RADIO FREQUENCY GAS
HOWARD PERSON WILLIAMS
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
The author will always be indebted to the people that were
influential in the completion of this work. Several were especially
helpful and their names should be mentioned.
Deepest appreciation is extended to Dr. James D. Winefordner,
Chairman of the author's Supervisory Committee, whose encouragement
and understanding have been essential throughout this investigation.
Technical assistance was received from Bill Luchhurst, the
Many thanks are due to Mr. Tom Glenn for his assistance in the
completion of this work.
The author also wishes to express his appreciation to the
U.S.A.F. whose financial support made this project possible.
This work could not have been completed without the love and
understanding of the author's wife, Joan, and his children, Howard
T V ii '-r-
.* * * . . . . . . . . .
LIST OF T . . . . . . . . . . . . iv
Introduction . . . . . . . . . . 1
:. :,cuiy .........................
II. Theory . . . . . . . . . . . . 5
C:r Co :. nts 5
Sci o a . . . .
C. -: 2co ncc 'er "- o .l . . . 7
D. The son:cc rec of n --L ui ar
-t0 C- to. -I of 0
~ C . . . . . . . . . .
F. Thorotical o.rion of t fr-
.clc ari 1-otc:tial cor 1 c chc for
*.Ca. .he C'- e in lect ron noncen:.ation
o a i. P sna . . . . . . . 13
u. i Ocn o ExrX,3ions to Gas Chreo: -ao-
III. 2.. .rijn . . . . . . . . . . 16
C. G ,-ra Cos . . . . . .
eo Cn tons . . . . . . .
C. C ron o h T r.f. Dt~ctors . .. 72
D. I-.-Or .Ar edcd on r.f. sectors . . 72
a o. o........... .... ...........
.. .. ,- -47
5 0.. . . .... son. . . . . . 7
w. . . . . . . . . . . . 79
a,~~~~~~~~~~~. li ..K~,.~ ut~=O
LIST OF TABLES
1. Limits of Detection and Useful Dynamic Range for Several
Gases Using the Frequency Detector . . . . . . 47
2. Limits of Detection and Useful Dynamic Range for Several
Gases Using the Potential Detector . . . . . . 65
3. Comparison of Frequency and Potential Detectors . . 73
4. Comparison of Frequency and Potential Detectors With
Other Detectors . . . . .. . ...... . 74
LIST OF FIGURES
1. Series R-C-L Circuit . . . . . .
2. Modified Series R-C-L Circuit . . . . . . .
3. Exponential Dilution Flask Used for Sampling . . . .
4. Block Diagram of Frequency Detector . . . . .
5. Schematic Diagram of Oscillator for Frequency Detector .
6. Schematic Diagram of Kixer Circuit for Frequency Detector
7. Schematic Diagram of Frequency Meter for Frequency
Detector . . . . . . . . . . . . .
8. Detector Cell for Frequency Detector . . . . . .
9. Schematic Diagram of Oscillator Circuit for Potential
Detector . . . . ..
. . . . . .
Potential Detector Cell . . . . . . .
Analytical Curve for NO2 . . . . ......
Analytical Curve for CH Cl . . .. . .. . ..
Analytical Curves for n-C3H8 and 02 . . . .
Analytical Curves for CO and H2 . . .
Analytical Curves for CO2 and NH3 . . .
Analytical Curves for N2, CH4 and Ar . . . . .
Analytical Curves for SO2 and C2H . . ..
Analytical Curves for n-C4I10 and Air . . . . .
Analytical Curves for Air, CO, and H2 . . .
Analytical Curves for SO2 and N2 . . . . . .
. . .
21. Analytical Curves for N20, CO2, and Ar . . . 71
22. Proposed Cell for the Potential Detector . . . .. 77
radio frc--cncy rmotr od of i-cion
date back to 19, (1). This early d.... tr I:-,-
dcsc ried and responded o chaCnes ine e
flovi. throyu h ca. aclo. r. o .
the s7 h.e b to eli; inate torau iL off e&;.
located in tan. circuits of t.o ;. atc o0cil.
dlczra of .;e oz ll..ato. circic t usc" a s
the mrixer circuit' not the f6,y rc
ficient data were shown to evaluate sr2d co::par
with.' oi er oas chror de ect s.
1n 1l, ,a v or detector b-e : on ch;
was:a3 ... rte -). T.l.. ih frecuency circuit
ad cquantitative rsuls ndicaedhe us .
c ic v .rs, c... accte, e 1 for.ate, ca
chl orofor: L* a b y, s tivl r; V r .A
;ount of orja c r inectu ore lie as
detector. This detecor wa slso nearly L. A
tion. The C i d ton cell consisted of a caac
reaorrnt circuit of a Clir: oscillator. T; .a
r:.ixer circuit aTr fr or: : eter co.rised Au.L
-h0 ous- of c ojcillator. was i'cd into a rix
v." difference fre. y to zhe f,- .. o tcr.
c:'"ic c ."
. Ao d&
. i 10',,m
; .0 ioeCtor
0 IC Co or
. 1 c*n. -
V!3 UO_,A- ViLn -
s of this uccuCor for
.rF:n tetrachloride, and
nse, and linearity w.ith
advan ;aos of this
"ive no flow rate varia-
wa. s .. of
c*^ c c.llators i'tih a
1.*hol, dcteteion syatcm.
or circIt which relayed
Th.s detector was
ideal for monitoring gases in flow streams.
Another class of radio frequency detectors uses the properties of
a r.f. glow discharge or plasma discharge. Helium or argon are generally
used as the carrier gases although other gases could conceivably be used.
Helium and argon will support a r.f. glow discharge at atmospheric
pressure but reduced pressures are needed for other gases in discharge
cell. The dielectric constant of a plasma or a r.f. glow discharge has
a dielectric constant less than unity. The relative dielectric constant
is given (2) by
6' =1 (o
where: Wp = angular resonance frequency of the plasma, and
W = angular frequency of the applied voltage.
The resonant frequency of a plasma is affected by many parameters, and
any change in this frequency changes the relative dielectric constant.
One detector described in this dissertation uses the change in dielec-
tric constant of a plasma as a means of detecting foreign gases in a
There are other means of monitoring changes in the plasma state
occurring in a discharge cell. One of these utilizes a probe or probes
immersed in the plasma. A probe will have a potential impressed on it
depending upon its location in the plasma. This potential is given (3)
for r: r where
X = number of electrons and ions
ro = radius of plasma space charge sphere
r = distance of probe from center of space charge
e = charge on electron or ion, and
E = permitivity of free space.
Two probes unsymmetrically situated in the plasma space charge will give
rise to a voltage between the two probes (4). It is interesting to note
that such an arrangement may be used as a battery. Any change in the
carrier gas results in a change in the number of electrons or ions and
consequently a change in potential difference between the t-.o electrodes
is immediately noted when a foreign gas alters the steady state discharge.
The potential method mentioned above was also used to detect gases.
This was very similar in principle to the work reported by Karmen and
Bowman (5) who measured the electrical characteristics of a r.f. excited
glow discharge in helium to detect organic samples. A d.c. current suf-
ficient to drive a d.c. recorder without amplification was obtained
using a highly unsymmetrical cell. A cylinder of wire mesh and a con-
centrically placed wire were used for the detection cell. The cell was
placed in series with a coil, and the d.c. current in this series was
monitored. The instrument was quite sensitive to temperature and was
90 percent destructive to organic samples (6). Essentially the same
device has been reported by Hampton (2). Winefordner, Williams, and
Miller (8) used the same type of circuit as Winefordner, Steinbrecher,
and Lear (2); however, the difference frequency corresponded to the
changes in a glow discharge occurring in the capacitor cell.
The purpose of this dissertation is to evaluate tio types of
sensitive r.f. gas chromatographic detectors. One detector has a dif-
ference frequency readout, and the other has a d.c. potential difference
readout of a probe inserted in the plasma. Both are quite simple in
construction, linear in response, and very sensitive.
A. ^ ^L.-.... ....
A pl as..a ray bc def:icd () .3 a 'nIy ii :c C. cc C
Sositive an/or .. av ons, elccro.-s, n.Cua :.olccul.. neutral
aco:i-c :c les. in ore -r or pl 1. .. o .1 V i.crc
cain-tai n c, .a n- .' vn s oAC n3 at. 1 ^,er
In -...:c" hclri .3* ic pr .. O r o.
cfroaency field, s-ich as n the Lar-nc c~ rc'lt of a oscllaor -ch
consists of a series R-C-L circuit (see -~re ).
r 3 C
. -tics R-C-L Circuit,
R is the d.c. resistance, L the inductance of the coil, and C the
capacitance of the components in Figure 1. This circuit has a resonance
X = X. (eqn.l)
X = 2fC= reactance of the capacitor (eqn. 2)
X, = 211 fL = reactance of the coil (eqn. 3)
The resonant frequency of an R-C-L circuit assuming the d.c. resistance
is negligible compared to X and XL is given by
f= (eqn. 4)
o 2 i(7
The total impedance of the resonant circuit is given by
z = [2 (XcXL)2] (eqn. 5)
The Q, or quality factor, of this circuit at resonance is given by
S= L (ecn. 6)
and in practical circuits Q can have large numerical values, e.g., 100
to 1000. The same amount of current flows through all components of a
series circuit, and so at resonance, the voltage drop across C (Ec) and
the voltage drop across L (EL) are given by
E = Xc, (eqn. 7)
EL= D. (eqn. 8)
The voltage, E, across the whole circuit (see Figure 1) is given by
E = IZ = IR, (eqn. 9)
X = ~, when the circuit is at resonance.
If equations 6, 7, and 9 are co '1bin tlen
-c (cm. = /f
and so the voltage across the c or, C, at res can be of tAo
order of hundreds to thousands of vol : oil tle liei voi
Since the : between the c aacior plat-es is al e., less than
1 r, the volt ae r rain en can exceed 10,! 3 volsi/c1. whiich is suffi-
cientto I produce a r.f. las a at at olheric pressure in hloliu-, -. :,,
and other gases.
S has resonance fr. cnces due t osci .on.s o ions
and electrons. 'he resonance frecqi due to electrons (_) () is
e = e
1. = .r:.s o,
A sitl, r f-Z .- *.cy 0
of electrons er unit volu e
on an election
>sion due to ion option is
fi- T ke
- 8 -
ni = number of ions per unit volume
M. = mass of the ion
D. The Resonance Freouency of an R-C-L Circuit and its Relationship to
Electron Concentration of Plasmas
The resonance frequency of a series R-C-L circuit (3) (4) is
changed when a r.f. glow discharge is introduced into the capacitor.
The capacitance of the oscillator circuit is given by
C = e'Co, (eqn. 13)
' = real part of the complex dielectric constant
C = capacitance with a vacuum between the plates of the
The dielectric constant for a plasma can be shown to be
e +2 2 2
W = 2rf
W = 21if
f = resonance frequency of plasma
f = resonance frequency of oscillator,
Solving for E results in
= 1- )2 = 1- -2-
Now, if the frequency of the plasma, f is taken as the electron
resonance frequency, fe, then
S=fe = ke(ne ) (eqn. 16)
Substitution of equation 16 into 15 yields
k' = 1 e e (eqn. 17)
Substitution of equation 17 and 4 into equation 13 leads to an imaginary
frequency when W p2>W2. For most plasmas at high frequency, this con-
dition is true. Therefore, the simple series circuit described is
inadequate to explain the presence of a r.f. discharge in the capacitor
of the circuit. However, this problem can be circumvented by using an
equivalent circuit to represent the simple series circuit illustrated in
Such a circuit is given in Figure 2. Since the plasma is local-
ized in only one portion of the capacitor, capacitor C can be repre-
sented as two capacitors (C1 and C2) in parallel, where C1 represents
that portion of the capacitor with no plasma (carrier gas only between
the plates), and C2 represents the portion of the capacitor with plasma.
R L C
Fig. 2 Modified Series R-C-L Circuit.
- 10 -
If R (in Figure 2) is very small, then the resonance frequency of the
equivalent circuit shown in Figure 2 is given by
f = 1 C (eqn. 18)
o 2f 1TL(C1+C2
Substituting for C2 in terms of dielectric constant according to equa-
tions 13 and 17 results in
k n I
27i L[C1+Co(1- e 2e
Rearrangement and solving for fo gives
2 1+412LC k n
f = oee
Differentiation of equation 20 results in
17"2LC k dn
2fdf = 4 o e e
0 0 4r1 2(LC1+Lco)
Division of equation 21 by equation 20 gives
2df 41f2LCk dn
o 0 e e
S- ee (eqn. 22)
f 1 + 4f02LC k n
Since ne is of the order of 10 electrons per unit volume, ke is of the
order of 10 cps, Cl and Cd are of the order of 10-1 farads, and L is
of the order of 10-3 henries, then 4Tr2L(C,+Co)k 2n e?1, and so
1 o e e
- 11 -
From equation 23, it can be seen that the change in frequency with
respect to a given reference oscillator frequency, i.e., df /f is
linear with a change in the number of plasma electrons, i.e., dn /2ne'
with respect to a reference number of electrons. If the relative change
in frequency is small, then for a finite change in frequency and elec-
tron concentration, the following equation results, namely,
df =( o f.Ane (eqn. 24)
where f /2ne is a constant. This is the equation of a straight line in
terms of f0 and ne having a slope of fo/2n
o e o e
By taking logarithms of both sides of equation 24, the following
log(fo) = log (n e) + log 2 (eqn. 25)
This equation indicates a linear plot on log-log coordinates with a
slope of unity and an intercept of log fo/2ne'
E. The Potential on a Probe Inserted Into a r.f. Plasma
The potential of a spherical plasma which is also related to the
electron concentration, n is given by (2).
2 22 2 22
Xe (3ro-r2) nee (3r -r2)
0i(r) = :- (eqn. 26)
4Tfr 32E 6e
for r r where:
0i(r) = potential on a probe as a function of r
n = --- = number of electrons per cm3
- 12 -
X = number of electrons in spherical plasma of
e = charge on the electron
r = radius of plasma
C = dielectric constant of free space
r = distance of a point from the center of the plasma
Thus, if a probe is placed at any point into the plasma, the potential
on this probe is given by equation 26. If a second probe is inserted
into the plasma, its potential will be described by a similar expression.
Therefore, the potential difference between the two probes is given by
nee2(3r-r 2) nee 2(3r2 r2)
V. = (ri)-0i(r2) = o 2 2)
.6 61 (eqn. 27)
V= e e 1 (eqn. 28)
By fixing the probes such that rl and r2 are constant and differentiating
with respect to ne, the following equation is obtained
2 r 2)
dVi = 6- dne. (eqn. 29)
If equation 29 is divided by equation 28, then the relative change in
potential difference is given by
-- e. (eqn. 30)
- 13 -
The interpretation of equation 30 is quite analogous to the interpreta-
tion of equation 23. It can be seen that a change in the number of
electrons, dn with respect to a reference number, n results in a
linear change in the potential difference between the tio probes, dVi,
with respect to a reference potential difference, Vi. If the relative
change in potential difference, dVi/Vi, is small, then for a finite
change in electron concentration,6 ne, and in potential difference,
AVi, the following equation results
AVi = (-) An (eqn. 31)
where Vi/ne is a constant for any given r.f. plasma. Equation 31 is
the equation of a straight line in terms of LVi and ne with a slope
of Vi/ne. By taking logarithms of both sides of equation 31, equation
log(WVi) = log(Ane) + log(Vi/ne). (eqn. 32)
A plot of log (AVi) versus log (Azne) should result in a straight line
of unity slope and an intercept of log (Vi/ne)
F. Theoretical Comparison of the Difference Freouency and Potential
Difference Methods for Measuring the Change in Electron Concentra-
tion of a r.f. Plasma
The two methods of monitoring changes in the number of electrons
of a plasma may be compared with respect to relative sensitivity by
rearranging equation 24 to give
ne (eqn. 33)
f/ ne 2'
- 14 -
and by rearranging equation 31 to give
4V,/n 1. (eqn. 34)
*V n e
From equations 33 and 34, it can be seen that the change in frequency
with respect to a reference frequency is half as great as the change in
potential difference with respect to a reference potential difference
for the same relative change in electrons. Therefore, the change in
potential difference method should be twice as sensitive as the change
in frequency method.
G. Application of Expressions to Gas Chronatographic Systems
The carrier gas used in most of the studies was helium (He)
although several other gases, including argon, could be used. The r.f.
glow discharge in helium results from the excitation of He atoms to an
excited state (He*) via the r.f. energy. Since even the highest purity
helium contains ionizable impurities (I), there is a background electron
concentration (n e) due entirely to the carrier gas, which can be repre-
He + I->He + I + e-(ne). (eqn. 35)
When a sample is put into the r.f. discharge, the sample may or may not
fragment into smaller radicals due to the high energy of the He par-
ticles, and the resulting particles can either ionize due to collisions
with the metastable heliums which will cause an increase of n to n +-
ne or capture electrons resulting in a decrease of ne to ne- one.
If more than one fragment results part of the resulting particles can
- 15 -
ionize and part can capture electrons, and so the change in ne cannot be
predicted unless the specific case is known. If the sample (S) is not
fragmented but is ionized, the process can be represented as
He + S-->He +* S + e(+An e). (eqn. 36)
If the sample (S) is not fragmented but captures electrons, the process
can be represented as
S + e"-- S- (-Ane) (eqn. 37)
Therefore, a change in electron concentration, n e, results due to the
sample (or its fragmented products) ionizing or capturing electrons.
The change should be a linear function of the amount of sample, Sa,
introduced into the gas chromatographic column if the above processes
are correctly depicted, i.e.,
n = K S (eqn. 38)
e c a
where K is a function of the plasma, the sample molecule, and the
column size, temperature, and characteristics. Therefore, as long as
all experimental conditions are maintained constant and change in ne,
V. and fo are small, then
o = oca (eqn. 39)
V6 = i ca (eqn. 40)
which indicates a linear relationship between the signal and amount of
sample being measured.
A. Sampling Technique for Gases
The sampling device chosen was an exponential dilution flask
first described by Lovelock (2). In principle, it consisted of a cham-
ber of a fixed volume with an inlet for the carrier gas, an injection
port for sample introduction, a mixing device to instantaneously mix
the sample plus carrier homogeneously, and an exit port. The expression
for concentration, C, at any time, t, should be given by:
C = Co exp(-) ,
where V = volume of flask (cm3)
U = flow rate (cm3/sec)
t = time (sec)
C = concentration at t = 0
C = concentration at any time, t
The plot of log C versus time gives a straight line with a slope of
-2.303V and an intercept of log Co
The construction of the dilution flask was non-critical. A
schematic drawing of the flask is given in Figure 3 and the construction
procedure is as follows. A phenolic cylinder@ was turned on a lathe
to the following dimensions: 3 in.(o.d.) x 1/4 in.thick x 2 3/4 in.high.
The end pieces @ were cut from 1/4 in.thick sheet LUCITE . An
11/64 in.x 1/8 in.deep hole was drilled in the center of each end piece
to act as a bearing for the paddle assembly.
- 16 -
Fig. 3 Exponential Dilution Flask Used for Sampling.
1. Magnetic Stirring Bar
2. Teflon Vanes
3. Phenolic Cylinder
4. Lucite End Pieces
5. Injection Port
6. Photo Spool With Teflon Bearings
7. Copper Tubing
- 18 -
- 19 -
The paddle assembly was constructed by cutting trapezoidal shaped
vanes, 1 1/2 in.x 1 in.x 1 in.high, from 1/32 in thick TEFLON sheet
(). These were inserted into a size 120 plastic roll film spool (
(Supreme Photo Supply Co., Inc., 1841 Broadway, New York,23, N. Y.). A
TEFLON covered magnetic stirring bar Q was inserted into a 5/16 in.
hole which had previously been drilled in the film spool perpendicular
to the slots and about 1/4 in.from the end. The vanes and stirring bar
were then cemented in place. A 1/4 in.TEFLON ( rod was turned to
5/32 in.diameter x 1/4 in,and inserted in the ends of the spool to act
as the shaft assembly.
The injection port was fabricated from a brass bulkhead adapter
SWAGELOK Q (Crawford Fitting Co., 884 East 140th Street, Cleveland,
Ohio, part no. 200-A1-2) (). A 1/8 in.hole was drilled in the side of
the injection port, and a piece of 1/8 in.(o.d.) copper tubing Q was
soldered in place. A small round file was used to smooth rough edges
inside the injection port, and an injection gasket (Wilkens Instrument
& Research, Inc., Box 313, Walnut Creek, California) was held in place
as shown in Figure 3. This injection port assembly was then fitted into
a 1/8 in.hole which was drilled in an end piece 3/8 in,from the edge
and cemented in place. Next, a 1/8 in.exit hole was drilled in the side
of the phenolic cylinder 3/16 in.from the base (see Figure 3), and a
piece of 1/8 in.copper tubing was cemented in place. Finally, the end
pieces were cemented to the phenolic cylinder containing the paddle
assembly. There was sufficient room between the vanes and the cylinder
wall to prevent any obstruction to a syringe needle. A magnetic stirring
- 20 -
motor completed the assembly. The volume of this particular flask was
All samples were introduced into the dilution flask using
Hamilton gas-tight syringes. With the exception of air, plastic bags
filled with the particular gas under investigation were used in conjunc-
tion with the syringes to avoid contamination of the gas by air. Sand-
wich size "Baggies" were flushed out several times with the sample gas
then the top was twisted tightly to seal the gas in. Next, a short
piece of masking tape was pressed around the twisted end to seal per-
manently the bag filled with gas, and within a matter of seconds, the
needle of a syringe was inserted through the plastic bag such that the
needle extended into the inner space of the bag. The syringe was
flushed several times by filling and subsequently emptying the syringe
with the needle inserted in the bag. A positive pressure was maintained
on the gas-filled bag to prevent diffusion of air into it through.the
opening caused by the needle. Next, the bag of gas with the syringe
needle still inserted was placed against the injection port of the
dilution flask such that the syringe was pointed at the injection port
through the bag. The injection was then accomplished by pushing the
needle through the other wall of the plastic bag and into the injection
port. The gas sample was thus injected with a minimal amount of con-
tamination by air. This method gave reproducible results and was much
more reliable than other sampling methods attempted. For injection of
samples near the lower limit of detection only one bag per injection
was used. However, at other ranges, two or three injections were made
- 21 -
using one gas-filled bag by patching the holes with masking tape.
Hamilton syringes with removable needles were found to leak at the
needle syringe junction. To prevent this, thin TEFLON tape was
wrapped about the male syringe fitting between needle and syringe.
When the needle was pressed on to the syringe with a twisting motion,
the TEFLON flowed and gave a gas-tight seal.
A sampling difficulty encountered with the syringes was due to
the soft injection port septums clogging the needles either partially
or completely. This problem was eliminated by changing needles often.
B. The Frequency Detector
A block diagram of the experimental setup is given in Figure 4.
The resonant frequency of the sample and reference oscillators was 72 Me.
Clapp oscillators were similar to the ones previously described by
Winefordner, Steinbrecher, and Lear (1). The outputs of each oscillator
were fed into a diode mixer-circuit. Due to the tendency for the
oscillators to capacitatively couple at these high frequencies, an
attentuation network consisting of C7 and R in Figure 5 was used to
lower the oscillator output voltage to 0.2 volts as measured with a
Hewlett Packard Model 410-B Vacuum Tube Voltmeter. This combined with
good shielding resulted in a minimal amount of "lock in." Cathode
followers were found to be useless at these frequencies.
The output of the mixer was fed into the frequency meter and then
into a recorder. The oscillator circuit used for both detectors is
shown in Figure 5. The circuit of the mixer circuit is given in Figure 6,
and the circuit of the frequency meter circuit is described in Figure 7.
- 23 -
-o (c )
r- \ C"%
- 25 -
HH 0 0
CO ) ( -
i .H4 4-o
c* .* C.H
Q 4 (H ,C to to
0 0 C z
o) .- -) *
C) C "0 D too
SC* O CO
(U r- 4 NH U
I II II II
o N \ NH N H 10
I I I 0
4 0 co 0 r-
\0 C -t O0 d- H
CO 0 N H H
too C 0
00 \ O
S I Il I N \O
0 CoO H 0
0 o t0 0 I
Q) r - a 0
0H 0 0 0 0 0 0 0
6Y t- b d
Q ) -S g
O V" O O O G 0
S)O O O O O O
S- O CC O * * * *
o o N < O O O O O O 0
o I I I I I I I I I
M O .i O O C
I--I I I
,O 0 H v ; H o
E" II~ I I I I I
* I I a a
-*- 0 N C- \u 1 03
"3 ~ ~ 2 rtl a^^ ?
- 29 -
- 30 -
The values of all components are listed in the captions for each diagram.
The Clapp oscillators were built on aluminum chassis (3 x 6 x 4 inches).
The coil L1 in the tank circuit was rigidly mounted on a porcelain plate
on the top of the chassis and was shielded by means of an aluminum box
(3 x 4 x 3 inches). The oscillator tubes were enclosed in tube shields.
The B and filament connections for the oscillators were connected to
500-if button-type feed-through condensers and r.f. chokes (Ohmite Z-50)
and then to regulated power supplies. The feed-through condensers and
the chokes are not shown in Figure 5. A regulated power supply (Model
407D, John Fluke Mfg. Co., Inc., Seattle, Washington) was used to supply
the power to the filaments and B of the oscillators. A regulated power
supply (Model PS-3, Heath Co., Benton Harbor, Michigan) was used for the
mixer and frequency meter. It was of great importance that the com-
ponents of the tank circuits be mounted and wired as rigidly as possible.
Square bus bar wire was used for all connections in the tank circuit.
The coils were wound from silver-coated copper wire (No. 10) and were
mounted rigidly on polystyrene rods (3/4 in.o.d.). Connections to the
coils were made through the base of the porcelain plate. The leads of
all components were kept short and positioned as far from the chassis
as possible to avoid loss of energy from the circuit due to capacitative
The cell capacitor, drawn to scale, is shown in Figure 8. It
consisted of a micrometer () (No. 263, L. S. Starrett Co.,) mounted in
a brass cylinder 1/2 in.i.d. x 3/4 in,o.d. x 3 5/8 in.long. The cylinder
was threaded to fit into a 3 in.o.d. by 1/2 in.thick brass mounting
Fig. 8 Detector Cell for Frequency Detector.
1. Carrier plus sample gas inlet
2. Carrier gas inlet
3. Counting flange
4. Coil side of cell
5. Housing for cell
7. TEFLOND bushing
- 32 -
___ ~ II' K
plate (). The actual effective capacitance was determined by the area
of the end of the gas inlet tubing, which consisted of an 1/8 in.o.d.
brass tube soldered coaxially inside a 1/4 in.brass tube, and the dis-
tance from the brass electrode in the recessed LUCITE insulator at
the bottom of the assembly. The concentric tube electrode was spring
loaded such that a positive force was always on the rear of the elec-
trode against the micrometer. A TEFLON Q spacer ( 1/2 in.o.d. by
5/16 in. thick with milled sockets for the anvil of the micrometer and
the rear of the concentric electrode was placed between the micrometer
and electrode. This maintained alignment of the movable electrode while
allowing ease of movement when necessary. The dual-gas inlets ) and
( were designed to allow pure helium to flow through the outer
cylinder and carrier helium plus sample to flow through the inner
cylinder. The purpose was to extend the working range of the detector
to very large concentrations. In practice, only the inner cylinder (
was used for gas flow, while the other gas inlet ( was sealed off.
The sensitivity was reduced considerably when helium flowed through the
outer cylinder. Gas entering the cell was vented to the atmosphere
through holes in the LUCITE insulator.
The frequency determining capacitor in the reference oscillator
consisted of a 1.3 3.1 puf (160 203 E. F. Johnson) butterfly
capacitor for fine tuning and 2.7 10.8 p.f (160 211 E. F. Johnson)
butterfly capacitor for coarse tuning. The two capacitors were mounted
in parallel on a fiberglass circuit board attached to the chassis on a
stand-off arrangement. The advantage of these capacitors over ordinary
variable capacitors was that no brush or bearing contacts were present
- 33 -
giving rise to additional impedance in the tank circuit. A phenolic
shaft was connected between the rotor shaft and a gear reduction system
such that fine tuning of the small capacitor could be easily accomplished.
A similar phenolic shaft without gears was attached to the larger
capacitor for coarse frequency adjustment. This arrangement was vory
convenient for adjustment of the reference oscillator frequency to that
of the sample oscillator.
The frequency meter and mixer circuit were mounted on the same
chassis in a sloping-panel cabinet 10 in.x 18 1/6 in.x 10 13/32 in.
The two oscillators were connected to the mixer circuit by means of two
short rigid air dielectric coaxial connectors. All three chassis were
firmly attached by brackets and screws to a 3/4 in.x 18 in.x 36 in.ply-
wood base. In order to minimize temperature variations, all components
were enclosed by construction of a box 12.in x 12.in x 36 in.from 3/8 in.
plywood. Hinged doors suitably arranged allowed adjustment of the
reference and sample oscillators. The resistors in the frequency meter
tended to heat; therefore, ventilation holes were drilled in the front of
the sloping-panel cabinet in the vicinity of these resistors, and a
3 1/2 in.hole was cut out of the top of the plywood cabinet. A high
velocity fan was mounted above this hole not touching the cabinet. This
arrangement worked rather well in cooling the frequency meter circuit.
The wooden cabinet was placed on a shock mount (Type 4995-1030 Barrymount,
Barry Controls, Watertown, Mass.). This virtually eliminated vibration
problems. The power supply connections were made using ordinary hook-
- 35 -
C. operating Conditions
The plate voltage supply for the frequency meter and mixer was
adjusted to 300 volts. There were several ways to affect the field
strength in the detector cell, and this was done by varying the elec-
trode distance in the detector capacitor, the gas pressure in the
capacitor, or the power from the oscillator tube. The working conditions
were optimized with respect to a 50 l injection of air into the dilution
flask. This corresponded to an initial concentration of 288 parts per
million. The best response was obtained with an electrode distance of
0.012 in and an oscillator plate-voltage of 390 v. A large increase in
field strength resulted in arcing between the electrodes and gave rise
to an increase in noise while decreasing the sensitivity drastically.
The pressure in the capacitor was assumed to be approximately atmos-
pheric pressure, which was supported by the observation that the re-
sponse was the same over a wide range of flow rates. The baseline,
however, was affected by changes in flow rate, but not the sensitivity.
The frequency chosen for the reference frequency was selected to
give greatest linear dynamic range for air and still be within the
frequency range of the mixer circuit. The coils (L, and LS) in Figure 4
were adjusted by means of a grid-dip meter until both oscillators were
operating at about 72 megacycles per second. The exact frequency was
not critical as long as both oscillators were oscillating at the same
frequency when the r.f. discharge (with only helium carrier) was sus-
tained in the sample cell.
The coils L2 and L3 of the diode mixer were extremely critical
- 36 -
and somewhat tedious to wind. It was necessary to wind several coils to
obtain a mixer with a wide difference frequency range. The diode mixer
and frequency meter circuits gave a linear output with difference fre-
quency from 0.3 to 200 kc.
With helium carrier gas flowing through the detector c)ll (30 cc -
60 cc per minute), the experimental setup was quite stable. Over a
period of several minutes the drift was negligible and the peak-to-peak
noise was approximately t 50 cps.
The carrier gas used was ordinary helium as supplied by Linde
Company. A Matheson pressure regulator was used in conjunction with a
five-foot piece of 1/8 in.x 0.02 in stainless steel capillary tubing to
maintain a well regulated helium flow-rate. In series with this, a
metering valve (B-25, Nuclear Products, Cleveland 10, Ohio) and a flow
meter (1A-15-1 rotameter, Ace Glass Inc., Vineland, N. J.) were used to
control and monitor the final flow rate to the dilution flask and 7.2
(glass ball) 4.2 (stainless steel ball) (points read at mid-ball
position) on the rotameter. This corresponded to a flow rate of 80
cm3 /min as read on a soap-film flow meter.
The filament voltages of the power supplies were turned on to
allow the oscillator, mixer, and frequency meter tubes to warm up for
about one minute prior to turning the plate voltages on to the values
indicated above. The instrumental setup was allowed to warm up for 30
minutes to assure maximum stability. The frequency meter was adjusted
to scale No. 6 which corresponded to 4.9 kc full scale, and condenser
CR was adjusted until a difference frequency of zero resulted (if CR
was turned either counter clockwise or clockwise from the position
- 37 -
resulting in a zero difference frequency, the frequency difference in-
creased), and then CR was turned to give a difference frequency 10 per-
cent of full scale. The sensitivity was then set to the desired value,
and CR was again adjusted to give signal 10 percent of full scale. Then
a 50 pl sample of air was introduced through the injection port into the
dilution flask by means of a hypodermic syringe, and the emergence of
the peak was observed on the frequency meter or recorder. If the signal
was first negative and then positive, it was necessary to turn CR (with
helium passing through CS) until the meter or recorder decreased to zero
and then increased again to 10 percent. 'Nhen the reference oscillator
was adjusted to have a higher frequency than the sample oscillator, the
samples would normally produce positive peaks with respect to response
for air. Once this reference state was established, then the frequency
meter was adjusted to the proper scale and working curves were obtained
for each sensitivity scale for each gas studied. The detector was
checked daily for sensitivity by injecting a reference sample of air,
50 pl by means of a Hamilton 100 )l gas tight syringe fitted with a
Chaney adapter adjusted to 50 l.
D. Potential Detector
The detector based on the change of relative potential differ-
ence between tvo electrodes immersed in the helium r.f. plasma is
schematically shown.in Figure 9. A Clapp oscillator was used to pro-
duce an r.f. discharge within the tank capacitor through which helium
carrier gas flows.
A probe inserted in this plasma and the grounded capacitor
H CNl -:t n
o0 0 0
0 U U) H
-) 0 0 0
(\I C \ V \10 ` 1
o C04 o o
11 II It I1 II II II
M c^\ -3- v^ II
K~~~ n^ a ia' a Sm
- 39 -
- 40 -
electrode furnished a voltage that could be read on a d.c, vacuum tube
voltmeter (odel 220, Keithly Instrument Inc., Cleveland, Ohio). A
bucking voltage in series with the probe lead allowed reaction of the
baseline voltage. An Ohmite Z-28 r.f. choke was placed in series with
the probe and the bucking voltage to attenuate any r.f. oscillations
on the d.c. voltage. A voltage dividing network consisting of a 10K
ohm fixed resistor and a 3 ohm variable potentiometer were used to
reduce the voltage output of the vacuum tube voltmeter and the small
voltage (1 mv or less) was measured by a potentiometer recorder. All
leads were carefully shielded by use of high quality low capacitance
Figure 9 shows the schematic circuit of the oscillator. C4 is
the sample cell, and R2 is a 300K ohm ten-turn Helipot precision resis-
tor which was used to control the power driving the oscillator circuit.
This was used to adjust peak voltage obtained from the cell probe.
The cell construction is shown in Figure 10. It consisted of a
1 in.x 1 in.x 1 1/2 in.LUCITE block in which two 1/8 in.holes were
drilled perpendicular to each other and intersecting at the center of
the block. The 1/8 in.o.d. stainless steel tube ) served as an inlet
for the carrier gas and sample. The flat electrode and pointed elec-
trode () were turned from Monel stock to 1/8 in.x 3/4 in. The point
of the tapered electrode was 1/8 in.long with a 35 taper. Copper wire
1/32 in.diameter was silver soldered to the external ends of the monel
electrodes to facilitate soldering these to the rest of the circuit
using rosin core Pb-Sn solder. A 0.0025 in.hole drilled perpendicular
Fig. 10 Potential Detector Cell.
1. LUCITE block 1 in.x 1 in.x 1 1/2 in.
2. Yonel electrodes
3. Gold plated No. 22 wire
4. Stainless steel tube 1/8 in.in o.d.
- 42 -
- 43 -
to the intersection of gas stream and electrodes allowed the insertion
of a No. 22 gold-plated wire between the electrodes. Spacing of the
three electrodes was not carefully measured and the design of this cell
was certainly not an optimum one, but it demonstrated the feasibility
of the method. Roughly the 1/8 in electrodes were spaced 1/8 in.apart
while the wire probe Q was positioned parallel to the face of the flat
electrode. The distance between the flat electrode and the wire probe
was about 1/32 into 1/16 in. with the tip of the wire slightly above
the center of the gap between the electrodes as shown in Figure 10.
E. Operating Conditions
The operating conditions for this arrangement were obtained by
maximizing response for a 10 pi injection of air into the dilution flask.
This corresponded to an initial concentration of 45.7 parts per million
of air in helium carrier. The optimum conditions for the cell described
was a flow rate of 5.2 (stainless steel ball) 9 (glass ball) of a
flowmeter (size 1A15-1, Ace Glass Inc.). The B+ voltage was adjusted
to 390 volts or 335 volts and 890 or 840 on the ten turn resistor dial.
The grid resistor, R2, had values ranging from 33K to 34.8K ohms. These
resistance values were arrived at by decreasing the grid resistor values
until a stable voltage was indicated on the VTVM. This meant an r.f.
discharge was then established in the cell. The r.f. discharge increased
in size when R2 was decreased. The resistance, R2, was decreased until
a maximum voltage of 40 to 60 volts was indicated. The resistance was
further decreased until the voltage again dropped to roughly 50 percent
of the peak voltage or to 20 to 30 volts. 'This meant that both the probe
and the ground reference electrode were then imnersed in the r.f. glow
When a sample of air was introduced in the carrier gas, an
increase in voltage was observed. If a very large sample was introduced,
the voltage first increased then decreased to the point where the dis-
charge was sometimes extinguished depending on the size of sample in-
jected. .Working curves were measured over a range of voltages varying
from reference voltage to maximum voltage.
A 225 volt dry cell battery, a 12 megohm variable potentiometer,
and a 20 megohm fixed resistor made up the bucking voltage network.
This was enclosed, together with the previously mentioned r.f. choke,
in a 7 in,x 6 in.x 6 in.aluminum box for shielding. An on-off switch
for the battery and TEFLON insulated feed through connectors com-
pleted the suppression network.
The frequency of the oscillator was 25 Ecs. This choice was not
completely arbitrary because the oscillator frequency determined the
visual size of the glow discharge for the cell arrangement and other
IV. RESULTS AND DISCUSSION
A. Frequency Detector
The response of the frequency detector was determined for the
following gases: Air, N2, Ar, CO, H2, CO2, 02 NH, NO2,S02, CH4, CH3C1,
C2H6, n-C3Hg, n-CqH10. In Table 1 the limit of detection and useful
dynamic range of 15 gases in helium carrier gas are given.
The analytical curves are plotted on log-log coordinates in
order to compress the carves to a convenient size. Any deviation from
unity slope indicates non-linearity in response. Figures 11 through 18
are the analytical curves obtained.
It seems evident that the gases with a negative (-) response
either decrease the number of electrons by a dilutional process and/or
electron capture mechanism. Ionization should lead to an increase in
the number of electrons in the plasma. Competition of all these mecha-
He + S -S e- +- He
St 4 e---S
S + e-- S-
S-->dilution of carrier
It should be noted that a 50 microliter sample of air was injected
daily to see if the response was reproducible. For a series of ten in-
jections of 50 l of air (injections were made over a 10-minute period),
- 45 -
the relative standard deviation was 3.4 percent. For a large number of
injections of 50 ul of air over a period of two months, the relative
standard deviation was slightly less than 9 percent.
- 47 -
Limits of Detection and Useful Dynamic Range for
Several Gases Using the Frequency Detector
Limit of Useful
Gas ppm Range Responsea
5 x 102
5 x 102
5 x lo3
5 x 102
a(+) means frequency increases due to sample gas in helium
carrier gas. Air was used as the reference gas for all measurements.
0 0 -
- 49 -
; I C~
(S C. 3
- 51 -
- 53 -
- 55 -
- 57 -
C C CD
--- O I.--
\- U- 0-
\ CM fl3 O
\8 ^ |
- 59 -
- 61 -
-------------- Q LJ.
- 63 -
- 64 -
B. Potential Detector
Working curves for the potential detector were determined for
the following gases: SO2, N20, CO2, Ar, CO, Air, i2, -and 2. The limit
of detection and useful dynamic concentration range for 1 .t gases or
helium are given in Table 2. The analytical curves were plotted on
log-log coordinate paper in order to compress the curves. Any deviation
from unity slope indicates non-linearity in response. Figures 19
through 21 are the working curves obtained for this detector. All gases
responded by increasing the voltage output. Again, air was chosen as a
This detector was not as reproducible as the frequency detector.
The baseline voltage drifted slowly over a several-day period. This was
attributed to electrode reactions at the gold-plated probe which showed
evidence of a deposition of some type on the probe. It is believed that
the electrodes should be "aged" longer to achieve a more drift-free
state. Drift was negligible over a one-day period, and each determina-
tion was made within a half day so that baseline drift was negligible
for each sample.
Limits of Detection and Useful D-mnadic Range for
Several Gases Using the Pott ..,-.i Detector
Gas Limit of Detection (ppm) Useful Dynamic Range
N20 0.1 104
SO2 0.07 10
CO 0.8 104
Air 1.0 5 x 103
CO 6.0 10-
N2 7.0 5 x 102
Ar 80.0 5 x 102
H2 70.0 102
- 65 -
- 67 -
- 69 -
- ?1 -
- 72 -
C. Comparison of the Two rf. Detectors
It is difficult to compare the two detectors because the potential
detector was not completely optimized. However, some comparisons can
still be made. In Table 3, the frequency detector is compared with
the potential detector. Within an order of magnitude the performance
of the two detectors is about the same for the gases listed.
In Table 4, the two detectors are compared with other detectors
commonly used in gas chromatography. The data in Table 4 indicate that
the two detectors are as sensitive as the most sensitive ionization
detectors available. They are inexpensive to construct and have wide
useful dynamic ranges.(10).
D. Further Work Needed on r.f. Detectors
If the Clapp oscillators were reconstructed using solid state
components, then noise level and drift should decrease greatly. Greater
overall stability would probably result by using a commercial mixer and
a commercial frequency meter.
A new cell design for the potential detector would be advanta-
geous. The cell depicted in Figure 10 is far from an optimum design.
In principle, this detector is nearly the same as that reported by
Karmen and Bowman (5) (6). A concentric cell as described by them could
be used if the vacuum tube voltmeter is arranged to measure the poten-
tial difference across the whole series coil-capacitance network. An
alternative cell could be constructed according to Figure 22. This cell
should have an even smaller volume than the cell described by Karmen and
Bowman (5) (6). Ideally, Invar or Elinvar should be used to make the
- 73 -
Comparison of Frequency and Potential Detectors
Useful Dyna.ic Range
5 x 102
5 x 103
5 x 102
5 x 102
5 x 103
5 x 102
- 74 -
a- 0 0 0
Q) O O O c nO O as
00 H *Hr H H o E0
C~ ot C r
ci- a o o o o
0 0 1- 4 i $ 4 -4 0 m $4
+ 0 C O O O OC ca
C) ) C) * OM C) 0)
mE 0 0 0 o0 -<0 <
0 O ci w 0 a W
0 O o O o O 0 0 0
+ r- C C)C C C C)
a) CC) c cc)
0 H ) 0 0 *H 0 H
S0 0 0 0 0
00 4 o l 4- 4 l
0H 0 O o o
0 04 -0 0 O O O i
o r I r- I I r
0 0 0 0 0o
0) 0 r
44 0 0 0 0
o 0O O O
EQ H H H 0 H H
a) 0 0
o2 a a a
U0 0 H 0 0 0
o a) H 0 a H H H
00 0 0 0 0 0
+3 C 3 0 +
$4 M 04 0 C C
C: 0: (X 0 H 0 0 f
) 0 0 0
r $i O 'U H O
- 75 -
tank circuit components in order to reduce thermal drift. It is
imperative that the cell should be unsymmetrical.
The r.f. discharge is (according to Karmen and Bowman)(6) 90 per-
cent destructive to organic samples passing through the discharge. Use
may be made of this property by using the discharge as a pyrolysis cell.
By injecting the sample into the discharge cell, fragmentation should
occur and then passing the effluent gases through a packed column and
then through a detector, the sample should give a "fingerprint" of the
organic material sampled.
If a flame is placed within the capacitor cell using a potential
readout as in the potential detector, this could result in a sensitive
gas chromatographic detector and a means of studying flame processes.
Such a detector has already been demonstrated to have considerable
- 77 -
1. Turner, D. W. NATURE, 181, 1265 (1958).
2. Winefordner, J. D., Steinbrecher, D., and Lear, W. E. Anal. Chem.,
32, 515 (1961).
3. Von Hippel, A. R. Dielectrics and Waves, John Wiley & Sons, New
York (1954), p. 262.
4. Francis, G. Ionization Phenomena in Gases, Butterworths Scientific
Publications, London (1960), p. 136.
5. Karmen, A., and Bowman, R. L. Anals. New York Academy of Sciences,
72, 714 (1959).
6. Karmen, A., and Bowman, R. L. Gas Chromatography, Sec. Int. Symp.
Anal. Instr. Div. Instr. Soc. of America, June 1959, Academic
Press, New York (1961), p. 65.
7. Hampton, W. C. Journal of Gas Chromatography, 3, 217 (1965).
8. Winefordner, J. D., Williams, H. P., and Miller, Anal. Chem.
32, 161 (1965).
9. Lovelock, J. E. Anal. Chem., 32, 162 (1961).
10. Seiyama, T., and Kagawa, S. Anal. Chem., 38, 1069 (July, 1966).
- 78 -
Howard Person Williams was born in Wilson County, North Carolina,
and attended public school in Wilson. In September of 1950, he
entered East Carolina College. He joined the United States Air Force
in May, 1953. On January 14, 1956, the author married Joan Calvert.
Upon separation from the service in May, 1957, he reenrolled at East
Carolina College and received a Bachelor of Arts degree in May, 1960,
after completing the requirements for a major in Chemistry and in
After entering the Graduate School of the University of Florida
in June, 1960, the author held the position of graduate assistant. In
April, 1963, he received a Master of Science degree with a major in
Chemistry. On January 4, 1964, he reentered the Graduate School of
the University of Florida in order to work toward a Doctor of Phil-
Mr. and Mrs. Williams have two sons, Howard Person Williams, Jr.,
and Michael Calvert Williams.
- 79 -
This dissertation was prepared under the direction of the
chairman of the candidate's supervisory committee and has been
approved by all members of that committee. It was submitted to the
Dean of the College of Arts and Sciences and to the Graduate Council,
and was approved as partial fulfillment of the requirements for the
degree of Doctor of Philosophy.
December 17, 1966
6^. f i
Dean, Colle e(of/ ArtI and Sciences
Dean, Graduate School
*_ (/ ------
/ ? .3