United States
               Environmental Protection
               Agency
Atmospheric Research and
Exposure Assessment Laboratory
Research Triangle Park, NC 27711
               Research and Development
EPA/600/SR-94/192   November 1994
EPA      Project Summary

               Sampling  Carbonaceous
               Aerosols: A  Review  of
               Methods and  Previous
               Measurements
               Roy L. Bennett and Leonard Stockburger
                Carbonaceous aerosols, a complex
               mixture of carbonate carbon, elemental
               carbon, and  organic carbon (organic
               compounds), are of environmental im-
               portance due to their impact on visibil-
               ity  and the toxicity  of some  of the
               organic compounds.  A knowledge of
               the relative vapor and particle concen-
               tration of semivolatile organic com-
               pounds is required to understand how
               the compounds are transported in the
               atmosphere and removed by deposi-
               tion. Also, the deposition in the lungs
               is dependent on the vapor/particle dis-
               tribution. Because of the multiplicity of
               compounds of varying volatility in the
               organic aerosols, the sampling of these
               particles  presents  an  extremely  diffi-
               cult challenge. This report, based on a
               literature search of measurement stud-
               ies  for the past 20 years, reviews the
               possible artifacts that can occur that
               result in  positive and negative errors
               owing to volatilization, sorption, or re-
               action during the sampling processes.
                The sampling approaches that  have
               been used range from single filters up
               to complex parallel  multiple component
               systems that employ diffusion devices
               to separate particle and vapors. The
               artifacts that were  addressed by  each
               approach are included. Representative,
               but not exhaustive, lists of previous
               carbonaceous aerosol measurements
               are presented based  on the literature
               review. The first list includes measure-
               ments of elemental carbon and organic
               carbon without resolution of individual
               organic compounds. The location, mea-
               sured concentrations, and comments
               on the artifacts addressed are presented
               for  each  study.  The  second list is a
               representative group of measurements
               of  specific organic compounds  or
               classes of organic compounds. Here,
               because of the large number of indi-
vidual compounds present, concentra-
tions of each compound are not listed,
but the artifacts that the employed sam-
pling method addresses are noted.
  This Project Summary was developed
by EPA's Atmospheric  Research and
Exposure Assessment Laboratory, Re-
search Triangle Park, NC, to announce
key findings of the research project that
is fully documented in a separate report
of the same title (see Project Report
ordering information at back).


Introduction
  Carbonaceous materials are usually a
major  component of ambient fine particle
mass concentration. For example, carbon-
aceous aerosols have been found to ac-
count  for typically 40% of the total fine
particle mass loadings in the Los Angeles
area. These  particles consist of a  multi-
tude of organic compounds, both molecu-
lar and  macromolecular, as well  as
elemental carbon.  Over 2800 compounds
have been identified in the ambient atmo-
sphere, but  in  those studies that have
quantified the individual organic compound
concentrations,  usually the sum  of the
measured compounds account for only a
few percent of the total organic mass con-
centration.
  The potential toxicity  of some of the
organic aerosols has made them the sub-
ject of environmental concern and exten-
sive investigation. Special attention has
been devoted to the hydrocarbon fraction
owing  to the  mutagenic activity exhibited
by compounds  such as polynuclear aro-
matic hydrocarbons (PAHs) and their po-
tential carcinogenic   effects.  The
predominance of organic aerosols in the
fine particle size range makes them effi-
cient transporters of any toxic compounds
into the respiratory system.

-------
  The carbonaceous  aerosols,  espe-
cially elemental  carbon, are also of en-
vironmental concern because they occur
in  the  visibility-reducing size  range
(0.1|j,m-1.0 |im)  and may therefore have
an important impact on visibility condi-
tions  and  possibly climate.  Elemental
(or black) carbon is the most important
absorbing species for sunlight in the at-
mosphere, and  its catalytic  properties
may result in  its contribution to  impor-
tant atmospheric chemical reactions.

Elemental Carbon  and Organic
Carbon Aerosols
  The complex mixtures of carbonaceous
aerosols are often classified  as  carbon-
ate, elemental  carbon (EC), and  organic
carbon (OC) that consists of organic com-
pounds. Ambient studies have shown that
carbonates account for only a negligible
fraction  of the total carbon  in urban aero-
sol.  Elemental  carbon has a chemical
structure similar to impure graphite and is
emitted  directly from fossil  fuel combus-
tion sources. Optical methods have been
used to measure EC, as it is  considered
the  primary component of atmospheric
aerosol that strongly absorbs light.
  Organic aerosol  compounds in the at-
mosphere may result from direct emis-
sions from a source (primary OC) or may
result from condensation of low vapor pres-
sure products  of gas phase reactions  of
hydrocarbons onto  existing atmospheric
particles (secondary OC)  or the reaction
product of primarily emitted organic com-
pounds with atmospheric oxidants.
  A method based on thermal/optical re-
flectance procedures has  been used  to
measure the EC and OC components  of
ambient aerosols.  Several variations  of
the  procedure have been  used that differ
in respect to the analysis time, tempera-
ture ramping rates, volatilization and com-
bustion, and  calibration  procedures.
Variations in the measured EC and OC
results  of  a variety of samples in an
interlaboratory comparison study have
been  reported. In general, the agreement
between laboratories for total carbon was
within 20% for all samples,  but there were
large interlaboratory differences in  the OC/
EC  ratio  for  all samples.  Differences
among sampling  methods were compared
with differences in analytical methods. For
EC, the observed differences  in side-by-
side  sampling were accounted for by dif-
ferences in the  analytical  methods. For
OC,  most methods  showed greater  vari-
ability in results than could  be accounted
for by the differences in analytical meth-
ods.
Carbonaceous Aerosols
Sampling Artifacts

  As with most ambient aerosol measure-
ments, the most frequently employed sam-
pling  technique  for organic aerosols  is
collection on filters, usually quartz fiber
filters. Because the organic aerosols are
a complex  mixture  of many compounds
with a wide  range of vapor pressure, sorp-
tion properties,  and chemical  stabilities,
the sampling process has the potential for
several sampling artifacts. These may in-
clude  the following:
  (1)  A portion  of the organic vapor phase
      may  sorb onto the filter surface  or
      on the deposited particulate mat-
      ter. This  is often  called  the "posi-
      tive"  artifact  because the apparent
      mass that would be  collected and
      measured is greater than the aero-
      sol mass that existed in the volume
      of the atmosphere sampled.
  (2)  The sampling process may change
      the gas-solid distribution of the
      sample being collected,  which  re-
      sults in volatilization of some of the
      particulate organic compounds. This
      produces a negative artifact.
  (3)  The collected particulate matter may
      chemically react with vapor compo-
      nents that pass  through  the filter
      during sampling  to change the
      chemical  composition  of the deposit.
      This  can  result in  positive or nega-
      tive artifacts, especially if individual
      chemical compounds or  chemical
      functionalities are  being measured.
  These sampling artifacts complicate the
organic aerosol measurements, and it is
difficult to separate the different compet-
ing artifact  processes  although this has
been the object of many studies.
  The positive  artifact has  been investi-
gated  extensively.  For example, samples
collected on a high-volume sampler quartz
filter indicated  a positive  bias  of about
15%.  With 24-h samples, positive errors
in organic carbon of 15-20% were found,
but the degree to which this was offset by
loss of volatile OC  is not known. A typical
glass  fiber filter  has a surface area that is
much  smaller than  that of collected atmo-
spheric particulate matter, so that sorption
of vapor on the particles  is believed  to
outweigh sorption  on  filters. Diminished
levels  of OC recovered from filters at higher
sampling velocities were attributed to de-
creased sorption rather than  increased
volatilization. The negative artifact has also
been investigated. Theoretical calculations
have indicated that  the evaporation losses
are probably most  important when filters
with high pressure drops are encountered.
Investigators who used a multiple parallel
collector with a diffusion denuder system
concluded that samples  from  a remote
location exhibited  negative  artifacts as
large as 40 to 80%.
  Changes in sample integrity have been
indicated by studies on the degradation or
reaction of PAH. A portion of particulate
phase  aldehyde has  been shown to be
oxidized  during  sampling to the corre-
sponding aliphatic  acid. Reaction of or-
ganic acids  in atmospheric samples with
basic components of glass fiber filter may
occur.
Gas/Particle Distributions of
Semivolatile Organic
Compounds
  A comprehensive understanding of the
environmental  impact of pesticides,  PAH,
PCS, and other semivolatile organic com-
pounds in  the atmosphere requires a
knowledge  of both  their gas  and particle
concentrations. The gas/particle ratios are
important in how the compounds are trans-
ported  in the atmosphere and removed by
deposition.  Particles are quite effectively
removed by rain  events. Health effects
also  depend on the gas/particle  ratios as
they  are related to how the  compounds
deposit in the lungs  and exert  potential
toxic effects.
  The  earliest systems  for measuring the
gas/particle  distributions were a particle
filter  (glass fiber, quartz fiber, or Teflon)
followed by  a solid sorbent trap such as
polyurethane  foam  (PUF),  Tenax, or
XAD-2. The measurement is a severe chal-
lenge since the compounds  are usually
present in low levels (nanograms per cu-
bic meter or less),  and the procedure is
subject to  positive artifacts  from vapor
sorption, negative artifacts from volatiliza-
tion  from the  filter, and degradation of
compounds owing to reactions during sam-
pling.
  The  potential for volatilization and sorp-
tion artifacts is related in part to the kinet-
ics of the sorption process. The magnitude
of the  artifacts is dependent  on  the vari-
ability  of the total particle concentration,
the gas phase concentration, the tempera-
ture  during the  sampling,  and  the
mass-transfer  reaction  time  scales.  The
equilibrium  vapor pressure concentration
(EVC) of atmospheric compounds and the
significance  of the efficiency  of collection
of the  compounds on  filters have  been
considered.  Absolute values of  the  ab-
sorbed state EVCs or the factor  by which
the  pure state  EVCs are commonly re-
duced  by sorption have been  deduced
from measurements of losses from actual
air samples. A critical  evaluation of  the
literature gave  a wide  range of factors
(2-500) by which the pure state  was re-

-------
duced by adsorption. This was attributed
to uncertainty in the literature data rather
than the difference in actual reduction fac-
tors. Comprehensive analyses have been
presented of the partitioning of organic
compounds between vapor and solid aero-
sol phases.  Estimates  of the slope and
intercept from plots obtained  from a data-
base of gas particle distribution data vs 1/
T were found to be useful for predicting
the distribution  in other urban environ-
ments. The effects of relative humidity on
gas/particle partitioning have been inves-
tigated and compared to field studies. The
log of the gas/particle  partition constant
was found to have linearly decreasing de-
pendence on the relative humidity  over
the range from 40 to 85% relative humid-
ity, based on semivolatile PAH compounds
measurements at Osaka,  Japan. A ther-
modynamic model has been developed to
predict the partition between vapor and
liquid  phases based on two parameters:
initial vapor mole  concentration  ratio and
the ratio of Henry's Law constant to initial
vapor partial pressure. As  opposed  to
soluble organic compounds,  the partially
soluble organics tend  to partition toward
the vapor phase owing to the large Henry's
Law constants.
Measurement Approaches for
Sampling Carbonaceous
Aerosols
  Sampling methods for determining or-
ganic  aerosol  concentration and,  if de-
sired,  the  corresponding vapor  phase
concentration of semivolatile compounds
have  evolved from  the simple arrange-
ment of a single filter followed by a sor-
bent trap. As investigators have attempted
to eliminate  or  quantify the  various arti-
facts that were encountered, their sam-
pling  approaches have  become  more
complex  with various  denuder arrange-
ments, multiple  combinations of filters  of
different sorbency (quartz, carbon impreg-
nated, Teflon) and various sorbent traps.
Sample arrangements that have been used
included the following:
  (1)  Filter only
      This approach assumed there were
      neither sorption nor desorption arti-
      facts.
  (2)  Filter/sorbent
      This approach was used for  total
      (vapor plus particle) concentration
      but does not  give correct gas/par-
      ticle  distributions except for  non-
      volatile organics  that are due  to
      artifacts.
  (3)  Multiple filters in  series
      Two quartz filters were employed
      back to back with the amount on
      the second filter used to correct the
      positive artifact (vapor adsorption)
      on the first. The vapor concentra-
      tion passing through the second fil-
      ter is assumed to be essentially the
      same as the first,  which may not
      be  true  if the vapor  is strongly
      sorbed by  the filter.  Also, it does
      not account for the  negative artifact
      or the amount adsorbed on the col-
      lected particles.
  (4)  Dual quartz tandem filters with  a
      parallel Teflon/quartz tandem.
      This is similar to the dual filter ap-
      proach above but had two parallel
      sets of tandem filters: the first  is a
      quartz filter with a quartz after-filter
      and  the  second is a Teflon filter
      with a quartz after-filter. The posi-
      tive artifact is corrected by subtract-
      ing  from  the  quartz of the first set
      the amount on quartz after-filter of
      the second set. The rationale is
      that the Teflon would absorb  much
      less of the vapor than quartz so the
      exposure of the quartz filter after
      the Teflon  would be closer to the
      atmospheric vapor  concentration.
  (5)  Denuders
      Diffusion  denuders  are devices with
      channels operated  at laminar flow
      conditions  in which vapors of the
      atmospheric sample are separated
      from the particles owing to the much
      greater diffusion coefficients  of the
      vapors.  High performance denud-
      ers,  compound annular denuders
      that consist of coaxial cylinders so
      that the air samples pass through
      the annular space, have been de-
      signed that allow the use of higher
      sampling rates.
  Various combinations  of  denuder ar-
rangements have been employed  in or-
ganic  aerosol  sampling.  A denuder
difference procedure for specific PAH com-
pounds  has been  developed in which two
parallel  sampling arrangements are  used.
On one  side was the sequence: denuder,
quartz  fiber filter,  sorbent  trap.  On  the
other was quartz  fiber filter, sorbent trap
(without the denuder).  For specific  PAH
compounds  the method  measures true
particle  phase, while the vapor concentra-
tion and the  degree of negative artifact
are determined by denuder difference cal-
culation, not  direct analysis  of the  de-
nuder.
  A compound  annular denuder system
has been employed to trap vapor con-
stituents by the denuder difference  ap-
proach.  Two filter/absorbent packs were
operated in parallel: one was preceded by
the denuder; the other was  not. The frac-
tion of analyte that occurred in the  vapor
phase was determined by subtraction of
the denuder sampler results from the con-
ventional sampler results.
  Diffusion denuders consisting of sepa-
rated  layers of quartz  filters  have been
used under the assumption that they re-
move only the gaseous organics that are
normally retained on the quartz filter. This
denuder was followed  by a dual  quartz
filter pack. Comparison  of the  results with
a parallel dual quartz pack led to the con-
clusion that the OC on the back filter of an
undenuded filter pack is primarily  a posi-
tive enhancement artifact. The  approach
does not address the negative artifact.
  An  annular denuder-based  sampler,
called the  integrated organic vapor/par-
ticle sampler (IOVPS), has been used for
the direct determination of gas and vapor
semivolatile PAH  compounds in tobacco
smoke.  Finely ground  XAD-4 was used
for the denuder wall coating since it has
been shown to be an efficient sorbent for
PAH and nicotine. The technique provides
a means for  determining phase concen-
trations for specific compounds where their
collection,  recovery, and  analysis have
been  adequately demonstrated, but it is
not yet  applicable to total organic vapor/
aerosol  phase distribution and concentra-
tion measurement.  Similarly,  a diffusion
denuder sampler made from  120 short
parallel  gas chromatographic  (GC) tubes
coated  with stationary  phase  has been
employed to measure PAH. The collected
gas on the  denuder was analyzed  directly
by desorption onto a GC column  without
solvent dilution. The denuder was followed
by a quartz fiber filter  and a PUF plug.
Particles  are collected  on  the filter.
Vapor-phase molecules  that break through
the denuder  as well as molecules that
volatilize off the particles are collected on
the PUF.
  One of the  most elaborate of  the semi-
volatile organic sampling systems consists
of three parallel samplers,  two  of which
contain  multichannel parallel  plate diffu-
sion  denuders that  were carbon-impreg-
nated filter paper to collect the gas  phase.
Sampler 1  has  the  following sequences:
cyclone, denuder, two  quartz  filters, sor-
bent filter (the same carbon-impregnated
material used in the denuder plates) and,
finally,  a quartz filter.  Sampler 2 is  the
same as Sampler 1 except the denuder is
placed after the two quartz filters  instead
of before. Sampler 3 is only a filter pack;
that is, the same as  Samplers  1  and 2 but
without  the denuder. Conclusions drawn
by the  investigators from sampling with
the system at Canyonland  National Park
are that the conventional quartz filter arti-
fact is a negative rather than positive arti-
fact and previous data measurements with
filters may  have underestimated the par-

-------
ticulate  organic  carbon by  more than
100%.
Survey of Carbonaceous Aero-
sol Measurements

Measurement of Total Organic
Aerosol Concentration
  A representative listing of field studies,
where only the elemental carbon and total
organic  aerosol  concentrations were de-
termined without organic compound reso-
lution,  is presented in the  report with
references.  Sampling  techniques  have
ranged from single filters to  complex, par-
allel multicomponent sampling systems.
Measurement of Specific
Organic Compounds
  A list of studies in which individual com-
pounds  or  classes of  compounds have
been specifically identified and measured
is also presented in the report with refer-
ences. Owing to the large number of stud-
ies  on   pesticides,  PCS,  and  PAH
compounds,  the  full  report  does not
present a comprehensive listing of all mea-
surements that  have been made.  Also,
the sum of the concentrations of the re-
solved compounds in most of the studies
does not account for  more than  a few
percent  of  the total organic mass. Be-
cause many compounds in these specific
organic classes exist as both vapor and
particles their measurement involves the
same sampling and artifact problems en-
countered with total organic aerosol mea-
surements.

Conclusions
  The sampling of organic aerosols in the
atmosphere is an extremely difficult task
owing to the very complex composition of
its organic molecular constituents. Exami-
nation of the  peer-reviewed  literature for
organic aerosol sampling studies  during
the last 20  years yields the following ob-
servations:
  (1) A large percentage of the reported
      measurements are subject to arti-
     fact errors. In a  few  cases the in-
     vestigations ignore the artifacts, in
     some they are acknowledged but
      unaddressed, and in other studies
     attempts are  made to estimate the
      magnitude of the artifacts by vari-
  The EPA authors, RoyL. Bennett (also the Project Officer, see below) and Leonard
    Stockburger, are with the Atmospheric Research and Exposure Assessment
    Laboratory, Research Triangle Park, NC 27711.
  The complete report,  entitled "Sampling Carbonaceous Aerosols: A Review of
    Methods and Previous Measurements," (Order No. PB95-129060; Cost: $17.50,
    subject to change) will be available only from
          National Technical Information Service
          5285 Port Royal Road
          Springfield, VA 22161
          Telephone: 703-487-4650
  The EPA Project Officer can be contacted at
          Atmospheric Research and Exposure Assessment Laboratory
          U. S. Environmental Protection Agency
          Research Triangle Park, NC 27711
    ous sampling  designs, many  of
    which yielded results that were later
    challenged.
(2)  In studies where the investigators
    attempted to measure the sampling
    error, the magnitudes of the deter-
    mined positive or negative artifacts
    have varied considerably.  This  is
    probably due, in part, to differences
    in the organic molecular composi-
    tion (hence  in the relative volatili-
    ties and  sorptivities)  and  in the
    concentrations that occurred at the
    different sampling locations.
(3)  Multiple  component sampling sys-
    tems employing vapor phase sepa-
    ration  devices  such as denuders
    require a thorough evaluation  to
    demonstrate efficient removal of the
    vapor at the operating  conditions.
    This  has been  done  for  specific
    PAH compounds. However, the ap-
    plication of  their multiple  compo-
    nent system to the determination of
    total OC aerosol for mixtures of at-
    mospheric organic compounds pre-
    sents a formidable challenge.
(4)  The use  of paired, parallel sam-
    plers (denuder difference method)
    is subject to the propagation of er-
    rors through subtraction, which de-
    grades  the   precision.  More
    development work with  control ex-
    periments needs  to be  done on a
    wide  range of organic compounds
    and concentrations before the sam-
    pling  approach becomes credible
    for extended field studies.
 United States
 Environmental Protection Agency
 Center for Environmental Research Information
 Cincinnati, OH 45268

 Official Business
 Penalty for Private Use
 $300
                                                          BULK RATE
                                                    POSTAGE & FEES PAID
                                                             EPA
                                                       PERMIT No. G-35
 EPA/600/SR-94/192

-------