less efficient than originally envisioned (Mackay and Cherry, 1989). Second, most common organic pollutants in the subsurface have low solubilities in water and tend to remain as either a separate organic phase liquid in the subsurface (nonaqueous phase liquid or NAPL) as in the case of chlorinated solvents or a separate solid phase as where chemical explosives have precipitated in the subsurface. Organic liquids that are denser than water are referred to as dense non-aqueous phase liquids (DNAPLs). During the late 1980s, it was recognized that the presence of DNAPLs made a site particularly difficult to remediate (Feenstra and Cherry, 1988; Mackay and Cherry, 1989; Mercer and Cohen, 1990; NRC, 1994). Before it was understood that DNAPLs commonly exist in source areas, it was assumed that by removing a few pore volumes of contaminated groundwater, the majority of the total contamination could be extracted.
At sites where they are present, separate phase or sorbed contaminants serve as a long-lived contamination source. That is, groundwater that flows through the volume of subsurface containing the contaminant—termed the source zone—will be contaminated by the small amount of contaminant that dissolves. This suggests that groundwater remediation to background levels will not be achieved unless the contaminant source is removed or physically isolated from flowing groundwater (NRC, 1994). Unfortunately, due to a lack of effective characterization tools and the tendency of DNAPLs to have a spatially limited but extremely heterogeneous distribution, it is very difficult to find contaminant sources within the subsurface. In addition, although numerous new technologies have been developed to remediate source zones, the difficulty in evaluating these technologies (due to the lack of data from pilot studies) makes prediction of their effectiveness for full-scale applications problematic.
Several NRC reports extend the findings of the 1994 report on pump-and-treat systems to include more comprehensive analysis and encompass new remediation technologies (NRC, 1997, 1999a, 2003). These reports have noted the general paucity of data available for evaluating remediation technology performance (including technologies for DNAPL sites). These and many other recent studies (e.g., ITRC, 2000, 2002; SERDP, 2002; EPA, 2003) have demonstrated that restoration of sites with DNAPL contamination to pre-contamination levels is rare and may not be practically achievable. Indeed, there are no reported cases of large DNAPL sites where remediation has restored the site to drinking water standards. At this time, most DNAPL sites have pump-and-treat systems in place to contain the dissolved phase plumes and thus minimize risk to the public. At only a small fraction of these sites has remediation of the DNAPL source actually been attempted.
Layered onto this issue of technical impracticability are the opinions of stakeholders, including those who live or work near contaminated sites, as well as the high cost associated with remediation efforts. There is often pressure from the public to remediate when pollution is found. This pressure to clean up sites is contrasted by the fact that remediation technologies for DNAPL sites are under-