Conversely certain components could be effectively se- Brunner, C. A., Stephan, D. G., Ind. Eng. Chem. 5 , 4 0 4 8 (1965).

Dick, W. L., M.A.Sc. Thesis, University of Ottawa, Canada,

questered by a ligand such as E D T A which would produce 1969.
a negatively charged chelate subject t o the repulsive forces in Glasstone, S., “Textbook of Physical Chemistry,” 2nd ed, p.
the diffuse layer. 1246, D. Van Nostrand Co., New York, 1959.
Haas, P. A., Oak Ridge National Laboratory Rept. ORNL-3527,
1 Q65.
Acknowledgment Keyworth, D. A., T a h n t a 2, 297 (1959).
Rubin, A. J., Johnson, J. D., Lamb, J. C., 111, Znd. Eng. Chem.,
The authors wish to thank the Department of Energy, Pro;. Des. ‘Develop. 5, 368i375 (1966).
Sandell, E. B., “Colorimetric Determination of Traces of Metals,”
Mines, and Resources for providing financial assistance for 2nd ed., pp. 309-311, Interscience, New York, 1950.
this work. Schonfeld, E., Sanford, R., Mazella, G., Ghosh, D., Mook, S.,
“The Removal of Strontium and Cesium From Nuclear Waste
Solutions by Foam Separation,” Report No. NYO-9577,
Nomenclature Radiation Applications Inc., New York, 1960.
rc = surface excess of copper, mol/cm2 WILLIAM L. DICK‘
CF = copper concentration in the foamate, mol/cm3 FRANK D. TALBOT2
CB = copper concentration in the bulk, mo1/cm3
VF = collapsed foam (foamate) rate, cm3/min Department of Chemical Engineering
S = surface rate, cm2/min University of Ottawa
D = distribution factor, ern Ottawa, Ontario, Canada
[i]= concentration of component i
1 Present address, Division of Industrial Wastes, Ontario Water
Resources Commission, Toronto, Ontario, Canada.
literature Cited 2 To whom correspondence should be sent.

Brunner, C. A., Lemlich, R., IND.ENG.CHEM.,FUNDAM.

4,297- RECEIVED for review October 27, 1969
300 (1963). ACCEPTED January 22, 1971

Solubility of Gases in liquids in Relation to the Partial

Molar Volumes of the Solute. Carbon Dioxide-Water

It i s shown that the Krichevsky-Kasarnovsky equation does not in fact truly represent the solubility data for
COZ in liquid water. However, the Krichevsky-llinskaya equation i s appropriate for this purpose at low COS
concentrat ions.

P a r k i n s o n and de Nevers (1969) recently showed that al- Fco2= vcOzm

= 37.6 ml/mole
though experimental solubility data for carbon dioxide in
us.OP~c)o Zat various
liquid water when plotted as In ( ~ C O ~ / X C a t temperatures between 0 and 50°C for values of xcoZ less
constant temperatures give straight lines in conformity with than 0.8 mole yo.Therefore, if we limit consideration t o this
the Krichevsky-Kasarnovsky equation (1935), the slopes of range of conditions, then any failure of the Krichevsky-
these lines are not in general equal to the values obtained for Kasarnovsky equation must reflect a variation of the activity
Fco,m/RT by independent experiment. Examination of the coefficient of the dissolved C o n with COz Concentration.
Krichevsky-Kasarnovsky equation One might suppose in this event that plots of In ( ~ C O ~ / X C O ~ )
us. Pcozcould not yield straight lines. This is not the case,
as we shall show.
A general equation for ~ C O ,as a function of x? is given by
Prausnitz (1969)
indicates that i f it i s a valid equation the slopes of the straight
lines shown by Parkinson and de Nevers must equal v c o Z m /
RT. The fact that they do not led these authors t o conclude
that “one must be cautious in applying the equation in this where by definition
way; internal agreement does not seem to guarantee ac-
curacy.” It is the purpose of this communication t o explain sco,
H = lim - (constant T )
the apparent anomaly. zc0,-0 xcoz
The Krichevsky-Kasarnovsky equation holds for the
solubility of a gas in a relatively nonvolatile liquid solvent and ycoi* is the activity of the dissolved COS normalized so
subject to the restriction t h a t both the partial molar volume that
and the activity coefficient of the solute in the liquid phase
lim ycOl* = 1
must be independent of solute concentration, and therefore of xc0,-0
pressure.
The data of Parkinson and de Nevers show that f c o , is The simplest realistic expression for yco9* as a function of
indeed essentially constant at a value of xcoZis given by (Prausnitz, 1969)

312 Ind. Eng. Chern. Fundarn., Vol. 10, No, 2, 1971

 0.10 - I I I I I
Table 1. Values of Characteristic Constants. IO'C
XCO~ < 0.008 -
0.08-
r, oc H In H A/RT #
.

0 670 6.507 -2.420 80"

D* 0.06-
-
10 1000 6.908 -2.520 I
15 1233 7.117 -1.748
20 1432 7.267 -0.700
8
0.04- -
25 1633 7.398 0.315
35 2098 7.649 1.478 -
50 2904 7.974 1.948
I
0 5 IO 15 20 25 30
-(x&,- I) lo3
A
In Y C O ~=* R~( z ~ , o-~ 1) (3) Figure 1 . Typical fit of data by Equation 4

where A is a constant for a given temperature. Equations 2

and 3 together with the rigorous thermodynamic equation Since PcO,/P is never far from unity, this procedure should be
entirely adequate.
These plots produced curves that became linear a t lower
vco,
pressures, including the region for which constancy of ~ C O ~
T,Z RT is indicated, and the slopes of the linear portions of these
curves give values for AIRT. The values of this quantity
lead to the Krichevsky-Ilinskaya equation (1945)
so determined are listed in Table I. An example of these
A plots is shown in Figure 1. The ultimate departure of these
In
bo,
--
xcoz
= In H + RT
- (ZH,O~ - 1) + plots from linearity reflects either the inadequacy of a con-
stant value for ~ C O ,a t higher COS concentrations or the
vc02mPC0, failure of Equation 3 a t higher COS concentrations, or both.
(constant T ) (4)
RT In any event the assumptions attendant to Equation 4 are
evidently valid for the range of COS concentrations of in-
It is again assumed in the derivation of this equation that terest, Le., for ZCO, < 0.8 mole %.
vco, is independent of xcOzand P , a condition that will be We can now construct plots of
satisfied by the data employed. The only other assumption
is that Y*CO, is given by Equation 3 for the range of condi- fco, A
tions employed, and adequate data are available to test this In - - - (zH2o2 - 1) U S . Pcoz
xco2 RT
assumption.
Note that in both Equations 1 and 4 PCO,is presumed to be for comparison with plots of In (~?co,/zco~) vs. PcoZas shown
given by by Parkinson and de Nevers. As seen from Equation 4, the
former plots should be straight lines with slopes equal to
PCOz = p - PHzO Vcozrn/RT. The comparison is shown in Figure 2. The solid
where P is the system pressure and Pazois the vapor pressure lines conform to Equation 4 and have been drawn independent
of water a t the temperature of interest. of the points (solid circles), starting a t intercept values equal
to In H and extending to the right with constant slopes
exactly equal to vcoZm/RT.These straight lines clearly
Procedure correlate the data, as they should. The plots of In ( ~ c o ~ / z c o , )
By rearrangement of Equation 4 we have us. PCO?are shown as dashed straight lines drawn by eye
through the data points (open circles). They obviously do not
fco, vCOzmPCo2 A in general have slopes equal to Vco,"/RT.
ln-- = F~( Z H ~ O-~ 1) (constant T )
H Z co, - RT
Discussion
Plots were made of the left-hand side of this equation us.
( z - ~1) for
~ temperatures
~ ~ of 0, 10, 15, 20, 25, 35, and It may seem surprising that both sets of lines in Figure 2
50°C. The fugacity data and solubilities were taken from the should be straight. If we take the Krichevsky-Ilinskaya
compilation of Houghten, et al. (19571, and VcoZrnwas taken equation, Equation 4, as valid, as has been shown for the
as constant as previously stated. Values of H were first deter- concentration region considered, then the solid lines must
mined as the intercepts on plots of ~ C O ~ / Z C Ous., Pco,. These be straight. There is, however, no evident reason why the
plots become almost linear a t low pressures, and therefore dahsed lines should be straight.
extrapolation to PCO,= 0 is readily accomplished. The values Differentiation of Equation 4 at constant T gives
of H so obtained are listed in Table I. Determination of
~ C O , was based upon the assumptions that for the vapor
phase dPco2
2 2 ~ ~ 0 1 4&H,O)
d In (fco2/Zcoz) -~
-
RT
(
~

dPco,
I 2;m
Pco At the concentrations considered x H -+~ ~1, and in addition
fco2 = fC02"YC0~ = fco," 7: we have the exact relation, dzazo = -dzco,. Therefore

Ind. Eng. Chem. Fundam., Vol. 10, No. 2, 1971 313

 Table II. Comparison of Slopes. See Equation 5

(1 1
r, oc 1 03,a h - ' ah-' ah-'
31 I
i 15
0
10
1.7
1.6
1.6
7.2
5.0
2.9
8.9
6.6
4.5
8.7
6.2
4.2
20 1.6 1.0 2.6 2.5
25 1.5 -0.4 1.1 1.1
35 1.5 -1.4 0.1 0.0
50 1.4 -1.3 0.1 0.0

- i The quality of this approximation is shown by the results

listed in Table 11, where values for d In ( ~ C O ~ / Z C O a~s ) / ~ P C O ~

6.6f L4-'4----u
ORDINATE AS INDICATED
--e- ORDINATE WITH Am0
- 4
O'C
determined from the slopes of the dashed lines of Figure 2
are compared with values calculated by the above expres-
sion. The comparison is between columns 4 and 5.
V l . r i I I
The values of A/RT determined in this work are shown as a
0 4 8 12
.-
pCo, ,atm function of temperature in Figure 3. At the lower tempera-
tures A/RT is negative, and according t o Equation 3, ycoZ*
Figure 2. Comparison of data correlations for the sol-
ubility of carbon dioxide in water will be greater than unity. This indicates a lower solubility
of COZin water than would be given by Henry's law. It may
be that the clustering of water molecules a t lower tempera-
tures is responsible for this effect. The hook a t the end of the
d 111 ( ~ c o 2 / ~ crv-
o J -2A/RT
-
BC O~~ curve near t = 0°C may also be due to the unusual behavior
dPco, ( d P c o l l d x c o 3 i- of water a t temperatures near its freezing point.
The only variable quantity on the right-hand side of this At higher temperatures A/RT becomes positive, making
expression is dPCO,/dzCoz. If the data are plotted as ycO2* < 1 and resulting in higher COa solubilities than would
P c O 2 / x c O 2us. Pco2, lines are obtained t h a t are nearly linear be expected from Henry's law. This suggests a specific
with positive slopes. Thus there is no region where interaction between the COZ and HzO molecules. The value
dPCO2/dxCO2is actually constant. However, this quantity of A/RT appears to level off at a value of about 2.0 for tem-
peratures above 5OoC.
changes by less than 30% a t 0°C and by less than 15% a t
50°C over the pressure ranges considered, and its systematic With respect to Equation 4 the two variable terms on the
increase nith pressure is small enough t o be masked by scatter right-hand side contribute differently a t different tempera-
in the data. tures. At lower temperatures both terms are positive and
both therefore serve to increase In ( ~ C O ~ / Z C O ~ )At
. about 23"C,
The limiting value of ~ P c o ~ / ~a zt PcoZc o ~= 0 is, of course,
Henry's constant H.Thus to a first approximation we may however, A is zero, and Equation 4 reduces to Equation 1
write for this temperature. This is the reason that the two lines
of Figure 2 nearly coincide for a temperature of 25"C. Above
d 111 ( f c o l / ~ c o nrv
) 5 4 Vconm this temperature A becomes negative, and the two variable
df'co, -HRT + RT (5) terms of Equation 4 tend to counteract each other. I n fact,
they counteract each other so completely that a t both 35
and 50°C In ( ~ c o ~ / z c is
o ~essentially
) independent of P c O z ,
and the dashed lines for these temperatures on Figure 2 are
horizontal.

Acknowledgment

Acknowledgment is made to the donors of The Petroleum

Research Fund, administered by the American Chemical
Society, for partial support of this research.

Nomenclature

A = a constant for given T

j c o Z v = fugacity of pure COZ in the vapor phase a t the
system T and P
1 I I I I I ~ C O , = fugacity of COSas a solute in the liquid phase
0 IO 20 30 40 50
TEMPERATURE ,*C
H = Henry's constant
P = absolute pressure of the system
Figure 3. Temperature dependence of A P ~ o ,= partial pressure of CO?

314 Ind. Eng. Chem. Fundom., Vol. 10, No. 2, 1971

 = partial pressure of HzO
PH~O Krichevsky, I. R., Ilinskaya, A. A,, Zh. Fiz. Khim. 19, 621
R = universal gas constant (1945).
T = absolute temperature Krichevsky, I. R., Kasarnovsky, J. S., J . Amer. Chem. Soc. 57,
Vcoz = partial molar volume of COZ as a solute in the 2168 (1935).
Parkinson, W. J., de Yevers, N., IND.ENG.CHEM.,F U N D ~ M .
liquid phase 8, 709 (1969).
pc02m = partial molar volume of COS a t infinite dilution Prausnitz, J. X , “Moleciilar Thermodynamics of Fluid-Phase
z c o 2 = mole fraction of COz in the liquid phase Equilibria,” Prentice-Hall, Inc., Englewood Cliffs, N. J.,
ZH%O = mole fraction of Hz0 in the liquid phase 1969, Chapter 8.
yco2 = mole fraction of COz in the vapor phase
y c o 2 * = activity coefficient of COz as a solute in the liquid R. E. GIBBS
phase, normalized with respect t o Henry’s law H. C. VAN NESS
Rensselaer Polytechnic Institute
Literature Cited Troy, New York 12181
Houghton, G., LIcLean, A. RI., Ritchie, P. D., Chem. Eng. Sci. RECEIVED for review September 24, 1970
6 , 132 (1957). ACCEPTEDMarch 3, 1971

Distribution of Particle Spacings in a Random Dispersion

The distribution function of the particle spacings in a randomly dispersed system of equal-sized particles was
derived. The probability of occurrence of a certain spacing increases with a decrease in the spacing. This
seems to b e the origin of agglomeration among particles. Spherical particles are more likely to agglomerate
if the particle volume concentration is greater than about 5%.

M a n y investigations have been made into the spatial
distribution of particles in randomly dispersed and packed
systems (Haughey and Beveridge, 1966; Leek, et al., 1969;
Mason, 1968; Scott, 1962; Smith, 1968) in which most in-
vestigators discussed the distribution of the local particle
concentration,
The object of this study is to consider the distribution
of the particle spacings in a random dispersion. Knowledge
of the distribution is needed to explain the homogeneous
state of particle dispersions and to discuss the origin of par-
ticle agglomeration.
Distribution of Particle Spacings and Discussion
possible states of the particle dispersion. In Figure l b , for
example, four empty cells are shared among the three spac-
ings, XI,X 2 , and X s . The total number of the possible states
becomes (Davidson, 1962)
This is mathematically analogous to the Bose-Einstein
statistics in which elementary particles are shared among
the states of the energy levels (Gurney, 1949). Similarly,
the number of the mags in which one of the spacings may
have z empty cells is

I n a randomly dispersed system of equal-sized particles,

consider the representative size L in a n arbitrary direction
(Figure l a ) . The size L consists of the line of the A‘ cubic
cells. The size of each cell, d , is equal to the particle diameter This is so because the remaining (S - Ji - x) iiidistinguish-
and the cell cannot contain more than one particle. It is able empty cells are shared among the remaining dis-
assumed that a cell is either full or empty. d particle volume tinguishable spacings. If each of the states of the particle
fraction greater than 0.5 is counted as unity; a fraction dispersion is equally probable, Le., if there is complete random-
smaller than 0.5 is counted as zero. Hence, the state of the ness, the following equation gives the probability of occur-
particle dispersion along the representative size L can be rence of the state in which the face-to-face spacing from a
expressed by a line of full and empty cells, as in Figure 1b. particle to its nearest neighbor (particle or wall) is xd.
If 54 cells are full, the ratio .li/S relates to the overall par-
ticle volume fraction C, as (4)
c =M / S = Go/$ (1)
AV-Af
where @ is the shape factor of the particles (= n/6 for spheres). Obviously P(zd) = 1. For large values of ‘Y,say more
Z=O
As is seen from Figure l b , the state of the particle dispersion
than 50, Equation 4 reduces to (Kubo, 1960)
can be described by the spacing or the number of empty cells
between a particle and its nearest neighbor (particle or wall). P ( z d ) = C(1 - cy (5)
For Ji particles there are -If + 1 spacings; these are dis-
m
tinguished one from another by denoting each by a value of
where C M/.V and P(zd) 1. If we need the prob-
+
X,(i = 1 , 2 , , J4 1). The number of the ways of showing =
z= 0
=

the (S - M)indistinguishable empty cells among the (-14 + ability of occurrence of the center-to-center spacing y d ,
1) distinguishable spacings expresses the number of the from Equation 5 we obtain

Ind. Eng. Chem. Fundam., Vol. 10, No. 2, 1971 315