Robert D. Waters, Patrick V. Brady and David J. Borns
Sandia National Laboratories

Sandia National Laboratories is developing guidelines that outline the technical basis for relying on natural attenuation for the remediation of metals and radionuclide-contaminated soils and groundwaters at US Department of Energy (DOE) sites for those specific cases where natural processes are effective at ameliorating soil and groundwater toxicity. Remediation by monitored natural attenuation (MNA) requires a clear identification of the specific reaction(s) by which contaminant levels are made less available as well as considerable long-term monitoring. Central to MNA is the development of a conceptual model describing the biogeochemical behavior of contaminant(s) in the subsurface. The conceptual model will be used to make testable predictions of contaminant availability over time. In many cases, comparison between this prediction and field measurements will provide the 'bright line' test of whether MNA is to be implemented. As a result, development of the conceptual model should guide site characterization activities as well as long-term monitoring. Sandia is in the process of developing a protocol and guidance document to help DOE site managers and their Environmental Protection Agency regulators determine if remediation by natural attenuation is appropriate for their specific environmental restoration projects and, if so, the additional characterization and analysis required to ensure that the approach is successful.

Natural attenuation is defined as "naturally occurring processes in the environment that act without human intervention to reduce the mass, toxicity, mobility, volume, or concentration of contaminants in those media". These in situ processes include "biodegradation, dispersion, dilution, sorption, volatilization, and/or chemical and biochemical stabilization of contaminants". Natural attenuation has been extensively documented and is increasingly relied on for the cleanup of soils and groundwaters contaminated with fuel hydrocarbons, polyaromatic hydrocarbons (PAHs), and even chlorinated solvents [1, 2]. Often natural attenuation leads to a net decrease in remediation costs while providing substantial reductions in contaminant levels and risks to human health and the environment.

Sites contaminated with metals and radionuclides pose special problems for application of the natural attenuation approach. Whereas natural attenuation of organic contaminants means breakdown by microorganisms, natural attenuation of metals often means sequestering or transformation by the soil matrix or dilution. Radionuclides, in turn, might be considered naturally attenuated if their interactions with soils result in transport times to possible receptors much greater than their radioactive half lives. Although laboratory and field evidence for the transformation and sequestering of inorganics is abundant, natural attenuation of metals and radionuclides has received less emphasis in the regulatory universe relative to the natural attenuation of organic contaminants. This is changing with the recent issuance of Environmental Protection Agency (EPA) guidance on monitored natural attenuation [3].

The increasing recognition of natural attenuation by EPA does not constitute a change in cleanup goals; nor does it imply a walk-a-way, or default solution. The burden of proof remains on the proponent, not the regulator. EPA expects implementation to require extensive site characterization, long-term monitoring, risk assessment, and contingency measures. The net result is that implementation is site-specific. Typically, natural attenuation has been applied in combination with more active remedial approaches, at sites where cleanup levels were not extensively exceeded in the first place, or after more proactive remediation efforts had been halted. Generally, contingency plans requiring active remediation were in place.

Site characterization for MNA is typically demonstrated, in decreasing order of importance, through:

1. historical data showing declining contaminant levels in soils or groundwaters;
2. geological and/or geochemical data demonstrating natural attenuation processes or rates; and
3. field or soil microcosm studies.

If the first criterion is satisfied, further effort is made to examine the other criteria. On the other hand, in the absence of historical evidence for reductions in contaminant levels, the argument for natural attenuation probably cannot be made solely on the latter two. In the end, the regulators make the decision whether natural attenuation is applicable.

The future use of the site must be taken into account if remediation by natural attenuation is considered. MNA often takes longer to achieve cleanup goals than more active remediation measures. Land use concerns may consequently bias cleanup towards proactive, as opposed to passive remediation. Isolated sites with great distances from potential receptors are, therefore, more likely to be candidates for MNA compared to sites connected by short travel times to potential receptors. Some advantages and disadvantages of MNA are outlined in Table I.

In summary, a number of milestones must be achieved to build support for MNA at a particular site. The source term must be controlled - either treated or removed, to limit subsequent contaminant fluxing. The plume and down-gradient areas must be monitored to establish plume dynamics. If contaminant levels are seen to decrease over time, a conceptual model to account for the decrease should be established, and if possible, refined to provide a basis for making defensible predictions of the long-term evolution of contaminant levels. These milestones are not necessarily easy or cheap to achieve, and in all cases the appropriate regulatory agency should be involved at the earliest stages.

Metals and radionuclides can be removed from soil solutions and groundwaters by (1) sorption to mineral surfaces and/or soil organic matter (SOM); (2) formation of insoluble solids; (3) uptake by plants and organisms; and occasionally (4) through volatilization (e.g. methylation of mercury). Focusing on the formation of adsorbed species ('surface complexes'), insoluble solids and uptake by plants, we note that metal/radionuclide speciation depends primarily on the ambient biogeochemical conditions of the soil or groundwater: pH, redox state (electron availability), alkalinity, and the presence of chelating (e.g. EDTA, natural organic acids) or solid-forming (e.g. phosphate) ligands are critically important.

At the same time, the sequestering of metals/radionuclides out of the aqueous phase often makes their engineered extraction problematic. Corrosive soil leaches, vitrification techniques, and grout curtains are examples of extreme measures to liberate or isolate metals in soils. In many cases the technical impracticability of metal/radionuclide extraction is a direct result of the natural attenuation processes. Nevertheless, contaminant immobilization cannot be assumed - some metals/radionuclides (e.g., chromate and pertechnetate) have very little interaction with the matrix, and can move rapidly through soils and groundwaters. It is, therefore, necessary to explore sorption, plant uptake, and solubility in substantial detail. Ideally, this will provide some basis for identifying the conditions where MNA might be plausible and where it clearly won't be.

Sorption is particularly effective at limiting the concentrations of metals/radionuclides that are present in trace quantities. At high contaminant levels the actual amount of contaminant in solution is typically determined by the presence of contaminant-containing insoluble minerals. There are obvious exceptions. For example, Cs⁺ and TcO₄^- form no insoluble solids.

Sorption can be characterized as being 'reversible' or 'irreversible'. Contaminants sorbed reversibly to a surface can be desorbed in response to a decrease in contaminant level in solution. In other words, the sorbed species remains in contact with the solution and responds to changes in solution composition. Irreversibly sorbed species typically do not reequilibrate rapidly with solutions once sorbed. Irreversible sorption may occur through a combination of occlusion (overcoating), diffusion into dead-end pores, or structural collapse of the mineral around the sorbed species.

Because equilibrium desorption cannot always be assumed it is important to split sorption into forward and backward reactions (respectively, adsorption and desorption) and treat them separately. Adsorption is very rapid and typically occurs over time spans less than a second, but sometimes longer. Adsorption from solution varies with pH, the type of mineral surface, the amount of surface coverage, the concentration of the trace element, and the composition of the soil solution. Ligands that form strong complexes with the contaminant may either decrease the total amount of sorption or form ternary complexes with the surface. At high pH, negative surface charge is maximal; at low pH positive surface charge is greatest. As a result cation sorption increases with pH; anion sorption increases with decreasing pH.

While adsorption has received the most attention, in many cases desorption may be the more important control over metal/radionuclide release at contaminated sites. Routinely the most contaminated sections of a site are removed and/or stabilized leaving a plume of contamination behind wherein the contaminants are primarily sorbed to mineral surfaces. Almost all performance assessment calculations assume that desorption is reversible. Hence, when fresh recharge comes into contact with sorbed contaminants the latter are predicted to instantaneously equilibrate, in effect setting contaminant levels in solution. In reality desorption rates are often relatively slow, sometimes vanishingly so. The actual desorption rate will in many cases determine the net export of metal/radionuclide toxicity.

Synthetic organic contaminants co-mingled with metals/radionuclides often give rise to the same observation. NTA, EDTA, and DTPA are all synthetic organics, which show up at DOE sites. Citrate and oxalate are two natural chelating agents of concern as well. Degradation rates of these chelates typically follows the trend: citrate ~ oxalate >> NTA > EDTA > DTPA. Estimating transport of chelated metals and radionuclides requires that the coupled processes of metal chelation, sorption, and chelate breakdown be understood.

Table II outlines likely natural attenuation mechanisms for most of the radionuclides and metals of concern. Also shown are the potential caveats that must be kept in mind for each contaminant. Specifically, we have sought to identify what soil chemical parameters control the natural attenuation pathway, and, by extension, what changes in soil chemistry would work against the given natural attenuation mechanisms.

Table III outlines the minimal geochemical data needed to determine if the particular natural attenuation mechanisms are operative. Data needs depend primarily on whether the likely fate of the compound is as a component of an insoluble solid, a sorbed contaminant, or, possibly, a species occluded on an iron hydroxide or carbonate mineral surface, or irreversibly sorbed to an interlayer clay site.

There is currently no protocol for implementing natural attenuation of metals or radionuclides. Typically technical protocols for implementing natural attenuation of organic contaminants follow a format along the following lines.

1. Review available site data. This should provide a useable hydrologic model, and roughly locate receptors, and pathways. This effort should also give some idea of whether or not natural attenuation is a possibility. Specifically, it should be determined whether existing data suggests a decrease in contaminant bioavailability over time. At this point an explicit consideration of showstoppers should be made.
2. Develop a preliminary conceptual model and assess the potential for natural attenuation. If contaminant levels in soil or groundwater have decreased or if the geochemical conditions appear to be favorable for natural attenuation (see e.g., Table II) a conceptual model outlining natural attenuation should be developed.
3. If needed, perform additional site characterization to support natural attenuation. Very rarely will the data required to test and calibrate a conceptual model be available at the outset. Instead additional data gathering (see e.g., Table III) will be required, particularly to support step 5 below.
4. Update conceptual model.
5. Simulate long-term site behavior. This step will require some estimate of the rate at which natural attenuation is occurring. Note that, whereas biodegradation of organic contaminants can relatively simply be fit to first order rate laws, there is less of a basis for predicting rates of the non-biodegradation natural attenuation mechanisms (e.g. irreversible sorption, solid formation, etc.).
6. Perform an exposure pathway analysis.
7. If natural attenuation is acceptable, prepare long-term monitoring plan and contingency plan.
8. Present results to regulators.

Natural attenuation of organic contaminants is generally demonstrated using a wealth of evidence argument pointing to reductions in contaminant mass. The four most effective components used to convince a regulatory agency are: evidence of contaminant loss in the field, variations in electron donor/acceptor levels, appearance of degradation byproducts, and soil microcosm studies done in the lab. However, the same approach probably cannot be used for inorganics. The appearance of byproducts, or variation in acceptor/donor levels, probably cannot be used to monitor irreversible sorption or the growth of contaminant-bearing insoluble minerals. When a contaminant, such as lead, sorbs it will displace some other cation such as Ca²⁺, which is likely to be far more abundant in solution. When Cs⁺ sorbs to clay, chances are that it will be present in only trace amounts and far less abundant in solution than the Na⁺ or K⁺ it displaces. As a result, while irreversible sorption of trace contaminants will dramatically affect solution levels of the latter, changes in other background metal concentrations will more than likely be minimal. The growth of contaminant-containing hydroxides, carbonates, and sulfides may also cause undetectable variations in hydroxide, carbonate, and sulfide levels in solution because the latter are typically present in initially greater concentrations than the metals with which they combine.

Standard geochemical codes [e.g., 5, 6] can be used to calculate whether contaminant levels are limited by the formation of an insoluble phase (e.g., Ba²⁺ by BaSO₄ growth). Geochemical modeling to support uptake by sorption is not far enough along to be a stand-alone demonstration of metal sorption. Instead, uptake by sorption can be demonstrated by: demonstrating that the sorbing phase is present in soils through a solubility calculation or direct observation; and showing that an appreciable fraction of the compound is associated with that phase. The latter is most directly done through sequential soil leaching procedures that dissolve specific minerals, along with any sorbed material. For example, citrate-Dithionate solutions remove iron hydroxides from soils. Hydrofluoric acid removes silicates. H₂O₂ removes organic matter. Acid acetate buffer solutions remove calcium carbonate.

Natural attenuation, or any other remediation strategy, can only be assessed with regard to clearly defined standards. It is important to consider what objectives can and cannot be attained by natural attenuation as well as the time scale over which various objectives may be attained. Environmental quality standards for the subsurface are defined for both the immobile phase (e.g., soil) and for groundwater.

Sorption processes, although they retard the migration of the contaminant toward potential receptors, necessarily involve association of the contaminant with the immobile phase. Soil quality criteria are commonly defined in terms of the total metal concentration in the soil. Since metals are naturally occurring substances, contamination can only be defined relative to some background level such as average crustal abundance. If total metal concentration in the soil is taken as the operative standard, then natural attenuation can only be applied if some zone of contamination is excluded from this standard for an extended period or even in perpetuity. Over the very long term, flushing of contaminated subsurface material with uncontaminated groundwater may decrease the total metal concentration in the soil to background levels. It may, however, be reasonable to define alternative standards for soil quality that correspond to the bioavailability of soil metals. Although the determination of the bioavailable fraction is a complicated problem, it is appropriate to address this question in the context of the applicability of natural attenuation. Note that different standards may need to be applied if surficially contaminated soils are subject to erosion or scouring by wind.

There are a number of technical obstacles, which might potentially limit the effectiveness of natural processes in controlling contaminant movement and availability in the subsurface, and consequently, regulatory acceptance of its implementation. To begin with, unlike the biodegradation of some organic contaminants (e.g., fuel hydrocarbons), which results in the contaminant of concern 'going away', typically metals and long-lived radionuclides will remain in the subsurface. (If radionuclides have sufficiently short half-lives they may 'go away' as well). In other words, many metals and radionuclides may still be present, though unavailable for biologic uptake. At the same time, dilution may lower contaminant levels to the point where they are acceptable in a regulatory sense, though there has been no net reduction in contaminant mass.

The transport of contaminants that exist as components of insoluble solids or sorbed (reversibly or irreversibly) to mineral surfaces may, because of the ambient geochemistry, be severely limited. Contaminants that are strongly sorbed or in solid form in soils are likely to see much larger volumes of fresh recharge. Consequently, the potential for dilution is heightened. The immobility of sorbed and/or solid phase contaminants makes them plausible candidates for MNA. Critical to such an assessment is a clear understanding of the sequestering mechanism. Specifically, the speciation of the contaminant needs to be known for three reasons: 1. to be able to predict the long-term stability of the sequestering in the face of possible changes in the ambient geochemistry; 2. to provide some clues as to how much time must elapse before the acceptable contaminant availability is achieved, and; 3. to allow an estimate to be made of the total attenuation capacity of a given soil/groundwater for the specific contaminant.

The potential for remobilization is a critical obstacle for acceptance of the remediation of metals and radionuclides. Obviously, time-spans are important. If remobilization of ⁹⁰Sr or ¹³⁷Cs (half-lives ~ 30 years) occurs over time spans much greater than a hundred years, a very significant fraction of the radiotoxicity will have decayed away. For long-lived radionuclides and metals, dilution may be the only process decreasing potential releases that might occur with remobilization. It is not hard to imagine scenarios leading to the remobilization of most, if not all of the contaminants of concern. Drastic changes in hydrologic conditions and/or subsurface water chemistry may adversely affect natural attenuation processes. For example, a natural attenuation remedy that relies on limited infiltration may be invalidated by irrigation for agricultural development. Cesium 'irreversibly' bound to interlayer clay sites in a soil could be very rapidly released if ammonium-rich fertilizer were subsequently applied for agricultural purposes. By the same token, lead sorbed to iron hydroxides in an initially aerated soil might be released if the soil became flooded, then anoxic, followed by destabilization and dissolution of the iron hydroxide host. On the other hand, the composition ranges of soil and groundwater are typically very limited, primarily because there are a host of biogeochemical processes which tend to control the pH, redox state, alkalinity, etc. of natural waters. Although drastic changes in the compositions of natural waters are more the exception than the rule, it will probably be impossible for site-owners to demonstrate that remobilization will never occur. This is a critical obstacle to the implementation of natural attenuation for metals and radionuclides.

The respective roles of site characterization and monitoring are two important considerations for use of MNA. The argument can be made that site characterization should specifically provide the means to develop a conceptual model of natural attenuation and, to the extent possible, calibrate that model so that contaminant availability can confidently be predicted in the future. Unless MNA is exceedingly fast (which is often not the case) it will be difficult to calibrate a kinetic model for MNA given the time allowed for a site characterization. Ultimately, long-term measurement might be required. Nevertheless, this should not be confused with long-term monitoring. Long-term monitoring should, quite simply, provide the means for assessing whether or not MNA is working. If the conceptual model for MNA is sufficiently effective at reproducing measured trends in contaminant levels, it should allow the frequency of monitoring to be significantly reduced.

An important consideration in evaluating the applicability of natural attenuation for a given site is its intended land use. Natural attenuation may be considered as part of the remediation strategy for a contaminated site or as a component of the permitting of an existing facility. The latter case necessarily involves some on-going release of contaminants into the environment and the relevant question is whether natural attenuation would afford sufficient protection to human health and the environment. In the former case, source control is probably (but not necessarily) a prerequisite to application of natural attenuation.

The efficacy of natural attenuation will depend on numerous factors including the type and extent of primary and secondary contamination (where primary contamination is associated with the original source and secondary contamination with dispersal of contaminants from the source), the hydrologic regime and hydrogeology, subsurface geology, and potential receptors. For a given site, these factors must be evaluated with regard to their likely effects on the sorption and dilution processes by which natural attenuation of metals and radionuclides may be accomplished.

Although natural attenuation encompasses several natural processes, it is important to recognize that only a few of these processes are operative for metals and radionuclides. For both metals and radionuclides, the operative processes are dilution and sorption. Dilution may occur by dispersion of dissolved contaminants in groundwater and/or by dilution of dissolved contaminants into surface water (e.g., upon interception of surface water by contaminated groundwater). Sorption may be defined generally to include the processes of adsorption, coprecipitation, precipitation, and diffusion into the matrix, processes by which solutes become associated with the immobile, solid phase. Sorption may either be reversible or slowly reversible. Slowly reversible sorption processes may be considered as effectively "irreversible" if the time scale for re-release of the contaminant from the solid phase is long relative to some time scale of interest or observation. Slowly reversible sorption of contaminants from solid phases exposed to uncontaminated groundwater may also contribute to dilution of the contaminant. The extent of dilution will be determined by the rate of contaminant release into solution relative to the velocity of groundwater flow.

For radionuclides only, radioactive decay is an additional process contributing to natural attenuation. This mechanism is important for only relatively short-lived radionuclides. In some cases, the in-growth of daughter nuclides may result in an increasing hazard over time that counterbalances or even outweighs the benefit due to loss of the parent nuclide.

We see the building of conceptual models for MNA as one of the primary challenges to its successful implementation. The most important sinks for metals and radionuclides in soil and groundwater are fairly well understood (microbiological effects less so). Nevertheless, field-based techniques for demonstrating that contaminants are being taken up into otherwise inaccessible and/or non-bioavailable fractions of the soil matrix are few and far between, and therefore a critical need. Scanning electron microscopy, isotope exchange techniques, and soil digestions may provide a means for addressing this need.

1. Wiedemeier T. H., Wilson J. T., Kampbell D. H., Miller R. N., and Hansen J. E. (1995) Technical protocol for implementing intrinsic remediation with long-term monitoring for natural attenuation of fuel contaminant dissolved in groundwater. Air Force Center for Technical Excellence, Technology Transfer Division 1 & 2.
2. Wiedemeier T. H., Swanson M. A., Moutoux D. E., Wilson J. T., Kampbell D. H., Hansen J. E., and Haas P. (1996) Overview of the technical protocol for natural attenuation of chlorinated aliphatic hydrocarbons in ground water under development for the U.S. Air Force Center for Environmental Excellence. EPA Symposium on Natural Attenuation of Chlorinated Solvents.
3. US Environmental Protection Agency (EPA) (1997) Use of Monitored Natural Attenuation at Superfund, RCRA Corrective Action, and Underground Storage Tank Sites. USEPA Office of Solid Waste and Emergency Response Directive 9200.4-17.
4. Brady, P. V, Brady, M. V. And Borns, D. J. (1997) Natural Attenuation: CERCLA, RBCAs, and the Future of Environmental Remediation. Lewis Publishers, Boca Raton, Florida.
5. Bethke C. M. (1994) The Geochemist's WorkbenchÔ . A Users Guide to Rxn, Act2, Tact, React, and Gtplot. University of Illinois.
6. Wolery T. J. (1983) A computer program for geochemical aqueous speciation-solubility calculations: user's guide and documentation. Lawrence Livermore National Laboratory Report UCRL-53414.