Ckmical kgimzerimg Science. Vol. 41, No. 9, pp. 2309-2319, 1986. WKE-2509186 53.00 + 0.

00
Printed in Great Britain. Pergamon Journals Ltd.

GAS DESORPTION FROM LIQUIDS: MASS TRANSFER AND

DRAG COEFFTCIENTS FOR SINGLE BUBBLES IN FREE
RISE THROUGH NEWTONIAN LIQUIDS

H. ISHIKAWA,+ T. MIKI, M. OKAMOTO and H. HIKITA

Department of Chemical Engineering, University of Osaka Prefecture, 804 Mozu-Umemachi 4-cho, Sakai,
Osaka 591, Japan

(Received 24 May 1985; in revised form 23 September 1985; accepted 25 November 1985)

Abstract-Small CO, bubbles have been generated in quiescent water and 8.0 and 21.0 wt.:/ aqueous
sucrose solutions supersaturated with CO*. Measurements have been made of the free rise velocity and
growth rate of the bubbles in the range&l 170 pm in diameter by a photographic method. The experimental
results for the COs-water system agree well with theoretical and empirical equations for rigid spheres in
creeping flow when d, is less than about 15Oj1m and begin to approach the equations for partially
contaminated or clean fluid spheres as d, or Re increases. Similar results are obtained for CO1 bubbles in
aqueous sucrose solutions. The critical Reynolds number at which the bubble surface begins to move
decreases with increase in liquid viscosity and this can be interpreted in terms of the stagnant cap model. As
Re further increases, experimental results agree well with the predictions for spherical bubbles free from
surfactant in potential flow before finally beginning to deviate from them.

INTRODUCTION for the bubble desorption rate. Thuy and Weiland

Gas is desorbed, when the concentration of the (1976) and Weiland et al. (1977) investigated the
dissolved gas in the bulk of a liquid is greater than that desorption of CO2 from supersaturated water sol-
at the surface. If the difference between the partial utions under agitated liquid conditions using a string-
pressure at the surface and the partial pressure of gas in of-spheres column and a ballled agitated vessel. They
equilibrium with the bulk liquid (in other words, the found that bubbling always occurred when the partial
degree of supersaturation) is modest, the gas will be pressure of the dissolved CO2 exceeded the system
desorbed by ditfusion from the liquid free surface in a pressure and showed that the desorption rate of CO2
way analogous to gas absorption (quiescent desorp- increased rapidly with increase in relative supersatur-
tion). However, if the degree of supersaturation is ation. Hikita and Konishi (1984) measured the desorp-
large, bubbles will form in the interior of the liquid and tion rate of CO2 from supersaturated water solutions
much of the gas will be released by diffusing from the in a baffled agitated vessel. They distinguished between
surface of the bubbles (bubble desorption). Thus, the contribution of desorption through bubble surface
bubble desorption is a process very different from generated in the liquid and desorption through free
absorption processes in which the area of surface surface and correlated the volumetric mass transfer
available for mass transfer is determined by external coefficients for each component. Pasiuk-Bronikowska
factors (Danckwerts, 1970). and Rudzinski (1980, 1981) applied the population
Supersaturation of gas in the liquid phase results balance approach to the bubble desorption of Nz
from depressurization, rise in temperature or chemical produced by the reaction between H3NSOJ and
reaction. Therefore, bubble desorption is quite HN02 in aqueous solutions under agitated liquid
common in industrial practices such as air flotation of conditions. They showed that the experimental time
solids (Takahashi et al., 1979; Ress et al., 1980), coal variations of desorption were simulated well with a
pyrolysis (Attar, 1978), decarbonization and vacuum model in which quiescent desorption and nucleation,
degassing in metallurgical processes (Szekely and growth and escape of bubbles were all taken into
Martins, 1969), CO2 desorption in fermentation pro- consideration. Unfortunately, assumptions for the
cess (Aiba et al., 1973), and regeneration of spent bubble growth and escape of bubbles from the liquid
absorbents in gas purification processes (Shah and phase on which the model is based are questionable.
Sharma, 1976). As mentioned above, past experimental investi-
A number of investigations of bubble desorption gations for bubble desorption were mainly concerned
have been conducted, mainly in terms of the rate of with the overall rate of desorption or the volumetric
desorption from supersaturated solutions. Burrows mass transfer coefficients and none were concerned
and Preece (1954) and Kamei et al. (1953) have carried with the mechanism of bubble formation or nucle-
out experiments on the desorption of gas on air and ation. The population balance technique is considered
He, and C02, respectively, from quiescent super- to be an effective tool in the fundamental study of the
saturated liquids and presented empirical correlations bubble desorption mechanism, but there is much to be
done before application. For example, investigations
for mass transfer and drag coefficients of bubbles have
‘To whom correspondence should be addressed. been carried out extensively (reviewed by Clift et al.,
CES 41:9-H 2309
2310 H. ISHIKAWA et al.

Fig. 1. Schematic diagram of experimental apparatus. (1), (2) CO, cylinder, (3) pressure regulator, (4)
pressure vessel, (5) sparger, (6) safety valve, (7) pressure gauge, (8) water vapor saturator, (9), (17) gas flow
meter, (10) motor and cam, (11) vibrating rod, (12) observation cell, (13) capillary (inlet of supersaturated
liquid), (14) lamp house, (15) stroboscope, (16) manometer, (18) gas analyser, (19) constant temperature room.
VI-VL2. valves.

1978), but there is little information available on Side view Front view
motion and mass transfer of very small bubbles less Ian 78
than 1 mm in diameter, which play an important role in
the bubble desorption process.
As a fundamental study to clarify the mechanism of
bubble desorption, we measure in this paper the mass
transfer and drag coefficients for single CO2 bubbles in
free rise generated in quiescent supersaturated water
and aqueous sucrose solutions with diameters from 60
to 1170pm. t 1

Fig. 2. Schematicdiagramof the observation cell and camera

EXPERIMENTAL
system for observation of bubbles less than about 400 pm. (1)
Apparatus camera, (2) microscope, (3) observation cell, (4) inlet of
Figure 1 shows a schematic diagram of the exper- temperature regulated water, (5) outlet of temperature re-
gulated water, (6) inlet of supersaturated liquid, (7) inlet of
imental apparatus which can be divided into two parts.
CO1, (8) outlet of COz, (9) vibrating rod, (10) motor and cam,
One is for the preparation of aqueous solutions (11) outlet of supersaturated water. VI, valve.
supersaturated with COz and is similar to that used in
previous work (Hikita and Konishi, 1984). The main
part was a 0.09 m3 pressure vessel made of steel of
which the inner wall was coated with epoxy resin, and
was equipped with a perforated plate gas sparger in it.
The other is for the generation and observation of CO2
bubbles. In this figure, an apparatus is shown for
observing bubbles of less than approximately 400 pm.
Figure 2 shows a schematic diagram of the ap-
paratus for the observation of bubbles of less than
about 400 Hrn in diameter. The observation cell is a
plexiglass box of 1 cm width, equipped with a vibrating
rod to generate CO1 bubbles and with a water jacket in
which temperature regulated water was circulated to 1
keep the temperature of the supersaturated liquid at
303 K. Photographs of bubbles were taken with a
camera (Olympus PM-6) mounted to a microscope
(Olympus SZ-Tr2). Illumination was by a flash illumi-
nation system (Sugawara PS-240), duration 25 ps and
interval 3.8-100 ms. The flash lamp was placed to face
1
the side wall of the cell to illuminate bubbles at a right Fig. 3. Schematic diagram of the observation cell and camera
angle to the axis of the lens. By the use of the system for observation of bubbles greater than about 400 pm.
illumination system, a number of images of a particular (1) observation cell, (2) vibrating rod, (3) wafer jacket, (4)
bubble can be captured on one frame of film. motor and cam, (5) inlet of supersaturatedliquid, (6) outletof
supersaturatedliquid, (7) inlet of temperature regulated
Figure 3 shows schematically an apparatus for the
water, (8) outlet of temperature regulated water, (9) inlet of
observation of bubbles greater than about 400 pm in COz, (10) outlet of CO2, (11) camera A, (12) camera B, (13)
diameter. The cell consists of a plexiglass column of stopwatch, (14) stroboscope.
 Gas desorption from liquids 2311
5 cm i.d. and 1 m height. The column was also without changing the solution. As the time required for
equipped with a vibrating rod on the bottom of the cell the above process was short, the change in the
and with a water jacket with a square cross-section to concentration of CO2 in the liquid was negligible.
keep the temperature of the supersaturated liquid at Observations were made at several different heights
303 K and to avoid distortion of the bubble image. above the vibrating rod. The lowest and the highest
Two 35 mm cameras were used to photograph bubbles heights of the camera measured from the rod were
at two different positions about 15 cm apart. The first 0.3 cm and 15 cm, respectively. Before each observa-
camera took the first picture of the bubbles which tion at a new position, the supersaturated solution was
passed an appropriate mark on the side of the column. replaced in the pressure vessel.
When the shutter button of the camera was pressed, an At least three images of a particular bubble were
electric stopwatch started to measure time. Then, by captured on each frame. The apparent diameter dB’ of
pressing the shutter button of the second camera, the each bubble and apparent distance L’ between ad-
image of the same bubble taken by the first camera was jacent bubble images were measured directly by a
captured. When the shutter of the second camera was reading microscope. Actual diameter d, and distance L
opened, the stopwatch stopped measuring time. Thus, were obtained as d,‘/w and L’lo, respectively, where w
the time required for the bubble to rise from one mark is the magnification defined by a ratio of apparent
to the next was measured. The actual distance between length on the film to actual length.
two positions where bubble images were captured by The bubble rise velocity was determined by
the two cameras was a little different than that between
u = Lfr. (1)
the two cameras. Therefore, correction for the distance
was made as explained later. While the distance The drag coefficient was calculated as
between the two cameras was kept constant at 15cm,
they were moved vertically to appropriate positions in C, = 4&+” 9/3u= (2)
order to observe bubbles of different sizes. where d, ay is an average bubble diameter given by
d B,av= (d,, + d,W- (3)
Procedure
Aqueous solutions supersaturated with CO2 were The mass transfer coefficient is defined by
prepared as follows: after the temperatures in the 1 dn
constant temperature room and the liquid reservoir k, = A (4)
became constant at 303 K, solvent, water or aqueous ndi,,(Ai -%) dt
sucrose solutions, of about 0.06 m3 was introduced where nA is number of moles of CO2 in the bubble and
into pressure vessel 4 through valve VS (valves V, can be calculated from
opened and Vf, Vs, V7 closed). Then, valve V5 was
closed and pure CO2 gas from cylinder 2 was fed nA = P,,VJRT (5)
continuously to the vessel through valve V2, while the P,, in eq. (5) is the partial pressure of CO2 in the
pressure of the gas phase in the vessel was kept bubble given by
constant at 0.29 MPa by means of valves Vz and Vq.
P AB = P,-P,
After about an hour and half, valves V7 and Vs were
opened and the aqueous solution supersaturated with = PO + p,gh -t4a/d,.,, -P,. (6)
COz was fed to the observation cell through valve V7
Differentiation of eq. (5) assuming constant tempera-
and capillary 13, while CO1 was being supplied
ture gives
continuously under the same condition mentioned
above. The feed rate of the solution to the cell was
(7)
controlled by valve VT_
For observation of bubbles less than about 400 m
From eq. (6), neglecting the change in the average
in diameter, we used the apparatus shown in Fig. 2. The
bubble diameter, the change in P,, with time can be
supersaturated aqueous solution from the pressure
related to the bubble rise velocity U as
vessel was fed to the observation cell through capillary
liquid inlet 6, and was drained through liquid outlet 11. dPAB _
--
Pure CO* from cylinder 1 was continuously supplied dt
from 7 and exhausted from 8. Feed of the liquid to the
cell was stopped after it was fed for a period equivalent Combining eqs (4), (7) and (8) gives
to 7-8 times of the average liquid residence time. The v,f%g”
dV,
liquid level was kept at 17 cm. Then, bubbles were
formed by starting the vibration of vibrating rod 9. At
k,= xd pAB
B,,‘(Ai-Ab)RT dt -P,, 1
*
(9)

least two to three images of bubbles were captured on The second term in the brackets is the correction
one frame by adjusting the flash interval. Vibration of necessary for the change in bubble size due to the
the rod was stopped just before taking photographs in change in hydrostatic pressure.
order to minimize the effect of liquid turbulence due to The mass transfer coefficient was calculated from eq.
vibration of the rod on movement and mass transfer of (9) by substituting the measured bubble rise velocity U,
bubbles. Usually five frames were taken at one position and the experimental bubble growth rate determined
2312 H. ISHIKAWA et al.

by using physical properties such as solubilities and diffusivities

of CO2 in aqueous solutions.
~dVi3= ri (4n3- 4u3l/67 (10) The solubility of CO2 in aqueous sucrose solution
dt
was calculated by the following van Krevelen and
and other experimental quantities such as ds+“, PAS, Hoftijzer empirical equation (1948)
etc. However, the values of the second term in the
brackets of eq. (9) were negligibly small in comparison log (A,/&) = -O.l48C, (12)
with those of the first term when the measurement was where Al and Ai, are the solubilities in the solution and
conducted in the apparatus shown in Fig. 2. in water, respectively, and C, is the concentration of
For observation of bubbles greater than about sucrose in solution. The numerical coefficient - 0.148
400 pm in diameter, we used the apparatus shown in in eq. (12) was determined by analysing the solubility
Fig. 3. The procedure was almost the same as that for data of Usher (1908) and Showalter and Ferguson
small bubbles. With each camera, five frames were (1936).
taken at an appropriate height from the rod without The value of Ai, was obtained from the partial
changing the solution. Observations were made at pressure P,, of CO, in gas phase by use of Henry’s law
several different positions. The lowest and the highest constant for the CO,-water system (Bohr, 1894).
positions of the lower camera was 5 cm and 65 cm The diffusivity D, of CO;! in water at 303K was
from the upper tip of the vibrating rod, respectively, obtained from the value reported at 298 K by
while the liquid level in the cell was kept 15 cm above Peaceman (1951) corrected for temperature and
the upper camera. Images of bubbles were captured on viscosity of water, according to the well-known
the two frames taken by the two cameras. Apparent Stokes-Einstein relation. Diffusivity of CO2 in
diameter d’, of particular bubbles on the two frames aqueous sucrose solutions was estimated from the
and apparent distances, 1; and l;, between the bubbles value in water by correcting for viscosity of the
and lower edge of the frame were measured by a solutions, according to the empirical relation (Hikita et
reading microscope. Actual distance between two al., 1978)
positions where two bubble images were captured was
then calculated by D,&13/T = const. (13)

L = 0.15-c (I; -&)/W. (11) The viscosity and density of water were taken from the
International Critical Tables (1963) and the Perry’s
The bubble rise velocity, the bubble growth rate, and Chemical Engineers’ Handbook (1963), and those of
the drag and the mass transfer coefficients were then aqueous sucrose solutions were measured by conven-
determined by the same method as mentioned above. tional methods. The surface tension of the three
However, the second term in the brackets of eq. (9) solvents were taken from the International Critical
could not be neglected in this case. Tables (1963). These physical properties are sum-
Liquid samples were taken from V, and Vs in Fig. 1 marized in Table 1.
and 6 in Fig. 3, and were analysed by either measuring
the CO, volume liberated in gas burette 17 in Fig. 1 or
Drag coeficients and mass transfer coeficients
by a titration method where a liquid sample of 10 cc
First we checked the shape of bubbles dealt with in
was discharged into a flask containing a mixture of
our experiments, because most theoretical equations
standard NaOH and BaCl,, and the excess NaOH was
for bubble motion and mass transfer have been derived
then titrated with standard HCl using phenoIphthalein
for spherical bubbles. Grace (1973) has given a con-
as indicator.
venient chart for determining the shape regime for
Solvents used were distilled water and aqueous
bubbles and drops in terms of the three dimensionless
sucrose solutions whose concentrations are shown in
numbers, i.e. Eotvos, Morton and Reynolds numbers.
Table 1.
Under the present experimental conditions, the maxi-
RESULTS AND DISCUSSION mum Eotvos number and the minimum Morton
Prediction of physical properties number were 0.191 and 1.11 x lo- ‘I, respectively, and
In order to analyse and correlate the experimental the bubbles in the present experiments were found to
results, it is necessary to know the values of the be complete spheres.

Table 1. Experimentalconditions and physicalpropertiesof systems used

Surface Solubility Diffusivity

Temp. Viscosity Density tension of co* of co2 Schmidt
Solvent (K) (kg/m s) (kg/m’) (N/m) (mol/m’) (m’/s) number

Distilled water 303 8.01 x 10-h 995.7 7.12 x IO-’ 28.5 2.24 x 10-g 364
8.0 wt. % aqueous
sucrose solution 303 9.85 x 10-b 1027 7.16 x lo-* 26.5 1.93 x 1o-9 49s
21.0 wt. y0 aqueous
sucrose solution 303 1.57 X 10-3 1098 7.22 x lo-’ 23.2 1.43 X 10-q 998
 Gas desorption from liquids 2313

By the use of the figures given by Clift et al. (1978), particle Reynolds number Re (= pcdsU/pc)_ The re-
the retardation of the velocity of the bubbles was sults of the mass transfer coefficients are reported as
estimated to be less than 2% under the present plots of Sherwood number Sh against Re in Figs 7-9.
experimental condition. Thus, the effect of natural
convection on the bubble rise velocity was neglected in
the present work. Transport models for bubbles
The effect of natural convection on heat transfer was In Figs 4-9, the theoretical equations for drag
analysed theoretically by Acrivos (1960, 1966). By coefficient C, and Sherwood number Sh are compared
analogy, the mass transfer coefficient due to natural with the experimental data.
convection is given by The gas inside a rising bubble usually circulates
because of the external liquid flow, so that a finite
Sh, = 0.5827(GrS~)“~ (14) velocity exists at the gas-liquid boundary. For creep-
where Gr represents the Grashof number for mass ing flow around a fluid sphere with its interface
transfer. In the present work, the ratio Sk/Sh,,, assumed to be completely free from surface-active
(= k,,,lk,,otJ ranged from 0.01 (large bubbles) to contaminants, Hadamard (1911) and Rybczinski
0.11 (small bubbles). The mass transfer coefficients (1911) independently obtained solutions to the
measured can be corrected for the contribution of mass Navier-Stokes and continuity equations. These lead to
transfer due to natural convection according to the the expression
relation (Acrivos, 1966)
C, = 16/Re (Re c 1). (16)
Sh = (Sh,,s4 + Sh.P)l14 . (15) For creeping flow around a fluid sphere with the
From eq. (1 S), the effect of natural convection on mass Hadamard-Rybczinski velocity distribution, Oellrich
transfer was also found to be negligible. et al. (1973) obtained the mass transfer coefficients
The effect of cell wall on the bubble rise velocity and numerically.
the mass transfer coefficient may be present in our The above theoretical equation and numerical sol-
experiments. The maximum ratio of the bubble di- ution hold for the creeping flow around a bubble at Re
ameter to the cell width or to the column diameter was -c 1. However, when the Reynolds number increases,
observed in the observation cell shown in Fig. 2 and the contribution of the inertia term in the
was about 0.04. The value was observed when the Navier-Stokes equation becomes considerable com-
Reynolds number ranged from 10 to 20, the value pared with the viscosity term. In the intermediate
depending on the system. In this range, the bubble Reynolds number region, Hamielec et al. (1962, 1963)
behaves as if it has almost no contamination, as applied the Galerkin method to fluid spheres, and
discussed later. According to Clift et al. (1978), the presented the correlation for the drag coefficient on
effect of the wall on terminal velocity is negligible in the spherical bubbles:
above range of Re and bubble-to-column diameter
C, = 13.75/Re0.‘4 (4 < Re < 100). (17)
ratio. Further, the effect of the wall on mass transfer is
considered to be negligible since the wall has a much In the region 10 c Re < 104, Lochiel and Calderbank
smaller effect on mass transfer than on drag. (1964) solved the thin concentration boundary layer
Figures 4-6 show plots of the drag coefficient C, vs. problem using surface velocity from boundary layer

10-
6-
4-
u”
2-
1. Stokes
1- 2. Hadamard-Rybczinski
6- 3. Levich
4. Hamielec et al.
4-

0.21 I III I I III I I III 1 I

0.2 4 6 1 2 4 6 10 2 4 6 100 2 400
Re

Fig. 4. Drag coefficients for the CO,-water system at 303 K.

2314 H. ISHIKAWA
et al.

100

6 Cot- 8.0 wt.% Aq. Sucrose Soh

4

10

u” 6
4

2 2. Hadamard-Rybczinski

I 4. Hamielec et al.

0::
0.t 2 4 6 1 2 4 6 10 2 4 6 100 200

Re

Fig. 5. Drag coefficientsfor the CO,-8.0 wt. % aqueous sucrose solution system at 303 K.

6- - 21.0 wt.% Aq. Sucrose Soln 1

4-

2-

IO-

U” 6-
4-

1. stokes
2- 2. Wamard-Rybczimki
3. Levi&
1- 4. Hamielei et al.

6-
0.41 b ’ I I I III
0.06 0.1 2 4 6 I 2 4610 2 4 6 100

Re

Fig. 6. Drag coefficients for the CO,-21.0 wt. % aqueous sucrose solution system at 303 K.

theory as known expression:

Sh = 1.13(1 -2.90/Re’~2)‘~2Pe’~2. (18) C, = 24/Re (Re c 1). (21)

In the limiting case of potential flow of Re -P cc, For the same flow, numerical and approximate sol-
where viscous effect is negligible compared with inertia utions for mass transfer are available. Clift et al. (1978)
effect, Levich (1962) derived the equation: proposed the following empirical equation which
correlates the numerical solution within 2 0/e-
C, = 48/Re. (19)
Sh = 1 + (1 + Pe)‘13 (Re < 1). (22)
The velocity distribution of Boussinesq (1905) leads to
the following equation for mass transfer: At intermediate and high Reynolds numbers, Ranz
and Marshall (1952) presented the following semi-
Sh = 1.13Pe112. (20) empirical equation for drops evaporating into gases:
When the bubble surface is covered with surface-
Sh = 2 + 0.6Re”2Sc1’3. (23)
active contaminants, the surface of a small bubble is
known to behave like a rigid surface. For creeping flow Equation (23) was originally obtained by experiments
around a rigid sphere, the velocity distribution cor- where Re > 2 and SC was around unity, but ap-
responding to Stokes flow (1880) leads to the well- proaches the theoretical equation for steady diffusion
 Gas desorption from liquids 2315

2
0 Present work
l Motarjemi -Jameson
100

6
4

5 2

10 1. Oellrich et al. _
2. Clift et al.
6 3. Ranz-Marshall
4 4. Boussinesa
5. Lochiel-Cal~erbank
6. Ueyama-Hatanaka
(TI”=4.6~10-~)

I
A.2 4I 6II I 21 4610
I III 2
I 4I 6,I1 100 2I 4

Re

Fig. 7. Mass transfer coefficients for the CO,-water system at 303 K.

“t co .,- 6.0 wt.% Aa. Sucrose Soln

0

2 46 1 2 4610 2 4 6 100 200

Re

Fig. 8. Mass transfer coefficients for the C02-8.0 wt. % aqueous sucrose solution at 303 K.

from a stagnant sphere, Sh = 2, as Re --* 0. Miyauchi partially covered with a stagnant layer of surfactant.
and Nomura (197 1) carried out an experiment on mass They derived a simple equation for the drag coefficient
transfer by using a single platinum sphere and showed on fluid spheres as
that eq. (23) could be extrapolated to low Re of order of
0.01 and high SC.
In systems of intermediate purity, the interface may
be partially covered with surfactants and there may be

1
2+3~
a non-uniform surface tension distribution. The effects -sin 34)+.2)2~
of surface-active contaminants on bubble motion and
mass transfer have been analysed by many investi- where 4 is the cap angle. Equation (24) reduces to eq.
gators (Savic, 1953; Levich, 1962; Harper, 1973; (16) when 4 = 0 (no surfactant) and to eq. (21) when
Agrawal and Wasan 1979). Sadhal and Johnson (1983) $J = rr (surface completely covered with surfac-
analysed the creeping flow due to the motion of a fluid tant). Thus, eq. (24) expresses the relation between
sphere in another immiscible fluid when the interface is the drag coefficient on bubbles with surfaces par-
2316 H. ISHIKAWA et al.

3.
4.
5.
1. Lochiel-
Oellrich
Boussinesq
Ram-Marshall
Caldertmnk
et al.
2. Clift et al.

6. Ueyama Hatanaka
,__?, ...n-5.

Re

Fig. 9. Mass Transfer coeficients for the CO,-210 wt. % aqueous sucrose solution system at 303 K.

tially covered with contaminants and the cap angle. experimental values decrease with Re in the same way
Ueyama and Hatanaka (1973, 1976) extended the as predicted by eq. (17), which is derived for a spherical
stagnant cap model for a contaminated fluid sphere at bubble with its surface completely free from surfact-
low Reynolds numbers and presented an equation for ant, but are somewhat larger than the predictions. As
mass transfer Re increases further, the experimental drag coefficients
come to agree with the Levich equation, eq. (19). and
Sh = 2 + 0.6~Re’1Z Scrf3 (25) finally begin to deviate from it.
Similar results were obtained for CO1 bubbles in
where j? is the ratio of the local mass transfer coefficient
aqueous sucrose solutions as shown in Figs 5 and 6.
at the front stagnation points of fluid spheres to that of
These figures show more clearly that the experimental
equivalent rigid spheres, and can be approximately
drag coefficients agree well with those predicted by the
expressed as
Hadamard-Rybczinski equation, at small Re.
jl= 1.13Pe’/6f’/2 (1 +0.717B7/g)3/7 (26) These results for the drag coefficient suggest that
when bubbles are very small, their surfaces are fully
where B is a correction factor for the effect of the
covered with surface-active contaminants and behave
interfacial velocity gradient on the velocity profile
like those of rigid spheres, and that the degree of
given by
contamination decreases with an increase in bubble
I3 = q/(8&f 3)1f2. (27) diameter or in bubble rise velocity. Similar results were
The parameters f and q in eqs (26) and (27) are the reported by Garner and Hammerton (1951) who
dimensionless interfacial velocity and the dimension- measured the bubble rise velocities in glycerol, white
less velocity gradient, respectively, at the front stag- hydrocarbon and water. According to their results,
nation point and are approximately expressed by drag coefficients for air bubbles in glycerol agreed with
the Stokes equation when bubble diameter was small,
f= 0.725(1 -A*/3)“’ (1 +K) (d* > 2.5) (28) and then deviated to reach the Hadamard-Rylxzinski
f = 0.5(1 -0.025j1*3)/(1 + K) (A* < 2.5) (29) equation, eq. (16), as bubble diameter increased. In the
present experiments, both the change of the flow
q = 2f+3K/(l+K) (30) regime from the creeping flow to the potential flow and
where A* is a dimensionless interfacial pressure at the the change of the degree of surface contamination
rear stagnation point of the fluid sphere defined by (from heavily to partially contaminated) are consid-
ered to have occurred simultaneously with the increase
A.* = lI*//&u . (31) in bubble diameter.
The Sherwood numbers for the CO1-water system
Discussion shown in Fig. 7 agree well with the Ranz-Marshall
As shown in Fig. 4, the experimental drag coef- equation, eq. (23), when Re is less than about 3. The
ficients for COs bubbles rising in quiescent water agree reasons why the experimental results of Sh in such a
with the Stokes equation, eq. (21), for rigid spheres low Re region as Re -K 1 were lower than the theor-
where Re is small, and approach the Hadamard- etical values given by eq. (22) and came to agree with
Rybczinski equation, eq. (16) at around Re = 3. the Ranz-Marshall equation are not clear. Further
In the intermediate Reynolds number region, the theoretical and experimental investigations are needed
 Gas desorption from liquids 2317
to clarify the mechanism of mass transfer in this low Re COz-water, and - 8.0 and - 21.0 wt. y_ for the
region. aqueous sucrose solution systems, respectively.
As Re exceeds 3, Sh begins to deviate sharply from Further, the surface pressure at the rear stagnation
eq. (23), and finally agrees with the Boussinesq equa- points for the CO*-aqueous sucrose solution systems
tion, eq. (20), or the Lochiel-Calderbank equation, eq. is 2.4 x low5 N/m, one half that for the CO,-water
(18), when Re > 20. This dependence of Sh on Re is in system.
accordance with that of C, on Re mentioned above, Let us try to interpret these experimental results by
that is, CO2 bubbles in water are contaminated with the stagnant cap model. According to the mode1
surfaceactive impurities almost completely and behave (Saville, 1973; Ueyama and Hatanaka, 1973; Sadhal
like rigid spheres when Re < 3 (corresponding to and Johnson, 1983), the cap angle or the extent of
bubble diameters less than about 150 pm). contamination depends only on 1* in the Stokes flow
Furthermore, the C, results revealed that when 3 region, with the behaviour of bubbles deviating from
< Re < 20, bubbles were partially contaminated, the that of solid spheres at the same R* value irrespective of
degree of contamination depending on Re, and the system. In the Stokes flow region, the terminal velocity
flow regime changed from creeping flow to potential of a spherical gas bubble (no circulation) is given by
flow. The solid line 6 in Fig. 7, which was calculated
from Ueyama-Hatanaka equations, eqs (25)-(3 l), cor- u = dg=pcsll&, (32)
relates well with our experimental results for 3 < Re Equation (3 1) and the equality of the 1* value yield for
< 100, if we assume a constant value of 4.8 two liquid systems the relation
x 10s5 N/m for the parameter, II*, which is the
surface pressure at the rear stagnation point. Such
agreement between the Ueyama-Hatanaka equations
and the present experimental results for the wide
By use of eqs (32) and (33), the ratio of the critical
Reynolds number region may be because their analysis
Rey;lolds numbers for two systems is
was conducted originally in the creeping flow region.
In Fig. 7, experimental mass transfer coefficient for
2 = (~>““(~>“‘(~>‘. (34)
the O,-water system measured by Motarjemi and
Jameson (1978) are also plotted for comparison,
For the 8.0 and 21.0 wt. % aqueous sucrose sol-
though the Schmidt number differs a little from that of
utions system, eq. (34) predicts:
the COz-water system. The results at Re of about 10
agree well with the Ranz-Marshall equation, eq. (23), Re,2 Re,(2 1.O wt. %)
and deviate from it to approach the Boussinesq and the
Lochiel-Calderbank equations, eqs (20) and (IS),
respectively, as Re increases. The Reynolds number at Rec1
which the values of Sh measured by Motarjemi and
Jameson begin to deviate from the Ranz-Marshall
The value of 0.41 agrees well with the experimental
equation is about 15, considerably larger than that of
value of 0.43. Furthermore, for the water and 8.0 wt. %
the present experiments (Re N 3). Furthermore, their
aqueous sucrose solution system, the ratio is estimated
values of Sh are smaller than ours at similar Re values.
as
These discrepancies cannot be attributed to the dif-
ference in the direction of the mass flux (Bird et al.,
1960), and probably occurred due to a different degree
of contamination of the water; Motarjemi and
Jameson used tap water while we used distilled \;ater.
The experimental results for the dependence of Sh
on Re for the Cot-aqueous sucrose solution systems This value is in good agreement with the experimental
shown in Figs 8 and 9 are similar to those for the value of 4.3. Thus, we consider that the stagnant cap
CO,-water system. When Re < 1, C, shown in Figs 5 model can interpret the experimental results for mass
and 6 indicates that bubble surfaces are almost free transfer.
from surfactants. However, almost all the data points
of the Sherwood number for Re < 1 fall between the Acknowledgments-The authors would like to thank Dr.
numerical solution obtained by Oellrich et al. and eq. Hatanaka and Dr. Ishimi for very helpful discussions.
(22) or the Ram-Marshall equation, eq. (23), indicat-
ing that the bubble surfaces are not completely free
from surfactants. These experimental results are
reasonable, because the effect of surfacecontamination NOTATION
on mass transfer is more sensitive than on bubble A concentration of CO2 in solution, mol/m’
motion. Ai solubility of CO2 in solution, mol/m3
The critical Reynolds numbers at which the CD drag coefficient, dimensionless
Sherwood number begins to deviate from the Ranz- C, concentration of sucrose in solution, mol/m”
Marshall equation are about 3, 0.7 and 0.3 for the d, bubble diameter, m
 2318 H. ISHIKAWA et al.

D, diffusivity of COz in solution, m2/s REFERENCES

Eo Eotvos number, = dB2Apg/o, dimensionless Acrivos, A., 1960, A.1.Ch.E. J. 6, 584.
f dimensionless surface velocity at front stag- Acrivos, A., 1966. Chem. Engng Sci. 21, 343.
nation point, dimensionless Agrawal, S. K. and Wasan, D. T., 1979, Chem. Engng J. 18,
215.
9 acceleration of gravity, m/s2
Aiba, S., Humphrey, A. E. and Millis, N. F., 1973, Biochemical
Gr Grashof number, = dB3pcApg/p:, dimen-
Engineering, 2nd edn. University of Tokyo Press, Tokyo.
sionless Attar, A., 1978, A.I.Ch.E. J. 24, 106.
h liquid depth, m Bird, R. B., Stewart, W. E. and Lightfoot, E. N., 1960,
liquid-phase mass transfer coefficient, m/s Transport Phenomena, p. 675, John Wiley, New York.
k,
Bohr, C., 1894, Ann. Phy. 68, 500.
1 distance between a bubble and a lower edge of
Boussinesq, J., 1905, J. Math. 6, 285.
frame on film, m Burrows, G. and Preece. F. H., 1954, Trans. Inst. Chem. Engrs
L distance that a bubble travelled in a given time 32, 99.
period, m Clift, R., Grace, J. R. and Weber, M. E., 1978, Bubbles, Drops
M Morton number, = pc4g/pca3, dimensionless and Particles. Academic Press, New York.
Danckwerts, P. V., 1970, Gas-Liquid Reactions, p. 264.
P pressure, MPa
McGraw-Hill, New York.
P total pressure in a bubble, MPa Garner, F. H. and Hammerton, D., 1951, C&m. Engng Sci. 3,
P; vapor pressure of water, MPa 1.
PO atmospheric pressure, MPa Grace, J. R., 1973, Trans. Inst. Chem. Engrs 51, 116.
Hadamard, J. S., 1911, C.r. Lebd. Sdanc. Acad. Sci. Paris 152,
Pe Peclet number, = d,U/D,, dimensionless
1735.
4 dimensionless velocity gradient at front stag- Hamielec, A. E. and Johnson, A. I., 1962, Can. J. them. Engng
nation point, dimensionless 40, 41.
R gas constant, MPa m3/molK Hamielec, A. E., Storey, S. H. and Whithead, J. H., 1963,Can.
Re Reynolds number, = p=d,U/C(c, dimensionless J. them. Engng 41, 246.
Harper, J. F., 1973, J. FIuid Mech. 58, 539.
SC Schmidt number, = p, DA/p,, dimensionless
Hikita, H. and Konishi, Y., 1984, A.I.Ch.E. J. 30, 945.
Sh Sherwood number, = k,d,/D,, dimensionless Hikita, H., Asai, S. and Azuma, Y., 1978, Can. J. them. Engng
t time, s 56, 371.
T absolute temperature, K International Critical Tables, 1963, Vol. IV. McGraw-Hill,
u bubble rise velocity, m/s New York.
Kamei, S., Kimura, K., Hochi, S. and Matsusaka, K., 1953,
v, volume of bubble, m3 Kagaku Kogaku 17. 309.
Levich, V. G., 1962, Physicochemical Hydrodynamics (trans-
lation from Russian). Prentice-Hall, New York.
Lochiel, A. C. and Calderbank, P. H., 1964, Chem. Engng Sei.
Greek symbols
19, 471.
B ratio of the mass transfer coefficients of the Miyauchi, T. and Nomura, T., 1971, Kugaku Kogaku 35,
fluid sphere to the equivaIent rigid sphere at 1234.
the front stagnation point, dimensionless Motarjemi, M. and Jameson, G. J., 1978, Chem. Engng Sci. 33,
viscosity ratio, = pJ,u_ dimensionless 1415.
F* dimensionless interfacial pressure at the rear
Oellrich, L., Schmidt-Traub, H. and Brauer, H., 1973, Chem.
Engng Sci. 28, 711.
stagnation point of the fluid sphere and de- Pasiuk-Bronikowska, W. and Rudzinski, K. J., 1980, Chem.
fined by eq. (31), dimensionless Engng Sci. 35, 512.
P viscosity, kg/m s Pasiuk-Bronikowska, W. and Rudzinski, K. J., 1981, Chem.
EnanQ Sci. 36, 1153.
I-I* surface pressure at the rear stagnant point,
P&m&, D. W., 1951, Sc. D. Thesis, MIT.
N/m Perry, J. H., 1963, Chemical Engineers’ Handbook, 4th edn, p.
density, kg/m3 3. McGraw-Hill, New York.
difference in density between the surface and Ranz, W. E. and Marshall, W. R., 1952, Chem. Engng Prog. 48,
bulk, kg/m3 173.
Ress, A. J., Rodman, D. J. and Zabel, T. F., 1980, J. Separ.
surface tension, N/m
Proc. Technol. 1, 19.
interval of flashing, s Rybczinski, W., 1911, Bull. Int. Acad. PO!. Sci. L&t., Cl. Sci.
stagnant cap angle, rad Math. Nat., Ser. A, 40.
magnification defined by a ratio of the ap- Sadhal, S. S. and Johnson, R. E., 1983, J. Fluid Mech. 126,237.
parent length on film to actual length, Savic, P., 1953, National Research Council of Canada. Report
No. MT-22.
dimensionless Saville, D. A., 1973, Chem. Engng J. 5, 251.
Shah, Y. T. and Sharma, M. M,, 1976, Trans. Inst. Chem. Engrs
54, 1.
Subscripts Sh;;ugalter, H. A. and Ferguson, J. B., 1936, Can. J. Res. 14(8),
b liquid bulk Stokes, G. G., 1880, Mathematical and Physical Papers, Vol. I.
B bubble Cambridge University Press, Cambridge.
continuous phase (liquid phase) Szekely, J. and Martins, G. P., 1969, Trans. Met. Sot. AlME
; dispersed phase (gas phase) 245, 1741.
Takahashi,T., Miyahara,T. and Mochizuki, H., 1979, J. them.
?lC natural convection
Engng Japan 12, 275.
obs observed Thuy, L. T. and Weiland, R. H., 1976, Ind. Engng Chem.
W water Fundam. 15. 286.
 Gas desorption from liquids 2319

Ueyama, K. and Hatanaka, J., 1973, J. them. Engng Japan 6, van Krevelen, D. W. and Hoftijzer, P. J., 1948, Chim. Ind.
167. XXIeme Congr. Int. Chim. Ind.. p. 168.
Ueyama, K. and Hatanaka, J., 1976, J. &em. Engng Jnpun 9, Weiland. R. H., Thuy, L. T. and Liveris, A. N., 1977, Znd.
155. Engng Chem. Fundam. 16, 332.
Usher, F. L., 1908, Z. phys. Chem. 62, 622.