AI10382-GC-MS-Food Safety-Analysis - page 70

12.2 Linearity, Response Factor, Matrix Effect
The calibration curves were created at six levels (matrix-
matched) and injected in duplicate. R
f
values for internal
standardization were determined from the calibration
curves for all matrices and internal standards by
calculating cumulative average response factor over the
whole calibration range. The linearity of calibration
curves was assessed in three groups of compounds
(depending on the relevant MRL values) in calibration
ranges of 0–200, 0–1000 and 0–2000 ng/g, respectively,
(details and results in Table 3). Calibration levels were
equidistantly distributed over the calibration range.
Linear function was evaluated according to Mandel’s
fitting test and plotting of residuals for which <20%
acceptance limit was set.
3
Correlation coefficient values
were additionally established for which an artificial
0.985 was set as an acceptance limit, as no legislative
limits are defined for them. The set value wasn’t met for
fenpropathrin and dichlofluanid (LK) and propargite
(WF) based on the high LOQ values related to the
calibration levels. No weighted function was applied.
Matrix effects were evaluated by (Youden-) plotting of
measured relative peak areas of calibration standards in
solvent against the areas in the relevant matrix. No
matrix effect is observed if the difference of the slope
(dif%) of the fitted line is less than 20% from the ideal
(y=x) curve, while matrix effects are observed when the
difference is between 20–50% (minor matrix effect) or
exceeds 50% (major matrix effect). Matrix effect results
are listed in Table 3. For the compounds with demon-
strated matrix effect application of matrix matched
calibration is required.
12.3 Accuracy
Method trueness was assessed by recovery studies using
blank matrices spiked at three concentration levels (L1,
L2 and L3) and injected in six individually prepared
replicates. (Table 4). Spiking of samples occurred prior
to sample preparation. Found concentrations, recovery
and relative standard deviation (% RSD) were calculated
(Table 5). According to SANCO requirements recovery
values are deemed acceptable if between 70–120%.
3
Values were calculated only for those cases in which
spiking levels were higher than the compound LOQ in
the particular matrix. Recovery values could not been
established for amitraz in WF and captan, chlorthalonil
and tolyfluanid in LK matrices due to the high LOQ
values measured relative to the spiked levels. Strong
influence of matrix on the results were observed in
several cases and results could not been established at
one or two spiking levels based on the measured
different LOD/LOQ values in the different matrices
(details in Table 4). For routine measurement these
compounds in these matrices have to be measured with
separate, specially optimized analytical methods.
Method bias was established by means of external
quality control materials obtained from FAPAS (York,
UK). Available FAPAS materials were #19140QC
(lettuce puree), #19141QC (green bean puree) and
#19142QC (melon puree). The available Fapas samples
represented only a limited number of the target com-
pounds and different matrices from those targeted.
However, measured values showed good agreement with
the assigned values in all cases except carbofuran, in
which the measured value was slightly below the
acceptance range. This could be due to differences
between the two different matrix characteristics. Details
on the measured FAPAS values are listed in Table 7.
12.4 (Intermediate) Precision
Instrument injection precision was tested for both
retention time and peak area for all target compounds
by subsequent injections (n=6) of low concentration level
(L1) standard solutions. Insturment injection precision
for retention time was below 0.5% for all compounds
and between 1.2–18.04% (fipronil and fenamiphos-
sulfoxide) for peak area without internal standard
compensation indicating reliable instrument perfor-
mance. Method within-day and between-day precision
values were determined for each matrix at middle
spiking level (L2) and expressed as %RSD over 3 days
with individually prepared samples (n=6). Mean
within-day precision values were determined as an
average of the 3 individual days’ mean precision, while
between-day precision was expressed as mean of the
overall precision data. According to SANCO require-
ments <20% was set as acceptance criteria for the target
compounds and matrices.
3
Measured values are shown
in Table 5.
12.5 Limit of Detection, Limit of Quantification
Limits of detection and quantification were estimated
following the IUPAC. Measured method LOD, LOQ
and the relevant legislative limits (MRLs) are listed in
Table 6.
7
An artificial MRL=10 ng/g was set as target
value for compounds, for which no MRL values are
legislatively defined. The expectation of the method was
to meet MRL values at least at LOQ level which was
achieved for the vast majority of target compounds. For
methiocarb (WF, LK), carbofuran (SW), oxadyxil (WF)
and propargite (WF, LK) the established LOQ values
were below the targeted MRLs’ value. However, with
exchanging of quantifier and qualifier ions the target
values can be reached. For fenpropathrin (WF, LK),
amitraz (WF) and tebufenocid (all matrices), the target
values could not be reached even when exchanging the
quantifier and qualifier ions.
6
1...,60,61,62,63,64,65,66,67,68,69 71,72,73,74,75,76,77,78,79,80,...225
Powered by FlippingBook