S. R. M. ELLIS and R. D.
GARBETT
Department of Chemical Engineering,
University of Birmingham, Birmingham, England
A New Equilibrium Still
for the Study of Partially
Miscible Systems
Increasing industrial use of azeotropic
systems for distillation has pointed up a lack
of data for partially miscible regions. This
new still can provide some ofthese data
M A N Y OF THE azeotropic systems,
particularly the aqueous ones, that are
being used for industrial distillations in-
volve partially miscible regions for which
few data are available. An equilibrium
still has therefore been designed which
permits the determination of vapor-
liquid equilibrium data for systems in-
volving both a two-phase liquid and a
two-phase vapor condensate. The still
is similar in principle to the recirculating
stills (3-5) for homogeneous systems but
uses vibratory stirrers for the liquid and
the condensate.
Experimental results show that data
obtained in the partially immiscible
region are in close agreement with the
correlating equation y = K / x . The
following discussion also shows that the
best over-all correlating equation for
the miscible and immiscible regions is
the four-suffix van Laar equation.
Description of Apparatus
The still consists essentially of a liquid
chamber, A , provided with a main heater
of approximately 150 watts, wound onto
a porcelain or glass tube and inserted
into the bayonet, B; a 20-mm. diameter
vapor space containing a saucer-shaped
baffle, C, which eliminates entrainment
by deflecting the liquid spray from con-
denser, D; and condensate chamber, E,
and a line back from this to the liquid
chamber.
The two-phase liquid is boiled in the
liquid chamber where it is completely
emulsified by the stirrer, details of
which are shown below. The vibration
of this stirrer pumps the liquid up be-
tween the inside of the stirrer and the
wall of the heating surface. The boiling
liquid mixture emerges as jets from the
four holes at the top of the stirrer. The
bevelled rings of Kel F promote a down-
ward current outside the stirrer which
reinforces the pumping action. One of
these rings is fitted as low on the stirrer
as possible in order to flush the liquid
round its lower edge. This flow is rein-
forced by the thermal currents, and an
equilibrium mixture of vapor and liquid
is sprayed onto the thermometer pocket.
This arrangement permits an accurate
boiling point measurement of the two-
phase liquid mixture and replaces the
*
JOINTS SILVER SOLDERED. -COPPER TU0E.
BRASS CONE
422
VERY THIN BRASS TUBE THICK COPPER WIRE
SOLDERED
TEFLON RINGS
RINGS 0OUND IN
OSlTiON
THIN BY WIRE
COPPER
BEATEN 0ELL.
Because of its pumping action in emulsi-
fying the two-phase liquid, this stirrer
replaces the Cottrell pump used in stills
for miscible liquids
Cottrell pump used in stills for miscible
liquids.
The equilibrium vapor condenses into
a two-phase liquid which collects in the
condensate chamber, E. The conden-
sate stirrer is powered by the same vi-
brator unit as the main stirrer and the
two stirrers are connected by a stainless
steel yoke.
The problem of maintaining a con-
stant level of liquid in the condensate
chamber and ensuring that the conden-
sate returning to the liquid chamber has
the same composition as that remaining
in the condensate chamber was a difficult
one, and a number of different methods
were tried before a satisfactory arrange-
ment was evolved. The returning con-
densate leaves the condensate chamber
via a sloping tube of I-mm. bore which
has a very fine constriction over the half
of its length next to the chamber. This
ensures that a minimum of separation
occurs in the return tube and that the
condensate leaving is representative of
the dispersed two-phase mixture in the
condensate chamber.
The bearing for the main stirrer con-
sists of three Teflon collars set in a
stout borosilicate glass tube which is
fitted with a B19 cone. Teflon was
chosen because it is resistant to boiling
solvents, has a very low coefficient of
friction, and is hard and tough compared
with other plastics. The collars were
fitted in position by winding unraveled
VOL. 52, NO. 5 M A Y 1960 385
glass thread around them, coating this
with a thin layer of molten polyethylene Results of Four Binary Systems on Samples Taken from the New Still
and allowing to cool, heating the glass
tube sufficiently to melt the polyethylene, B.P., O C. Yw
and forcing the collars in. T o provide
Ethyl Acetate-Water Systemb
additional retention the borosilicate glass
tube is slightly tapered at its lower end. 98.85 0.056 4.055 37.3 0.996
96.2 0.114 12.56 60.3 1.00
The collars were then ground to fit the 86.5 0.489 59.1 86.8 0.68
copper stirrer shaft by means of jewelers' 76.2 0.858 66.8 79.7 0.843
rouge. This bearing provides a smooth 70.55 4.59 71.4 19.5 0.952
steady mount for the stirrer shaft. When 70.55 14.4 70.25 6.11 1.101
70.5 26.9 70.6 3.29 1.28
the collars are new this arrangement 70.45 35.4 70.9 2.515 1.436
alone provides an adequate seal, but as 70.55 40.8 70.9 2.175 1.56
an extra precaution a rubber seal is also 70.5 51.4 71.0 1.73 1.90
fitted. The collars can be replaced when 70.5 60.8 70.25 1.45 2.41
70.45 69.0 70.7 1.29 3.01
they become worn. 70.5 77.5 69.9 1.132 4.25
The condenser stirrer is very light and 71.45 87.37 76.5 1.06 5.69
is mounted on a l/*-inch brass shaft. 73.35 94.445 86.50 1.037 6.85
The condenser itself provides a sufficient 75.55 98.07 94.11 1.008 7.85
Still 1
seal, and the glass is protected from frac- 100.0 0.0216 0.457 10.53 0.996
ture by a thin-walled copper tube which 100.0 0.0221 0.171 3.83 0.996
is supported inside the condenser tube, 100.05 0.02236 0.16 3.54 0.990
between it and the shaft, by means of a 100.05 0.02245 0.2752 6.07 0.992
99.9 0.03103 0.0667 1.074 1.00
piece of rubber tubing. Still 2
In addition to the main heater, there is 100.05 0.01875 0.1860 4.91 0.990
a subsidiary heating coil wound round 100.05 0.02495 0.1940 3.87 0.993
the still body. The body and the side- Still 3
100.0 0.009325 0.119 6.00 0.998
arm are lagged. The lower end of the 99.95 0.01475 0.0382 1.29 1.00
liquid chamber is left unlagged to ensure 99.85 0.0232 0.0527 1.14 1.00
that, with suitable control of stirrer 99.9 0.0263 0.0682 1.30 0.998
power, no stagnant pocket of the denser
phase collects there. Two windows in Methyl Ethyl Ketone-Water Systemb
the lagging are provided so that the spray 98.5 0.05 6.07 70.0 0.991
onto the thermometer pocket can be 92.3 0.487 27.1 38.2 0.970
observed. 87.55 0.931 38.7 32.8 0.985
83.25 1.465 47.9 29.3 0.995
The vibrator is mounted above the 73.7 5.39 61.9 13.85 1.12
still on sponge rubber on a small steel 73.65 8.62 64.5 9.03 1.08
platform to which it is firmly held by two 73.65 19.9 63.9 3.87 1.255
strong springs. The still is provided with 73.65 29.0 64.1 2.665 1.41
73.6 39.5 64.1 1.96 1.655
suitable shock-absorbing supports.
a The subscript represents the system indicated in each group.

Procedure

The still is charged with 200 ml. of

liquid of the desired composition such
that the lower thermometer pocket just
touches the surface of the liquid. T o
prevent a stagnant pocket of liquid from
being formed in the short tube connecting
the bottom of the condensate chamber
and the sampling tap, this space is filled
with clean mercury. The heaters and
stirrer are switched on and the latter is
adjusted to a suitable power. The main
heater is adjusted to give a boilup rate of
3 to 5 ml. per minute, and the jacket
heater is adjusted to give 2 to 3' C. of
superheat. The still is allowed to cir-
culate steadily for 2.5 to 3 hours. After
this period, the boiling point is read, and
the heaters but not the stirrers are turned
Off.
Samples of 10 ml. each of the liquid
and the condensate are then run off into
chilled receivers. Immediately before
each sample is taken the appropriate tap
is flushed into a flask. This ensures a
representative sample and removes the
mercury from the condensate chamber.
For the systenis which have been in-
vestigated, boiling points were deter-
mined with a 0' to 120' C. thermometer
graduated to 0.1' C. which had been
calibrated against an NPL standard
thermometer. Stem connections were
applied and boiling points were corrected
to 760 mm. of mercury.
Vapor liquid Equilibria
in Partially Miscible Binary Systems
Four binary systems have been in-
vestigated-namely, ethyl acetate-water,
methyl ethyl ketone-water (1tEK):
n-butyl alcohol-water, and isobutyl alco-
hol-water. The reagents used were
purified by fractionation, with boiling
points, densities, and refractive indices
agreeing with published values.
The samples taken from the still were
analyzed as follows.
Ethyl Acetate-Water. Homogeneous
samples containing up to about 65 wt.
7,ethyl acetate were analyzed to within
i0.57, by saponification and back titra-
tion.
The standard method of analyzing for
water is that of Fischer (6). However,
this is primarily intended for the detec-
tion and estimation of small quantities of
water. For rhe analysis of equilibrium
still samples of about 10 grams, the cal-
cium carbide method of Klockmann ( 9 )
was modified for use with liquid samples.
Water in the ethyl acetate-water sys-
tem could be determined by this method
to 10.005 gram. The carbide method
was used for mixtures containing more
than about 65 wt.% ethyl acetate.
Below about ly0 by weight of water
the Fischer method (6) was used, because
at this concentration the accuracy of the
carbide method falls off.
Methyl Ethyl Ketone-Water, n-
Butyl Alcohol-Water, a n d Isobutyl
Alcohol-Water. Miscible mixtures were
analyzed by density determinations using
10-ml. pyknometers. Because of the
large difference in density between water
and the organic compounds, this method
is particularly good.
For two-phase mixtures the sample was
first made homogeneous with one of the
pure components.
In regions of very low water content,

386 INDUSTRIAL AND ENGINEERING CHEMISTRY

 EQUILIBRIUM STILL FOR MISCIBLE SYSTEMS

this region is uneven, resulting in some

Results o f Four Binary Systems on Samples Taken from the New Still (continued) scatter of the experimental results. The
xsustern,a YsYsternla
operation of the still gradually improved
B.P., O C . Mole yo Mole % Yaystema YW again as even lower liquid concentrations
73.6 48.2 64.0 1.605 1.94 of ethyl acetate were reached. This
73.55 55.55 64.2 1.375 2.25 difficulty was also experienced at low
73.6 66.2 66.2 1.21 2.79 concentrations of isobutyl alcohol in
73.65 72.8 67.8 1.122 3.295 water.
74.05 79.9 72.3 1.078 3.90
75.00 87.8 76.75 1.01 5.00 The resulting experimental hetero-
76.95 94.48 86.65 1.00 5.63 geneous azeotropic values are compared
with literature values in the table below.
n-Butyl Alcohol-Water Systemb
For methyl ethyl ketone-water there
99.35 0.0997 2.74 55.5 0.996 appears to exist both a heterogeneous
95.8 0.695 15.35 51.6 0.995
94.4 1.05 17.35 41.0 1.02 azeotrope which is referred to in the
93.1 1.945 22.2 29.9 1.021 table above and a homogeneous azeo-
92.75 6.4 26.2 10.9 1.03 trope of composition 66.2 mole70 ketone
92.75 15.4 26.3 4.54 1.14 with a boiling point of 73.6" C.
92.7 23.8 26.35 2.94 1.26
92.8 37.2 25.5 1.82 1.545
93.1 47.7 27.2 1.495 1.79
94.0 57.2 28.3 1.25 2.08 Correlation of Results
95.05 64.8 30.1 1.12 2.38
96.9 71.2 32.4 1.018 2.63 In the region where both the liquid
100.05 79.1 40.9 1.018 2.82 and the vapor condensate are two-phase,
106.2 89.2 58.6 1.015 3.09 the vapor composition is fixed at the
111.1 94.36 73.8 1.00 3.17
115.6 98.20 90.27 0.995 3. 18 heterogeneous azeotrope and the boiling
temperature; hence, the vapor pressure
Isobutyl Alcohol-Water Systemb of each component is constant. Thus
95.8 0.510 14.95 47.3 0.997 the activity coefficient of component 1,
92.05 1.16 24.1 39.0 1.025 based on over-all liquid and condensate
90.0 3.6 33.1 18.8 1.003
90.0 6.4 33 * 45 10.7 1.04 compositions as suggested by Herington
90.0 13.2 33.5 5.2 1.108 ( 8 ) ,is given by
90.05 22.4 33.5 3.05 1.237
90.05 33.5 33.95 2.07 1.432
90.1 44.1 34.25 1.58 1.70
90.35 52.5 35.7 1.37 1.93
90.75 62.9 37.6 1.185 2.37 and K is a constant depending on the
92.45 70.8 42.5 1.11 2.60 azeotrope composition and the vapor
95.5 81.2 53.0 1.067 2.94
100.1 90.05 67.6 1.02 3.24 pressure of the pure component a t the
106.6 98.37 93.05 0.997 3.39 azeotropic boiling point.
All systems were corrected to 760 mm. of mercury. For each system, K has been calculated
from the mean value for the azeotrope
and the lines y = K / x plotted in the two-
phase liquid region. As was expected,
all the experimental results in this region
the Fischer method (6) was again used to Experimental Results are accurately represented by this equa-
obtain higher accuracy. This was not tion. Figure 2 shows the agreement be-
possible for the methyl ethyl ketone- The results for the four systems are tween experimental and correlated re-
water system, for which no satisfactory tabulated above. sults for n-butyl alcohol-water and ethyl
end point could be obtained. . The vapor liquid compositions for acetate-water.
n-butyl alcohol-water are plotted in The results of other workers (7, 7, 74)
Figure 1. seem to indicate that not only does the
Some difficulty was experienced using two-suffix van Laar relationship not hold
this equilibrium still for the system ethyl in the two-phase liquid region but also
acetate-water containing less than 1.24 that the data in the two miscible regions
mole % ethyl acetate, where the boiling cannot be satisfactorily related by one
still mixture was homogeneous and the pair of simple van Laar equations. This
condensate two-phase, with a difference in is demonstrated in Figure 2. The data
boiling point of up to 30' C. between the have therefore been correlated as sug-
condensate and liquid. The boiling in gested by Robinson and Gilliland (76)

Experimental Results Literature Results

System B.P. 760 Mole % B.P. 760 Mole %
A B mm., O C. A mm., O C. A Determination
Ethyl acetate Water 70.5 70.5 70.4 69.7 (14)
73.4 66.2 (11)
MEK Water 73.6 64.0 73.2 65.1 (13)
ea 100
%-Butyl alcohol
HOLE 'lo n - BUTYL ALCOHOL IN LIQUID Water 92.7 26.3 92.7 24.8 (10)
92.7 24.6 (17)
Figure 1. These data for vapor-liquid Isobutyl alcohol Water 90.0 33.5 89.8 33.2
89.8 33.0 (18)
composition were obtained with the 89.6 32.4 (21
new still

VOL. 52, NO. 5 MAY 1960 387


-- 2- SUFFIX V A h L A A 2 EON
INDIVIDUAL VAN L A A R i Q h S
... . . 4 - S U F F I X VAN L A A R ECN
The correlating equations fit the data
well in the middle composition regions:
but badly in the extremities.
For the data to be rated as thermo-
dynamically consistent by this method.
the areas above and below the datum
line must be equal. These areas have
been estimated; the differences between
the positive and the negative areas are
+17y0, - 6 7 , : + 1 6 7 # , and +12% for
the systems ethyl acetate-water, methyl
ethyl ketone-water, n-butyl alcohol-
water, and isobutyl alcohol-water, re-
spectively, which are fairly high.

Nomenclature
end values of log yl at x1 =
0 and log y2 at X P = 0
hypothetical end values
constant in four-suffix van
Laar binary equation
constants in Redlich and
Kister binary equation
20 40 60 vapor pressure of com-
HOLE'Is ETHYL ACETATE IN L I Q L I D
ponent 1
Figure 2. Activity coefficient-concentration plots for n-butyl alcohol-water (left) pressure
and ethyl acetate-water (right) show that the four-suffix van Laar equation mole fraction of component
gives best correlation 1 in the liquid
mole fraction of component
1 in the vapor
using a separate pair of van Laar equa- mental A and B values were used and effective volume fraction
tions for each miscible region. the values of D were chosen to give the activity coefficient of com-
The following constants were employed best fit. The D values used were 2.39, ponent 1
in these calculations : 0.79, 1.15, and 0.97 for ethyl acetate-
SUBSCRIPTS
O denotes pure component

System A B' A' B IC1 Kn ..

1, 2, 3 . . . i . IZ refer to components

Ethyl acetate-water 2.19 *.. 1.54 0.924 0.885 0.938

MEK-water 1.591 0.515 1.25 0.806 0.774 1.004 Literature Cited
n-Butyl alcohol-water 1.845 0.275 1.32 0.544 0.701 0.9625
Isobutyl alcohol-water 1.727 0.545 1.29 0.568 0.686 0.961 (1) Carlson, H. C., Colburn: A . P., IND.
ENG.CHEM.34, 581 (1942).
(2) Colburn, A . P., Shoenhorn, E. M.:
Shilling, D.: Ibid.,35, 1250 (1943).
The values A and B are the experi- 11 ater, methyl ethyl ketone-\\ater, n-butyl (3) Ellis, S.R. M., Trans. Inst. Chem. Engrs.
mental end values, A referring to the alcohol-water, and isobutyl alcohol- 30, 58 (1952).
organic component and B to water. \\rater, respectively. ( 4 ) Ellis, S. R. M., Froome, B. A,: Chem.
The constants A' and B' are hypothetical Typical curves are given in Figure 2. & Ind. 1954, 237.
( 5 ) Ellis, S . R. M., Thwaites, J. M., J .
end values chosen to give the best fit. For the ethyl acetate-water system, the A p p l . Chem. 7, 152 (1957).
These methods of correlation require fit is not too good, but it is a consider- (6) Fischer, K., Angew. Chem. 48, 394
three separate equations to represent a able improvement on the two-suffix van (1935).
single binary system. An attempt to fit Laar equations. For the methyl ethyl (7) Garner, F. H.: Ellis. S.R. M., Pearce.
C. J., Chem. Eng. Sci. 3, 48 (1954).
the more complex four-suffix van Laar ketone-water and the other systems the (8) Herington, E. F. G., J . Appl. Chem.
equations has therefore been made. fit is satisfactory 2. 19 119523.
These are given by Wohl ( 7 9 ) as The four binary systems have also been (9) 'Klockmann, R., Ciienziker Ztg. 7 6 ,
correlated by the Redlich and Kister 706 (1952).
log 71 t ? [ A - D(221 - 3ti2)] (2) (10) Lecat, M., Ann. SOC. sci. Bruxelles 47B,
method (15). Curves have been fitted 1 , 149 (1927).
to the plots of log y l j y ~us. x i . The (11) Ibid., 60, 155 (1940-1946.)
correlation curves were then plotted (12) Lecat, M.? "Tables Azeotropiques,"
where 10th ed., Vol. I, Brussels, 1949.
from the equation 3) Othmer, D. F., Ricciardi, L. G.,
B Thaker, M. S.,ISD. ENG. CHEM.45,
X? 7
log xY' = Blr(1 - 2x1) + C:? [6x: (1 - xi)
2
1815 (1953).
4) Pearce, E. J., Gerster, J. A , Ibid.;
- 11 + D12{(1 - 2x1) 8 x 1 ( l - 11) - XI) 42, 1418 (1950).
5) Redlich, O., Kister, A. T., Ibid., 40,
(4) 341, 345 (1948).
These equations have been plotted for
611 Robinson. G. S.. Gilliland, E. R.,
the systems ethyl acetate-water and using the mean Redlich and Kister con- "Elements of Fractional Distillation,';
methyl ethyl ketone-Lvater. Experi- stants which are given below. Chap. 4, McGraw-Hill, New York,
1950.
(17) Smith, T. E., Bonner, R. F.: IND.ENG.
System CHEM.41,2867 (1949).
1 2 Biz Cl2 Dl2
(18) Stockhardt, J. S., Hull, C. M.: Zbid.,
23,1438 (1931).
Ethyl acetate Water 1.025 0 0.287 (19) Wohl, K., Trans. Am. Inst. Chem.
MEK Water 0.920 -0.220 0.110 Engrs. 42, 215 (1946).
n-Butyl alcohol Water 0.860 -0.230 0.132
Isobutyl alcohol Water 0.785 -0.240 0.090 RECEIVED
for review December 12, 1958
h C E P T E D June 12, 1959
~

388 INDUSTRIAL AND ENGINEERING CHEMISTRY