injection volume could be obtained (benzo[a]pyrene: 25 fg/?L), leading to a 100–3500-fold higher sensitivity compared with common GC-EI-MS (Stader et al., 2013; Große Brinkhaus et al., 2017). The concentrations were corrected for recoveries and dilution effects. Coeluting pairs of PAH are marked in Table 1 for both analytical techniques. For the non- target PAC screening using GC-APLI-MS, a list (peaks with correspond- ing mass-to-charge ratio, retention time and intensity) of features with a minimum intensity of 50,000 counts per second was established which contains only aromatic compounds due to the aromatic-selective ioni- zation process. Masses of features which were assigned according to Xiao et al. (2017) were identified by m/z  0.005. For heavy fuel oil analyses, flash chromatography (adapted SARA fractionation, EPA method 3611B, 1996) was used followed by analysis of the aromatic fraction eluted with dichloromethane (Thianer and Achten, 2017). 3. Results and discussion 3.1. Concentrations of 71 PAH including 16 EPA PAH in inlet and scrubber discharge water 3.1.1. Open loop-inlet water As expected, only traces of PAH were found in the OL-inlet samples (sampling point 1, Fig. 1) of the four ships (for single PAH compounds <0.087 ?g/L (S1), <0.85 ?g/L (S2), <0.020 ?g/L (S3) and < 0.019 ?g/L (S4), Suppl. Inform. S4). They are in line with the studies of Teuchies et al. (2020) and Du et al. (2022), being in the same range. The separated dissolved fractions of the inlet samples showed low concentrations for the EPA PAH of 0.05–1.3 ?g/L (P71 PAH: 0.05–4.6 ?g/L) and for the particulate fractions even lower concentrations for the EPA PAH of 0.01–0.10 ?g/L (P71 PAH: 0.02–0.27 ?g/L) (Suppl. Inform. S4). 3.1.2. Discharge water of open loop and closed loop systems The OL-discharge water samples were colorless and the CL samples showed a light yellowish color with visible black soot particles and flocs (Fig. 2). If for the CL samples the sum of HMW PAH (>fluoranthene) of dissolved fractions (which are expected to be sorbed to very small par- ticles) are compared with corresponding sum of dissolved and particular fractions, 2.9, 81, 6.6 and 5.5 % of HMW PAH for S1, S2, S3 and S4 result, respectively. If S2 is assigned an outlier as only low amounts of particles were present in the original discharge sample (Fig. 2), roughly 5 % of the PAH in particles are found in the dissolved fractions, hence very roughly 5 % particles present in the dissolved fractions can be concluded. Notably, some black soot particles were still visible in the so- called dissolved fractions after centrifugation in all discharge water samples (very little in OL and more in CL samples), indicating that the separation into the two fractions in the sample preparation procedure (centrifugation) was not fully successful due to the small size of the soot particles. The P EPA PAH ( P 71 PAH) concentrations in OL-discharge water samples ranged from 1.13 to 24.2 ?g/L (3.20–105.6 ?g/L) with median concentration of 20.9 ?g/L (73.8 ?g/L) and in CL samples from 6.04 to 96.5 ?g/L (24.3–371.0 ?g/L) with median concentration of 30.8 ?g/L (109.7 ?g/L) (Fig. 3). In general, the PAH concentrations in ships 1, 2 and 4 are in the same range, while ship 3 shows concentrations lower by an order of magnitude. Ship 1 and 2 resemble each other in the results (except for the particulate fraction) and this can be due to the similarity in the model of the engines and fuel used on board (see Section 3.2). Other relevant parameters to be considered are presented in Table 2. On the one hand, engine type, type of scrubber installation and engine load influence the emissions of PAH in the exhaust gas by affecting the engine performance and combustion efficiency. For instance, a two- stroke engine commonly leads to higher emissions of unburnt fuel than a four-stroke engine. This was not reflected here: ship 3 with a two- stroke main engine had the lowest PAH concentrations. It is assumed that in this case the fuel quality was the dominating influence. Regarding the type of scrubber installation, retrofit installations (ship 1 and 2) result in higher exhaust gas backpressure, if the scrubber is not adapted to the existing engine, and this impairs the engine performance leading to increased emissions. Regarding engine load, low load oper- ation tends to lead to higher emissions (Grigoriadis et al., 2021). The engine loads did not differ substantially between sampled ships but between operation mode: OL showed higher engine loads (63–86 % MCR) than CL (16–38 % MCR). On the other hand, specific flowrate and water treatment affect the concentration of PAC that end up in the discharge water. At lower specific flowrates (discharge water volume per engine's energy output) the pollutants show a higher concentration. In addition, the continuous re-circulation in CL-operation of the wash water, which causes higher temperatures than in OL-operation, can lead to volatilization of highly- volatile compounds such as 2-ring PAH like naphthalenes. A lower specific flowrate in CL-operation implies a more intensive recirculation, therefore a lower fraction of naphthalenes in the discharge water is expected (ship 3 and 4). The water treatment in CL-operation has been reported to be very efficient in the removal of hydrocarbons, especially for those of higher molecular weight (Magnusson et al., 2018). Ac- cording to Marin-Enriquez et al. (2023), the flocculant addition (ship 1 and 4) seems to enhance the removal efficiency in CL; the water treat- ment in OL, in turn, did not show a clear removal effect so that a sig- nificant impact on the PAH concentrations can be excluded. In OL- and CL-discharge water samples, P 71 PAH concentrations were three to four times higher compared with the P EPA PAH. Par- ticulate PAH were apparent in both OL and CL systems, however they constituted 69–90 % (except for ship S2: 1 %) PAH in CL systems in contrast to only 0.1–66 % PAH in OL systems. It has to be noted that comparison of dissolved and particulate PAH concentrations should be done with care because complete homogeneity during sample prepara- tion cannot be guaranteed. In comparison, Du et al. (2022) detected P EPA PAH (total of unsubstituted and alkylated PAH) concentrations of 4.1–16.0 ?g/L (16.6–102.8 ?g/L), respectively, and a median concentration of 7.5 ?g/L Fig. 2. Open loop (OL, left) and closed loop (CL, right) discharge water samples of ships 1–4 (S1–4). C. Achten et al. Marine Pollution Bulletin 208 (2024) 116790 5