Preconcentration

The second critical step during the analysis of iron in seawater with both Fe(II) and Fe(II)+(III) techniques, is the preconcentration procedure. This stage of the analysis is

important in that: i) it allows separation of iron from some of the interfering metal

cations for the CL reaction, with quantitative recovery of the element; ii) it permits a

large enrichment factor, thus lowering the limit of detection with a high sample throughput; and iii) it removes the sea-salt matrix which, at higher pH, may lead to

precipitates in the manifold.

Conventional Chelex-100 resin, which has often been used to separate metals from solutions, is not appropriate in FIA-CL systems because of the swelling and contraction of the resin itself when the pH changes (Obata et al., 1993). Chelating resins containing 8-hydroxyquinoline (8-HQ) have been made, which is a well-characterised reagent that reacts with over 60 metal ions to form stable complexes, and can be immobilised on a support matrix. In early studies 8-HQ was immobilised onto silica gel, which has a good mechanical strength, resistance to swelling and rapid overall exchange kinetics in column application (Sturgeon et al., 1981). It was however found unstable at high pH (> 9), the chelating group potentially “bleeding” by hydrolysis and subsequently potentially showed contamination for iron from the newly exposed silica surface (Sturgeon et al., 1981). Therefore in later systems the silica substrate was replaced by a polymer such as Fractogel TSK which is a highly porous, mechanically and chemically stable hydrophilic organic gel more stable at high pH (Landing et al., 1986). However, this synthesis was time-consuming (> 20 h) and sometimes failed for unknown reasons. A new single- or double-step protocol (depending on the starting chemical) was found to link 8-HQ to the TSK polymer via an amino link instead of an ester linkage, which reduced the “bleeding” of 8-HQ from the resin (Dierssen et al., 2001). Later studies suggested that TSK resins may leach colour (8-HQ bleeding) when eluted with a concentration of hydrochloric acid higher than 0.1 M as used with the Fe(II)+(III) technique, making the determination of Fe(III) at low concentrations impossible due to the masking effect of the leached functional group (Obata et al., 1993; Weeks and Bruland, 2002). The choice of the resin therefore depends on the technique used and the compromise made between loading capacity, elution profiles and stability to acids (Obata et al., 1993; Weeks and Bruland, 2002).

The seawater matrix is complex and may potentially create interferences during the detection step. Sea-salt ions such as Mg2+, Ca2+ and Cl- tend to significantly suppress (for cations) or increase (for halides) the chemiluminescence signal (Chang and Patterson, 1980; Bowie et al., 1998). These ions may also precipitate (e.g. to Mg(OH)2)

after mixing with basic luminol solution at a pH > 10 and clog the detector (de Jong et al., 1998). A rinsing step with ultra pure water after passing the sample through the column is therefore necessary to remove sea-salts still present in the dead volume of the column. According to the results of Obata et al. (1993) and de Jong et al. (1998), Fe(III) was quantitatively collected at a pH between pH 2.6 and 4, and Fe(II) was completely recovered at pH 5 and above. Using a basic pH (> 8) may lead to the formation of iron colloids so that iron was not fully recovered from the sample stream and may even precipitate (Weeks and Bruland, 2002). Chromium(III), Co(II), Cu(II), and Mn(II) are the few elements susceptible of interfering with the chemiluminescence reaction (see below), however only Co(II) and Cu(II) are collected onto the resin at a pH of 5 – 5.5 (Obata et al., 1993; de Jong et al., 1998; Weeks and Bruland, 2002). Both Fe(II) and Fe(III) can thus be selected from some of the interfering trace metals by carefully buffering the pH to 5.5 (Bowie et al., 2002a; Weeks and Bruland, 2002). At this pH, Fe(II) is susceptible of oxidising to Fe(III) on the order of few minutes, but this reaction can be minimised by adding the buffer just prior loading the sample onto the 8-HQ resin, when using the Fe(II) technique (see below).

The preconcentration column is therefore important for separating iron from many trace-metals, lowering the limit of detection, and removing sea-salts. A limitation may be that, while the reagent blank could be made negligible, there still might be a column blank that may be non-negligible when measuring sub-nanomolar iron levels (Weeks and Bruland, 2002). Additionally several factors can impact on the chelating efficiency onto the 8-HQ resin, such as the pH of the buffered sample, the loading flow rate, the eluent concentration, the column preconditioning, the column size, and the organic speciation of iron in the sample (Bowie et al., 2003; Bowie et al., 2004). Factors such as the particle size, porosity, and texture of the resin will also have an impact on the extraction efficiency from the 8-HQ resin (Bowie et al., 2003). It was also recently shown that the presence of organic ligands in seawater samples modify the quantity of iron collected onto preconcentration resins (Ndung'u et al., 2003; Ussher et al., 2005). In order to ensure that all Fe complexes are destroyed and dissolved iron is loaded onto the resin, micro-wave treatment (Weeks and Bruland, 2002) or UV-digestion (Guéguen et al., 1999; Ndung'u et al., 2003) of the acidified sample prior analysis have been recommended. Alternatively, it has been suggested that stored samples should not be analysed before a minimum of 6-months after collection to allow complete release of iron from organic complexes and colloids (Bowie et al., 2004).