Speciation and solubility of copper along a soil contamination gradient

Abstract

The bonding of copper (Cu) in soil is important for bioavailability of Cu and toxicity towards plants and animals. The hypothesis of the present study is that variations in molecular Cu bonding can explain previously reported variations in soil ecotoxicity along a Cu contamination gradient when parent material and mineralogy are similar. The specific aim is to understand molecular bonding at the different levels of bioavailability, and compare this with results from a sequential extraction. The studied field samples were retrieved from an anthropogenic contaminated site (Hygum, Denmark) dated to the 1920s. This fallow field shows a steep Cu gradient within a confined area where other soil forming factors are constant. Five surface soil samples (0–15 cm) with Cu levels ranging from background (25 mg Cu kg−1) to strongly contaminated levels (3920 mg Cu kg−1) were analyzed by X-ray absorption spectroscopy [Cu K-edge X-ray absorption near edge structure (XANES)], sequential chemical extractions by a three-step BRC method, particle-size and density fraction, microprobe mapping (2 × 2 μm resolution) and bulk soil chemical analysis. Only Cu and the soil organic C contents covaried, and both bulk chemical analyses and microprobe mapping showed strong co-localization of Cu and natural organic matter (NOM). The distribution of Cu between particle-size fractions does not vary significantly, and the clay fraction accounted for 62–75 % of the bulk soil Cu content. The largest Cu fraction (78–81 %) was found in the light density fraction (<1.8 g cm−3). The acid-extractable and the reducible fractions are well explained by a simple linear regression (R 2 = 0.97 and R 2 = 0.99, respectively). Both the oxidizable fraction and the residual fraction level out at the highest contaminations. The molecular environment indicated by XANES was in all the contaminated samples best described by Cu(II) bound to humic acids only. The natural samples’ XANES revealed that a part of the Cu here was mineral bound as Cu(0)/Cu(I). The microprobe, physical fractionations, and XANES results all supported previous reports that Cu contamination is bound to NOM in soils, and we here show that this is independent of the studied contamination levels when all soil properties except NOM content are constant. Current understanding of sequential chemical extraction protocols does not incorporate this consistently, as the large observed shifts in our extractability of Cu fractions were most likely caused by differences in submicrometer molecular Cu–NOM bonding and not differences in Cu–mineral phase interactions.