are identical and are a unique function of pL (Bell, 1970). electric field strength were uniquely determined by pL.

Curves 3 and 5 demonstrate t h a t at 2000 XHz and no =
1011 cmP3the distribution is sensitive to t h e individual values
of both p and L. However, if the central electron density is
increased, (curve 4), the distribution becomes coincident with
t h a t obtained a t lower frequencies and is again dependent
only on the product pL.
The distribution of electric field strength is shown in Figure
2. Here again the curves for 20 and 200 M H z coincide. At
2000 V H z there is a very pronounced effect of n+ and p , a s
can be seen b y comparing curves 3, 4,and 5. The maximum
in the distribution is reduced and shifted towards the center
of the discharge for a decrease in noor a n increase in pressure.
The behavior demonstrated in Figure 2 can readily be
understood in terms of Equation 4,which governs the mag-
nitude of E , as a function of electron density. When r >> a,
then
(5)
except for a very narrow region near x* = 1 where (T - 1)
approaches the magnitude of CY.
The central electron density, no, was shown to be a function of
pL and the ratio of t h e power density to the gas pressure.
These results were combined to illustrate that the distribu-
tion of rates for electron-molecule reactions occurring in two
discharges of different size could be made identical if the
power density and the product pL were both held constant.
The original results were deduced for a discharge operating
a t 20 MHz. The present analysis allows the same criteria of
similarity t o be extended t o discharges operating a t other
frequencies, provided the inequality ro >> CY is satisfied for
both discharges.
Nomenclature
D, = ambipolar diffusion coefficient, cm2/sec
e = charge on electron, coulomb
E = electric field strength, volts/cm
L = plate separation, cm
m = mass of electron, kg
n = electron density, electrons/cm3
n* = nlna
p = pressure, torr
r = dimensionless electron density
2id = drift velocity, cm/sec

Where Equation 5 is valid, the electric field distribution is x = distance, cm

x* = 2x/L
insensitive t o the magnitude of n+, the pressure, or the driving a = Townsend ionization coefficient, ionization/cm
frequency. By contrast, when r 2 a , all three variables have CY = dimensionless elastic collision frequency
a n effect upon the field distribution. Since t h e local value of e, = permittivity of free space, faradayslm
the effective field strength determines the local value of the v = elastic collision frequency, sec-1
v i = ionization frequency, sec-l
ionization frequency, t h e relative magnitudes of r and a will w = driving frequency, sec-l
also influence the behavior of the electron density distribution.
The effect of electric field strength distribution on the local literature Cited
rates of a reaction is illustrated in Figure 3. The normalized
Allis, W. P., “Handbuch der Physik,” S. Flugge, ed., Vol. 21,
rate of ionization is plotted as a function of position. As p. 283, Springer-Verlag, Berlin, 1956.
can be seen, the distributions of ionization rates for 20 and Bell, A. T., IND.ENG.CHEM.FUNDAM. 9, 160 (1970).
200 RIHz are identical. For the same value of pL the dis- Dinan, F. J., Fridmann, S., Schirmami, P. J., Adaan. Chenz.
Ser. No. 80,289 (1969).
tribution a t 2000 NHz is different and is, furthermore, sensi- Eremin, E. N., Vasil’ev, S. S., Kobozev, K.I., Zh. F i z . Khim. 9,48
tive to variations in p and L , even though their product is (1937).
held constant. When no = 10l2ern+ (curve 4),the distribu- Streitweiser, A., Ward, H. R., J . Amer. Chem. SOC.85, 539 (1063).
tion of rates is almost identical t o t h a t a t the lower frequencies. ALEXIS T. BELL
Based on a simplified model of a high frequency discharge, University of California
i t has been shown recently (Bell, 1970) t h a t the value of Eoe/p Berkeley, Calif. 94720
RECEIVED for review January 14, 1970
and the dimensionless distributions of electron density and ACCEPTED July 20, 1970

Correlation of Kinetic Data from laminar Flow-Tubular Reactors

By expressing the residence time of annular elements of the fluid in a laminar flow reactor as a function of
reactor length and radial position it i s possible to relate the fractional conversion of reactant to dimension-
less groups containing the rate constant, inlet concentrations, reactor volume, and flow rate. The functional
dependence will vary with the order of the kinetic expression relating the rate of disappearance of reactant
to concentration. For kinetic orders other than zero, the fractional conversion obtained in the laminar flow
reactor will be less than that calculated by the plug-flow assumption. This analysis extends the previous
treatments to include the general nth-order rate expression and first-order consecutive reaction rate ex-
pressions.

In many experimental investigations using liquid reactants,
the velocity and concentration profiles are not constant across
the reactor and integrated rate expressions based on the plug-
flow concept, that relate conversion and bulk contact time,
are not applicable. I n the laminar flow regime, present in
tubular reactors at Reynolds numbers less than 2100, a
parabolic velocity profile can exist over a major portion of
the reactor and the residence time of the annular elements of
fluid within the reactor will vary from some minimal value a t
the center line of the tube where the velocity is a maximum
to some high value near the wall where the fluid velocity is
zero, This distribution of contact times resulting from laminar

Ind. Eng. Chem. Fundom., Vol. 9, No. 4, 1970 681

:iow and the correct relation between conversion and flow or expressed in terms of reactor volume and flow rate ob-
;ate was first considered by Bosworth (1948), who developed tained by multiplying and dividing by rRw2,
Ihe case for zero-order kinetics. Somewhat later this approach
,vas extended to second-order reactions by Denbigh (1951),
while the first-order case was developed by Cleland and
Kilhelm (1956). This is the same result obtained for plug-flow reactors.
As there is a continued interest in this area, it was desirable For second-order reaction n equals 2 and Equation 7 be-
to pull this information together and to extend the analysis to comes, after integration,
include the general nth-order reaction and a treatment of
first-order consecutive reactions.
f = 2p - 2p2 In ( 1 + l/p) (11)
The reactor is assumed to be long compared to its diameter, A plot of conversion us. p is given in Figure 1. Also shown,is
ith no change in density due to reaction, and a parabolic the curve for plug flow.
.elocity profile established over the entire length.
f = 2P/(1 + 20) (12)
For reactions whose rates are first-order, the reactant con-
centration at any time is given by
l'hese are the simplifications required for a tractable analysis CA = CAoeekt (13)
,f the problem. The fluid in any annular element behaves like
The molar flow rate a t the exit of the reactor becomes, after
:iconstant volume-isothermal reactor with the molar flov
substitution of Equations 13, 5, and 1 into Equation 2 ,
;ate of reactant in the reactor given by
d,Ta = c~*U*2rRdR (2)
,vhere C A = concentration of A and K A = moles of A per
mit time.
Expressing the rate of disappearance of A by
2
[ - ( 3 1R",
1 - - 2dR (14)

,ives after integration for n # 1,

(3)
- h'a
=--
2NAo
1 -f
2
- so1
Collection of terms into dimensionless groups gives
e- a/V
VdV

C'41-n = Cao'-" + k ( n - 1)t where a = kL/U,,, = kV,/2Qp Substituting 2 for l / V leads

to the tabulated function E3 (a).
1 (4)
CA = cAo(1 + k[n - ~ ] C A O " - ' ~ ) ~
Since the time required for the reactant in each annular ele-
ment to reach the exit is Values for the third-order exponential integral, &(a), may
L L be found in the "Handbook of Mathematical Functions"
(1964).For most purposes, the plots found in Figures 2 and 3
should be adequate.
I n many systems of interest, consecutive, irreversible re-
substituting Equations 5, 4, and 1 into Equation 2 gives actions occur and intermediates are involved, as depicted in
the following sequence.
~-2132~ (17)
-d-c a - klCA
- (18)
dt
[ - (g)2]
1 2rRdR (6)

1 hich leads to Equation 7 when the above terms are collected

to dimensionless groups and integrated.
_ - kzCB
dCc
dt

LYA dNA = TRm'UrnaxCAo l1+ (1 p/V)l'l-"VdV (7) Integration of the above and a material balance for component
A gives
'i'he group of terms in front of the integral sign on the right- CA = CAoe-k'l (21)
*land side is twice the molar feed to the reactor, and Equa- ki
.ion 7 when written in terms of the fraction reacted becomes C B = CAo- (e--Iclt - e-kzt) (22)
kz - ki

3
2
= s, 1
(1 +'p/V)lI1-nVdV CA, = CA + CB + CC
The fractional conversion of starting material X is given by
(23)

For a zero-order reaction n is zero and integration of Equation Equation 16 and similar substitutions for CB and t give the
8 gives related equation for N B with k l kz *
(9)

382 Ind. Eng. Chern. Fundam., Vol. 9, No. 4, 1970

 10
8.0
6.0

0.4

2 .o

?I
<O
V

Q
,“ 1.0
, 0.8
0.6

0.4

0.2

0.1
0.2 0.4 0.6 0.8 1.0
f = FRACTION REACTED
Figure 1 Second-order reaction 2A -+ P
1

0.5 0.8
0.4 0.6

0.3
0.4

0.2
m
Y

I9 0.2
W

0.1 m
Y

N
0.08 W

0.1
0.06
0.08
0.04 0.06

0.04
0.02 0.03
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 0 0 2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
,Umrx
.A
Figure 2. Third-order exponential integral Figure 4. Second-order exponential integral

If kl and k2 are equal, the concentration of B is obtained by

(25) integrating
dCe
where - = kl(CA - C B ) (28)
dt
(26) C B = CA0(klt)e-”‘ (29)
with the molar flow rate of B a t the exit given by

Ind. Eng. Chcm. Fundom., Vol. 9, No. 4, 1970 683

 which was obtained from Equation 2 by substitution for f = fraction of A reacted, dimensionless, f = 1 -
C B and t and setting Z equal to 1/V. NA/NA,
(moles/cm3j1-~
From the material balance for reacted A, the molar flow rate k = rate constants,
sec
of C is given by L = reactor length, cm
A‘*, = .Y* + S B + A’c (31)
NA
ATA, =
= molar flow rate of reactant A at exit, moles/sec
molar flow rate of reactant a t inlet, moles/sec
and the fraction of A , appearing as C is given by ATB,Sc = molar flow rate of products a t exit, moles/sec
Qf = bulk flow rate, cm3/sec
R = radius, cm
_ -- 1 - 2[E3(a) + a E z ( a ) ]IC1
, = IC2 (32) RW = radius of reactor, cm
h-Ao 1 = time, sec
U = velocity, cm/sec
U,,, center line velocity, cm/sec

(a
=
V = dimensionless group, 1 -
(33) V, = reactor volume, rR,*.L, cm3
Z = 1/v
Plots of &(a) and &(a) are given in Figures 2 and 4. k(n - ljCa,”-’L
P = for nth-order reaction =
Urn,,
Nomenclature
Literature Cited
a = dimensionless woun for first-order reaction.
kiL - kiVr Bosworth, R. C. L., Phd. Mag. 39, 847 (1948).
Cleland, F. A., Rilhelm, R. H., A.Z.Ch.E. J . 2 , 4 6 9 (1956).
C,,, 2Qj Denbigh, K. G., J . A p p l . Cheni. 1, 227 (1951).
a’ = dimensionless group for first-order reaction, “Handbook of Mathematical Functions,” Milton Abramowitz
_kzL -_kzV,- ~ and I. A. Stegun, eds., National Bureau of Standards Applied
Urnax 2Qf hlathematics Series, 55 (1964).
C* = concentration of reactant in react.or, moles/cm3 MARVIN RI. JOHYSON
CAO = concentration of reactant a t inlet of reactor,
moles/cm3 Phillips Petroleum Co.
C B ,Cc = concentration of products in reactor, moles/cm3
Bartlesville, Okla. 74003
&(a) = second-order exponential integral RECEIVED for review February 24, 1970
&(a) = third-order exponential integral ACCEPTED *July27, 1970

CORRESPONDENCE

Parametric Pumping and Cycling Zone Adsorption

SIR: I have what I believe are important points regarding
the separation techniques of “parametric pumping,” “cycling
zone adsorption” (Pigford et al., 1969b), and the like.
Mechanism of Separation
The cause for separation in such devices is not nearly as
complicated as has been implied by most of the papers in
the area-e.g., “coupling of oscillatory thermal and mass
fields with alternating flow displacements” (Wilhelm et al.,
1966), “periodic, synchronous, coupled transport system”
(filhelni et al., 1968), “The origin of the separation effect
lies in the ability of the column to store solute temporarily
on the solid adsorbent particles, withdrawing it from the
lean bottom product and subsequently adding it to the top
product as each of these penetrates portions of the column in
turn” (Pigford et al., 1969a).
Much of the confusion has arisen, I think, through efforts
to contrast parapumping ideas with equilibrium stage con-
cepts (Sweed and Rilhelm, 1969) rather than to show how
they fit into equilibrium stage concepts, which are funda-
mental idealizations of all separation processes involving more
than one phase. First of all, the cause of the separation is
what has been for years called “regeneration”-Le., the sorp-
tion of a solute out of a volume of fluid by a volume of sor-
bent, followed by the desorption of the solute from the volume
of sorbent into a second volume of fluid. It is most common in
molecular sieve, chemisorption, and ion exchange systems
where in practice the sorption equilibrium is easy to shift by
large amounts, so that relatively large volumes of fluid are
in contact n i t h relatively small volumes of sorbent with a
relatively large amount of solute transfer before each shift of
the equilibrium. The equilibrium may be shifted in any num-
ber of ways -e.g., by changing temperature or some property
of the two fluid phases such as p H or concentration of a third
component. To effect a separation no column or columnar
shape is required. For example, a n apparatus shaped as in
Figure 1 could be used to effect a separation (however small)
between the two reservoirs. T o effect a separation no reversal
of velocity is required; even in a closed system, only recycle or
regeneration is required. An apparatus as shom-n in Figure 2
could be used to effect a separation (however small) by proper
altering of the temperature.
Such devices are parametric pumps. However, they will not
give very large separations, because they are essentially “one-
stage” devices. The use of a long column “stages” or “se-
quences” the parametric pumping operation in an efficient
way, just as the sequencing of stages in series in a column
greatly increases the efficiency of distillation, extraction, and
chromatography. By “staging” here I mean staging in the
usual sense of staging in space and not in time. Staging in time
has been implied as the difference between parametric pump-
ing and other types of separation (Sweed and Wilhelm, 1969;
Wilhelm et al., 1968). This is not true. The increase in separa-
tion with time of a batch parametric pump (to some limiting
value) is analogous to the increase in separation with time of
a total reflux batch distillation column (to some limiting
value). Thus the separation is caused by “regeneration” and
is enhanced in the usual way by staging the cause of separa-
tion in a long column. Neither regeneration nor staging in a
column is a new concept. Their use together is new and this
is the contribution of Wilhelm.

684 Ind. Eng. Chcm. Fundam., Vol. 9, No. 4, 1970