Using Mercury Volatltity To Measure Redox Potential in Oxic Aqueous

02-H20 redox couple in oxic aqueous systems. The redox potential of this couple was evaluated indirectly by cal- culating the redox status of mercury ...
1 downloads 0 Views 504KB Size
Environ. Sci. Technol. 1989, 23, 828-831

Using Mercury Volatltity To Measure Redox Potential in Oxic Aqueous Systems James J. Bisognl, Jr.

Department of Environmental Engineering, Cornell University, Ithaca, New York 14853 An experimental laboratory study was conducted to determine the effective redox potential exerted by the 02-H20 redox couple in oxic aqueous systems. The redox potential of this couple was evaluated indirectly by calculating the redox status of mercury via volatility measurements. The results of this study show that the redox potential exerted by the 02-H20 redox couple is -0.4 V below that predicted by the Nernst equation. 1. Introduction

Determination of the speciation of metals in aquatic environments is often dependent on redox conditions (redox potential) of these aquatic environments. Measurement or prediction of redox potential in aquatic environments, particularly oxic environments, is a very difficult task. Some form of the 02-H20 redox couple usually poises the redox potential of oxic aqueous systems. However, the actual redox potential that this couple exerts is obscure. There have been various attempts to quantify the 02-H20 redox couple by techniques such as measurement with an inert "redox" electrode, thermodynamic calculation, and measurement by equilibration with known redox couples. This last technique can provide an accurate estimate of redox potential if an electroactive redox couple is used and if the relative concentrations of the redox species can be determined without disturbing the existing equilibrium of the system. The mercuric/metallic mercury redox couple satisfies these criteria because it is electroactive and it has a unique volatility characteristic that can be used to "measure" the relative concentration of mercuric and metallic mercury without disturbing redox conditions in the system. Hence, this mercury couple was employed in an experimental study to assess the redox potential of an oxic aqueous system. 1.1. Redox Control in Oxic Aqueous Systems. For oxic aqueous systems that contain low levels of electroactive species, the 02-H20 redox couple can be considered the dominant or poising redox reaction (1). This redox couple is represented in eq 1. The redox potential, O2 4H+ 4e- G 2H20; E H o = 1.229 V (1)

+

+

established by the redox reaction shown in eq 1is given by the Nernst equation, shown as eq 2, where R is the RT EH= 1.229 + - In ([H+]Po,)1/4 F universal gas constant, F the Faraday constant, T the absolute temperature, [H+] the molar H+ concentration, and Po, the partial pressure of oxygen (atm). For oxic water eq 2 predicts that pH is the primary factor controlling redox. In fact, theoretical prediction of redox potential for oxic waters at 30 "c can be approximated by eq 3. Over the pH range of 4-10 redox potential would then be expected to vary from -0.99 to 0.64 V. E H = 1.229 - 0.059pH (3) Unfortunately, quantitative evaluation of the redox potential of aquatic systems dominated by the 02-H20 couple is extremely difficult. Vetter (2) describes how the

EH,

828

Environ. Sci. Technol., Vol. 23, No. 7, 1989

full redox potential of the 02-H20 couple is seldom measured when using an inert platinum electrode system. Whitfield (3)and Stumm (4) discuss in detail the problems associated with measurement of redox potential using the "standard" inert platinum electrode. They conclude that this electrode system is responsive to an oxide or sulfide coating on the platinum electrode rather than the 02-H20 couple. Indeed, the actual oxidizing strength of the 02-H20 couple may be lower than predicted by thermodynamics. Garrels and Christ (5) state that aqueous systems exposed to air tend to exhibit electrode potentials -0.5 V below the theoretical potential. Breck (6) has proposed that the E H of sea water is controlled by the second of a two-step wateroxygen redox couple which has a standard potential of 0.68 V. Nordstrom et al. (7) show that the 02-H20 couple is not in equilibrium with measured E H or E H predicted from the Nernst equation using the ferrous/ ferric ion ratio measured in an acid stream. Instead, they found that the effective redox potential was lower then that theoretically predicted. In general, the redox potential of oxic aqueous systems is difficult to measure and is not readily predicted by theoretical Nernstian response. Of course, even if the redox potential of a system can be determined, there is no assurance that the pertinent redox couple will be in equilibrium with this potential (8). On the other hand, measurement of E H and theoretical correlation of this E H with redox couples under anaerobic conditions has been successful (9). 1.2. Redox of the Mercury Couple. The redox reactions of inorganic mercury are known to be kinetically rapid (10,ll). The redox couples of inorganic mercury are, therefore, generally electroactive and will respond quickly to ambient redox conditions. They are accurate indicators of redox conditions. Rapid reaction kinetics also allow accurate use of equilibrium calculations to predict redox reactions. Elementary mercury, HgO, exhibits the lowest oxidation state (0) of mercury. Redox equilibria between elemental Hg and mercuric mercury Hg2+are given by eq 4. Hg2++ 2e-

Hgo; E H o = 0.659 V

(4)

An intermediate oxidation state of mercury exists in the mercurous(1) form. However, the mercurous oxidation state is usually dominated by elemental and mercuric mercury, although it may play a part as an intermediate in the oxidation or reduction of the elemental and mercuric forms (12). The degree of oxidation or reduction shown in eq 4 is strongly influenced by the presence of mercury complexing ligands. In the I1 oxidation state mercury can complex with a variety of ligands including hydroxide, chloride, sulfide, and organic matter (13).Therefore, the fraction of total aqueous mercury that is in any particular oxidation state is a function of concentration and type of complexing ligands as well as the redox potential. Chemical equilibrium models such as MINEQL (14) can be used to predict speciation, once pH, EH,and the ligand stability constants have been quantified.

0013-936X/89/0923-0828$01.50/0

0 1989 American Chemical Society

1.3. Volatility of Inorganic Mercury. Volatility can be used to estimate the concentration of certain aqueous mercury species. The dimensionless Henry's constant, H, can be used to assess this volatility. For example, if the volatile species is HgO, H is defined as

H

= [Hg(,,Oj/ [Hgbqfl

gas supply

rotameter

(5)

where [Hg( ,O] is the gas-phase elemental mercury concentration f m ~ l - c m -and ~ ) [Hg(aqf]is the aqueous-phase elemental mercury concentration (m~ls-cm-~). Iverfeldt and Lindquist (15) report H for HgC1, and Hg(OH), equal to 2.9 X and 3.2 X lo*, respectively at 25 "C. Sanemasa (16)reports an H of 0.32 for elemental mercury at the same temperature. Based on these values of H, Hgo has the highest potential to volatilize from the aqueous phase. It follows that the most important factor in determining volatility of inorganic mercury from aqueous phases is the fraction of the total aqueous mercury that is in the form of elemental mercury. This fraction is defined as P . The rate of volatilization of compounds from the aqueous phase has commonly been described by diffusion film models (17,18).When applied to elemental mercury the film model yields eq 6 where F is the flux rate of

2. Research Plan The objective of the experimental research reported here is to evaluate the effective redox potential exhibited by oxic aqueous systems where the 02-H20 redox couple dominates. This evaluation was done indirectly by estimating the aqueous elemental mercury concentration by measuring volatilization rates of mercury. Using this estimate and known concentrations of ligands, equilibrium calculations were performed to estimate the ratio of mercury(0) and mercury(I1) oxidation states. Redox potential was then estimated from this ratio. 3. Experimental Methods 3.1. Experimental Apparatus. A program of laboratory experimentation was designed to quantify the volatilization of inorganic mercury under a variety of redox and speciation conditions. Figure 1 shows the experimental apparatus used to measure mercury volatilization. A steady flow of gas (air, nitrogen, or oxygen) was passed through a Pyrex reactor a t a flow rate of 12 L/h. Initial reactor head space was 0.72 L and reactor liquid volume was 4.0 L. Sampling of the reactors caused these volumes to change slightly during the course of an experiment. Before entering the reactor system, all gases were passed through special mercury-adsorbing activated carbon (Coleman Instruments, Maywood, IL). In some experiments the mercury-free gas was discharged through a submerged diffuser, in others (experiments 5 and 6) gas was discharged into the head space above the reactor liquid phase. Mixing of the reactors was accomplished for all but two reactors (experiments 5 and 6 ) by the gas flowing

w

reactor vessel

redoxJpH

moniionng vessel

Flguro 1. Mercury volatilization apparatus.

Table I. Experimental Volatilization Reactors and Conditions

expt

PH

Cl-, pM

TOC

1 2 3 4 5

4.3

0.5 0.5 0.5 0.5 0.5 0.5 0.5 1000

30 30 30 30 30 30 20 30 30 30 30

6

mercury across air-water interface (m~l.cm-~.h-l) and K is the overall mass transfer coefficient (h-l). Henry's constant of H$ is much greater than the Henry's constant of other common inorganic mercury species, such as HgCl, or Hg(OH),. Unless these latter species have solution concentrations on the order lo7 and lo5, respectively, greater than the solution concentration of Hgo, their rate of volatilization will be insignificant.

m

effluent gas

7

a 9

10 11

6.9

6.9 7.1 7.2 7.2 7.1 7.1 5.1 7.3 10.3

0 0 0

gas air air 0 2

N2

air air air air air air air

mixing mode diffuser diffuser diffuser diffuser surffstir surface diffuser diffuser diffuser diffuser diffuser

through the submerged diffuser. In experiment 5 mixing was enhanced with a magnetic stirrer. Mixing in experiment 6 was solely due to gas flowing over the surface of the liquid. The reactor effluent gas was passed through two mercury absorption traps. The purpose of the traps was to provide a means to account for all volatilized mercury. Each trap contained 300 mL of a solution comprised of 1%(w/v) KMnO, and 10% (v/v) concentrated H2S04. Gas flow rates were controlled by a rotameter. Continuous monitoring of reactor pH and electrode potential was accomplished by recycling a portion of the reactor contents through a sealed monitoring vessel which housed a glass electrode (for pH) and a platinum electrode (for electrode potential). The vessel also contained a double-junction reference electrode, which was used for both pH and electrode measurements, although these electrode potentials are not reported in this paper. A peristaltic pump was used to recycle the reactor contents at a flow rate of 25 mL/min. Table I is a list of the experimental conditions under which these reactors were operated. Experimental conditions were selected to attain a large range in redox potential and mercury speciation. Selection of specific experimental ligands has no effect on general applicability of results. The ligands serve only as a concentration buffer for Hg(I1). However, it is important that the value of the stability constants for the selected ligands be known accurately. Hence, chloride and hydroxide were selected as the major mercury ligands in these experiments. Initial mercury concentration was adjusted to 0.25 pM by adding HgClz or Hg(N03),. This mercury concentration was chosen to be below the solubility limit of aqueous elemental mercury (19). Redox conditions were adjusted by varying the pH (by addition of NaOH or HN03) and the gas type (air, nitrogen, or oxygen). Environ. Sci. Technol., Vol. 23, No. 7. 1989

829

It was important to verify that the rate of volatilization of mercury was limited by the mass transfer of mercury rather than by the rate of oxidation or reduction of mercury (this is an inherent assumption of the reactor analysis described below). To make this verification, K was intentionally varied in experiments 5-7 by varying agitation (mixing levels) and temperature. 3.2. Sampling and Analysis. The reactors were analyzed for total mercury concentration twice a day. During reactor sampling the gas inflow and outflow lines were clamped and the samples were removed with volumetric pipets and transferred to 300-mL bottles containing 1mL of 5% (w/v) KMnOl in 250 mL of deionized water. The effluent gas traps were analyzed for mercury and recharged with fresh acid permanganate solution each time the reactors were sampled. All mercury samples were analyzed in less than 24 h from the time of sampling. pH and electrode potential were measured continuously. 3.3. Analytical Techniques. Determination of pH was made by the glass electrode method. Electrode potential measurements were made with a platinum electrode and an Orion 90-02 double-junction reference electrode. Total mercury analysis was performed by a cold vapor (flameless) atomic absorption method (20). 3.4. Reactor Analysis. A mass balance on mercury about the experimental reactors can be obtained if it is assumed that the reactor gas and liquid phases are completely mixed and that the transfer of mercury out of the aqueous phase is limited by mass transfer and not by redox reactions of mercury. It is also assumed that elemental mercury is the only species that exhibits significantly volatility. The extent of mercury adsorption onto the reactor walls and tubing was determined in a separate set of experiments. These experiments employed reactors of the same type, including the solution composition, as used in the volatility experiments. The reactors were sealed with rubber stoppers and the liquid phase was periodically analyzed for total mercury. Since there was no gas flow through these reactors it was assumed that during the short time period that the reactors were open for sampling very little mercury escaped via volatilization. The results of the adsorption experiments showed that adsorption of mercury was first order with respect to total aqueous mercury. The first-order adsorption coefficient, kah, was found to vary with pH. These values for kads were found to be 0.0092,0.0065 and 0.0039 h-l for pH = 4.3, 7.0, and 10.3, respectively. Mass balance of mercury about the liquid phase is given by

where [Hg,] is the total aqueous mercury concentration (m~ls.cm-~). Mass balance of mercury about the reactor gas phase, assuming no mercury in the inflow, is given by

where V , is the volume of reactor liquid phase (cm3), V , the volume of reactor gas head space (cm3), and Q the volumetric gas flow rate (cm3.h-'). Equations 7 and 8 are a set of first-order, homogeneous, linear differential equations, with constant coefficients. All of these coefficients, except P and K , can be directly 830

Environ. Sci. Technol., Vol. 23, No. 7, 1989

Table 11. Experimental Results: Volatilization and Speciation expt

P

K , h-l

MSE

1 2 3 4 5 6 7 8 9 10 11

0.44 0.57 0.51 0.58 0.40 0.42 0.30 0.35 0.30 0.54 0.66

0.15 0.15 0.14 0.15 0.09 0.03 0.07 0.13 0.15 0.15 0.15

18.3 42.6 81.0 33.0 76.0 52.0 4.8 70.8 32.6 30.5 80.1

measured. Aqueous-phase elemental mercury concentration can be solved as a function of time only if all the coefficients are known. However, it is possible to estimate P and K by solving the mass balance equations with trial values of these parameters and by comparing the predicted and measured [Hg,] values. This procedure can be repeated untiI P and K values are found that give the best statistical fit of the experimental data. The results of this procedure are given below. 4 . Results 4.1. Mercury Volatility and Speciation. A com-

puter-based trial and error procedure was used to find the combination of P and K values that yielded the lowest mean squared error (MSE) between the observed and predicted [Hg,] values. The results of this procedure are shown in Table 11. Mercury recovery in the gas effluent traps was about 65-85% for all the reactors. Loss of mercury was probably due to sorption onto tubing between reactors and traps as well as trapping inefficiencies. From the data in Table I1 it can be seen that the reduced mixing level of experiments 5 and 6 resulted in an expected lowering of the overall mass transfer coefficient for these experiments. In experiment 7,the overall mass transfer coefficient was also reduced due to the lower temperature (20 "C) of this experiment. The results of experiments 5 and 6 strongly suggest that mass transfer was the limiting step in the volatilization of mercury, since nothing was done in these experiments to affect the rate of oxidation or reduction of mercury. The change in temperature in experiment 7 could have resulted in a decrease in both mass transfer and redox kinetics so no conclusions can be drawn from this experiment. Given the results of experiments 5 and 6 and the small temperature change (10 "C), it is likely, however, that mass transfer is also limiting in experiment 5. The remaining experiments yielded overall transfer coefficients, K , that averaged 0.146 h-' with a standard deviation of 0.007 h-l. This small standard deviation suggests that the mixing regime and gas-transfer conditions were practically the same for these experiments. Since K was statistically the same for all experiments (except experiments 5-7), any differences in mercury volatilization rate must be attributed to differences in elemental mercury concentration in the reactor. Initial total mercury in each reactor was the same; therefore, speciation of the mercury and redox potential was responsible for the different elemental mercury concentrations. 4.2. EHEvaluation. The value of the apparent redox potential, EHP, can be calculated by using MINEQL because concentration and stability constants of all ligands are known and the values for P have been estimated from volatility measurements. This has been done and the results are given in Table 111. Also tabulated in this table

results confirm, using an independent experimental approach, the findings of others ( 1 4 ,7). The redox response of this couple is Nernstian to the extent that redox potential is directly proportional to pH, which is the dominant variable in the Nernst equation. Registry No. Hg, 7439-97-6; 02,7782-44-7; H20, 7732-18-5.

Table 111. Comparison of EBPand E n expt

E H ~V ,

EH,V

expt

1 2 3 4 5 6

0.58 0.42 0.43 0.42 0.44 0.44

0.98 0.82 0.82

7 8 9 10

0.81

0.80 0.80

11

EHP,V EH,V 0.44 0.40 0.55 0.41 0.22

0.81 0.81 0.93 0.80 0.62

0.6 I

I

I

02

06

Literature Cited

/*

.

1

I

1

07

08

0.8

E,

I

1

(VOW

Flgure 2. Comparison of calculated (EnP)and theoretical (EH)redox

potential.

is the redox potential computed from eq 3, E H . Figure 2 is a plot of E H P versus Ew Also shown on this graph is a linear regression between EHPand E H . The correlation is statistically significant at the 99% confidence level with a correlation coefficient, f,equal to 0.97. The slope is essentially unity (1.00) and the intercept is -0.39 V. This confirms observation by others (2,3,5,7) that the water-oxygen couple exerts a redox potential on the order of 0.5 V below the theoretical Nernstian redox potential. The unit slope suggests that pH is an important factor in determining redox potential, since E H is directly related to pH (eq 3). This is in agreement with the findings of Lindberg and Runnells (8) and Thorstenson (21).

5. Summary and Conclusions Based on the results of this experimental study it was found that the apparent redox potential exerted by the water-oxygen couple in fresh oxic water is -0.4 V below that predicted by the theoretical Nernst equation. These

(1) Stumm, W.; Morgan, J. J. Aquatic Chemistry; Wiley-Interscience: New York, 1981;Chapter 7. (2) Vetter, K.J. Electrochemical Kinetics; Academic Press: London, 1967;pp 615-616. (3) Whitfield, M. Limnol. Oceanogr. 1974,19(5),857-865. (4) Stumm, W. Adv. Water Pollut. Res. 1967,1, 283-307. (5) Garrels, R. M.; Christ, C. L. Solutions, Minerals, and Equilibria; Harper and Row: New York, 1965;pp 136-137. (6) Breck, W. g. In The Sea; Goldberg, E. D., Ed.; Wiley-Interscience: New York, 1974;Vol. 5,p 153. (7) Nordstrom, D. K.;Jenne, E. A.; Ball, J. W. In Chemical Modeling in Aqueous S y s t e m ; Jenne, E. A,, Ed.; ACS Symposium Series 93; American Chemical Society: Washington, DC, 1979;pp 51-79. (8) Lindberg, R. D.; Runnells, D. D. Science 1984,225,925-927. (9) Berner, R. A. Geochim. Cosmochim.Acta 1963,27,563-575. (10) Cotton, F.A.; Willdnson, G. Advanced Inorganic Chemistry; Wiley-Interscience: New York, 1980;pp 593-596. (11) Wollast, R.; Billen, G.; Mackenzie, F. T. In Ecological Toxicology Research; McIntyre, A. D., Mills, C. F., Eds.; Plenum: New York, 1975;pp 145-166. (12) Monnier, D.; Loepfe, L. Anal. Chim. Acta 1967,37,339. (13) Gilmour, J. T. Environ. Lett. 1971,2, 143-152. (14) Westall, J. C.;Zachary, J. L.; Morel, F. M. M. MINEQL: A Computer Program for the Calculation of Chemical Equilibrium Composition of Aqueous S y s t e m , Technical Note 18;Dept. of Civil Engineering, M I T Cambridge, MA, 1976. (15) Iverfeldt, A.; Lindquist, 0. Atmos. Enuiron. 1982, 16, 2917-2925. (16) Sanemasa, I. Bull. Chem. SOC.Jpn. 1975,48,1795-1798. (17) Lewis, W. K.;Whitman, W. G. Ind. Eng. Chem. 1924,16, 1215. (18) Liss, P.S.;Slater, P. G. Nature 1974,247,181-184. (19) Glew, D. N.; Hames, D. A. Can. J. Chem. 1971, 49, 3114-3118. (20) Hatch, W.; Ott, W. Anal. Chem. 1968,40,2085. (21) Thorstenson, D.C. Geochim. Cosmochim. Acta 1970,34, 745.

Received for review July 11,1988.Revised manuscript received February 8, 1989. Accepted March 6,1989.

Environ. Sci. Technoi., Vol. 23, No. 7, 1989 831