Sie sind auf Seite 1von 9

Chemi~ Eneitleeti Science,lSKS, Vol. 10, pp. C&34to !W2. Pegamon Press Ltd. London.

Printed in Great Britain

The evaporation of two -component liquid mixtures


Department of Chemical Engineering, Imperial College, London, SW.?.

(Received 14 December 1958)

Abstract-The ~ffu~~ties of acetone, carbon tetraehloride and water vapours in air have been
determined experimentally using the method first developed by WINKELMANNand values have
been obtained over a range of operating conditions to an accuracy of within f 2 per cent. The
evaporation of two-component liquid mixtures, consisting of a volatile and a non-volatile material
has also been been studied in the same apparatus. It has been shown that, when the non-vo~t~e
liquid is the denser, the liquid remains completely mixed and the evaporation rate can therefore
be simply calculated for rm ideal mixture. When the non-volatile component is the less dense,
convective mixing is negligible and the mass transfer in the liquid is a process of unsteady state
molecular diffusion.

R&urn&-L’auteur a d&ermin6 exp&imentalement les diffusibilitbs dans l’air des vapeurs

d’acdtone de tetrachlorure de carbone, et d’eau, en utilisant la mCthode de WINKELMANN. Les
valeurs obtenues dans certaines conditions d’opt!ration ont une prt!cision& + ou - 2%. Le m&e
appareillage sert ir Studier 1’6vaporation de mdlanges liquides B deux constituants comprenant
un produit volatil et un non volatil.
Quand le produit non volatil est le plus dense, l’auteur montre qui le liquide reste enti&ement
B I’&& de mklange, et la vitesse d’6vaporation peut done &re caleultie simplement eomme pour
un mhlange idPal. Quand le corps non volatil est le moins dense, le m&ngeage par convection est
n$#geable et le transfert de masse dans le liquide est un pro&d6 de diffusion mol&ulaire en
r@ne transitoire.

Zusammenfassung-Die Diff~ionskoeffizienten von Azeton, Tetrachlorkohlenstoff und

Wasserdampf in Luft werden experimentell in einervon WINKELMANNangegebenenApparatur mit
einer Genauigkeit von f 2% bestimmt. Die Verdampfung von fliissigen Gem&hen BUS zwei
Komponenten, einer fiiichtigenund eine~chtfl~chtigen, wurden ebenfalls in der gleichenApparatur
untersucht. Wenn die nichtfl~ohtige Fliissigkeit die gr&sere Dichte bat, dann bleibt die Fliissigkeit
vollst%ndig gemischt und die Verdampfungsgeschwindigkeit kann daher auf einfache Weise fiir
ein ideales Gem&h berechnet werden. Wenn die nichtfliichtige Komponente die geringere Dichte
hat, kann man die konvektive Vermischung vernachliissigen und den Stofftransport in der
Fl~sigkeit als einen niehtsta~io~r~n Vorgang der molekularen Diffusion auffassen.

will sink into the bulk of the liquid and considerable

WHEREAS there have been many investigations mixing will occur, whereas in the latter, the residue
relating to the evaporation of a pure liquid into a will tend to concentrate in the upper layers and
gas stream, little attention has been paid to the the more volatile liquid must diffuse through this
evaporation of a liquid consisting of two or more region before it can evaporate.
components of differing volatilities. The more For the present study, two simple two-com-
volatile components will then evaporate prefer- ponent systems have been selected. Each con-
entially leaving a residual liquid which may be of tained a virtually non-volatile liquid (dibutyl
greater or lower density than the original liquid, phthalate) ; in the first mixture the volatile liquid
according to the values of the densities of the was of lower density (acetone), and in the second
components. In the former case the heavy residue the volatile liquid was of higher density (carbon
The evaporationof two-componentliquid mixtures

tetrachloride). The method of investigation which bath. This air stream entered through the horizon-
was adopted was to allow the mixture to evaporate tal section B of the evaporation tube and the air
at constant temperature in a vertical tube across leaving the tube was metered. The walls of the
the top of which was passed an air stream in bath were of glass and the level of the liquid in the
which the vapour was carried away. The rate of tube was measured by means of a cathetometer to
evaporation was measured by observing the rate of within 0.002 cm. At the beginning of an experi-
fall of the liquid surface. ment the liquid was introduced into the tube and
A series of pre~nary experiments was carried allowed to reach the temperature of the bath
out using the pure volatile liquid, in order to before the passage of air was commenced. Read-
examine the reproducibility of results and the ings of liquid level were then obtained as a function
effect of experimental conditions. As a result of of time.
these experiments, values of the diffusivity of three The effect of the operating variables was studied
vapours in air-acetone, carbon tetrachloride and using acetone as the liquid. At a given tempera-
water-have been obtained. ture, the two important variables were the rate of
passage of air and the diameter of the tube. The
APPARATUS AND EXPERIXENTAL air rate was varied between 05 and 2.1 cms/sec
METHOD and three tube diameters were used (4, 6 and
The method adopted was based on that origin- 8 mm). The lower permissible limit of air rate is
ally used by STEFAN [l] and W~N~EL~ANN [z], set by the necessity of main~ning an approxima-
and subsequently modified by a number of workers tely zero concentration of vapour at the mouth of
including LEE and WILKE [3]. The liquid to be the tube, and the upper limit by the production of
evaporated was contained in the vertical limb of excessive turbulence in the upper parts of the
the tube shown in Fig. 1, the upper section A, diffusing vapour. In tubes of too small diameter,
serving merely for the introduction of the liquid. the overall rate of diffusion will be affected by the
velocity profile over the cross-section, and in
excessively large tubes, turbulence is induced too


Under conditions of molecular di~usion, the rate

cAir stream of mass transfer per unit area of liquid surface is
given by Stefan’s law as :--

c P
!!!=D _..!c%‘In_!i!?
‘dt “h pB~
where h is the distance between the air stream and
Tube diameters the liquid surface at time t, PBS, PB1 are the partial
pressures of air in the air stream and at the surface,
p is the molar density of the liquid, 0, is the
I'm.1. tivaporationtube. diffusivity of the vapour in air, and CDTis the total
molar concentration of gas and vapour in the
diffusion region ; this will be constant and equal
The tube was immersed in a the~ostatieally to the number of moles of gas plus vapour per
controlled bath. Air from a cylinder of compressed unit volume in the tube. In a given experiment,
gas passed through a copper coil immersed in the h and t are the only variables and therefore on
bath and its temperature was raised to that of the integration :


h2 - ho2
-__ = (2) Thus, if r-& is plotted against h - h,, a straight
where h, is the initial value of h. line should be’obtained. The diffusivity D, can
Now the effective distance, h, over which then be calculated from the slope s ;
molecular diffusion is occurring cannot be estim-
ated accurately because of end-effects ; the position i.e. (4)
Dv = 2s C,, ln’P,/P,,
of the liquid surface cannot be defined precisely
because of the curvature of the meniscus, and at and the effective initial distance over which
the upper end of the tube turbulence is induced in diffusion is occuring can be calculated from the
the diffusing vapour. On the other hand, values of intercept on the ordinate axis, A typical curve is
h - h, can be measured very accurately and it is shown in Fig. 2,
therefore convenient to rearrange equation (2) as :
@ - ho)(h - ho+
--- 2ho)_ 2Wm ln pB2 Results
t P pB~ The effect of tube diameter and air flow rate
t was studied using acetone at 30°C. In each case,
@ - ho)
Isee h = -20, G, 1,” PB&‘BI the results were corrected to standard atmos-
P ho pheric pressure by multiplying by the factor
(3) P/760.

FIG. 2. Experimental results. Acetone at 50.Q°C in 6 mm tube.

Table 1. l$ cct sf tube diuwzeter and air .flkm rates on apparent &ff&vitl/

Tube Air floret Slope of l?rembre

diameter rate curve (9) P
9 q;
Temp. %
(mm) (cm3/sec) (“C) (hr/cm%) (cmz/sec) (mm W (cmz/sec)
6 0.94 30.0 0.957 0.105 780 0.168
6 l-10 30.0 0.903 o-111 760 0.111
6 2.10 30.3 ow70 0.113 767 0.114
6 0.42 30.3 0.947 o-105 781 0.108
8 1 ,ar 30.3 O-825 6.120 752 O-118
4 0.53 30.5 0.830 O-118 753 0.116
8 1.41 so*4 0.878 0.110 764 0.111
The evaporation of two-component liquid mixtures

It will be seen that the results do not vary by In a number of the experiments in the 6 mm
more than f 2 per cent from the mean value of tube, the intercept on the t/(h - h,) axis was
0*112 cm2/sec, and it is therefore concluded that measured, and the effective position of the top of
neither tube diameter nor air rate has a critical the region of molecular diffusion was determined.
effect for the ranges investigated. The succeeding
experiments were all carried out in the 6 mm tube, TWO-COMPONENT MIXTURES-VOLATILE LIQUID
with an air flow of about 1 cm3/sec. The variation LESS DENSE
of diffusivity Dr with temperature B is seen in
The evaporation of a mixture of equal volumes
Fig. 3, and the diffusivities of acetone and carbon
of acetone and dibutyl phthalate in the 6 mm tube
0.14, 1 I was studied at 42°C. Measurements were taken in
the same way as for single component liquids.
Dibutyl phthalate can be considered as entirely
non-volatile (vapour pressure at 42°C is
2 x 1O-4 mm Hg), and on the assumption that the
mixture is ideal, the relation between the partial
pressure at the interface and the concentration of
acetone in the liquid at the interface is given by
Raoult’s Law. As the residue is considerably
denser than the evaporating liquid, it will be
assumed that it mixes completely with the bulk of
the liquid which will therefore be of uniform
I I I I I I I I composition throughout at any given time, so that
Q’s10 20 30 40 50 60 70
the partial pressure of acetone vapour at the inter-
Temperature, “C
face, and hence the evaporation rate, can be cal-
FIG. 3. Diffusivities of acetone and carbon tetrachloride in
air as function of temperature. culated in terms of the amount of evaporation
which has taken place up to that time. These
tetrachloride vapours in air are given to within various assumptions will be justified if the meas-
approximately 2 per cent by the relations : ured and calculated amounts of evaporation
Acetone D, = 0.087 + O-00086 8 (5) correspond during the whole course of the experi-
Carbon ment. The calculated amount of evaporation is
tetrachloride D, = 0.076 + 0.00032 B (6) obtained as follows.
After time t, the number of moles of the more
GUSH [5] obtained a value of O-111 cm2/sec for volatile component which have evaporated
the diffusivity of acetone for temperatures
= (h - h,) pv A. Thus, if n, and n, are the num-
between 30 and 50°C. bers of moles of the volatile and non-volatile
No values for the diffusivities are quoted in the components initially, the mole fraction of the
literature, but approximate values have been
fir - (h - h,) POA
calculated from Gilliland’s equation and compared volatile liquid at time t =
n, + n2 - (h - h,) p,A’
below with those calculated from equations (5) The molar concentration of vapour at the surface
and (6) at 42°C the temperature at which the of the liquid is therefore
experiments with 2-component mixtures were
carried out. n, - (h - ho)~1, A
C” = cfxl
(n,+n,-(h--h,)p,A (7)

From equations (5) and (6) O-1 23. 0490 cm2/sec where C,, is the corresponding value for the pure
From Gilliland’s equation 0.113 0.081 cm2/sec
volatile liquid at the same temperature.
An isolated measurement for water at 46°C gave Now the rate of diffusion through the gas phase
a value of 0.258 cm2/sec. is strictly given by equation (1). However, the


ratio P/P,, was equal to 1.18 at the commence- between t and h (in hours and centimetres respect-
ment of the experiment, and fell rapidly during the ively) becomes :
initial stages to a value of approximately unity. t = Oq641h2- adoh + 35.6 -
Since an algebraic relation between h and t can be
- 413 log,, (756 - h) (10)
obtained only when the ratio is unity, this sim-
plification was introduced, because it was con-
sidered that the resulting errors would be less
than those introduced by a graphical solution of
the problem.
Thus, from Fick’s Law,

dh f), Go n, - (h - ho)~0 A X Experimental points from 2 runs

pV~= (8) E 5 a Calculated points
nl + n2 - (h -44 PCA> 0

Rearranging, this gives

lit _&Lh--&-I)
dh D,C, IJo0

On integration :
f, hr
t = $jE; (he - ho%)- ~
Do;:o A (h - h,) - PIG. 4. Experimentaland calculated resultsfor evapor-
ation of equivolume mixtures of acetone and dibutyl
phthalate at 42°C.

The experiments were carried out at 42°C with In Fig. 4 is shown the ex~~mentaliy deter-
mixtures consisting initially of equal volumes of mined relationship between h and t together with
dibutyl phthalate and acetone. Under these that calculated from equation (10). The agree-
conditions, the values of the quantities in equation ment is seen to be within 4 per cent over the whole
(9) were as follows : range, with the theoretical curve slightly below
the experimental one. This small difference could
0,764 be attributed to small inaccuracies in experiment
PO= x = 0.0132 g moles/cm3
or to the assumption that P/P,, is equal to unity.
D, = 0,123 cmz/sec = 443 cm2/hr (from equa-
n, = 2 x 0.764 = 0.0263 g moles. DENSER
58 When the volatile liquid is denser than the
non-volatile component, the residue after evapor-
n2 = 2 x y7G = 0.00753 g moles. ation is of relatively low density and therefore
shows little tendency to mix with the bulk of the
A = 0.312 cm2
liquid. Under these circumstances, therefore,
h, = 1.15 cm diffusion must take place through the bulk of the
c ~ E y, EC!y, l liquid phase before further evaporation can occur..
760 315 22,400 Since the capacity of the vapour phase is very
= 2.32 x 1O-5 g moles/cm3. much less than that of the liquid phase, it will be
assumed that in the gas phase the rate of transfer
On substitution and simplification, the relation is proportional to the overall concentration

The evaporation of two-component liquid mixtures

gradient and that in the liquid phase the transfer is intervals. This procedure was necessary because
an unsteady state diffusional process. The liquid it was not found possible to express C, as a simple
depth was assumed to be effectively infinite. algebraic function of t.
Measurements of the rate of evaporation of a The solution of equation (14) is given in Appen-
mixture of equal volumes of carbon tetrachloride dix 2 as:
and dibutyl phthalate in the 6 mm tube at 42”C,
should then enable the liquid phase diffusivity, D,
to be calculated. If the calculated value of D
over the whole duration of the experiment remains
constant the above assumptions concerning the
mechanism of the process would appear to be
justified. (for t > t,) (15)
The rate of diffusion in the gas phase is given
At time t, the molar rate of evaporation per unit
by :
area is given by
o,c, (11)
= P”Z (16)
where C, is the molar concentration of vapour at
the liquid surface at time t. If the mixture is ideal, Then from equations (13), (15) and (16), at time t,
Raoult’s Law will be applicable and the molar
concentration of the volatile component in the (h WWn-, - (h WW,,,
liquid at the surface C, will be proportional to l/q 1
the concentration in the vapour.
(hW’Wo - (hWW, = p fj
i.e. C, = kc,, (12)
6 1 ’ dt

From equations (12) and ,(I3),

Thus C, = ‘F. h f = bh g (say) (13) k = CL/C, for pure carbon tetrachloride,
and b = k p,/D,.
For diffusion in the liquid : Under the conditions of the experiment-atmos-
3C iJ2c pheric pressure and WC-the vapour pressure

(14) of carbon tetrachloride is 251 mm Hg, so that

3t Y l-54/155
where C is the molar concentration of the volatile k= = 780
l/22,400 x 273/315 x 251/760
liquid at a depth y below the surface.
This equation must be solved to give the con- and b = ‘g4 = 2.4 hr/cm2
centration gradient at the surface for the following PO
boundary conditions : [D, = 0.090 cm9/sec = 324 cma/‘hr from equation
t F 6, 0 < y < co, C = C, (uniform constant value) (6)l
Substituting in equation (17),
t>o, y= 00, E=o
y=o, t=t,, c=c,

t=t, c=c, (h dh/dt),_, - (h dh/dt),+,
t=t, C=C, etc. l/t - t,

where C,, C, . . . . C,, are values of C, calculated + @ Wdt)o - (h d4’4 -’ (18)

from equation (13), using the experimentally df

determined values of h and dh/dt at regular time (units of centimetres and hours) (t > t,)


0 200 400 600 800 1000 120014001600 1800 2OOO22002400
FIG. 5. Experimental figures for the evaporation of equi-volume
mixture of carbon tetmchloride and dibutyl at 42%

tration was starting to change rapidly at the
c!$! 2 (say)
= 2.2
i 1 (1% bottom of the tube and it was no longer justifiable
to consider the liquid depth as infinite. A rough
Ex~rimen~l measurements of a mixture of
calculation has shown that the concentration at
equal volumes of carbon tetrachloride and dibutyl
the bottom of the tube would have changed by
phthalate were made over a period of 4400 hr and
about 20 per cent after 1000 hr. The mean value
values of a were calculated at various times from
of (dh/dt)/a over the period was O-113 hr*/cm.
the experimental results. The results are given in
Hence the mean value for the diffusivity for the
Fig. 5 and Table 2,
carbon tetrachloride - dibutyl phthalate system
was given by equation (19) as 0.028 cm2/hr.
Table 2. Calculation of liquid phase diffusivity for
carbon tetrachloride in dibutyl phthalate CONCLUSIONS
The WI~K~L~ANN method has been used with
Time, t dh/dt (I
varying experimental conditions to determine the
W (cm/hr) (cmZ/hr1,5) CT
diffusivity of a vapour in air. It has been con-
10 0*0095 0.138 0.069
cluded that the results obtained are substantially
20 o@l70 0.0943 O-074 independent of both air rate and tube diameter,
50 0%)060 0.0585 0.103 over the range investigated. Values of diffusivities
100 0.0052 o-041 0.127 have been obtained for acetone, carbon tetra-
500 0.0022 0.0204 0.108
chloride and water vapours to within an estimated
1000 O-0016 o-0133 0.120
2606 0.0011 o+m94 0.117 accuracy of & 2 per cent.
2660 0~00085 0.00324 0.103 The evaporation of a mixture of acetone and
3Oocl omo72 0.00768 0.094 dibutyl phthalate has been studied in the same
4mO 0*06652 O-30688 0,076 apparatus and it has been shown that the heavy
- residue sinks and completely mixes with the
remaining liquid. The levels of the liquid surface
Over the period from 50 to 2600 hr, the value of calculated on this assumption were found to agree
(dh/dt)/a remained substantially constant. For closely with the experimental values throughout
lower values of time, the gradient dh/dt was the course of the experiment.
changing very rapidly so that accurate measure- When a mixture of carbon tetrachloride and
ments were difficult to obtain. After more dihutyl phthalate evaporates, the evaporation rate
prolonged periods than about 2600 hr, the concen- falls off very rapidly with time because the residual

The evaporation of two-component liquid mixtures

liquid is of a lower density and therefore remains

close to the surface. The value of the liquid phase SOLUTION OF THE LIQUID PHASE DIFPLWION
diffusivity has been calculated at various times EQUATION
during the process, on the assumption that the 2
amount of mixing in the liquid due to convection =lIbZ (14)
is negligible and that the transfer process in the
liquid approximates to that which would occur When t = 0, C = C, o<y<a,
under conditions of molecular diffusion in a liquid Putt = c - c,
of infinite depth. The value of the diffusivity so 3C 2%
Then z=Day2 (20)
calculated was constant for a period of over 2500 hr.
At the beginning of the experiment, accurate
and when t = 0, c = c,, = 0. (O<Y< a)
values were difficult to obtain because of the rapid
3c iJc
change in the evaporation rate. Towards the end Wheny=co, -=-=0 (t>O)
of the experiment, low values of diffusivity were 3Y SY

obtained because the concentration was changing When y = 0, C=Crandc=cratt=t,

appreciably in the liquid at the lower end of the C = C,, and c = cn at t = t, etc.
tube. Mean value of c over time interval tn_1to tn = * (cnel + co)
In this study, the behaviour of two simple ideal = an (w).
systems has been elucidated.

haviour of the more complex mixtures which are

normally encountered in the petroleum
gations are needed in order to determine the be-


Taking Laplace Transforms of both sides of equation (20),
e-N dt = D
e-Pt dt

0 0
chemical industries, both under conditions of
natural evaporation and during the combustion :. (_ )-=+p-=De
Ct 0
of the liquid in bulk.
Acknozoledg<ments-The author is indebted to Professor
D. M. NEVITT for providing facilities for the experimental F = B, e~d~/D + B2 e-~d/b/D
work. :. (21)

and? = B ~~D/dp’D - - B, dpfi e-y’p’D (22)

>Y 1
When y = co, -=OandEi:O
:. B, = 0
Carbon tetrachloride
:. wheny=0,s=B2,
Density (g/cm3)
= 1.63255 - 1.9110 x 1O-30 - 6.90 x 10-‘02 and Fy = - B,, l/p/D
Vapour pressure (P, - mm Hg) is given by :
logr, P, = 79.875 - -F Thus B, = c e-Pf dt
% ‘n
Density (g/cm3)
= 0.81248 - 1.1 x 1O-3,9 - 8.58 x IO-‘@ ‘= fale-Ptdt+j a2 e-Pt + . . . . + an e-pi dt + . . .
Vapour pressure (P, - mm Hg) is given by : 0 11 'n-1

1660 _ 3 (1 _ C-Ptl) + z (e-Ptl - e+e) + .. ..

log,, P, = 7.930 - 7

Dibutyl phthalate + 5 (e-P&* - e-P%) + ....

Density (g/cm3) at 42°C = 1.048


=- : aI + $ e-PI1 (as - a,) + . . . .
A = cross-sectional area of evaporating surface
a, = mean value of c over internal ta_l to t,
B,, B, = integration constants
6 = k&D,
= $ (co -j- cl) + $ e-P*1 (ca - co) +. . . C = molar concentration of volatile liquid at time t,
depth y below surface
Co = uniform constant value of C at t = 0
f 2.
e-P’n(c,+, - CrI) -I- . . . .
CL = molar concentration of volatile liquid at surface
C, = value of C at t = 1,


in equations (21) and (22) gives

(C,+ 1 -

1 C, = molar concentration of vapour at liquid surface
Cvo = molar concentration of vapour at surbce of pure
C oT = total molar concentration of gas and vapour
D = liquid phase diffusivity
- D,, = vapour phase diffusi~ty
- C& e-y’p’D (25) h = distance over which vapour is diffusing at tima t
I ho = initial value of h
_ (C, - Co) ,--PdP/D -
k = Henry’s constant CL/CV
and !!? = _ ’
2l/~D N = molar rate of transfer per unit area
fil = number of moles of volatile component initially
eWPta(Cn+l - C& e (24) n2 = number of moles of non-volatile component
-I initially
Taking the inverse transform [4] of equiltion (24), P = total pressure
P, = partial pressure of air
dC 1
-=- B (C1 - CO) e-Y*/*~~ - P Bm = logarithmic mean ptlrtial pressure of air
3Y 22/&t P,, = vapour pressure

72=7$ 1 P = parameter in Laplace transformation

-r: (%+1- C*+) e-Y’/4D(tU s = slope of curve in single component experiments
n-1 22/wD (t - t,J T = absolute temperature
(25) t=time
When t = t,, the corresponding term is zero for all value y = distance below surface
of 8. p = moIar density of single component liquid
Thus, when y = 0, pu = molar density of more volatile component in
liquid form
o = function of h, t, dh/dt as defined by equations
(18) and (19)
(fort > tn) (26) 0 = temperature in degrees centigrade

STEFAN J. Wien Ber. II 1873 68 385.
WINKELMANN A. Ann. Phys. 1884 22 1, 154 ; 36 1889 93.
LIXE C. Y. and WILKE C. R. Indwtr. Engng. Chm. 1954 46 2881.
MICKLEY H. S., SHERWOODT. K. and &ED C. E. Applied Mathematics in Chedcal h’ngineer-ing p. 284. McGraw
Hill, New York 1957.
GUSH L. L. Trans. Inst. Ckem. Engrs. 1958 26 142.