EPA-600/2-78-008
February 1978
Environmental Protection Technology Series
CHEMIUIM1NESCEMT MONITOR FOR
VINYL CHLORIDE
Environmental Sciences Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development. U S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL PROTECTION TECH-
NOLOGY series. This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-78-008
February 1978
CHEMILUMINESCENT-MONITOR FOR VINYL CHLORIDE
M. W. Greene, S. G. Riccio
W. D. Dencker, and R. I. Wilson
Beckman Instruments, Inc.
Advanced Technology Operations
Anaheim, California 92806
Contract No. 68-02-1770
Project Officer
Ralph Baumgardner
Atmospheric Chemistry and Physics Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences Research
Laboratory, U.S. Environmental Protection Agency, and approved for publica-
tion. Approval does not signify that the contents necessarily reflect the
views and policies of the U.S. Environmental Protection Agency, nor does
mention of trade names or commerical products constitute endorsement or
recommendation for use.
ii
-------
ABSTRACT
A monitor for vinyl chloride monomer (VCM) in ambient air was constructed
using commercially available components of a gas chromatograph (GC) coupled
with a chemiluminescence ozone analyzer slightly modified to make it suitable
for use as a GC detector. The specificity for VCM is enhanced by use of a
chemiluminescence detector because saturated hydrocarbons do not chemilumi-
nesce with ozone. A custom GC oven and monitor cabinet were used to make it
possible to transport the monitor in a standard station wagon for purposes
of field evaluation.
A custom absorbtion trap capable of concentrating the VCM from one liter
(or more) of ambient air was used to extend the lower detection limit. Using
a custom trap heating circuit, the concentrated sample can be thermally eluted
into the GC in a gas volume of 10 ml or less, providing a 100-fold increase
in sensitivity compared to that obtained with a 10 ml sample loop.
The results of testing at the component level are reported in detail.
These results imply that the ultimate sensitivity of the technique may be set
by the stability of the ozone concentration delivered to the chemiluminescence
detector. Specifically, ozone appears to decompose with photon emission at
a rate of about four percent per hour.
Preliminary system and final acceptance test results are also reported in
detail. These test results indicate that a monitor of this type can detect
less than one part per billion (ppb) of VCM in ambient air operating on a
fifteen-minute cycle. If the concentration trap is replaced by a 20 ml sample
loop, the limit of detection is estimated to be about 35 ppb VCM when oper-
ating on a five-minute cycle.
iii
-------
CONTENTS
Abstract iii
Figures vi
Tables vii
1.0 Introduction 1
1.1 Purpose of Contract 1
1.2 Scope of Contract 1
1.3 General Approach and Background Information... 1
2.0 Summary, Conclusions, and Recommendations 4
2.1 Summary 4
2.2 Conclusions 4
2.3 Recommendations 4
3.0 Testing of Components of FCM Monitor 6
3.1 Trap Temperature Profile Data 6
3.2 Trap Capacity as a Function of Ambient Temperature..7
3.3 Trap Elution Test 7
3.4 Column Elution Times for Several Components 8
3.5 Ozonator Tests .8
3.6 Reaction Chamber Design 12
4.0 System Testing 13
4.1 Linearity of Peak Height with Sampled Volume 13
4.1.1 Comparison of Nitrogen and Helium as Carrier Gas.19
4.1.2 Comparative Chromatograms for Air and Oxygen
for the Ozonator Supply 22
4.2 Linearity of VCM Peak Height with VCM Concentration24
4.3 System Tests Using Sample Loops in Place of the
Trap 24
4.4 The Effects of Varying Ozonator and Carrier Flow
Rates 30
5.0 Final Acceptance Test 32
5.1 Initial Functional Tests 32
5.2 Tests Conducted with Carbowax Columns 36
5.3 Tests Performed in Final Configuration. 40
-------
FIGURES
_Number
1 Design of VCM Trap. ...
2 Chromatograms of Six-'Component Mixture in Nitrogen,
Identifying Peaks where Possible . , 14
3 Chromatogram of Six-Component Mixture in Nitrogen for
(A) One-Half Minute Sample Collection Period at 200
m£/min and (B) One-Minute Sample Collection Period
at 200 mA/min 15
4 Chromatogram of Six-Component Mixture in Nitrogen for
(A) Two-Minute Sample Collection Time Period at 200
m£/min, and (B) Three-Minute Sample Collection Time
Period at 200 mJl/min 16
5 Chromatograms of Six-Component Mixture in Nitrogen for
(A) Four-Minute Sample Collection Period at 200 m£/min
and (B) Five-Minute Sample Collection Period at
200 m£/min n
6 Output for 1-ppm VCM vs Sample Collection Time 18
7 Linearity of all Peak Heights vs Sample Collection Time . . 20
8 Comparison of VCM Monitor Chromatograms when Using
Nitrogen and Helium as the Carrier Gas 21
9 Comparison of Chromatograms Obtained when Using Air
instead of Oxygen for the Oxonator Supply 23
10 Chromatogram of Six-Component Mixture in Nitrogen Obtained
Using a 10.7 mi (Total) Sample Loop in Place of the Trap. 26
11 Chromatogram of Six-Component Mixture in Nitrogen Obtained
Using a 32-mJl Sample Loop in Place of the Trap 27
12 Chromatogram of Six-Component Mixture in Nitrogen Obtained
Using a, 32-m£ Sample Loop in Place of the Trap, and
Using 120 mJl/min of Nitrogen Carrier Gas instead of
75 m£/min .28
13 Chromatogram of Six-Component Mixture in Nitrogen Obtained
with 18.5 mi (Total) Sample Loop in Place of the Trap,
and with the Model 950 Sensitivity and Response Speed
Changes Indicated , 29
14 The Effect of Varying Carrier Gas Flow Rate upon the Peak
Height Observed for 5-ppm VCM in Nitrogen, with Ozonator
Flow Rate as a Parameter .11
15 Three of Many Chromatograms Obtained at EPA Facility Using
a Permeation Tube Source for 0.75-ppm VCM Concentration . 33
16 Two of Many Chromatograms Obtained at EPA Facility Using a
Permeation Tube Source for 0.75-ppm VCM Concentration .,. 34
17 Chromatogram Showing Linearity of the Monitor Below
1 ppm VCM 35
18 Chromatograms Indicating Presence of Interfering
Pollutants 37
19 Chromatogram Indicating the Presence of Interfering
Compounds in Cigarette Smoke. 38
va.
-------
FIGURES
20 Two Chromatograms Obtained after Modifying Operating
Conditions to Eliminate Interference of Cigarette
Smoke and of Unknown Atmospheric Pollutants with VCM
Determination 39
21 Relevant Portions of Two Consecutive Chromatograms
Obtained Using 0.25-ppm VCM in Nitrogen from Permeation
Tube, Collected for Five Minutes at 150-m£/minute. ... 41
22 Two Successive Chromatograms Obtained for Five-Minute
Collections of 0.125-ppm VCM in Nitrogen from Permea-
tion Tube Source at Flow Rate of 150 mJl/minute 42
TABLES
Number Page
1 Temperature vs Time Profile of Empty Trap 6
2 Trap. Capacity for VCM vs Trap Temperature While Absorbing. 8
3 Elution Times of Several Hydrocarbons for Two Stripper
and Two Analysis Columns Provided by the Beckman
Application Engineering Department . 9
4 The Effect of Using Oxygen Instead of Air as the
Ozonator Supply. 22
5 Ratio of Peak Heights for Chromatogram Obtained Using
10.7-m£ Sample Loop (Figure 10) to those Obtained for
a 400-m£ Sample Concentrated in the Trap (Figure 2). 25
vii
-------
1.0 INTRODUCTION
1.1 PURPOSE OF CONTRACT
The basic purpose of this contract was to combine existing instrumentation and
known techniques to produce a vinyl chloride monomer (VCM) monitor of very
high sensitivity and specificity, packaged in a form suitable for transport in
a standard station wagon to facilitate field evaluation of the monitor for the
measurement of VCM in ambient air. EPA personnel had previously demonstrated
the basic sensitivity and specificity for VCM of a gas chromatographic sep-
aration technique combined with a chemiluminescence detector. Absorption of
VCM on a charcoal bed as a means of sample collection and concentration had
also been previously demonstrated.
1.2 SCOPE OF CONTRACT
The contract required interfacing and packaging commercially available process
gas chromatograph (PGC) components with a commercially available chemilumines-
cence analyzer as the detector. The sensitivity of this analysis approach was
to be extended by using a "trap" to extract VCM (and other hydrocarbons) from
a relatively large volume of air sample. A previously developed trap with a
temperature control capable of causing a very rapid thermal step change for
rapid elution of the collected sample was selected. With this trap the de-
tection limit was extended to five parts per billion (ppb) of VCM, with an
analysis cycle time of less than 30 minutes. Off-the-shelf equipment was used
to the maximum extent permitted by size constraints. Special components were
fabricated with minimal documentation to ensure that the bulk of the funds
expended would be available for initial tests, optimization, and further
testing.
1.3 GENERAL APPROACH AND BACKGROUND INFORMATION
Gas chromatograph (GC) techniques are well known and require no detailed
description here. A Beckman Instruments, Inc., Model 6700 Process Gas Chro-
matograph Programmer was used to provide the basic programmable timing and
control functions. Two standard Beckman PGC 10-port valves were used for
valving a sample trap, a stripper column, and an analysis column into various
independent and/or series gas flow circuits upon commands from the programmer.
The stripper and an analysis column were designed and fabricated by the
Applications Engineering Department of the Process Instruments Division of
Beckman Instruments, Inc. In final acceptance testing, an analysis column
previously tested by the EPA was substituted for the Beckman analysis column.
-------
A Beckman Model 950 Ozone Analyzer (chemiluminescence type) was modified for
monitoring vinyl chloride. The major modifications were:
• Addition of a standard Beckman ozonator, consisting of a quartz .
envelope mercury vapor lamp, properly housed,, and a ballast trans-
former power supply.
• Modification of the internal plumbing to provide for selection of
either air (supplied by a built-in pump) or oxygen flow through the
ozonator, and to permit use of one of the two built-in flowmeters
•for monitoring the ozonator supply flow rate, and use of the other
flowmeter for monitoring the total flow being discharged from the
chemiluminescence reaction chamber.
• The addition of an absorber to remove trace contaminants from the
ozonator supply, and of an ozone absorber in the gas effluent line
to prevent the discharge of ozone to the atmosphere.
• Rewiring of the Model 950 power switch to make it control the ozona-
tor power supply only.
• Modification of the Model 950 reaction chamber to make it about three
times the standard depth, which increased the sensitivity to vinyl
chloride about three-fold.
In the modified instrument the carrier gas effluent from the GC section is
blended with an ozone-rich air (oxygen) stream in the Model 950 reaction cham-
ber. VCM, as well as other components which chemiluminesce when they react
with ozone, is detected quantitatively by a photomultiplier tube positioned
adjacent to the reaction chamber.
The major advantage of a chemiluminescence detector for this application lies
in the fact that saturated hydrocarbons, which might not be satisfactorily
separated by the GC portion of the monitor, do not produce a chemiluminescence
signal. The specificity of the monitor is, therefore, greatly improved by the
use of this detector instead of one of the ordinary GC detectors of comparable
sensitivity, such as the flame ionization detector, which responds to all
hydrocarbons.
A custom GC oven was purchased in order to meet the contractual size con-
straints for the monitor. The GC valves and columns were mounted within the
custom oven. A hot-air-bath-type temperature control was employed. A
standard Beckman circuit board (with minimal modification) was used to control
the oven temperature.
The VCM. trap and trap thermal control circuitry were adapted from a previous
contract performed for the EPA by Beckman (Contract No. 68-02-0778). Hand-
wired breadboard circuitry was employed, as on the prior contract, to avoid
unnecessary cost for an evaluation prototype.
-------
All custom components were mounted in one compartment of the monitor console,
which is referred to as the control module in this report. These included:
• The trap and the trap temperature control circuit board.
• The GC oven and its temperature control circuitry.
• All gas pressure gages, flow control valves, and flow indicators
(except flow indicators and controls included in the Model 950
chemiluminescence monitor).
• Power distribution terminal blocks and fuses for each component of
the monitor.
The trap can be easily removed and replaced by an ordinary GC "sample loop" if
desired. The maximum volume of the sample loop is determined by the ability
of the GC columns to handle such hydrocarbons as ethylene and propylene, which
normally elute before VCM. If the concentrations of these hydrocarbons are
too high and the sample loop volume is too large (approaching 20 m£), the
columns are saturated by the other constituents and the VCM peak is obscured
by tailing of the prior peaks. However, since the trap employed has a lower
capacity for ethylene and propylene, these hydrocarbons may saturate the trap
and break through during the trap loading cycle without affecting the trap
retention of the VCM. When the trap is valved into series flow with the GC
columns and heated to elute the hydrocarbons, the limited ethylene and propy-
lene capacity of the trap prevents overloading of the column (up to much higher
concentrations). The relatively high VCM capacity of the trap permits con-
centration of the VCM from more than one liter of sample. The net effect
permits concentration of the VCM by a factor of at least 50 without producing
overlapping peaks due to comparable concentrations of ethylene, propylene,
etc. which occur when larger volume sample loops are employed.
The use of an ordinary sample loop (5- or 10-mJl volume) instead of the trap
may be beneficial if relatively high (20 ppm or more) VCM concentrations are
to be monitored.
-------
2.0 SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS
2.1 SUMMARY
This essentially hardware-oriented project was conducted to provide a highly
specific and sensitive monitor for vinyl chloride monomer (VCM) in ambient
air. Standard, commercially available instrumentation was combined with a
minimum quantity of custom designs to provide a monitor suitable for transpor-
tation in a standard station wagon to facilitate field evaluation.
The monitor developed uses gas chromatographic (GC) equipment and techniques
to isolate VCM from related and potentially interferring compounds, and a \
chemiluminescence analyzer as the GC detector to further increase the speci-
ficity of the system. A charcoal (or equivalent) trap is used to concentrat/e
the VCM contained in up to one liter of sample into a volume of a few milli-
liters suitable for injection (by thermal elution) into the GC carrier gas
stream.
The results of tests performed on the major components before integration into
the monitor are reported, with emphasis on the theoretical and practical impli-
cations of the test results. The tests performed on the monitor prior to
delivery to the EPA, and the results obtained, are also reported in detail.
The results of Final Acceptance Testing at the EPA facility are also con-
tained in this report. In the final test configuration the VCM analysis was
free from interferences from pollutants in the prevailing ambient air. An
analysis cycle of less than 15 minutes would detect less than .2 parts per
billion (ppb) of VCM. The long-term stability was such that when operated
with a full-scale operating range of 0-100 ppb VCM or more the monitor should
not require calibration more often than once per week.
2.2 CONCLUSIONS
,,It may be concluded from the results of the work performed on this contract
that a VCM monitor capable of reliably detecting less than 5 ppb VCM in
ambient air with less than a 15-minute operating cycle can be produced.
Commercially available major components plus some custom components which
are not available off the shelf are required. Monitors of this basic design
,' will fit in a station wagon for convenience in monitoring the ambient air in
selected areas for VCM concentration.
Most of the VCM-related compounds which may be expected to be present when VCM
is detected do not interfere with the VCM analysis. It is more difficult to
discriminate against several unknown atmospheric pollutants and components of
cigarette smoke than it is to separate VCM from ethylene, propylene, 1-butene,
and the dichloroethylenes.
-------
2.3 RECOMMENDATIONS
It is recommended that the VCM monitor be subjected to additional laboratory
interference testing. In particular, identification of those unknown atmos-
pheric pollutants and cigarette smoke components which were eluted from the GC
column at very nearly the same time as the VCM would be of value for future
field evaluation.
Field testing of the monitor in localities where traces of VCM are expected to
exist in the ambient air is of major interest. The results of all tests per-
formed on this contract indicate that the field tests should be highly
successful.
Sample collection trap absorbers other than charcoal should be evaluated. In
particular, it is known that the trap provided with the monitor has very
little capacity for ethylene, which can be of advantage when large concen-
trations of ethylene are present because overloading of the GC columns is
prevented by the small ethylene capacity. However, this is a disadvantage
when and if it is desired to monitor ethylene too. Similarly, the propylene
peaks are not linear with sampled volume, which might be due to marginal trap
capacity, making quantitative analysis for propylene difficult. (Nonlinearit}
of the chemiluminescence detector itself for propylene has been reported else-
where, making it difficult to estimate the charcoal trap efficiency for propy-
lene.) Logically, other trap materials will have complementary advantages,
broadening the applicability of the technique.
Finally, other applications of the instrumental techniques used in the VCM
monitor should be considered. All compounds convertable to NO, H2S, or mer-
captans, for example, can be monitored by this technique. The specificity,
sensitivity, and stability of this analysis technique make it very attractive.
5
-------
3.1
3.0 TESTING OF COMPONENTS OF VCM MONITOR
TRAP TEMPERATURE PROFILE DATA
The details of the trap design are shown in FIGURE 1. A special trap tempera-
ture controller, capable of rapidly ramping the trap temperature, was employed.
in all tests. A test was performed on an empty trap, using a small thermo-
couple inserted at various distances into the trap to determine the time re-
quired to reach 2509C after initiation of the heat cycle. The trap temperature
controller was adjusted to provide a steady-state temperature of about 250°C at
a central location in the trap. The sequence was: locate thermocouple; apply
heat and record temperature vs time; turn power off; relocate the thermocouple
for the next test as the trap cooled back to ambient temperature. The data,
tabulated in TABLE 1, are in units of distance from the end of the trap tube
farthest from the temperature controlling thermistor, and in units of time
required to reach 250°C after application of power.
While the temperature profile was not completely uniform, subsequent measure-
ment of the elution peak time indicated that elution occurred with a half-
bandwidth of about six seconds.
TABLE 1. TEMPERATURE VS TIME PROFILE OF EMPTY TRAP
Thermocouple Location Relative
to One End of Trap Tube
(Centimeters)
0.64
1.27
1.91
2.54
3.17
3.81
4.44
(Inches)
0.25
0.50
0.75
1.00
1.25
1.50
1.75
(Adjacent thermistor)
Time after Application
of Power Required to
Reach 250°C
(Seconds)
12
24
36
48
60
60
48
Remarks
— Start trap fill
at 1 cm
Position of
Coconut Shell
Charcoal in
Typical Filled
Trap
—End trap fill
at 3.8 cm
-------
LENGTH OF HEATER WIRE
-.3 CM (1/8 IN)
STAINLESS STEEL
TUBE
COCONUT SHELL CHARCOAL PACKING
3.81 CM (1-1/2 IN) LONG
Figure 1. Design of VCM Trap
3.2
TRAP CAPACITY AS A FUNCTION OF AMBIENT TEMPERATURE
A trap charged with coconut shell charcoal was connected in series with the
inlet to a flame ionization detector. A gas mixture of 2.6 ppm VCM in nitro-
gen was passed through the trap at 150 m£/min. Between tests at various trap
temperatures (simulating various ambient temperatures), the trap was heated
and stripped free of hydrocarbons by a flow of nitrogen. After cooling to the
new ambient temperature, the trap was reconnected to the 2.6-ppm VCM sample at
a flow rate of 150 m£/min. The flame ionization analyzer output, which was
recorded continuously, served to indicate the time to breakthrough of VCM.
The results of these tests are given in TABLE 2.
The trap temperature during the collection was not precisely known because the
air-bath oven could not be closed during the test. The tabulated maximum con-
centration factors for VCM are computed with a 10-m£ sample loop as a reference.
This is a reasonable size since tests indicated that an 18.5-m£ sample loop
gave reasonably satisfactory results, but that a 32-mJl sample loop made the
system performance too dependent on ethylene and propylene concentrations. It
may be concluded, therefore, that a concentration factor of at least 50, up to
a trap temperature of 55°C during sample collection, can be obtained.
3.3
TRAP ELUTION TEST
A sample gas containing 2.6 ppm VCM in nitrogen was passed through the trap
at a flow rate of 150 m£/min for six minutes and 45 seconds, giving a total
sample of 1.01 liters. The trap was then connected in series with a standard
flame ionization detector (FID) with a nitrogen carrier gas flowing. After
the FID output stabilized, the trap temperature controller was energized
(set for 250°C). The recording of the FID output indicated that the VCM peak
reached the FID two seconds after the trap heat was applied. The peak re-
sponse occurred at 12 seconds, and the response was below 10% of the peak
-------
TABLE 2. TRAP CAPACITY FOR VCM VS TRAP TEMPERATURE WHILE ABSORBING
Approximate Trap
Temperature during
Absorption Cycle
(° Centigrade)
.30
40
50
55
Approximate Time
to Breakthrough of
2.6 ppm VCM in
Nitrogen at
150 m£/min
(Minutes)
12
9
9
7
Total Sample .
Volume Before
Breakthrough
(Milliliters)
1800
1350
1350
1050
Approximate
Maximum
Concentration
Factor for VCM
180
135 .
135
105
value in 18 seconds. The VCM peak was reasonably symmetrical, with a half-
bandwidth of about six seconds.
3.4
COLUMN ELUTION TIMES FOR SEVERAL COMPONENTS
Stripper and analysis columns were obtained from the Beckman Applications
Engineering Department. The stripper column was 1.6 m long, standard stain-
less-steel, GC grade, 1/8" OD tubing. It consisted of an upstream section
1.3 m long packed with 30% dioctyl adidate on Chromosorb P, 45/60 mesh, and a
downstream section 0.3 m long containing Poropak (200-2) Q, 50/80 mesh. The
analysis column was a 1.0 m long, standard stainless-steel, GC grade, 1/8"
OD tubing containing Poropak (200-2) R, 50/80 mesh.
Two columns of each type were obtained. They were preconditioned by heating
to 120°C for one hour with a flow of 40 m£/min of nitrogen. They were then
controlled at 95°C for subsequent testing.
Each column was tested using the trap (with thermal ramp elution) for sample
injection, and the standard FID analyzer as the detector. The results of
these tests are given in TABLE 3.
3.5
OZONATOR TESTS
The ozone generator (ozonator) employed is of Beckman design, using a Western
Quartz, Inc., mercury vapor, quartz envelope lamp. A stream of air (or
oxygen) flows through a Teflon tube surrounding the lamp, and then to the
reaction chamber through small-bore Teflon tubing. Part of the oxygen, in
passing through the lamp housing, is converted to ozone by action of the
ultraviolet (UV) light emitted by the lamp. Tests performed under this con-
tract indicated that the following conditions prevail:
-------
TABLE 3. EWTION TIMES OF SEVERAL HYDROCARBONS FOR TWO STRIPPER AND TWO
ANALYSIS COLUMNS PROVIDED BY THE BECKMAN APPLICATION ENGINEERING
DEPARTMENT
Column Identification
Analysis Column — No. 1
No. 2
Stripper Column — No. 1
No. 2
Elution Times (at Peak) for Several Hydrocarbons
in Minutes after Applying Power to Trap
VCM
1.5
1.5
1.4
1.2
*1 , 2-dichloroethylene
cis
5.5
3.7
3.2
*1 , 2-dichloroethylene
trans
8.5
5.5
4.8
*Identification of these compounds was later found to be doubtful. These data
are consistent with the vendor tag on the cylinder received, but this was not
what was ordered. . Several other mixtures were tested later, but it was not
possible to prove whether the cylinder tag, the purchase order, or neither
was correct.
The ozone concentration of the effluent gas is independent of gas
flow rate up to about 160 m£/min, above which the concentration of
ozone decreases.
The maximum achievable ozone concentration is approximately propor-
tional to the square root of the oxygen concentration of the gas
used.
•When stored in a glass spectrophotometer cell with rock salt windows,
the ozone decomposes at a rate of roughly 4%/hour.
The tests which resulted in these observations were neither repeated nor pur-
sued further. Any conclusions drawn are, therefore, merely speculative
pending further investigation. A thorough literature search, which was beyond
the scope of this effort, might provide a complete understanding of the ozone
formation and decomposition mechanisms. However, the implications of these
observations are sufficiently interesting to warrant further mention in this
report.
These observed results may be predicted from a simple theory which assumes
reasonable mechanisms for the ozone formation and decomposition reactions. In
particular, the constancy of ozone concentration with flow rates below about
160 mJl/min suggests that a steady state (equality of formation and decompo-
sition ra.tes) is being attained in less time than the gas spends in the
-------
bzonator. The rate of formation of ozone will be proportional to the inci-
dent ionizing radiant energy density and to the concentration of oxygen if
the rate-governing step involves interaction of photons with isolated oxygen
molecules. The rate ol: decomposition will be proportional to the square of
the ozone concentration if decomposition involves the simultaneous collision
of two ozone molecules (either in space or at a wall). Under steady-state
conditions the two rates must be equal and, if these assumed mechanisms are
correct, it follows that the concentration of ozone will be proportional to
the square roots of the oxygen concentration and of the ionizing radiant
energy density. A single test (discussed further below) of the effect of
lamp current on concentration indicated roughly a 40% increase in ozone for
a factor of two increase in lamp current, which is not conclusive, but is
supportive evidence.
No evidence regarding decomposition of ozone by molecular collisions at the
walls was found. This is a possible, but considered improbable, reaction
mechanism. If a simultaneous collision of two ozone molecules with the wall
is involved, a higher concentration of ozone should result if the inside diam-
eter of the Beckman ozonator is increased, since this would increase volume
more rapidly than surface area. Increasing the length of the ozonator should
have no effect, since the volume and surface both increase linearly with the
length of a cylinder. If collisions of two ozone molecules in space are in-
volved in the decomposition reaction, then altering the reaction chamber
dimensions will have no effect on the maximum ozone concentration. The slow
rate of decomposition of ozone in a glass cell, discussed further below,
suggests that neither the walls nor merely the simultaneous collision of two
ozone molecules is involved in the rapid decomposition rate occurring within
the ozonator. Another factor is probably involved, as explained below.
In order to account for a rapid ,rate of decomposition in the ozonator com-
pared to that outside the ozonator (discussed further below), it is suffi-
cient to assume that the rate of decomposition is proportional to the square
of the ozone concentration and to the lamp intensity. It would then be
necessary that the ozone formation rate be proportional to Pc>2 and to the
square of the lamp intensity in order to account for the observed proportion-
ality between equilibrium ozone concentration and the square root of both PQ2
and the lamp current.
Measurements of the IR absorbance at 1020 and 1052 cm~l wave numbers indicate
that up to about 1% ozone can be obtained from the Beckman ozonator when using
oxygen and the larger of two ballast transformers. .This estimate is based .on
published values for the specific absorbance of ozone at these wave numbers.
Absorbance in the UV region was not measured because a suitable spectropho-
tometer was not available at the time. A measurement of the apparent ozone
concentration due to the decrease in paramagnetic susceptibility of air passed
through the ozonator indicated that about 0.32% ozone (maximum) was obtained
using air (up to 80 mJl/min or more) and a smaller ballast transformer, which
fixed the lamp current at about one-half that of the larger transformer used
to obtain about 1% ozone with oxygen. Using the square root of both the lamp
current and of the oxygen concentration to calculate the ozone concentration
to be expected from the larger ballast transformer, and for 100% oxygen instead
10
-------
of air, one obtains 0.32% ozone \/(2)(100/21) = 0.99% ozone, which is an amount
in good agreement with the spectroscopic data noted above. Using oxygen and
the smaller ballast transformer (which is used in the monitor), one would ex-.
pect about 0.7% ozone maximum. A similar test, using several similar ozona-
tors, was made subsequently at Beckman on a different project. These IR
absorption tests indicated that about 0.7% ozone was the maximum attainable,
rather than 1%, with the larger transformer. In the configuration delivered
it may be concluded that between 0.5 and 0.7% ozone can be obtained using an
oxygen supply, and that between 0.2 and 0.3% ozone can be obtained using air.
The spontaneous decomposition rate of ozone was roughly determined by noting
that the IR absorbance decreased to about 1/3 of the initial value in 17 hours,
indicating a decomposition rate of about 4% of the ozone per hour when stored
in a glass cell with rock salt windows. It is also possible to obtain an
estimate of the spontaneous ozone decomposition rate in the reaction cell of
the Model 950. The zero offset of the Model 950 with about 0.7% ozone, but
with no VCM or other known reactant present, is known. This output is equiva-
lent to that which would be observed if a sample containing about 0.05 ppm
ozone were to be blended with an excess of ethylene in the reaction cell at
a flow rate of 13.3 mSL/s (which is the normal operating mode of the Model 950
as an ozone monitor). It is known that 50 to 100% of the ozone reacts in the
cell under these conditions and that the sample residence time in the cell is
about one second. It may be inferred that an equivalent amount of ozone spon-
taneously decomposes in the cell each second when 0.7% ozone is present. It
is necessary to assume that the spontaneous decomposition of ozone also re-
sults in photon emission. It is then possible to estimate the spontaneous
decomposition rate to be about (0.5 x 10~6)/(7 x 10~3) per second. This
corresponds to 7 x 10~^% per second, or 2.6% per hour. This agrees (within
about a factor of two) with the spontaneous decomposition rate observed in
the IR absorption cell, discussed above.
One implication of these data is that the observed zero offset of the VCM moni-
tor is due to an ozone decomposition reaction accompanied by chemiluminesence.
This reaction must be slow compared to the decomposition rate which occurs
within the Beckman ozonator. Further implications are that the decomposition
reaction outside the ozonator is probably first order rather than second order
with the ozone concentration, and that the high rate of decomposition of ozone
in the ozonator is due, therefore, to the interaction of two ozone molecules
with a UV photon, rather than to merely the collision of two ozone molecules.
The UV photons causing decomposition would have a flux density proportional to
that of those causing ozone formation for given lamp operating conditions. The
spectral energy distribution of a lamp of this type would be expected to remain
constant over a fairly wide range of lamp current, for which the lamp voltage
drop is constant. Over this current range the energy output is, therefore,
linear with lamp current. While it is probably the case, it was not shown that
.the Beckman ozonator lamp operates in this current range for both transformers
tested (30 and 60 mA nominal currents).
11
-------
3.6 REACTION CHAMBER DESIGN
The standard reaction chamber of the Beckman Models 950, 951, and 952 chemi-
luminescence analyzers is only approximately 2.5 mm (0.1 inch) deep. This is
an adequate depth for the standard analyzers, but it did not appear to be ade-
quate for the vinyl chloride/ozone reaction. It was found that the same net
signal was obtained for flow rates of about 16 to 160 m£/min each of ozonated
air and 2.6 ppm vinyl chloride in nitrogen. The technique of making corres-
ponding changes to retain a constant ratio of the two flow rates provides
fixed ozone and VCM concentrations. The gases were passed directly into the
Model 950 reaction chamber in this series of experiments. The 2.6-ppm VCM in
nitrogen sample was replaced by nitrogen for each flow setting to provide a
zero VCM signal. The major observations resulting from these experiments were:
• The sensitivity to vinyl chloride is dependent on the product of
the concentrations of vinyl chloride and ozone only.
• The maximum signal occurs when the flow rates of sample and ozonated
air are equal, which maximizes the product of VCM and ozone
concentrations.
• The zero offset is essentially constant for equal flow rates of VCM
and ozone-rich gas from 16 to 160 m£/min, indicating that the offset
is dependent on the ozone concentration only. (Note that the ozone
concentration in the stream from the ozonator was constant for flow
rates below about 160 mJl/min.
• The above results indicate that only a small percentage of the vinyl
chloride reacts in the chamber, probably less than 10% even at the
low flow rate of 16 m^/min of each gas.
The chamber depth was subsequently increased by about a factor of three to
increase the percentage of VCM which reacted in the chamber.
12
-------
4.0 SYSTEM TESTING
4.1 LINEARITY OF PEAK HEIGHT WITH SAMPLED VOLUME
The Model 6700 Process Gas Chromatograph (PGC) Programmer provides for con-
venient selection of sample collection time. A measure of the efficiency of
the trap for VCM collection is provided by maintaining a fixed sample flow
rate while varying the time of collection. A plot of VCM peak height versus
sample collection time should be a straight line passing through the origin,
provided the flow rate and collection times employed are within the range of
100% trap efficiency. A test was performed using a 200 m£./min flow rate (room
conditions) of nitrogen containing 1 ppm each of ethylene; propylene; 1-butene;
VCM; 1,1-dichloroethylene; and trans 1,2-dichloroethylene. The sample was
collected in the trap for 0.5, 1, 2, 3, 4, and 5 minutes. The chromatogram
for a two-minute collection time using a recorder chart speed of one inch per
minute is shown in FIGURE 2. FIGURES 3, 4, and 5 are chromatograms for the
0.5-, 1-, 2-, 3-, 4-, and 5-minute collection times, using a chart speed of
0.1 inch per minute. The operating conditions are fully described on FIGURE 2
for convenience of reference. The data on the other figures stress the delib-
erate changes made in operating conditions.
It should be noted that peak number 5 of these figures is believed to be an
impurity in the sample, and that (a) either one of the two dichloroethylenes
does not chemiluminesce, or (b) they elute together, or (c) the wrong material
was employed in making the sample. Manual injection of additional 1-butene
during the trap-loading cycle caused an increase in peak number 4 without
altering peak number 5 significantly. Similar tests indicated that peak 5
was not due to the dichloroethylenes, and that the two dichloroethylenes were
possibly eluting simultaneously, but data from two other mixtures merely in-
creased the confusion. One mixture called for 1,1- and 1,2-dichloroethylene
on the purchase order, but its label indicated that cis and trans 1,2-
dichloroethylene were used. This mixture gave two distinct peaks during com-
ponent testing when using the FID, but only one peak in the system tests with
the chemiluminescence detector. It is probable that one of the samples in
that case was a dichloroethane, which does not chemiluminesce. It is quite
possible that a similar error occurred with the other mixtures. It is con-
cluded that the results for the dichloroethylenes were essentially
inconclusive.
The peak heights for 1 ppm of VCM, taken from FIGURES 3,4, and 5, are plotted
as a function of collection time in FIGURE 6. It is apparent that zero peak
height corresponds to a very small collection time (about 0.1 min), rather
than zero time. This was caused by the particular valving and interconnection
arrangement employed. Specifically, at the end of a sample collection cycle a
solenoid valve positioned upstream of the trap closes and remains closed until
13
-------
OPERATING CONDITIONS
RECORDER!
MODEL SHSO SETTINGS i
GC OVEN TEMPERATURE:
FLOW RATESi
SAMPLE COMPOSITION!
CHART SPEED OF ONE INCH/MIN (NOTE THAT
3 MINUTES WERE REMOVED)
100 mV FULL-SCALE SENSITIVITY
2.5 PPM RANGE AND 0.1-VOLT OUTPUT
ZERO POT AT 7.00 TURNS
SPAN POT AT 7.00 TURNS
RESPONSE SWITCH IN "FAST" POSITION
90°C
SAMPLE AT 200 mJl/MIN, COLLECTED FOR
TWO MINUTES
CARRIER FLOW RATE AT 75 mVMIN (ROOM
CONDITIONS)
OZONATOR FLOW RATE AT 200 mJl/MIN
(ROOM CONDITIONS)
NITROGEN, CONTAINING ONE PPM EACH OF
THE FOLLOWING (PEAKS ARE IDENTIFIED
WHERE POSSIBLE BY CORRESPONDING
NUMBERS)
(1) ETHYLENE
(2) PROPYLENE
(3) 1-BUTENE
(4) VINYLCHLORIDE MONOMER
(5) UNIDENTIFIED—PROBABLY IMPURITY
(6) 1,1-DICHLOROETHYLENE AND/OR trans
1,2-OICHLOROETHYLENE, EACH
PRESENT AT 1 PPM CONCENTRATION
Til
... i
-i-i-M-
11.
.: ——-TIME
Figure 2. Chromatograms of Six-Component Mixture in Nitrogen, Identifying
Peaks where Possible
14
-------
OTHER CONDITIONS
SAME AS FIGURE 2, EXCEPT RECORDER^]
CHART SPEED IS 0.1 INCH/MIN i
B
-TIME
Figure 3.
Chromatogram of Six-Component Mixture in Nitro*
gen for
(A) One-half-Minute Sample Collection Period
at 200 ml/min
(B) One-Minute Sample Collection Period at
200 mH/min
15
-------
-TIME
B
OTHER CONDITIONS
SAME AS FOR FIGURES 3 AND 5
Figure 4. Chromatogram of Six-Component Mixture in Nitrogen for
(A) Two-Minute Sample Collection Time Period at 200 mi/min, and
, (B) Three-Minute Sample Collection Time Period at 200 mSL/min
16
-------
OTHER CONDITIONS
SAME AS FOR FIGURES 3
-TIME
Figure 5. Chromatograms of Six-Component Mixture in Nitrogen for
(A) Four-Minute Sample Collection Period at 200 mH/min, and
(B) Five-Minute Sample Collection Period at 200 ml/min
17
-------
l-PPM VINYLCHLORIDE OUTPUT vs
TRAP COLLECTION TIME. SAMPLE
WAS 6-COMPONENT MIXTURE, AT
200 M&/MIN.
2 3
MINUTES COLLECTION
Figure 6. Output for 1-ppm VCM vs Sample Collection Time •
-------
the beginning of the next collection cycle. The sample pump is connected to
draw sample through the trap, and operates at all times. Thus, the pump
evacuates the trap and interconnecting lines during the period of solenoid
valve closure. When the valve opens at the beginning of each sample
collection cycle, the sample rushes into the trap and interconnecting lines
with almost no flow resistance (up to the pump exhaust valve). The net
effect is that a volume of sample equal to that of the system between the
solenoid valve and the sample pump exhaust valve is added to the calculated
volume collected (calculated as flow rate multiplied by time for each col'-
lection time). From the data presented, this volume is about 15 m&, which is
a reasonable value. This defect was not corrected because it was not con-
sidered significant for purposes of this contract. However, proper use of a
three-way solenoid valve to eliminate this effect would be appropriate for
any further development effort.
FIGURE 7 is a plot of all peak heights from FIGURES 3, 4, and 5. The plot
for peak number 1, ethylene, clearly indicates that the trap has very little
capacity for ethylene. The plot for peak 5 (unknown impurity) indicates that
the trap reaches capacity for this component at about four minutes, corres-
ponding to 800 m& at room temperature and pressure (RTF). Peak 2, propylene,
is nonlinear with sample volume, but complete saturation of the trap is not
indicated for up to one liter of sample. The peak height curve for peak
number 6 (probably dichloroethylene of one or more types) is not a smooth
curve. In retrospect, it is quite possible that a part of this peak was being
lost by initiating stripper column back flush prematurely. This would make
the peak height very dependent on minor changes in carrier gas flow rate, GC
column temperature, etc.
A subsequent test with helium carrier at a 50% higher flow rate yielded a 52%
higher value for peak 6, without affecting the other peaks proportionately.
This suggests that back flush of the stripper was being initiated too soon
when the data of FIGURES 2 through 7 were obtained. The helium carrier
test results are discussed further below.
4.1.1 Comparison of Nitrogen and Helium as Carrier Gas
.After obtaining the data presented in FIGURES 2 through 7, a new reference
chromatogram using nitrogen as carrier was obtained. The major change was
that the oven temperature indicated 98°C instead of 90°C at the time of this
test. No reason for this change could be determined, but it is probable that
the oven cover was not properly closed during this test. Immediately follow-
ing this test the nitrogen carrier was replaced with helium and a second
chromatogram was obtained. A later calibration check of the carrier flow-
meter with helium indicated that the helium flow rate was 120 m£/min, com-
pared to the nitrogen rate of 75 m£/min. No other conditions were different.
The two comparative chromatograms are shown in FIGURE 8.
No large improvement in performance was obtained by using helium instead of
nitrogen, but the inadvertent use of a higher helium flow rate clouds the re-
sults somewhat. There was a 30% increase in VCM and 1-butene peak heights,
19
-------
OTHER CONDITIONS!
gSAME AS FIGURE 2
(5) UNKNOWN
(1 ) ETHYLENE
AT 200 M&/MIN (RTP)
COLLECTION
0 1
Figure 7. Linearity of all Peak
2 3
Heights vs Sample Collection Time
-------
,
in
! i
•Ml o
-f—H-
s^
. j:
— TIME
OTHER CONDITIONS
SAME AS FOR FIGURES 3 AND 5, EXCEPT AS FOLLOWS:
SAMPLE COLLECTION PERIOD: TWO MINUTES AT 200 mA/MIN
OZONATOR FLOW AT 105 m£/MIN (FROM 100 m£/MIN)
GC OVEN TEMPERATURE: 98°C (FROM 90°C)
Figure 8. Comparison of VCM Monitor ChromatO"
grams when Using Nitrogen and Helium
as the Carrier Gas
21
-------
and a 54% increase In Lhc sixth peak (dichloro
-------
OTHER CONDITIONS
SAME AS FOR FIGURE 4, CHROMATOGRAM A
Figure 9. Comparison of Chromatograms Obtained
when Using Air instead of Oxygen for
the Oxonator Supply
23
-------
It was beyond the scope of this contract to pursue such questions of reaction
mechanisms further. It should be noted that most of the tests reported here
were not repeated, and that no firm conclusions should be drawn based on
these results.
4.2 LINEARITY OF VCM PEAK HEIGHT WITH VCM CONCENTRATION
The trap concentration time was held at one minute at a flow rate of 150 m£/
min to provide useful sensitivity to VCM in the 0.5- to 5.0-ppm range, while
permitting a rapid cycle time. Samples containing 0.5, 5.0, 1.0, and 2.6 ppm
VCM were measured in the system. An apparent nonlinearity was traced to the
error in calculated sample volume passed through the trap, explained above.
(It was initially thought that the carrier nitrogen contained vinyl chloride,
but additional testing demonstrated that no detectable levels of unsaturated
hydrocarbons were present in the nitrogen carrier supply.) The agreement
between the four available samples, when corrected for the constant volume
sampling error, was within ±5%. This fact is noteworthy because the four
samples were obtained from three sources, which supports recent US NBS reports
that samples containing low concentrations of VCM are stable. Furthermore,
this test provided direct confirmation of the linearity of the monitor to VCM
in the 0.5- to 5.0-ppm range. Coupled with the demonstration of linearity
versus sample collection time, discussed above, these results indicated that
low VCM concentrations could be monitored with excellent results. Direct
demonstration of the monitor performance on lower concentrations of VCM was
left for final acceptance testing, the results of which are discussed in
Section 5 of this report.
4.3 SYSTEM TESTS USING SAMPLE LOOPS IN PLACE OF THE TRAP
It was of considerable interest to determine the limit of sensitivity which
could be achieved without concentration in a trap. Significant preliminary
work had been performed by EPA personnel. Using a 25-m£ sample loop and a
chemiluminescence detector of greater depth (which is a better design than the
Beckman chamber for slow reactants such as VCM) , it had been possible to
detect 50 ppb VCM in air. A good figure of merit for the VCM monitor de-
veloped on this contract was, therefore, its sensitivity to VCM independent
of the ability to concentrate the sample in the trap.
The trap was replaced by several sample loops fabricated from standard, GC
grade, I/8-inch OD stainless-steel tubing. The volume of the lines con-
necting the trap (or loop) to the GC valves was about 5 m&, and constitutes
a portion of total loop volume in every case. The loops tested had volumes
(total) of 10.7, 18.5, and 32 m£. The major effects noted were:
• The peak heights were not significantly greater for the larger
loops than for the small.
• The peak areas increased with sample loop volume, but the peaks
tended to overlap.
24
-------
• Increasing carrier flow rate for the larger sample loops helped
sharpen the peaks, but did not significantly affect the peak height.
• With proper choice of the Model 950 output filter time constant,
about 50 ppb VCM can be detected using the 18.5-mJi sample loop.
(The built-in Model 950 response switch "slow" position was used.)
These general conclusions are supported by the results presented in FIGURES 10
through 13, which are discussed further below.
FIGURE 10 shows the chromatogram obtained with the 10.7-m£ volume sample loop,
using the six-component mixture as before. The Model 950 gain was increased
25-fold to partially make up for the loss of concentration factor provided by
the trap, which was 37.4 to 1 in the case of the chromatogram of FIGURE 2. By
direct calculation using the simplest assumptions, the peak heights of FIGURE
10 would be expected to be 0.67 as large as those of FIGURE 2. To the extent
that the concentration factors for the various peaks are not proportional to
trap collection time (sampled volume), this simple calculation will not be
valid. The ratios of peak heights for the two chromatograms are given in
TABLE 5. Based on the nonlinearity of most of the peak height vs sample
collection time curves of FIGURE 7, only VCM and 1-butene would be expected to
give ratios of about 0.67. While the VCM ratio is lower than expected by about
24%, the 1-butene ratio is higher than expected by about the same percentage.
TABLE 5. RATIO OF PEAK HEIGHTS FOR CHROMATOGRAM OBTAINED USING 10.7-mH SAMPLE
LOOP (FIGURE 10) TO THOSE OBTAINED FOR A 400-mH SAMPLE CONCENTRATED
IN THE TRAP (FIGURE 2). WITH KNOWN SENSITIVITY CHANGES, THE RATIO
EXPECTED FOR COMPONENTS HAVING A LINEAR RESPONSE WOULD BE 0.67.
Peak No.
1
2
3
4
5
6
Compound
Ethylene
Propylene
Vinyl Chloride Monomer
1-butene
Unknown Impurity
1,1- and/or 1,2-dichloroethylene
Ratio of Peaks
(Figure 10 to Figure 2)
3.54
0.83
0.53
0.78
0.27
0.89
The chromatogram of FIGURE 11 was obtained under the same conditions as that
of FIGURE 10, except that the sample loop was increased to 32 m£ from 10.7 m£.
The recorder zero was also shifted slightly as a matter of convenience. The
important features of this chromatogram are: (a) the peak heights did not in-
crease three-fold compared to that of FIGURE 10; (b) all peaks were broadened
by roughly a factor of three; and (c) the peaks overlapped significantly.
25
-------
-« TIME
THIS CHROMATOGRAM COMPARES DIRECTLY TO THAT OF FIGURE 2 EXCEPT THAT THE MODEL
950 SENSITIVITY WAS INCREASED TWENTY-FIVE TIMES AND THE RECORDER CHART SPEED
WAS REDUCED TO ONE-HALF FOR THIS FIGURE. THE CHROMATOGRAM OF FIGURE 2 WAS OB-
TAINED USING A CONCENTRATION FACTOR OF 37.4. THE PEAK HEIGHTS FOR THIS FIGURE
WOULD BE EXPECTED TO BE 67% AS LARGE AS THOSE OF FIGURE 2, BY DIRECT CALCULA-
TION, IF NO OTHER PARAMETERS CHANGED.
OTHER CONDITIONS
RECORDER: CHART SPEED ONE-HALF INCH/MIN
100 mV FULL-SCALE SENSITIVITY
MODEL 950 SETTINGS: 1.0 PPM RANGE (2.5 TIMES MORE SENSITIVE THAN FIGURE 2)
10 mV OUTPUT (10 TIMES MORE SENSITIVE THAN FIGURE 2)
ZERO AND SPAN POTS AT 7.0 TURNS (AS FOR FIGURE 2)
RESPONSE SWITCH IN "FAST" POSITION (AS FOR FIGURE 2)
FLOW RATES: SAME AS FOR FIGURE 2, EXCEPT 10.7 m£ SAMPLE LOOP
INSTEAD OF TRAP
SAMPLE: SAME AS FOR FIGURE 2
Figure 10. Chromatogram of Six-Component Mixture in Nitrogen Obtained Using a
10.7 mH (Total) Sample Loop in Place of the Trap
26
-------
THIS CHROMATOGRAM COMPARES DIRECTLY TO THAT OF FIGURE 10, THE LARGER
SAMPLE LOOP BEING THE ONLY DIFFERENCE.
Figure 11. Chromatogram of Six-Component Mixture in Nitrogen Obtained Using a
32-mH Sample Loop in Place of the Trap
27
-------
I'Lrl'l.n i i ! <",''• \\ • •
— TIME
THIS CHROMATOGRAM COMPARES DIRECTLY TO THAT OF FIGURE 11, WITH ONLY A
CHANGE IN CARRIER GAS FLOW RATE.
Figure 12. .Chromatogram of Six-Component Mixture in Nitrogen Obtained Using a
32-mZ Sample Loop in Place of the Trap, and Using 120 mH/min of
Nitrogen Carrier Gas instead of 75
28
-------
'OTHER CONDITIONS
SANE AS IN FIGURE 10, EXCEPT AS FOLLOWS:
MODEL 950 SETTINGS: 0.25 RANGE (FOUR TIMES MORE
SENSITIVE THAN FIGURE 9).
SAMPLE LOOP:
RESPONSE SWITCH IN "SLOW11
POSITION
18.5 m£ (TOTAL)
-100-
----ispt—-t~r--— HTJ; ,-rrr;
'- :!''•;. ' ] ' '• • • I ' 1 i ! I
iiy-Ui'iil! !H L! i :-L_Ll4!lLULLL
THPftiit1lt![!;MrliiTl{-Nl1{ir
•TIME
Figure 13. Chromatogram of Six-Component Mixture in Nitrogen
Obtained with 18.5 m$ (Total) Sample Loop in Place of
the Trap, and with the Model 950 Sensitivity and Response
Speed Changes Indicated
29
-------
The .chromatogram of FIGURE 12 was taken with all conditions the same as for
FIGURE 11, except that the carrier gas flow rate was increased from 75 to 120
m£/min. The higher carrier flow rate improved the peak resolution markedly,
but did not significantly improve the peak heights compared to those obtained
with the 10.7-m£ sample loop at the lower flow rate (FIGURE 10).
An intermediate sample loop (18.5 mJl total) was then tested, with the carrier
flow rate set back to 75 m£/min, but with the Model 950 sensitivity increased
four-fold, and with the Model 950 response switch set in the "slow" position
to reduce the noise level. The chromatogram obtained is shown in FIGURE 13.
The "slow" response setting of the Model 950 was a little too slow for opti-
mum peak resolution. (Replacement of a single capacitor would permit optimi-
zation.) However, FIGURE 13 illustrates that it would be possible to detect
33 ppb VCM (signal-to-noise ratio of only one to two) when using the 18.5-m&
sample loop.
4.4 • THE EFFECTS OF VARYING OZONATOR AND CARRIER FLOW RATES
As noted in Section 3, Component Testing, the output of the Model 950 chemi-
luminescence monitor was maximum for equal steady-state flow rates of ozone-
rich and VCM containing gases. Those tests were performed by passing the two
gas streams directly into the Model 950 reaction chamber. It was believed
that roughly equal ozonator and carrier gas flow rates would, therefore,
provide the optimum (maximum) VCM peak when the Model 950 was combined with
the GC section for system testing. The situation was not that simple, how-
ever, presumably because of the nonlinear effect of carrier gas flow rate upon
VCM peak heights. The family of curves presented in FIGURE 14 does verify
that an optimum, ozonator flow rate exists for each carrier flow rate, but the
total variation in sensitivity for all data obtained was only about a factor
of two.
Several times in the course of the total test program it was discovered that
one or more of the flow indicators had become fouled by particulate matter.
It is quite possible that part of the data used to obtain the curves of
FIGURE 14 were in error due to faulty flow indicators. The minor features of
these curves, in particular, should be considered suspect since none of the
tests were repeated. The general excellent stability and reproducibility of
the monitor throughout the system tests suggests, however, that the basic
operational features indicated in FIGURE 14 are valid.
30
-------
20
70 80
CARRIER FLOW
9O
100
110
120
THE CHOICE OF 75 mfc/MIN CARRIER FLOW RATE AND 100 m£/MIN OZONATOR FLOW RATE
FOR MOST OF THE TESTS REPORTED WAS BASED UPON THESE CURVES AND UPON THE DE-
SIRE FOR A CONVENIENT ELUTION TIME FOR THE VCM PEAK. THE CHOICE OF 40 m£/
MIN FOR CARRIER AND 60 m£/MIN FOR THE OZONATOR FLOW RATE IS ACTUALLY OPTIMUM
WITH REGARD TO SENSITIVITY AND FREEDOM FROM DEPENDENCE UPON MINOR VARIATIONS
OF THE TWO FLOW RATES.
Figure 14. The Effect of Varying Carrier Gas Flow Rate upon the Peak Height Observed for
5-ppm VCM in Nitrogen, with Ozonator Flow Rate as a Parameter
-------
5.0 FINAL ACCEPTANCE TEST
The Final Acceptance Test was performed at the EPA facility at Research Tri-
angle Park, North Carolina, with contractor participation. The tests performed
may be conveniently divided into three classifications:
• Initial functional checkout, in the same configuration used for the
before-delivery test.
• Step-by-step test, modification, and retest to achieve virtual
freedom from interference with the VCM peak due to ambient air
pollutants.
• Testing in final configuration, which indicated that a 1-ppb VCM
detection limit is feasible.
Each of the above classifications of tests is discussed in the following
subsections.
5.1 INITIAL FUNCTIONAL TESTS
FIGURE 15 is a reproduction of a portion of recorder chart showing three suc-
cessive chromatograms obtained near the end of a four-hour run. All operating
conditions are noted on FIGURE 15. These three chromatograms are typical of
the reproducibility of the system in all tests. The VCM concentration was
approximately 0.75 ppm. A slow zero drift, which is barely noticeable over
the time period shown in the figure, occurs whenever the ozonator has been off
for a time. The initial zerd offset of the chemiluminescence analyzer due to
ozone decreases with time for the first day or so of use following a shutdown
of a day or more.
Two successive chromatograms obtained with a five-minute (instead of a one-
minute) sample collection time are reproduced as FIGURE 16. The only other
change made was to reduce the recorder sensitivity five-fold (to 50 mV from
10 mV full scale). The peak heights are slightly smaller in FIGURE 16 than
discussed fully in Section 4 of this report.
FIGURE 17 shows chromatogram B, obtained under conditions identical to those of
FIGURE 16, and chromatogram A, which was obtained with a tube having a lower
(about one-third) VCM permeation rate, but with the recorder gain increased
five-fold (to 10 mV from 50 mV full scale. The theoretical rate of VCM peak
heights A to B would be 5/3 = 1.67. That observed is within 1% of this value,
demonstrating good linearity for 250 and 750 ppb VCM.
Operation of the monitor on room air samples indicated that there were signif-
icant interferences in the 10-ppb VCM range due to unknown atmospheric
pollutants. Furthermore, it was found that cigarette smoking in the
32
-------
OTHER CONDITIONS;
DATE t
MODEL 6700 TIMINGt
TIME
DEC 15, 1975
(SEE OPERATING INSTRUCTIONS)
A ON AT 60 S, OFF AT 250" S (GC VALVE A)
B ON AT 60 S, OFF AT 539 S
C ON AT 0 S, OFF AT 60 S
D ON AT 60 S, OFF AT 90 S
MASTER CYCLE TIMER, 540 S
(GC VALVE B)
(SAMPLE COLLECTION)
(TRAP HEATING)
(9-MINUTE CYCLE)
RECORDER!
MODEL 950 SETTINGS»
SAMPLE COLLECTION*
0 TO 10 mV, 16 INCHES/HOUR
1.0 RANGE
100 mV OUTPUT
RESPONSE "FAST"
ONE MINUTE AT 150 m£/MIN, 0.75 PPM VCM
Figure 15. Three of Many Chromatograms Obtained at EPA Facility using a
Permeation Tube Source for 0.75-ppm VCM Concentration
33
-------
— TIME
THESE CHROMATOGRAMS WERE OBTAINED WITH OPERATING CONDITIONS IDENTICAL
TO THOSE FOR FIGURE 15, EXCEPT THAT THE RECORDER SENSITIVITY WAS RE-
DUCED FIVE-FOLD, THE SAMPLE COLLECTION TIME WAS INCREASED FIVE-FOLD,
THE CYCLE TIME WAS 16.7 MINUTES, AND THE MODEL 950 ZERO WAS RESET .
SLIGHTLY. THE SLIGHT DISCREPANCY IN PEAK HEIGHTS OF FIGURES 15 AND
16 IS DUE TO A FIXED-VOLUME (15 mH) SAMPLING ERROR WHICH ARISES FROM
EVACUATION OF THE TRAP AND SOME INTERCONNECTING LINES PRIOR TO
INITIATION OF SAMPLE COLLECTION. SEE SECTION IV OF THIS REPORT.
Figure 16. Two of Many Chromatograms Obtained at EPA Facility using
a Permeation Tube Source for 0.75-ppm VCM Concentration
34
-------
TIME
B
CHROMATOGRAM B IS IDENTICAL TO THOSE OF FIGURE 16, WHILE CHROMATOGRAM
A WAS OBTAINED WITH ONE-THIRD THE LENGTH OF VCM PERMEATION TUBE IN THE
LINE, AND WITH THE RECORDER SENSITIVITY INCREASED FIVE-FOLD (TO 10 mV
- FROM 50 mV FULL SCALE). THE INCREASE IN VCM PEAK HEIGHT IS WITHIN 1%
OF THE VALUE CALCULATED, ASSUMING PERFECT LINEARITY. ALSO NOTE THAT
THE SMALL PEAK, PROBABLY PRDPYLENE, INCREASED FIVE-FOLD. THIS
SUGGESTS THAT THE PROPYLENE IMPURITY IS IN THE ONE REMAINING PER-
MEATION TUBE ONLY.
Figure 17. Chromatogram Showing Linearity of the Monitor Below 1 ppm VCM
-------
laboratory increased both the number of interfering substances and, possibly,
the concentration of some interferants. Typical chromatograms illustrating
these points are shown in FIGURES 18 and 19.
FIGURE 18 is a reproduction of the relevant portions of two chromatograms.
Note that a different recorder was in use, which makes time increase to the
right. Chromatogram A was obtained using a 5-minute collection of room air
at 150 m£/min. Chromatogram B was obtained immediately after A, using 4.25
minutes of room air and 0.75 minute of 0.25-ppm VCM in air from the permea-
tion source, both at a 150-m£/min flow rate. The VCM peak shown in FIGURE 18B
is due to an equivalent of 38 ppb VCM for 5 minutes of sample collection. The
interfering peak, FIGURE 18A, is equivalent to about 12 ppb VCM. Both these
chromatograms were obtained with no cigarette smoking in the laboratory.
FIGURE 19 shows a complete chromatogram obtained on a 5-minute sample of room
air collected while there was cigarette smoke in the laboratory. Note that
the propylene peak increases markedly, and that a substance eluting very soon
after VCM is in cigarette smoke. The three later peaks are due to unknown
cigarette smoke components. The last two peaks were not detectable in the
room air; corresponding portions of FIGURES 18A and B were featureless.
While the original intent had been to provide a monitor capable of monitoring
VCM and two related compounds such as the dichloroethylenes, it was decided
that elimination of the direct interferences with VCM should receive priority
for the balance of the Final Acceptance Test
5.2 TESTS CONDUCTED WITH CARBOWAX COLUMNS
The GC oven temperature was increased from 90°C to 98°C to determine the
effect upon the air pollutants which were found to interfere with the VCM
peak. No significant improvement over the data of FIGURE 19 was obtained.
The Beckman analysis column was then replaced with a column previously eval-
uated by the EPA. This analysis column was 1 m long and consisted of 3-mm ID
Teflon tubing packed with 0.4% Carbowax 1500 (or Carbopack A, Supelco, Inc.)
The Beckman stripper column was left in place. This configuration was em-
ployed for the balance of the Final Acceptance Test. The only parameters
subsequently varied to eliminate interference were the GC oven temperature
and the time after initiation of trap heating at which back flush of the
stripper was initiated.
Tests of the new column configuration were conducted at about 30, 54, and
70°C for the GC oven temperature. It was found that at 70°C and with back
flush of the stripper column initiated at 120 to 130 seconds after start of
trap heating (equivalent to normal GC sample injection time), the troublesome
interferences were eliminated. This is illustrated by chromatograms A and B
of FIGURE 20. In retrospect, it is possible that the back flush was being
initiated too soon, which would cause a loss of part of the VCM peak as well
as the immediately following interferants. The test record does not clearly
indicate that the time could be reduced further without reducing the VCM peak.
36
-------
TIME
B
ROOM AIR (5-MINUTE COLLECTION) CHROMATOGRAM (A) AND ROOM AIR
(4.25 MIN) PLUS ABOUT 45 SECONDS COLLECTION OF 0.25 PPM OF VCM FROM
PERMEATION TUBE SOURCE (B). THESE CHROMATOGRAMS CLEARLY INDICATE
THE PRESENCE OF ONE OR MORE SUBSTANCES IN THE ROOM AIR WHICH WOULD
INTERFERE WITH VCM AT A LEVEL OF A FEW PPB. THE RECORDER SENSI-
TIVITY WAS ROUGHLY 2 PPB VCM PER DIVISION FOR A 0.75-LITER SAMPLE
(5-MINUTE COLLECTION AT FLOW RATE OF 150 mfc/MIN).
Figure 18. Chromatograms Indicating Presence of Interfering Pollutants
37
-------
ROOM AIR (5-MINUTE COLLECTION) WITH
DELIBERATE SMOKING IN LABORATORY. THIS '
CHROMATOGRAM COMPARES DIRECTLY TO THAT .
OF FIGURE ISA IN ALL RESPECTS EXCEPT '
THAT NO SIGNIFICANT FEATURES OCCURRED
IN THE PORTION OF THE CHROMATOGRAM OF
FIGURE ISA NOT SHOWN. SMOKE CONTAINS :
SOME LIGHT COMPONENTS AND SOME WHICH ;
ELUTE MUCH LATER, BUT NOT ALL OF THE
TROUBLESOME ATMOSPHERIC COMPOUNDS ARE ;
IN SMOKE. --i
-»4i-r •••!--f
.,; m :.LLLI. -;-..
i I ' J ' i ' i
TIME —
Figure 19. Chromatogram Indicating the Presence of Interfering Compounds
in Cigarette Smoke
38
-------
..-.*—,,-r~.«.-.—** -. .
ii ; i . .!u • i :
.'.-.I!'.-,: i.
i : i -! ' t • i - r -
•/,:. m
TIME-
B
CHROMATOGRAM A WAS FOR 10 SECONDS OF 0.25-PPM VCM IN CLEAN AIR WITH
BALANCE OF FIVE MINUTES COLLECTION OF ROOM AIR. CHROMATOGRAM B WAS
FOR ROOM AIR ONLY. SMOKING WAS NOT CONTROLLED DURING THESE TESTS,
WHICH ACCOUNTS FOR THE MAJOR VARIATION IN THE PROPYLENE PEAKS.
NOTE THAT THERE IS LESS THAN ONE CHART DIVISION PEAK (PROBABLY
NOISE) AT THE VCM ELUTION TIME FOR CHROMATOGRAM B. THE INTERFER-
ENCE IS, THEREFORE, LESS THAN 2 PPB VCM EQUIVALENT, AND PROBABLY
LESS THAN 0.5 PPB.
Figure 20. Two Chromatograms Obtained after Modifying Operating Conditions
to Eliminate Interference of Cigarette Smoke and of Unknown
Atmospheric Pollutants with VCM Determination
39
-------
It should be emphasized that, as implemented for these tests, the back-flush
method of eliminating the interfering compounds precluded the possibility of
measurement of any components which elute from the stripper column after VCM.
However, the Model 6700 programmer can be used to delay these components by a
double back-flush technique. It is only necessary to plug in an extra timing
board (no wiring required) to provide two valve-A driver functions. It will
then be possible to start back flush at the proper moment (as for these tests)
by having valve timer A-l turn off 120 to 130 seconds after trap heat is
initiated. The new timer, A-2, can be set to operate valve-A to resume for-
ward flush at an appropriate later time, until the desired peaks are out of
the stripper column. Timer A-2 may then initiate permanent back flush, as
A-l did in these tests, to prevent very slow peaks from ever reaching the
analysis column.
5.3 TESTS PERFORMED IN FINAL CONFIGURATION
The final configuration provided for virtually no interference with the VCM
peak in the prevailing ambient air. Smoking was permitted in the laboratory,
and only the propylene peak varied due to the variations in smoke loading.
The operating conditions of the permeation tube VCM calibrator were verified
and a series of stable chromatograms was obtained. The two successive chro-
matograms of FIGURE 21 are typical of the noise level and peak reproducibility
obtained for 0.25-ppm VCM in precleaned air collected in the trap for five
minutes at a flow rate of 150 m£/min (750-m& sample).
A second test result is indicated by comparison of the base-line noise levels
in FIGURES 20 and 21. The Model 950 sensitivity was increased ten-fold, with
a corresponding decrease in recorder sensitivity for FIGURE 21 compared to
FIGURE 20. The approximate equality of noise for the two cases indicates that
the dominant noise comes from the PMT. It is very probable that variations in
ozone concentration are the major cause of low frequency noise. The ozonator
lamp current varies with line voltage. This causes variations in the ozone
concentration, which affects both the base-line and the VCM peak heights.
The permeation tube calibrator parameters were then adjusted to provide a
sample of 0.125-ppm VCM in air. The chromatogram of FIGURE 22A is typical of
several obtained. Note that the Model 950 gain was increased about a factor
of two to provide almost full-scale peaks for 0.125-ppm VCM, and that the
noise increased proportionately in FIGURE 22A compared to that of FIGURE 21,
as would be expected. FIGURE 22B is a reproduction of a chromatogram obtained
with all conditions the same as for FIGURE 22A except that the Model 950 re-
sponse switch was placed in the "slow" position. The distortion and attenu-
ation of the peaks indicate that the "slow" time constant is four to five
times too large for optimum performance in this application. This situation
could be corrected by changing only one capacitor in the Model 950 preampli-
fier board.
40
-------
TIME —
THESE CHRQMATOGRAMS ALSO SHOW THAT THE MAJOR NOISE IS FROM THE PMT, SINCE
THE MODEL 950 SENSITIVITY IS UP TEN-FOLD AND RECORDER SENSITIVITY IS DOWN
TEN-FOLD COMPARED TO FIGURE 20
Figure 21. Relevant Portions of Two Consecutive Chromatograms Obtained Using
0.25-ppm VCM in Clean Air from Permeation Tube, Collected for
Five Minutes at 150-mH/minute
41
-------
A TIME —
THE MODEL 950 RESPONSE SWITCH WAS SET "FAST" FOR A AND "SLOW" FOR B. THE
SLOW TIME CONSTANT SHOULD BE REDUCED BY A FACTOR OF ABOUT FOUR FOR OPTI-
MUM FILTERING IN THIS APPLICATION
Figure 22. Two Successive Chromatograms Obtained for Five-Minute Collections
of 0.125-ppm VCM in Nitrogen from Permeation Tube Source at Flow
Rate of 150 mH/minute
42
-------
The low frequency variations in baseline still seen in FIGURE 22B would then
be no more than about five times as large, or about +1% of recorder scale
maximum. This would result in a detectable limit of about 1.5-ppb VCM for
a single measurement (signal-to-noise ratio of about one). Comparison of
several chromatograms to visually screen out the random noise effect would,
of course, permit a still lower detectable limit. Furthermore, by increasing
either the sample flow rate or the collection time, the VCM concentration fac-
tor could be doubled (provided the trap temperature during sample collection
was less than about 35°C) to make it possible to detect less than 1 ppb VCM
with a single observation.
43
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-78-008
2.
3. RECIPIENT'S ACCESSIOI»NO.
4. TITLE AND SUBTITLE
CHEMILUMINESCENT MONITOR FOR VINYL CHLORIDE
5. REPORT DATE
February 1978
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
S.G. Riccio, M.W. Greene, W.D. Dencker, and
R.I. Wilson
8. PERFORMING ORGANIZATION REPORT NC.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Beckman Instruments, Inc.
Advanced Technology Operations
Anaheim, California 92806
10. PROGRAM ELEMENT NO.
1AD605
11. CONTRACT/GRANT NO.
68-02-1770
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory - RTF, NC
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final 3/75 - 6/76
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT
A monitor for vinyl chloride monomer (VCM) in ambient air was constructed using
commericially available components of a gas chromatograph (GC) coupled with a chemi-
luminescence ozone analyzer slightly modified to make it suitable for use as a GC
detector. The specificity for VCM is enhanced by use of a chemiluminescence detector
because saturated hydrocarbons do not chemiluminesce with ozone.
A custom absorption trap capable of concentrating the VCM from one liter (or more)
of ambient air was used to extend the lower detection limit. Using a custom trap
heating circuit, the concentrated sample can be thermally eluted into the GC in a gas
volume of 10 ml or less, providing a 100-fold increase in sensitivity compared to that
obtained with a 10 ml sample loop.
Preliminary system and final acceptance test results are reported in detail. These
test results indicate that a monitor of this type can detect less than one part per
billion (ppb) of VCM in ambient air operating on a 15-minute cycle. If the concentra-
tion trap is replaced by a 20 ml sample loop, the limit of detection is estimated to
be about 35 ppb VCM when operating on a 5r-minute cycle.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
*Air pollution
*Vinyl chloride
*Monitors
*Chetniluminescence
*Gas chromatography
13B
07C
07D
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (ThisReport)
UNCLASSIFIED
21. NO. OF PAGES
52
20. SEC
iECURITY CLASS IThispage)
UNCLASSIFIED
22. PRICE
EPA Form 2220-1 (9-73)
44
------- |