Controls on uranium distribution in lake sediments

Available online at www.sciencedirect.com

Geochimica et Cosmochimica Acta 74 (2010) 203–214 www.elsevier.com/locate/gca

Controls on uranium distribution in lake sediments Anthony Chappaz 1, Charles Gobeil *, Andre´ Tessier INRS-ETE, Universite´ du Que´bec, 490 de la Couronne, Que´bec, Que´., Canada G1K 9A9 Received 21 January 2009; accepted in revised form 18 September 2009; available online 29 September 2009

Abstract Uranium geochemistry has been investigated in three acid lakes located on the Canadian Shield and one circumneutral lake in the Appalachian Region of Eastern Canada. In all Shield lakes, dissolved U concentrations were higher in the porewater than in the overlying water. In one of them, whose hypolimnion is perennially oxic, U released to porewater at depths of Fe remobilization was removed from the porewater at depths of Fe oxyhydroxides precipitation; these similarities in the U and Fe profiles indicate that part of the U becomes associated to Fe oxyhydroxides. The dissolved U and Fe profiles in the other two Shield lakes, whose hypolimnions were anoxic when sampled, did not show any significant recycling of these elements in the vicinity of the sediment–water interface and both elements diffused from the sediment to the overlying water. In contrast, in the Appalachian Lake, dissolved U concentrations were higher in the overlying water than in porewater, strongly decreased at the vicinity of the sediment–water interface and then remained relatively constant with sediment depth. Diagenetic modeling of the porewater U profiles, assuming steady-state, reveals that authigenic U always represented 63% of the total U concentration in the sediments of all lakes. This observation indicates that diagenetic reactions involving U are not quantitatively important and that most of the U was delivered to the sediments at our study sites as particulate U and not through diffusion across the sediment–water interface, as is seen in continental margin sediments. Comparison of the U:Corg and U:Fe molar ratios in diagenetic material collected across the sediment–water interface with Teflon sheets and in surface sediments (0–0.5 cm) of the lake having a perennially oxic hypolimnion suggest that solid phase U was mainly bound to organic matter originating from the watershed; a strong statistical correlation between sediment non-lithogenic U and Corg in the Appalachian Lake supports this contention. Thermodynamic calculations of saturation states suggest that dissolved U was not removed from porewater through precipitation of UO2(s), U3O7(s) and U3O8(s) as previously proposed in the literature. Crown Copyright Ó 2009 Published by Elsevier Ltd. All rights reserved.

1. INTRODUCTION The observation in the marine environment that uranium accumulates at higher rates in reducing than in oxidized sediments raised considerable scientific interest for its use as a proxy of past environmental changes of the ocean (e.g., Francßois et al., 1993; Mangini et al., 2001; Tribovillard et al., 2006; Arnaboldi and Meyers, 2007). The most commonly identified pathway to explain the accu* Corresponding author. Tel.: +1 418 654 2589; fax: +1 418 654 2600. E-mail address: [email protected] (C. Gobeil). 1 Present address: Department of Earth Sciences, University of California Riverside, CA 92521-0423, USA.

mulation of authigenic U in reduced sediments is the microbially mediated reduction of U(VI), which exists as a soluble carbonate complex in oxygenated seawater, to less soluble U(IV) at redox conditions close to those allowing Fe(III) or SO4 reduction (Klinkhammer and Palmer, 1991). As a result of this redox transformation, a U concentration gradient develops across the sediment–water interface, driving a flux of dissolved U into the sediments whose magnitude is believed to depend on the depositional rate of organic matter at the seafloor and/or the O2 level in bottom water (e.g., Lovely et al., 1991; Francßois et al., 1997; Fredrickson et al., 2000; Zheng et al., 2002; Sani et al., 2004; McManus et al., 2005, 2006). However, the quantitative use of U as a paleoceanographic tracer is still limited by our lack of knowledge of the exact U fixation mechanism in

0016-7037/$ - see front matter Crown Copyright Ó 2009 Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.gca.2009.09.026

204

A. Chappaz et al. / Geochimica et Cosmochimica Acta 74 (2010) 203–214

sediments, as well as of its post-depositional redistribution (McManus et al., 2005). As for other redox sensitive elements, current knowledge of aquatic U geochemistry comes mainly from research in the oceanic environment. However, previous investigations on Mo and Re in lake sediments have proven to be useful in identifying reactions and processes affecting redox sensitive elements during diagenesis (Chappaz et al., 2008a,b); this study complements these previous investigations. Our specific goals were to clarify the geochemical processes responsible for U accumulation in freshwater sediments, to quantify authigenic U in these sediments, and to improve our knowledge on the geochemical reactions that control porewater U concentrations during early stages of diagenesis. 2. METHODS 2.1. Study areas and sampling Four Eastern Canada lakes were sampled between July 2003 and August 2005. Lakes Tantare´, Vose and Despe´riers lie on the Canadian Shield in Eastern Canada and have acidic water, whereas Lake Holland is part of the Appalachian Region and its slightly alkaline water is enriched in carbonate, calcium, and magnesium relative to that of the Canadian Shield lakes. Geographic coordinates of all lakes are given by Chappaz et al. (2008a). The hypolimnion of the sampled basin in Lake Tantare´ is perennially oxygenated, whereas those of the three other lakes become seasonally anoxic and were anoxic at the sampling time. Porewater samples for pH, U, Al, Fe, Mn, major cations and anions, sulfide (RS(II)), and inorganic and organic carbon measurements were collected by in situ dialysis, using acrylic peepers that were deployed by divers at the deepest site of each basin for 21 days. Those used in Lakes Vose and Despe´riers, and for the July and September 2003 campaigns in Lake Tantare´, were designed to collect samples at a uniform vertical resolution of 1 cm, from 5 cm above the sediment–water interface to 10 cm depth in the sediments, whereas those used in Lake Holland and for the August 2004 campaign in Lake Tantare´ were designed to collect samples at a vertical resolution of 0.5 cm, from 2.5 cm above the sediment–water interface to 2.5 cm depth in the sediments, and then at a resolution of 1 cm down to 10 cm depth. Preparation and procedures to sample the peepers were described in detail by Chappaz et al. (2008a). Sediment cores were also collected by divers at the porewater sampling site in each lake using 9.5-cm internal diameter butyrate tubes; they were extruded within 2 h and sectioned at 0.5-cm intervals to 10 or 15 cm depth and then at 1-cm intervals to the bottom of the cores. To quantify the adsorption of U onto authigenic Fe oxyhydroxides, Teflon sheets (7.5 cm  15 cm) that had been inserted vertically in the sediments of Lake Tantare´ in October 1993 were recovered in August 2006 (Chappaz et al., 2008a,b). During this 12-yr period, the sheets have collected Fe oxyhydroxides that precipitated across the sediment–water interface; these authigenic Fe oxyhydroxides were identified as poorly crystallized ferrihydrite and lepidocrocite (Fortin et al., 1993). Following their retrieval,

the Teflon sheets and their diagenetic Fe oxyhydroxide deposits were rinsed with lake water and transported in polyethylene containers. 2.2. Analyses Uranium measurements were carried out on the sediment core samples, the Fe-rich diagenetic material collected on Teflon sheets and the porewater samples collected by Chappaz et al. (2008a,b). Uranium concentrations in the solutions resulting from complete acid dissolution of the sediments (McLaren et al., 1995) were measured with an inductively coupled plasma quadrupole mass spectrometer (ICP-MS; Thermo Instrument X7) using external calibration. Replicate analyses (n = 5) of U concentration in the reference materials MESS-3 (recommended value of 4 lg g1) and PACS-2 (recommended value of 3 lg g1) from the National Research Council of Canada yielded 3.8 ± 0.1 lg g1 and 2.9 ± 0.2 lg g1, respectively. Dissolved U in porewater samples, as well as in solutions resulting from the solubilization of the Fe-rich diagenetic material, were also measured by quadrupole ICP-MS. Analytical precision and accuracy, determined from replicate analyses of the certified solution TM-DWS from Environment Canada, were better than 5%. The methods used for the other measurements carried out on the solid phase (Al, Ca, Fe, total N, total C, 210Pb, 137Cs, and 214Pb) and the porewater (pH, Ca, Fe, Mg, Mn, sulfide, SO4, and inorganic and organic C) were described by Chappaz et al. (2008a). Sediment ages and mass accumulation rates (x) at the sampling sites have been previously determined by 210Pb geochronology, assuming negligible sediment mixing (Chappaz et al., 2008a,b). Ranges found for x were 3.5– 6.7 mg cm2 yr1, 4.5–15.7 mg cm2 yr1, 3.8 mg cm2 yr1 and 7.8–16.4 mg cm2 yr1 for Lakes Tantare´, Holland, Despe´riers and Vose, respectively. Speciation of porewater U was calculated with the computer programs WHAM 6 (Tipping, 2002) and MINEQL+ (Schecher and McAvoy, 1998) using pH and total concentrations of dissolved U, Al, Fe, Mn, K, Mg, Ca, Na, inorganic carbon, organic carbon, SO42, Cl and RS(II) as inputs. The thermodynamic databases of WHAM 6 and MINEQL+ were updated for the reactions given in Table 1. 3. RESULTS 3.1. Porewater The low sulfide (<0.02 lM; Fig. 1m–o) and Fe (<2 lM; Fig. 1g–i) concentrations measured in the water overlying Lake Tantare´ sediment are consistent with the relatively high O2 concentrations (>120 lM; data not shown) measured in the bottom water of this lake when sampling occurred; collectively, these data indicate that the hypolimnion and the sediment surface were oxic. In contrast, the overlying water of the three other lakes showed lM concentrations of sulfide (Fig. 1p–r) as well as high Fe concentrations (Fig. 1j and k), except in the Fe-limited Lake Holland. Furthermore, O2 concentrations in the

Uranium in sediments

205

Table 1 Equilibrium constants and corresponding reactions used in the calculations. Reaction 2+

+

+

UO2 + H2O = UO2OH + H UO22+ + 2H2O = UO2(OH)2(aq) + 2H+ UO22+ + 3H2O = UO2(OH)3 + 3H+ UO22+ + 4H2O = UO2(OH)42 + 4H+ 2UO22+ + H2O = (UO2)2OH3+ + H+ 2UO22+ + 2H2O = (UO2)2(OH)22+ + 2H+ 3UO22+ + 4H2O = (UO2)3(OH)42+ + 4H+ 3UO22+ + 5H2O = (UO2)3(OH)5+ + 5H+ 3UO22+ + 7H2O = (UO2)3(OH)7 + 7H+ 4UO22+ + 7H2O = (UO2)4(OH)7+ + 7H+ UO22+ + CO32 = UO2(CO3)(aq) UO22+ + 2CO32 = UO2(CO3)22 UO22+ + 3CO32 = UO2(CO3)34 3UO22+ + 6CO32 = (UO2)3(CO3)66 2UO22+ + CO32 + 3H2O = (UO2)2CO3(OH)3 + 3H+ 3UO22+ + CO32 + 3H2O = (UO2)3CO3(OH)3+ + 3H+ 11UO22+ + 6CO32 + 12H2O = (UO2)11(CO3)6(OH)122 + 12H+ Ca2+ + UO22+ + 3CO32 = CaUO2(CO3)32 2Ca2+ + UO22+ + 3CO32 = Ca2UO2(CO3)3(aq) Fe2+ + HS = H+ + FeSm(s) (mackinawite, disordered) „Fe(s,w)OH = „Fe(s,w)O + H+ „Fe(s,w)OH + H+ = „Fe(s,w)OH2+ „Fe(s,w)OH + 2H+ + CO32 = „Fe(s,w) CO3H0 + H2O „Fe(s,w)OH + H+ + CO32 = „Fe(s,w) CO3 + H2O „Fe(s)(OH)2 + UO22+ = („Fe(s)O2) UO20 + 2H+ „Fe(w)(OH)2 + UO22+ = („Fe(w)O2) UO20 + 2H+ „Fe(s)(OH)2 + UO22+ + CO32 = („Fe(s)O2) UO2CO32 + 2H+ „Fe(w)(OH)2 + UO22+ + CO32 = („Fe(w)O2) UO2CO32 + 2H+ UO22+ + 2e = UO2(s) 3UO22+ + 4e + H2O = U3O7(s) + 2H+ 3UO22+ + 2e + 2H2O = U3O8(s) + 4H+ UO22+ + 4e + HS + 3H+ = US(s) + 2H2O UO22+ + 2e + 2HS + 2H+ = US2(s) + 2H2O UO22+ + 3HS + H+ = US3(s) + 2H2O UO22+ + [HA]F = [UO2  HA] (L. Tantare´) UO22+ + [FA]F = [UO2  FA] (L. Tantare´) UO22+ + [HA]F = [UO2  HA] (L. Vose) UO22+ + [FA]F = [UO2  FA] (L. Vose) UO22+ + [HA]F = [UO2  HA] (L. Despe´riers) UO22+ + [FA]F = [UO2  FA] (L. Despe´riers) UO22+ + [HA]F = [UO2  HA] (L. Holland) UO22+ + [FA]F = [UO2  FA] (L. Holland)

log K

Ref.

5.25 12.15 20.25 32.40 2.70 5.62 11.90 15.55 32.20 21.90 9.94 16.61 21.84 54.00 0.86 0.65 36.93 25.4 30.6 3.5 9.13 6.51 19.58 11.59 2.34 6.14 4.37 0.56 14 26 6.6 25 11 17 2.143.07b 2.223.09b 2.933.14b 3.203.72b 2.863.20b 3.033.23b 6.237.09b 7.067.92b

Guillaumont et al. (2003) Guillaumont et al. (2003) Guillaumont et al. (2003) Guillaumont et al. (2003) Guillaumont et al. (2003) Guillaumont et al. (2003) Guillaumont et al. (2003) Guillaumont et al. (2003) Guillaumont et al. (2003) Guillaumont et al. (2003) Guillaumont et al. (2003) Guillaumont et al. (2003) Guillaumont et al. (2003) Guillaumont et al. (2003) Guillaumont et al. (2003) Guillaumont et al. (2003) Guillaumont et al. (2003) Bernhard et al. (2001) Bernhard et al. (2001) Rickard (2006) Davis (2001)a Davis (2001)a Davis (2001)a Davis (2001)a Davis (2001)a Davis (2001)a Davis (2001)a Davis (2001)a Guillaumont et al. (2003) Wagman et al. (1982) Wagman et al. (1982) Guillaumont et al. (2003) Guillaumont et al. (2003) Guillaumont et al. (2003) Calculated Calculated Calculated Calculated Calculated Calculated Calculated Calculated

a In the adsorption reactions provided by Davis (2001), “„” refers to adsorption sites, the reported constants are intrinsic constants, and subscripts s and w are for “weak” and “strong” adsorption sites, respectively. b Conditional constants calculated from the output of WHAM 6.

bottom water of these lakes were below detection limit when sampling occurred. All of the results thus indicate that the sediment surface of these lakes was anoxic. Dissolved U concentrations in water overlying the sediment of the three Canadian Shield lakes (<30 pM) were much lower than in that of the Appalachian Lake (up to 290 pM) (Fig. 1a–f). These U values are 2–3 orders of magnitude lower than those in seawater, i.e., 14 nM (e.g., Ku et al., 1977). Porewater U concentrations in all lakes fall within the range of the few values available for lake sediment porewater (42– 416 pM; Nagao et al., 2002). In each of the lakes, the three replicate profiles at a given sampling date are always consistent indicating the absence of sampling and analytical artefacts. In Lake Holland, dissolved U concentrations were

higher in the overlying water than in porewater, strongly decreased at the vicinity of the sediment–water interface due to diffusion into the sediments and then remained relatively constant with sediment depth; this pattern is usually observed in reducing marine sediments (McManus et al., 2005; Morford et al., 2007). In contrast to Lake Holland, however, dissolved U concentrations were higher in the porewater than in the overlying water of the three Canadian Shield lakes (Fig. 1a–e). In the perennially oxic basin of Lake Tantare´, mobilization of uranium in the porewater occurred roughly at depths where Fe was mobilized from the sediment and it was removed from porewater approximately at depths where Fe was precipitated as Fe oxyhydroxides (Fig. 1a–c and g–i). It should be noted that we assumed the presence of a thin

206

A. Chappaz et al. / Geochimica et Cosmochimica Acta 74 (2010) 203–214 [U] (pM) 0 10 20 30 40 -4 -2 0 2 4 6 8 (a) 10

L. Tanataré July 03

Depth (cm)

log IAP/Ks -4 0 4 8 12 16

2

r =0.99

pH

FeSm UO2

(g)

0

50

100

5 0

6 4

2

7 8

6

(y)

(s)

(m)

L. Tantaré

0

2

4

6

8

-2 -1 0 1 2 3 0 10 20 30 40

(ε)

-4 0 4 8 12 16

2

r =0.97

pH

FeSm UO2

ΣS(-II)

Sept. 03

(t) (h)

0 10 20 30 40 -4 L. Vose Sep 03 -2 0 2 4 6 8 (d) 10 0 10 20 30 40 -4 -2 0 2 L. Despériers 4 June 01 6 8 (e) 10 0

[U] (pM) 0 10 20 30 40

U3O7

0 10 20 30 40 L. Tantaré -4 Aug. 04 -2 0 2 4 6 8 (c) 10

-4 -2 0 2 4 6 8 10

-1

[DOC] (mg L ) 0 2 4 6 8

ΣS(-II)

0 10 20 30 40 -4 -2 0 2 4 6 8 (b) 10

[Fe] (µM) ΣS(-II) (µM) 25 50 75 0 2 4 6 8

0

0

(z)

(n)

25

50

5 75 0

6 4

2

7 8

6

0

2

4

6

U3O7

8

-2 -1 0 1 2 3 0 10 20 30 40

(φ)

-4 0 4 8 12 16

2

r =0.88

UO2 U3O7

pH

FeSm

ΣS(-II)

0

100

5 200 0

6 4

2

7 8

6

(α)

(u)

(o)

(i)

0

5

10

15

-2 -1 0 1 2 3 0 10 20 30 40

ΣS(-II)

(γ)

-4 0 4 8 12 16 FeSm

2

r =0.88

pH UO2

0

100

(β)

(v)

(p)

(j)

5 6 7 200 0 5 10 15 20

0

5

10

15

-2 -1 0 1 2 3 0 10 20 30 40 2

r =0.99

ΣS(-II)

U3O7

(η)

-4 0 4 8 12 16 FeSm

pH

UO2

U3O7

(χ) (q)

(k)

(ι)

(w)

-2 -1 0 1 2 3

5 6 7 100 200 300 0 2 4 6 8 10 0 10 20 30 40 0

5 10 15 20 0

100 200 300 -4 0 4 8 12 16 FeSm

pH

U3O7

ΣS(-II) L. Holland Aug. 05

(f)

2

r =0.96

(r)

(l)

6

(x)

7

pH

8

(δ)

UO2

(ϕ)

-10-8 -6 -4 -2 0

Rates (10-20mol cm-3 s-1)

Fig. 1. Triplicate porewater profiles of U, Fe, RS(II), pH and DOC are given for all lakes and sampling dates in the first four columns (panels a to x). In the fifth column (panels y to d), comparison is made between the measured (open circles) and modeled (solid line following the open circles) average (n = 3) porewater U concentrations; the other solid line in panels y to d represents the U net reaction rate (RU net ) profile. In the sixth column (panels e to u), calculated values of log IAP/Ks for U3O7(s) (filled circles), UO2(s) (open circles) and disordered mackinawite FeSm(s) (gray circles) are reported (IAP is the ion activity product and Ks is the solubility product); the vertical broken line represents equilibrium. The horizontal broken line in all panels indicates the sediment–water interface and the thin gray line above the sediment–water interface in panels b, f, z and d represents the upper limit of the nepheloid layer.

Uranium in sediments

lakes, ranging between 2 and 4 nmol g1 in Lakes Vose and Despe´riers and between 7 and 9 nmol g1 in Lake Tantare´ (Fig. 2a–c); the U:Al molar ratios in the sediments of these lakes (Fig. 2i–k) were within a factor 3 of the average lithogenic U:Al ratio (2.0  106; McManus et al., 2006). In contrast, the sedimentary U concentrations in Lake Holland were significantly higher than in the other lakes, and varied with sediment depth (Fig. 2d). The U profile exhibited a subsurface maximum at 1.75 cm depth; below this horizon, the U concentrations decreased to 5.75 cm depth and then constantly increased down to the bottom of the core, reaching concentrations above 20 nmol g1 below

nepheloid layer above the sediment–water interface in September 2003 for this lake (Chappaz et al., 2008a) and for Lake Holland. The dissolved U and Fe profiles in the anoxic Lakes Vose and Despe´riers did not show any significant recycling of these elements at the vicinity of the sediment–water interface (Fig. 1d, e, j and k) and both elements diffused to the overlying water. 3.2. Solid phase Sediment uranium concentrations were low and relatively constant with depth in the three Canadian Shield -1

{Corg} (%)

{U} (nmol g ) 0

2

4

6

8 10

207

0

10

20

6

{U}/{Al} (x10 )

30

40

0

2

4

6

8

0 5 Al

10 15

Corg

20 25 L. Tantaré

30

(a)

(e)

0 0

2

4

6

8 10

0

1 10

2 20

(i)

3

30

40

0

2

4

6

8

0 Al

5 10 15

Corg

20

Depth (cm)

25 L. Vose

30

(f)

(b)

0 0

2

4

6

8 10

0

1 10

2 20

(j)

3

30

40

0

2

4

6

8

0 5

Corg

Al

10 15 20 25 L. Despériers

30

(g)

(c)

0 0

5 10 15 20 25

0

1 10

2 20

(k)

3

30

40

0

10

20

50 100

0 5 10

Corg

Al

15 20 25 30

L. Holland

(d)

(h)

0

1

2

(l)

3 -1

{Al} (mmol g ) Fig. 2. Depth profiles of U (a–d), Corg and Al (e–h) concentrations and of U:Al molar ratio (i–l) in the sediments of all lakes.

208

A. Chappaz et al. / Geochimica et Cosmochimica Acta 74 (2010) 203–214

16

and dissolved Ca2+ and CO32 concentrations in Lake Holland than in the three Canadian Shield lakes, the predicted dominant aqueous species was the mixed complex Ca2 UO2 ðCO3 Þ3 0 (P97% of total dissolved U).

14

4. DISCUSSION

20 2

r = 0.75

-1

{UNL} (nmol g )

18

12

4.1. Modeling of U porewater profiles 10 8 Lake Holland

6 15

20

25

30

35

40

{Corg } (%)

Fig. 3. Plot of non-lithogenic (UNL) versus {Corg} in Lake Holland sediments.

Table 2 Values of U:Fe, U:Corg, Corg:Fe and C:N molar ratios measured in surface sediments (0–0.5-cm layer) and iron-rich material collected on Teflon sheets in Lake Tantare´. Molar ratio

Surface sediments

Teflon sheets

U:Fe U:Corg Corg:Fe C:N

46  107 3.7  107 13 16 ± 1

8 ± 2  107 3.3 ± 0.3  107 2.6 ± 0.3 15 ± 3

25 cm depth. The U:Al molar ratio in Lake Holland sediments was thus up to 47 times higher than the average lithogenic U:Al ratio (Fig. 2l). It should be noted that the sediment organic carbon concentrations in this lake vary with depth in a similar fashion to that of the U concentrations (Fig. 2d and h). Fig. 3 indicates that non-lithogenic U (UNL), defined as UNL = UMeasured  [AlMeasured  (U:Al)Lithigenic], where (U:Al)Lithogenic is 2.0  106 (McManus et al., 2006), was significantly correlated with Corg (r2 = 0.75; n = 30). The mean (±SD) C:N molar ratios in sediment from Lakes Tantare´, Despe´riers and Holland were 16 ± 1, 16 ± 1 and 13 ± 1, respectively (N data unavailable for Lake Vose). These ratios are much higher than the C:N ratio for phytoplankton (6.6; Redfield, 1934) but within the range of those of humic substances in soils (range of 12–23; Buffle, 1988), indicating that the sediment organic matter was mainly derived from the surrounding watershed. The mean (±SD) U:Fe, Corg:Fe, U:Corg and C:N molar ratios for the iron-rich deposits collected by the Teflon sheets were 8 ± 2  107, 2.6 ± 0.3, 3.3 ± 0.3  107 and 15 ± 3, respectively (Table 2). 3.3. Speciation of dissolved U(VI) Speciation calculations with the code WHAM 6 predict that the concentrations of two U species were distinctly preponderant, depending on the study lake. In the three lakes located on the Canadian Shield, the neutral complex UO2 CO3 0 was predicted to dominate U speciation in the sediment overlying waters and in the porewaters (41–86% of total dissolved U). In contrast, due to the higher pH

Porewater U profiles are sensitive indicators of reactions involving U in the solid phase, and their modeling can lead to a better understanding of U diagenesis. The following one-dimensional mass conservation equation (Boudreau, 1997) can be used to define the depth intervals (zones) where U is produced to or consumed from porewater and to estimate the net reaction rate for the processes responsible for U production or consumption:     @/½U @ @½U ¼ /ðDS þ DB Þ @t @x @x x   þ /a ½Uburrow  ½U þ RU ð1Þ net ¼ 0 In this equation x is depth (cm; positive downward from the sediment–water interface), / is sediment porosity, t is time (s), DS is the effective diffusion coefficient (cm2 s1), DB is the biodiffusion coefficient (cm2 s1), ½Uburrow is the concentration of dissolved U in the burrows of benthic animals (mol cm3 of porewater), which we assume to be identical to that in the water overlying the sediments, ½U is the U porewater concentration (mol cm3), a is the bioirrigation coefficient (s1), and RU net is the net reaction rate (mol cm3 of whole sediment s1) of porewater U producU tion (RU net > 0) or consumption (Rnet < 0). We neglected advective fluxes due to sediment compaction and groundwater flow. With the assumption of steady-state, Eq. (1) was solved numerically for RU net by applying the code PROFILE (Berg et al., 1998) to the average dissolved U profiles obtained for each lake and sampling period. We assumed that DS ¼ /2 Dw (Berner, 1980), where Dw is the tracer diffusion coefficient of the U species present in porewater and / is our measured values of sediment porosity. Estimates of the Dw values for the two predominant species: the UO2 CO3 0 complex for the Canadian Shield lakes and the Ca2 UO2 ðCO3 Þ3 0 complex for Lake Holland were obtained by the following equation (Schwarzenbach et al., 1993): Dw ¼

2:7  104 M 0:71

ð2Þ

where M represents the molecular mass. Dw values of 4.41  106 cm2 s1 and 3.14  106 cm2 s1 were obtained for UO2 CO3 0 and Ca2 UO2 ðCO3 Þ3 0 at 25 °C, respectively, and were subsequently corrected for the effect of temperature with the Stokes–Einstein equation (Li and Gregory, 1974), assuming an in situ temperature of 4 °C. The value used for the biodiffusion coefficient DB in Lake Tantare´ was 2.2  109 cm2 s1 (Gallon et al., 2004). This estimation was based on the work of Hare et al. (1994) who determined that the most abundant benthic animal at this site was the chironomid Sergentia coracina (1080 ± 140 individuals m2), and the laboratory

Uranium in sediments

209

Table 3 Modeling results for Lakes Tantare´, Vose, Despe´riers and Holland: net rates of porewater U production or consumption as a function of U depth in the sediments (RU net ), as well as fluxes of U deposited with settling particles (J dep ) and fluxes of dissolved U across the sediment–water U U interface due to diffusion (J U D ), bioirrigation (J I ) and biodiffusion (J B ). Lake

Zone

Depth interval

RU net 20

JU dep

JU I

JU B

2 1

cm

10

1.14 0.89 0.03

151

0.22

0.26

3.85  104

() (+) () (+)

2.0 to 0.72 0.72 to 0.55 0.55 to 1.83 1.83 to 9.50

1.60 2.09 0.81 0.08

151

0.25

0.53

4.11  104

() consumption (+) production (+) production

0 to 1.17 1.17 to 2.33 2.33 to 9.50

0.22 0.94 5  104

151

0.61

0.39

1.09  103

Vose

(+) production (+) production

0 to 1.05 1.05 to 9.50

1.10 0.01

111

–1.24

Despe´riers

(+) production () consumption

0 to 4.22 4.22 to 9.50

0.25 0.16

28

–0.75

Holland

() consumption () consumption

2.0 to 1.05 1.05 to 9.50

9.97 0.02

492

8.07

Tantare´ September 03

Tantare´ August 04

measurements of biodiffusion coefficients per individual for chironomids (Matisoff and Wang, 2000). The bioirrigation coefficient a was assumed to vary linearly from a0 at the sediment–water interface to zero at 10 cm depth (Gallon et al., 2004), since chironomids are generally not found below this sediment depth (Matisoff and Wang, 1998). Eq. (3) was used to estimate a0 (Boudreau, 1984): a0 ¼

DS r 1 ðr22  r21 Þðra  r1 Þ

ð3Þ

where r1 is the radius of an animal’s tube (0.1 cm), r2 is half the distance between adjacent tubes (1.5 cm), and ra is equal to r2/2. Because seasonal anoxia in the bottom waters of Lakes Vose, Despe´riers and Holland prevents the development of benthic communities, we assumed that the influences of bioturbation and bioirrigation were negligible in these lakes and thus DB and a in Eq. (1) were assumed to be equal to zero. In all lakes, the best fits provided by PROFILE are in very good agreement with the measurements (r2 = 0.88– 0.99; Fig. 1y–d). In Lake Tantare´, PROFILE consistently defined two zones (depth intervals) where reaction rates are relatively fast: one just below the sediment–water interface (July 2003 and August 2004) or in the nepheloid layer (September 2003), where dissolved U was consumed, and the other, just below the consumption zone, where dissolved U was produced. At greater depths, U was either slowly produced (July 2003) or consumed (August 2004); in September 2003, U was consumed between 0.5 and 1.8 cm depth and slowly produced below this horizon (Fig. 1). In the two Canadian Shield lakes whose hypolimnion was anoxic at sampling time, the model determined a zone of U production between 0 and 4 cm followed by a

10

JU D

0 to 1.05 1.05 to 2.11 2.11 to 9.50

consumption production consumption production

s

20

() consumption (+) production (+) production

Tantare´ July 03

mol cm

3 1

mol cm

s

zone of either slow production (Lake Vose) or consumption (Lake Despe´riers) (Fig. 1). In contrast, PROFILE defined a zone of intense consumption rate of dissolved U between 2 and 1 cm depth in Lake Holland sediments. The net reaction rates (RU net ) obtained for each zone are given in Table 3 for each lake and time period. Note that for the modeling with PROFILE of the Lake Holland and September 2003 Lake Tantare´ results, the sediment–water interface was shifted to the upper limit of the nepheloid layer. 4.2. Influence of diagenesis on sedimentary records and fluxes of U Since sediment mixing was negligible at all our sampling sites (Chappaz et al., 2008a,b), the measured sediment U concentration represents the sum of the U concentration in the settling particles deposited at the sediment surface and that of the U added to or removed from the solid phase by diagenetic reactions during sediments burial. To estimate the latter concentration, we used the following equation (Chappaz et al., 2008a) which relates the net removal or production rate of porewater U to the net rate of U fixation to or release from the solid phase:     d½U dfUg U Rnet ¼ / ¼ m ð4Þ dt reaction dt reaction In this equation, m is the dry bulk density (g cm3 of whole sediment), and fUg is the solid phase U concentration (mol g1). The subscript “reaction” indicates reaction rates in solution and solid phases. From Eq. (4), it follows that: dfUg ¼ 

RU RU net dt ¼  net dx m mvS

ð5Þ

210

A. Chappaz et al. / Geochimica et Cosmochimica Acta 74 (2010) 203–214

and fUgauthigenic ¼ 

Z

x¼xi

x¼0

x¼xi X RU net dx   Dx mvS mv S x¼0

RU net

ð6Þ

where xi is the depth of a sediment layer and vS is the sedimentation rate (cm s1). {U}authigenic can be viewed as the solid U concentration gained or lost by a layer of sediment during its burial. To obtain the U concentration at the time of sediment deposition (fUgdeposited ), the values of {U}authigenic are subtracted from those of measured U (fUgmeasured ). Using the RU net values obtained with PROFILE our calculations show that {U}authigenic represents at the most in all lakes 3.3% of {U}measured. This result indicates that the influence of diagenetic processes on the sedimentary records of U in all lakes was negligible, and that the {U}measured profiles reflect closely the historical record of U deposition at the sediment–water interface. The present-day total flux responsible for U accumula´ is the sum of tion in the sediments (J U acc ) of Lake Tantare the fluxes of U deposited at the sediment surface with settling particles (J U dep ) and those of dissolved U transported across the sediment–water interface by molecular diffusion U U (J U D ), bioirrigation (J I ) and bioturbation (J B ): U U U U JU acc ¼ J dep þ J D þ J I þ J B

ð7Þ

Due to the seasonal anoxia in Lakes Vose, Despe´riers and Holland, Eq. (7) can be reduced for these lakes to: U U JU acc ¼ J dep þ J D

ð8Þ

The present-day values of J U dep for all lakes were obtained by multiplying the sediment mass accumulation rate (x; mg cm2 yr1) obtained from the 210Pb geochronology for the top 0.5-cm sediment layer by the value of fUgmeasured since fUgauthigenic was negligible in this sediment U U layer. The values of J U D , J I and J B were calculated with the code PROFILE (Table 3). In all lakes, the contribution of U JU dep represents more than 95% of J acc . This finding is consistent with the negligible contribution of fUgauthigenic to fUgmeasured and it clearly indicates that U in our lacustrine sediments derives essentially from deposition of particulate U. The great importance of settling particles in the transport of U to lake sediments contrasts with the preponderance of U diffusion across the sediment–water interface observed in the marine environment where dissolved U concentrations in the water column are higher and sedimentation rates lower (e.g., McManus et al., 2005, 2006). 4.3. Uranium diagenesis and association to sediment phases In this section, we use thermodynamic calculations, in combination with our results obtained from the analysis of porewater, sediments and the Fe-rich authigenic material collected on Teflon sheets to constrain fixation mechanisms of dissolved U to the solid phases and U association to sediment phases. 4.3.1. Diagenesis under oxic condition Modeling the porewater U profiles obtained at all sampling dates in Lake Tantare´ consistently shows that U was

released to porewater a few cm below the sediment–water interface and that it was removed from it just above this production zone (Fig. 1y–a). The depths of U production to and removal from porewater coincide with those reported for Mo at the same sampling dates in this lake (Chappaz et al., 2008a); these U production and consumption zones also coincide grossly with the depths of Fe(II) production to the porewater and its removal as authigenic Fe oxyhydroxides following Fe(II) oxidation. Chappaz et al. (2008a) demonstrated that Mo was closely coupled to Fe, being released to porewater as a consequence of the reductive dissolution of Fe oxyhydroxides and removed from porewater by adsorption onto authigenic Fe oxyhydroxides. Based on the above observations, we thus conclude that, like Mo, the distribution and mobility of dissolved U in the sediments is tightly coupled to Fe redox recycling in Lake Tantare´. It should be noted, however, that the sediment U profile in Lake Tantare´ (Fig. 2a) does not display a sharp surface maximum as observed in the Mo and Fe solid phase profiles (Chappaz et al., 2008a); this peculiarity of U is to be attributed to the low contribution of diagenesis (i.e., Fe redox cycling) to the measured sediment U profile (see Section 4.2). Moreover, the porewater U profiles of the three anoxic lakes are consistent with our interpretation of a close relation between porewater U and Fe in Lake Tantare´. Indeed, in Lakes Despe´riers and Vose, the depths at which U was released to porewater corresponded grossly to the depths at which Fe was also released. No removal of U above the production zones was observed in these lakes simply because there was no formation of Fe oxyhydroxides when our porewater sampling occured since the bottom waters were anoxic (Fig. 1j, k and p, q). As for Lake Holland, the low and almost depth unvariable dissolved Fe concentrations (Fig. 1l) indicate that redox recycling of Fe in the sediments of this lake is unimportant. The mean (±SD) U:Fe molar ratio measured in the Ferich diagenetic material (8 ± 2  107; Table 2) that we collected on Teflon sheets in Lake Tantare´ was much smaller than that measured in the top 0.5-cm layer of the sediments (46  107; Table 2). Adsorption of U(VI) on ferrihydrite has been well studied and it has been interpreted in terms of the surface complexation model (e.g., Waite et al., 1994). To compute the U:Fe molar ratio predicted for U(VI) adsorption on authigenic Fe oxyhydroxides under Lake Tantare´ geochemical conditions, we have used the two-layer version of the surface complexation model (DLM) developed by Dzombak and Morel (1990) which is incorporated into the computer code FITEQL (Herbelin and Westall, 1994). For this calculation, we updated the FITEQL database with the intrinsic constants for the adsorption of U on ferrihydrite reported by Davis (2001; Table 1); we also adopted the intrinsic constants for H+ and carbonate adsorption (Table 1), as well as the specific surface area (600 m2 g1) and the concentration of strong (1.8  103 mol/mol of ferrihydrite) and total (0.875 mol/ mol of ferrihydrite) adsorption sites used by Davis (2001) to extract adsorption constants for UO22+. It should be noted that, in Davis (2001), the bidentate complexes of UO22+ with ferrihydrite, for structural reasons, were as-

Uranium in sediments

cribed an exponent of one for „FeOH in their respective mass action law equations and a coefficient of two for „FeOH in the mass balance equation. As a consequence, we had to use the code FITEQL for the calculation of the {Fe–UO2}/{Fe–ox} ratio because, to our knowledge, no other speciation code allows calculations with different coefficients in the mass action law and mass balance equations. Calculations described above were performed for each porewater data set at depths where removal of dissolved U occurred in Lake Tantare´ (i.e., at 0.5 cm, 1.5 cm and 0.5 cm depth in July 2003, September 2003 and August 2004, respectively). The results predict a mean (±SD) U:Fe molar ratio of 5.6 ± 0.2  107, which is 1.5-fold lower than that measured in the Fe-rich diagenetic material and much lower (about 8-fold) than that measured in the 0–0.5-cm layer of the sediments (Table 2). The relatively small difference between the U:Fe molar ratio predicted by the surface complexation model and that measured in the diagenetic material could be due to the fact that our modeling exercise did not take into account the formation of ternary complexes of Fe oxyhydroxides with humic substances and U(VI). Several experimental studies (Lenhart and Honeyman, 1999; Murphy et al., 1999) have shown that natural organic matter increases U(VI) adsorption onto Fe oxyhydroxides, but no usable surface complexation constants are available to take into account this effect. Another possibility is that a significant portion of U measured in the Fe-rich material was bound to the organic matter contained in the diagenetic material in other forms than ternary complexes (see Section 4.3.2 below). The large differences (7- to 8-fold) between the U:Fe molar ratio measured in the top 0.5-cm sediment layer and that measured in the diagenetic material or predicted by the surface complexation model clearly indicate that the proportion of the total U in the sediments of Lake Tantare´ associated with Fe oxyhydroxides is minor. This point will be elaborated in the next section. 4.3.2. Association of non-lithogenic U to sediment phases The values of the U:Al ratio in the sediments of Lake Tantare´ and Holland are significantly above the average lithogenic U:Al ratio (Fig. 2i–l). This observation indicates that the sediments of these lakes contain non-negligible amounts of non-lithogenic U. Assuming, as previously discussed, that the amount of U associated to Fe oxyhydroxides is minor, one can therefore raise the question to what sediments phase UNL is bound to. The Fe-rich diagenetic material contained large amounts of organic C in addition to Fe oxyhydroxides, the mean (±SD) Corg:Fe molar ratio being 2.6 ± 0.3 (Table 2). Corg:Fe molar ratios of similar magnitude have been reported for Fe-rich particles in lake water (Tipping et al., 1981). Although we cannot exclude that the Fe-rich material collected on Teflon sheets incorporates bacterial remains (Fortin et al., 1993), based on the high C:N molar ratio of this material (Table 2) and given the strong propensity of humic substances to adsorb onto Fe oxyhydroxides (Tipping, 1981; Gu et al., 1994), it is likely that the organic matter associated with the diagenetic Fe-rich material is mainly humic substances.

211

The organic matter present in the top 0.5-cm layer of Lake Tantare´ sediment has a very similar C:N molar ratio (16; Table 2) to that found in the diagenetic material collected on Teflon sheets (15 ± 3; Table 2), indicating a common source, which, given the high C:N values, is probably allochthonous. However, the Corg:Fe molar ratio is much larger in the sediment top layer (13; Table 2) than in the diagenetic material (2.6 ± 0.3; Table 2), revealing that only a small fraction (around 20%) of the sediment organic matter is associated with the Fe oxyhydroxides; the remaining organic matter is likely present as organic coating on other mineral phases than Fe oxyhydroxides and as discrete particles. Despite this large difference in the Corg:Fe molar ratio, the U:Corg molar ratio measured in the diagenetic material (3.3 ± 0.3  107) collected on Teflon sheets is very similar to that measured in the top 0.5-cm layer of the sediments (3.7  107; Table 2), suggesting that UNL is mainly associated with organic matter in the sediments of Lake Tantare´. Our finding that diagenetic reactions involving U are quantitatively negligible (see Section 4.2) shows that the association of U to the particulate organic matter is preserved during burial of Lake Tantare´ sediments. The strong correlation observed in Lake Holland between {UNL} and {Corg} (Fig. 3) also supports this contention for other lakes. 4.3.3. Diagenesis under anoxic condition Diagenetic modeling of the porewater U profiles reveals a zone of relatively fast removal of dissolved uranium just below the sediment–water interface for Lake Holland (Fig. 1d) and zones of slower removal below 0.5–4 cm depth for Lake Despe´riers (Fig. 1v) and for Lake Tantare´ in September 2003 (Fig. 1z). In the other cases, no removal of porewater U is predicted by PROFILE below the zone where Fe oxyhydroxides are reduced (Fig. 1y, a, b and Table 3). Removal of U in anoxic sediments has been attributed generally to the biologically-mediated reduction of U(VI) followed by the precipitation of low-solubility U(IV) solid (Klinkhammer and Palmer, 1991; Morford et al., 2001; McManus et al., 2005). Laboratory batch experiments, in combination with spectroscopic measurements, have also shown that U(VI) can be partially reduced by sulfide minerals such as galena, pyrite, mackinawite and amorphous FeS(s) and precipitated as UO2(s) or mixed U(VI) and U(IV) solids (Wersin et al., 1994; Moyes et al., 2000; Scott et al., 2007; Hua and Deng, 2008). Of particular interest is the study of Hua and Deng (2008) who reported that U(VI) removal by amorphous FeS(s) occurs in two steps: (i) a rapid exchange of Fe(II) for U(VI), followed by (ii) a slow and partial reduction of U(VI) by structural Fe(II) or S(II), leading to the formation of UO2(s) and other solids involving U(IV). Reactions for the formation of the various U(IV) solids, and their respective equilibrium constants, calculated from the DG0f values provided by Guillaumont et al. (2003) and Wagman et al. (1982), are listed in Table 1. To calculate the ion activity product (IAP) and thus test the saturation state of porewater with respect to these solids, we had to take into account both reduction of U(VI) to U(IV) and

212

A. Chappaz et al. / Geochimica et Cosmochimica Acta 74 (2010) 203–214

complexation of U(VI) by humic substances. To achieve this task, we run successively the codes WHAM 6 and MINEQL+ in the following manner. Conditional equilibrium constants for U(VI) complexation by humic (K cond UHA ) and fulvic (K cond UHA ) acids were first estimated with the code WHAM 6 under the geochemical conditions prevailing at each depth of a porewater profile: ½UO2  HA   K cond ð9Þ UHA ¼ UO2 2þ ½HAF and ½UO2  FA   K cond UFA ¼ UO2 2þ ½FAF

ð10Þ

where [UO2  HA] and [UO2  FA] are the concentrations (mol L1) of the humic and fulvic acid complexes with UO2 2þ , respectively, and [HA]F and [FA]F are the concentrations (g L1) of free (uncomplexed with UO2 2þ ) humic and fulvic acids, respectively. For estimating (Kcond UHA ) and 2þ (Kcond UFA ), we used the values of [UO2 ], [UO2  HA] and [UO2  FA] provided by the output of WHAM 6 and we assumed that a negligible fraction of HA and FA was complexed with UO2 2þ , i.e., that [HA]F and [FA]F where idencond tical to [HA] and [FA]. The ranges of (Kcond UHA ) and (KUFA ) values thus calculated for each lake are given in Table 1. cond The conditional constants Kcond UHA and KUFA were then incorporated into the code MINEQL+ and U speciation was calculated again, allowing the redox state of U to vary. In this latter calculation, we assumed that the redox potential was determined by the couple SO42/HS in porewater samples with sulfide values greater than 0.1 lM. Comparison of the IAPs with the solubility products (Ks) for the various solids indicates that porewaters were undersaturated (log IAP/Ks < 0) by 4–58 orders of magnitude with respect to U3O8(s), US(s), US2(s) and US3(s) and thus that these solids could not be formed in the sediments; in contrast, porewaters were supersaturated (log IAP/Ks > 0) with respect to UO2(s) and U3O7(s) (Fig. 1e–u) indicating that the formation of these solids was thermodynamically possible. However, the saturation state of porewater with respect to these solids does not vary in a consistent manner. For example, porewater remained supersaturated by 4–5 orders of magnitude with respect to UO2(s) in all lakes, including Lake Holland at the depth where U was consumed, an unexpected result if precipitation of this solid was occurring. Similarly, it is difficult to explain why U3O7(s) would precipitate in Lake Holland (log IAP/Ks P 8) and not in the other lakes where log IAP/Ks is also P8. These inconsistencies with a precipitation mechanism may indicate that these solids do not form or simply that the equilibrium constants for the formation of these solids are inaccurate. Uranium sequestration by adsorption onto sediment solid phases is another possible fixation mechanism that has been suggested for this element during early diagenesis (Anderson et al., 1989; Zheng et al., 2002; Morford et al., 2007). Adsorption has also been shown to be the first step in U removal by amorphous FeS(s) in the laboratory experiments (Hua and Deng, 2008). This process would be expected to occur at depths where U is removed from porewater and where porewater is saturated or oversaturat-

ed with respect to FeS(s); indeed, FeS(s) undersaturation should be associated with a release of U to porewater (i.e., U production zone) since FeS(s) would be dissolving. Comparison in Fig. 1 of panels y–d showing U production/consumption zones with panels e–u displaying porewater saturation index for disordered mackinawite (FeSm(s); Rickard, 2006), indicates that these conditions are fulfilled only in Lake Holland at depths between 2 and 0.5 cm (Fig. 1d and u) where a relatively rapid removal rate of U is observed. In contrast, results for Lake Vose and for Lake Tantare´ in July 2003 and August 2004 show U production zones at depths where porewater is saturated with respect to FeSm(s) and those for Lake Despe´riers and for Lake Tantare´ in September 2003 show a U consumption zone at depths where porewater is undersaturated with respect to FeSm(s). These inconsistencies suggest that U adsorption onto FeSm(s) is not an effective mechanism of U fixation in our lake sediments. However, such a conclusion is equivocal since there are uncertainties about (i) the forms of Fe sulfide present in our sediments, (ii) the equilibrium constant for the formation of the different forms of Fe sulfide, and (iii) the RU net values. Clearly, further field and experimental work is required to unequivocally identify the mechanism of U removal from porewaters in anoxic sediments. Study of several additional anoxic lakes where U is removed from porewater at a fast rate, such as in Lake Holland, would provide a valuable insight. Concomitant laboratory experiments also need to be achieved in order to determine more accurate formation constants for the U(IV) solids and to investigate U(IV) adsorption to various sediment phases (e.g., sulfides, sulfidized organic matter). 5. CONCLUSION Modeling porewater U profiles with a transport-reaction diagenetic equation indicates that authigenic U formation was negligible compared to U input with settling particles in the four studied lakes. In agreement with this contention, present-day fluxes of dissolved U by diffusion, bioturbation and bioirrigation are insignificant compared to the flux of U delivered with settling particles. Thus, U accumulation in our freshwater lakes is notably different from marine systems where U is, for the most part, delivered to the sediments by diffusion across the sediment–water interface and is thus mostly authigenic. Based on several observations we infer that U deposited to bottom sediments at our sampling sites was mostly associated with allochthonous organic matter and that this association was preserved during sediment burial. These observations are: the strong correlation between non-lithogenic U and Corg in the Appalachian Lake sediments; the similarity of the U:Corg ratio in the diagenetic material and in the surface sediments of Lake Tantare´, whose hypolimnion is perennially oxygenated; and the negligible impact of diagenesis on the sediment U profiles. Comparison of the U and Fe porewater profiles indicates that some U was bound to Fe oxyhydroxides in the three Canadian Shield lakes and was released to porewater upon reductive dissolution of these oxyhydroxides. Even if the proportion of total sediment U associated with Fe oxy-

Uranium in sediments

hydroxides was relatively small, the increased porewater U concentration upon dissolution of Fe oxyhydroxides was sufficient to generate an upward U flux across the sediment–water interface because the water column U concentrations were extremely low (<30 pM). In the marine environment, such an upward U flux generated by the reductive dissolution of Fe oxyhydroxides cannot be observed due to much higher U concentration in the water column (10 nM). In this regard, and in contrast to the Shield lakes, the Appalachian Lake Holland appears as an analogue of the marine environment since it exhibits a downward U flux across the sediment–water interface. This dissimilarity in U behavior is likely due to water column U concentrations almost one order of magnitude higher in L. Holland than in the other lakes. ACKNOWLEDGMENTS Financial support from the Natural Sciences and Engineering Research Council of Canada and the Fonds de Recherche sur la Nature et les Technologies du Que´bec are acknowledged. We thank L. Rancourt, R. Rodrigue and P. Fournier for their technical assistance and three anonymous reviewers and the Associate Editor for their critical comments on the manuscript. Permissions from the Que´bec Ministe`re de l’Environnement to work in the Tantare´ Ecological Reserve and from Faune et Parcs Que´bec to work in the Aiguebelle Provincial Park are gratefully acknowledged. REFERENCES Anderson R. F., Lehuray A. P., Fleisher M. Q. and Murray J. W. (1989) Uranium deposition in Saanich Inlet sediments, Vancouver Island. Geochim. Cosmochim. Acta 53, 2205–2213. Arnaboldi M. and Meyers P. A. (2007) Trace element indicators of increased primary production and decreased water-column ventilation during deposition of latest Pliocene sapropels at five locations across the Mediterranean Sea. Palaeogeogr. Palaeoclimatol. Palaeoecol. 249, 425–443. Berg P., Risgaard-Petersen N. and Rysgaard S. (1998) Interpretation of measured concentration profiles in sediment pore water. Limnol. Oceanogr. 43, 1500–1510. Berner R. A. (1980) Early Diagenesis. Princeton University Press. Bernhard G., Geipel G., Reich T., Brender V., Amayri S. and Nitsche H. (2001) Uranyl(VI) carbonate complex formation: validation of the Ca2UO2(CO3)3(aq.) species. Radiochim. Acta 89, 511–518. Boudreau B. P. (1984) On the equivalence of nonlocal and radial diffusion models for porewater irrigation. J. Mar. Res. 42, 731– 735. Boudreau B. P. (1997) Diagenetic Models and Their Implementation. Springer-Verlag, Berlin. Buffle J. (1988) Complexation Reactions in Aquatic Systems. Ellis Horwood Ltd., New York. Chappaz A., Gobeil C. and Tessier A. (2008a) Geochemical and anthropogenic enrichments of Mo in sediments from perennially oxic and seasonally anoxic lakes in Eastern Canada. Geochim. Cosmochim. Acta 72, 170–184. Chappaz A., Gobeil C. and Tessier A. (2008b) Sequestration mechanisms and anthropogenic inputs of rhenium in sediments from Eastern Canada lakes. Geochim. Cosmochim. Acta 72, 6027–6036. Davis J. A. (2001) Surface complexation modeling of uranium(VI) adsorption on natural mineral assemblages. Report NUREG/ CR-6708. U.S. Nuclear Regulatory Commission, Rockville, MD.

213

Dzombak D. and Morel F. M. M. (1990) Surface Complexation Modeling: Hydrous Ferric Oxide. Wiley. Fortin D., Leppard G. G. and Tessier A. (1993) Characteristics of lacustrine diagenetic iron oxyhydroxides. Geochim. Cosmochim. Acta 57, 4391–4404. Francßois R., Bacon M. P., Altabet M. A. and Labeyrie L. D. (1993) Glacial interglacial changes in sediment rain rate in the SW Indian sector of sub-antarctic waters as recorded by Th-230, Pa-231, U, and d15N. Paleoceanography 8, 611–629. Francßois R., Altabet M. A., Sigman D. M., Bacon M. P., Frank M., Bohrman G., Bareille G. and Labeyrie L. D. (1997) Contribution of Southern Ocean surface-water stratification to low atmospheric CO2 concentrations during the last glacial period. Nature 389, 929–935. Fredrickson J. K., Zachara J. M., Kennedy D. W., Duff M. C., Gorby Y. A., Li S. W. and Krupka K. M. (2000) Reduction of U(VI) in goethite (a-FeOOH) suspensions by a dissimilatory metal-reducing bacterium. Geochim. Cosmochim. Acta 64, 3085–3098. Gallon C., Tessier A., Gobeil C. and Alfaro-De La Torre M. C. (2004) Modeling diagenesis of lead in sediments of a Canadian Shield lake. Geochim. Cosmochim. Acta 68, 3531–3545. Gu B., Schmitt J., Chen Z., Liang L. and McCarthy J. F. (1994) Adsorption and desorption of natural organic matter on iron oxide: mechanisms and models. Environ. Sci. Technol. 28, 38–46. Guillaumont R., Fanghanel T., Neck V., Fuger J., Palmer D. A., Grenthe I. and Rand M. H. (2003) Update on the Chemical Thermodynamic of Uranium, Neptunium, Plutonium, Americium, and Technetium. Elsevier, Amsterdam. Hare L., Carignan R. and Huerta-Diaz M. A. (1994) A field study of metal toxicity and accumulation by benthic invertebrates; implications for the acid-volatile sulfide (AVS) model. Limnol. Oceanogr. 39, 1653–1668. Herbelin A. L. and Westall J. C. (1994) FITEQL: a computer program for the determination of chemical equilibrium constants from experimental data, Version 3.1. Report 94-01, Chemistry Department, Oregon State University, Corvallis, Oregon. Hua B. and Deng B. (2008) Reductive immobilization of uranium(VI) by amorphous iron sulfide. Environ. Sci. Technol. 42, 8703–8708. Klinkhammer G. P. and Palmer M. R. (1991) Uranium in the oceans: where it goes and why. Geochim. Cosmochim. Acta 55, 1799–1806. Ku T. L., Knauss K. G. and Mathieu G. (1977) Uranium in the open ocean: concentration and isotopic composition. Deep Sea Res. 24, 1005–1017. Lenhart J. J. and Honeyman B. D. (1999) Uranium(VI) sorption to hematite in the presence of humic acid. Geochim. Cosmochim. Acta 63, 2891–2901. Li Y.-H. and Gregory S. (1974) Diffusion of ions in sea water and in deep-sea sediments. Geochim. Cosmochim. Acta 38, 703–714. Lovely D. R., Phillips E. J. P. and Woodward J. C. (1991) Microbial reduction of uranium. Nature 350, 413–416. Mangini A., Jung M. and Laukenmann S. (2001) What do we learn from peaks of uranium and of manganese in deep sea sediments? Mar. Geol. 177, 63–78. Matisoff G. and Wang X. (1998) Solute transport in sediments by freshwater infaunal bioirrigators. Limnol. Oceanogr. 43, 1487– 1499. Matisoff G. and Wang X. (2000) Particle mixing by freshwater infaunal bioirrigators: midges (Chironomidae: Diptera) and mayflies (Ephemeridae: Ephemeroptera). J. Great Lakes Res. 26, 174–182. McLaren J. W., Methven B. A. J., Lam J. W. H. and Berman S. S. (1995) The use of inductively coupled plasma mass spectrom-

214

A. Chappaz et al. / Geochimica et Cosmochimica Acta 74 (2010) 203–214

etry in the production of environmental certified materials. Mikrochim. Acta 119, 287–295. McManus J., Berelson W. M., Klinkhammer G. P., Hammond D. E. and Holm C. (2005) Authigenic uranium: relationship to oxygen penetration depth and organic carbon rain. Geochim. Cosmochim. Acta 69, 95–108. McManus J., Berelson W. M., Severmann S., Poulson R. L., Hammond D. E., Klinkhammer G. P. and Holm C. (2006) Molybdenum and uranium geochemistry in continental margin sediments: paleoproxy potential. Geochim. Cosmochim. Acta 70, 4643–4662. Morford J. L., Russell A. D. and Emerson S. R. (2001) Trace metal evidence for changes in the redox environment associated with the transition from terrigenous clay to diatomaceous sediment, Saanich Inlet, BC. Mar. Geol. 174, 355–369. Morford J. L., Martin W. R., Kalnejais L. H., Francßois R., Bothner M. and Karle I.-M. (2007) Insight on geochemical cycling of U, Re and Mo from seasonal sampling in Boston Harbor, Massachusetts, USA. Geochim. Cosmochim. Acta 71, 895–917. Moyes L. N., Parkman R. H., Charnock J. M., Vaughan D. J., Livens F. R., Hughes C. R. and Braithwaite A. (2000) Uranium uptake from aqueous solution by interaction with goethite, lepidocrocite, muscovite and mackinawite: an X-ray absorption spectroscopy study. Environ. Sci. Technol. 34, 1062–1068. Murphy R. J., Lenhart J. J. and Honeyman B. D. (1999) The sorption of thorium (IV) and uranium (VI) to hematite in the presence of natural organic matter. Colloids Surf. A 157, 47–62. Nagao S., Yanase N., Yamamoto M., Kofuji H., Sohrin Y. and Amono H. (2002) The geochemistry of uranium in pore waters from lake sediments. J. Radioanal. Nucl. Chem. 252, 225– 232. Redfield A. C. (1934) On the proportions of organic derivatives in seawater and their relation to the composition of plankton. In James Johnson Memorial Volume (ed. R. J. Daniel). Liverpool University Press, Liverpool. Rickard D. (2006) The solubility of FeS. Geochim. Cosmochim. Acta 70, 5779–5789. Sani R. K., Peyton B. M., Amonette J. E. and Geesey G. G. (2004) Reduction of uranium (VI) under sulfate-reducing conditions in

the presence of Fe(III)-(hydr)oxides. Geochim. Cosmochim. Acta 68, 2639–2648. Schecher M. E. and McAvoy D. (1998) MINEQL+. Environmental Research Software. Schwarzenbach R. P., Gschwend P. M. and Imboden D. M. (1993) Environmental Organic Chemistry. Wiley-Interscience, New York. Scott T. B., Riba Tort O. and Allen G. C. (2007) Aqueous uptake of uranium onto pyrite surfaces; reactivity of fresh versus weathered material. Geochim. Cosmochim. Acta 71, 5044–5053. Tipping E. (1981) The adsorption of aquatic humic substances by iron oxides. Geochim. Cosmochim. Acta 45, 191–199. Tipping E., Woof C. and Cooke D. (1981) Iron oxide from a seasonally anoxic lake. Geochim. Cosmochim. Acta 45, 1411– 1419. Tipping E. (2002) Cation Binding by Humic Substances. Cambridge University Press. Tribovillard N., Algeo T. J., Lyons T. and Riboulleau A. (2006) Trace metals as paleoredox and paleoproductivity proxies: an update. Chem. Geol. 232, 12–32. Wagman D. D., Evans W. H., Parker V. B., Schumm R. H., Halow I., Bailey S. M., Churney K. L. and Nuttall R. L. (1982) The NBS tables of thermochemical properties. J. Phys. Chem. Ref. Data 11 (Suppl. 2. Am. Inst. Phys., Melville, NY). Waite T. D., Davis J. A., Payne T. E., Waychunas G. A. and Xu N. (1994) Uranium(VI) adsorption to ferrihydrite: application of a surface complexation model. Geochim. Cosmochim. Acta 58, 5465–5478. Wersin P., Hochella M. F., Persson P., Redden G., Leckie J. O. and Harris D. W. (1994) Interaction between aqueous uranium (VI) and sulfide minerals: spectroscopic evidence for sorption and reduction. Geochim. Cosmochim. Acta 58, 2829–2843. Zheng Y., Anderson R. F., van Geen A. and Fleisher M. Q. (2002) Remobilization of authigenic uranium in marine sediments by bioturbation. Geochim. Cosmochim. Acta 66, 1759– 1772. Associate editor: David J. Burdige