EPA-650/4-74-050
DEVELOPMENT OF SAMPLING DEVICES
FOR GASEOUS ATMOSPHERIC TRACERS
by
C. L. Deuel andR. M. Roberts
Analytical Research Laboratories, Inc.
Monrovia, California 91016
Contract No. 68-02-1235
ROAP No. 26AAI
Program Element No. 1AA003
EPA Project Officer: Francis Pooler, Jr.
Meteorology Laboratory
National Environmental Research Center
Research Triangle Park, North Carolina 27711
Prepared for
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
July 1974
-------
This report has been reviewed by the Environmental Protection Agency
and approved for publication. Approval does not signify that the
contents necessarily reflect the views and policies of the Agency,
nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.
11
-------
ABSTRACT
This report discusses the development and testing of an air sampler
to be used to collect over protracted periods and to facilitate the
measurement of atmospheric tracer compounds released in meteoro-
logical diffusion studies. Tests were conducted to determine the ad-
sorptive capacity of various sorbents potentially suitable for collect-
ing such electronegative tracer compounds as SF, and CF-SF,-. A
field-practical sampler, incorporating the best sorbent found, a high
surface-area coconut charcoal, was then designed and subjected to
laboratory tests. The effects on sampler performance of various
atmospheric influences, such as composition, pollutants, tempera-
ture, and tracer loading and level were determined. Desorption
techniques allowing quantitation of the tracer were developed.
This report was submitted in fulfillment of contract number 68-02-
1Z35 by Analytical Research Laboratories, Inc. , under the sponsor-
ship of the Environmental Protection Agency. Work was completed
as of June, 1974.
ill
-------
CONTENTS
Abstract iii
List of Figures iv
List of Tables v
Sections
I. Conclusions 1
II. Recommendations 3
III. Introduction 4
TV. Test Program 7
Laboratory Test System 7
Adsorbent Screening Approach 11
Adsorbent Screening Results 14
Site Competition Studies 17
Influence of Temperature on Adsorber Capacity 20
Other Atmospheric Tracers Tested 22
Atmospheric Tracers Response and Calibration 24
Atmospheric Tests 24
Effect of Concentration on SF, Adsorption 30
Analytical System 31
Stripping Tracer from Sampler 32
Samplers 39
Field Operation Requirements 54
V. References 56
-------
FIGURES
No. Page
1 Laboratory Test System - Flow Diagram 8
2 Laboratory Test System - Front View 9
3 Laboratory Test System - Back View 10
4 Retention Volume of SF, on Type AC Charcoal 21
5 Retention Volume of Freon 11 on Type AC Charcoal 23
6 Retention Volume of CF^SF,- on Type AC Charcoal 26
7 Linear Portion of Calibration Curve for SF, 27
8 Calibration Curve for CF3SF5 28
9 ECD Response of 3 x 10~10 g Standard SF^ Sample 33
10 ECD Response of Duplicate 3x10 g SF/ Samples
Following Charcoal Adsorption-Desorption 34
11 Sampler Desorption Unit 35
12 Field-Type Tracer Sampler 41
13 Field-Type Tracer Sampler with Adsorption Tubes
in Place 42
14 Disassembled Field-Type Tracer Sampler with
Adsorption Tubes Removed 43
15 Field-Type Tracer Sampler with Desorption Unit 44
16 Atmospheric Tracer Sampler Sketch 46
17 Sampler Body - Side View 47
18 Sampler Body - End Views 48
19 Atmospheric Tracer Sampler - Side-View
Photograph 50
20 Atmospheric Tracer Sampler - End-View
Photograph 51
21 Atmospheric Tracer Sampler, Assembled for
Field Testing 52
v
-------
TABLES
No. Page
>^_B^^B^^ MBMVV-&IBM
1 SF/ Capacity of Various Adsorbents, p-g/g 15
2 Effect of Hydrocarbon Mixture on VR of SF/ on
Type AC Charcoal at 24°C 19
3 Comparison of the Retention Volume (VR) of SF/
on One Lot of Type AC Charcoal at Various
Temperatures 25
4 Sulfur Hexafluoride Desorption Recovery Data 38
5 Desorption Recovery Data for CF3SF5 40
6 Atmospheric Tracer Sampler - Parts List 49
VI
-------
SECTION I
CONCLUSIONS
Sulfur hexafluoride, the lowest boiling of the atmospheric-tracers,
is not retained well by any adsorption media. Of the materials
tested, only charcoal was found to retain this compound sufficiently
well to permit sampling at expectable tracer levels for the sought-
for period of 4 hours.
There is considerable difference in the isothermal capacity of vari-
ous charcoals and even the same charcoal, when in different condi-
tions, to adsorb SF, and other tracers. Activation prior to use is
mandatory if maximum sorptive capacity is to be realized.
Charcoal was found not to be irreversibly affected by large doses
of acid vapors, atmospheric pollutants, water, and other atmos-
pheric components.
Input of large quantities of water and hydrocarbons into the sampler
will affect its capacity for SF, and other tracers. If sufficient char-
coal is incorporated in the sampler, worse-case situations can be
compensated for. Removal of particulates, water, and most hydro-
carbons from the atmospheric sample stream are, however, success-
fully accomplished by the precolumn trap incorporated within the sam-
pler itself.
Quantitative collection of SF, and other tracers over a 4 hour sam-
pling period with a time-averaged tracer level as unrealistically
high as 0. 1 ppm can be accomplished at air temperatures as high as
40°C (104°F).
Vacuum /thermal stripping is necessary for quantitatively removing
SF, from the charcoal sampler. Analytical error considered, 100%
recoveries by this approach are readily achievable.
The use of a charcoal column in the chromatograph used for quanti-
tating the desorbed sampler contents separates the SF/ and other
-------
tracers from any hydrocarbons desorbed from the sampler. This
column can readily be backflushed to avoid contamination of the
detector.
The sampler, when used as recommended, will perform for up to
100 sampling operations, involving different urban environmental
conditions, without evidence of performance decay.
The short linear dynamic range of the conventional electron capture
detector (ECD) makes tracer quantitation unnecessarily involved.
The broader-range (constant current) Maggs ECD system now com-
mercially available would greatly facilitate such measurements.
-------
SECTION II
RECOMMENDATIONS
This program was limited to laboratory study and development.
Although polluted urban air was routinely employed, it was not
within the scope of the contract to conduct widespread field tests
to verify the laboratory results. It is therefore recommended that
field tests be performed using suitable verification systems. The
sampler developed on the program, which is not a deliverable item,
would be donated for this purpose.
The development of the extended-range Maggs ECD will greatly
enhance the measurement of tracers in time-Integra ted samples.
The great sensitivity of the conventional ECD is normally of no value
with such samples and, in fact, large dilutions must be practiced to
bring the desorbed tracer within linear ECD range. The Maggs ECD
would largely obviate such manipulations and should be evaluated in
any further test work performed with the ARLJ sampler design.
-------
SECTION III
INTRODUCTION
OBJECTIVES
With the growth of urbanization and its associated industrialization,
it has become increasingly imperative for man to control his environ-
mental pollution. To develop the principles and mechanisms necessary
to achieve effective air pollution control, the diffusion of pollutants
must be determined. The number and complexity of the variables
affecting the movement of air pollutants, including the chemical and
physical nature of the pollutants, the structure of the air pathways,
local topography, and prevailing meteorology, are such that mathe-
matical modeling and computer calculations are necessary for the
prediction of pollution patterns.
Monitoring of specific pollutants provides valuable information con-
cerning the overall distribution of a specific pollutant, but, since the
monitored chemical will have originated from many sources, is of
limited value in pinpointing emitters. In order to associate pollutants
with their sources, unique inert chemical tracers that can be detected
at very low concentrations are frequently released to the atmosphere
at points of interest. Compounds generally employed are those of the
generic groups of electronegative compounds incorporating S-F or C-F
structures. These compounds tend to be low boiling, non-toxic, odor-
less, colorless, chemically stable, are easily dispersed into the at-
mosphere, are not likely to be present in normal atmospheres, and
can be readily measured with the ECD in the parts per billion range.
The very characteristics that make such compounds invaluable as at-
mospheric tracers for instantaneous detection prevents their effective
field use when time-integrated concentrations need to be known. To
date, large evacuated containers fitted with calibrated slow leaks have
been the only functional means of collecting time-averaged gaseous
tracer samples. A more ideal system that fulfills the criteria for
-------
extended-time sampling periods would be based upon adsorption
methods. In this type of system, the tracer-containing air stream
would be passed through a column containing the adsorption medium.
The tracer would be retained on the adsorbent over the sampling in-
terval, after which the column would then be sealed and preserved
for conventional analysis at a later time in the laboratory.
The purpose of this program then was to develop and test a small,
light weight sampler compatible for operation with conventional field
gear, such as gas pumps etc. This device would be capable of collect-
ing atmospheric tracers, such as sulfur hexafluoride (SF/) and tri-
fluorom ethyl sulfur pen tafluo ride (CF,SF,-), at levels ranging from the
limit of ECD detectability to four magnitudes of higher concentration.
The sampler would be capable of operating for sampling intervals
i
ranging from a few rt^inutes to several hours. The device would be
capable of sampling under adverse atmospheric conditions, including
those of temperature, humidity, and pollutant (including particulate)
levels. The sampler would be of a reusable configuration and be
capable of extended cycle usage without decay of performance.
SUMMARY OF RESULTS
Initial experimentation involved the screening of twenty possible adsor-
bents to determine their retention capacity for tracers. Sulfur hexa-
fluoride was used exclusively in this process since not only is it the
most universally used, it is the most difficult to retain. The materials
investigated included porous polymer beads (both coated and uncoated),
various combinations of fluoroethylene polymers and fluorinated coat-
ings, desiccant materials, catalytic substances, and various activated
charcoals. Of these, only activated charcoal was found to retain SF/
in quantities sufficient to satisfy program requirements. The sorptive
capacity of the charcoal selected for the sampler was measured as >1
mg SF/ per gram charcoal. Based on maximum atmospheric tracer
levels expected and sampling period (4 hrs) at design flow rate (25
-8
cc/min), a capacity for 10 g SF/ was the projected requirement.
-------
After a suitable adsorbent was found, the effects of atmospheric pollu-
tants were examined at various concentrations and temperatures.
These included such constituents as hydrocarbons, carbon monoxide,
oxides of nitrogen, oxides of sulfur, and water. Urban air was also
used as the tracer diluent in a series of verification sampling tests.
In addition to SF,, other tracers including various Freons and a new
EPA developed compound, CF.-SF , were studied for retention volume
and recovery.
After determining that Barnebey-Cheney Type AC activated charcoal
would retain sufficient quantities of SF, for time-integrated sampling,
methods for recovery and quantitation of the tracers were developed.
By use of vacuum thermal stripping in combination with cryogenic
trapping, essentially 100% recoveries were regularly achieved. The
recovered quantities generally were sufficiently large to exceed the
range of linear response of the ECD's used. Dilution of these con-
centrated samples was necessary to achieve quantitation of the re-
covered tracers.
The final sampler design recommended consists of a train commencing
with a Millipore filter (for the removal of particulates), followed by a
reservoir of mixed adsorbent/desiccant to remove water and many
hydrocarbons, then a U-tube filled with ground, screened, activated
charcoal.
-------
SECTION 3V
TEST PROGRAM
LABORATORY TEST SYSTEM
A gas train system, Figures 1 to 3, was constructed for evaluation
of the candidate adsorbents. The hardware consisted of stainless
steel Hoke toggle valves with Kel-F seats, Brooks flow controllers,
and stainless steel lines, all mounted on a 25 x 750 cm panel. Separate
input lines were provided for the premixed tracer, diluent gas, hydro-
carbon, NO , and SO gas mixtures. Flow rates were individually
Ji, J*.
set through the flow controllers and/or needle valves and were mea-
sured by a bubble flow meter.
At the beginning of the train, a standard fritted-glass gas scrubber was
installed to furnish gas humidification. Because the system was oper-
ated above atmospheric pressure, humidification to 100% saturation
was not possible, but this limitation was not considered to be signifi-
cant. Moisture was introduced at the head of the train because: (1)
if the air contaminants to be added in controlled amounts to the stream
were to react with each other or with water, no products would be
formed that would not be formed in water saturated air; and (2) the
use of a bubbler after the addition of the other ingredients would result
in loss of some of the gases, especially NO and SO .
^i 3±.
The accumulator consisted of a short column of the adsorbent of inter-
est. This unit was either placed in a Transite container with nichrome
wire heaters and a thermocouple temperature readout for high temper-
ature operation, or immersed in a dewar with a suitable coolant for
low temperature studies.
The cryotrap system consisted of multiple loops of 3. 2 mm SS tubing
having an internal volume of ~7 cc. These loops allowed the gas flow
to pass into and out of the cryogenic fluid several times to break up
aerosols and furnish essentially 100% trapping efficiency.
-------
Dilute
Tracer
Ozone
Hydro-
carbons
NO /SO
x x
Vent
Cryotrap
Figure 1. Laboratory test system - flow diagram
8
-------
0)
•1-1
>
-p
fl
o
a
0)
-(->
CO
>s
01
-M
CO
O
,D
rt
tv]
(U
-------
o
g
CQ
>s
ra
4->
CO
0)
-1-1
>s
fn
O
-P
rt
M
o
0>
be
• l-t
h
10
-------
The detector used for the adsorbent screening studies was a tritium
foil ECD originally designed for use with a P&E Model 880 gas chro-
matograph. Since this detector was designed as a hang-on unit, it
was well suited for incorporation into the gas train.
ADSORBENT SCREENING APPROACH
An obvious approach to conducting evaluation tests of candidate ad-
sorbents would be to pass tracer-containing gas through them until
breakthrough occurs. This is essentially a simple form of frontal
analysis. If the tests are conducted within the linear portion of the
adsorption isotherm and since irreversible adsorbents would be ex-
cluded, the tracer concentration would not be important. Thus the
capacity of each candidate could be related to the breakthrough volume,
V , or the air flow-rate times the time to breakthrough, tB. The latter
is usually graphically defined as the vertical intercept of the break-
through signal front that furnishes equal area segments formed by the
intercept, the signal, and tangents to the baseline and plateau of the
signal, viz:
Start
I
Time
X
X = Y
"ti
rt
FRONTAL ANALYSIS BREAKTHROUGH POINT
11
-------
If, at breakthrough, the tracer were eliminated from the input gas,
it would take in excess of an additional tg for all of the tracer remain-
ing on the adsorbent to come off. Although the stripping process can
of course be expedited, frontal analysis is not a convenient approach.
For screening purposes, one need only inject a small volume of the
tracer-containing gas that would have been used for frontal analysis.
If the carrier flow rate is the same, the signal for the discretely in-
jected sample will, under ideal conditions, occur at the same time
(retention time or tR) as tg, viz:
Inject
Time
GAS CHROMATOGRAPHIC ANALYSIS SIGNAL TRACE
rt
ti
uo
•t-t
CO
This is to say that retention volume, VR, which is flow rate times
tp , is equal to VR. This equivalence, which is well known to gas
chromatographers and has been mathematically derived from basic
principles, is conditional. The factors that can introduce offsets
are not important in the present context, however, and can be ignored
for purposes of preliminary screening work. In the text that follows
12
-------
the terms breakthrough and retention volume or V,-. and V_ are thus
interchangeably used.
Because V„ and V-. are determined by the dead volume within the
D t\
adsorber device tested, the quantity of active material packed
therein, and other factors, use of a capacity term would be more
definitive. This would permit the comparison of the adsorbents
screened on the basis of weight of tracer retained per unit weight
of adsorbent. Determination of ultimate capacity, however, would
have no significance for the present work since the tracer concen-
trations required would far exceed the micrometeorological test
situation. For the adsorbent screening evaluations described here,
therefore, the capacity values given are relative to the input con-
centration employed. That is, capacity equals tracer concentration
times Vp divided by the weight of adsorbent used.
A commercially prepared and analyzed mixture of 0. 9 ppm SF, in
nitrogen was purchased for this program. For adsorbent screening
purposes, however, a 50 ppm mixture of SF, in nitrogen was pre-
pared and used. Nitrogen rather than air was selected as the diluent
because of the necessity for continuous monitoring for breakthrough
with the ECD. The latter produces an appreciable signal for oxygen.
For testing, a series of small adsorption tubes each 500 x 3. 2 mm
were packed with the sorbent of interest. These tubes each contained
about 0. 7 g of active packing.
Retention volume studies were made by establishing a steady state
base line with nitrogen carrier gas flowing through the adsorber.
A "small" sample of SF,, consisting of 1 or 2 cc of 50 ppm SF/ in
nitrogen, was then introduced into the carrier stream. The retention
volume, V_, was calculated as the product of the elapsed time from
K
the introduction of the sample to the maximum detector response and
the flow rate in cc/min. In tests to verify that VR = V_, the nitrogen
carrier stream was stopped and a stream of nitrogen containing
13
-------
SF/ substituted. The appearance of a signal at breakthrough required,
within experimental error, the same volume of gas.
In addition to simplifying the preliminary screening work, the slug
addition of tracer to measure VR saved time in clearing the adsorbers
of tracer and helped maintain a low tracer level in the laboratory air.
ADSORBENT SCREENING RESULTS
Table 1 contains a listing of the observed SF, capacities for the vari-
ous adsorbents tested on this program. The porous polymer Porapak
series comprises various crosslinked polymers which have found uni-
versal acceptance in gas chromatographic applications. Of these,
Porapak P, a styrene-divinylbenzene copolymer is the least polar,
while Porapak T has a crosslinkage of ethylene glycol dimethacrylate,
making it the most polar of the series. Porapak Q, ethylvinylbenzene-
divinylbenzene, is slightly more polar than P, while Porapak N, con-
taining vinyl pyrollidone, is slightly less polar than T. There were
no appreciable differences in the SF, capacity of any of the polymer
beads, and certainly no pattern of capacity that related to polarity.
Polypak II was an early contribution of Hewlett-Packard to the porous
polymer bead adsorbent field. This material is quite similar to Pora-
pak Q. Since the porous polymers are readily liquid coated, the Poly-
pak II was coated with 10% tetrahydroxyethylethylenediamine (THEED).
This coating improved the retention of SF/, but was still much less than
the poorest of the activated charcoals. Tenax, another porous polymer
with a 2,6-diphenyl-p-phenylene oxide base, exhibited very low capacity
for SF/. Carbosieve B is a charcoal with controlled pore size that
performs much as molecular sieves. Like the latter, it too furnished
only minimal retention of SF/.
Dietz and Cole have reported evidence of SF/ adsorption on Teflon
thread sealant tape and on Teflon valve components. Based upon this
report, a series of adsorbers involving a variety of fluorocarbons was
prepared. Fluoropak 80 and Teflon VI, both finely divided Teflon pow-
ders, showed little affinity for SF,. A partially fluorinated substance
14
-------
TABLE 1. SF, CAPACITY OF VARIOUS ADSORBENTS
D
(M-g/g)
Adsorbent SF/ Capacity
,
tetrahydroxyethylethylenediamine
Barnebey-Cheney type AC charcoal,
activated 1022
Witco 888 Charcoal, activated 546
Barnebey-Cheney type AC charcoal,
as received 404
Barnebey-Cheney type GI charcoal,
as received 221
Polypak II with 10% THEEDa 11. 9
Porapak T 10.4
Porapak P 8. 9
Porapak Q 8. 3
Porapak N 7. 5
Fluoropak 80 7, 5
Teflon VI 6.0
Carbo sieve B 4. 4
Kel-F 300 L. D. 3.0
TenaxGC 3.0
Fluoropak 80 with 5% carbowax
4000 and 2% carbowax 600 0. 3
Diatomaceous earth with
15% Fluorolube and 5% Kel-F 300 L. D. 0. 3
5A Molecular Sieve 8. 5
Silica gel 10. 7
Hopcalite < 1
Calcium chloride < 1
Lithium hydroxide < 1
15
-------
Kel-F 300 L. D. , proved to have even less capacity. To achieve a
more intimate contact, an adsorption tube packed with diatomaceous
earth coated with partially fluorinated Fluorolube and Kel-F 300 was pre-
pared. This and a similar tube containing Fluoropak 80 coated with
carbowax had the least adsorption of SF, of all candidates tested.
Of the adsorbents tested, only the activated charcoals retained SF/
in sufficient capacity to permit long term sampling. These charcoals
were ground and screened, with the 40-60 mesh cut being retained
for use. Two Barnebey-Cheney coconut shell charcoals, types AC
and GI, were tested as received. The type AC indicated nearly twice
the capacity of type GI.
A petroleum-based charcoal, Witco 888, was ground, screened, and
activated in a vacuum oven at 280 C for 8 hours. The apparent capac-
ity for SF/ was greater than 0. 5 mg/g. A second lot of type AC
charcoal was then ground and screened. The 40-60 mesh cut was
activated in a vacuum oven at 280 for 8 hours. Activation more than
doubled the capacity of type AC over the as-received condition. An
SF/ capacity value of 1. 02 mg/g was calculated. Based upon test
2
values obtained by Turk and coworkers , one gram of activated, type
AC charcoal should retain all of the SF/ contained in over 15 cubic
meters of air when at concentrations of 11 ppb. Obviously, such large
volumetric samplings are not required in the present context, but
the capacity reported does tend to corroborate its selection, as was
done on the present program.
A series of reactive materials were also considered for potential use
in the removal of possibly interfering air constituents. These con-
sisted of hopcalite (a group of metal oxides) for CO conversion, cal-
cium chloride and magnesium perchlorate for water abstraction,
lithium hydroxide for CO7 and, possibly, NO /SO removal, silica
w X X.
gel and 5A molecular sieve for organic s and water removal. All
materials were ground and screened so they could be readily packed
in a 50x0. 32 cm adsorption tube. The hopcalite, lithium hydroxide
and calcium chloride tests all indicated retention volumes only
16
-------
slightly greater than the system dead space. The magnesium per-
chlorate sample adsorbed sufficient moisture to cake and plug the ad-
sorption column. Silica gel and 5A molecular sieve both retained
very little SF,, with retention volumes comparable to those obtained
from porous polymer beads.
SITE COMPETITION STUDIES
Air constituents considered as possibly having deleterious effects
upon the adsorption of atmospheric tracers were carbon monoxide,
carbon dioxide, water, hydrocarbons, and the oxides of nitrogen and
sulfur. A series of mixtures of CO, CO?, NO^, SO?, and CH, in
nitrogen were prepared, each at ~325 ppm. Introduced into the gas
train, a ZOcc/min flow from any of the mixed gases, when blended
with the 90 cc/min nitrogen carrier flow, resulted in a concentration
of ~60 ppm in the system. With the possible exception of CO and CO2,
these levels were considerably greater than would be expected in air.
Ozone was omitted because it is instantaneously eliminated by charcoal.
A series of determinations established a 24 C retention volume (V^,)
±v
of 2500 cc for the type AC charcoal adsorber used for these tests. To
determine the effect of each contaminant, a sample of SF/ was added
to the carrier stream composed of the ~ 60 ppm dilution of the appro-
priate chemical as described. Each of the contaminant mixes in turn
was blended with the nitrogen carrier. The VR for SF/ remained at
a constant 2500 cc despite the presence of the other components
except methane, which caused the VR for SF/ to be reduced to 2450 cc.
Combinations of mixes were also introduced into the carrier with no
detectable changes in SF/ adsorption capacity, except again, for a
slight decrease when methane was present in the mixture. Finally,
to simulate a very dirty air sample, all the pollutants were combined
together in the carrier stream. The V_ for SF/ was reduced to
lx b
2460 cc, a reduction of less than 2%. Following these tests, the V,.,
XV
for SF/ was checked using uncontaminated carrier. It was found to
be normal at 2500 cc. Analyzing these test results, it would seem that
of the contaminants chosen methane alone offers adsorption site com-
17
-------
petition with SF/. Subsequent tests indicated methane had a retention
volume of 100 cc on the adsorption column tested.
Partially because of the low retention volume of methane and to test
for the effects of the various classes of hydrocarbons, a mixture of
nominally 180 ppm each of representative n-alkanes (propane)
branched alkanes (neopentane), alkenes (cis and trans isomers of
2-butene), aromatics (benzene), oxygenated hydrocarbons (methyl
alcohol), and chlorinated hydrocarbons (chloroform) was prepared in
nitrogen. This mixture was blended with the carrier stream at a
1:9 ratio. After a series of SF/ retention determinations, it was
noted that the retention volume for SF/ at 25 C had dropped from an
initial 2200 cc to 1150 cc.
A fresh mixture of the six hydrocarbons representing the various
classes was prepared at 300 ppm in nitrogen. A fresh adsorber
column loaded with type AC charcoal that had been activated by
thermal-vacuum stripping at 285 C for 8 hours was prepared. This
column had a retention volume for SF/ of 3325 cc at 24 C. A series
D
of tests were performed in which the V0 of SF/ was first determined
.TV. D
using uncontaminated nitrogen carrier. Following this, a determina-
tion of the VR of SF/ was made with the hydrocarbon mix blended into
the carrier stream at a 1:6 ratio to produce a 5 0 -ppm level of each
contaminant. These tests were followed by a series of 8 additional
VT, determinations, in which pure nitrogen and hydrocarbon-loaded
K.
nitrogen carrier were alternated. The results of these tests are
shown in Table 2. In each test, the retention volume (capacity for)
of SF/ was greater with the pure nitrogen than with the contaminated
nitrogen. The retention volume decreased for each member in the
series so that, after 5 cycles, the V_ with pure nitrogen was 3040 cc
rv.
(an 8. 6% decrease) and that with added hydrocarbon had declined from
3180 cc to 2610 cc (nearly 18% decrease). Following the 5th cycle, the
gas train was switched to bypass the detector and the adsorber was
heated to 85 C while being purged with nitrogen for 20 minutes. After
cooling, a test cycle was repeated with the Vra for SF, being deter-
XX D
18
-------
V0 OF SF, ON TYPE AC CHARCOAL AT 24"C
K D
TABLE 2. EFFECT OF HYDROCARBON MIXTURE ON
;HARCOAL AT 2^°-
(cc's)
Retention Volume
^
Jtv,
Test Number Without hydrocarbons With hydrocarbons
1 3325 3183
2 3277 2800
3 3230 2700
4 3135 Z688
5 3040 2610
6* 3230 3164
7b 3495
cl O
Adsorber heated to 85 C and nitrogen purged 20 minutes
Adsorber heated to 85 C and nitrogen purged 30 minutes
19
-------
mined at 3230 cc with clean nitrogen and 3165 cc with the hydrocarbon
mixture present. The adsorber was again heated to 85 C with nitrogen
purging for 30 minutes. After cooling, the SF, VR was measured at
3495, seemingly a level of activation greater than that initially indicated.
These data indicate that continuous adsorption of hydrocarbons will
definitely lower the capacity for SF,. By adjusting the quantity of char-
coal, however, sufficient sorptive capacity for both can be provided
for a single run. For recycle, the hydrocarbons are readily desorbed
with the SF,, thus reactivating the charcoal adsorbers for additional
use. In any case, the final version of the sampler contains a pre-
section for the removal of most hydrocarbons.
Passage of the carrier gas through a water-filled gas scrubber to
achieve nearly 100% relative humidity did not affect SF, adsorption.
The V^ of water on the sampler charcoal was found to be only 21 cc,
K.
producing an almost immediate, swamping signal from the ECD. Be-
cause of the effect of water on EC detection, water was removed from
the carrier stream by a short calcium chloride desiccant column in-
serted between the adsorber and the detector. A series of determina-
tions using this system indicated no detectable changes in the adsorber
capacity for SF, when collected from a nearly water saturated
carrier. This will be discussed in further detail in a later section.
INFLUENCE OF TEMPERATURE ON ADSORBER CAPACITY
Temperature studies were performed through a range of -17 C (O jT)
to 71.1 C (160 F). These results are presented in Figure 4. In study-
ing the plot of retention volume vs. temperature, a break is seen to
occur between 30 and 40 C. This behavior is markedly similar to
that evinced in adsorption isotherm plots wherein a transition from
monolayer to multilayer adsorption is indicated. In any case, the dim-
inished capacity of charcoal for SF, above 30 C (86 F) clearly estab-
lished that the weight of adsorbent theretofore used should be increased.
The combined effect of temperature and the addition of the six compon-
20
-------
u
o
W*
H
W
w
H
70
60
50
40
30
20
10
0
I I I I 1 I I I I 1 1 I I
1000 2000
RETENTION VOLUME (V,,). cc
K,
Figure 4. Retention volume of SF, on type AC charcoal
21
-------
ent hydrocarbon mixture was next determined. In these tests, the
V,, for SF, with pure carrier was compared to that with synthetically
J\. D
polluted carrier ( ~18 ppm for each contaminant). The VR values
determined (2200 cc at 24 C) agreed with one another within exper-
imental error (2-15 cc variation). After several determinations at
increasingly lower temperatures, the VR at -10 C was still only
2200 cc. Upon warming the adsorber to 25 C, it was found that the
VT, had been reduced to only 1150 cc. This indicated that at lower
K.
temperatures, the adsorption capacity for hydrocarbons is greatly
enhanced to the detriment of SF, adsorptive capacity. These low
temperature studies supported the conclusion arrived at after operat-
ing at elevated temperatures, namely, a greater quantity of adsorbent
would have to be provided for in the sampler design.
OTHER ATMOSPHERIC TRACERS TESTED
In addition to SF/, limited studies of other potential tracers were
made. Among these were trifluoromethylsulfur pentafluoride (CF-SFr),
trichlorofluoromethane (Freon 11), octafluorocyclobutane (Freon C-318)
and difluorodibromomethane (Freon. 12B2).
Minimal attention was given the first two Freons named because, due
to their background levels and physical characteristics, there was a
question as to their appropriateness. Freon 12B2 was suggested late
in the program and the sample submitted by the National Oceanic and
Atmospheric Administration had somehow been compromised such
that it was received as essentially an air sample containing only a
trace of the Freon.
Dilutions in nitrogen of the above named tracers were prepared in the
ppm range. Freon 11 (b.p. 24 C) did not breakthrough after 1 hour at
20°C and a carrier flow rate of 90cc/min (V-^ 75400 cc). At 30°C, the
K
retention volume was 3200 cc and, at 100 C, it was still greater than
2000 cc. Figure 5 displays VR vs temperature for this material.
Charcoals' capacity for CF,SF,- was determined to be greater than
for SF^ throughout the ambient temperature range. This increased
capacity is most pronounced at lower temperature, being 50% greater
22
-------
1 I
o
o
o
CO
o
o
o
CO
o
o
w
3
3
o
s
H
2
W
H
W
o
o
o
o
o
M
O
oo
CJ
o
o
o
00
«J
o
o
o
O
<
Q)
IT)
a)
23
-------
at 20°C. As the temperature is increased, the capacity differential
for CF3 SF5 over SF^ rapidly decreases to only 5% at 50°C. Table 3
contains a direct comparison of the retention of SF/ and CF,, SF,- at
selected temperatures on the same adsorber. Figure 6, displaying
the retention volume of the latter compound at various temperatures,
does not show the sharp break in adsorption at 30 -35 C noted with SF/.
ATMOSPHERIC TRACERS RESPONSE AND CALIBRATION
The 0. 9 ppm (5. 4 mg/m ) commercial mix of SF/ purchased proved
to be too concentrated for determination of response. It was there-
fore further diluted to 0. 9 ppb. A linear ECD calibration for this
-11 -9
material was obtained from 5x10 g through 2x10 g, as shown in
Figure 7. The lower end of the linear segment of the ECD response
curve may well extend to less than 5xlO~ g, but this was not determined.
A 1 ppb dilution of CF- SFp. was also prepared for ECD response
3
calibration ( 8. 0 mg/m ). Calibration was carried out only through
a range that was required for the laboratory tests. The response was
_9
found to be linear within the range of interest (1x10 g through
- 8
2. 5xlO~ g), as is seen in Figure 8.
/ o
Although these calibrations were performed with a Ni electron
3
capture detector, the laboratory also uses H radioactive- source ECD'S
and a non- radioactive photoionization ECD. Each of these detectors
- 8
was found to become saturated when sample sizes greater than 2x10 gm
SF/ were used. This indicated that no advantages of extended range
was available from any alternate type of detector configuration.
ATMOSPHERIC TESTS
The samplers designed for this program will be discussed in detail
later. However, the initial version was used for a series of atmos-
pheric tests that are to be discussed at this point. Weather conditions
varied from clear, dry, and windy, through dry and still with light
smog, through moderate rain. The only problems encountered were
with the magnesium perchlorate desiccant used in the sampler. Under
humid conditions, it caked and interfered with sample flow. This
24
-------
TABLE 3. COMPARISON OF THE RETENTION VOLUME
(V^) OF SF, and CF,SFC ON ONE LOT OF TYPE
K D 6 D
AC CHARCOAL AT VARIOUS TEMPERATURES
(cc's)
VR
Temp. ,°C SF
6
20 2420 3710
25 2290 3030
30 2110 2460
40 1050 1540
50 880 930
25
-------
o
o
o
OO
o
o
o
(M
O
o
o
o
o
W
o
(-7=-
Q
H
W
W
tf
ri
O
o
o
U
0)
CO
V
U
*o
o
"ri
-------
CO
H
O
O
10
9
8
ANALYTICAL CONDITIONS
SF/ cone.
Carrier flow
Chart speed
Column temp.
Retention Vol.
Atten.
Detector
0. 9 ppb
100 cc/min
1. 27 cm/min
80°C
700 cc
x4 x,
F&M Ni
o
w
H
O
fe
w
rt
Q
o
0
0.2 0.4 0. 6 0.8 1.0 172 174 176 178 2.0
ANALYTE CONCENTRATION, ng/cc
Figure 7. Linear portion of calibration curve for SF/
27
-------
9
H
i—i
Q
O
O
3
t~*
<
ti
* 1
• H C
£ -51?
ft O O CJ s
* °!>U °
o o roo o
r-? i-l i-I 00 CO X
o
rt
O
u
CO 5 P
h Rj Jd 'O
O U UO
CO
Q
U
o
ri
S
S
W
O
HT
<"—1
O
O
W
H
in
CO
fc
U
O
f-l
3
O
fl
O
• H
-4->
rf
U
oo
O
CO
O
CM
uio *V3HVT:VNDIS
28
-------
material was therefore replaced with calcium sulfate which precluded
any further problems.
Adding controlled trace levels of SF, proved somewhat of a problem,
even using a 0. 9 ppb mix, but adjustments of technique led to reliable
additions. For these tests, the sampler was connected to a vacuum
source. The inlet was tee'd, one leg leading to outdoor air. The
other leg led to a reservoir of 0. 9 ppb SF/ in nitrogen. Introduction
of the dilute SF/ was adjusted by a flow controller set to deliver 5 cc/
min, which is about the lowest practical flow level for these units.
The actual flow measured was found to be 5. 5 cc/min. The air intro-
duced brought the total flow to 25 cc/min. The density of SF, at the
-3
temperature of the experiment was 6. 5 x 10 g/cc, so that SF/ mass
-11
flow was 3. 2 x 10 g/rc>.
With flow durations of 15. 3, 30. 8, and 61. 3 minutes, recoveries of
3. 4 x 10"10, 8. 6 x 10"10, and 1. 8 x 10"9 g SF6, respectively, were
made, for recoveries of 69%, 86%, and 89%, respectively. Desorp-
tions were performed with a helium purge at 50 cc/min at 150 C for
25 to 60 minutes with cryogenic trapping using liquid nitrogen. As
is discussed elsewhere, this method of desorption is not quantitative,
vacuum stripping being the preferred approach, as was subsequently
employed. The low recovery obtained was not entirely due to the de-
sorption process, however. Another source, which was quickly cor-
rected was a fixed volumetric error. This caused the apparent re-
covery efficiency to increase with sampling time or volume. Actually
the volumetric error was merely being "diluted". This unexpected
problem, once identified, could have been corrected for, but a redesign
of the dilution system was considered preferable.
The implications of using different types of detectors were evaluated.
First, the photoionization ECD was used for a series of samplings as
previously described. Recoveries were 83-86%, indicating no major
differences from the Ni detector.
It is of interest to note that all atmospheric test samples displayed at
least 3 ECD signals in addition to that for SF,, although there were no
29
-------
interferences (peak overlap) in the analyses. To verify that retained
hydrocarbons did not contribute analytical interferences with respect
to SF/ detection, portions of the desorbates from 2 and 3 hours atmos-
pheric testing were passed through a flame ionization detector (FID).
Numerous large signals were obtained, but none eluted near the SF/
or other tracer peaks. Since some ECD's operate over limited temper-
ature ranges, they may become contaminated from condensates acquired
over continuous exposure to some higher boiling hydrocarbons. As dis-
cussed later, steps were therefore taken to anticipate this potential
problem.
The adsorbers used for these tests were subjected to more than 100
exposures to atmospheres artificially contaminated at high levels of
concentration. In addition, they were subjected to many hours of
urban atmospheric sampling. There were temporary diminutions of
sorptive capacity for SF/ under these conditions, but the capacity was
readily restored when the adsorber was heated and purged between
tests. The stripping process to remove the tracer from the adsorber
will effectively remove any normal atmospheric constituents that would
temporarily lessen adsorption capacity.
EFFECT OF CONCENTRATION ON SF6 ADSORPTION
During long term sampling, tracer concentration over a given time
interval will be subject to wide variation. In determining the possible
effect upon adsorption capacity, three mixtures of SF, at 0. 9 ppm,
0. 9 ppb, and 145 pp'm were connected to a manifold, which was in line
of a carrier gas stream, which, in this case, was helium. When the
adsorber discharge was connected to an ECD, VR of SF/ for any one
of or a combination of the SF/ concentrations could be determined.
The flow rates of the tracers were controlled by flow controllers.
Although the lowest SF/ mixture was 0. 9 ppb, the sample was further
diluted with the helium carrier in performing these concentration effect
studies. In practice, a flow rate of 5. 5 cc/min of 0. 9 ppb SF/ tracer
into a total carrier flow of 100 cc/min produced an effective concentra-
tion of 0. 05 ppb.
30
-------
3
Altshuller reported from his work that 0. 01 ppb is the minimum
concentration of SF, detectable. This represents an SF/ mass of
-14 6 ^6
6. 5 x 10 g/cc. Four orders of magnitude in sample concentra-
tion above detectable was the range of sampling capability sought.
If six liters of air were sampled, which would correspond to fulfill-
ment of the development objective of sampling at a rate of 25 cc/min
for as long as four hours, then the mass of SF/ to be handled would
be 3. 9 x 10"10 to 3. 9 x 10"6 g.
In the performance of the concentration effect studies, discrete
samples were introduced by opening and closing the manifold valving
for measured time intervals. Thus, with the flow controller for the
0. 9 ppb SF, mixture set for 5.5 cc/min delivery, a 5 second sample
-12
introduced a mass of 2. 4 x 10 g SF,. This is equivalent to sam-
-14
pling, for 4 hours, air having an SF, concentration of 6. 5 x 10
g/cc, the minimum level detectable.
Before each test run to determine the effect of tracer concentration
variation, the charcoal was freshly activated by thermal vacuum
stripping. Initially, equal volumes of the three SF, mixtures were
run. This represented a concentration span of 146, 000 or five orders
of magnitude. The V,-. obtained for a 40 second input of a flow of 40
cc/min of each of the mixtures was 3240 cc in each case. Tests were
then conducted with the 0. 9 ppb mixture diluted in helium to 0. 05 ppb
to furnish, as described in the previous paragraph, a mass equivalent
to that of the minimum detectable level collected over a 4 hour period.
Finally, by manipulating the flow inputs from the three SF/ mixtures,
inputs having various concentration patterns, including step-changed
ones in which the levels varied by the factor of 146, 000 cited above,
were effected. In all of these tests, the variations in VR were small,
all falling within a range of 3200 to 3500 cc.
ANALYTICAL SYSTEM
Mention has been made of the column used in the GC system. For
analytical purposes, a 50 x 0. 32 cm column of the same charcoal as
used in the adsorber was employed in the chromatograph. The pur-
31
-------
pose of this column was to provide separation of the tracer from other
electron capturing materials, and to retain hydrocarbons sufficiently
to avoid any possibility of cell contamination. The incorporation of a
six port valve in the system permitted easy backflushing once the
tracer had passed through the detector. An alternative method con-
sisted of disconnecting the column from the detector and heating the
oven to 150-200 C for 15-30 minutes, thus discharging the hydrocar-
bons into the oven rather than through the detector. Because the
sampler itself incorporated a presection of molecular sieve, the need
for either procedure was questionable.
The use of a charcoal column does preclude the use of an argon-methane
carrier. For the analyses on this program, helium was used as the
carrier at a flow of 100 cc/min. Column temperature was 80 C. Five
percent methane in argon at 100 cc/min was used as a purge gas in
the detector. The analytical instrument was an F & M 5756 chromato-
/ "3
graph with a Ni ECD operated at 250°C.
Figure 9 shows the response obtained from 3xlO~ g of SF, from a
-10
standard mix. Figure 10 shows duplicate runs in which 3x10 g
SF, was adsorbed on charcoal and desorbed by vacuum-thermal
methods, which are discussed later. The chromatograms show con-
siderable peak broadening, which is acceptable in view of the separa-
tion advantage gained from the charcoal column. The duplication and
recovery are also demonstrably quite satisfactory.
STRIPPING TRACER FROM SAMPLER
The desorption unit initially designed for use with the original sampler
is shown as Figure 11. In this photograph, four adsorption tubes are
in place. Connected to the two end pieces are an inlet line on one
block and an exit line on the other. These allow for purging of the
jtracer from the charcoal adsorber. To use, an adsorption tube of
choice is aligned between the entrance and exit ports of the end pieces.
In this manner, the other three adsorption tubes are blanked off. The
system is then connected to a helium line inside an oven (an unused
3Z
-------
o
77%
of Scale
Instr.
Detector
C olumn
Carrier
Purge
CONDITIONS
Inlet
Detector
C olumn
Range
Atten
Chart
Carrier
Purge
235°C
235°C
80°C
10 mv
x8
1. 27 cm/min
100 cc/min
100 cc/min
SPECIFICS
F&M 5756
63Ni ECD
AC Charcoal
Helium
5% Methane
in Argon
111 I I I I I I I I
CHART TRAVEL, cm
Figure 9. ECD response of 3 x 10" g standard SF/ sample
33
-------
74% of
Scale
I I 1 1 I
74% of
Scale
I I
I I
CHART TRAVEL, cm
Figure 10. ECD response of duplicate 3 x 10"10 g SF6 samples
following charcoal adsorption-desorption
34
-------
a
o
o
CO
0)
-a
r-l
I
35
-------
chromatograph is ideal) and a cryogenic trap system is connected to
the exit line. In the present tests, the purging was conducted at
150°C with a helium flow rate of 50 cc/min. for times varying from
25-60 min. Recovery by this method ranged from 70 to 90%. Repeat-
ing the desorption operation, however, indicated residual SF, in the
desorption unit. Repeated purging at up to 200 C indicated a contin-
ued residual of SF,. Because of these difficulties, vacuum.-thermal
stripping was investigated. Unfortunately, the desorption unit was
not amenable to vacuum operation.
To evaluate vacuum-thermal desorption, a single stainless steel
tube, 11x0. 63 cm, was used to replace the collection unit. Mixes of
the 1 ppm and/or 1 ppb SF, were passed through the collector at flow
rates varying from 20 to 45 cc/min for time intervals of 5 to 60 min.
Following exposure, the adsorber was connected to a vacuum rack
and desorbed at 150 C at 1x10 torr for 1 hour. The cyrogenic trapping
system consisted of a pre-cooler or U-trap at -40 C followed by a
high efficiency Schultz trap immersed in liquid nitrogen. After de-
sorption, the Schultz trap was isolated from the rest of the system and
its contents transferred ( by LN-, condensation) to a suitable receiver.
The trapped contents were then either analyzed directly or diluted to
reduce the tracer level to within the detector's linear range. Typical
recoveries were:
Concentration added SF^. delivered, g SF^ recovered, g % recovery
——^———————————_ —— o / w D > v
0. 9 ppm 1.76x10 1.66x10 94.3
0.9 ppb 3.86xlO"9 3.66xlO"9 94.8
The use of large, high-efficiency vacuum racks for desorption cannot
be considered routine procedure for most laboratories. The desorp-
tion apparatus was therefore reworked for simpler vacuum stripping.
This was accomplished by milling a small recess in both end plates
in alignment with the 4 adsorption tubes. By placing a viton 0-ring in
each recess, vacuum tight seals were produced when the 0-rings were
compressed. The line that had been used to admit the purge gas was
plugged. The unit was then placed in a GC oven and the effluent tube
36
-------
connected to a shut-off valve outside the oven. Two stainless steel
traps were then connected in series. Each trap consisted of 3m of
0. 32 cm tubing coiled into a 3-loop "race-track" configuration small
enough (4 cm x 11 cm) to be immersed in a 250 ml wide-mouth Dewar.
Each trap was terminated with shut-off valves on both ends. The
first trap was immersed in an alcohol bath maintained at -40 C with
dry ice and the second in a liquid nitrogen bath. The loops were
situated in the baths so as to pass in and out of the cryogenic liquid.
This creates a thermal gradient and flow turbulence such that
vapors (and aerosols if involved) are more efficently trapped. The
free end of the trapping train was connected to a vacuum pump.
Desorption was accomplished by heating the sampling fixture to 150
and transfering the tracer in vacuo for one hour. This system proved
to be only slightly less efficient than what was achieved on the vacuum
rack. Recoveries ranged from 80 to 93%. Variations in trapping
efficiency, occasionally falling to as low as 50%, were traced to a
permanent set of the O-ring seals, which promoted leakage after 2 to
4 desorption cycles. Frequent replacement of the O-rings controlled
the problem, and allowed completion of desorption efficiency testing.
To enhance reliability, the sampler was redesigned to eliminate the
need for O-rings and flat surface seals.
Table 4 contains the data from a series of SF, loadings and recoveries
using vacuum-thermal stripping with the redesigned fixture. Loading
was performed by flowing SF, -containing nitrogen at concentrations
of 0. 9 ppm and 0. 9 ppb at carefully controlled rates for varying periods
of time through the 4-segment sampler. The test series were gener-
ally performed in groups of three since one adsorption tube that had
been previously stripped after a preceding test series was included as
a control blank. No residual SF, was detected from any of these blank
desorptions.
Test series 1 through 4 were intended to verify desorption efficiencies
in the ranges likely to be encountered in field situations. Tests 5 and
6 were to determine desorption efficiency at arbitrarily selected extremes.
37
-------
TABLE 4. SULFUR HEXAFLUORIDE
DESORPTION RECOVERY DATA
Test No.
6A
6B
2A
2B
2C
3A
3B
3C
1A
IB
1C
ID
IE
4A
4B
4C
5A
5B
5C
SF6
Adsorbed, p,g
0. 00045
0. 00060
0.200
0.200
0.200
0. 300
0. 300
0. 300
0.450
0.450
0.450
0.450
0.450
0. 600
0. 600
0. 600
45.0
45.0
45.0
SF6
Desorbed, p,g
0. 00042
0. 00056
0.202
0.198
0.199
0. 300
0. 307
0.298
0.454
0.443
0.447
0.452
0.450
0. 590
0.610
0. 600
43. 5
45. 1
45.2
Avg.
Recovery, % Recovery, %
93. 3
93. 3 93. 3
101.0
99. 0 99. 8
99.5
100. 0
102.3 100.5
99. 3
100.9
98.4
99.3 99.8
100.4
100. 0
98. 3
101.6 100.0
100. 0
96.7
100.2 99.1
100.4
38
-------
Special care must be taken, when dilution of the trapped tracer is
necessary, to insure thorough mixing. A test series in which incom-
ing turbulance and diffusion were depended upon for mixing produced
recoveries of only 39 to 43%. The mixing bottles had to be heated
and sufficient time allowed for adequate diffusion.
Obviously, insufficient dilution can also produce low results. A
test series in which a 45 p,g sample of SF/ was diluted for analyses
is cited to demonstrate this. In this series, a 45 p,g sample was
diluted but insufficiently to enter the linear range of ECD response.
The dilution factor can be assigned a relative value of 1. 0. Incre-
mental volumetric dilutions of a like sample size are shown to indi-
cate the nonlinear response effect.
NONLINEAR ECD RESPONSE ON APPARENT RECOVERY
Test No. Relative Dilution Apparent SF, Recovery
recovery, jj, g
1 1. 0
2 1.4
3 1.7
4 2.5
These tests demonstrated the magnitude of error that can result from
lack of attention to detailed laboratory procedure.
25.3
29.9
31.3
43. 5
56.2
66.4
69.6
96.7
Table 5 contains recovery efficiency of a test series using CF, SFr
tracer, in which detector saturation •was not involved. The recovery
reproducibility was not quite as precise as with SF/ .
SAMPLERS
Figure 12 presents a sketch of the sampler initially fabricated for this
program. Figures 13 and 14 are photographs with the adsorption tubes
in place and removed. Figure 15 is a display of the desorption system
with a loosely assembled sampler.
The sampler unit consisted of 4 stainless steel adsorption tubes, each
11x0. 63 cm, evenly spaced around and attached to cylindrical blocks at
39
-------
TABLE 5. DESORPTION RECOVERY DATA FOR
CF,SFK CF,SF_ .
35 35 Avg.
Test No. Adsorbed, |Jig Desorbed, yg Recovery, % Recovery, %
1A 0.150 0.150 100.0
IB 0.150 0.146 97.3 98.7
2A 0.300 0.290 96.6
2B 0.300 0.281 93.7 97.0
2C 0. 300 0. 302 100. 7
3A 0. 600 0. 586 97. 7
3B 0.600 0.595 99.1 98.4
40
-------
s
rt
CO
0)
o
rt
0)
•l-l
h
0)
bJD
•I-l
41
-------
(D
O
CO
0)
O
CQ
£
CD
t-H
I
CO
0)
O
CD
i—I
CD
•i-i
CO
i—I
-------
o
s
0)
co
0)
fl
o
•r-l
OH
O
CO
rt
jd
a)
T— I
A
B
d
ca
(D
O
CD
&
4->
T!
i—i
CD
S
-------
4-1
1
ti
o
PH
^
O
ra
0)
-d
0>
i-H
A
a
rt
ra
0)
o
ri
(U
• r-4
44
-------
either end. Each tube was intended to be an individual adsorber.
This assembly was in turn fastened between two larger cylindrical
sections which can be called the head piece and end piece. The
center of the head piece was bored out and a shoulder milled to
accept a standard 25 mm millipore filter with its mounting hardware.
The filter was included to remove particulates from the stream sampled.
A glass tube, packed with desiccant, runs from the center of the filter
recess to the center of the end piece. Ground and screened indicating
Drierite proved satisfactory as the desiccant, with the indicator
showing when replacement was required.
A side bore in the end piece connected with an opening on the upper
surface of the end plate. In the head piece was a similar opening
directly over the lower one. This passage led to a discharge opening
on the side of the head piece. The end piece and head piece were
joined by means of two guide bars which sealed the sampler across
teflon discs. In use, one adsorber tube bridged the openings in the
head and end pieces, while the other 3 were blanked off by the teflon
sealing discs. In this manner, when allowable sampling time was
exceeded, or when a change in location was made, the end piece could
be loosened and a new adsorption tube moved into line by rotating
the column segment.
This design operated satisfactorily and was used in generating much
of the data in this report. However, the change to vacuum-thermal
stripping created sample recovery problems. These difficulties
were discussed previously and, for the reasons stated, a new sampler
configuration was designed and fabricated. Figure 16 is a sketch of
the essential components. It will be noted that, except for the body of
the sampler which is further detailed for machinists' use in Figures
17 and 18, all component parts are standard stock items available in
most gas sampling laboratories. The parts used are itemized in
Table 6. Figures 19 and 20 are photographs of the sampler with the
adsorber and filter holder removed. In Figure 21, the sampler is
shown attached to a mass flowmeter in series with a Millipore
45
-------
O
+->
0)
J4
M
?H
Q)
rt
to
0)
o
o
•H
!4
0)
in
O
0)
!H
3
bfl
46
-------
NI
ri
w
Z
Q
to
£
W
Q
•
rt
.3
g
o
s
5
1— 1
O vO
CNJ r-l
00 O
O
U Q
o
i— i
^
tn
CM
W
IT)
O
CO
•*
1— t
fa
CM
i— I
00
I— 1
CO
o
CM
o
•^
v£3
o
•r1
MH
II
w
CO
(U
0)
Z
O CQ
(0
Tl
•1-1
CD
I
0)
t-
•—I
O
47
-------
A. TOP VIEW
Notes & Dimensions per Figure 17
B. BOTTOM VIEW
Dia 3. 02 cm (1. 19 in.)
Note 3. - Use No. 38 drill for inter-
connecting holes
Figure 18. Sampler body - end views
48
-------
w
•*
?H
CD
Manufactur
P/N
o
o
U->
CM
0
1
o
•*
X
X
B
o
•i-l
•4->
^
O
ft
,3-S
g^
>E °
o
o
1
in
CM
O
&
vO
X
X
•B
.I-J
Supplied w
ti
CD
CD
f-t
support sc
CO
GO
cu .-1
(!)
W =3
co
nJ
1
CD
CD
w
j
PH
5
O
ri
1
rt
fH
O
U
CD
a
bfi
fl
£0
oU
0
CO
0
O
•i-i
f-t
u
PJ W
W H
O
W
J
CQ
D O
B ^
rtQ
in
CD
-p
s-s
cc
nJ
O
CD
fn
O
CD
sa
•H
M-l
-(->
CD
.—I
•<->
^
O
o
o>
B
B
O
o
CD
•§
CD
I
•l->
B
CD
O
co
T)
-------
bfi
O
0)
'>
I
•1-1
CO
O
rt
CD
cr-
r—I
(U
50
-------
bfi
O
0>
•1-1
I
T)
fl
a>
-------
ra
CD
CD
•r-l
Tl
0>
i—I
,Q
E
CO
co
s
rt
CO
CD
O
rt
O
•I-l
^1
CD
CO
O
s
CM
CD
?H
3
W)
52
-------
restricted orifice vacuum, pump for sampling. This system would be
suitable for field use, if AC power were available.
The sampler body was fabricated from a nominal 50 mm (2 in. )
diameter rod. The laboratory models were made of brass, although
a stainless steel body might prove more desirable for field use.
One end of the rod was reduced to 32 mm and threaded to accept a
standard filter holder. The center was bored and milled to 15 mm
diameter by 31 mm deep. This formed a reservoir for desiccant
approximately 5 cc in volume. Above the reservoir, a 5 mm lip was
milled out to accept a 25 mm filter, its accompanying stainless steel
support screen, and two teflon gaskets. In the bottom of the sampler
body, two holes 3 cm apart were drilled and tapped to accept Swagelok
400-1-2 connectors. A 2 mm hole is drilled through one of these
latter holes into the bottom of the desiccant receiver. Another hole
is drilled at 90° and tapped into the other bottom part. This opening
may receive either a Swagelok 401-A-2 adapter or a Cajon 2HN nipple.
The adsorber consists of a length of 0. 63 cm(0. 25 in.) 304 SS tubing
bent into a U shape over a 1. 51 cm (0. 6 in. ) radius. The adsorber
is packed with 40-60 mesh ground and screened Barneby-Cheney type
AC charcoal. The adsorbent is retained by glass wool plugs in either
end. The adsorber tube is connected to the two Swagelok 400-1-2
fittings in the bottom of the body. The Swagelok connector is supplied
with Swagelok nuts and ferrules, 402-1, 403-1, and 404-L needed on
both the sampler body and the U-tube. The length of the adsorbent
tube can be varied with the intended application. For generalized
sampling, a 30 cm length is recommended. This will hold ~ 3 g of
charcoal, which, in turn, will adsorb over ~3mg SF, when the former
is in a moderately active state. With the use of Molecular Sieve
desiccant to remove the water and a majority of the hydrocarbons, this
capacity should exceed any anticipated atmospheric loading. Filling
the desiccant reservoir with 1.6 mm (1/16 in.) pelleted 5A Molecular
Sieve provides for water and hydrocarbon removal. By intermingling
a few granules of indicating Drierite with the Molecular Sieve, the
53
-------
need for bed replacement can be detected.
For desorption, the U-tube is removed from the body and the exit-
end sealed with a Swagelok 400-C cap. The adsorber is then placed
in a suitable oven, with the free end attached to a cryogenic vacuum
train as previously described. In most instances it will be necessary
to dilute the desorbate to a level that can be analyzed by conventional
gas chromatographic means. Sampling rate is recommended at
20-50 cc/min. Tests verifying these recommended procedures were
conducted at 25 cc/min sampling rate.
FIELD OPERATION REQUIREMENTS
On the present program, no field testing was attempted or required.
The system shown in Figure 21, which was tested only in the labora-
tory, could be used in the field. The vacuum was obtained from a
Millipore vacuum-pressure pump, Cat. Number XX60000 000. The
flow was controlled by using a limiting orifice, Cat. Number XX 50
000 01 and adjusting the flow through control of differential pressure
by means of the vacuum and pressure regulating valves on the pump.
The pump is 38. 1 x 20. 3 x 21. 6 cm and weighs 10. 8 kg.
Flow was measured by a general purpose mass flowmeter. A Matheson
Model Number 8110-0112 with a range of 0-100 cc/min is ideal for this
use. This unit with its transducer weighs 3. 4 kg. The meter is 19. 7
x 12. 7 x 12. 7 cm and the transducer, with its connecting cable mea-
sures 14 cm high by 5 cm in diameter.
Both the pump and the flowmeter require 115 volts AC, 60 Hz which
can be supplied by a small portable generator (0. 5-1 Kw), if standard
AC is not available.
Before sampling begins, the flowmeter is connected to the vacuum inlet
of the pump with its limiting orifice installed. The flow is then adjusted
to the desired rate by the vacuum regulator valve. The system is then
shut down. The inlet to the mass flowmeter transducer is connected to
the outlet of the sampler. The pressure drop through the sampler is
so small that only minor adjustment of the regulator valve is necessary
54
-------
to maintain the preset flow. In fact, because of this, no attempt was
made to measure the actual pressure drop, since it would expectedly
vary considerably with minor variables of the packing procedure (e. g. ,
the degree of compaction used in inserting a glass wool plug).
It is realized that meteorologists prefer battery operated, low power-
draw equipment. It is obvious from the observed characteristics of
the sampler that it will be compatible with the small battery-driven
gas pumps that are used in the field. An example is the Spectrex
pump. This device operates on 6 Volt batteries, uses a potentiometer
for flow control, weighs only 225 g, and has a cube volume of only 125
cc. By precalibrating such a pump with a sampler over various time
intervals using, say, a wet test meter, it should be possible to field
sample in an open loop mode using only the two components and battery
power.
55
-------
SECTION V
REFERENCES
1. Dietz, R. N. , and E. A. Cole, "Tracing Atmospheric Pollutants
by Gas Chromatographic Determination of Sulfur Hexafluoride, "
Environ. Sci. Technol. , 7, No. 4, pp 338-342 (1973)
2. Turk, A. , S. M. Edmonds, and H. L. Mark, "Sulfur Hexafluoride
as a Gas-Air Tracer," Environ. Sci. Technol. , 2, No. 1, pp 44-
48 (1968)
3. Altshuller, A. P. , "Atmospheric Analysis by Gas Chromatography"
in Advances in Chromatography, Volume 5, Giddings, J. C. and
R. A. Keller, Editors, Marcel Dekker, Inc. , New York, 1968
56
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
\. REPORT NO.
EPA-650/4-74-050
3. RECIPIENT'S ACCESSION-NO.
4. TITLE AND SUBTITLE
Development of Sampling Devices for Gaseous
Atmospheric Tracers
5. REPORT DATE
July 1974
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
C. L. Deuel and R. M. Roberts
8. PERFORMING ORGANIZATION REPORT NO.
2601-F
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Analytical Research Laboratories, Inc.
160 Taylor St.
Monrovia, Ca. 91016
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-02-1235
12. SPONSORING AGENCY NAME AND ADDRESS
13. TYPE OF REPORT AND PERIOD COVERED
Final Report 6/1/73-6/1/74
EPA, National Environmental Research Center
Research Triangle Park, N. Carolina 27711
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This report discusses the development and testing of an air sampler to be used to
collect over protracted periods and to facilitate the measurement of atmospheric
tracer compounds released in meteorological diffusion studies. Tests were con-
ducted to determine the adsorptive capacity of various sorbents potentially suitable
for collecting such electronegative tracer compounds.as SF, 'and CF~SF~.' A field-
practical sampler, incorporating the best sorbent found, a nigh surface-area coconut
charcoal, was then designed and subjected to laboratory tests. The effects on sam-
pler performance of various atmospheric influences, such as composition, pollutants,
temperature, and tracer loading and level were determined. Desorption techniques
allowing quantitation of the tracer were developed.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C. COSATI Field/Group
Air pollution, tracer, sorption, measuring
instruments, gas sampling, gas detectors,
trace elements, electronegativity.
Atmospheric tracers,
electrophilic materials
7/D. 13/B,
14/B
3 DISTRIBUTION STATEMENT
19. SECURITY CLASS (Tins Report)
Unclassified
2). NO. OF PAGES
56
Unrestricted
20. SECURITY CLASS (Tills page)
Unclassified
22. PRICE
EPA form 2220-1 (9-73)
------- |