The role of colloidal material in the fate and cycling of trace metals in estuarine and coastal waters
Trace metals in natural colloidal material (l-400nm) were investigated in the River Beaulieu, the Trent- Humber system and the Celtic Sea. Colloidal and truly dissolved (<10,000molecular weight) fractions were separated by cross-flow filtration (CFF). Filtration artefacts (conventional and CFF) involving filter blockage and colloid aggregation/disaggregation transformations were minimised through appropriate protocols. 65Zn equilibration experiments were carried out with Beaulieu River colloids (<0A^m). The initial uptake of 65Zn onto colloidal material (10 to 15 %) was very rapid (seconds to minutes) but a large fraction of the tracer remained in the truly dissolved phase. Preliminary modelling to estimate the fraction of colloidal zinc under variable SPM concentrations and Kc sensitivities was completed using different partition coefficients (Kc and Kf) derived during the equilibration experiments. Initial 65Zn-colloid partitioning was related to the mass concentrations (and theoretical surface area) of colloids and colloidal aggregates. The 65Zn radiotracer/colloid association was exchangeable but overall equilibrium partitioning of colloidal/truly dissolved 65Zn did not change with increased total zinc concentration. Loose hydrophobic colloidal aggregates (filaments) with Mn hydroxide coatings formed during sample storage and were capable of removing ~60% of the 65Zn. Field flow fractionation (FFF) of colloidal Beaulieu River water highlighted the significance of the smaller colloidal fractions (<0.08^m and 0.08 to 0.134m) which contained the highest concentration of Fe, Mn and Zn when normalised to an assumed geometric surface area. A spatial and seasonal investigation of colloidal trace metals in the Trent/Humber system identified groups of metals with varying colloidal associations in the order (high to low); Fe> Pb, Mn > Cu, Zn, Ni > Cd. Colloidal Fe, Pb and Mn all illustrated removal in the low salinity region of the estuary. Ni (mainly truly dissolved) showed somewhat conservative behaviour. Total dissolved Cd (and Zn) consistently showed a mid-estuarine maximum (truly dissolved), which was attributed to chloro-complex formation or ionic exchange with major seawater ions. For most metals, the positioning and intensity of the turbidity maximum zone (TMZ) appeared to have greatest control on their removal. Cu is controlled seasonally by input of organic ligands (truly dissolved/microcolloidal). The existence of separate organic and inorganic colloidal pools was proposed, which may have seasonally variable signatures. The presence of lower molecular weight (<10,000Da) colloids is significant for Pb and Cu. Investigations of colloidal Al in the Celtic Sea highlighted problems with colloid disruption within the cross-flow system, and detection of Al colloids using a surface active fluorometric technique. Water column profiles of reactive and dissolved (<0.4/u.m) Al gave insights into processes controlling Al distribution and concentration. Resuspension of Celtic Sea sediment led to an instantaneous release of truly dissolved Al derived from authigenic mineral dissolution. FFF analysis of samples from the resuspension experiments showed that colloidal Al was not a significant fraction (<20% of total dissolved Al). On the basis of these experiments it was possible to estimate that, in the absence of scavenging, the input of truly dissolved Al from typical resuspension events could be sufficient to account for the 5-8nM increase in dissolved Al concentrations observed close to the bed. This research has contributed towards a new approach to interpretation of trace metal speciation and processes in a variety of natural waters. Assessment of the role of colloidal and truly dissolved phases within the 'dissolved' fraction has enabled better understanding of particle/water/metal interactions. The description of trace metal/colloid partitioning and associations within these multicomponent systems has the potential to enhance their modelling in the future.