FT-ICR MS analyses were performed at the COBRA Laboratory (Infranalytics FR2054 research infrastructure) using a 12 T Solarix XR mass spectrometer (Bruker Daltonics).

Water samples for solid phase extracted DOM (SPE-DOM) from 5 and 500 m were transferred to HDPE carboys (acid washed and rinsed with MilliQ water) until their processing in the lab, within 2 hours. The samples were successively filtered through 3 then 0.2 µm polycarbonate filters using a peristaltic pump. The first 1 L, used to rinse the filters and tubing, was discarded. The filtered sea water was then recovered in 2 L combusted glass bottles and acidified to pH 2 with 37% HCl. DOM was extracted using styrene divinyl benzene polymer cartridges (Agilent Bond Elut PPL, 1 g). After extraction, cartridges were rinsed with acidified MilliQ water (200 mL, pH 0.01N). The cartridges were then dried with pure air using sterile air-filters and stored at -20 °C until elution. Elution was done by gravity using 6 mL HPLC grade methanol (MeOH). Extracts aliquots were dried under pure air (sterile air-filtered) until complete evaporation of MeOH for mass spectrometry analysis. However, the presence of chloride ions from the rinsing process caused DOM signal suppression. To resolve this, all samples underwent a second SPE process: dried DOM extracts were resuspended in 5 mL of LCMS water and run through 1g PPL cartridges. After this second extraction step the cartridges were rinsed with 3 mL of LCMS water and dried with N2. Elution was done using 2 mL of LCMS MeOH in amber glass vials and stored at -20 °C. Fourier transform ion cyclotron resonance mass spectrometry FT-ICR MS analyses were performed at the COBRA Laboratory (Infranalytics FR2054 research infrastructure) using a 12 T Solarix XR mass spectrometer (Bruker Daltonics). SPE-DOM MeOH extracts were diluted in LCMS grade MeOH and ran in FIA (flow injection analysis) in analytical triplicates and several MeOH blanks were randomly run with each sample set. Sample replicates analysis order was randomized to rule out running order effects.  A quality control sample consisting of an equi-volume mixture of all samples was also analyzed. An electrospray ionization source was used in negative mode. Mass spectra were acquired in a mass range of m/z 98.2-1200. Accumulation time was set to 0.02 s and 200 scans were accumulated. 

The instrument was first externally calibrated with sodium trifluoroacetate (NaTFA), and then internally calibrated with a calibration list of CxHyOz molecular formulas (MFs, n=90). This internal list of calibrants was made using a test sample spiked with an Agilent tune mix mass calibration solution. A series of ions repeatedly observed in previous DOM studies (m/z 339.10854, 369.11911, 411.12967, 469.13515, 541.15628) were also identified and checked based on their fine isotopic structure. These ions were then added to the calibrants list. The alkylate series of these ions were then identified and included in the calibrants list. 

MFs were assigned using the Bruker DataAnalysis software in a mass range of m/z 98.3-1200 respecting the following restrictions: CxHyN00-2O0-15S0-1, double bond equivalent (DBE) ≤ 80, 0.4≤ H/C ratio≤ 3, only peaks with signal to noise ratio (S/N) > 9 were considered and a 0.5 ppm maximum mass error tolerated. Since there was very low signal at m/z above 850, attributed formulas were filtered to keep only m/z values ≤ 850. An intense peak at m/z 265.14; likely contamination; was removed from all samples. Only the MFs identified in all analytical replicates were kept for each sample replicate. With these cleaning steps, we obtain a set of 3789 of MFs identified across all samples. For each sample, a mean MF list is created by merging the analytical replicates.