Chapter 1: introduction
In this thesis an account is given of a research project dealing with the chemical speciation and bioavailability of copper in Lake Tjeukemeer, a lake in the north of the Netherlands. The reason for the initiation of this project was a lack of knowledge about the speciation of copper and the influence of copper on the behaviour of algae. This influence may be important, because copper is an essential element for algae, but becomes toxic when it is present at too high concentrations. From a literature search it was concluded that free copper can be assumed to control bioavailability for algae. Therefore, it was decided to focus attention on free copper. Currently, no techniques are available that can measure the concentration of free copper directly at the concentration levels typically found in Lake Tjeukemeer. However, some techniques may give useful information, providing that the results are carefully interpreted. As a consequence of the low concentrations of copper, present in natural waters, determinations of copper concentrations are very prone to interferences due to contamination.
Chapter 2: copper concentration in Lake Tjeukemeer and external influences
Lake Tjeukemeer is a shallow, alkaline and eutrophic lake. It is part of the Frisian "boezem". The hydrology of the lake is man- made: in winter water from the surrounding polders is pumped into the lake and dominates the water composition, while in summer water originating from Lake IJsselmeer is let in. The total "dissolved" (i.e. after filtration through a 0.2 μm pore size filter) copper concentration in Lake Tjeukemeer is typically around 30 nM. Although no evidence could be obtained, it is likely that the copper concentration in the lake is not increased by letting in water from Lake IJsselmeer. However, water originating from the polders probably does increase the copper concentration in Lake Tjeukemeer. It is shown that surface run-off was more likely to cause this elevation than leaching, though during laboratory experiments leaching occurred under certain conditions in the upper layer of the soil column. Sediment samples from the lake were subjected to sequential extractions and to extractions using EDTA. It was concluded that release of copper from the sediment to the water phase is not likely to be an important process.
Chapter 3: ultrafiltration
Ultrafiltration was used to gain an insight into the size of the copper species present in Lake Tjeukemeer. Interpretation of ultrafiltration results is not simple because of several potential artefacts, among which contamination. Free copper, having a size of about 0.6 nm, will be found in the fraction smaller than 5 nm, the smallest pore size which was used, together with other species smaller than 5 nm. Though some samples appeared to have been contaminated, a critical examination allowed drawing of some conclusions. About half of the dissolved copper concentration was found in the fraction smaller than 5 nm (typically a few tens of nM), implying that at least half of the total copper was present in complexes.
Chapter 4: voltammetric techniques
Two voltammetric techniques (anodic and cathodic stripping voltammetry) were applied to obtain information on the fraction of electrochemically labile copper.
Anodic stripping voltammetry (ASV) is an often used technique in speciation studies, but the results are not easy to interpret. Under the conditions applied during the measurements (pH, deposition time, buffer) all inorganic and some organic copper species can be assumed to be detected ("ASV-labile"). The concentration of ASV- labile copper was usually below 2 nM, and often not detectable at all in spite of the long deposition time. This means that at most only a few percent of copper was ASV-labile, and therefore by far the largest part must have been present in organic complexes. At natural pH the concentration of ASV-labile copper must have been even lower. Although most copper was concluded to be organically complexed, no correlations were found with parameters representing organic carbon.
Cathodic stripping voltammetry (CSV), using catechol as synthetic ligand, was applied under non-equilibrium conditions. However, the CSV-labile copper concentration found, was often more than 100% of the dissolved copper concentration. Based on additional experiments, it was concluded that the concentration of organic material in the lake probably was too high to allow useful application of this technique to Lake Tjeukemeer.
Chapter 5: ligand competition techniques
To assess the concentration of free copper itself, two ligand competition techniques were applied.
The MnO 2 method was one of these. MnO 2 was added to standards of known composition, allowing the determination of the copper adsorption as a function of the free copper concentration. Three methods were used to process these adsorption data:
1) Langmuir behaviour was assumed and the linearisation method recommended by Van den Berg (1982a) was applied;
2) a different linearisation suggested in this thesis was used to process Langmuir data;
3) finally a linear model supposing copper adsorption (Γ) to be proportional to the free copper concentration; this may be considered an isolation of the first, nearly linear part of a Langmuir curve.
The obtained values for the free copper concentration were valid only in the presence of MnO 2 , but they could be converted to values valid in the absence of MnO 2 after a series of assumptions. The van den Berg linearisation method did not yield satisfactory correlation coefficients; it was reasoned that this linearisation method should be used for higher free metal concentrations than those used here. The second linearisation method and the linear model gave quite satisfactory correlations. Contamination in several cases hindered the calculation of free copper concentration in lake water samples. In other cases, the average of the free copper concentration varied from 0.013 nM (second linearisation) via 0.018 nM (first linearisation) to 0.050 nM (linear model), all in the presence of MnO 2 . These values were a few tens of percents higher when converted to the situation without MnO 2 . The free copper concentrations found were not systematically correlated with the concentration or uv-absorbance of organic matter. It was calculated that more than 95% of copper in Lake Tjeukemeer was organically complexed when the linear model was used; in the case of Langmuir linearisations this percentage was 99.
A voltammetric version of the SEP-PAK C 18 adsorptionmethod (Sunda & Hanson, 1987) was developed here. Labile copper was measured at a rotating mercury film electrode using differential pulse ASV in two copper addition series; one also had EDTA included. Plotting the labile versus the total copper concentration allowed quantification of the concentration of the
copper-EDTA complex, which in turn allowed the calculation of free copper concentration. Since the ASV-signal in lake water samples without copper addition increased upon addition of EDTA, the free copper concentration could not be determined without adding copper. However, it was argued that extrapolation to these conditions was reasonable. Thus for free copper
concentration a value between 10 -13and 10 -12M was obtained. The corresponding concentration of organically complexed copper was calculated to be more than 99% of the dissolved copper concentration. It was concluded that this new version of a ligand competition technique appears to be promising.
Chapter 6: copper titrations
Copper titrations were used to obtain information on copper complexation in Lake Tjeukemeer. Several type of titrations and data processing were applied.
Using a cupric ion selective electrode (ISE), the copper complexation capacity in Lake Tjeukemeer was monitored fortnightly during three years. Both Van den Berg-Ruzic-Lee and Scatchard analysis showed ligand concentrations of several tens of μM. For Van den Berg-Ruzic-Lee analysis log(K) varied from about 4.5 to about 7.5, for Scatchard analysis from about 4.5 to about 10. Affinity spectra yielded log(K) values ranging from 4.5 to 8, and suggested at least three groups of binding sites. The highest log(K) values coincided with relatively low humus concentrations and blooms of Cyanobacteria.
One titration curve was subjected to a more detailed look. Using Van den Berg-Ruzic-Lee analysis it was shown that the assumption of two groups of ligands; is necessary to mimic the titration data adequately. To mimic the Scatchard curve of this titration satisfactory, the presence of a third group of ligands had to be assumed. It was shown that both Van den Berg-Ruzic-Lee and Scatchard analysis had to be applied to obtain adequate complexation parameters.
obtaining information about copper complexation at natural pH could not be done directly, because copper precipitated at high pH- values. Therefore, copper titration data of lower pH-values were fitted using a pH-dependent Freundlich isotherm, and extrapolated to the natural pH. At low pH (3, 4) more active sites were assessed than at higher pH (5, 6). The free copper concentration at pH = 7.55 (the pH used for the experiments described in chapter 5) was calculated to be around 10 -10M; when only the more active sites were assessed, this value was about 10 -14M.
Because of the relatively high detection limit of an ISE, ASV, in combination with Scatchard analysis, was used to gain insight into the complexation at low copper additions (up to 157 nM). The log(K) at pH = 5 was found to range from 8 to 9. virtually all copper was organically complexed. If these titrations were performed in ultrafiltrates, it could be shown that the apparent ligand concentration decreased with pore size, while the log(K) increased with pore size, i.e. the smallest ligands form the strongest complexes with copper.
Chapter 7: modelling
An equilibrium model was used to calculate the speciation in four samples. Meaningful calculations can only be done if sufficient knowledge is available about the species which are present and the corresponding equilibrium constants. The data which were used as input for the model were therefore subjected to a critical examination.
When only in organic complexation was included in the calculations, the free copper concentration varied from 10 -12to 10 -10M, representing at most 2% of the dissolved copper concentration. Copper was mainly present as carbonate complexes, while a few percent of copper was present as copper hydroxide complexes.
When the pH-dependent fit of copper titration data (as described in chapter 6) was used to model complexation of copper by organic ligands, the free copper concentration decreased, ranging from 10 -12to 10 -11M. More than 90% of copper was calculated to be organically complexed, except for one very alkaline sample which displayed a considerable fraction of carbonates. Hydroxides were of minor importance.
When the fit using only the more active site was used, the free copper concentration ranged from 10 -16to 10 -14M. All copper (i.e. more than 99.98%) was computed to occur in organic complexes.
Chapter 8: speciation of copper in relation to its bioavailability
Bioassays were performed in water samples from Lake Tjeukemeer, two other lakes and a polder site, to assay the biologically available fraction of copper. A green alga, Scenedesmusquadricauda, was grown in batch cultures to which fixed concentrations of phosphorus and nitrogen, and varying concentrations of copper were added. Electrochemically labile copper was determined in samples with the same copper additions. For Lake Tjeukemeer, equilibrium calculations were performed as well, as described in chapter 7. For all samples, adding 0.16 μM of copper resulted in a higher biomass in the stationary phase; for Lakes Tjeukemeer and IJsselmeer this increase was even significant. Copper additions of 1.6 μM and higher displayed growth inhibition. Growth inhibition was well correlated with electrochemically labile and computed free copper, but since these latter two parameters were almost proportional to the total copper concentration, these correlations are not very meaningful.
Scenedesmus quadricauda was also grown in synthetic media of known composition. The copper and EDTA concentrations were varied to obtain a wide range of free copper concentrations. Speciation calculations showed that not only the free copper concentration varied, but also the free cobalt and free zinc concentrations, because of competition of these three metals for EDTA. This competition made it difficult to draw unambiguous conclusions. However, it was clearly shown that the free instead of the total concentrations of copper, cobalt and zinc determined the bioavailability. Cobalt and zinc were not toxic in our assays, but copper became toxic above free copper concentrations of about 10 -10.5M. At high EDTA concentrations, some element became limiting, probably either copper or cobalt or zinc. The concentration at which copper became limiting could not be quantified exactly because of the competition with cobalt and zinc, but must have been lower than 10 -12.5M.
The growth stimulation which was found in Lakes Tjeukemeer and IJsselmeer when copper was added, therefore implies that the free copper concentrations in those lake water samples must have been lower than 10 -12.5M.
In natural samples higher values for the biomass were obtained than in synthetic media. It was not understood why.
Chapter 9: looking back and ahead
Looking back upon this research project, it is concluded that the results obtained by the various methods were at least qualitatively consistent. The techniques which claim to determine only free copper, displayed a wide range of free copper concentrations, but all indicated that only a very small fraction of the dissolved copper concentration was present as free copper. The relation between the copper speciation and bioavailability could be determined for one algal species.
To allow this type of research on a routine basis, more knowledge is needed about the interaction of metals with organic matter. This work clearly demonstrated the importance of those interactions. Ligand competition techniques may be most promising. Better knowledge of inorganic solutions is required to obtain more accurate results.