Determination of metal–humic complexes, free metal ions and total concentrations in natural waters Petra K. Appelblad,ab Douglas C. Baxter*a and Jonas O. Thunbergc aDivision of Inorganic Chemistry, Lulea° University of Technology, S-971 87 Lulea° , Sweden. E-mail: douglas.baxter@km.luth.se bSGAB Analytica, Lulea° University of Technology, S-971 87 Lulea° , Sweden cDivision of Applied Geology, Lulea° University of Technology, S-971 87 Lulea° , Sweden Received 8th February 1999, Accepted 22nd March 1999 A comparison of two systems for the quantitation of metal–humic complexes and free metal ions, consisting of the separation by coupled ion exchange columns followed by detection by inductively coupled plasma mass spectrometry or cold vapour atomic fluorescence spectrometry, is presented.The systems evaluated comprised the serially coupled anion and cation exchangers, Sephadex A-25/Chelex 100 and Dowex 1X8/Chelamine Metalfix. Separation and preconcentration of the species studied were accomplished with both systems, elution being carried out using 2 MHNO3. Total concentrations, metal–humic complex fractions and free metal ion fractions of Al, Ba, Cd, Co, Cu, Fe, Hg, Mn, Pb, Sr, U and Zn in nine natural waters were determined.Statistical evaluation of the data from the two cation exchange materials, including results for additional elements, showed better precision (for Al, Ba, Cr, Cu and Mo) and higher recoveries (Al, Ba, Cd, Fe, Sr and Zn) for Chelex 100 than Chelamine Metalfix for free metal ions. On the other hand, Chelamine Metalfix recovered a significantly greater amount of Ni.The amounts of metal–humic complexes were compared with modelled distributions of these species, and one advantage of the preferred Sephadex A-25/Chelex 100 system is that the elements studied are all correctly classified with respect to their binding strengths to humic substances, which is not the case with the Dowex 1X8/Chelamine Metalfix pair.With the preferred system, metal–humic complexes can be reliably determined, as indicated by the results of equilibrium speciation modelling. However, comparison with the total concentrations showed statistically significant, non-quantitative recoveries of Al, Cu, Hg,Mn, U and Zn from some samples. Thus a combination of speciation and total concentration measurements is required to obtain a complete representation of the distribution of trace elements in natural waters.Humic substances are a general category of naturally occurring at least 94% of the latter neutral complexes were retained on a reversed phase non-polar C18 packing placed upstream of heterogeneous macromolecules, formed through the breakdown of plants and animal tissues by chemical and biological the Dowex 1X8 anion exchanger in a three-column system.The final column contained Chelamine Metalfix, in which the processes. They form both soluble and insoluble complexes with polyvalent cations, depending on the degree of saturation, chelating group, 1,4,7,10,13-pentaazatridecane, is attached to an organic polymer. This resin was investigated by Blain as is well documented by many experimental1–5 and modelling6 –8 studies.Complexation of trace metals by humic sub- et al.,12 who reported that Chelamine Metalfix allows the preconcentration of and matrix elimination for Cd, Cu, Ni, stances may aVect the aquatic system in two ways: by decreasing the toxicity of certain metals and by increasing the Pb and Zn over a broad pH range in saline waters without prior treatment such as buVering.The combination of C18, availability of others.9 KerndorV and Schnitzer10 reported, in one study, the relative strengths of the bonds to humic acid Dowex 1X8 and Chelamine Metalfix permitted the isolation of neutral complexes, anionic complexes and free metal ions, for metal ions. Depending on the stabilities of the bonds, the metal ions may be divided into three groups exhibiting strong, respectively, from fresh water samples.One disadvantage inherent in the use of strong ion intermediate or weak interactions with humic substances. The elements that were found to be most tenaciously bound by exchangers, such as Dowex 1X8, is irreversible sorption to the material. Kim et al.13 attributed this phenomenon to the humus were iron and mercury, while lead, copper and aluminium formed complexes of intermediate strength.Cobalt, high aYnity of phenolic compounds, which would include humic substances, for quaternary amine sites. An eYcient cadmium and zinc displayed weak interactions with humic substances. Chromatographic techniques,1,3–5,11 such as solid weakly basic resin is diethylaminoethylcellulose (DEAE Sephadex A-25), which has a macroporous structure and a phase extraction (SPE), have proven to be eVective experimental approaches to the isolation of humic substances.The hydrophilic matrix to which secondary amines are attached. This resin has been found to be useful for speciation studies advantages of SPE include easy handling of large sample volumes and convenience for automated sample processing.of metal–humic complexes in natural waters.14,15 In this work, Sephadex A-25 was used in combination with Chelex 100, in Groschner and Appriou1 employed Dowex 1X8, a strongly basic resin with quaternary amine functional groups, for the which the functional group is iminodiacetate. Although a great number of chelating sorbents have been used for the preconcen- preconcentration of anionic complexes in aqueous solutions, using 2 M HNO3 as eluent.They reported that the resin tration of trace metals from diVerent matrices,16,17 Chelex 100 has been a popular choice in many analytical applications.18,19 exhibited a high retention eYciency (98.8%) for ethylenediamine tetraacetate-bound copper, but for complexes with fulvic The performance of Sephadex A-25/Chelex 100 was compared with that of Dowex 1X8 Chelamine Metalfix for acid, 8-hydroxyquinoline, nitrilotriacetic acid and tryptophan, the recoveries of the copper species were rather poor.However, natural water samples. To evaluate the eYciency of the two- J. Environ. Monit., 1999, 1, 211–217 211column system for complete recoveries of metal ions, total water to obtain the required concentration.SPEX standard solutions were added to yield a concentration of 100 mg l-1 in concentrations were also determined by inductively coupled plasma mass spectrometry (ICPMS) or cold vapour atomic solution. BuVer solutions of 1 M ammonium acetate were added to adjust the pH. The 2 M HNO3 eluent was made by fluorescence spectrometry (CVAFS) for Hg.The experimental results were also compared with data calculated using equilib- diluting purified HNO3 (p.a., Merck, purified in the laboratory by subboiling distillation in a quartz still ) with Milli-Q water. rium speciation models. The resins used in the experiments were Chelamine Metalfix, (p.a., 40–80 mm, Fluka, Sigma-Aldrich Sweden AB, Experimental Stockholm, Sweden), Chelex 100 resin (analytical grade, 100–200 mesh, Bio-Rad Laboratories AB, Sundbyberg, Instrumentation Sweden), Dowex 1X8 (p.a., 100–200 mesh, Fluka) and DEAE Two diVerent ICPMS instruments (Table 1) were used for the Sephadex A-25 (Pharmacia Fine Chemicals AB, Uppsala, measurements, one equipped with a quadrupole mass filter Sweden).The columns used were made of poly(chlorotrifluor- (ICPQMS; VG Plasma Quad, VG Elemental, Winsford, UK) oethylene) (Malmo� Fluorocarbon, Malmo� , Sweden) and polyand the other with a high-resolution sector field mass spec- (etheretherketone) (SeQuant, Umea°, Sweden), and the dimentrometer (ICPSMS; ELEMENT, Finnigan MAT, Bremen, sions of the columns were 2 mm id×30 mm and 4 mm Germany).For the mercury determinations, a Millennium id×50 mm, respectively.The columns were equipped with Merlin System CVAFS instrument (PS Analytical, polypropylene frits to retain the resin, and were packed by Analys-konsult AB, Lidingo� , Sweden) was employed. using a slurry of ion exchange resin and Milli-Q water which In the preconcentration system, a peristaltic pump ( XV-4, was transferred with a pipette into the columst apply- Alitea, Sweden) was used.For the breakthrough tests, a ing suction. BischoV HPLC pump model 2250 (Coricon, Knivsta, Sweden) Prior to ICPMS determination of total analyte concendelivered sample to a BischoV Lambda 1010 UV detector for trations, all samples were acidified to obtain solutions in monitoring the eZuent absorbance at 254 nm.A Perkin-Elmer 0.14 MHNO3. Column eluates were diluted to obtain threefold Lambda 1 UV/VIS spectrophotometer (Perkin-Elmer, preconcentration in the solutions to be analysed by ICPMS. Norwalk, CT, USA) was used with 10 mm quartz cells for the For mercury determinations, KBr (p.a., Merck, Spa°nga, measurements of the absorbances at 400 nm to facilitate Sweden), KBrO3 (p.a., KEBO Lab AB, Spa°nga, Sweden) and estimation of the amounts of humic substances in the collected HCl (s.p., Merck) were added to the samples to concentrations water samples.of 0.47 mM, 1.43 mM and 1.5% (v/v), respectively. The samples were then allowed to stand for an hour. Immediately Reagents and solutions before analysis, 25 ml of 1.73 M hydroxylammonium chloride (p.a., KEBO Lab AB) was added to destroy residual oxidizing Standard solutions were prepared by diluting 1 mg ml-1 single agent and then a solution of 4% (w/v) SnCl2 (Merck) in 10% element standard solutions (SPEX Plasma Standards, Edison, (v/v) HCl was used to reduce Hg2+ to Hg0.The generated NJ, USA) with high purity water (Millipore, Milli-Q, Bedford, Hg0 was purged from the solution and detected by AFS.MA, USA). 1 M ammonium acetate buVer solutions (made in the laboratory by mixing 57 ml of anhydrous acetic acid with Preconcentration and elution 75 ml of purified 25% ammonia liquor and diluting to 1000 ml with high purity water; pH adjusted with HNO3 and NH4OH) All synthetic and natural water samples were handled on a were used to obtain the desired pH in the solutions.Model clean bench. Before preconcentration and elution, the anion solutions containing humic substances (Aldrich, Stockholm, exchange columns were flushed with 10 ml of 2 M HNO3, Sweden; batch number 8101816) were prepared from a stock followed by 20 ml of Milli-Q water. For the cation exchange solution containing 500 mg l-1 and diluted with high purity columns, 10 ml of 2 M HNO3 was passed, followed by 20 ml of Milli-Q water and, in the case of Chelex 100, 5 ml of 1 M Table 1 Operating parameters of the ICPMS instruments pH 6.0 ammonium acetate buVer.An aliquot of 20–30 ml of the filtered samples or model solutions was pumped through Parameter ICPQMS ICPSMSb the coupled columns with the aid of a peristaltic pump at a flow rate of 0.3 ml min-1 for the Dowex/Chelamine system Rf power/W 1348 1500 Sample uptake/ml min-1 0.8–0.9 0.1–0.3 and 1.0 ml min-1 for Sephadex/Chelex.Thereafter the columns Gas flow rates/l min-1 were disconnected from each other and the flow direction Coolant 14 15 reversed. The columns were washed with 5 ml of Milli-Q water Auxiliary 1.1–1.3 0.69 followed by elution with 2 M HNO3. Nebulizer 0.81–0.90 1.0–1.12 Ion lens settings a a Collection of water samples Torch Fassel torch Fassel torch Spray chamber Scott type, Scott type, Water samples were collected in 1998 from a bog-aVected double pass double pass drainage area near the Aitik copper mine in northern Sweden. Nebulizer Meinhard Meinhard Sample cone Nickel, Nickel, The samples were collected on the Julian dates 137 (17/05), 1.0 mm orifice 1.1 mm orifice 170 (19/06), 180 (29/06), 203 (22/07), 233 (21/08) and 260 Skimmer cone Nickel, Nickel, (17/09).In the field, the water samples were filtered through 0.7 mm orifice 0.8 mm orifice acid washed 0.45 mm filters (Millipore, type HA) and collected Acquisition mode Peak jumping E-scan into 1.5 l acid washed PTFE bottles. No preservation of the Dwell time/ms 10.24 10*, 20** samples was used to prevent any disturbance of the natural Samples per nuclide 3 30*, 25** Sweeps per replicate 20 15*, 15** equilibrium.Trip blanks were carried out to determine poten- Windows (%) tial contamination. In situ measurements of conductivity and Acquisition 30*, 100** pH values for the samples are given in Table 2. pH and Integration 30*, 80** conductivity measurements were performed using a Surveyor Search 30*, 60** II sonde (Hydrolab Corp., Austin, TX, USA).To obtain aAdjusted to obtain maximum signal intensity. bThe asterisks denote reliable pH values, the pH-electrode was allowed to stabilize the resolution mode: *low resolution and **medium resolution. for more than 30 min in the water. The temperature was also 212 J.Environ. Monit., 1999, 1, 211–217Table 2 Concentrations of humic substances (HS), pH, conductivities, temperatures and ionic strengths recalculated from conductivity20 of the natural water samples HS/ Conductivity/ Ionic strength/ Temperature/ Sample mg l-1 pH mS cm-1 10-4 °C 137 11.7 6.37 32.4 5.2 0 170 8.9 6.55 45.3 7.2 8.3 180 8.2 6.86 49 7.8 10.9 203 8.2 6.76 54.3 8.7 11.4 233 8.4 7.08 58 9.3 11.9 260 8.5 6.93 65.3 10.4 6.4 7967 14.1 6.52a b — b 7968 13.8 6.38a b — b 8891 12.8 5.81a b — b aMeasured in the laboratory. bNot measured.monitored as it aVects both the conductivity and pH. The pH and conductivity values in Table 2 are corrected for temperature dependence. Water samples (7967, 7968 and 8891) were also collected in the Lavsjo�n Lake, Sweden in September 1998.Estimation of humic substance concentrations in natural waters The concentrations of humic substances in the natural water samples were estimated using spectrophotometric measurement at 400 nm. Filtered (0.45 mm Millipore, type HA) standard solutions containing 0.5, 1.0, 2.0, 5.0, 10.0, 20.0 and 50.0 mg l-1 of the Aldrich humus were used for the calibration. Natural water samples were also filtered before the spectrophotometric determination. The results of the determinations of humic substances in the natural water samples are presented in Table 2.Equilibrium speciation modelling Fig. 1 Fractions of humic-bound metals (calculated as concentration retained on anion exchanger divided by the sum of the concentrations Individual models were established for each of the analytes, retained on the anion and cation exchangers) obtained using Sephadex calculations being performed using the free energy minimiz- (open bars) and Dowex (hatched bars).Errors bars are±one standard ation algorithm implemented in the computer program deviation (n=3), computed using the rules of error propagation for SOLGASWATER described by Eriksson.21 Thermodynamic including the uncertainties in both the determined humic-bound and summed concentrations.data were primarily taken from Gunneriusson and Sjo� berg,7 Lo�vgren and co-workers8,22 or the literature cited therein. In some cases, stability constants had to be adjusted to zero ionic Al and Co complexes. Three of the elements, Cr, Mo and V, strength conditions, which was accomplished using the are present, at least to some extent, as complex oxoanions in extended Debye–Hu� ckel expression proposed by Davies.23 natural waters, which means that the fractions retained on the Stability constants for metal–humic complexes were initially anion exchangers cannot be unambiguously classified as estimated from the formation constants of the corresponding organically bound. However, a significantly greater fraction oxalates, and the previously established relationship between of V is recovered by Sephadex than by Dowex (t=7.57; these values.7,8 As literature values from diverse sources may P>0.01).vary considerably, data were arbitrarily selected on the basis of their providing equilibrium speciation distributions that agreed, within a factor of two, with the experimental results.Humic substances were treated as a simple diprotic acid (H2L), and assuming one H2L group per 30 carbon atoms.7 The concentrations of humic substances, expressed in mg l-1, were estimated from the absorbance measurements described above and converted to mol l-1 active sites on the basis of the observation that Aldrich humic acid contains 50% by mass of carbon.24 Results and discussion Comparison of the ion exchange resins A compatween the two anion and the two cation exchange materials was performed.Selected results are depicted in Fig. 1. For Cd, Co, Mn, Ni and Sr, no significant diVerences in the amounts of humic complexes were obtained using either Sephadex or Dowex, so these data have been Fig. 2 Breakthrough curves for humic substances (HS) using omitted. With respect to precision, F-tests showed that Sephadex (&) and Dowex (+). Sample processed contained 5 mg l-1 HS. Sephadex was significantly better (P>0.05) than Dowex for J. Environ. Monit., 1999, 1, 211–217 213Table 3 Statistical comparison of the cation exchange materials in two-column speciation systems with respect to retained free metal ion concentrations and precisions (n=3) t-test F-test Cation exchanger P<0.01 P<0.02 P<0.05 P<0.01 P<0.02 P<0.05 Chelex Al, Ba, Sr, Zn Cd Fe Mo Ba Al, Cr, Cu Chelamine Ni — — — — — Table 4 Comparison of the speciation data with independent measurements of the total concentrations (n=3).Error terms are one standard deviation.Numbers in the Element column indicate the samples described in Table 2 Concentration/mg l-1 Elementa Sephadex Chelex Sumb Total t-statisticc Al (0.2) 137 32.58±0.29 22.02±0.33 54.60±0.44 68.57±0.79 -26.76*** 170 17.87±0.33 14.38±0.16 32.25±0.37 32.62±0.16 -1.59 180 10.81±0.37 7.75±0.16 18.56±0.40 18.92±0.20 -1.39 203 10.96±0.04 10.44±0.19 21.40±0.19 24.31±0.42 -10.93*** 233 5.53±0.22 6.48±0.17 12.01±0.28 12.18±0.06 -1.03 260 4.12±0.14 6.16±0.06 10.28±0.15 10.79±0.16 -4.03** (0.08) 7967 38.18±0.47 100.39±8.14 138.57±8.16 173.33±0.31 -7.37** (0.08) 7968 32.59±0.23 86.76±0.72 119.35±0.75 160.32±1.51 -42.14*** (0.08) 8891 31.11±5.05 93.81±6.70 124.92±8.39 168.51±0.73 -8.96** Ba (0.01) 7967 <0.003 5.04±0.30 5.04±0.30 4.81±0.01 1.33 7968 <0.003 5.04±0.05 5.04±0.05 4.62±0.04 11.72*** 8891 <0.003 4.75±0.17 4.75±0.17 4.78±0.08 -0.26 Cd (0.02) 137 0.02±0.01 0.02±0.01 0.04±0.01 0.04±0.03 0.00 170 0.02±0.01 0.04±0.02 0.06±0.02 0.04±0.01 1.55 203 0.03±0.01 0.03±0.01 0.06±0.01 0.05±0.01 1.22 233 0.03±0.01 0.10±0.01 0.13±0.01 0.13±0.03 0.00 Co (0.02) 170 0.22±0.02 0.29±0.02 0.51±0.03 0.05±0.01 25.20*** 180 0.07±0.02 0.03±0.01 0.10±0.02 0.04±0.01 4.65** 233 0.04±0.02 0.07±0.03 0.11±0.04 0.06±0.01 2.10 260 0.02±0.01 0.03±0.01 0.05±0.01 0.04±0.01 1.22 Cu (0.2) 137 5.63±0.15 0.31±0.02 5.94±0.15 5.81±0.19 0.93 170 3.13±0.07 0.30±0.02 3.43±0.07 3.56±0.21 -1.02 180 3.14±0.02 0.51±0.05 3.65±0.05 3.71±0.09 -1.01 203 2.10±0.06 1.44±0.01 3.54±0.01 4.11±0.06 -16.23*** 233 2.02±0.08 0.66±0.05 2.68±0.09 3.06±0.07 -5.77*** 260 2.08±0.04 1.48±0.05 3.56±0.06 4.02±0.21 -3.65 Fe (0.4) 7967 443.4±128.5 22.69±4.13 446.1±128.6 535.3±14.56 -1.19 7968 366.6±95.3 53.7±9.21 420.3±95.7 505.4±3.18 -1.55 8891 370.7±21.3 119.8±27.3 490.5±34.6 519.4±4.21 -1.44 Hgd (2.0) 7967 3.22±0.06 <0.67 3.22±0.06 3.60±0.04 -9.20*** 7968 2.96±0.06 0.77±0.00 3.74±0.06 3.30±0.06 9.09*** 8891 4.00±0.01 2.45±0.00 6.44±0.01 7.00±0.11 -9.12** Mn (0.3) 137 0.07±0.01 0.17±0.02 0.24±0.02 0.29±0.02 -3.06 170 0.44±0.02 0.48±0.02 0.92±0.03 0.21±0.01 38.89*** 233 0.05±0.02 0.52±0.04 0.57±0.05 0.54±0.02 0.96 (0.03) 7967 0.58±0.08 1.24±0.11 1.81±0.13 2.96±0.12 -11.22*** (0.03) 7968 0.43±0.09 0.95±0.11 1.38±0.14 2.92±0.06 -17.09*** (0.03) 8891 0.15±0.09 1.30±0.15 1.45±0.18 2.12±0.07 -6.07*** Pb (0.1) 203 0.23±0.01 1.03±0.03 1.26±0.03 1.12±0.05 4.16** Sr (2.0) 7967 <0.67 10.38±0.67 10.38±0.67 10.77±0.24 -0.94 7968 <0.67 10.91±0.29 10.91±0.29 10.70±0.09 1.18 8891 <0.67 10.18±0.75 10.18±0.75 10.66±0.34 -1.01 U (0.0005) 7967 0.115±0.009 0.005±0.001 0.120±0.009 0.139±0.002 -3.74 7968 0.098±0.001 0.004±0.000 0.102±0.001 0.134±0.001 -45.40*** 8891 0.113±0.007 0.006±0.002 0.119±0.007 0.136±0.001 -4.11 Zn (0.3) 137 0.46±0.38 4.01±0.07 4.47±0.39 4.55±0.58 -0.20 170 0.39±0.04 2.49±0.09 2.88±0.09 1.16±0.09 23.41*** 203 0.14±0.06 0.91±0.06 1.05±0.06 1.65±0.39 -2.63 (0.2) 7968 0.57±0.17 3.86±0.23 4.43±0.29 3.73±0.13 3.83 (0.2) 7968 0.21±0.03 6.15±0.97 6.36±0.97 3.46±0.24 5.03 (0.2) 8891 0.26±0.05 3.94±0.10 4.21±0.12 5.14±0.22 -6.56*** aDetection limits (mg l-1) in parentheses if not otherwise noted. bThe sum of the concentrations isolated by the Sephadex and Chelex columns.cNegative values indicate that the sum is less than the total concentration and positive values that the former is larger. The asterisks denote significant diVerences (two-tailed t-test) at the 95% (*), 98% (**) or 99% (***) confidence level. dMercury concentrations in ng l-1. 214 J. Environ. Monit., 1999, 1, 211–217Of the remaining seven elements, both anion exchange Sephadex resin, 360 ng humic complexes per ml resin are retained, whilst Dowex retains only 160 ng ml-1. materials provide statistically equivalent humic-bound fractions for Al, Fe and U. Clearly, U should be considered, along When comparing the cation exchange materials, Chelex shows a better performance than Chelamine, both with respect with Fe and Hg, as one of the elements exhibiting strongest association with humic substances in fresh water systems, to recovery and precision, for a range of elements (Table 3).The only exception is Ni, for which Chelamine exhibits a according to KerndorV and Schnitzer.10 Significantly greater fractions of the total Cu and Pb concentrations were recovered highly significantly (P<0.01) greater recovery.Both cation exchange columns proved to be equally viable for the other by the Sephadex column (t-test; P<0.01 for both elements). The Sephadex results are in accordance with the intermediate elements considered (Co, Mn, Pb, U, V). The Chelex column shows good recovery (84–100%) for many elements if the binding strengths observed for the humic complexes of these elements.Finally, for Ba and Zn, whose interactions with column is prewashed with 100 mM ammonium acetate pH 6.0 buVer before subsequent loading of the sample. Pai et al.18 naturally occurring organic ligands are expected to be negligible or weak, the use of Dowex resin results in serious observed that, when resins in the Na+, K+ or NH4+ form are used, it is advisable to apply external buVer systems, usually overestimations of the humic-bound fractions. For Ba, this can be explained by the low recovery of the free metal ions acetate.Chelamine shows good recovery for several elements, but has poor flow characteristics, flow rates as low as on the Chelamine column, as the sum of the concentrations collected by the anion and cation exchange materials was used 0.3 ml min-1 being needed to obtain a good recovery of trace metals.12 There is, however, no need for an external buVer to derive the humic-bound fraction. Chelamine is known to discriminate better against alkaline earth elements than system with Chelamine.As the anion exchange resins can be used at a flow rate of 1.0 ml min-1, it would only be time- Chelex.12 When measuring the absorbance of the acid Sephadex eluate consuming to use Chelamine instead of Chelex 100.With these results in mind, the conclusion can be drawn that Sephadex at 400 nm, a significant response was obtained. This finding suggests that the metal ions eluted are associated with humic and Chelex columns should be coupled if simultaneous preconcentration and separation of metal–humic complexes and free substances.When the same procedure was performed on the acid Dowex eluate, no (or very small compared to Sephadex) metal ions are to be accomplished. response was obtained, indicating eYcient retention of humic substances. An eVect visually observed is the discolouration Measurement of metal–humic complexes and free metal ions in of the resin, further supporting this assertion.There is a natural waters possibility that the humic substances are irreversibly retained on the resin, while the metal ions are eluted. When the waters were processed through the coupled columns, a fractionation of the total amount of metals in the water was The capacities of the anion exchange materials were investigated by performing a breakthrough test, using a model attained.Some of the fractions contained extremely low concentrations and are therefore reported as being less than the solution containing 5 mg l-1 Aldrich humic acid and 100 mg l-1 of the test elements. Breakthrough curves are shown in Fig. 2, limit of detection. Several results listed for the two-column method are below the detection limits for conventional where, in order to see small changes in the response, the wavelength used was 254 nm, as it provides higher sensitivity ICPQMS determinations, but quantitation is possible here due to the threefold preconcentration factor.Trip blanks were for humic substances than detection at 400 nm. For the Table 5 Thermodynamic data used in the calculations.Data are valid at 25 °C and zero ionic strength Metals/log bp, q, ra Species Al3+ Ba2+ Cd2+ Co2+ Cu2+ Hg2+ Mn2+ Pb2+ Sr2+ UO22+ Zn2+ MeOH+ -13.4 -10.08 -7.80 -3.40 -7.70 -13.2 -5.80 -9.00 Me(OH)2+ -5.00 Me(OH)2+ -10.52 Me(OH)3 -16.25 Me(OH)4- -23.46 Me(OH)2 -20.35 -15.00 -6.17 -17.10 -17.80 Me2(OH)22+ -10.60 -5.60 Me(OH)3- 28.30 Me2(OH)3+ -6.40 Me3(OH)42+ 23.90 -12.70 Me3(OH)5+ 15.60 MeH-1L -7.32 MeL+ -3.22 MeL2- -7.31 MeL 6.36 -4.26 -3.95 -3.10 2.73 -4.42 -3.84 -6.13 -2.07 -4.87 MeSO4 2.46 2.40 2.30 2.30 2.80 2.30 Me(SO4)22- 3.50 Me(SO4)34- 3.00 MeHCO3+ -2.90 Me(CO3)22- -18.36 Me(CO3)34- -29.04 MeCl+ 7.21 MeCl2 13.98 MeClBr 16.63 MeBr+ 9.54 MeBr2 17.91 aFormation constants are defined according to the general equilibrium: pH++qMen++rXm-uHpMeqXr p+nq-mr.J. Environ. Monit., 1999, 1, 211–217 215performed during the sampling procedure by processing To evaluate the capabilities of the two-column system for recovering trace elements, total concentrations determined Milli-Q water. These blanks were prepared and analysed as for the other samples and, for the elements studied, no utilizing accredited analytical methods were compared with the sums of values obtained from analyses of the eluates from contamination could be detected by ICPQMS. It can be seen in Table 4 that, for some analyte/matrix the Sephadex and Chelex columns.Quality control of total concentration determinations by ICPMS includes analysis of combinations, some fraction of the total concentration is not retained on either the Chelex or the Sephadex columns, which the riverine certified reference material SLRS-3 supplied by the Natural Research Council of Canada, for which accurate probably depends on losses during the washing step. It is, however, necessary to perform this washing step to ensure results are routinely obtained (results not shown here).The results of the statistical analysis are also reported in that the species eluted have been retained by the sorbent.To accomplish measurements of ultratrace levels of Hg and U in Table 4. The sum of the speciation results for Al is consistently lower than the total concentration. This also applies to the natural waters, a higher preconcentration factor would have to be used together with ICPQMS. Instead, other instrumental majority of data for Fe, Cu and Mn, although the diVerences are not always statistically significant.For the last two techniques (CVAFS and ICPSMS) were applied, thus providing the improved limits of detection required (2 ng l-1 Hg and elements, Groschner and Appriou1 found that 40–55% of Cu and 6–23% of Mn were present as complexes retained by C18 0.5 ng l-1 U, respectively, for direct measurement).Iron was measured in only three of the samples, as the technique used in their three-column speciation system. In the present twocolumn arrangement, at most 14% Cu and 53% Mn are not for six of the samples (ICPQMS) suVers from severe spectral interferences, e.g. 40Ar16O+ on 56Fe+. Significant portions of retained by either Sephadex or Chelex. It is improbable that this unretained fraction corresponds to neutral humic com- the total concentrations of most elements were retained on the Sephadex column (Table 2), illustrating the importance of plexes, because all water samples analysed had a low ionic strength and pH values above 5.8.Under such conditions, humic substances in determining the speciation of numerous trace metals in natural waters. humic substances are negatively charged.25 Thus, analyte escap- Fig. 3 Fractions of humic-bound metals, determined using the Sephadex anion exchange column (open bars) or calculated using equilibrium speciation models (hatched bars), in the water samples described in Table 2: (a) 137; (b) 170; (c) 180; (d) 203; (e) 233; (f ) 260; (g) 7967; (h) 7968; (i) 8891. Errors bars for the experimental measurements are±one standard deviation, computed using the rules of error propagation for including the uncertainties in both the determined humic-bound and total concentrations.The asterisks denote significant diVerences (twotailed t-test) at the 95% (*), 98% (**) or 99% (***) confidence level, and ‡ indicates that data were below the limit of detection for the experimental procedure. 216 J.Environ. Monit., 1999, 1, 211–217ing from the two-column system is more likely to be in the provides a reliable measure of the humic-bound fraction, and thus lends credibility to the results obtained. However, using form of neutral inorganic complexes, such as hydroxides, sulfates or carbonates.7 It was found to be impractical to only an anion exchange material for the separation of metal– humic complexes could give an incorrect estimation of the couple a third column in the present system, owing to the excessive back-pressure that compromised successful operation metals complexed by organic matter in the event that some neutral species exist in natural waters.To study neutral metal– of the peristaltic pump. As the concentrations of the humic substances were humus species, and if ICPMS is the detector of choice, a reversed phase material has to be used together with a mobile estimated, the pH determined and, in six of the samples, the ionic strength computed from the measured conductivities phase that will not cause interferences in the determination.(iv) Utilization of multi-element analytical techniques in using Langelier’s approximation20 (Table 2), calculations of the fractions of metal–humic complexes could be established studying trace element–humus interactions is also to be recommended. Even elements forming weak complexes with using the constants in Table 5.The sulfate concentrations in the samples were not measured; instead a reference value of humic substances will contribute to a depletion of available binding sites, potentially aVecting the toxicity and 0.1 mM7,26 for Swedish rivers was adopted.In Fig. 3, the measured and calculated fractions of the metal–humic com- bioavailability of other metals. plexes are compared. As the sums of the fractions of Al and Mn isolated by Chelex and Sephadex are significantly non- Acknowledgements quantitative, the separately determined total concentrations TFR (Technical Research Council ), Sweden is acknowledged were used to estimate the humic-bound fractions from the for financial support.SGAB Analytica, Lulea°, Sweden contrib- experimental results. Otherwise, the larger of the concenuted with both financial and technical assistance. Hans trations denoted sum and total in Table 4 was used to derive Hultberg, IVL, Go� teborg, Sweden is thanked for providing the experimentally determined fractions.The pH variation the Lavsjo�n samples used in this study. observed (5.81–7.08) in the samples analysed did not have any discernible eVect on the distribution of the metals. Instead, the changes in metal ion concentration and amount of humic References substances seem to be the most important factors.14 The 1 M.Groschner and P. Appriou, Anal. Chim. Acta, 1994, 297, 369. comparison of the experimental and calculated humic-bound 2 B. Aster, P. Burba and J. A. C. Broekaert, Fresenius’ J. Anal. fractions for Cd, Co, Mn, Pb and Zn in Fig. 3 reveals few Chem., 1996, 354, 722. significant diVerences, and it is interesting to note the impor- 3 M.Norde�n and E. Dabek-Zlotorzynska, J. Chromatogr. A, 1996, tance of humic substances for the speciation of Cd and Co, 739, 421. despite the weakness of these complexes. The equilibrium 4 K. G. Heumann, L. Rottmann and J. Vogl, J. Anal. At. Spectrom., speciation model is unreliable for Al and Hg, for which three 1994, 9, 1351. 5 L. Rottmann and K. G. Heumann, Anal.Chem., 1994, 66, 3709. and two possible humic complexes, respectively, have been 6 E. Tipping, Aquat. Geochem., 1998, 4, 3. proposed.7,8,22 For these elements, the aqueous speciation is 7 L. Gunneriusson and S. Sjo�berg, Nordic Hydrol., 1991, 22, 67. complicated by the formation of hydroxide and halide com- 8 L.Lo� vgren and S. Sjo�berg, Wat. Res., 1989, 23, 327. plexes, which suggests that greater attention should be paid 9 F.J. Stevenson, Humic Substances, Wiley, New York, 2nd edn., towards the input data requirements (i.e. determination of 1994, pp. 19–20. chloride and bromide concentrations) and to studying the 10 H. KerndorV and M. Schnitzer, Geochim. Cosmochim. Acta, 1980, 44, 1701. eVects of pretreatment of the Chelex column with buVer on 11 V. I.Esteves, N. M. A Cordeiro and A. Costa Duarte, Mar. the results. Chem., 1995, 51, 61. The experimentally determined fractions of humic-bound 12 S. Blain, P. Appriou and H. Handel, Anal. Chim. Acta, 1993, Cu were significantly lower than the modelled distribution in 272, 91. three of the six samples collected near the Aitik copper mine 13 B. R. Kim, V. L. Snoeyink and F. M. Saunders, J.Water Pollut. in northern Sweden. These samples constitute the latter half Control Fed. 1976, 48, 120. 14 C. Pettersson, B. Allard and H. Bore�n, Estuar. Coast. Shelf Sci., of a series in which the water conductivity increased with 1997, 44, 533. sampling date (Table 2). This would indicate that increasing 15 M. Hiraide, S. P. Tillekeratne, K. Otsuka and A. Mizuike, Anal. concentrations of major cations are competing for available Chim. Acta, 1985, 172, 215. binding sites on the humic substances.14 According to the 16 C. C. Huang and M. H. Yang, Anal. Chem., 1997, 69, 3930. model calculations, the 11 elements considered (Table 5) 17 K. Akatsuka, J. W. MacLaren, J. W. Lam and S. S. Berman, occupied as much as 80% of these sites. J. Anal. At. Spectrom., 1992, 7, 889. 18 S. Pai, P. Whung and R. Lai, Anal. Chim. Acta, 1988, 211, 257. 19 L. A. Ellis, J. Anal. At. Spectrom., 1998, 13, 631. Conclusions 20 V. L. Snoeyink and D. Jenkins, Water Chemistry, Wiley, New York, 1980. (i) For the quantitation of humic-bound metals in natural 21 G. Eriksson, Anal. Chim. Acta, 1979, 112, 375. waters, Sephadex is preferable to Dowex. The latter has a 22 L. Lo� vgren, T. Hedlund, L.-O. O� hman and S. Sjo�berg, Water lower capacity and irreversibly retains humic substances, and Res., 1987, 21, 1401. 23 C. W. Davies, Ion Association, Butterworths, London, 1962. incorrectly classifies Ba and Zn as forming fairly strong 24 S. Boggs, Jr., D. C. Livermore and M. G. Seitz, Rev. Macromol. complexes with naturally occurring organic ligands. Chem. Phys., 1985, C25, 599. (ii) Parallel determinations of total concentrations are 25 F. J. Stevenson, Humic Substances, Wiley, New York, 2nd edn., necessary, as in the preferred configuration of Sephadex/ 1994, p. 350. Chelex complete recoveries of Al, Cu, Mn and U were 26 K. B. Berner and R. A. Berner, The Global Water Cycle, Prentice not possible. Hall, Englewood CliVs, 1987, p. 397. (iii) The comparison of experimentally determined and calculated equilibrium distributions suggests that Sephadex Paper 9/01070J J. Environ. Monit., 1999,