PCBs AND ASSOCIATED AROMATICS 925
volatility of PCBs increases with temperature and as the
amount of chlorine in the molecules decreases. Aroclor
1016, because it contains a high proportion of the less
chlorinated congeners is more volatile than, say, Aroclor
1254 which has a higher average number of chlorines per
molecule. The effect is that at the normal GC operating
temperature, Aroclor 1016 is eluted much faster, in about
10 minutes, than Aroclor 1254, in about 20 minutes. The
interfering components of the mixture tend to be more vol-
atile than even the constituents of Aroclor 1016 and conse-
quently these are eluted quickly and tend to spill over into
the early part of the PCB “fingerprint”. Since the amount
of interfering substances can be very much higher than the
amount of PCB, and even though the detector is less sensi-
tive to the interfering substances, the co-elution of PCBs
with other detector responsive materials causes a combina-
tive effect on the peak area and a high quantitation bias is
obtained. The relative effect of the interfering compounds
on the quantitation of Aroclor 1254 is usually much less
because the peaks which are used for quantitation elute
much later than the interfering materials and the extent of
overlap of the interference and PCB peak areas is usually
much less.
Some of the problematic effects of interferences can be
avoided by a combination of rigorous sample cleanup and
subsequent analysis by capillary column GC. Capillary
column technology allows for much greater resolution of the
components of mixtures and consequently PCB components
tend to become separated from interfering materials. There
is still a very strong need for adequate sample cleanup but
the presence of interfering non-PCB components becomes
much more easily recognized.
The following analytical results illustrate the discussion:- the correlation coefficient for the standard curve,
which is a measure of the linearity of the detec-
tor response in the concentration range of interest,
is sometimes better if one chooses peaks which
appear later in the chromatogram than those nor-
mally used in the early part of the chromatogram; - the correlation coefficient tends to become less
than satisfactory for three early peaks usually
chosen for quantitation when co-eluting interfer-
ences are present; - the apparent concentration of PCBs in the samples
can fall dramatically when the standard curve is
calculated on the basis of peaks which appear later
in the chromatogram than the early peaks which
may be the ones routinely chosen for quantitation.
This indicates that there is a substantial difference
between the chromatograms of the standards and
the samples.
The analytical data in Table 48 were predicated on the
assumption that the PCB contamination in the samples were
due to a single Aroclor, namely Aroclor 1016, and that the
relative peak intensities in the chromatogram matched those
of an available standard Aroclor 1016. This can clearly leadto a high bias in the result when the quantitation method
uses peak areas which contain a significant contribution
from interfering substances. Such a simplistic approach
is not suitable for the quantitation of PCB contamination
derived from partially decomposed Aroclors, mixtures of
Aroclors, or PCBs which may not have originated from an
Aroclor.
The chromatograms of the samples used for the above
discussion do contain a distribution of PCBs which match
the pattern found in Aroclor 1016. However, the relative
amounts of PCB congeners in the standard and the samples
do not match. The samples may have been derived, for
example, from a batch of Aroclor 1242 which had been dis-
tilled to yield only the most volatile fraction as Aroclor 1016.
This might explain why there is relatively little of the less
volatile, more highly chlorinated congeners. Chemical deg-
radation, on the other hand, tends to dechlorinate the more
highly chlorinated material first and then, as the chlorines
are removed, the mixture becomes more concentrated in
the less chlorinated material until it too eventually becomes
dechlorinated.
Different isomers of PCB with the same number of chlo-
rine substituents can cause substantially different responses
from electron capture detectors. The accurate quantitation of
PCBs therefore relies on the similarity between the sample
chromatogram and the standard Aroclor.
A calibration technique was proposed by Webb and
McCall which employed individual peak response factors.
A table was developed for each Aroclor in which the
weight-percent composition of each peak in the chromato-
gram was identified by whole numbers which represented
their retention times relative to a reference compound defined
as 100. The weight-percent compositions of Aroclors were
determined using GC/MS.
The column used was a packed column with poor chro-
matographic resolution compared with today’s capillary col-
umns. The individual peak calibrations were valid only for
the specific lots of Aroclors tested.
Sawyer^26 has conducted a similar characterization to Webb
and McCall and concluded that individual peak calibration
is the most reliable approach for samples containing a non-
Aroclor residue.
The method of peak by peak comparison is used in
ASTM D-4059. The concentration of PCB in each peakTABLE 48
The apparent PCB concentration of samples
is drastically altered by the
method of calculation
Calculated PCB concentration
(ppm Aroclor 1016)Sample # Early peaks Late peaks Later peaks
A 61.7 51.7 8.7
B 16.7 D.L. D.L.
C 19.0 D.L. D.L.C016_003_r03.indd 925C016_003_r03.indd 925 11/18/2005 1:12:42 PM11/18/2005 1:12:42 PM