I$VCLIBAIIL INSTAUKENTS AND METHODS 32 (1965) 224-128 ; C NORTH-HO,LLAND'P*UDL 1, .1 ;1.41NG Co.

XISION WSS SpECTROPArMC

SPE4~~~~ ISOTOPE DIEUMON A~WS
P. J. DE BIPWRE and G. H. DEBUS

Received 14 July 1964

à
=~! ~ of propa"on 3 mass spocuDtnetric quantitative determinations bY tneans of isotope dilution.
Diagmms are given allo%-ing a choice ar optimal conditions .
1his linear relationship between R, R and K is
iques exist for the quantitative de- assumed to 4 sWently accurate for the, purpose of
tcrrnination. of small sampi=, but the precise quantita- this paper.
*an 0'USUf3M11t1gMM q0antltleS 19 111n1ted A 2JA 1 = absolute mass rauo (A second to llrst
etl~'L-')ds. isotope,
spectrometry in isotope dilution p = MTIMs sample to spike mast
w a 1 s 11 em pin enables high precision to be A)btained q: = NT/A ~ -~!- sample., to.4
P vike ~ . . _ : OfAV',
rmm examination of minute quantities of material . standard:de-4iatiot,'
1 coWi~ 4bW~ 1 eleme~" sample) a = - #(R)IR "ve standard *vi on ~4 :'reta';-
M W a spike Q, a known amouil a aeprecision , ofan isotope ratiomeasurement . a is
ek-mnt but with a different isotope com- assumed,to be constant").
I p

9f sample, The f6ll" basi6:relations applrt ,

as
,ofthe numbers ofaton~~ oftwo isotopes of an
13010pa IBM IS ~j ~W

1 A21
i , I V I &AT 41)
~~~ M~= " U prese~ = L T
1

vaemical form in sample and spme an any rxT i

in(2)
the isolation of the aliquots to be I -W 2 1 = KRs
should be avoided. Several aspects of this S
its precision have been treated in the rN21 - rN2 + rN21, = KRO.
attempts have been made for a [NII . FNj js + [NI IT (3)
y of the sta'fistical error propagation. This
NNith the attainable precisions by mass By defini--ion the itotal amount of sample expressed
ent ofisotope dilution. in number of atoms is
mple case ofa bi-isotopic sample and
ud. T 11 S k ~
ion for a polyisotopic sample and/or and expre:~,sed as a total mass :
xfhere'-). This general relation shows
precision of any isotope dilution 'V'T = X~4- W
a t,~N-- o-isotope case when the Knowl(Age of the spike and of or q hence defines
topes are chosen as dilution isotopes. The the unknewn sample si --,e.
ioned report also contains a detailed treat-
oil special cases. 3. Calculation of q and p :
d bas4c relations used NT [N I + N2]T [NIIT 1 + KRT
absoluteI
[N mass of an i-qAope,, JVS ., + N 2]s I]s + KRs
-V toial mass of the element considered,
n=" of atoms of element or isotope, for N1 0 0 in both sample and spike;
ts referring respectively to the Blend, [N,]T,s
(71prg-et) and o the Spike, calculated from (1), (2) and (3)
[N I]s
true ratio ,of second to first isotope,
measured isotope ratio, [N IITK-'~T + (N I]sKRs
= mass disc-iimination factor. = KRIB
[N11T + [NI]s
224
 . PRECISION .MAS~ SPECTROMETRIC ISOTOPE DILUTION ANALYSIS 225

and Eq. (14) expresses the expected

all dependance of the
MIT Rs - RID relative precision on p, from isotope diiution para
(7)
[Nl Is RB - RT meters, of which one has an approximate knowledge
before starting an isotope dilution procedure.
Using eq. (0) and (7)
The relative precision on q is obtained by replacing p
-A, -7 R, I + KRT
by q in eq. (14) and putting j equal to unity. It is
1+ KAs (8)
RT graphically presented in figs . 1-8 assuming K= 1 .
which is only a,'ftction of the measured isotope ratio [a(q)lq]' is given in units s', as a function of q and Rs.
n blend, sample andzpike and ofthe mass discrimina- The typical RT-values here presented are 0, 0.1/99.9,
tion factor. 0.5/99 .5, 1/99, 4/96, 10/90, 20/80 and 50,150. The pre-
is calculated from : cision on the definition of the spike and the precision
MT = [N I]TA, +[N2]TA2 = [N11TA10 + yK RT) indicated by the graphs allow the determination of the
MS = [N j]SAj(1 + yKRS) expected precision of the quantitative determination of
~!T = I + yKRjr . an element by the isotope dilution technique. It may
P=
MS , 1+ 7KRs be mentioned that an isotope ratio may be defined as
N2.IN, or NIN,2 and therefore it is always possible to
Using eq. (7)
choose RT ;5 1 . The assumption a(R) = sR is not -p-
I + yKRT plicable to very enriched and depleted samples, and
P= (10)
RB - RT I + yKRs * in those cases the precisions are worse than indicated
q can be derived froni p 1by putting y equal to unity. by the graphs. However, for modcrn mass spectro-
metry the assumption holds at least for R values be-
4. Attainable preeMons tween 0.001 and 1000 .
For MT = pMs and
- assuming the determination of
As, to be independent of the mass spectrometric 5. Examples
measurement of p, the statistical errer is : 5.1 A precise quantitative determination of a large
tT2 p2or2(M + Mj a 2( p) natural boron sample (R, = 20/80) estimated at = 0. 1
(MT) ~_- S)
or mmot (= I mg) is required with a I% relative pre-
2 cision .

I ~tT'
!'_MTT '1
P lf(ms
~S)] '-
(12)

Similarly for 7. fpike given in number of atoms :

Available spikes range typically from 90 to 99 .9%
B' O, i.e . Rs = 10 to 1000. A boron isotope ratio
measurement can be performed with a relative pre-
N1. = qA's
2+ cision i; = 0.5%, hence [a(q)1qP.]' ;~; 4 . Fig. 7 sho~vs
[«NS)
(13) different interesting parameter combinations :
[ NT 1 q ] [_Ns Rs = 10 : q = 0.7 yields an optimal relative pre-
The relative errors [(p(p)/p)2 and [a(q)lq] 2 can be cal- cision of 0.7% since [a(q)lq&]' is minimum at 2 .! .
2
culated from eq. (8) and (10) according to the normal The required 1% precision i.e . [a(q)1'q8] ;~j 4 is ob-
law of propagation of statistical errors . tained for 0.16 < q < 3 and the spike added should
After elimination of R13 with the aid of eq. (8) and be between 0.6 and 0.033 mmol (about 6-0.33 mg).
taking into account the constant relative precision on Rs = 100 : optimum condition at q = 0.2 with pre-
an isotope ratio measurement [cr(R) = &R], the fol- cision of 0.6% ; 1 % precision maintained for 0.016 <
lowing result is obained : q :!~ 3.8 .

I A + D
12(l +,vKR-)"
P2 (R s - R T) 2 0 + I~KRs
( I + P)2(j + )!KRs)2
+ - - 2 +
)2 I _-)2 ERT
(R% - R T ' '+ yKR T
)]2[p(l + yKRc,) + (I +
2 I [pRT(l + yKRs) + Rs(I + jYKRT yKRT)1 2
2 )2 I )2(j + y, .R S)2
P (R, - RT ( + yKRT
2 S)2
(R T - R
(ric2 . (14)
)2(l yKRS)'
+(I + )?KR T
 P. 1. DE IBITAVRE AND G. H. DEBUS,

L 1''. L
4-
01MV I 1 ~

J V ` *en-
W -- OMMou"W5
__ " % ..
a i sm,
go11 ,01 a of

1MINCH
,

Waffirll in _ ,IIR
mi "it,
-M- Oku
M a Im
I
UP

10
all

I mass$,"'

251-
mail 0 "1 .1" so 0
M
=1111.1 Me

.__~W

10
9

0Fqr, 1. RT = 0. Fig. 2. Rjr .1/99.9fto.001 .

no- . .--.. % go"

21"
I - :-~

ING a

_4 J111 ,

T 4 1-
.4
D
L_

I
-
.44i,~f, r _+_
i ; i - -1- . 4
H4

14.

T imp,

Filz_ 3. -;? T = 0.5 *99.51,-- 0.005 .

Er7 Fwors (sqLared) [a(q) ;'qej2 as functi,)n L4'q . discrete values of RS and forseveral values of RT .

um condition at q=0.1 ; pre-
Precision maintained for zt 0.001 ::5;
fizures, demonstrate that no substantial im-
ment of L-recision is the use of a highly
enriched B 10 spike, but that the precision, is niai ntained
for a wf Jer range of q-values . Highly enriched Mate ria I
should in this case only be used when there is no way
of making a fair estimate of q.
5 .1 A Very SMEL11 natural boron sample (R I. = 20/80)
 P ECISION MASS SPECTRO TRIC ISOTOPE DILUTION ANALYSIS 227

7WK 7 . - -4 IT- t 11-11 __ ,

t~ t
j f_1
t,
I
4 1~7. -11, -
1 - , "

i kr
l

10 3

7711- ill ;41,T

~W- 1 1 ut W PH ~ 7, ;
fitTO,

, I

Al

4 t -4

to 1
:1 7: -
V 7- - -

i
I, - I ;
0A
-0, ff,

_T,
4h,
W-A ,

le lo t
q
RT = 4/96As 0.04. Fig. 6. RT ~ 10/90~~ 0- 1 -

W,

to

10

Io '

le
4'
Fig. 7. 0/80 ;e0.25,
1
2 Fig~ nT, 7= 50,'50--,: i .

Error magnification factors (squared) [, les of and for several values of Ry .

has to be dctcrmined,
10 for example 10 - 10 mol ( :-- 10 - 9 Rs : the preferred spike amount of 10' mol
g) . One would like to add as much spiG as pwsible in (A . q 10 - ') will only allow to reach a Native pre
order to dispose of a sufficient amount ef material for cision of 9% since [cr(q)lq B]' =- 300 ; 1 '/,, precision
Mol ( 0 10-7
isotopic analysis, say 10' g) O .e . would be obtained for [a(q)qr_j2 = 4 but requires
10-2 Examination q = 0. 16 . A .spike only 6 times the sample can be ad&
). of fig 7 yields :
 P. J. DIE BIÈVRE AND G. H. DEBUS
10 -8 MI spike (q = 10-2)
to neck 1.3% precision; the I % is ob-
and this means addition of a
Tn this case -tt mol spike added,
relative n which is better than
Sion bably still attained for
An axw%-Às 0, leads to some interesting
to conditions needed in order to
aclh 'hi chest povqible precis-ion .
I To atuin high pTectsion on the quantitative de-
re" . Ination of the sample. a prcper choice of the
ta mP~c to spike ratio has to be made. This implies the
~rnatlort of an unknown sample amount. However,
this estimafim is not c6tical when the proper spike,
~e. FoT instance (cf. fig. 4) for an un-
of an element with isotope ratio
ision sN.~'5 is obtained with R. ~-- 10
.
the estimated q value is found to
limits~ then the precise analysis has to
. using the first determination in
first estimate. It is obvious however,
for this first approximation by another less
suitable spike, i.e. Rs~, can be more
fa
,!-!~-c for asurements than the measuring
ccn_t to tile technique, e.g. fig. 8 shows
I " ision of &-, 5 is only obtained
:!~- 6 and Rs ~~, 100 or Rs <- 0 .01 i .e. the
carich-ed in one of the two dilu-
wevei*, a spike with Rs = 0. 1 or
the same result if q can be estim-
.,; + I .
depleted spike material is not
n-c-ded, gm a variety of parameter com-
ws to reach a precision close to the
M.
s %vhen the difference be-
ea-st.; and the error is infinite
g. 6 for RI = 10,1'90 and Rs = 1011100);
however very small differenecs between P's and Rjr
still allow a rough determination.
5. For very small sample sizes on which direct iso-
topic measurement cannot be made, it is still possible
to choose q :5; 10' in order to reach a sufficient
amount for analysis. However the isotopic composition
of the sample must be known approximately, :knd for
is
ftnatural" samples this condition in fact, fulfilied'.
r
For example a 10-13 mol natural U sample can be
"spiked"' with -a hundredfold amouni of 99%. enriched
u235 . The measurement can be done with it , precisjzpe.
1
6 .
of approximately e-v'5 (see fig. 3).
It should be emphasized thatthetotal precision on the
determination of theelement inasample is always worse
than the given precision on the spike (eq. 12 and 13).
The autbors. are indebted to Dr. L . SImepen for,
initiating the problem., for his constant -.intemt. and,
for his suggestions.
The autlimrs also acknowledge gratifully T. L.
Collins (KAPL, Schenectady), N. R. Dal,, (AW RE,
Aldermaston), A. N. Hamer (UKAEA,4,eactor'f
gineering La& Risley), G. Nief (CtA, Sai:-Iay), C. J.
Rodden (NBL, New York), H. M. 'Rosenstock (NBS,
Washington), C. M. Stevens (ANI, Chicago), A.- H.
Turnbull (AERE Harwell) and especially FL K. Web-
ster (AERE, Harwell) for critical reading and for the
useful suggestions.
References
H. Ewald H . Hintenberger, Methoden und Anwendungen der
Massensp,~Oroskopie (Verlag Chemie, Weinheim 1953).
2,) R. T. Jamieson, G. D. L. Schreiner in Electromagnericalky
Enriched Isotopes and Mass Spectrometry (ed. 1q. L. Smith,
Butterworths 1956) 1.69. '
3) L. A. Dietz, Anal. Chem. 34 (1962) 709 .
4) L. A. 1)ictz, Anal. Chem . 35 (1963) 797.
5) M. H. .:)odson, J. Sci. Instr. 40 (1963) 289.
6) R.. K. Webster in Advances in Mass Spectronietr~y (ed. by
J D. Waldron Pergamon Press, 1959).
7) E. A. C. Crouch and R. K. Webster, J. Chem. Six . (1963) 11 .
8) R. K. Webst,-r, J. Chem. Soc. (1963) 118.
9) P. J. 1: ,e Bi6vre and G . If. Debus, Optimal conditions for mass
spectrometric quantit~,tivc detertainations by isotope dilution
( EUR Repcrt, :i-. preparation) .
10) G - . H. Debus~ Nucl . Instr. and Meth. 16 (1%2) 63.