for each compound class has been given in Tables 2 to 5
under the heading % Blanks. This column expresses the
amount of an analyte found in the blank as a percentage of
the mean amount in the samples before the amount in the
blank has been subtracted.
Results and Discussion
Fuels. The fuel properties highlighted the different natures
of coal and wood. Coal had the higher gross calorific value
(33 kJ/kg vs 19 for wood), but less volatile matter (37% -
wood 83%). Coal had about 10 times more chlorine than
wood (0.35 vs 0.04% for wood). These values are within the
range typically found for wood (e.g. ref 15). For all EFs on
a per weight basis, the burning of coal exhibited much higher
emissions than the burning of wood (Tables 2-6). The
precision of three independent test runs for both fuels was
very satisfactory. Relative standard deviations were < 20%
for most PCDD/Fs, PCNs, PAHs and PM10. The determination
of EFs for PCBs was less reproducible, which was likely a
result of the occurrence of a greater number of PCB congeners
in the blanks and the lower concentrations of PCBs in the
samples.
PAHs. PAH analysis was carried out by Harwell Scientifics.
GFFs were spiked with 1000 ng each of D8-naphthalene, D8-
acenaphthylene, D10-acenaphthene, D10-pyrene, D12-benzo-
[b]fluoranthene and D14-dibenzo[ah]anthracene. GFFs and
XAD-2 traps and condensate/rinsings fractions were extracted
with dichloromethane (DCM). The extracts for each test were
combined and evaporated to below 10 mL volume, then made
up to 10 mL in a volumetric flask. An aliquot volume of 2 mL
was then taken for PAH analysis. After addition of a recovery
spike the aliquot was cleaned up by column chromatography,
and finally analyzed by GC/MSD (Hewlett-Packard 5970
instrument equipped with a 50 m DB5 0.2 µm film capillary
column operated in SIM) after addition of a syringe standard.
PM10. A proprietary stainless steel in-stack cyclone was
used as a classifier, separating PM10 from non-PM10 particles.
A Graseby Andersen 90 mm filter holder was located externally
to the duct and unheated, therefore maintained at ambient
temperature (or slightly higher, as dictated by the flue gas
temperature as it passed through the filter). This was clearly
not in compliance with US EPA Method 201a but was a
necessary expedient in view of the prevailing conditions (12).
Whatman GF/C 90 mm filters were used for collection of
PM10. The PM10 material collected on the filters and the
associated PM10 particles which deposited upstream of the
filter during the PM10 tests were recovered and weighed.
Blank levels were in general < 10% of PCDD/F, PAH and
PM10 emissions, with higher blank levels for the higher
chlorinated PCDDs, some PCB and PCN congeners. Mean
recoveries for all samples of the labeled 13C12 standards were
as follows: PCDD/Fs - 58-76% except for one hepta-furan
(52%); PCBs - 68-100% except congener 28 (42%); coplanar
PCBs - congener 77-55%; congener 126-28%; congener
169-38%; PCB recoveries for the PCN results - 64 to 123%,
except congener 28 (46%).
Representativeness of Fuels and Burning Practice. The
vast majority of households burning bituminous coal in the
U.K. do so on appliances which are similar in general design
to the Fulham Mk.III grate and which are set in openings
roughly similar in size to that used for the tests. Combustion
air to the undergrate is controlled by means of an adjustable
flap or damper and it is unlikely that in the normal course
of events the fire would be allowed to burn away with no
means of controlling the undergrate air supply and hence
the rate of burn. It follows that the concentrations of various
pollutants generated during the house coal tests could be
viewed as typical of those generated from a household
situation, from the equivalent mass of the same fuel charged
to an established fire.
Comparison with Published EFs. Cl1-8DD/Fs. To our
knowledge, this is the first study reporting all homologue
groups of PCDD/Fs in such samples (Table 2). The lower
chlorinated DFs dominated the overall emissions from both
coal and wood, with EFs decreasing with increasing chlorine
number for the PCDFs. EFs were highest for Cl1DFs (11,000
ng/kg coal - 2,100 ng/kg wood), followed by Cl2DFs (4,100-
250) and Cl3DFs (460-30). For both fuels, EFs of PCDDs were
highest for Cl2DDs, followed by OCDD and Cl3DDs.
For ΣCl4-8DD/Fs, emissions were 230 ng/kg coal and 50
ng/kg wood. Much higher EFs were determined by Chagger
et al. for coal (2700 ng/kg), even under good combustion
conditions (cited in ref 19), and by Broeker et al. for various
coals (60-1800 ng/kg (20)). The EFs for wood were com-
parable to results from Gullett et al. (15) for wood combustion
in fire stoves in the San Francisco Region (10-76 ng/kg wood),
but much lower than results reported by Gullett and Touati
(21) for forest fires (100 up to 7000 ng/kg wood) and by Broeker
et al. for beech wood (500-2400 ng/kg (20)).
PCBs. As a general trend, EFs decreased with increasing
degree of chlorination (Table 3), similar to the pattern
observed for the PCDFs (see above). However, Cl3Bs had
much lower EFs than Cl4Bs (900 vs 7200 ng/kg fuel for coal
and 76 vs 410 ng/kg fuel for wood, respectively). For the
higher chlorinated PCBs, EFs decreased with increasing
degree of chlorination. Major PCBs emitted from both coal
and wood were #49 and #41/64 (>3,000 ng/kg coal; >100
ng/kg wood). Other congeners showed lower EFs, by 1-2
orders of magnitude. EFs for coplanar PCBs #77, #126 and
#169 were also determined, and decreased from #77 (4 Cl)
to #169 (6 Cl). Data on PCB emissions from combustion is
scarce. Gullett et al. (15) determined EFs for PCBs in wood
fuels - their values exceed our measurements by about an
order of magnitude. We cannot explain the differences, which
probably reflect the inherent variability in different fuels,
procedures and analytical methods. Part of the discrepancy
might be that the measurements by Gullett et al. (15) may
indeed represent ‘worst-case’ scenarios with unusually dry
fuel and high burn rates (22).
Regarding the combustion of wood, the traditional inset
open fire is not the most suitable appliance on which to burn
wood, although there are a significant proportion of house-
holds possessing this type of fire on which wood is burned
either by itself, or perhaps more commonly, in admixture
with coal. The main influence on differences between the
patterns of pollutant emissions from the tests as carried out
compared with hardwood burned in a domestic situation on
the same appliance type will probably be the moisture
content. It is unlikely that seasoned hardwood prepared for
domestic consumption would ever be as low in moisture as
the wood used in the test work. Therefore, it is reasonable
to suggest that certainly PAH, PM10 and possibly PCDD/F,
etc. EFs derived from the tests on wood will be lower than
those which would be obtained from burning the quality of
seasoned hardwood normally charged to a householder’s
fire.
PCNs. A wide range of PCN congeners was detected in
the test runs (Table 4). Similar to the EFs for PCDFs and
PCBs, higher EFs were determined for the lower chlorinated
congeners. Cl3Ns and Cl4Ns showed highest EFs (around 300
and 50 ng/kg fuel for coal and wood, respectively) with much
lower EFs for the higher chlorinated congeners. PCN #24
dominated with an EF of 120 ng/kg (coal) and 32 ng/kg
(wood). EFs for PCNs (total 680 and 120 ng/kg for coal and
wood, respectively) were lower than those determined for
PCBs (8,900 and 600 ng/kg). We believe these are the first
published measurements of EFs for PCNs, so no comparison
is possible.
9
VOL. 39, NO. 6, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 1439