Appendix-B-to-40-CFR-Part-136-MDL

e-CFRDataiscurrentasofJanuary4,2011Title40:ProtectionofEnvironmentPART136—GUIDELINESESTABLISHINGTESTPROCEDURESFORTHEANALYSISOFPOLLUTANTSBrowsePrevious|BrowseNextAppendixBtoPart136—DefinitionandProcedurefortheDeterminationoftheMethodDetectionLimit—Revision1.11DefinitionThemethoddetectionlimit(MDL)isdefinedastheminimumconcentrationofasubstancethatcanbemeasuredandreportedwith99%confidencethattheanalyteconcentrationisgreaterthanzeroandisdeterminedfromanalysisofasampleinagivenmatrixcontainingtheanalyte.ScopeandApplicationThisprocedureisdesignedforapplicabilitytoawidevarietyofsampletypesrangingfromreagent(blank)watercontaininganalytetowastewatercontaininganalyte.TheMDLforananalyticalproceduremayvaryasafunctionofsampletype.Theprocedurerequiresacomplete,specific,andwelldefinedanalyticalmethod.Itisessentialthatallsampleprocessingstepsoftheanalyticalmethodbeincludedinthedeterminationofthemethoddetectionlimit.TheMDLobtainedbythisprocedureisusedtojudgethesignificanceofasinglemeasurementofafuturesample.TheMDLprocedurewasdesignedforapplicabilitytoabroadvarietyofphysicalandchemicalmethods.Toaccomplishthis,theprocedurewasmadedevice-orinstrument-independent.Procedure1.Makeanestimateofthedetectionlimitusingoneofthefollowing:(a)Theconcentrationvaluethatcorrespondstoaninstrumentsignal/noiseintherangeof2.5to5.(b)Theconcentrationequivalentofthreetimesthestandarddeviationofreplicateinstrumentalmeasurementsoftheanalyteinreagentwater.(c)Thatregionofthestandardcurvewherethereisasignificantchangeinsensitivity,i.e.,abreakintheslopeofthestandardcurve.(d)Instrumentallimitations.Itisrecognizedthattheexperienceoftheanalystisimportanttothisprocess.However,theanalystmustincludetheaboveconsiderationsintheinitialestimateofthedetectionlimit.2.Preparereagent(blank)waterthatisasfreeofanalyteaspossible.Reagentorinterferencefreewaterisdefinedasawatersampleinwhichanalyteandinterferentconcentrationsarenotdetectedatthemethoddetectionlimitofeachanalyteofinterest.Interferencesaredefinedassystematicerrorsinthemeasuredanalyticalsignalofanestablishedprocedurecausedbythepresenceofinterferingspecies(interferent).Theinterferentconcentrationispresupposedtobenormallydistributedinrepresentativesamplesofagivenmatrix.3.(a)IftheMDListobedeterminedinreagent(blank)water,preparealaboratorystandard(analyteinreagentwater)ataconcentrationwhichisatleastequaltoorinthesameconcentrationrangeastheestimatedmethoddetectionlimit.(Recommendbetween1and5timestheestimatedmethoddetectionlimit.)ProceedtoStep4.(b)IftheMDListobedeterminedinanothersamplematrix,analyzethesample.Ifthemeasuredleveloftheanalyteisintherecommendedrangeofonetofivetimestheestimateddetectionlimit,proceedtoStep4.Ifthemeasuredlevelofanalyteislessthantheestimateddetectionlimit,addaknownamountofanalytetobringthelevelofanalytebetweenoneandfivetimestheestimateddetectionlimit.Ifthemeasuredlevelofanalyteisgreaterthanfivetimestheestimateddetectionlimit,therearetwooptions.(1)Obtainanothersamplewithalowerlevelofanalyteinthesamematrixifpossible.(2)Thesamplemaybeusedasisfordeterminingthemethoddetectionlimitiftheanalyteleveldoesnotexceed10timestheMDLoftheanalyteinreagentwater.ThevarianceoftheanalyticalmethodchangesastheanalyteconcentrationincreasesfromtheMDL,hencetheMDLdeterminedunderthesecircumstancesmaynottrulyreflectmethodvarianceatloweranalyteconcentrations.4.(a)Takeaminimumofsevenaliquotsofthesampletobeusedtocalculatethemethoddetectionlimitandprocesseachthroughtheentireanalyticalmethod.Makeallcomputationsaccordingtothedefinedmethodwithfinalresultsinthemethodreportingunits.Ifablankmeasurementisrequiredtocalculatethemeasuredlevelofanalyte,obtainaseparateblankmeasurementforeachsamplealiquotanalyzed.Theaverageblankmeasurementissubtractedfromtherespectivesamplemeasurements.(b)Itmaybeeconomicallyandtechnicallydesirabletoevaluatetheestimatedmethoddetectionlimitbeforeproceedingwith4a.Thiswill:(1)Preventrepeatingthisentireprocedurewhenthecostsofanalysesarehighand(2)insurethattheprocedureisbeingconductedatthecorrectconcentration.ItisquitepossiblethataninflatedMDLwillbecalculatedfromdataobtainedatmanytimestherealMDLeventhoughthelevelofanalyteislessthanfivetimesthecalculatedmethoddetectionlimit.Toinsurethattheestimateofthemethoddetectionlimitisagoodestimate,itisnecessarytodeterminethatalowerconcentrationofanalytewillnotresultinasignificantlylowermethoddetectionlimit.Taketwoaliquotsofthesampletobeusedtocalculatethemethoddetectionlimitandprocesseachthroughtheentiremethod,includingblankmeasurementsasdescribedabovein4a.Evaluatethesedata:(1)IfthesemeasurementsindicatethesampleisindesirablerangefordeterminationoftheMDL,takefiveadditionalaliquotsandproceed.UseallsevenmeasurementsforcalculationoftheMDL.(2)Ifthesemeasurementsindicatethesampleisnotincorrectrange,reestimatetheMDL,obtainnewsampleasin3andrepeateither4aor4b.5.Calculatethevariance(S2)andstandarddeviation(S)ofthereplicatemeasurements,asfollows:where:Xι;i=1ton,aretheanalyticalresultsinthefinalmethodreportingunitsobtainedfromthensamplealiquotsandΣreferstothesumoftheXvaluesfromi=lton.