United States
Environmental Protection
Agency
Environmental Monitoring
Systems Laboratory
Research Triangle Park NC 27711
Research and Development
EPA/600/S4-86/041 Mar. 1987
&EPA Project Summary
Evaluation of Photovac 10S50
Portable Photoionization Gas
Chromatograph for Analysis of
Toxic Organic Pollutants in
Ambient Air
Richard E. Berkley
The Photovac 10S50 portable pho-
tionization gas chromatograph was
evaluated as a monitor for fourteen
selected toxic organic vapors in ambient
air, including benzene, toluene, bromo-
and chloro-benzene, o-xylene, and nine
halo-methanes, ethanes, and ethylenes.
Such analyses have usually been done
by gas chromatography using a mass
spectrometer as detector (GC/MS).
This requires preconcentration of
analytes, which is an important source
of analytical error. A portable chromato-
graph with a detector sensitive enough
to detect pollutants without precon-
centration could usefully supplement
or complement data obtained by pre-
concentration/GC/MS and avoid such
errors. The Photovac 10S50 is a truly
portable instrument which can be
operated on battery power. It incorpo-
rates a sampling pump, a column
compartment, a 10.6 electron volt
photoionization detector, and a micro-
computer with printer-plotter. Ultra zero
air is used as carrier gas. A benzene
detection limit of 0.1 parts per billion
by volume is claimed, as well as the
ability to detect many hazardous pol-
lutants at ambient concentrations in
air.
An extrapolated benzene detection
limit of 95 femtograms was found, and
response to benzene was linear up to at
least 750 parts per billion. Field sam-
pling near Research Triangle Park, North
Carolina found apparent levels of ben-
zene and toluene up to 17 parts per
billion by volume. Trichloroethylene and
1,1,1-trichloroethane also apparently
were present. Installation of a fused-
silica capillary column and modification
of the sample loop to minimize dead
volume improved resolution at the cost
of slightly increased retention times.
The unit lived up to most of the manu-
facturer's claims. It actually is portable
and meets the claimed detection limit
for benzene of 0.1 parts per billion by
volume. It also can detect chloroethy-
lenes at similar concentrations and
chloromethanes and chloroethanes at
10- to 100-fold higher concentrations.
Unattended operation will require some
improvement in identification of the
calibrant peak.
This Project Summary was developed
by EPA's Environmental Monitoring
Systems Laboratory, Research Triangle
Park, NC, to announce key findings of
the research project that Is fully docu-
mented In a separate report of the same
title (see Project Report ordering In-
formation at back).
Introduction
Analysis of vapor phase organic com-
pounds in ambient air usually is done by
gas chromatography using a mass spec-
trometer as detector (GC/MS). The prin-
cipal advantage of GC/MS is its selectivity
and range of applicability, and its principal
disadvantage is its relatively high detec-
-------
tion limit. Preconcentration of analytes is
required because large quantities of
oxygen and water cannot be tolerated
and because mass spectrometers are not
sufficiently sensitive to detect ambient
levels of analytes in samples small enough
to be chromatographed. Unfortunately,
preconcentration produces many errors.
Efficiency of pollutant collection and
delivery to the GC/MS is low and variable
for many compounds, and numerous arti-
facts are formed. This method is at best
semiquantitative. It is also expensive and
labor-intensive. However, because of its
effectiveness in qualitative analysis it is
virtually indispensable and likely to re-
main so.
Analyzing pollutants directly without
separating them from air could avoid
these problems, but detection of one part
per billion by volume of benzene in one
milliliter of air requires a detection cap-
ability of three picograms. An analytical
method of such sensitivity could supply
data unspoiled by gross sampling errors.
Photoionization gas chromatography can
detect many toxic organic pollutants
without preconcentration. It is less selec-
tive than GC/MS because only retention
times are used for identification, but since
it does not use preconcentration, it avoids
many sampling errors. Samples are ex-
tremely small compared to the capacity
of the column, and the instrument can be
recalibrated frequently. This optimizes
accuracy of retention times. A peak which
appears within the "window" for a certain
compound could be produced either by
that compound or by another of similar
retention time. Though positive identifi-
cation is not established, the "identified"
compound cannot actually be present in
a concentration larger than indicated
unless there is a negative interference.
Thus the data can generally be taken as
credible estimates of the upper limits of
concentrations.
The unit is self-contained, mounted in
an aluminum case with internal electric
power and carrier gas supplies. It mea-
sures 46 x 16 x 34 centimeters (18 25 x
6.25 x 13.25 inches) and weighs 11.8
kilograms (26 pounds). It is equipped with
a microcomputer which controls sampling
and analysis and processes data, and it
has a built-in printer-plotter. Calibration
gas can be delivered from an aerosol can
mounted on the control panel inside the
case. It possesses significant advantages
over conventional gas chromatographs:
a. It is portable.
b. Its detector is sensitive enough to
respond to ambient background
levels of several toxic organic com-
pounds without preconcentration
(benzene, toluene, and the chloro-
ethylenes, for example). It could be
used as is to screen for these com-
pounds in ambient air or near haz-
ardous waste sites. It also could be
a valuable supplement or comple-
ment to GC/MS analysis.
c. Analyte concentrations are far below
levels that could overload the column
and lead to tailing or distortion of
peaks.
d. It is tolerant of oxygen, so that
separation of analytes from air is
unnecessary.
e. It is blind to water.
f. It can be transported without special
preparation.
There are also some disadvantages:
a. Photoionization detectors are not as
specific as mass spectrometers.
"Identification" of a compound by
Photoionization gas chromatography
is evidence but not proof of its
presence.
b. Because the column is exposed to
large quantities of air, it must be
operated at ambient or only slightly
elevated temperatures. This reduces
the quality of chromatographic
resolution and limits analytes to
compounds with high vapor pres-
sures at ambient temperatures.
In this study the Photovac 10S50 was
evaluated for direct detection of benzene,
toluene, o-xylene, chlorobenzene, bromo-
benzene, trichloroethylene, tetrachloro-
ethylene, dichloromethane, trtchloro-
methane, tetrachloromethane, 1,2-
dichloroethane, 1,2-dibromoethane,
1,1,1-trichloroethane, and 1,1-
dichloroethylene.
Results
Laboratory evaluation of the Photovac
10S50 began at the end of February,
1986 and continued until August, 1986.
Its capabilities were tested initially rby the
examination of its performance features
as described in the operator's manual.
These included peak resolution and in-
tegration, calibration, benzene detection
limit, reproducibility of analyses, and
operation with a capillary column. Per-
formance under field operating conditions
also was observed. Most of the work was
done while operating on battery power,
and no battery-related failures occurred.
As supplied, the instrument was equip-
ped for precolumn backflush operation.
Relatively volatile compounds passed into
the main column, whereas heavier ones
were eliminated by backflushing them
from the precolumn, which preventec
delay or disruption of subsequent anal-
yses. Potentially this could be a significam
convenience when sampling grossly con-
taminated air.
It was necessary to determine how
long to wait after injection before back-
flushing. Valve times which cut off al
peaks after toluene late one afternoon
were usually too short to pass toluene
the following morning, because the
column temperature was lower. Withoul
temperature control of the column enclo-
sure, even small changes in ambient
temperature caused problems.
Peak integration was controlled by three
parameters set by the operator and stored
in the microcomputer library. The way in
which these parameters influenced in-
tegrations was not obvious and was no1
explained in the Operator's Manual. Ap-
parently reasonable settings seemed to
keep the microprocessor from recognizing
the end of one peak until the next one
appeared. That often kept integration of
the last peak from being completed. Use
of settings recommended by the manu-
facturer solved this problem. It would
have been helpful if the Operator's
Manual had given more guidance for
setting these parameters and more in-
formation about the way they affected
integrations. When integrations were
performed while operating at high gam,
perpendiculars were dropped from the
beginning and end of each peak to the
level where the baseline had been at the
beginning of the run. That was a rea-
sonable procedure where early-elutmg
peaks were partially overlapped, but not
for late-eluting peaks that were riding on
an elevated baseline In that case, true
peak area could be a small fraction of
reported peak area, resulting in large
concentration errors. This problem is
potentially very serious when operating
the instrument unattended or while ob-
taining time-averaged results without
plotting chromatograms.
The Photovac 10S50 is intended nor-
mally to be calibrated by passing a
standard calibrant mixture from a pres-
surized container through the sample
loop. This method of calibration is com-
patible with the way data are reported by
the microprocessor, and the calibrant is
treated in the same way as any other
sample, which eliminates many potential
sources of systematic error. Recalibration
during field operation must be done in
this manner. Standards consisting of part
per billion to part per million mixtures ol
analytes in dry nitrogen are available
from vendors. This method of calibration
-------
is convenient to use and consistent with
the design of the instrument, but ob-
taining such standards may require sub-
stantial time and expense. Storage sta-
bility of calibration mixtures is very short
unless proper containers are used. Pres-
surized standards used in this study were
Standard Reference Materials from the
National Bureau of Standards (NBS SRM
No. 1805 and No. 1811).
When pressurized standard mixtures
were unavailable, calibration standards
prepared by static dilution of headspace
vapors were used. This kind of calibration
is useful for work in which syringes are
used in sampling, but standard mixtures
prepared by static dilution of headspace
vapors may not be very reproducible. Use
of certified pressurized standards is
clearly preferable.
Nonvolatile memory in the micropro-
cessor includes four Libraries, each of
which controls operating parameters and
has a capacity of 25 compounds. The
library entry for each compound includes
its name, retention time, response factor,
and warning level. Library entries are
created by the operator, after injecting a
sample, by entering information for in-
dividual peaks. Library entries can be
updated after each calibration by specify-
ing the peak number of the calibrant, its
library number, and its concentration.
Recalibration of one compound results in
proportional corrections to all library en-
tries. This makes it possible to use one
compound in each library as a standard
for the entire library. During continuous
cycling, all corrections are made auto-
matically after each calibration run.
The detection limit was determined by
injecting aliquots of NBS SRM No. 1805
(0.254 parts per million of benzene in
nitrogen) with gas tight syringes at a
constant gain of 50. Output signal was
sampled by computer at intervals of 0.2
second. Start and end points of peaks
were estimated by inspection and peak
area was determined by summing the
heights of all data points above the line
between them. The standard deviation of
the 40 baseline points immediately before
the beginning of the peak plus the 40
baseline points immediately after the end
of the peak was calculated. Peak areas
were subjected to linear regression
analysis. The slope of the plot was 0.502
volt*second/picogram with an intercept
of 6.4 volt*second and a correlation co-
efficient of 0.999. The aggregate standard
deviations of the baselines of all peaks
was 0.0238. The detection limit was taken
to be twice the aggregate standard devia-
tion of the baselines divided by the slope,
or 95 femtograms, equivalent to a con-
centration of 0.03 parts per billion by
volume of benzene at 25 C. This is lower
than the manufacturer's claim of 0.1 part
per billion and much lower than required
to detect 1 part per billion of benzene in
air. It should be remembered that no
estimate of detection limit can anticipate
complications that may arise during
operation in the real world. Nevertheless,
it is apparent that this unit, as is, should
easily be capable of detecting the part per
million or lower levels of benzene, ben-
zene derivatives, and haloethylenes that
raise concern about public health when
they are found near hazardous waste
sites and other pollutant sources.
Tiny extraneous peaks were found in
all of the calibration gases. Their number
was variable and unpredictable. Because
the calibrant peak was identified solely
by elution order, and because area limits
could not be used to suppress peak
recognition, misidentifications resulted.
This limited the feasibility of unattended
operation. If some peak other than the
calibrant received the peak number as-
signed to it, then its area was matched
with the calibrant concentration in the
library, and the unit was put catastrophi-
cally out of calibration until the next time
it happened to find the right peak during
a calibration run. Such errors could be
avoided only by timely recognition of the
problem by the operator, followed by ap-
propriate corrective action. Miscalibration
could cause serious errors in time
weighted averages of results obtained
during unattended operation.
After calibration, the unit was used to
sample laboratory air. Concentrations of
benzene between 0.001 and 0.003 parts
per million by volume were found, but no
other compounds. The unit also was op-
erated at nine locations near Research
Triangle Park, North Carolina. Background
levels usually could not be measured
because gam could not be set higher
than 20 without sending the calibrant
peak off scale. Use of a lower calibrant
concentration would have enabled detec-
tion of ambient vapors at lower levels,
but the unit, as presently configured,
could not report concentrations below
0.001 parts per million. Benzene, toluene,
1,1,1-tnchloroethane, and trichloroethy-
lene apparently were detected in the field
study. These identifications are based
only on retention times. Nevertheless,
"identified" compounds could not have
been present in concentrations larger
than indicated
A 0.53 millimeter X 25 meter fused
silica column with a 5 micrometer chemi-
cally bonded methyl silicone phase was
installed after completion of field evalua-
tion. The backflush valves were bypassed
and the sample loop was replaced with a
length of narrow-bore tubing which had
the same volume (1 milliliter). A slight
improvement in resolution was achieved
at the cost of a small increase in retention
times.
Conclusions and
Recommend ations
The Photovac 10S50 is a truly portable,
self-contained instrument, which gen-
erally performs as claimed. It is capable
as is of monitoring, under operator
supervision, ambient background levels
of benzene, volatile substituted benzenes,
and haloethylenes. Since it is earily cap-
able of detecting part per million or lower
levels of these compounds, it could be
very useful for monitoring volatile ef-
fluents from hazardous waste sites, at
temperatures between 20 and 38 C (68
to 100F). Since it operates without pre-
concentration, the results should not be
subject to preconcentration artifacts.
Because the calibration procedure reliably
divides the chromatogram into "windows"
in which each compound must appear if
it appears at all, it can provide estimates
of the upper limits of concentrations re-
gardless whether those identifications
actually are correct.
Response to benzene was linear over a
range from the detection limit to 750
parts per billion by volume. An extra-
polated detection limit of 95 femtograms
of benzene was calculated. This cor-
responds to an atmospheric concentration
of 30 parts per trillion by volume at 25 C,
which is better than the manufacturer's
claim of 100 parts per trillion by volume.
Automatic column backflush was found
to be of limited use without a temperature-
controlled column enclosure. Multiple
component calibration mixtures, prepared
by static dilution of headspace vapors in
ultra zero air, were used to calibrate the
unit for fourteen aromatic and halocarbon
compounds. Bromobenzene, the least
volatile of these, eluted in less than thirty
minutes at a carrier flow rate of 40 mil-
liliters per minute.
Field sampling at several locations in
the vicinity of Research Triangle Park,
North Carolina found apparent ambient
background levels of benzene and toluene
between the detection limit and 17 parts
per billion by volume. Trichloroethylene
and 1,1,1-trichloroethane were also ten-
-------
tatively identified. Use of a 25 meter
fused-silica capillary column afforded a
slight increase in resolution at the cost of
a slight increase in retention times.
Several problems were encountered.
The instrument could not be successfully
operated unattended due to miscalibration
problems. The Operator's Manual pro-
vided limited information about how the
instrument worked. Instructions for set-
ting integration parameters told nothing
about how they influenced the integration
process.
The Photovac 10S50 ought to be
modified to provide the following
improvements:
a. Controlled temperature (up to 40 C)
in the column enclosure.
b. Standard use of fused-silica capillary
columns.
c. A corrected and expanded Operator's
Manual.
d. Revised software to provide:
i. More reliable means of locating
the calibrant peak.
ii. Reporting of analyte concentra-
tions lower than 0.001 parts per
million.
iii. An escape option when a wrong
key has been pressed.
iv. Increased operator control over
peak integration methods.
It is also recommended that the modified
instrument by subjected to further field
testing to contrast results with those
obtained simultaneously by Tenax/GC/
MS and other methods.
The EPA author P. E. Berkley is with the Environmental Monitoring Systems
Laboratory, Research Triangle Park. NC 27711.
The complete report, entitled "Evaluation of Photovac 10S50 Portable
Photoionization Gas Chromatograph for Analysis of Toxic Organic Pollutants
in Ambient Air," (Order No. PB 87-132 8S8/AS; Cost: $13.95, 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 author can be contacted at:
Environmental Monitoring Systems Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
^^ U.S. OFFICIAL
"EMA1.TY
U.S.POSIAGE
fti't!2 i'B/ j ' ,•>/,!-£ '•
\ I )hL MOO! "~ f\ *t O "
/ * -1 r li ,4. t. ::
>S-- .Oi-MO X' " "PTTB;
Official Business
Penalty for Private Use $300
EPA/600/S4-86/041
0000329 PS
U S iNVIR
------- |