Redox potentials and redox reactions in deep ... - Science Direct

ChemicalGeology, 98 (1992) 131-150 Elsevier Science Publishers B.V., Amsterdam

131

[3]

Redox potentials and redox reactions in deep groundwater systems I. G r e n t h e a, W. S t u m m a, M. Laaksuharju b, A.-C. Nilsson b and P. Wikberg b =EAWAG, CH-8006 Diibendorf, Switzerland bDepartment of Inorganic Chemistry, The Royal Institute of Technology, S-I0044 Stockholm, Sweden.

(Received June 19, 1990;revised and accepted March l, 1991 )

ABSTRACT Grenthe, I., Stumm, W., Laaksuharju, M. Nilsson, A.-C. and Wikberg, P., 1992. Redox potentials and redox reactions in deep groundwater systems. Chem. Geol., 98:13 l- 150. Both laboratory investigations and field studies of deep groundwater systems indicate that stable and reproducible redox potentials can be measured with a precision of +_25 mV. The Eh data measured in the field are in good agreement with the half-cell reaction: "Fe (OH)3"(s) + 3H + +e- ~Fe 2+ + 3H20 involving hydrous ferric oxide and Fe2+ in solution, as indicated both by the slope of a plot of Eh vs. (3pH + log [ Fe2+ ] ), and the value of the solubility product for "Fe (OH)3"is), calculated from these data. The solubility product: pK~ = - log[ Fe3+ ] [OH- ]3 =40.9 _+1.1

falls in the range 37.3 < pKs< 44. l, as previously given by D. Langmuir and D.O. Whittemore for amorphous ferric hydroxide and crystalline goethite, respectively. The measurement and interpretation of field redox data from borehole investigations are complicated by the mixing of waters of different origin. Most groundwaters studied are not characteristic of the geochemistry of the particular section sampled; they are mixtures of water of different origin and may contain significant amounts of surface water (indicated by the tritium content) and drilling water used to cool the drill-bit. The initial oxygen content in these components is rapidly reduced by Fe (II) minerals with the formation of Fe (OH)3,), resulting in mixed waters which are anoxic and containing dissolved Fe (II). Surface-mediated redox reactions seem to play an important role both in the reduction of oxygen and of uranium and other trace elements.

1. Introduction T h e r e d o x p o t e n t i a l a n d the o x i d a t i v e ( O X C ) a n d r e d u c t i v e c a p a c i t i e s ( R D C ) (cf. Scott a n d M o r g a n , 1 9 9 0 ) o f g r o u n d w a t e r are i m p o r t a n t p a r a m e t e r s for the safety assessm e n t o f n u c l e a r w a s t e r e p o s i t o r i e s b e c a u s e the s p e c i a t i o n , solubility a n d the s o r p t i o n c h a r a c teristics o f a c t i n i d e s a n d r e d o x - s e n s i t i v e fis-

Correspondence to: I. Grenthe, Department of Inorganic Chemistry, The Royal Institute of Technology, S-10044 Stockholm, Sweden.

sion p r o d u c t s are d e p e n d e n t o n these q u a n tities. The redox potential provides i n f o r m a t i o n o n the d r i v i n g force for r e d o x reactions, while the O X C a n d R D C p r o v i d e a m e a s u r e o f the r e d o x c a p a c i t y o f the system, w h i c h is i m p o r t a n t b o t h for the possibility o f m e a s u r i n g r e d o x p o t e n t i a l s a n d for d e t e r m i n ing the extent to w h i c h r e d o x processes can take place in the system. I f we take as a reference p o i n t the r e d o x int e n s i t y w h i c h separates oxic a n d a n o x i c envir o n m e n t s (i.e. P o 2 < l 0 -6 a t m ) , O 2 - M n O E Fe ( I I I ) - t y p e m i n e r a l s m a k e u p m o s t o f the ox-

0009-2541/92/$05.00 © 1992 Elsevier Science Publishers B.V. All rights reserved.

132

i. GRENTHE ET AL.

idation capacity, whereas organic material, Fe (II) and S ( - I, - II) components constitute most of the reduction capacity. The OXC and RDC are dependent on the redox reactions that may take place in the system. In the present case the O X C / R D C of the bedrock is determined both by heterogeneous electron transfer processes involving pyrite, Fe(II) and Fe(III) in oxide and silicate minerals (cf. White, 1990), and weathering which results in a release of Fe(II) and M n ( I I ) from bed-rock minerals to the water. The bedrock and the aqueous phases form two different compartments in the groundwater system, where the O X C / R D C in the latter is in general more rapidly accessible for redox reactions than the O X C / R D C in the bedrock. Fracture-filling minerals have an intermediate role, they are often very efficient and rapid electron transfer agents (cf. White, 1990 and references therein). The redox processes taking place at the interface between solid and solute are dependent both on sorption of the redox-active solutes and on the available wetted surface. Of particular importance for the discussion of the redox reactions in groundwater systems are the coatings of hematite, goethite or "rust" sometimes encountered in the fractures even at large depth and otherwise strongly reducing conditions. An important example of a rustformation process is the reduction of oxygen, as exemplified by the oxidation/weathering of biotite: KMgFe2AISi3 O1o (OH)2 + ½02 + 8H20 + 3H+~AI(OH)3 +2Fe(OH)3 +K

+ +Mg

2+ +

3Si(OH)4

Hematite is common in fractured rocks that have suffered hydrothermal alteration during oxidizing conditions. Hematised fractures that are water-conducting, often carry Fe-oxyhydroxides (possibly due to dissolution/precipitation, cf. Bruno et al., 1992a). Such hematite/Fe-oxyhydroxide bearing zones can be

found at great depth (at least 1000 m below surface) and do not indicate oxidizing conditions at present. However, Fe-oxyhydroxides are also present in fractures in the near-surface sections of the bedrock, probably as a result of an ongoing percolation of oxygen containing surface water (cf. Tullborg, 1989 ). The focus of this communication is on heterogeneous electron transfer reactions in deep groundwater systems. The specific issues discussed are: ( 1 ) redox potential measurements, and the problems of measuring and interpreting such data; (2) the reduction of dissolved oxygen and metal ions, mainly actinides, through electron transfer from Fe(II) minerals; (3) the reduction capacity of deep groundwater systems and the redox evolution of infiltrating surface water. The discussion is based on experimental data from the Fennoscandian Shield obtained from the hydrogeological investigations made by the Swedish Nuclear Fuel and Waste Management Co. (SKB) as one part of an effort to develop a repository for spent nuclear fuel in deep bedrock formations. The experimental data in this study have been obtained from a total of 50 boreholes, located at eight different sites in Sweden. The groundwater characteristics have usually been studied in three to six water-carrying sections with a hydraulic conductivity of 10-9 to 10-6 m s- ~, from the surface to several hundred meters depth. Field procedures and local hydrology are important for the correct collection and interpretation of the experimental data. Variations in hydraulic conductivity in combination with topography and meteorological conditions will give rise to different hydraulic heads within the bedrock. A borehole which passes through sections with different hydraulic heads may shortcircuit the groundwater through the borehole, with mixing of different waters as a result. The pressure gradient in the undisturbed rock is also small compared to the gradient introduced by the pumping. Hence, the pumping

REDOX POTENTIALS AND REDOX REACTIONS IN DEEP GROUNDWATER SYSTEMS

of water from a specific section, which is sealed off from the rest of the borehole, might cause inflow and mixing of waters from other watercarrying sections with different chemical characteristics. The water flow in most deep groundwater systems is small, hence the water used to cool the drill-bit is a possible source of contamination. The experimental program has been going on during a ten-year period, hence the procedures used in the field investigations have been successively improved. A summary of the investigations and the geology and mineralogy of the various test sites have been given by Smellie et al. ( 1985, 1987 ). All experimental data are accessible through the GEOTAB database (cf. the Appendix). A general conclusion from these data is that the main redox-active minerals in the bulk bedrock are "gray" Fe ( I I ) / F e (III)-silicates, such as chlorite, hornblende and biotite, indicating that weathering reactions should result in anoxic deep groundwaters, which contain dissolved Fe (II). The field data also show that these waters contain < 0.1 mg l- ~02, and that they are strongly reducing (as indicated by redox potentials in the range - 0 . 4 to - 0 . 1 V), with RDC >> OXC, unless there has been a recent intrusion of oxygen through mixing with surface water. Evidence for reducing conditions is also obtained from the concentrations and isotope disequilibria of uranium, which indicate the presence of U (IV).

1.1. Redox potential measurements in laboratory and field systems Redox potential measurements may provide a direct measure of the redox intensity of pertinent redox components, and together with an assessment of the total concentrations of the redox components, also of the O X C / R D C . However, there are several difficulties involved, both experimental and conceptual;

133

some aspects have been discussed by Lindberg and Runnels (1984). Eh may be measured experimentally and/or calculated from chemical analysis of the various redox couples present. Stable electrode potentials are in general obtained if: (1) the system contains electrode-reactive species, that can exchange electrons with the measuring electrode; (2) the redox reactions involve one-electron transfers such as between Fe (II) and Fe (III); (3) the system has a sufficiently large redox capacity. The latter does not necessarily require high concentrations of the redox-active species in solution, as demonstrated by systems where the electron exchange reactions with the electrode involve one or more solid phases. However, disturbances in the form of irreversible reactions at the electrode, and "mixed" potentials are more important at lower concentrations of redox-active species. Many of the redox couples present in surface and deep groundwater systems involve multielectron transfers, often with structural reorganization between the red and ox forms, and such reactions are kinetically slow (Katakis and Gordon, 1987). Hence, it is not surprising that there are disequilibria among some redox couples in aquatic systems in nature, and that Eh-values computed from analytical data on the various redox couples do not agree with one another (Lindberg and Runnels, 1984). Of particular importance for the groundwater systems investigated herein, is the evidence that the Fe (II)/Fe (III) system gives stable and reproducible redox potentials also under the conditions encountered in nature. Evidence for this has also been obtained from laboratory investigations, e.g. Berner (1963), Doyle (1968), Macalady et al. (1990), and references therein. Nordstrom et al. (1979 ) studied the redox equilibria in acid mine drainage and found an excellent fit between measured Eh-values and those calculated from the analytical F e ( I I ) / Fe (III) ratio in solution.

134

Boulegue and Michard (1979) studied the redox conditions in sulphur-rich brines and found that the measured redox potentials were controlled by polysulphide species. Stumm and Morgan (1981, p.490), and Morris and S t u m m ( 1967 ) have discussed the Eh concept and the reliability of experimental determinations. Whitfield ( 1974 ) studied the behavior of Pt electrodes theoretically and concluded that they act as inert electrodes only within a rather narrow Eh interval. Grundl and Macalady (1989) have studied the performance of Pt and wax-impregnated graphite electrodes in anaerobic ferric/ferrous chloride systems. Macalady et al. (1990) have also discussed F e ( I I ) / F e ( I I I ) redox data in anaerobic systems; of particular relevance for the present study is their conclusion that: "lab and field Eh measurements using Pt or wax-impregnated-graphite electrodes can provide Nernstian potentialsin the presenceof measurableFe (II) at pH's as high as 6.6." This literature survey indicates that measurement of redox potentials in the laboratory and in the field, and the interpretation of these data in terms of chemical reactions are possible in systems where the redox properties, as measured by the electrode, are determined by the F e ( I I ) / F e ( I I I ) system. However, such studies require care and control of the experimental conditions; the kinetics of the various processes in the system play an important role. There are two types of kinetic constraints which affect Eh-measurements. One is a slow redox equilibrium in the potential determining reaction, e.g. between SO42- ~-HS-, resulting in such low exchange currents, that stable potentials are not attained. The other are disequilibria in the bulk phase caused by the mixing of waters of different redox status, e.g. oxic and anoxic. The latter type of process is very important in field studies and may explain part of the experimental difficulty in obtaining reproducible data.

1. G R E N T H E ET AL.

Field redox potentials are expected to reflect information mainly on the part of the Fe ( I I ) / Fe (III) redox system which is rapidly accessible, i.e. the parts present in the solution and in the fracture minerals.

2. Experimental 2.1. Laboratory investigations In order to ascertain if it is possible to obtain reproducible redox data, which can be related to known chemical processes, we performed a set of laboratory investigations before undertaking field studies. From the laboratory studies we were able to identify some of the factors that might disturb redox measurements in the field. In the course of this work we also improved existing sampling and analytical procedures. These technical improvements made it possible to measure redox potentials, that usually are reproducible within + 20 mV or better, i.e. the same precision as in the laboratory studies. This is sufficient both for the assignment of potential-determining reactions and for determination of the redox speciation of actinides and other redox-sensitive trace elements. Two sets of laboratory measurements were made: the first by measuring the redox potential in a system containing solid Fe (III)-oxyhydrate and Fe (II)-carbonate in equilibrium with an aqueous phase of known pH and carbonate concentration; the second by potential measurements in a simulated field system obtained by circulating water through a column of coarse-crushed rock.

2. I. 1. Electrode system and measurement procedures. The electrode system was the same as in a modified form used in the field measurements. The electrode set contained three inert electrodes (Au, Pt and glassy carbon) for

REDOX POTENTIALSAND REDOX REACTIONSIN DEEPGROUNDWATERSYSTEMS TABLE 1 pH-, Eh- and pS-values of the calibration solution at 10 °C Calibration solution

Buffer solution + quinhydrone Buffer solution + quinhydrone Buffer solution 0.05 M Na2S in 0.1 M NaOH 0.01 MNa2S in 0.1 M H C O 3 - - C O ~ 2 - buffer

pH

Eh (mV)

4.0

+487

7.0

+316

pS

10.0 (13.3)

2.0

(10.5)

5.5

redox potential measurements, a Ag/Ag2S electrode for sulphide measurements (this was only used in the field measurements up to 1986), a glass electrode with a pressure compensation device for pH measurements, and a gel-filled triple junction Ag/AgC1 electrode as a common reference. We used different materials in order to establish whether or not the potentials are dependent on the electrode material. Such a dependence makes it less likely that Eh can be related to one particular redox reaction. Each electrode was connected to a separate amplifier. These were connected to a multiplexer, which in the field equipment was operated from a computer on the surface. The multiplexer connected one electrode at the time to an analog/digital converter. The digital word, converted to serial form, was transmitted to the surface computer as a frequencyshifted signal. The field equipment also contained a thermistor for temperature measurements and a conductivity cell. The equipment has been described in more detail by Alme'n et al. (1986) and by Wikberg (1987). The long-time stability of the sulphide electrode was not satisfactory and it was only used in the early phase of the site investigations. The sulphide concentration was also determined by chemical analysis (the methylene blue method; SIS, 1976a, detection limit 0.01 mg 1-i ). In laboratory studies we also tested a porous carbon electrode for redox potential measure-

135

merits (see p. 136). The electrodes were standardized using buffers as shown in Table 1.

2.1.2. Function test using the hydrous Fe(III)oxide/Fe(I1)-carbonate system. This study was made to establish whether the measured redox potential in a model system of high pH agreed with the value calculated from known thermodynamic data for the two solids, or not. This part of the experiment thus provides an extension of the studies of Doyle (1968) and of Macalady et al. (1990), both made at pH 0. The COz formed there during the oxidation of the dissolved organic material will participate in weathering reactions, where alteration of plagioclase, K-feldspar, biotite and muscovite results in the release of Fe (II) to the solution. The main result of redox and weathering reactions along the transport path of the groundwater is then a redistribution of RDCfrom the bedrock minerals to RDC in the aqueous phase and the fracture minerals. Some support for a decrease of DOC with depth, presumably a result of an oxidation, is given by the field data (Fig. 7 ).

3.5. On the occurrence ofhematite and hydrous iron oxides in fractures at large depth

The presence of hematite has been discussed by Tullborg ( 1989 ), and is probably a result of very old hydrothermal events, while the formation of Fe-oxyhydroxides may be a result of more recent low-temperature alterations, possibly a precipitation caused by a decrease in solubility of Fe (III) at the high pH-values and low carbonate concentrations encountered at larger depth. Support for this view is obtained from a recent study of the solubility of hematite in carbonate media by Bruno et al. (1992a). The equilibrium data reported by these authors indicate that the solubility of Fe (III)-oxide/hydroxide decreases by a factor of between 10 and 100 with a decrease of the partial pressure of CO2 by a factor of 10. A decrease of the partial pressure of CO2 takes place with increasing depth when the groundwater system transforms from open to closed conditions with respect to CO2. This chemical gradient is sufficient for the precipitation of Fe (III) if there is flow from the upper levels to the depth.

REDOX POTENTIALS AND REDOX REACTIONS IN DEEP GROUNDWATER SYSTEMS

145

12 (a)

:" 6

(b)

10 8

I

:.

i

•; "b'"

x~ 6 E

2 J

t

0 Et (rag/t}

Ix 1000 )

12E (YEARS)

(xlOO0)

Fig. 7. a. Statistics of DOC vs. salinity. The salinity is correlated both with depth and the "age" of the water. b. Correlation between the amounts and age of TOC.

3.6. Electron transfer processes at mineral surfaces Electron transfer processes involving mineral surfaces are of key importance for redox and sorption reactions in most ground- and surface-water systems. As we have indicated in the previous text such reactions may also be decisive for the electron exchange reactions at inert electrodes and the possibility of measuring redox potentials. General surveys of this topic have been given by Hering and Stumm (1990) and Wehrli (1990). White has reviewed the redox reactions of Fe ( I I ) / F e (III)oxide and -silicates (1990). Tronc et al. (1984) and Jolivet et al. (1988, 1990) have studied redox reactions involving magnetite and hematite. The electron transfer reactions between magnetite and sorbed species are dependent on the surface area and are very fast for colloidal systems, as exemplified by the reduction of Ag+, and the "refilling" of electrons to the Fe203 thus formed through sorption of Fe 2+ (Jolivet et al., 1990). Jolivet et al. have properly described the minerals participating in these reactions as "electron tanks". The fast reactions may be related to the rapid electron exchange between Fe (II) and Fe (III) with a half-life of 10 -1° s in magnetite (Allen et al., 1982). Examples of redox reactions coupled to surface sorption are abundant (Hering and

Stumm, 1990; Wehrli, 1990; White, 1990). White and co-workers have made extensive studies of the coupling between electron transfer reactions and sorption/weathering reactions (e.g., White, 1990). White has also discussed the surface redox potential of Fe ( I I ) / Fe (III), i.e. reactions of the type: Fe (III)-silicate