Iron isotope fractionation in river colloidal matter
Johan Ingri a,⁎, Dmitry Malinovsky a, Ilia Rodushkin b, Douglas C. Baxter b,Anders Widerlund a, Per Andersson c, Örjan Gustafsson d,
Willis Forsling e, Björn Öhlander a
a Division of Applied Geology, Luleå University of Technology, SE-971 87 Luleå, Swedenb Analytica AB, Aurorum 10, SE-977 75 Luleå, Sweden
c Laboratory for Isotope Geology, Swedish Museum of Natural History, Box 50007, SE-104 05 Stockholm, Swedend Institute of Applied Environmental Research (ITM), Stockholm University, SE-106 91 Stockholm, Sweden
e Division of Chemistry, Luleå University of Technology, SE-971 87 Luleå, Sweden
Received 22 August 2005; received in revised form 20 March 2006; accepted 20 March 2006Available online 2 May 2006
Editor: H. Elderfield
Abstract
Temporal variations in the iron isotopic composition, δ56Fe between − 0.13‰ and 0.31‰, have been measured in thesuspended fraction in a Boreal river. The major mechanism behind these variations is temporal mixing between two types ofparticles–colloids, Fe-oxyhydroxides and Fe–C colloids. Data in this study indicate that these two types of colloids have differentFe-isotope composition. The Fe–C colloid has a negative δ56Fe value whereas the Fe-oxyhydroxide colloid is enriched in 56Fe.
Keywords: Fe isotopes; Kalix River; suspended matter; Fe-colloids; temporal variation
1. Introduction
Zhu et al. [1] demonstrated systematic changes in theFe-isotope composition for the last 1.7 million years in aFe–Mn crust from the North Atlantic Ocean. Theysuggested that the variations reflect isotopic variabilityin the continental sources of Fe fluxes to the oceans.However, Levasseur et al. [2] found no evidence that theobserved oceanic Fe isotopic heterogeneity in marinehydrogenetic ferromanganese deposits is controlled by
variations in continental sources. Most of the ferroman-ganese nodules and crusts analysed in the open oceanshow clearly negative δ56Fe values [2,3], thus implyingan overall negative isotope signal in seawater. However,Levasseur et al. [2] concluded that the overall negativeisotope signal and the variations are induced locallywithin the ocean, although the exact processes for thefractionation remain unclear.
Fantle and DePaolo [4] suggested that the net effectof continental weathering processes is to mobilize smallamounts of isotopically light Fe in an “exchangeable”,or dissolved form. They inferred that continental wea-thering under modern oxidising earth surface conditions
Table 1Fe-isotopic composition of suspended matter in the Kalix River
Sample Date δ56Fe(‰)
2 STD(‰)
δ57Fe(‰)
2 STD(‰)
LeachableFe ⁎
(%)
91-42 March 30,1991
0.18 0.04 0.25 0.05 98
91-54 May 8 0.05 0.03 0.06 0.0391-63 May 13 − 0.05 0.04 − 0.07 0.06 9591-73 May 17 − 0.05 0.04 − 0.08 0.0991-80 May 27 − 0.06 0.01 − 0.07 0.0291-93 June 7 0.00 0.04 0.01 0.08 9991-137 July 10 − 0.13 0.05 − 0.18 0.08 9691-164 August 8 − 0.04 0.04 − 0.05 0.0591-207 September 9 0.08 0.03 0.12 0.03 9791-231 October 1 0.20 0.03 0.27 0.04 9691-243 October 8 0.22 0.03 0.32 0.10 9891-266 November 5 0.16 0.06 0.23 0.0891-278 December 3 0.12 0.01 0.18 0.0491-280 December 9 0.04 0.03 0.07 0.01 9992-2 January 10,
19920.12 0.05 0.18 0.07
92-10 February 6 0.22 0.01 0.31 0.03 9992-21 March 13 0.20 0.03 0.30 0.0692-43 April 9 0.31 0.03 0.44 0.05 9992-60 April 17 0.23 0.05 0.34 0.0492-72 May 14 0.14 0.06 0.22 0.09 9192-73 May 18 0.01 0.09 0.01 0.13 9692-79 May 22 − 0.07 0.02 − 0.09 0.03 9692-120 June 18 − 0.08 0.02 − 0.12 0.03 9792-123 June 22 − 0.07 0.03 − 0.09 0.03
⁎ Fe released by reductive HCl extraction (see text).
793J. Ingri et al. / Earth and Planetary Science Letters 245 (2006) 792–798
preferentially releases dissolved Fe with negative δ56Fe,which is transported in rivers to the ocean, and suggestedthat riverine Fe has a substantial role in determining theδ56Fe of both the modern and ancient oceans.
It is not yet clearly established if the relatively highdissolved Fe concentrations (b 0.45 μm) observed inmany organic rich freshwater systems are attributed to Fe-organic (humic) complexes [5] and/or small Fe-oxyhydr-oxide particles [6]. Recently, both types of colloids, Fe-oxyhydroxides and Fe–C colloidal matter, have beenidentified as important carriers of dissolved Fe in borealriver systems [7–11]. Data in this study indicate that thesetwo types of colloids have different Fe-isotope compo-sition. This finding provides a mechanism for a variableFe-isotope signal in river suspended matter and henceopens up the possibility for a changing river introducedFe-isotope signal to the ocean during geological time. Therelative river discharge and conservativeness during es-tuarinemixing need to be understood for each of these twoFe-colloid forms to properly evaluate the role of riverintroduced Fe to the Fe-isotope variations preserved in theenvironmental archive of the ocean.
2. Sampling
Sampling was performed in the Kalix River, NorthernSweden. Detailed description of the river system, generalhydrogeochemistry and the sampling procedures for thecollection of the suspended phase in the river have beendescribed elsewhere [12–14]. The samples were filteredin situ and filters were kept frozen until analyses. Fe-isotope measurements were performed during 2003 and2004 on samples collected during the period March 1991to June 1992. The filters were deliberately clogged duringsampling to detect both particles and colloidal matter.Substantial amounts of colloidal particles (b 0.45μm) richin Fe–C are enriched on the filters during periods of highTOC (total organic carbon) in the river [7].
3. Analytical methods
The suspended matter was prepared for Fe-isotopeanalyses using a modified microwave-assisted extractionprocedure [13]. The filters were transferred to microwavedigestion PFA Teflon vessels, and 5 ml of nitric acid(16 M) and 1 ml of hydrochloric acid (10 M) were added.The samples were then heated in a microwave oven for60 min. This procedure dissolves most of the Fe in thesamples, usually more than 95% of the total Fe (Table 1).Some samples during spring flood inMay have up to 10%of total Fe left in un-dissolved particles. A mixture of24 M HF and 16 m HNO3 (3:1) was used to dissolve
remaining Fe in the silicate residue obtained from the HClreductive extraction, and the relative amount of Fe in theHCl-leachable fraction was calculated (Table 1).
Anion-exchange chromatography was used for chem-ical purification of Fe.With minor adjustments, the anion-exchange closely followed that employed by Malinovskyet al. [15]. A mass balance estimate of Fe contents onaliquots of each sample before and after anion-exchangeseparation demonstrated that Fe recovery from the co-lumn was always higher than 95%. To eliminate highconcentrations of HCl, which were found to affect preciseMC-ICPMS measurements of Fe, the Fe fractions afteranion-exchange separation were transferred into Teflonbeakers, evaporated on a hot plate to dryness and re-dissolved in 0.3 M HNO3. The samples were then dilutedto 5.0±0.5 mg Fe l− 1 with 0.3 M HNO3 and spiked withNi at 5 mg l− 1, followed by isotope ratio measurementsusing multi-collector inductively coupled plasma massspectrometry (MC-ICPMS).
Procedural blanks were prepared for all samplemanipulations and found to contain insignificant quanti-ties of Fe relative to that extracted from the filters. Totalprocedural blanks of Fe for the filters after digestion and
794 J. Ingri et al. / Earth and Planetary Science Letters 245 (2006) 792–798
anion-exchange procedures including evaporation stepsconstitute 15±6 μg (n=6). This blank level at the mostcorresponds to ∼3% of Fe extracted from the sampleswith low Fe concentrations. For most of the samples theblank contribution does not exceed 1.5%. We therefore
assume that the influence of blank contribution duringsample preparation on Fe isotopic composition of thesamples is insignificant.
Fe-isotope ratio measurements were performed withdouble focusing high resolution MC-ICPMS instrument(Neptune, Thermo Finnigan, Germany). The typicaloperation conditions for the instrument and the perfor-mance of Fe-isotope ratio measurements are described indetail in [15]. Nickel was used as a normalising elementfor instrumental mass bias correction of the Fe isotopesusing 62Ni/60Ni ratio. The suitability of Ni as a normal-ising element for reduction of the instrumental mass biashas been demonstrated in previous studies [15,16]. Theanalyses were made in a sequence of isotope standard,sample, isotope standard and so on.
Results are presented using the δ-notation, defined as
where (56Fe/54Fe)standard is the ratio for IRMM-014,corrected for instrumental mass discrimination using Ni.Similar notations were also used for the 57Fe/54Fe and57Fe/56Fe ratios. As the samples and standard were dopedwith Ni, having the 58Ni isobar with 58Fe, no attempts tomeasure isotope ratios including 58Fe were made.
4. Results and discussion
Systematic changes in the δ56Fe value were detectedin the suspended fraction in the Kalix River (Table 1,Fig. 1a). The data indicate that there must exist at leasttwo types of Fe particles–colloids (in addition to Fe indetrital particles, mechanically weathered primary rockfragments), each carrying a different and distinct isotopesignature. We suggest that these are Fe-oxyhydroxidesand organically-complexed colloidal Fe, respectively.
4.1. The origin and influence of Fe-oxyhydroxideparticulate–colloidal matter
Data in Fig. 1 show the seasonal variations for somekey parameters, including the suspended Fe concentra-tion, in the Kalix River. The suspended phase in the riveris strongly enriched in Fe, mainly due to high
Fig. 1. a) Temporal variations in the δ56Fe value in suspended matter inthe Kalix River. Error bars show 2 SD of the mean. b) Suspended totalFe (clogged filters, 0.45 μm nominal cut-off). c) Water discharge andashed suspended load (ASL). d) Dissolved Al (b 0.45 μm) and totalorganic carbon (TOC). e) The Ce-anomaly, defined as 3(Ce/Cetill) / [2(La/Latill)+ (Nd/Ndtill)], in suspended matter from the Kalix River(clogged filters, 0.45 μm nominal cut-off, Cetill value for Ce in localtill, C-horizon).
Fig. 2. The relationship between total Fe (in percent of ashed weight)and the δ56Fe value in suspended matter from the Kalix River.
Fig. 3. The relation between the suspended Fe/Al ratio (a), the dissolvedAl concentration (b 0.45 μm) (b), the total suspendedMn concentration(c) and the δ56Fe value in suspended matter from the Kalix River.
795J. Ingri et al. / Earth and Planetary Science Letters 245 (2006) 792–798
concentrations of Fe-oxyhydroxides. It is clear from Fig.1 that the variation of the δ56Fe value is closely relatedto the total suspended Fe fraction in the river. A linearrelationship between these two variables is clearly seen(Fig. 2). The steady increase of Fe in the suspendedfraction during the ice-covered period, from Decemberto late April (Fig. 1b), reflects increasing influence ofFe-oxyhydroxide particles [7,12] precipitated frominflowing oxygen deficient groundwater enriched indissolved Fe(II). It has been shown by Bullen et al. [17]that Fe-rich oxyhydroxides formed by the inflow of Fe(II) into oxidised stream water produce particles en-riched in 56Fe. A similar process can explain the positiveδ56Fe values in the Fe-rich particles in the Kalix River.
4.2. The influence of detrital particles on the δ56Fevalue in the suspended fraction
There is a pronounced decrease in the relative Feconcentration in the suspended phase during melt-waterdischarge in May (Fig. 1b) because of re-suspension ofdetrital particles with comparably low Fe concentration.It is well established that igneous terrestrial rocks haveδ56Fe values close to zero or slightly positive in relationto the IRMM-14 standard [3,18]. Detrital particles sus-pended in the Kalix River (mainly igneous rocks in thecatchment) should therefore have a δ56Fe value close tozero or slightly positive. Thus, the decreasing δ56Fe valueduring spring flood is partly an indication of increasedamounts of leachable (Table 1) suspended detrital par-ticles. This is consistent with the suggestion by Fantleand DePaolo [4] that rivers with large suspended loads(much detrital particles) tend to have δ56Fe values nearzero. However, several lines of reasoning suggest that thedata cannot be explained by a simple mixing between Fe-oxyhydroxide particles and detrital particles, directlyregulated by water discharge. Assuming that most ofparticulate Al reflects detrital particles the amount ofdetrital Fe of the total suspended Fe concentration is
around 10% to maximum 20% in the analysed samplesduring spring flood [12], as indicated by relatively highFe/Al ratios in May (2 to 5 (Fig. 3a), compared with 0.62in local till [12] and 0.52 in average continental rock).Hence, the influence of detrital Fe on the 56Fe signaleven during spring flood should be of minor importancein the Kalix River. From Fig. 1c it is clear that the ashedsuspended load, ASL, increases during melt-waterdischarge in May, mainly indicating increased concen-trations of detrital material. However, in October andNovember no significant increase of ASL can beobserved, in spite of clearly decreasing δ56Fe values(Fig. 1a,c). Furthermore, as seen in Fig. 3a, the δ56Fevalue varies significantly at almost identical Fe/Al ratios(especially below a ratio of approximately 12), indicat-ing that the δ56Fe changes inspite of an almost constantamount of detrital particles (suspended Al is mainly an
Fig. 4. The correlation between the δ56Fe value and the Ce-anomaly insuspended matter from the Kalix River.
796 J. Ingri et al. / Earth and Planetary Science Letters 245 (2006) 792–798
indicator of detrital particles). Hence, a third suspendedfraction depleted in 56Fe must be present in the river.
4.3. The origin and influence of Fe–C colloids
Brantley et al. [19,20] showed that the δ56Fe of Fedissolved from a silicate soil mineral by siderophore-producing bacteria is as much as 0.8‰ lighter than bulkFe in the mineral. The δ56Fe of the exchange fraction onsoil grains is also lighter by approximately 0.6–1.0‰than Fe from both hornblende and iron oxyhydroxides.Data presented from a soil profile by Fantle and Depaolo[4] suggest that organic Fe is significantly lighter thanthe mineral Fe. Fantle and Depaolo and recentlyEmmanuel et al. [21] furthermore found the mostnegative δ56Fe values at the base of the O-horizon. It istherefore reasonable to suggest that dissolved Fe–C-richcolloidal matter leached from the upper horizons ofpodzols (O and E), reaching streams and subsequently ariver, has a negative δ56Fe value.
During the weathering process of primary minerals inboreal regions, Fe andAl are complexed by organicmatterin the upper soil horizons, transported downwards andsubsequently precipitated in the B-horizon (the podzoli-zation process). Recent studies, summarised inLundströmet al. [22], have shown that very high concentrations ofdissolved organically-complexed Al and Fe are present inthe organic surface horizon (O-layer) in podzols. Duringsnowmelt and heavy rains the water levels in streams andrivers rise and the upper soil horizons are flushed withwater. At these storm events Fe–C colloidal matter can bemobilised and reach streams and subsequently the KalixRiver [14]. This phasewas discussed by Ingri et al. [7] andhas recently been verified by FIFFF (flow field-flowfractionation) in the Kalix River [11]. Colloidal Cincreases dramatically in concentration at spring floodmaximum, and the C-rich phase has a large influence onthe transport of Fe during this period [11].
Dissolved Al is closely associated with organic ma-terial in the river [7] (Fig. 1d), in contrast to particulateAl, which mainly reflects detrital particles [12]. Becausedissolved Al, Fe and organic C to a large extent originatefrom the podzolization process during high waterdischarge, dissolved Al in the river can be used as anindicator of colloidal organically-complexed Fe [7]. Thedrop in δ56Fe seen during melt-water discharge in Mayand in October and November correlates with increasedTOC concentrations and increased dissolved Al (Fig. 1).With increasing amounts of dissolved Al an overalldecreasing trend for 56Fe is observed although the scatteris large (Fig. 3b). Hence, more Fe–C–Al colloids aretrapped on the clogged filters when the dissolved Al
concentration in the river goes up. Decreasing influenceof a Fe–C colloid can explain the systematic increase ofδ56Fe during winter, because dissolved Al decreasescontinuously during winter. In contrast, the ASL is moreor less constant during winter (Fig. 1).
The C-rich colloidal carrier phase has been shown tobe important also for the transport of the rare earthelements (REE) during increased water discharge in theKalix River [7,11]. Based on the geochemistry of theREE and especially on the Ce-anomaly (defined as 3(Ce/Cetill) / [2(La/Latill)+ (Nd/Ndtill)]) in the suspendedphase (Fig. 1e), the influence of the Fe–C colloidalphase for the δ56Fe value can be further verified. Alinear negative correlation (r=−0.73, four samples withvery high Mn values excluded, see below) betweenδ56Fe and the Ce-anomaly is observed in the riversuspended matter (Fig. 4). Observe that the Ce-anomalyin this case is not an indicator of redox status in the river,but more an indicator of fractionation processes duringthe podzolisation process in the soil [7]. Ingri et al. [7]suggested that the decreasing Ce-anomaly during winter(Fig. 1e) indicates more and more influence ofgroundwater (oxyhydroxide particles), and the relativeenrichments of Ce at storm events are caused by C-richcolloids. The linear relationship in Fig. 4 thereforeindicates mixing between Fe-oxyhydroxides (positiveδ56Fe), formed by the precipitation of inflowing anoxicgroundwater, and input of Fe–C-rich colloidal matter(negative δ56Fe) from the upper soil horizons. Mixingbetween these sources is therefore a plausible explana-tion for the temporal variations observed for the δ56Fevalue in the suspended fraction in the river.
4.4. The influence of precipitation of dissolved Mn andthe formation of Mn-rich particles in the river
There are four samples which show very high Mnconcentrations, above 1% Mn (Fig. 3c). These four
797J. Ingri et al. / Earth and Planetary Science Letters 245 (2006) 792–798
samples do not follow the linear trends observed for theother samples plotted in Figs. 3 and 4. These δ56Fevalues are more negative than would be expected fromthe mixing line between Fe-oxyhydroxides and Fe–Ccolloids. Manganese rich particles are formed in theKalix River during summer due to oxidation of dis-solved Mn, likely triggered by enhanced bacterial acti-vity in the river [23]. It remains to be shown how theformation of the Mn-rich particles influences thefractionation of the iron isotopes, but it is clear thatthe four relatively large negative δ56Fe values are notrelated to an increased concentration of dissolved Al andTOC or decreased concentration of suspended Fe.
4.5. Concluding remarks
Due to sampling and analytical difficulties thedissolved Fe reactive fraction, or the small sized colloidalFe fraction, in river water has not yet been directly deter-mined. By using an indirect method, clogging of filters,we have in this study shown that the small sized colloidalFe fraction in a Boreal river has a clear negative δ56Fevalue and that the signal varies depending on the majorpathways of water in the drainage basin. We believe thatthis type of Fe–C colloid is present in many other riversystems, and therefore might have implications for the Fe-isotope composition of Fe in the open ocean.
There are indications that Fe can reach the openocean via the small sized Fe–C colloidal material ob-served in boreal and arctic river systems. The size of theFe–C colloid is smaller than 10,000Da (10 kDa),whereas colloidal Fe-oxyhydroxides are generally largerthan 10 kDa [9,11]. A hydrodynamic diameter of 1.2 nmhas been calculated for the Fe–C carrier colloid [9],which suggests that the Fe–C complexes most likely canbe found also below the 1 kDa cut-off.
Dai and Martin [24] measured a relatively conserva-tive behaviour of ultrafiltered (b 10 kDa) Fe in theestuaries of both Ob and Yenisey rivers, in contrast tolarger colloids of Fe (0.4 μm–10 kDa) that showed rapidremoval. Furthermore, data from the Kalix River estuaryshowed a conservative behaviour of Fe over the 60 kmlength of the low-salinity plume outside the river, in spiteof significant aggregation [10]. Gustafsson et al. [10]hypothesized that this was a result from high organic-to-detrital matter characteristics of the aggregates. A lowspecific density of mineral-poor amorphous organicaggregates may lead to transport of these particles furtheraway from the river mouth. The Fe–C colloids reach theKalix River during storm events, and especially duringsnowmelt in spring. This facilitates the colloidal matterto reach far out from the river mouth, because the dis-
charged freshwater forms a layer on top of the estuarinewater at the river mouth. This layer glides far out in theestuary before the freshwater is thoroughly mixed withseawater. Similarly, Dai and Martin [24] measured moreFe removal in the Ob estuary compared with Yeniseywhich they related to the larger total suspended matterconcentration in Ob (20–143 mg l− 1) compared withYenisey (5.4–5.7 mg l− 1). Hence, data from boreal andarctic rivers indicate that river introduced Fe can reachfar out in the coastal zone and possibly be a source forreactive Fe to the open ocean.
Data in this study suggests that the δ56Fe value in thesmall sized Fe–C colloidal fraction that eventually reachthe open ocean from boreal rivers should be negative.The importance of this for the overall negative δ56Fe inopen ocean ferromanganese crusts remains to be shown,but the results of this study underline the need to studythe pathways of colloidal Fe–C from the river mouth outinto the open ocean before river input can be totallydisregarded as a source for Fe-isotope variations inocean water, especially in the North Atlantic.
Acknowledgements
This study was founded by the European Commis-sion, the Swedish Research Council and Kempestiftel-serna. We also wish to thank C. Pontér and AnalyticaAB for financial and logistic support.
References
[1] X.K. Zhu, R.K. O'Nions, Y. Guo, B.C. Reynolds, Secular varia-tion of iron isotopes in North Atlantic deep water, Science 287(2000) 2000–2002.
[2] S. Levasseur, M. Frank, J.R. Hein, A.N. Halliday, The globalvariation in the iron isotope composition of marine hydrogeneticferromanganese deposits: implications for seawater chemistry?Earth Planet. Sci. Lett. 224 (2004) 91–105.
[3] B.L. Beard, C.M. Johnson, J.L. Skulan, K.H. Nealson, L. Cox, H.Sun, Application of Fe isotopes to tracing the geochemical andbiological cycling of Fe, Chem. Geol. 195 (2003) 87–117.
[4] M.S. Fantle, D.J. DePaolo, Iron isotopic fractionation during con-tinental weathering, Earth Planet. Sci. Lett. 228 (2004) 547–556.
[5] E.M. Perdue, K.C. Beck, J.H. Reuter, Organic complexes of ironand aluminium in natural waters, Nature 260 (1976) 418–420.
[6] A.J. Cameron, P.S. Liss, The stabilization of “dissolved” iron infreshwaters, Water Res. 18 (1984) 179–185.
[7] J. Ingri, A. Widerlund, M. Land, Ö. Gustafsson, P.S. Andersson,B. Öhlander, Temporal variations in the fractionation of the rareearth elements in a boreal river; the role of colloidal particles,Chem. Geol. 166 (2000) 23–45.
[8] O.S. Pokrovsky, J. Schott, Iron colloids/organic matter associatedtransport of major and trace elements in small boreal rivers andtheir estuaries (NW Russia), Chem. Geol. 190 (2002) 141–179.
[9] B. Lyvén,M.Hassellöv,D.R. Turner, C.Haraldsson,K.Andersson,Competition between iron- and carbon-based colloidal carriers for
798 J. Ingri et al. / Earth and Planetary Science Letters 245 (2006) 792–798
trace metals in a freshwater assessed using Flow Field-FLOWFractionation coupled to ICPMS, Geochim. Cosmochim. Acta 67(2004) 3791–3802.
[10] Ö. Gustafsson, A. Widerlund, P.S Andersson, J. Ingri, P. Roos, A.Ledin, Colloid dynamics and transport of major elements througha boreal river-brackish bay mixing zone, Mar. Chem. 71 (2000)1–21.
[11] K. Andersson, R. Dahlqvist, D. Turner, B. Stolpe, T. Larsson, J. Ingri,P.S. Andersson, Colloidal rare earth elements in a boreal river:changing sources and distributions during spring flood, Geochim.Cosmochim. Acta (in press).
[12] J. Ingri, A. Widerlund, Uptake of alkali and alkaline-earth elementson suspended iron and manganese in the Kalix River, northernSweden, Geochim. Cosmochem. Acta 58 (1994) 5433–5442.
[13] F. Ödman, T. Ruth, C. Pontér, Validation of a field filtrationtechnique for characterization of suspended particulate matterfrom freshwater. Part 1. Major elements, Appl. Geochem. 14(1999) 301–317.
[14] J. Ingri, A. Widerlund, M. Land, Geochemistry of major elementsin a pristine Boreal river system; hydrological compartments andflow paths, Aquat. Geochem. 11 (2005) 57–88.
[15] D. Malinovsky, A. Stenberg, I. Rodushkin, H. Andrén, J. Ingri,B. Öhlander, D.C. Baxter, Performance of high resolution MC-ICP-MS for Fe isotope ratio measurements in sedimentarygeological materials, J. Anal. At. Spectrom. 18 (2003) 687–695.
[16] A. Stenberg, D.Malinovsky, I. Rodushkin, H. Andrén, C. Pontér, B.Öhlander,D.C.Baxter, Separation of Fe fromwhole bloodmatrix forprecise isotopic ratio measurements by MC-ICP-MS: a comparisonof different approaches, J. Anal. At. Spectrom. 18 (2003) 23–28.
[17] T. Bullen, A.F. White, C.W. Childs, D.V. Vivit, M.S. Schultz,Demonstration of significant abiotic iron isotope fractionation innature, Geology 29 (2001) 699–702.
[18] B.L. Beard, C.M. Johnson, K.L. von Damm, R.L. Poulson, Ironisotope constraints on Fe cycling and mass balance in oxygenatedEarth oceans, Geology 31 (2003) 629–632.
[19] S.L. Brantle, L. Liermann, T.D. Bullen, Fractionation of Fe isotopesby soil microbes and organic acids, Geology 29 (2001) 535–538.
[20] S.L. Brantle, L.J. Liermann, R.L. Guynn, A. Anbar, G.A. Icopini,J. Barling, Fe isotopic fractionation during mineral dissolutionwith and without bacteria, Geochim. Cosmochim. Acta 68 (2004)3189–3204.
[21] S. Emmanuel, Y. Erel, A. Matthews, N. Teutsch, A preliminarymixingmodel for Fe isotopes in soils, Chem.Geol. 22 (2005) 23–34.
[22] U.S. Lundström, N. van Breemen, D.C. Bain, P.A.W. van Hees, R.Giesler, J.P. Gustafsson, H. Ilvesniemi, E. Karltun, P.-A. Melkerud,M. Olsson, G. Riise, O. Wahlberg, A. Bergelin, K. Bishop, R.Finlay, A.G. Jongmans, T. Magnusson, H. Mannerkoski, A.Nordgren, L. Nyberg, M. Starr, L. Tau Strand, Advances in under-standing the podzolization process resulting from a multidisciplin-ary study of three coniferous forest soils in the Nordic Countries,Geoderma 94 (2000) 335–353.
[23] C. Pontér, J. Ingri, K. Boström, Geochemistry of manganese inthe Kalix River, northern Sweden, Geochim. Cosmochim. Acta56 (1992) 1485–1494.
[24] M.-H. Dai, J.-M. Martin, First data on trace metal level andbehaviour in two major arctic river-estuarine systems (Ob andYenisey) and in the adjacent Kara Sea, Russia, Earth Planet. Sci.Lett. 131 (1995) 127–141.
