United States
                     Environmental Protection
                     Agency
Environmental Monitoring and
Support Laboratory
Cincinnati OH 45268
                     Research and Development
EPA/600/S4-88/029 Sept. 1988
SEPA          Project  Summary
                     Heated   Purge  and  Trap  Method
                     Development  and Testing

                     Samuel V. Lucas, Hazel M. Burkholder, and Ann Alford-Stevens
                      A heated purge-trap-desorb (HPTD)
                     analytical method for eight polar, water
                     soluble, volatile organic analytes was
                     developed and tested. This method uses
                     a standard 5-mL purge vessel with an in-
                     tegral, low volume, water-cooled con-
                     denser, and a Tenax trap. A commercially
                     available PTD apparatus can be used,
                     but a water bath (90°C) Is required to
                     heat the sample as it is being purged. A
                     5-mL sample aliquot heated to 90°C was
                     purged with helium gas (flow  of 40
                     mL/min); purged analytes were trapped
                     on a Tenax trap and desorbed through
                     a small volume condenser into  a gas
                     chromatograph (GC) equipped with a
                     wide bore (0.53  mm), fused  silica
                     capillary column coated with a polar sta-
                     tionary phase  (Supelcowax-10).  As
                     analytes eluted from the GC, they were
                     detected and measured with a  flame
                     ionization detector during  method
                     development and with a mass  spec-
                     trometer during method testing.
                      Of  33  compounds tested, 8 were
                     amenable to determination with HPTD
                     procedures. Those 8 compounds were
                     acetonltrile,  acrolein,  acrylonitrile,
                     2-butanone,  1,4-dioxane,  isobutanol,
                     methylacrylonitrile,  and propionitrile.
                     Analyte recoveries ranged from  about
                     80% for  acrylonitrile and  methylacry-
                     lonitrile to about 30% for 1,4-dioxane.
                     With the chromatographic conditions
                     used, calculated method detection limits
                     were 2-9 /
-------
  Some Appendix VIII and  Michigan list
compounds are so volatile that they are lost
during sample preparation procedures in
methods  using  liquid/liquid extraction
followed by extract concentration or are too
water soluble to be removed efficiently by
ambient  temperature  PTD  procedures.
Although some of these coumpounds have
been successfully determined by injection
of a water sample aliquot directly into a GC,
this procedure provides much higher detec-
tion limits (DLs) than are achievable with
PTD procedures for compounds that are ef-
fectively purged at room temperature.
  The goal of this project was to test the
feasibility of determining some or all of the
33 selected compounds by purging sam-
ples at temeratures  higher than ambient
laboratory  temperature  and, for  those
amenable to PTD, to  develop standardized
procedures for their  determinations.
Because determination  of as many com-
pounds as possible in one sample aliquot
was desired, a fused silica! capillary col-
umn was used instead of the packed col-
umns currently used in SW-846 methods,
because the former provide increased
analyte separation.
  The 33 compounds included in this study
were:

Nitrites
  Acetonitrile
  Acrylonitrile
  Propionitrile
  Malononitrile
  3-Chloropropionitrile
  Methacrylonitrile
  2-Hydroxypropionitrile
  2-Hydroxy-2-metnylpropionitrile

Alcohols
  Isobutanol
  Propargyl alcohol
  Chloral  hydrate
  2-Chloroethanol
  1,3-Dichloro-2-propanol

Aldehyde and Ketones
  Acrolein
  2-Butanone
  Bromoacetone

Thiol
  Trichloromethanethiol
  Thiophenol
  Methyl mercaptan

 Nirtogen Bases
   Pyridine
   2-Picoline
   Methylhydrazine
   1,1-Dimethylhydrazine
   1,2-Dimethylhydrazine
  n-Propylamine
  Aziridine
  Methylaziridine
  N-(2-Hydroxyethyl)aziridine

Miscellaneous
  Acreylamide
  1,4-Dioxane
  -Propiolactone
  Tetranitromethane
  2-Butanone peroxide

Procedures

Method Development
  Experimental parameters expected to af-
fect analyte performance were sequential-
ly examined. Two fused silica capillary col-
umns were used. One was a  30 m X 0.53
mm i.d. column coated with a 1.0-^m film
of polar stationary phase, Supelcowax-10,
and the other was a 60 m X 0.75  mm i.d.
column coated with a 15-^m film of relative-
ly  nonpolar siloxane stationary  phase,
Vocol. Both columns  (obtained  from
Supelco,  Bellefonte,  Pennsylvania) were
tested by directly injecting an aliquot of a
water  solution  containing all  candidate
analytes, and the column judged to provide
better overall performance was used in all
subsequent work. Analytes were detected
and  measured  with a  flame  ionization
detector (FID).
  Each day that aliquots of analyte fortified
reagent water were analyzed, two  or more
aliquots of a calibration solution were also
directly injected into the GC. The  injected
quantity of each analyte was the same as
the quantity used to fortify 5-mL samples
that were purged. Typically, 250 ng of each
analyte was injected: this corresponded to
an analyte concentration of 50 ng/L in a
5-mL  sample.  Results  of duplicate or
triplicate  analyses  of 5-mL  aliquots of
reagent water containing analytes were us-
ed to calculate absolute recovery of each
analyte. (Recovery was 100 times the GC
peak area for a  purged analyte divided by
the GC peak area measured when an equal
quantity of that analyte was injected directly
into   the  GC.)   The effect  of  purging
temperature on analyte recovery was tested
at five temperatures (22, 40, 60,  85, and
99°C).
  Analytes were screened for hydrolytic
stability; those not detected by direct injec-
tion of an aliquot of a pH 6.8 buffered (0.01
M phosphate in reagent water) standard
solution held at 85°C for 15 minutes were
not included in subsequent work. Analytes
that were hydrolytically stable were further
tested using purging conditions similar to
those described  in  Method 8030  for
acetonitrile, acrolein, and acrylonitrile. In
most cases, analytes not recovered at con-
centrations >1000 iigIL were eliminated
from further testing.
  Also tested was the possible enhance-
ment  of analyte recovery by shifting the
aqueous/vapor phase equilibrium constant
toward the vapor phase by adding salt to
the aqueous solvent. Salts were selected
on the basis of their solubility, commercial
availability,  and high ionic  strength.
Chloride and  sulfate  salts of sodium and
magnesium were used at concentrations
that provided 80% saturation at 85°C.
  Three alternatives were tested to control
water vapor that exits the heated aqueous
sample along with analytes. A condenser
was attached to the purge vessel outlet via
a 0.25-in. Swagelok union with Teflon fer-
rules. One condenser had a 0.25 in. o.d. X
8 cm cold zone that was packed with 3-mrn
glass helices. Another condenser had an
8 mm o.d. X 10 cm cold zone with the tem-
perature of the condenser cooling watei
controlled at 20°C. The condenser was im-
mersed in a heated water bath up  to the
beginning of the condenser cold  zone
Each  sample was purged for 15 min at 4C
mL/min (total volume 0.6 L) was used with
the condenser water maintained at 20°C
and the trap at 23-25°C during the purge
step.  Two  non-condenser alternatives foi
control of  the water evaporated from the
purge vessel  were examined.  One ap
proach  used  a molecular sieve (Linde
zeolite type 3A) between the purge vesse
and the trap. In the other approach, twc
high-retention  trapping materials,  Car
bosieve and  Carbotrap (Supelco),  were
used  when the trap temperature was hek
above the dew point of the purge gas. This
temperature was 90°C when purge gas was
not diluted with a post-purge make-up gas
and 70°C with 1:1 purge gas/make-up gas
A combination trap consisting of 4:1 Car
bopack: Carbosieve (with the Carbotrap a
the trap inlet) was also tested at 90°C trap
ping  temperature with no purge carrie
dilution.

Method  Performance
  The final chromatographic and HPTD ex
perimental conditions (90°C sample purge
temperature)  resulting  from  the  methoc
development activities were tested using ar
MS detector. Samples  were  purged in ;
5-mL  purge vessel contained in a commer
cially  available apparatus (LSC-2, Teckma
Co.,  Cincinnati,  Ohio).  After  passinc
through a 10-cm X 8-mm o.d. glass Vigreu;
condenser, purged analytes were collecte(
in a Tenax trap. Sample temperature was
maintained at 90°C while helium purge gai
flowed  at 40  mL/min  and  the tra|

-------
temperature was maintained below 25°C.
Sorbed analytes were desorbed from the
trap at 180°C for 4 min after a 50°C preheat.
After desorption, the trap was baked at 210°
for 10 min to prepare it for  the next sam-
ple. Desorbed analytes were introduced in-
to a GC equipped with a 30-m x 0.53-mm
i.d.  fused silica capillary column coated
with a  1.0-pm film of Supelcowax-10
(Supelco) with helium carrier gas flow of 10
mL/min. The  oven temperature program
was 40°C for 4 min (during trap desorption),
then 8°C/min to 130°C. The GC was inter-
faced to  an  MS through  a glass-jet
separator. The quadrupole MS was scann-
ed from m/z 35 to 250 at 1  scan/sec, and
the MS was calibrated according  to the
standard USEPA criteria for bromofluo-
robenzene. The MS detector response to
each analyte was determined relative to
50  fjg/L  of  an  internal standard,
benzene-d6, which  was added to each
sample just before purging.
  The GC/MS system was calibrated with
solutions containing the eight analytes over
a 100-fold concentration range beginning
near the estimated detection limit (DL) of
each compound. Data were obtained for
ten replicates of a water sample fortified
with each analyte  at a  concentration
thought to be low enough to allow computa-
tion of its method detection  limit (MDL) ac-
cording to a standard  Agency procedure
(Appendix B to Part 136, 49 FR 26189). An
MDL is defined as the lowest concentration
at which an analyte can be  measured and
reported  with  99%  confidence that the
analyte concentration is greater than zero.
The MDL is the product of Student's t fac-
tor and the standard deviation of seven or
more replicate measurements at a concen-
tration near but greater than the estimated
analyte DL. Method performance data were
also obtained for ten additional replicates
of a water sample fortified with each analyte
at a concentration 10 times greater than
that used to calculate MDLs.

Results and Discussion

GC Testing
  Only megabore (>0.53 mm o.d.) fused
silica columns were used  in this study,
because a wide bore column is required to
accommodate the 10 mL/min desorption
flow rate, unless prefocusing sorptive or
cryogenic traps are used. Two types of GC
column stationary phases were tested; one
was a highly polar phase (Supelcowax-10)
and  the other was a relatively nonpolar
siloxane phase (Vocol) that is often used for
Method 8240 determinations. Because the
polar phase   (Supelcowax-10)  provided
significantly greater analyte selectivity and
substantially reduced GC peak tailing of
polar analytes, it was used for all subse-
quent  work.  Unfortunately, many  polar
analytes (especially alcohols) for which the
Supelcowax-10 provided the most dramatic
chromatographic improvement compared
to the Vocol column, were not sufficiently
recovered by HPTD. Therefore, the choice
of GC  column was principally driven by
candidate analytes that were later shown
to be inappropriate for  HPTD determina-
tions. Furthermore, all candidate analytes
that  were appropriate  for HPTD eluted
before water on the Supelcowax-10 column,
but the effect of this desorbed water was
not apparent during methdd development
activities.

Hydrolytic Instability Testing
  Six candidate analytes were eliminated
from further study because they were not
sufficiently stable in reagent water buffered
to pH 6.8 and held at 85°C. Those com-
pounds were N-(2-hydroxyethyl)aziridine,
methylaziridine,    methylhydrazine,
1,1-dimethylhydrazine,  tetranitromethane,
and  thiophenol. Although acrolein and
3-chloropropionitrile   displayed  some
hydrolytic instability, they were retained for
subsequent method development studies.

Effect of Salt on Analyte
Recoveries
  Analyte recovery with  HPTD procedures
should be enhanced through a shift  in the
aqueous phase-vapor phase equilibrium
constant toward the vapor phase. This shift
is produced by using high salt concentra-
tions to lower the activity coefficient of the
aqueous solvent. Initial testing showed that
sodium and  magnesium chlorides dis-
solved  fairly rapidly with  good reproducibili-
ty. The sulfate salts, however, were highly
exothermic on dissolution, produced salt
cakes  in the purge vessel, and did not
always completely dissolve. The high con-
centration of magnesium sulfate required
to achieve 80% saturation  produced a
viscous liquid that apparently formed a li-
quid plug at the purge  vessel outlet/con-
denser inlet.
  The  greatest enhancement of analyte
recovery with salt addition was observed for
the two most polar, water soluble analytes,
1,4-dioxane  and  acetonitrile.  Dioxane
recovery increased from 21% without salt
to 56% with sodium sulfate, and acetonitrile
recovery increased from 59% without salt
to 88% with magnesium chloride. Although
2-chloroethanol was well  recovered from
traps when spiked directly onto them, it was
not detected  in  any HPTD experiments.
  Although  the  use  of  salt  increased
recoveries of all analytes that were not
quantitatively recovered without salt, the
advantages of using salt were judged to be
less important than the disadvantages  of
using salt. Salt particles transported as an
aerosol from the  purge  vessel can  be
deposited in the PTD apparatus and sorb
or chemically degrade some analytes.
Deposited salt can also block the transfer
line; this results in a plugged  and  ruined
column. Because of these difficulties, the
use  of salt  with HPTD procedures was
discontinued.

Effect of Purge Temperature on
Analyte Recoveries
  No optimal temperature between room
temperature and  100°C  was  observed,
because recoveries of all analytes increas-
ed uniformly with temperature. The three
most volatile  and least  water  soluble
analytes (i.e., those with  the  highest re-
coveries at  22 and  40°C), 2-butanone,
methacrylonitrile,  and acrylonitrile, were
recovered  essentially  quantitatively  at
>85°C. The least volatile and  most water
soluble  analytes,  1,4-dioxane and
acetonitrile,  were  not  expected  to  be
recovered quantitatively at any temperature
using a  5-mL sample and a  purge gas
volume of 600 mL. On the basis of these
results, 90°C was  adopted as the purge
temperature for the GC-MS method perfor-
mance evaluation  study.  A 90°C purge
temperature should allow use of a water
bath in all but the most extreme altitudes
and  hypobaric climatic  conditions.

Trap Breakthrough
  Because methyl  mercaptan was  not re-
tained in the trap  when only  200 mL  of
purge gas was used, this  compound was
not  appropriate  for  determination with
HPTD procedures. At the trap temperature
(25°C) proposed for  GC-MS  determina-
tions, acetonitrile and acrolein began  to
breakthrough at about 510 and 525 mL,
respectively. This was considered to be
adequate for a 450-mL purge volume. With
reasonable control of HPTD  conditions,
these two analytes should be  trapped.

Control of Purged Water Vapor
  The major difficulty for HPTD determina-
tions is created by the large  amount  of
water vapor that exits  in the  purge gas
along with analytes from the heated sam-
ple purge vessel. For a typical 11-min purge
at 90°C and purge gas flow of 40 mL/min,
about 0.75 g of water vapor exit the purge
vessel. Condensation of this relatively large
amount of water in the PTD  apparatus,

-------
especially the trap, can cause extremely
poor reproducibility, and can even prevent,
measurement of analyte concentrations.
Three  main  effects  are observed:  (1)
analytes are sequestered in water droplets
in the connecting lines and valve; (2) GC
separation is degraded when water is in-
troduced into the  GC  column; and  (3)
analytes are lost when dissolved in water
exiting the trap during sample  purging.
Therefore, a major concern was to minimize
the effect of water.
  Three main approaches were investi-
gated:
  • A condenser was attached to the purge
    vessel outlet to return most water vapor
    to the purge vessel.
  • A trap that could be  used  at  an
    elevated temperature during purging
    was used with  a post-purge make-up
    gas to reduce the purge carrier dew
    point and with the trap temperature
    maintained  above the dew point to
    eliminate any possibility of condensa-
    tion of water  vapor from the purge
    carrier.
  • A dessicant was placed between the
    purge vessel and the trap to adsorb
    water vapor from the heated purge car-
    rier while allowing purged analytes to
    pass through.

Effect of Condenser Design
  The packed condenser that was used for
initial method development work retained
essentially all of the water condensate. The
purge gas percolated  through the liquid
water and equilibrated with a large volume
of cold aqueous phase. Recovery of purg-
ed  analytes was  significantly  reduced.
Subsequent modifications in condenser
design focused on minimizing the volume
of cold  condensate  to reduce  purged
analyte recapture and determining the ef-
fect  of  the  condenser  cooling water
temperature on analyte recovery. As ex-
pected, analyte recoveries increased with
increasing temperature of the condenser
cooling water and  decreasing volume of
cold condensate.  In  the improved con-
denser, which was used during later  GC-
MS  testing,  condenser cooling water
temperature was maintained at 20°C to en-
sure that condensate would not form in the
PTD apparatus.
  A clean, dry condenser was required for
each analysis, but thorough rinsing of the
purge vessel in place  did not eliminate
analyte carryover, probably because some
analytes  remained  in the small volume of
cold water on the condenser surfaces.
When the condenser was also rinsed, a
relatively large amount of water was left
behind and apparently recaptured analytes
purged during the next sample analysis.
The result was reduced analyte recovery,
compared to that obtained with a dry con-
denser. The 10-cm long condenser used for
the GC/MS  experiments  is  probably
somewhat longer than actually necessary,
but no experiments have been performed
to determine the optimum length.

Purged Water Control Options Not
Involving A Condenser

  Because three  candidate  analytes
thought to be recoverable by HPTD were
not recovered using a condenser, two other
approaches to control the water vapor were
investigated. All three of those compounds,
propargyl alcohol, 2-chloroethanol,  and
acrylamide, had been recovered when plac-
ed directly onto traps before a normal PTD
procedure. The lack of recovery of these
compounds with a condenser was thought
possibly to be caused by recapture of purg-
ed analyte in the small amount of cold con-
densate present in the condenser.

  The two noncondenser approaches were
using a molecular sieve to remove water
and using trapping materials that could
withstand temperatures above  the dew
point of the purge stream. With a molecular
sieve, water vapor was removed between
the purge vessel and the trap. Isotherms
for  water  adsorption  provided  by   the
manufacturer were used to predict that 3.0
g of molecular sieve would adsorb all of the
purged  water at a sieve temperature of
100°C. The use of  a  near.stoichiometric
quantity of molecular sieve was expected
to minimize analyte adsorption  onto  the
molecular sieve surface. With standard H-
PTD  conditions,  however,  analyte
throughput was totally unacceptable, with
<1.0% throughput of 2-butanone, metha-
crylonitrile  and  acrylonitrile,  and  no
throughput of the other 11 analytes tested.

  The use of trap packing material that
could be operated at higher temperatures
was not successful. Carbotrap did not re-
tain  the  most volatile  analytes  at  the
elevated trapping temperature. While Car-
bosieve apparently retained analytes at the
elevated temperature, GC separation of
early  eluting analytes was unacceptable.
When analytes were placed directly onto
the  combination  (4:1  Carbopack/Car-
bosieve) trap, a normal HPTD sequence us-
ing reagent water produced severely split
GC  peaks for  early eluting  analytes
(especially acrolein  and 2-butanone) and
broadened peaks  for mid-elution range
analytes.
Method Performance
  Method performance was determined by
GC/MS analysis of 10 replicate aliquots of
reagent water fortified with each analyte at
a concentration estimated to be near its DL
and 10 replicates fortified at a concentra-
tion 10 times the estimated  DL (Table 1).
Relative  standard deviation (RSD)  was
about 5% for the higher concentration and
about 7% for the lower concentration. In the
low   concentration  samples,  mean
measured concentration bias ranged from
-18% to -1% for seven of the eight analytes.
For 2-butanone, background  concentra-
tions of 6-10 /»g/L prevented measurement
of the fortified concentration of 6 pg/L In
the high  concentration  samples,  mean
analyte  bias ranged from -34% to 0%.
Without  the acrylonitrile bias of  -34%,
mean analyte bias ranged from -15% to
0%.

  Results from analyses of reagent watei
fortified with low concentrations of the eight
compounds produced calculated MDLs ol
2-9 pg/L for seven compounds. With low
concentration data, an MDL could  not be
calculated for  the  eighth  compound
2-butanone, because a high backgrounc
concentration prevented accurate measure
ment of  a fortified concentration of 6 pg/L
When data obtained from analyses o
reagent water containing 2-butanone at i
concentration of 60 pg/L were  used, ar
MDL of 85 ngJL was calculated. Four of the
seven MDLs calculated from low concen
tration data may not, however, be realistic
because four compounds were fortified a
concentrations  that were >5  times thi
calculated MDLs.  Additional data are re
quired to assess MDLs  for those com
pounds. Such data should be acquired wit)
different  GC  conditions  that product
sharper GC peaks. All method analyte!
eluted'before desorbed water, which was
in effect, the injection solvent. This wate
caused a reverse solvent effect for analyte
eluting before the solvent, and peak shap<
and  resolution were impaired.

Conclusions
  Acceptable method performance wa
observed for the eight  compounds the
were amenable to removal from aqueou
samples with HPTD procedures. Future e
forts should be directed toward modifyin
Method 8240 to incorporate these eight ac
ditional   analytes,  because a separat
method for only eight analytes is probabl
not cost  effective.  In addition, using
heated  sample purge vessel  and a cor
denser to trap purged water vapor may in
prove recovery of  some Method 824

-------
Table 1.    Method Performance Data
High Concentration
Analytes/
Surrogate Stds.
Acrolein
2-Butanone
Methacrylonitrile
Acrytonitrile
Acetonitrile
Propionttrile
1,4-Dioxane
Isobutanol
2-Butanone-ds
Acetonitriie-ds
1,4-Dioxane-d8
p-Bromofluorobenzene
Fort.
Cone.
pg/L
300
60
200
200
400
200
600
600
60
400
300
50
Mean
Meas.
Cone.8
rt/L
300
52s
180
132
340
174
552
510
64
408
315
50
Std.
Dev.
X9/L
5
3
5
3
4
5
9
5
5
5
6
3
Low Concentration
Fort.
Cone.
?9/t-
30
6
20
20
40
20
60
60
60
400
300
50
Mean
Meas.
Cone."
pg/L
26
14°
20
17
34
18
55
49
62
380
303
48
Std.
Dev.
pg/L
7
16
10
5
8
5
7
4
4
4
4
2
MDL
P9/L
6
6
2*
9
3d
4d
7*
-
-
* Mean of 10 determiniations.

b Method detection limit, where MDL = Std. Dev. X t, where t = 2.821 for 10 measurements

0 Uncorrected for the background concentration, which was 6-10 pg/L

" Fortified concentration was <5 time calculated MDL.
analytes   (such   as  naphthalene,
trichlorobenzenes,  tetrachlorobenzenes,
and hexachloropropene) while not adverse-
ly affecting recovery of more volatile sam-
ple components. Capillary columns other
than the two used in this study should be
evaluated to select a column providing bet-
ter chromatographic characteristics.
   Samuel V. Lucas and Hazel M. Burkholder are with Battelle Columbus Division,
     Columbus. OH 43201-2693; the EPA author Ann Alford-Stevens is with
     the Environmental Monitoring and Support Laboratory, Cincinnati, OH 45268.
   Robert O'Herron is the EPA Project Officer (see below).
   The complete report, entitled "Heated Purge and Trap Method Development
     and Testing," (Order No. PB 88-242 6O7/AS; Cost: $14.95, 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:
           Environmental Monitoring and Support Laboratory
           U.S. Environmental Protection Agency
           Cincinnati,  OH 45268

-------
United States                     Center for Environmental Research                                      BULK RATE
Environmental Protection            Information                                                   POSTAGE & FEES PAI
Agency                         Cincinnati OH 45268                                                  gpA

                                                                                           PERMIT No. G-35


Official Business
Penalty for Private Use $300

EPA/600/S4-88/029
                                     OOOC329  f»S

                                     U $  SHVlf*  WtOTfCTIOK  HGEMCY
                                     REGION  5 H»Ri*f
                                     230  S  OEARtCtH  STREET
                                     CHICA60              IL

-------