volume with H
2
O, mix and transfer to acid-washed
polypropylene bottle. (
6
)
Std1
.—Pipet 0.5 mL intermediate
standardsolutionintoa100mLacid-washedvolumetricflask.
Add 10 mL analytical grade HNO
3
, dilute to volume with
H
2
O, mix, and transfer to acid-washed polypropylene bottle.
All calibration solutions shouldbe stable for 1week.
(
g
)
Test portion preparation
.—Three single test portions
were first extractedusinga smallweighing spatula fromeach
of the five amber PVC 100 mL boxes (Greiner Bio-one)
containing 10 g of test sample (Table 2), thenweighed, and
finally prepared as described for SLVbeforeMCDorMDO
microwave digestions.
(
h
)
Sample digestion
.—All 15 test portions (i.e.,
triplicates extracted fromeach of the five test samples) were
digested according to MCD or MDO programs adapted or
similar to those described for SLV (Mars Xpress andMDS
2000 systems, respectively). Information was provided to
collaboratorsconcerningthemaximumconcentrationforeach
element in the five delivered food matrixes in order to
pre-adapt the final dilution of digest test portions before
ICP-AESanalysis.
(
i
)
Detection
.—Digested test solutions, or appropriate
dilutions were presented to the ICP-AES instrument
calibrated as described for SLVwith acid-matched standard
calibrant solutions
.
(
j
)
Calculations
.—Element concentrations in the five
food matrixes were calculated using the equation described
for SLVin the
ICPAnalysis Section
, part (
c
).
(
k
)
Statistical calculations
.—Medians, SD
r
, and SD
iR
,
repeatability (r) and reproducibility (R) limits,
z
-scores, and
HorRat values for nine elements in five blind samples
prepared in triplicate were determined by a statistical
treatment of 45data sets.
(
l
)
Repeatability and reproducibility limits
.—
Repeatability limit r,which is themaximumtolerated relative
variationbetween twomeasurements takenbyasingleperson
or instrument on the same matrix and under the same
conditions, is expressed as follows:
2.772 SD
r
where SD
r
is the repeatability standard deviation.
Reproducibility limit R is the maximum variation between
two measurements taken by several persons in different
laboratories,ondifferentdays,underdifferentconditions, and
is expressed as follows:
2.772 SD
R
whereSD
R
is the reproducibility standarddeviation.
(
m
)
Z-score
.—Evaluation of the whole laboratory
performance in termof accuracywas obtained based on the
calculation of the
z
-score. The
z-
score “
z
” is given by the
following equation:
z
x X
RSD
R
where x is the foundmean value of analyte concentration in
the test material calculated from the ninemeans reported by
thenine laboratories;Xis the certified (or reference) valueof
the certified (or in-house) reference material; RSD
R
is the
relative reproducibility standarddeviation.
If
z
2, the result is satisfactory; if 2 <
z
<3, the result is
questionable; if
z
>3, the result is unsatisfactory.
(
n
)
HorRat values
.—HorRat values for each element in
eachof the fivematrixeswere calculated as follows:
HorRat value=
RSD , %
2C , %
R
0.1505
whereRSD
R
is the relative reproducibility standarddeviation
of each element in each of the five testedmatrixes; C is the
relative concentration (expressed as dimensionless mass
fraction) of the analyte in question (valid for concentration
ratio above 10
–9
).
The original data developed from interlaboratory studies
were assigned a HorRat value of 1.0 with limits of
acceptability of 0.5–2.0. Consistent deviations fromthe ratio
on the low side (values <0.5) may indicate unreported
averaging or excellent training and experience; consistent
deviations on the high side (values >2) may indicate
inhomogeneity of the test samples, need for further method
optimization or training, operating below the LOD, or an
unsatisfactorymethod.
(
o
)
Horwitzequation
.—Horwitzequationexpressedusing
logscale(logSD
R
=0.8495logC1.6991)wascomparedwith
asimilarequationfoundusingthe45SD
R
valuesandmedians
of concentration (i.e., SD
R
and median values for nine
elements in fivematrixes) obtained fromthedata treatment of
theRT.
ResultsandDiscussion
SLV
(
a
)
Closed-vessel microwave digestion optimization
.—
Average values of each element concentration in whole egg
powder, baby food composite, and baking chocolate fell
within confidence intervals of reference value for the three
matrixes and are in agreement withAOAC recovery criteria,
except for ironandcopper inbakingchocolate,with recovery
values of 123 and 125%, respectively, and for phosphorus in
whole eggpowder,with a recoveryvalue of 116%(Table 6).
Significant variation of results for iron and copper in baking
chocolate have previously been observed (34). The higher
valueof recovery for phosphorus inwholeeggpowder isdue
to theuseof theoptimizedmicrowaveprograminCEMMars
Xpress, whereas generally lower values are obtained with a
program leading to incomplete digestion (6, 8). This single
optimized microwave digestion program on a CEM Mars
Xpress systemis adapted tobe applicable tovariousmatrixes
covering theAOACfood triangle.
(
b
)
Linearity
.—The calibration curves constructed by
plotting element concentration versus peak ratio response
(element/IS) showed good linearity either in linear or in
weightednonlinear regression (Table 7).Weightednonlinear
1494
P
OITEVINETAL.
: J
OURNALOF
AOACI
NTERNATIONAL
V
OL
. 92,N
O
. 5, 2009
MTE-01
FORWORKINGGROUP/ERPUSEONLY
DONOTDISTRIBUTE
