Main Content

August 2018 Article

In 2012, EPA released a report titled, “EPA’s Vapor Intrusion Database: Evaluation and Characterization of Attenuation Factors for Chlorinated Volatile Organic Compounds and Residential Buildings”. The information in the database was critical to EPA’s derivation of soil-gas and groundwater Vapor Intrusion Screening Levels (VISLs).

Vapor Intrusion (VI) is the process in which chemicals associated with gasoline, dry cleaning fluid, and various other volatile substances, migrate from soil or groundwater sources and enter buildings. Readers of this column will recognize the term “attenuation”, which is the decline in concentration as vapors migrate from source to receptor. Understanding how much attenuation takes place during vapor migration is useful in two ways. If we understand the amount of attenuation between, say, a groundwater source and an overlying building, we can use groundwater concentration data to predict indoor-air vapor concentrations. Groundwater is usually far less costly to analyze than indoor air. Conversely, if we start with a chemical’s Regional Screening Level (RSL) or some other maximum allowable concentrations in indoor air, we can use our understanding of attenuation to determine the screening levels for soil gas and groundwater.

In EPA’s 2012 Draft VI guidance, EPA derived soil-gas and groundwater screening levels by modeling the amount of vapor attenuation. They estimated that vapor concentrations in indoor air equaled approximately 1/10^th of concentrations directly beneath the floor in subslab soil gas, and expressed this ratio as an Attenuation Factor (AF) of 0.1. Deep soil gas and groundwater AFs were assumed to equal 0.01 and 0.001, respectively. Accordingly, subslab soil-gas, deep soil-gas, and groundwater screening levels equaled indoor-air screening levels times 10, 100, and 1,000, respectively. (Groundwater VISLs were also adjusted for groundwater-to-air vapor partitioning using Henry’s constants, as discussed in the January 2018 Focus on the Environment newsletter). Due to the complexities of soil-to-indoor-air attenuation, EPA only briefly attempted to predict VI from soil, with their Johnson & Ettinger spreadsheets, last updated in 2004.

To refine their understanding of AFs, EPA compiled data from their own investigations and from outside investigators, which would allow them to compare indoor vapor concentrations to subsurface concentrations. The dataset was limited to chlorinated volatile organic compounds (CVOCs) from residential settings. The results were published in EPA’s 2012 Vapor Intrusion Database publication.

A graph showing the ratio of indoor air to subslab soil gas, (EPA’s Figure 12), is shown below.

Had the AF for subslab soil gas actually equaled 0.1, all of the data points would have fallen on the second diagonal line from the top. Instead, the points fall, on average, between the third and fourth diagonal lines, corresponding to an AF of 0.003. So one’s best guess of indoor air vapor concentrations would equal subslab concentrations x 0.003, and the subslab screening levels would equal indoor-air screening levels / 0.003. Because there’s a fair amount of scatter in the data, EPA updated their default subslab soil-gas AF and attenuation factors using a ratio of 0.03, to allow for uncertainty.

The ratio of indoor air to exterior soil gas, which is typically collected below 5 feet with a drilling rig, (EPA’s Figure 20), is shown graphically below.

The exterior soil-gas plot shows approximately the same 0.003 ratio of indoor air / soil gas concentrations as subslab soil gas. Accordingly, EPA revised both the subslab and the exterior soil-gas default AFs to 0.03, so that all soil-gas screening levels equal indoor-air screening levels / 0.03. Notice that the exterior soil-gas plot also shows far more scatter than the subslab-soil-gas plot, which suggests that subslab is better for predicting VI. Subslab soil gas is also cheaper and easier to collect than exterior soil gas, if you don’t mind putting holes in the floor.

The ratio of indoor air to vapor concentrations directly above groundwater, (EPA’s Figure 16), can be seen in the plot below.

Had the AF for groundwater vapor actually equaled the default AF of 0.001, the data points would have fallen on the fourth diagonal line from the top. Instead, the data points surround the fifth line, corresponding to an actual groundwater AF of 0.0001. EPA kept the default AF of 0.001 to allow for uncertainty.

Thus, EPA’s indoor-air VISLs are divided by 0.03 (multiplied x 33) to derive soil-gas VISLs, and indoor VISLs are divided by 0.001 (multiplied x 1,000) to derive groundwater VISLs.

But there’s more to the story. Besides EPA’s tenfold multiplier to account for uncertainty, other factors make soil-gas and groundwater VISLs unrealistically low in many situations. Among them:

• The VI database was based on residential data. Commercial/Industrial (CI) buildings often have far higher indoor-air exchange rates, and higher ceilings, which dilutes soil gas and reduces VI more than in residential settings.
• The database was built on data from CVOCs, which resist chemical breakdown. The attenuation of Petroleum Hydrocarbons (PHCs) is often far greater.
• Most of EPA’s groundwater data did not meet the Agency’s own criteria for inclusion in the database. Yao, Verginelli, Suuberg, and Eklund’s article in Groundwater Monitoring & Remediation, (2018) describes how 70% of the indoor-air-to-groundwater pairs were separated by more than 30 meters (98 feet) laterally. EPA’s 2015 VI Guidance considers the VI lateral separation distance, or “footprint”, to be 100 feet from the source for CVOCs, and 30 feet for PHCs (EPA’s 2015 Petroleum VI Guidance). Yao, et al. estimate that vapor attenuation from groundwater is 10 x greater than EPA’s estimate.
• VI-risk is usually estimated on the basis of the highest subsurface concentration. Actual VI results from a mix of lower and higher soil-gas vapor concentrations, especially near source areas typical of industrial facilities.

Accordingly, in our experience, the risk of VI based on subsurface data is hugely exaggerated, at least at CI sites. Assuming that indoor-air vapor concentrations equal the maximum subslab concentration x 0.03 overestimates risk by 1,000 or more. Also, consider that the risk from constituents in indoor air may already be exaggerated to account for uncertainties. For example, the cancer risk for tetrachloroethene (perchloroethene, PCE) is exaggerated by 1,000, after applying three uncertainty factors of 10x. It’s appropriate to take subsurface contamination seriously and investigate the risk of VI, but an exceedance of subsurface VISLs by a factor of 10x or 100x, especially at CI sites, rarely indicates a VI problem, and treating it as a crisis brings fear to occupants and unnecessary expense to responsible parties.