Field trial of ion-exchange resin columns for removal of metal contaminants, Thala Valley Tip, Casey Station, Antarctica

Cold Regions Science and Technology 48 (2007) 105 – 117 www.elsevier.com/locate/coldregions

Field trial of ion-exchange resin columns for removal of metal contaminants, Thala Valley Tip, Casey Station, Antarctica Penny Woodberry a , Geoff Stevens a,⁎, Kathy Northcott a , Ian Snape b , Scott Stark b a

Particulate Fluids Processing Centre, Department of Chemical and Biomolecular Engineering, University of Melbourne, Melbourne, Australia b Australian Antarctic Division, Hobart, Australia Received 15 March 2006; accepted 15 August 2006

Abstract A field trial for removal of dissolved metal contaminants in water from an abandoned waste disposal site, using ion-exchange columns of an iminodiacetic acid chelating ion-exchange (IDA) resin, was conducted at Casey Station, Antarctica. An on-site monitoring technique, employing 3M Empore chelating disks for preconcentration for analysis by atomic absorption spectrometry (AAS), was also trialled for measurement of dissolved metal concentration in the treated water. The field trial indicated that the ionexchange columns of Amberlite IRC748 were suitable for the application, with concentrations of contaminant metals in treated waters successfully reduced. The observed order of selectivity for contaminant metals was markedly different to that observed in previous laboratory tests; the presence of organic ligands and colloidal iron in the natural waters on-site interfered with the retention of Cu and Fe, reducing selectivity for these metals. The on-site monitoring technique was considered capable of producing results of sufficient accuracy and reproducibility to identify that the ion-exchange columns were effectively reducing metal contaminant concentrations, and to identify breakthrough of the ion-exchange columns. Concentrations of Cd, Ni and Zn corresponded well with the dissolved metal fraction measured by ICP–MS, although Pb analysis was less precise. However, interference from organic ligands and colloids probably caused underestimation in the measurement of these metals. The on-site monitoring technique was useful for providing rapid feedback during clean-up at this remote site. However, the interferences would require careful evaluation in order to relate the measurements to water quality standards or objectives. © 2006 Elsevier B.V. All rights reserved. Keywords: Antarctica; Ion-exchange resin; Water treatment; Metal contaminants

1. Introduction Amberlite IRC748 (Rohm and Haas Ltd), an iminodiacetic acid (IDA) chelating cation-exchange resin, is being assessed for the removal of dissolved contaminant metals from the surface waters discharged from abandoned coastal waste disposal sites in cold regions. One such site is the Thala Valley Tip, an abandoned waste disposal site associated with Old Casey Station, Antarc⁎ Corresponding author. 0165-232X/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.coldregions.2006.08.001

tica. Investigations have indicated that metal contamination present at the tip site is being dispersed into the marine environment of adjacent Brown Bay, largely via surface drainage, during the summer melt (Snape et al., 2002; Scouller et al., 2002; Stark et al., 2003). Thala Valley is being developed as a case study for the investigation and remediation of coastal waste disposal sites in cold regions. During the summer melt, the temperature of surface waters is typically between 1 and 4 °C, therefore low temperature is an important sitespecific factor which will affect the uptake of metals by an

106

P. Woodberry et al. / Cold Regions Science and Technology 48 (2007) 105–117

ion-exchange resin. Another site-specific factor is variable salinity of the surface waters. The site has been subject to periodic inundation by seawater during extreme high tides, therefore surface waters may potentially have high salinity at the point of discharge to Brown Bay. Clean-up of the Thala Valley Tip was undertaken during the 2003–04 season at Casey Station. Water management formed an integral part of clean-up operations to limit dispersal of contaminants mobilised into meltwaters during excavation and earth-moving activities. Contaminated water was treated using a water treatment plant prior to discharge to Brown Bay. This study considers the operation of the ion-exchange columns of Amberlite IRC748 for removal of dissolved metals during clean-up activities as a field trial. Our previous laboratory batch equilibrium studies (Woodberry et al., 2006) have indicated that dissolved metal contaminants can be successfully removed by Amberlite IRC748 in the presence of the main seawater matrix ions (Na, Ca and Mg) at 4 and 20 °C, and that increased salinity actually lowers the selectivity of the resin for Ca and Mg. Our previous laboratory dynamic flow studies (Woodberry et al., 2006) have also indicated that dissolved metal contaminants can be successfully removed by Amberlite IRC748 under variable salinity and low temperature conditions, and that the order of selectivity for the contaminant metal ions under field conditions (10% seawater and 4 °C) would be expected to be: Cu N Pb N Ni N Zn N Cd. As part of this field trial, an on-site monitoring technique for measuring dissolved metal contaminants

was also investigated. On-site monitoring of water quality during contaminated site remediation is an integral component of most remediation plans. This involves careful measurement of chemical parameters at very low concentrations, often using expensive techniques such as ICP–MS. Hence preconcentration techniques which are simple and reproducible can be very useful in situations where results are required quickly and laboratory infrastructure may be limited. 2. Water treatment plant 2.1. Ion-exchange columns The water treatment plant (WTP) was used to treat contaminated meltwaters retained in the tip site during excavation activities and drained from waste containers (excavated materials contained large quantities of ice and snow which melted during the summer). A multistage treatment process was employed for the removal of particulate material and dissolved metal contaminants, contained within a modified sea-container for ease of mobility and deployment (Northcott et al., 2007). The layout of the WTP is presented in Fig. 1. The water to be treated enters the mixing tank into which chemicals are dosed for the removal of particulate materials in the flocculation tanks and the separator. The water is then pumped from the header tank into the ion-exchange columns for the removal of dissolved metal contaminants. The ion-exchange columns were originally designed to be operated as two trains of columns in parallel; each

Fig. 1. Layout of water treatment plant (WTP).

P. Woodberry et al. / Cold Regions Science and Technology 48 (2007) 105–117 Table 1 Metals concentrations in waters sampled from the Thala Valley a Metal

Total metals (μg L− 1)

Copper 442.6 Iron 19 968.2 Manganese 1686.5 Lead 1476.4 Zinc 3044.7 a

Particulate contribution (%)

NEPM (1999) guidelines (μg L− 1)

60 95.3 10.8 98.9 29.8

5 N/A N/A 5 50

Data sourced from Snape et al. (2001).

train comprising two columns in series. Before entering the ion-exchange columns, water from the settler is pumped through bag filters for removal of particulates down to 1 μm. However significant quantities of particulate material, comprised of colloidal ferric hydroxides, still entered the columns potentially damaging and clogging the resin bed. Contaminated meltwaters can contain high concentrations of ferric iron, and this problem is significantly exacerbated by the dosing with ferric chloride as a coagulant in the particulate removal section. To address this problem, two columns were converted to sand filters. Therefore operation of the section for removal of dissolved metals involved three columns in series: the first in the series filled with 50 L of sand, the second and third in the series filled with 50 L of ionexchange resin (in the sodium form). 2.2. On-site monitoring On-site monitoring was required to determine inlet and outlet metal concentrations from the ion-exchange columns to confirm the successful removal of contaminant metals and to detect column breakthrough. An assessment of the treated water quality was also attempted. There are currently no water quality guidelines specifically for Antarctica. As a benchmark, the Australian Groundwater Investigation Levels (GILs) specified in the National Environment Protection (Assessment of Site Contamination) Measure 1999 (the NEPM) for the protection of marine aquatic ecosystems were used. Table 1 gives typical metals concentrations (particulate and dissolved fractions) for waters sampled from the Thala valley (Snape et al., 2001). Dissolved metal concentrations in surface waters discharged from the waste disposal site range from μg L− 1 to nearly mg L− 1, and GIL target concentrations for many metal contaminants are in the μg L− 1 range. Salinity typically ranges from near freshwater to salinity approaching that of seawater during extreme tides and marine inundation. Chemical analysis of trace concentrations in such a variable saline matrix would normally be undertaken by

107

high resolution ICP–MS. Typically, samples would be diluted prior to analysis to overcome some of the matrix effects inherent when measuring saline water. However, Casey Station has very limited laboratory and analytical facilities, and the most advanced instrument on site is an atomic absorption spectrometer (AAS). Many Antarctic research stations have an AAS, but few would have instruments with appreciably better detection limits. There are a range of techniques which can be used for preconcentrating trace metals sufficiently for analysis by atomic spectrometry; many involve retaining metals on a chelating resin or some other form of adsorption media (sometimes with the addition of a complexation agent to aid adsorption) (Warnken et al., 1998; Wells and Bruland, 1998; Anthemidis et al., 2001). 3M Empore chelating extraction disks were selected for preconcentration of samples in the field trial, as they are readily available commercially. The Empore disks can also be used at high flow rates, practicable for treating large sample volumes, and are suitable for use with saline samples. The Empore disks utilise Chelex-100, an IDAfunctional chelating cation exchange resin, similar to the resin used in the ion-exchange columns. Ion-exchange resins containing the IDA functional group have been investigated and successfully employed for preconcentration of trace metals and separation from matrix ions in seawater to aid in analysis (Nicolai et al., 1999; Jimenez et al., 2002; Grotti et al., 2002). Prior to the preconcentration tests, an 0.01 mol L− 1 ammonium acetate buffer solution (pH 5.4) was added to the aqueous samples. The glacial acetic acid and ammonia used in preparation of the buffer solution, along with all other reagents were analytical grade. The chelating disk

Fig. 2. Filter apparatus for preconcentration of on-site monitoring samples.

108

P. Woodberry et al. / Cold Regions Science and Technology 48 (2007) 105–117

(3M Empore) was preconditioned with 50 mL of 3 mol L− 1 nitric acid and 50 mL of MilliQ water (Millipore Corporation) before conversion to the ammonium form with 100 mL of 0.1 mol L− 1 ammonium acetate. Buffered samples were passed through a series of in-line filter units at a flow rate of 40 mL min− 1 using a peristaltic pump (Fig. 2). The first and second units comprised a pair of glass fibre filters (Whatman GF/C) and a 0.45 μm membrane filter (Sartorius, cellulose nitrate), respectively. This allowed filtration to 0.45 μm (the pore-size of the chelating disk) prior to passage through the chelating disk contained in the third and final in-line filter unit. Filtration was required owing to the significant quantities of particulate material b 5 μm in diameter in the samples. The disk was eluted with 10 mL of 3 mol L− 1 nitric acid. A single elution was used in order

to maintain a high concentration factor to achieve the lowest possible detection limit for the samples. After dilution to 1.5 mol L− 1 nitric acid, the solution was analysed for Ni, Cu, Zn, Cd, Fe and Pb by AAS. 2.3. Sampling The WTP was operated approximately 12 h per day, 5 to 6 days per week (weather permitting) from the end of November 2003 until mid January 2004. Samples for laboratory analysis were collected on a daily basis (approximately every 10 kL of water treated) from the inlet and outlet of the ion-exchange columns, and also from four sample points within the resin bed of the first ionexchange column in the series. The 50 mL samples were filtered to 0.45 μm and preserved with 50 μL of nitric

Fig. 3. Operating conditions for ion exchange columns: (a) temperature, (b) pH, (c) flow rate, (d) seawater matrix ions in column inlet water, and (e) capacity.

P. Woodberry et al. / Cold Regions Science and Technology 48 (2007) 105–117

acid before freezing for storage and transport to Hobart. Analysis of contaminant metals (Cd, Cu, Pb, Ni and Zn) and seawater matrix ions (Na, Ca and Mg) was undertaken by inductively coupled plasma–mass spectrometry (ICP–MS) using a Finnigan Element sector field instrument at the Central Science Laboratory, University

109

of Tasmania. These samples were used to assess the performance of the ion-exchange columns. Samples were also collected from the inlet (1 L) and outlet (10 L) of the ion-exchange columns on a weekly basis for on-site monitoring purposes i.e. for preconcentration and analysis by AAS at Casey Station.

Fig. 4. Inlet and outlet dissolved concentrations for contaminant metals: (a) Cd, (b) Cu, (c) Fe, (d) Pb, (e) Ni, (f ) Zn.

110

P. Woodberry et al. / Cold Regions Science and Technology 48 (2007) 105–117

Table 2 Metal species present in solution in tip waters modelled by PHREEQC (8 °C, equilibrated with atmospheric carbon dioxide and oxygen) pH 5.4

7.0

2+

Cd (amount precipitated = 0%) Free ion Inorganic complex a Cu2+ (amount precipitated = 0%) Free ion Inorganic complexa 3+ Fe (amount precipitated ∼100%, as FeOOH) Inorganic complexa Pb2+ (amount precipitated = 0%) Free ion Inorganic complexa Zn2+ (amount precipitated = 0%) Free ion Inorganic complexa

86.5 13.5

86.4 13.6

99.8 0.2

37.6 62.4

99.9

100

95.4 4.6

67.1 32.9

99.8 0.2

98.6 1.4

The WATEQ4F database was used for calculations (Ball and Nordstrom, 1991). a Combined percentages of metals complexed by inorganic ligands, e.g. hydroxides and carbonates.

Corresponding samples collected for analysis in Hobart to assess the performance of the ion-exchange columns were used to validate the on-site monitoring technique.

not required, suggesting that the tip waters were naturally buffered. The target flow rate for the ion-exchange columns was 10 bed volumes per hour or equivalent to 1000 L h− 1. Over the period of operation, the flow rate was less than this target level at an average of approximately 900 L h− 1; flow rate was limited by the capacity of the particulate removal section of the WTP. Pressure drops over the sand filters and ion-exchange columns were also monitored and used as indicators for the replacement of bag filters and sand. The sand filters required changing on a weekly basis (approximately every 50 kL to 60 kL of water treated), and the bag filters every one to two days of operation (approximately every 10 kL to 20 kL of water treated). Na, Ca and Mg concentrations (Fig. 3d) in the inlet waters indicated that the salinity spiked in the first 20 kL of water treated and again at 150 kL of water treated. Apart from these spikes, the salinity was generally equivalent to 5–10% seawater. During the season metal concentrations in feed water were such that resin loading was relatively constant (Fig. 3e). At around 150 kL of water treated, the rate of metals uptake suddenly increased and the resin was rapidly exhausted. 3. Characterisation of tip waters

2.4. Water treatment plant operational data 3.1. Concentrations of metals Temperature, pH and flow rate were recorded hourly during operation of the water treatment plant. For the majority of the period of operation, temperature of the column outlet water fluctuated between 6 °C and 10 °C (Fig. 3a). This is warmer than normally expected for meltwaters at the site (2 °C to 4 °C). The higher temperature is in part due to warming of the water as it passes through the WTP (temperature increase of up to 2 °C observed), and in part to the effect of heat-tracing on the holding container for water to be treated; heattracing was required to prevent the water from freezing overnight. However, water temperatures in the tank were consistently below 10 °C, lower than temperatures that would be typical of temperate regions. The pH of column waters was reasonably constant over the period of operation, fluctuating from pH 6.8 to 7.3, with the average being approximately pH 7. Apart from the first day of operation (outlet pH elevated due to protonation of the resin), little change in pH between the column inlet and outlet water was observed during operation. Little pH change was also observed in the particulate section even upon dosing with ferric chloride; sodium hydroxide dosing for pH control was generally

The main dissolved metal contaminants identified to be present in meltwaters treated by the ion-exchange columns were Cd, Cu, Pb, Ni and Zn, with some colloidal iron present. Their concentrations in the inlet water to the columns can be separated into three distinct periods of water treatment during clean-up operations in Thala Valley (Fig. 4a–f ). The first period (0 to 50 kL treated) was at the beginning of the summer when small Table 3 Percentage removal of metals by IDA in a batch reaction modelled by PHREEQC (pH 5.4, 8 °C, equilibrated with atmospheric carbon dioxide and oxygen) IDA/metal molar ratio % Removed

Cd Cu Fe Pb Zn

IDA in excess

→

20

1

2

IDA below capacity 0.7

0.5

0.2

99.14 86.54 17.17 4.45 1.48 0.29 100.00 100.00 99.99 99.97 99.12 99.58 100.00 99.41 73.84 35.72 13.51 2.47 99.98 99.72 91.98 72.04 45.48 13.55 99.98 99.60 89.04 64.61 37.15 10.00

P. Woodberry et al. / Cold Regions Science and Technology 48 (2007) 105–117

volumes of meltwater were percolating through contaminated material within the tip-site. Levels of dissolved metal contamination generally exceeded the GIL target concentrations. This period corresponds with the first observed spike in salinity (Fig. 3d). As the melt progressed (50 to 150 kL treated), the quantities of meltwater increased and the contaminated water mixed with ‘clean’ water flowing into the collection pond from non-contaminated areas within the Thala Valley. Consequently, levels of dissolved metal contamination were diluted (and salinity decreased), however levels of Cu still exceeded the GIL target concentration. At the height of the melt (200 kL treated), containers filled with frozen waste from the tip-site started to thaw and the majority of water treated comprised contaminated water drained from these containers. The concentration of dissolved metal contaminants in this water generally exceeded the GIL target concentrations. The second observed spike in salinity corresponds to the beginning of treatment of waters from the waste containers. The change in rate of the resin capacity usage also corresponds to the beginning of treatment of waters from the waste containers. 3.2. Metal species in tip waters The inorganic metal species likely to be present in tip waters were investigated using the geochemical computer model PHREEQC (U.S. Geological Survey). Typical concentrations and conditions (i.e. temperature and alkalinity) were specified and the PHREEQC model was used to predict the proportions of metal species at pH 7.0 (typical pH in ion-exchange columns) and pH 5.4 (samples acidified to pH 5.4 for the on-site monitoring technique), and the percentage removal of metals from a multi-component solution by IDA (stability constants for IDA–metal complexes sourced from Smith and Martell (1976). The results of the PHREEQC modelling, summarised in Tables 2 and 3, aided in the understanding of the results of the field trial by showing how the inorganic and organic species of the metals present in the aqueous phase affected the uptake of metals onto the resin. 4. Results: ion-exchange columns

111

The ion-exchange columns substantially reduced the concentrations of dissolved metal contaminants in contaminated meltwaters, despite the low temperature and salinity levels, confirming the results of previous laboratory studies. The high concentrations of metals in the meltwaters from the waste containers (from 150 kL to 200 kL water treated) rapidly used up the capacity of the resin and breakthrough of metal contaminants began to occur down through the column bed. (Data collection for the field trial was concluded once operational monitoring identified that breakthrough of the columns had occurred, however, the ion-exchange resin was replaced to continue effective treatment of contaminated water.) Although concentrations of Cu were successfully reduced to below the GIL target concentration, the effluent concentration levels achieved were consistently higher than those for the other contaminant metals. In natural waters, Cu may be present in the form of hydroxy or carbonate complexes (Langmuir, 1997). If a significant proportion of Cu is present as stable inorganic complexes, retention of the IDA resin in the chelating disks may be affected. It is also possible that natural organic matter present in meltwaters will compete with IDA resin for Cu. Although Antarctic soils are characteristically low in organic matter (Beyer and Bolter, 1998), those at the waste disposal site were found to be almost 10 times that found in background soil samples taken from the Casey station area. This is most likely a result of dumping of domestic waste, such as kitchen scraps, and construction materials such as wood.(Northcott et al., 2003). Pb was the only metal for which dissolved concentrations were not substantially reduced by the ionexchange columns, however the inlet waters were very low in Pb concentration for most of the season. By the time higher Pb concentrations were being treated (waste container meltwater with much greater levels of dissolved lead), breakthrough of the ion-exchange columns was occurring. It was therefore not possible to assess the performance of the ion-exchange resin for the removal of Pb, based on these results. A study of the uptake of metal ions from soil leachates using an iminodiacetic acid resin by (Manouchehri and Bermond, 2006), found that lead uptake was inhibited by high concentrations of calcium ions in solution, more so than for other metals species such as copper and cadmium.

4.1. Column outlet concentrations 4.2. Breakthrough curves Dissolved contaminant metal concentrations of the inlet and outlet water of the ion-exchange columns during operation over the 2003–04 season are presented in Fig. 4a to f.

Dissolved contaminant metal concentrations at each sampling point within the resin bed and at the outlet of the ion-exchange columns during operation over the

112

P. Woodberry et al. / Cold Regions Science and Technology 48 (2007) 105–117

Fig. 5. Breakthrough curves for contaminant metals: (a) Cd, (b) Cu, (c) Fe, (d) Pb, (e) Ni, (f) Zn.

2003–04 season are presented in Fig. 5a to e. The ion exchange columns were fitted with four sample ports, spaced at equal distances along the 1 m depth of the resin bed. The results of sampling from the sample

points and outlet of the columns indicate that breakthrough occurred for Zn and Cd at the top two sample points only (to half way down the resin bed). Both Ni and Cu achieved breakthrough at all sample points, at

P. Woodberry et al. / Cold Regions Science and Technology 48 (2007) 105–117 Table 4 On-site monitoring technique method blank and calibration errors for column outlet samples (μg L− 1) Errors a

Ni Cu Zn Cd Pb Fe

Method blank

Calibration

0.02 0.01 0.61 0.02 0.01 0.23

0.02 0.003 0.01 0.01 0.03 0.01

c

Detection limit b

NEPM GIL

0.09 0.03 2 0.05 0.03 0.6

15 5 50 2 5 NA

a

Based on S.D. Based on method blank ± 95% confidence interval, df = 2. c Estimated for measurement of the NEPM GIL by the AAS (50 μg L− 1 for Fe). b

lower concentration with depth within the column. The breakthrough of Fe occurred at all sample points nearly simultaneously. This may be due to the sand filters upstream reaching capacity and allowing some colloidal iron into the columns. Similar to Fe, all sample points showed breakthrough of Pb at approximately the same time. 5. Results: on-site monitoring 5.1. Detection limits and reproducibility Dissolved metal concentrations in outlet water from the ion-exchange columns were in the very low μg L− 1 range, and therefore likely to be approaching the detection limit of the on-site monitoring technique. The calibration error of the AAS was estimated for its contribution to the overall error. The method blanks generally exceeded instrument detection limits and variability, and thus dictated the overall method detection limit, as summarised in Table 4. It is noted that the analytical data available is limited and for improved error estimates, further data would need to be collected. The most significant contamination problem with the technique was for Zn. The source of the contamination was identified to be the filters used for prefiltration of the samples. It is possible that this source of contamination could have been reduced if pre-treatment of filters (for example, dilute acid-leaching) had been undertaken. This contamination source leads to a fixed quantity of Zn in the results. Therefore as the inlet samples required better prefiltration than the column outlet samples, the effect of filter contamination on Zn concentration in column inlet samples was found to be 10 times higher than in column outlet samples. This resulted in greater errors and detection limits for column inlet samples.

113

Dissolved metal concentrations for the triplicate column inlet sample (Table 5) indicated that the onsite monitoring technique provided acceptable reproducibility for the application (within 10% of the average of the three replicates). Most metals were within 6%, with the exception of Pb. In the case of Pb, poor reproducibility was attributed to a number of factors including: low concentrations approaching the method detection limit, the effect of calibration drift and possible inhibition of ion exchange due to varying amounts of calcium ions in the feed water (Manouchehri and Bermond, 2006). 5.2. Comparison of on-site monitoring and ICP–MS results The dissolved metal concentrations measured by the on-site monitoring technique for Ni, Zn and Cd were generally consistent with the ICP–MS results of the validation samples (Fig. 6a, c and d). Underestimation by the on-site monitoring technique was observed when concentration levels were high, such as for column inlet samples on the 1st of December 2003 and the 5th of January 2004. This suggests that for some samples the retention of metals by the chelating disks was limited by the capacity of the Chelex-100 resin. Copper concentration was consistently underestimated by the on-site monitoring technique, by as much as 20 μg L− 1, in both column inlet and column outlet samples (Fig. 6b). The low retention of Cu by the extraction disks is consistent with the field trial of the ion-exchange columns, where it was observed that outlet concentrations achieved for Cu were not as low as those achieved for other contaminant metals. Iron concentration was generally underestimated by the preconcentration technique, by as much as 100 μg L− 1, particularly at the high concentration levels measured on the 1st of December 2003 and the 5th of January 2004 (Fig. 6f ). This may be due to the presence of a significant proportion of Fe in colloidal form. Significant quantities of Table 5 Triplicate dissolved metal concentrations for inlet column sample (μg L− 1) 12-Dec-03

Ni Cu Zn Cd Pb Fe

Replicate 1

Replicate 2

Replicate 3

Average

4.00 8.00 168 0.40 1.40 26.00

4.00 8.20 175 0.40 1.20 24.00

3.60 8.20 160 0.40 0.40 26.00

3.87 8.13 168 0.40 1.00 25.33

114

P. Woodberry et al. / Cold Regions Science and Technology 48 (2007) 105–117

Fig. 6. Comparison of preconcentration technique and validation analysis by ICP–MS, for operational monitoring of ion-exchange columns: (a) Ni, (b) Cu, (c) Zn, (d) Cd, (e) Pb, (f ) Fe.

P. Woodberry et al. / Cold Regions Science and Technology 48 (2007) 105–117

suspended Fe, believed to be ferric oxyhydroxides (FeOOH), were observed in water being treated by the ion-exchange columns during the field trial, a problem exacerbated by the dosing of ferric chloride in the water treatment stage (removal of particulates) prior to the ionexchange columns. While large quantities of this material were filtered out by bag (10 μm) and sand filters before entering the ion-exchange columns, outlet waters still contained colloidal Fe. Much of the remaining colloidal Fe would have been removed during the filtration step in the preconcentration technique, however, a proportion is also likely to have passed through the chelating disk. Owing to the short contact time between the solution and disk, these colloids would not have been fully adsorbed, resulting in an underestimation of the Fe concentration in comparison to the validation samples. The Pb concentration was underestimated by the onsite monitoring technique on the 19th of December and the 2nd of January, and overestimated for the rest of the samples collected (Fig. 6e). As noted previously the reproducibility of the on-site monitoring technique for Pb was relatively poor. 6. Discussion The breakthrough data for the ion-exchange columns was used as an indication of the selectivity of the resin for the different contaminant metals; less breakthrough indicating higher selectivity. The order of selectivity observed for the field trial was: Cd∼ZnNNi∼CuNFe which is markedly different to the order of selectivity observed in the laboratory column breakthrough tests (Woodberry et al., 2006). The order of selectivity observed for the contaminant metals at 4 °C for the non-buffered experiments was: 0% seawater Cu N Pb N Ni N Cd N Zn 10% seawater Cu N Pb N Ni N Zn N Cd Note that Pb is not shown in the field trial sequence due to its low concentrations in the dissolved phase and hence lack of reproducibility of results. A low retention of Cu was observed in both the field trial of the ion-exchange columns and in the on-site monitoring, and it was suggested that the presence of inorganic or organic ligands interfered with the retention of this metal by the IDA resins. The PHREEQC model predicts that while a significant proportion of Cu is present as inorganic complexes at pH 7.0, Cu will be

115

present predominantly as the free ion under the pH 5.4 conditions of the on-site monitoring technique (Table 2). Therefore the presence of an organic ligand in meltwaters is more likely for the low retention of Cu in the ion-exchange columns and the underestimation of Cu concentrations. It is also considered that the presence of an organic ligand may be responsible for the low selectivity of Cu observed in the field trial of the ion-exchange columns compared to the laboratory column breakthrough tests. An acetate buffer was used in some of the laboratory column breakthrough tests to investigate the effect of buffering on the selectivity of the metal. While no effect was observed for the main contaminant metals present in the tip waters, the retention of Cr was significantly inhibited due to interaction with the organic acetate ligand (Woodberry et al., 2006) illustrating that it is possible an organic ligand with a preference for Cu may be inhibiting the retention of this metal in the ion-exchange columns. Due to the low solubility of Fe at the concentrations required for the laboratory column breakthrough tests, the relative selectivity of Fe between the laboratory and the field experiments can not be assessed. However, the low selectivity for Fe in the ion-exchange columns and the poor retention exhibited in the on-site monitoring technique was unexpected as the Fe–IDA complex has a high stability constant (Smith and Martell, 1976). For the on-site monitoring technique, it was proposed that the low retention was due to the presence of a significant proportion of Fe in colloidal form. The PHREEQC model supports this, predicting that at both pH 7.0 and 5.4, most of the Fe present in tip waters is suspended FeOOH, with the small dissolved fraction present predominantly as hydroxide complexes. The PHREEQC model also supports the suggestion that the presence of dissolved Fe as hydroxide species may inhibit the uptake of Fe by the IDA resins. The PHREEQC batch ionexchange model (Table 3) predicted that when IDA is present in excess, the percentage removal of all metals is very high (close to 100%). As the loading of the resin approached capacity, Fe was no longer retained on the resin, despite the high stability constant of Fe–IDA. However, the retention of the other metals species reflected the order of the published stability constants for IDA (Smith and Martell, 1976). The PHREEQC model also predicts that Cd and Zn are present in tip waters predominantly as the free ion, and the high affinity of the IDA resins for these metals observed in the field trial of the ion-exchange columns and on-site monitoring technique indicated little interference by inorganic or organic complexes. Calculations of percentage metal removed at different IDA/metal ratios based on the

116

P. Woodberry et al. / Cold Regions Science and Technology 48 (2007) 105–117

results of the field trial compared to the results of the PHREEQC modelling showed the expected reduction in retention for Fe as compared to Zn. However, the field trials also showed a reduction in the retention of Cu which was not predicted by the PHREEQC model, indicating again that there is another factor affecting selectivity of the resin for this metal. The concentrations of Cd and Pb were too low in the field trial to sensibly perform similar calculations for comparison with the PHREEQC model. It is interesting to note that the order of selectivity based on the stability constants of the IDA–metal complexes (Smith and Martell, 1976) (Cu ∼ Fe N Ni N Pb ∼ Zn N Cd) closely matches the order of selectivity observed in the laboratory column breakthrough tests (Woodberry et al., 2006), using laboratory prepared contaminated waters. This suggests that while the stability constants (generally published for specific solution conditions) and simple equilibrium modelling may be useful as a guide for selectivity in the laboratory, the inorganic and organic metal species present in the natural on-site waters will significantly affect the selectivity illustrating the need for site-specific trials for water treatment technologies. 7. Conclusions The field trial indicated that prior to breakthrough, the concentrations of contaminant metals in meltwaters discharged from the Thala Valley Tip during clean-up operations were successfully reduced by ion-exchange columns of Amberlite IRC748, despite the low temperature and variable salinity levels. The overall order of selectivity exhibited by the resin in the field trials was: Cd ∼ Zn N Ni ∼ Cu N Fe. This order of selectivity was unexpected based on observations of laboratory column breakthrough tests. The main differences were the markedly reduced selectivity for Cu and Fe. It is considered that the presence of organic ligands and colloidal iron, respectively, in the natural tip waters interfered with the retention of these metals by the IDA resin. This interference was also observed in the measurement of these metals by the on-site monitoring technique, which used an IDA resin for concentration of metal contaminants prior to AAS analysis. The on-site monitoring technique, employing the 3M Empore chelating disks for preconcentration prior to analysis by AAS, was considered to be capable of producing results of sufficient accuracy and reproducibility to identify that the ion-exchange columns were effectively reducing metal contaminant concentrations, and to identify column breakthrough. Generally the technique allowed assessment relative to GIL target concen-

trations for Cd, Ni and Zn, although a small bias in Ni and Zn was observed, caused by underestimation of concentrations when the capacity of the disk was exceeded. For Cu, a systematic underestimation of the metal concentrations was observed by the on-site monitoring technique, likely because of the influence of organic ligands. Further empirical data may allow for a systematic correction to be applied to column outlet concentrations, or it may be necessary to apply a sample pre-treatment procedure to destroy any organic complexes. The technique was not considered to be successful for the measurement of Fe or Pb, although the main problem for Pb was considered to be due to the AAS analysis, rather than the preconcentration technique. Acknowledgements The authors would like to acknowledge funding from Australian Antarctic Science Grant 1300 and the support of the Particulate Fluids Processing Centre, a Special Research centre of the ARC at the University of Melbourne. The authors would also like to thank Dr Ash Townsend for ICP–MS analysis at the CSL, University of Tasmania and Mr Nick Graham for AAS analysis at Casey Station. References Anthemidis, A.N., Zachariadis, G.A., Kougoulis, J.S., Stratis, J.A., 2001. Flame atomic absorption spectrometric determination of chromium(VI) by on-line preconcentration system using a PTFE packed column. Talanta 57, 15–22. Ball, J.W., Nordstrom, D.K., 1991. Users' Manual for WATEQ4F, with Revised Thermodynamic Data Base and Test Cases for Calculating Speciation of Major Trace and Redox Elements in Natural Waters. U.S. Geological Survey, Denver CO. 192 pp. Beyer, L., Bolter, M., 1998. Formation, ecology, and geography of cryosols of an ice-free oasis in Coastal East Antarctica near Casey Station (Wilkes Land). Australian Journal of Soil Research 37, 209–244. Grotti, M., Abelmoschi, M.L., Soggia, F., Frache, R., 2002. Determination of trace metals in sea-water by electrothermal atomic absorption spectrometry following solid-phase extraction: quantification and reduction of residual effects. Journal of Analytical Atomic Spectrometry 17, 46–51. Jimenez, M.S., Velarte, R., Castillo, J.R., 2002. Performance of different preconcentration columns used in sequential injection analysis and inductively coupled plasma-mass spectrometry for multielemental determination seawater. Spectrochimica Acta Part B: Atomic Spectrometry 57 (3), 391–402. Langmuir, D., 1997. Aqueous Environmental Chemistry. Prentice-Hall Inc., New Jersey. Manouchehri, N., Bermond, A., 2006. Study of trace metal partitioning between soil–EDTA extracts and Chelex-100 resin. Analytica Chimica Acta 557, 337–343. Nicolai, M., Rosin, C., Tousset, N., Nicolai, Y., 1999. Trace metals analysis in estuarine and seawater by ICP–MS using on line

P. Woodberry et al. / Cold Regions Science and Technology 48 (2007) 105–117 preconcentration and matrix elimination with chelating resin. Talanta 50, 433–444. Northcott, K.A., Snape, I., Connor, M.A., Stevens, G.W., 2003. Water treatment design for site remediation at Casey Station, Antarctica: site characterisation and particle separation. Cold Regions Science and Technology 37, 169–185. Northcott, K.A., Woodberry, P., Snape, I., Stevens, G.W., 2007. Water treatment to prevent contaminant dispersal during remediation of cold regions contaminated sites. Cold Regions Science and Technology 37, 92–104. Scouller, R.C., Snape, I., Stark, S.C., Gore, D.B., 2002. Accumulation and dispersion of terrestrial contaminants in the marine environment: geochemical monitoring to identify potentially impacted sites. In: Snape, I., Warren, R. (Eds.), Proceedings of the Third International Conference on Contaminants in Freezing Ground. Australian Antarctic Division, p. 141. Smith, A.E., Martell, R.M., 1976. Critical Stability Constants. Plenum Press, New York. Snape, I., Riddle, M.J., Jonathon, S.S., Coleen, M.C., King, C.K., Duquesne, S., Gore, D.B., 2001. Management and remediation of contaminated sites at Casey Station, Antarctica. Polar Record 37 (202), 199–214.

117

Snape, I., Gore, D., Cole, C., Riddle, M., 2002. Contaminant dispersal and mitigation at Casey Station: an example of how applied geoscience research can reduce environmental risks in Antarctica. Royal Society of New Zealand Bulletin 35, 641–648. Stark, J.S., Riddle, M.J., Scouller, R.C.L., Snape, I., 2003. Human impacts in Antarctic marine soft-sediment assemblages: correlations between multivariate biological patterns and environmental variables. Estuarine, Coastal and Shelf Science 56, 717–734. Warnken, K.W., Gill, G.A., Wen, L.-S., Griffin, L.L., Santschi, P.H., 1998. Trace metal analysis of natural waters by ICP–MS with online preconcentration and ultrasonic nebulization. Journal of Analytical Atomic Spectrometry 14, 247–252. Wells, M.L., Bruland, K.W., 1998. An improved method for rapid preconcentration and determination of bioactive trace metals in seawater using solid phase extraction and high resolution inductively coupled plasma mass spectrometry. Marine Chemistry 63, 145–153. Woodberry, P., Stevens, G.W., Snape, I., Stark, S., 2006. Removal of metal contaminants from saline waters at low temperature by an iminodiacetic acid ion-exchange resin, Thala Valley Tip, Casey Station, Antarctica. Solvent Extraction and Ion Exchange 24, 603–620.