\

W

W

i:
M

W)

H
HI

fl
N

W

W

lell

i

t

w m
\l—3 }
CDN

 

.THS_

mummy a? A was mmmrocsmmm
$EPAEAT30N ma DETECTEON
rem-swam: FOR WATER vma
Tmmwon mm mamas:

The”: {as 95m Dogma «:5 M. S.
MICHIGAN STATE UNIVERSITY
James CL leaking
1965

 

 

 

 

ABSTRACT

EVALUATION OF A GAS CHROMATOGRAPHIC SEPARATION AND
DETECTION TECHNIQUE FOR WATER VAPOR TRANSMISSION RATE
STUDIES

A chromatographic method for the separation and detec-
tion with quantitative measurement of water vapor was devel-
oped and evaluated. The applicatinns of this specific method
to packaging research are described. The applications of gas
chromatography to packaging in general are reviewed.

Evaluation of a porous polymer chromatographic column
packing material was carried out utilizing thermal conduc—
tivity and ionization detector gas chromatographs. The re-
spective sensitivities to water and water vapor of the gas
chromatographs were investigated and tabulated.

Several instrumental anomalies were encountered. ex-
plained. and effectively controlled.

The major finding of this research was that separa-
tions. detections, and measurements of water vapor in air
by chromatography is a successful method. It is indicated
that further deve10pment utilizing this method should be
pursued actively. Also, many other separatinns may be
uncovered by further investigation of the porous polymer
beads, and packaging research should benefit greatly from

these separations.

EVALUATION OF A GAS CHROMATOGRAPHIC SEPARATION AND
DETECTION TECHNIQUE FOR WATER VAPOR TRANSMISSION
RATE STUDIES
By

James G. Jenkins

A Thesis

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

School of Packaging
1966

TABLE OF CONTENTS

Acknowledgements ..............................
List of Tables ................................
List of Figures ...............................
Introduction
Purpose of this research ................
Present WVTR detection method ...........
Theory of Gas Chromatography

General theory of gas-solid and gas-
liquid Chromatography 00.000000000000000.

Columns: Their efficiency and resolution
limitations 00.0.0.000...OOOOOOOOOOOOOOOO

Instrumentation: Thermal conductivity
detectors and ionization detectors ......

Experimental Procedure

Effect of Water on Ionization Detector

response 0.0.0.000...OOOOOOOOCOOOOOCIOOOO

Sensitivity determinations of chromato-
graphy with reSpeot to water ............

Conclusions and Future Work
Results and conclusions .................

Future worli .0.00......OOOOOOOOOOOOOOO...

Page

ii
iii

U1 Kn #1 +4

10
16

16

23
“5
45
1+9

ACKNOWLEDGEMENTS

The writer is deeply indebted to H. E. Lockhart for
much helpful criticism plus much encouragement during
the course of this research, to L. B. Westover,

O. L. Hollis, D. Pegaz, and J. L. Mautz who gave
freely of their time and efforts, and to my wife

Mary Carolyn who worked long and hard to make this
thesis possible. I shall never forget.

I.
II.
III.

IV.

V.

VI.

VII.

LIST OF TABLES

Filament Stripping Experiment 0000000000..
conditioning EXperiment 00.000000000000000

Burrell K-7 Minimum H20 Detection Limit
Experiment Using a 6.5 ft. x 3/16" Column.

Burrell K-7 Minimum MeOH Detection Limit
Experiment Using a 6.5 ft. x 3/16" Column.

Burrell K-7 Minimum H20 Detection Limit
Experiment Using a 2.5 meter x 2.0 mm.
Stainless Steel Column ...................

Thermal Conductivities of Various Gases

at 0°C O00......00.0.0000...00.00.0000...

Compilation of Data from the Fisher Gas
Partitioner coo000000000000000000000000000

ii

Page
27
29

30

31

37

38

1+2

II.

III.
IV.

V.
VI.

VII.

VIII.

IX.
X.
XI.
XII.

LIST OF FIGURES

Weight of Water Vapor in 1.0 cc. Air vs.
Percent Relative Humidity OOOOOOOOOOOOOOOOO

Illustrations of Normal. Langmuir, and
Anti-Langmuir curves OOOOOOOOOOOOOOOOOOOOOO

Composition Profiles of Tungsten Filaments.

Conductance of Various TUngsten-Carbon
Compositions Measured at Room Temperature..

Structure of Porapak R and Porapak Q ......

Performance Data on a New Ionization
Detector. Burrell Corporation .............

Measured Area vs. Weight Loaded for Water

Sensitivity Determination on the Burrell
K-7 Instrument Using a 6.5 ft. x 3/16"

column OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO

Measured Height vs. Weight for Burrell K-7
Water Sensitivity Determination on 6.5 ft.
X 3/16" COlumn OOOOOOOOOOOOOOOOODOO00......
A Typical Chromatogram using Porapak Q ....
Present Chromatographic syStem 000000000000

Fisher Chromatogram of Water in Air .......

PrOposed Chromatographic System ...........

iii

Page

13

14
17

2A

32

33

35
no

4#
5O

INTRODUCTION

EEEPPSB of this research

The purpose of this research was to investigate and
evaluate the practicality of using a chromatographic tech-
nique. preferably compatible with existing equipment at the
School of Packaging, so that a statistically reliable method
of measuring water vapor transmission rates would be avail-
able to the packaging engineer. A second goal of this re-
search was to find a method by which a single measurement
of the components. 02. C02 and water vapor. in atmospheric
air would be possible. The final area of interest was to
investigate and describe applications of gas chromatog-
raphy to packaging research in general.

Any gas chromatographic separation is dependent on
the efficiency of the columns. The following character-
istics must be inherent to the column material.

1) The column material must be easily prepared or

commercially available.

2) The column material must be stable at temperatures

above 200°C.

3) The separation agent cannot bleed or react with

the instrument or components to be separated.

U) The column material must be easily packed into

small diameter columns and it must allow

necessary carrier-flow characteristics.

5) The material must separate a wide range of com-
pounds such as N2. 02. 002. water vapor, and
non-polar organic vapors.

6) The material must not only separate water but
be able to yield smooth symmetrical peaks.
Briefly. the column must minimize tailing
@dsorption-desorption) of water.

At present. there is no permanent record made of water
vapor measurements; this is not the case with the gas per-
meants. The gas chromatograph will enable absolute measure-
ments to be performed. Simultaneous comparison of water
vapor content and gas permeant content are but one reason
for attempting this research project.

If a gas chromatographic technique may be developed
for both gas permeants and water vapor measurements a
source of instrumental error would be eliminated. This
source of error is in using electric hygrometry for water
vapor measurements. The many advantages of using one de-
tection and measuring instrument for any given set of
measurements has been demonstrated by Lockhart (13) in
the utilization of gas chromatography for measuring
permeability rates through barrier materials in the
Davis cell (A).

Present WVER detection methods
A degree of laboratory SOphistication rarely found
in technicians is involved in classical methods of measur-
ing water vapor. Among these methods are gravimetric, chem-
ical. electric hygrometric, dessicant-cup. and pressure in-
crease methods (8.12.lh.15.16). Gas chromatography has
been demonstrated to be a simple straightforward measurement
device. It is used for routine testing as well as research.
The School Of Packaging is presently using a testing
system developed by H. E. Lockhart that measures gas per-
meants by gas chromatography (13). The water vapor is
measured by electric hygrometry (29). The level of detec-
tion is indicated by Figure I (20.26); a sensitivity of at
6

least 1.0 x 10' gm./cc. for water must be obtained for gas
chromatography to be a practical method of measuring water
vapor in air.

The collection and tabulation of permeability data
is essential for future design of packages as the main

cause of product degradation in many cases is due to water

vapor transmission and gas permeability.

FT SUP? I

‘- L4

Height of Water Vapor in 1.0 cc. Air vs. Percent
Relative Humidity

 

o 0 .
23 C. (73.4 F.) 20.23 gms. 190 vapor
cubic meter of air

100a

Percent 601
Relative
Humidity

 

04’ 2 Q 2 A? 10 12 I3 6 18 20
Gms. x 10" of 0 Vapor/cc. of Air

 

‘ I I
b-A—d \.
{‘0

THEORY OF GAS CHROMATOGRAPHY

General Theory of GaseLiguid and Gas-Solid_Chromatoggaphy.
Small diameter columns of various solid supports,
which may or may not have been previously coated with a thin

layer of some organic compound have been used for chemical
separations for years. If the solid support is coated, this
coating is usually a high boiling liquid or solid. This
coating is commonly called the liquid phase.

A sample of various components may be inserted into
the front of the column by some suitable method. The sam-
ple is usually transported through the column by the use of
an inert gas called the carrier gas. Separations are
effected by differencesin vapor pressure, solubility of
the sample components in the high boiling liquid coating the
column solid support. or in the case of using a solid sup-
port alone the elution and separations are dependent upon
surface adsorption. molecular interaction between the sample
and this support.

The eluting components are detected by some means.
the more pOpular being thermal conductivity cells and ioni-
zation detectors. A series of signals are given by these
detectors; when these signals are diSplayed as signal in-
tensity versus time symmetrical peaks approximating Gaussian

distribution curves are given. and the areas under these

curves may then be normalized to obtain the percentages of

the sample's components.

Columns: Thei£_efficiency and resolution limitations.

Sufficient column length must be present to separate
the desired components completely and quickly.

Means of describing a column's efficiency are approached
by many workers in various ways (11.23). The method des-
cribed here is based on the theoretical plate treatment
of James and Martin in which N. the number of theoretical
plates. is found by using peak dimensions.

(A) N = 16(X/I)2
N may be calculated using Equation (A) where X is the re-
tention time and Y is the baseline width intercepted by
tangents to the points of inflection of the peak.. (See
Figure II-A). In general. the separation requiring N
plates by normal distillation is given by N2 theoretical
plates. One other comparative method of measuring column
efficiency is commonly used (11.23). This measurement H
is defined as the height equivalent to a theoretical plate.
It is obtained from N and the length of the column. L.

(B) E = L/N
In Equation (B) a measure of band broadening is essentially
given as width of peak base. I. during time of elution. X.

See Figure II-A. A means of comparing column resolution is

FIGURE II

Illustrctions of Normal, Langmuir,

signal

s i gnal

Curves

 

 

 

 

 

II-B

 

L_____P time

and Anti-Langmuir

elutinr
——-———————-&
sample

 

given by
(c) a = (II - Uta/x1 - 1)

This expression for R, the distance between peak maxima, is
expressed as a multiple of the standard deviation of the
first peak. It is clear from Equation (B) and Equation (0)
that if the resolution, R, is to be doubled, the cOlumn length
must be increased by a factor of 4. Also the effective
resolution may be increased by increasing the ratio of
liquid phase to solid support within practical limits.

Along with the resolution the symmetry of the peak
for the eluted component is dependent on adsorption-desorp-
tion effects taking place on the surface of the solid support
and on the instrument walls. The instrument must be non—
reactive, clean, and it must possess small hold up volume
in order to minimize tailing and spreading effects (2.10.11).

The shape of the curve obtained in gas chromatography
may be varied in three ways (11,23,2h). The first varia-
tion is shown in Figure II-A. The nearly Gaussian peaks
are obtained when the concentration of the solute in the
stationary phase is directly prOportional to the concentration
of the solute in the gas phase. It is seen that the retention
time is independent of concentration. The type called
Langmuir or Classical Isotherm as shown in Figure II-B is

caused by the main portion of the solute band being eluted

more rapidly than the leading front edge due to a limited
area of adsorption available. Langmuir curves are given by
all columns to some degree. Peaks of the Langmuir type are
usually Observed in gas-solid chromatography. In this type
of column the retention time is a function of sample size
and normal narrow peaks are Obtained with micro amounts. A
third type is shown in Figure II-C. Its shape is due to
the main portion eluting more slowly than the extremities.
Again, retentinn time is a function of sample size which is,
of course, dictated by sample volume available and the sen-
sitivity of the detector.

The column is the heart of a gas chromatograph be-
cause the actual separations are Obtained in the column.
In gas—liquid chromatography high thermal stability and
low vapor pressure at the Operating temperature of the
instrument must be possessed by the liquid phase. The
separation of homologous mixtures is easy, and, of most
importance, azeotropes are not formed in homologous series.

The injection of samples must be done in such a manner
as to approximate a "plug" introduction of vapor. For
optimum results the smallest possible sample consistent
with detection sensitivity is desirable, but this efficiency
is reduced if the sample vapor is excessively diluted with
carrier gas.

The rate of elution as well as the degree of separation

10

is affected by the temperature of the column and the flow
rate of the carrier gas. It is necessary to Optimize these
variables in such a manner that a rapid, precise, and yet
simple technique may be available to the packaging engineer.
Throughout the remainder of this paper the term "gas
chromatography" is used instead of the limiting terms "gas-
solid" or "gas-liquid". Since the development of solid
supports which will separate both polar and non-polar substi-
tuents the scope of the field is widened in gas chromatog-

raphy--both gas-solid and gas-liquid.

Instrumentation
Thermal conductivity detectors.
A thermal conductivity cell is basically a balanced
differential bridge by which the unbalance caused by various
characteristic thermal conductivities of materials flowing
over one arm of the electrical bridge is measured. These
are of two types--the hot wire filament and the thermistor
thermal conductivity cell. Samples as low as 10'9 to 10"10

grams can be detected using small volume thermal conductivity

cells (6,11,23,24).

Ionization detectors.
There are two pOpular types used: the direct ioni-
zation detector and the hydrogen flame detector. The hydro—

gen flame is completely insensitive to water (23) and will

11

not be discussed here. The argon detector is a type of
ionization detector which Operates on the principle of
forming metastable argon ions (6.23) as a source of energy
by which free electrons similar to those described by
Ryce and Bryce (22) are produced. The separated components
are ionized by the free electrons not much differently from
the way a sample is ionized by a mass spectrometer (3). Ion-
ization detectors are characterized by their fast response
in the range of milliseconds (23,2#) and high sensitivity;
10"12 gram determinations are not uncommon fOr non-reactive
organic compounds (23).

Any material capable of being heated to incandescence
and emitting free electrons will constitute a filament
for an ionization source. These free electrons are accel-
erated by a positive potential which gives them enough kinetic
energy to ionize molecules. The molecules are ionized by re-
moving electrons from the electron orbits of the molecules.
The ions formed migrate to the collectorgiate by virtue of
their positive charge. Upon striking the negative collector
the ions discharge by accepting electrons from the collec-
tor. This electron flow amounts to a minute current and
this current, when amplified by an electrometer, is a sta-
tistical representation of the sample introduced into the

device.

12

In the case of a filament manufactured from pure
tungsten it is necessary to precondition it with a hydro-
carbon. The basic reaction is to form two distinct tung-
sten carbides.

1W + ZHZC=CH2 .9 WZC + WC + w + 2CH4
A diagramatic presentation of this may be seen in Figures
III-A, III-B, and III-C. Figure III-B is the ideal filament
which has been described by Sharkey and many others
(3,7,23,27). It is possible to overcondition a filament
and this filament structure is represented by Figure III—C.
The reaction involved is Similar to the above but it is
driven further to the right causing a larger percentage of
WC to be formed at the surface. Supposedly the carbon
permeates (diffuses) through the tungsten metal and WC
forming a large build-up of W2C in the center of the filament
(3).

A graph of conductance (reciprocal of resistance) is
shown in Figure IV. The minimum conductance, which is the
maximum resistance, is found at the point where the filament
has the greatest amount of W2C present at the emitting sur-
face. Briefly, this is the point where the electron emission
is Optimum with reSpeot to a constant filament heating
current. As previously mentioned, the filament is the

source of free electrons. These free electrons when given

13

 

1,. _ -...

BEACH f‘om YeYdOn. 1““s ~n~3tro~otnr nwa th
" ‘1 . n4“ am 1 A 4— , ”1’ «L .‘ q
(LP: -1" I") ‘ r) LIA-4") 1n dii L‘lejo P. 1000

 

 

III-A Pure tungsten before any conditioning

 

VJ 2 C

NW

 

III-B A properly conditioned filament
(III-A + hydrocarbons)

% 'IT
M

III-C An overconditioned filament
(III-B + excess hydrocarbons)

 

A

\\\\\\\\\‘1

\\X\\\\\\\

 

\\

 

 

N

14

*3. ’n v
rlpure If

Conductance of Various Turgeten-Ca.bon Compositions
Ieasurcd at Room Temperature

 

7

 

 

O
r—
{‘0
b)
4:
Um
O\

on

\O
\Q
\0
C3
\0
-\)
*0
C\
\ C)
Kn
\

100

7 [4‘3

 

.711 - o T) , v. r: 4- 7.-4- - —~., A -~ ° '-

ragea LTOW neynon. Lass cpectr01etrg anfl Its npgllcatlooe
4- -...~'. n1...‘..,- 4. , H- a n
LKD‘DTPfihwtg ulfljnlSLJéK. IAiQC?.LOy.

 

15

enough kinetic energy will ionize sample molecules.

As part of the description of the ionization source
it must be stated that the ionization of materials present
in the source will give rise to one ionization signal un-
less the ions are separated by some electrical or magnetic
means. A chromatograph will. in most cases, allow Just one
single component to be eluted at a time; in turn, this sin-
gle component is ionized and the ions detected are representa-
tive of that particular Specie plus any residual components
present in the ionization source at the time of ionization.

There are two gas chromatographs available at the
School of Packaging. The Fisher Gas Partitioner is a ther-
mal conductivity instrument which Operates at ambient
temperatures (stabilized by a constant differential heater).
The other instrument available utilizes an ionization detec-
tor and is equipped with a programmed column heating element.
It is the Burrell Kromo-tog K-7. Most of the evaluation
was performed using the Burrell instrument in order to
utilize its high sensitivity and column heating equipment.

A tungsten filament is utilized in the Burrell Kromo-
tog K-7. This tungsten filament is heated by a potential of
18 volts. and an emission of 0.8 milliampere is obtained
when 5.0 to 6.0 amperes current is passed through the tung-
sten filament causing it to become an electron emitter

(Edison effect).

16

EXPERIMENTAL PROCEDURE

Effect of water on ionization detector_response

The first step in this procedure involved the selec-
tion and preparation of a suitable chromatographic column.
Initially. 5% Carbowax coated on Teflon powder was con-
sidered to be the best water separation column. It has been
reported by Kirkland (10) that this column packing separated
liquid water with minor tailing. Kirkland also describes
cooling the Teflon powder (after coating) to 0°C. and
vibration packing 3/16” and l/h" one meter columns success-
fully. Attempts to pack columns at room temperature invar-
iably result in aggregated Teflon and undesirable plugging.

In the past few years at The Dow Chemical Company gas
- chromatograph methods of measuring water were tried and
evaluated using the common technique of coating a fluorine
containing polymer with a suitable liquid phase. That much
of the tailing (Langmuir effect) might be eliminated by the
use of a uniformly sized support (microspheres) was confirmed
in a personal interview with L. B. Westover (30). A solid
porous support had been developed by 0. L. Hollis (9) and
was being released by Waters Associates, Inc. of Framingham.
Massachusetts under the name of Porapak was also made known
in this interview. Two of the more common Porapak structures

are graphically demonstrated in Figures V-A and V—B (9. 17.

17

Frans: v

Structure of Porapak R and Porapak Q

V-A
Porapak B----a copolymer of
divinylbenzene and
styrene. The styrene
molecules are those with

solid rings.

 

V-B
Porapak Q-—-—a polyethyl-

vinylbenzene,

18

18.19.30). Both the liquid phase and solid support functions
in the column are actually served by the porous polymer beads
(17) for chromatographic separations.

Due to the availability of this chromatographic column
packing. Porapak, which packs easily and needs no liquid
phase coating. Teflon powder support was not investigated
further. The Porapak column had been recommended by West-
over (30) and Hollis (9) for the separation Of water vapor
from air. Hollis suggested using a small diameter column
at least 6 feet long and Operating the column above 100°C.
Past experience with coated Teflon indicated that bleeding
Of the liquid phase would be another disadvantage with the
Teflon column. The Porapak does not bleed even at temper-
atures of 250° C for it has no liquid phase tO bleed. Pora-
pak exhibits the unique prOperty Of separating polar and
non-polar compounds with virtually no tailing of either type.

A column 6.5 ft. I 3/16 in. 0.D. aluminum was packed
with 18.3 m1. of Porapak Q by vibration. This column was
conditioned at 230° C. for 4 hours while being continuously
purged with helium. This column was bent in the hairpin
shape common to the Burrell instrument and attached to it.
While Optimum column conditions were being sought. the emis-
sion current was noticed to drOp by a factor Of 25% (0.8 to
0.6) during detector response. Due to this loss Of emission

current. the chromatOgrams were in nO way reproducible.

19

The reason for this was that the tungsten filament was
being stripped by the oxygen-containing samples. This
tungsten stripping effect is well known in mass Spectro-
metry and is referred to frequently in the literature
(1,3,11,27,28). Also, insulating deposits will be
released by accelerated decomposition in the ionization
chamber due to the presence Of oxygen-containing compounds.
Erratic Operatinn will be caused by the release Of these
deposits, especially if these deposits are subsequently exposed
to electron bombardment from the filament, and the back-
ground will be high. The filament itself will be damaged
by some of these materials in this process Of decomposition.
As an example, water vapor is decomposed when contact is
made with the filament; the oxygen is chemically combined
withthe filament and hydrogen is evolved as atomic hydrogen.
H20 4- WZC —-—-> we + WC + 2H

H O + W --Ԥ W0 + 2H

2
230+wc —-—-> wo+co+uH

2
At the Operating temperature of a tungsten filament
the tungsten oxide is distilled to the walls Of the ioniza-
tion chamber. There it is reacted with the atomic hydrogen
to reform tungsten metal and water so that the process is

repeated. The process is not reversible because the tung-

sten metal is now deposited on the waIlsurface.

20

NO + 2H -—e> w + HZO
This effect is readily noticed in a mass Spectrometer after
many water samples are run, and the effect is shown as a
loss Of total emission and then the filament must be
reconditioned by purging with a hydrocarbon. The loss of
tungsten is irreversible, and in Order to Obtain the same
total emission level the filament current must be contin-
uously increased. This type of reaction proceeds readily at
pressures Of 10'12 mm. Hg. and as the pressure Of the ioniza-
tion chamber is increased the loss Of emission becomes notice-
ably greater.

If carbon-containing compounds alone were present, the
surface Of the filament would be chemically changed to main-
ly WC, the other carbide of tungsten, and the carbon would
permeate to the center Of the filament forming WZC. This
type Of filament is shown in Figure III-C. An entirely
different work function Of electron emission is possessed
by this filament, and this emission will be much less than
that Of the case shown in Figure III-B for the same amount
Of input current. All these reactions take place in mass
spectrometers Operating at pressures Of 10'12 mm. Hg. In
the case Of the Burrell K-7 this deleterious effect will
be accelerated due to higher ionization pressures Of 0.1

and 1.0 mm Hg (6).

21

The effect would be Observed for small amounts Of an
oxygen-containing sample. When a large amount is suddenly
introduced the emission current decreases sharply, and
erratic non-reproducible chromatograms are caused by this
decrease. TO Obtain reproducible ionization of samples some
source Of hydrocarbon must be introduced to maintain the
filament as shOwn in Figure III-B. The chemical composition
Of a properly conditioned filament has been determined by
Sharkey (13.27.28) as previously mentioned (See Figure III).

The Optimum emission will take place when the carbon
content is 3.16%. This may be seen in Figure IV. ‘One is not
concerned with the exact chemical composition of the tungsten
filament being exactly 3.16% carbon, but rather that the
filament easily reaches some stable reproducible level of
carbonization. It may be readily demonstrated that when
a slightly conditioned tungsten filament is exposed to an
excess Of any hydrocarbon a further production Of W20 will
take place and the resistance Of the filament will increase
(1.27). This resistance increase will cause the filament
to be maintained at a satisfactory emission level by succes-
sively lower quantities Of filament heating current needed.
Conversely, if oxygen is introduced, the filament heating
current must be increased to maintain this same level Of
electron emission. The exact reverse behavior would be ob-

served with an over-conditioned filament (3,28). The

22

Burrell instrument has no provisions for automatic electronic
control of the emission level (6) and drastic decreases and
increases in the total emission current are noted during

the Operation Of the instrument. In this research, water
vapor decarbonized and methanol vapor carbonized the tungsten
filament in the Burrell K-7.

Of interest is the section entitled "Analytical He-
sults" found in Performance_pata on a New_;pnization Detectgg,
a Burrell Corporation publication (6). The analysis of
hydrogen, oxygen, nitrogen, methane, carbon monoxide, argon,
carbon dioxide,ethy1ene, and acetylene is referred tO as
the analysis Of a "fixed gas." These gasses were adsorbed
on molecular sieve and silica gel giving reproducible quanti-
tative chromatograms. The reproducibility is caused by the
presence Of hydrocarbons in this "fixed gas" sample. In
general usage, the constituents N2, 02, C02, and water vapor
are found in atmospheric samples, and are denoted by the term
"fixed gas." Atmospheric samples dO not generally contain
methane, ethylene, or acetylene in any great amount. A fil-
ament having a coating Of tungsten carbides wzc and wc coated
on pure w is Obtained in the presence Of those hydrocarbons.

Also shown in the Burrell publication (6) is a chromato-
gram which contained 90% water and the remaining 10% made

up of the hydrocarbons methanol and ethanol. The ionization

23

energy was reduced after the methanol and ethanol were eluted;
Figure VI is an exact scale drawing of reference (6) with

the dotted line drawn in to indicate water. If this exper-
iment were to be repeated with ionization energy great enough
to ionize the water, erratic chromatograms should be obtained.
It is not stated in the Burrell publication anywhere that
water-containing compounds may be quantitatively analyzed in
the absence Of hydrocarbons. The non-reproducible Situa-
tion noted in this research may be overcome by prOperly con-
ditioning the filament by the use Of hydrocarbons during the
analysis Of small amounts Of oxygen-containing compounds.

It is seen that, unless a prOportional amount Of hydrocar-
bon is present when oxygen-containing samples are ionized,

the electron emission will not be linear; thus, the signal

measured will be non-linear also.'

Sensitivitydetermination Of chrOmatograph gith respect to water
Even though the Burrell ionization instrument gave less

than completely successful results, the anomalous conditions

are instrumental in nature.and would be eliminated by care-

ful conditioning or instrumental deSign changes. These

changes are covered in detail in the section on further

work. '

Excessive high background was Observed at maximum

sensitivity and minimum attenuation Settings. The background

24

flT.flTTH ‘1 T I
d .- ‘ .
l —L J iv1‘AAJ \]

a -n.m ~ 1“ L ‘ T - c . r'r
Performance Data on a let Ionics

Turrcll Corporation

100

vita—nirv-
1'1 1. L LLJ.

/

\O
O

CO
0

[1’0 AIR I

 

 

ZING “vs "V

1.41. ".11. K1 .1.

PD
til
U
C:
O
is
C)
H
O
n y
- 4

A

tion Detector

ALCOHOL

G—-ETHYL ALCOHOL

 

0 1+ 8

 

MTYJW H 1T Tzfi'fpa
Lilla: "' liIiQL‘ldtJ

Fl
0\

12

25

was in the same order Of magnitude as that of water vapor
in air. Due to high background levels the water vapor de-
tection experiments were performed at lower sensitivity
and higher attenuation settings. The values Obtained are
relative values,and, in fact, will be 50 times lower than
could be Obtained by Operating at the maximum sensitivity
of the Burrell instrument.

Various solutions Of water in methanol were prepared
to determine the minimum practical limit of sensitivity Of
the Burrell K-7 for water. The use Of methanol was dictated
by the Observed deleterious effect Of water alone upon the
tungsten filament in the Burrell KromO-tog instrument. There
was still a noticable drOp in emission current, but this was
due to the cooling of the filament below its most efficient
emission temperature, and, as there would always be a hydro-
carbon (methanol) present, the normal behavior of a properly
conditioned filament will be experimentally created.‘

A series Of chromatographic experiments were per-
formed on the Burrell K-7 using a 3/16 in. 0. D. aluminum
column 6% ft. long packed with 18.3 ml. Of Porapak Q resin.
The column was tested between 75° C. and 200° C. with
varying helium flow rates in order tO Obtain relatively
"tail-free" curves. All tables have the exact instrument
Operating conditions Specified on them. The first experi-
ments (EXP BK-7-1A and EXP BK-7-1B) were performed to

26

Observe the variation Of the peak area for pure water when
no organic compounds were present. The 1.0/.1. sample was
injected by a microliter syringe through the Burrell syringe
port. TO eliminate any question of loading error, the
syringe was loaded, weighed to the nearest 0.1 of a milli-
gram, and injected; the syringe was immediately weighed
again. The 1.0,al. load Showed no detectable variation to
0.1 milligram when the loads were weighed and injected
carefully. 'Also the sample was drawn up through the syringe
by a Squeeze bulb. This was to insure that no air bubbles
were trapped in the syringe needle. Due care was taken in
injecting the sample through the syringe port, as this ~
port was Operated at temperatures in excess Of 100° C. for
all Of this project. The method used was to insert the
syringe fully into the port, immediately to depress the
plunger, and to remove it in one swift motion. If this
were not rapidly done. the needle would become hot and
preferentially volatilize lower boiling constituents.
The results Of these two eXperimentS are Shown in Table I.
A variation Of -h9.h% to +31.3% is demonstrated by these
experiments, and this variation is due to the.water stripping
problem mentioned previously.

A second experiment was designed to demonstrate the
effect an organic compound has upon stabilizing the emission

characteristics of the filament. The results of this

.4.

.e u a ego on u m n.2_ one eta mo cnoanaeeoo oe.m ”mfi-s-mm mum

.pgomfiaem.. clear:cc .p,omoag candoao on
Q\. I a. u .. 3.1 .343 '3‘ A 0.1_\|O\lrs
cm I a or m wg r oH_t:mc r H Hfi5\. 00 u c

\u :1... :1 I)... o)“ . e I..- 4.4 ...1...
CUP NH paw. CEDHOO Eigw rnrfidw :wfi\m 1r. .CCQCW ..Crlhwlwttr \Li.

ssfiaoz once Beam ..0

 

27

Cl
C 1
C x)
l
\C
If x
\r‘)
+
Q
0‘
O
1. '\
0
€ \
\f \
U\
0
O
O
H
:9-

J
a

\"1‘
C
0’
+
C
1
C
I
\Q
O
0
CK)
H
C-
L'\
0
;)
r-I
O:
C’ \
(\3
C \
O
C \
(\1
O
O
O
H
O
O
H
OH

0..
C’ )
C‘\
+
W
V—I
kW
.2
O
( N
H
C?
11W O
O
.D
( J
O
O
0
r1
C)
O
H
(‘1

\.O\1II ,J\o.1)\l \\ o .1 \o\.\. 0 0|. I

..i c; c ec+ woo Hr 0L Cc eve o o e r
-..n e .d1. .-1.
(If AL kknF L:

C \3
H
C J
"t‘
if
C \
I
~)
{\
\3
Q
as
C‘ \
H
O
C)
r1
C)
O
H
L;

ee.m . em.mfi+ \. em.m cm.ofi 02.0 o.a m
FULL Smear
e.Ce- mfi.e + om.“ eo.wfi me.o o.H m

((\
r4
0 \
+
CD
0
O
H
I
(\I
o
D-
(\1
O
("x
H
C)
H
o
H
O
o
T‘I
v—I

 

 

4a-nnwm an
..no 3. .r .....3. 3.5 ram .M a
cone .o: .am Wong oamecane .pa .nm ogmaom seem nsmao: Hodge
r... 3.0.40. ,3.. EQHP CO H3 ma 3an . ..1. >0 .mo $0.0m. mtflm @055... Heurim Him 9b mHQuH Gm...

 

28

experiment are given in Table II, and it is interesting to
note that between Trial #1 and #2 no organic was introduced
to the previously conditioned system, and a drOp of 15.00 cm.2
in the water peak area is noted. This is calculated to be
a variation of 45.5%. The organic (methanol) 1.0/#1. load
was injected between all of the other trials and after
four injections of methanol (each followed by a 1.0/“l.
load of water) steady state conditions were again attained
by the instrument. The importance of prOperly condition-
ing this instrument before attempting analysis work cannot
be overstated.

An experiment designed to determine the minimum water
sensitivity of the Burrell was carried out by using a blend
of methanol and water. A conditioned filament was created
by the introduction of a source of organic material (methanol)
and progressively smaller amounts of water. These injections
were performed without sacrificing the previously demon-
strated precision obtained by using 1.0/“l. sample loads.

The results of these experiments are given in Tables III and
IV. The practical minimum sensitivity of the Burrell K-7
instrument with electrometer set at a sensitivity of 10.

and the attenuation set at 5 was no greater than 3.000 x IO'u
gm./5 cm.2 for water in the presence of an ionizable hydro-

carbon. A graphic diSplay of this data from Tables III and

IV is shown in Figure VII. It is shown in Figure VIII where

29

 

 

Conditiofiir“ ‘"f0?‘u“flt
F

- ’ o v --_v o ‘T’ZC‘I (‘1
“I h ‘ ..‘I '4... 1 ‘1 ‘1' 7’) ,f! T" .,.1.. . ‘ ”ML: 4...» x ..a 1....
--1 .oignt Fa-“ a r--- “71‘.” Triccrl.

1 1.0 n:. 2.;0 cm. 23.30 on. 33. 9 cm.

- ('3

2 1.0 ug. 1.75 on. 21.00 cm. 3.38 cm.

3 1.0 Mg. 2.7. cm. 16.30 on. 27.00 cm.

I .

4 1.0 V”. 3.00 on. 19.20 on. 28.00 cm.

/ . ,.
5 1.0 T‘f . 3.05 0:71. I‘LJQEO CH1. 33.76 0171.

‘\‘I0\

H
O

O

O

k.) I
I

\ 1
O
[...L
CO

0

\. n
O

'3 '30
‘ZQJIEJ T11.

\ .3)
O
C\
O
O

a
I—A
CO

KR
O
9
...3

In F“

‘U.

 

U? rT1-j.
-. ‘&J_J.

TVP P"
IJJ‘A. JiL

introduced a
introduced
until instrument

\ ,3 J. .1.
rar- .1r _ _
.)t\.d.uu. ESL/8.00.

‘ 1
a.proac:,o

~7—2: Speed 2, 3/16” aluminum column at

2000 C., Holiur flow 68.5
3 =10, .51 =2 10.

oc./min.,

(\3

0.)

NM

(\D

PO

30

.mUmz Sp
psmasmfiu .m u a .ofi u m ..o own ..sflg\.oo m.
.fi wmmmm .6 xmmmpom Spas QESHCO ESCHEfiHm .pm

C)

 

 

 
 
   

 

ofi.ofin mm.m- mow.m OH.: om.fi mmfim.o :
Hm.m - mm.m- mam.m om.: m.H mmfim.o m
:fl.m + mo.n+ :wm.m moa.m m:.: 00.: mm.fi mmHN.o m
oo.flfi+ mm.@+ omm.m om.d o¢.H mmfim.o H
om.fi . oo.m+ omo.m om.m om.fi mumm.o m
\ mnfi.m mm.m . \
CU.H + 00.3- omm.m om.u mq.a vmmu.o a
oq.u + Hfim.o+ omm.mfl mfi.wfi om.fi mam:.o m
cm.qfi- mmo.o- amo.mfi mam.ofi mo.mH oa.mfi om.fi mum:.o m
(u.m + Hum.o+ omm.mfi ofi.wfi ou.fi mam:.u H
EL .50 .80 .FC .50 .bh
mmu< .pm .xm mmg< mammmfiue .pm .xm psmfimm ozmm unmamz Hmfipa
mmmgmbm anm SOprabmmkfi .®>< .mo ampm mmmhm>< mmmw Hafim maaidm
msfil.

sazfloo :ofi\m x .pm m.@ m
pgmaflhamwm pfiaflq Sofiuompmm o m ESEHSHE mam Hammpsm

HHH mum.v

 

31

 

 

.m M Q .OH H m ..o emu ..:H_\ 00 m.mQ C H5HHo:

.H woman .w Mammhom 55H: SeSHoo EDSHHSrd .pm moo \ M :mfi\m Hfiumumm NH
55.5 - :H.o- oom.mm OH.mH om.m oQNQ.o :
HN.H - mu.m- - oQo.mm 0H.wH cm.m onQ.o m

mmr.or mm.mH
--- :H.:+ oun.om 0Q.0H oH.m o CQNQ. o m
B.HH+ mm.m+ omm.:m om.om oq.m :5 Q. o H
035.0 + --- QQQ. m OH.mH om.m onwm.o m
Hmw.om H.mH -
omm.o - --- mwo.o« 0H.mH VH.m oumu.o H
:Q.N - mc.m- ouo.mm OH :H 03.x gmmm.o m
NQ.m - Mo.m- mHm.Hm cum.mm m..HH 0H m5.m wmmm.o m
mm.Q + mo.m+ 0mm.Qm om.QH oQ.m arm .0 H
.,:g m.n:u .-o :50 .50 .mwu
m¢5< .5m .55 Wm5< QHQQQHHB .55 .mm pgmHmm Qsam pgmHm: HmHHm
mfldhwbfl thh HHOHWZ. >63 1 .O>< .mo deS vmfih®>d Mdom HHSm vdmfidm

x?

 

casHom =QH\m x .55 Q.Q m mgHmp
pCmEHMHw 9H1 pHEHA :chompcm Mom: has; H: mum Hamppfim

L

>H Hand?

I';('\f\ ."1
A 42'.“ \.

natnr
in

milligram

“- 7‘ ‘1' V“
‘PLJ ~.'.’v -,l.

o
6

L)

,_J

FIC

' T ,. A ”,1...
.d' t" |_ . 7-
1L

ff.

32

T

4..

.L/

7:

f3 "1
l.

0;-HJ

Loadcfl
v-7

coljmui

1 7" - A
UK} .0: L",1.‘-J“"‘= J '1

L I.—
3/1

VTI

6n

A)

Q? 4—
.r 3“
.4..3.LIO-L

rument

uifiitlviby
Tfii‘r

15.).L;
\/

 

 

15

("‘3

M

1

33

10.0

lm.o mamamflaaflg
CH
no.0 nSchmaoo

 

\m N .um m.m :o COHQGQHEpmme
m Hamupsm 50% psmfim: .m> psmflmm wmuzmdmm

HHH> mmerm

34

sample weight in milligrams is plotted against peak height
in cm. that 0.05/11. (5 x 10"5 gm.) of water must be present
before water detection (any signal) is observed. There

must be enough signal to measure area before the water per-
centage may be calculated in any case.

A graph of grams of water per cubic centimeter versus
relative humidity at 23° C. (73.40 F.) was given in Figure I;
it is indicated that at least a sensitivity of 1.0 x 10'5 gm.
of water at 50% relative humidity must be present in any
instrument for detection at the 50% B. H. level.

In order to compare the resolution efficiency of the
Porapak column Figure IX is included. This is a reduced
scale drawing of Trial #1 (EXP BK-7-3). The elution time
for water is 2.58 minutes at 750 C. column temperature with
a flow of c468.5 cc./m1n. of helium in the Burrell K-7
instrument using the 3/16 in. x 6% ft. aluminum column.

A second column of stainless steel. 2.0 mm. I. D.

x 2.5 meters long was packed with 10.2 cc. of Porapak Q

and conditioned at 230° C. for 0 hours constantly being
purged with helium. On this column an experiment was run
to determine the relative sensitivity of this column versus
the 3/16 in. column. The chart Speed was set at i. the
sample load was 2.0/41. of distilled water. and the column
flow was 40.0 co./min. with a column temperature of 125° C.

The average of three determinations was 160.1 cm.26al. of

35

mafia
msflmmopcsfi

 

Hmcmflw we
huHmSmpCm

 

 

[J

 

 

  

 

 

 

 

d xmamaom mQHmS Emamopmsoaso HmOHmmg d

 

oom

C C
a. \Ja.4 O 7.7-1 N.)- .1
U00 t...- t...... 0 r

 

 

firmHQ amass-HchSuem

.0 0mm .9
on .Qflfi\.oo
.SHE\.Qfl w e.

at» CSSHOO
mqu seHm
toom padso
WHH...

<

OH H m

36

water loaded. These figures may be found in Table V.

One last experimental procedure was attempted with
the Burrell instrument. The instrument was stabilized
at 125° C. column temperature and a flow rate of 00.0 cc./
min. He was established. The sensitivity was set at 10 and
the attenuation at 1. These settings gave very high noise
to signal ratios. A sample of 1.0 cc. of room air was
injected. The instrument did not give a response for water.
The emission current drOpped to such a low level that it
was not possible to increase the filament heater current
high enough to obtain 0.8 milliampere emission current.

This experiment was attempted 5 times and no meaningful
water separation chromatograms were obtained.

At this point it was necessary to evaluate the thermal
conductivity detector using the same column packing material
in order to complete column and instrumental evaluations.

A survey of the literature indicated that thermal conduc-
tivity is not as sensitive as ionization detection for
organic vapors. A study of thermal conductivities of var-
ious gases and vapors indicated that the detection of water
vapor appeared possible by thermal conductivity provided

a large enough sample is used. See Table VI.

Table VI is a compilation of selected thermal conduc-
tivities taken from International_Critical TablesI Vol. 5

"Thermal Cnnductivity: Gases and Vapors." pages 213 - 216.

37

H n m ..e emmH ..eHe .ee
N CESHCO HCCPm mam-fi-SH?P

C ‘
UJ

H! \I ‘14.
o —l t»,
4 s
\1 \l _) o 4 4 4.)! A44 0‘1!
.l. u.. ' I -l .
n1. f- (..IU 0r Hep B KFr-r r.|rFP

 

.\. .‘IJ

CH eon mm.m- m

H
@H.mH+ em.m+ Hs.QH mm.:H
m

 

HQ.OH+ mH.e+ s.eH s.om ms.H o.m H
0.20 d.:He .ro .To .so AW;
mess .pm .mm moms mfiwsm a .pm .xm psmficm Usmm psmHmm adage
mnsaebs moan Sempmemu . .mbd .so swam enmaobm ysem HHSh mflamdn

.

L

 

 

:uufloo Hompm amoflsflmpm .EL Ohm
lacmflucmum. .vfih Hmoflpoopofl -cww

w momma m.u

(

s5sHsHs a-a

H‘H

inermal Conductivities of Various

12’ ..° ..
n = Lila-Hr

 

 

"1.31" 1"

Fe 13.9
Air rxz-.23
21:2 ~15.9
1120 ”209.0

: i
T‘s.)
[\J
.
(\)

CO

39

The values are given at 0° C. which is a common reference
point for which all could be compared meaningfully. That
thermal conductivity is very sensitive to water vapor has
been verified by much previous experimentation (2,9,10,11,23).

Due to favorable thermal conductivity of water vapor
and the fact that thermal conductivity does not destroy
the sample being analyzed as does ionization detection, an
experiment to determine the Fisher Gas Partitioner (Model 25V)
water sensitivity was performed.

An aluminum column 3/16 in. 0. D. x 3 ft. was packed
with 10.0 cc. of Porapak Q and it was conditioned at 230° C.
for 0 hours. This cOlumn was installed in the Fisher Gas
Partitioner. This instrument was used as a one column in-
strument by removing the drying tube and column #1 (see
Figure x). A short 7% in. copper tube was substituted
for column #1 and the Porapak Q column substituted for
column #2. The instrument was run at room temperature with
the temperature held constant by a thermal stabilizer Oper-
ating at 50° C. The cOlumn flow (helium) was adjusted be-
tween flow rates of 80 cc./ min. and 125 cc./min. for optimum
separation and minimum elution times.

Three distinct component areas on the chromatogram
were given at the 100 percent sensitivity range with the
introduction of 1.0 cc. of laboratory room air. The first

area is due to the composite peak measured by thermistor Sl,

00

*3 a"??? 'f
I'I‘ka ‘ '

.‘...'. -1.

- ‘4 “1,. ‘fl -\ . r‘v r V\1‘- .
Present chromatographic

 

 

 

He

 

 

sample
inlet

f

 

drying
tube

 

 

 

 

 

 

column
'I

 

 

 

 

 

 

 

co umn
exhaust ”10

-f a
Column f 1: 30 in. x i in.

30K hexanethylphosphoroamide

l
6.

wi th

F~
F)
N
O O

COlUJI‘L‘ 7

detectir: the

filled

with

’3 ‘f

3
’1L

rmistors.

ft. X 3/16 in. 0. D. filled
Molecular Sieve 1

01

the second component is air measured by S and the third

2.
small component is water also measured by thermistor

82. The quantitative confirmation of the third component
was done by using three separate measurement techniques.
The first technique was to carefully set the helium carrier
flow at 125 cc. per minute and inject multiple air samples,
carefully measuring the elution time of the third component
which was observed to be 0.60 minute. The second measure-
ment technique was to introduce the same sample size of air
from the humidity chest. The peak for the third component
was observed to have the same elution time, 0.60 minute,
but increasingly larger area. The third confirmation was
to introduce a small sample of pure water and air by a
microsyringe and to observe the elution time of the third
component. It was -0.60 minute.

Great care mustbe used in this latter verification
step due to the necessity of over-loading this instrument
by the introduction of such a-large sample; damage to
the measuring thermistors may be caused. The controlled
environmental room was at 07% relative humidity and 75° F.
when the air samples were taken; this correSponds to 1.0 x
10-5 gm. of water in 1.0 cc. of air. The results of this
data are tabulated in Table VII and instrumental conditions
are Specified in the table.

A reduced scale drawing of the chromatogram obtained

. "(lT'I'l 3‘.- ~.
'- VLU 4

I...

.1 0...".
\
hrea

n .H
i

a

I

T‘\H-~--'.( 1.:
.. 1 .
‘.-‘ \. .LCHtL'lCIi'

--_Y

VII

7?
L4

TABL'
ta from the
lit.
cm.

L/'

'eight Pk.

.,...
ii

.1: 1 -11.

9

l
Full

Ti
3and

“mple

«JUL

c9.

Volume

F‘I

-

 

 

 

Trial

42

C C"? C\

M’\ (:2 km 0 \ C-\.:, r»: Cri-

. O O O O I O O

UNLW 1“- t- H U3 0‘] 0')
(\J 1"! (\3 V1 Y—“I

+-t+-I II I I

U-\ L“ L“ ”W L“ UN I C)

C3 01‘! CC- C.‘ C7" Cl 3' | (\-

o o o o o o I o

0\0\c\0\0\0\ It»
I

I I + I I + I +

(7- 1r\ 0 \O \O C {\H (\2
L" \ Cm ("N \f\ U\ C \ C‘ (\-
\C' U\\O :1 :t 42' :r

O O O O O O O O

COOC‘CC‘OO

O O c O C 0 mm
Mom: ("\ ms,- (W3
0 I O O O I O I

HHHHHrit—IH

OCDCTC>OCDU\U\
C>O\C\C~DMD~P»V3
O I O O O I O O

r-ICTCOCC‘CZIO

C‘OOCOCDOC‘J
0000.000

HV—{r‘lr-Iri'r'lr-iv'i

v—4 N (“x—Ll" UN)- Ce 0::

G:-
i'vi
c..-
O 0H
C. C}-
0 or‘i H
0 E; +3
C) \ .C-
\ . C"
o 0 PL:
:50 a
in) c...
VHH
\fl (\1 -r'I
I H '11:
C.
VI 0
O L)
1 H
(H O
(\I C)
L) a L'\
. ("r‘
71 0’3
.531
CS >3
U) {.1- ,CJ
r’l 03
C5 $4 Pd
Z O 0)
CW:- N
G) --—-1
I: r-I
“4-3 .,—:
o "—2 ff:
[1* I; C."
+3
0 5:: a
L'W E—
L\- "‘ 0..»
r4 $4
a O 2:
r} O '-5
3 L";
""1 12$ S—i
re :5 C,"
""3 5:; M
E u": {—3
.H E G)
#3 La 4—)
r4
1‘ 0..“- E3
'P O
----«i Z 0
PK.) $.-
05 H
F‘I\4—)
C; 0“» c3
3;:
pt:e
‘39. H
L\~ . :5
:J 4.5 +3
e-m
. 5.1
.1! O O
'r' ' 9-1
5.5 m E
Q)
- c3 4—)
O
053::
0 2‘5
C‘ S
C F—l r4
0 .r1 0
H :3 o

J-f‘
,.) I.
b-

”’3’?

I.

1
x.

o.

q

..—

leu '1

 

H (\3 C"\\'l7

\C: \C’ \. (‘fl\
0 o o o

C\\./') (P\ O‘\

(\I ('1'

+ I I I

C \ (\‘\-t' Cx’
o o o o
H V-{
+ I + l

C13

-:1'

V'I

0

7‘1

0. :3 O O
\O N T-I L‘r\
:i C‘\ H O

O (3 C C)
\i’\ (“\(T \lz“‘

o o o o
H H ‘4 r1

OOCJO

r-I' T’I H r'l

H No\;‘*

{tit-“lit
. U“

‘x
a DOVE

.n the

Wee

-et'

d run b

tl'l

,—
C

4-
Li

A
'3

'1‘211’3
temperature of the

MI

*1
A4-

I‘O'

1

taken 1
-no control over

4
.a

syring-

he

t

43

by the Fisher instrument may be seen in Figure XI. It
should be noted that the noise to signal ratio is high
and some loss of precision may be inherent in the
calculation of peak areas by using the normal chromato-
graphic calculation method based on triangulation of the
peak areas (11.23). I

4h

TH flTT’fififi ‘8”:
i l"- .
L \JViéL—l «L

“Hun. 4 r“- +- w, ..AV 0 7n - - -°- 1'-
JlLHOf vpfOJJqurifl OL hater in air

...J-

Air - water 3 (_ Air + mate“

measured by 32 meas‘red by 31

 

 

Hater
measured by

32 \I
,\

Helium flow 125 oo./min.
1.0 00. load of air
from atmosphere at

u7; R. H. and 750 F

      

 

 

45

CONCLUSIONS AND FUTURE WORK

Results and conclusions

A support material Porapak has been evaluated and found
to possess excellent stability in a wide temperature range
(up to 250° C.). These porous polymer beads are physically
stable and they vibration pack well in small diameter columns
at room temperature. Porapak did not bleed and it yielded
excellent high temperature sensitivities. This material
can be conditioned completely in four hours at 230° C.
with a helium purge. Porapak has demonstrated the ability
to separate water both in the liquid and vapor phase from
other polar and non-polar compounds; the packing has shown
the ability to separate water cleanly with minimum tailing
and rapid elution times (0.60 to 5.0 minutes) dependent on
column dimensions, carrier flow, and temperature. Porapak
may be coated as any other column packing material and by
possessing geometrical uniformity excellent chromatographic
separations result.

The research has yielded water sensitivity data for
both of the instruments presently used at the School of
Packaging. The data from the evaluation of the Burrell
instrument supports the statement that water in liquid form
as well as water vapor may be detected. The Burrell was

operated at a sensitivity level of 10 and an attenuation of

#6

5. The data shows that a level of 3.00 x 10'“ gm./5 cm.2

(See Figure VII) would be obtained with the 6.5 ft. x 3/16 in.
O. D. aluminum column. The data obtained using the 2.5

meter x 2.00 mm. I. D. stainless steel column further indicates
separations using this column yield approximately 6.8

times as much signalas the 3/16 in. aluminum column does.
From this data it appears that the 2.5 meter x 2.00 mm. I. D.
stainless steel would have an easily attainable sensitivity
of 2.1 x 10'5 grams per 5 cm.2 for water. This figure is
relative in that an arbitrary area (5 cm.2) is defined as the
necessary minimum area required for chromatographic cal-
culations. Even with this requirement the increase of the
Burrell sensitivity by 50 times (the electronic maximum sensi-
tivity level) would put the level of detection at 1.05 x

10"7 gm./5cm.2, well below the minimum necessary to detect
water between 1 - 100 % Relative humidity range in 1.0 cc.

air at 73° F. This sensitivity indicates that indeed one

may use the Burrell for water vapor measurements in air.

It must be stressed that long and arduous conditioning and
many instrumental corrections will be necessary before the
Burrell instrument may be used quantitatively for either
water or water vapor measurements. The Burrell instrument

is extremely sensitive and reproducible to most organic
compounds, and, with the non-bleeding characteristics of

Porapak, this instrument appears better suited at present

47

for organic vapor separation and detection applications.

The Fisher instrument uses a thermistor thermal con-
ductivity cell for detection and from the experimentation
and the reduced scale drawing (Figure XI) there is undeni-
able evidence of the ability to measure water vapor in air.
The obvious candidate for a chromatographic detection system
of water vapor is a small cavity thermal conductivity cell
fitted with nonreactive thermistors.

A reason for obtaining different instrumentation is
that the preferred level of detection isin the order of 5 -
10% relative humidity. The Fisher Gas Partitioner
appears to be limited at higher levels of water vapor
due to the overloading and damaging of the metallic oxide
thermistors. Liquid water sample measurements are present-
ly prohibited using the Fisher instrument as there are no
provisions for heating the columns and many adsorption-
desorption (tailing) effects would be noted with liquid water
samples. Because the column could not be heated, a flow
rate of 125 cc./min. helium was necessary to minimize tail-
ing when water vapor was separated on the Fisher instrument.
Another reason for using thermal conductivity that is Just
as important as the sensitivity requirements is that ioniza-
tion detectors destroy the sample whereas thermal conduc-

tivity cannot. One other advantage in thermal conductivity

48

detection may be in the intrinsic simplicity of this detector
which means that lower sensitivity but more stability and re-
liability in operation are usually noted. I

As a means of comparing the Fisher with the Burrell
K-7 the minimum sensitivity of the Fisher was calculated
from the 8 trials of room air and using the arbitrary re-
quirement of 5 cm.2 area for the water peak. The Fisher

4 grams per 5

will yield a maximum sensitivity of 1.01 x 10‘
cm.2 compared with the maximum theoretical Burrell sensi-
tivity of 1.05 x 10"7 gm./5 cm.2.

From the summarized results four major conclusions may
be stated.

1. Detection of water vapor in air has been accom-
plished using the Fisher Gas Partitioner.

2. The Burrell experiments indicate that under ideal
instrumental Operation the detection of water vapor in
air would be easily accomplished.

3. The preliminary work With the Burrell indicates
much develOpmental and instrumental conditioning will be
necessary for use of this instrument for water measurement
studies.

4. The reproduction of the Fisher chromatogram pro-
vides the School of Packaging with "evidence of thermal

conductivities sensitivity and ease of water vapor detection.

49

The best choice of detection for water vapor in air appears
to be a small cavity thermal conductivity cell fitted to a
temperature programmed two column gas chromatograph utiliz-

ing a system such as shown in Figure XII.

Future work

This section will first deal with suggestions to im-
prove existing equipment, taking into consideration that
the suggestion of acquisition of a temperature programmed
small cavity thermal conductivity gas chromatograph appears
to be the best approach to obtaining the desired sensitivity
and versatility. .

(The suggestions for improving the Fisher apparatus
are as follows: First, a system of columns such as shown
in Figure XII would be successful if some external means
of temperature programming the column were to be developed.
Secondly. the sensitivity increase noted when column volume
was reduced on the Burrell would also be noted in a thermal
conductivity instrument such as the Fisher. The sensitivity

1‘ em./

of the Fisher Gas Partitioner was found to be 1.01 x 10'
5 cm.2 using a 3.0 ft. x 3/16" column. If the volume of
both the column and the thermal conductivity cell were re-
duced by a factor of 10. the sensitivity would be increased
by 10. The cell volume may be changed easily and will not

affect the separation efficiency; however, the column dimen-

sions must be such that clean separations will be made.

 

50

FIGURE XII

V

 

Reference

- -~ ..- ., .1“ r: .J-
iroposed ChrodatOwlaphlc gyfibed

 

A f\
U K.)

 

sample
inlet

 

l

 

 

 

 

temperature
programmed
Porapak
column

for

H20, CO
separatIon

 

 

 

 

exhaust tube

 

 

column
a 2
I Ir

 

—\

' 0" ’ y o J_ _ ~
31 and 82 are detecting thermist0°s.

 

51

Water vapor and carbon dioxide would be trapped after de-

tection at the S detector by the drying tube and column #2;

1
other separations (02, N2) would be done on the #2 column
utilizing the $2 -detector.

If the Burrell instrument is to be used for measuring
oxygen-containing compounds (such as water vapor using
the standard tungsten filament equipment), it must have
some means of electronically maintaining stable electron
emission levels. Another method less preferable would be
to incorporate a method of injecting a hydrocarbon to
chemically create a suitable filament environment.

Another way to eliminate this emission level insta-
bility would be to substitute a non-reactive material for
the reactive tungsten. The tungsten filament may be changed
to a non-reactive .material such as rhenium (21,22) which
does not form stable carbides; its nitrides are also un-
stable; its oxides are conducting and it is not involved
in a water cycle in the same way as tungsten. A rhenium
filament is limited by one drawback. This is that its
vapor pressure is 156 times that of tungsten at the same
electron emission. but the electrical resistivity of the
material is higher so a larger diameter wire may be used.

The life of rhenium is limited by evaporation. but it may

be used for long periods quite well if prOper electronic

52

controls are observed. The life of the tungsten filament

is not normally limited by evaporation, but rather by the
fragility of tungsten recrystallized by prolonged heating

at high temperature. and tungsten is caused to distill as
the oxide by the forementioned water cycle phenomenon.
Acceleration of the normal (brittle) hardening is also
caused by this water cycle. With this filament material
change one may also want to increase the amount of sample
introduced into the ionization detector which would increase
detector sensitivity.

One last instrumental change is recommended. The
Burrell electronics. especially the electrometer circuit,
should be stabilized.

This research dealt primarily with the chromato-
graphic separation and detection of water vapor, and second»
1y with the application of this chromatographic separation
and detection method to packaging research. The results
of this work may be directly applied to further the technique
of water vapor transmission rate measurements. This specific .
chromatographic technique may be also used in determining
the water content of packaging materials. It may be used
to analyze water absorbed by hygros00pic products as well
as materials giving the package researcher another way of
studying package performance. A third area of interest was

to determine the other area in which gas chromatography may

53

be applied to packaging research.

The time spent with the ionization detector instrument
indicated that the Porapak material coupled with ioniza-
tion detection will be an excellent way to study product de-
gradation such as those found in food products (25). This
area alone would yield a wealth of information and develop-
mental data. The study of odor transmission rates as well
as permeants such as 02, N2, C02, and water vapor would
give design data to extend shelf life. The use of chroma-
tography in studying residual solvents by S. G. Gilbert (5)
illustrates but another area of application to packaging
research. One more tool for the packaging engineer is
found in gas chromatography. This research has enabled the
School of Packaging to evaluate and plan for further use of
this valuable tool in packaging research. Hopefully. this
type of research will continue and the results utilized for

solving packaging problems.

1.

2.

7.

9.

10.

ll.

12.

BIBLIOGRAPHY

Books, Articles. and Periodicals

Andrews, M. R. JournalZQf PhysicalTChemistry.
Vol. 27 (1923) 270.

Bennett, 0. F. Analytical Chemistry. Vol. 36
(1964) 684.

Beynon, J. H. Mass Spectrometry and Its Applications
to Organic Chemistry. New York: Elsevier Publishing
Company. 1960.

Davis, Donald w. "Gas Permeability, An Isostatic Test
Method." Modern Packaging. (May 1946) 145.

Gilbert, Seymour G. Modern Packaging. (February 1966)
129.

Guild, L. V.. M. I. Lloyd. and F. Arrl. "Performance
Data on A New Ionization Detector." Instrument
Society of America. A paper presented at the 2nd
International Symposium of Gas Chromatography."

E. Lansing, Michigan, June 1959.

Hall. L. D. Review of Scientific Instrumentation.
Vol. 28 (l957)7653.

"Heat Sealing Characteristics of Materials."
Modern Packaging Encyclopedia 1966. 475.

Hollis, o. L. Analytical Chemistry. Vol. 38 (1966)
309. '

Kirkland, J. J. Analyticavahemistry. Vol. 35
(1963) 6.

Littlewood, A. B. Gas Chromato ra h . New York:
Academic Press, 1962. .

Lockhart. H. E. "A Review of the State of the Art
and Summary of the Permeability Theory." Multi-
sponsor Technical Report #1 (March 30, 1962)
Michigan State University. School of Packaging,
E. Lansing, Michigan.

13.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24.
25.

Lockhart, H. E. Design and Application of an Iso
tactic Permeability Test Method for the Study
of Cellulosic and Plastics Barrier Materials.
Packages, and Closures." A thesis. (1965)
Michigan State University. E. Lansing, Michigan.

"Method of test for Water Vapor Rates Materials
in Sheet Form, E-96-63T." American Society
for Testing and Materials.

"Method of Test for Gas Transmission Rate of Plastic
Film and Sheeting, D-l434-63." American Society
for Testing_and Materials.

"Method of Test for Water Vapor Permeability of Pack-
ages, D-895-62T." American Society for Testigg
and Materials.

Moore, J. C. Jou;na1_of PolymeygScience. Vol. 2
(1964) 835.

Porapak. A commercial publication of-Waters Associates,
Inc.. Framingham, Massachusetts. 1965.

Pprapak and AerOpak. A commercial publication of the
Varian Aerograph Company, 1966.

"Properties of Saturated Steam." gandbook of Chem-
istry and Physicsgg41§t Ed. Cleveland, Ohio:
Chemical Rubber Publishing Co. (1959) 2448.

Robinson. C. F. and A. G. Sharkey. Review of Scien-
tific Instrumentation. Vol. 29 (1958) 250.

Ryce, S. A. and w. A. Bryce. "An Ionization Gauge
Detector for Gas Chromatogra hy." Canadian Journal

of Chemistry. Vol. 35 (1957 1293.

Saframski, Leo w. and Stephen Dal Nagare. C & E
News: Vol. 39 (June 26. 1961) 102 - 114.

o C & E News. Vol. 39 (July 3. 1961) 76 " 830

Schwecke, w. M. and J. H. Nelson. Analytical Chem-
istry. Vol. 36 (1964) 689.

26.

27.

28.

29.

30.

Semat, Henry. Fundamentals of Physics. 3rd Ed.
New York: Holt et als, 1960.

Sharkey, A. G. A report. 5th Annual Meeting American
Society for Testing and Materials Committee E-14.
New York, 1957.

Sugimoto, A. and H. Hoshino. Reports to Scientific
Research Institute, Tokyo. Vol. 27 (1951) 431.

Wexler, Arnold. "Electric Hygrometers." N B S
Circular 586, September 3. 1957.

Other Sources

. Personal interviews with L. B. Westover.

The Dow Chemical Company. Midland. Michigan.

‘immw