Analytical Chemistry Detection Limits and the Evaluation of Dredged Sediment

Transcription

1 Analytial Chemistry Detetion Limits and the Evaluation of Dredged Sediment by Robert P. Jones and Joan U. Clarke PURPSE: This tehnial note presents information larifying the proper use of analytial hemistry detetion limit terminology with respet to the evaluation of dredged material prior to disposal in oean or inland waters of the United States. This doument is intended to support guidane presented in the Inland Testing Manual (USEPA/USACE 1998) and the ean Testing Manual (USEPA/USACE 1991). Valuable guidane regarding detetion limit terminology an be found in Appendix 1 of EM (USACE 2001) and Appendix D of the Department of Defense's (DoD's) Quality Systems Manual (USDoD 2002). BACKGRUND: ne of the most important aspets of evaluating dredged sediment prior to disposal is determining the presene or absene of environmental ontaminants. With modern analytial hemistry tehniques, ontaminants of onern (Cs), when present above minimum levels, may be onfirmed with a high degree of ertainty. For a given analytial method, when C onentrations fall below these minimum levels, it is impossible to know for ertain whether suh ompounds are present in the environmental matrix being tested. Therefore, analytial hemistry data for non-detets should always be viewed relative to a predefined numerial level (i.e., onentration value) generially termed the "detetion limit" (DL). Unfortunately, a wide array of terms have been oined that relate to analytial detetion limits, and it is ritial that data users understand the meaning and proper use of these terms when interpreting analytial hemistry data. BASIC TERMINLGY AND NCEPTS: In order to suessfully interpret analytial hemistry data, it is ritial to understand the meaning of ommon terms used to desribe analytial detetion limits. With respet to dredged material evaluation, Projet Ation Level (PAL), Target Detetion Limit (TDL), Method Detetion Limit (MDL), Method Quantitation Limit (MQL), and Laboratory Reporting Limit (LRL) are partiularly important terms. It is also important to understand the relationship of these terms to absolute instrument sensitivity (AIS). These terms are disussed below. Projet Ation Level. Deisions regarding disposal of dredged material with respet to eah C should be based on some predetermined onentration level or projet ation level (PAL) for the C. Bakground ontaminant levels at the disposal site may serve as a useful point of referene for establishing PALs. Sediment Quality Guidelines (SQGs) that are derived numerial values representing onentrations of Cs thought to adversely affet benthi organisms ould also be used in developing PALs. Sediment Quality Guidelines have been developed by EPA through its Contaminated Sediment Management Strategy (Southerland et al. 1992). ther entities have also developed SQGs as referene values. Cs in elutriate samples may be ompared to Water Quality Standards issued by states under Setion 404 of the Clean Water At DISTRIBUTIN STATEMENT A Approved for Publi Release Distribution Unlimited

2 (USEPA/USACE 1991, 1998). Regardless of the benhmark used, it is ritial that ontrat labs report C onentrations to detetion levels well below PALs. Target Detetion Limit. TDL is a term used in the Inland Testing Manual (ITM, USEPA/USACE 1998). The TDL is a performane goal set greater than the lowest, tehnially feasible detetion limit for routine analytial methods and less than the available regulatory riteria or guidelines (i.e., the PAL) for evaluating dredged material. TDLs are values defined as part of the projet planning proess, and should be seleted suh that detetion limits reported by the analytial lab are low enough to ensure that the presene of Cs an be ruled in or ruled out at or below the predetermined PALs. Typially TDLs should be no lower than one-tenth prevailing regulatory guidelines (USEPA/USACE 1998). Method Detetion Limit. The MDL is a statistially derived expression of theoretial method detetion apability. The MDL is defined as the minimum onentration of a substane that an be measured and reported with 99 perent onfidene that the analyte onentration is greater than zero. Detailed proedures for determining MDLs are desribed in 40 CFR Part 136 Appendix B (Federal Register 1995), Appendix I of EM (USACE 2001), and in Appendix D of the DoD's Quality Systems Manual (2002). In general, MDL values are determined by performing the omplete analytial proedure (extration/digestion, leanup, and instrumental analysis) on repliate spiked samples (7 or more) in an otherwise lean, interferene-free matrix representative of the environmental matrix to be tested. For sediment MDLs, lean sand or lay is typially used as the interferene-free matrix. MDLs are alulated by multiplying the standard deviation of these repliate (w) measures by the Student's f-value at the 99-perent onfidene level (at n - 1 degrees of freedom). If deisions with respet to disposal of dredged material are to be based on MDLs, it is highly reommended that the data user require the laboratory to verify its alulated MDL values by extrating and analyzing a dupliate sample (spiked at about twie the alulated MDL for eah method target). For a given target ompound, the MDL hek is aeptable if it produes a detetable signal at least three times that of the bakground noise (USDoD 2002). Sine MDL hek samples do not test for real world sample matrix effets, an aeptable MDL hek does not guarantee similar detetion apability for field samples. MDL studies are performed annually or when signifiant hanges in method standard operating proedures (SPs) our. Additionally, a new MDL study is typially performed when an analytial instrument is replaed for a given method. MDLs are based on a disrete set of measurements, and in the data reporting proess, MDLs are not typially adjusted for sample-speifi parameters suh as sample weight, perent solids, or dilution. MDLs are estimates of detetion apability and are valuable as referenes. Careful onsideration should be given to the inherent unertainties (as disussed above) assoiated with MDLs before using these values in making deisions regarding the disposal of dredged material. Method Quantitation Limit. The Method Quantitation Limit (MQL) is a term used in urrent Corps guidane (USACE 2001). The MQL is set at a fator five to ten times the MDL for most target analytes in a method, but no lower than three times the MDL for any single target analyte. The MQL represents the value at whih the laboratory has demonstrated the ability to reliably measure targets within presribed performane riteria, and it establishes the lowest onentration at whih data may be reported without qualifiation. In the absene of projet-speifi requirements to the ontrary, the MQL is set at the level of the lowest alibration standard for the method, and the lowest alibration standard for eah target must be at least three times the MDL

3 or greater. All target analytes deteted below the MQL would be flagged as estimates (i.e., J- flagged). The MQL is a fixed referene value based on some multiple of the MDL, and it is not adjusted for sample-speifi parameters. Laboratory Reporting Limit. LRLs are the minimum levels at whih a lab will report analytial hemistry data with onfidene in the quantitative auray ofthat data. LRLs are threshold values below whih the laboratory reports a given result as non-deteted (i.e., U-flagged) and are presented as the "less than onentration value" (i.e., < ###, where ### equals the value of the LRL). LRLs are laboratory-determined values that may be based on projet-speifi reporting limits, regulatory ation levels, or multiples of the MDL, but they should be no lower than the lowest alibration level. Typially, LRLs are no lower than the low alibration standard for the method. In general, LRLs should be 3-5 times the MDL for a given target analyte. LRLs are typially quoted by laboratories based on a default set of method onditions. For example, with sediment or soil for pestiides analysis by gas hromatography using a soxhlet extration proedure, LRLs might be expressed based on a 30-gram dry, undiluted sample. However, for field samples, LRLs are adjusted for sample-speifi parameters suh as sample weight, perent solids, or dilution. These an have a signifiant influene on final reported values for the LRL, espeially if dilution of the sample is required due to high levels of target analytes or interferenes. Absolute Instrument Sensitivity. Eah and every ompound-speifi method that employs the use of modern analytial instrumentation has some form of detetor that produes a signal in response to the presene of target ompounds. The instrument itself, as well as some sample matrix omponents, also produes measurable detetor output alled noise. In general, absolute instrument sensitivity (AIS) is based only on the system noise and represents a level at whih a referene signal should be measurable. In the ontext of analytial detetion limits, the lowest onentration of a target analyte that produes a signal that an be reliably distinguished from the bakground noise is the true detetion limit of the instrumental system (Willard et al. 1988). This an be expressed quantitatively as the minimum analyte onentration that produes a signal 2-3 times the standard deviation of the blank signal (noise). The signal-to-noise ratio (S/N) is often used to gauge instrument sensitivity, and as a rule of thumb, S/N values in the range of 3-5 would be onsidered suffiient to distinguish signal from noise. Being stritly a funtion of instrument noise, AIS does not aount for non-instrument related fators (suh as extration/leanup tehniques or sample amount) that also impat overall method detetion apability. Thus, absolute instrument sensitivity is a ritial fator influening the ability of a method to detet target analytes, but sine it is not the only fator, AIS should not be used as the sole determinant when establishing analytial detetion or reporting limits. Additionally, the AIS may vary from day to day and may be altered by the input of dirty samples into the instrument. For this reason, most routine methods employed by ommerial labs do not attempt to measure AIS on a daily basis. Instead, they estimate sensitivity through periodi performane of Method Detetion Limit (MDL) studies desribed above. The relationship between PALs, TDLs, MDLs, MQLs, LRLs, and AIS is ritially important. This relationship is summarized shematially in Figure 1. TDLs are projet-speifi data quality objetives that should be defined before the ontrat lab is seleted, and the ability of the ontrat lab to meet projet TDLs should be onsidered as part of the seletion proess. As a general rule, MDLs should not be used as the final determinant for whether or not TDLs have been met.

4 MDL values should be used for referene purposes only. MDL levels may be useful in gauging the presene or absene of a given C, but quantitative auray of values reported near the MDL is unertain. MQLs an be viewed as referene values against whih field sample LRLs are ompared. The LRL for a given target may be greater than or equal to the MQL, but as a general rule, it should not be lower than the MQL. Beause LRLs represent quantitatively reliable onentration levels, they are the most appropriate values to determine if TDLs have been met by the laboratory. PAL B q o *-> "E o D) C "(/) L_ o A K ) DC r DC -J C5) C o 1 C3) C ) C 1/2 PAL Typial TDL Typial LRL (1/10 PAL) MQL (3-5 X MDL) MDL Chek (2-3 X MDL) MDL AIS (3-5 X Noise) Noise Zero Figure 1. Shemati summary of detetion limit terminology

5 ERC/TN EEDP THER MMN DETECTIN LIMIT TERMS: In addition to the terminology disussed above, a number of other related terms may be enountered by the data user. Some of these are similar in meaning to terms disussed above, but differenes are apparent. It is ritial to understand the similarity and differenes to ensure proper interpretation of analytial data. Many of the definitions below are taken verbatim from ited guidane. Method Reporting Limit. The Method Reporting Limit (MRL) is also used in urrent USACE (2001) guidane. It is defined as the threshold value below whih the laboratory reports a given result as non-deteted (i.e., U-flagged) and is presented as the "less than onentration value" (i.e., < ###, where ### equals the value of the MRL). MRLs are set based on projet-speifi fators that inorporate the needs of the data user, the sensitivity of the method/instrumentation, and the unertainties assoiated with low-level data the user is willing to aept. As with LRLs, MRLs are adjusted based on sample matrix, weight/volume, perent solids, and dilutions. Depending on the planned use of the data, the MRL may be set as low as the sample adjusted level equivalent to the MDL or as high as one-half the PAL. In general, USACE (2001) reommends that MRLs be established at approximately one-half the PAL. While multiple fators are onsidered in establishing MRLs, to say they are set arbitrarily would be a vast oversimplifiation; Further details on the appropriate use and designation of MRL values are desribed in urrent guidane (USACE 2001). MRL and LRL are very similar terms, but unlike LRLs, MRLs an, for ertain data uses, be equivalent to MDLs for individual ompounds. The key differene is that MRLs are set with projet-speifi onsiderations in mind, but in ontrast LRLs are defined by the ontrat lab and are generally based on method performane. Instrument Detetion Limit. The instrument detetion limit (IDL) is onsidered the minimum detetion onentration for the instrument only, and unlike the MDL, it ignores sample preparation effets (USACE 2001). EPA SW-846 Methods 6010B, 6020, and 7000 Series methods for metals analysis make referene to IDLs. For Method 6020, the IDL is estimated by alulating the average of the standard deviations of three runs on three non-onseutive days from the analysis of a reagent blank solution with seven onseutive measurements per day. Beause IDLs do not assess the impat of sample preparation proedures or sample matrix effets, they should not be ahievable in environmental samples and should not be used as benhmarks when evaluating samples against projet ation levels. Estimated Quantitation Limit. The estimated quantitation limit (EQL) is a term used in the EPA SW-846 ompendium. It is defined as the lowest onentration that an be reliably ahieved within speified limits of preision and auray during routine laboratory operating onditions. The EQL is generally 5 to 10 times the MDL, but it may be nominally hosen within these guidelines to simplify data reporting. In SW-846, for many target analytes, the EQL is seleted as the lowest non-zero alibration standard. EQLs in SW-846 are provided as guidane values and may not always be ahievable. "Pratial Quantitation Limit. The pratial quantitation limit (PQL) is also used in the SW- 846 ompendium and is defined as the lowest level that an be reliably measured by routine laboratory operating onditions within speified limits of preision and auray (USACE 2001). PQLs are guidane values essentially synonymous with EQLs.

6 Sample Quantitation Limit. The sample quantitation (SQL) is a term established within the USEPA Risk Assessment Guidane for Superfund (RAGS) and is the limit of interest when reporting data for use in a risk assessment (USACE 2001). The SQL is defined as the MDL adjusted for sample-speifi parameters suh as dilution or sample aliquot sizes. The terminology disussed above is summarized in Table 1. It should be noted that detetion limit terms not presented in this artile may be enountered by the data user. For this reason it is ritial that the ontrat lab learly define all detetion limit terminology used in data reporting and that the data user understand the appropriate use of reported detetion limits. Table 1 Detetion Limit Terms Term Projet Ation Level Target Detetion Limit Laboratory Reporting Limit Method Detetion Limit Absolute Instrument Sensitivity Method Quantitation Limit Method Reporting Limit Instrumental Detetion Limit Estimated Quantitation Limit Pratial Quantitation Limit Sample Quantitation Limit Abbreviation PAL TDL LRL MDL AIS MQL MRL IDL EQL PQL SQL Definition Conentration level of Cs that ultimately diret final deisions regarding disposal of dredged material. Performane goal set between the lowest, tehnially feasible analytial detetion limit and available regulatory guidelines for evaluating dredged material. Lowest level lab reports with onfidene in auray of quantitation and without qualifiation (i.e., J-flagged). May be set at low alibration standard for the method; Should be 3-5 times the MDL or greater. Adjusted for sample-speifi parameters when reported. Statistially derived expression of theoretial detetion apability; Based on standard deviation of repliate spiked samples taken through all steps of analytial proedure; Based on standard proedure from 40 CFR Part 136 Appendix B Lowest level that reliably produes signal distinguishable from bakground noise. Level lab has shown ability to reliably measure targets within presribed performane riteria; Lowest onentration reported without qualifiation (i.e., J-flagged); May be set at low alibration standard for the method; Typially 5-10 times the MDL, but must be at least three times the MDL. ' Projet-speifi, sample adjusted threshold values below whih lab reports data as nondeteted and less than the threshold value (< ###). May or may not imply a high level of analytial ertainty (e.g., when MRL = MDL). Minimum detetion onentration for instrument only. Lowest level that an be reliably measured under routine onditions. Lowest level that an be reliably measured under routine onditions. Used in risk assessment; MDL adjusted for sample-speifi parameters. FACTRS THAT INFLUENCE DETECTIN LIMITS: A number of fators impat the final detetion limits reported for environmental samples. Some of these are summarized below. Sample Amount. Conentrations of target analytes in environmental samples are typially reported in units of mass of the target (mg) per mass (kg) or volume (L) of sample. For a given method, as the amount (weight or volume) of sample is redued, the LRL is proportionally elevated. Sediment and soil methods typially report data on a dry weight basis, and all methods have physial limitations on the sample amount that an be handled by the sample prep apparatus. With sediment samples, the perent moisture an signifiantly impat the atual amount of dry sample analyzed. For example, if a typial method requires 30 g (dry weight) of sediment to

7 yield an LRL of 1.0 mg/kg, but 30 g of wet sediment with a moisture ontent of 33 perent is analyzed, the resulting LRL for the sample would be 1.5 mg/kg. For this partiular sample, 45 g wet sediment would have been required to ahieve the 1.0 mg/kg LRL, but physial onstraints of the sample extration equipment might make this impratial. Also, inreasing the sample size to 45 g in this ase ould lead to inreased interferenes that might not be adequately removed through method leanup proedures. Sample Matrix. For dredged material evaluations, hemial analysis may be required on sediment, site water, and tissue samples. Eah of these matries will yield different detetion limits even though the exat same instrumental tehnique may be used. For instane, a 30-g wet sediment sample as noted above might yield a 1.5 mg/kg (1.5 ppm) sample-speifi LRL for the analysis of pestiides by gas hromatography (GC). The sample preparation proess produes a 10-mL extrat for analysis by GC. If a 1-L water sample were prepared for the same analysis with a resulting 10-mL extrat, a sample-speifi LRL of 0.03 mg/l (0.03 ppm) might be reported. Dilution. Analytial methods typially work best over a range of onentrations for a given target ompound. Just as there is a lower limit of detetion, there is also an upper limit above whih aurate analytial quantitation is unreliable. This upper limit is ditated primarily by the dynami range of the instrument detetor, and very high levels of any material that produes a signal ould overload the detetor. This results in a signal maximum that fails to inrease with inreasing target ompound onentration. This situation an our when high levels of target ompounds or unwanted interferenes are present in the sample. ne remedy is simply to dilute the sample extrat to a point that the instrument signal is within its dynami range. With extrat dilution, however, omes a proportional inrease in LRL. If dilution of 1:100 was required of the sediment sample that otherwise would have produed an LRL of 1.5 mg/kg in the example above, then the LRL would inrease to 150 mg/kg. Extrat Volume. After the initial sample extration or digestion and leanup steps have been ompleted, the resulting solution (solvent extrat, aid digestate, or other solution for analysis) must typially be adjusted to a final volume prior to instrumental analysis. The onentration of this solution is then determined and used to alulate the final onentration of target analytes in the sample. The final volume of the sample extrat an therefore impat analytial detetion limits. In the example above referening a 10-mL solvent extrat, it is possible to redue the extrat volume through ontrolled evaporation tehniques. In this manner the extrat volume ould be redued to 1.0 ml, lowering the LRL for the sediment sample from 1.5 mg/kg to 0.15 mg/kg. Analytial Tehnique. Different analytial tehniques may show signifiant differenes in detetion apability. A prime example of this is seen with gas hromatography/mass speetrometry (GC/MS) as ompared to gas hromatography with eletron apture detetion (GC/E). Both tehniques may be used to detet the same ompounds in some ases, but for E-ative ompounds, GC/E an give LRLs one or more orders of magnitude lower than GC/MS. HANDLING NN-DETECTS: So what happens when non-detet data are reported as "less than onentration value"? Guidane for handling non-deteted targets is desribed in the Inland Testing Manual (USEPA/USACE 1998), and detailed supporting information is given elsewhere (Clarke 1995, 1998). Briefly, several options are available inluding the following simple

8 approahes: (1) substitute the reported detetion limit (DL) for all non-detets, (2) substitute onehalf the reported detetion limit (DL) for all non-detets, (3) substitute a value of zero for all non-detets. ther, more omplex methods using tehniques suh as regression and maximum likelihood have been reommended in the statistial literature. In general, the simple substitution methods listed above work better than more omplex tehniques when sample sizes are very small, as is typially the ase in dredged sediment evaluations (Clarke 1998). With small sample sizes, substitution of the DL when up to 40 perent of the data are ensored (reported as less than a detetion limit), or substitution of one-half the DL when greater than 40 perent are ensored, are methods that work reasonably well (USEPA/USACE 1998). Alternatively, ensored data sets may be statistially analyzed using non-parametri proedures suh as Dunn's test (Hohberg and Tamhane 1987) for multiple omparisons. In general, when ensoring exeeds 60 to 80 perent, any statistial analysis is likely to result in unaeptably high error rates. While the above guidane is reasonably straightforward, the term DL has not been well-defined. Does DL in this guidane refer to MDL, SQL, LRL, MRL, or MQL? The answer to this question is simple and omplex, and it really depends on the information available for a given data set as well as the level of unertainty the data user is willing to aept. For instane, if an MDL value is reported, has the MDL been verified through the analysis of the appropriate MDL hek sam : pies, and was the verifiation performed reently? As noted above, the MDL hek sample should be spiked at 2-3 times the stated MDL for eah target. If MDL values have been satisfatorily verified (USACE 2001), then substituting values approahing the MDL may be aeptable when the data are being used to support a risk assessment, but suh values should be adjusted for sample-speifi parameters (inluding dilutions) that in effet transform MDL values into SQLs. For non-risk-related data uses, substitution values should go no lower than the sample-adjusted onentrations equivalent to the orresponding level of the MDL hek sample for eah respetive target analyte. Half-value substitutions (i.e., one-half DL) for non-detets should not be employed when DL refers to either the SQL or the MDL hek sample onentrations due to the inherent unertainties assoiated with these values. If MDL values are not available, or reported MDL values have not been appropriately verified, then substitution values for non-detets should be limited to detetion threshold onentrations that are assoiated with a high degree of analytial ertainty (i.e., LRL, MRL, or MQL). Sine MQL is not a sample-adjusted quantity, it should not be used for field samples. By definition, MRL ould be defined as equal to the MDL, so the use of MRL for substitution would in this ase be related to the level at whih it is defined. Thus the LRL would be the most appropriate substitution value to selet, and aording to the guidane desribed above, the substitution of one-half LRL would be appropriate for non-detets in dredged sediment evaluations. Guidane in the Inland Testing Manual (USEP A/US ACE 1998) for substituting non-detets should be read arefully and should be fully understood before employing any of these tehniques. NCLUSIN: Analytial detetion limits are an integral part of environmental hemistry data olleted through dredged sediment evaluations. It is ritial that data users understand the meaning and appropriate use of detetion limits reported by the ontrat lab. Sine many different terms are ommonly used by ommerial laboratories, the data user should insist that ontrat labs learly define the exat meaning of reported detetion limits. The data user should have this information prior to submission of samples to the laboratory, so that the ability of the lab to meet projet-speifi TDLs and PALs an be assessed. The data user must be aware that many

9 fators an influene final LRLs issued by the ontrat lab, and should set TDLs to allow for unexpeted sample-speifi elevation of LRLs. Finally, the assignment of numerial values to non-detet data should be approahed autiously, and areful onsideration should be given to the unertainties assoiated with this ativity. Depending on the projet, substitution of the MDL, the LRL, or one half the LRL may be aeptable. Ultimately, the projet manager must make an informed deision regarding the appropriate use of analytial hemistry data with respet to the disposition of dredged material. A lear understanding of laboratory detetion apability, and the language used to onvey it, will greatly failitate the deision-making proess, and this in turn will inrease the likelihood that resulting deisions will be firmly grounded on solid analytial data. PINTS F NTACT: For additional information ontat one of the authors, Dr. Robert P. Jones, Robert.PJones@erd.usae.army.mil; or Ms. Joan U. Clarke, (601) , Joan.Clarke@erd.usae.army.mil, or the manager of the Environmental Effets of Dredging Programs, Dr. Robert M. Engler, (601) , Robert.M.Engler@erd.usae.army.mil. Funding to support preparation of this paper was provided through the Dredging perations Tehnial Support (DTS) Program, Dr. Douglas G. Clarke, Program Manager, (601) , Douglas. G. Clarke@erd. usae. army.mil. This tehnial note should be ited as follows: Jones, R. P., and Clarke, J. U. (2005). "Analytial hemistry detetion limits and the evaluation of dredged sediment," ERDC/TN EEDP-04-36, U.S. Army Engineer Researh and Development Center, Viksburg, MS. REFERENCES: Clarke, J. (1995). Guidelines for statistial treatment of less than detetion limit data in dredged sediment evaluations. Environmental Effets of Dredging Tehnial Note EEDP-04-23, U.S. Army Engineer Researh and Development Center, Waterways Experiment Station, Viksburg, MS. Clarke, J. (1998). "Evaluation of ensored data to allow statistial omparisons among very small samples with below detetion limit observations," Environmental Siene and Tehnology 32: Federal Register. (1995). "Definition and proedure for determination of the Method Detetion Limit," 40 CFR Part 136, Appendix B, Revision Hohberg, Y., and Tamhane, A. C. (1987). Multiple omparison proedures. John Wiley & Sons, New York, Southerland, E., Kravitz, M., and Wall, T. (1992) "Management Framework for ontaminated sediments (the U.S. EPA sediment management strategy)." Sediment Toxiity Assessment. G.A. Burton, Jr., ed., Lewis Publishers, Chelsea, MI, U.S. Army Corps of Engineers (USACE). (2001). "Engineering and Design Requirements for the Preparation of Sampling and Analysis Plans, Appendix I - Shell for Analytial Chemistry Requirements," EM , Washington, DC.

10 U.S. Department of Defense (USDoD). (2002). "Quality systems manual for environmental laboratories, Version 2, Final," U.S. Environmental Protetion Ageny/U.S. Army Corps of Engineers (USEPA/USACE). (1991). "Evaluation of dredged material proposed for oean disposal - Testing Manual." EPA-503/8-91/001. Washington, DC. U.S. Environmental Protetion Ageny/U.S. Army Corps of Engineers (USEPA/USACE). (1998). "Evaluation of dredged material proposed for disharge in waters of the United States - Testing manual (Inland Testing Manual)," EPA-823-B Washington, DC. Willard, H., Merritt, L., Dean, J., and Settle, F. (1988). "An introdution to instrumental methods." Instrumental Methods of Analysis. 7th ed., Wadsworth Publishing Co., Belmont, CA. NTE: The ontents of this tehnial note are not to be used for advertising, publiation, or promotional purposes. Citation of trade names does not onstitute an offiial endorsement or approval of the use of suh produts. 10

11 o o z> oo =*S 5 z *r*k r