are extended t o high pressures t o make

possible more intelligent control over

certain industrial processes. Y - X - P - T
curves are evaluated up t o a temperature of 275" C. a t saturation pressures.
Critical temperatures and pressures of
the system are also obtained.

ETHANOL-WATER
SYSTEM

JOHN GRISWOLD, J. D. HANEYl,

AND v. A. KLEIN2

Vapor-Liquid Properties
at High Pressures

The Jniversity of Texas, Austin, Texas

ERTAIK operations used in the manufacture of an-

system determined with it are reported in another article

(1A). The pressure gage had a total range of 1500 pounds
and was graduated in 10-pound divisions. Experimental
pressures were read to 1 pound, and the gage was checked
against a dead-weight tester. Temperatures were determined
by calibrated iron-constantan thermocouples and a low-range
potentiometer. The accuracy of the temperature observations
was approximately 0.5"
A combination check on thermocouples and gage was obtained by observing the vapor pressure-temperature curve for water in the same apparatus, up
to 1500 pounds pressure. The ethanol used throughout the
work was U. S. P. grade material. Distilled water was taken
from the laboratory supply. Analysis of samples was obtained from densities at 20" C., determined by the balanceplummet-thermostat method of Osborne, McKelvy, and
Bearce (8).

hydrous alcohol, ethylene from alcohol, silica aerogel,

and other products depend on the high-pressure vaporliquid behavior of ethanol-water. Data of satisfactory accuracy and pressure range have not heretofore been available.
VAPOR-LIQUID EQUILIBRIA

e.

Vapor-liquid equilibria up to 15 atmospheres were reported by Grumbt (3). However, a study of his data shows
serious scattering of the points and self-inconsistencies which
he attributed to refluxing in the vapor line of his apparatus.
The vapor-liquid equilibria of the system were determined
a t several constant temperatures with an all-steel recirculation
type apparatus. The development of this apparatus and the
high-pressure vapor-liquid equilibrium of the benzene-toluene
The photograph shows a control panel for continuous rectification of
190 proof alcohol from wheat mash, a t the plant of Joseph E.Seagram &
Sons, Inc.: section of column may be seen in the background.

1
2

701

Present address, Joseph E. Seagiam & Sons, Inc., Lawrenceburg, Ind.

Present address, Dow Chemical Company, Freeport, Texas.

INDUSTRIAL AND ENGINEERING CHEMISTRY

702

-A . VAPOR- L iaum

EQUIL IERIUM

OF

Vol. 35, No. 6

C R I T I C A L TFhPER.4 TURES

*-f-f3''

E THANO1 - WA TER
CONSTANT TEMPERAT U K S

4.VAPOR-LIQUID

PER CCNT ETHANOL

ro

20

30

40

so

60

70

80

30

10

20

30

40

50

60

70

80

90

EQUILIBRIUM

OF

FTHA NOL- WA TER

CONSTANT PRESSURES

-7

--b CAREY 8 LEWIS

20

30

4,O

50

Figure 1

CALCULATED F
EXPERIMENTAL

THANOL IN

IO

(1

NOYES 8 WARFEL (

iiauio
60

70

80

90

The experimental vapor-liquid equilibrium data are summarized in Table I and plotted as isothermal Y-X curves on
Figure 1A. To develop the constant pressure Y-X plot of
Figure lB, pressure isotherms were plotted and then curves
of vapor composition against pressure for several constant
liquid compositions were constructed. Interpolation of the
latter curves gave the isopiestic Y-X graph (Figure 1B).
CRITICAL TEMPERATURES

Critical temperatures were determined b y observing the

behavior of mixtures of known composition when sealed into
glass tubes and heated. The sealing technique and the heater
were described in an earlier article ( 2 ) . The sample tubes
were of Pyrex, 4 mm. 0.d. with a 1-mm. wall thickness and

an over-all length of about 60 mm. The observed temperatures (after emergent stem corrections) were accurate
to approximately 1" C.
The constant volume behavior of this system as the critical
state is approached was found to differ somewhat from that
characteristic of pure compounds and hydrocarbon mixtures.
The tubes were charged with alcohol solution to approximately one third their volume a t room temperature and were
then sealed. I n the determination idiich followed, the liquid
volume or meniscus level rose with temperature. TT'ithin
the last 1" C. below the critical, the meniscus rose from about
two thirds of the tube height to completely fill the tube.
The vapor phase apparently became zero, and disappearance
of the meniscus could not be observed. On slow cooling from
1' C. above this temperature, a white cloud 1%-ouldsuddenly
appear, quickly condense, and reveal a liquid meniscus.
These rising and falling temperatures differed by less than
1' C. for all cases in which the tubes were charged to between
20 and 40 per cent of their volume a t room conditions. This
temperature is therefore taken as the true critical. Further
support of this hypothesis is obtained from a study of relations for relative vapor and liquid volumes a t constant total

INDUSTRIAL AND ENGINEERING CHEMISTRY

June, 1943

volume. Experimental critical temperatures are summarized

in Table I1 and plotted on Figure 1C. The figure also shows
the recent data of White (IO), which lie 2" to 5" C. lower
than the new values.
The critical temperature of this system approaches linearity with composition (on thc
weight basis) much more closely than doe?
that of binary hydrocarbon mixtures ( 7 ) .

703

point. The ethanol-water azeotrope, which occurs a t 95.6

weight per cent alcohol at atmospheric pressure, lies at 95.3
weight per cent alcohol a t a total pressure of 1450 mm. (9)
This indicates only a slow change in the azeotropic compo-

CRITICAL PRESSURES

Critical pressures were determined by the

same procedure and with an apparatus similar to that used in an earlier article (9). The
equipment (Figure 2) consisted of a steel
bomb connected to a pressure gage by a
loop of l/*-inch 0.d. annealed steel tubing.
The bomb was suspended in a fused salt
(IETS) bath by a sliding linkage to a motordriven cam. This gave the bomb vertical
reciprocating agitation. The pressure gage
described under "Vapor-Liquid Equilibria"
was utilized when the pressures were below 1500 pounds. At higher pressures a
3000-pound gage graduated in 20-pound
divisions and checked against a deadweight tester was used.
The bomb was charged with 100 cc. of
alcohol solution and heated to the boiling
point to eliminate air. The gage tubing
(used to vent air) was then connected to
Figure 2. Apparatus for Determination of Critical Pressures
the gage, and temperature-pressure data
were taken through the critical temperature
of the mixture, as read from Figure 1C.
The usual range was from 20" below to 20" C. above the
sition with temperature and pressure. Azeotropic composicritical temperature, in which six to eight observations were
tions and behavior a t higher temperatures and pressures have
made. The bath temperature and observed pressure were
apparently not been reported. The critical temperature of
constant for at least 10 minutes prior to readings.
anhydrous ethanol is 243" C. From Figure 1C the critical
composition at 250" C. is approximately 80 mole per cent
alcohol, and a t 275" it is approximately 45 mole per cent.
A minimum-boiling azeotrope must exhibit (a) no difference
in composition between liquid and equilibrium vapor, and
EQUILIBRIUM
OF ETHANOLWATER
TABLE
I. VAPOR-LIQUID
( b ) an isothermal maximum pressure a t some definite compoTemperature,
Mole %,Ethanol
Mole 7% Ethanol
Pressure
c.
in Liquid
in Vapor
Lb./Sq. In. kbs.
sition. To show the aeeotropic behavior more clearly at
7.3
33.4
107
150
250 O C. and above, the pressure-composition diagram of
13.8
41.4
119
26.5
Figure 3 was constructed. It is evident that at 275" C. no
48.7
130
51.4
61.6
145
azeotrope exists. At 250" an equilibrium determination
63.9
69.5
147
gave 68.8 mole per cent alcohol in the liquid and 69.6 in the
5.8
24.7
300
200
11.4
33.8
331
vapor at a pressure of 1010 pounds. Since the critical
23.7
43.3
370
composition is 80 mole per cent alcohol and the critical
49.7
58.5
415
63.3
68.1
424
pressure is slightly above 1000 pounds a t 250" C., the presence
250

275

12.6
26.0
40.0

24.5
34.4
42.5

1176
1341
1492

The data were plotted and the pressures at the critical

temperatures read from the plots. The results are included
in Table I1 and plotted on Figure 1D. The critical pressure
is seen to be substantially linear with weight per cent ethanol
at concentrations below 70 per cent.
BEHAVIOR OF AZEOTROPE

Separation of a mixture b y fractional distillation may be

limited b y the existence of either an azeotrope or a critical

TABLE11. SUMMARIZED
DATAFOR CRITICALTEMPERATURES
AND PRESSURES
Weight
Ethanoy

Mole 7
Ethan07

100
94.0
88.7
84.0
79.0
74.1
69.2
64.4
61.0
55.1
49.6
45.1
40.0
35.8
26.9
24.0
18.7

100
86.1
75.5
67.3
59.4
52.8
46.8
41.5
38.0
32.4
27.8
24.4
20.6
17.9
12.6
11.0
8.3

Critical
Temp
C."

243.0
248.0
253.1
259.2
264.2
270.3
277.3
284.4
288.4
296.5
307.5
311.6
317.6
325.7
334.8
339.8
344.9

Weight %
Ethanol
100
86.5
80.0
63.9
46.0
30.9
16.3

Critical
Mole 7 Pressure
Etheno? Lb./Sq. Ih.
100
71.5
61.0
40.9
25.0
14.9
7.1

925
1100
1220
1618
2060
2440
2830

704

INDUSTRIAL AND ENGINEERING CHEMISTRY

Vol. 35, No. 6

is an exceptional rather than a general situation (4). The

system thus does not fall into either of two classes which
exhibit retrograde condensation as enunciated by Katz and
Kurata (6). A necessary condition for retrograde condensation is the existence of separate points of maximum
pressure and of niaxiniuin temperature during the coexistence
of liquid and rapor, by a mixture of some definite composition. This requires that the peak of a border curve be round
rather than sharp. Although few of the present l--X data
lie close to the critical locus, vapor and liquid isotherms
extrapolate so near the same point on the locus curl-e that
zones of retrograde condensation must be either extremely
small or nonexistent.
ACKNOWLEDGMEhT

S. S.Sutherland assisted in the construction of the diagrams.

LITEHATUHE CITED
I

M O L E PER C h N T E T H A N O L

Figure 3

or absence of a n azeotrope between 70 and 80 mole per cent

alcohol a t this temperature cannot be definitely ascertained.
PRESSURE-TEMPERATURE RELATIONS

The most widely used method of obtalning high-pressure

vapor-liquid equilibrium data on binary systems heretofore
has been to observe dew and
bubble points on samples of
definite composition in a variable-volume calibrated glass pressure tube (6).
The resulting P-V-T data are
customarily plotted directly as
envelope curves. The envelope
for each composition is tangent
to the critical locus curve.
Vapor-liquid equilibria may be
calculated from P-X-Y plots
of dew and bubble point curres.
On the other hand, the present
method yields direct Y-X data.
Hon-ever, a P-T diagram was
constructed for ethanol-water
(Figure 4) since i t is of interest
for comparison with other systems similarly plotted. To develop this diagram, the data
were interpolated from a P-X
chart with pressure on a logaritlimic scale, The spacing of the
isotherms was nearly linear with
temperature, and P-T values for
compositions of 25, 50, and 'it?
niole per cent alcohol were read
from the plot. The results along
with vapor pressures of water,
ethanol, and the critical locus
are shown on Figure 4.
The critical locus contains
no point of higher pressure or
higher temperature than the
critical values for aater. This

( l j Carey and Lewis, IND.

E B G . CHEM.,24, 882 (1932).
(1A) Griswold, Andres, and Klein, Trans. A m . Inst. Chem. Engis.,
39, 223 (1943) ; Petroleum Refinerg, 22, No. 6 (1943).
(2) Griswold and Kasch, 1x11.EBG. CHEM.,34, 804 (1942).
(3) Grumbt, J. A., Tech. Mech. Thermodgnam., 1, 309, 349 (1930).
(4) Hougen and Watson, "Industrial Chemical Calculations",
2nd ed., pp. 406, 407, New York, John TTiley & Sons, 1936.
(5) Katz and K u r a t a , IND.
ENG. CHEM.,32, 817 (194CI)
(6) K a y , W. B., Ibid., 30, 459 (1938).
(7) Mayfield, F. D., I b i d . , 34, 844 (1942).
(74.) Noyes and Warfel, J . A m . Chem. Soc , 23, 463 (1901).
(8) Osborne, McKelvy, and Bearce, Bur. Standards, B d l . 9, 371
(1913).
(9) Wade and Merriman, J . Chem. Sac., 99, 997 (1911).
(10) White, J. F., Trans. A m . I n s t . Chem. Engrs., 38, 435 (1942).

Figure 4