Aerosol deposition is one of the major processes providing bioavailable Fe to the surface ocean. However, the quantification of aerosol Fe flux in the surface ocean is highly challenging operationally. In this study, we measured both Fe isotopic composition and specific elemental ratios in 5 size-fraction aerosols collected over the East China Sea (ECS) to quantify the relative contribution of lithogenic and anthropogenic aerosol Fe. Both the isotopic and elemental ratios indicate that anthropogenic aerosol Fe mainly originates from high temperature combustion activities, with the end member of the δ⁵⁶Fe to be -4.5 ‰. We found that the Cd/Ti ratio is a much more reliable proxy to quantify the contribution of anthropogenic aerosol Fe in coarse aerosols than δ⁵⁶Fe in the ECS. Attributed to extremely high deposition velocities and low solubilities for large size aerosols, lithogenic aerosols are still the dominant dissolved aerosol Fe source in the ECS.

Method

Sampling site and procedures

The aerosol samples were collected in a small islet in the ECS, Pengjia islet (PJ, 1.1 km², 25.63°N, 122.08°E), which is located at 66 km from the northernmost of Taiwan (Fig. 1). There are only about 30 governmental staff stayed on the volcanic islet for meteorological observation and coastal guarding. Based on aerosol optical depth data, PJ receives aerosols from the north during the northeast monsoon season and from the south during the southwest monsoon season (Fig. 1). This makes it an ideal time series sampling site to investigate the impact of East Asian aerosols on the surface water of the Northwestern Pacific Ocean (NWPO). Size-fractionated aerosol samples were collected by polytetrafluoroethylene filters (TE-230-PTFE, Tisch Environmental Inc., US) installed on the high volume aerosol sampler (TISCH Environmental Inc., US, MODEL-TE-5170) and coupled with a cascade impactor (TISCH Environmental Inc., US, Series 235). The cascade impactor separated aerosols into five size fractions, including stage 1 (>7.3 mm), stage 2 (3.1-7.3 mm), stage 3 (1.6-3.1 mm), stage 4 (1.0-1.6 mm), and stage 5 (0.57-1.0 mm). The stage 5 and stage 1-4 aerosols are referred to as fine and coarse aerosols in this study, respectively. The collection time for one single filter generally lasted for 7-8 days each month from September 2019 to August 2020 to ensure sufficient mass for isotopic composition analysis in different treatments. Detailed sampling information, including sampling dates and volumes, is shown in Table S1. The filters with aerosol samples were freeze-dried and weighed after collection. The aerosol samples were then stored at -20 °C freezer prior to further chemical processing.

Quantification of dissolved and total concentrations, and isotopic composition

All sample pretreated procedures were carried out in an ISO Class 4 positive pressure cleanroom by operators wearing powder-free polyvinyl chloride (PVC) gloves. We followed the suggested protocols of GEOTRACES Cookbook to clean sample vials (Cutter et al., 2017). Regarding the dissolved fraction, we conducted two different leaching protocols, including instantaneous ultrapure water leached (ultrapure water) and acetate buffer leached (buffer) treatments. The ultrapure water Fe was obtained by passing ultrapure water through 0.2 mm filters instantaneously (Buck et al., 2006; Morton et al., 2013). For buffer leached Fe (buffer Fe), an ammonia acetate buffer solution at pH 4.7 was used to mimic aerosol metals dissolution processes by rainwater (Baker and Jickells, 2006; Sarthou et al., 2003) or short period complexation processes by seawater ligands (Perron et al., 2020). For total digestion (referred as ‘total’ hereafter), samples were heated for 4 h at 120°C (heater temperature) in 2 ml of a freshly prepared mixed solution containing 4M HF, 4M HCl, and 4M HNO₃ (Eggimann and Betzer, 1976). Filter blanks for each treatment were obtained by subjecting new filters to the same leaching procedures as the samples. In over 90% of the cases, the sample concentrations were at least two orders of magnitude higher than the concentrations of the blanks, and in all cases, the samples were at least one order of magnitude higher than the blank value. The detailed information about the pretreatment procedures, analytical method, blank test, and the validation of precision and accuracy were described in our previous studies (Hsieh et al., 2022; Hsieh et al., 2023).

Iron isotopic compositions were determined by using multi-collector Inductively coupled Plasma Mass Spectrometer (MC-ICP-MS, Neptune Plus, Thermo Fisher Scientific) equipped with a sample inlet system APEX-IR (no gas added), normal Ni sampler cone, and X-type skimmer cone from Elemental Scientific. Samples were measured in high-resolution mode with ⁵⁴Cr correction on ⁵⁴Fe and ⁵⁸Ni correction on ⁵⁸Fe. The d⁵⁶Fe measurements were conducted using the double-spike technique, involving the addition of mixed spike (⁵⁷Fe and ⁵⁸Fe), with a sample-to-spike ratio of 1:2, as described by Dauphas et al. (2017). The δ⁵⁶Fe data (⁵⁶Fe/⁵⁴Fe) ratios are reported in per mil notation (‰) relative to the IRMM-014 Fe isotope reference material (Institute for Reference Materials and Measurements) and described below:

To minimize the Cr and Ni interferences, we purified all samples by using an anion exchange resin (AG1-X8, Bio-Rad, 100-200 mesh). The samples were loaded in 1.4 mL 7 N HCl solution onto perfluoroalkoxy alkanes (PFA, Savillex) microcolumns filled with the resin. We then applied ultrapure 0.4 ml 7N HCl thrice to remove Ni and Cr. The Fe was retained on the resin and later eluted out with 0.7 N HCl (Fig. S1). The eluted off purified Fe samples were dried in open cap vials then were redissolved subsequently in 1 ml of 0.5 M HNO₃ at 120 °C for 2 hours with closed caps, and then were ready for isotopic analysis. We measured IRMM-014 right after 4 sample analysis and measured NIST-3126a right after 8 sample analysis to validate analytical accuracy. Our long-term δ⁵⁶Fe value of NIST-3126a is 0.36±0.03 ‰, averaged from 5 different analytical periods spanning for half year. The precision and accuracy of δ⁵⁶Fe data in this study were validated by using reference materials BCR2 and BHVO2 (Table 1, Craddock and Dauphas (2010)).