sample.Powersettingsandnebulizergasflowwereoptimized
toensure that theMg280.271:Mg285.213ratiodemonstrates
robust operating conditions (18) in accordancewith the ratio
established by the instrument manufacturer. Five replicate
readings of the same samplewere performed, with relatively
long integration times to minimize noise. The solution
presented to the nebulizer contains at the most 5000 mg/L
cesium(
Note
:500mg/LCs for ruggedness testsMD1,MD2);
20 mg/L indium; and 5 mg/L strontium, yttrium, and
chromium; less thanhalf of eachelement concentrationof the
higher working standard Std6 and <0.5 g/L total dissolved
minerals.All responsesforbothrecommendedandalternative
lines for each element are corrected using one IS line
(Table 5).
(
b
)
ICP-AES spectrometers (MD3 andMD4 ruggedness
tests)
.—Operatingconditionsofbothinstruments (Table1)as
well as sample and IS pump tubes and peristaltic pump
rotationspeedusedwereselectedaccordingtomanufacturer’s
recommendations and were adapted to optimize aerosol and
maximize precision, and to demonstrate robust operating
conditions. They were capable of determining at least the
recommended multiple lines for each element of interest
(Table 5). Combined ionization buffer/IS solution was
prepared such that after mixing digest sample and
IS/ionization buffer solutions using the instrument’s
peristaltic pump, the combined solution presented to the
nebulizer contained at least 500 mg/L Cs; 20 mg/L indium;
5mg/L strontium, yttrium, and chromium; and less than half
of eachelement concentrationof thehigherworkingstandard
and <.5 g/L total dissolved minerals. All sample instrument
responses of recommended and alternate lines for each
element are correctedusingone IS line (Table 5).
(
c
)
Quantification
.—The concentration (C) of each
element, inmg/kg, was calculated as follows:
C
a V
m
F
whereC=concentration in the test portionsample (mg/kg); a
=concentration(mg/L)of theelement inthedigestsolution;V
= volume (mL) of the test solution after beingmade up [i.e.,
50mLforclosedvesselmicrowavedigestion(MDC) forSLV
and 100 mL for open-vessel microwave digestion (MDO)
usedfor testMD2];F=dilutionfactorof the test solution;m=
weight of the test portion (g).
StatisicCalculations
All data used for selectivity, accuracy and precision
performance tests were treated using robust statistics (29)
based in the concept of Rousseeuw and Croux (28). The
presenceof somesuspectvalues (outliers) canstronglydistort
classical estimations.However, resultsmustnotbeeliminated
without a valid justification. For that reason, robust statistics,
whichprovidegoodestimations evenwithout the elimination
of suspect values, have been used in the SLVandRT. These
robust estimations are insensitive to extreme values and
depend only slightly on data distribution. It is then neither
necessary to test for outliers nor to exclude suspect values.
Themedianhasbeenusedasarobust estimationof thecentral
value. The robust standard deviationSD(Sn) drawn fromthe
algorithmSnofRousseeuwandCroux(28)hasbeenusedasa
robust estimation of spread. This algorithm uses median
absolutedistances betweenall the coupleswe can set upwith
the data:
Sn=medi {medj
*
xi –xj
*
}
SD(Sn) =1.1926 Sn
Statistical comparative
t
-test is the robust adaptationof the
classical
t
-test byreplacing themeanwith themedian, and the
SD with the robust SD(Sn). Robust
t
-test was used for
selectivity (median comparison) and accuracywas evaluated
to check if the recovery is significantly different from100%
(at 95 and99%confidence intervals).
1490
P
OITEVINETAL.
: J
OURNALOF
AOACI
NTERNATIONAL
V
OL
. 92,N
O
. 5, 2009
Table 7. Regressioncoefficientsobtained from
weighed linear andnonlinear regression for
recommendedandalternate lines
Calibrationcurve
a
Concn range,mg/L
Element line R² nonlinear
b
R² linear
c
Low
High
Ca II
d
0.99997 0.99980
0
240
Cu I
d
0.99999 1.00000
0
2.1
Fe II
d
0.99997 0.99999
0
8.5
K I
d
0.99924 0.99326
0
320
Mg I
d
0.99999 0.99891
0
75
Mn II
d
1.00000 0.99998
0
0.04
Na I
d
0.99996 0.99666
0
160
P I
d
0.99998 0.99997
0
160
Zn I
d
0.99999 0.99998
0
3.2
Ca II
e
0.99994 0.99993
0
240
Cu I
e
1.00000 0.99995
0
2.1
Cu I
e
0.99999 0.99999
0
2.1
Cu I
e
0.99999 0.99998
0
2.1
Fe II
e
0.99996 0.99997
0
8.5
Fe II
e
0.99999 0.99998
0
8.5
Mg I
e
0.99997 0.99929
0
80
Mg II
e
1.00000 0.99999
0
80
Mn II
e
1.00000 0.99997
0
0.04
Mn II
e
1.00000 0.99998
0
0.04
P I
e
0.99999 0.99967
0
160
P I
e
1.00000 0.99993
0
160
Zn I
e
0.99999 0.99991
0
3.2
Zn I
e
1.00000 1.00000
0
3.2
a
Calibrationwithsix standardsolutionsandablank.
b
Weighedseconddegreequadratic regression.
c
Weighed linear regressionwithout forcingzero.
d
Recommended lines for ICP-AESanalysis.
e
Alternateconfirmatory lines for ICP-AESanalysis.
MTE-01
FORWORKINGGROUP/ERPUSEONLY
DONOTDISTRIBUTE
