EPA-650/2-75-050
August 1975 Environmental Protection Technology Series
DEVELOPMENT OF SELECTIVE
HYDROCARBON SAMPLING SYSTEM
AND FIELD EVALUATION WITH
CONVENTIONAL ANALYTICAL SYSTEM
U.S. Environmental Protection Agency
Office of Research and Development
Washington, D. C. 20460
-------
EPA-650/2-75-050
DEVELOPMENT OF SELECTIVE
HYDROCARBON SAMPLING SYSTEM
AND FIELD EVALUATION WITH
CONVENTIONAL ANALYTICAL SYSTEM
by
Dr. Arthur F. Isbell, Jr.
Analytical Research Laboratories, Inc.
160 Taylor Street
Monrovia, California 91016
Contract No. 68-02-1201
ROAP No. 26AAP-30
Program Element No. 1AA010
EPA Project Officer: James B. Homolya
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park, N. C. 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
WASHINGTON, D.C. 20460
August 1975
-------
EPA REVIEW NOTICE
This report has been reviewed by the National Environmental Research
Center - Research Triangle Park, Office of Research and Development,
EPA, and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Environmental
Protection Agency, nor does mention of trade names or commercial
products constitute endorsement or recommendation for use.
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environ-
mental Protection Agency, have been grouped into series. These broad
categories were established to facilitate further development and applica-
tion of environmental technology. Elimination of traditional grouping was
consciously planned to foster technology transfer and maximum interface
in related fields. These 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
9. MISCELLANEOUS
This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to
develop and demonstrate instrumentation,,equipment and methodology
to repair or prevent environmental 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 for sale through the National
Technical Information Service, Springfield, Virginia 22161.
Publication No. EPA-650/2-75-050
11
-------
ABSTRACT
A sampling and analytical system was designed to permit the
routine determination of the saturate, unsaturate, aromatic, and
oxycarbon content of stationary-source hydrocarbon vapor emissions.
The intent has been to lend great definition to the traditional
total hydrocarbon analysis.
A Joy Manufacturing Company Method 5 stack sampler has been modi-
fied to incorporate a vapor collector cartridge which quantita-
tively traps all volatile hydrocarbons present in the sampled
gas stream.
The Class Analyzer desorbs the sample from this Universal Collector
and passes it through a system of class abstractors which remove
certain hydrocarbon classes while permitting the remainder of the
sample to pass into cryogenic traps.
The contents of each trap are desorbed sequentially through an
external hydrocarbon (flame ionization) detector. The quantity
of each class present in the sample can be calculated from the
detector responses to the effluents from the various class
abstractor streams.
Detailed evaluations of the Universal Collector and class abstrac-
tor candidates are presented. Test results of the complete
sampling and analytical system are included.
The stack sampler and Class Analyzer serve as models for instru-
mentation which will permit technicians to sample and analyze
stationary source emissions by hydrocarbon class.
This report was submitted in fulfillment of Contract Number
68-02-1201 by the Analytical Research Laboratories, Inc., under
the sponsorship of the Environmental Protection Agency. Work
was completed as of December 1974.
ill
-------
CONTENTS
age
Abstract
List of Figures v
List of Tables vi
Sections
I Conclusions 1
II Recommendations 3
III Introduction 4
IV Hardware Design and Construction
and Experimental System Fabrication 7
V Experimental Procedures 25
VI Discussion 32
VII References 59
iv
-------
FIGURES
No. Page
1 Schematic Diagram of Universal Collector Test
Apparatus 8
2 Universal Collectors 10
3 Joy Method 5 Stack Sampler 12
4 Schematic Diagram of Modified Method 5 Sample
Train 13
5 Front View of Sampling Unit 15
6 Sampling Unit Impinger Compartment 16
7 Schematic Diagram of Class Analyzer 21
8 Front View.of Class Analyzer 22
9 Rear View of Class Analyzer 23
10 Schematic Diagram of Class Abstractor Test
Apparatus 27
11 Universal Collector By-pass . 30
12 Schematic Diagram of Universal Collector 39
13 Typical Class Analyzer Output Signal 57
v
-------
CABLES
No.
Paqe
1 Model Compounds for Class Abstractor Evaluation 29
2 Candidates for Heavy Ends Sorbent of Universal
Collector 35
3 Candidates for Light Ends Sorbent of Universal
Collector 37
4 Class Abstractor Candidates 45
5 Class Analyzer Analysis of Sample EPA-1 53
6 Class Analyzer Analysis of Sample EPA-2 56
vi
-------
SECTION I
CONCLUSIONS
The combined collection, class-separation, and individual deter-
mination of the saturate, unsaturate, aromatic, and oxycarbon
fractions of volatile hydrocarbons in stationary source emissions
is feasible. The principle employed in the Class Analyzer, how-
ler, will have to be further explored and developed before it can
be made applicable for the selective determination of halocarbons.
The operation of the Class Analyzer directly on stream is not
Practical. The volatile hydrocarbons must first be accumulated
°n an adsorbent and then desorbed into the Class Analyzer. The
'Universal Collector" developed on this program successfully
ulfills this requirement.
Simultaneous volatile hydrocarbon sampling (with the Universal
c°Hector) and parti'culate collection can be achieved using a
m°dified EPA Reference Method 5 stack sampler. Sample streams
"aving high moisture levels are troublesome. Decoupling the
Universal Collector from the stack sampler for interfacing with
tne Class Analyzer is reasonably facile.
The use of a field-practical detector in the modified stack sampler
to signal hydrocarbon breakthrough from the Universal Collector
d°es not appear feasible. Careful adherence to proper sampling
Procedures should, however, obviate premature overloads of the
Universal Collector.
Several design refinements of the bread-board Class Analyzer are
necessary before it can be advanced to prototypic hardware quality.
°ntamination from the Universal Collector and the Class Analyzer
c°mprise intrinsic problems requiring further attention.
of the very steps of the selective hydrocarbon sampling
analysis procedure will not require extensive training or
qualifications on the part of the user. The results of the program
-------
lead to the conclusion that routine sampling and analysis of
stationary source emissions by technicians will be possible after
minor refinements of the designs thus far developed have been
effected.
-------
SECTION II
RECOMMENDATIONS
e Class Analyzer design should be modified to incorporate several
eeded changes. The plumbing layout can be improved and less
entive wall materials employed. The use of heated, teflon-lined
ubing could be used to provide the latter enhancement. System
°w rates downstream of the traps should be stabilized by incorpo-
lng suitable flow controllers. Any such devices used should
> however, contaminate or hold up the sample. Means should be
ovided for heating the trap more rapidly once the cryogenic
hs are removed. An electrical configuration would probably be
est suited for this purpose. A flame ionization detector (FID)
a signal-processing system should be made an integral part
ne Class Analyzer. Also, it may be desirable to incorporate
andem
an electron capture detector for the purpose of estimat-
In9 halocarbon contributions.
optimum performance of the Class Analyzer, water vapor should
removed. A desiccant could be placed in the filter oven of
stack sampler upstream of the Universal Collector. Studies
needed, however, to establish the efficiencies of various
ccants and their inertness toward sample constituents.
imum preconditioning of the AC charcoal used in the Universal
ector must be established. Thermal conditioning at high
Peratures causes partial sample loss while low temperature con-
oning does not remove surface contaminants which continually
leed-0ff into the sample.
-------
SECTION III
INTRODUCTION
The heavy contribution of hydrocarbon emissions (in the present
context, hydrocarbon emissions includes all volatile organics)
to air pollution has been recognized for some time. Abating the
effects of these emissions must necessarily involve identifying
specific pollutants with their sources. Even when the nature of
the emissions of a given stack is known, and means of controlling
them are applied, monitoring on a continuous or periodic basis is
essential to assure proper functioning of the emission control
devices. Laborious and complex laboratory identification and
quantitation must be replaced in part by more rapid test methods
in order to cope with demands put upon the monitoring organization,
Field monitoring now involves the use of the Total Hydrocarbon
Analyzer (TEA) or the collection of a total representative sample
of the emissions followed by detailed laboratory analysis of the
collected sample. The latter analysis generally relied upon gas
•chromatographic techniques for the separation of the various com-
ponents of the sample with identification of the fractions by
different instrumental methods, or by correlation with known reten-
tion data. Such techniques are not .readily amenable to field use,
and do not furnish rapid results. Detection by optical disper-
sion and absorption or by flame ionization techniques provide
rapidly attainable results for total organic carbon content, but
without identifying the nature of the organic compounds. Much
improvement would be realized if pollutants could be sampled and
analytically catalogued into classes of compounds, such as
saturates, unsaturates, aromatics, oxycarbons, halocarbons, etc.
If an abstraction device could be designed that is unique for a
specific class of compounds while rejecting others, similar to
many of the semiquantitative Kitagawa-tube test kits, then a total
carbon determination of the rejected fraction would permit the
calculation, by difference, of the total organic emission that
-------
falls in the abstracted class. Since much of the effect of a
specific pollutant is attributable to the functionality of the
substance, classification by functional group should be adequate
for rapid and reasonable assessment of the pollution potential,
without resorting to identification of each individual member
of the class.
The purpose of the present program has been to develop and field
test a system that will collect, separate, and release for analysis
atmospheric hydrocarbon pollutants as discrete molecular-type
classes. The overall system, on being interfaced with conventional
analytical devices (a gas chromatograph or a THA) would permit
selective determinations of such hydrocarbon classes as: saturate,
unsaturate, aromatic, halocarbon, and oxycarbon fractions. The
system would be used for sampling over various time periods and
environments, including hot stack gases, hydrocarbon pollutants
associated with stationary emission sources.
A non-cryogenic, isokinetic sampler capable of collecting all hydro-
carbon vapors without wall losses, preventing the breakthrough
of light hydrocarbon components from the collector, and furnishing
isolated collection of preformed particulates was developed from
standard sampling hardware. A special separation system was also
developed into which the volatile contents of the sampler can be
transferred by thermal or other forms of desorption. By means of
stream splitting and selective abstraction of hydrocarbon classes
on different, selected sorbents, analytical measurements of the
classes is achievable.
The initial phase of the program consisted of a comprehensive
literature survey to determine the extant knowledge in the field
of sorption chemistry. The results of this survey permitted the
selection of the best candidates for a hydrocarbon sampling device,
the Universal Collector, and for hydrocarbon class abstractors for
the Class Analyzer.
-------
Phase II involved laboratory screening of suitable Universal
Collector candidates. Based on the results of these experiments,
six identical Universal Collectors were constructed.
The third phase consisted of the modification of a commercially-
available EPA Reference Method 5 stack particulate sampler.
The incorporation of the Universal Collector into the Method 5
Sampler resulted in a stack sampler capable of collecting both
particulates as well as volatile hydrocarbons from a stationary
source.
Laboratory evaluation of suitable class abstractor candidates com-
prised the fourth phase. A detailed analysis of the sorption
properties of the various sorbents permitted the selection of a
series of class abstractors which, when properly arranged, pro-
vided a class analysis of a collected hydrocarbon sample.
Phase V included the incorporation of the class abstractors into
an analytical instrument, the Class Analyzer. This device separates
the collected sample into various fractions which, when passed
through an external THA or FID provides quantitative information
about the class composition of the sample.
In the final phase of the program, the total system was subjected
to field testing. Hydrocarbon samples from actual and simulated
stationary sources were collected, using the modified Method 5
stack sampler. These samples were then introduced into the
Class Analyzer for the quantitative determination of hydrocarbon
class composition.
-------
SECTION IV
HARDWARE DESIGN AND CONSTRUCTION AND EXPERIMENTAL SYSTEM FABRICATION
UNIVERSAL COLLECTOR (UC)
The test, apparatus shown schematically in Figure 1 was designed
for a comprehensive evaluation of UC candidates. The breakthrough
and desorption volumes of a variety of volatile hydrocarbon samples
were determinable at temperatures ranging from 0 to 400 . Cryo-
genic trapping of the sample itself or of the effluent from the UC
candidate permitted further gas chromatographic analysis for
sample composition or contamination by the UC candidate.
Sample flow control and measurement were provided by Brooks Model
8744B NRS 'flow controllers and a Fischer and Porter Lab-Crest
Model 10A1460NB22B rotameter with a 55 ml/min maximum flow rate
for air. Temperature control and sample detection were provided
by a Perkin-Elmer 880 gas chromatograph equipped with a dual FID
system. The FID response was recorded by a Leeds and Northrup
Speedomax G strip chart recorder capable of chart speeds varying
from 0.635 to 30.48 cm/min (0.25 to 12 in/min).
The tubing was made of 304 stainless steel with Swagelok brass
fittings. All valve bodies were either 316 stainless steel, brass,
or plated brass. Buna-N 0-rings were used in the backflush valve.
Hydrocarbon samples were either purchased from Matheson Company
as analyzed standards or mixed and analyzed gas chromatographically
by ARLI. ARLI-prepared samples were mixed using a stainless steel
vacuum-pressure rack using standard vacuum rack techniques.
The UC candidates usually required preparation prior to testing.
Exceptions were uncoated 80-100 mesh Porapak Q and 60-80 mesh
Tenax GC. Porapak Q coated with 3% OV-1 (methyl silicone gum)
was prepared by dissolving 0.06g of OV-1 in chloroform in a 100 ml
round-bottom flask. Two grams of Porapak Q were added to this
solution. The chloroform was removed using a rotary vacuum
evaporator with the round-bottom flask submerged in warm water.
-------
r
HYDRO-
CARBON
MIXTURE
N.
FLOVv'-CONTROL BREADBOARD
ROTAMETER
BACKFLUSH
-CL
TTT
VALVE
COLUMN
BY-PASS
VALVE
VENT
TRAP
UNIVERSAL
COLLECTOR
COLUMN
OVEN
1
FID
PERKIN-ELMER 880 GAS CHROMATOGRAPK
Figure 1. Schematic diagram of Universal Collector test apparatus
8
-------
Most of the molecular sieve and charcoal candidates were manu-
factured in pellet form. These pellets were ground to either
60-80 or 80-100 mesh using a mortar and pestle.
For testing purposes, the UC candidates were packed in 20 by
0.64 cm O.D. stainless steel tubes. Each candidate was condi-
tioned for about 2 hours at 200 under a 30 ml/min helium flow.
The final configuration of the UC is shown in Figure 2. Swagelok
fittings are used throughout. Connecting fittings are 0.64 cm O.D.
double-end shut-off quick-connect fittings. All fittings which
might come into contact with corrosive substances are 316 stain-
less steel. The remaining fittings are brass or plated brass.
All tubing is 304 stainless steel. The sorbent tubes measure 20
by 0.95 cm O.D. The remainder of the tubing is 0.64 cm O.D.
STACK SAMPLER
The original program plan called for the design and construction of
a stack sampler. The Project Officer subsequently suggested that
a more cost-effective approach would be to modify an off-the-shelf
EPA Reference Method 5 particulate sampler.
Solicitations for Method 5 hardware were accordingly sent to a
number of vendors. Bid specifications required the hardware to
incorporate all necessary components to conform with EPA/APCO
standards for Reference Method 5 as outlined in the Federal Register,
Vol. 36, No. 247, except as'noted below. The system should be
encased, portable, operable on 115 VAC (60 Hz), and include (or
exclude) the following options or dimensions:
o The sampling unit should include a filter assembly and
cyclone. The filter assembly should be of- the largest
size normally provided by the vendor and should include
three fritted glass discs and six gaskets. Filtering
media should not be supplied.
o The sampling unit should not include the four liquid
impingers and interconnecting hardware usually provided.
-------
Figure 2. Universal Collectors
10
-------
o The pitobe assembly should have an effective length
of 1.5 m (5 ft). The internal surface of the
sampling tube should be Pyrex glass or its equivalent.
One each of three different sized sampling nozzles
[0.64, 0.95, and 1.27 cm dia. (0.25, 0.375, and 0.50 in)]
should be provided.
o Two nomographs should be supplied.
o The umbilical cord length should be 30.5 m (100 ft).
o The pitobe support system should be of a length appro-
priate to handle the specified pitobe.
After considering shipping costs, equipment specifications, some
minor added requirements, and vendor location, a decision was made
to purchase the Joy Method 5 stack sampler (Figure 3). Most of
the other models considered would probably have served as well.
The Joy stack sampler is composed of two basic units: (1) the
control unit, which contains the pump, heater controls, tempera-
ture gauge, and air flow measuring and control devices; and (2)
the sampling unit, which contains the heated pitobe, the. filter
oven, and the impinger train.
Modifications were made to the sampling unit only (Figure 4).
Initially, these modifications were confined to the impinger com-
partment. The intersection of the particulate and organic sample
streams was in the cold impinger compartment. Condensate. from
wet air streams could enter the UC. The bulky UC connection flange
Required the elimination of two of the four impingers. The
organic-sample slip stream flow-rate was determined by measuring
the pressure drop across a restrictor using a sensitive Magnehelic
gauge. Flow rate changes caused exaggerated gauge fluctuations.
These shortcomings were corrected by redesigning the system.
The intersection of the particulate and organic sample streams
was moved into the heated filter oven. A desiccant tube was
Placed in the oven between the intersection and the UC to remove
11
-------
rm
~rrn
Figure 3. Joy Method 5 stack sampler
12
-------
TO VACUUM PUMP
ON-OFF VALVE
ROTAMETER
NEEDLE
VALVE
TO STACK
QUICK-
CONNECT
FITTINGS
\ A
B
l
A
FILTER
FILTER OVEN
JJ_NIVE_RSAL COLLECTOR JEMPINGERS . .'
ICE BATH
A - DESICCANT
B - AC CHARCOAL
C - TENAX GC
Figure 4. Schematic diagram of modified Method 5 sample train
13
-------
enough water to prevent condensation. Swagelok bulkhead quick-
connect fittings were installed in the wall between the filter
oven and the impinger compartment. This compact design left room
for three of the four standard Method 5 impingers. The three
impingers and their connecting glassware were also obtained from
Joy Manufacturing Company. The restrictor and Magnehelic gauge
were replaced by a 15 cm Fischer and Porter Lab-Crest Model
10A1460NB22B rotameter to permit more precise flow rate determina-
tions. Figures 5 and 6 show various views of the final sampling
unit design.
CLASS ABSTRACTOR COLUMNS
For evaluation experiments, the class abstractor substrates were
packed in stainless steel tubing. Swagelok brass fittings were
used to connect the columns to a Perkin-Elmer 880 gas chromato-
graph for abstraction testing. A series of class abstractors was
tested. These abstractor candidates were selected from the litera-
ture, and on the basis of past experience.
In 1963, a hydrocarbon class analyzer built for the analysis of
automobile exhaust was reported. The class abstractors described
therein were prepared by impregnating 1 g of diatomaceous earth
(80-100 mesh) with 1 ml of various solutions. The unsaturate
abstractor used a 20% HgS04, 20% H2S04 solution- The unsaturate
and aromatic abstractor used a 4%AgSO , 95% H2S04 solution.
To prepare the HgSO., H-SO, solution, one adds 0.2 g of HgSO. to
0.6 ml of water in a small (5 to 10 ml) Erlenmeyer flask. The
HgSO. will be yellow but highly insoluble. The mixture is then
heated to incipient boiling while stirring. Enough concentrated
HpS04 is next added to completely dissolve the yellow HgS04 and to
start the formation of a white precipitate. Water is added drop-
wise until the white precipitate dissolves completely. The hot
solution is poured into a small Erlenmeyer flask containing 1 g
of 60-80 mesh GC-22 firebrick (or the equivalent). This is mixed
thoroughly until all firebrick particles are moist. The coated
14
-------
Figure 5. Front view of sampling unit
15
-------
Figure 6. Sampling unit impinger compartment
16
-------
firebrick is packed into a 20 by 0.64 cm O.D. stainless steel tube.
The column is then conditioned under 30 ml/min helium flow at
ambient temperature for 2 hours.
The Ag SO., H.SO. substrate is prepared by making at least 1 ml
of a solution containing 4% Ag2S04, 95% H_S04, and 1% water.
The AgpSO. dissolves quite slowly so that continuous stirring is
required to effect complete solution. A 1 ml portion of this
solution is pipetted into a small Erlenmeyer flask containing 1 g
of 60-80 mesh Gas-Chrom P firebrick. This is mixed thoroughly
until all firebrick particles are moist. The coated firebrick is
then packed into a 20 by 0.64 cm O.D. stainless steel tube. The
column is then conditioned under 30 ml/min helium flow at ambient
temperature for 2 hours.
Coulson used HgCClO.)^ and HC10. on firebrick to remove olefins
and acetylenes from automobile exhaust gases as an identification
2
Method. A similar abstractor was used for the analysis of
3 4
gasoline hydrocarbons. ' This substrate is prepared by making
at least 1 g of a solution 1 M in Hg(C104>2 and 2M in HC10.. One
roixes 1 g of this solution with 1 g of 60-80 mesh Gas-Chrom P
firebrick in a small Erlenmeyer flask until all firebrick particles
are moist. The coated firebrick is packed into a 20 by 0.64 cm O.D.
stainless steel tube. The column is conditioned under 30 ml/min
helium flow at 110°C for 2 hours.
also experimented with various combinations of Hg(C104)2
and HCIO. on firebrick for the abstraction of unsaturates from
automobile exhaust gases. He found that 40% Hg(C104)2 on GC-22
firebrick was an excellent unsaturate abstractor. This substrate
is prepared by dissolving 0.8 g of Hg(C104)2 in about 10 ml of
water in a 100 ml round bottom flask. One then adds 2 g of 60-80
ttesh GC-22 firebrick to this solution. The water is removed by
rotary vacuum evaporation. Again the coated firebrick is packed
into a 20 by 0.64 cm O.D. stainless steel tube. The column is
Conditioned under 30 ml/min helium flow at 130°C for 2 hours.
17
-------
Laub has suggested that a strong electron acceptor might form
sufficiently strong charge-transfer complexes with electron donors,
such as aromatic compounds, to provide selective abstractions.
A suitable electron acceptor, 2,3-dichloro-5,6-dicyanobenzoquinone
(DDQ), was both coated directly on a solid support and dissolved
in a non-aromatic chromatographic liguid phase prior to coating.
A 10% DDQ coating on Anakrom ABS firebrick is prepared by dissolv-
ing 0.1 g of DDQ in about 5 ml of chloroform in a 100 ml round
bottom flask. One then adds 1 g of 100-110 mesh Anakrom ABS. The
chloroform is removed by rotary vacuum evaporation. The coated
firebrick is packed into a 20 by 0.32 cm O.D. stainless steel
tube and conditioned under 30 ml/min helium flow at ambient
temperature for 2 hours. -
To prepare a non-aromatic liquid phase containing DDQ, one saturates
about 0.1 g of d'i( 2-ethylhexyl) sebacate with DDQ. One dissolves
0.1 g of this solution in about 5 ml of acetone in a 100 ml round
bottom flask. Next 1 g of 100-110 mesh Anakrom ABS is added. The
acetone is removed by rotary vacuum evaporation. The coated fire-
brick is packed into a 20 by 0.32 cm O.D. stainless steel tube
and conditioned under 30 ml/min helium flow at ambient temperature
for 2 hours.
7
Brenner, Cieplinski, Ettre, and Coates reported the adsorptive
properties of various synthetic zeolites. Rowan used Molecular
Sieve 5A as a class abstractor to aid in gas chromatographic
8 A
identifications. The molecular sieve pellets are ground and sieved
to 60-80 mesh and packed in a 20 by 0.64 cm O.D. stainless steel
tube. The column is conditioned under 30 ml/min helium flow at
200°C for 2 hours.
The author's past experience has indicated that salt-coated porous
9 10
silica beads and aluminas possess class abstractor properties. '
Both substrates are prepared in a similar manner. One dissolves
the required weight of salt in about 5 ml of water in a 100 ml
round bottom flask. Then 1 g of 100-150 mesh Porasil C or 100-120
mesh F-20 alumina is added. The water is removed by rotary vacuum
18
-------
evaporation and the dried substrate is packed into a 20 by 0.32
cm O.D. stainless steel tube. The column is conditioned under a
30 ml/min helium flow at 300°C for 2 hours.
A boric acid-coated substrate has been found to be a quantitative
11 12
abstractor of primary and secondary alcohols. ' Three sub-
strates were prepared. The first substrate is prepared by
dissolving 0.4 g of Carbowax 20M in about 10 ml of chloroform
in a 100 ml round bottom flask and 2 g of 60-80 mesh Chromosorb P
is added. The chloroform is removed by rotary vacuum evaporation.
In another 100 ml round bottom flask, 0.06 g of boric acid is
combined with about 10 ml of petroleum ether and the Carbowax-
coated Chromosorb is added. After the petroleum ether is removed
by rotary vacuum evaporation, the substrate is packed in a 20
by 0.64 cm O.D. stainless steel tube. The column is conditioned
under 30 ml/min helium flow at 225 C overnight. The preparation
procedure for the second substrate is identical to the first sub-
strate with boric anhydride substituted for boric acid. The third
substrate is prepared by dissolving 0.1 g of boric acid in about
10 ml of water in a 100 ml round bottom flask. One then adds
2 g of GC-22 firebrick, removes the water by rotary vacuum evapora-
tion, and packs the substrate in a 20 by 0.64 cm stainless steel
tube. The column is conditioned under 30 ml/min helium flow at
300°C overnight.
Several substrates designed for the abstraction of oxycarbons
12
have been reported. Two of these abstractors are prepared by
merely placing 100 mg of either. LiBH or LiAlH in a short section
of 0.64 cm stainless steel tubing. These columns are then ready
for immediate use.
A modification of this abstractor requires coating GC-22 firebrick
with LiAlH.. Because LiAlH. reacts rapidly with water, precautions
must be taken to prevent this reaction. To do this, one places
1 g of 60-80 mesh GC-22 firebrick in a 20 by 150 mm test tube.
The firebrick is dried in' a vacuum oven at 110 overnight. The
firebrick is removed and sealed immediately with a stopper. When
the firebrick has cooled, the stopper is quickly removed and
19
-------
2 ml of a 1.3 M LiAlH solution in diethyl ether is quickly
introduced into the test tube. The stopper is replaced and the
test tube agitated until all firebrick particles are moist.
The stopper is again removed and a wad of glass wool inserted to
cover the firebrick particles. A one-hole stopper with a glass
tube inserted is then fitted to the test tube. The diethyl ether
is removed by connecting a low-humidity vacuum source to the
glass tube. A section of 0.64 cm stainless steel tubing is
guickly packed with the coated firebrick. The column is then con-
ditioned under 30 ml/min dry helium flow at ambient temperature
for 2 hours.
CLASS ANALYZER
The final configuration of the Class Analyzer is shown in Figures
7 to 9. This breadboard system was designed and constructed to
test the feasibility of the routine analysis of air samples for
hydrocarbon class content.
The Class Analyzer was designed to desorb a sample from the
Universal Collector. The sample is split into six streams contain-
ing selective hydrocarbon class abstractors. These class abstractor-
selectively remove certain hydrocarbon classes from each stream,
allowing the remaining classes to pass into cryogenic traps. The
trapped sample components are volatilized sequentially and swept
into a hydrocarbon detector. The quantities of saturates,
unsaturates, aromatics, and oxycarbons can be calculated from the
detector responses to the effluents from the various class
abstractor streams.
The Class Analyzer was constructed in a Bud Prestige steel cabinet
type C-1552. All components are mounted on the removable 47.5 by
23 cm front panel.
The plumbing consists of 304 stainless steel tubing and Swagelok
brass fittings. Teflon tubing is used to connect the Class Analyzed
to the FID. The main needle valve and all on-off valves are
Whitey valves. The individual stream needle valves are integral
20
-------
H-4 H-3
H-2 H-l
L-l L-2
>
^
<
-
i
>
r
7 NEEDLE ,-
%H VALVES HJ
F
•••
•
*
<: ?
*
ROTAMETERS
4
i
<
h-
>
~1
-
.
»
?
•<
•
"
i-^£ X""1 VALVES h-X XH h-y X~J
)
>
1
^
E
T
-<
>-
p
CRYOGENIC
TRAPS
1
!
j
1 D
— — — —
*^~
l!
\
c
i
-"^
i;c
1
— •
u
/ CONNECT \
_\ FITTINGS '
4
t
j
] TO FID
COLUMN OVEN
A - TENAX GC
B - AC CHARCOAL
C - 40% Hg(C104)2 ON GC-22
D
E
- 10%
LiAlH. ON GC-22
- HgSO ,H2S04 ON GC-22
F - QUICK-CONNECT FITTING
FOR
MA
, He
11
M
^
VALVE5
^^^
1
1
1
i
»
!
-
/'
*
k
\
^^
A
\
^
X1
4
\.
N
7
H
^
H
>
/i
j
f
j
L
-X
T
i
j
i
C
k QUICK-CONNECa
^N
FITTINGS
| | UNIVERSAL
B j COLLECTOR
—
il
1
—
MEASUREMENT OF FLOW TO FID
DESORPTION OVEN
Figure !• Schematic diagram of Class Analyzer
21
-------
Figure 8. Front view of Class Analyzer
22
-------
Figure 9. Rear view of Class Analyzer
-------
parts of the 15 cm Fischer and Porter Lab-Crest 10A1460NB22B
rotameters. These rotameters are equipped with black glass floats
which provide a measurement capability of 0 to approximately 55
ml/min of helium. Swagelok quick-connect fittings are used for
connecting the UC to the Class Analyzer and for connecting the
Class Analyzer to the external FID.
The cryogenic traps are of the "race track" design. The heavy-
ends traps are 0.32 cm 304 stainless steel tubing. In order to
enhance the methane trapping efficiency, the light-ends traps
13
are 0.32 cm copper tubing packed with 60-80 mesh Devarda's metal.
Two specially-built ovens are used in the Class Analyzer. The UC
desorption and class abstractor ovens are heated to 200° and 125°,
respectively, by resistance elements. Oven temperatures are
adjusted by means of standard laboratory autotransformers and
measured with laboratory thermometers. The UC desorption oven has
a hinged door to permit easy attachment and removal of the UC.
The class abstractor oven top is removable for class abstractor
substrate replacement.
LABORATORY TEST STACK
A laboratory test stack was designed and constructed to permit
testing of the modified stack sampler in the lab (see Figure 3).
The stack itself is a 152 by 30 cm section of galvanized pipe.
A short section of 8.9 cm pipe was welded to a hole'in the side
of the stack to serve as a test port.
A high-capacity squirrel cage blower provides stack air flow which
is varied by restricting the blower inlet area. The entire stack
and blower assembly was mounted on a mobile structural frame.
24
-------
SECTION V
EXPERIMENTAL PROCEDURES
UNIVERSAL COLLECTOR EVALUATION
The various UC candidates described in Section IV were evaluated
for their performance characteristics. The UC test apparatus (see
Figure 1) was employed for these evaluations.
The evaluation procedure involved the installation of a candidate
column in the gas chromatograph oven. After column preconditioning,
the temperature of the candidate column was maintained at either 30°
or 0°. The column by-pass valve was opened to permit the carrier
gas (nitrogen) to flow directly to the FID. When a stable recorder
baseline was established, the nitrogen was turned off and a dilute
(ca. 50 ppm) stream of hydrocarbon-containing gas was introduced,
also at 30 ml/min. The detector amplifier attenuation was adjusted
so that the recorder pen position was at a maximum while remaining
on scale. Then the nitrogen flow was restored and the by-pass valve
turned off to force the carrier gas flow through the column.
When a steady recorder baseline was again established, the dilute
hydrocarbon gas was then passed into the column. After the hydro-
carbon was detected by causing a recorder pen deflection, the nitro-
gen flow was restored causing the recorder pen eventually to return
to the baseline. The volume of gas which flowed from the time the
dilute hydrocarbon stream was turned on until the first detectable
pen deflection occurred was the breakthrough volume for that parti-
cular hydrocarbon. Because this evaluation procedure was intended
to establish comparisons between various UC candidates, the system
void volume could be ignored as a common factor.
The desorption volumes were then determined at 200°. The candidate
column was filled to just under its breakthrough volume with a hydro-
carbon sample at 30° or 0°. The sample flow was turned off and the
column heated to 200°. Then the backflush valve was turned to re-
verse the gas flow through the column. The nitrogen carrier gas was
25
-------
turned on. The desorption volume was the volume of carrier gas
which flowed through the column from the time the carrier gas was
turned on until the recorder pen returned to its baseline.
The hydrocarbons used for UC candidate evaluation were approximately
50 ppm mixtures of methane, hexane, benzene, and ethanol, each in
nitrogen. These hydrocarbons represented a range of class types and
boiling points. The determination of the methane breakthrough volume
for the final configuration of the UC was performed by substituting
the UC for the trap in the test apparatus. The column by-pass valve
was opened and the trap valve positioned so that gas flowed through
the UC before entering the FID. The UC was submerged in an ice bath
and various methane mixtures (ca. 25-2500 ppm) introduced into the
system. The methane breakthrough volume was then determined as pre-
viously described.
BREAKTHROUGH DETECTOR EVALUATION
The Laboratory Data Control, Inc. piezoelectric detector was tested
for its sensitivity to methane. The test procedure involved estab-
lishing a steady baseline flowing 30 ml/min of nitrogen through the
detector. Then the nitrogen was turned off and the methane mixture
turned on also at 30 ml/min. A standard 45 ppm mixture of hexane in
nitrogen was also used as a test gas.
CLASS ABSTRACTOR CANDIDATE EVALUATION
Early class abstractor evaluation was performed by merely installing
a candidate column in a Perkin-Elmer 880 gas chromatograph and in-
jecting a series of model compounds into the column to determine
whether abstraction occurred. The test apparatus was modified for
later candidate evaluation to allow the detection of partial abstrac-
tion situations (Figure 10). Comparing the detector response to a
model compound first injected into an open tube leading to the detec-
tor followed by an identical injection into the abstractor candidate
column gave a better indication of abstractor performance. Whenever
possible, the detector response was integrated by an Autolab System
I computing integrator.
26
-------
FLOW
CONTROLLERS
ABSTRACTOR
He
FID
INTEGRATOR-
RECORDER
INJECTORS
OVEN
Figure 10. Schematic diagram of class abstractor test apparatus
27
-------
Table 1 is a list of model compounds which were used for most
of the abstractor candidate evaluations. For each class, repre-
sentatives of low and higher boiling compounds were selected. In
some cases, additional compounds were used as models to better
define the performance of a particular abstractor candidate.
CLASS ANALYZER EVALUATION
The performance of the Class Analyzer was evaluated using known
samples. With the UC by-pass (Figure 11) connected to the Class
Analyzer, samples were injected with a syringe.
In order to measure sample recovery percentages and abstraction
efficiency, the effluent from the Class Analyzer was directed to
the FID in an Aerograph Hy-FI Model 600-B gas chromatograph. The
FID was also connected to the chromatograph injector by an open
tube. This arrangement permitted the injection of a sample directly
into the FID so that a specific detector response could be deter-
mined. An identical injection into the Class Analyzer produced a
detector response which could be compared with the specific response
measured previously. Any reduction in response could be attributed
to sample loss. Whenever possible, the detector response was inte-
grated by an Autolab System I computing integrator.
LABORATORY TEST STACK
The laboratory test stack was used to introduce synthetic samples
to the stack sampler. Gaseous samples were introduced directly into
the blower inlet at a controlled flow rate. Liquid samples were
infused into a glass fiber wick placed in the blower inlet. An ad-
justable speed Sage infusion pump drove a 10 ml syringe containing
the sample. The air stream flowing over the wick evaporated the
sample. The air flow was adjusted to about 6700 1/min by restricting
2
the blower intake area to 64 cm .
The usual procedure observed involved positioning the stack sampler
inlet nozzle in the center of the stack diameter. The stack blower
and stack sampler pump were turned on and sample introduction begun.
28
-------
Table 1. MODEL COMPOUNDS FOR CLASS ABSTRACTOR EVALUATION
Class
Model compounds
Saturate
Unsaturate
Aromatic
Halocarbon
Oxycarbon
Methane, hexane
Ethene, 1-octene
Benzene, 1,2,4-trimethylbenzene
Chloroform, 1,1,2,2,-tetrachloro-
ethane
Methanol, 4-methylbutanol
Acetaldehyde, 2-ethylhexanal
Acetone, cyclohexanone
Ethyl acetate, isopentyl acetate
Diethyl ether, tetrahydrofuran
29
-------
I//'/i;'fV
-------
An equilibration period was allowed before sample collection began
to ensure the most constant sample concentration possible. Grab
samples were collected immediately before and immediately after
sampling. The contents of the stainless steel bombs used for grab
sampling were analyzed by gas chromatography in order to character-
ize the sample.
The samples, which were collected in 6 UC's, were then analyzed by
the Class Analyzer. Multiple samples were collected to determine
the analytical precision. The accuracy of the sampling and analyti-
cal methods was determined by comparing the results with those ob-
tained from the gas chromatographic analysis of the grab samples.
MODIFIED STACK SAMPLER AND CLASS ANALYZER
The operating procedures for the modified stack sampler and the Class
Analyzer are described in their respective Operation Manuals.
31
-------
SECTION VI
DISCUSSION
LITERATURE SURVEY
An extensive literature survey was completed prior to the begin-
ning of the technical effort. The subjects sought were based on
two-step sampler-analyzer design. The first step provides for the
collection of total organic emissions while the second step pro-
vides for the separation of the collected sample into chemical
classes.
Because the total collection of the sample must be nondestructive,
emphasis was placed on obtaining information on available solvents
whose adsorption mechanism is based on physical processes.
The separation of the collected sample into chemical classes, on
the other hand, needs not exclude selective sample destruction.
Hence, in seeking unique means of trapping a specific chemical
class without collecting other chemical species, both physical and
chemical means were investigated. Again, sorbents were sought as
class abstractors. The headings of chemisorption and analytical
means of determining specific classes of compounds were searched.
Principal interest in specific class determinations was focused
on saturates, olefins, aromatics, halocarbons, alcohols, ketones,
aldehydes, esters, ethers, mercaptans, and carboxylic acids.
The pursuit of pertinent information followed several pathways.
For current journal publications (1971 to present), a computer
search was requested through NASA's Western Research Application
Center (WESRAC) at the University of Southern California. This
search was based on key word indexing of Chemical Abstracts. An
additional search, covering principally government publications,
was requested through the Commerce Department's National Technical
Information Service (NTIS) facilities. However, NTIS felt that thei*
survey would not be productive because of our subject matter, so the
survey request was cancelled.
32
-------
Our own personnel conducted a survey of literature up to 1971
by use of Chemical Abstracts cumulative indices and Air Pollution
Abstracts. Because cumulative indices were available through
1971, computerized searching of earlier titles would have been
less economical than a personal one.
The literature survey produced about 80 pertinent references.
These references were obtained and studied to determine their
applicability to the present effort.
UNIVERSAL COLLECTOR
Ideally, the Universal Collector (UC) is a device which will quan-
titatively collect all volatile organic compounds from a stationary
source. The UC will store these collected organic compounds without
sample decomposition. The sample can then be ouantitatively released
from the UC with no contamination added to the sample by the UC itself,
Finally, the UC should be a compact device.
Past experience and information from the literature survey indicated
that no single sorbent could fulfill the reouirements of an ideal
UC. Typical gas chromatographic sorbents do not have sufficient
capacity for the more volatile organic compounds (light ends), such
as methane. Highly active adsorbents such as activated charcoal
have a high capacity for the light ends but the less volatile organic
compounds (heavy ends) are irreversibly adsorbed. Therefore, the
envisioned UC was a dual-bed unit: an active adsorbent to remove the
light ends preceded by a sorbent whose purpose is to prevent the
heavy ends from reaching the more active adsorbent.
The evaluation procedure for the various UC sorbent candidates approxi-
mated the proposed operating conditions for the UC. Sample collection
(sorption) was to occur at ambient or ice-bath temperature. Sample
release (desorption) was to occur at an elevated temperature (ca.
200°C) to ensure rapid and complete desorption.
Past experience and the results of the literature survey indicated
the existence of a variety of suitable candidates for the heavy ends
33
-------
sorbent. The relatively stable, inert surface of a porous organic
polymer seemed to be ideal.
By far the most common type of porous polymer sorbent is a cross-
linked polymer produced, for example, by copolymerizing styrene
and divinyl benzene. This group is represented by the Porapak
series (Waters Associates) and the Chromosorb Century series
(Johns-Manville Corp.). The different properties of members of
each series are due to variations in pore size, surface area, and
chemically-bound functional groups. Porapak Q was chosen as a
representative of that sorbent type because it is widely used as
a gas chromatographic support for the analysis of hydrocarbon
mixtures.
The sample capacity of these porous polymer supports can usually be
increased by coating the surface with a suitable chromatographic
liauid phase. This liquid phase should have a negligible vapor
pressure to prevent the contamination of the collected sample by
the liquid phase. A liquid phase which has the necessary proper-
ties is OV-1, a methyl silicone gum with a recommended maximum
temperature limit of 350°C. A 3% coating (by weight) was applied
on 80-100 mesh Porapak Q.
A porous polymer that is significantly different from both the
Porapak and Chromosorb Century series is Tenax GC. This support
is a polymer based on 2,6-diphenyl-p-phenylene oxide. It is more
thermally stable than any members of the Porapak or Chromosorb
Century series. No significant volatilization occurs below 320°.
Zlatkis has demonstrated its usefulness in trapping volatile con-
stituents of urine, human breath, automobile exhaust, ambient air,
^ a. 14
and etc.
The results of the experimental evaluation of the heavy end sor-
bents are summarized in Table 2. Coating Porapak Q with OV-1 does
indeed increase the capacity of the sorbent. However, both the
retention and desorption volumes are increased. The disadvantage
of an increased desorption volume largely cancels the increased
34
-------
Table 2. CANDIDATES FOR HEAVY ENDS SORBENT OF UNIVERSAL COLLECTOR
to
en
So r bent
Porapak Q
3% OV-1 on
Porapak Q
Tenax GC
Sorbate
Methane
Hexane
Benzene
Ethanol
Methane
Hexane
Benzene
Ethanol
Methane
Ethane
Propene
Butane
Specific retention
volume (temperature),
ml/q (6C)
6.2
161.
>6550.
193,
>8510.
213.
> 7340.
5.7
201.
231.
> 6000.
260.
30.
79.
578.
167.
(30)
(100)
(30)
(100)
(30)
(100)
(30)
(30)
(100)
(100)
(30)
(100)
(0)
(0)
(0)
(0)
Specific desorption
volume at 200°C,
ml/q
—
43
-
50
-
65
-
_
108
101
-
92
-
-
-
-
-------
retention volume. The possibility of sample contamination by
the OV-1 coating makes this support somewhat inferior to un-
coated Porapak Q.
Inspection of Porapak Q after repeated temperature cycling from
ambient to 200 revealed a change in its color from white to
brown. Although its sorptive properties did not appear to de-
teriorate, its long-term performance was subject to suspicion.
Tenax GC performed well in all tests. It has sufficient capacity
to prevent heavy ends from reaching the light ends sorbent. Its
thermal stability seems to be excellent with no signs of deterior-
ation after several months' usage. Therefore, Tenax GC was selected
as the heavy ends sorbent.
Selecting the light ends sorbent was a more difficult job than the
heavy ends sorbent. The quantitative retention of lower hydrocar-
bons, such as methane, requires an extremely active surface. Normal
gas chromatographic supports are not acceptable. Past experience
with aluminas and silicas indicated that these sorbents would be
inferior to charcoals and molecular sieves for the retention of
methane.
The eight charcoals and molecular sieves shown in Table 3 were
selected for experimental evaluation because of past experience
with them or because of advantageous properties claimed for them.
Most of these materials are not designed for chromatographic appli-
cations. They are manufactured as pellets which required grinding
and sieving to produce the proper size particles (ca. 60-100 mesh).
The Linde Molecular Sieves are synthetic zeolites. The numerical
designation (4A, 5A) is related to the approximate diameter (in
angstroms) of the pores in the open zeolite structure. The AW-500
molecular sieve has been treated by acid washing.
Barnebey-Cheney AC charcoal has been used by NASA in the air puri-
fication systems of manned spacecraft. Both the AC and GI type
charcoals have demonstrated exceptionally high capacities for
36
-------
Table 3. CANDIDATES FOR LIGHT ENDS
SORBENT OF UNIVERSAL COLLECTOR
Sorbent type
Sorbent
Specific retention
volume of
methane at 30°C,
ml/q
Molecular sieve
Activated carbon
Linde AW-500
Linde 4A
Linde 5A
Barnebey-Cheney AC
Barnebey-Cheney GI
Nuchar WV-H
Cabot Graphon
Carbosieve-B
5.2
3.2
4.9
85.
46.
6.8
6.7
4.5
37
-------
organic compounds. Nuchar WV-H is a controlled porosity charcoal
whose capacity for specific organic compounds was unknown to the
project. Graphon is a charcoal with a highly uniform, graphitized
surface. Carbosieve-B is a controlled porosity charcoal which is
claimed to have properties similar to molecular sieves.
Methane retention at 30° was determined by introducing a 50 ppm
mixture of methane in nitrogen into a 20 by 0.64 cm O.D. stain-
less steel tube filled with the test sorbent. The instant that
methane breakthrough was detected determined the retention volume.
The specific retention volumes were calculated by dividing the
actual retention'volumes by the weight of sorbent used for each
test. Barnebey-Cheney AC charcoal clearly had the highest speci-
fic retention volume.
The two preferred sorbents, Tenax GC and AC charcoal, were arranged
in series for sample collection (Figure 12a). However, special
requirements for the separation and cryogenic trapping of the light
ends necessitated keeping the light and heavy ends separated during
sample recovery. Thus, a third port was installed in the UC for the
introduction of a carrier gas during sample recovery (Figure 12b).
The capacity of the UC was of critical importance. Sensitivity of
the analytical technique would suffer if the capacity were too low.
Maintaining the UC at 0° during sample collection and incorporating
the greatest practical amount of sorbent would maximize the capacity-
Packing each sorbent in a 20 by 0.95 cm O.D. stainless steel tube
results in sufficient capacity. This permits about 2.8 g of Tenax
GC and 6.6 g of AC charcoal to be used in each UC. Methane break-
through does not occur until over 400 ml of sample have passed
through the UC.
A convenient means of attaching the UC to the stack sampler and the
Class Analyzer required careful consideration. The UC ports must
be closed when the UC is not connected to either the stack sampler
or the Class Analyzer. This prevents sample contamination or loss.
38
-------
SAMPLE
INLET
OUTLET
HEAVY CARRIER LIGHT
ENDS GAS ENDS
OUTLET INLET OUTLET
CLOSED
/t\ A
QUICK-
CONNECT
FITTINGS
B
/\ A\
A
! B
a. Sample collection
b. Sample recovery
A - TENAX GC
B - AC CHARCOAL
Figure 12. Schematic diagram of Universal Collector
39
-------
Initially, a threaded flange was designed and made as the con-
nector for the UC. This flange was somewhat bulky and expensive.
It did not provide automatic shut-off of the UC ports.
The next connector design tested incorporated individual double-
ended automatic shut-off quick connect fittings. Two disadvan-
tages of this system do exist. Some sample loss or contamination
can result due to the Buna-N 0-rings which are used in the fittings.
Also, perfect alignment of.the three fittings on each UC with the
fittings on the stack sampler and Class Analyzer is difficult.
This causes more Difficulty in connecting the UC than is desirable.
However, the system does work and is the best alternative found to
the present time.
STACK SAMPLER
The modifications made to the Joy Manufacturing Company Method 5
stack sampler have converted a particulate sampler into a sampler
capable of collecting volatile hydrocarbons as well as particulates
in stack gases. The UC capacity and the hydrocarbon sample train
flow rate capabilities permit a 400 ml stack gas sample to be col-
lected in a minimum of 7 minutes and a maximum of about 230 minutes.
Replacement of a saturated UC with an empty one requires less than
15 seconds, so the actual maximum sampling period depends on the
number of UC's available for sample collection. Therefore, both
short-term and long-term sampling can be consecutively performed
during the same test-period on site.
One possible problem might occur during the sampling of hot, humid
stack gases. Condensation of water vapor in the UC could occur.
Excess moisture can cause a loss of hydrocarbon capacity. A des-
iccant tube has therefore been installed upstream of the UC in the
filter oven. The literature indicates that anhydrous potassium
carbonate is a desiccant which does not abstract most hydrocar-
bons. ' However, the project staff did not have an opportunity
to test potassium carbonate for its inertness toward hydrocarbons
and its efficiency as a desiccant at the 120°C temperature of the
filter oven. Therefore, further work needs to be done in this
40
-------
Incorporation of a breakthrough detector downstream of the UC in
the hydrocarbon sample train was planned. The purpose of this
detector was to alert the operator when any hydrocarbon eluted
from the UC.
The performance characteristics required of the breakthrough
detector were quite demanding. The detector had to be capable
of detecting methane in the low ppm range. Inorganic compounds
would have to be nonresponsive. It had to be compact, rugged,
and capable of operating in the field.
A flame ionization detector (FID) has the necessary sensitivity
properties. Operation of a FID in the partial vacuum of the stack
sampler train would be inconvenient. Providing sources of pure
hydrogen and air would be difficult. A small, portable hydrogen
generator could not be found.
A thermal conductivity detector (TCD) is compact, portable, and
rugged, but detects inorganic as well as organic compounds. Its
sensitivity is greatest when a carrier gas with a thermal conduc-
tivity greatly different from the sample compounds is used. Air
is a poor carrier gas in this respect.
International Sensor Technology of Costa Mesa, California, devel-
oped a solid state electrolytic cell hydrocarbon gas sensor. The
conductivity of the solid electrolyte changes when hydrocarbon
gases are adsorbed on the electrolyte surface. Unfortunately, gas
flow over the electrolyte cannot be tolerated. Diffusion of the
sample to the detector is the reauired mode of operation. A major
drawback of this device was its ultimate sensitivity to methane.
Response to 100 ppm methane was assured, but no experience with
lower levels was available.
Laboratory Data Control (LDC) of Riviera Beach, Florida, has devel-
oped a piezoelectric detector. The detector is a quartz crystal
oscillating at about 10 MHz. A thin layer of a suitable gas chro-
ma tographic liquid phase is coated on the crystal. When a sample
41
-------
passes over the coated crystal, organic components dissolve in
the liquid coating increasing the crystal mass and lowering its
oscillating frequency. Because this is a mass detector, heavier
compounds are more easily detected. Nevertheless, LDC saw no
problems meeting our sensitivity requirements.
Because the piezoelectric detector seemed to be the most promising
candidate breakthrough detector available, one was purchased as
part of a complete, although inexpensive gas chromatograph. Opera-
tion of the detector on either 115 VAC or 12 VDC did not affect its
performance. Operation of the detector at 0° rather than ambient
temperature did not cause any problems and may have even improved
its sensitivity somewhat. Sensitivity was further increased by
using a 1 mv rather than a 10 mv recorder. However, the increase
in background noise partially nullified the sensitivity increase.
Initially, the crystal was coated with Carbowax 400, a polyethylene
glycol. A mixture of 45 ppm hexane in nitrogen was detectable, but
not without operating at the lower attenuation settings. More sen-
sitivity would be required for the detection of low levels of methane*
During experimentation, the crystal oscillation ceased, probably due
to irreversible crystal surface contamination. The crystal was
cleaned and recoated with Carbowax 400. Obtaining the proper coat-
ing thickness was a trial-and-error process. The crystal cracked
before an optimum coating layer could be deposited.
A new crystal was obtained from LDC. In order to improve the detec-
tor sensitivity toward methane, LDC suggested a crystal coated with
SE-30, a silicone gum rubber. This new detector was thermally in-
sulated to minimize baseline drift.
With the new crystal, a 45 ppm mixture of hexane in nitrogen (mixed
and analyzed by The Matheson Company) flowing at 30 ml/min at 25°
produced at 0.49 mv signal at an attenuation of 16. According to
theory, the piezoelectric detector response is proportional to con-
centration, molecular weight, and the partition coefficient of the
sample in the crystal coating. Based on empirical chromatographic
42
-------
data, the detector response to a 45 ppm methane mixture should
have been about 650 times less than for a comparable hexane sam-
ple. The resulting signal, 0.01? mv at an attenuation of 1, is
less than the noise level. The response of the detector to syn-
thetic mixtures of methane in nitrogen in the 30 to 100 ppm range
was independent of the methane concentration.
In a continuing effort to gain the required sensitivity, LDC
factory technical representatives and Dr. David Bonner, a piezo-
electric detector specialist at Texas Tech University, were con-
sulted. Detection of hydrocarbons lighter than hexane has not
been investigated. Therefore, we concluded that a considerable
research effort would be required to obtain sufficient methane
sensitivity. It was decided, therefore, to eliminate the piezo-
electric detector from consideration as a possible breakthrough
detector during the present effort.
Our research direction was accordingly changed to determine
whether a breakthrough detector was really renuired. The break-
through volumes of methane samples ranging from 25 to ?. 500 ppm
were determined. The breakthrough volume from the UC at 0° and
50 ml/min flow rate remained constant at about 425 ml. Therefore,
if sample sizes are kept below 400 ml, no breakthrough should occur
under any circumstances. This is true because the UC operates by
a reversible adsorption mechanism. Therefore, the breakthrough
volume is unaffected by past exposure to organic molecules because
essentially all molecules will eventually elute. By establishing
a set of sampling conditions, premature breakthrough can thus be
avoided.
CLASS ABSTRACTORS
The central goal of this research effort is to develop a method
of analyzing a gaseous sample for its content of saturates, un-
saturates, aromatics, halocarbons, and oxycarbons. The approach
we have chosen is to assemble a series of chromatographic columns
which selectively abstract certain hydrocarbon classes.
43
-------
The ideal class abstractor column quantitatively abstracts a
class (or classes) of hydrocarbons. Those classes which are
not abstracted should pass through the column without being
retained. Because the abstracted compounds accumulate in the
column, a simple regeneration of the column would be ideal.
Table 4 lists the 17 different abstractor candidates which were
evaluated. Most of these substrates consist of materials coated
on solid supports. With the exception of the Molecular Sieve
(synthetic zeolites), the Porasil (porous silica), and the alumina,
all supports are some form of calcined diatomaceous earth. The
GC-22 and Chromosorb P are diatomaceous earths which have been cal-
cined without a flux. They generally have a larger surface area
and are more adsorptive and efficient than those diatomaceous
earths which have been calcined with a sodium carbonate flux.
Gas-Chrom P is flux calcined, washed with acid to remove inorganic
impurities, washed with base to remove organic impurities, and
washed with water until neutral. Anakrom ABS is similar to Gas-
Chrom P except that it has been deactivated and made hydrophobic
by silanization with dimethylchlorosilane.
These treatment procedures are designed to improve chromatographic
peak shape., resolution, etc. Such considerations are unimportant
for a class abstractor. Therefore, the results would probably
remain identical had all diatomaceous earth supports been the
same. The use of different supports merely reflects the prefer-
ence of each research group that published its work with a certain
substrate.
A substrate designed to abstract unsaturates and aromatics, Ag2S04»
H^SO on Gas-Chrom P , was unsatisfactory for use in the Class
Analyzer. Oxycarbons and certain unsaturates were abstracted, but
aromatics were merely retained strongly at 30°. Two separate sub-
strate preparations were made with no appreciable difference in
performance.
An unsaturate abstractor, HgSO , H2SC>4 on GC-22 , abstracted all
unsaturates tested with the exception of ethene. Oxycarbons were
44
-------
Table 4. CLASS ABSTRACTOR CANDIDATES
Ag2SO4, H?SO4 on Gas-Chrom P
HgS04, H9S04 on GC-22
Hg(C104)2, HC104 on Gas-Chrom P
40% Hg(ClO4)0 on GC-22
10% 2,3-dichloro-5,6-dicyanobenzoquinone (DDQ)
on nnakrom ABS
DDQ, di(2-ethylhexyl)sebacate on Anakrom ABS
5A Molecular Sieve
10% Cr2
-------
also abstracted but all other classes tested passed through the
column quantitatively at 30 . In an effort to minimize retention
volumes for those classes not abstracted, the column was heated
to 125 . Apparent dehydration occurred which seriously degraded
the column's performance. Therefore, this abstractor candidate
can be used as a heavy ends unsaturate and oxycarbon abstractor.
2-4
Another unsaturate abstractor, Hg(C10.)2, HC10. on Gas-Chrom P ,
did not Quantitatively abstract ethene, propene, and pro-
pyne at 100°. The higher unsaturates, as well as most oxycarbons,
were quantitatively abstracted. However, some loss of aromatics
occurred as well. The mixed behavior of this substrate makes it
a poor candidate for use in the Class Analyzer.
Coating GC-22 with 40% Hg(ClO4)2 produced a substrate which quan-
titatively abstracted all unsaturates, aromatics, and oxycarbons
tested. The saturates were only slightly retained at 125°. Some
loss of halocarbons occurred, however. With the exception of halo-
carbons, this substrate behaved well enough to be used as a light
ends unsaturate abstractor and a heavy ends abstractor for unsatur-
ates, aromatics, and oxycarbons.
The charge transfer complex formed between lower aromatics and 2,3-
dichloro-5,6-dicyanobenzoquinone (DDQ) was too weak to cause ab-
straction of aromatics. The behavior was similar when DDQ was eithe*"
coated" directly on Anakrom ABS or dissolved in di( 2-ethylhexyl)
sebacate before coating. DDQ is temperature sensitive, so both
columns were tested at 30° initially. When no aromatic abstraction
occurred, the temperature was raised to 75 for the liquid solution
coating and 225° for the solid coating. The only change was partis-1-
decomposition of DDQ. Neither column abstracted any hydrocarbon
class, so no further evaluation was performed.
Molecular Sieves have been used for selective hydrocarbon separa-
7 R
tions by several investigators. ' However, our experience with
5A Molecular Sieve was that quantitative separations were extremely
46
-------
difficult over a temperature range from 30° to 200°. Almost
every hydrocarbon tested was partially abstracted. Such per-
formance does not meet the criteria of an ideal class abstractor.
Salt-coated silica and alumina have shown promise as being selec-
9 10
tive hydrocarbon class abstractors. ' Four columns were made:
10% Cr0(S04). on Porasil C, 10% Na2S04 on Porasil C, 20% NaBr on
F-20 alumina, and 20% NaOH on F-20 alumina. In terms of their
performance as class abstractors, all columns performed similarly.
Oxycarbons were abstracted but all other classes were strongly
retained, even at 200°. This strong retention greatly increases
the time reauired for analysis. Attempts to deactivate Porasil
C and alumina with water injected into the columns had no bene-
ficial effect. The conclusion is that these adsorbents are much
too active for use in the Class Analyzer and no suitable deactiva-
tion procedure could be found.
Boric acid reacts with most alcohols to form non-volatile borate
esters. This reaction has been utilized in designing alcohol
11 1 ?
abstractors. ' " A substrate of 3% boric acid and 20% Carbowax
20M on Chromosorb P merely retained alcohols at 175°. This reten-
tion was strong enough to permit a separation of alcohols from
other hydrocarbon classes tested, but such performance is clearly
inferior to total abstraction.
Because partially dehydrated boric acid is thought to be the re-
actant, a 3% boric anhydride and 20% Carbowax 20M substrate was
prepared. However, all classes tested passed through the column
with no abstraction and very weak retention occurring. Obviously,
totally dehydrated boric acid (boric anhydride) is unreactive
toward alcohols.
Preconditioning a 5% boric acid on GC-22 column at 300° permitted
substantial, but not complete dehydration to occur. The resulting
column abstracted primary and secondary alcohols quantitatively at
125°. Tertiary alcohols appeared to be dehydrated to alkenes
rather than being abstracted. All other classes pass through the
47
-------
column auite rapidly with no apparent losses. This column was
not incorporated into the Class Analyzer only because a better
column was subsequently found.
The reducing agents, LiBH. and LiAlH., react with oxycarbons to
form non-volatile complexes. L,iBH. is more thermally stable than
LiAlH., but is more hygroscopic and difficult to handle. Its per-
formance as an oxycarbon abstractor is inferior also. At tempera-
tures from 100° to 200°, LiBH. abstracts only higher oxycarbons such
as cyclohexanone, 4-methylbutanol, and 2-ethylhexanol. Lower oxy-
carbons are not abstracted at all.
LiAlH. is reported to decompose from 130 to 187 , depending on
the source of information. At 130 , LiAlH. abstracts all oxycarbons
tested initially, but after about 40 hours of operation at 130°,
acetone and ethyl acetate are not quantitatively abstracted. This
may be due to decomposition. However, operation at 120° results
in no appreciable acetone abstraction, so the operating temperature
must be kept near 130 for ideal performance. All other classes
unaffected by LiAlH..
A problem with using LiAlH. powder as a substrate is its high back
pressure and propensity for clogging. Coating a chromatographic
support with LiAlH. was attempted by dissolving it in diethyl
However, the moisture present in the solvent and the atmosphere
rapidly diactivates LiAlH4 and forms an insoluble precipitate. A
source of 1.3M LiAlH4 in diethyl ether was found. This made
GC-22 with a 10% layer of LiAlH. possible. This procedure produces
a substrate which does not cause any gas flow problems.
An additional problem with LiAlH. is its reactivity with moisture
in the sample. Not only does the moisture deactivate LiAlH , but
the reaction appears to release previously abstracted oxycarbons.
This problem was discovered only after a LiAIH. column had been
installed in the Class Analyzer. A solution to the problem would
be the installation of a desiccant in the stack sampler and/or
immediately upstream of the LiAlH. column. Further research is
necessary to optimize this system.
48
-------
CLASS ANALYZER
The final configuration of the Class Analyzer is shown in Figure
7. The selection of abstractor columns was based on their per-
formance during the evaluation phase.
Testing of the Class Analyzer was conducted by making syringe in-
jections of model compounds into the Class Analyzer. This was
made possible by connecting the UC by-pass (with its septum) to
the Class Analyzer.
Testing revealed several problems. Initially, sample losses ex-
ceeded 50%. The cause of these excessive losses was a series of
small but troublesome leaks in the swaged tubing connections.
Correction of these leaks eliminated all but roughly 10% of the
losses. These losses can only be attributed to wall losses within
the system itself. By enclosing as much of the system as possible
in a heated compartment, these losses should be minimized.
Tests involving cryogenic trapping revealed a background level of
hydrocarbons in the system which would prohibit the detection of
hydrocarbons in a sample below about 0.1 ppm. The source of this
background is probably those hydrocarbons which are slowly being
released from the walls of the system tubing. Carrier gas con-
tamination was minimized by passing 99.996% helium through a
Molecular Sieve and charcoal trap at -196°. Another source of
background hydrocarbons is from the LiAlH. column. Moisture in
the sample can release previously abstracted oxycarbons.
The loss of trapped methane was a concern. Devarda's metal in the
cryogenic trap minimizes this loss. Because of methane's sub-
stantial vapor pressure even as a solid at -196 , the amount of
methane lost depends on the volume of carrier gas which flows
through the trap. Our tests indicate that very little methane is
lost during the trapping phase, when 300 ml of carrier are passed
through the trap. Methane is apparently displaced along the length
of the trap as carrier gas flows through the trap. As long as the
trap is sufficiently long, little methane should be lost. Quanti-
tative studies have not been performed.
49
-------
TOTAL SYSTEM TESTING
Testing of the entire sampling-analysis system was attempted.
Ideally, actual samples from a variety of industrial operations
were to be collected.
Field testing locations were sought at the beginning of the pro-
gram. The Project Officer suggested that early arrangements be
made to avoid later complications. A list of over 75 potential
field test sites was compiled from suggestions made by the Los
Angeles Air Pollution Control District, from ARLI clients, and
from companies listed in the Telephone Directory. These test
sites included chemical manufacturing plants, paint and varnish
works, degreasers, solvent reclaimers, printing operations,'petro-
leum refineries, and power plants.
The most serious difficulty in finding suitable test sites was the
problem of using the Method 5 stack sampler for sampling. The
Method 5 equipment is quite heavy and bulky. Substantial support
must be provided to make sampling physically possible and to pro-
tect the delicate glassware in the sampling unit. The support rail
of the Joy Method 5 sampler attaches to a clamp which fits over a
standard 7.6 cm (3 in.) test port. In all the many contacts made
with representatives of potential test sites, a stack equipped with
a standard test port was never found.
The second problem was the lack of cooperation of many company
executives. Those at the larger plants having high capacity stacks
suitable for sampling were not agreeable to having EPA contractors
sampling their emissions.
Some potential test sites did not emit hydrocarbons that were suit-
able for our sampling needs. Freouently, the emissions contained
only one or two components, rather than the variety that was so
In several instances, sampling access was arranged only to be post-
poned indefinitely or cancelled due to alleged breakdowns, changes
in the company's attitude toward the agreed-upon arrangements etc.
50
-------
In order to test the sampler, small companies whose emissions
emanated from small roof vents, had to be relied upon. An un-
heated Pyrex glass sample probe was fabricated to replace the
long stack sampler pitobe for this type of sampling.
When the difficulties of finding suitable test sites became
critical, a laboratory test stack was constructed. The test
stack permitted the operation of the Method 5 sampler in its
intended mode. Also, a variety of different types of samples
could be prepared for testing. Unfortunately, use of this
simulator was not initiated until late in the program, so that
only a limited number of tests could be undertaken.
Our first test mixture (EPA-1) consisted of an equimolar mixture
of hexane, 1-hexene, benzene, and methanol. This liquid mixture
was infused into the test stack at the rate of about 0.38 ml/min.
The test stack air flow was about 6700 1/min. This provided a
stack sample containing approximately 14 ppm total hydrocarbons
in addition to the comparatively minor contribution from the
ambient air.
During the actual test sampling, the stack gas temperature and
relative humidity were 24.5*0.5° and 50± 5%, respectively. The
pitobe and filter oven were maintained at 125 - 10°. The Universal
Collector and the Method 5 impingers were kept at 0°. The total
sampling rate through the Method 5 train was 28 1/min with 55 ml/
min passing through the Universal Collector. Each Universal Col-
lector was in operation for 7 min 16 sec resulting in a sample
volume of 400 ml. The pitobe inlet was positioned at the center
of the stack for all six Universal Collectors.
Prior to installing the first Universal Collector, the stack
blower, the syringe infusion pump, and the Method 5 sampler pump
were activated to permit a stable equilibrium condition to be
established. As the first Universal Collector was installed, a
grab sample was collected in an evacuated stainless steel bomb .
as closed to the pitobe inlet as possible.
51
-------
Gas chromatographic analysis of the grab samples indicated
that the sample composition was not uniform during the sam-
pling period. Probable causes were non-uniformities in the sample
volatilization and dispersion in the stack gas stream.
The results of the analysis of sample EPA-1 are tabulated in
Table 5. The data for two of the collectors are not shown.
The contents of UC 6 were lost because of problems in operating
the Class Analyzer. Pressure surges during valve actuation
caused FID extinguishment. Plumbing modifications alleviated
this problem. UC 2 apparently leaked during the storage
period between sampling and analysis. The automatic shut-off,
quick-disconnect fittings can leak if the end of the fitting is
placed against a flat surface.
Operator error caused FID extinguishment during the analysis of
UC 5. The entire sample was not lost, however, and those data
that were obtained are shown.
The method of calculation is as follows:
CS = (PL-2 + PH-2)RS/V
CU = (PL-1 - PL-2 + PH-3 - PH-4)RU/V
t
-P )R/v
; /V
C - ( P ' - P ) R /V
U0 ~ V H-l H-3; 0'
P ! = P . ( F /F . )
i x ave i
3
where C = concentration of a specific hydrocarbon class, mg/m
2
P = raw peak area, cm
i 2
P = corrected peak area, cm
6 2
R = detector response factor, 10 mg/cm
V = sample volume, ml
F = flow rate, ml/min
S = saturate class
U = unsaturate class
A = aromatic class
52
-------
Table 5. CLASS ANALYZER ANALYSIS OF SAMPLE EPA-1
uc
1
3
4
5
Hexane
3.75
3.67
3.74
3.22
Concentrat:
1-Hexene
6.43
6. 32
9.33
-
.on of compor
Benzene
0.46
3.98
2.47
5.11
3
lent, mg/m
Methanol
10.20
-0.23
-0.90
-
Total
20.84
13.74
14.64
-
53
-------
O = oxycarbon class
L = light end stream
H - heavy end stream
The precision of the saturate analysis (represented by hexane)
was reasonably good. This is partly due to the mode of sample
concentration calculation. When no light end components are
present, the saturate calculation is the only calculation which
does not require two peak area determinations and the inherent
increase in error. The saturate determination is calculated
directly from the effluent of stream H-2.
The aromatic and oxycarbon data are clearly not satisfactory.
This is probably due to moisture in the sample which causes the
LiAlH column to elute oxycarbons which had been abstracted
earlier. The same LiAlH. column is in both streams H-3 and H-4,
thus causing erratic results in both the aromatic and oxycarbon
determinations. The unsaturate determination is fairly good.
Here the difference in effluent from H-3 and H-4 is involved in
the unsaturate determination, so that the above-mentioned error
cancels.
Modifications were made to the stack sampler, the Class Analyzer,
and the sampling and analytical procedures as a result of the
experience with sample EPA-1. After these modifications had been
completed, sample EPA-2, a 24-hour ambient air sample, was col-
lected.
The 24-hour ambient air sample was collected continuously on 7
and 8 November 1974. The lowest flow rate attainable through the
hydrocarbon sample train at present is 2 ml/min. Using the sample
size limitation of 400 ml, only 3 hr 20 min of continuous sampling
is possible. With only 6 UC's available and about 10 minutes
allowed for a change of UC's, only 21 hours of sampling could be
accomplished. The decision was made to increase the sample size
limit to permit a full 24 hour sampling period at the possible
of losing a small amount of methane.
54
-------
Each UC collected about 454 ml of sample over a 3 hour 47
minute period. Allowing about 13 minutes for a change of UC's,
the total sampling period was 24 hours. The meteorological con-
ditions varied from 6 C and 86% relative humidity to 26 C and
32% relative humidity. The air was clear and still during the
entire sampling period. The Los Angeles Air Pollution Control
District reported maxima of 0.08 - 0.09 ppm 03, 13-15 ppm CO,
and 0.12 - 0.36 ppm NO during the sampling period.
X
The results of the analysis are tabulated in Table 6. An integra-
tion error prevented the determination of the FID response which is
used to calculate the light end unsaturate concentration in UC 1.
Otherwise, the analysis went smoothly. The saturate levels
followed a pattern that parallels the level of automobile traffic
in the Los Angeles Basin. The saturate levels are at a minimum
during the late night and early morning hours (UC's 4 and 5) and
at a maximum during daylight hours (UC's 1, 2, and 6). The light
unsaturates and the aromatic levels appear to be reasonable. The
oxycarbon and heavy unsaturate levels reflect too much effluent
from stream H-3. Again, water vapor in the air sample is probably
responsible for this difficulty.
Typical examples of the output signal from the Class Analyzer are
shown in Figure 13. The ideal, sharp, Gaussian peak shape does
not occur in all cases. The multiple or broadened peaks are
doubtless due to fractionation of the sample as it is being vapor-
ized from the cryogenic traps. More rapid vaporization would
probably improve the peak shape and thus simplify the concentration
determination.
One additional problem was accentuated during this analysis: ex-
cessive sample contamination by the AC charcoal bed in the UC.
This can be seen as the broad peak in the chromatogram of stream
L-l in Figure 13. The next sample to be analyzed was a synthetic
light hydrocarbon sample. However, the analysis was postponed
55
-------
Table 6. CLASS ANALYZER ANALYSIS OF SAMPLE EPA-2
uc
1
c.
3
4
5
6
Sampling
period, PST
10-14
14-18
18-??
22-0?
02-06
06-10
Sature
Light
0.99
0.81
0.25
0.24
0.21
0.82
Cc
ttes
Heavy
0.39
0.55
0.32
0.20
0.08
0.13
:>mponent c
Unsatx
Light
_
0.57
0.08
0.12
0.47
0.68
:oncentra1
arates
Heavy
-0.14
2.57
2.14
0.43
0.13
0.13
:ions, mg/ffi
Aroma tics
0.56
0.49
0.54
0.56
0.21
0.15
Oxycarbons
0.46
-2.33
-3.04
-0.60
-0.09
0.41
.01
Ch
-------
vjl
0
0
0 2 4 C
Time, minutes
0
Figure 13. Typical Clnss Analyzer output signal
-------
until the charcoal contamination could be investigated.
Conditioning of AC charcoal at 200° under helium flow does not
sufficiently clean the charcoal surface of adsorbed hydrocarbons.
Conditioning at 300 in vacuo overnight reduces but does not eli-
minate the contamination. Conditioning at 550 in vacuo eliminates
the contamination but results in such an active charcoal surface
that C-2 and C-3 hydrocarbons exhibit some irreversibility of
adsorption (i.e., sample loss occurs). The ideal conditioning
procedure has not been found as yet.
In addition to the light hydrocarbon sample, two additional
samples were to be collected and analyzed. These were the
effluents from: (1) a paint spray booth; and (2) from a printing
operation. However, scheduling and budgetary constraints forced
the cancellation of these tests.
The sampling and analytical hardware were being improved with
each sample that was collected. Presently, a sample can be
broken down into saturate and non-saturate fractions with good
reliability. This in itself is a valuable analysis because
saturates are generally quite innocuous as pollutants while
non-saturates tend to be more photo-reactive and noxious.
Samples were stored for 30 days in several UC' s in a storage
freezer (-5°C) with no apparent changes in the samples noted.
retention characteristics of the UC's themselves did not appear
to change over the course of the research effort. If the UC is
overheated, physically damaged, or irreversibly contaminated,
of usage should be indefinite.
With additional development effort and equipment redesign, the
determination of hydrocarbon class content in stationary source
emissions can become a routine analytical procedure.
58
-------
SECTION VII
REFERENCES
1. Innes, W. B., W. E. Bambrick, and A. J. Andreatch, "Hydro-
carbon Gas Analysis Using Differential Chemical Absorption
and Flame lonization Detectors," Anal. Chem., _35t No. 9,
pp 1198-1203 (1963).
2. Coulson, D. M., "Hydrocarbon Compound-Type Analysis of
Automobile Exhaust Gases by Mass Spectrometry," Anal. Chem.,
31, No. 5, pp 906-910 (1959).
3. Martin, R. L., "Determination of Hydrocarbon Types in Gaso-
line by Gas Chromatography," Anal. Chem., 34, No. 7, pp 896-
899 (1962).
4. Robinson, R. E., R. H. Coe, and M. J. O'Neal, "Rapid Hydro-
carbon-Type Analysis of Gasoline by Dual Column Gas Chro-
matography," Anal. Chem., 43, No. 4, pp 591-594 (1971).
5. McEwen, D. J., "Automobile Exhaust Hydrocarbon Analysis by
Gas Chromatography," Anal. Chem., 38, No. 7, pp 1047-1053
(1966).
6. Laub, R. J., Private Communication, Department of Chemistry,
University College of Swansea, Wales, UK.
7. Brenner, N., E. Cieplinski, L. S. Ettre, and V. J. Coates,
"Molecular Sieves as Subtracters in Gas Chromatographic
Analysis. II. Selective Adsorptivity with Respect to
Different Homologous Series," J. Chroma tog. . _3_, No. 3, pp
230-234 (1960).
8. Rowan, R., Jr., "Identification of Hydrocarbon Peaks in Gas
Chromatography by Sequential Application of Class Reactions,"
Anal. Chem.. .33, No. 6, pp 658-665 (1961).
9. Isbell, A. F., Jr., and D. T. Sawyer, "Gas-Solid Chromatography
with Salt-Modified Porous Silica Beads," Anal. Chem.. 41. No.
11, pp 1381-1387 (1969).
59
-------
10. Brookman, D. J., and D. T. Sawyer, "Specific Interactions
Affecting Gas Chromatographic Retention for Modified Alumina
Columns," Anal. Chem.. 40, No. 1, pp 106-110 (1968).
11. Ikeda, R. M., D. E. Simmons, and J. D. Grossman, "Removal
of Alcohols from Complex Mixtures During Gas Chromatography,"
Anal. Chem.. 36, No. 11, pp 2188-2189 (1964).
12. Regnier, F. E., and J. C. Huang, "Identification of Some
Oxygen-Containing Functional Groups by Reaction Gas Chro-
matography," J. Chromatogr. Sci., _8, No. 5, pp 267-271 (1970),
13. Zocchi, F., "Analysis of Methane in Helium, Hydrogen, and
Neon in the Part Per Billion Range," J. Gas Chromatoqr.,
6_, No. 4, pp 251-253 (1968).
14. Zlatkis, A., W. Bertsch, H. A. Lichtenstein, A. Tishbee, F.
Shunbo, H. M. Liebich, A. M. Coscia, and N. Fleischer,
"Profile of Volatile Metabolites in Urine by Gas Chromato-
graphy-Mass Spectrometry," Anal. Chem.T 45. No. 4, pp 763-
767 (1973).
15. Farrington, P. S., R. L. Pecsok, R. L. Meeker, and T. J.
Olson, "Detection of Trace Constituents by Gas Chromato-
graphy," Anal. Chem., 31, No. 9, pp 1512-1516 (1959).
16. Williams, I. H., "Gas Chromatographic Techniques for the
Identification of Low Concentrations of Atmospheric Pollu-
tants," Anal. Chem.. 37t No. 13, pp 1723-1732 (1965).
60
-------
TECHNICAL REPORT DATA
fl'lcasc read Instructions on the reverse before completing)
REPORT NO.
JPA-650/2-75-050
4-TlTLE ANDSUBTITLE
Development of Selective Hydrocarbon Sampling System
and Field Evaluation with Conventional Analytical
System
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
August 1975
6. PERFORMING ORGANIZATION CODE
• AUTHOR(S)
Dr. Arthur F. Isbell, Jr.
8. PERFORMING ORGANIZATION REPORT NO.
2501-F
'•PERFORMING ORGANIZATION NAME AND ADDRESS
Analytical Research Laboratories, Inc.
160 Taylor Street
Monrovia, California 91016
10. PROGRAM ELEMENT NO.
1AA010
11. CONTRACT/GRANT NO.
68-02-1201
'2- SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
Chemistry and Physics Laboratory
National Environmental Research Center
Research Triangle Park, N.C. 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final Report
14. SPONSORING AGENCY CODE
•SUPPLEMENTARY NOTES
°. ABSTRACT • ' —
A sampling system was designed to permit the determination of the saturate, unsat-
Urate, aromatic, and oxycarbon content of stationary-source hydrocarbon vapor emis-
sions. A Joy Manufacturing Company Method 5 stack sampler was modified to incor-
porate a vapor collector cartridge that quantitatively traps all volatile hydro-
carbons present in the sampled gas stream. The Class Analyzer desorbs the sample
from this Universal Collector and passes it through a system of class abstractors
that remove certain hydrocarbon classes while permitting the remainder of the sample
to pass into cryogenic traps. The contents of each trap are desorbed sequentially
through an external flame ionization detector. The quantity of each class present
in the sample is calculated from the detector responses to the effluents from the
various class abstractor streams.
Detailed evaluations of the Universal Collector and class abstractor candidates are
presented. Test results of the complete sampling and analytical system are included.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
Gas sampling, flue gases, concentrating,
sorption, chromatographic analysis,
alkanes, unsaturated hydrocarbons, aro-
toatic hydrocarbons, alcohols, aldehydes,
ketones.
Selective hydrocarbon
sampling system, hydro-
carbon class analyzer
07/03,
07 M,
lH/02
iUTION STATEMENT
Unrestricted
19. SECURITY CLASS (This Report}
Unclassified
20. SECURITY CLASS (fnis page)
Unclassified
21. NO. OF PAGES
60
22. PRICE
Form 2220-1 (9-73)
------- |