to the analyte and sample characteristics. In this study, five different SPME coatings (i.e., polydimethylsiloxane, divinylbenzene, carboxen, polyacrylate, and polyethylene glycol) were tested and compared in terms of extraction efficiency, evaluated by the number of detected peaks (signal-to-noise ratio > 3) and the total peak area of the VOCs profile in indoor air samples, used as a reference model, before starting the analysis of working environments inside the manufacturing plant. Then, two fibers (monophasic and biphasic, based on polydimethylsiloxane and carboxen/polydimethylsiloxane, respectively) were chosen and used for all subsequent analyses to recover the greatest number of compounds. Therefore, for each air sample, two parallel 30 min-SPME extraction processes with the two different fibers were carried out and the total list of the identified compounds was obtained by combining the results of each analysis. The SPME procedure was also optimized in terms of adsorption and desorption time, ranged from 10 -60 min and 3-9 min, respectively. Then, optimal values of 30 min (sampling/extraction time) and 6 min (desorption time) were applied for all analyses.
The optimization of the temperature gradient was performed to achieve a good compromise between low chromatographic times and adequate peak resolution. Details of the employed analytical method are summarized in the experimental section, and, as an example, a typical VOC chromatographic profile is shown in Figure 1.
VOC profiles
The evaluation of indoor air quality was performed by a passive sampling process through SPME fiber exposure in different zones of the manufacturing factory. VOC measurements were conducted for three months during the summer, with a sampling frequency of fifteen days; the total number of sampling times is 6, for each SPME fiber, both when the system is on and when the system is off. The complete list of the compounds identified in the inner and outer chamber, at system on and off, is shown in Table S1 (Online Resource – Supplementary Information). A total of 212 compounds were identified: 146 and 52 in Zone A at system ON and OFF, and 105 and 37 in Zone B at system ON and OFF, respectively.
As reported in Table 1, only 17 compounds were always present, both in Zone A and B, regardless of system conditions (ON or OFF). When the system was ON, 58 common compounds were observed both in Zone A and B (Table 2), while only 24 substances were concurrently found in both zones when the system was OFF (Table 3). Considering the Zone A, a compound comparison was performed between the system ON and OFF (Table 4). A total of 33 common compounds were found. Among them, higher levels of benzene and phenol-based molecules were observed when the system was ON. Similar evaluations were carried out about the data relating to the outer chamber (Table 5) at system on and off; fewer common compounds (i.e. 24) have been identified, at lower levels than those in the inner zone.
A Venn diagram of the identified compounds, observed in specific conditions or common to the different situations, is shown in Figure 2. Among the 17 common compounds, identified in all cases (zone A and B and system ON and OFF), a semi-quantitative evaluation, shown in Figure 3, was performed for the potentially toxic compounds: benzene, (1-methyldodecyl)-, benzene, 1,3-dimethyl-, phenol, 4,4'-(1-methylethylidene)bis-, and p-xylene.
Considering that the working conditions are different inside and outside the painting chamber when the system is on or off, multivariate analyses (i.e., Principal Component Analyses) were carried out to gain insight into these differences. The peak area values of a shortlist of 17 selected (presumably toxic) compounds such as benzene, xylene, and phenol derivatives, specifically: benzene, (1-butylnonyl)-; benzene, (1-ethylundecyl)-; benzene, (1-methyldodecyl)-; benzene, (1-methyltridecyl)-; benzene, (1-methylundecyl)-; benzene, (1-pentyloctyl)-; benzene, (1-propyldecyl)-; benzene, 1,2,3-trimethyl-; benzene, 1,3,5-trimethyl-; benzene, 1,3-dimethyl-; benzenesulfonamide, N-butyl-; o-xylene; phenol; phenol, 2,4'-isopropylidenedi-; phenol, 4,4'-(1-methylethylidene)bis-; p-xylene; toluene) were determined and used for the statistical analysis. In the data matrix, empty gaps, corresponding to missing data for VOCs not found in some of the samples examined have been filled with a virtual peak area value equal to the LOD signal, estimated as the tenth part of the minimum peak area within the entire data set. After data pre-treatment (centering and normalization), the PCA model showed two principal components; the PC1 axis justifies 82% of the total variance, while PC2 covers a further 17%. Each variable positively affects the regression parameters of the PCA model (R2= 0.99 and Q2 0.99), and no outliers were found, as confirmed by the DModX plot (i.e., the Distance to the Model, measuring how well each observation fits the model). From the biplot of Figure 4, it is possible to observe a dense distribution of all the compounds investigated in the region corresponding to the switched-on plant, along the outer edge. Indeed, as expected, the observations ON fall in the region occupied by a high number of variables, and, specifically, the Biplot region occupied by A and B OFF does not contain any compounds.
From the data in the correlation matrix, reported in Table S2 (Online Resource – Supplementary Information) Pearson coefficients higher than 0.90 were obtained for almost all compounds belonging to the class of benzene derivatives, except for the following three couples: benzene, (1-butylnonyl)-/benzene, 1,2,3-trimethyl-, benzene, (1-methylundecyl)-/benzene, 1,2,3-trimethyl-, and p-xylene/benzene, 1,2,3-trimethyl-. With the exception of two couples of compounds (phenol/benzene, 1,2,3-trimethyl- and o-xylene/benzene, 1,2,3-trimethyl-) with negative Pearson coefficients, positive correlations between the two areas throughout the operating phases ON/OFF were observed for all other compounds, suggesting that the VOCs in the internal and external chambers shared same sources (volatilization of paint solvents, personal care products, building materials, etc.) and underwent similar processes, showing comparable chemical lifetimes in the atmosphere.
The proposed experimental plan did not allow a quantitative analysis of each VOC found in the environment, but it could detect the presence of these substances at very low levels (ppb or ppt), Specifically, within the scope of this study, only a semi-quantitative comparison of the levels of some selected volatile organic compounds in different working areas and stages was carried out. The results indicate, as expected, that when the manufacturing system is active, the relative amounts of organic compounds in the environment increase. Anyway, it can be assumed that the values obtained do not differ from the levels generally observed in urban areas, both residential and industrial sites. Numerous studies have reported exposure to air pollutants and often refer to single cities, short time periods, and specific air pollutants (Fassò et al. 2023; Gilardi et al. 2023). In particular, the phenomena of air pollution from benzene-based molecules (benzene is a ubiquitous pollutant of indoor and outdoor air, as well as class 1 carcinogenic substance, according to IARC - International Agency for Research on Cancer) have been extensively described in the recent literature, referring to various regions of Italy (Martellini et al. 2020; Toscano and Murena 2021; Cattaneo et al. 2021; Ielpo et al. 2021; Di Gilio et al. 2021; Cucciniello et al. 2022; Manco et al. 2022; Urbano et al. 2023). In addition to highly industrialized areas, high pollution levels are frequently reported in the most densely populated areas, in close correlation with geographic characteristics, climate, seasons, number of inhabitants, urban traffic, and time slots of the day (Battista et al. 2021; Badaloni et al. 2023; Ciacci et al. 2023). Therefore, in the complex air quality scenario, it is reasonable to assume that the levels of VOCs, found in indoor work environments, comply with occupational limits and air quality guidelines, considering the practices adopted by the company for continuous air exchange and all necessary measures for the health of workers, in full compliance with current legal provisions. Moreover, the effect of VOCs on indoor air quality and potential health consequences do not only depend on the concentration levels, but also on the exposure time which, generally, is limited to short periods of time for each worker, thanks to adequate work shifts.