Detailed site investigations to assess potential inhalation exposure and risk to human health associated with the migration of petroleum hydrocarbon vapors from the subsurface to indoor air are frequently undertaken at leaking underground storage tank (UST) sites, yet documented occurrences of petroleum vapor intrusion are extremely rare. Additional assessments are largely driven by low screening-level concentrations derived from vapor transport modeling that does not consider biodegradation. To address this issue, screening criteria were developed from soil-gas measurements at hundreds of petroleum UST sites spanning a range of environmental conditions, geographic regions, and a 16-year time period (1995 to 2011). The data were evaluated to define vertical separation (screening) distances from the source, beyond which, the potential for vapor intrusion can be considered negligible. The screening distances were derived explicitly from benzene data using specified soil-gas screening levels of 30, 50, and 100 µg/m3 and nonparametric Kaplan-Meier statistics. Results indicate that more than 95% of benzene concentrations in soil gas are ≤30 µg/m3 at any distance above a dissolved-phase hydrocarbon source. Dissolved-phase petroleum hydrocarbon sources are therefore unlikely to pose a risk for vapor intrusion unless groundwater (including capillary fringe) comes in contact with a building foundation. For light nonaqueous-phase liquid (LNAPL) hydrocarbon sources, more than 95% of benzene concentrations in soil gas are ≤30 µg/m3 for vertical screening distances of 13 ft (4 m) or greater. The screening distances derived from this analysis are markedly different from 30 to 100 ft (10 to 30 m) vertical distances commonly found cited in regulatory guidance, even with specific allowances to account for uncertainty in the hydrocarbon source depth or location. Consideration of these screening distances in vapor intrusion guidance would help eliminate unnecessary site characterization at petroleum UST sites and allow more effective and sustainable use of limited resources.
Subsurface to indoor-air vapor intrusion has been investigated for many chemicals, including radon, municipal landfill gas, chlorinated solvents, and petroleum hydrocarbons. The fate and transport of these chemicals can vary substantially in the unsaturated zone, which influences their respective persistence and potential risk for vapor intrusion into buildings. This article describes the development of robust, evidence-based screening criteria for petroleum hydrocarbons, which can be applied in the preliminary stages of vapor intrusion risk assessment.
Assessments of vapor intrusion, human exposure by inhalation and chronic risk to human health are common place at leaking petroleum underground storage tank (UST) sites; however, confirmed occurrences of subsurface vapor intrusion to indoor air at these sites are rare (Fitzpatrick and Fitzgerald 2002; Hers et al. 2003; Tillman and Weaver 2005; McHugh et al. 2010). The exception cases reported in the literature generally involve sites where high-concentration hydrocarbon sources (e.g., light nonaqueous-phase liquids [LNAPL]) are located in relatively close proximity (less than 10 ft or 3 m) to, in contact with, or have entered building foundations (e.g., sumps, basements, or elevator pits) (McHugh et al. 2010). These types of sites tend to be readily identified through complaints of petroleum odors by residents of affected buildings (McHugh et al. 2010). The noted discrepancy between predicted and actual vapor intrusion risk has been attributed to low (e.g., ppb-level) screening-level concentrations in soil gas or groundwater derived using the Johnson and Ettinger (1991) (J&E) model or empirical attenuation factors derived primarily from chlorinated hydrocarbon data (US EPA 2012b), neither of which take into consideration biodegradation in the unsaturated zone (Tillman and Weaver 2005). Biodegradation of petroleum hydrocarbons is readily acknowledged by the regulatory community (US EPA 2002; Tillman and Weaver 2005; ITRC 2007; US EPA 2011) as an important factor affecting the potential for petroleum vapor intrusion, yet uncertainties exist on how to best incorporate the attenuation process in site-screening methodologies.
Several model (Abreu and Johnson 2006; DeVaull 2007; Abreu et al. 2009) and field studies (Hers et al. 2000; Roggemans et al. 2001; Fitzpatrick and Fitzgerald 2002; Ririe et al. 2002) have helped document the significance of biodegradation for petroleum vapor intrusion. In general, biodegradation limits the potential for vapor intrusion if aerobic conditions are present in the unsaturated zone between the hydrocarbon source and building foundation. O2 concentrations in soil gas necessary to support aerobic biodegradation are generally reported to be in the range of 1 to 4% vol/vol (DeVaull 2007). For this condition to occur, O2 availability in the unsaturated zone must exceed the metabolic demand for O2 associated with hydrocarbon biodegradation. Hydrocarbon reaction rates in the unsaturated zone in the presence of O2 are rapid (e.g., half lives on the order of hours or days—DeVaull 2007) and essentially instantaneous with respect to the rates of physical transport (molecular diffusion and advection) commonly associated with vapor intrusion (Davis et al. 2009a). This difference in rates promotes the development of sharp reaction fronts in the unsaturated zone at locations where concentrations of O2 and hydrocarbon are optimal for aerobic biodegradation (Abreu and Johnson 2006; DeVaull 2007; Abreu et al. 2009; Davis et al. 2009a). The aerobic reaction front is typically characterized by hydrocarbon concentrations in soil gas that decrease by several orders of magnitude and O2 concentrations in soil gas that increase by several percent within relatively short (e.g., 3 to 6 ft) (1 to 2 m) vertical distances from the source (DeVaull et al. 1997; Lahvis et al. 1999; Abreu et al. 2009; US EPA 2012a). In general, the hydrocarbon vapor attenuation is predicted to increase exponentially with increasing distance from the source (DeVaull 2007; API 2009). Similar O2/hydrocarbon relations are observed at the edges or fringes of petroleum hydrocarbon plumes in groundwater where aerobic biodegradation becomes a dominant attenuation mechanism (Gutierrez-Neri et al. 2009).
Aerobic reaction fronts have been shown to develop in relatively close proximity to the water table (e.g., capillary zone) above dissolved-phase petroleum hydrocarbon sources (Roggemans et al. 2001; Abreu et al. 2009). In such cases, the microbial metabolic demand for O2 is insufficient to drive conditions in the unsaturated zone extensively anaerobic even in the presence of relatively impermeable surface covers, such as, wet clay soils, or building foundations, which can potentially restrict O2 ingress from the atmosphere (McHugh et al. 2010; US EPA 2012a). The lack of vapor intrusion cases reported in the literature involving dissolved-phase petroleum hydrocarbon sources separated vertically from building foundations supports this finding (McHugh et al. 2010). The position of the aerobic reaction front above LNAPL sources is more variable. The reaction front is predicted to develop some distance above the capillary zone as a result of higher hydrocarbon vapor mass flux and greater metabolic demand for O2 (Abreu et al. 2009). The hydrocarbon mass flux also tends to be greater from LNAPL sources than dissolved-phase sources because LNAPL sources are apt to be distributed above the capillary fringe by water table fluctuations, where soils are less moisture-saturated and less resistant to vapor transport. General relations between petroleum hydrocarbon source type, O2, and the position of the reaction front in the unsaturated zone are conceptually depicted in Figure 1.
The aerobic reaction front may never develop between the vapor source and building foundation at certain sites. This condition is mainly associated with high-concentration hydrocarbon sources (e.g., LNAPL) where the vertical source-separation distance is less than the (finite) reaction-zone thickness. These conditions have primarily been observed at sites where LNAPL is located within 15 ft (5 m) of a building foundation and at certain terminal, pipeline, and manufacturing sites with large-volume petroleum releases in the subsurface (McHugh et al. 2010). This finding is underpinned by transport modeling that indicates that vertical separation distances of more than 23 ft (7 m) may be necessary for development of the aerobic reaction front between high-hydrocarbon concentration (e.g., LNAPL) sources and impervious building foundations with cracks (Abreu and Johnson 2006; Abreu et al. 2009). The modeling does not, however, account for advective flow of O2 in, around, and through concrete building foundations, which can be significant (Fischer et al. 1996; McHugh et al. 2006; Lundegard et al. 2008; Patterson and Davis 2009; Tittarelli 2009). Conversely, O2 may be more limited below large building (e.g., industrial/commercial) foundations, which tend to be thicker and more intact than typical residential building foundations (Eklund and Burrows 2009). For example, anaerobic conditions (O2 concentrations less than 1%) have been observed directly below a large, 243 m2 building foundation, 19 cm thick (Patterson and Davis 2009). Certain petroleum fuel types may also affect the location of the reaction front between the source and building foundation. For example, gasoline containing more than 10% vol/vol ethanol can generate significant concentrations of methane (CH4) in soil gas (1 to 20%) upon biodegradation in groundwater (Ma et al. 2012). Subsequent oxidation of the CH4 in the unsaturated zone can, in turn, reduce the amount of O2 available for biodegradation of petroleum hydrocarbons and increase the separation distance between the source and aerobic reaction front. O2 availability can also be limited by the presence of low-permeability or highly water saturated soils in the unsaturated zone (US EPA 2012a). Such soils are also, however, likely to impede hydrocarbon transport. Therefore hydrocarbon attenuation is predicted to be greater in lower permeability soils (e.g., silts/clays) than higher permeability soils (e.g., sand) for similar ranges of source concentrations and biodegradation rates (API 2009).
Vapors emanating from dissolved-phase sources are likely to be composed primarily of benzene, toluene, ethylbenzene, xylenes (BTEX), and other aromatic hydrocarbons and relatively water soluble petroleum hydrocarbons. Vapors emanating from LNAPL sources will tend to be the same constituents plus a sizeable fraction of aliphatic and relatively insoluble hydrocarbons, especially if the source is large or un-weathered. Benzene is estimated to be the primary constituent of potential concern (COPC) for gasoline and diesel fuels based on a chronic (30-year duration) human health residential exposure scenario (GSI 2012). Benzene is the primary COPC for these fuel types because of its volatility, recognized toxicity, mass fraction in common fuels (gasoline and diesel), and fate and transport in the unsaturated zone. Other key petroleum-related COPCs for vapor intrusion include 1,2 dichloroethane (EDC) and 1,2 dibromoethane (EDB) (lead scavengers used in gasoline primarily from the 1920s to the mid-1990s and early 2000s inside and outside the United States, respectively) and methyl tert-butyl ether (MTBE), ethyl tert-butyl ether (ETBE), diisopropyl ether (DIPE), and tert amyl methyl ether (TAME) (ether oxygenates used primarily from the 1990s to mid-2000s in the United States and to present day in certain countries outside the United States). Instances of lead scavenger vapor intrusion are not, however, found reported in the literature. Documented cases of ether oxygenate vapor intrusion are also uncommon, even though MTBE has historically been included in many vapor intrusion risk assessments conducted in the United States. The only reported case of MTBE vapor intrusion found in the literature involves a gasoline LNAPL source in relative close proximity (5.25 ft or 1.6 m) to a building foundation (Sanders and Hers 2006). Chronic (30 years) inhalation risks are also not expected to be a common occurrence for MTBE at locations where usage has been banned (such as the United States) because of its rapid attenuation in groundwater (McHugh et al. 2012).
Vapor intrusion screening is commonly performed using generic attenuation (alpha) factors that do not factor in biodegradation in the unsaturated zone. The attenuation factor is a proportionality constant relating the hydrocarbon concentration in indoor air to the hydrocarbon concentration in shallow or deep soil gas. For screening purposes, the US EPA (2002) recommends a generic attenuation factor of 0.1 (corresponding to a 10× decrease in concentration) to account for hydrocarbon attenuation between shallow and deep soil gas. Actual (observed) and predicted hydrocarbon attenuation associated with aerobic biodegradation of petroleum hydrocarbons in the unsaturated zone is often much greater, however, (e.g., less than 0.0001—corresponding to a more than 10,000× decrease in vapor concentrations) (Hers et al. 2003; Abreu and Johnson 2006; DeVaull 2007; Abreu et al. 2009). To help address this conservatism, some regulatory agencies apply “bio-attenuation” factors of 0.1, or in rare cases 0.01 or 0.001 provided aerobic conditions in the unsaturated zone can be demonstrated (Golder Associates 2008; Friebel and Nadebaum 2011; California State Water Resources Control Board 2012). The California State Board's bioattenuation factor of 0.001 was largely supported by interim results from this study.
The studies also show that the hydrocarbon attenuation can be highly variable, especially at the aerobic/anaerobic interface, where vapor attenuation factors can vary by several orders of magnitude over relatively short vertical distances (e.g., less than 3 ft or 1 m). The development of sharp hydrocarbon concentration gradients within the unsaturated zone supports the use of source-separation distances as an alternative to attenuation factors for vapor intrusion screening at petroleum hydrocarbon sites. Beyond the source-separation or screening distance, the potential for vapor intrusion can be considered negligible.
The use of screening distances in vapor intrusion risk assessment is not new. The U.S. Environmental Protection Agency (EPA) has proposed vertical and lateral screening distances of 100 ft (30 m) (US EPA 2002). These distances are based on an empirical dataset showing no vapor intrusion at residences separated by more than 100 ft (30 m) laterally from the interpolated edge of a chlorinated hydrocarbon groundwater plume. The 100-ft (30-m) criterion has subsequently been adopted by many U.S. states for application at both chlorinated and petroleum hydrocarbon sites (Eklund et al. 2012). Some state and federal agencies (Atlantic PIRI 2006; ASTM E2600–10) have adopted lesser screening distances for petroleum hydrocarbon sources. The screening distances range from 10 to 30 ft (3 to 10 m) for dissolved-phase sources and from 30 to 100 ft (10 to 30 m) for LNAPL sources. Use of screening distances less than 100 ft (30 m) often requires documentation of aerobic conditions in the unsaturated zone (e.g., O2 concentrations in soil gas more than 2% or 5% vol/vol) and the absence of preferential pathways for vapor migration (Davis et al. 2009b; Wisconsin DNR 2010; California State Water Resources Control Board 2012; New Jersey DEP 2012). Sewer lines intersecting high-concentration dissolved-phase hydrocarbon or LNAPL plumes and building foundations are the only preferential pathway for vapor intrusion at petroleum UST sites found reported in the literatrue (e.g., Pennsylvania DEP 2001; Riis et al. 2010).
In recent years, efforts have been made to validate vertical screening distances for dissolved-phase and LNAPL hydrocarbon sources using empirical soil-gas and groundwater data (Davis 2009; Peargin and Kolhatkar 2011; Wright 2011; US EPA 2013). Davis (2009) estimated that 5 ft (1.5 m) of clean soil were needed to attenuate petroleum hydrocarbon vapors emanating from dissolved-phase sources to non-detect levels. A 30 ft (10 m) thickness of clean soil was estimated for LNAPL sources. The analysis was based on an evaluation of 259 benzene and 210 total petroleum hydrocarbon (TPH) vapor samples from 53 geographical locations in the United States and Canada. Dissolved-phase sites were defined on the basis of benzene concentrations in groundwater less than 1000 µg/L. Similar conclusions were reached by Peargin and Kolhatkar (2011). Benzene concentrations in soil gas in excess of a 10−5 risk-based level of 300 µg/m3 were only observed within 15 ft (5 m) of relatively “high” benzene concentrations in groundwater (more than 1000 µg/L) suspected of being affected by residual-phase LNAPL. In contrast, benzene concentrations in soil gas were generally less than 30 µg/m3 above dissolved-phase sources of benzene less than 1000 µg/L, regardless of the vertical source-separation distance. The analysis was based on 218 pairs of benzene soil-gas and groundwater concentration data from 25 sites (20 in California). Both Davis (2009) and Peargin and Kolhatkar (2011) differentiated dissolved and LNAPL sources on the basis of current or historical presence of free-product LNAPL in groundwater monitoring wells and dissolved-phase benzene concentrations in excess of specified groundwater concentration (e.g., 1000 µg/L). Simple presence/absence of free-phase LNAPL in groundwater monitoring wells may not necessarily be a reliable indicator of residual-phase LNAPL, however (Figure 2). As illustrated, residual-phase LNAPL and dissolved-phase sources will “look the same” based on the absence of free-phase liquid in groundwater monitoring wells; yet residual-phase LNAPL sources will pose a far greater risk for vapor intrusion than dissolved-phase sources, for reasons previously noted. Residual- and free-phase LNAPL sources will roughly “act the same,” however, with respect to their potential for vapor intrusion. The vapor intrusion risk is similar because residual-phase sources will be present at LNAPL sites regardless of whether free-phase LNAPL is visible in nearby groundwater monitoring wells. Wright (2011) examined 1080 pairs of benzene soil vapor and groundwater concentration data collected at 124 primarily UST sites in Australia. Forty-one percentage of the soil-gas data were associated with fractured rock aquifer systems and 12% were collected below building foundations (i.e., sub slab). The soil-gas samples were collected from permanent vapor probes at 120/124 (97%) of sites. Wright's analysis resulted in vertical screening distances of 5 to 10 ft (1.5 to 3 m) for relatively “low-strength” dissolved-phase sources (benzene less than 1000 µg/L and TPH less than 10,000 µg/L) and around 30 ft (10 m) for LNAPL and “poorly characterized” dissolved-phase sources. The screening distances were defined assuming the petroleum vapor intrusion risk is negligible at benzene concentrations in soil gas less than 50 µg/m3, a value that is 5% of the lowest Australian Health Screening Level defined in Friebel and Nadebaum (2011). The empirical soil-gas data described by Wright (2011) and Davis (2009) are currently being used to develop screening distances for the Cooperative Research Centre for Contamination Assessment and Remediation of the Environment (CRC Care) (Wright 2013).
The US EPA Office of Underground Storage Tanks (OUST) is also developing screening distance criteria for UST sites based on the Davis (2009) database and some recent additional site data (US EPA 2013). The data base evaluation involved the derivation of vertical screening distances for dissolved-phase and LNAPL sources and differing VOCs (e.g., benzene, xylenes, hexane, C5-C8 aliphatics, C9-C12 aliphatics, and C9-C18 aromatics). The US EPA database included 893 benzene and 782 TPH soil-gas samples from 74 sites (mainly UST) located primarily in North America. The vast majority of soil-gas samples (more than 99.5%) were collected from permanent vapor probes. Vertical screening distances were determined using two different methods:
a “clean soil” method described by Davis (2009) based on estimating the 95th percentile of unsaturated-zone soil thicknesses (unaffected by residual-phase LNAPL sources) required to attenuate benzene below a specified soil-gas screening level of concern for vapor intrusion of 100 µg/m3; and
a “vertical distance” method based on estimating the source-separation distance required to attenuate benzene below a specified soil-gas screening levels of 50 µg/m3 in more than 95% of cases.
Although the “vertical distance” method involved less uncertainty than the “clean soil” method, the two methods generally resulted in similar vertical screening distance estimates. The screening distances computed for UST sites ranged from 0 to 5 ft (0 to 1.5 m) for dissolved-phase sources to 13 to 15 ft (4 to 5 m) for LNAPL sources. The study also examined effects of hydrocarbon constituent type, soil type (fine and coarse-grained), surface cover (pavement, building foundation, and open ground), and site type (UST vs. non-UST—e.g., refinery, terminal, and petro-chemical) on screening distances. The vertical screening distances derived from benzene data were greater than for other petroleum hydrocarbons and thus assumed to be conservative for establishing screening distances for dissolved-phase and LNAPL sources which contain mixtures of hydrocarbons. The distances were also relatively unaffected by soil type or surface cover. Slightly greater screening distances (18 to 20 ft or 5.5 to 6 m) were determined for LNAPL sources and non-UST sites. The sensitivity to site type was uncertain, but assumed related to larger source sizes (release volumes) and increased demands on O2 availability at non-UST sites than UST sites. The US EPA (2013) database analysis is currently being used to support screening distance development for the US EPA OUST (Walker 2012).
This study examines a combined empirical database of petroleum UST data from US EPA (2013) and Wright (2011). Vertical screening distances are derived specifically for dissolved-phase and LNAPL sources using benzene soil-gas data. The screening distances are derived assuming specified screening-level concentrations for benzene in soil gas (30, 50, and 100 µg/m3) below which the potential for vapor intrusion is deemed negligible. The screening criteria are intended for use in chronic human health vapor inhalation risk assessments. Short-term (acute) safety risks associated with the generation of flammable gases (such as CH4) in the subsurface are not addressed. These risks are typically assessed independently and before chronic vapor inhalation screening.
Screening Distances for Petroleum Hydrocarbons
The US EPA (2013) and Wright (2011) databases contain soil-gas and groundwater data amassed from UST and non-UST (i.e., terminal, pipeline, and manufacturing) sites across North America (primarily) and Australia, respectively. As noted previously, the soil-gas samples were collected primarily from permanent soil-gas vapor probes. The soil-gas and groundwater data were quantified using standard hydrocarbon analytical methods, including, EPA Methods 8015, 8021, 8260, TO-3, and TO-15; Canadian Council of Ministers of the Environment (CCME) Tier I Method; Massachusetts Department of Environmental Protection (MADEP) Air-Phase Petroleum Hydrocarbon (APH) Method; and C6-C9 (by Purge-and-Trap Gas chromatography–mass spectrometry—GC/MS) and C10-C14 (by Gas Chromatography-Flame Ionization Detector—GC FID) TPH methods by various Australian laboratories), and fixed gas methods (ASTM D1946 and EPA Method 3C). Both databases have undergone extensive quality control and quality assurance (QA/QC) measures to ensure objectives were met for data interpretation and assessment. QA/QC of the data included reviews of site investigation reports; groundwater monitoring, soil-gas, and borehole log data; site plans; locations of sample locations relative to USTs and other potential unsaturated-zone sources; sample methods and analyses; data quality testing (e.g., pneumatic and tracer testing, purging procedures, sample breakthrough results); and general relations between hydrocarbon and fixed gas (O2, CO2, and CH4) concentrations in the unsaturated zone (i.e., broad consistency with the conceptual model for aerobic and anaerobic hydrocarbon biodegradation). The site data contained in both US EPA (2013) and Wright (2011) were also ranked independently using similar methods to assess overall data quality and confidence in the conceptual model for vertical hydrocarbon transport from a source at depth. Suspect data associated with unacceptable QA/QC (e.g., tracer) tests, sample methods and analyses, or likely presence of hydrocarbon sources in the unsaturated zone were flagged and eliminated from further consideration. Of note, US EPA (2013) found no significant difference (less than 1%) in screening distance estimates derived using higher- and lower-ranked soil-gas data for a population of UST sites contained in the US EPA database with LNAPL sources. The databases are publically available at http://www.epa.gov/oust/cat/pvi/index.htm#group.
The databases contain information on:
soil-gas probe location (depth below land surface)
depth to groundwater or free product (LNAPL)
reported observed presence of free product (LNAPL)
type of hydrocarbon (gasoline/petrol and/or diesel fuel)
soil-gas concentrations: BTEX); naphthalene (US EPA 2013 only); hexane; heptane; TPH; and fixed gases (O2, carbon dioxide [CO2], and CH4)
land surface condition (sub-foundation, bare ground, and asphalt)
building foundation size
facility type (UST, terminal, and refinery)
Information pertaining to the released hydrocarbon product type was not always well-documented and thus assumed to be variable (i.e., gasoline, diesel, or mixtures of both) even though the majority of petroleum UST sources were reported as gasoline. The gasoline composition was unknown and assumed to vary with respect to fuel oxygenate composition, given the relatively broad time span for data collection (1995 to 2011) and vast geographic retail markets (North America and Australia) covered in the data base. The gasoline sources are assumed to contain up to 15% vol/vol MTBE and 10% vol/vol ethanol.
The combined data base was processed to remove:
all non-UST (refinery, terminal, or pipeline) site data—to focus the analysis on petroleum UST sites;
soil-gas samples collected from unsaturated-zone locations designated as containing fractured porous media—to eliminate data potentially affected by preferential vapor transport through fractures;
paired soil-gas and groundwater samples collected (1) within 20 ft (6 m) of potential unsaturated-zone petroleum sources (UST systems and boreholes where LNAPL was observed in the unsaturated zone) and (2) in designated “source area” locations—to minimize potential effects from vapor and residual-phase LNAPL sources in the unsaturated zone on the interpretation of hydrocarbon attenuation;
soil-gas data with paired benzene concentrations in groundwater below reporting levels (less than 1 to less than 40 g/L)—to restrict the analysis to sites with meaningful sources of benzene in groundwater; and
data of unacceptable quality based on review of QA/QC data—to limit inclusion of poor or suspect data in the analysis.
The resulting dataset comprised 828 paired and concurrent measurements of soil gas (hydrocarbon and fixed gas—O2, CO2, CH4) and groundwater from 120 UST sites and 332 sampling locations. Approximately 37% of the soil-gas samples were associated with unconsolidated sands, gravels, and fill material; the remaining 63% comprised silts, silty clay, and clay. Eleven percent of the soil-gas samples were collected from locations below building foundations (i.e., below concrete floors in contact with soils), 44% below pavement (e.g., concrete, asphalt), and 45% below bare ground.
The soil-gas data was further classified by the petroleum hydrocarbon source type, either LNAPL or dissolved-phase. Residual- and free-phase hydrocarbon sources were defined using the following criteria:
current or historical presence of LNAPL (including sheens) in proximate groundwater or soil;
benzene concentrations more than 15,000 µg/L in adjacent groundwater samples; and
n-hexane concentrations in soil gas more than 100,000 µg/m3 (This soil-gas concentration equates to less than 0.02% of the saturated gas phase concentration in equilibrium with unweathered gasoline as defined by Baehr .)
The benzene and n-hexane criteria were purposely chosen as upper-bound concentrations for dissolved-phase petroleum sources to impart conservatism in the derivation of dissolved-phase screening distances. As such, some of the soil-gas population attributed to dissolved-phase sources may include soil-gas data associated with residual-phase LNAPL sources. In particular, the 15,000 µg/L criterion for benzene in groundwater is only slightly less than the effective aqueous solubility for benzene, assuming groundwater is in equilibrium with unweathered gasoline containing 1% vol/vol benzene for the period of data collection. Actual constituent concentrations in groundwater rarely approach effective solubility limits, however, (Newell et al. 1995) due to weathering and dilution during sampling. The n-hexane soil-gas concentration criterion (100,000 µg/m3) is over an order of magnitude higher than the maximum n-hexane concentration measured in 86 soil-gas samples (7000 µg/m3) observed above 20 separate dissolved-phase hydrocarbon sources (US EPA 2013). The n-hexane criterion only affected a small population of soil-gas samples (5 of 31) where n-hexane concentrations were reported. After applying these criteria, the LNAPL dataset comprised 467 soil-gas measurements from 73 sites and 204 sample locations; the dissolved-phase dataset comprised 261 soil-gas measurements from 47 sites and 128 sample locations.
High-methane concentrations in soil gas could also be used as a metric to differentiate LNAPL and dissolved-phase sources given the latter are unlikely to generate extensively anaerobic conditions in the unsaturated zone. A methane criterion was not established in this case, however, because of uncertainty in defining a soil-gas concentration attributable to dissolved-phase or LNAPL sources.
Derivation of vertical screening distances for dissolved-phase and LNAPL sources required knowledge of (1) the benzene vapor concentration measured at a specified soil-gas sample depth below-ground surface, and (2) the vertical distance in the unsaturated zone between the soil-gas sample depth and the top of the petroleum hydrocarbon source. For dissolved-phase sources, the top of source was defined as the water table elevation recorded in an adjacent monitoring well at the time (or nearest the time) of soil-gas sampling. For LNAPL sources, the top of the source (smear zone) was defined from the upper residual-phase LNAPL boundary inferred from boring logs and maximum water table elevations recorded in adjacent groundwater monitoring wells. Approximately 39 and 61% of the LNAPL and dissolved-phase datasets, respectively, were non-detect for benzene in soil gas at specified reporting limits ranging from 0.1 to 370,000 µg/m3. Approximately 70% of the non-detect benzene data were below a reporting limit of 10 µg/m3.
Kaplan-Meier statistics (Kaplan and Meier 1958) were used to define the probability that benzene soil-gas concentrations are less than specified screening-level concentrations of concern for vapor intrusion (Climit = 30, 50, and 100 µg/m3) at finite vertical separation distances above the source (e.g., ≥ 0, ≥ 2 … ≥ 10 ft). Kaplan-Meier statistics are appropriate for analyzing large populations of non-detect soil-gas data with variable reporting limits (Helsel 2005, 2006; Singh et al. 2006). The method is nonparametric and thus does not require that the data fit a particular distribution (e.g., normal, log-normal). Application of the method resulted in cumulative distributions of benzene concentrations in soil gas for each population of data falling within each specified vertical separation distance range for which a percentage of concentration data ≤ Climit was determined. In this article, screening distances were defined such that 95% (or more) of benzene soil-gas concentration data were less than or equal to specified screening-level concentrations of Climit = 30, 50, and 100 µg/m3. These Climit values are consistent with the median benzene soil-gas screening limit (36 µg/m3) recommended by 21 of 35 U.S. states with vapor intrusion guidance (Eklund et al. 2012). Statistical parameters (mean, standard deviation, median, upper 95th confidence limits, and 25th and 75th percentiles) were also calculated for each specified source-separation distance and Climit value.
Dissolved-Phase Petroleum Hydrocarbon Sources
Measured benzene concentrations in soil gas above dissolved-phase sources are plotted in Figure 3 vs. the vertical separation distance in the unsaturated zone between the soil-gas sampling location and dissolved-phase source (water table). Nearly half (47%) of the soil-gas data were associated with relatively high-benzene concentrations in groundwater ranging between 1000 and 14,400 µg/L (Figure 3, histogram inset). The two highest measured benzene concentrations in soil gas (1000 and 1300 µg/m3) occurred at a distance 1.5 ft (0.5 m) above a perched-water unit, where dissolved-phase concentrations are unknown and the TPH concentration in soil (250 mg/kg) indicates a possible LNAPL source (Alaska DEC 2011). Including these two potential misidentified data points, more than 95% of the benzene soil-gas concentrations were less than or equal to the specified soil-gas screening levels of 30, 50, and 100 µg/m3, as illustrated in Figure 4 and summarized in Table 1. The probability of exceeding the thresholds for vapor intrusion above a dissolved-phase petroleum hydrocarbon source at a UST site is thus ≤5%. Consequently, vapor intrusion is not likely to occur where dissolved-phase sources are present unless groundwater (including capillary fringe) comes in contact with a building foundation. A similar screening distance (0 ft or 0 m) was derived in the US EPA (2013) study for UST sites based on analysis of a significantly smaller population of empirical data and similar specified soil-gas screening criteria (50 and 100 µg/m3). The vertical screening distance for dissolved-phase sources is also relatively insensitive to Climit for the specified values (Climit = 30 to 100 µg/m3) (Figure 4) as only 8% of the benzene soil-gas data fall within the Climit concentration range.
Table 1. Kaplan-Meier Conditional Probability Statistics for Benzene Soil-Gas Concentration Data as a Function of Distance Above Dissolved-Phase and LNAPL Hydrocarbon Sources
Distance above Source (ft)
Probability That Benzene Soil-Gas Concentration Is Ä Specified Screening-Level Concentration—Climit (%)
Number of Samples
Note: Statistics are included for benzene soil-gas concentration data collected above LNAPL hydrocarbon sources and below building foundations for comparative purposes.
LNAPL source—collected below building foundation (sub slab)
LNAPL source—collected below pavement
The vertical screening distance is also relatively independent of the source concentration of benzene in groundwater, as implied by the soil-gas/groundwater data plotted in Figure 5. The measured benzene concentrations in soil-gas are poorly correlated with benzene concentrations measured in co-located or nearby groundwater monitoring wells and approximately 2 to 5 orders of magnitude lower than soil-gas concentrations predicted by equilibrium gas/liquid-phase partitioning according to Henry's Law. Similar findings are reported by US EPA (2013), in particular, for dissolved-phase concentrations less than 5 mg/L. The poor correlation between soil gas and groundwater can be attributed to a number of factors, including (1) challenges measuring dissolved-phase concentrations at the water table (i.e., the true source of hydrocarbon vapors in the unsaturated zone) using conventional groundwater monitoring wells screened across the water table, (2) hydrocarbon biodegradation between the water table and the lowermost soil-gas sampling location, and (3) potential differences in actual vs. interpolated dissolved-phase source concentrations at locations where groundwater and soil-gas samples are not co-located. The latter is not considered a significant issue, however, because 82% of reported groundwater samples were collected within 10 d of the paired soil-gas samples and 93% of reported groundwater samples were collected within 20 ft (6 m) of paired soil-gas samples. It is also important to note that soil-gas and groundwater concentrations can be poorly correlated in the presence of residual-phase LNAPL sources that act as additional hydrocarbon sources above the water table (Zemo 2006). Collectively, the results imply that shallow groundwater concentration measurements may have limited value as indicators or predictors of indoor-air concentrations at petroleum UST sites. Vapor intrusion risks may be better defined through detailed source characterization, including the identification of source type (i.e., residual-phase LNAPL) and source-separation distances during the initial phases of site assessment (e.g., borehole/monitoring well installation).
LNAPL Hydrocarbon Sources
Benzene concentrations in soil gas identified as originating from LNAPL hydrocarbon sources are plotted in Figure 6 vs. the vertical separation distance between the soil-gas sampling location and hydrocarbon source (assumed to be the highest water table elevation). The maximum observed benzene soil-gas concentrations are several orders of magnitude higher than those associated with dissolved-phase sources. Greater than 95% of the benzene soil-gas concentrations are ≤Climit values (30, 50, or 100 µg/m3) beyond a distance of approximately 13 ft (4 m) above the LNAPL source (Figure 7 and Table 1). Again, the vertical screening distance is relatively insensitive to the benzene soil-gas screening-level concentration (Climit = 30 to 100 µg/m3), as only 12% of concentration data fall within this range. Similar screening distances (13 to 15 ft or 4 to 5 m) are reported by US EPA (2013) for LNAPL sources at UST sites.
Screening distances are also relatively unaffected by the presence or absence of a building foundation (sizes ranging from 400 to 10,000 ft2 where reported) at land surface (Table 1). Some sensitivity to pavement is observed for vertical source-separation distances less than 13 ft (Table 1). This sensitivity may be attributed to (1) relatively higher hydrocarbon concentrations in soil gas (noting that sub-pavement samples are likely collected on-site and near release areas), and (2) limitations on O2 ingress from the atmosphere (noting that sub-pavement soil-gas samples are likely collected in areas with more laterally extensive surface cover). There were insufficient soil-gas data to assess whether pavement would significantly affect the screening distance estimate beyond a 13-ft (4-m) vertical source-separation distance, however. US EPA (2013) also reports some sensitivity of screening distances to pavement (for non-UST sites), but no sensitivity to building foundations. The vertical screening distances derived from this analysis are thus considered generally applicable for both current and future building construction scenarios at petroleum UST sites provided source-building separation distances remain relatively unchanged by redevelopment.
Oxygen and Petroleum Hydrocarbon Distributions in the Unsaturated Zone
The distributions of O2 in the unsaturated zone shown in Figure 8a and 8b generally support the vertical screening distance estimates derived from benzene data for dissolved-phase (0 ft or 0 m) and LNAPL (13 ft or 4 m) hydrocarbon sources, respectively. In particular, more than 97% of the 206 O2 soil-gas concentrations measured in the unsaturated zone above dissolved-phase sources exceed 4% vol/vol, which as previously noted represents a relatively conservative lower-bound for aerobic biodegradation. Aerobic conditions are observed even though approximately 35% of the O2 soil-gas samples were collected below building foundations and pavement where O2 ingress from the atmosphere is thought to be more limited. This finding is consistent with the modeling of Abreu et al. (2009) which shows that O2 concentrations in the unsaturated zone are relatively unaffected by benzene vapor sources less than 10 mg/L (10,000,000 µg/m3) located 3 m or more from building foundations. Aerobic conditions are also generally observed at distances more than 13 ft (4 m) above LNAPL sources. Approximately 93% of 37 O2 soil-gas concentrations are more than 4% vol/vol. Although the population of data is small (n = 37), aerobic conditions are generally observed in the unsaturated zone at much shorter source-separation distances than predicted (23 ft or 7 m) by transport modeling (Abreu et al. 2009). The model assumptions used to simulate biodegradation in the unsaturated zone and/or advective flow of O2 around and through building foundations may thus be conservative. It is important to note, however, that benzene concentrations in soil gas can still exceed vapor concentrations of concern for vapor intrusion, even though conditions in the unsaturated zone are conducive for aerobic biodegradation. Over 25% of the benzene soil-gas data exceed the 30 µg/m3 screening level at locations where O2 concentrations in soil gas exceed 4% O2 vol/vol.
Co-located O2 and TPH concentrations in soil gas are plotted in Figure 9. In certain cases, the data are seemingly inconsistent with general aerobic biodegradation relations (i.e., low-hydrocarbon [TPH]/high-O2 concentrations and vice-versa). This finding is not unexpected given that aerobic reaction stoichiometry is defined by mass flux (i.e., concentration gradients), not concentration. For similar mass fluxes, O2 soil-gas concentration gradients on the order of 1 to 2% per foot can correspond with inverse TPH concentration gradients on the order of 1E6 to 1E7 µg/m3 per foot assuming approximately 4,400,000 µg/m3 of TPH is consumed for every 1% (~275,000,000 µg/m3) of available O2. Detectable TPH concentrations in soil gas less than 1E7 µg/m3 coincident with O2 concentrations more than 4% vol/vol may thus be indicative of soil-gas samples collected within the aerobic reaction front.
Only a small percentage (~7%) of the soil-gas data shown in Figure 9 exhibit TPH concentrations more than 1E7 µg/m3 and O2 concentrations more than 4% vol/vol. These data do not intuitively fit the aerobic biodegradation paradigm and may indicate sampling bias. Approximately one-half of these data were obtained at one site (Hal's Chevron) with low-permeability soil (clayey silt) where entrainment or channeling of atmospheric air into soil-gas samples (McAlary et al. 2009) may have occurred during active (pumped) sampling. These specific data had no bearing on the derivation of screening distances for LNAPL, however, because the soil-gas samples were collected within only a few feet of the hydrocarbon source. Finally, an underdetermined error may also be affecting the quantification of relatively low-O2 soil-gas concentrations. Nearly half (42%) of the O2 data reported in Figure 9 at concentrations in soil gas less than 10% vol/vol are associated with calculated total gas pressures of hydrocarbon vapors and fixed gases less than 95% vol/vol. Additional analysis and reporting of relatively conserved gases (Nitrogen: N2; Argon: Ar) along with O2, CO2, and CH4 in soil gas may help resolve this issue.
The resultant screening distances (0 ft [0 m] for dissolved-phase and 13 ft [4 m] for LNAPL petroleum sources) are expected to be applicable at most petroleum UST sites given the wide range of environmental (i.e., soil types, surface conditions [e.g., building foundations and pavement], subsurface temperatures, and fuel types/compositions) intrinsic to the soil-gas data base used in their derivation. Nevertheless, additional validation may be warranted at certain sites not captured in the data base or by this statistical analysis where O2 availability for aerobic biodegradation may be more limited. These sites include those with high-organic-matter soils (e.g., peat), areally extensive building foundations at land surface (e.g., large commercial facilities or apartment complexes), high-ethanol content fuel (e.g., E85) releases, historic releases of gasoline containing lead scavengers (EDB and EDC), and sites with excessively dry unsaturated-zone soils (exhibiting water saturations less than the wilting point) capable of affecting biodegradation rates (Guymon 1994). The latter is not expected to be a common occurrence given that unsaturated-zone soils are seldom drier than the wilting point, with the exception of near surface soil layers under arid conditions that are open to the atmosphere (DeVaull 1997). Sites with very low-permeability soils (e.g., wet clays) may also affect O2 availability; however, low-permeability soils are also predicted to significantly retard vapor transport (US EPA 2012a). Any sensitivity to low-permeability soil may actually be inherent in the screening distances calculated in this study given that 63% of the analyzed soil-gas samples were from sites with predominantly low-permeability soil (silts, silty clays, and clays).
The screening distances may also not be applicable at the sites and locations filtered from the data base, including:
non-UST (e.g., refinery, terminal, and pipeline) petroleum release sites, especially those with larger volume (e.g., hundreds of thousands to millions of gallon) releases than might be encountered at typical UST sites,
fractured rock unsaturated-zone systems (i.e., fractured geologic media) between the petroleum hydrocarbon source and a building foundation, and
locations within 20 ft or 6 m of a former or current UST systems (e.g., tanks and dispensers) or known release areas where there is increased potential for encountering unsaturated-zone LNAPL sources (relevant for vertical screening distance application only).
Two critical factors required for screening distance application are the vertical separation distance between a petroleum hydrocarbon source and building foundation and the hydrocarbon source type (dissolved-phase or LNAPL). The screening distance may be based on the highest recorded or predicted water table elevation and the elevation of the base of the building foundation. For dissolved-phase sources, it may be practical to set buffer distances (e.g., 5 ft or 1.5 m) to account for uncertainties in water table elevation (including capillary fringe) at sites where the temporal variability is unknown or historical records are inadequate. Water table variability may be less of an issue for LNAPL sources where the top of the residual-phase source (smear zone) is likely to become established over time at historic highs in the water table elevation. Detailed characterization will, however, be required during the initial phases of a site investigation (e.g., borehole/monitoring well installation) to identify residual-phase LNAPL sources. It may be worthwhile in such cases to try and identify LNAPL using a multiple lines of evidence approach using several of the indicator criteria provided in Table 2.
Table 2. Direct and Indirect Indicators of Residual-Phase LNAPLa
Measures and Screening Values
Note: Concentrations lower than the reference values can also be indicative of LNAPL sources.
Adapted from Garg and Beckett (2009, written communication).
•Hydrocarbon and CO2 concentrations in soil gas that show no decrease (or O2 concentrations that show no increase) or remain relatively constant with distance from source
The vertical screening distances derived in this study are also expected to apply laterally in the absence of (1) mobile/expanding hydrocarbon plumes or (2) significant hydrogeologic barriers (e.g., perched-water tables or low-permeability soil lenses) or preferential pathways that could potentially enhance lateral hydrocarbon vapor migration (Abreu and Johnson 2006; http://www.epa.gov/oust/cat/pvi/index.htm). Preferential pathways are not commonly reported at petroleum UST sites, however. The potential for vapor intrusion to occur from dissolved-phase and LNAPL hydrocarbon sources displaced more than 0 ft (0 m) and13 ft (4 m), respectively, from building foundations is therefore limited. Lateral screening distances of this magnitude, however, provide little to no safeguard against hydrocarbon sources potentially contacting or being present below building foundations at depths less than the vertical screening distances should uncertainty exist in plume delineation or groundwater flow direction. A lateral off-set buffer distance could be invoked from the edge of a dissolved-phase source to the building foundation in such cases.
Finally, the results of this study can aid the design of additional site characterization strategies at petroleum UST sites in the event that screening distances cannot be met. For example, hydrocarbon concentrations in soil gas associated with dissolved-phase sources (in the absence of residual-phase LNAPL) are likely to be below concentrations of potential concern and relatively independent of constituent concentrations in shallow groundwater. In such cases, it may be practical to focus additional monitoring or data collection on the potential for groundwater (including capillary fringe) to contact a building foundation rather than groundwater sampling. Additional soil-gas sampling of hydrocarbons and signatures of hydrocarbon biodegradation (O2, CO2, and CH4) between the source and building foundation may be beneficial at sites with LNAPL sources.
Several field and model studies have shown that current screening methods for petroleum hydrocarbon vapor intrusion that do not account for biodegradation can drive unnecessary sampling, data interpretation, and vapor mitigation measures. A statistical analysis of empirical soil-gas data collected at over a 100 UST sites in North America and Australia provides an opportunity to improve the screening paradigm. The data base encompasses a 16-year period (1995 to 2011) of sample collection and a wide range of environmental conditions and geographic regions. Key results of the analysis are summarized as follows:
hydrocarbon source type (e.g., dissolved-phase or LNAPL) and the vertical separation distance between the source and building foundation are key metrics defining the vapor intrusion risk at petroleum UST sites;
dissolved-phase hydrocarbon sources in the absence of residual-phase LNAPL are unlikely to pose a significant risk for vapor intrusion unless groundwater directly contacts a building foundation (i.e., more than 95% of the benzene soil-gas concentrations associated with dissolved-phase sources are less than a specified soil-gas screening limit of 30 µg/m3);
the poor correlation of benzene concentrations in shallow groundwater and deep soil gas implies that use of shallow groundwater concentration measurements as predictors of indoor-air concentrations may be of little benefit in vapor intrusion risk assessment at petroleum UST sites;
the risk of vapor intrusion above an LNAPL source is relatively negligible beyond a vertical screening distance of 13 ft (4 m) (e.g., more than 95% of the benzene soil-gas concentrations associated with LNAPL sources are less than a specified soil-gas screening limit of 30 µg/m3 beyond a 13 ft [4 m] source-separation distance);
although O2 concentrations in soil gas generally support the screening distances derived from benzene soil-gas data and the aerobic biodegradation reaction model, more than 25% of benzene concentrations in soil gas exceed a specified risk-based screening level of 30 µg/m3 at unsaturated-zone locations assumed to be aerobic (i.e., where O2 concentrations in soil gas more than 4% O2 vol/vol); and
the screening distances derived in this study are relatively insensitive to a) the presence/absence of building foundations at land surface (which may have relevance for the future redevelopment provided the new foundation depth does not reduce the vertical source-separation distance) and b) the specified benzene soil-gas screening limits (30, 50, and 100 µg/m3) at which vapor intrusion risks are assumed to be negligible.
The screening distances derived in this study, including vertical and lateral buffer distances to account for uncertainty in the source depth (water table elevation), exact edge of a hydrocarbon plume, or groundwater flow direction are consistent with those determined in other studies using independent methodologies (Davis 2009; Peargin and Kolhatkar 2011; Wright 2011; US EPA 2013). The screening distances are also deemed broadly applicable for a wide range of petroleum UST sites. Screening distance application will necessitate detailed source characterization during the initial phases of site investigation (borehole development, monitoring well installation) to determine the hydrocarbon source type (dissolved-phase or LNAPL) and separation distance between the source and building foundation. Certain environmental or site conditions not reflected in the data base or captured in the analysis may preclude screening distance application at petroleum UST sites, however. These conditions include preferential pathways (e.g., underground sewers and fractured rock), high-organic-rich (e.g., peat) soils or excessively dry soils in arid environments, large building foundations (e.g., associated with industrial/commercial and apartment complexes), high-ethanol content fuel (e.g., E85) releases, and historical gasoline release containing lead scavengers (EDB and DCA). In addition, vertical screening distances may not be appropriate for application within 20 (6 m) of unsaturated zone sources (e.g., residual-phase LNAPL, operating petroleum UST system) given that soil-gas data collected within this separation distance were excluded from the analysis. Further research on the potential effect of these factors on screening distances and petroleum vapor intrusion at UST sites is recommended. The results of this study can also be used to support further site assessment strategies in cases where the screening distances are not met. It may be advantageous, for example, to focus additional characterization on the potential for groundwater to contact a building foundation rather than on additional groundwater or soil-gas sampling at locations with dissolved-phase sources or on hydrocarbon attenuation in the unsaturated zone (i.e., soil-gas sampling) at locations with LNAPL sources. Successful implementation of screening distance methodology is expected to lessen the occurrence of unnecessary data collection at petroleum UST sites and promote more effective and sustainable use of limited resources.
The authors gratefully acknowledge the U.S. EPA Office of Underground Storage Tanks Vapor Intrusion Working Group, RTI International, and the American Petroleum Institute for helping support this effort.
Matthew A. Lahvis, corresponding author, is at Shell Global Solutions (US), Shell Technology Center Houston, 3333 Hwy 6 South, Houston, TX 77082;Matthew.Lahvis@shell.com.
Ian Hers, is at Golder Associates Ltd., 4260 Still Creek Drive, Burnaby, British Columbia, Canada V5C 6C6;firstname.lastname@example.org.
Robin V. Davis, is at Utah Department of Environmental Quality, 195 N. 1950 W. Salt Lake City, UT 84116;email@example.com.
Jackie Wright, is at Environmental Risk Sciences Pty Ltd, 6 Wilshire Ave, Carlingford, NSW, 2118, Australia;Jackie@enRisks.com.
George E. DeVaull, is at Shell Global Solutions (US), Shell Technology Center Houston, 3333 Hwy 6 South, Houston, TX 77082;George.DeVaull@shell.com.