Mineralogical control of rare earth elements in acid sulfate soils

Available online at www.sciencedirect.com

Geochimica et Cosmochimica Acta 73 (2009) 44–64 www.elsevier.com/locate/gca

Mineralogical control of rare earth elements in acid sulfate soils Susan A. Welch a,b,c,d,*, Andrew G. Christy a,c, Lloyd Isaacson a,b,e, Dirk Kirste a,b,f a

Department of Earth and Marine Sciences, Cooperative Research Centre for Landscape Environments and Mineral Exploration (CRC LEME), The Australian National University, Canberra ACT 0200, Australia b Cooperative Research Centre for Landscape Environments and Mineral Exploration (CRC LEME), The Australian National University, Camberra ACT 0200, Australia c Research School of Earth Sciences, The Australian National University, Canberra ACT 0200, Australia d School of Earth Sciences, The Ohio State University, Columbus, OH 43210, USA e Southern Cross University, P.O. Box 157 Lismore, NSW 2480, Australia f Department of Earth Sciences, Simon Fraser University, Burnaby, BC, Canada Received 17 December 2007; accepted in revised form 17 October 2008; available online 1 November 2008

Abstract Major, trace and rare earth element concentrations were measured in porewater, surface water and sediments at an acid sulfate soil site. The concentrations of La and Ce in porewater are up to 1–3 ppm. There is a strong correlation between REE concentration and acidity, except that the maximum concentrations were consistently found below the horizon of maximum acidity, associated with an increase in pH (to ca. 4) and change in mineralogy from jarosite-dominated to goethite-dominated mottles. Jarosite replacement by goethite is as expected with the rise in pH, which in turn is due to the occurrence of a fossil shell bed just below. The rare earth element patterns in the porewaters are enriched in the MREE with respect to Post-Archaean Australian Shale (PAAS). Measurements and calculations show that this is in accord with experiments on low-degree partial dissolution of jarosite, even when the jarosite itself is highly enriched in LREE. There is a clear fractionation in the patterns between the clay-rich soil matrix, which is slightly depleted in the LREE when normalized to PAAS (La/YbPAAS 0.5), and the secondary mineral phase jarosite, which is enriched in the LREE (La/YbPAAS = 15–50). The REE pattern in the porewater changes with the transition from jarosite- to goethite-rich mottles, becoming relatively more enriched in the LREE compared to the HREE, which is consistent with the incongruent dissolution of jarosite to form goethite and the release of greater amounts of jarosite REE to solution, including proportionately more of the jarosite-compatible LREE. Maximum surface water REE concentrations in acidic water were 100–200 ppb La and Ce. REE patterns in surface water were very similar to the porewater transition zone, enriched in the MREE, but asymmetric, relatively enriched in the LREE compared to the HREE. Ó 2008 Elsevier Ltd. All rights reserved.

1. INTRODUCTION Due to their similar geochemical behavior, the rare earth elements (REE) have been used as environmental tracers

* Corresponding author. Address: Research School of Earth Sciences, The Australian National University, Canberra ACT 0200, Australia. Fax: +614 292 7688. E-mail addresses: [email protected], [email protected] (S.A. Welch).

0016-7037/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.gca.2008.10.017

for solid earth, sedimentary and aqueous systems (cf. Nance and Taylor, 1976; Taylor and McLennan, 1985; McLennan, 1989; Johannesson et al., 2006). Previous work on the concentrations and distributions of REE in natural waters has shown that the concentrations are a complex function of the source of REE, geochemical processes such as adsorption, precipitation, ion exchange or mixing, solution composition (particularly pH), and concentrations of complexing ligands (Brookins, 1989; Elderfield et al., 1990; Johannesson et al., 1996; Gammons et al., 2003, 2005a,b). The ultimate source of REE in natural waters is

REE in acid sulfate soil

the weathering of bedrock. Therefore, in systems with low water/rock ratios, such as soils and groundwaters, the extent of water–rock interaction has a profound impact on the geochemistry of the REE in both solution and the solid ˚ stro¨m, 2001; Hanniphase (cf. Johannesson et al., 1996; A ˚ stro¨m and Corin, 2003; Haley gan and Sholkovitz, 2001; A et al., 2004; Verplanck et al., 2004). In most other natural waters such as rivers, lakes and seawater, where there is limited water–rock interaction, the major controls on REE composition and concentration are solution variables, predominantly acidity, and the extent of carbonate, sulfate and phosphate complexation (cf. Elderfield et al., 1990;Wood, 1990; Johannesson et al., 1996, 2006). However, a significant fraction of the total REE can be associated with a suspended or colloidal phase even in these systems, and this can result in fractionation between the truly ‘dissolved’ and ‘solid’ phase (Elderfield ˚ stro¨m, et al., 1990; Dia et al., 2000; Ingri et al., 2000; A ˚ 2001; Astro¨m and Corin, 2003; Gammons et al., 2005a,b). There have been several recent studies specifically focusing on the concentration and distribution of REE in natural and anthropogenically altered acidic environments (Johannesson et al., 1996; Lewis et al., 1997; Verplanck et al., 1999, 2004; Leybourne et al., 2000, 2006; Serrano et al., ˚ stro¨m, 2001; Worrall and Pearson, 2001; Protano 2000; A ˚ stro¨m and Corin, 2003; Gammons and Riccobono, 2002; A et al., 2003, 2005a,b; Bozau et al., 2004; Merten et al., 2005; Olias et al., 2005). Reported REE concentrations in these environments are extremely high: up to ppm levels of La ˚ stro¨m, 2001; Merten et al., 2005; and Ce in water (e.g., A Wood et al., 2006). Most of these studies show that the REE distribution has a relative enrichment in the middle rare earths (MREE) compared to the light rare earths (LREE) and heavy rare earths (HREE) when normalized to an Earth surface reservoir. There are many possible mechanisms for this, and the predominant mechanism is a function of the environmental conditions. MREE enrichments have been attributed to the original composition of the source water or rock, fractionation onto colloidal (Fe) phases, exchange between solution and mineral phases that are enriched in MREE, dissolution of MREE-rich mineral phases, phosphate complexation or sulfate complexation, ˚ s(Johannesson et al., 1996; Verplanck et al., 1999, 2004; A tro¨m and Corin, 2003; Gammons et al., 2003). Although MREE enrichment is common in acidic environments, it is not exclusive to these environments (e.g., Hannigan and Sholkovitz, 2001; Haley et al., 2004) nor is it always found in acidic environments (Lewis et al., 1997; Bozau et al., 2004; Merten et al., 2005). In fact, many river systems have a distinct MREE enrichment that is attributed to fractionation between ‘dissolved’ and colloidal phases (Elderfield et al., 1990). This study focuses on the REE geochemistry of sediments, porewater and surface water at an acid sulfate soil site that is undergoing remediation by managing water levels. The REE have been used to determine the extent of water–rock interaction, the fate of trace metals that were leached by oxidation and acidification, and how the chemistry of the site has responded as a result of flooding. Previous work on jarosite sourced from this site has

45

demonstrated that jarosite incorporates the LREE into its structure (Welch et al., 2007). However, when jarosite was reacted in slightly acidic solutions, there was a fractionation of the REE between solution and the solid phase, with solutions being enriched in the MREE. 2. FIELD SITE AND HISTORY The field area is Mays Swamp, an acid sulfate soil site near Kempsey, New South Wales, Australia (Fig. 1) approximately 400 km north of Sydney (30° 550 4500 S 152° 550 4800 E). This area has been identified as one of the acid sulfate soils hot spots (Tulau and Naylor, 1999) by the NSW Department of Natural Resources and has been the focus of several other studies (Somerville et al., 2004; Beavis et al., 2005; Isaacson et al., 2006; Welch et al., 2007, 2008). The lithology of the area is a sulfidic grey clay deposited in an estuarine or lagoonal environment during the last glacial high stand. The swamp itself is a semicircular depression with an area of approximately 4 km2. Surface elevation decreases approximately 40 cm from north to south. The southernmost edge of the swamp is separated from the Seven Oaks Drain by an elevated dirt track, though there are several culverts that undercut the track which allow for exchange between surface water in the swamp and the Seven Oaks Drain. The East Drain to the west of the swamp limits surface water runoff into Mays swamp from the adjacent acidic field and acidic areas along the Clybucca Creek, although during periods of flooding there is surface water exchange in these low-lying areas. In the 1960s and 1970s, drains were constructed in the area to increase agricultural and pastoral productivity and mitigate flooding. This has resulted in oxidation of the underlying sediments, acidification and intense scalding. In the 1990s the site was severely degraded, extremely acidic and almost devoid of vegetation (personal communication from R. Yerbury, landowner at the field site). The site has changed dramatically over the last several years as a result of environmental management aimed at limiting pyrite oxidation and managing acidic conditions. In 2001, a weir was installed in the Seven Oaks Drain near the eastern edge of the swamp to induce flooding and raise the water table. The vegetation of the lower lying areas of Mays Swamp has changed significantly since the implementation of hydrological management and a reduced grazing regime. Periods of prolonged ponding and higher water tables have induced revegetation of previously scalded and denuded areas of the swamp with the water tolerant species of water couch and spike rush (Paspalum distichum and Eleocharis equisetina) dominating. Surface water pH is predominately nearneutral in vegetated areas, however, subsurface sediments are still extremely acidic: pH 3–4 down to approximately 1 m depth, due to the oxidation of pyrite and the dissolution of abundant jarosite mottles (Beavis et al., 2005; Welch et al., 2007, 2008). 3. METHODS Surface water and sediment samples were collected on several occasions between 2003 and 2007 for study. An

46

S.A. Welch et al. / Geochimica et Cosmochimica Acta 73 (2009) 44–64

Fig. 1. Map of the field site. (A) Coast of Australia showing location of field site near Kempsey NSW and insert of Australia. Map image is Ó copyright Commonwealth of Australia (Geoscience Australia) [2008] (B) Aerial photo showing Mays swamp (outlined in white), East Drain which is a major conduit for acidic water from low-lying areas to the west of Mays swamp, the Seven Oaks drain to the south of the field area which flows into the Clybucca Creek and the area south of Mays swamp (SS or southern scald) where an acid scald had recently formed.

extensive survey of sediment and water analysis was conducted in February 2006, after a period of prolonged inundation of the field site. Several water samples were also collected in April 2006 and April 2007. Surface water samples in the flooded pasture and in the Seven Oaks and East drains were collected in HDPE bottles. Water pH, electrical conductivity (EC), dissolved oxygen (DO) and Eh were measured using field meters calibrated with standard solutions. Surface conditions, water depth, turbidity, plant species, density and health were also noted when surface standing water samples were collected. Water samples were filtered in the field with 0.45 lm membrane syringe filters. Two aliquots were collected, one sample was acidified with nitric acid to pH <2 for cation and trace metal analysis, the second aliquot was not acidified and analyzed for anions. Sediment cores were collected from the field site in April 2005 and February 2006 from approximately the same area (Fig. 1). The core taken in 2005 was obtained using a 1.5 m long, 6 cm diameter PVC pipe fitted with a core catcher in the bottom. The core was cut in half lengthwise was allowed to air dry and then impregnated with resin and cut into polished sections. A section of the core that had abundant jarosite mottling at approximately 40 cm below the surface was examined by light and electron microscopy. The composition of major ions in the clay matrix and jarosite mottles was determined on approximately 40 spots with a JEOL JSM6400 Scanning electron microscope (SEM). Due to the high porosity of the sediments and the hydrous minerals present, total oxides were generally less than 80%, so concentrations were normalized to 100%. The block was also analyzed using a Cambridge S360 SEM equipped with both backscattered and secondary electron detection to determine the spatial relationship between the jarosite mottles and the clay matrix, and to determine the abundance of rare earth element bearing minerals such as monazite or allanite. Semi-quantitative analyses were done on selected spots using Energy Dispersive X-ray Analysis (EDXA). Major and trace element composition of clay and jarosite on the

polished section was then determined by laser ablation ICP-MS using an Agilent 7500 ICP-MS with a Resonetics ArF+ excimer laser operating in the far UV (193 nm). Samples were analyzed using an 86 lm spot size. Isotopes analyzed were 24Mg, 27Al, 29Si, 31P, 43Ca, 45Sc, 49Ti, 51V, 53Cr, 55 Mn, 56Fe, 57Fe, 59Co, 60Ni, 63Cu, 65Cu, 66Zn, 75As, 85Rb, 88 Sr, 89Y, 138Ba, 139La, 140Ce, 141Pr, 146Nd, 147Sm, 153Eu, 157 Gd, 163Dy, 166Er, 169Tm, 172Yb, 175Lu, 208Pb. Concentrations were initially calculated with respect to a standard reference glass BCR-2g. Major and trace elements concentrations were then recalculated based on major element composition obtained from the JEOL JSM6400 SEM. Trace element concentrations were normalized with respect to the sum of SiO2 + Al2O3 + MgO = 90% for the clay matrix and to average Fe in the jarosite mottles using data obtained from analysis by the JEOL SEM. Three cores collected along a transect perpendicular to the Seven Oaks Drain in Feb. 2006 were sealed in the field and stored upright under cool conditions until they were cut. Sediment cores were described, and then porewater in the different soil horizons was extracted by filter centrifugation using a 0.45 lm filter. Porewater pH and EC were immediately measured using micro-electrodes calibrated with standard reference solutions. Aliquots of the extracts were then diluted in either milli Q water or 2% nitric acid for analysis. Dilution factors were on the order of 5- to 200fold depending on the conductivity, the analytical method and the amount of pore water collected. In some instances, the pore water extracts were diluted twice to obtain sufficient sample volume for analysis of trace metals. Anion samples were analyzed using a Dionex 4000i ion chromatograph to determine F, Cl, Br, NO3  , SO4 2 and PO4 3 . Cations including Si and trace elements were analyzed using a Varian VISTA AX CCD Simultaneous Inductively Couple Plasma Atomic Emission Spectrometer (ICP-AES) and a Varian Ultra Mass Inductively Coupled Plasma Mass Spectrometer (ICP-MS) to determine major and trace element composition respectively. Detection limits are element

REE in acid sulfate soil

ching and Brune (1991) and from the Visual MINTEQ vers 2.53 data base (Allison et al., 1991). 4. RESULTS The distribution of major and trace element composition in porewater extracts from the cores have been described in detail in Isaacson et al. (2006) and Isaacson (2006) and will only be described briefly here. Chemical analysis of sediments and water from this site have been described elsewhere (Somerville et al., 2004; Beavis et al., 2005; Isaacson et al., 2006; Welch et al., 2007, 2008). 4.1. Sediment description In the 2006 survey, sediments from Mays Swamp were collected along three transects perpendicular to the Seven Oaks Drain (Fig. 1). Intact sediment cores were collected from 9 sites with 1.5 m long 6 cm OD core tubes, although data will only be presented from transect 2. Additionally, a 1.5 m long gouge auger was used to collect sediment cores that were described on site and separated based on stratigraphic unit. The soil/sediment profile is similar throughout the site, though the thickness of the units is variable (Fig 2). Working downwards, the profile comprises: (i) (ii) (iii) (iv) (v) (vi)

a dark, peaty, organic-rich A1 horizon. an acidic alluvial A2 horizon. an oxidized jarositic mottled clay zone . a red/brown iron oxide/hydroxide mottled clay zone. a shell layer, and a thick, unoxidized glauconitic estuarine sub-soil.

In core 1, the jarosite mottled zone had significantly less jarosite towards the top of the profile compared to the other core sections that were obtained. Analysis of the bulk sediment and clay fractions by XRD shows that the clay matrix contains traces of quartz, K-feldspar and gypsum. The chemical and mineralogical composition of the clays change down profile: in the upper part of the profile the clay is dominated by illite, but it grades to an illite/glauconite composition with depth. This lithology has partly evolved in response to the hydrological

core

1

2

3

0 Organic rich reducing A horizon

20 Alluvial A horizon

depth cm

specific for both the ICP-AES and ICP-MS. Detection limits range from 1 to 50 ppb for most metals analyzed on the ICP-AES. Detection limits in the ICP-MS range from 0.1 ppb down to 0.01 ppb for some of the REE. Major and trace element concentrations were determined using calibration curves with standard concentrations bracketing the samples concentrations. Corrections for mass interferences on the mass spectrometer were done as necessary using the calibration software. The rare earth isotopes that were quantified were 139La, 140Ce, 141Pr, 146Nd, 147Sm, 153 Eu, 157Gd, 159Tb, 163Dy, 165Ho, 166Er, 169Tm, 172Yb, 175 Lu. These were chosen to limit isobaric interference with the other REE and oxides of Ba and the LREE Trace element concentrations were obtained by both ICP-AES and ICP-MS analysis when possible, and compared. For samples whose concentrations fell within the working range of both instruments (ca. 50–500 ppb), there was good agreement between the two analytical methods, generally <20% difference. A quality control standard was run as a sample to check accuracy of the analysis for both the ICP-AES and ICP-MS analysis. Selected sediment samples from the three cores used for pore water extraction were analyzed by X-ray diffraction to confirm mineralogy. Small samples of the cores were also analyzed by scanning electron microscopy using a Cambridge S360 SEM and Hitachi 4300 SE/N SEM. Mineral phases were not quantified on these specific core samples, but clay and bulk soil mineralogy was determined on similar cores from this area using a Siemens D501 X-ray Diffractometer (Somerville et al., 2004; Beavis et al., 2005). Solution speciation and saturation indexes for mineral phases were initially calculated using PHREEQC Interactive V 2.12.5.669 with the LLNL (EQ3/6) database (Parkhurst and Appelo, 1999). The database was updated for the REESO4 þ complexes using the stability constants for I = 0 from Schijf and Byrne (2004), (their Table 3). Because there were numerous errors for the REE species noted in the LLNL (EQ3/6) database, the stability constants for the major expected REE species were replaced. The 1:1 REE-Cl, PO4 3 and F- complexes were updated with the values from the Visual MINTEQ version 2.53 database (Allison et al., 1991). The REE(OH)2+ stability constants were updated from Klungness and Byrne (2000). The geochemical modelling calculations were also redone using stability constants for REEðSO4 Þ2  species sourced from the Visual MINTEQ version 2.53 data base, from Wood (1990); and with the constants from the LLNL (EQ3/6) data base (Parkhurst and Appelo, 1999). The error for the NdðSO4 Þ2  and the omission of the CeðSO4 Þ2  stability constants in the LLNL (EQ3/6) data base were corrected by estimating values from the other REEðSO4 Þ2  stability constants, assuming that there would be a systematic trend across the series. Geochemical modelling calculations were all done with the stability constants for REEðSO4 Þ2  stability constants omitted from the data base. Since previous studies have shown that the solubility of REEPO4 mineral phases are important in controlling the concentration of REE in natural waters (Firsching and Brune 1991; Byrne and Kim, 1993; Johannesson et al., 1995), the database was updated with the logK values for REEPO4 from Firs-

47

40

Oxidized B horizon

60

Jarosite mottled B horizon

80 100 120

Iron hydroxide mottled B horizon Shelly transition B horizon Glauconite reduced C horizon

140

Fig. 2. Schematic diagram of sediment profiles in the three cores.

48

S.A. Welch et al. / Geochimica et Cosmochimica Acta 73 (2009) 44–64

management that has been a dominant influence on the hydrology and hydrogeochemistry of the swamp since the construction of Seven Oaks Drain. The most marked recent change is the formation of the sulfidic organic-rich A horizon, which has developed over the last two years prior to sampling in the area of core 2. 4.2. Porewater chemistry The chemistry of the porewater changes as a function of depth in the profiles and of distance from the drain (Fig. 3). Sulfate is the major ion in these porewaters, and comprises approximately 50–70 wt% of the total dissolved solids (TDS). The sulfate is the product of pyrite oxidation that occurred when the site was drained and oxidized (Tulau and Naylor, 1999). In all of the cores sulfate concentration is lowest at the surface and then increases with depth. In cores 1 and 2, sulfate concentration is nearly constant below 20 cm. However, in core 3, sulfate concentration increased down to ca. 60 cm and then decreased below 80 cm. The pH profile for all the cores has the same general C-shaped trend: there is an acidic maximum zone in the subsurface in the middle of the profile, from which pH increases slightly towards the surface and also increases with depth. The acid zone varies in thickness and intensity across the field site (Fig. 3a). The increase in pH towards the surface could reflect simple dilution of the porewater with the standing water on the surface. This is consistent with the decrease in sulfate and TDS that is also observed over this zone in the three cores. The decrease in acidity is also due in part to biologically mediated reduction of sulfate and iron and formation of pyrite in the organic-rich soils. Previous field work in this area in 2004 (Beavis et al., 2005) showed that the extremely acidic conditions persisted up to the land surface at this site, and reduced sulfur minerals (pyrite and iron monosulfides) were not observed. The zone of maximum acidity is coincident with extensive jarosite mottling in core 2 and 3 (Figs. 2 and 3). The dissolution of jarosite and precipitation of goethite buffers the acidity at approximately pH 3.5–3.7 (White et al., 1997; Welch et al., 2008). Porewater pH is slightly higher in core 1, at approximately 4–4.5. Note that this area is near

4

5

6

7

8

0

0

The concentrations of selected rare earth elements and Y in the porewater are depicted in Figs. 4 and 5. The LREE along with Y were detected in most of the porewater extracts. In many of the samples, the less abundant middle and heavy rare earth elements were below detection limits. The concentrations of La, Ce and Y are plotted as a function of depth in the three cores (Fig. 4). The concentrations are low near the surface, and then increase with depth to the acidic zone. There is a good correlation of REE concentration with increasing acidity (Fig. 5) indicating that solution pH is a major control on concentration. However, the maximum concentrations of La, Ce, Y and Gd do not correspond with the lowest pH, but were found at pH of about 4. The concentration of Y is similar to the concentration of La in core 1 and 2, but in core 3 the Y concentrations are ca. 20% higher. All of the profiles show a sharp increase in concentration of these elements at the bottom of the acid zone, just as pH is beginning to increase (Fig. 4). This horizon also corresponds to a change in mineralogy from jarosite-rich to goethite-rich mottles (Fig. 2). The maximum concentrations in this layer are around 1 ppm for La and Y and 3 ppm for Ce. Below the maximum-concentration zone, the concentrations decrease dramatically to low ppb levels. This is coincident with a large increase in pH in the porewater from approximately 4 to 7, due to buffering by the carbonate minerals of the fossilized shell layer present. The measured REE in the porewater extracts were normalized to Post-Archaean Australian Sediment (PAAS, Nance and Taylor, 1976) for each depth interval in each core (Fig. 6). Although there is some variability,

2000

4000

TDS mg/L

6000

core 1 core 2 core 3

0

8000

0

a

5000

10000

15000

0

b

c

20

20

40

40

40

60

60

60

80

80

80

100

100

100

120

120

120

20

depth (cm)

4.3. REE in porewaters

SO4 mg/L

pH 3

the Seven Oaks Drain and is lower in the landscape and therefore would have experienced much more frequent and persistent periods of inundation. Although jarosite is present in this core, it is not as abundant as in core 2 and 3. The increase in pH with depth towards the bottom of the three cores reflects the buffering capacity of the shell layer, though in core 1 the shell layer is significantly reacted compared to the other two cores.

Fig. 3. (a) pH, (b) SO4 2 and (c) total dissolved salts (TDS) in porewater extracts versus depth for three sediment cores.

REE in acid sulfate soil

49

ppb 0

500

1000

1500

depth cm

0

0

2000

1000

2000

3000

4000

a

0

20

20

40

40

40

60

60

60

80

80

80

100

100

pH Ce

5

6

7

8

c

core 2

core 3

120 4

3000

100

La Y

core 1 3

2000

b

20

120

1000

0

0

120

3

4

5

6

7

8

3

4

5

6

7

8

pH

Fig. 4. LREE and Y in porewater concentrations as a function of depth for the LREE and Y in (a) core 1 closest to the drain (b) core 2 and (c) core 3.

10000

10000

a

b

1000

Ce ppb

La ppb

1000

1 2 3

100 10 1

100 10 1

0

0 2

4

6

8

2

4

1000

10000

c

8

6

8

d

1000

Y ppb

100

Gd ppb

6 pH

pH

10

100 10

1

1

0

0 2

4

6 pH

8

2

4 pH

Fig. 5. (a) La, (b) Ce, (c) Gd and (d) Y concentration in the porewater extracts versus pH. All the trace metals show a strong correlation with acidity, except the maximum concentrations occur at the bottom of the oxidized profile.

all the plots show enrichment in the MREE, typically peaking at Gd. The shape of the pattern changes slightly down profile (Fig. 6 and Table 1). Nearer the surface, where pH is higher and concentrations are lower, the patterns are flatter ((La/Gd)PAAS from 0.65 to 0.78, Table 1). As concentrations increase, the patterns become progressively enriched in the MREE ((La/Gd)PAAS 0.35 to 0.44 and (Yb/Gd)PAAS 0.34 to 0.41; Table 1). In core 3, the REE patterns are relatively symmetric, even for the highest concentrations ((La/Gd)PAAS = 0.35 and (Yb/

Gd)PAAS = 0.40; and Table 1, Fig. 6c). However, in cores 1 and 2, the normalized REE become asymmetric as concentrations increase, being preferentially enriched in the LREE over HREE ((La/Gd)PAAS 0.54 to 0.56 and (Yb/Gd)PAAS 0.29 to 0.34; Table 1). This also corresponds to the change in mineralogy of the core, from jarosite- to goethite-rich mottles. The LREE enrichment is most apparent for the two maximum-concentration samples in cores 1 (>53 cm) and 2 (35–42 cm).

50

S.A. Welch et al. / Geochimica et Cosmochimica Acta 73 (2009) 44–64 10-1 0-8 8-15 15-41a 15-41b

-2

10 REE/PAAS

41-53 53-85 53-85b 53-85c 85-120

10-3

85-120b

10-4 La

Ce

Pr

Nd

Sm

Eu

Gd

Tb

Dy

Ho

Er

Tm

Yb

Lu

10-1

0-6a 0-6b 6-16a

REE/PAAS

10

-2

6-16b 17-20a 17-20b 20-26

10

-3

26-31 31-35a 31-35b 35-42

10-4 La

Ce

Pr

Nd

Sm

Eu

Gd

Tb

Dy

Ho

Er

Tm

Yb

Lu

-1

10

10-2 REE/PAAS

0-10 10-16 16-56a 16-56b 16-56c

10-3

56-77a 56-77b

10-4 La

Ce

Pr

Nd

Sm

Eu

Gd

Tb

Dy

Ho

Er

Tm

Yb

Lu

Fig. 6. REE in the pore water (in ppm) normalized to Post-Archaean Australian Sediment (PAAS, Nance and Taylor, 1976) (a) core 1, (b) core 2 and (c) core 3. The numbers in the legends represent depth intervals in cm.

4.4. Surface water chemistry Surface water samples were collected on several occasions from the site. Nine surface water samples were collected during February 2006, when the site was extensively flooded. Most of these samples were near-neutral pH, and had low to below detection levels of REE (Table 2). In the neutral pH surface water, the redox chemistry was variable over a few cm water depth, with sulfidic flocs apparent, both suspended in water and coating the grasses

deeper in the water, but with oxidized iron flocs suspended in water and coating grasses at shallower depth. The water surface also had an oily sheen that is characteristic of iron oxidation. The distribution of the oxidized and reduced zones was variable over the surface, with the reduced material being relatively more abundant at the sites closer to the drain, where the standing water was deeper, or where vegetation was relatively dense. However, three of the samples collected were acidic: two were from standing water in a scald on the southern side of

REE in acid sulfate soil Table 1 Values for [La/Gd]PAAS, [Gd/Yb]PAAS, [Yb/Gd]PAAS, and [La/ Yb]PAAS. La/Gd

Gd/Yb

Yb/Gd

La/Yb

Pore water Core 3 10–16 Core 3 16–56c Core 3 56–77b Core 2 17–20a Core 2 20–26 Core 2 31–35 Core 2 35–42 Core 1 15–41 Core 1 41–53 Core 1 53–85c Core 1 85–120

0.66 0.35 0.34 0.65 0.39 0.44 0.54 0.78 0.44 0.55 0.56

1.21 2.44 2.51

0.83 0.41 0.40

0.80 0.86 0.87

2.97 2.41 3.36

0.34 0.42 0.30

1.17 1.05 1.81

2.76 2.97 3.41

0.36 0.34 0.29

1.21 1.63 1.90

Surface water ms2c1 ms2c2 ms3a4 drain ss1 ss2 dead

0.46 0.38 0.32 0.51 0.70 0.62

2.23 6.69 3.34 2.43 5.66 4.21

0.45 0.15 0.30 0.41 0.18 0.24

1.03 2.57 1.06 1.23 3.96 2.62

the Seven Oaks Drain, approximately 500 m from the end of transect 2, that had recently formed and was covered with dead vegetation, and one from the Seven Oaks Drain at the weir. The acidic sites tended to have abundant oxidized iron flocs suspended in the water and sorbed to vegetation surfaces. The acidic samples had concentrations of La and Ce of 10–20 ppb (Table 2 and Fig. 7). Concentrations in the surface water show a strong pH dependence, similar to what was observed for the porewater, though REE concentrations were at least an order of magnitude lower in the surface water. In April 2007, surface water and drain water samples were collected during a flood event that occurred after prolonged dry conditions. In contrast to the conditions found the previous year, most of the surface water samples collected in the area of the three transects and in the East Drain were acidic, though the water in the Seven Oaks Drain was predominantly near-neutral pH (Table 2). The

51

concentrations of La and Ce in the surface water ranged from low ppb levels in the near-neutral pH samples to several hundred ppb in the acidic samples. Again, there was a good correlation between acidity and REE concentration in all the surface water samples (Fig. 7). However, concentrations were approximately an order of magnitude higher during the 2007 sampling trip than for the 2006 samples. The concentrations of REE in the surface water samples collected in 2006 and 2007 were normalized to PAAS (Fig. 8). In many of the neutral pH water samples, the MREE to HREE element concentrations were below detection limits. Where data was available, the patterns for the neutral pH samples were relatively flat to slightly enriched in the MREE (Table 1 and Fig 8). The REE patterns for the acidic samples (pH <5) were very similar to the patterns seen for the highest porewater concentrations: all were enriched in the MREE, but the pattern was asymmetric, relatively more enriched in the LREE, especially La and Ce. 4.5. REE speciation in water The speciation of major and trace elements in selected pore and surface waters was calculated with PHREEQC Interactive v2.12.5 using the LLNL (EQ3/6) database (Parkhurst and Appelo, 1999) that had been modified to include new stability constants for REE F, Cl, OH and SO4 species (see Section 3). Speciation calculations were also done for selected water samples both with and without stability constants for the REEðSO4 Þ2  complexes sourced from the LLNL (EQ3/6) database, the Visual MINTEQ Vers 2.53 database and from Wood (1990). All of these calculations produced similar results. However, we present here only the outputs with the REEðSO4 Þ2  stability constant sourced from Wood (1990) and those with the stability constants for the REEðSO4 Þ2  species omitted from the data base. The most abundant REE species (>1% of total dissolved element) are depicted in Fig. 9 for six representative porewater samples and Fig. 10 for three surface water samples. The patterns are similar for all the samples. When the REEðSO4 Þ2  complexes are included in the modelling, the REE composition is dominated by the REEðSO4 Þ þ species. The proportion of the REEðSO4 Þþ species increases across

Table 2 Measured surface water chemistry parameters for samples with detectable levels of rare earth elements. Sample

Date

pH

SO4 2 mg/L

EC (lScm1)

La (ppb)

Ce (ppb)

ms2c1 ms2c2 ms3a4 drain ss1 ss2 dead

Feb Feb Feb Feb Feb Feb

6.02 5.82 5.66 3.67 3.17 3.00

1.65 47.7 20.1 272 1406 946

346 369 144 823 2800 2000

1.3 0.5 0.2 9.1 9.2 9.3

3.3 1.3 0.5 18.1 14.9 17.0

ED1 ED3 ED5 ms11 ms15 ms20 ms30 ms37

April April April April April April April April

6.8 5.4 4.85 3.77 3.01 2.82 2.92 3.86

4483 1851 5703 3930 3853 2788 3565 3951

1045 1235 1786 4970 5780 3850 4220 5900

0.1 6.1 57.6 0.6 112 301 154 43.3

0.3 10.5 96.1 1.0 290 605 320 95.4

2006 2006 2006 2006 2006 2006 2007 2007 2007 2007 2007 2007 2007 2007

52

S.A. Welch et al. / Geochimica et Cosmochimica Acta 73 (2009) 44–64 1000 La 2006 Ce 2006 La 2007

ppb

100

Ce 2007

10

1

0.1 2

4

6

8

pH

Fig. 7. Concentration of La and Ce as a function of pH for surface water samples collected in February 2006 and April 2007.

10-3

a ms2c1

10-4

REE/PAAS

ms2c2 ms3a4 drain ss1 ss2

10-5

10-6 La

Ce

Pr

Nd

Sm

Eu

Gd

Tb

Dy

Ho

Er

Tm

Yb

Lu

-1

10

b 10-2

ED3

REE/PAAS

ED5 MS11 MS15 10-3

MS20 MS30 MS37

10

-4

10-5 La

Ce

Pr

Nd

Sm

Eu

Gd

Tb

Dy

Ho

Er

Tm

Yb

Lu

Fig. 8. REE of surface waters normalized to PAAS collected in (a) Feb 2006 and (b) April 2007. Sample names for (a) are ms2c1 and ms2c2 which are surface waters taken near core 1 and core 2, respectively, ms3a4 was taken approximately 300 m west of core 3 (Fig. 1), drain is the from the Seven Oaks drain at the weir, ss1 and ss2 are from a scalded area on the southern side of the drain approximately 500 m from the end of transect 2. Sample names for (b) are ED taken from the East Drain and MS from the surface of Mays Swamp near the transect 1 and 2 from 2006.

the series from ca. 75 to 80% from La to Pr, and then decreases systematically to approximately 50% to 60% for LuSO4 þ . The one notable anomaly is for Eu where the

abundance of the EuSO4 þ dips to approximately 60% with a corresponding sharp increase in the EuðSO4 Þ2  species. The stability constants for the monosulfate complexes from

REE in acid sulfate soil 100

% species

80 60

40

a +

REESO4 3+ REE REE(SO4)2 REE-P REE-ST + REESO4 b

20

100

d 80

60

40

20

0

0

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

% species

100

b

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu 100

80

80

60

60

40

40

20

20

0

% species

e

0

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu 100

53

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu 100

f

c

80

80

60

60

40

40

20

20

0

0

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Fig. 9. Speciation of REE in porewater extracts from (a) core 1, 0–8 cm, pH 6.3, (b) core 1, 41–53 cm, pH 4.7, (c) core 1, 85–97 cm, pH 4.35, (d) core 2, 0–6 cm, pH 4.56, (e) core 2, 17–20 cm, pH 3.6 and (f) core 2, 36–42 cm, pH 4.57. REE-ST is the total of REESO4 þ and REEðSO4 Þ2  . REE-P is the total of REEPO4 and REEHPO4 þ . REESO4 þ b is for speciation calculations with the disulfate complexes omitted.

Schijf and Byrne (2004) vary systematically across the series, increasing slightly across the series (0.03 log units) from La to Eu and then decreasing (0.2 log units) from Eu to Lu. However, the values for the disulfate REE complex given in Wood (1990) do not show this same systematic trend across the series, and have anomalously high values for the EuðSO4 Þ2  species (0.2 log units) compared to the other REE, hence the depicted deviations. The total REE sulfate complexes (REESO4 þ + REEðSO4 Þ2  ) comprise 80–90% of the REE, and decrease systematically with increasing atomic number. There is a corresponding increase in the sulfate-free REE3+ from approximately 10% to 20% across the series. The only other complexes that were abundant (>1%) were REEPO4 and from the surface porewater sample of core 1 (Fig. 9) and REECl2+ and REEF2+ species from two of the surface water samples collected (Fig. 10). Results of the modelling without the stability constants for the REEðSO4 Þ2  complexes show that when this com-

plex is not considered, the abundance of the REESO4  species is still dominant, and is similar to the total for the (REESO4 þ + REEðSO4 Þ2  ). Saturation indices (SI) with respect to REE minerals were also determined from the speciation calculations. The data base was altered to include the solubility constants for REEPO4 phases from the Visual MINTEQ Vers 2.53 database (Allison et al., 1991) as well as with the values from Firsching and Brune (1991). Solutions were greatly undersaturated with respect to most of the REE-bearing mineral phases in the LLNL (EQ3/6) database, however, many of the solutions are supersaturated with respect to REEPO4 and REEPO410H2O compounds (Table 3). The REEPO410H2O phases are not known as minerals but evidently crystallize readily under laboratory conditions, and function as proxies for less hydrated, less soluble mineral phosphates of the monazite, xenotime, brockite or rhabdophane groups. The saturation indices for the REEPO4 phases are approximately 0.5–1 log units higher than for

54

S.A. Welch et al. / Geochimica et Cosmochimica Acta 73 (2009) 44–64 100

% species

80 60 40

pH 5.41

20

REESO4+ REE3+ REE(SO4 )2REE ST REEF2+ REECl 2+

0 La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu 100

approximately 20 to 30% across the series. REE-Cl2+ and REEF2+ species were also present in these surface samples at levels around 1–2% of the total REE. Phosphate concentrations in the surface waters were much lower than in the porewater, from undetectable (<0.1 mg/L) up to 1 mg/L PO4 3 in surface waters compared to 1–3 mg/L PO4 3 in porewater. The near-neutral pH sample which had the lowest concentration of REE was saturated with respect to LaPO410H2O and CePO410H2O (Table 3) and supersaturated with respect to the LREEPO4 phases. The more acidic surface water samples were all undersaturated with respect to the REEPO410H2O and REEPO4 compounds.

80

% species

4.6. Sediment analysis 60

pH 3.86 40 20 0 La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

100

% species

80 60

pH 2.82 40 20 0 La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu

Fig. 10. Percent REE speciation for selected surface waters for (a) East Drain 3, pH 5.41, (b) Mays Swamp 37, pH 3.86 and (c) Mays Swamp 20, pH 2.82. REE-ST is the total of REESO4 þ and REEðSO4 Þ2  . REESO4 þ b is for speciation calculations with the disulfate complexes omitted.

the REEPO410H2O compounds. The surface porewater sample in core 1 was near-neutral pH and has the lowest REE concentrations, but has highest degree of supersaturation with respect to REEPO4 and REEPO410H2O. All the other porewater samples are acidic and have higher total REE, but lower SI due to the pH dependence on the solubility of these compounds. All the porewaters were supersaturated with respect to REEPO4 and REEPO410H2O for the LREE (SI = 1–2). SI generally decreased with increasing atomic number, and alternated higher and lower across the series reflecting the relative abundance of the REE at the Earth’s surface. The REE speciation in surface waters was generally similar to that in porewaters (Fig. 10). REESO4 þ species dominated. The relative abundances of REESO4 þ and REEðSO4 Þ2  were lowest in the least acidic (pH 5.41) water sample, decreasing from 77% for La to 70% for Lu. This sample also had the highest relative abundance of the ‘‘free” (aquated) REE3+ complexes, increasing from

In 2005, an intact core was collected in the same area as core 2. This core was allowed to dry in air, was impregnated with resin, and then cut into sections. A jarosite-rich section from approximately 40 cm depth was analyzed by light and electron microscopy and laser ablation ICP-MS to determine the concentration and distribution of REE phases. A single submicron REE-bearing mineral grain was found within the clay matrix with backscattered electron imaging in the SEM. The composition of this area determined by spot analysis using EDXA showed peaks for Ce, La, Nd, Th and P along with Si, Al and Fe. It is not possible to identify the mineral species from this spot analysis because the EDXA data was only semi-quantitative and because of interference from the background clay matrix. It is possible that this mineral was an authigenic REE phase such as florencite or rhabdophane but due to the presence of detectable Th, it is more likely that this grain was a detrital monazite. Major and trace element concentrations in the clay and the jarosite mottles were determined by laser ablation ICPMS operating in scanning mode. Data was obtained from five transects that were approximately 86 lm wide and 1 cm long. The data from the scans were then broken into intervals for segments that were dominated by clay, as evidenced by high Al and Si count rates, and jarosite, evidenced by high Fe. Concentrations of major and trace elements were initially calculated with respect to a standard reference glass, and the data for each segment was then normalized using major element data obtained from quantitative spot analysis with the SEM. For the clay fraction, all the laser ablation data was normalized assuming the sum of (SiO2 + Al2O3 + MgO) was 90 wt%. For the jarosite mottled zones, all of the laser ablation data was recalculated assuming the jarosite contained 33 wt% Fe (Welch et al., 2008). The concentrations of La in the clay intervals were all similar, about 15–20 ppm, or about half the average crustal abundance (Nance and Taylor, 1976; Taylor and McLennan, 1985). The concentrations of REE normalized to PAAS for La, Ce, Pr, Nd, Sm, Eu, Gd, Dy, Er, Tm, Yb, and Lu are plotted in Fig. 11. All of the clay intervals have a similar pattern: slight depletion in LREE, with a linear increase to slight enrichment in HREE. The ratios (REE/REEPAAS) range from 0.4 to 0.5 for La up to 1.0–1.4 for Lu, so (La/Lu)PAAS = 0.3–0.5. Two of the scanned intervals for

REE in acid sulfate soil

55

Table 3 Saturation index with respect to REEPO410H2O for six representative porewater and three acidic surface water samples, calculated using PHREEQCi V 2.12.5.669 (Parkhurst and Appelo, 1999). Saturation indices relative to anhydrous phosphates calculated for data of Firsching and Brune (1991) and also for the logK values from Visual MINTEQ Version 2.5.3 database. Sample

Core 1

Core 1

Core 1

Core 2

Core 2

Core 2

ED3

MS37

MS20

Depth cm pH LaPO410H2O CePO410H2O PrPO410H2O NdPO410H2O SmPO410H2O EuPO410H2O GdPO410H2O TbPO410H2O DyPO410H2O HoPO410H2O ErPO410H2O TmPO410H2O YbPO410H2O LuPO410H2O

0–8 6.3 4.17 4.49 3.66 3.91

41–53 4.7 1.78 2.04 1.06 1.5 0.85 0.54 0.71 0.11 0.65 0.28 0.09 0.64 0.08

85–97 4.35 2.28 2.5 1.53 1.93 1.23 0.37 1.11 0.36 1.05 0.2 0.64 0.47 0.23 0.68

0–6 4.56 0.97 1.2 0.12 0.6 0.13 0.73 0.07

17–20 3.6 0.2 0.41 0.57 0.05 0.91 1.61 1.07 1.86 1.08 1.86 1.37 2.33 1.69 2.68

36–42 4.57 2.91 3.18 2.24 2.66 1.93 1.08 1.74 0.9 1.73 0.81 1.15 0.24 0.87 0.18

–

–

–

5.41 0.08 0.06 0.78 0.47 1.33 1.95 1.39 2.33 1.61 2.41 1.94 4.43 2.21 4.6

3.86 3.13 2.85 3.74 3.31 4.08 4.94 4.23 5.06 4.37 5.19 4.79 5.7 5.11 6.02

2.82 4.3 4.06 4.99 4.57 5.33 6.14 5.43 6.22 5.49 6.3 5.9 6.85 6.25 7.24

SI for REEPO4 using K values from Firsching and Brune (1991) 5.6 3.22 3.72 2.4 LaPO4 CePO4 PrPO4 5.1 2.5 2.98 1.56 5.35 2.93 3.36 2.03 NdPO4 2.33 2.71 1.34 SmPO4 0.8 1.71 0.6 EuPO4 1.79 2.19 1.01 GdPO4 0.64 1.11 TbPO4 1.48 1.88 DyPO4 1.07 1.56 HoPO4 1.66 2.21 ErPO4 1.2 1.37 TmPO4 1.98 2.29 YbPO4 0.7 LuPO4

1.64

4.35

1.35

1.69

2.86

0.89 1.4 0.57 0.26 0.02 1.1 0.24 0.49 0.21 0.49 0.38 1.29

3.69 4.1 3.41 2.42 2.82 1.66 2.57 2.17 2.72 2.08 2.93 1.2

0.66 0.95 0.13 0.62 0.32 1.58 0.78 1.06 0.38 2.6 0.17 3.23

2.29 1.87 2.6 3.6 3.15 4.31 3.53 3.83 3.23 3.86 3.05 4.64

3.54 3.14 3.86 4.81 4.36 5.47 4.66 4.95 4.34 5.02 4.2 5.87

SI for REEPO4 using K values from Visual MINTEQ 5.2 2.82 3.32 LaPO4 6.17 3.73 4.19 CePO4 5.44 2.84 3.32 PrPO4 5.6 3.18 3.61 NdPO4 2.53 2.91 SmPO4 1.01 1.92 EuPO4 2 2.4 GdPO4 0.96 1.43 TbPO4 1.53 1.93 DyPO4 0.6 1.09 HoPO4 0.98 1.53 ErPO4 0.15 0.32 TmPO4 0.71 1.02 YbPO4 0.11 LuPO4

1.24 2.11 1.23 1.65 0.77 0.05 0.23 0.78 0.19 0.96 0.47 1.54 0.89 1.88

3.95 4.87 4.03 4.35 3.61 2.63 3.03 1.98 2.62 1.7 2.04 1.03 1.66 0.61

0.95 1.74 1 1.2 0.33 0.41 0.11 1.26 0.73 1.53 1.06 3.65 1.44 3.82

2.09 1.16 1.95 1.62 2.4 3.39 2.94 3.99 3.48 4.3 3.91 4.91 4.32 5.23

3.26 2.38 3.2 2.89 3.66 4.6 4.15 5.15 4.61 5.42 5.02 6.07 5.47 6.46

clay show small negative anomalies for Ce and small positive anomalies for Eu, one shows a small positive Ce anomaly. Both of these elements have available oxidation states other than +3 (Eu2+ and Ce4+) that could fractionate them from the other REE under appropriate redox conditions, but the slight fractionation observed for Eu probably reflects an anomaly in the source rock. Ce is readily oxidized under Earth surface conditions, and commonly undergoes changes in redox state similarly to Fe or Mn (cf. Moffet, 1994), so the small Ce anomalies could be the result of recent changes in redox chemistry.

2 2.89 1.9 2.28 1.54 0.81 1.22

The concentration of La in the jarosite sections was more variable but in general higher than for the clays, ranging from around 20 up to 70 ppm. (La/LaPAAS = 0.5–1.8). However, the REE patterns for four of the jarosite sections were very similar, enriched in the LREE and progressively less enriched in HREE (Fig. 11). The enrichment factor (La/Lu)PAAS was 15–50. In contrast, a scan across the other jarosite-rich mottle (Fig. 11b) had a different pattern: this segment was enriched in the LREE and also slightly enriched in the HREE, with a rather flat MREE minimum. It is likely that this section was comprised of jarosite and FeOOH phases.

56

S.A. Welch et al. / Geochimica et Cosmochimica Acta 73 (2009) 44–64

a

10

1

REE/PAAS

clay 36-56 clay 133-140 clay 151-157 clay 150-194 jar 1 jar 2

0.1

0.01 La

b

Ce

Pr

Nd

Sm

Eu

Gd

Dy

Er

Tm

Yb

Lu

10

REE/PAAS

1 jar 42-44 jar 49-62 jar 82-85 clay 0.1

0.01 La

Ce

Pr

Nd

Sm

Eu

Gd

Dy

Er

Tm

Yb

Lu

Fig. 11. (a and b) REE normalized to PAAS for laser ablation data for sections of jarosite mottles and clay matrix for two different areas within a 2  4 cm block. The numbers in the legend refer to collection time intervals in seconds.

5. DISCUSSION The composition and concentrations of REE in natural waters are a complex function of: (i) solution composition which affects solubility, speciation, (ii) the composition and concentration of REE in the source rock/sediment and the reactivity of the REE-bearing mineral phases within the solid phase, (iii) geochemical and biogeochemical processes such as mixing, sorption–desorption, precipitation, ion exchange, uptake by biota or changes in redox state. 5.1. Effect of acidity on REE sorption and dissolved concentrations The major control on REE concentration in water at this site is pH. Both surface water and porewater concentrations show a strong positive correlation between REE content and acidity. Increasing acidity increases the solubility and dissolution rate of REE-bearing minerals, and also des-

orbs REE from surfaces (e.g., Bau, 1999; Verplanck et al., 1999, 2004; Coppin et al., 2002; Gammons et al., 2003, 2005a,b; Quinn et al., 2004; Wood et al., 2006). Several other studies have shown similar REE-pH relationship in ˚ stro¨m and coworkers natural waters. For example, A showed strong correlation between acidity and REE concentration for streams discharging from an acid sulfate soils ˚ stro¨m, 2001; A ˚ stro¨m and Corin, 2003). site in Finland (A Similarly, Gammons et al. (2005b) showed that REE concentrations decreased with increasing pH in streams discharging from an acid lake. Olias et al. (2005) and Verplanck et al. (2004) found similar results for a groundwater and river water impacted by acid mine drainage. All these studies showed that the predominant mechanism that controlled REE concentration and the fractionation was the pH-dependent desorption of REE from Al-Fe oxyhydroxide mineral phases. Several experimental studies have shown that the REE are strongly adsorbed onto solid phases (Bau, 1999; Verplanck et al., 1999, 2004; Coppin et al., 2002; Pourret et al., 2007a,b). The surface water at our field site contains flocs of iron oxyhydroxide (predominantly ferrihydrite and

REE in acid sulfate soil

schwertmanite: Sullivan and Bush, 2004) and iron sulfide, as well as abundant organic matter both as live grasses and detrital organic carbon, all of which can act as reservoirs for sorbed REE. However, porewaters are in intimate contact with clays, iron oxyhydroxides, jarosite, buried organic matter and roots, and have an even greater solid surface:solution volume ratio than the surface water. Clearly, the pH dependence of sorption onto mineral and organic surfaces must be an important control on REE in solution at this site. In experimental studies of sorption of the REE onto iron oxyhydroxides, the REE behave conservatively in solution and do not adsorb significantly at pH <5 (Bau, 1999; Verplanck et al., 1999, 2004). Above pH 5, the REE start to become significantly partitioned onto Fe oxyhydroxide surfaces. The HREE are preferentially sorbed onto Fe oxyhydroxides, and they start to sorb significantly at lower pH than the LREE. In contrast to this, the adsorption edge of the REE onto humic compounds occurs at a lower pH than for iron oxyhydroxides, between pH 2 and 4 (Pourret et al., 2007a,b). The fractionation onto humates differs from that onto iron minerals: the sorbed fraction shows slight enrichment in the MREE compared to the LREE and HREE. Although iron flocs are abundant in the surface water, the concentrations and normalized patterns for the REE in the surface water are not entirely consistent with sorption onto iron oxyhydroxides. The observed relationship between pH and REE concentration is relatively linear from pH 3 to 7, whereas if sorption onto iron oxyhydroxides was a major control, we would expect dissolved REE concentrations to remain relatively constant below pH 5, but to decrease sharply at pH >5. Furthermore, although the linear decrease in the relative abundance of dissolved REE from Gd to Lu is consistent with sorption onto iron oxyhydroxides, the depletion of the LREE is not. The sorption of the REE onto humates does not fit the observations well either, although suspended/colloidal humate complexes could partially explain the observed MREE enrichment if the humate particle sizes were small enough to pass through ˚ stro¨m and Corin, 2003; Pourret et al., 2007b and the filter (A references therein). Although the REE concentrations and patterns are similar in the surface water and the porewater, the principal geochemical controls may be very different. Since the sediment contains >90% clay, sorption reactions in the porewater are likely to be largely controlled by reactions with clay surfaces. The results of sorption studies on clays show that the behavior of REE is more complex than in FeOOH sorption studies, and that sorption behavior depends on acidity, salinity and clay composition (cf. Sinitsyn et al., 2000; Coppin et al., 2002). All experiments by Coppin et al. (2002) with kaolinite and smectite showed that REE sorption increased with increasing pH. At high ionic strength, there was a clear sorption edge for both kaolinite and smectite that occurred between pH 6 and 8. The results the experiments of Coppin et al. (2002) indicate that the REE concentration in solution should be high and essentially independent of pH in acidic solutions (pH <5), a prediction which again is not in accord with observed behavior in our porewater samples. However, at lower ionic strength,

57

similar to the TDS concentration measured in the porewater, Coppin et al. (2002) found an almost linear variation of sorption with acidity for kaolinite, and much greater sorption of the REE overall, especially under acidic conditions. Note that this behavior is consistent with that observed for the porewater at our site, suggesting that the primary reservoir of sorbed REE is clay, in agreement with the observed soil mineralogy. However, the LREE/MREE/ HREE enrichment pattern must also be considered. The REE concentrations of selected porewater samples were normalized to those of a typical clay composition from this site, REEpore/REEclay (Figs. 11a and 12a). Although there is some variability in the patterns due to the MREE enrichment in the porewater, overall there is a decrease in the REEpore/REEclay across the series from light to heavy REE. The slope of the fractionation curves become progressively steeper for elements heavier than Gd, indicating that sorption onto clays is particularly important for the heavier elements, consistent with their small ionic radius and stronger bonding. 5.2. Speciation effect on total dissolved REE The speciation of REE elements in natural waters have been found to be a major control on both the concentration and the fractionation of the REE (e.g., Brookins, 1989; Johannesson et al., 1996). Although sulfate complexes comprise only a small fraction of the total REE in most natural waters (e.g., Johannesson and Lyons, 1994; Byrne and Li, 1995; Byrne et al 1996; Johannesson et al., 1996), studies have shown that in acidic high sulfate waters, such as those impacted by oxidation of pyrite, the sulfate complexes are important (Gammons et al., 2003, 2005a; Vercouter et al., 2005; Wood et al., 2006). The inorganic speciation of the dissolved REE is often dominated by the REESO4 þ complex in acidic sulfate-rich waters. The change in the stability of this monosulfato complex across the series is similar to that observed for the PAAS-normalized REE in the surface and porewaters at our field site. The stability constants for the complexes increase slightly from 3.61 ± 0.02 (La) to 3.64 ± 0.02 (Eu), and then decrease linearly to 3.44 ± 0.02 for Lu (Schijf and Byrne, 2004). Hence, complexation by sulfate has been suggested as a possible, albeit, minor cause for the MREE enrichment that is commonly observed in acid sulfate waters (Johannesson et al., 1996). However, the speciation calculations for both the surface and pore waters do not show a significant MREE enrichment that would be attributable to the formation of sulfate complexes (Figs. 9 and 10). The sum of all REESO4 species concentrations is nearly constant for the LREE, and then decreases systematically from Sm to Lu. One source of ambiguity in the speciation calculations and the role of sulfate on both the concentrations and fractionation of the REE in acid water is the stability of the REEðSO4 Þ2  complex. The values given by Wood (1990), as well as those in the Visual MINTEQ (Allison et al., 1991) and LLNL (EQ3/6) (Parkhurst and Appelo, 1999) data bases are all similar but do not vary systematically across the series, and all have anomalously high values (0.2 log units) for the stability of the EuðSO4 Þ2  species.

58

S.A. Welch et al. / Geochimica et Cosmochimica Acta 73 (2009) 44–64 10 -1

REE pore/REE clay

a c3 16-56 c3 56-77 c2 17-20

10 -2

c2 35-42 c1 53-85 c1 85-120

10 -3 La

Ce

Pr

Nd

Sm

Eu

Gd

Dy

Er

Tm

Yb

10

Lu

b

REE pore/REE jar

1 c3 16-56 c3 56-77 c2 17-20

10 -1

c2 35-42 c1 53-85 c1 85-120

10 -2

10 -3 La

Ce

Pr

Nd

Sm

Eu

Gd

Dy

Er

Tm

Yb

Lu

1000

REE pore/REE jardis

c c3 16-56 c3 56-77 c2 17-20 100

c2 35-42 c1 53-85 c1 85-120

10 La

Ce

Pr

Nd

Sm

Eu

Gd

Dy

Er

Tm

Yb

Lu

Fig. 12. REE in selected porewater samples from the three cores normalized to (a) a representative clay composition from the site, (b) a representative jarosite mottle, and (c) jarosite dissolution from Welch et al. (2007).

However, using these values, our calculations show that the REEðSO4 Þ2  complexes should comprise 5% to 40% of the total REE in solution. These are consistent with the recent measurements by Vercouter et al. (2005) where both mono- and disulfate complexes of Eu and La were measured by nanoESI-MS. The presence of negatively charged REE complexes could be very important in fractionating the REE, since these species should have very different sorption characteristics than the positively charged REE3+, REECl2+, REESO4 þ . Cleary, the high SO4 2 concentrations in these waters is important in controlling over-

all concentrations of the REE but role of SO4 2 on the REE fractionation patterns is not well constrained. 5.3. REE buffering by phosphates The formation of hydrous REE phosphate minerals such as rhabdophane has been suggested as a major control limiting the REE concentrations in natural waters (Byrne and Kim 1993; Byrne et al., 1996; Johannesson et al., 1995; Cetiner et al., 2005; Ko¨hler et al., 2005) and has been suggested as a mechanism for the MREE enrichment ob-

REE in acid sulfate soil

59

served in many natural waters (cf. Firsching and Brune, 1991). Solution saturation indices for REEPO4 phases (Firsching and Brune, 1991; Allison et al., 1991) and for LREEPO410H2O phases for selected water samples were calculated (Table 3). It should be noted that the decahydrates are unknown as minerals, and are presumably metastable under natural surficial conditions with respect to less soluble minerals such as the anhydrous phosphates (monazite and xenotime groups) and dihydrates (rhabdophane). Many of the porewater extracts from this site were greatly supersaturated with respect to the LREEPO410H2O phases, which may account for the LREE depletion in the porewaters (Table 3). However, with the exception of the most acidic sample from core 2, all were greatly supersaturated (SI 0.5 to 5) with respect to REEPO4 indicating the REE in porewaters at this site are not in equilibrium with these phases. The surface waters have similar REE patterns and concentrations but much lower phosphate, and with the exception of the near-neutral pH sample taken from the East Drain, were undersaturated with respect to LREEPO410H2O and REEPO4 phases. However, the phosphate minerals can only buffer dissolved REE concentrations if it is possible to nucleate them. Note that no unambiguous secondary REE minerals have been observed.

2004; Pourret et al., 2007a,b; Pe´drot et al., 2008) as well as MREE enrichments in the organic-bound REE fraction at pH conditions similar to those measured at our site (Johannesson et al., 2004; Pourret et al., 2007a,b) indicating that organic complexation could be an important factor controlling REE concentrations and fractionation in the surface waters at our site as well. In the pore water samples, however, it is unclear if there was a substantial fraction of colloidal material in the filtrate. The sediments are comprised largely of micron-sized clay (flat to HREE enrichment pattern), jarosite (LREE-enriched) and iron oxyhydroxides. The REE concentration and REE patterns were not determined for the iron oxyhydroxides, however, these should preferentially sorb the HREE under acidic conditions. If a significant amount of sediment passed through the filter, then the REE signature of the pore water should reflect the solid phase, and display a REE normalized pattern that is flat due to clays, LREE enriched from jarosite, or HREE enriched from iron oxyhydroxides. Instead, the REE in the pore water extracts show MREE enrichment, inconsistent with solid phase, indicating that there is no significant particulate fraction in the pore water extracts.

5.4. Role of colloidal phases

The original REE composition of the sediments before the oxidation and acidification that occurred in the 1970s and 1980s is unknown. Considering that these are detrital clays deposited in a coastal shallow marine environment, it is likely that the REE concentrations and pattern would have been very similar to an average Earth surface reservoir such as PAAS when they were deposited. However, chemical weathering and diagenesis can result in large fractionation in the REE (cf. McLennan, 1989, and references therein). The REE pattern in the clay matrix of the oxidized zone currently shows a systematic depletion of the LREE compared to the HREE when normalized to PAAS. If the assumption that the REE pattern was flat is correct, then the depletion is extreme: half of the LREE would have to have been leached from the clays over the last 40 years. Even though the REE are typically considered to be immobile, this extensive leaching is not unprecedented. For ˚ stro¨m (2001) showed that in a 2 m thick acid sulexample, A fate soil profile in Finland, La was depleted by ca. 40% in the upper part of the acid zone and enriched by about 40% in the lower part of the acid profile, compared to the unoxidized sediments. Given a typical LREE concentration of 500 lg/L in porewaters at our site, the LREE depletion in the clays would be equivalent to only about 1 ml of acidic water leaching each gram of sediment per year, which is eminently feasible. However, the LREE have not been leached to that extent from the profile as a whole, because the LREE depletion in the clay matrix is compensated by LREE enrichment in the jarosite mottles. Visual inspection of the cores shows that jarosite can comprise up to 10–20% of the sediment in the jarosite-rich mottled zone. The concentrations of the LREE in the jarosite are greater than or

Several studies have shown that a significant fraction of the total REE in natural waters (10s to 100%) is associated with a suspended or colloidal phase, and this can result in fractionation between the truly ‘dissolved’ and ‘solid’ phase (cf. Elderfield et al., 1990; Dia et al., 2000; Ingri ˚ stro¨m, 2001; A ˚ stro¨m and Corin, 2003; Gamet al., 2000; A mons et al., 2005a,b; Pourret et al., 2007b; Pe´drot et al., 2008). Although the water samples collected in this study were filtered with 0.45 lm filters, this procedure is inadequate to separate the REE associated with colloids, and it is probable that some fraction of the REE measured in our samples are associated with a nanoparticulate phase. However, although the concentrations and the REE normalized patterns are similar between the surface water and porewater, the contribution of the colloidal fraction is probably different. The REE in the surface waters at our site were probably affected by both iron-oxyhydroxide phases and organic matter. Experimental studies have shown that REE in water are readily scavenged by FeOOH phases (Bau 1999; Verplanck et al., 2004; Gammons et al., 2003, 2005a,b). However, under acidic conditions such as those at our field site, the HREE should be preferentially sorbed relative to the LREE in the absence of strong complexing ligands such as carbonate ions or organic ligands, which is inconsistent with the REE pattern observed for our water samples. Dissolved organic carbon (DOC) was not analyzed for our water samples, although based on the dense vegetation it is expected that the surface waters would have had high concentrations of DOC. Several studies have shown good correlations between the REE and DOC in natural waters (Dia et al., 2000; Ingri et al., 2000; Johannesson et al.,

5.5. REE in sediments

60

S.A. Welch et al. / Geochimica et Cosmochimica Acta 73 (2009) 44–64

equal to the LREE in the clays, indicating that 10–20% of the LREE could have been mobilized from the clays and sequestered in the jarosite as a result of acid weathering. The mobilization of REE from the clays could be the major control of REE content in the secondary jarosite phases. Due to their large ionic radius, the LREE tend to be incompatible in silicate mineral structures compared to the HREE (Taylor and McLennan, 1985). Studies of REE mobility in weathering profiles show that the LREE are preferentially mobilized (e.g., Nesbitt, 1979; Banfield and Eggleton, 1989). The LREE are also more weakly sorbed to mineral surfaces. Coppin et al. (2002) found that HREE sorbed more strongly on illite and smectite surfaces, and the pH of the sorption edge decreased with increasing atomic number. Similarly, experiments and observations of natural environmental processes show that in acid conditions, the HREE are preferentially sorbed to Fe oxyhydroxide phases (Bau 1999; Verplanck et al., 2004; Gammons et al., 2003, 2005a,b). The observed LREE-enriched pattern in jarosite may in part reflect the preferential solubilization of the LREE associated with the acid weathering of the clay sediments fractions. However, there are crystal chemical reasons for believing that this effect is amplified by the strong preference for LREE exhibited by the jarosite crystal structure. LREE enrichment in the jarosite phase has been found in all the jarosite samples analyzed from this site (Welch et al., 2007) and has also been noted in other naturally occurring jarosite minerals (Dutrizac and Jambor, 2000 and references therein; Dutrizac, 2004). The jarosite-group minerals are isostructural with several hydrous AluminumPhosphate-Sulfate minerals (APS), including florencite (REE Al3(PO4)2(OH)6 with dominant REE = La, Ce or even Nd: Milton and Bastron, 1971) that forms commonly in weathering environments (cf. Banfield and Eggleton, 1989; Taunton et al., 2000a,b). The preference of florencite for LREE is consistent with these larger ions fitting better in the large A (K+) site of the jarosite structure than do the smaller HREE. So, precipitation of jarosite under acidic conditions could preferentially incorporate the available LREE into the structure, leaving the smaller HREE to be sorbed back on the clay surfaces with repeated wetting and drying of the site. The sulfate and ferric iron that are essential for jarosite crystallization would have formed from the rapid oxidation of pyrite. Another factor contributing to the LREE enrichment in the jarosite mottles could be biologically mediated uptake of the REE. We note that the jarosite mottles at this site form preferentially around root channels. These microenvironments would have been a source of organic matter supporting the sulfate- and iron-reducing microbial populations that formed pyrite. Once the field site was drained and dried, the root channels would have been zones of oxygen diffusion into the sediments, and would have become sites of rapid pyrite oxidation. Several studies have shown that plants preferentially take up the LREE (Tyler, 2004; Merten et al., 2005; Stille et al., 2006) so the LREE enrichment in the jarosite mottles might be the result of degradation of plant material.

The REE composition of the major mineral phases within the sediments clearly does not match the REE composition of the acidic waters. Stoichiometric leaching of the clays and jarosite would produce LREE-enriched water from jarosite and HREE-enriched from clays, contrary to what is observed. This suggests that the pore water composition reflects equilibration with, rather than complete dissolution of, the sediments. Leaching of sediments, even under extremely acidic conditions, often results in the non-stoichiometric solubilization of the REE (Elderfield et al., 1990; Johannesson and Zhou, 1999; Bozau et al., 2004; Merten et al., 2005). At our site, jarosite is enriched in the LREE. However, experiments showed that jarosite dissolution is not stoichiometric at pH similar to that of the porewater and surface water of this study (Welch et al., 2007). For low degrees of dissolution (<5%), LREE-enriched jarosite incongruently dissolved to produce a solution that was enriched in the MREE, very similar to porewaters in the jarosite-rich zone. The REE concentrations in selected porewaters from the three cores were normalized to a typical jarosite composition, and also to a representative composition from the jarosite dissolution experiments (Welch et al., 2007) (Fig. 12b and c). The differing fractionations of the REE relative to La can be quantified using the equilibrium constant for REE ion exchange reactions between the solid and aqueous REE: REE-jarosite þ LaðaqÞ ! La-jarosite þ REEðaqÞ k ¼ ðREEaq =REEjar Þ=ðLaaq =Lajar Þ The K values normalized to La are listed in Table 4. Although the REE concentrations were variable, the K values for the exchange reaction were usually similar from sample to sample (errors were large for some very lowabundance HREE), indicating that the REE in jarosite was in equilibrium with the porewaters. K increased from La to a maximum value at Er or Tm in all cases, implying that REE consistently partition into the porewater over jarosite with increasing atomic number up to these elements. Decrease in K for the remaining HREE may indicate that these can substitute on the Fe3+ site in jarosite, but more likely implies that they sorb strongly onto Fe oxyhydroxides that are produced by incongruent dissolution of jarosite (Bau, 1999; Verplanck et al., 1999, 2004). The REE in the porewaters, when normalized to the REE released to solution in the short-term (two week or 1% jarosite dissolved) experiments of Welch et al. (2007), follow two distinct trends (Fig. 12c). The patterns for the two porewater extracts in core 3 are flat, indicating that the REE patterns in the porewaters are consistent with short-term dissolution of jarosite. Although many of the MREE and HREE elements were not measured in the jarosite dissolution experiments, the data that are available suggest that the patterns stay flat across the series. However, the trends observed for the samples in cores 1 and 2 show a systematic decrease across the series from La to Eu. The difference between the core 1 and 2 trend and that of core 3 is probably due to differing extent of inundation. Core 3 is the farthest from the drain and highest in the landscape (as evident from the depth of standing water) and would

REE in acid sulfate soil

61

Table 4 K = (REEaq/REEjar)/(Laaq/Lajar) values for determined for the exchange reaction:REE-jarosite + La(aq) ? La-jarosite + REE(aq) for two depth intervals for each of cores 1, 2, and 3.

La Ce Pr Nd Sm Eu Gd Dy Er Tm Yb Lu

Core 3 16–56c

Core 3 56–77 b

Core 2 17–20b

Core 2 35–42

Core 1 53–85b

Core 1 85–120

1.0 2.9 6.0 8.6 11.8 13.7 21.0 24.5 44.1 33.2 26.7 21.3

1.0 2.8 5.3 7.9 11.2 15.7 21.3 25.0 41.3 32.4 26.4 26.2

1.0 2.1 4.2 5.1 6.4 10.8 9.6 19.1 25.5 27.5 13.3 27.0

1.0 2.3 4.0 5.6 7.4 10.3 13.7 19.9 19.4 20.3 12.6 12.2

1.0 2.2 4.6 6.3 6.0 13.5 13.6 13.6 36.7 29.9 8.0 16.1

1.0 2.0 3.2 4.3 6.2 8.5 13.2 16.5 24.5 16.3 12.0 15.8

have been the driest, whereas the two closer cores would have been wetter, both more frequently and for longer periods of time, and therefore would likely have had experienced more extensive jarosite weathering. Release of REE from jarosite dissolution is clearly non-stoichiometric, showing a MREE enrichment in solution from a LREE-enriched mineral phase for a low degree of partial dissolution, <1% reacted (Welch et al., 2007). However, if a large enough proportion of the original jarosite is dissolved, the LREE should be released into solution and flatten out the MREE enrichment trend. Thus, the enrichment in the LREE in core 1 and 2 porewaters relative to core 3 and the jarosite dissolution experiments in the laboratory probably reflects more extensive dissolution of the mineral phase. The MREE/LREE enrichment that results from jarosite dissolution is predicted by simple quantitative models. Suppose that jarosite decomposes by the reaction: KFe3 ðSO4 Þ2 ðOHÞ6 ! 3FeOðOHÞ þ Kþ þ 2SO4 2 þ 3Hþ For very small molar ratios of water:jarosite, as for the case of porewater in jarosite mottle, the concentrations of any given REE will be such that cj = Djcw, where cj = concentration in jarosite, cw = concentration in water, and Dj = jarosite/water distribution coefficient. The highest porewater REE concentrations, for sample ms2c2 (35–42), in PAAS-normalized units, are La = 0.0363, Nd = 0.0376, Gd = 0.0674, Lu = 0.0123. Corresponding DjLa/DjREE are 1, 5.6, 13.7, 12.2 (Table 4). For La, PAAS-normalized cj = 1.84 (cf. Section 4.6). From these data, we can reconstruct Dj = 50.7, 9.1, 3.7, 4.2 for La, Nd, Gd, Lu respectively, and corresponding normalized cj = 1.84, 0.34, 0.25, 0.05. Clearly, normalized concentrations in jarosite show a monotonic trend of depletion with increasing atomic number. The distribution coefficients also decline with increasing atomic number, but this trend may flatten out or reverse slightly at the HREE end of the series. Table 4 suggests that the compatibility trends of HREE may vary slightly from site to site. The difference in cj and Dj across the REE results in MREE enrichment for small ratios of water:jarosite (as appropriate for porewater and low degrees (<<10%) of partial dissolution. Nevertheless, increasing dissolution of the jarosite ultimately results in enrichment of La and the other LREE.

5.6. Mineralogical control of REE maximum Although the REE concentrations correlate with pH of the porewater in general, the maximum concentration of REE and Y are consistently found lower in the profile, below the zone where the minimum pH occurs. This zone is coincident with a change in the modally dominant mineral from jarosite to goethite. The REE patterns at maximum concentration are distinctly different from those at shallower depth: they show MREE enrichment, but the pattern is asymmetric with LREE-enriched relative to HREE (Table 1). There are several possibilities why the REE maximum occurs below the pH minimum, and why there is a change in the REE patterns: (i) There may be an enrichment in REE-bearing mineral phases at this depth. The maximum-REE layer occurs above the fossil shell layer, suggesting that the depositional environment changed. For example, the elevated REE in the porewater in this zone may reflect acid leaching of a heavy mineral sand containing minerals such as monazite. (ii) Another possibility is that the REE in the porewater could be from the dissolution of fossil shells. The shells in the layer below the acid zone are extensively etched, and there are gypsum filled casts a few centimeters above the current shell layer indicating that some of the shells have been completely dissolved by acid. The REE are readily complexed by carbonate ions (Cantrell and Byrne, 1987), the stability constants increasing with atomic number, and are readily incorporated into carbonate minerals. Incorporation into biogenic molluscan shell aragonite is in excess of that expected for equilibrium between inorganic aragonite and water, with La = 4–25 ppm and the shale-normalised that may show either moderate HREE or LREE enrichment [(La/Yb)norm = 0.28–16; (Whittaker and Kyser, 1993 and references cited therein)]. Data from rarer calcite shells such as Cretaceous inoceramids shows overlapping but lower REE content and a similarly variable range of enrichment trends (La = 1–11 pm, (La/Yb)norm = 0.35–2.5; Jime´nezBerracoso et al., 2006). Other experimental studies of natural carbonate samples have shown that concentrations can range over orders of magnitude, with enrichment of LREE, MREE or HREE, or no significant fractionation, under ‘marine conditions’ (Zhong and Mucci, 1995; Le´cuyer et al., 1998; Webb and Kamber, 2000).

62

S.A. Welch et al. / Geochimica et Cosmochimica Acta 73 (2009) 44–64

(iii) The elevated REE could be the result of leaching of ˚ sthe REE from the sediments higher up in the profile. A tro¨m (2001) showed a significant depletion in La concentration in sediments in the top of an oxidized acid sulfate soil profile combined with enrichment in La at the bottom of the acid zone above the redox boundary, very similar to the profile we observed in the porewater from the three cores at our site. The intermittent wetting and drying of the site over decades could have leached the REE from the sediments and mobilized them downward, accounting for the elevated concentrations and also for the relative enrichment in the LREE, since the LREE are not as readily sorbed to surfaces as the HREE. Note that this mechanism implies downward flux of REE, in contrast to (ii) which implies upward flux of REE from the shell bed. (iv) The most plausible mechanism for producing high REE concentrations and the observed enrichment patterns is simply the incongruent dissolution of jarosite to form goethite. Jarosite incorporates and concentrates the REE into its structure (Dutrizac 2004; Welch et al., 2007). The MREE-enriched pattern in the porewater is similar to the REE pattern in the early stages of jarosite dissolution (Welch et al., 2007). The calculations above demonstrate that the measured MREE-enriched porewater concentrations across the REE are consistent with LREE being both most abundant and most compatible in jarosite, but that we would expect flattening of the LREE depletion as the extent of partial dissolution increases. The spatial variation of REE in porewater is what we would expect if partial dissolution of jarosite were the main control on porewater REE, as per scenario (iv) above. The location of the maximum of dissolution is certainly controlled by the presence of the shell layer, in that the increase in pH near the shells destabilizes jarosite relative to goethite, but it does not seem necessary to invoke the shells as a major reservoir of REE as in scenario (ii). We note that there is no evidence for the REE minerals of scenario (i), but that downward leaching of REE as per scenario (iii) may be a contributor to the total REE content of the jarosite at deeper levels. Nevertheless, jarosite breakdown remains the simplest comprehensive explanation of the observed REE distribution. 6. SUMMARY AND CONCLUSIONS The concentrations of REE measured in the porewaters at this acid sulfate soil site are up to ca. 3 ppm (3000 ppb) and are among the highest values ever reported under surficial conditions (Gammons et al., 2003, 2005a; Merten et al 2005). The concentrations are strongly dependent on pH of the porewater and surface water, but the pH dependence is different between surface water and porewater. The REE patterns normalized to PAAS for water samples show a distinctive MREE enrichment, very similar to the patterns produced in the early stages of jarosite dissolution experiments. This suggests that the REE of the porewater are in equilibrium with jarosite, which is the principal solid repository of REE. Jarosite itself is strongly LREE-enriched, and more extensive dissolution would tend to change the enrich-

ment pattern of the porewater in that direction. The maximum REE in the porewater occur at the bottom of the oxidized zone, and are coincident with the transition from jarosite- to goethite-rich mottles. Flattening of LREE depletion in this horizon is as expected for increasing partial dissolution of jarosite. Decomposition of jarosite to goethite occurs because of pH increase, due to the presence of a shell layer just below. Solid solution in jarosite and adsorption to clays and goethite all provide important reservoirs for REE in acid sulfate soils. The REE concentrations and fractionation pattern can be as local probes of the extent of jarosite dissolution. However, the very high concentrations of REE in associated porewater implies that these elements can be highly mobile under acid sulfate conditions, and the importance of lateral and vertical transport of REE still needs to be assessed. ACKNOWLEDGMENTS This work was supported by the CRC LEME. We thank Russell and Georgina Yerbury for allowing us access to their property, in spite of the massive chaos we always create. Al Cockburn and Dr. Sara Beavis collected a few of the surface water samples for us during their frog surveys. We thank Charlotte Allen at the RSES for the laser ablation analysis. We also thank Ulli Troitzsch and Linda McMorrow for their unending help in analyzing sediment and water samples. Our correspondence with Karen Johannesson and Scott Wood was extremely enlightening. We also thank Karen Johannesson for serving as editor for this manuscript, as well as Matthew Leybourne and an anonymous reviewer whose comments improved the manuscript.

REFERENCES Allison A. W., Brown D. S. and Novo-Gradac K. J. (1991) MINTEQA2/PRODEFA2, a geochemical assessment model for environmental systems: version 3.0 User’s Manual. U.S. Environmental Protections Agency Report EPA/600/3-91/021. ˚ stro¨m M. (2001) Abundance and fractionation patterns of rare A earth elements in streams affected by acid sulphate soils. Chem. Geol. 175, 249–258. ˚ stro¨m M. and Corin N. (2003) Distribution of rare earth elements A in anionic, cationic and particulate fractions in boreal humusrich streams affected by acid sulphate soils. Water Res. 37, 273– 280. Banfield J. F. and Eggleton R. A. (1989) Apatite replacement and rare-earth mobilization, fractionation, and fixation during weathering. Clay. Clay Miner. 37, 113–127. Bau M. (1999) Scavenging of dissolved yttrium and rare earths by precipitating iron oxyhydroxide: Experimental evidence for Ce oxidation, Y-Ho fractionation, and lanthanide tetrad effect. Geochim. Cosmochim. Acta 63, 67–77. Beavis S., Beavis F., Somerville P. and Welch S. (2005) Spatial and temporal variability of acidity at a coastal acid sulfate soil site II: A time for change. In Regolith 2005—Ten Years of CRC LEME (ed. I. C. Roach) CRC LEME, pp. 22–26. Bozau E., Leblanc M., Seidel J. L. and Sta¨rk H.-J. (2004) Light Rare Earth Elements enrichment in an acidic mine lake (Lusatia, Germany). Appl. Geochem. 19, 261–271. Brookins D. G. (1989) Aqueous geochemistry of rare earth elements. In Geochemistry and Mineralogy of Rare Earth Elements, Reviews in Mineralogy, vol. 21 (eds. B. R. Lipin

REE in acid sulfate soil and G. A. McKay). Mineralogical Society of America, Washington, pp. 201–225. Byrne R. H. and Kim K.-H. (1993) Rare-earth precipitation and coprecipitation behaviour—The limiting role of PO43 on dissolved rare-earth concentrations in seawater. Geochim. Cosmochim. Acta 57, 519–526. Byrne R. H. and Li B. Q. (1995) Comparative complexation behavior of the Rare-Earths. Geochim. Cosmochim. Acta 59, 4575–4589. Byrne R. H., Liu X. W. and Schijf J. (1996) The influence of phosphate coprecipitation on rare earth distributions in natural waters. Geochim. Cosmochim. Acta 60, 3341–3346. Cantrell K. J. and Byrne R. H. (1987) Rare earth element complexation by carbonate and oxalate ions. Geochim. Cosmochim. Acta 51, 591–605. Cetiner Z. S., Wood S. A. and Gammons C. H. (2005) The aqueous geochemistry of the rare earth elements Part XIV. The solubility of rare earth element phosphates from 23 to 150 degrees C. Chem. Geol. 217, 147–169. Coppin F., Berger G., Bauer A., Castet S. and Loubet M. (2002) Sorption of lanthanides on smectite and kaolinite. Chem. Geol. 182, 57–68. Dia A., Graua G., Olivie´-Lauquet G., Riou C., Mole´nat J. and Curm P. (2000) The distribution of rare earth elements in groundwaters: assessing the role of source-rock composition, redox changes and colloidal particles. Geochim. Cosmochim. Acta 64, 4131–4151. Dutrizac J. E. (2004) The behaviour of the rare earths during the precipitation of sodium, potassium and lead jarosites. Hydrometallurgy 73, 11–30. Dutrizac J. E. and Jambor J. L. (2000) Jarosites and their application in hydrometallurgy. In Sulfate Minerals—Crystallography Geochemistry and Environmental Significance, Reviews in Mineralogy and Geochemistry, vol. 40 (eds. C. N. Alpers, J. L. Jambor and D. K. Nordstrom). Mineralogical Society of America, Washington, pp. 405–452. Elderfield H., Upstill-Goddard R. and Sholkovitz E. R. (1990) The rare earth elements in rivers, estuaries, and coastal seas and their significance to the composition of ocean waters. Geochim. Cosmochim. Acta 54, 971–991. Firsching F. H. and Brune S. N. (1991) Solubility products of the trivalent rare-earth phosphates. J. Chem. Eng. Data 36, 93–95. Gammons C. H., Wood S. A., Jonas J. P. and Madison J. P. (2003) Geochemistry of the rare-earth elements and uranium in the acidic Berkeley Pit Lake, Butte. Montana. Chem. Geol. 198, 269–288. Gammons C. H., Wood S. A. and Nimick D. A. (2005a) Diel behavior of rare earth elements in a mountain stream with acidic to neutral pH. Geochim. Cosmochim. Acta 69, 3747–3758. Gammons C. H., Wood S. A., Pedrozo F., Varekamp J. C., Nelson B. J., Shope C. L. and Baffico G. (2005b) Hydrogeochemistry and rare earth element behavior in a volcanically acidified watershed in Patagonia. Argentina Chem. Geol. 222, 249–267. Haley B. A., Klinkhammer G. P. and McManus J. (2004) Rare earth elements in pore waters of marine sediments. Geochim. Cosmochim. Acta 68, 1265–1279. Hannigan R. E. and Sholkovitz E. R. (2001) The development of middle rare earth element enrichments in freshwaters: weathering of phosphate minerals. Chem. Geol. 175, 495–508. ¨ rjan Gustafsson O ¨ ., Andersson Ingri J., Widerlund A., Land M., O ¨ hlander B. (2000) Temporal variations in the P. and O fractionation of the rare earth elements in a boreal river; the role of colloidal particles. Chem. Geol. 166, 23–45. Isaacson, L. (2006) The distribution of acidity, salts and metals in a highly modified acid sulfate soil backswamp, Mays Swamp, Kempsey, NSW. Honours Thesis, Australian National University.

63

Isaacson L., Kirste D., Beavis S. and Welch S. (2006) Controls of acid, salt and metal distribution at a coastal acid sulfate soils site. In Regolith 2006: Consolidation and Dispersion of Ideas. (eds. R. W. Fitzpatrick and P. Shand) Proceedings of the CRC LEME Regolith Symposium, November 2006, pp. 166–170. Jime´nez-Berracoso A., Oliveiro E. and Elorza J. (2006) New petrographic and geochemical insights on diagenesis and paleoenvironmental stress in Late Cretaceous inoceramid shells from the James Ross Basin, Antarctica. Antarctic Sci. 18, 357–376. Johannesson K. H., Hawkins, Jr., D. L. and Alejandra Corte´s A. (2006) Do Archean chemical sediments record ancient seawater rare earth element patterns? Geochim. Cosmochim. Acta 70, 871–890. Johannesson K. H. and Lyons W. B. (1994) The rare-earth element geochemistry of mono-lake water and the importance of carbonate complexing. Limnol. Oceanogr. 39, 1141–1154. Johannesson K. H., Lyons W. B., Stetzenbach K. J. and Byrne R. H. (1995) The solubility control of rare earth elements in natural terrestrial waters and the significance of PO43 and CO32 in limiting dissolved rare earth concentrations: a review of recent information. Aquatic Geochem. 1, 157–173. Johannesson K. H., Tang J., Daniels J. M., Bounds W. J., David J. and Burdige D. J. (2004) Rare earth element concentrations and speciation in organic-rich blackwaters of the Great Dismal Swamp, Virginia, USA. Chem. Geol. 209, 271–294. Johannesson K. H., Lyons W. B., Yelken M. A., Gaudette H. E. and Stetzenbach K. J. (1996) Geochemistry of the rare-earth elements in hypersaline and dilute acidic natural terrestrial waters: complexation behavior and middle rare-earth element enrichments. Chem. Geol. 133, 125–144. Johannesson K. H. and Zhou X. (1999) Origin of middle rare earth element enrichments in acid waters of a Canadian High Arctic lake. Geochim. Cosmochim. Acta 63, 153–165. Klungness G. D. and Byrne R. H. (2000) Comparative hydrolysis behavior of the rare earths and yttrium: the influence of temperature and ionic strength. Polyhedron 19, 99–107. Ko¨hler S. J., Harouiya N., Chaı¨rat C. and Oelkers E. H. (2005) Experimental studies of REE fractionation during water– mineral interactions: REE release rates during apatite dissolution from pH 2.8 to 9.2.. Chem. Geol 222, 168–182. Le´cuyer C., Grandjean P., Barrat J.-A., Nolvak J., Emig C. and Paris F. (1998) D18O and REE contents of phosphatic brachiopods: a comparison between modern and lower Paleozoic populations. Geochim. Cosmochim. Acta 62, 2429–2436. Lewis A. J., Palmer M. R., Sturchio N. C. and Kemp A. J. (1997) The rare earth element geochemistry of acid-sulphate and acidsulphate-chloride geothermal systems from Yellowstone National Park, Wyoming, USA. Geochim. Cosmochim. Acta 61, 695–706. Leybourne M. I., Goodfellow W. D., Boyle D. R. and Hall G. M. (2000) Rapid development of negative Ce anomalies in surface waters and contrasting REE patterns in groundwaters associated with Zn–Pb massive sulphide deposits. Appl. Geochem. 15, 695–723. Leybourne M. I., Peter J. M., Layton-Matthews D., Volesky J. and Boyle D. R. (2006) Mobility and fractionation of rare earth elements during supergene weathering and gossan formation and chemical modification of massive sulfide gossan. Geochim. Cosmochim. Acta 70, 1097–1112. Merten D., Geletneky J., Bergmann H., Haferburg G., Kothe E. and Bu¨chel G. (2005) Rare earth element patterns: a tool for understanding processes in remediation of acid mine drainage. Chemie der Erde 65(S1), 97–114. McLennan S. M. (1989) Rare-earth elements in sedimentaryrocks—Influence of provenance and sedimentary processes. In Geochemistry and Mineralogy of Rare Earth Elements Reviews

64

S.A. Welch et al. / Geochimica et Cosmochimica Acta 73 (2009) 44–64

(eds. B. R. Lipin and G. A. McKay). Mineralogical Society of America, Washington, pp. 169–200. Milton D. J. and Bastron H. (1971) Churchite and florencite (Nd) from Sausalito. California Miner. Rec. 2, 166–168. Moffett J. W. (1994) The relationship between cerium and manganese oxidization in the marine environment. Limnol. Oceanogr. 39, 1309–1318. Nance W. B. and Taylor S. R. (1976) Rare earth patterns and crustal evolution, I. Australian post-Archean sedimentary rocks. Geochim. Cosmochim. Acta 40, 1539–1551. Nesbitt H. W. (1979) Mobility and fractionation of rare-earth elementsduringweatheringofagranodiorite. Nature 279, 206–210. Olias M., Ceron J. C., Fernandez I. and De la Rosa J. (2005) Distribution of rare earth elements in an alluvial aquifer affected by acid mine drainage: the Guadiamar aquifer (SW Spain). Environ. Pollut. 135, 53–64. Parkhurst, D. L., Appelo, C. A. J. (1999) User’s Guide To Phreeqc (Version 2)—A Computer Program For Speciation, BatchReaction, One-Dimensional Transport, and Inverse Geochemical Calculations. Water-Resources Investigations Report 99– 4259 312 pp. Pe´drot M., Dia A., Davranche M., Bouhnik-Le Coz M., Henin O. and Gruau G. (2008) Insights into colloid-mediated trace element release at the soil/water interface. J. Colloid Interface Sci. 325, 187–197. Pourret O., Davranche M., Gruau G. and Dia A. (2007a) Rare earth elements complexation with humic acid. Chem. Geol. 243, 128–141. Pourret O., Davranche M., Gruau G. and Dia A. (2007b) Organic complexation of rare earth elements in natural waters: evaluating model calculations from ultrafiltration data. Geochim. Cosmochim. Acta 71, 2718–2735. Protano G. and Riccobono F. (2002) High contents of rare earth elements (REEs) in stream waters of a Cu–Pb–Zn mining area. Environ. Pollut. 117, 499–514. Quinn K. A., Byrne R. H. and Schijf J. (2004) Comparative scavenging of yttrium and the rare earth elements in seawater: competitive influences of solution and surface chemistry. Aquatic Geochem. 10, 59–80. Schijf J. and Byrne R. H. (2004) Determination of SO4b1 for yttrium and the rare earth elements at I = 0.66 m and t = 25 °C—Implications for YREE solution speciation in sulfate-rich waters. Geochim. Cosmochim. Acta 68, 2825–2837. Serrano M. J. G., Sanz L. F. A. and Nordstrom D. K. (2000) REE speciation in low-temperature acidic waters and the competitive effects of aluminium. Chem. Geol. 165, 167–180. Sinitsyn V. A., Aja S. U., Kulik D. A. and Wood S. A. (2000) Acid–base surface chemistry and sorption of some lanthanides on K+-saturated marblehead illite: I. Results of an experimental investigation. Geochim. Cosmochim. Acta 64, 185–194. Somerville P., Beavis S. and Welch S. (2004) Spatial and temporal variability of acidity at a coastal acid sulfate soil site. In Advances in Regolith (ed. I. C. Roach). CRC LEME, pp. 335– 339. Stille P., Steinmann M., Pierret M.-C., Gauthier-Lafaye F., Chabaux F., Viville D., Pourcelot L., Matera V., Aouad G. and Aubert D. (2006) The impact of vegetation on REE fractionation in stream waters of a small forested catchment (the Strengbach case). Geochim. Cosmochim. Acta 70, 3217–3230. Sullivan L. A. and Bush R. T. (2004) Iron precipitate accumulations associated with waterways in drained coastal acid sulfate landscapes of eastern Australia. Marine Freshwater Res. 55, 727–736.

Tulau, M. J., Naylor, S. D. (1999) Acid sulfate soil management priority areas in the lower Macleay floodplain. NSW Department of Land and Water Conservation. Taunton A. E., Welch S. A. and Banfield J. F. (2000a) Microbial controls on phosphate and lanthanide distributions during granite weathering and soil formation. Chem. Geol. 169, 371– 382. Taunton A. E., Welch S. A. and Banfield J. F. (2000b) Geomicrobiological controls on light rare earth element, Y and Ba distributions during granite weathering and soil formation. J. Alloy. Compd. 303, 30–36. Taylor S. R. and McLennan S. M. (1985) The Continental Crust: Its Composition and Evolution. Oxford, Blackwell, p. 312. Tyler G. (2004) Rare earth elements in soil and plant systems—A review. Plant Soil 267, 191–206. Vercouter T., Amekraz B., Moulin C., Giffaut E. and Vitorge P. (2005) Sulfate complexation of trivalent lanthanides probed by nanoelectrospray mass spectrometry and time-resolved laserinduced luminescence. Inorg. Chem. 44, 7570–7581. Verplanck P. L., Nordstrom D. K. and Taylor H. E. (1999) Overview of rare earth element investigations in acid waters of U.S. Geological Survey abandoned mine lands watersheds. U.S. Geol. Survey, Water-Resources Investigations Report, vol. 994018A, pp. 83–92. Verplanck P. L., Nordstrom D. K., Taylor H. E. and Kimball B. A. (2004) Rare earth element partitioning between hydrous ferric oxides and acid mine water during iron oxidation. Appl. Geochem. 19, 1339–1354. Webb G. E. and Kamber B. S. (2000) Rare earth elements in Holocene reefal microbialites: a new shallow seawater proxy. Geochim. Cosmochim. Acta 64, 1557–1565. Welch S. A., Christy A. G., Kirste D., Beavis S. G. and Beavis F. (2007) Jarosite dissolution I—Trace cation flux in acid sulfate soils. Chem. Geol. 245, 183–197. Welch S. A., Kirste D., Christy A. G., Beavis F. R. and Beavis F. (2008) Jarosite dissolution II—Reaction kinetics, stoichiometry and acid flux. Chem. Geol. 254, 73–86. White I., Melville M. D., Wilson B. P. and Sammut J. (1997) Reducing acid discharges from coastal wetlands in eastern Australia. Wetlands Ecol. Manage. 5, 55–72. Whittaker S. G. and Kyser T. K. (1993) Variations in the neodymium and strontium isotopic composition and REE content of molluscan shells from the Cretaceous Western Interior seaway. Geochim. Cosmochim. Acta 57, 4003–4014. Wood S. A. (1990) The aqueous geochemistry of the rare-earth elements and yttrium 1. Review of available low-temperature data for inorganic complexes and the inorganic REE speciation of natural waters. Chem. Geol. 82, 159–186. Wood S. A., Gammons C. H. and Parker S. R. (2006) The behavior of rare earth elements in naturally and anthropogenically acidified waters. J. Alloy. Compd. 418, 161–165. Worrall F. and Pearson D. G. (2001) Water–rock interaction in an acidic mine discharge as indicated by rare earth element patterns. Geochim. Cosmochim. Acta 65, 3027– 3040. Zhong A. and Mucci A. (1995) Partitioning of rare earth elements (REEs) between calcite and seawater solutions at 25 °C and 1 atm, and high dissolved REE concentrations. Geochim. Cosmochim. Acta 59, 443–453. Associate editor: Karen Johannesson