Our study approach focused on characterizing MWF sources (new and used MWFs in sumps) and identifying important exposure determinants explaining the measured personal particulate and gaseous MWF exposures. This approach would help to better anticipate workers’ exposures to straight and water-based MWF and implement exposure reduction strategies.One main result of this study was that the oil mists in workshops are not simply aerosolised MWF. Rather, a compositional change between the MWF and the generated aerosols was observed for both OC and metal content. For OC, these changes were dependent on the MWF type. Results presented in Fig. 1 and Fig. S6 in Supplementary Material suggest that machining with straight MWFs generated an aerosol with increased amounts of compounds with intermediate thermal stability compared to used MWFs. A possible explanation is an oxidative degradation of hydrocarbons present in the MWFs20. On the contrary, machining with water-based MWF tended to generate aerosols with lower amounts of volatile components and greater amounts of higher thermal stability components (Supplementary material, Fig. S7). This suggests that water-based MWFs were more prone to pyrolysis at the tool-machined interface with high temperatures (350–750 °C13) compared to straight MWFs. Since oil mist is relatively stable and long lasting in indoor air16, aging processes could also occur and bring additional changes to the aerosolized MWF compositions.Some metals (particularly Cu, Pb and Zn for straight MWF and Al, Fe and Zn for water-based MWF) tended to concentrate in the sumps during use (Table 2). Clearly, the presence of these metals was the result of machined metal, mainly brass and bronze for straight MWFs, and stainless steel and aluminium for water-based MWF (Table 1). Contrary to the sumps, airborne MWF contained fewer metallic elements and if at all, at very low concentrations (Table 3). The metal distribution in used MWF collected from the sumps were quite different from the air concentrations collected at the same time and were not correlated. This result suggests that metals are mainly quenched in the MWF. Only one study has determined the relationship between metal concentrations in MWF sumps and air concentrations in the workshops34. They reported concentrations of Fe, Mn and Zn in sumps in the range of 1–100 mg/L in workshops machining stainless steel and very low metal air concentrations, which concur with our results. Although 84% of Pb concentrations in the air of the workshops was below LOQ (35 ng/m3), Pb concentrations were very high in straight MWF sumps. In the studied workshops machining non-ferrous metals (e.g., brass), the sumps concentrations for Cu and Pb were between 100 and 200 mg/L. Based on the Basel convention on the Control of Transboundary Movements on Hazardous Wastes and Their Disposal35, used oil with Pb levels > 100 ppm must be considered as hazardous wastes and be treated. Skin exposure to MWF and degreasing solvents were often observed during the sampling campaigns. These solvents might alter the skin barrier and render skin more permeable to metals already known to penetrate skin, such as Cu, Pb, Cr, and Ni36. Pb concentration in our study were similar to a study showing significant skin permeation of Pb (29–132 µg/g) and thus, contribute to the overall Pb blood levels37. This exposure route should be considered in further studies.This study confirms the importance of measuring not just the particulate, but also the gaseous MWF fraction in air to properly assess workers’ exposures to MWF. The median inhalable particulate MWF mass fraction measured in this study was similar to others recently reported values9,19,29,38, which were all below 230 µg/m3. Characterisation of the gasous MWF fractions have so far been poorly investigated. Previous published scientific studies have targeted specific compounds such as ethanolamine39 or formaldehyde29, but only Lillienberg19 and Koller9 measured the gasous MWF fraction in parallel with the inhalable particulate MWF fraction. The concentration range was 0.2–29.4 mg/m3 for gasous MWF compounds found in Swedish workshops using straight, water-based, and synthetic MWF and had a mean ratio (gaseous mass)/(particulate + gaseous mass) between 84 and 94%19. This ratio is similar to our study (88% for workshops using straight MWFs) but higher than the 66% for water-based MWFs in this study. Lower gaseous to total airborne MWF mass ratio for water-based MWFs were observed by Koller et al.9 who reported ratios between 40 and 95% in samples collected in Swiss workshops. The presence of a larger mass of gaseous MWF compounds in the air observed for straight MWFs compared to water-based MWFs could result from differences in MWF viscosity. Indeed, evaporation of MWFs from filters was less than 5% for cutting fluids with viscosities greater than 18 Cst (at 40 °C)15. Materials safety data sheet for all water-based MWF encountered in our study indicated viscosities above this value, whereas most straight MWFs presented viscosities below 18 Cst (Supplementary material Table S1).Aldehyde concentrations were quite low (< 80 µg/m3, Table 3) and similar between workshops using straight or water-based MWFs. Such concentrations are similar to concentrations found in ambient air40 and are comparable with formaldehyde concentrations in workshops using MWF19,20,41. As the European regulation has prohibited the use of formaldehyde in the formulation of MWF42, the presence of this carcinogenic compound could result from the oxidative degradation of MWFs20.The European standard (EN 68943) describes a strategy for assessing compliance with occupational exposure limits (OEL) for single chemical substances as well as for chemical mixtures. We defined two similar exposure groups (SEG): one for workers exposed to straight MWFs and a second for workers exposed to water-based MWFs. Workers for both SEGs were also exposed to formaldehyde (as a representative of all aldehydes) and metals including boron, calcium, iron, and nickel as predominant elements in the aerosol (Table 3). Based on the EN 689 procedure to compare results with OELs, we calculated an exposure index for each selected pollutant (Tables S2, S3 in Supplementary material). Using this approach, most of the chemical agents contribute less than 0.1 of the OEL, except for the particulate fraction of straight or water-based MWF. We thus tested the compliance of these measurements with the OEL based on the comparison of the 70% upper confidence limit with the 95th percentile of the distribution of the results. The results obtained (Table S4 in the Supplementary material) indicated that there was still a greater than 5% probability of MWF exposures (particulate or particulate + gaseous fraction) exceeding the existing OELs or recommended values in the surveyed workshops. Although aerosol concentrations of formaldehyde, aluminum, and nickel were very low relative to their respective OELs (Tables S2, S3, Supplementary material), they are associated with skin sensitization (for Ni), asthma (for Al and Ni), and cancer (formaldehyde and Ni) where no toxicological threshold exists. Furthermore, possible synergetic or antagonistic effects of such co-exposures cannot be excluded44. Therefore, reducing MWF exposures and other pollutants in these workshops is still warranted29.Process types are considered key determinants for exposure to MWF aerosols32. The processes we followed in this study were always in combination, such as turning combined with drilling and deburing to produce the metal piece, making it challenging to pinpoint process type as a precise exposure determinant. Consequently, we preferred constructing clusters of activities, as country-dependent differences were also observed (Table 4). These clusters were important determinants for explaining personal exposure to the particulate or gaseous fractions (Tables 5, 6).For particulate exposure when using straight oil, about three times higher MWF concentrations were observed during machining in Switzerland compared to France. Differences in machines type and enclosure (more CAM use in Switzerland compared to processes using stamping press in France, Table 1) could explain these differences. As expected, particulate exposure reduced (2.3–3 times) when machines stopped for setting or maintenance. Considering water-based MWF, handling tasks generated greater (about 2 times) aerosolized particules compared to CNC machining (used as the reference in our models). This could result from covering the machined parts by a thin MWF film for corrosion protection before being transported or using the compressed air gun before packing and shipping the metal pieces.This study is the first to our knowledge applying the exposure determinant approach to the gaseous fraction of MWF (Table 6). For straight MWF, setting activities including changing tools and machines maintenance generated very high levels of gaseous components and were the only determinants significantly associated with such an exposure. This surge could be associated with the use of compressed air gun to clean the inside of the machine or rapidly opening the CNC’s doors after testing new settings.For water-based MWF, the task of handling packaged parts was again an important exposure determinant for the gaseous fraction, but of borderline significance. The evaporation of this protective MWF layer could be at the origin of this increased exposure.The particulate MWF exposure was greater with CAM machines compared to CNC and could be related to the absence of enclosure for the CAM. Indeed, we showed this to be a significant determinant explaining particulate exposure for workers using straight MWF (Table 5). The absence of enclosure induced up to three times more particulate exposure to straight MWF. This value is in line with Lillienberg19, who report an increase of the inhalable particulate fraction of about 1.5 times for partly open machines. The importance of enclosed systems is also stressed for transfer lines, with a reported reduction of about 90% of the particulate MWF concentration28.Machine type and enclosure were not significant exposure determinants for the gaseous fraction (Table 6). This illustrates the difficulty to control the emissions of volatile compounds without considering ventilation and elimination of recirculating air29.The use of hydraulic fluids significantly increased the gaseous fraction of water-based MWF (Table 6). This has previously been reported in laboratory experiments where the presence of tramp oils in water-based MWF increased the misting potential45. This could explain the significant contribution of this determinant for the gaseous fraction measured in workshops using water-based MWF.MWF management was a significant exposure determinant for managing gaseous fraction from water-based MWF. A large reduction of the exposure to the gaseous fraction was achieved when the used MWF was not recycled but eliminated through the fluid purchaser.Increased workshop temperatures reduced the mass of both particulate and gaseous MWF fractions, especially for water-based MWF. Such temperature exposure effect has been reported previously23.This study belongs to a larger and broader Franco-Swiss epidemiological research focussing on occupational exposures to MWF with the aim to assess whether MWF exposure is associated with biomarkers of oxidative stress, genotoxic effects as well as respiratory symptoms30. The consideration of the sumps as sources of contaminants in addition to the joint characterization of the particulate and gaseous fraction should also be noted. Finally, exposure variables like the organic carbon and the metal concentrations are not often reported and bring new insight on the MWF transformations and emissions.Like all field studies, these results also present limitations. The first one relates to the selection of companies, which was based on a willing to participate rather than to be representative of the present exposures, and thus, our results might not be generalizable to all companies using MWF. A further limitation in the modelling of the determinants is the collinearity between some determinants. In particular, the machine type is highly correlated with the machine enclosure as all CNCs had an enclosure, whereas the grinding wheels mostly hadn’t. This might have led to the absence of the expected effect of enclosure. Some determinants were difficult to collect like the use of compressed air, the machining parameters, or the frequency of MWF changing. These determinants could not be considered in the statistical models. Whereas the recording grid for the exposure determinants was done jointly, differences in recording activities or other determinants by both hygienist teams could be present and thus induce an observational effect. This could have an influence on the activity clusters for example. A stepwise approach has been used for the inclusion of each exposure determinant in the statistical models. When not statistically significant, the added determinant was not considered. This lead us for example to not consider the ventilation parameter, which in other studies has been reported as an important determinant.
