Bringing the ocean into the laboratory to probe the chemical complexity of sea spray aerosol

Abstract

The production, size, and chemical composition of sea spray aerosol (SSA) particles strongly depend on seawater chemistry, which is controlled by physical, chemical, and biological processes. Despite decades of studies in marine environments, a direct relationship has yet to be established between ocean biology and the physicochemical properties of SSA. The ability to establish such relationships is hindered by the fact that SSA measurements are typically dominated by overwhelming background aerosol concentrations even in remote marine environments. Herein, we describe a newly developed approach for reproducing the chemical complexity of SSA in a laboratory setting, comprising a unique ocean-atmosphere facility equipped with actual breaking waves. A mesocosm experiment was performed in natural seawater, using controlled phytoplankton and heterotrophic bacteria concentrations, which showed SSA size and chemical mixing state are acutely sensitive to the aerosol production mechanism, as well as to the type of biological species present. The largest reduction in the hygroscopicity of SSA occurred as heterotrophic bacteria concentrations increased, whereas phytoplankton and chlorophyll-a concentrations decreased, directly corresponding to a change in mixing state in the smallest (60-180 nm) size range. Using this newly developed approach to generate realistic SSA, systematic studies can now be performed to advance our fundamental understanding of the impact of ocean biology on SSA chemical mixing state, heterogeneous reactivity, and the resulting climate-relevant properties.

Keywords: biologically active; cloud condensation nuclei; clouds; ice nucleation; marine aerosols.

Fig. 1.

(A) Bubble radius distributions for breaking waves (gray squares), plunging waterfall (blue circles), and sintered glass filters (red line). [Reproduced with permission from ref. (Copyright 2002, Macmillan Publishers)]. (B) Probability density function of the resulting SSA number distributions (dN/dlogd_p, with the d_p at 15 ± 10% RH) produced by these three methods. The SSA distribution recovered from the plunging waterfall is consistent with that produced by breaking waves, whereas the SSA distribution by the sintered glass filters (red triangles) is considerably narrower.

Fig. 2.

Concentrations of seawater TOC (green circles), chlorophyll-a (orange line), heterotrophic bacteria (blue squares), and photosynthetic eukaryotic phytoplankton (black triangles) measured over the 5-d mesocosm experiment. Times for the four major additions of growth medium (GM), heterotrophic bacteria (HB), and phytoplankton (PH) are noted (A1–A4) with green arrows (SI Text, section 2.1 and Table S1). Dashed boxes mark two sampling regions (R1 and R2).

Fig. 3.

(A) Size-resolved chemical mixing state for R1 (Fig. 2). Integration of two single particle analysis methods [TEM with energy-dispersive X-ray (EDX) analysis < 562 nm and ATOFMS > 562 nm] shows the existence of four major particle types. (B) STXM chemical spatial maps of the two most dominant submicrometer particle types (types 2 and 4) highlight the differences in the inorganic-to-organic ratios (Left), abundance of chloride (Center), and carboxylates (Right).

Fig. 4.

(A) Single particle chemical mixing state for particles with a diameter between 60 and 180 nm for sampling regions indicated in Fig. 2 and Table S3 (the full dataset in these regions is discussed in SI Text, section 6). Num. fraction, number fraction. (B) Same as A, but for the next largest size regime (180 < d_p < 320 nm). (C) GF at 92% RH for the hygroscopic fraction of SSA (GF_{act. frac.} = d_wet/d_dry, act. frac. refers to the active fraction that took up water, GF >1.2), (D) d_act measured at 0.2% supersaturation for cloud droplet formation, and (E) IN concentrations (Conc.) at −32 °C as a function of seawater TOC levels. The shaded regions are provided as a visual guide. The data represent mean values, and the vertical error bars for GF and IN are 2σ. For CCN, the lower error bars are 2σ, whereas the upper bars account for potential errors in the counting mechanism used (SI Text, section 5.1). The horizontal error bars represent 1σ.