Scribd
Upload
486 views9 pages

Iso 8466-1 VP

Uploaded by

Director Laboratorio MAHT Ltda
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
486 views9 pages

Iso 8466-1 VP

Uploaded by

Director Laboratorio MAHT Ltda
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 9
MERCK KGaA Darmstadt INTERNATIONAL Iso STANDARD 8466-1 First edition 1990-0301, Water quality — Calibration and evaluation of analytical methods and estimation of performance characteristics Part 1: Statistical evaluation of the linear calibration function Gualte de ren — Etalonnage et évatuation des méthodes d'analyse et estimation ties caractores de performance Parte 1: Evaluation statistique dela fonction linéaire détalonnage Normenstelle Reference numbor 150 8466-1 : 1960 (=) ISO 8466-1 : 1990 (E) Contents Foreword 1 Scope 2 Definitions 3 Symbols 4 Performance 4.1 Choice of working range. 42 Calibration ané characttiatos ofthe method 43 Assessment 5 Bamps 5.1. Choice of workina rane 52. Callbrationand characteristics ofthe method 53. Evaluation Annex A Bibliography 's0 1980 {Al rights reserved No art ot tis pubicaton may be reproduce ot utzed any former any ‘means, electors or mecranizel,Ieuding phlocopyng and microti without perma ‘ait trom the pablane Interationa Organization for Standardization Caso portne Ss CH 1211 Gondve 20 « Swterand Printed in Switzeiand 10 8466-1 : 1990 (E) Foreword 180 {the Intemational Organization for Standardization) is ¢ wordwde federation of rational standards oocies ISO member bodies!, The work of preparing Intemational Standards normally caved out through ISO technical commitees, Each member Dory intersted in @ subyect Ko” which a technical cornmittes has been etabzhed has the right to be vepresented on that cornitte. International organizations, govern mental and non-governmental, in liaison with ISO, also taks port in tne work. ISO Colaborates closely with the Intemational Ekectrorechn'cal Commission (IEC! on all matters of lectrotechnica! stanaarciation Draft International Standards adopted by the technical committees are circulated to ‘the member bodies for approval before ther acceptance as Inetnational Srandards by the ISO Counc They are approved in accordance with ISO procedures requiring a least 76 % approve’ by the member bodies voting, Intemational Stancate 150 8466-1 was prepared by Technical Committee ISO/TC 147, Woter qual 180 8466 consists of the follawing pasts, under the genera tile Water quality — Calibration and evawation of analytes! methods and estimation of portormance enaracrerstis = Part 1 Stavsteat evaluation of the tinea eatbration tanction — Part 2: Caliation strategy for non-linear calibration functions Pars 3° Method of standard addition — Part 4’ Estimation of fm of detection an limit of determinstion of an anaiyicat basis method INTERNATIONAL STANDARD 180 8466-1 ; 1990 (E) Water quality — Calibration and evaluation of analytical methods and estimation of performance characteristics Part 1: Statistical evaluation of the linear calibration function 1 Scape This part of ISO #466 describes the steps to be taken in evaluating the statistical enaractersties of te ines calibration ‘unetion. ti appicable to methods requiring a calioration. Fur ther gars of this international Standard will cover the deter: mination of limit af detection anc limit of determin he wfect of imerferences and other performance characteristics Itis intended especialy for the evaluation of the pure analytical methad and for the calewation of performance charactretics (of the ealoraion function In o10er 19 dorive comparable analytical results and as a basis for anaivical quality contol tho calibration anc evaluation of analyical methods have 10 9¢ performed sn form 2. Definitions For the purposes of this part of IS0 8466, the folowing defn tions apply 2.1 analytical method: An analytical method is composed ff procedural, measuring, calibrating anc evaluating instruc tions Ise Figure 1) Whereas the procedural and measuring instructions cepend on ‘the mathod, ane are sheretare tho abject of standardisation of tne respective method, the calbrating and evaluating instruc ‘ons ste valid for any analytical method requiring calibration 2.2 calibrating instruction: Describes the apprasch to determine the calibration function fram information values,» ‘obtained by measuring given standard concentations, x. The slope of the callvasion functian, b, a8 a measure of sensitivity fof the arslytical method and the standard deviation of the ‘method. 5,q, a8 figuras of mest end charactenstice which result from the calibration experment ‘The standard deviation, 5, allows the comparison of indepen ent analvtca! methods For the user of the method, these charactenstice present itera forthe meznal laboratory quaity contra. Original sam Measuring sample measuring inseuctions tallorating ane evalsting iets 1 Analytical result Figure 1 ~ The analytical method 23. evaluating insttuction: A caleulation quide for the ‘computation of cancentratons from the measured values by Us use of Ue earian function. Aaairanaly, the contcence rango permits an objective assessment of the mpracison ofthe ‘analytical vault? 2.4 measured values: The concentration-dependent intial values (e.g) extinction! of @ measuring system NOTE — Information value ant mussurad volume are smonynous 2.5. residual standard deviation, s,: The residual standard evistion describes the scatter ofthe information values about the calelates regression ine, (ta figure of merit. describing the precision of the calibration method means the standard of deviation of the calloation pro eodure 1SO 8486-1 : 1990 (E) 2.8 standard deviation of the method ,: The rato of the residual standard deviation, s,, 10 the sensitivity of the calibration function, b. It a figure of merit for the perfor mance of the analytical method, ands valid within tne working range (see equation 13) For the purpose of this standard, the standard deviation of the method means the stancard of deviation of the caibration pro cedure 2.7. costticient of varistion of the method, V9: The ratio ‘ofthe standard deviation of the method 5,10 the appertaining ‘mean, x, which ig the contre of the working range. ‘Soe alsa note to 2.6 and 26. 2.8 working range (of an anslvtical method): The interval being experimentaly estabishod anc statically proved Oy he calibration of the method, between the lowest and highest ‘Quantity or mass concentration. The lowest possible init of 8 ‘working ange isthe Imt of detection ofan analytical mothod. 2.9 homogeneity of v Homogeneity of vances ‘of pooled data, uch as those resulting from replicate analyses ‘at different levels, is confiemed f those variances are not Significantly correlated t0 thor appertaining concentrations. the ealivration function ofthe compete analytical method, in clusive of all procedural steps, within the working range in ‘question 2.11 measuring sample (reaction sample): A sample wh ‘can be directly submited to the measurement of the determi rand, A measuring sample ie normaly obtained by acding the required reagents tothe analytical sample. Ooviously, measur ing and anaiytca sample are identical fro reagents nave tobe ‘added to the analytical sample 3. Symbols x Concentation of the standard sample : Subscript of the concentation love's, where Pa AQ WN N ‘Number of concantration lavels (for this par of 180 Ba8R, N= 10 ~ Concentration of the standard samp atthe lower lave! of the working range (1st standard sarmpel Xo Concentration of the standard sample at the upper level of the working range (Ith standard sample) My {J information value for the concentration», j Subscript ofthe replicate jaf eval, where, = 1 Bonet n, Number of eplcates per level x y ‘Mean of the information values y,, of standard samples, having the concentration 3, Hy tmal Foal Keo vate vain Information value of the standard concentra tion x caleuated from the caloration function Variance of the information values for the analyses of standard samples, raving the cor Degrees of freedom for the calculation oft vatiance = n=. Calculates blonk (ordinate intereopt ofthe ca bration straight line) Sensitivity of the method Islope of the ca bation line; covtficiant of ragrassion! Moon of the standard concentrations resulting from the calibration experimen [Mean of the information values », resulting fom the calibration experiment Residual standard devition, Residual stancare deviation obtained by ines regression calculation Residual standard deviation obtained by nor linear regression calculation Diterence of variances Information valve of an analysed sample Number of replicates on the same analysed sample. Moan of information values, resulting from replicates. Concentration af the analtical sample calculated trom the information vale» Concentration of the analytical sample, calculatze from the mean of the information values LN ~ 2 degrees of freedom anc a confide level of (7 — a! l-factorof Student's dstibi tion Tabled value of the Fasthbution (Fisher Snedecor! with f, and fy degrees of freedom and a confidence level of (1 — a Standard deviation of the method. Coetficent of variation of the method Confidence intarval forthe concentation Confidence interval of the mean Tf the con: centration 4 Performance 4.4 Choice of working range Each calbrtion expeimant i started with the choice of a preliminary working range ‘The warking range depends on 8) the practce-elated objective of the calibration The working range shall cover, as fer as possible, the appiiation range for water, waste water, and sludge analy- sis, The most frequently expectec sampe concentation ‘should lain the centre of the working range. 1b) teasibiities of tochnical reaizaiity. The measured values obtained must be Lneariy correlates to the concentrations. This requires thatthe measured values obtained near the lower limit of the working range can be istingu'shed trom the blanks of the method. The lower limit of the working range should therefore be equal to or areater then the limit of detection of the method. Diution and concentrating st9ps should be feasible without the Fisk of 85 )_the vatiance ofthe information values must be inden dent of ene concentration ‘The independence 's vertied by a statistical test on the linearry’® 4.11 Proparation of the calibration After establishing the preliminary working range, measured values. of at loast five (recommended N= 10) standard samoles are determined. The concentrations, x, of these stan atc samples shal be dstibuted equidistant over the working range. In order to check for the homogeneity of the varances, ten repicates of each of the lowest and the highest concentra tions lx; and 9) af the working range are determined. Ten in formation values, , e8u' from these series of measurements Isee table 1 ISO 8466-1 : 1990 (E) Both dat sets of the cancentrations x, and xp are used t0 ccsleulate the variances s? and sf as given in equation (1) with the mean Lo tori tert = 10 2 The variances are tested (F-test for significant diMerences at the limits of the working range 5 The test value PG is determined for the F-test from equation (3 Pos? tors >t @ Po Wrst > oy is compared withthe tabled values ofthe F-distibutions! Desi: 8) PG < Fa. 5, om the difference between the vvanances s] and 23 s ot significant DI IPG > Fy. og the diference between the variances «2 and 52s signiican. If the differance between the variances is significant, the preliminary working "ange should be made smaller until the di ference between the variances is found to be random ony Table 1 — Data sheet for the calibration 180 6466-1 : 1990 (€) 4.13. Test for linearity’? 6.8! rity! ‘The easost test forthe lingacty is the graphical reaesentation| of the calbration data witn the caleuiaced regression line. Any tinkinesriy ie ant fe gure 2 123456789 Figure 2 — Graphical linearity check In the stetisical linearity test the cabbation data are used to Calculate @ linear calibration turetion 3s well as a nor hinear ‘calbration function, bath withthe residual standord dewtion “The citferance of the variances DS? s calculated from equation ai DS? = I= 2182, IN - 318, “) agraes of freedom: f= 1 (0? and the variance of the non inar calbration function 52 fare submited toa Fast n order to examine fr srificant ct ferences. ‘The test value PG rexqliod for the Fest is ealeulated trom equation 8) os? Pc : 6 Decision: 8) EPG < F: The non-linear calioration function does rot lead toa signtficamiy beter adjustmert €.. the caloration function i near bi IFPG > F: The working range shouldbe reduced as tar 35 possible to receive linea callbratin function; otnervise the information values of analyzed sampies must be ‘evaluated using the non lies" callration function. 4.2 Calibration and characte: ties of the method [After the final working range \s established, ten standarg samples are analyzed in aecoreance with all the step of the Analytical method in order to obtain ten IN’ = 10) measures ‘values (son table 2). “The mwasurement against a blank isnot alowed, since thereby \aluabi information on the magritude ofthe blank wil be lost “The comparison medium for zeoing the nstrument is always possible, 3 pure solvent {e.g pure water! Teble 2 ~ Dow set for simple ner eeresion ST a 7 | 2 a4 | 5 | é I ‘The ten date sets, consisting of the values of x, andy, are sub mitted to # linear regression analysis to abtan the cootficionts «| 4nd bof the calation ntion which describe the ine cor felation between the concentration as an indogendent| ‘arable, snd the measured value y 2s & dependent varinle ‘The calibration function as well as the characteristic of the ‘ethad shoul el ftom data obained from a wotk'ng range 4 f0.%,9 35 teceved from the measurement and not corectes {or blanks. Generaly, no biank value (concentration x = 0) is tobe included in the calibration experiment and, consaquently, 1 te least squares ft of the regression The tnear calibration function is given by equation (6! yeas he ‘6 ‘The cocticints ne obtained frarn equatione (7 for sanity fsloce of the caliraion function) and (Bi for the ordinate intercept tcalculated blank} o-5 be 8 ‘The cooHicients prove an estimate ofthe rue function, whieh s ivited by the unavoidable procedural scatter. The precsion fof the estimates quant fied by the residual standard deviascn, S. Which is 2 measure of the scare ofthe information values about the calibration ine ane is given by aquation [9 ® 4.3 Assessment “The concentration of an analyzed sample is abtained ‘al trom the measured value », 10 give soe 9} ‘bl trom the mean ofa series of replicates, ¥, performed on the same original sample, to give © a 3 {As tothe uncerainty of an analytical result, keepin mind that the analytical ertor is @ combination of the uncertainty of the determination of the measured valve, and the uncanainty of the estimation of the regrossion cootticents” ¥ ss ISO 8466-1 : 1990 (E) From the law of error propagation it follows that, for wach value 4, 8 confidence interval for the true value ) exists whose limiting points are an two hyperbolic paths bracketing the caltation line Between these paths the tue calibration func tion can be expected with a significance level af Uf): N- 2 contidence lever = 1 ~ al, cetermined by Student's factor. The confidence intervals for analytical results, calculated from the calibration function, are gven by the intersections with the respective hyperbolic patns in fiqute 3. The estimation of the confidence intervals are given by equation (12) 7 f= vet ua Equation (12! indicates that the confidence interval VBI) brackets the true analytical value wth a range governed by the statistical security of Student's sistibution, The magnitude of VBL! is mainly determined by the numberof eplicates and Figure 3 — Working range x; t0 x ca apperta ¥ ration line with confidence band and a single analytical result with ite ing confidence interval 180 8466-1 : 1990(E) Annex A (informative) Bibliography IIL Vonoenie, ©, Davey, V., Ouse, W., Funk, W. and Knurz, H., Statistical methods ane performance characteristics for the assessment and comparson of arsiyticel procedures. Ar approach to standardaation, Vom Wasser 87. 8811, pp. 58-74 |} Manos. J., The statistical analysis of experimental dats, Interscience PUBL, J. Wiley & Sons, (1968), New York 131 Gorrscaa.s, G., Standardization of quantitative analytical procedures, Z. Anal, Chem. 275, 11876), pp. 110 |) France, J.P, de Zetw, A, and Haxkeny, A, Evaluation and optinzation of the standard adaition method for stom tion spectrometry and anodic etiaping voltammetry, Anal Cher. 80, (7978), pp. 138 1380 sore IB] Grae, U., Hewunc, HJ. and SraNat, K., Formulae and tales of mathematica statistics, 2nd Edtion, Savinger Verlg, (1966) Berlin, Heidelberg, New York I6]_ Sacis,L., Methods for statistical evaluation, 3a Edition, Springer Verlag, (1871), Berlin, Heidelberg, New York 171 Browycee, K.A., Statistical theory and methodology in scence an engingerng, J. Wiley & Sons, (1965), New York 18). Doesese., K., Statistics in chemi 5 analysis, VEB-Vertag fir de Grundstoffindusre, (196), Leipzig, (9) acne, A, Evaluation of BOD-

You might also like