Aerosol iron solubility is a major uncertainty in the global biogeochemical cycle of iron and, via its impact on ocean productivity, the carbon cycle and their influence on global climate. Previous studies have reported widely different values for this solubility (0.01 – 80%). Here we show that the primary control on aerosol iron solubility is the surface area to volume ratio of mineral aerosol particles, which changes during atmospheric transport as mineral aerosol concentration decreases due to preferential removal of larger particles (assuming particle morphology to be relatively constant with particle size). This important result indicates that aerosol iron solubility is not fixed, but will change predictably as an inverse function of dust concentration on both spatial and temporal (e.g. glacial – interglacial) scales.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 The solubility of aerosol Fe in seawater is a key uncertainty in our understanding of the global biogeochemical cycle of Fe and, through the influence of seawater Fe concentrations on phytoplankton growth, of the global carbon cycle and climate [Jickells et al., 2005]. Numerous studies, using a wide variety of starting materials and analytical techniques, have reported aerosol Fe solubility values over a very broad range (0.01 – 80%) [Mahowald et al., 2005] and there has been much debate as to which of these values are applicable to the environment [Duce et al., 1991; Fung et al., 2000; Bopp et al., 2003; Jickells et al., 2005]. Studies on the solubility of Fe in soils [Sillanpaa, 1982; Bonnet and Guieu, 2004] generally report lower values than observed in atmospheric aerosols [Zhuang et al., 1990; Chen and Siefert, 2004; Baker et al., 2006b] and several workers have used laboratory results to suggest that chemical or photochemical mechanisms during transport through the atmosphere might be responsible for this effect [Pehkonen et al., 1993; Spokes and Jickells, 1996; Desboeufs et al., 2001]. However, field and modelling studies on the influence of atmospheric processing on Fe solubility have generally been unable to confirm that these processes have a significant effect [Hand et al., 2004; Luo et al., 2005; Baker et al., 2006a, 2006b]. The majority of work in this area has been carried out on desert soils or aerosols because mineral dust generated in deserts constitutes by far the largest source of Fe to the ocean. We recently conducted the first systematic study of the solubility of Fe in aerosols from diverse source regions between 50°N and 50°S in the Atlantic Ocean using a weak acid leach and reported that Fe solubility varied systematically with aerosol source, Saharan dust having the lowest values [Baker et al., 2006b]. Here, however, we reanalyse that, and other, data to elucidate the fundamental processes regulating aerosol solubility.
2. Materials and Methods
 Aerosol samples were collected during 3 research cruises in the Atlantic Ocean, using high volume (1 m3 min−1) aerosol samplers. Aerosol collection substrates (Whatman 41, cellulose) were acid-washed before use to reduce trace metal blanks. During two of the cruises (JCR and M55) aerosol samples were size-segregated using cascade impactors. Bulk samples were collected during IRONAGES III. Results are reported here for the bulk (sum of fine and coarse modes for JCR and M55) aerosol. Soluble aerosol Fe and other trace metals were determined using a weak acid leach (1.1M ammonium acetate, pH 4.7) and total metals (and total P) using a strong acid digestion (concentrated HF/HNO3). Soluble P and Si concentrations were determined using aqueous extraction at pH 7. Percentage solubilities for aerosol components were calculated by Solx = 100 * Xsol/Xtotal, where superscripts ‘sol’ and ‘total’ refer to the soluble and total concentrations of each aerosol component (X). Full details of analytical procedures are given elsewhere [Baker et al., 2006a, 2006b]. Mineral aerosol atmospheric mass loadings (md) were calculated from total aerosol Al concentrations by assuming that all aerosol Al was derived from mineral dust, which comprises 7% by mass Al. In order to illustrate the variation in mineral aerosol surface area to volume ratio (A/V) with changing particle size we make a (simplistic) assumption of spherical particles and hence A/V = 3/r, where r is the particle radius.
3. Results and Discussion
 Using the data of Baker et al. [2006b] and Baker et al. [2006a] we show that aerosol Fe solubility is an inverse function of the atmospheric concentration of mineral aerosol (Figure 1). Similar relationships can be obtained for the solubility of aerosol Al, P and Si (Figure 2), elements whose principal atmospheric source is also mineral dust. The solution chemistry of Fe is characterised by low solubility under oxic conditions and active photochemistry [Jickells et al., 2005] while neither of these effects are observed for Al, P and Si. These factors indicate that the observed solubility behaviour is controlled by dissolution from the solid phase, rather than individual element solution phase chemistry (although, as we note below, solution phase chemistry has a significant influence on subsequent Fe dissolution in seawater).
 It is well known that the average size of the mineral aerosol population decreases with distance away from desert dust sources [McTainsh and Walker, 1982], as a result of higher deposition rates for larger particles [Duce et al., 1991], although large particles (>20μm) can also be transported great distances [Mahowald et al., 2005]. Particle diameters decrease from ∼ 60 – 80 μm for particles deflated during saltation [Mahowald et al., 2005] to ∼ 7 – 28 μm just off the coast of West Africa [Stuut et al., 2005] and ∼ 1 – 3 μm for Saharan-origin mineral particles over the Caribbean [Prospero et al., 1970; Talbot et al., 1986]. Typical dust concentrations during this transport also decrease from 400 – >3000 μg m−3 at, or near, dust storm sources [Gillies et al., 1996; Offer and Goossens, 2001] to 9 – 104 μg m−3 off the coast of West Africa [Baker et al., 2006a, 2006b] and 3.2 – 18.7 μg m−3 at Barbados (annual mean concentrations for the period 1965 – 1992) [Prospero et al., 1996]. Demonstration of the decrease in size of mineral aerosol particles with transport outside of the main dust export routes is more difficult, because the processes of internal and external mixing with other aerosol components combine to make specific determination of the mineral particle size very difficult. Zhang et al.  recently described how adsorption of seaspray on mineral aerosol leads to particle growth, which in turn enhances gravitational settling rates. They noted that the effect of these two processes was to maintain the modal size of the overall aerosol population at ∼3 – 4 μm, but that the size of the mineral component of the aerosol decreases with increasing transport time over the ocean. Thus we hypothesise further size reduction, concurrent with decreasing concentration, during transport to the remote (from desert sources) atmosphere.
 We examine the likely change in mineral aerosol particle A/V with changing mineral aerosol atmospheric concentration using the data for near-source, West African coastal and Caribbean mineral aerosol concentrations and particle sizes (hence A/V) noted above (Figure 3). Figure 3 also shows example data for mineral aerosols taken from the literature for which both concentration and particle size were reported, together with an envelope which we suggest describes the relationship over the mineral aerosol concentration range encountered for our field data. Note however that we have not been able to obtain specific information on mineral aerosol particle size (as opposed to overall mixed composition aerosol particle size) data at low mineral concentrations, such as are present over High Nutrient Low Chlorophyll areas. Thus we cannot confirm the relationship at very low concentrations, although mineral aerosol particle radii of <0.1 μm (A/V > 30 μm−1) have been reported [Satheesh and Moorthy, 2005]. The A/V-concentration plot reproduces the overall form of Figure 1. Particle solubility is fundamentally a surface process and we therefore suggest that changes in the mean surface area to volume ratio of the mineral aerosol population, caused by progressive removal of larger mineral aerosol particles during transport, account for the trend in aerosol Fe solubility shown in Figure 1. Here we assume that particle morphology is similar across the mineral aerosol size spectrum. Some differences in morphology (for instance, very rough large particles versus smooth small particles) might counteract the trend in A/V with concentration that we propose. In general however, we suggest that as mineral aerosol particles become smaller a greater proportion of their volume is exposed and therefore available for dissolution.
 The weak acid leaching methods employed in the studies noted above do not reproduce the dissolution characteristics of aerosol Fe in seawater directly, primarily because the solution pH is too low (∼ pH 4 vs ∼ pH 8 for seawater) [see Baker et al., 2006b]. As a result, the absolute magnitude of Fe solubility in Figure 1 is probably over-estimated somewhat. However, the relative solubility characteristics of the aerosol samples are recorded by these methods. In terms of the global Fe and C cycles, the important parameter is how much aerosol Fe is soluble in seawater (or bioavailable to marine organisms) [Jickells et al., 2005]. The solubility of Fe in seawater is extremely low [Liu and Millero, 2002], but may be enhanced by organic ligands with a high affinity for Fe [Kraemer, 2004]. Thus the overall solubility of aerosol Fe on deposition to the ocean will be dependent on the pre-existing dissolved Fe concentration in the seawater and the presence and nature of Fe-binding organic ligands there, as well as the size of the aerosol mineral particles. Our results (through application of a single analytical method to real aerosol samples collected over a huge gradient in dust concentrations) reveal the nature and extent of the atmospheric control on aerosol Fe solubility over such large concentration gradients. This will be subsequently modified by effects in seawater after deposition to the ocean.
 Our observation that aerosol Fe solubility is higher at low mineral concentration is consistent with recent studies on seawater Fe chemistry which imply relatively high Fe solubility (up to 40%) at sites remote from large deserts [Boyle et al., 2005; Sedwick et al., 2005]. The effect of this trend in Fe solubility will be to smooth out the strong gradients in atmospheric iron (dust) supply to the ocean, with proportionately less soluble Fe delivered to areas close to major dust sources and more to remote areas (e.g. the Southern Ocean), relative to most current models, which use a fixed value for aerosol Fe solubility [Fung et al., 2000; Bopp et al., 2003]. However, the observed variation in aerosol solubility is at least an order of magnitude lower than that in dust deposition, so that gradients in soluble Fe inputs to the ocean, consistent with global patterns of iron limitation in ocean waters [Jickells et al., 2005], will still be present. More recent models do incorporate variable aerosol Fe solubility, parameterised as a function of atmospheric chemical processing [e.g. Fan et al., 2006] which is computationally expensive. Our results suggest that the link between atmospheric processing and Fe solubility is primarily physical rather than chemical in nature, and can be readily parameterised in global models as a function of transported mineral dust concentration. Variations in atmospheric iron supply to the oceans over temporal scales (e.g. glacial – interglacial timescales) will also be modulated, because of the implied non-linearity in inputs of soluble iron with inputs of total iron. Our results remove some of the uncertainty involved in discussion of aerosol Fe solubility and imply that debates over the “correct” value for this parameter are inappropriate.
 This study was supported by the UK Natural Environment Research Council through the Atlantic Meridional Transect consortium (NER/O/S/2001/00680). This is contribution 134 of the AMT programme. We thank two anonymous reviewers for their contributions to the manuscript.