Federal Facilities Forum Issue

Transcription

1 United States Offie of Offie of Solid Environmental Protetion Researh and Waste and Ageny Development Emergeny Response EPA/540/R-97/501 November 1996 Federal Failities Forum Issue FIELD SAMPLING AND SELECTING ON-SITE ANALYTICAL METHODS FOR EXPLOSIVES IN SOIL A. B. Crokett, H. D. Craig, T. F. Jenkins, and W. E. Sisk The Federal Failities Forum is a group of It is imperative that any persons working U.S. Environmental Protetion Ageny on sites believed to be ontaminated with (EPA) sientists and engineers, representing explosive residues thoroughly familiarize EPA regional offies, ommitted to the themselves with the physial and toxi identifiation and resolution of issues affet- properties of the materials potentially ing the haraterization and remediation of present and to take all measures as may be federal faility Superfund and Resoure prudent and/or presribed by law to protet Conservation and Reovery At (RCRA) sites. life, health, and property. This publiation Current forum members are identified in the is not intended to inlude disussions of the text. The forum members identified a need to safety issues assoiated with sites ontamprovide Remedial Projet Managers (RPMs) inated with explosive residues. Examples of and other federal, state, and private personnel safety issues to be onsidered inlude but are working on hazardous waste sites with a not limited to: explosion hazards, toxiity of tehnial issue paper that identifies sreening seondary explosives, and/or personal proedures for haraterizing soils ontaminated protetive equipment. Information pertaining with explosive and propellant ompounds. to these onerns an be found in Roberts and Forum members Sott Marquess and Paul Hartley (1992) and Yinon (1990). Speifially, Leonard provided tehnial guidane and this paper is not intended to serve as a guide diretion in the development of this Issue paper for sampling and analysis of unexploded and other Forum members provided omments. ordnane, bulk high explosives, or where seondary explosives onentrations in soil This paper was prepared by A. B. Crokett, exeed 100,000 mg/kg (10%). These H. D. Craig, T. F. Jenkins, and W. E. Sisk. onditions present a potential detonation Support for this projet was provided by the hazard, and as suh, safety proedures and EPA National Exposure Researh Labora- safety preautions should be identified tory's Charaterization Researh Division before initiating site haraterization ativwith the assistane of the Superfund Projet's ities in these environments. Finally, this Tehnology Support Center for Monitoring paper does not address primary explosives or and Site Charaterization. For further initiating ompounds, suh as lead azide, lead information, ontat Ken Brown, Tehnology styphnate, or merury fulminate, whih are Support Center Diretor, at (702) , extremely unstable and present a substantial Alan B. Crokett at (208) , or Harry safety risk at any onentration. Craig at (503) Idaho National Engineering and Environmental Laboratory, Lokheed Martin Idaho Tehnologies Company 2 U.S. Environmental Protetion Ageny, Region 10 3 U.S. Army Cold Regions Researh and Engineering Laboratory 4 U.S. Army Environmental Center Tehnology Support Center for Monitoring and Site Charaterization, National Exposure Researh Laboratory Charaterization Researh Division Las Vegas, NV Tehnology Innovation Offie Offie of Solid Waste and Emergeny Response, U.S. EPA, Washington, D.C. Walter W. Kovalik, Jr., Ph.D., Diretor Printed on Reyled Paper 381asb96

2 PURPOSE AND SCOPE Historial disposal praties from manufaturing, The purpose of this issue paper is to provide spills, ordnane demilitarization, lagoon disposal of guidane to Remedial Projet Managers regarding explosives-ontaminated wastewater, and open burn/ field sampling and on-site analytial methods for open detonation (OB/OD) of explosive sludges, waste deteting and quantifying seondary explosive om- explosives, exess propellants, and unexploded pounds in soils (Table 1). The paper also inludes ordnane often result in soils ontamination. Common a brief disussion of EPA Method 8330 (EPA munitions fillers and their assoiated seondary explosives 1995a), the referene analytial method for the inlude Amatol (ammonium nitrate/tnt), Baratol (barium determination of 14 explosives and o-ontaminants nitrate/tnt) Cylonite or Hexogen (RDX), Cylotols in soil. (RDX/TNT), Composition A-3 (RDX), Composition B (TNT/RDX), Composition C-4 (RDX), Explosive D or This issue paper is divided into the following Yellow D (AP/PA), Otogen (HMX), Otols (HMX/TNT), major setions: (1) bakground, (2) an overview Pentolite (PETN/TNT), Piratol (AP/TNT), tritonal (TNT), of sampling and analysis for explosives in soil, tetrytols (tetryl/tnt), and Torpex (RDX/TNT). (3) data quality objetives, (4) unique sampling design onsiderations for explosives, (5) a Propellant ompounds inlude DNTs and single base summary of on-site analytial methods, and (6) a (NC), double base (NC/NG), and triple base summary of the EPA referene analytial method. (NC/NG/NQ) smokeless powders. In addition, NC is While some setions may be used independently, frequently spiked with other ompounds (e.g., TNT, joint use of the field sampling and on-site DNT, DNB) to inrease its explosive properties. AP/PA analytial methods setions is reommended to is used primarily in Naval munitions suh as mines, develop a sampling and analytial approah that depth harges, and medium to large aliber projetiles. ahieves projet objetives. Tetryl is used primarily as a boosting harge, and PETN is used in detonation ord. Many of the explosives listed in Table 1 are not speifi target ompounds of sreening methods, yet A number of munitions failities have high levels of they may be deteted by one or more sreening soil and groundwater ontamination, although on-site methods beause of their similar hemial struture. waste disposal was disontinued 20 to 50 years ago. Also listed are the explosive and propellant Under ambient environmental onditions, explosives are ompounds targeted by high performane liquid highly persistent in soils and groundwater, exhibiting a hromatography (HPLC) methods inluding EPA resistane to naturally ourring volatilization, biodeg- SW-846 Method 8330, the standard method radation, and hydrolysis. Where biodegradation of TNT required by EPA regions for laboratory onfirm- ours, 2-AmDNT and 4-AmDNT are the most ation. ommonly identified transformation produts. Photohemial deomposition of TNT to TNB ours in the presene of sunlight and water, with TNB being BACKGROUND generally resistant to further photodegradation. TNB is subjet to biotransformation to 3,5-dinitroaniline, whih Evaluating sites potentially ontaminated with has been reommended as an additional target analyte in explosives is neessary to arry out EPA, U.S. EPA Method Pirate is a hydrolysis trans- Department of Defense, and U.S. Department of formation produt of tetryl, and is expeted in Energy poliies on site haraterization and environmental samples ontaminated with tetryl. Site remediation under the Superfund, RCRA, investigations indiate that TNT is the least mobile of Installation Restoration, Base Closure, and Formerly the explosives and most frequently ourring soil Used Defense Site environmental programs. ontamination problem. RDX and HMX are the most Failities that may be ontaminated with explosives mobile explosives and present the largest groundwater inlude, for example, ative and former ontamination problem. TNB, DNTs, and tetryl are of manufaturing plants, ordnane works, Army intermediate mobility and frequently our as ammunition plants, Naval ordnane plants, Army o-ontaminants in soil and groundwater. Metals are depots, Naval ammunition depots, Army and Naval o-ontaminants at failities where munitions proving grounds, burning grounds, artillery impat ompounds were handled, partiularly at OB/OD sites. ranges, explosive ordnane disposal sites, bombing Field analytial proedures for metals, suh as x-ray ranges, firing ranges, and ordnane test and fluoresene, may be useful in sreening soils for metals in evaluation failities. onjuntion with explosives at munitions sites. 2

3 Table 1. Analytial Methods for Commonly Ourring Explosives, Propellants, and Impurities/Degradation Produts. Field Laboratory Aronym Compound Name Method Method Cs N TNT 2,4,6-trinitrotoluene Cp, Ip N TNB 1,3,5-trinitrobenzene Cs, Is N DNB 1,3-dinitrobenzene Cs N 2,4-DNT 2,4-dinitrotoluene Cp, Cs N 2,6-DNT 2,6-dinitrotoluene Cs, Is N Tetryl Methyl-2,4,6-trinitrophenylnitramine Cs N 2AmDNT 2-amino-4,6-dinitrotoluene N 4AmDNT 4-amino-2,6-dinitrotoluene Is N NT Nitrotoluene (3 isomers) N NB Nitrobenzene N Nitramines Cs N RDX Hexahydro-1,3,5-trinitro-1,3,5-triazine Cp, Ip N HMX Otahydro-1,3,5,7-tetranitro-1,3,5,7-tetrazoine Cs N NQ Nitroguanidine Cs G Nitrate Esters Cs NC Nitroellulose Cs *L NG Nitroglyerin Cs *P PETN Pentaerythritol tetranitrate Cs *P Ammonium Pirate/Piri Aid AP/PA Ammonium 2,4,6-trinitrophenoxide/2,4,6-trinitrophenol Cp, Is A Cp = Colorimetri field method, primary target analyte(s). Cs = Colorimetri field method, seondary target analyte(s). Ip = Immunoassay field method, primary target analyte(s). Is = Immunoassay field method, seondary target analyte(s). N = EPA SW-846, Nitroaromatis and Nitramines by HPLC, Method 8330 (EPA 1995a). P = PETN and NG (Walsh unpublished CRREL method). G = Nitroguanidine (Walsh 1989). L = Nitroellulose (Walsh unpublished CRREL method). A = Ammonium Pirate/Piri Aid (Thorne and Jenkins 1995a). *The performane of a number of field methods have not been assessed utilizing "approved" laboratory methods. It is reommended that verifiation of the performane of any analytial method be an integral part of a sampling/analysis projets quality assurane program. 3

4 The frequeny of ourrene of speifi explosives in soils was assessed by Walsh et al. (1993), who ompiled analytial data on soils olleted from 44 Army ammunition plants, arsenals, and depots, and two explosive ordnane disposal sites. Of the 1,155 samples analyzed by EPA Method 8330, a total of 319 samples (28%) ontained detetable levels of explosives. The frequeny of ourrene and the maximum onentrations deteted are shown in Table 2. TNT was the most ommonly ourring ompound in ontaminated samples and was deteted in 66% of the ontaminated samples and in 80% of the samples if the two explosive ordnane disposal sites are exluded. Overall, either TNT or RDX or both were deteted in 72% of the samples ontaining explosive residues, and 94% if the ordnane sites are exluded. Thus, by sreening for TNT and RDX at ammunition plants, arsenals, and depots, 94% of the ontaminated areas ould be identified (80% if only TNT was determined). This demonstrates the feasibility of sreening for one or two ompounds or lasses of ompounds to identify the initial extent of ontamination at munitions sites. The two ordnane sites were predominantly ontaminated with DNTs, probably from improper detonation of waste propellant. The table also shows that NB and NTs were not deteted in these samples; however, NTs are found in waste produed from the manufature of DNT. OVERVIEW OF SAMPLING AND ANALYSIS FOR EXPLOSIVES IN SOIL The environmental harateristis of munitions ompounds in soil indiate that they are extremely heterogenous in spatial distribution. Conentrations range from nondetetable levels (< 0.5 ppm) to perent levels (> 10,000 ppm) for samples olleted within several feet of eah other. In addition, the waste disposal praties at these sites, suh as OB/OD, exaerbate the problem and may result in onditions ranging from no soil ontamination up to solid "hunks" of bulk seondary explosives, suh as TNT or RDX. Seondary explosives onentrations above 10% (> 100,000 ppm) in soil are also of onern from a potential reativity standpoint and may affet sample and materials handling proesses during remediation. An explosives hazard safety analysis is needed for materials handling equipment to prevent initiating fores that ould propagate a detonation throughout the soil mass. Reliane on laboratory analyses only for site haraterization may result in a large perentage of the samples (up to 80% depending upon the site) Table 2. Ourrene of Analytes Deteted in Soil Contaminated with Explosives. % Sample with Maximum Level Compound Analyte (g/g) Present Nitroaromatis TNT ,000 TNB DNB ,4-DNT ,6-DNT AmDNT AmDNT 7 11 Tetryl Nitramines RDX 27 13,900 HMX TNT and/or RDX 72 Derived from Walsh et al. (1993). with nondetetable levels. The remaining samples may indiate onentrations within a range of four orders of magnitude. Analyzing a small number of samples at an off-site laboratory may result in inadequate site haraterization for estimating soil quantities for remediation and may miss potentially reative material. Laboratory analytial osts vary depending on the turnaround time required. Typial osts for EPA Method 8330 analysis range from $250 to $350 per sample for 30-day turnaround, $500 to $600 for 7-day turnaround, and approximately $1,000 per sample for 3-day turnaround, if it is available. Beause of the extremely heterogeneous distribution of explosives in soils, on-site analytial methods are a valuable, ost-effetive tool to assess the nature and extent of ontamination. Beause osts per sample are lower, more samples an be analyzed and the availability of near-real-time results permit redesign of the sampling sheme while in the field. On-site sreening also failitates more effetive use of off-site laboratories using more robust analytial methods. Even if only on-site methods are 4

5 used to determine the presene or absene of mean that more samples will be required to ontamination (i.e., all positive samples are sent haraterize the exposure pathways of interest. off-site for laboratory analysis), analytial osts an be Sampling variability typially ontributes muh more redued onsiderably. Beause on-site methods provide to total error than analytial variability (EPA 1990, near-real-time feedbak, the results of sreening an be 1992a). Under these onditions, the major effort used to fous additional sampling on areas of known should be to redue sampling variability by taking ontamination, thus possibly saving additional more samples using less expensive methods (EPA mobilization and sampling efforts. This approah has 1992a). been suessfully used for a Superfund remedial investigation of an OB/OD site (Craig et al. 1993). EPA's Guidane for Data Useability in Risk Assessment (EPA 1992a) indiates that on-site During site remediation, suh as Superfund remedial methods an produe legally defensible data if ations, data are needed on a near-real-time basis to appropriate method quality ontrol is available and assess the progress of leanup. On-site methods an be if doumentation is adequate. Field analyses an be used during remediation to guide exavation and used to derease ost and turnaround time as long as materials handling ativities and to evaluate the need supplemental data are available from an analytial for treatment on inremental quantities of soil (EPA method apable of quantifying multiple explosive 1992b). Final attainment of soil leanup levels should analytes (e.g., Method 8330) (EPA 1992a). be determined by an approved laboratory method, suh Signifiant quality assurane oversight of field as EPA Method This approah was effetively analysis is reommended to enable the data to be used at a Superfund remedial ation for an explosives widely used. The auray (orretness of the washout lagoon (Oresik et al. 1994; Markos et al. onentration value and a ombination of both 1995). systemati error [bias] and random error [preision]) of on-site measurements may not be as high in the field as in fixed laboratories, but the quiker DATA QUALITY OBJECTIVES turnaround and the possibility of analyzing a larger number of samples more than ompensates for this The EPA Data Quality Objetives proess is fator. Remedial projet managers, in onsultation designed to failitate the planning of environmental with hemists and quality assurane personnel, data olletion ativities by speifying the intended should set auray levels for eah method and use of the data (what deision is to be made), the profiieny standards for the on-site analyst. deision riteria (ation level), and the tolerable error rates (EPA 1994; ASTM 1996). Integrated use of On-site methods may be useful for analysis of on-site and laboratory methods for explosives in soil waste treatment residues, suh as inineration ash, failitate ahieving suh objetives as determining ompost, and bioslurry reator sludges. However, the horizontal and vertial extent of ontamination, on-site methods should be evaluated against obtaining data to ondut a risk assessment, laboratory methods on a site and matrix-speifi identifying andidate wastes for treatability studies, basis beause of the possibility of matrix identifying the volume of soil to be remediated, interferene. Treatability studies are used to evaluate determining whether soil presents a potential the potential of different treatment tehnologies to detonation hazard (reative aording to RCRA degrade target and intermediate ompounds and to regulations), and determining whether remediation evaluate whether leanup levels may be ahieved for ativities have met the leanup riteria. site remediation. Treatability study waste for explosives-ontaminated soils should be of higher Environmental data suh as rates of ourrene, than average onentration to evaluate the effets of average onentrations, and oeffiients of variation heterogeneous onentrations and for potential are typially highly variable for ontaminants toxiity effets for proesses suh as bioremediation. assoiated with explosive sites. These differenes are a funtion of fate and transport properties, ourrene in different media, and interations with During remediation of soils ontaminated other hemials, in addition to use and disposal with explosives, monitoring the rate of praties. Information on frequeny of ourrene degradation and determining when treatment and oeffiient of variation determines the number of riteria have been met are neessary so that samples required to adequately haraterize exposure residues below leanup levels an be disposed pathways and is essential in designing sampling of and additional soil treated. Soils plans. Low frequenies of ourrene and high ontaminated with explosives are urrently oeffiients of variation, suh as with explosives, being treated by inineration, omposting, and 5

6 solidifiation/stabilization (Noland et al. 1984; extreme short-range heterogeneity, sampling error Turkeltaub et al. 1989; EPA 1993; Craig and overwhelmed analytial error. Contaminant Sisk 1994; Miller and Anderson 1995; Channell distributions were very site speifi, dependant on a et al. 1996). Other biologial treatment systems number of variables suh as waste disposal history, that have been evaluated for treating the physial and hemial properties of the speifi explosives-ontaminated soils inlude anaerobi explosive, and the soil type. The onlusion was that bioslurry, aerobi bioslurry, white rot fungus, to improve the quality of site haraterization data, and land farming (Craig et al. 1995; Sundquist the major effort should be plaed on the use of et al. 1995). higher sampling densities and omposite sampling strategies to redue sampling error. UNIQUE SAMPLING DESIGN There are several pratial approahes to reduing CONSIDERATIONS FOR EXPLOSIVES overall error during haraterization of soils ontaminated with explosives, inluding inreasing Heterogeneity Problems and Solutions the number of samples or sampling density, olleting omposite samples, using a stratified The heterogeneous distribution of explosives sampling design, and reduing within sample in soil is often alluded to but seldom quantified. heterogeneity. Beause explosives have very low The problem is probably onsiderably greater volatility, loss of analytes during field preparation of for explosive residues in soil than most other omposite samples is not a major onern. organi waste. From available Superfund site data, the median oeffiient of variation (CV) (standard Inreasing the Number of Samples - One simple deviation divided by the mean) for volatiles, way to improve spatial resolution during extratables, pestiides/polyhlorinated biphenyls haraterization is by olleting more samples using (PCBs), and tentatively identified ompounds in a finer sampling grid suh as a 5-m grid spaing soils ranges from 0.21 to 54% for individual instead of a 10-m spaing. Though desirable, this ontaminants (EPA 1992b). Data from 10 munitions approah has been rejeted in the past beause of the sites show the median CV for TNT was 284%, and higher sampling and analytial laboratory osts. the TNT CV ranged from 127% to 335% for When inexpensive on-site analytial methods are individual sites. Comparable data for RDX are used, this approah beomes feasible. The slightly median CV of 137% with a range of 129% to 203%, lower auray assoiated with on-site methods is and the median CVs for 2,4-DNT and AP/PA were more than ompensated for by the greater number of 414% and 184% respetively. If the natural samples that an be analyzed and the resultant variability of the hemials of potential onern is redution in total error. large (e.g., CV > 30%), the major planning effort should be to ollet more environmental samples (EPA 1992b). Colletion of Composite Samples - The olletion of omposite samples is another very Jenkins et al. (1996a, 1996b) reently onduted effetive means of reduing sampling error. Samples a study to quantify the short range sampling are always taken to make inferenes to a larger variability and analytial error of soils ontaminated volume of material, and a set of omposite samples with explosives. Nine loations, three at eah of from a heterogeneous population provides a more three different failities, were sampled. At eah preise estimate of the mean than a omparable loation, seven ore samples were olleted from a number of disrete samples. This ours beause irle with a radius of 61 m: one from the enter ompositing is a "physial proess of averaging" and six equally spaed around the irumferene. (adequate mixing and subsampling of the omposite The individual samples and a omposite sample of sample are essential to most ompositing strategies). the seven samples were analyzed in dupliate, Averages of samples have greater preision than the on-site, using the EnSys RIS olorimetri soil test individual samples. Deisions based on a set of kit for TNT (on-site method) and later by Method omposite samples will, for pratial purposes, 8330 at an off-site laboratory. Results showed always provide greater statistial onfidene than for extreme variation in onentration in five of the nine a omparable set of individual samples. In the study loations, with the remaining four loations showing disussed above by Jenkins et al. (1996a, 1996b), the more modest variability. For sites with modest omposite samples were muh more representative of variability, only a small fration of the total error was eah plot than the individual samples that made up beause of analytial error, i.e., field sampling error the omposites. Using a omposite sampling dominated total error. For the loations showing strategy, usually allows the total number of samples 6

7 analyzed to be redued whih redues osts while with RSDs below 3% at two other sites]. improving haraterization. Compositing should be Subsampling in the field is muh more hallenging used only when analytial osts are signifiant. An beause omplete sample proessing is not feasible. Amerian Soiety for Testing and Materials (ASTM) However, most sreening proedures speify guide was developed on omposite sampling and relatively small samples, typially a few grams. field subsampling (Gagner and Crokett, 1996), (ASTM, 1997). To redue within-sample heterogeneity, two methods an be employed: either homogenization Stratified Sampling Designs - Stratified sampling and extration or analysis of a larger sample. Unless may also be effetive in reduing field and direted otherwise, an analyst should assume that subsampling errors. Using historial data and site information representative of the entire ontents of knowledge or results from preliminary on-site the sample ontainer is desired. Therefore, the methods, it may be possible to identify areas in subsample extrated or diretly analyzed should be whih ontaminant onentrations are expeted to be representative of the ontainer. The smaller the moderately heterogeneous (pond bottom) or volume of that subsample removed for analysis and extremely heterogeneous (open detonation sites). extration, the more homogeneous the entire samples Different ompositing and sampling strategies may should be before subsampling (e.g., a representative be used to haraterize different areas that may result 0.5-g subsample is more diffiult to obtain than a in a more effiient haraterization. 20-g subsample from a 250-g sample). Colleting representative 2-g subsamples from 300 g of soil is Another means of stratifiation is based on partile diffiult and an require onsiderable sample size. Beause explosive residues often exist in a wide proessing suh as drying, grinding, and riffle range of partile sizes (rystals to hunks), it is splitting. Even in the laboratory, as disussed above, possible to sieve samples into various size frations obtaining representative subsamples is diffiult. An that may redue heterogeneity. If large hunks of ASTM guide is being developed to help in this explosive are present, it may be pratial to regard (Gagner and Crokett 1996). While oarse-sieve a relatively large sample (many sample-mixing proedures suh as sieving to kilograms), medium-sieve a portion of those fines, disaggregate partiles, mixing in plasti bags, et., and subsample the fines from medium sreening as an and should be used to prepare a sample, well. This would yield three samples of different extrating a larger sample is perhaps the easiest partile size and presumes that heterogeneity method of improving representativeness. For this inreases with oarseness. Eah fration would be reason, 20 g of soil is extrated for the Cold Regions analyzed separately but not neessarily by the same Researh and Engineering Laboratory (CRREL) method (visual sreening of the oarser frations for method, and the same approah may easily be used hunks of explosive may be possible) and then ould to improve results with most of the on-site methods be summed to yield the onentration on a weight or shown in Table 3. The major disadvantage of area basis. In addition, aqueous disposal of explosive extrating the larger sample is the larger volume of wastewaters suh as washout lagoons or spill sites waste solvent and solvent-ontaminated soil that often results in preferential sorption to fine-grained needs disposal. materials, suh as fines or lays, partiularly for nitroaromatis. The effetiveness of proper mixing in the field is illustrated in the reent report by Jenkins et al. Reduing Within Sample Heterogeneity - The (1996a, 1996b). Dupliate laboratory analyses of the heterogeneity of explosives in soils is frequently same samples, inluding drying, grinding, mixing, observed during the use of on-site analytial methods and areful subsampling resulted in an RSD of 11%. in whih dupliate subsamples are analyzed and Beause this field-mixing proedure was so effetive differ by more than an order of magnitude. Grant et in homogenizing the sample, the sampling and al. (1993) onduted a holding time study using subsampling proedure is presented here (Jenkins et field-ontaminated soils that were air-dried, ground al. 1996a). Soil ores (0 to 15 m in length and 5.6 with a mortar and pestle, sieved, subsampled in m in diameter) were olleted into plasti resealable tripliate, and analyzed using Method Even bags, and vegetation was removed. The sample of with suh sample preparation, the results failed to dry soil, a mixture of sand and gravel, was plaed yield satisfatory preision [the relative standard into 23-m aluminum pie pans, the soil was broken deviations (RSDs) often exeeded 25% ompared up using gloved hands, and large roks were 7

8 Table 3. Comparative Data for Seleting On-Site Analytial Methods for Explosives in Soil a. Criteria Method/ Method Type Detetion Range and Type of Results Samples per Bath Soil Sample Analysis Time - Kit Analytes and EPA Range Fator Sample Preparation Prodution Rate Method No. Size & Extration (one person) CRREL Colorimetri TNT: 1 to 22 mg/kg (22 X) TNT, RDX: Quantitative TNT: Bath or single 20 g 3 min shaking in minute extrat 6/samples; TNT, RDX, 2,4-DNT, RDX: 1 to 20 mg/kg (20 X) 2,4-DNT: Semiquantitative RDX: 6 to 7/bath or single ml aetone; settling; TNT: 5 minutes/sample; Ammonium Pirate /Piri 2,4-DNT: 2 to 20 mg/kg (10X) AP/PA: Quantitative 2,4-DNT & AP/PA: Single filtration. RDX: 30 minutes/6 RDX samples; Aid AP/PA: 1.3 to 69 mg/kg (53 X) or bathed 25 samples/day for TNT + RDX DNT: 30 minutes/6 samples AP/PA: 15 minutes/sample EnSys RIS Colorimetri TNT: 1 to 30 mg/kg (30 X) Quantitative Single 10 g Dry < 10% moisture TNT: 30 to 35 minutes/10 samples in TNT: Method 8515 draft RDX: 1 to 30 mg/kg (30 X) (optional); 3 min lab; estimated 40 to 45 minutes in RDX: Method 8510 shaking in 50 ml field. proposed aetone; 5 min settling; RDX: 60 minutes/6 samples. filtration. Optional drying time not inluded. 8 USACE Colorimetri 6 to 100 mg/kg (17 X) Quantitative Single or bathed 6 g 1 min shaking in 35 ml10 to 20 samples/day depending on TNT methanol; settling; soil harateristis filtration as needed. TM D TECH Immunoassay - ELISA TNT: 0.5 to 5.0 mg/kg (10 X) Semiquantitative 4 (single or bath) 3 ml 3 min shaking in minutes for 1 to 4 samples for TNT: Method 4050 draft RDX: 0.5 to 6.0 mg/kg (12 X) (onentration range) (~4.5 g) ml aetone; settle 1 TNT or RDX. RDX: Method 4051 draft to 10 min. TM Idetek Quantix Immunoassay - ELISA TNT: 0.25 to 100 mg/kg (400 X) Quantitative 20 to 40 (bath only) ~4.2 g 3 min shaking in to 3.5 hours for 20 to 40 samples. Antigen-Antibody ml aetone; settle Idetek estimates - 2 hours for up to 40 TNT several minutes. TNT samples. TM EnviroGard Immunoassay - ELISA Plate kit: 1 to 100 mg/kg (100 X) Plate: Quantitative Plate: bath of 8 2 g Air dry soil, 2 min Plate: 90 minutes for 8 samples TNT: Plate kit Tube kit: 0.2 to 15 mg/kg (75 X) Tube: Semiquantitative Tube: bath of 14 shaking in 8 ml Tube: 30 minutes for 14 samples TNT: Soil (tube) kit (onentration range) aetone; filter. Drying time not inluded. Ohmiron RaPID Assay Immunoassay - ELISA TNT: 0.07 to 5 mg/kg (71 X) Quantitative 5 to 51 (bath only) 10 g 1 min shaking in 20 1 hour for 20 extrations; 45 minutes Magneti partile/tube ml methanol; settle for analysis (51 samples) kit 5 min; filter TNT: Method 4050 proposed a Expanded and modified from EPA 1995b

9 a Table 3. Comparative Data for Seleting On-Site Analytial Methods for Explosives in Soil (ontinued). Criteria Method/ Interferenes and Cross-reativities > 1% based on IC50 (see text) Reommended QA/QC Storage Conditions and Skill Level Kit Shelf Life of Kit or Reagents CRREL TNT = TNT + TNB + DNB + DNTs + tetryl; Blank and alibration standards Store at room temperature. Medium - detetion limits (ppm); TNB 0.5; DNB < 0.5; 2,4-DNT 0.5; 2,6-DNT 2.1; tetryl 0.9 analyzed daily before and after RDX = RDX + HMX + PETN+ NQ + NC + NG sample analyses. Blank and spiked - detetion limits (ppm); HMX 2.4; PETN 1; NQ 10; NC 42; NG 9 soil run daily. Soil moisture > 10%, and humis interfere with TNT and RDX; nitrate and nitrite interfere with RDX. 2,4-DNT = 2,4-DNT + 2,6-DNT + TNT + TNB + tetryl; high opper, moisture and humis interfere. AP/PA = relatively free of humi and nitroaromati interferenes. EnSys RIS TNT = TNT + TNB + DNB + DNTs + tetryl; Method and soil blanks and a Store at room temperature. TNT: Low - detetion limits (ppm); TNB 0.5; DNB < 0.5; 2,4-DNT 0.5; 2,6-DNT 2.1; tetryl 0.9 ontrol sample daily, one Shelf life: RDX: Medium RDX = RDX + HMX + PETN + NQ + NC + NG dupliate/20 samples. TNT = 2 to 24 months at 27C - detetion limits (ppm); HMX 2.4; PETN 1; NQ 10; NC 42; NG 9 Some positive field results (1:10) RDX = 2 to 12 months at 27C Soil moisture > 10%, and humis interfere with TNT and RDX; nitrate and nitrite interfere with RDX. should be onfirmed. 9 USACE TNB interferes by raising minimum detetion limit. Blank soil sample, and alibration Store at room temperature Medium standard prepared from lean site soil. TM D TECH Cross reativity: Samples testing positive should be Store at room temperature or Low TNT: tetryl = 35%; TNB = 23%; 2AmDNT = 11%; 2,4-DNT = 4%; onfirmed using standard methods. refrigerate; do not freeze or exeed AP/PA unknown but ~100% at lower limit of detetion 37C for prolonged period. Shelf RDX: HMX = 3% life 9 months at room temperature TM Idetek Quantix Cross reativity: Dupliate extrations Refrigerate 2 to 8C, do not freeze or Medium-high, initial TNB = 47%; tetryl = 6.5%; 2,4-DNT = 2%; 4AmDNT = 2% 1 in 10 repliate exeed 37C. Shelf life 9 to 12 training reommended 2 sample wells/extrat months. Avoid diret light. TM EnviroGard Cross reativity: Plate: Samples run in dupliate. Store 4 to 8C; do not freeze or Plate: Medium-high Plate: 4-AmDNT = 41%; 2,6-DNT = 41%; TNB = 7%; 2,4-DNT = 2% exeed 37C. Do not expose Tube: Medium Tube: 2,6-DNT = 20%; 4AmDNT = 17%; TNB = 3%; 2,4-DNT = 2% substrate to diret sunlight. Shelf life: Plate 3 to 14 months. Tube 3 to 6 months. Ohmiron RaPID Cross reativity: Dupliate standard urves; positive Refrigerate reagents 2 to 8C. Medium-high, initial Assay TNB = 65%; 2,4-Dinitroaniline = 6%; tetryl = 5%; 2,4-DNT = 4%; 2AmDNT = 3%; ontrol sample supplied. Positive Do not freeze. training reommended DNB = 2% results requiring ation may need Shelf life 3 to 12 months. onfirmation by another method. a Expanded and modified from EPA 1995b

10 a Table 3. Comparative Data for Seleting On-Site Analytial Methods for Explosives in Soil (ontinued). Criteria Method/ Training Costs Comparisons to Method 8330 Other Developer Additional Considerations Kit Availability (not inluding labor) Referenes Referenes Information CRREL Free video for TNT and $15/sample plus $1,500 for Brouillard et al. 1993; EPA 1993, 1995a Jenkins Dr. Thomas F. Jenkins Large work area (2 large desks); requires the most setup time; RDX, see text for address. Hah spetrometer. (Method 8515), 1995b; et al. 1995; CRREL possible TNB interferene, no eletriity or refrigeration None available for 2,4- Jenkins 1990; Jenkins and Walsh 1992; Thorne and 72 Lyme Road required; deionized water required; must assemble materials; DNT, AP/PA. Markos et al. 1995; Lang et al. 1990; Jenkins Hanover, NH glassware must be rinsed between analyses; larger volume of Walsh and Jenkins 1991; 1995b (603) aetone waste, olor indiative of ompounds. Jenkins et al. 1996a; Jenkins and Walsh 1991, 1992; Thorne and Jenkins 1995a EnSys RIS Training available. $21/sample for TNT, EPA 1995a (Method 8515); EPA 1995b; Strategi Diagnostis, In. Large work area (desk size) power supply required to harge Appliable video on $25/sample for RDX plus IT 1995; Jenkins et al. 1996a, 1996b; 375 Pheasant Run Hah spetrometer; possible TNB interferene; olor indiation CRREL method available,$160/day or $430/wk for lab Markos et al. 1995; Myers et al Newtown, PA of other ompounds; requires aetone and deionized water; address in text. station. Lab station ost = (800) uvettes must be rinsed between analyses. Nitrate and nitrate $1,950 interferenes with RDX kit an be orreted using alumin-aartridges from EnSys. 10 USACE None available. $4/sample or $5/sample if IT 1995; Medary 1992 Dr. Rihard Medary Large work area (2 large desks); requires the most setup time; filtered plus $1,500 for U.S. Army Corps of Eng. possible TNB interferene; no eletriity or refrigeration Hah spetrometer 601 E. 12th Street required; must assemble materials; glassware must be rinsed Kansas City, MO between analyses. (816) TM D TECH 2 to 4 hours free on-site $30/sample for TNT or RDX EPA 1995a (Methods 4050 and 4051); Teaney et al. Strategi Diagnostis, In. Small working area; few setup requirements; no eletriity or training. plus $300 for DTECHTOR EPA 1995b; Haas and Simmons 1995; Pheasant Run refrigeration required; temperature dependent development (optional) Markos et al. 1995; Myers et al. 1994; Calif. EPA Newtown, PA time (effet an be redued by hanging DTECHTOR setting); Teaney and Hudak a and (800) signifiant amount of paking; relatively narrow range; no 1996b hek on test; easy to transport or arry; kits an be ustomized. Out-or range reruns require use of another kit. Idetek 1 day free on-site $21/sample for TNT plus EPA 1995b; Haas and Simmons 1995; Idetek, In. Large work area (desk); requires setup time, eletriity, TM Quantix training. $5,880 for lab station or Markos et al Reamwood Ave. refrigeration and deionized water; requires areful washing of $500/month rental. Sunnyvale, CA mirowells; repliate run for eah sample, average of the two (800) is the result; less temperature dependent. Out of range reruns require use of another kit. Enviro- Free training available. Plate: $17/sample plus $4129 Haas and Simmons 1995 Calif. EPA Strategi Diagnostis, In. Large work area (desk size); requires setup time, refrigeration TM Gard for equip. & small supplies Pheasant Run and power; aetone not supplied. Out-of-range reruns require Tube: $20/sample plus $2409 Newtown, PA use of another kit. for equip. & small supplies. (800) Ohmiron 4 hours free on-site $13 to $20/sample plus $5,500 EPA 1995b; Haas and Simmons 1995; Calif. EPA Strategi Diagnostis, In. Large work area (desk); requires setup time, eletriity and RaPID Assay training. for equip. (purhase) or $800 Markos et al. 1995; Rubio et al d 375 Pheasant Run refrigeration; less temperature dependent; low detetion limit; for first month, $400 eah Newtown, PA all reagents supplied; reagents and kit need refrigeration. additional month (rental). (800) Out-of-range reruns require use of another kit. a Expanded and modified from EPA 1995b

11 removed (sieving may work well too). A seond pie Another study (Bauer et al. 1990) has shown that pan was used to over the sample, whih was then explosives in spiked, air-dried soils are stable for a shaken and swirled vigorously to disperse and 62-day period under refrigeration. Data from the homogenize the soil. The sample was then oned Grant et al. (1993) study indiate that air drying of and quartered, and 5 g subsamples were removed field-ontaminated soils may not result in from eah quarter and omposited to form the 20-g signifiant losses of explosive ontaminants. sample for analysis. Splits of the same sample were Explosives in air-dried soils are stable at room obtained by remixing the soil and repeating the temperature if they are kept in the dark. oning and quartering. Aetonitrile extrats of soil samples are expeted Wilson (1992) studied sample preparation to be stable for at least 6 months under refrigeration. proedures for homogenizing ompost prior to Aetone extrats also are thought to be stable if the analysis for explosives. Wilson (1992) method extrats are stored in the dark under refrigeration involves maerating air-dried ompost using a No. (aetone enhanes photodegradation of explosives). 4 Wiley mill followed by sample splitting using a Jones-type riffle splitter. The improved method Explosion Hazards and Shipping Limitations dereased the RSD from more than 200% to 3% for TNT analyses. The Department of Defense Explosive Safety Board approved the two-test protool (Zero Gap and Sample Holding Times and Preservation Deflagration to Detonation Transition tests) in Proedures Marh 1988 for determining the explosive reativity of explosive-ontaminated soil. Tests on TNT and The EPA-speified holding time for nitroaromati RDX in sands with varied water ontent showed that ompounds in soil is 7 days until extration and soils with 12% or more explosive are suseptible to extrats must be analyzed within the following 40 initiation by flame, and soils ontaining more than days (EPA 1995a). The speified sample pre- 15% explosives are subjet to initiation by shok servation proedure is ooling to 4C. This riterion (EPA 1993). Explosives exist as partiles in soil was based on professional judgment rather than ranging in size from rystals to hunks, whih an experimental data. detonate if initiated. However, if the onentration of explosives is less than 12%, the reation will not Two signifiant holding time studies have been propagate. The water ontent of the soil has minimal onduted on explosives (Maskarine et al. 1991; effets on reativity. The test results apply to total Grant et al. 1993, 1995). Based on spiking lean weight perent of seondary explosives suh as soils with explosives in aetonitrile, Maskarine TNT, RDX, HMX, DNT, TNB, and DNB. The tests reommended the following holding times and do not apply to primary or initiating explosives suh onditions: TNT immediate freezing and 233 days as lead azide, lead styphnate, and merury at -20C; DNT 107 days at 4C; RDX 107 days fulminate. As a onservative limit, the EPA Regions at 4C; and HMX 52 days at 4C. Grant spiked and the U.S. Army Environmental Center onsider soils with explosives dissolved in water to eliminate soils ontaining more than 10% seondary any aetonitrile effets and also used a explosives, on a dry weight basis, to be suseptible field-ontaminated soil. The results on spiked soils to initiation and propagation (EPA 1993). If showed that RDX and HMX are stable for at least 8 hemial analyses indiate that a sample is below weeks when refrigerated (2C) or frozen (-15C) 10% explosives by dry weight, that sample is but that signifiant degradation of TNT and TNB onsidered to be nonreative. In most ases, this degradation an our within 2 hours without eliminates the requirement to ondut the expensive preservation. Freezing provides adequate two-test reativity protool. preservation of spiked 2,4-DNT for 8 weeks or longer. The results on field-ontaminated soils did In sampling to determine whether an explosion not show the rapid degradation of TNT and TNB hazard exists, a biased sampling approah must be that was observed in the spiked soils, and adopted (Sisk 1992). Soils suspeted of having high refrigeration appeared satisfatory. Presumably, the onentrations of explosives should be explosives still present in the field soil after many grab-sampled and analyzed to determine whether years of exposure are less biologially available than the level of explosives exeeds 10%. Samples to be in the spiked soils. shipped for off-site analysis must be subsampled 11

12 and analyzed on-site. Explosive residues are usually onentrated in the top 5 to 10 m of soil; therefore, deep samples must not be olleted, blended, and analyzed to determine reativity. Vertial ompositing of surfiial soils with high levels of explosives with deeper, relatively lean material provides a false indiation of reativity. Soils ontaining explosive residues over the 10% level an, using proper preautions, be blended with leaner material to redue the reativity hazard and permit shipment to an off-site laboratory. The dilution fator must be provided with the sample. If analytial results indiate that explosives are present at a onentration of 10% or greater, the samples must be shipped to an explosives-apable laboratory for analysis. The samples must be pakaged and shipped in aordane with appliable Department of Transportation and EPA regulations for reative hazardous waste and Class A explosives (AEC 1994). In addition to the above information, the Army Environmental Center requires ertain minimum safety preautions, as summarized below, for field sampling work at sites with unknown or greater than 10% by weight of seondary explosives ontamination (AEC 1994). An extensive reords searh and historial doumentation review must be onduted regarding the ontaminated area to identify the speifi explosives present, determine how the area beame ontaminated, estimate the extent of ontamination, and determine the period of use. Personnel responsible for taking, pakaging, shipping, and analyzing samples must be knowledgeable and experiened in working with explosives. Soil samples must be taken using nonsparking tools, and wetting the sampling area with water may be neessary. If plasti equipment is used, it must be ondutive and grounded. Sample ontainers must be hemially ompatible with the speifi explosive, and srew tops are prohibited. Samples are to be field sreened for explosives if possible. Suffiient soil samples must be olleted to haraterize the site in a three-dimensional basis in terms of perent seondary explosives ontamination with partiular attention paid to identifying hot spots, hunks of explosives, layers of explosives, disolorations of the soil, et. In sreening samples for reativity, it should be remembered that most sreening proedures test for only one analyte or lass of analyte. Without other supporting knowledge, onluding that a soil is not reative based upon just one analysis ould be dangerous. For assessing reativity when multiple ompounds are present at high levels, the CRREL 12

13 and EnSys RIS olorimetri methods for TNT and (shaking with aetone versus ultrasoniation with RDX are more appropriate than immunoassay test aetonitrile) rather than the analytial methods kits beause olorimetri tests detet a broader whih may also produe different results. However, range of explosive analytes. Some onservatism in if a group of aetone extrats are analyzed by two evaluating potential reativity using olorimetri different on-site methods, the subsampling and exmethods is appropriate. For example, Jenkins et al. tration errors are minimized and any signifiant (1996) reommended using a limit of 7% differenes should be from the analytial methods. explosives for onservatively estimating the lower limit of potential reativity. High levels of Preision and Bias Tests for Measurements of explosives in soils may result in a low bias for Relatively Homogenous Material - When multiple on-site methods beause of low extration splits of well-homogenized soil samples are effiienies. Colorimetri tests of hemial analyzed using different analytial methods, omposition are used only to estimate potential statistial proedures desribed in Grubbs (1973), reativity. There are no on-site methods available to Blakwood and Bradley (1991), and Christensen atually determine explosive reativity. Explosive and Blakwood (1993) may be used ompare the reativity is a determination made from validated preision and bias of the methods. Grubbs (1973) laboratory analyses. desribes a statistial approah appropriate for omparing the preision of two methods that takes PROCEDURES FOR STATISTICALLY into aount the high orrelation between the COMPARING ON-SITE AND REFERENCE measurements from eah method. An advantage of ANALYTICAL METHODS Grubbs approah is that it provides unbiased estimates of eah method s preision by partitioning When on-site methods are used, their performane the variane of the measurement results into its needs to be evaluated and this is ommonly done by omponent parts (e.g., variane aused by analyzing splits of some soil samples by both the subsampling and by the analytial method). on-site method and a referene method (ommonly Blakwood and Bradley (1991) extend Grubbs Method 8330). The performane of the on-site approah to a simultaneous test for equal preision method is then statistially ompared to the and bias of two methods. Christensen and referene method using a variety of methods, Blakwood (1993) provide similar tests for depending upon the objetive and the harateristis evaluating more than two methods. of the data. In most ases, measures of preision and bias are determined. Preision refers to the For omparisons involving bias alone, t-tests or agreement among a set of repliate measurements analysis of variane may be performed. For and is ommonly reported as the RSD (standard omparing two methods, paired t-tests are deviation divided by the mean and expressed as a appropriate for assessing relative bias (assuming perent), the oeffiient of variation (standard normality of the data, otherwise data deviation divided by the mean), or the relative transformations to ahieve normality must be perent differene. Bias refers to systemati applied, or nonparametri tests used). A paired t-test deviation from the true value. an be used to test whether the onentration as determined by an on-site method is signifiantly The following disussion of statistial methods different from Method 8330 or any other referene applies to omparisons of analytial results based on method. For omparing multiple methods, a paired sample data, e.g., soil samples are analyzed randomized omplete blok analysis of variane an by both an on-site method and a referene method, be used, where the methods are the treatments and or soil extrats are analyzed by two different on-site eah set of split samples onstitutes a blok. methods. Care must be taken in interpreting the result. For example, if subsamples of a jar of soil These tests are best applied when the (splits) are analyzed by an on-site and referene onentrations of explosives are all of method, the differenes deteted may be aused by approximately the same magnitude. As the subsampling error (sample was not homogeneous variability in the sample onentration inreases, the and the splits atually ontained different apability of these tests for deteting differenes in onentrations of explosives), extration effiieny preision or bias dereases. The variability in the 13

14 true quantities in the samples is of onern, and high whether the slope is signifiantly different from 1 variability in sample results aused by poor an rejet the null hypothesis even when there is in preision rather than variability in the true fat no differene in the true bias of the two onentration is well handled by these methods. methods. A similar argument applies to tests of the interept value being equal to zero. Preision and Bias Tests for Measurements over Large Value Ranges - When the To perform a proper errors-in-variables regression onentrations of explosives over a large range of requires onsideration of the measurement errors in values, regression methods for assessing preision both variables. The appropriate methods are and auray beome appropriate. Regression outlined in Mandel (1984). These methods require analysis is useful beause it allows haraterization estimating the ratio of the random error variane for of nononstant preision and bias effets and the on-site method to that of the referene analytial beause the analysis used to obtain predition method. With split sample data, suitable estimates intervals for new measurements (e.g., the results of of these ratios may generally be obtained by using an on-site method an be used to predit the variane estimates from Grubbs test or the related onentration if the samples were analyzed by a tests mentioned above. referene method). If the variane ratio is not onstant over the range In a regression analysis, the less preise on-site under study, more ompliated models than those method is generally treated as the dependent analyzed in Mandel (1984) must be employed. variable and the more preise referene analytial Alternatively, transformations of the data might method (e.g., SW-846 Method 8330) as the stabilize the variane ratio. Note that it is the independent variable. To the extent that the variane ratio, not the individual varianes, that relationship is linear and the slope differs from a must remain onstant. The ratio of varianes for two value of 1.0, there is an indiation of a onstant methods with nononstant absolute varianes but relative bias in the on-site method (i.e., the two onstant relative varianes will still have a onstant methods differ by a fixed perentage). Bias should variane ratio. be expeted if on-site methods based on wet-weight ontaminant levels are ompared to laboratory Two other aveats about the use of regression methods based on the dry weight of soil samples. tehniques also are appropriate. First, standard Similarly, an interept value signifiantly different regression methods produe bias regression from zero indiates a onstant absolute bias (i.e., the parameters estimation and may produe misleading two methods differ by a fixed absolute quantity). unertainty intervals. Similarly, the interpretation There, may of ourse be both fixed and relative bias of R-squared values also is affeted. Seond, peromponents present. forming regressions on data sets in whih samples with onentrations below the detetion limit (for When unertainty is assoiated with the one or both methods) have been eliminated may also onentration of an explosive as measured by the result in biased regression estimates, no matter referene method, standard least squares regression whih regression analysis method is used. analysis an produe misleading results. Standard least squares regression assumes that the Comparison to Regulatory Thresholds, Ation independent variable values are known exatly as in Limits, et. - When the purpose of sampling is to standard referene material. When the on-site make a deision based on omparison of results to a method results ontain appreiable error ompared speifi value suh as an ation level for leanup, to the referene method, regression and variability on-site and referene analytial method results may estimates are biased. This is known as an be ompared simply on the basis of how well the errors-in-variables problem. two methods agree regarding the deision. The appropriate statistial tests are based on the Beause of the errors-in-variables problem, the binomial distribution and inlude tests of equality of slope oeffiient in the regression of the on-site data proportions and hi-square tests omparing the on the referene data will generally be biased low. sensitivity and speifiity (or false positive and false Hene a standard regression test to determine negative rates) of the on-site method relative to the 14