-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing!
1. REPORT NO.
EPA/600/4-85/Q63
3. RECIPIENT'S ACCESSION NO.
b 120631"
«- TITLE AND SUBTITLE
A Cryogenic Preconcentration - Direct FID (PDFID)
Method for Measurement of NMOC in Ambient Air
5. REPORT DATE
October 1985
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Frank F. McElroy and Vinson L. Thompson
(EMSL/RTP); Harold 6. Richter (OAQPS/RTP)
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Quality Assurance Division/EMSL
Office of Research and Development
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Monitoring Systems Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVE RE D
Final
14. SPONSORING AGENCY CODE
EPA/600/08
15. SUPPLEMENTARY NOTES
16. ABSTRACT
Accurate measurements of atmospheric concentrations of non-methane
organic compounds (NMOC) are necessary in the application of photochemical
models that are used by states in developing the control strategies needed
to achieve compliance with ambient air quality standards for ozone. NMOC
measurements obtained with available continuous NMOC analyzers have often
been of inadequate quality. Speciated gas chromotographic measurements,
though adequate, are excessively difficult and expensive where speciated
data are not needed. A simplified cryogenic preconcentration, direct flame
ionization detection (PDFID) method that is sensitive and provides accurate
measurements of ambient NMOC concentrations has been developed and standard-
ized sufficiently to be recommended for use by state and local air polluton
control agencies in the development of their ozone control plans. Recent
refinements to the method are discussed, an automatic remote sampling
system is described, and the performance (precision and accuracy) of the
method is characterized, based on results from utilization of the method
for NMOC analysis of 1375 air samples collected from 22 sites during the
summer of 1984. A complete description of the method is also provided in
an appendix.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
COSATI 1 icld'GiOup
Air pollution measurement
analysis of organic
compounds
Non-methane organic
compounds, NMOC,
EKMA, Ozone Control
18. DISTRIBUTION STATEMENT
RELEASE UNLIMITED
EPA Form 2220-1 (R«v. 4-77) PREVIOUS EDITION
19. SECURITY CLASS (TIlis Report}
UNCLASSIFIED
21. NO. OF PAGES
84
20 SECURITY CLASS (This page)
UNCLASSIFIED
22. PRICE
is OBSOLETE
-------
DISCLAIMER
The information in this document has been funded wholly or in part by
the United States Environmental Protection Agency under contract 68-02-3431
to Research Triangle Institute and contract 68-02-3513 to Radian Corporation.
It has been subject to the Agency's peer and administrative review, and
it has been approved for publication as an EPA document.
Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
-------
FOREWORD
Measurement and monitoring research efforts are designed to anticipate
potential environmental problems, to support regulatory actions by developing
an in-depth understanding of the nature and processes that impact health and
the ecology, to provide innovative means of monitoring compliance with regu-
lations, and to evaluate the effectiveness of health and environmental
protection efforts through the monitoring of long-term trends. The Environ-
mental Monitoring Systems Laboratory, Research Triangle Park, North Carolina,
has responsibility for: assessment of environmental monitoring technology
and systems; implementation of agencywide quality assurance programs for air
pollution measurement systems; and supplying technical support to other groups
in the agency, including the Office of Air, Noise and Radiation, the Office
of Toxic Substances, and the Office of Enforcement.
Development of a better and more easily utilized method for measuring
ambient levels of nonmethane organic compounds will help states to determine
more precisely the extent of control of such compounds that is necessary to
achieve and maintain compliance with applicable air quality standards for
ozone. This document is intended to substantially standardize such a method,
characterize its performance, and improve its availability and applicability
by assisting potential users in its implementation.
Thomas R. Mauser, Ph.D.
Director
Environmental Monitoring Systems Laboratory
Research Triangle Park, North Carolina
m
-------
ABSTRACT
Accurate measurements of atmospheric concentrations of non-methane or-
ganic compounds (NMOC) are necessary in the application of photochemical
models that are used by states in developing the control strategies needed
to achieve compliance with ambient air quality standards for ozone. NMOC
measurements obtained with available continuous NMOC analyzers have often
been of inadequate quality. Speciated gas chromotographic measurements,
though adequate, are excessively difficult and expensive where speciated data
are not needed. A simplified cryogenic preconcentration, direct flame ioniz-
ation detection (PDFID) method that is sensitive and provides accurate mea-
surements of ambient NMOC concentrations has been developed and standardized
sufficiently to be recommended for use by state and local air pollution
control agencies in the development of their ozone control plans. Recent
refinements to the method are discussed, an automatic remote sampling system
is described, and the performance (precision and accuracy) of the method is
characterized, based on results from utilization of the method for NMOC
analysis of 1375 air samples collected from 22 sites during the summer of
1984. A complete description of the method is also provided in an appendix.
IV
-------
CONTENTS
Disclaimer ii
Foreword iii
Abstract „ iv
Figures vi
Tables vii
Acknowledgments viii
1. Introduction 1
2. Basic PDFID Method 4
3. Refinements
Analytical instrument . . 6
Trap ' 7
Water interference 7
Sample canisters 11
Sample collection system . . 15
4. Field Test and Method Performance
Description 19
Results 20
Calibration drift . ....... 22
System reproducibility 22
Analytical precision 22
Overall precision . 23
Accuracy 25
References 30
Appendix
Cryogenic Preconcentration, Direct FID Method Description
-------
FIGURES
Number Pa9e
1. Schematic diagram of the PDFID analytical system. 5
2a. Ratio of the responses of various compounds to propane 8
response, using an open trap.
2b. Ratio of the responses of various compounds to 8
propane response, using a beaded trap.
3. FID baseline shift at various moisture levels. 10
4. Technique used to minimize water interference. 12
5. Sample flow rate into a canister as the canister 16
pressure rises for two simple flow control devices.
6. Sampling system for integrated samples. 17
7a. Plot of the differences between repeat analyses for 24
28 ambient samples versus concentration.
7b. Plot of the percent differences between repeat analyses 24
for 28 ambient samples versus concentration.
8a. Plot of the differences between duplicate samples for 58 26
duplicate sample pairs versus concentration.
8b. Plot of the percent differences between duplicate samples 26
for 58 duplicate sample pairs versus concentration.
9. Comparison of PDFID Measurements to GC speciation 28
measurements for 336 ambient samples.
10. Comparison of NMOC measurements of 120 ambient samples 29
by 2 independent PDFID analytical systems.
VI
-------
TABLES
Number Page
1. Stability of Ambient Samples in Stainless Steel 14
Canisters.
2. Summary information for 1984 NMOC Monitoring Project. 21
3. Analytical precision. 23
4. Overall precision. 25
5. Accuracy relative to propane standards. 27
Vll
-------
ACKNOWLEDGMENTS
The authors acknowledge the contributions of Dr. William A. McClenny
and Mr. J. Marvin McBride of the Environmental Protection Agency, Dr. R. K.
M. Jayanty of the Research Triangle Institute, and Andrew Blackard, formerly
with the Research Triangle Institute, in the original development of the
PDFID method. We also acknowledge the work of Andrew Blackard, David Dayton,
Dr. Denny Wagoner and many others associated with the Radian Corporation in
carrying out the 1984 summer NMOC monitoring project, and particularly the
invaluable assistance of Dr. Robert McAllister of Radian in providing statist-
ical analysis of the project data. Finally, we acknowledge the contributions
of Dr. Edwin L. Meyer and Mr. Stanley F. Sleva of the Office of Air Quality
Planning and Standards, Dr. R. K. M. Jayanty and Mr. Maurice Jackson of the
Research Triangle Institute, and Dr. William A. McClenny and Mr. Joseph E.
Knoll of the Environmental Monitoring Systems Laboratory who technically
reviewed this manuscript and provided many constructive comments.
VII 1
-------
Section 1
INTRODUCTION
Accurate measurements of ambient concentrations of nonmethane organic
compounds (NMOC) are important to the control of photochemical smog because
these organic compounds are primary precursors of atmospheric ozone and other
oxidants. Achieving and maintaining compliance with the National Ambient Air
Quality Standards for ozone thus depends largely on control of ambient levels
of NMOC.
A number of photochemical dispersion models have been developed to des-
cribe the quantitative relationships between ambient concentrations of NMOC
and other compounds (e.g. NOX) and subsequent downwind concentrations of
ozone. 1 An important application of such models is to determine the degree of
control of NMOC that is necessary, in a particular area, to achieve compliance
with applicable ambient air quality standards for ozone.1»2 To achieve this
purpose, the models require input of accurate data on ambient concentrations
of NMOC.
The more elaborate theoretical models generally require detailed organic
species data.2 Such data must be obtained by analysis of air samples with a
sophisticated, multicomponent gas chromatographic (GC) species analysis
system.2»3 simpler empirical models such as the Empirical Kinetic Modeling
Approach (EKMA)l require only total NMOC concentration data—specifically the
average total NMOC concentrations from 6:00 AM to 9:00 AM.2
Until recently, ambient NMOC measurements for EKMA were often obtained
with commercial, continuous NMOC analyzers.2 However, measurements from these
instruments have been shown to be unreliable, particularly at concentrations
below about 0.5 ppmC*, due to a variety of instrument-related problems. These
problems included the indirect, subtractive nature of the measurement process
employed (total organic compounds minus methane),4,5,6,7,8,9 non-uniform per-
carbon response for different compounds due to oxygen interference,8'9 inade-
quate sensitivity,5'6 and interference from water vapor.5 A technical
*By convention, concentrations of NMOC are reported in units of parts per
million carbon (ppmC), which for a specific compound is the concentration
'by volume (ppmV) multiplied by the number of carbon atoms in the compound.
-------
assistance document was prepared to assist users in minimizing these pro-
blems,10 but the usefulness of NMOC measurements obtained with these instru-
ments is nevertheless limited.
The GC speciation method3 provides much more accurate and reliable
ambient measurements. Utilizing cryogenic preconcentration followed by GC
separation and flame ionization detection (FID), the technique provides quan-
titative, identified, species concentrations of the C2 to CIQ compounds typi-
cally observed in ambient air.9 NMOC measurements may be obtained by summing
the individual species concentrations. The cryogenic sample preconcentration
greatly enhances the sensitivity of the method while effectively minimizing
interference from methane and oxygen, allowing direct measurement of various
organic species with little variation in the per-carbon response for most
compounds of interest.
Although capable of providing adequate ambient NMOC measurements, the GC
speciation method requires sophisticated analytical equipment, a high level of
operator skill and experience, and'considerable time per analysis, rendering
such measurements expensive and labor-intensive. This expense is often not
justified in EKMA applications where speciated data are not required.
The obvious solution that suggests itself is to develop a simplified
method derived from the speciation method that eliminates the GC separation
and much of the operator skill required but retains the cryogenic precon-
centration for good sensitivity, the FID for selectivity, and the inert
carrier gas for uniform per-carbon response. Such a simplified method employ-
ing a GC instrument without a column was used by Cox et al, to measure NMOC
near Louisville, Kentucky and Nashville, Tennessee in I960.11 A similar
system, described by Jayanty et al. utilized a simple, commercial continuous
total hydrocarbon analyzer rather than a GC instrument.12,13 These methods
are indeed much simpler, faster, less expensive, and considerably easier to
use than the GC speciation technique.
In its efforts to standardize this simplified preconcentration direct
FID (PDFID) method, the Methods Standardization Branch (MSB) of the Quality
Assurance Division, Environmental Monitoring Systems Laboratory (EMSL) con-
ducted extensive testing of the method. Refinements to reduce interference
from water vapor and improve measurement precision were developed and incor-
porated into the method. Also, sampling apparatus for automatic collection
-------
of remote samples was developed and tested. Tests of both the analytical and
sampling systems included laboratory performance tests and comparison of
measurement performance to that of the highly regarded GC speciation method.
Finally, an extensive 22-city NMOC monitoring project in the summer of 1984
provided an excellent opportunity to assess method performance and monitor
operational problems during an actual field measurement application of the
methodology. The results of this work are contained in the remainder of this
report.
-------
Section 2
BASIC PDFID METHOD DESCRIPTION
A detailed description of the PDFID method and operating procedure,
incorporating changes and refinements developed to date, is provided in the
Appendix. A brief description of the method follows.
Figure I shows a schematic diagram of the basic analytical apparatus.
Major components include a sample volume metering system, a six-port gas
valve (Seiscor Model VIII, Seismograph Service Corporation, Tulsa, OK), a
cryogenic preconcentration trap, and an FID-integrator-recorder system.
The vacuum pump is used first to reduce the pressure in the vacuum
reservoir to a known selected absolute pressure setting, usually about
10 kPa (75 mm Hg). With the gas valve in the sample (trapping) position,
sampling is initiated by opening the sample valve. Air is drawn through the
trap, which is immersed in liquid argon (-186° C). The volume of air passing
through the trap is measured by the increasing pressure (diminishing vacuum)
in the reservoir, and the sample flow is stopped (by closing the sample
valve) when a second selected pressure is reached. The trap condenses
NMOC while permitting air and methane to pass through. During the trapping
mode, helium carrier gas passes through the gas valve directly to the FID.
Following the trapping mode, the gas valve is switched to the inject
position, directing the helium carrier gas through the trap, in the direction
opposite to the previous sample flow, before passing to the FID. The cryogen
is then removed, and the trap is heated to approximately 90° C. Organic
compounds collected in the trap revolatilize and are swept into the FID by
the carrier gas. The resulting peaks are integrated and converted to ppmC by
an NMOC calibration curve, prepared previously with propane-in-air concentra-
tion standards. Use of the same precise reservoir pressure readings for each
trapping cycle results in a constant sample volume and allows calibration of
the system with known concentration standards without knowledge of the actual
sample volume.
-------
VACUUM
VALVE
ABSOLUTE
PRESSURE GAUGE
SAMPLE
LOW
PRESSURE
REGULATOR
VACUUM
PUMP
CANISTER
VALVE
SAMPLE
METERING
VALVE
PRESSURIZED
CANISTER
SAMPLING
GLASS
BEADS
CRYOGENIC
SAMPLE TRAP
(LIQUID ARGON)
Figure 1. Schematic diagram of the PDFID analytical system.
-------
Section 3
REFINEMENTS
ANALYTICAL INSTRUMENT
The method described by Jayanty, et al.13 utilized a modified commercial
total hydrocarbon analyzer, rather than a chromatograph, in the analytical
system to minimize equipment cost. However, this instrument lacked a signal
integrator, temperature control for the FID, and precision flow control for
carrier and FID support gases. Most recent model gas chromatographic instru-
ments contain a very sophisticated signal integrator-recorder, a high-quality
temperature-controlled FID, and precision gas flow controls, as well as a
temperature-programmable oven, provision for controlling the valve tempera-
ture, and a sequence timer. These features make such an instrument very
advantageous for the method to improve analytical precision even though no
chromatographic column is required.' Gas chromatographs are commonly available
in many laboratories, and a chromatograph could be used on an intermittent
or time-share basis, if necessary.
Several other potential advantages of a gas chromatograph may also be
realized, depending on the degree of integration or modification of the
instrument to the method requirements. If the chromatograph has a multiport
valve, it may be suitable for use with the method. In this case, or if any
pneumatically operated valve is used, the chromatograph may be used to control
the valve as well as the entire analytical cycle. The chromatograph oven may
also be conveniently used to heat the trap during analysis of the trapped
NMOC.
In the system used by MSB, a Seiscor Model VIII pneumatic 6-port valve
and the trap were mounted inside the oven of a Hewlett-Packard Model HP-5840A
gas chromatograph. Rather than using a beaker of warm water to heat the
trap, as described by Jayanty, et al.,13 the oven was programmed to heat to a
maximum temperature of 90° C. However, as long as the oven door was open, the
oven did not heat. As soon as the dewar of liquid cryogen was removed from
the trap and the oven door closed, the trap was heated to about 90° C in
approximately one minute in a manner that was repeatable from cycle to cycle.
-------
Mounting the valve inside the oven allowed it to operate at an average
temperature somewhat above ambient to avoid possible loss of organic compounds
in the valve. In a subsequent, more sophisticated system used in the field
monitoring project (described later), the valve was mounted outside the oven
on an insulated block thermostatically controlled by the chromatograph
(Hewlett-Packard Model 5880) such that the valve was maintained at a constant
temperature of 95° C. In both systems, the valve and trap were located close
to each other and close to the FID to minimize the lengths of interconnecting
lines.
TRAP
Although most cryogenic hydrocarbon traps described in the literature
are packed with glass beads or other packing,14 the system described by
Jayanty, et al.^ specified an open (unpacked) tubular trap 120 cm long and
coiled into 8 loops. Using the same open trap used by Jayanty, we were unable
to duplicate the equal response to 'known concentrations of various paraffinic
and olefinic hydrocarbons reported by Jayanty. Our system's response to
similar compounds, presented in Figure 2a, showed considerable and unpredict-
able variability from compound to compound. However, when the open trap was
replaced with a short (30 cm), single-loop, stainless steel trap packed with
60/80 mesh glass beads, the response was much more uniform, as shown in
Figure 2b. The packed trap was also more convenient to use and appeared to
warm more uniformly, resulting in symmetrical peaks that were repeatable.
Peaks with the open trap often had an unsymmetrical shape and the peak shapes
were not repeatable, as was also reported by Jayanty.
WATER INTERFERENCE
In testing the method, a marked decrease in precision was noted when
ambient samples were analyzed compared with analysis of laboratory standards.
Similarly, preliminary comparisons with the GC speciation technique showed
greater variation for ambient samples than expected. The obvious difference
between laboratory standards and ambient samples is in moisture content, the
former being very dry while the latter usually contain considerable moisture.
Jayanty reported negligible water vapor interference,^ and in routine opera-
-------
t)
M
C
O
a
v>
tt
a:
1
1
9
•7
6
5
4
3
2
1
n
-
-
1 .00
1 .03
.78
.62
.67
.68
c o
. b J
.62
.60
V
c
a
a
o
t.
a.
u
c
a
X
+j
u
u
c
t)
f —
X
0.
0
t.
CL
0)
C
1)
-,-
•o
A
+J
3
CD
1
m
u
c
u
•(->
3
m
u
c
0
•»->
3
CQ
U
C
a
X
u
I
a>
c
u
3
. —
0
H
0)
C
V
^~
X
X
1
z
*—4
Figure 2a. Ratio of the responses of various compounds to propane
response, using an open trap.
1 .2
1 . 1
.
i
(U
U) • ^
c
o .8
a
M .7
Q)
^ .6
u _
> .5
^ .4
"^ -3
o
. 2
. 1
.
•
-
Ink n
m ^3 O
U
C
10
Q.
O
c
a.
1CT4
• tJ"
(U
c
K)
t~
4-1
U
1.11
U
c
t)
• —
X
a
0
CL
1 05
(U
c
0)
•—
T3
(0
•I-1
m
i
n
1 . 08
0)
c
0)
^>
3
m
^ ^
. y~o
0)
c
(0
^_)
3
CD
R7
• Or
QJ
C
10
X
flj
I
. 90
c
c
u
3
• —
O
r-
Q Q
. y c.
A)
c
u
, —
X
X
1
~
Figure 2b. Ratio of the responses of various compounds to propane
response, using a beaded trap.
8
-------
tion of the analytical system, no obvious effects due to moisture were indi-
cated. However, when the recorder scale was greatly expanded and traces
recorded for humidified zero air were compared to traces recorded for dry
zero air, it was clear that moisture caused a definite positive shift in the
FID baseline. This shift started as soon as the trap was warmed, and is
illustrated in Figure 3. The amount of the baseline shift seemed to be
constant for different levels of humidity, but the duration of the shift
varied, being roughly proportional to the total moisture content. The base
line returned to normal after the moisture had apparently been completely
flushed out of the trap by the flow of carrier gas.
In our system, the chromatograph was used to control the valve and ana-
lytical cycle. A manifestation of this configuration was that the instrument
automatically returned the valve to the trapping mode at the end of each
analytical cycle. This required that the duration of the analytical cycle be
long enough (~ 3 minutes) to allow complete flushing of the moisture from
the trap by the carrier gas before the valve returned to the trapping position.
The integrator continued to operate during this time, which would present no
problem with a stable baseline. However, with the positively shifted base-
line, the integrator continued to accumulate area, and even though the shift
was small, the area accumulated over the extended duration of the analytical
cycle caused a significant positive bias proportional to moisture level.
Further investigation indicated that for very humid samples, the dura-
tion of the baseline shift could extend past the end of the analytical cycle.
This situation resulted in two detrimental effects: (1) the trap was not
completely dry at the beginning of the next cycle, and (2) the integrator is
designed to interpret the reading at the beginning and end of the cycle as
baseline references, and it constructs a baseline based on those points.
When the baseline was still shifted due to moisture at the end of the analyt-
ical cycle, the integrator constructed an upward sloping baseline, which
caused a substantial difference in the peak area calculated by the integrator.
This difference in the baseline for different moisture levels obviously caused
considerable variability in the analytical results—even among replicate ana-
lyses of the same sample canister—because the moisture content of successive
samples from a canister could increase as the canister pressure decreased if
the canister contained condensed water. Since the canisters were pressurized
-------
DRY ZERO AIR
13,000 ppm WATER
o
Q.
C/9
LU
oc
Q
16,000 ppm WATER
27,000 ppm WATER
TIME, min
Figure 3.
FID baseline shift
(expanded vertical
at various
scale).
moisture levels
10
-------
to about three atmospheres (absolute), condensation in the canister would be
expected for any sample where the ambient relative humidity was over about 30
per cent.
Attempts were made to remove the moisture content of the sample air by
passing it through a Nafion tube (Purma-Pure) dryer, but without success. A
number of small Purma-Pure dryers were tested and did reduce the moisture
sufficiently to avoid a significant zero shift. However, all the dryers out-
gassed organic compounds. After long purging with dry zero air, the outgassing
could be reduced to near zero with dry air; but passing wet air through the
dryer would cause the outgassing to start again. It may be possible to clean
such a dryer to the point where it could be used for reducing moisture.
However, we made no further attempts to use the Nafion dryers.
Instead, the moisture problem was minimized by reducing the duration of
the integration period to a value just sufficient to include the NMOC peaks.
This value, determined empirically, allowed integration of the NMOC peak area
of the ambient samples tested but 'eliminated most of the area caused by the
water offset. More of the water-related area was eliminated by the integrator,
which constructed a positively sloped baseline as illustrated in Figure 4.
There was enough moisture even in fairly dry ambient air samples to insure
that the shift always extended beyond the end of the integration period.
Thus, the area was always calculated the same way. This technique reduced
the effect of the moisture to such a degree that the system showed little
bias even though calibrated with dry gases. The short analytical cycle did
not allow the trap to dry out before the valve switched back to the trapping
mode, so to adequately purge the trap, the valve was manually changed to the
inject mode for two minutes (determined empirically) at the end of each
analytical cycle. This technique was adopted and resulted in good precision
and good agreement with the speciation technique, as described later.
SAMPLE CANISTERS
Although the method can be used for direct, in situ ambient NMOC measure-
ments, greater utility is achieved by collecting ambient air samples remotely
for analysis at a central site. The most reliable type of sample container
appears to be the passivated stainless steel canister.3 The relative sta-
bility of hydrocarbon compounds in such canisters has been demonstrated.3
11
-------
01
Z
o
a.
LU
ff
O
a!
^
OPERATIONAL
^END
INTEGRATION
i
.---'-
' /
WATER-SHIFTED
BASELINE
J,
"\^
t
BASELINE NORMAL BASELINE
CONSTRUCTED BY INTEGRATOR
TO DETERMINE CORRECTED AREA
TIME
Figure 4. Technique used to minimize water interference. The amount
of baseline shift shown is exaggerated with respect to
the area of typical ambient NMOC peaks to more clearly
illustrate the construction of the operational baseline.
Nevertheless, we wanted to (a) verify that our canisters were not contaminated,
(b) verify that they could be easily and reliably cleaned after each sample,
(c) establish a reliable cleaning procedure, and (d) confirm short- and
long-term stability of ambient samples in the canisters.
The canisters used were an all stainless steel, 6-liter, spherical type
manufactured by Demaray Scientific Instruments, Ltd., Pullman, Washington.
The inside surfaces of these canisters had been passivated by the manufacturer
using the SUMMA process (Molectrics, Carson, CA).
Analyses of canisters filled with zero air compared well with direct
analysis of the zero air, confirming that the canisters were not contaminated
at the start of testing.
A basic cleaning procedure consisted of alternately evacuating the
canister to a vacuum of < 5 mm Hg, then pressurizing it to about 200 kPa (30
psig) with zero air, repeating this cycle three times. The last fill with
zero air was analyzed to verify that the cylinder was clean before the final
evacuation. To test this procedure, the canisters were used to collect
several ambient air samples near a high-traffic roadway and then subjected
12
-------
to the cleanup procedure after each sample. Some of the canisters were
heated in an oven to 150° C during the cleaning, while others were cleaned at
room temperature, in all tests, the analysis of the final zero air fill
showed little difference from the direct analysis of the zero air, indicating
that the cleaning procedure was effective and reliable. Since no advantage
was apparent from heating the canisters during cleaning, the room temperature
procedure was adopted.
The stability of various hydrocarbon compounds in stainless steel cani-
sters has been tested by many researchers. There is substantial evidence to
indicate that the stability of such compounds is excellent for periods of
several days to two to three weeks.3 The stability of 11 ambient NMOC samples
stored in canisters for periods of 7 to 21 days is shown in Table 1. The
samples were obtained during June, July and August of 1985 from a variety of
eastern and southern areas of the country and ranged in concentration from
about 0.15 ppmC to about 2.5 ppmC. Each sample was analyzed at seven-day
intervals with the analyses compared to the original (day 0) analysis.
Both the actual differences and the percent differences from the original value
are shown.
Each time sample air is withdrawn from a canister for an analysis, the
pressure in the canister decreases. Thus, the test results in Table 1 include
any effect that may be caused by the change of sample pressure in the canister
(which would also be expected to affect the amount of moisture in the sample).
As a control experiment, each of seven canisters were analyzed repeatedly on
the same day as the canister pressure decreased. All of these analyses
results were within 6% of the original analysis except one very low level
sample for which a difference of 18% was observed, but this difference was
only 0.032 ppmC. From this data and from Table 1, it can be concluded that
relatively little degradation of the sample occurs, either with pressure
change or with storage of the NMOC sample in the canister for periods of up
to three weeks. In spite of this evidence indicating general stability,
however, it seems prudent to analyze canister samples as soon as possible
after collection.
13
-------
Table 1. STABILITY OF AMBIENT SAMPLES IN STAINLESS STEEL CANISTERS
Analysis, ppmC
CAN
21
41
146
167
174
196
179
77
154
107
43
SAMPLED
6/19/85
6/26/85
6/19/85
5/22/85
5/22/85
6/06/85
7/09/85
7/17/85
7/24/85
8/01/85
7/26/85
DAY 0
.413
.154
.475
1.076
1.170
2.115
1.050
.440
.695
2.284
2.500
DAY 7
.399
.178
.425
1.089
1.151
2.055
1.014
.420
.676
2.257
2.472
DAY 14 DAY 21
.447
.202
.426
1.168 1.041
1.180 1.100
1.780
.362
.678
2.070
2.590
N =
Mean =
Standard Deviation =
Di
DAY 7
-.014
.024
-.050
.013
-.019
-.060
-.036
-.020
-.019
-.027
-.028
11
-.021
.024
fference, ppmC
DAY 14 DAY 21
.034
.048
-.049
.092 -.035
.010 -.070
-.335
-.078
-.017
-.214
.090
10 2
-.042 -.053
.137 .025
Difference, percent
DAY 7
- 3.4
15.6
-10.5
1.2
- 1.6
- 2.8
- 3.4
- 4.5
- 2.7
- 1.2
- 1.1
11
- 1.3
6.3
DAY 14
8.2
31.2
-10.3
8.6
0.9
-15.8
-17.7
- 2.4
- 9.4
3.6
10
-.3
14.5
DAY 21
-3.3
-6.0
2
-4.6
1.9
-------
SAMPLE COLLECTION SYSTEM
EKMA models generally require 3-hour integrated NMOC measurements over
the 6 AM to 9 AM period. A technique used previously for collection of such
integrated samples is to first collect the sample in a Mylar or Tedlar bag
over the required time period. At the end of the sample period, the sample
is transfered into a stainless steel canister3 because the integrity of the
sample can degrade rapidly in the bag. Disadvantages of this technique in-
clude added risk of contamination from the bags, permeation of compounds into
or out of the bags, difficulty in cleaning the bags, possible bag leaks, and
the inconvenience of the additional steps necessary to transfer the sample to
a second container.
It would be much better and far more convenient if the sample could be
collected directly into the canister. The chief problem is the difficulty of
maintaining a constant, low flow rate of sample air into the canister as the
canister pressure rises during the'collection period. Since the sample must
pass through any flow control device used, it must be of a simple, non-contam-
inating design. Electronic flow controllers are expensive and maybe subject
to calibration drift. Pressure regulators pose significant potential for
sample contamination, and capillaries do not control the flow rate adequately
without pressure regulation. .Critical orifices control flow rates better,
but all of these control devices require a large pressure drop to maintain
criticality. This large pressure difference requires a large pump (which
must be of a non-lubricated, non-contaminating design), and the high pressure
causes the pump to operate at considerably elevated temperatures, causing
possible changes or contamination of the sample as it passes through the pump.
Figure 5 compares the flow control characteristics of a capillary
configured between an air pump and a 6-liter canister with a short capillary
(hypodermic needle) located in the inlet of the pump. In both cases, the
pump was a stainless steel, metal bellows type (model MB-151, Metal Bellows
Company, Sharon, MA) with a maximum outlet pressure of about 275 kPa (40 psig).
With the capillary (or orifice) located between the pump outlet and the
canister, the flow drops considerably (-37% for the capillary) as the canister
pressure rises to 100 kPa (15 psig). However, when a short capillary is
15
-------
located on the vacuum side of the pump, the flow rate remains quite constant,
dropping only -9.8% for the same condition. If a lower final pressure can be
accepted in the canister (i.e., lower total sample volume), then the flow
decrease is less (-5.3% to 76 kPa (11 pslg)). Also, temperature rise in the
pump is minimal when the orifice is located on the vacuum side of the pump.
The flow rate can be adjusted by selecting a capillary or orifice of
appropriate size. Glass or sapphire orifices are available, and hypodermic
needles, although not really orifices, work acceptably and are inexpensive
and readily available. A 30 gauge needle, 2.54 cm long provided about the
140
120
100
c
1
80
0 60
40
20 -
O HYPODERMIC NEEDLE ON VACUUM SIDE OF PUMP
D CAPILLARY ON PRESSURE SIDE OF PUMP
-15
-10
-505
CANISTER PRESSURE, psig
10
15
20
Figure 5. Sample flow rate into a canister as the canister pressure
rises, for two simple flow control devices.
16
-------
right flow rate to fill a 6-liter canister to 100 kPa (15 psig) over a three-
hour period (-67 cm3/min). Figure 6 illustrates the sampling configuration.
A final minor problem with direct sampling into canisters concerns
unattended operation. A timer is easily configured to start and stop the
pump at the desired times. But since the canister is evacuated prior to the
sampling period and pressurized following the sampling period, it must be
tightly valved off during non-sampling periods to prevent leakage and valved
on during the sampling period. A conventional solenoid valve would serve this
purpose, but its temperature rises substantially when energized during
sampling, jeopardizing the integrity of the air sample. A special type of
bistable solenoid valve (Skinner model V52d2A1100 valve with Viton seal and
Magnelatch coil) that requires energization only briefly to open or close
overcomes this problem. This valve may be operated with a special (e.g.
electronic) timer that can be programmed to energize the valve for brief
SAMPLE
IN
CRITICAL
ORIFICE
VACUUM IN
PUMP
PRESSURE
GAUGE
METAL
BELLOWS
PUMP
CANISTERvS;
Figure 6. Sampling system for integrated samples
17
-------
periods at the appropriate turn-on and turn-off time. Alternatively, the
valve may be controlled with the same conventional-type mechanical timer that
operates the pump, by using a simple capacitive circuit to provide the brief
power pulses needed for the valve. More details on the sampling system are
provided in the method description in the Appendix. Twenty-two sampling
systems based on the configuration described above were used very successfully
in the 1984 summer NMOC monitoring project.
18
-------
Section 4
FIELD TEST AND METHOD PERFORMANCE
DESCRIPTION
During the summer of 1984, the PDFID method was used to obtain 6 AM
to 9 AM NMOC concentration measurements in 22 urban sites in the eastern and
central areas of the United States.15.16 These data were needed by partici-
pating states or local agencies to prepare State Implementation Plans (SIP)
for ozone control, using EKMA to calculate the reduction in NMOC levels
necessary to achieve compliance with the National Ambient Air Ozone Standards.
MSB participation in this project provided an excellent opportunity to assess
the performance of the method in an actual, typical field application and
with a large data base.
Early in the planning of the project, it was decided that all NMOC
samples from all 22 cities would be collected in canisters and shipped to a
common laboratory located at Research Triangle Park (RTP) rather than equip
each city with a separate analytical system for local analysis. The rationale
for this decision was based on several considerations: (1) the method was
new and unproven, and any operational problems could be resolved more expedi-
tiously at a common RTP laboratory; (2) considerable experience and expertise
was needed to assemble, set up, and test two new dual analytical systems and
train the analysts, and these functions could best be handled at a common RTP
laboratory; (3) the analytical system had not been cost-optimized, and the
cost of fabricating 22 analytical systems greatly exceeded the costs associ-
ated with shipping and analysis at an RTP laboratory; (4) shipping all samples
to RTP allowed better control of the analytical variability and assessment of
method performance; and (5) central analysis at RTP allowed for re-analysis of
some samples by GC for comparison, for quality assurance, and for use of the
speciated data by EPA and the participating state agencies.
The sampling systems were configured as shown in Figure 6 and used a
Metal Bellows Model MB-151 pump, Skinner Magnelatch solenoid valve, mechanical
timer with capacitive pulse circuit, and 30 gauge hypodermic needle with a
stainless steel filter on the pump inlet to control the sample flow rate at
19
-------
about 65 cm3/min for a 3-hour integrated sample. Occasionally, a larger
hypodermic needle was substituted so that two canisters could be filled
simultaneously over the 3-hour period for assessment of overall measurement
precision. Canisters containing the collected air sample were shipped daily
to the RTP laboratory in special, all-metal shipping containers. Additional
details of the project are available in Reference 16.
To accommodate the 22 daily samples plus calibration and quality control
samples, four identical PDFID analytical systems were fabricated and imple-
mented on two Hewlett-Packard Model 5880 dual chromatographs. Samples were
randomly assigned to one of the four analytical systems for analysis, and
analysts were regularly alternated between the two dual systems to avoid
systematic bias.
Each 6-liter canister was pressurized to approximately one atmosphere
gauge pressure and thus contained about six liters of available sample above
atmospheric pressure. Since less than 0.5 liter of sample was needed for
analysis, multiple analyses of each sample were possible. To reduce variab-
ility, two replicate sequential analyses were made routinely on each sample
and averaged, with a third analysis made if the first two were not in reason-
able agreement (standard deviation _< 0.2 ppmC). Some randomly selected
samples were re-analyzed later the same day—again in duplicate--on either
the same or a different system, for assessment of analytical precision.
Further, some samples were analyzed on a fifth, independent system operated
by MSB. Selected samples were also subjected to a complete species analysis
by GC. After all analyses had been performed, the canisters were flushed,
re-evacuated, and sent to the sampling sites for additional samples.
RESULTS
In all, 1375 valid samples, ranging in concentration from 0.06 ppmC to
4.75 ppmC, were analyzed from the 22 sites.16 Fifty-eight of these were
duplicate samples collected simultaneously in two paired canisters. Table 2
lists the 22 cities and the maximum and mean of the NMOC concentrations
measured at each site. Also listed in Table 1 are the sampling dates and
the per cent completeness (number of valid samples obtained divided by the
number expected) for each site. Overall completeness for the project was
90.6 per cent.
20
-------
Table 2. SUMMARY INFORMATION FOR 1984 NMOC MONITORING PROJECT
ro
Site Location
Akron, OH
Atlanta, GA
Beaumont, TX
Birmingham, AL
Charlotte, NC
Chattanooga, TN
Cincinnati , OH
Clute, TX
Dallas, TX
El Paso, TX
Fort Worth, TX
Indianapolis, In
Kansas City, MO
Memphis, TN
Miami, FL
Philadelphia, PA
Richmond, VA
Texas City, TX
Washington, DC
West Orange, TX
West Palm Beach, FL
Wilkes-Barre/Scranton, PA
TOTAL
OVERALL AVERAGE
First
Scheduled
Sampl ing
Date
6/27/84
7/11/84
6/18/84
7/11/84
7/11/84
7/11/84
6/27/84
6/18/84
6/18/84
6/18/84
6/18/84
7/11/84
6/27/84
7/11/84
8/13/84
6/27/84
6/27/84
6/18/84
6/27/84
6/18/84
6/18/84
6/27/84
Final
Scheduled
Sampl ing
Date
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
9/28/84
Total No.
Scheduled
Sampl ing
Days
68
58
75
58
58
58
68
75
75
75
75
58
68
58
35
68
68
75
68
75
75
68
1459
Total No.
Dupl icates
3
3
2
1
4
2
1
5
4
4
3
3
2
1
3
2
1
4
3
3
2
2
58
No. Valid
Samples
68
55
66
56
60
46
65
72
74
69
69
59
68
52
31
63
64
69
63
71
72
63
1375
62.5
Percent
Completeness
96
90
86
95
97
77
94
90
94
87
88
97
97
88
82
90
93
87
89
91
94
90
90.6
NMOC,
Mean
0.79
0.79
0.89
0.99
0.54
1.34
0.93
0.82
0.97
0.93
0.97
0.80
0.79
1.43
1.32
1.02
0.53
0.92
0.81
0.69
0.54
0.45
0.853
ppmC
Max
2.95
4.27
3.84
2.91
2.52
4.08
3.93
4.31
2.57
2.45
3.81
2.50
2.76
4.75
3.70
3.10
1.22
4.74
2.93
3.35
2.63
0.96
-------
CALIBRATION DRIFT
Each analytical system was calibrated at the beginning and at the end of
each day's analyses, using clean, dry, zero air and propane standards traceable
to an NBS propane Standard Reference Material (SRM), No. 1665b. Zero drift
during the day, defined as the final zero reading minus the initial zero read-
ing, ranged from -0.016 ppmC to +0.013 ppmC using data from all four analyt-
ical systems. The mean of the zero drift was 0.00016 ppmC, with a standard
deviation of 0.00219 ppmC. Daily span drift, defined as the difference
between each day's end calibration value and the initial calibration value
divided by the initial value, ranged from -10.1% to +14.1%. Mean span drift
was +1.47%, with a standard deviation of 4.06%.
SYSTEM REPRODUCIBILITY
To learn if there were any differences between the four analytical sys-
tems, four samples of local ambient air were collected on each of four days
(June 14, 15, 18, and 19, 1984) before the monitoring program began. Each
sample was analyzed by each of the four systems for a total of 60 analyses
(4 analyses were missing).* The NMOC concentrations ranged from about
0.14 ppmC to 1.25 ppmC. An analysis of variance (ANOVA) of these measurement
data showed that there was no significant difference among the four instruments
at the 0.05 level. In addition, the ANOVA indicated that the canisters did
not contribute significantly to the differences in analytical results, con-
firming that the canisters could be used, cleaned, and reused successfully.
ANALYTICAL PRECISION
Known concentration standards of specific compounds are not adequately
representative of ambient mixes of NMOC. Therefore, analytical precision for
the four analytical systems combined was assessed from the differences ob-
served between the original analyses and repeat analyses later in the sane
day (or the next day) for 28 of the samples. As shown in Table 3, the 28
differences ranged from -0.14 to +0.270 ppmC with a mean of 0.0250 ppmC and a
standard deviation of 0.1067 ppmC. Since the mean is substantially smaller
*As noted previously, each such analysis was actually the average of two
or possibly three sequential, replicate analyses.
22
-------
than the standard deviation, we can conclude that the mean is not significantly
different than zero. This suggests that the time delay between the first and
second analyses had no significant effect on the concentration measurement.
Table 3. ANALYTICAL PRECISION
Number of repeat analyses 28
Range of differences -0.14 to +0.27 ppmC
Mean difference +0.0250 ppmC
Standard deviation of differences 0.1067 ppmC
Range of per cent differences -38.3 to +30.7%
Mean per cent differences 0.2%
Standard deviation of per cent differences 12.7%
When the absolute values of the differences in ppmC are plotted against
concentration (Figure 7a), there is indication of a proportional relation-
ship--!'^., the differences tend to be proportional to the concentration.
However, when the absolute value of the per cent differences (concentration
difference divided by the average of the two analyses) are plotted (Figure
7b), little or no concentration dependence is indicated. Accordingly, the
analytical precision may be described as the standard deviation of the per
cent differences, which is 12.7%. The average of the per cent differences is
0.2, which, as expected, is not significantly different from zero.
OVERALL PRECISION
Overall precision was assessed from the differences between the analyses
of 59 pairs of duplicate samples collected simultaneously in duplicate,
paired canisters. Accordingly, this overall precision assessment includes
both analytical variability and variability contributed by collection and
storage of the air samples in the canisters. The absolute value of one of
23
-------
.3 -
_
Q-
Q.
V
U
c
u
.2 -
.5 1 1.5 2
five. Concentration, ppmC
2.5
Figure 7a. Plot of the differences between repeat analyses for
28 ambient samples versus concentration.
c
-------
the 59 differences was much larger (0.94) than the absolute value of the
other differences (see Table 4) and was clearly an outlier. Table 4 gives
overall precision statistics for the remaining 58 sample pairs, after removing
the outlier pair. The differences ranged from -0.41 to +0.22 ppmC, with a
mean difference of -0.026 ppmC and standard deviation of 0.119 ppmC.
Table 4. OVERALL PRECISION
Number of duplicate sample pairs 58
Range of differences -0.41 to +0.22 ppmC
Mean difference -0.026 ppmC
Standard deviation of differences 0.119 ppmC
Range of per cent differences -67.8% to 47.8%
Mean per cent difference -3.1%
Standard deviation of per cent differences 17.4%
A plot of the absolute value of the differences (Figure 8a) shows very
little concentration dependence, and a plot of the absolute values of the per
cent differences (Figure 8b) shows some negative correlation. Thus, the
standard deviation of the differences (not per cent differences) is probably
the best way to describe overall precision. However, Table 4 lists both the
difference and per cent difference statistics for comparison with Table 3.
Note that the overall precision (Table 4) is only slightly worse than the ana-
lytical precision (Table 3), indicating that collecting and storing the air
samples in the canisters added relatively little to the overall variability.
ACCURACY
Because the NMOC measurements encompass an unspecified mixture of various
organic compounds, absolute accuracy is undefined. Accuracy relative to
internal propane standards was assessed with audit samples. These audit
25
-------
.5
.1
u.
°- .3
u
(J
c
.2
.1 -
0.0
Vv.;*^
.5 1 1.5 2 2.5
flve. Concentration, ppmC
Figure 8a. Plot of the differences between duplicate samples
for 58 duplicate sample pairs versus concentration
V
u
-------
samples were prepared by diluting NBS-traceable propane standards with zero
air into clean canisters at a pressure similar to the pressure of the ambient
samples. Table 5 lists the regression slopes and intercepts for .the measured
concentrations versus the calculated propane audit concentrations for the
four analytical systems and for all systems combined. The combined data
suggest that the slope was slightly greater than 1.0000, indicating a slight
overall positive bias in the measurement system.
Table 5. ACCURACY RELATIVE TO PROPANE STANDARDS
Slope
Channel N^ (95% Cont. Interval)
A 21 1.02329 (±0.02474)
B 21 1.01689 (±0.01540)
C 15 1.02905 (±0.04208)
D 16 1.02331 (±0.05264)
Intercept
(95% Cont. Interval)
0.01461 (±0.02585)
0.02045 (±0.01610)
0.04442 (±0.07158)
0.03655 (±0.08711)
Correlation
Coefficient
0.9987
0.9995
0.9977
0.9960
All
Channels
Combined
73
1.02962 (±0.01539) 0.02096 (±0.0277)
0.9980
*Number of audit concentrations.
Accuracy relative to GC speciation analysis was assessed by comparing
the results from 336 samples that were independently reanalyzed by GC specia-
tion (sum of species) analysis by EPA's Atmospheric Sciences Research Labor-
atory. A linear, orthogonal regression of these data is illustrated in
Figure 9. The resulting slope of 1.081 and intercept of 0.015 indicated a
modest bias of approximately +8% for the PDFID method compared to GC specia-
tion analysis.
Accuracy relative to the fifth independent PDFID analysis system,
operated by the MSB, was assessed by comparing the results from 120 samples
that were reanalyzed by the independent MSB system. An orthogonal linear
regression resulted in a slope of 1.032 and an intercept of -0.1891. This
comparison is illustrated in Figure 10, and shows the generally good agreement
between the two independently operated PDFID analytical systems.
27
-------
LT>
cxi
'QUQd AE SISA1UNU DOWN
Figure 9. Comparison of PDFID Measurements to GC speciation
measurements for 336 ambient samples.
28
-------
_i i
tn
\x +
+ T1++ *
J I
LD
O_
CL
LJ
I—
in
in
a
LJ
LJ
D_
OJ LJ
CQ
LO
t—i
LO
>-
_J
a:
z
CE
n
r\j
3ujdd '
3NIinOd AB SISAIdNU
Figure 10. Comparison of NMOC measurements of 120 ambient samples by
2 independent PDFID analytical systems.
29
-------
REFERENCES
1. U. S. Environmental Protection Agency, "Uses, Limitations, and Technical
Basis of Procedures for Quantifying Relationships Between Photochemical
Oxidants and Precursors." EPA-450/2-77-021a (Nov. 1977).
2. U. S. Environmental Protection Agency, "Guidance for Collection of
Ambient Non-methane Organic Compound (NMOC) Data for Use in 1982 Ozone
SIP Development." EPA-450/4-80-011 (June 1980).
3. H. B. Singh, "Guidance for the collection and use of ambient hydrocarbons
species data in development of ozone control strategies." U. S.
Environmental Protection Agency, EPA-450/480-008 (April 1980).
4. F. F. McElroy, V. L. Thompson, "Hydrocarbon Measurement Discrepancies
Among Various Analyzers Using Flame-Ionization Detectors."
EPA-600/4-75-010 (Sept. 1975).
5. J. W. Harrison, M. L. Timmons, R. B. Denyszyn, C. F. Decker, "Evaluation
of the EPA Reference Method for the Measurement of Non-methane
Hydrocarbons." U. S. Environmental Protection Agency, EPA-600/4-77-033
(June 1977).
6. F. W. Sexton, R. M. Michie, F. F. McElroy, V. L. Thompson, "A Comparative
Evaluation of Seven Automated Ambient Non-methane Organic Compound
Analyzers." U. S. Environmental Protection Agency, EPA-600/54-82-046
(August 1982).
7. H. G. Richter, "Analysis of Organic Compound Data Gathered during 1980 in
Northeast Corridor Cities." U. S. Environmental Protection Agency,
EPA-450/4-83-017 (August 1983).
8. Coordinating Research Council, CY65 Report Project No. CM-4-58, New York,
NY (1966).
9. M. W. Jackson, "Analysis for Exhaust Gas Hydrocarbons--Nondispersive
Infrared Versus Flame lonization." JAPCA 11:697 (1966).
10. F. W. Secton, F. F. McElroy, "Technical Assistance Document for the
Calibration and Operation of Automated Ambient Non-methane Organic
Compound Analyzers." U. S. Environmental Protection Agency,
EPA-600/4-81-015 (March 1981).
11. R. D. Cox, M. A. McDevitt, K. W. Lee, G. K. Tannahill, "Determination
of Low Levels of Total Non-methane Hydrocarbon Content in Ambient Air."
Environ. Sci. Techno!. 16(1):57 (1982).
12. R. K. M. Jayanty, A. Blackard, F. F. McElroy, J. A. McBride, W. A.
McClenny, "Determination of Non-methane Organic Carbon (NMOC) in Ambient
Air by Cryogenic Preconcentration and Flame lonization Detection,"
presented at the 75th Annual Meeting of APCA, New Orleans, LA (June 1982)
30
-------
13.
14.
15.
16.
R. K. M. Jayanty, A. Blackard, F. F. McElroy, W. A. McClenny, "Laboratory
tvaluation of Non-methane Organic Carbon Determination in Ambient Air by
tryogenic Preconcentration and Flame lonization Detection." U. S.
tnvironmental Protection Agency, EPA-600/54-82-019 (July 1982).
Battelle Columbus Laboratories. "Literature Survey of Cryogenic Sampling
Techniques," Contract No. 68-02-3487.
M' r' Renter, F. F. McElroy, V. L. Thompson, "Measurement of Ambient
NMOC Concentrations in 22 Cities During 1984," presented at 78th Annual
Meeting of APCA, Detroit, MI (June 1985).
Radian Corporation, Final Project Report, "Nonmethane Organic Compounds
Monitoring Assistance for Certain States in EPA Regions III, IV, V, VI,
and VII. Phase II." EPA Contract No. 68-02-3513,
DCN No. 85-203-024-12-01. February, 1985.
31
-------
APPENDIX
CRYOGENIC PRECONCENTRATION AND DIRECT
FLAME IONIZATION DETECTION (PDFID) METHOD
FOR MEASUREMENT OF ATMOSPHERIC CONCENTRATIONS
OF NON-METHANE ORGANIC COMPOUNDS (NMOC)
August 1985
Quality Assurance Division
Environmental Monitoring Systems Laboratory
Office of Research and Development
U. S. ENVIRONMENTAL PROTECTION AGENCY
Research Triangle Park, North Carolina 27711
Preceding page blank 33
-------
CONTENTS
Page
INTRODUCTION 37
1. APPLICABILITY 40
2. PRINCIPLE 40
3. PRECISION AND ACCURACY 41
4. APPARATUS 42
4.1 Air Sampling 42
4.2 Sample Collection in Pressurized Canisters 42
4.3 Sample Canister Cleaning 43
4.4 Analytical System 44
4.5 Other Materials 46
5. SUPPLIES 46
5.1 Helium 46
5.2 Combustion Air 46
5.3 Hydrogen 46
5.4 Propane Calibration Standard 47
5.5 Zero Air 47
5.6 Cryogen 47
6. SYSTEM DESCRIPTION . ..... 47
6.1 Direct Sampling 47
6.2 Sample Collection in Pressurized Canisters 47
6.3 Analytical System 51
34
-------
CONTENTS (continued)
Page
7. PROCEDURE 58
7.1 Recommended Procedure for Canister Cleaning 58
7.2 Procedure for Collection of Samples in Canisters ... 59
7.3 Analysis Procedure 60
8. CALIBRATION 64
8.1 Calibration Frequency 64
8.2 Calibration Standards 64
8.3 Calibration Procedure 64
9. METHOD MODIFICATIONS 65
9.1 Sample Metering System 65
9.2 FID Detector System 65
9.3 Range 66
10. REFERENCES 57
FIGURES 58
APPENDIX 75
35
-------
FIGURES
Figure " Page
1 Schematic of Analysis System Showing Three Sampling
Modes . 68
2 Sample System for Collection of Integrated Field
Samples 69
3 Canister Cleaning System 70
4 Cryogenic Sample Trap Dimensions ... 71
5 Construction of Operational Baseline and Corresponding
Correction of Peak Area 72
6 Suggested filter-hypodermic needle assembly for
canister sampling system 73
7 Electrical pulse circuit for driving Skinner
Magnelatch solenoid valves with a conventional
mechanical timer ..... 74
Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
36
-------
INTRODUCTION
A variety of photochemical dispersion models have been developed to
describe the quantitative relationships between ambient concentrations of
precursor organic compounds and subsequent downwind concentrations of
ozone. An important application of such models is to determine the degree
of control of such organic compounds that is necessary in a particular area
to achieve compliance with applicable ambient air quality standards for
1 9
ozone.-"-j^- For this purpose, the models require measurements of atmospheric
concentrations of non-methane organic compounds (NMOC).
The more elaborate theoretical models generally require detailed
organic species data.^>3 5UC^ species data must be obtained by multi-
component gas chromatographic (GC) analysis of air samples.2>3 Simpler
empirical models such as the Empirical Kinetic Modeling Approach (EKMA)
require only total NMOC concentration data, specifically the average total
NMOC concentrations from 6 AM to 9 AM daily.2
For many EKMA applications, NMOC measurements are required at urban,
center-city-type sites.^ The NMOC concentrations typically found at such
urban sites may range up to 5-7 ppmC* or higher. If transport of precur-
sors into an area is to be considered, then NMOC measurements upwind of the
area are necessary.1 Upwind NMOC concentrations are likely to be very low,
(less than a few tenths of 1 ppm). Continuous commercially available NMOC
analyzers have been used to measure NMOC for EKMA applications, but the
*By convention, concentrations of NMOC are reported in units of parts per
million carbon (ppmC), which for a specific compound is the concentration
by volume (ppmV) multiplied by the number of carbon atoms in the compound.
37
-------
measurements have generally been only marginally adequate for urban sites
and unacceptable for upwind sites.4'5 NMOC GC species measurements can be
used by summing the various components to obtain a total NMOC concentra-
tion.2 But for EKMA, the species data are not needed, and the cost and
complexity of species analysis is very high.
The method described herein can be used to obtain both the requisite
urban, as well as upwind, NMOC measurements.6»7»8 This method is a simpli-
fication of the GC speciation method mentioned above. It combines the cryo-
genic concentration technique used in the GC method for high sensitivity
with the simple flame ionization detector (FID) for total NMOC measurements,
without the GC columns and complex procedures necessary for species separa-
tion. And because of the use of helium carrier gas, the FID has less res-
ponse variation to various organic compounds than a conventional NMOC
analyzer with air carrier or direct sample injection into the FID.4>8
This method can be used either for direct, in situ ambient measure-
ments or for analysis of integrated samples contained in metal canisters.
Making direct measurements at the monitoring site avoids the need for
collection of air samples in canisters. However, the analyst must be
present during the 6 AM to 9 AM period, and repeated measurements
(approximately six per hour) must be taken to obtain the 6 AM to 9 AM
average NMOC concentration. A separate analytical system and analyst is
needed for each monitoring site. (Further development of the method may
eventually allow for automatic operation, for on-line semi-continuous
analysis in the future.)
The use of sample canisters allows the collection of integrated air
samples over the 6 AM to 9 AM period by automated samplers at unattended
38
-------
monitoring sites. One analytical system can then be used to analyze the
samples from several monitoring sites. Degradation or contamination of the
air samples by the canister or sample collection system could be a potential
problem. However, tests indicate that the use of properly cleaned stainless
steel canisters, as described in the procedure, is practical and adds
relatively little additional variability to the method.8 Although storage
of the air samples for several weeks in the stainless steel canisters appears
to result in no appreciable degradation of the sample,8 good practice would
suggest that the samples be analyzed as soon after collection as practical.
39
-------
CRYOGENIC PRECONCENTRATION AND DIRECT FLAME IONIZATION DETECTION (PDFID)
METHOD FOR MEASUREMENT OF ATMOSPHERIC CONCENTRATIONS OF
NON-METHANE ORGANIC COMPOUNDS (NMOC)
1. APPLICABILITY
This method is applicable to measurement of concentrations of total
gaseous non-methane organic compounds (NMOC) in the atmosphere for use
with atmospheric photochemical models such as EKMA1 or for other appropriate
applications. Measurements may be obtained either in situ, or by subsequent
analysis of integrated air samples collected over a fixed time period, such
as the 3-hour (6 AM to 9 AM) measurements specified for EKMA. Collection
of integrated samples also allows for central analysis of samples from
multiple sites. The high sensitivity and low detection limit of the method
make it suitable for upwind measurements, while the wide dynamic range
allows analysis of urban air samples as well.
2. PRINCIPLE
An air sample is taken either directly from the ambient air at the moni-
toring site, where the analytical system is located, or from a special sample
canister filled previously at a remote sampling site. A fixed-volume portion
of the sample is drawn at a low flow rate through a glass beaded trap that is
cryogenically cooled to approximately -186° C (liquid argon temperature).
At this temperature, all organic compounds in the sample other than methane
are collected (either via condensation or adsorption) in the trap, while
methane, nitrogen, oxygen, etc., pass through. The system is dynamically
calibrated so that the volume of sample passing through the trap does not
have to be quantitatively measured, but must be precisely repeatable between
the calibration and analytical phases.
40
-------
After the fixed volume air sample has been drawn through the trap, a
helium carrier gas flow is diverted to pass through the trap, in the opposite
direction to the previous sample flow, and into a flame ionization detector
(FID). When the residual air and methane have been cleared from the trap
and the FID baseline becomes steady, the cryogen is removed and the tempera-
ture of the trap is raised to approximately 90° C. The organic compounds
previously collected in the trap revolatilize and are carried into the FID,
resulting in a response peak or peaks from the FID. The area of the peak
or peaks is integrated, and the integrated value is translated to concentra-
tion units via a previously obtained calibration curve relating integrated
peak areas with known concentrations of propane.
The cryogenically cooled trap simultaneously concentrates the non-
methane organic compounds while separating and removing the methane from
air samples. Thus the technique is direct reading for NMOC and, because of
the concentration step, is more sentitive than conventional NMOC analyzers.
Also, operation of the FID detector with a helium carrier results in less
response variation to different organic compounds'^ than is observed with
conventional NMOC analyzers having air carriers or direct air injection.4
Quantitative trapping has been shown for most compounds tested.6»8
3. PRECISION AND ACCURACY
The analytical precision, assessed in an actual field monitoring pro-
ject, was estimated to be 12.7%.8 The overall precision estimate for the
method, including the effect of collecting and storing the ambient samples
in stainless steel canisters, was found to be about 0.12 ppmC (approximately
17.4%).^ Because of the number and variety of organic compounds included in
41
-------
the NMOC measurement, determination of absolute accuracy is not practical.
Based on comparison with manual GC speciation analysis—a technique regarded
as the best available for the measurement of atmospheric organic compounds--
the proportional (per cent) bias was determined to be +8.1%, with a neglig-
ible fixed bias (intercept).8 Although the 8.1% bias was statistically
significant, no correction factor is proposed for the method because this
bias is modest, and the speciation technique is not an absolute standard.
Experimental tests indicate some degree of FID baseline shift from
water vapor in ambient samples, which could result in positive bias,
variability, or both. These problems can be adequately minimized by careful
selection of the integration termination point and appropriate baseline
corrections, as described in Section 7.3.
4. APPARATUS
The following components and materials are required or recommended.
Sources for the more specialized components are given in the Appendix. An
overall schematic diagram of the analytical system is shown in Figure 1,
and a suggested system for collecting ambient samples in canisters is shown
in Figure 2. A canister cleaning system is shown in Figure 3.
4.1 Air Sampling
4.1.1 Sample manifold or sample inlet line, to bring sample air into
the sampling or analytical system.
4.1.2 Vacuum pump or blower, if needed, to draw sample air through a
sample manifold or long inlet line to reduce inlet residence
time.
4.2 Sample Collection in Pressurized Canisters (See Figure 2)
4.2.1 Sample canisters. Stainless steel pressure vessels of 4 to 6 L
42
-------
volume, with one or two leak-free shut-off valves (see
Appendix). Interior surfaces of the canisters should be
passivated using the SUMMA process (Molectrics, Carson, CA).
Each canister should have a unique identification number.
4.2.2 Sample pump. Stainless steel, metal bellows type (Metal
Bellows model MB-151 or equivalent) capable of 2 atmospheres
(200 kPa, 30 psig) minimum output pressure. Pump must be free
of leaks, clean, and uncontaminated by oil or organic compounds.
4.2.3 Pressure gauge. 0-200 kPa (0 - 30 psig).
4.2.4 Shut-off valve, for gauge.
4.2.5 Stainless steel orifice or short capillary, capable of main-
taining a substantially constant flow over the sampling period
(see Section 6.2).
4.2.6 Particulate matter filter. (2 micron stainless steel
sintered in-line type)
4.2.7 Timer (for unattended sample collection). Capable of control-
ling pump(s) and solenoid valve (see Section 6.2).
4.2.8 Solenoid valve. Normally closed, bubbletight, electrically-
operated valve. A special bi-stable solenoid valve that requires
enerigizing only briefly for turn-on and turn-off (Skinner
Magnalatch or equivalent) is recommended to minimize tempera-
ture rise in the valve (see Section 6.2).
4.2.9 Needle valve. Optional fine metering valve may be needed
to adjust flow rate of sample from canister during analysis.
43
-------
4.3 Sample Canister Cleaning (See Figure 3)
4.3.1 Vacuum pump. Capable of evacuating the sample canisters to
an absolute pressure of <5 mm Hg.
4.3.2 Vacuum manifold. A metal manifold with connections for
several canisters to be simultaneously cleaned.
4.3.3 Shut-off valves (3), as shown.
4.3.4 Vacuum gauge. Capable of measuring the vacuum in the vacuum
manifold to an absolute pressure of 5 mm Hg or less.
4.3.5 Cryogenically cooled trap. U-shaped open tubular trap cooled
with liquid nitrogen or argon, to prevent contamination from
back diffusion of oil from the vacuum pump.
4.3.6 Pressure gauge. 0-50 psig (0-345 kPa), to monitor canister
pressure.
4.3.7 Flow control valve, to regulate flow of zero air into canisters.
4.4 Analytical System (See Figure 1)
4.4.1 FID detector system, including flow controls for the FID fuel
and air, temperature control for the FID, and signal processing
electronics.
4.4.2 Chart recorder, compatible with the FID output signal, to
record FID response signals for visual interpretation.
4.4.3 Integrator, electronic, compatible with the FID output signal
and capable of integrating the area of one or more FID response
peaks and calculating peak area corrected for baseline drift
(see Section 6.3.10).
NOTE: Items 4.4.1, 4.4.2, and 4.4.3 are conveniently provided
by a current model laboratory chromatograph (such as the
Hewlett-Packard model 5880 or similar). See also Sec-
44
-------
tions 6.3.7 and 6.3.10. A chromatograph may also provide
other convenient features such as an oven for warming
the trap and valve, automatic control of the valve and
integrator, pressure or flow regulators, etc. (see below).
4.4.4 Six-port chromatographic valve. Seiscor model VIII
(pneumatic), Valco 9110 (manual), or equivalent.
4.4.5 Trap (See Figure 4). Fabricated from 0.3175 cm (1/8") o.d.,
0.21 cm i.d. chromatographic grade stainless steel tubing
to the approximate dimensions shown. A 7 to 10 cm section
in the center of the trap is packed with 60/80 mesh glass beads,
held in place with dimethyldichlorosilane-treated glass wool
at both ends.
4.4.6 Cylinder pressure regulators. Standard, two-stage cylinder
pressure regulator, with pressure gauges, for helium, air
and hydrogen cylinders.
4.4.7 Low pressure regulators. Single stage, with pressure gauge,
if needed, to maintain constant helium carrier gas and
hydrogen flow rates (see Section 6.3.5).
4.4.8 Needle valve. Fine metering valve to adjust sample flow
rate through trap.
4.4.9 Cryogenic Dewar, to hold liquid cryogen sized to contain
submerged portion of trap.
4.4.10 Absolute pressure gauge. 0 - 400 mm Hg, Wallace & Tiernan
model 61C-ID-0410, or equivalent (see Section 6.3.1).
4.4.11 Vacuum reservoir. Vacuum tank of about 1 to 2 L capacity
(see Section 6.3.1).
4.4.12 Gas purifiers. Gas scrubbers containing Drierite or silica
45
-------
gel and 5A molecular sieve to remove moisture and organic
impurities in the helium carrier gas, air, and hydrogen.
4.4.13 Shut-off valves (2). Leak free, for vacuum valve and sample
valve.
4.4.14 Vacuum pump. General purpose laboratory pump capable of
evacuating the vacuum reservoir to an appropriate vacuum
that allows the desired sample volume to be drawn through
the trap.
4.4.15 Trap heating system. Chromatograph oven, hot water, or
other means to heat the trap to 80° to 90° C.
4.4.16 Vent, to keep the trap at atmospheric pressure during trapping
when using pressurized canisters, with means to detect and
verify positive vent flow, such as a rotameter or bubbler
(see Section 6.2).
4.5 Other Materials
4.5.1 Various connecting tubing and plumbing fittings. All such
items in contact with the sample, analyte, or support gases
prior to analysis should be stainless steel or other inert
metal. Do not use plastic or Teflon tubing or fittings.
4.5.2 Various mechanical mounting fixtures, as necessary.
5. SUPPLIES
5.1 Helium. Cylinder of high purity grade helium.
5.2 Combustion air. Cylinder of air containing less than
0.5 ppm hydrocarbons, or equivalent air source.
5.3 Hydrogen. Cylinder of ultra high purity grade hydrogen, or
equivalent hydrogen source.
46
-------
5.4 Propane calibration standard. Cylinder containing 1 to 100 ppm
(3 to 300 ppmC) propane in air. The cylinder assay should be
traceable to a National Bureau of Standards (NBS) propane in air
Standard Reference Material (SRM) or to a commercially available
Certified Reference Material (CRM).
5.5 Zero air, containing less than 0.01 ppmC hydrocarbons. Zero air may
be obtained from a cylinder of zero-grade compressed air scrubbed with
Drierite or silica gel and 5A molecular sieve or activated charcoal,
or by catalytic cleanup of ambient air. All zero air should be passed
through a cryogenic cold trap for final cleanup.
5.6 Cryogen. Liquid argon or liquid oxygen. (Observe appropriate
safety precautions with liquid oxygen.)
6. SYSTEM DESCRIPTION
6.1 Direct Sampling
For direct ambient sampling, the cryogenic trapping system draws the
air sample directly from a pump-ventilated distribution manifold or
sample line. The connecting line should be of small diameter
(1/8" o.d.) stainless steel and as short as possible to minimize its
dead volume. With direct sampling, multiple analyses will have to be
taken over the sampling period to establish hourly or 3-hour NMOC
concentration averages.
6.2 Sample Collection in Pressurized Canisters
Collection of ambient air samples in pressurized canisters provides a
number of advantages, including (1) convenient integration of ambient
47
-------
samples over a specific time period, e.g., 1 or 3 hours; (2) remote
sampling and central analysis; (3) storage and shipping of samples, if
necessary; (4) unattended sample collection; (5) analysis of samples
from multiple sites with one analytical system; and (6) collection of
replicate samples for assessment of measurement precision. However,
great care must be exercised in selecting, cleaning, and handling the
sample canisters and sampling apparatus to avoid losses or contamination
of the samples.
Figure 2 shows a schematic diagram of a recommended sample col-
lection system. The small auxiliary vacuum pump purges the inlet
manifold or lines with a flow of several liters/minute to minimize the
sample residence time. The larger, metal bellows pump takes a small
portion of this sample to fill and pressurize the sample canister.
Both pumps should be shock-mounted to minimize vibration.
A critical orifice or hypodermic needle connected to the inlet
of the metal bellows pump is used to maintain a substantially constant
flow into the canister(s) over the sample period and must be selected to
provide the desired flow rate. This flow rate is chosen so that the
canisters are pressurized to at least one atmosphere above ambient
pressure (2 atmospheres absolute pressure) over the desired sample
period. The flow rate can be calculated by
F = P V N
T x 60 (1)
where
48
-------
F = Flow rate, cm3/min,
P = Final canister pressure, atmospheres absolute*,
V = Volume of the canister, cm3,
N = Number of canisters connected together for simultaneous
sample collection,
T = Sample period, hours.
For example, if one 6 L canister is to be filled to 2 atmospheres
absolute pressure (15 psig) in 3 hours,
F = 2 x 6000 x 1 = 67 cm3/min
3 x 60
A 30 gauge hypodermic needle 2.5 cm long provides a flow of approxi-
mately 65 cm3/min with the Metal Bellows Model MB-151 pump. Such a
needle will maintain approximately constant flow up to a canister
pressure of about 10 psig, after which the flow drops with increasing
pressure. At 15 psig (2 atmospheres absolute pressure), the flow is
about 10% below the original flow.
The hypodermic needle is protected with a 2.0 ym stainless steel
in-line particulate filter, which also keeps particulate matter from
depositing in the pump, lines, and canister. A suggested filter-
hypodermic needle assembly can be fabricated as shown in Figure 6.
For automatic operation, the timer is wired to start and stop the
pump or pumps at the appropriate times for the intended sample period.
*Absolute pressure in atmospheres = Pg/Pa + 1» where Pg = gauge pressure in
the canister, psig, and Pa = standard atmospheric pressure (14.7 psi).
49
-------
The timer must also control the solenoid valve. The Skinner
Magnelatch solenoid valve specified avoids the substantial tempera-
ture rise that would occur with a conventional normally closed sole-
noid valve, which would have to be energized during the entire sample
period. This temperature rise in the valve could cause outgasing of
organics from the Viton valve seat material. The Magnelatch valve,
however, requires only brief electrical pulses to open and close at
the appropriate start and stop times and therefore experiences no
temperature increase. The pulses may be obtained with an electronic
timer that can be programmed for short (5 to 60 seconds) ON periods
or with a conventional mechanical timer and a pulse circuit such as
the one shown in Figure 7.
The canisters are originally evacuated. The connecting lines be-
tween the sample pump and the canister(s) should be as short as pos-
sible to minimize their volume. Check to see that the flow rate into
the canister remains relatively constant over the entire sampling
period. (As previously noted, some drop in the flow rate may occur
near the end of the sample period as the canister pressure approaches
two atmospheres absolute pressure.)
Simultaneous collection of duplicate samples decreases the possi-
bility of lost measurement data from samples lost due to leakage or
contamination in either of the canisters. Two (or more) canisters can
be filled simultaneously by connecting them in parallel (see Figure 2)
and selecting an appropriate flow rate to accommodate the number of
canisters (Equation 1). Duplicate (or replicate) samples also allow
assessment of measurement precision based on the differences between
50
-------
duplicate samples (or the standard deviation among replicate samples).
Prior to field use, each sampling system should be tested for pump
contamination (see Section 7.2), leaks, and proper flow rate. The
plumbing on the outlet side of the metal bellows pump can be checked
for leaks by shutting off the canister valves, pressurizing the system,
and checking fittings, etc. with a nonhydrocarbon-based leak detector
fluid. The metal bellows pump should also be leak-checked by plugging
its outlet and ensuring that there is no flow into its inlet side.
The canisters must be cleaned and checked for contamination before use
(see Section 7.1).
During analysis, a pressurized canister containing an air sample
is connected to the six-port valve with a vent, as shown in Figure 1.
The canister valve or an optional flow control valve installed between
the canister and the vent is used to reduce the canister pressure and
adjust the canister flow rate to a value slightly higher than the trap
flow rate set by sample metering valve. The excess flow exhausts
through the vent, which assures that the sample air flowing through
the trap is at atmospheric pressure. The vent is connected to a flow
indicator such as a rotameter or is submerged in water so that the
escaping bubbles provide a visual indication of vent flow to assist in
adjusting flow control valve.
6.3 Analytical System
6.3.1 Sample volume metering system. The vacuum reservoir and
pressure gauge (see Figure 1) are used to meter precisely
repeatable volmes of sample air through the cryogenically
51
-------
cooled trap. With the sample valve closed and the vacuum valve
open, the reservoir is first evacuated with the vacuum pump to
a predetermined pressure (e.g. 80 mm Hg). Then the vacuum valve
is closed and the sample valve is opened to allow sample air to
be drawn through the cryogenic trap and into the evacuated reser-
voir until a second predetermined reservoir pressure is reached
(e.g. 180 mm Hg). The (fixed) volume of air thus sampled is de-
termined by the pressure rise in the vacuum reservoir (difference
between the predetermined pressures) as measured by the absolute
pressure gauge. This volume can be calculated by
„ _ AP Vr
Vs - Ps (2)
where
o
Vs = Volume of air sampled, standard cm0,
AP = Pressure difference measured by gauge, mm Hg,
Vr = Volume of vacuum reservoir,
Ps = Standard pressure (760 mm Hg).
For example, with a vacuum reservoir of 1700 cm3 and a pressure
change of 100 mm Hg (80 to 180 mm Hg), the volume sampled would
be 225 cm3.
The sensitivity of the method is proportional to the sample
volume. However, sample volumes over about 500 cm3 may lead to
loss of sample flow during trapping due to clogging of the trap
from ice. Sample volumes below about 100 - 150 cm3 may cause
increased measurement variability due to dead volume in lines
and valves. For most typical ambient NMOC concentrations,
sample volumes in the range of 200 - 500 cm3 appear to be
52
-------
appropriate. If a response peak obtained with a 500 cm3 sample
turns out to be off-scale or to exceed the calibration range, a
second analysis can be carried out with a smaller volume. The
actual sample volume used need not be accurately known if it is
precisely repeatable during both calibration and analysis.
Similarly, the actual volume of the vacuum reservoir need
not be accurately known. But the reservoir volume should be
matched to the pressure range and resolution of the absolute
pressure gauge so that the measurement of the pressure change
in the reservoir—and hence the sample volume—is repeatable
within 1%. A 1700 cm3 vacuum reservoir and pressure change of
20 to 200 mm Hg, measured with the specified pressure gauge,
has proven adequate. A smaller volume reservoir may be used
with a greater pressure change to accommodate absolute pressure
gauges with lower resolution, and vice versa.
6.3.2 Trap. The trap should be carefully constructed from a single
piece of tubing in the shape shown in Figure 4. The central
portion of the trap (7 to 10 cm) is packed with 60/80 mesh glass
beads with small glass wool plugs to retain the beads. The
trap must fit conveniently into the Dewar flask (Section 4.4.9),
and the arms must be of an appropriate length to allow the
beaded portion of the trap to be submerged below the level of
liquid cryogen in the Dewar. The trap should connect directly
to the six-port valve, if possible, to minimize line length
between the trap and the FID. It must be mounted to allow the
Dewar to be conveniently slipped on and off the trap and also
to facilitate heating of the trap (see Section 6.3.4).
53
-------
6.3.3 Liquid cryogen. Either liquid oxygen (bp -183.0° C) or liquid
argon (bp -185.7° C) may be used as the cryogen; experiments
have shown no difference in trapping efficiency between the two
cryogenic liquids.6 However, appropriate safety precautions
must be taken if liquid oxygen is used. Liquid nitrogen
(bp -195° C) should not be used as it causes condensation of
oxygen and methane in the trap. It may be possible to use liquid
nitrogen in an automated system if an automatic temperature con-
troller is used to obtain an operational temperature in the range
of -180° to -185° C. The level of the cryogenic liquid should be
maintained constant with respect to the trap (see Section 7.3.11)
and should completely cover the beaded portion of the trap.
6.3.4 Heat source. To facilitate integration of the NMOC response
peak, a hot bath or other heating source is used to heat the
trap and volatilize the concentrated NMOC such that the FID
produces one (or only a few) sharp and easily integrated peak
(or peaks). The trap should be heated to a temperature in the
range of 80° to 90° C. A simple heating source for the trap is
a beaker or Dewar filled with water maintained at 80° to 90° C.
Other types of heat sources include a temperature-programmed
chromatograph oven, electrical heating of the trap itself, or
any type heater that brings the temperature of the trap up to
80° to 90° C in 1 to 2 minutes. A uniform trap temperature rise
rate (above 0° C) may help to reduce variability and facilitate
more accurate correction for the moisture-shifted baseline
(see Section 6.3.10). If a programmable chromatograph oven is
used to heat the trap, the following parameters have been found
54
-------
to be acceptable: initial temperature, 30° C; initial time,
0.20 minutes (following start of the integrator); heat rate,
30°/min.; final temperature, 90° C.
6.3.5 Carrier gas. Helium is used to purge residual air and methane
from the trap at the conclusion of the sampling phase and to
carry the revolatilized NMOC from the trap into the FID. A
single-stage auxiliary regulator between the cylinder and the
analyzer may not be necessary but is recommended to regulate
helium pressure better than the two-stage cylinder regulator.
When an auxiliary regulator is used, the secondary stage of the
two-stage regulator must be set at a pressure higher than the
pressure setting of the single-stage regulator.
6.3.6 Construction. The six-port valve and as much of the inter-
connecting tubing as practical should be located inside an oven
or otherwise heated to 80° to 90° C to minimize wall losses or
adsorption/desorption in the connecting lines. All lines
should be kept as short as practical.
All tubing used for the system should be chromatographic
grade stainless steel connected with stainless steel fittings.
Pneumatic damping may be needed between the six-port valve and
the FID to dampen the effect of valve actuations, which may
otherwise cause upsets in the FID signal or extinguish the
flame. A stainless steel capillary may be used for damping,
but its length should be as short as possible to prevent
broadening of the peak.
After assembly, the system should be pressurized to about
80 psig (550 kPa) and checked for leaks. During this procedure,
55
-------
disconnect the absolute pressure gauge and cap the line to
prevent damage to the gauge. If the system is leak free,
depressurize the system and reconnect the gauge.
6.3.7 FID Detector. The FID burner air, hydrogen, and carrier helium
flow rates should be set according to the manufacturer's
instructions to obtain an adequate FID response while main-
taining as stable a flame as possible throughout all phases of
the analytical cycle. Typical flow rates are as follows:
hydrogen, 30 cm3/min; carrier (He), 30 cm3/min; burner air,
400 cm^/min.
6.3.8 Linearity. Response has been shown to be linear over a
wide range (0 to 10,000 ppb C).6
6.3.9 Range. Some FID detector systems such as those associated with
laboratory chromatographs may have autoranging. Others may
provide a "range" (attenuator) control and internal full-scale
output voltage selectors. An appropriate combination should be
chosen so that an adequate output level for accurate integration
is obtained down to the detection limit, yet the electrometer or
integrator must not be driven into saturation at the upper end
of the calibration. Saturation of the electrometer may be
indicated by flattening of the calibration curve at high concen-
trations. Additional adjustment of range and sensitivity can
be provided by adjusting the sample volume used, as discussed
in Section 6.3.1.
56
-------
6.3.10 Integrator. The integrator must be electrically compatible
with the output signal of the FID detector so that sufficient
resolution is available at low concentrations without over-
ranging on high concentrations. If both an integrator and a
separate chart recorder are used, care must be exercised to
be sure that these components do not interfere with each other
electrically. Range selector controls on both the integrator
and the FID analyzer may not provide accurate range ratios,
so individual calibration curves should be prepared for each
range to be used.
The integrator should be capable of marking the beginning
and ending of peaks, constructing the appropriate baseline
between the start and end of the integration period, and
calculating the peak area accordingly (see Figure 5). This
capability is necessary because the moisture in the sample,
which is also concentrated in the trap, will cause a slight
positive baseline shift. This baseline shift starts as the
trap warms and continues until all of the moisture is swept
from the trap, at which time the baseline returns to its normal
level. The shift generally continues longer than the ambient
organic peaks. If possible, the integrator should be programmed
to correct for this shifted baseline. Alternatively, analyses
of humidified zero air should be used as blanks to correct the
ambient air concentration measurements accordingly.
57
-------
7. PROCEDURE
7.1 Recommended Procedure for Canister Cleaning
7.1.1 Leak-test all canisters by pressurizing them to about 40 psig
(275 kPa) with zero air and immersing them in water. Defective
canisters should be returned to the manufacturer for repair.
7.1.2 Connect canisters to the vacuum manifold as shown in Figure 3.
7.1.3 Open the vacuum shut-off valve and evacuate the canisters to
5.0 mm Hg or less for one hour or more, using a cryogenically-
cooled trap in the vacuum line to eliminate back diffusion of
hydrocarbons and oil from the vacuum pump.
7.1.4 Close the vacuum and vacuum gauge shut-off valves. Open the zero
air valve to pressurize the canisters with zero air to about 35
psig (240 kPa). If a zero gas generation system is used, the
rate of flow may need to be limited to maintain the zero air
quality.
7.1.5 Close the zero air valve and allow the canisters to vent down to
atmospheric pressure through the vent valve. Then close the
vent valve.
7.1.6 Repeat steps 7.1.3 to 7.1.5 two additional times.
7.1.7 Fill the canisters with zero air and analyze the contents as
a blank check of the canisters and of the cleanup system and
procedure. This step should be performed on 100% of the
canisters until the cleanup system and procedure are proven
to be reliable. The check can then be reduced to a lower
percentage unless problems arise. Any canister that does not
test clean (compared to direct analysis of zero air) after
repeated cleaning should not be used.
58
-------
7.1.8 Re-evacuate the canister after the analysis and leave it
evacuated until used.
7.1.9 Attach a paper tag to each canister for field notes. The
canister is now ready for collection of an air sample.
7-2 Procedure for Collection of Samples in Canisters
7.2.1 Clean and test the canisters according to the procedure in
Section 7.1.
7.2.2 Assemble a sample collection system such as the one shown in
Figure 2.
7.2.3 Check the pump for contamination by filling two evacuated,
cleaned canisters with zero air through the sampling system
and analyzing them.
7.2.4 Check the flow control orifice on each sampling system to
make sure the sample flow remains relatively constant up to
about 15 psig (2 atmospheres absolute pressure).
7.2.5 Install the pump at the site. If the inlet line is long
(over about 3 meters), use an auxiliary pump as shown in
Figure 2 to ventilate the line.
7.2.6 Verify that the timer, pump(s), and solenoid valve are
connected and operate properly.
7.2.7 Verify that the timer is correctly set for the desired sample
period, and that the solenoid valve is closed. Connect
the evacuated canister(s) to the solenoid valve.
7.2.8 Open the canister valve. A small rotameter temporarily connect-
ed to the sample inlet can be used to verify that there is no
flow. (Flow detection would indicate a leaking solenoid valve.)
59
-------
7.2.9 After the sample period, close the canister valve, disconnect
the canister from the sampling system and connect a pressure
gauge to the canister. Briefly open and close the canister
valve, and note the canister pressure. If the canister pressure
is not approximately 2 atmospheres absolute (15 psig), determine
and correct the cause of the low or high sample pressure before
the next sample.
7.2.10 Fill out the identification tag on the sample canisters as
necessary. Take the canisters to the analytical system for
analysis.
7.2.11 Complete records of the sampling should be entered in a labora-
tory notebook. The sampling operator should be alerted to take
note of any activities or special conditions in the area (rain,
smoke, etc.) that may affect the sample contents.
7.3 Analysis Procedure
7.3.1 Assemble the analytical system as shown in Figure 1 and as
discussed in Section 6. Allow the FID detector to warm up
and stabilize for several hours before analysis.
7.3.2 Check and adjust the helium carrier pressure to provide the
correct carrier flow rate for the system (see Section 6.3.7).
Also check FID hydrogen and burner air flow rates.
7.3.3 Close the sample valve and open the vacuum valve to evacuate
the vacuum reservoir.
7.3.4 With the trap at room temperature, place the six-port valve
in the inject position.
60
-------
7.3.5 Open the sample valve and adjust the sample metering valve
for an approximate sample flow of 50 - 100 cm3/min. (The flow
will be lower later, when the trap is cold.)
7.3.6 Connect a sample canister or direct sample inlet to the six-
port valve as shown in Figure 1. For a canister, open the
canister valve and adjust the canister valve and/or the
sample metering valve to obtain a moderate vent flow as
indicated by the flow indicator or by constant bubbles.
Then close the sample valve.
CAUTION: Do not allow water to be drawn into the six-port
valve. The sample flow will be lower (and hence the
vent flow wilT be higher) when the trap is cold.
7.3.7 Open the vacuum valve (if not already open) to evacuate the
vacuum reservoir. With the six-port valve in the inject
position and the vacuum valve open, open the sample valve
for a few minutes to flush and condition the inlet lines.
7.3.8 Close the vacuum valve and allow the reservoir pressure to
rise to the predetermined sample starting pressure on the
absolute pressure gauge (see Section 6.3.1). Then quickly
close the sample valve at the starting pressure reading.
7.3.9 Switch the six-port valve to the sample position.
7.3.10 Submerge the trap in the cryogen and allow a few minutes for
the trap to cool completely (indicated when the cryogen stops
boiling). Check and adjust the initial cryogen level to the
same level used during calibration (Section 8).
61
-------
7.3.11 Open the sample valve and observe the increasing pressure on
the pressure gauge. When it reaches the pressure representative
of the desired sample volume (see Section 6.3.1), close the
sample valve.
7.3.12 Add a little cryogen or elevate the Dewar to raise the liquid
level to a point slightly (1 to 5 mm) higher than the initial
level at the beginning of the trapping (see Section 7.3.8).
Then switch the 6-port valve to the inject position, keeping the
cryogenic liquid on the trap. Also close the canister valve to
conserve the remaining sample in the canister.
7.3.13 Start the chart recorder and wait until the FID response
baseline has stabilized (about 20 to 60 seconds). Do not wait
longer than one minute.
7.3.14 Start the integrator. Remove the liquid cryogenic bath from
the trap and smoothly but not too quickly replace it with a
Dewar of hot water (approximately 80° to 90° C) or, if the trap
is in an oven, start heating the oven. Use the same tempera-
ture and level of hot water or a consistent heating sequence
for both calibration and sample analyses. Heating the trap too
quickly may cause an initial negative-going response which
could hamper accurate integration. Some initial experimenta-
tion may be necessary to determine the optional heating proce-
dure for each system, but once established, the procedure
should be consistent for each analysis.
62
-------
7.3.15 Continue the integration only long enough to include all of
the organic compound peaks and to establish the end point
FID baseline, as shown in Figure 5 (probably 1 to 2 minutes,
depending on rate of trap heating). The end point baseline
will be shifted somewhat higher than the initial baseline due
to moisture in the sample. Construct an operational baseline
from the initial baseline at the beginning of the first peak to
the end point baseline as shown in Figure 5, and correct the
peak area reading according to this operational baseline.
Electronic integrators either do this automatically or they
should be programmed to do this correction.
NOTE: Be sure that the 6-port valve remains in the inject
position until all moisture has purged from the trap (3 minutes
or longer).
7.3.16 Use the calibration curve (Section 8.3) to convert the inte-
grated peak area reading into concentration units (ppmC). Note
that the NMOC peak shape may not be precisely reproducible due
to variations in heating the trap, but the total NMOC peak area
should be reproducible.
7.3.17 Duplicate Analysis - Analyze each canister sample at least two
times and report the average NMOC concentration. Problems
occasionally occur during an analysis that will cause improper
or inconsistent results. If the first two analyses do not agree
closely, additional analyses should be made to identify in-
accurate measurements and produce a more accurate average.
63
-------
8. CALIBRATION
8.1 Calibration Frequency
Initially, a complete dynamic calibration at five or more concen-
trations should be carried out on each range to define the calibration
curve. Subsequently, the calibration should be verified with two- or
three-point calibration checks (including zero) each time the analytical
system is used to analyze samples.
8.2 Calibration Standards
Propane calibration standards may be obtained directly from low con-
centration cylinder standards or by dilution of high concentration
cylinder standards with zero air. Dilution flow rates must be measured
accurately, and the combined gas stream must be mixed thoroughly.
Calibration standards should be sampled directly from a vented manifold
or tee. Remember that a propane NMOC concentration in ppmC is three
times the volumetric concentration in ppm.
8.3 Calibration Procedure
8.3.1 Select one or more combinations of FID attenuator setting, out-
put voltage setting, integrator resolution (if applicable), and
sample volume to provide the desired range or ranges (e.g.,
0 to 1.0 ppmC or 0 to 5.0 ppmC). Each such range should be
calibrated individually and have a separate calibration curve.
(Modern GC integrators may provide automatic ranging such that
several decades of concentration may be covered in a single
range.)
8.3.2 Analyze each calibration standard three times according to the
procedure in Section 7.3. Be sure that flow rates, pressure
gauge start and stop readings, initial cryogen liquid level in
64
-------
the Dewar, timing, heating and other variables are the same as
will be used during analysis of ambient samples.
8.3.3 Average the three analyses for each concentration standard and
plot the calibration curve(s) as integrated peak area reading
versus concentration in ppmC. The relative standard deviation
for the three analyses should be less than 3% (except for zero
concentration). Linearity should be expected; points that
appear to deviate abnormally should be repeated. If non-
linearity is observed, an effort should be made to identify and
correct the problem. If the problem cannot be corrected,
additional points in the non-linear region may be needed to
adequately define the calibration curve.
9. METHOD MODIFICATIONS
9.1 Sample Metering System
Although the vacuum reservoir and absolute pressure gauge technique
for metering the sample volume during analysis is efficient and
convenient, other techniques should work also. For example, a constant
sample flow could be established with a vacuum pump and a critical
orifice, with the six-port valve being switched to the sample position
for a measured time period. Or a gas volume meter such as a wet test
meter could be used to measure the total volume of sample air drawn
through the trap. However, these alternate techniques have not been
tested or evaluated.
9.2 FID Detector System
FID detector systems other than the Hewlett-Packard Model 5840A may
65
-------
also be adaptable to the method. The specific flow rates and necessary
modifications for the helium carrier for any alternate FID instrument
would have to be worked out by the user.
9.3 Range
It may be possible to increase the sensitivity of the method by
increasing the sample volume. However, limitations are likely to
arise, such as plugging of the trap by ice; hence, any attempt to
increase the sensitivity should be tested carefully.
66
-------
10. REFERENCES
1. U. S. Environmental Protection Agency, "Uses, Limitations, and
Technical Basis of Procedures for Quantifying Relationships
Between Photochemical Oxidants and Precursors." EPA-450/2-77-021a
(Nov. 1977).
2. U. S. Environmental Protection Agency, "Guidance for Collection
of Ambient Non-methane organic Compound (NMOC) Data for Use in
1982 Ozone SIP Development." EPA-450-/4-80-011 (June 1980).
3. H. B. Singh, "Guidance for the collection and use of ambient
hydrocarbons species data in development of ozone control
strategies." U. S. Environmental Protection Agency,
EPA-450/480-008 (April 1980).
•
4. F. W. Sexton, R. M. Michie, F. F. McElroy, V. L. Thompson,
"A Comparative Evaluation of Seven Automated Ambient Non-methane
Organic Compound Analyzers." U. S. Environmental Protection Agency,
EPA-600/54-82-046 (August 1982).
5. H. G. Richter, "Analysis of Organic Compound Data Gathered During
1980 in Northeast Corridor Cities." U. S. Environmental Protection
Agency, EPA-450/4-83-017 (August 1983).
6. R. K. M. Jayanty, A. Blackard, F. F. McElroy, W. A. McClenny,
"Laboratory Evaluation of Non-methane Organic Carbon Determination
in Ambient Air by Cryogenic Preconcentration and Flame lonization
Detection." U. S. Environmental Protection Agency,
EPA-600/54-82-019 (July 1982).
7. R. D. Cox, M. A. McDevitt, K. W. Lee, G. K. Tannahill, "Determin-
ation of Low Levels of Total Non-methane Hydrocarbon Content in
Ambient Air." Eniron. Sci. Technol. 16(1):57 (1982).
8. F. F. McElroy, V. L. Thompson, H. G. Richter, "A Cryogenic
Preconcentration - Direct FID (PDFID) Method for Measurement
of NMOC in the Ambient Air." U. S. Environmental Protection
Agency, EPA-600/ (August 1985).
67
-------
ABSOLUTE
PRESSURE GAUGE
VACUUM
VALVE
SAMPLE
VALVE
LOW
PRESSURE
REGULATOR
VACUUM
PUMP
VACUUM
RESERVOIR
SAMPLE
METERING
VALVE
He
CANISTER
VALVE
PRESSURIZED
CANISTER
SAMPLING
-a-'
VENT
GLASS
BEADS
CRYOGENIC
SAMPLE TRAP
(LIQUID ARGON)
Figure 1.
Schematic diagram of analysis system showing two
samp!ing modes.
68
-------
SAMPLE
IN
PRESSURE
GAUGE
VACUUM
PUMP
METAL
BELLOWS
PUMP
CANISTER(S)
Figure 2. Sample system for automatic collection of 3-hour
integrated field air samples.
69
-------
VACUUM
PUMP
ZERO AIR
SUPPLY
V
CRYOGENIC
XTRAP
SHUT-OFF
VALVE
\
VENT
•VACUUM
GAUGE
SHUT-OFF
VALVE
PRESSURE
GAUGE
SHUT-OFF
VALVE
FLOW
CONTROL
VALVE
VACUUM
MANIFOLD
SAMPLE CANISTERS
Figure 3. Canister cleaning system,
70
-------
TUBE LENGTH: ~30 cm
O.D.:0.32cm
I.D.: 0.21 cm
LIQUID LEVEL-
BO/BO MESH GLASS BEADS
Q
•GLASS WOOL
13 cm
(TO FIT DEWAR)
Figure 4. Cryogenic sample trap dimensions
71
-------
ID
UJ
tr.
o
NMOC
PEAK
START
INTEGRATION
END
INTEGRATION
WATER-SHIFTED
BASELINE
J
T
OPERATIONAL BASELINE
CONSTRUCTED BY INTEGRATOR
TO DETERMINE CORRECTED AREA
NORMAL BASELINE
TIME
Figure 5. Construction of operational baseline and
corresponding correction of peak area.
72
-------
•F' SERIES COMPACT, INLINE FILTER
W/2 ^m SS SINTERED ELEMENT
FEMALE CONNECTOR, 0.25 in O.D. TUBE TO
0.25 in FEMALE NPT
HEX NIPPLE, 0.25 in MALE NPT BOTH ENDS
30 GAUGE x 1.0 in LONG HYPODERMIC
NEEDLE (ORIFICE)
FEMALE CONNECTOR, 0.25 in O.D. TUBE TO
0.25 in FEMALE NPT
THERMOGREEN LBI 6 mm (0.25 in)
SEPTUM (LOW BLEED)
0.25 in PORT CONNECTOR W/TWO 0.25 in NUTS
Figure 6. Suggested filter and hyopdermic needle assembly for
sample inlet flow control.
73
-------
TIMER
SWITCH
O_
115 VAC
PUMP
100 K
w
BLACK
40 Aifd, 450 V DC
100 K
RED
40 pfd, 450 V DC
WHITE
MAGNELATCH SOLENOID VALVE
Figure 7. Electrical pulse circuit for driving Skinner Magnelatch
solenoid valve with a conventional mechanical timer.
-------
APPENDIX
ADDITIONAL INFORMATION CONCERNING SPECIAL PDFID METHOD COMPONENTS
COMPONENT
1. Sample canister
Absolute pressure
gauge
3. Six-port valve
4. Gas purifiers
Chromatographic
grade stainless
steel tubing
Laboratory gas
chromatograph,
with FID, flow
controls, and
integrator/
recorder
Metal bellows
pump
8. Cryogenic Dewar
IDENTIFICATION
4-6 Liter
Model 61C-ID-0410
(0-410 mm Hg),
6" Face
Seiscor Model VIII
Cat. #8125
Cat. #30101
(1/8" x 0.085")
HP Model 5840A or
equivalent
Model MB-151
8600 (285 ml)
SUPPLIER
Demaray Scientific
Instruments, Ltd.
N. 1218C Grand Ave.
Pullman, WA 99163
(509) 332-3684
Wallace & Tiernam
Div. of Pennwalt Corp.
25 Main Street
Belleville, NJ 07109
Seismograph Service Corp.
Seiscor Division
P. 0. Box 1590
Tulsa, OK 74102
Alltech Associates
Deerfield, IL
Alltech Associates
Deerfield, IL
Hewlett-Packard Corp.
Avondale, PA 19311
9. Magnelatch V52dA1100/
solenoid valve CV5-LAJVF24
Metal Bellows Corp.
1075 Providence Highway
Sharon, MA 02067
Pope Scientific Inc.
Menomonee Falls, WI 53051
Skinner Valve
New Britain, CT
75
-------
*—* \^r'
^ o •£
0 fc 8 oJ
.ti
OS
VH ,"* Ctf
Ufe 5 U
VH
£ -5 a -
a '£ « a
IT! r^ >y ru
S £ o o
. . £ H rT
•
a a
U
C S
irj O
§£~
c/3 < .g g
a
•0 *Q
Reproduced by NTIS
National Technical Information Service
U.S. Department of Commerce
Springfield, VA 22161
This report was printed specifically for your
order from our collection of more than 2 million
technical reports.
For economy and efficiency, NTIS does not maintain stock of its vast
collection of technical reports. Rather, most documents are printed for
each order. Your copy is the best possible reproduction available from
our master archive. If you have any questions concerning this document
or any order you placed with NTIS, please call our Customer Services
Department at (703)487-4660.
Always think of NTIS when you want:
• Access to the technical, scientific, and engineering results generated
by the ongoing multibillion dollar R&D program of the U.S. Government.
• R&D results from Japan, West Germany, Great Britain, and some 20
other countries, most of it reported in English.
NTIS also operates two centers that can provide you with valuable
information:
• The Federal Computer Products Center - offers software and
datafiles produced by Federal agencies.
• The Center for the Utilization of Federal Technology - gives you
access to the best of Federal technologies and laboratory resources.
For more information about NTIS, send for our FREE NTIS Products
and Services Catalog which describes how you can access this U.S. and
foreign Government technology. Call (703)487-4650 or send this
sheet to NTIS, U.S. Department of Commerce, Springfield, VA 22161.
Ask for catalog, PR-827.
Name
Telephone
- Your Source to U.S. and Foreign Government
Research and Technology.
------- |