SURFACE TENSION AND THE

PRINCIPLE OF CORRESPONDING
STATES
James R. Brock and R. Byron Bird, University of Wisconsin, Madison, Wisconsin

Experimental surface-tension data for pure substances have been correlated suc- P.C.S. characteristic of its own
cessfully by the use of two recent modifications of the principle of corresponding group.
states. The results are expressed in terms of simple analytical relations which allow
one to calculate the surface-tension curves either from the critical constants P,, V , , Actually very little has been done
and T , or from P,, T , , and the boiling point TI, and also provide a method for esti- to modify the P.C.S. by the intro-
mating critical properties from surface-tension measurements. In addition, a method duction of additional parameters t o
for estimating the surface tension of molten metals is suggested. characterize the chemical structure
of the constituent molecules. The
There is a t present considerable sults given here also serve as a n reason for this is that an enormous
interest in fluid mechanics prob- additional test for t h e two modifica- amount of very accurate experi-
lems involving free surfaces in con- tions of the P.C.S. which a r e used. mental data is required in order t o
nection with the reexamination of deduce t h e effects of dipole mo-
t h e unit operations in terms of the PRINCIPLE OF CORRESPONDING ments, quadrupole moments, polar-
fundamental transport processes STATES AND ITS MODIFICATIONS izabilities, and molecular shape on
taking place. It is therefore im- The P.C.S. in its original form t h e bulk physical properties. Be-
portant to have experimental data as enunciated by van der Waals cause of the impracticality of any
on surface properties or in t h e was applied specifically to t h e equa- such attempt at the present time,
absence of such data to be able to tion of state i n the form one might well ask whether some
make intelligent predictions of sort of empirical correlating para-
them. P r = Pr (Vq Tr) (1) meter can be found-one which
The current status of the experi- could be expected to work for a
mental data and t h e theory of sur- That is, f o r all substances the re- large number of substances. Re-
face tension is not too satisfactory. duced pressure should be, a uni- cently two such empirical methods
Many of the data in t h e standard versal function of the reduced vol- have been suggested, in which a
handbooks a r e more than thirty ume and reduced temperature. This single additional “correlating para-
years old and have never been cor- principle has been used in an alter- meter” has been introduced :
roborated by subsequent investi- nate form by Hougen and Watson 1. Meissner and Seferian(20)
gators. Also the discrepancies be- ( 1 1 ) in t h e preparation of their have suggested t h a t all substances
tween measurements on a given generalized compressibility chart, with the same value of the para-
substance by two different methods which is a graphical representation meter 2, = P,V,IRT, obey t h e same
have not been properly explained. of the function P.C.S., that is, t h a t Equation (2)
Considerable progress has been be replaced by
made, however, in developing the Z = Z (Pr, Tv) (2)
theory of surface tension in terms 2 = Z (Pq Tr;23 (3)
of intermolecular forces by means In addition to the generalized com-
of statistical mechanics ( 1 0). Two pressibility chart (and the derived indicating the explicit parametric
main approaches have been used: charts f o r t h e thermodynamic func- dependence on 2,. Similar expres-
t h e free-volume theories, based on tions) charts have been prepared sions can be written f o r t h e trans-
an approximate evaluation of the f o r two of the transport coeffi- port coefficients and other proper-
partition function(5,15), and the cients: viscosity(ll,Z4), f o r which ties. This modification of the P.C.S.
radial-distribution-function method p,. =.p: (P,., T,.), and thermal con- has been used by Lydersen, Green-
(2,3,13). Thus f a r these theories ductivity(8,12), f o r which k,.= korn and Hougen(l8) to make a
have been developed only for mon- kr(Pr, T r ) * very extensive analysis of data on
atomic substances, and much more All these charts have been of equation of state and thermo-
needs t o be done before t h e statisti- considerable use f o r purposes of dynamic properties of gases and
cal theories will be generally useful instruction and rapid calculation. liquids.
for predicting surface tensions of It is, however, inherently assumed 2. Riedel(2l) has suggested t h a t
complex compounds. in these generalized representa- all substances with the same value
Hence the methods presented tions that the constituent molecules of Y,= ( d In P,/d In 2‘) ,follow the
here, based on t h e principle of cor- of all substances a r e mechanically same P.C.S., t h a t is, t h a t Equation
responding states (hereinafter ab- similar. This drawback led Kamer- (2) be replaced by
breviated P.lC.S.), a r e offered as lingh Onnes to modify the original
a n aid f o r practical calculations PJC.S. by suggesting t h a t each 2 = 2 (Pr,Tr;Y,) (4)
pending f u r t h e r experimental and group of mechanically similar sub-
theoretical developments. The re- stances-for example, t h e halogens, with analogous relations f o r t h e
t h e alcohols, the straight chain other physical properties. Riedel
J. R. Brock is at present with Humble Oil
and Refining Company, Houston, Texas. paraffins-should obey a separate has achieved particularly good suc-

?age 174 A.1.Ch.E. Journal June, 1955

cess with this method in the corre- Recently Rowlinson ( 2 3 ) succeeded pressure of liquids, rectilinear di-
lation of vapor-pressure data.” in showing theoretically that a t ameters, Trouton’s constants, heat
The application of either of these least f o r certain types of molecular capacities (at constant volume) of
methods raises t h e important ques- interactions a single correlating liquids, and surface tension. The
tion a s to whether a single corre- parameter is indeed adequate. To success which t h e authors have had
lating parameter-such as 2, or Yo substantiate his theory he has in correlating surface-tension meas-
-can be expected to be sufficient studied t h e following properties of urements on the basis of methods
for describing the behavior of sixteen widely varying substances : 1 and 2 above can be regarded as
many different physical properties. second virial coefficients, vapor an additional justification of Row-
*The quantity here called Y,, is given equation linson’s generalization of the P.C.S.
the symbol 5cI, by Riedel(ZO), but the In P, = Y e In T,, - O.O364(Y, -
present authors prefer not t o use CL 3.75) [ ( 3 6 / T , , )4-42 In ThT- 35 - APPLICATION OF THE MODIFIED
since this letter is frequently used p. C. S. TO SURFACE TENSION
for the coeficient of expansion (a Tbr61
in which Tar is the reduced boiling Neither of the above discussed
In VIaT).. Riedel has shown by 6 temperature corresponding t o the re- modifications of the P.C.S. has yet
very careful analysis of vapor-pres- duced pressure p,..If many values of been fully examined a s to how well
sure data that the quantity Y o can Y o are needed, they can be calculated various physical properties can be
be calculated from the boiling point quickly from tables prepared by
T n a t a pressure P by means of the Riedel(Z1) . described. It is of interest to test
these methods by applying them to
12, ’ I’ ’ I
the correlation of surface-tension
data. F o r this property it is well
known that the P.C.S. in its origi-
nal form is definitely inadequate
( 9 ) , probably because of t h e align-
ment effects a t the interface, which
a r e of course highly sensitive to the
molecular shapes and molecular in-
teractions. The surface tension of
07 - a pure substance is particularly
a easy to study, since it is a function
e of temperature but not of pressure.
F o r the purpose of this work a
“reduced” surface tension has been
defined byn
0 4 1 ’ “ 1 8 1 1 1 1 1 * 1 I I I I ~ Y
32 33 34 35 36 37 38 3.9 40 41 42 43 44 45 46 47 48 49 YR = (5)
I/Zc p2/3T t / 3
FIG.I. DETERMINATION OF THE FUNCTION r (2,)IN EQUATION(8). THEVARI- ( I t should be noted that the
OUS TYPESOF CHEMICALSUBSTANCES USEDIN THE CORRELATIONAREINDICATED quantity y /Po% ( x T )% is dimension-
BY THE FOLLOWING SYMBOLS: less but t h a t quantity yR, defined
@ noble gases (> alkyl benzenes in Equation ( 5 ) , is not, because of
@ diatomic molecules (D nitrogen-containing organic compounds t h e missing factor x 5 , which was
@ simple inorganic molecules 0 halogen-containing organic compounds omitted to simplify calculations.)
0 straight-chain paraffins @ miscellaneous organic compounds
According t o the original P.C.S.
one would expect yIz to be a unique
0 branched paraffins function of T , for all substances.
esters 8 *One might also define a reduced
unsaturated hydrocarbons surface tension thus :
LINEREPRESENTSTHE LEASTSQUARE
THEDOTTED FIT, I’ = -0.951+ (0.432/2,).
12
= YV:’~/ KT, (5a)
This definition suffers from the dis-
advantage that V , is not known ac-
curately for many substances. Rowlin-
son (23) has, however, used this defi-
nition in his investigation.
According to the two modified
P.C.S., however,
Y R = Y R (Tr;Zc) (6)
Y R = Y R (Tr;Y J (7)
The temperature dependence of t h e
surface tension for a great many
substances has been shown ( 9 ) to
be proportional to ( 1-T,) 1119. The
0.4‘ I
assumption t h a t this temperature
6.0 6.5 70 75 8.0 8.5 9.0 9.5 10.0 10.5 dependence is universal allows one
YC to rewrite Equations (6) and (7)
FIG. 2. DETERMINATIONOF THE FUNCTION r ( Y c ) IN EQUATION (9). thus
THESYMBOLS USED FOR THE VARIOUS TYPESOF CHEMICAL SUBSTANCES
ARETHE SAMEAS THOSEGIVEN IN FIGURE! 1. THE DOTTED LINE REP- 11/9
RESENTS THE LEASTSQUARE FIT, r = -0.281 0.133Y0. + YR = r(Zc)(I- Tr) (8)

Vol. 1, No. 2 A.1.Ch.E. Journal Page 175

 d
11/9
YR = r(Y,) ( 1 - T J (9)
The experimental surface-tension
data used f o r obtaining r(Z,) and
r ( Y e ) a r e those given in Landolt-
Bornstlein's Physikalisch4hemis-
che Tabellen, the International
Critical Tables, t h e Annual Table
of Physical Constants, Tables An-
nuelles de Constantes e t DonnCes
NumCriques and the Tables of t h e
American Petroleum Institute. For
many substances i t was not: possi-
ble to find the surface tension of
t h e pure substance against its ow,n
saturated vapor; i t was necessary
t o use the value of surface tension
against air or some inert gas. The
error involved in the use of t h e
latter data is believed to be quite
small in comparison with the aver-
age deviation from the correlations
based on t h e principle of corre-
sponding states. The critical con-
stants needed f o r the correlation
were obtained from a recent com-
pilation made by Lydersen ( 1 7 ) and
from the extensive tabulation pre-
pared by Kobe and L y n n ( l 4 ) .
Eighty-four substances of widely
iffering character were selected
for which both surface-tension and
critical data were available. These
included two noble gases, five di-
atomic molecules, seven simple
polyatomic inorganic substances,
the first twenty normal paraffins,
ten branched paraffins, five alkyl
benzenes, six unsaturated hydrocar-
bons, eight esters, eight organic
halogen compounds, six organic ni-
trogen compounds, and seven other
organic compounds. The experi-
mental values of yRI ( l-Tr) 1119
were plotted against l i Z , (as shown
in Figure 1 ) and against Y , (as
shown in Figure 2 ) . Through the
point so plotted, the best straight
lines were drawn by means of a
least-squares fit. These linear rela-
Equations (10) and (11) could
tions then allow Equations (8) and
be used for estimating critical
(9) to be rewritten a s constants from surface-tension
measurements. If P , and T, a r e
TABLE ANOMALOUS DEVIATIONO F known, then V , can be obtained
EXPERIMENTAL SURFACETENSIONS f rorn
FROM GENERALIZEDEQUATIONS
Average %
Substance deviation from
Eq. (10) Eq. 11
Polar inorganic sub-
stances
++
Water. . . .. . . . . . . . . .
Ammonia. . . . . . . . . . .
90.6
35.5
+
+
44.0
6.2
++ 31.4 + 13.2
Hydrogen chloride. ..
Hydrogen bromide.. . 13.4 + 5.0
Alcohols
Methanol. . . . . . . . . . . +I79 + 58.2
++
Ethanol. , . . . . . . . . . .
Butanol . . . . . . . . . . . .
89 +I17
39.4 + 87
Carboxylic acids
Acetic acid.. . ... . . . . $166 + 70
. ++
Propionic acid. . . . . . . +
88 91
Butyric acid., . . . . . 21 +I15
Page 176
YR = (-0.281 + A trial-and-error method has to be
0.133 Y,) (1- TT)l1" (11)
which are the final generalized re-
lations for surface tension based
on the two modified P.C.S. de-
scribed above.*
GENERALIZED RELATIONS FOR
SIIJRFACE TENSION
Applicability. Over the tempera-
t u r e range of the available data f o r
the eighty-four substances con-
sidered, i t was found t h a t the aver-
age error for Equation (10) is
6.2% and for Equation (11, 3.0%.
The smaller deviation for t h e Y ,
correlation does not necessarily in-
dicate that Y, is fundamentally a
better choice of correlating para-
meter than 2,. Many of the devia-
tions in the 2, correlation a r e quite
likely due to erroneous values of 2,
(which in t u r n frequently can be
traced to faulty V , values).
Equations (10) and (11) a r e
sufficiently accurate for most en-
gineering calculations and provide
a quick method for estimating sur-
face tension from critical data. If
experimental critical data are not
available, they can be estimated by
the group-contribution methods de-
vised by Riedel(21,ZZ) and ex-
tended by Lydersen ( 1 7 ) . Lydersen
has also given an empirical rela-
tion whereby Z, may be estimated
from heats of vaporization.
V , = 0.432 (RT,/P,) [0.9.51 + and the various heavy isotopes of
And if T, and a boiling point a r e
known, then P, can be found from
*The constants in Equations (10
and (11) are slightly different from
those given previouslyl, since only
30 substances were used in the origi-
nal correlation.
A.1.Ch.E. Journal
P, =
Y / T/ ~
(- 0.281+O. 133U,) (1 - T,)
(13)
used to obtain P,, as Y , depends on
P,. Equation (13) is probably about
as good f o r estimating P , as is t h e
group-contribution formula given
by Lydersen ( 1 7 ) .
Limitations. The generalized rela-
tions for surface tension are rea-
sonably successful for simple in-
organic substances and for a wide
variety of organic compounds.
There a r e several groups of sub-
stances f o r which the correlations
cannot be expected to hold: very
light atoms and molecules (the
isotopes of hydrogen and helium) ,
molten metals, certain highly polar
inorganic substances ( H20, NH,,
HCl, H B r ) , associating substances
(alcohols, carboxylic acids, etc.) ,
and fused salts. The order of mag-
nitude of the deviations of some
of these substances from the gen-
eralized equations is indicated in
Table 1.
The only class of substances
listed above which has been suc-
cessfully studied is t h e first one-
the isotopes of hydrogen and helium
-which exhibit quantum effects.
These substances should always be
excluded from generalized correla-
tions inasmuch as there a r e other
preferable ways to handle them.
The first method was suggested by
B y k ( 4 ) , who proposed a quantum
mechanical modification of t h e
original van der Waals' principle
of corresponding states. A theo-
retically sounder method is the use
of a quantum mechanical P.C.S.
based on the reduction of variables
with molecular parameters ( r a t h e r
than critical parameters), as pro-
posed by de Boer(7) and discussed
extensively by Lunbeck i19 ) . The
application of this latter approach
t o surface tension has been carried
out by one of the a u t h o r s ( 6 ) . I n
this way the surface tension of He"
hydrogen. Recently the surface ten-
sion of He3 was measured by Love-
joy(16) in t h e range 1.0" to 2.5"K.
Above about 1.8"K. the experi-
mental values agree almost exactly
with the predicted curve. The dis-
crepancy below 1.8"K. is undoubt-
edly due to statistics effects, which
were not taken into account in the
P.C.S. prediction. As yet no ex-
perimental values of the surface
tension for the heavy isotopes of
hydrogen are available.
June, 1955.

‘og,,Pe
FIG. 3. SURFACE TENSIONAT 600°C.
I N DYNE CM.-l O F VARIOUS MOLTEN
METALS PLOTTED AGAINSTTHE LOG-
ARITHM O F THEIR ELECTRICAL RE-
SISTIVITY AT 20°C. I N MICROHMCM.
Much needs to be done in con-
nection with t h e other anomalous
substances. F o r the molten metals
there seems to be a relationship
between the surface tension and
the electrical resistivity, as can be
seen in Figure 3. F o r the materials
in the last three groups above, no
workable generalizations have been
obtained. It is not known a t present
whether Equations (10) and (11)
apply generally to the fluorocar-
bons. The only one of these sub-
stances for which both surface-
tension and critical data could be
found was perfluoroheptane, f o r
which the agreement was not par-
ticularly good. It will be interest-
ing to see whether this class of
compounds exhibits further devia-
tions.
SURFACE TENSION AND THE
PARACHOIR
The calculation of the surface
tension from the modified P.C.S.
methods is not presented as a sub-
stitute for the method of calculat-
ing surface tension from the para-
chor. The latter calculation requires
a knowledge of the liquid and vapor
densities, whereas the former re-
quires that one know either p,, v,,
T, or P,, V,, T,. Thus t h e two
schemes of estimating surface ten-
sion are complementary. Further-
more it is frequently useful to have
two different methods available f o r
estimating physical properties un-
der unusual conditions. A P.C.S.
approach to surface tension is in a
Vol. 1, No. 2
sense more satisfying, in t h a t t h e
correlation is obtained by a general
method, capable of being applied t o
a variety of properties as already
discussed. The parachor, on the
other hand, involves a specific
method.
The parachor is f a r from being
well understood theoretically. It
was introduced as a useful em-
piricism by Sugden, and its use
in structural determinations is well
known. At the present time there
is no satisfactory explanation of
the fact that the parachor is very
nearly constant over a wide range
of temperature. Lennard-Jones and
Corner ( 1 3 ) have shown that the
parachor is given within about 4%
for simple substances by the em-
pirical relation :
[PI = (7.1 X lo3 05” (14)
in which o and E a r e the parameters
in the Lennard-Jones potential ( a
is the low-energy collision diameter
in AngstrGm units and E is the
maximum energy of intermolecular
attraction in e r g s ) . This relation
shows that the parachor is almost
proportional to 03, which is in
agreement with Sugden’s conten-
tion t h a t the parachor is some sort
of modified molecular volume.
ACKNOWLEDGMENT
The authors are indebted to Aksel
Lydersen for numerous interesting
discussions and for the use of some
of his unpublished work. They also
wish to thank Arthur Stroud for per-
forming some of the calculations. In
addition, R. B. Bird wishes to
acknowledge the financial support of
the National Science Foundation.
NOTATION
k = coefficient of thermal conduc-
tivity
P = pressure
P , = vapor pressure
R = gas constant
T = temperature
V = molar volume
Y = ( d In P , l d In T )
2 = compressibility factor, PVIRT
y = surface tension
y s = “reduced” surface tension
[see Equation ( 5 ) ]
l? = quantity defined in Equations
(8) and ( 9 )
x = Boltzmann’s constant
p. = coefficient of shear viscosity
pe = electrical resistivity
Subscripts
b = quantity evaluated a t t h e boil-
ing point
A.1.Ch.E. Journal
c = quantity evaluated a t the criti-
cal point
r I quantity made dimensionless
by division by the correspond-
ing quantity a t the critical
point
LITERATURE CITED
1. Brock, J. R., MS. thesis, Univ.
Wisconsin (1954).
2. Buff, F. P., and J. G. Kirkwood,
J. Chem. Phys., 17, 338 (1949).
3. Buff, F. P., J . Chem. Phys., 19,
991 (1950).
4. Byk, A., Ann. Physik, 66, 157
(1921); 69, 161 (1922).
5. Corner, J., Trans. Faraday SO-
ciety, 44, 1036 (1948).
6. De Boer, J., and R. B. Bird,
“Quantum Theory and the Equa-
tion of State,” Chapter 6 in
Reference 10, p. 430.
7. de Boer, J., Physica, 14, 139
(1948).
8. Gamson, B. W., Chem. E n g . Progr.
45, 154 (1949).
9. Guggenheim, E. A., “Thermo-
dynamics,” 2 ed., North-Holland
Publishing Company, Amsterdam
(1950).
10. Hirschfelder, J. O., C. F. Curtiss,
and R. B. Bird, “Molecular
Theory of Gases and Liquids,”
John Wiley and Sons, New York
(1954).
11. Hougen, 0. A., and K. M. Watson,
“Chemical Process Principles,”
John Wiles and Sons, New York
(1947).
12. Hougen, 0. A., Chem. E n g . Progr.
Monoarawh Series, No. 1, 47,
(1951).
13. Kirkwood, J. G., and F. P. Buff,
J. Chem. Phus.. 17. 338 (1949).
14. Kobe, K. A., and R.’E. Lynn, Jr.,
Chem. Revs., 52, 117 (1953).
15. Lennard-Jones, J. E., and J. Cor-
ner, Trans. Faraday Society, 36,
1156 (1940).
16. Lovejoy, D. R., Can. J. Plays., 33,
49-53 (1955).
17. Lydersen, A. L., “Estimation of
Critical Properties of Organic
Compounds by the Method of
Group Contributions,” Univ. W i s -
consin Eng. E x p t . Sta. Rept.
(1955).
18. Lydersen, A. L., R. A. Greenkorn,
and 0. A. Hougen, “Thermody-
namic Properties of Pure Fluids
from the Theorem of Correspond-
ing States,” Univ. Wisconsin Eng.
E x p t . S t a . Rept. (1955).
19. Lunbeck, R. J., Doctoral disserta-
tion, University of Amsterdam
(1951).
20. Meissner, H. P., and R. Seferian,
Chem. E n g . Progr., 47, 579
(1951).
21. Riedel, L., Chem. Ing. Tech., 26,
83 (1954).
22. Ibid., p. 259.
23. Rowlinson, 3. S., Trans. Faraday
Society, 50, 647 (1954).
24. Uyehara, 0. A., and K. M. Wat-
son, N u t . Petroleum News, 36, 764
(October 4, 1944).
Page 17‘1-