Tiêu chuẩn iso tr 18231 2016

© ISO 2016 Iron ores — Wavelength dispersive X ray fluorescence spectrometers — Determination of precision Minerais de fer — Spectromètres à fluorescence à rayons X à longueur d’onde dispersive — Déte[.]

Iron ores — Waveleng th dispersi ve

Mine ais de fe — S ectromètre s à flu r scenc à ra o s X à lo g eur

d’o de dispe sive — Déte mination de la précision

Refer ence n mb r

ISO/TR 1 2 1:2 16(E)

Fir t edition

2 16-0 -0

COPYRIGHT PROTECTED DOCUMENT

© ISO 2016, P blshed in Sw itz rlan

A ll rig hts r eserved Unles otherw ise spe ified, nopar of this p blc tion ma y be r epr od c d or utilz d otherw ise in an form

or b an me ns, ele tr onic or me hanic l, inclu in p oto opying , or postin on the internet or an intranet , w ithout prior

written permis ion Permis ion c n be req esed from either ISO at the ad r es below or ISO’s member bod y in the c u try of

F reword i v

Introduction v

1 Sc ope 1

2 F requency of testing 1

3 Co nter tests 2

3.1 Cou te r esolution 2

3.1.1 General 2

3.1.2 Pr oc d re 4

3.1.3 A sses ment of results 6

3.2 Cond ctivity of the gas flow pr op rtional cou te win ow 6

3.2.1 General 6

3.2.2 Pr oc d re 7

3.2.3 A sses ment of results 7

3.3 Pulse shif corr ector 7

3.3.1 General 7

3.3.2 Pr oc d re 8

4 Spectromete tests 8

4.1 General 8

4.2 Pr ecision 9

4.2.1 General 9

4.2.2 C lculation of cou ting s atis ical e r or 1

4.3 Tes spe imen 1

4.3.1 General 1

4.3.2 S q ential spe tr omete s .1

4.3.3 Simultaneous spe tromete s 1

4.4 Ins rumental con itions 1

4.4.1 General 1

4.4.2 S q ential spe tr omete s .1

4.4.3 Simultaneous spe tromete s 1

4.5 Stabi ty tes 1

4.6 Spe imen r otation tes 1

4.7 Carousel repr od cibi ty tes 1

4.8 Mou ting an lo ding r epr od cibi ty tes 1

4.9 Comp rison of sample holde s 1

4.1 Comp rison of carousel p sitions 14

4.1 A ng ular r epr od cibi ty 14

4.1 Col mator repr od cibi ty (or seq ential spe tr omete s fited w ith an inter chang ea le col mator) 14

4.1 Dete tor chang ing r eprod cibi ty (or seq ential spe tromete s fitted with mor e than one dete tor) .14

4.14 Crys al chang ing r epr od cibi ty 14

4.1 Othe tests 1

4.1 Note on glas bead curv tur e 1

5 Deter minatio of the de d time and the max imum usable c ount rate of the equipment 15 5.1 General 1

5.2 Methodsof dete mination of dead time

1 5.2.1 General 1

5.2.2 Re ommended method for dete mining dead time 1

A nne x A (informative)Calculation of the c oeficient of var iation of dupl cates

24

Biblog raphy 26

ISO (he Int ernational Org nization for Stan ardization) is a worldwidefede ation of national s an ards

b dies (ISO membe b dies) The work of pr p ring Int ernational Stan ards is normaly car ied out

through ISO t ech ical committ ees Each membe b dy int er st ed in a subje t for w hich a t ech ical

committ ee has be n es a lshed has the right t o be r pr sent ed on that committ ee Int ernational

org nizations, g ove nmental an non-g ove nmental, in laison with ISO, also take part in the work

ISO cola orat es closely with the Int ernational Ele trot ech ical C mmis ion (IEC) on al matt ers of

ele trot ech ical s an ardization

The proc d r s used t o develo this document an those int en ed for it furthe maint enanc ar

desc ibed in the ISO/IEC Dir ctives, Part 1 In p rticular the dife ent a pro al c it eria ne ded for the

dife ent ty es of ISO document should be not ed This document was draft ed in ac ordanc with the

edit orial rules of the ISO/IEC Dir ctives, Part 2 ( e www.iso.org dir ctives)

A tt ention is drawn t o the p s ibi ity that some of the element of this document ma be the subje t of

p t ent right ISO shal not be held r sp nsible for identifying any or al such p t ent right Detais of

any p t ent right identified d ring the develo ment of the document wi be in the Introd ction an / r

on the ISO ls of p t ent de larations r c ived ( e www.iso.org p t ent )

Any trade name used in this document is information given for the convenienc of use s an does not

cons itut e an en orsement

F or an ex lanation on the meaning of ISO spe ific t erms an ex r s ions r lat ed t o conformity

as es ment, as wel as information a out ISO’ s adhe enc t o the WTO principles in the Te h ical

Bar ie s t o Trade (TBT) se the fol owing URL: F or word - Sup lementary information

The committ ee r sp nsible for this document is ISO/TC 1 2, Iro or a d dir c t r duc ed iro ,

Subcommitt ee SC2, Chemic al a alys i s

If an X-ra fluor sc nc spe tromet er is t o be used for pr cise analy ses, it ne ds t o be fu ctioning

cor e tly t o spe ification, that is, the e rors as ociat ed with the v rious functions of the ins rument

ha e t o be ve y smal It is important the efor that the spe tromet er be t est ed t o ensur that it is

in e d fu ctioning t o delve the r q ir d pr cision The o je tive of this Te h ical R ep rt is t o set

out t es s that can be used t o asc rtain the ext ent of the e rors an t o sugg es proc d r s for their

r ctif ication These t es s ar not used t o asc rtain w hethe the ins rument is o e ating o timaly but

t o det ermine w hethe the ins rument is ca a le of giving a pr sele t ed pr cision

Iron ores — Waveleng th dispersi ve X-ray fluor esc enc e

This Te hnical R ep rt desc ibes methods of t es that can be a pled t o wa elength dispe sive X-ra

fluor sc nc (WD- XRF) spe tromet ers t o ensur that the spe tromet ers ar functioning in a manne

that alow s pr cise analy ses t o be made

The t es s outl ned ar designed t o measur the e rors as ociat ed with the o e ation of c rtain p rt of

the spe tromet er They ar not designed t o che k eve y part of the spe tromet er but only those part

that ma be the common sour es of e ror

It is as umed that the pe formanc of the ins rument has be n o timiz d ac ording t o the

man factur r’ s ins ructions F or al t es s, the two-theta angle should be car fuly set for the lne being

measur d The pulse height win ow should be set ac ording t o the man factur r’ s ins ructions an

should ha e a bro d set ing w hich ma also inclu e the esca e peak for g s pro ortional count ers The

ins rument an det ect or g s environment should be as spe if ied b the man factur r, as should the

p we sup ly t o the ins rument

NOTE Where no distinction has b en made, it is as umed that a test is ap lica le t o b th seq ential an

simultane us spectrometer

2 F requency oftesting

Tes ing is not r q ir d t o be car ied out with each b t ch of analyses The fr q ency of t es ing v ries

depending on the t es inv lved Ta le 1 l s s the sugg est ed fr q ency with w hich each t es should be

car ied out Whe e spe if ic pro lems ar encou t er d, mor fr q ent t es ing ma be r q ir d an

r mediation work pe formed

Table 1 — Sugg ested frequency of precisio tests

Monthly R esolution of the g s-flow proportional cou ter

R esolution of the s intil ation an se led g s cou ter

Operation of the p lse height shif cor ector

Half yearly Con uctivity of g s-flow proportional cou t er win ow

General sta ility

Col imat or reprod cibility

Detector changing reprod cibility

Cry stal changing reprod cibility

Angular reprod cibility

Ye rly Carousel reprod cibility

Comparison of carousel positions

Comparison of sample holder

S mple lo ding an u lo ding

a

The position of the p lse heig t pe k should also be che k ed afer chan in

a botle of dete tor gas sinc a variation in the methane c ntent of the gas wi

chan e the position of the pe k

The fr q encies with w hich the t es s lst ed in Ta le 1 ar car ied out ar sugg est ed on the b sis that

the e ha e be n no chang es t o the spe tromet er If me hanical or ele tronic maint enanc of a major

natur is car ied out, the a pro riat e t es s should be made befor the spe tromet er is taken b ck int o

Impurities in the flow g s and contamination of the anode wir ma ca use g s flow pro ortional

count ers t o grad al y det eriorat e, w hich wi l r sult in b th a shif an a bro dening of the ene g

dis ribution (pulse height) curve Simi arly, scinti lation cou t ers an sealed g s cou t ers ma , for

v rious r asons, exhibit the same grad al det erioration This can, ultimat ely, adve sely afe t the

measur ment Impurities in det ect or g scan be minimiz d b the use of g s filt ers

The r solution ( RES)of a cou t er is r lat ed t o it ene g dis ribution curve, an is given b the measur d

peak width at half height (W ) ex r s ed as a pe c ntag e of the ma imum of the pulse ampltu e

dis ribution (V ), using F rmula (1) w he e the v lues of Wan Var in t erms of arbitrary u it (w hich

v ry betwe n ins rument man factur rs) o tained from the X-a is ( e Figur 1):

calculat ed using the folowing formulae:

RES

th

n is then mbe of primary ele trons pe incident phot on (g scount ers) or n mbe of

phot oele trons cole t ed b the f irs dynode of the phot omultiple tube ( cinti lation

coun-t ers), calculat ed using F ormula (5):

Henc , for Cu K α ( E = 8,04 ke V), the theor tical r solution of an Ar g s cou t er is 1 ,5 %

3.1.1.2 Scintilation c ounter

F or a scinti lation cou t er:

R

EES

m

w he e

k is a fact or that v ries with the design of the count er, phosphor eff iciency ( cinti lation

cou t ers), diamet er, cleanl nes an comp sition of the anode wir (g s cou t ers)

F or a wel-designed an clean g s-flow pro ortional cou t er, k should be les than 1,1 Henc , for such

a cou t er, RES

mshould be les than 1 ,6 % for Cu K α radiation F or the scinti lation count er, this v lue

should be les than 5 %

3.1.2 Proc edure

This t es should be car ied out on al cou t ers used in the spe tromet er Mos mode n ins rument

pro ide the faci ity t o measur pulse height dis ributions an t o print out the count er r solution an

thisfaci ity should be used if a ai a le

F or seq ential spe tromet ers, it is r commen ed that he t es be car ied out using eithe Cu Kα or F e K α

radiation for b th det ect ors Howeve , if these l nes ar measur d using only the scinti lation count er

in actual analy sis, measur an X-ra lne of a major element analysed with the g s pro ortional count er

for t es ing

If the spe tromet er does not pro ide a ut omatic fu ctions t o det ermine RES

mthen the folowing

proc d r should be used

a) S le t a sample containing the a pro riat e analyt e an , using a lowe level set ing an the pulse

height analy se (PHA) win ow set t o “ hr shold” (no up e level), adjus the X-ra tube p we t o

give a count rat e of a out 2 × 1

4

cps (cou t pe se on )

b) S le t a nar ow pulse height win ow (2 % t o 4 % of the peak v ltag e V of Figur 1) an de r ase

the lowe level set ing u ti the cou t rat e dro s t o es entialy z ro

c) Inc ease the lower level st epwise, noting the cou t rat e at each st ep, unti the peak has be n p s ed

an the cou t rat e dro s a ain t o a ve y low v lue Each st ep should be of the same width as the

pulse height window width, i.e if the pulse height win ow width cor esp n s t o 0,2 u it , then

each st ep of the lowe level should be 0,2 u it

d) Plot the cou t rat e o tained at each st ep a ains the lowe level v lues An ex mple is shown in

Figur 1

Key

1 half pe k heig ht

2 pe k w idth at half pe k heig ht

Figure 1 — Intensity as a functio of lower level set ing (arbitrary energy units) displayed for

F e K α radiation me sured o a flow counter

The measur d cou t er r solution RES

m( in % ) iso tained from the plot as folow s:

RES

Vm

=

−

w he e

B an A ar the lowe level set ings at the half height positions on eithe side of the peak ;

V is the lowe level set ing at he ma imum of the pulse height dis ribution

The pulse height dis ribution should be det ermined at a cou t rat e in icat ed b the ins rument

man factur r, or jus below 2 × 1

4

cps ( e Figur 1)

When det ermining the pulse height dis ribution aut omaticaly, A, B , Vand W ar not usualy displa ed,

but the gra h, peak p sition an det ect or r solution ar shown t og ethe with ins rument pe if ic data

3.1.3 A sses ment of results

3.1.3.1 Gas flow propor tio al c ounter

Sinc the r solution ofa pro ortional count er depen s on design, an a solut e r solution v lue cannot be

given but he v lue should be near t o that given in F ormula (6) The o timum r solution of a p rticular

count er ma be spe if ied b the man factur r or it ma be fou d with use When RES

minc eases b a

fact or of 1,2 times that v lue, the count er should be se vic d

3.1.3.2 Scintilation and se led gas c ou ter

These cou t ers usualy do not exhibit signif icant chang es in r solution u les the cou t er is beginning

t o fai , w hich can be q it e early for some sealed g s count ers for lght element an w he e they ha e

be n used at ve y high flux for ext en ed pe iods Their r solution can be calculat ed ac ording t o

F rmula ( 9) but the k fact or is ins rument spe ific Henc , it is suff icient t o measur their r solution

ex e imental y on a r gular b sis If a signif icant inc ease in r solution is not ed or the pulse height

dis ribution ext en sout idethe ene g win ow w hen it did not on ins alation, then the cou t er should

be r plac d

S inti lation cou t ers can fai d e t o leaks in the be y llum win ow, the eb admit ing mois ur t o the

very hy grosco ic NaI c y stal The efe t of such mois ur r d c s cou t rat es for the long er wavelengths,

e.g ~2Å but ma ha e r latively lt le efe t on short wa elengths, ~0.6Å Asmor mois ur r act with

the c y stal, it sensitivity wi l be r d c d for al wa elengths

Simi arly, the sensitivity of sealed count ers wi r d c d e t o leaks in the det ect or win ow A leaking

sealed g s cou t er win ow wi l r sult in a los of g s density an th s g sampl fication

Although a r d ction in sensitivity of the scinti lation an sealed g s cou t ers does not afe t cou ting

pr cision, the signal-t o-noise ratio wi be r d c d an long er counting times wi l be r q ir d t o achieve

a given pr cision Furthe mor , if the r d ction in sensitivity is d e t o faiur of the win ow, then this

r d ction wi be contin ous an t otal det ect or faiur wi be imminent

Thebes method of det ecting det erioration of such count ers is t o set aside a s a le spe imen ( uch as an

ins rument monit or) an t o pe iodicaly measur the int ensity of a r asona ly long wa elength l ne (for

ex mple, F e K α for the scinti ation cou t er) Any pulse height shif or de r ase of int ensity o e time

should be inves ig t ed t o asc rtain w hethe the scinti ation (or sealed g s)count er r q ir s r placing

3.2 Conducti vity of the gas flow propor tional counter window

3.2.1 General

The g s-flow pro ortional count er win ow consis s of a thin la e of plas ic mat erial ( uch as My lar or

p ly ro y lene)co t ed with a ve y thin ele tricaly con uctive la e ( y icaly aluminium) F or practical

conside ations, the window mat erial should be a le t o sup ort atmosphe ic pr s ur without br akag e

Whe e air or helum p ths ar fr q ently used o e a pe iod of time, this mat erial ma s r t ch an ca use

the aluminium co ting t o sufe from mic o-c acks, r sulting in the win ow be oming non-con ucting

A g s-flow pro ortional cou t er ha ing a p orly con ucting or non-con ucting win ow a pears t o

fu ction normal y in mos r spe t Howeve , measur ment made with such a count er wi l be in e ror,

as the count er wi l give r d c d cou t rat es for samples w hich would normal y give a high cou t rat e

This is espe ial y notic a le w hen low an high count rat es ar measur d conse utively

Whe e 6 μm win ow s ar used, los of con uctivity is seldom encou t er d, as the win ow usualy has

a lfe of some years, r q iring r plac ment only w hen the count er is se vic d for othe r asons The

useful l fe of thinne win ow s (≤ μm p ly ro y lene win ow s ar commonly used in mos mode n

spe tromet ers) v ries from seve al months t o mor than a year Whe e the spe tromet er chambe

is contin ously u de v cu m, the win ow lfe is long er than for those w he e the spe tromet er is let

down t o normal air or Hel um pr s ur fr q ently

NOTE If the g s-flow proportional cou t er has a b ry llium win ow, this test may b neces ary to oc asional y

check for the formation of hairline c acks

3.2.2 Pr oc edure

3.2.2.1 S quential spectr ometer s

The proc d r for seq ential spe tromet ersis as folow s

a) Car y out the t es using K Kα radiation

b) Using low X-ra tube p we (low kV an mA), sele t a sample that gives a cou t rat e of between

10 0 cps and 5 0 0 cps

c) S t the 2θ angle an pulse height analyse using this sample an then, using these set ings, measur

thecount rat e for 1 s

d) R eplac the sample with one ha ing a high potas ium conc ntration (a briq ett e of p tas ium

chloride or p tas ium hydrog en phthalat e is p rticularly suita le or a sample advised b the

ins rument man factur r); set the X-ra tube t o such a p we that he ma imum cou t rat e alowed

b theman factur r is achieved Maintain these con itions for 2 min

e) R ed c the X-ra tube p we t o it s arting v lue an measur the count rat e for the original

sample a ain

3.2.2.2 Simultaneous spectr ometer s

F or simultaneous spe tromet ers, the t es is made on those channels ha ing g s-flow pro ortional

count ers fitt ed with thin plas ic win ow s The t es is car ied out in a simiar manne t o the proc d r

spe if ied in 3.2.2.1, ex cept that the radiation used is that spe if ic t o the sele t ed channel

NOTE Tests for simultane us spectromet er are also ap lica le t o f ix ed chan els in a seq ential

spectromet er

3.2.3 A sses ment of results

If the win ow is con ucting pro e ly, the initial cou t rat e wi l be within 5 % of the final count rat e

w hen the X-ra tube p we is r turned t o the original set ings If the win ow is po rly cond cting, a

r d c d cou t rat e wi be o se ved In such cases, the cou t rat e should slow ly r turn t o the original

rat e but or non-con ucting win ow s, the time inv lved ma be in ex ces of 1 h The win ow should be

r plac d if the e is any evidenc of p or con uction

3.3 Pulse shif c or r ector

3.3.1 General

A t high cou t rat es, high det ect or g ins wi l r sult in a shif t o lowe pulse ampltu es In mode n

spe tromet ers, the efe t of pulse shif ha e be n minimiz d b designing cou t ers with a lowe

dependenc on g s density an comp sition an b the use of a ut omatic pulse shif cor e tion

ele tronics Neve theles , the efe t of pulse shif e rors on analytical ac uracy can be signif icant

Al mode n spe tromet ers ar f itt ed with a ut omatic pulse shif cor e tion but the o e ation of

det ect ors should also be che ked w hen the a ut omatic shif cor e tion is not o e ating at low cou t

rat es The thr shold cou t rat e at w hich the aut omatic shif cor e tion o e at es can v ry betwe n each

det ect or ty e, ins rument model an spe tromet er man factur r This ma even be pro ramma le

with aut omatic pulse shif disa led or ena led in sof war or hardwar

F or g s-flow det ect ors, sinc the pulse shif is depen ent on the g s comp sition (pro ortion of arg on

t o methane an the level of contaminant ) an g s density (t empe atur or pr s ur of the g s within

the det ect or), pulse shif should be che ked w heneve the flow g s sup ly cy lin e has be n r plac d or

w heneve the g s flow det ect or or g s density s a i iz r has be n r p ir d

NOTE Le ks in the g s-flow cir uitry may result in the er one us entry of air, or helium in the case of liq id

me surements, which wil alter the flow g s composition

As mode n spe tromet ers ar fitt ed with g sdensity s a i iz rs of signif icant v lume, det ect or g s lnes

should be alowed t o purg e prior t o che king pulse height aft er any of the a o ework has be n car ied

out As the v lume of flow g s cir uit ma be up t o 3 l, the purging oflnes ma take up t o 3 h u les

the p s ibi ty exis s for a highe purg e rat e for a short pe iod of time A dvic from the man factur r

should betaken he e

3.3.2 Proc edure

a) F or al XRF ins rument , use a sample w hose radiation wa elength is suita le for the det ect or an

c y stal combination Below ar the r commen ed lnesfor each c y stal:

1) TlAP – Al K α or Mg Kα;

2) 2,5nm t o 5 nm 2d c y stals ( ynthetic multia e )— Al Kα, Na Kα or Mg Kα;

3) 0,1 nm t o 0,41nm 2d c y stals (LiF

2 0, LiF

2 0

& LiF

42) — Fe K α or Cu Kα;

4 ) Ge — SKα;

5) P — Al K α, Si K α or P K α;

6) InSb — Si K α

b) Using high X-ra tube p we (with high mA), sele t a sample, or samples, that give( ) a cou t rat e

near t o that pe ified b the man factur r as the ma imum for the cou t er

c) Measur the pulse height using a smal st ep siz (1 % t o 2 % of the peak v ltag e)an a st ep time of

not les than 1 s R ecord the peak ma imum p sition

d) Che k man factur r ins ructions for the cou t rat e at w hich the aut omatic pulse shif cor e tion is

ena led R ed c the X-ra tube mA set ing so that a count rat e below this level is o tained or use a

se on sample

e) R epeat the proc d r c)an r cord the peak ma imum position

f) R epeat the w hole proc d r for each det ect or

If the peak ma imum p sitions at the low cou t rat e and the high count rat e dife b mor than 5 %

then the high v ltag e g in of the det ect or r q ir s adjus ing This adjus ment ma r q ir the ski s of

a q alf ied t ech ician

If the ins rument uses pulse shif cor e tion at al cou t rat es, the t es wi l che k that the cor e tion is

working

4 Spectrometer tests

X-ra spe tromet ers ar mainly used for q antitative analy ses The degr e of pr cision r q ir d

in v rious a plcations v ries conside a ly an the folowing t es s ar t o det ermine w hethe the

spe tromet er can del ve the r q ir d pr cision

Det ermining the measur ment e ror component of the t otal analytical e ror can be achieved b

measuring one sample 2 times (for ex mple) an then proc s ing the data t o o tain the mean, s an ard

deviation an % coeff icient of v riation (%Co V ) of the conc ntration or int ensity v lues The %Co Vof

the conc ntration v lues wi be that of the int ensity v lues multipled b the matrix cor e tion t erm

for the element of int er s The s an ard deviation of the int ensities should be close t o the cou ting

s atis ical e ror

If the measur ment e ror is out ide of the s atis ical e ror lmit given in Ta le 2 an also out ide of

the ac epta le e ror for the analy ses being car ied out, then the folowing t es s can be used t o f in the

ins rumental v ria les w hich ar causing the pro lem an they can then ha e their influenc on the

t otal analytical e ror minimiz d

F or ty ical routine analy ses, a %Co Vof 0,1 % is g ene aly satis act ory for analytical r q ir ment

If highe pr cision is r q ir d, then the ins rument should be car ful y maintained an the t es ing

fr q ency of p ramet ers lst ed in Ta le 1 might ne d t o be inc eased

Tes ing of the spe tromet er is car ied out b making r peat ed measur ment t o det ermine the b sic

ins rument s a i ity an then the r peata i ity as v rious ins rumental p ramet ers ar chang ed, one

v ria le at a time, so that any e ror sour e can be fou d

The s atis ical b u ds of t es s con uct ed pr sent a pro a i ty dis ribution a out the cou ting

s atis ical e ror ( %CSE) F or t es s con uct ed at low levels of pr cision, it isfr q ently p s ible t o o tain

r sult that ar les than the %CSE The pro a i ty of this oc ur ing de r ases w hen inc easing the

n mbe of cou t ac umulat ed pe measur ment an w hen inc easing the n mbe of measur ment

A t highe levels of pr cision, r sid al ins rumental e rors wi l ha e an inc eased impact on r sult an

the %Co Vwi g ene aly be gr at er than the %CSE

Pr cision measur ment ar made on the major element of the mat erial t o be analy sed using a sensitive

l ne In the case below, the F e K α lne is used

Dead-time cor e t ed int ensities should be used for measur ment sinc these int ensity v lues ar used

t o convert int ensity data int o conc ntration v lues d ring the analytical proc d r Such int ensities

ar also used in calculating measuring times r q ir d t o achieve the desir d cou ting s atis ical e ror

w hen set ing upan analytical method

The %CSE should theor tical y be b sed on the actual count a det ect or r gist ers A t high cou t rat es

a o e a out 1 0 0 0 0 cps, dead-timeefe t ha e an inc easing efe t on the calculat ed %CSE an this

v lue is bett er es a l shed using the non-dead-time cor e t ed count rat e

A t high cou t rat es, w he e dead time an pulse pie upefe t ar high, the time int erv l dis ribution of

count also v ries from a Pois on dis ribution Howeve , such dife enc s ne d not beconside ed w hen

simply trying t o det ermine if a spe tromet er is ca a le of o taining a r q ir d degr e of analytical

pr cision or if some comp nent of the spe tromet er ne ds t o be fix ed

4.2 Precision

4.2.1 General

The pr cision of WD- XRF analytical methods is depen ent on the e ror of sub-samplng the mat erial

brought t o the la orat ory for analy sis, the sample pr paration e ror an the sample measur ment e ror

The t otal analytical e ror ( T) is det ermined from contributing e ror comp nent using F ormula (1 ):

T= ( subsamplnger r or ) +(pr ep r ationer r or ) +( measur ementer r o

r

r )2

(1 )

Fact ors contributing t o measur ment pr cision e rors ar counting s atis ical e rors, ins rumental

s a i ty e rors, e rors as ociat ed with mo ing ins rumental p rt an e rors d e t o measur ment

paramet er set ings These e ror sour es wi l be ad r s ed in the subclauses below

The a o e e rors ar based on 1 s an ard deviation (s d) Gene aly, the pr cision of an analytical

method is desc ibed using 2 s d, w hich cor esp n s t o a 9 % pro a i ty that a r sult wi l be within

those b u ds The ma imum cou ting s atis ical e ror r q ir d for an analytical proc d r can be

calculat ed b r ar anging F ormula (1 ) an ha ing minimiz d an k owing the sub samplng an

sample pr p ration e rors These can be o tained ex e imentaly

4.2.2 Calculatio of c ounting statistical er r or

F or an inf init e n mbe of measur ment an as uming no ins rumental e rors:

N is the t otal cou t ac umulat ed pe measur ment

F or a finit e n mbe of measur ment an as uming Pois on s atis ics a ply:

%

Stan ar d deviation

Mean

An the s an ard deviation = Mean , as shown in F ormula (1 )

If the n mbe of measur ment (n ) is < 0, then the sample s an ard deviation (b sed on n 1) is used

inst ead of the p pulation s an ard deviation

Whe e 2 measur ment ar made, the o se ved %Co Vshould not ex ce d 1,4 times (1 % pro a i ty

[2]

)

the %CSE, as shown in Ta le 2

Table 2— %CSE and u pe statistical lmit for %CoV

Total cou t

( N)

%CSEa

1,4 × %CSE

a

%CSEb

Usin non-de d-time c r e ted c u t at a c u t ate of 14 1636 cps

Counting s atis ical e rors should be calculat ed on the cou t actualy det ect ed b the det ect ors an

not the dead-time cor e t ed count A t low cou t rat es < 0 0 0 0 cps, the e wi l be l t le dife enc

in the dead-time cor e t ed an non-dead-time cor e t ed %CSE A t high cou t rat es, the dife enc

be omes notic a le

The non-dead-time cor e t ed cou t rat e can be calculat ed using the formula for ext en a le dead time

[F ormula (14 )] A dead time of 0,1 μs was used in the calculation, with the v lue ent er d in se on s, i.e

0,0 0 0 01 s an the cou t rat e in cou t pe se on

The t otal cou t ( N) in Ta le 2 ar the dead-time cor e t ed count ac umulat ed Calculation of %CSE

b

is b sed on the same cou ting time used t o ac umulat e the dead-time cor e t ed cou t with a

non-The dife enc betwe n %CSE an %CSE wi de r ase with de r asing count rat e an inc ease with

inc easing count rat e The e is lt le dife enc between the v lues below cou t rat es of 1 mi lon cps

The %CSE

b

v lues of Ta le 2 should be close a pro imations on the s atis ical l mit o tained on flow

count ers at 2 00 0 0 cps for mos mode n ins rument

Tes measur ment ar alway s car ied out using the dead time cor e t ed count Only the cou ting

s atis ical lmit ar adjust ed for dead-time efe t

If the %Co Vfor a pr cision t es is larg er than the v lues given in Ta le 2, w hich ar lmit based on

s atis ics alone, then the measur ment should be r peat ed If the measur ment ar out ide the l mit

a ain, then cor e tive measur s should be taken t o minimiz the pro lem if the r sult do not satis y

analytical r q ir ment If the spe tromet er can ot me t the r q ir d pr cision for an analy sis, t es s

can be car ied out at a lowe cou t rat e or it use should be lmit ed t o les pr cise work

When t es ing spe tromet ers at the a o e levels of pr cision, it should be r membe ed that al

spe tromet ers ha e some r sid al ins rument e rors an , the efor , wi l not pas pr cision t es s w hen

t est ed at some ve y high level of pr cision such as 1 0 mi ion ac umulat ed cou t pe measur ment

Howeve , mos spe tromet ers ar ex e t ed t o be within the s atis ical lmit or plus 2 % of the

ma imum lmit w hen ac umulating betwe n 2 mi ion t o 40 mi lon cou t pe measur ment un e

s an ard la orat ory con itions sinc measur ment e rorsshould be and ar g ene aly smal

4.3 Test specimen

4.3.1 General

The t es spe imen should be ro us an s a le and should ha e a flat analytical surfac A metal alo

or a glas disc int o w hich the analyt e() has/ha e be n incorporat ed b fusion with a b rat e flux can

be used The use of compact ed powde pelet is t o be actively discourag ed asthese ar not ta le o e

long pe iods of time The sample should be firmly f ix ed in the sample holde so that no mo ement is

p s ibled ring measur ment

4.3.2 S quential spectrometer s

Any analytical l ne that gives a high cou t rat e can be used for the pr cision t es an it is a pro riat e

t o use the lne of a major comp nent of a ty ical sample In the ex mples given below, F e Kα is used

an the amount of iron in the t es spe imen should be such that, u de normal con itionsof X-ra tube

p we , the ma imum count rat e s ipulat ed b theman factur r is not ex ce ded

4.3.3 Simultaneous spectr ometer s

F or simultaneousspe tromet ers, a t es spe imen that alow s t es ing of seve al channels simultaneously

is desira le S ve al samples ma be r q ir d t o co e al element chan els The t es spe imen should

be such that, u de normal o e ating conditions of X-ra tube power, a cou t rat e of arou d 1

5

cps

or highe is o tained in each of the channels un e t es F or channels measuring very lght element ,

w he e such int ensity ma not be p s ible, use as high a cou t rat e as is achiev ble Ve y high cou t

rat es should be a oided as the e wi then be signif icant e rors in the r corded cou ting times used

unles thr e de imal plac s ar a ai a le Alt ernatively, use the fix ed time method t o col e t set of

count w hich ar ve y close t o the targ et v lue ( N) for al p we set ings

4.4 Instrumental c onditions

4.4.1 General

The X-ra tube should be o e at ed at the normal working p we Thesample spin e should be used, if

p s ible, u les othe wises at ed in the spe if ic t es