ESTIMATION OF PARAMETERS FOR THE

NRTL EQUATION FOR EXCESS GIBBS ENERGIES OF

STRONGLY NONIDEAL LIQUID MIXTURES

H E N R l R E N O N ’ A N D J . M. P R A U S N I T Z
Department of Chemical Engineering, University of California, Berkeley, Calif. 94720

Charts are given for estimating NRTL (nonrandom, two-liquid) parameters from
limiting activity coefficient data or from mutual solubilities. Calculated and
experimental results are given for binary and ternary systems.

FORquantitative studies on nonideal properties of liquid
mixtures it is convenient to express these properties
with excess functions. For liquid mixtures a t modest pres-
sures remote from critical conditions, the excess functions
are insensitive to pressure; a t constant temperature, there-
fore, the excess functions of such mixtures depend only
on liquid composition.
The excess Gibbs energy is the excess function of pri-
mary interest in chemical engineering and numerous pro-
posals have been made for relating the excess Gibbs energy
to liquid composition; the best known, perhaps, are those
by Margules and van Laar. A generalization of such pro-
posals was made over 20 years ago by Wohl (1946), but
more recently it has been shown that some useful relations
for the excess Gibbs energy could not be obtained from
Wohl’s generalization. I n particular, Wilson’s “local
composition” concept (1964) resulted in an equation quali-
tatively different from those proposed earlier. The local
composition concept provides a convenient method for
introducing nonrandomness into the liquid-mixture model.
Wilson (1964) and Orye and Prausnitz (1965) showed
that Wilson’s equation is in many respects more useful
and more directly applicable to strongly nonideal mixtures
than any other two-parameter equation, and a particular
advantage of Wilson’s equation for binary systems follows
from its straightforward generalization to multicomponent
mixtures without need for ternary (or higher) parameters.
However, as pointed out by Wilson (1964) and Orye and
Prausnitz (1965), Wilson’s equation also contains some
undesirable features; especially important is its inap-
plicability to liquid mixtures with only partial miscibility.
I n an effort to overcome this disadvantage while
retaining the useful local composition concept, we have
developed a new equation for the excess Gibbs energy
based on Scott’s two-liquid model of mixtures (1956).
This new equation, called the N R T L (nonrandom, two-
liquid) equation, was described in detail by Renon (1968),
who showed that it appears to be applicable to a wide
variety of mixtures for calculating vapor-liquid and liquid-
liquid equilibria. Robert L. Pigford suggested that the
practical utility of the N R T L equation could be enhanced
if we made available convenient charts for estimating
’Present address, Institut Francais du Petrole, Rueil-Malmaison,
France.
NRTL parameters from limited experimental data. A
number of such charts are presented here.
NRTL Equation
The NRTL equation for the molar excess Gibbs energy
of a binary mixture is a function of mole fractions x1
and x 2 :
where
with
g
12 = gz,
Equation 1 contains two temperature-dependent
parameters, (g12 - gZ2)and (921 - gll), in addition to a
nonrandomness parameter, a 1 2 ,which, to a good approx-
imation, does not depend on temperature and can often
be estimated with sufficient accuracy from the nature
of components 1 and 2 (Renon and Prausnitz, 1968).
For some systems it appears that (g~, - gll) and ( g 1 2
- g22) are linear functions of temperature. From Equation
1 the activity coefficients, yl and y2, are obtained by
differentiation (see Equations 21 and 22 of Renon and
Prausnitz, 1968). The N R T L equation is readily gen-
eralized to multicomponent mixtures; the result for such
mixtures contains only binary parameters.
When experimental phase-equilibrium data are available
over a range of composition, the optimum parameters
(gzl - g11) and (g12 - gZ2)can be determined by a least-
squares fit. [Such a calculation is tedious but can be
performed easily with an electronic computer, as shown by
Prausnitz et al. (1967).] However, it is also possible to
estimate these two parameters from limiting activity-
coefficient data, or, for partially miscible systems, from
mutual solubilities.

VOL. 8 N O . 3 J U L Y 1 9 6 9 413
NRTL Parameters from limiting Activity Coefficients and
The limiting activity coefficients are designated by rl" 712 =S -D (5)
and y Z . I n many cases these can be obtained by a mini-
mum experimental effort, and for a large variety of I n Figures 1 to 5, for given values of ( ~ 1 2 , we present
mixtures they can be estimated from the correlation of the relations between rl",7;) and S and D.
Pierotti et al. (1959). At a given temperature T and a To use Figures 1 to 5 we proceed as follows. First,
fixed a 1 2 , values of yl"(T) and r Z ( T ) uniquely determine experimental or estimated values of rl" and y2m a t the
values of parameters 712 and 7 2 1 . T o obtain these param- same temperature must be available. Second, depending
eters from rl" and y;, let on the chemical nature of the mixture, a value of a12
must be chosen; the values recommended by Renon and
s E %(721 + 712) (2) Prausnitz (1968) provide a convenient guide. Upon cal-
and culating loglo (yPy2m) and log,, (yl"/y2m), Figures l to 5
give values of S and D, which, in turn, yield 7z1 and
D E ?4(721 - 712) (3) r12as given by Equations 4 and 5.
I t then follows that T o illustrate such a calculation, we consider three sys-
tems: ethanol-hexane a t 55" C., methyl ethyl ketone-
hexane a t 60°C., and propanol-water a t 60°C. For each

Figure 1. Parameters in NRTL equation from activity coefficients a t infinite dilution for
a12 = 0.20

Figure 2. Parameters in NRTL equation from activity coefficients at infinite dilution for
a12 = 0.30

414 I & E C PROCESS D E S I G N A N D DEVELOPMENT

Figure 3. Parameters in NRTt equation from activity Coefficients at infinite dilution for
ai2 = 0.40

Figure 4. Parameters in NRTL equation from activity coefficients a t infinite dilution for
a12 = 0.47

Figure 5. Parameters in NRTL equation from activity coefficients at infinite

dilution for a12 = 0.30
Small or negative deviations from Raoult's law

VOL. 8 N O . 3 JULY 1 9 6 9 415

 Table 1. NRTL Parameters from Limiting Activity Coefficients
T., D g12 - g22,
Cal./Mok gzi - gll,
Cul./Mole 6P/Pb 6y'
System OC. loglor; log10y2" a12 S
Ethanol(l)-n-hexane(Z) 55 1.322 1.013 0.47 1.940 0.386 1010 1515 0.8 1.2
Methyl ethyl
ketone (1)-n-hexane(2) 60 0.565 0.507 0.20 0.669 0.286 253 630 0.7 0.8
l-Propanol(l)-water(2) 60 1.147 0.533 0.30 1.327 1.300 18 1735 1.7 2.1
a From Pierotti's correlation. * R.m.s. relative deviation in total pressure x 100. e R.m.s. absolute deviation in vapor male fraction x 100.

1.0 I I 1

- CALCULATED I Or , I --
, 0 EXPERIMENTAL
CALCULATED
, o EXPERIMENTAL

075 I -
075 300 I"
E
E

J
K
W
8 >
050 200
a W
%
K
L

J
2
0 25 too e

OY I I I I o 00 L 5
0.2 0.50
. A
0.75 1.0 O
0 025 0 50 0.75 10
X~~~~~~~~
X~~~~~~~

Figure 6. Vapor-liquid equilibria for ethanol-n-hexane at Figure 8. Vapor-liquid equilibria for 1-propanol-water
55" c. at 60" C.

positions and total pressures (Kudryavtseva and Susarev,

I
I."
n, , I 1963, Figure 6; Hanson and Van Winkle, 1967, Figure 7;
- CALCULATED Murti and Van Winkle, 1958, Figure 8). Agreement between
< O EXPERIMENTAL
calculated and experimental results is good for all three
systems; mean deviations between calculated and observed
0.75b
/ ' vapor compositions and total pressures are reported in
Table I.
NRTL Parameters from Mutual Solubilities
For binary mixtures with only partial miscibility, it
is possible to estimate NRTL parameters from mutual
solubilities a t the same temperature. Let x7 be the mole
300 E fraction of component 1 in phase a, where i t is smaller,
and let x i be the mole fraction of component 2 in phase
b, where it is smaller, such that x?/x! 5 1. For a fixed

0 1
I/ I I I I o
value of aI2 we may obtain S and D from Figure 9,
10, or 11; the NRTL parameters are then found from
0 0.25 0.50 0.75 1.0 Equations 4 and 5.
X~~~~~~
Figures 9 to 11 were prepared by solving simultaneously
the equilibrium equations
Figure 7. Vapor-liquid equilibria for methyl ethyl ketone-
n-hexane at 60"C. 7% = 7:xih (6)
and
of these three binary systems we estimate yl" and 7; a a b b
from Pierotti's correlation, with results shown in Table 72x2 = 72x2 (7)
I. Using the recommended values of a I 2 ,we then obtain,
for each system, S and D and finally (g12- g2?) and and the stoichiometric relations
- gil), as shown in Table I. With these estimated x; x; = 1 +
NRTL parameters, y-x and P-x diagrams were calculated
over the entire composition range (Figures 6, 7, and 8). and
Also shown, for comparison, are experimental vapor com- x? + x; = 1

416 l & E C PROCESS D E S I G N A N D DEVELOPMENT

 I5

- I3
D N

-
-- ;- 0
D
i
' 05

00 6 I O I5 20 25 30 4 5 6 8 IO I2
- L o * , o l x g x;1

Figure 9. Parameters in NRTL equation from mutual solubilities for a12 = 0.2

15

IO
I

-N
-..
0 -
I

0
rn
i
1 05

0
06 IO I5 20 25 30 4 5 6 8 IO I2
-LoPlo(xp x;)

Figure 10. Parameters in NRTL equation from mutual solubilities for 0112 = 0.3

I3

-
DN
10
x
\-
0

-9
01

05

00 6 IO 15 20 25 30 4 5 6 8 10 I2

Figure 11. Parameters in N R T L equation from mutual solubilities for aiz = 0.4

T o illustrate the use of mutual solubility data we con- Calculated and experimental vapor-liquid equilibria for
sider the system n-hexane-nitroethane. Using mutual solu- this system a t 45OC. are shown in Figure 12. The r.m.s.
bility data a t 5" and 25°C. reported by Hwa, Techo, (absolute) deviation in vapor-phase mole fraction is 0.002
and Ziegler (1963) we obtain the parameters shown in and the r.m.s. (relative) deviation in total pressure is
Table 11. Assuming linear temperature dependence, we 0.006. Experimental results are taken from Edwards
obtain a t 45" C.: (1962).
(g12 - gzz) = 474 cal./mole Ternary Systems
and
For an additional illustration of Figures 9 t o 11 we
(gZ1- gll) = 680 cal./mole consider the ternary system n-octane-isooctane-

VOL. 8 N O . 3 JULY 1 9 6 9 417

 <-OCTANE

Table II. NRTL Parameters from Mutual Solubilities

g
12 - g21 -
A
g22, gll, - CPLCULPTED
Temp., Solubilities Cal./ Ca1.l
EXPERIMENTPL
WEIGHT F R 4 C T l O N S
System O C . a12 rl(in 2) rz(in I) Mole Mole
n-Hexane(1)- 5 0.2 0.100 0.095 930 880
nitroethane(2) 25 0.2 0.260 0.275 702 780
n-Octane(1)- 25 0.2 0.120 0.191 604 1124
nitroethane(2) 35 0.2 0.187 0.297 478 1115
Isooctane(1)- 5 0.2 0.101 0.128 759 980
nitroethane(2) 25 0.2 0.206 0.305 497 1026

Figure 13. Liquid-liquid equilibria in system n-octane-

isooctane-nitroethane at 25" C.

E
i
W

200
Table 111. NRTL Parameters for Three Binary Systems at 7OOC.

-I
+
-1
4 i
100 2 Ethyl acetate Water 0.4 1335 2510
Water Ethanol 0.3 976 88
Ethanol Ethyl acetate 0.3 301 322

0
0 025 0 50 0 75 IO
'HEXANE
ETMANOL

Figure 12. Vapor-liquid equilibria for n-hexane- A

nitroethane at 45" C.
i /- ClLCULATED
EXPERIMENTAL
M O L E FR4CT10NS

nitroethane. Mutual solubility data for each binary

octane-nitroethane system are given in Table I1 a t two
temperatures. From these (using a l p = 0.2) we obtain
the NRTL parameters shown. We then proceed to cal-
culate the ternary liquid-liquid equilibria a t 25" C. using
the multicomponent form of the N R T L equation (Renon
and Prausnitz, 1968) and the assumption that isooctane
and n-octane form an ideal mixture. The results of this
calculation are shown in Figure 13,along with experimental E T H Y L ACETATE WATER
data of Techo (1962).Further, assuming a linear relation
between (9, - gJ and temperature for the system
isooctane-nitroethane, we calculated vapor-liquid equilibria Figure 14. Liquid-liquid equilibria in system ethyl acetate-
for the ternary system a t 35°C. When the calculated ethanol-water at 70"C.
vapor compositions are compared with experimental
values, the r.m.s. absolute deviation is 0.019 mole fraction
and the r.m.s. relative deviation in total pressure is 0.033.
For a final example, we consider the ternary system ments of Griswold et al. (1949).Predicted ternary proper-
ethyl acetate(l)-ethanol(2)-water(3) a t 70" C. The binary ties are in satisfactory agreement with experimental
parameters used in our calculations are given in Table results.
111; they were obtained as follows: Mutual solubility data
Conclusions
a t 70°C. for ethyl acetate-water are given by Griswold
et al. (1949);activity coefficients a t infinite dilution for The NRTL equation provides a useful description of
water-ethanol were estimated using Pierotti's correlation the excess Gibbs energies of strongly nonideal mixtures
(1959);and parameters for ethanol-ethyl acetate a t 70" C. that may or may not be completely miscible. Parameters
were estimated from limiting activity coefficient data of in the equation may be estimated from limiting activity
Murti and Van Winkle (1958) a t 40" to 60°C. Figure coefficients or from mutual solubilities; charts to facilitate
14 compares calculated results with the ternary measure- such estimates are presented. While best results are cal-

418 I & E C PROCESS D E S I G N A N D D E V E L O P M E N T

culated when the nonrandomness parameter, a12,is deter-
mined from experimental data, good results may often
be obtained by setting C Y ~ Z equal to a fixed value suggested
by the molecular nature of the components.
Acknowledgment
The authors are grateful to the Computer Center, Uni-
versity of California, Berkeley, for the use of its facilities.
literature Cited
Edwards, J. B., dissertation, Georgia Institute of Tech-
nology, 1962.
Griswold, J., Chu, P. L., Winsauer, W. O., Ind. Erg.
Chem. 41, 2352 (1949).
Hanson, D. O., Van Winkle, M., J . Chem. Eng. Data
12, 319 (1967).
Hwa, S. C. P., Techo, R., Ziegler, W. T., J . Chem. Eng.
Data 8,409 (1963).
Kudryavtseva, L. S., Susarev, M. P., Zh. Prikl. Khim.
36, 1471 (1963).
HIGH TEMPERATURE PYROLYSIS OF ISOBUTENE
K A R L S C H U G E R L ’A N D
Chemical Engineering Department, New York University, Bronx, N . Y .
The product distribution of the high temperature pyrolysis of isobutene was
experimentally determined. The decomposition can be represented as a system
of simultaneous and/or consecutive reactions.
METHYLACETYLENE
and/or allene can be obtained by
pyrolysis of isobutene (2-methylpropene) or propene.
Rice et a / . (Rice, 1948; Rice and Wall, 1950; Rice et
al., 1948) and Szwarc (1949) were among the first to
report the synthesis of methylacetylene and/or allene by
the pyrolysis of isobutene. Szwarc (1949) and later Sakak-
ibara et al. (Kunichika et al., 1965; Sakakibara, 1964)
and Amano and Masao (1964) investigated the pyrolysis
of propene. A number of patents (Happel and Marsel,
1956, 1962; Hogsed, 1959, 1960; Keith et al., 1961; Rice,
1948) deal with the industrial application of these reac-
tions. Radical sources can be used to increase the yield
of methylacetylene and/or allene (Happel and Marsel,
1965; Peer and von Lohiuzen, 1958).
The pyrolysis of isobutene has not been reported a t
temperatures above 1000”C. and very short reaction times.
I n this study, isobutene, as well as allene, methylacet-
ylene, and propene, which are intermediate decomposition
products of isobutene, were pyrolyzed by admixture with
hi,gh temperature steam.
Experimental
High temperature steam was generated by means of
a hydrogen-oxygen torch and diluted with low pressure
secondary steam. The olefin to be cracked was introduced
I Present address, Institut fur Chemische Technologie, T.U., 33
Braunschweig, Hans-Sommer-Str. 10, Germany.
Murti, P. S., Van Winkle, M., Ind. Eng. Chem., Chem.
Eng. Data Ser. 3, 72 (1958).
Orye, R. V., Prausnitz, J. M., Ind. Eng. Chem. 57, No.
5, 18 (1965).
Pierotti, G. J., Deal, C. A., Derr, E. L., Ind. Eng. Chem.
51, No. 1, 95 (1959).
Prausnitz, J. M., Eckert, C. A., Orye, R. V., O’Connell,
J. P., “Computer Calculations for Multicomponent
Vapor-Liquid Equilibria,” Prentice-Hall, Englewood
Cliffs, N. J., 1967.
Renon, H., Prausnitz, J. M., A.I.Ch.E. J . 14, 135 (1968).
Scott, R . L., J . Chem. Phys. 25, 193 (1956).
Techo, R., dissertation, Georgia Institute of Technology,
1962.
Wilson, G. M., J . A m . Chem. SOC.86, 127 (1964).
Wohl, K., Trans A.1.Ch.E. 42, 215 (1946).
RECEIVED
for review November 11, 1968
ACCEPTED March 21, 1969
Work supported by the National Science Foundation.
JOHN H A P P E L
from above the reaction section. The reaction mixture
was quenched rapidly by means of water sprays a t the
bottom of the reactor. The pyrolysis equipment (Figure
1) consisted of a hydrogen-oxygen torch, a, a water-cooled
zirconia-lined high temperature steam generator, b, a
5-p.s.i.g. steam inlet section, c , an olefin inlet section,
d , a reactor, e, with the quenching section, f , a knock-
out tank, g , packed with steel wool, liquid gas separators,
h, a gas meter, i, and a sampling device, j .
The steam superheater, k , olefin preheater, 1, and steam
line trace were warmed prior to use. The reactor was
flushed with nitrogen, and the spark plug igniter, m , dried.
Water to the quenching sprays, cooling jackets, and water
seals was turned on. The igniter was started and first
the oxygen and then the hydrogen check valves were
opened. When the flame was stabilized, the secondary
steam was turned on and the amount of hydrogen-oxygen
mixture as well as the ratio of hydrogen to oxygen was
gradually increased. A slight excess of hydrogen was used
to avoid burning the olefin. The desired temperature was
maintained by a proper ratio of primary and secondary
steam.
When the olefin flow rate was established for the run,
the nitrogen flow was stopped. The temperature i~the
reactor was measured a t three positions by P t - P t / I r
thermocouples, p . Flow rates were determined by
rotameters, q. Volume of effluent gas was measured by
a gasmeter, i. A bypass protected meter i during starts
VOL. 8 N O . 3 J U L Y 1 9 6 9 419