J. Phys. Chem.

1982, 86, 393-400 393

rank tensorB and would therefore be the same. The nu-
merator depends upon the structure and dimension of the
aggregates and would be proportional to the square of the
dimension of the cubic unit cell.
A useful model, for which simplicity is the chief justi-
fication, for the translational motion in these isotropic
liquid crystalline phases is that it occurs on a sphere in-
scribed in the cubic unit cell. In this model 2R, as given
by eq 11,determines the dimension of the cubic unit cell.
As described above, for a cubic phase consisting of rodlike
aggregates, D L = 3 D , , d , and, for lamellar aggregates,
DL = 3/@meas-+ Thus,the method gives different values
of R depending on the structural model. Through a com-
parison with X-ray diffraction data on the dimension of
the cubic unit cell, it is possible to discriminate between
the two models.
In Table TV calculated values of R are given for the cubic
phase at 84% C12TACand the cubic phase in the C,,TAF
system assuming rod- or lamellar-shaped aggregates. As
can be seen from Table IV, calculated values of R are in
better agreement with X-ray investigations assuming a
rodlike aggregate structure. This can be taken as further
support for the assumption of ascribing T,S to the trans-
lational diffusion.
A better description of the relaxation in terms of lateral
diffusion from these continuous cubic phases should be
given by a random-walk model of the diffusion on a surface
equivalent to the proposed model for the aggregate
structure. Structural information would thereby be ob-
tained by-fitting the dimenjion of the cubic unit cell to the
value of J,(O)+ J,(oo)- 2Js(2w0)in eq 5. Computer sim-
ulations along these lines are currently being carried out.
Acknowledgment. Thanks are due to Bertil Halle and
to Hilkan Wennerstrom for valuable discussions. This
work was supported by the Swedish Natural Science Re-
search Council and by the Foundation of “Bengt
Lundqvists Minne” (P.O.E.).

Patterns of Three-Liquid-Phase Behavior Illustrated by

Aicohol-Hydrocarbon-Water-Satt Mixtures

B. M. Knlckerbocker,t C. V. Perheck,’ H. T. Davls, and L. E. Scrlven

Dep8-t of Chmkal E n ~ h e r l n gend htatcwlele Sclence, University of Minnesota, Mlnneapoik, Minnesota 55455
( R e c e W : June 18, 1981; I n Final F m : October 5, 1981)

Ten salts were each dissolved in water and the solutions mixed with equal volumes of one of six hydrocarbons
and one of ten monohydric alcohols. The resulting multiphase mixtures were examined for the number of
coexisting liquid phases and, in wme cases, for the partitioning of alcohol among them. Several unusual patterns
of phase behavior have been observed. For example, increasing concentrations of sodium chloride induce the
widely observed 232 pattern of phase behavior in equal-volumemixtures of n-propyl alcohol, octane, and brine.’
However, with lithium chloride, calcium chloride, or magnesium chloride, the induced pattern is 2323. All of
the patterns observed in the alcohol-hydrocarbon-brine systems chosen for this study can be summarized by
two basic quaternary phase diagrams. It is conjectured that any three- or four-component system exhibiting
three liquid phases in equilibrium can be described in terms of these two basic phase diagrams.

Introduction wateraodium chloride mixtures mimic the phase behavior

The phase behavior of alcohol-hydrocarbon-brine
mixtures is made rich and varied by hydrogen bonding and
other effects of polar intermolecular forces in water and
salt solutions, nonpolar forces in hydrocarbons, and both
types in alcohols, which are amphiphilic. The dynamic
structures caused by hydrogen bonding and other polar
interactions make these mixtures difficult to model ther-
modynamically with empirical equations of solution state,
much less in terms of statistics of molecular mechanics.
A useful step toward understanding the thermodynamics
of such mixtures is to profile their phase behavior exper-
imentally, as we did for certain systems with sodium
ch1oride.l In this paper we take up other salts and the
basic features of all of the phase diagrams.
As amphiphiles, the lower molecular weight monohydric
alcohols employed here can be considered to be proto-
surfactants even though they form neither micelles nor
liquid crystals in aqueous solutions and so display neither
property often taken to distinguish surfactants.2 Previ-
ously we reported’ that numerous alcohol-hydrocarbon-
‘Mobil Research and Development Corp. Paulsboro, NJ 08066.
of surfactant-oil-watersalt systems, including those en-
countered in certain surfactant-based processes for en-
hancing petroleum recovery. At low salinities, certain
mixtures of alcohol, hydrocarbon, and brine split into two
phases with most of the alcohol in the lower water-rich
phase. At intermediate salinities the split in many in-
stances is into three phases, the alcohol residing chiefly
in the middle phase, which also contains substantial
fractions of water and hydrocarbon. At higher salinities
the split is into two phases with most of the alcohol in the
upper, hydrocarbon-rich phase. This so-called 232 pattern
is typical of surfactant-oil-water-salt systems, but our
results showed that the pattern is not restricted to am-
phiphiles which are highly surface active.
In what follows we show that, whereas systems with
potassium chloride, sodium sulfate, and sodium bromide
also follow the 232 pattern, those with lithium chloride,
magnesium chloride, and calcium chloride behave some-
what differently. Notwithstanding the differences, all
(1)B. M.Knickerbocker, C. V. Pesheck, L. E. Scriven, and H. T.
Davis, J. Phys. Chem., 83, 1984 (1979).
(2) R. G. Laughlin, Adu. Lip. Cryst., 3,99 (1978).

0 1982 American Chemical Society

 KF
NAA
T44
IBA
? ? ?
SB 4
3 2 2
- -
YBA
_ -
TB4 2- 2 2
~-
“UP4 ?- 2 2
2 3 2
~.
IP4
-2 _2 _2
E4 _2 3 3
11 4 1 2 1
i
known patterns of liquid phase behavior can be generated
from just two basic types of three-dimensional phase di-
agrams in which three coexisting fluid phases occur.
The monohydric alcohols are ten: methyl alcohol (MA),
ethyl alcohol (EA), n-propyl alcohol (NPA), isopropyl
alcohol (IPA), n-butyl alcohol (NBA), sec-butyl alcohol
(SBA), isobutyl alcohol (IBA), tert-butyl alcohol (TBA),
n-amyl alcohol (NAA),and tert-amyl alcohol (TAA). The
hydrocarbons are the six even-numbered n-alkanes having
from 6 to 16 carbon atoms in the molecule.
Experimental Procedures
Procedures were as described earlier,’ the hydrocarbons
and alcohols were from the same sources as before.
Lithium chloride, magnesium chloride, barium chloride,
potassium bromide, lithium sulfate, and calcium chloride
were from Baker Chemicals; sodium chloride, calcium
fluoride, and lithium bromide were from Fisher Scientific;
magnesium sulfate, sodium bromide, and potassium
chloride were from Allied Chemical; sodium iodide and
potassium sulfate were from Mallinckrodt; and sodium
sulfate was from Merck. All chemicals were of 99% or
higher purity except for their water content, and were of
ACS or USP grade.
Salt solutions were prepared by diluting saturated (at
25 OC) solutions of the salts with doubly distilled, deionized
water. Lithium chloride solution concentrations were
checked by using the Fiske Model 404 chlorocounter and
by using the published saturated value3 as a standard.
Lithium chloride solutions of 1 , 2 , 3 , ...,and 14 M and 1.4,
1.5, 1.7, and 2.3 M were mixed. The lithium chloride-
octane-n-propyl alcohol system was analyzed by gas
chromatography and chlorocounter as described earlier.’
Experimental Results
Phase behavior can be classified according to the number
of liquid phases that form with increasing salinity up to
the concentration at which solid salt precipitates. Our
notation’ is a sequence of numbers of liquid phases, an
underbar indicating that the amphiphile resides primarily
in the lower, water-rich phase an an overbar that it resides
mainly in the upper, hydrocarbon-rich phase. An underbar
(3) H. Stephen and T. Stephen, Ed., ‘Solubilities of Inorganic and
Organic Compounds”, Macmillan, New York, 1963.
T44
.*\A
IRA -~
- 2 2 2 _ -
3 2 2
SB4
hB4
- - _ _ ~ _ _ ~
TBA 2 2 2 2 1 2
- 1 3 2
_ _
‘IPA 2 2 2
~ ~
2
-2 3 2
IPI 2 2 1 2 3 3
_ _
E4
\I4 2_2 2_ .
I I I I I
and an overbar together signify nearly equal partitioning.
The fmt number signifies the zero-salt phase behavior, the
last number signifies the phases present when solid salt
precipitates, and the middle number(s) respresent(s) the
number of phases present a t any intermediate salinities.
Phase boundaries are crossed only when three phases ap-
pear or disappear. Seven different phase behavior patterns
were seen in mixtures of equal volumes of alcohol, hy-
drocarbon, and NaCl brine:’ 222,233,232,222,222,322
and 222.
Table I shows the patterns of equal-volume alcohol-
hydrocarbon-KF brine and alcohol-hydrocarbon-NazS04
brine systems. These two salts induce sodium chloride like
patterns, though a particular alcohol-hydrocarbon com-
bination may display a different pattern with KF or
Na2S04than with NaC1. For example, the EA-hexane-
NaCl brine system gives 222 behavior, whereas the EA-
hexane-KF brine system gives 232 behavior.
Some salts other than sodium chloride, e.g., lithium
chloride, magnesium chloride, and calcium chloride, induce
patterns of phase behavior not observed for any sodium
chloride mixture. Table I1 shows the patterns for the
equal-volume alcohol-hydrocarbon-LiCl brine mixtures
and the salinity ranges (in wt %) over which three liquid
phases occur. The patterns in Table I1 that do not occur
in NaCl-brine systems’ are as follows:
(a ) 2323
-- Mixtures (Figure l a ) . This pattern resembles
the 232 pattern except at high salinities. At low salinities
these systems are &phase mixtures. Over an intermediate
salinity range an alcohol-rich phase is formed, and at
somewhat higher salinities the alcohol resides primarily
in the hydrocarbon-rich upper phase of a 2-phase mixture.
However, at very high salinities an alcohol-rich phase again
appears as one of three liquid phases. NPA shows this
pattern in equal-volume mixtures of octane and LiC1,
CaClZ,MgClZ,LiBr, and MgBrz brines.
( b )g23 Mixtures (Figure l b ) . Similarly, this pattern
resembles the 222 pattern except at high salinities. At low
salinities these systems are - p h a s e mixtures. Over a
narrow intermediate range of increasing salinity, there is
a gradual shift of alcohol from the lower to the upper phase
without the appearance of a third liquid phase at the
equal-volume mixing point. At very high salinities the
upper phase splits into alcohol-rich and hydrocarbon-rich
Patterns of Three-LlquM-Phase Behavlor
TABLE 11: Phase Behavior in Mixtures of Equal Volumes
of Alcohol, Hydrocarbon, and LiCl Brineu
TAAl
NAk
_ _
IPA
39-60 -
2 ?2 3
45-60
50.60 ~
I
50.60
222
I
‘6 ’ c8 ‘10 ‘12
The bold-face labels indicate the patterns of phase be-
havior. Numerical ranges are the g of LiC1/100 mL of
brine at which the mixture splits into three liquid phases.
-SALINITY - __ S A L I N I T Y
-SALINITY- -SALINITY-+
- - - -
2 2 3 3 2 3
(C ) (d)
Figure 1. Patterns of phase behavior observed in mixtures of equal
volumes of alcohol, hydrocarbon, and UCI brlne. Stippling qualitively
Indicates the cmcentrationof alcohd In each phase. Numercai labels
indicate the number of liquid phases: bars lndicate the phase(s) with
a hlgh concentratlon of alcohol: upper (2), lower (2), or both (2).
phases to give a three-phase mixture. IPA shows this
pattern with hexane and octane and LiCl brine.
(c) 223 Mixtures (Figure I C ) . This pattern is the same
as the 2 2 2 3 except that the alcohol is nearly equally
partitioned in the absence of salt. NPA shows this pattern
with hexane and LiC1, CaC12, MgC12, LiBr, and MgBr,
brines.
( d ) 323 Mixtures (Figure I d ) . This pattern resembles
the 322 pattern except at high salinities, where the upper
phase splits into alcohol-rich and hydrocarbon-rich phases.
NBA, SBA, and IBA show this pattern with hexadecane
and LiCl, CaC12, MgC12, LiBr, and MgBrz brines.
( e ) 223 Mirtures (Figure l e ) . In these mixtures the
alcohol partitions strongly into the hydrocarbon phase
except at high salinities, where the upper phase splits into
alcohol-rich and hydrocarbon-rich phases. With LiC1,
CaClZ,MgC12,LiBr, and MgBr2brines, NBA and IBA show
this pattern with dodecane and tetradecane, SBA does so
with tetradecane, and NAA does so with hexadecane.
(f) Mixtures Containing TBA and Very High Concen-
trations (above 55 wt 9%) of LiC1. All such equal-volume
The Journal of Physical Chemistry, Vol. 86, No. 3, 1982 395
TBA-hydrocarbon-LiC1 brine mixtures contain a solid,
effectively frozen phase rich in TBA, indicated by the letter
F in Table 11. In each case, this is the third phase, which
ri 57.60
makes the patterns analogous to the patterns a-c and e
above except that the alcohol phase is’present as a solid
rather than a liquid.
The NPA-octane-LiC1 brine system was studied in
detail. Phase compositions for the salinity range where
three liquid phases coexist in equal-volume mixtures are
plotted in Figure 2. The abscissas on each plot represent
the salinity, in molar units, of the brine used in mixing.
(The numerical data are tabulated in the supplementary
material associated with this paper. See paragraph at end
of text regarding supplementary material.) There are two
salinity ranges of three-phase behavior for equal-volume
mixtures, one from approximately 1 to 2.25 M LiCl, the
other from 5.8 M to saturated LiC1. As overall salinity
increases in the first three-phase region, the alcohol moves
53.60 I 53-60
out of the lower phase into the middle and then the upper
phase, the water becomes less soluble in the middle phase
and moves to the lower phase, and more octane moves to
’ ‘I4 ’ ‘16 the middle phase. The LiCl concentration increases in the
lower phase, whereas it decreases in the middle phase. In
the second three-phase region, as overall salinity climbs,
the alcohol moves from the upper phase into a middle
- phase, the octane concentration in the middle phase de-
creases as the octane shifts into the upper phase, and the
water concentration in the middle phase increases slightly.
Meanwhile, the alcohol concentration in the lower phase
falls, and the salt concentration in the middle phase rises
dramatically.
A set of 36 samples of NPA-octane-4.5 M LiCl were
mixed, one at each interior intersection of lines of 10 vol
% (80% brine, 10% NPA, 10% octane; 70% brine, 20%
NPA, 10% octane; ...; 10% brine, 10% NPA, 80% octane).
-SALINITY +
These mix points all lie on one plane in the tetrahedral
phase diagram between the two regions that had been
shown to contain three coexisting liquid phases. If those
two three-phase regions were connected, there would have
-
to be a three-phase region in this plane. All 36 of these
2 2 3 samples contained only two phases, as far as we could tell
(el by eye. Hence, we concluded that the two three-phase
regions are probably unconnected, although a possibility
remains that they are connected by a small three-phase
region that passes between the mix points that we chose.
Interpretation of Results
Figure 3 shows the three-liquid-phase region of the
TBA-tetradecane-water-NaC1 system at 25 “C in the
triangular prismatic coordinates that we used previously.’
This system shows the 232 pattern in mixtures of equal
volumes of alcohol, hydrocarbon, and brine. The three-
phase region, of course, consists of a sequence of tie tri-
angles, of which just one (LMU) is drawn. The arrows at
the points L, M, and U show the directions of change of
the respective phase compositions caused by adding salt.
Similar three-phase regions for other quaternary systems
are described by Widom: Lang and Widom: Griffiths?
and Hartwig et al.7
The line BD in Figure 3 is a tie line between stable
phases of compositions B and D in equilibrium. It is a
critical tie line, for the water-rich phase and the alchol-rich
phme reach identical compositions at the critical end point
(4)B.Widom, J. Phys. Chem., 77,2196 (1973).
(5)J. C.Lang, Jr., and B. Widom, Physica A (Amsterdam), 81, 190
(1975).
(6)R.B. Griffiths, J. Chem. Phys., 60,195 (1974).
(7)G. M.Hartwig, G. C. Hood, and R. L. Maycock, J.Phys. Chem.,
59, 52 (1955).
 Knickerbocker et ai.

50 601 11 i
s t , ..,nn, c

LiCl CONCENTRATION. Molar L i C l CONCENTRATION, Molar

100 , , , , , , , , , , I , , I I
0

MIDDLE

P
40
0
>
20
LOWER

'
0 2 4 6 S 10 12 14 16
LiCl CONCENTRATION, Molor
LiCl CONCENTRATION, Molar

Figure 2. phase composltions for mixtures of equal volumes of n-propyl alcohol. octane, and LiCl brine at 25 "C. The culves have be%n mted
to the data by visual inspection.

Alcohol

Fbus 3. Thdlquibphase region of the TEA-tet'adecane-watw- Figure 4. Schematic representation of a 311 phase diagram. The
NaCX system at 25 "C in tiangular pdsmatk madhates. Cuve ABCD zerwallnHy. three-phase region is thB triangle abd. AC i s h upper
is the Iocu8 of all three-phase equilibrium composltlons: AC and ED crlthxi tie line. Curves Aa and bcd represent all possible aWee-phase
are the two crlticai tie lines. equillbrlum compositions.

B. S i m i i l y AC is a critical tie line; the alcohol-rich phase
and the hydrocarbon-rich phase reach identical compo-
sitions at the critical end point C. The locations of the
critical tie lines and critical end points are extrapolated
from compositional data.' The curve BA is the locus of
compositions of the water-rich phase L of the continuum
of tie triangles. The curve BC is the locus of compositions
of the alcohol-rich phases M, and the curve DC is the locus
of compositions of the hydrocarhon-rich phase U. Altog-
ether the space curve ABCD is the locus of all of the
equilibrium phase compositions in the three-phase region.
Because each point on line ABCD represents a vertex of
a tie triangle and thus a stable phase, there must he ad-
jacent to ABCD a one-phase region over its entire length.
According to Gibbs' phase rule, the three-phase region
must he enveloped by a region of twc-phase equilibria (not
shown), except along the curve ABCD. This two-phase
region is simply connected it is possible to proceed by a
suitable path of component additions from one two-phase
point to any other without leaving the region. This two-
phase region reaches to the salt-free base; adding salt
causes the alcohol amphiphile to partition from the
aqueous into the hydrocarbon phase and to do so via the
middle phase when it is present (i.e., the 232 pattern).
Owing to the limited solubility of NaCl in this system,
there must he a three-phase region of solid/liquid/liquid
equilibria at high salinities, hut it is not shown.
The three-phase region may he cut off by the salt-free
base of the quaternary diagram to leave the 322 pattern,
as in the NBA-hexadecanewatepNaC1 system. This type
of diagram is illustrated in Figure 4. Its hallmarks are
the tie triangle ahd in the base and the loci of tie-triangle
Patterns of ThresLiquWPhase Behavior
A I c i hol
ngun 5. Sdwmatic representah of a 233 phase diagam. The tie
Mangle atd is ths lower face of a four-phase tetrahedron wim salt as
ths fm-th PhaJe (off scale to me top of ths diagram). BD is ths lower
cinkal tie Ihe. C w e s am ani d)repesent al ttrswhase equRkdun
composnions.
vertices: aA for the water-rich phase, bC for the alco-
hol-rich phase, and dC for the hydrocarbon-rich one. The
arrows at L, M, and U again show directions of change
caused by adding salt. C is again a critical end point and
AC a critical tie line. However, because there is no lower
critical tie line-it has in effect disappeared into the
physically unrealizable region of negative salinities-the
locus of all three-phase equilibrium compositions is no
longer a single m e . The two branches, aA and bCd, must
be contacted by one-phase regions that plow inward to
them through the two-phase regions of liquid/liquid
equilibria that otherwise envelop the three-phase region
above the base-two unconnected two-phase regions.
The three-liquid-phaseregion of Figure 3 may be trun-
cated at higher salinities by equilibria with a coexisting
solid, salt-rich phase, Le., by a four-phase solid/liquid/
liquid/liquid tie tetrahedron, to leave the 233 pattern, as
in the NPA-hexadecanewater-NaCl system. This type
of diagram is shown in Figure 5. The tie triangle abd is
the lower face of the four-phase tetrahedron; far off scale
to the top is the vertex representing the salbrich solid. B
is a lower (with respect to salinity) critical end point and
BD a lower critical tie line; in this case it is the upper
critical tie line that is missmgswallowed, in effect, by the
tie tetrahedron. The locus of all three-liquid-phase
equilibrium compositions is again divided, and each of the
branches, aBb and dD, must be contacted by one-phase
regions. Elsewhere, the threeliquid-phase region must be
enveloped by two-liquid-phase regions, except for tie tri-
angle abd, which faces into the tie tetrahedron. Inciden-
tally, the other faces of the tie tetrahedron must join
solid/liquid/liquid three-phase regions, and all six of its
edges must make contact with two-phase, solid/liquid or
liquid/liquid regions, none of which are shown.
The three-liquid-phaseregion of Figure 5 can be further
truncated by the salt-free base to give a 333 pattern, with
three liquid phases occurring at all salinities. This pattern
occurs in mixtures of water with NaCl, benzyl alcohol, and
n-alkanes.' It will also occur in other mixtures of salt and
water with two immiscible liquida that are both immiscible
with water.
The phase diagram of the NPA-octane-water-Licl
system is more complicated than in Figure 3.4, or 5. Its
pattern 2323 shows two three-phase regions. At very high
salinities, a LiC1-rich solid coexists with the three liquid
phases in a four-phase equilibrium like that shown in
Figure 5. The lower three-phase region has phase-com-
position changes paralleling those found by Lind et al.8 in
The Journal of PhVslcal chemistry. Vd. 66. No. 3, 1982 397
npure 6. Idealized twwphase region in a threecomponent phase
diagram wim an intensive varhble as ths vertical dkecfion. The upper
c r b i sduhn point (UCSP)ani ths lower ddcal solumn point (LCSP)
are connected by a pian-point Iwp (Ppl).
a system possessing two critical end points and suggests
a three-phase region similar to that shown in Figure 3, with
critical tie lines at approximately 1and 3 M LiC1. While
a possibility remains that these two three-phase regions
are connected, as explained above, the available evidence
indicates a phase diagram that is qualitatively a combi-
nation of Figure 3 with a skewed version of Figure 5. In
the combination are three critical tie lines, two associated
with the lower three-phase region and one at the lower edge
of the upper three-phase region.
Generalization: Three-Liquid-Phase Behavior
Patterns
To understand all of the possible patterns of phase
behavior in quaternary systems that can form three
coexisting liquid phases, it is necessary to consider the most
general cases of three-phase regions in a four-component
system. As a prelude it is very instructive to study the
most general cases of three-noncrystalline-phaseregions
in a three-component system as a function of a field var-
iable such as temperature or chemical potential or of a
molecular parameter such as carbon number. Field var-
iables have been introducedg to describe those thermo-
dynamic quantities which are the same in all phases at
equilibrium since many textbooks label as intensive var-
iables these quantities plus specific quantities such as
density, mole fraction, and the like. The triangular pris-
matic coordinate system of Figure 6 is appropriate, with
salinity replaced by a field variable as the vertical axis.
Because the field variable must be the same in all coex-
isting bulk phases at equilibrium, tie lines and tie triangles
must now lie in horizontal planes of the stacked ternary-
diagram version in Figure 6. In contrast, the concentration
of a fourth component such as salt varies from phase to
phase. Aa it does, the tie lines and tie triangles are tipped
askew as in the isothermal quaternary diagrams, Figures
3-5. Nevertheless, the topological properties of the two
versions are usefully equivalent, as brought out below.
A three-phase region must be enveloped in a region of
two-phase equilibria except along the loci of phase com-
positions. Moreover, the coexistence of noncrystalline, or
(8) C. V. Pesheck. M. I. Lind, L. E. Serivsn, and H. T. Davis. manu-
script to be submitted to J. Phya. Chem.
(9) R. B. Griftiths and J. C. Wheeler, Phys. Re". A, 1041 (1970).
990 The JownaI of Physical Chemlsby, Vol. 88, No. 3. 1982
Flgwa 7. schematic rqwesentatlon oi a -phase regbn 88811 in
ternary system.
disordered, phases can be thought of as terminating, in
principle a t least, in critical points. Therefore, we start
with the nonisothermal case of two-phase equilibrium in
a ternary system pictured in Figure 6. This representation
goes back to the pioneering work of Sehreinemakers.'O
The two-phase region is a free-standingsolid ovoid, a more
or less deformed ellipsoid, composed of a stack of hori-
zontal disks, each of which is a planar ruled surface of tie
lines. On the surface of each disk are two tie lines of zero
length corresponding to ordinary critical points, or plait
points, and these on successive disks define a closed space
curve, the plait-point loop (PPL), which at its highest and
lowest points defines the upper critical solution point
(UCSP) and the lower critical solution point (LCSP), re-
spectively. From this archetypal diagram a whole catalog
of phase diagrams with two-phase regions can be generated
by (a) deformations of the ovoid that leave tie lines as
rulings and disks horizontal and convex with respect to tie
lines and (b) horizontal and vertical translations of the
resulting body. (such deformations and translations do
not tear or puncture the surface and so define a class of
topological transformations that preserve the essential
thermodynamic features.) Only those parts of the ovoid
that remain within the triangular prism are physically
significant, however. Figure 7 is an example of a section
of an ovoid truncated on two vertical sides by the com-
positional edges of the prism. Other examples in a tet-
rahedral coordinate system have been reported."
The catalog can be extended to diagrams with more than
one two-phase region by starting with multiple ovoids but
keeping them out of contact. The complications brought
by the existence of ordered, crystalline phases are unim-
portant for present purposes, except that such a phase may
truncate an ovoid in a fan of solid/liquid/liquid three-
phase tie triangles between a pair of critical tie lines, each
running between the crystalline state and a plait point.
There are only a few ways that a three-phase region can
occur in a free-standing two-phase ovoid. Figure 8 illus-
trates one way, which we call achiral. At a lower critical
end point B, the @ phase bifurcates to and phases,
and, at an upper critical end point C, these fuse again into
the 0 phaae. B and C are linked by a plait-point segment
PPS that is unconnected with the plait-point loop PPL
that belts the altered ovoid. Because this loop closes on
the ovoid, it divides the hinodal surface into two distinct
(10) F.A. SEhrsinSmaLara,"Die Hetercgeaen GleichgewickW, VoL 3,
part 2, H. W. B. Rooseboom, Ed., Viswsg, Braunschweig, 1913.
(11) K. P. Myasnikova, N. I. Nikurashina, and R.V. Mertslin, Ruas.
J. Phyr. Chom. (Engl. Trannl.), 43, 223 (1969).
Knickerbocker et al.
a
Rgun 8. Achiral eruptim of a threephase regbn In a two-phase
regbn. AC and BD are critical tie lines. The two phases @, and &
are kwdved h both dtlcald pdntsC and D. The~nguiardiaqams
show ths sequence of phase behavior in the ovoid.
N ICTT
i NE
*re 9. Schematic representation of ths water-nioctlnemercuy
phase diagram. This phase diagram is a truncatbn of the type of
diagram shown in Figure 8.
regions. The situation is analogous to an island micibility
gap in a binary system. What is significant is that the a
phase is not involved in either critical end point, although
it is connected to each by a critical tie line, BD and AC,
respectively. Each of these amounts to a tie triangle of
infinitesimal width, just as the critical end points amount
to tie lines of infinitesimal length. Between the two lies
the three-phase region. The locus of three-phase equilib-
rium compitions is divided into segment AD and the loop
B@lC&,At each level (value of intensive variable or mo-
lecular parameter) of three-phase equilibria, there are three
plait points, each one corresponding to a different side of
the tie triangle-each side is necessarily bordered by a
two-phase region. There are two unconnected regions of
two phases, one connected with the plait-point loop PPL,
the other with the plait-point segment PPS. A truncated
version of this type of diagram would be the nicotine
water-mercury diagram sketched in Figure 9;the closed
miscibility gap in the nicotine-water binary system is
well-known.'*
By imagining the critical end points B and C in Figure
8 moving together (as pressure changes, or as molecular
structures are varied) and coalescing in a critical double
(12) W. J. Moore, "Physical Chemistry", 3rd ed.. Prentice-Hall, En-
glewwd Cliffs, NJ. 1962. p 114.
pattsms of Three-Liqu&F%aseBehavlor
Figure IO.
(a) (b)
Chlral eruption of a three-phase region In a twwphase
region: (a) Um levo fwm: (b) the dextro form. AC and BD are the
two dtlcal Ue Ilnes. The LII and 6 phases bewme ldentlcal at 8, and
Um 0 and y phases at C. The triangular diagrams give the seqtmnce
of phase behavior as the lntenslve variable Is Increased for (a).
point at a particular temperature, the reader can picture
the reverse of the proem by which this type of three-phase
region erupts from a tie line at one end of which there is
an incipient pair of plait points at infinitesimally different
values of temperatures or analogous variable. This is not
a tricritical point, since only two phases become identical.
A special case is to suppose that both B and C migrate to
and fuse into either the upper or lower critical solution
point, thereby shrinking the whole three-phase region to
iniiiteaimal size that can, in effect,fold up into the critical
solution point turned tricritical point. This, however,
requires two additional variable parameters, such as
pressure and salinity, as well as the three components and
temperature or analogous variable.
The alcohol-hydrocarbon-brine three-phase regions do
not resemble the type just described, but there is a second
way that a threephase region can occur in the ovoid in the
prismatic ternary diagram (see Figure 10). Here the third
phase that appears is comparatively rich in an amphiphilic
component, the relative affinity of which for each of the
other two components shifts regularly as temperature or
analogous field variable or parameter rises or falls. Again
the three-phase region lies between a lower critical tie line,
BD, B being the lower critical end point where a single
phase bifurcates into a and @ phases, and an upper critical
tie line, AC, C being the upper critical end point where the
@ and y phases fuse into a single phase. The locus of
tie-triangle vertices, i.e., three-phase equilibrium compo-
sitions, is a single curve ABCD. The plaibpoint curve PPC
is an open curve that conneds the two critical end points
but in doing so passes through the lower and upper critical
solution points (LCSP and UCSP) and swings around each
of the critical tie lines, in such a way as to provide three
plait points, one for each side of the tie triangle, at each
level of three-phase equilibrium. Because the locus of
three-phase compositions is not closed, the surface of the
ovoid has only one region (in contrast with Figure 8,whose
surface is divided into two regions), and the two-phase
region is thus simply connected.
Another consequence of the open curve, or equivalently
of the involvement of the @ phase with the y phase in the
lower critical end point and the a phase in the upper one,
is that the chiral ovoid of Figure 10 exists in two enan-
tiomorphic forms, levo (Figure loa) and dextro (Figure
lob). Hence, there me two poasible sequences of triangular
diagrams as temperature or analogous field variable or
The Journal of PhVJlcal Chemktry. Vol. 66, No. 3. 1982 399
UCSP
PPC
'pc A
A
LCSP A
mum 11. Schematlc representation of Um completely symmetrlc
regular solution model showing three trlcrklcal points (TCPs) and a
four-phase equlllbrlum plane. The trlcrklcal points arise from the
symmew 01 tha model. but the four-phase equlllbrlum plane remains
In Um phase diagram wRh a move away from tha symmetrical system.
molecular parameter is changed. The sequence of trian-
gular diagrams shown corresponds to Figure loa. While
the experimental results reported earlier in this paper a!J
show alcohol shifting from the aqueous phase to an alcc-
hol-rich phase and on to the hydrocarbon phase as salinity
increases, the progression with respect to some parametera
is the opposite in certain amphiphilehydrocarbon-brine
~ystems.'~
If one imagines the critical end points Band C in Figure
10 moving together, the critical tie lines AC and BD
shorten to zero length, and the whole three-phase region
with its progreeaion of tie triangles shrinks to infinitesimal
size,flattens, and folds up into the fusion of B and C, which
is a tricritical point. This type of tricritical point has been
carefully studied experimentally by Lang and Widoms and
compared to a Landau model developed by Griffiths!
Note that, when the tricritical point is reached, there is
no three-phase region in the freestanding egg. The reader
can imagine the converse unfolding by which the three-
phase region achieves detectable proportions in the free
standing ovoid.
There is a particular ternary model, the regular solution
model examined by Meijering," which has a three-phase
region that terminates in a trimitical point in the same
composition-temperature space. However, this tricritical
point appears only when the regular solution model is
symmetrical with respect to two of the binary interaction
parameters; furthermore, in this model, if the parameters
are equal at one temperature, they are equal at all tem-
peratures. Were the two parameters to vary differently
with temperature, the tricritical point would be isolated,
unconnected to a three-phase region. But were the two
parameters to depend on pressure as well, the tricritical
point would terminate a three-phase region in the four-
dimensional composition-temperaturepressure space.
Figure 11 shows the sequence of the phase diagrams
obtained from regular solution theory when all three in-
teraction parameters are exactly equal. This diagram
shows three ovoids, each containing a tricritical point and
(13) M. L. Robbina in 'Micellization, Solubilization. and
Micmemulsiom", K. L. Mittal, Ed., Plenum Press, New York, 1977, pp
713-44, R. N. H d y and R. L. Reed. Soe. Pet. Eng. J., 16, 147 (1976).
(14) J. L. Meijering, Philips Res. Rep., 6,183 (1951).

Flpn 12. Fusion of twolhree-pkasa owl&. Fl@re 10. a and b. wiW7
accounts fa2323 phase behavior. This figure has two mresphase
regbns. two plan-point segments. and four crnid tka lines.
a three-phase region, fusing in a collapsed four-phase
tetrahedron, which could inflate if there were more than
three components. The three ovoids containing tricritical
points suffer from the same model-dependent objections
as above, but the appearance of the four-phase plane does
not. A change in the interadion parameters hetween the
three components will turn the tricritical points into
critical tie lines, while maintaining the existence of a plane
which contains four phases. A less symmetric case is
presented in Figure 2 of ref 15, showing isothermal slices
of a nonisothermal ternary phase diagram, with the ovoids
truncated by the composition edges of the prism and, at
low temperatures, by a solid phase.
Concluding Remarks
Figure 8 and Figure 10a (with its mirror image Figure
lob) are the basic ways-achiral and chiral-that three
coexisting liquid phases can occur in phase diagrams of the
sort used here, whether the vertical axis is an intensive
variable or molecular parameter common to all phases, or
a compositional variable such as salt concentration, which
generally differs from phase to phase. In this latter situ-
ation tie lines remain straight but are tilted out of hori-
zontal; two-phase regions become twisted bundles of tie
lines, upper and lower critical solution points need no
longer be uppermost or lowermost; and tie triangles,
though still planar, are tipped askew within the diagram.
Still, all of the contact requirements of Gibbs’ phase rule
are preserved, and more complicated diagrams can be
assembled from the basic ones by deformations, now in-
cluding certain modes of tipping and twisting, by trans-
lations, by truncations, and by fusion of more than one
(15) A. W. Francis, J . P h p . Chem., 60.20 (1956).
Knickerbocker et al.
basic two-phase or three-phase body.
The skewing effect of salinity is obvious in the com-
parison of Figures 3 and 10. Nonetheless, each of the
pattern of phase behavior reported in this paper or earlier‘
232, 233, 322, etc., can he represented by starting with
-
Figure 10, truncating it a t high salinity with a solid,
salbrich phase, and otherwise manipulating it in allowable
ways. For example, fusing the solid ovoid of Figure 10a
onto the top of its mirror image, Figure lob, with due
regard for joining plait-point loops and segments, leads to
the body shown in the prismatic ternary diagram in Figure
12. Plainly its topology accounts for 2323 phase count
pattern, and only moderate topology-preserving manipu-
lations can convert it to the combinations of Figures 3 and
5 called for above. Moreover, truncating and skewing the
resulting diagram would produce diagrams for 233 and 323
systems that have been reported. Figure 11also shows the
possibilities of fusing ovoids to form more complicated
diagrams, including a four-phase region (which must be
planar when a field variable is used as the vertical axis,
hut will inflate to a tetrahedron when salinity is the vertical
axis.)
An intriguing possibiity worth pursuing where, as in the
science and technology of enhancing petroleum recovery,
there is justification, is to use computer methods to gen-
erate the basic ternary and quaternary phase diagrams,’+”
to manipulate them systematicallyto fit experimental data
for interpolation and extrapolation, and to display them
for research and development purposes. Tricritical points
are of particular interest in computations of multiple
fluid-phase equilibria from equations of state. The reason
is that by parameter variation it is often possible to trace
ordinary critical points from the tricritical points out of
which they have unfolded, the tricritical points being easier
to locate in the first place. Understanding the thermo-
dynamic principles behind patterns of three-phase be-
havior will lead to deeper, molecular-level research as well
as more efficient representations of multiphase equilibria.
Acknowledgment. This research was largely supported
by the Fossil Energy Division of the Department of En-
ergy. B.M.K. started working as an NSF Undergraduate
Research Participant. W.R. Rossen’s insights into the
achiral diagram and P. K. Kilpatrick’s into tricritical points
were very helpful.
Supplementary Material Available: Compositions of
the three equilibrium phases in the two three-phase regions
of the water-NPA-octaneLiCI phase diagram (3 pages).
Ordering information is availahle on any current masthead
page.
(18) R. G. B m m . M. S.Bidner, H. T. Davis,S. Prager, and L. E.
Scriven. SPE Paper No. 6122, Bakersfield, CA, April 14, 1977.
(17) D.Y. Peng and D. B. Rabinson, Ind. Eng. Chem. Fundom., IS,
59 (1976).
(18) J. M. Soreman, T. Msgnuaaen, P. Rasrnuasen and A. Freden-
elund. Fluid Phoge Equilib. 2. 297 (1979); 3. 47 (1979); 4, 151j1980).
(19) W. R. Rossen. Ph.D. Thesis. University of Minnesota, Minneap
olis, MN, 1982.