United States
                    Environmental Protection
                    Agency
Environmental Monitoring
Systems Laboratory
Research Triangle Park NC 27711
                    Research and Development
EPA/600/S4-85/075 Jan. 1986
v>ERA          Project  Summary

                    Laboratory  and Field
                    Evaluation of the
                    Semi-VOST  Method
                    J. Bursey, M. Hartman, J. Homolya,
                    R. McAllister, J. McGaughey, and D. Wagoner
                     An initial laboratory and field evalua-
                    tion has been completed which as-
                    sessed the performance of the Semi-
                    Volatile Organic Sampling Train
                    Method (Semi-VOST) for measuring
                    concentrations of principal organic haz-
                    ardous constituents (POHCs) with boil-
                    ing points greater than 100°C that are
                    emitted from hazardous waste inciner-
                    ators. A draft formulation of the
                    methodology  was tested through an
                    experimental  laboratory program to
                    evaluate method performance charac-
                    teristics. The program consisted of:
                    (1) determination of minimum com-
                    pound detection limits; (2) simulation
                    of a flue gas stream to determine the
                    impacts of water vapor, SO2, NOX, and
                    HCI on sampling; and (3) determination
                    of precision and bias for the sample
                    preparation and analytical elements of
                    the method. Results of the simulation
                    study  are reported  for 12 test com-
                    pounds (POHCs) representing a range
                    of physical and chemical properties
                    which could affect sample train collec-
                    tion and recovery performance. Three
                    of the test compounds (toluene,
                    chlorobenzene, and 1,1,2,2-
                    tetrachloroethane) were selected for
                    further investigation to evaluate
                    method precision and bias under field
                    conditions through  the use  of four
                    simultaneously-operating Semi-VOST
                    trains.  Cylinder gas standards of each
                    compound were blended and dynami-
                    cally spiked into paired sample trains
                    for determining method bias. Estimates
                    of bias are presented for toluene and
                    1,1,2,2-tetrachloroethane. Method pre-
cision for chlorobenzene is calculated
from unspiked sample train results
which also included distributive vol-
ume experiments to determine sample
train breakthrough of POHCs. Project
results are presented in Volume I along
with recommendations on elements of
the method requiring further research.
Volume II includes appendices giving
detailed descriptions of the method for-
mulation, laboratory and field studies,
along with quality assurance data.
  This Project Summary was devel-
oped by EPA's Environmental Monitor-
ing Systems Laboratory, Research Tri-
angle Park, NC, to  announce key
findings of the research project that is
fully documented  in two  separate
volumes of the same title  (see Project
Report ordering information at back).

Introduction
  The Solid Waste Disposal Act, as
amended by the Resource Conservation
and Recovery Act of 1976 (RCRA), re-
.quires that the Environmental Protec-
tion Agency establish a national regula-
tory program to ensure that hazardous
wastes are managed in a manner which
does not endanger human health or the
environment. The statute requires EPA
to promulgate  performance standards
for hazardous waste management. In-
cluded in the promulgated regulations
are provisions for waste disposal by in-
cineration and requirements that haz-
ardous waste incinerators be so oper-
ated that the principal  organic
hazardous constituents (POHCs) are de-
stroyed or removed with a minimum ef-

-------
ficiency of 99.99 percent. To determine
destruction and  removal  efficiency
(ORE),  EPA has designated the Semi-
Volatile Organic  Sampling  Train
Method (Semi-VOST) as the method for
measuring concentrations  of POHCs
with boiling points greater than 100°C
that are emitted from hazardous waste
incinerators.
  The Quality Assurance Division of the
Environmental Monitoring Systems
Laboratory at Research Triangle Park,
North Carolina, has a program to evalu-
ate and standardize EPA source test
methods.  While participating in this
program, Radian Corporation has un-
dertaken an evaluation of  the  Semi-
VOST Method, which is the subject of
the full report. The following were the
objectives of the program:
  • assemble and  refine a draft formu-
    lation of the Semi-VOST sampling
    and analysis methodology;
  • select a series of Appendix VIII test
    compounds for method  evaluation
    based  on the characteristics of in-
    cinerability, water solubility, and
    adsorption strength on XAD-2®
    resin;
 •  design and conduct an experimen-
    tal  laboratory program to evaluate
    method  performance characteris-
    tics in  terms of minimum com-
    pound detection limits,  sampling
    flow rate and volume, and the for-
    mation  of artifact compounds dur-
    ing the sampling of a simulated flue
    gas stream containing water vapor,
    S02, NOX, and HCI;
 •  determine  precision and bias at-
    tributable to each of the sample
    preparation steps and the final GC/
    MS analysis step;
 •  design, construct, and evaluate the
    performance of a dynamic spiking
    system for determining method
    bias using experimental field data;
    and
 •  carry out a field test study at a haz-
    ardous waste incineration facility to
    evaluate method precision and bias,
    and the potential of POHC sample
    train breakthrough.
Procedure
  The initial Semi-VOST method formu-
lation contained no analytical protocols
but included several sections that were
discretionary or that required clarifica-
tion. Therefore, prior to laboratory ex-
perimentation, a refinement  in the for-
mulation was developed  by surveying
members of the technical community
who had experience in the use of the
Semi-VOST method. Based on the sur-
vey responses, a working copy of the
method was drafted to encompass both
sampling and analysis procedures.
  In  the  application of Semi-VOST,
gaseous  and particulate components
are isokinetically withdrawn from  an
emission source and collected in a mul-
ticomponent sampling  train. Key ele-
ments of the train  include  a  high-
efficiency particulate filter and a packed
bed of a porous polymeric adsorbent
resin (XAD-2®). The filter is used to sep-
arate stack gas particles from gaseous
substances, which are  then adsorbed
on XAD-2®. As a collection sorbent,
XAD-2®,  a polystyrene-divinylbenzene
copolymer, has the advantage of high
surface area (300 rq,2/g) with an average
pore volume of 90 A, which permits col-
lection of semi-volatile compounds. Fol-
lowing sample collection, the train com-
ponents  are extracted with solvent,
concentrated, and identified/quantified
using high resolution gas chromatogra-
phy coupled with low resolution mass
spectrometry.
  After return of the samples to the lab-
oratory, all fractions are spiked with sur-
rogate standards prior to extraction,
concentration, and analysis. The use of
surrogates provides a measure of the
efficiency of the sample preparation
procedures. In the sample analyses, an
examination of surrogate recovery is
used to qualify and possibly correct the
analytical results. For example, poor or
variable surrogate recovery would sug-
gest difficulties with sample prepara-
tion/analysis as opposed to incinerator
operation. The spiking compounds
should  be the stable, isotopically-
labeled analogs of the  compounds of
interest or a compound that would  ex-
hibit properties similar to  those of the
compounds of interest,  be easily  chro-
matographed, and not interfere with the
analysis. Potential surrogate spiking
compounds include deuterated  naph-
thalene,  chrysene, phenol, nitroben-
zene, chlorobenzene, toluene,  and car-
bon 13-labeled pentachlorophenol.

  The condensate knockout trap  solu-
tion is spiked with the surrogates, trans-
ferred to a separatory funnel, acidified,
and extracted with methylene chloride.
The  aqueous phase  is made alkaline
and re-extracted with methylene  chlo-
ride. This procedure is used to optimize
the recovery of POHCs with different
functional groups. The condensate may
contain compounds that show great
water solubility and low organic solvent
solubility. The methylene chloride solu-
tions are combined in a Kuderna-Danish
(K-D)  evaporator/concentrator and re-
duced in volume usually to less than
10 ml, and a standard is added for quan-
titation by gas chromatography coupled
with mass spectrometry (GC/MS). Im-
pinger contents are  extracted with
methylene chloride  under both acidic
and basic conditions and then concen-
trated. The methylene  chloride/
methanol rinses of the probe and filter
housing are prepared  by spiking with
surrogate standards and adding suffi-
cient water to a separatory funnel to
force methylene chloride to become im-
miscible with  water and form  two
phases. The XAD-2®  resin is spiked with
surrogate standards and  transferred
with the filter to an all-glass Soxhlet ex-
traction thimble. The module is rinsed
with methylene chloride, the solution is
added to the Soxhlet, and extraction is
conducted for 16 hours.
  Analysis is performed by GC/MS
using fused silica capillary GC columns.
The mass spectrometer is operated in a
full (40 to 450 daltons) scanning mode.
For most purposes,  electron ionization
(El) spectra are collected, since a major-
ity of the potential POHCs give  reason-
able El spectra. Also  El spectra are com-
patible with the NBS Library of Mass
Spectra  and other reference data that
aid in identification.
  For quantitation by GC/MS, each sam-
ple fraction is spiked  with a  known
amount of an internal standard  prior to
analysis. The response ratio of the inter-
nal standard to each  compound of inter-
est in the sample is then compared to a
similar ratio established with a calibra-
tion curve.

Results and Discussion

Laboratory Evaluation
  A laboratory program consisting of
three  stages was designed and imple-
mented to provide an initial estimate of
method precision and bias and to inves-
tigate method interferences. The initial
portion of the laboratory program in-
volved the  selection of compounds to
test the limits of the  method and for the
determination of minimum detection
limits (MDL) of the GC/MS. Compound
selection was followed by a series of
simulated flue gas  generation experi-
ments to identify external effects of the
sampling environment that would affect
the analytical results and to evaluate the
overall method precision and bias. Fi-
nally, a series of experiments was con-

-------
ducted to determine specifically the pre-
cision and bias of the laboratory
preparation and  analytical portions of
the method.
  Results from the laboratory studies
indicated the following:

  • Method precision and bias are
   compound-specific and cannot be
   generalized for all classes of chemi-
   cals.  Laboratory  experiments
   demonstrated that  method preci-
   sion and bias are  related to the
   distribution and recovery of a par-
   ticular POHC  throughout the com-
   ponents of the sampling train. Sam-
   ple distribution and recovery of a
   POHC is related to its boiling point,
   water solubility, and chemical func-
   tion.
 • Compounds  adsorbed on  XAD-2®
   resin exhibited better precision and
   recovery than compounds collected
   in the condensate and impingers.
 • Test compound recoveries were not
   affected by either simulated flue gas
   variations in compound concentra-
   tion by HCI, S02, or water vapor, or
   by variations in sample flow rate
   and volume.
 • Increases in generated flue gas NOX
   levels were  associated with de-
   creases in the recoveries of 1,4-
   dioxane, urethane, and phenol.
 • For toluene,  chlorobenzene, and
    1,1,2,2-tetrachloroethane,  labora-
   tory experiments showed that the
   precision of the resin extraction and
   concentration steps was not signifi-
   cantly different from the precision
   of the concentration step alone.
   However, the  precision of the sepa-
   ratory funnel extraction and con-
   centration steps was significantly
   poorer than either the precision of
   the concentration step alone or the
   resin extraction and concentration
   step.
 •  For chlorobenzene, toluene, and
    1,1,2,2-tetrachloroethane, no signif-
   icant losses were detected when the
   K-D concentrate was reduced from
   5 ml to 2.5 ml using an inert gas for
   blowdown.
Field Evaluation
  Following completion of the labora-
tory studies, a field  test program was
conducted to evaluate method perform-
ance at an operational hazardous waste
incineration facility. A test plan was pre-
pared which included the collection of
 matrixed sequence of samples for the
 evaluation of method precision, bias,
 and POHC sample train breakthrough.

  Results from  the  field evaluation
study included the following:
  • A preliminary estimate of method
    bias of -13 percent for toluene was
    calculated from field  test data ob-
    tained using the dynamic spiking
    system.
    -  The 87  percent  recovery of
      toluene from the field test data
      compares favorably  with the
      mean  recovery of 91 percent
      measured  from the laboratory
      baseline  performance test of the
      dynamic spiking apparatus.
    -  The identification  of toluene in
      the  methylene chloride recovery
      solvent returned from the field
      necessitates analysis of addi-
      tional sample fractions from the
      test to confirm the toluene bias
      estimate.
 •  A preliminary estimate of method
    bias of -16 percent was calculated
    for 1,1,2,2-tetrachloroethane.
    -  The 84  percent  recovery of
      1,1,2,2-tetrachloroethane is in
      agreement with the 85 percent
      value from the baseline dynamic
      spiking test.
 •  The dynamic spiking  concept was
    demonstrated as a viable approach
    for determining method bias in field
    evaluation studies.
 •  A method precision of 19.9 percent
    for chlorobenzene was  calculated
    from a field test data set consisting
    of a series of paired, unspiked Semi-
    VOST trains sampling  the  emis-
    sions from a hazardous waste incin-
    erator.
 •  No sample train  breakthrough of
    chlorobenzene could be determined
    from the field test distributive vol-
    ume experiments. Therefore, the re-
    tention volume for chlorobenzene
    was not exceeded  and demon-
    strates the adequacy of this aspect
    of the current method formulation.

Recommendations
  Based on the results of the method
evaluation studies presented in this re-
port,  the following recommendations
are  made for further investigation:
 •  Method bias and precision have
    been  shown to be compound
    specific. An investigation should be
    conducted to determine  if Ap-
    pendix VIII compounds could  be ag-
    gregated according to properties
 such as water solubility, chemical
 similarity, or  adsorption strength
 (retention volume) on XAD-2 to pro-
 vide estimates of method precision
 and bias for compound categories.

 Pyridine,  pentachlorophenol, and
 resorcinol exhibited poor recoveries
 during the flue gas generator exper-
 iments. Preliminary studies have in-
 dicated that compound losses may
 be  attributed  to chromatographic
 properties, poor solvent extraction,
 or  compound reaction  in the
 aqueous components of the sam-
 pling train which  become acidified
 during sample collection. Also, the
 low recovery of pentachlorophenol
 may have been due to loss within
 the flue gas generator. Further labo-
 ratory studies are warranted to clar-
 ify  these  observations and to de-
 velop  simplified screening
 procedures to  demonstrate the ap-
 plicability of the method's sample
 preparation (extraction) and analy-
 sis  (GC/MS conditions) steps  to
 other Appendix VIII compounds or
 classes  of compounds.
 The recovery  losses of urethane,
 phenol, and 1,4-dioxane associated
 with NOX flue  gas generator levels
 should be studied using solutions to
 simulate condensate and impinger
 contents for evaluating compound
 reactivity with  NOX. Findings could
 then be used to plan a second field
 evaluation of the method which  in-
 cludes  POHC artifact formation
 study objectives.
 The particulate filters from the
 Semi-Vost samples collected during
 the  field test program were stored
 and solvent extracted for later anal-
 ysis. The  extracts should be ana-
 lyzed for semi-volatile compounds
 in preparation for future  studies
 to evaluate POHC-particulate  mat-
 ter  interactions  during sample
 collection.
 The preliminary estimates  of
 method bias for toluene and 1,1,2,2-
 tetrachloroethane should be final-
 ized by analyzing the remaining
 sample fractions from the field test.
 The presence of toluene  in the
 methylene chloride rinse  solvent
should also be accounted for.
 The dynamic  spiking system for
 determining method bias should  be
 redesigned to improve its reliability
in the field. The  use  of a  stable,
single gas mixture containing multi-
 ple  components would simplify op-

-------
     eration and delivery of a dynamic
     spike  to  two  simultaneously-
     operating Semi-Vost trains. Stable,
     isotopically-labeled  POHCs would
     be desireable spiking components
     as it is unlikely that they would be
     present in the flue gas.
     The existing field study GC/MS ana-
     lytical data set  for unspiked  sam-
     pling runs should be re-examined to
     determine if additional flue gas or-
     ganic compounds could be used to
     evaluate sample  train break-
     through.
     Future field  method evaluation
     studies should include the addition
     of spiked  waste feed materials to
     permit measurement of method
     precision and sample train break-
     through for several pre-determined
     compounds contained in the spiked
     waste.
        J. Bursey, M. Hartman, J. Homolya, R. McAllister, J. McGaughey, andD. Wagoner
          are with Radian Corporation, Research Triangle Park, NC 27709.
        John Margeson is the EPA Project Officer (see below).
        The complete report consists of two  volumes, entitled "Laboratory and Field
          Evaluation of the Semi-VOST Method:"
            "Volume I," (Order No. PB 86-123 551 /A S; Cost: $11.95)
            "Volume II. Appendices." (Order No. PB 86-123 569/AS; Cost: $22.95)
        The above reports will be available only from: (cost subject to change)
               National Technical Information Service
               5285 Port Royal Road
               Springfield. VA 22161
               Telephone: 703-487-4650
        The EPA Project Officer can be contacted at:
               Environmental Monitoring Systems Laboratory
               U.S. Environmental Protection Agency
               Research Triangle Park, NC 27711
United States
tnvironmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
~U.S>!)STAGt  !•

". 0 Z211:
Official Business
Penalty for Private Use $300

EPA/600/S4-85/075
           0000329   PS

           U  S  ENVIR PROTECTION AGENCY
           REGION  5  LIBRARY
           230  S  0EAR80RI*  STREET
           CHICA60               It   60604

-------