United States Environmental Protection Agency Robert S. Kerr Environmental Research Laboratory Ada OK 74820 Research and Development EPA/600/S2-86/042 June 1986 Project Summary Dissolved Oxygen and Oxidation-Reduction Potentials in Ground Water Thomas R. Holm, Gregory K. George, and Michael J. Barcelona Water samples were collected from various depths in a pristine sand and gravel water table aquifer at monthly intervals over a period of one year. Dissolved ox- ygen concentrations were near saturation 9 feet below the water table and decreas- ed to nearly zero at 78 feet below the water table. Changes in the Eh values were consistent with changes in the dissolved oxygen concentrations. Hydrogen perox- ide was detected in nanomolar concentra- tions at all depths, but not on every sampling run. Of all oxidation-reduction potentials calculated from substituting analytical results into the Nemst equation, only the Fe3+/Fe2+ couple in the deepest well agreed with the measured Eh within 50 millivolts. For the O2/H202 and NO3 /NH/ couples the range of poten- tials calculated from one year's data overlapped the range of measured Eh values, so there was some agreement on the average. However, for a given sampl- ing run, the various calculated potentials spanned several hundred millivolts, which means that the activity ratios for the various couples differed by many orders of magnitude (i.e. the system was not in redox equilibrium). The concentration pro- files of many solutes, including dissolved oxygen, suggest mixing of shallow and deep ground waters. The observed con- centration profiles were relatively constant over the duration of the sampling. This Project Summary was developed by EPA's Robert S. Kerr Environmental Re- search Laboratory, Ada, OK, to announce key findings of the research project that is fully documented in a separate report of the same title (see Project Report order- ing information at back). Introduction If ground-water quality is defined in terms of the concentrations and properties of substances dissolved in the water, then the presence or absence of dissolved oxygen affects many aspects of ground- water quality. The mobilities, reactivities, and toxicities of many elements depend on their oxidation states. Microbial popula- tions are distinctly different in oxic and anoxic waters and, therefore, the rates of microbial degradation of organic com- pounds are also quite different. Thus, dissolved oxygen affects both the geo- chemical and microbial processes which are likely to influence water quality. In this project we have begun the chemical char- acterization of a shallow aquifer with dissolved oxygen concentrations ranging from near-saturation to near-anoxic, con- ditions that are typical of many aquifers which are susceptible to contamination. The results of this study should, therefore, be useful to microbiologists, geochemists, and engineers studying ground-water quality and the fate of contaminants in ground waters. Aquifer contamination may be partially mitigated by natural physical processes, but the time scale for flushing a conser- vative substance from an aquifer is propor- tional to the hydraulic residence time, which can be hundreds of years in some aquifers, and the time required for the flushing of a hydrophobia contaminant that is strongly sorbed by the aquifer solids may be much longer than the residence time. Clearly, physical processes for aquifer self-purification can be very slow. However, natural chemical and biological processes, many of which are redox- ------- sensitive, can also contribute to aquifer self-purification. In view of the importance of the redox status of ground water and other natural waters it is desirable to have a convenient, reliable redox indicator. The potential of a platinum electrode, or Eh, is often used as a geochemical redox indicator because the Eh of a water sample can be readily mea- sured and, in certain aquatic environ- ments, the Eh can be quantitatively related to chemical speciation. However, using the Eh as a redox potential for an entire aquatic system is only meaningful in systems that are in chemical equilibrium and that contain electroactive solutes at approximately millimolar concentrations. Most ground waters are not in equilibrium, and the concentrations of electroactive species are often less than one micromolat Despite its limitations, the Eh can be a useful qualitiative indicator of the redox status. Consistently measured or calcu- lated Eh values can indicate relative redox levels in a single system, such as suc- cessive depths in sediment or zones in flow systems with similar concentrations of major electrolyte ions and electroactive minor ions. On the other hand, comparison of the Eh values of very different waters, e.g., of ground waters that are well-poised and anoxic with those that are poorly- poised and oxic, is probably not mean- ingful. A more reliable characterization of the redox level of a natural water involves a complete chemical analysis, including all redox-active species as well as Eh. This approach has been taken in this project. Materials and Methods Four monitoring wells were installed in the Sand Ridge State Forest near Havana, Illinois, in the Havana Lowlands water table aquifer. Three wells were completed at depths of 35, 50, and 65 feet in October, 1984, and the fourth well was completed at 104 feet in September, 1985. A deter- gent- and acid-cleaned Teflon positive- displacement bladder pump and Teflon tubing were installed in each well. For pH, Eh, specific conductance, and temperature measurements, ground water was pumped from the well and through a flow-cell. When all parameters reached stable values (i.e., less than 0.05 pH units, 10 mV, or 10 pS cm"1 change in suc- cessive casing volumes) the water that was being pumped from that point on was considered to be representative of the aquifer water. These stable values were recorded. After well flushing, unfiltered water samples were collected for organic car- bon, ammonia, hydrogen peroxide, hydrogen sulfide, and dissolved oxygen determinations. Mass transfer calculations indicated that oxygen diffusing through the sampling tubing probably did not ap- preciably contaminate any of the samples from the 35-, 50-, or 65-foot wells but may have contaminated the samples from the 104-foot well. After collection of the unfiltered samples, filtered samples were collected for determination of alkalinity, metals, and anions. In-line filtration was used to pre- vent air contact. Standard preservatives and holding times were used to preserve and store the water samples. Field blanks were collected by filtering deionized water. Determinations of the unstable solutes, dissolved oxygen (DO), alkalinity, and hydrogen peroxide were performed im- mediately after sample collection. In water samples from 35, 50, and 65 feet, DO was determined using the azide modification of the Winkler method. In samples from 104 feet a field colorimetric method (Chemetrics Inc., Calverton, VA) was used to determine DO. Hydrogen peroxide concentrations were determined using the fluorimetric scopo- letin/horseradish peroxidase method. Rea- gent blanks were run substituting distill- ed, deionized, and freshly distilled water for ground water. Hydrogen peroxide con- centrations in the blanks were equal to or up to three times greater than those of the samples for many sampling runs. Varying the reagent concentrations had no effect on the blanks. The H202 signal in the blanks was eliminated by addition of the enzyme prior to the scopoletin. Storing distilled water did not increase the concen- trations of H2O2 over those in freshly distilled water. The stoichiometry of the reaction in distilled water blanks was close to the accepted value of 1.0, which sug- gests that side reactions were not respon- sible for fluorescence quenching. There- fore, the measured quenching in the reagent blanks was due to H2O2 present in the distilled water and was not intro- duced with the reagents or produced by the analytical method. Blanks have not been addressed in the literature on this method. However, a num- ber of authors have suggested that H2O2 can be produced in deionized distilled water as a result of microbial activity, photochemical reactions, or by sparging with air. It is likely that hydrogen peroxide is a normal trace component of our distil- led water, produced in the distillation pro- cess or carried over from the feed water and varying with inputs of radiation and dissolved oxygen. Therefore, while a large > fraction of our H2O2 determinations do ' not satisfy the conventional analytical criterion of low blanks, we feel that we have positively identified H202 in ground waters from the Sand Ridge site. Manual colorimetric methods were used to determine ammonia in all samples and iron in samples from 35, 50, and 65 feet. Manganese determinations in the 35, 50, and 65-foot samples were performed using cathodic stripping and anodic strip- ping voltammetry. For the 9/19/85 and 10/17/85 samples, Fe and Mn were deter- mined by atomic absorption spectrophoto- metry. Automated adaptations of standard colorimetric methods were used to deter- mine orthoposphate, dissolved silica, sul- fate, nitrite, and nitrate. Chloride was determined by automated potentiometric titration. Volatile and nonvolatile organic carbon fractions were determined by wet oxidation and infrared CO2 detection. Results and Discussion The profiles of DO and Eh are shown in Figure 1. The shallowest ground waters were nearly saturated with DO and the deepest waters were nearly anoxic. The measured DO concentration of 0.2 mg L "1 in the 104-foot samples may have been an artifact of sampling (see below). The highest Eh values were measured in the shallow oxic ground waters and the lowest Eh values were measured in the nearly anoxic deep waters. Dissolved oxygen concentrations fluctuated up to 1.5 mg L ~1 between some sampling runs at all depths (Figure 2), which is greater than the uncertainty in the Winkler titra- tion; therefore the observed fluctuations were not an artifact of the analyses. Most of the Eh values in the three shallow wells varied between +330 and +430 mV (Figure 3) with a fairly constant difference of less than 50 mV between the Eh values at 35 and 65 feet. Iron and Mn concentrations were below the detection limits of approximately 1 ng L~1 in all samples from the 35-, 50-, and 65-foot wells. In the samples from the 104-foot well, Fe and Mn concentrations were approximately 0.5 and 0.2 mg L"1, respectively. Assuming that most of the dissolved Fe is in the ferrous form and using the published rate law for ferrous iron oxidation by dissolved oxygen, the half-life of ferrous iron in the aquifer should be less than two hours, which is much shorter than the hydraulic residence time. Therefore, the DO measured in the 104-foot samples was probably the result of oxygen contamination during sampling. ------- 0 -10- -20- -30- -40- -50- -70 - -50- -50 - -100 - -110 4 6 + Dissolved Oxygen m/L 10 12 0.2 0.4 0.6 OEhfV) Figure 1. Concentrations of dissolved oxygen and Eh values in ground waters of Havana lowlands aquifer, 9/19/85. 11 10 . 9 - 8 - I" 100 200 300 Time (days) Figure 2. Time series graph of dissolved oxygen concentrations in ground waters of Havana lowlands. D 35, + 50, O 65 feet. The time series graph of hydrogen per- oxide is shown in Figure 4. The highest concentrations were measured on 1/17/85. The 2/19/85 concentrations were some- what lower than those observed on 1/17. There was no consistent profile for H2O2. The 02/H202 couple may be a signifi- cant participant in the redox chemistry of certain natural waters, but its influence has not been studied. In the Sand Ridge ground waters, H202 concentrations may be comparable to those of other electroac- tive solutes. For example, dissolved Fe con- centrations were below the detection limit of approximately 2 \HQ L~1, or 36 nM, which is similar to H2O2 concentrations measured on some sampling runs. In prin- ciple, the oxidizing power of dissolved oxgen can be controlled by the kinetics of its reduction, if the rate of reduction of H202 is slower than the rate of its forma- tion, then the potential is effectively that of O2/H202, and O2becomes a weaker oxidant than if it were directly reduced to H20. There is now evidence that H2O2 is formed and accumulates, in the photooxi- dation of organic compounds in surface and ground waters. However, the presence of H202 in untreated ground water has not previously been reported. Since H202 concentrations in precipitation are fre- quently orders of magnitude greater than in ground water due to atmospheric photoproduction, rain water may also be a source of peroxide in ground water (i.e. relict H202 from recharge). Despite the wide-ranging sampling and detailed analysis of H202 photoaccumulation rates, no measurements of H2O2 in ground waters prior to irradiation have been reported. The few reported ground water H202 measurements were below detection limits. The concentration profiles of nitrate and orthophosphate (Figure 5) were represen- tative of most sampling runs. Nitrate con- centrations decreased with depth, while phosphate concentrations increased. The decrease to a nearly undetectable nitrate concentration in the 104-foot well is con- sistent with the sharp decrease in DO and Eh between 65 and 104 feet. However, reduced nitrogen species, nitrite or am- monia, were not detected. Nitrate concen- trations tended to decrease with time (Figure 6), while phosphate concentra- tions showed no temporal trend. The concentration profiles of DO, NO3", and, possibly, o-POA had two inflection points. Such profiles can be produced by diffusion/dispersion. The con- ceptual model is that upgradient from the ------- 600 0 100 200 300 Time (days) D35 +50 O 65 A 104 Figure 3. Time series graph of Eh values of ground waters of Havana lowlands. D 35, + 50, 65, A 104 feet. roo 200 Time (days) 300 400 Figure 4. Time series graph of hydrogen peroxide concentrations in ground waters of Havana lowlands. D 35, + 50, 65 feet. monitoring wells' two layers of ground nitrate, are separated by a thin transitional water, i.e. oxic recharge waters that leach region. Diffusion and dispersion causes nitrate from fertilized fields and deep the transitional layer to broaden and pro- anoxic ground waters that are low in duces the observed profile. It may be possible to estimate the average vertical dispersion coefficient in the aquifer by modeling the concentration profiles of DO, nitrate, and other solutes. Calculating Redox Potentials The redox couples considered were O2/H2O2, Fe(lll)/Fe(ll), NO3VNH4+ Mn(lll)/Mn(ll) and S042VHS~. For each couple a range of potentials was calculated by substituting analytical values of the various solutes into the Nernst equation. These calculated potentials were then compared to the observed range of Eh values. As an example, the maximum calculated 02/H202 potential was for water samples collected from the 35-foot well with a DO of 9 mg L-1 and pH of 7.5. The minimum potential was for 65-foot samples with DO of 3 mg L~1 and pH of 8. The DO concentrations mea- sured in the 104-foot well were not con- sidered because the values were question- able. A range of H2O2 concentrations of 0.1-10.0 nM was assumed. The calculated DO/H2O2 potentials ranged from +430 to + 535 mV, which overlaps the observed range of most of the Eh measurements for the 35- to 65-foot ground waters. For the N03~/NH4+ couple only nitrate was detected. As a result, the Nernst equation becomes an inequality (equation 1) (E(mV) > 884 + 7.08 (log (N03~) - 10 pH - log DC) (1) where the standard potential at 12°C is 884 mV, braces indicate ionic activities and DL is the detection limit for ammonia nitrogen (O.O5 mg L"). The calculated lower bound for NO3VNH4+ potentials was +324 mV, which is consistent with the observed range of Eh values. The calculated potentials would fall in the observed Eh range if the NH4+ concen- tration were less than 7 ng L"1. In order for NO3"/NH4+ potential to be consistent with the observed Eh, NO3~ must be below detection and NH/ must be de- tected. However, NH/ was not detected. Thus, the NO3VNH4+couple was appar- ently not in equilibrium in the deep waters. Iron and manganese were below detec- tion in all samples collected from the 35-, 50-, and 65-foot wells. However, hydrous oxides of Fe and Mn were extracted from aquifer and sand samples. Thus, lower bounds for calculated potentials of couples involving these metals were calculated assuming equilibrium with the oxides. For the couple Fe(OH)3/Fe2+ the ------- 0 -10 -20 -30 -40 | -60 -80 - -90- -100 - -110 0.00 0.02 0.04 0.06 Concentration (mM) + Phosphate* 100 0.08 0.10 D Nitrate Figure 5. Profile of nitrate and orthophosphate concentrations in ground waters of Havana lowlands. 9/19/8S. D nitrate, + orthophosphate. 100 200 Time (days) 300 Figure 6. Time series graph of nitrate concentrations in ground waters of Havana lowlands. D 35, + 50, 65 feet. Nernst equation becomes the inequality given by equation 2 E(mV) > 756 + 56.6 (4.88 - 3 pH - log DL) (2) where a detection limit of 1 ^g L 1 is assumed for Fe determination by the fer- rozine method, all soluble Fe is assumed to be Fe2+, and 4.88 is the log of the solubility product of Fe(OH)3. (Note that the only variable in equations ispH.)The lower bound for the Fe(OH)3/Fe2+ poten- tial was estimated to be +155 mV, which is not close to but is consistent with the observed Eh values. The calculated Fe(OH)3-Fe3+ potential would fall in the observed Eh range if the Fe2+ concentra- tions were less than 20 ng L"1, and Fe3+ levels were in equilibrium with Fe(OH)3 as assumed. The calculated Fe(OH)3/Fe2+ potential in the 104-foot ground waters was +38 mV. An uncertainty in the value of the solubility product of Fe(OH)3 of one logarithmic unit, a pH reading that was high by 0.33 pH units, or a combination of smaller uncertainties in these parameters could account for the disagreement be- tween the calculated and measured po- tentials. The solubility product determined at 25°C was used for the potential calculation because the enthalpy change of the reaction was not listed, which precluded a temperature correction. Thus, the calculated Fe(OH)3/Fe2+ potential agrees reasonably well with Eh. The lower bound for calculated MnOOH (Manganite)/Mn2+potentials in the 35- to 65-foot ground waters was calculated as for Fe(OH)3/Fe2+. The resulting value of +570 mV is greater than all but two extreme Eh measurements. In order for the calculated Mn potential to fall in the measured Eh range the Mn2* concentra- tion would have to be greater than 40 mg L~1, which would have been easily de- tectable. Assuming the presence of MnOOH at 104 feet, the calculated MnOOH/MN2+ potential is +474 mV, which is much higher than the observed value of +100 mV. Apparently the MnOOH-Mn2+couple is not at equilibrium in the aquifer. Measured and calculated potentials are compared in Figure 7. The best agreement between calculated and measured poten- tials was for Fe in the nearly anoxic 104-foot well. The potentials of the other couples differed from the Eh by amounts that were not possible to explain by ex- perimental uncertainty. In the shallower, oxic wells, the range of potentials calcu- lated for the O2-H202 couple overlapped the range of Eh values measured over one year. However, for individual samples, the differences between calculated and mea- sured potentials correspond to differences in the ratio of O2 to H202 concentrations of many orders of magnitude. Similar lack of agreement between measured poten- tials and potentials calculated from analytical data has been noted in an ex- amination of published ground water data ------- 600 500 400 300 200 roo Eh 02/H20 NOl/NHl (35-65) \f35-65) — Fe(OH)3/Fe** Figure 7. (35-65) JL Eh -(104) FefOHk/Fe* -(104) Comparison of Eh values and calculated redox potentials of ground waters of Havana low- lands ground waters. water system is not in redox equilibrium at any depth sampled because calculated redox potentials for several couples differ by up to hundreds of millivolts, corres- ponding to differences in activity quotients of orders of magnitude. The concentration profile of DO is fairly constant and is con- sistent with mixing of layers of oxic and anoxic water. Modeling the DO profile may help understand mixing in the aquifer. Recommendations The field studies should be expanded to include very anoxic ground waters in a similar hydrologic environment, preferably in the same aquifer. Solid oxidants and reductants in the aquifer should be studied to determine their role in redox processes. Geochemical extractions can give an esti- mate of the aquifer oxidizing or reducing capacity. Redox titrations of aquifer sed- iments can estimate the redox buffer cap- acity of the aquifer system. Respirometry experiments may estimate kinetics of aquifer redox processes. Characterization of the organic matter in ground water is essential to understanding aquifer redox processes in both oxic and anoxic waters. This characterization should include com- plexation of electroactive metals, which may influence measured Eh values, and molecular weight and functional group determinations, which can indicate poten- tial substrates for microbial respiration. (Linberg, R.D. and D.D. Runnels. 1984. Science 225:925-927). In certain special situations, eg. anoxic waters at low pH values, measured Eh values have been found to correspond to speciation of one or more elements. However, in oxic waters at near neutral pH, the measured Eh is usually far from that calculated from elemental speciation. The probable reason for the lack of agreement between measured and calculated Eh is that the system is not in redox equilibrium. Conclusions The relative redox status of ground water at the Sand Ridge site is related to the dissolved oxygen concentration as shown by Eh measurements and chemical speciation calculations. The Eh qualitative- ly indicates the relative redox status of the ground waters, i.e. the lower the DO the lower the Eh. In the deepest waters sam- pled, the Eh is quantitatively related to Fe speciation. However, for other redox couples and at other depths, there is no quantitative relationship between redox speciation and Eh. The aquifer-ground U. S. GOVERNMENT PRINTING OFFICE:1986/646-l 16/20848 ------- Thomas R. Holm. Gregory K. George, and Michael J. Barcelona are with Illinois State Water Survey. Champaign. IL 61820-7407. Bert E. Bledsoe is the EPA Project Officer (see below). The complete report, entitled "Dissolved Oxygen and Oxidation-Reduction Potentials in Ground Water,"{Order No. PB 86-179 678/AS; Cost: $11.95, subject to change) will be available only from: National Technical Information Service 5285 Port Royal Road Springfield, VA 22161 Telephone: 703-487-4650 The EPA Project Officer can be contacted at: Robert S. Kerr Environmental Research Laboratory U.S. Environmental Protection Agency P.O.Box 1198 Ada, OK 74820 United States Environmental Protection Agency Center for Environmental Research Information Cincinnati OH 45268 BULK RATE POSTAGE & FEES PAII EPA PERMIT No. G-35 Official Business Penalty for Private Use S300 EPA/600/S2-86/042 0000329 PS U S ENVIR PROTECTION AGENCY REGION 5 LIBRARY 230 S DEARBORN STREET CHICAGO IL 60604 ------- |