Chemical Engineering Science, Vol. 51, No. 10, pp.

1837-1846, 1996
Copyright © 1996 Elsevier Science Ltd
Pergamon Printed in Great Britain. All rights reserved
S0009-2509(96)00227-8 0009-2509/96 $15.00 + 0.00

DESIGN, MODELING AND PERFORMANCE OF A NOVEL SONOCHEMICAL REACTOR

FOR HETEROGENEOUS REACTIONS

C. HORST, Y.-S. CHEN, U. KUNZ AND U. HOFFMANN

Iustitutfar ChemischeVerfahreustechnik,SFB 180, TechnischeUniversihrttClausthal, LeibnizstraSe17, D-38678 Clausthal-Zellerfeld

Abstract - We constructed a novel sonochemicalreactor for heterogeneoussolid fluid reactions which us~ high intensity
ultrasound from a concentratorhorn. It can b¢ shownthat the conceptof a conical funnel fits the demands for a nearly perfect
radiation effectivenessand a good reaction management. Theoretical investigationsconcerningthe sound field in both the
sonotrode and the reactor for differentcases give some insight in the fundamentalrelationshipbetween consumed electrical
power, convertedmechanical energy in the transducer and radiated sound energy. The main energy conversionpaths in the
liquid are presentedas well. A Grignardreactionwas investigatedin the reactor and showedextremereactionrate en.haneements
which are due to a mechanochemicalreactionkineticcaused by implodingbubblesand the thereby generatedshock waves and
mierojets. A simulation of the particle size distributionduring simultaneousreaction and fragmentationin an ultrasonic leop
reactor was developedto correlatemeasuredsize distributionswith modelequations.

INTRODUCTION
Ultrasound is known to have great effects on chemical reactions. According to Suslick et al. (1988), especially
some heterogeneous solid fluid reactions gain extreme reaction rate enhancements by sonication. These effects are
explained by the treatment of ultrasound on solid surfaces. Reactive surfaces are always present by eroding the
surface and by cutting the solid into smaller pieces caused by the strong impact of cavitating bubbles in the vicinity
of solid boundaries. These cavitating bubbles undergo an asymmetric collapse creating shock waves and liquid
microjets with pressures up to some hundred megapascals at the neighbouring boundary. Such extreme conditions
cause plastification or breakage depending on the material properties and treatment conditions (Young, 1989).
Our aim was to construct a suitable sonochemical reactor with a concentrator horn at a frequency of 20 kHz and
intensities in the range from 14 to 160 W/cm 2 with radiation surfaces of 1.23 and 4.91 cm 2. The sonicated volume
varies between 0.1 and 0.5 dm 3, temperatures from -20 to 200 °C and pressures up to 1.6 MPa can be used. The
experimental reactor can be extended to a technical scale volume of about 10 dm 3 and may be used to close the still
present gap between laboratory and production in sonochemistry (Martin, 1993).
The chosen test reaction is the Grignard reaction between 2-chlorobutane and magnesium in tetrahydrofuran as
solvent. This reaction shows no significant side reactions but all relevant complications chemists normally
encounter when carrying out Grignard experiments like slow reaction rates, long induction times and high
exothermic behaviour.

REACTOR DESIGN
Almost every reactor design for
(1) reactor (0.5 dm3) with sonotrode gas exit
(2) stirred tank (2.5 dm3)
technical scale sonochemical reactors (3) flow loop pump
uses the reactor as an external (4) dosing pump
module in a loop. This arrangement (5) cryostat
has certain advantages: Several (6) gas flow meter
modules can be retrofitted to existing T thermo couple
plants, smaller units are easily cut off P manometer I 1

when reactions are running out of I

control, the number of modules can

be fitted for the desired production
volume and maintenance is simple.

REACTOR SYSTEM
gas feed
(0 - 1 dm3/h)
~ ]
B
I
L.
f
,3,
15 I

The laboratory system for a year

production of about 4 tons Grignard liquid feed
(4)
solution is shown in Fig. 1. (0 - 0.6 dm3/h, 2.1 mol/dm 3 R-CI)
The reactor (1) is coupled with a liquid product (2.0 mol/dm3 Grignard)
stirred tank (2) of about 2.5 dm 3.
Both containers are equipped with Fig. 1. Flow chart for the experimental0.5 dms sonochemicalreactor systemwith
cooling jackets. The loop pump (3) a productionof 4 tons Grignardsolutionper annum.
maintains a constant liquid flow from
the stirred tank to the sonochemical reactor. Typical recycle ratios vary from 30 up to 500. A feed pump (4)
conveys all liquid reactants. The gas feed allows the addition of inert or reactive gases. Liquid products can be
drawn off in the pipe between the stirred tank and the reactor. Gaseous reactants leave the system at the top of the
stirred tank through a pressure valve. Temperatures up to 200 °C and overpressures of 1.6 MPa can be applied. All
used materials are either stainless steel (AISI 316Ti) or Teflon. For further information see Horst (1995).
1837
1838 C. HORSTet al.
REACTOR DESIGN
The application of concentrator horns in chemical reactors demands a
X
careful analysis of the adequate vessel geometry. The tip of the transducer e g r e e s

horn acts in most cases as a point source and radiates spherical waves into [mm]
the liquid (Meyer and Neumann, 1967). The radiation pattern depends on
the product of the radiation surface radius R and the wave number k. Two xL=320
extreme cases can be evaluated: Plane waves with k.R>>l and spherical reaction
waves with k.R<<l. All wavetypes can be described by the specific wave chamber
impedance _Z:

v
I,gl
~. (1)
The real part of the complex quantity 7, is often called radiation resistance xT=12o
and describes the part of the radiated energy emitted as compression waves.
The imaginary portion depicts the inertial media oscillation. Easy formulas
for the plane and the spherical waves can be evaluated:
d/2
Z : PLCL, Re{Z} = PLCL for plane waves, (2)
15 4o [mm]
Z = PLCL " i ~ '
ikr Re{Z} < OLCL
for spherical waves. (3)
F i g . 2. Sketch o f the funnel geometry.
The maximum value of 7. is the product of the liquid density and sound
speed, the so-called wave resistance. In plane waves this value is always present. Spherical waves have a smaller
radiation resistance than plane waves. The main aim in constructing a sound radiating device is the utilization of
the full radiation resistance pLCLat the sound source, For the given concentrator horn with sonotrode diameters of
12.5 and 25 mm the radiation resistance is given to be:
Re{Z} k2R 2 t0.22 for R = 6.25 mm
PLcL k2R-------~
1+ - L0'53 forR = 12.5 nm~ (4)
with k=84.7 m ~ (water, 20 kHz).
The bad radiation effectiveness of the mentioned horns can be improved by applying a pipe or a funnel.
Funnels can be calculated as a one dimensional problem when the following assumptions are made:
1. The funnel diameter d is smaller than the
wavelength XL I
2. Walls are ideally sound reflecting
A combination of material and energy balance leads
to the well known equation of Webster (Meyer,
1967):
c32P dinS(x) c3p 1 02p
- + - - - - - (5)
a~x2 dx ax c~, ~ 2 "
The cross section S(x) of a conical funnel is defined
by
S(x) = sx 2 . (6)
Fig. 2 depicts the geometry of a conical funnel with
entrance and exit co-ordinates xT and XL.
The solution for the sound pressure 12 and the
displacement velocity v can be evaluated:
x T i(cot - kx)
p(x) = P T - - e , (7)
X

Xw fl 1 lei( t-kx)
0LcL ~" ~ - ~ - ~ ) . (8)
id exit with filter
By dividing these equations the radiation resistance id feed
)rmo couple
at the sonotrode follows: ng© bolt
nge closunB
Re{Z}= (kxw) ~ ~eflange
deTed f l ~ e
(9) ~lant joint
PLCL 1 + (kXT) 2 " ler jacaket
,. _, __ aling j ~ket
Comparing this result with equation 4 it is evident (11) ~¢1"
that the radiation impedance in a funnel is (12) adapter
(13) l~rode
determined by the co-ordinate xT of the sonotrode 04) lnmsduccx with boolter
and not by its radius R.
For optimum radiation resistance the term should be F i g . 3. Experimental sonochcmical reactor.
equal to 1. This can be achieved with xx=oo causing
 A novel sonochemical reactor for heterogeneous reactions 1839
the funnel tO degenerate to a pipe. A proper radiation resistance can be obtained with finite values of xr. We choose
a value of 0.99 and get a co-ordinate of xr =120 ram. With a diameter dr of 30 mm the opening angle 7 is 7.1 o. The
exit diameter dL should be greater than the wavelength and was chosen to 80 mm. The length of the funnel is 200
nlln.
The funnel geometry allows a sonication 1 ~ .............. spherical wave
from the bottom of the vessel. The 0,9 \~ ~ --conical horn
radiation resistance is nearly optimal and 0.8 \ \ \ ~ • 40 watt, m e a s u r e d
furthermore plane waves are present in
the reactor, o,7
:,'~,\
~,~ ~ ~ • 8 0 watt, measured
1 2 0 w a t t , measured
Fig. 3 shows the constructed laboratory o,s
scale reactor module. The concentrator ~.°-°'s
horn can be seen at the bottom of the
figure. All fluid reactants enter the 0,4 .
reaction chamber through a ring crevice 0,3
around the sonotrode. The conical
0,2
reaction chamber is covered by a cooling
jacket. A filter at the top of the reactor 0,1
prevents solid rinsing. The solid
reactants remain as a fluidized bed in the 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
distance x[cm]
reactor. This bed classifies the solid and
guarantees that the largest particles are Fig. 4. Comparisonbetweentheoreticaland measurednormalizedsoundpressure
in the most reactive zone above the distributionsin the experimentalsonochemicalreactor.
sonotrode. A clear zone at the top of the
reactor can be created with an adequate liquid flow velocity.
Fig. 4 compares the enhancement effect of tee funnel geometry with spherical and plane waves (p/p°=l). The
normalized sound pressure distribution is far better than for the spherical wave. Measured sound pressures at
different intensities are depicted in the same figure. The differences between theoretical and measured sound
pressures are due to the fact that in cavitating sound fields strong absorption of energy is observed. The higher the
cavitation intensity the larger the absorption.

SONOTRODE-LIQUID COUPLING
The generated amplitude and pressure at the
sonotrode tip can be evaluated by the so-called four
pole analysis of stepped transducer horns according to
~i.l = ~i= ~.~i+i =
Beyer and Letcher (1969). Imagine a part of the
sonotrode as shown in Fig. 5. The amplitude at every Ai.lei(,a'kx) Aiei(,n'Icx) Ai+Iei(mt-I°0
point in the slab can be described by two opposite ) )
waves: + Bi.lei(al÷kx) + Biei(td+kx) + Bi.l ei(~t+l°0
= Ac i(°~t-kx)+ Be i(a~t+kx). (10) (
At each of the two faces of the slab the boundary
conditions for continuity of mechanical displacement
and force must be satisfied. The elastic stress in a
segment can be evaluated by:
T = - p c 2 /~. (11) Fig. 5. Sketch of a concentrator element of the sonotrode and amplitudes by
& opposite wavetrains.
For a multicomponent concentrator horn of N
segments 2(N+ 1) boundary conditions for 2(N+ 1) unknown parameters Ai and Bi must be satisfied.
Conditions at the piezo crystal or ceramic can be represented by the current I and the voltage V:
do
I = Spie~o ~ = icoo, (12)

V = - h l l (A~piezo) + O" dpiez° (13)

KI1
with o as the surface charge density, Sp~ozoas the surface, dp~e~oas length of the ceramic and A~p~c~oas displacement
difference between the two ceramic faces, hl~ is the stiffness constant and ~q~ the permittivity.
The stress in the piezo crystal can be evaluated by:
T =c]l ~ - hllo. (14)
These equations can be solved with standard numerical methods. The amplitudes in the used horn and the conical
funnel are shown in Fig. 6 for a current of 0.05 A and water as medium.
A finite element analysis of the reactor is shown in Fig. 7. The differences between the analytical solution and the
FEM solution are due to the fact that the theory does not include the ring crevice between the sonotrode and the
1840 C. HORSTet al.
reactor wall. At this point a disturbed zone is observed in the FEM analysis which approaches the analytical
solution after a distance of some centimetres.
The load on the sonotrode by the coupled liquid is frequently considered when coping with energy quantification.
Most sonotrodes are built for resonance conditions without external load. In this case power and amplitudes reach
their best values and an optimum power conversion from electrical to mechanical energy can be achieved. The
frequency band around resonance frequency is very sharp. By applying an external load the resonance frequency is
shifted to different values and energy conversion becomes bad. Fig. 8 shows the optimum frequencies for either
maximal amplitude (higher load causes lower fopt) or power (higher load shifts fort to higher values) for a given
current I. With higher loads on the sonotrode the resonance maxima decrease and broaden. Furthermore the energy
conversion from electrical to mechanical energy becomes less effective. To achieve optimum energy yield a
frequency regulated generator and a load optimized sonotrode geometry should be used.

p/pO XL
0.5

t
d >k

0.1

d <k

1.0

~(x,t=O) [~tm] xT

Fig. 6. Evaluated amplitudes in the concentrator horn and in the Fig. 7. Amplitudes in the liquid phase, evaluated by a finite element
liquid phase of the funnel. analysis.

LENGTH OF CAVITATION ZONE

Any application of ultrasound causing cavitation in a liquid is accompanied by strong absorption of sound and a
change in the average density and sound speed of the bubbly liquid (Young, 1989). ff we denote pO as the sound
pressure in the liquid at the radiating surface then a decreasing sound pressure p over the distance x is observed.
This decrease is caused by the geometric dispersion of the conical funnel (Fig. 4, upper line) and an additional
decrease by absorption (Fig. 4, measured points). A lower sound pressure generates less intensive cavitation and
the absorption is smaller. After reaching the so called threshold pressure PT, cavitation is not longer possible. The
threshold pressure describes the lowest sound pressure causing cavitation in a liquid. PT can be evaluated by Blakes
formula:
/-

30
PT = Poo + 8 o l A (t5)
9 "~2(P~o+(2o/RT))R~
The absorption in a bubbly liquid may be expressed by the complex wavenumber
k = k - ict (16)
with ct as absorption coefficient.
In a bubbly liquid the absorption coefficient can be evaluated by:
ot = Im(k) = Im(copLf~Km). (17)
is the complex compressibility of the mixture defined by:
K m = K L +K a. (18)
KL is the compressibility of the pure liquid and ~ the additional compressibility due to bubbles.
If we assume small amplitudes of the bubble around an equilibrium value of 1% and take g(l%) as the number
distribution of bubbles per unit volume then the formula for the time derivative of the bubble volume is
0o

1 dV-4n~R:.-~-.g(R~)dR~. (19)
V dt
o
 A novel sonochemical reactor for heterogeneous reactions 1841

The bubble wall velocity may be deduced by: 200so,

dR iop 1 no detectable :
-- = - (20) power optimum '
dT PLR~ ((o~-co; + 2ibco) .............. fr
2~000 [- fo~(power) ..
with b as d a m p i n g c o n s t a n t a n d o , as r e s o n a n c e ~ ....... fopt(amplitude)
frequency of the bubble. The compressibility follows ~ I ~ .......
~ I~50

j(, _g(Ro)dRo
,+. ). (2,)
PL o O:'r -CO 21boo I~oooF_. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ....:::.::_Z... ~..........
The resonance frequency is equal to
/ x

Or p L R e ~. " 5 PtCL[kg 'm "2 s" ]'

Thermal, viscous and radiation losses define the FIg. 8. Optimum frequencies for a given concentrator horn at different loads
damping constant b for a bubble of radius ~ PLcLatthesonotrode. Resonance frequency f~=19870 [sql with no load.
(Prosperetti, 1977):
Poo + 2rlL + r'° 2 Re [~] (23)
b = bth + by + br 2OOLR~Im{cp} PLR2 2eL
with ¢pas phase shift between the sound field and the bubble motion.
Theoretical values of the absorption coefficient ct lie in the range of 1 to 3 dB/cm which is in good agreement with
the measured absorption coefficients of 2. l dB/cm for 80 Watt and 2.6 dB/cm for 120 Watt.
According to Young (1989) the sound speed of a bubbly liquid may be evaluated by:
2 P~o (24)
Cm =
P L g b ( l - - ~b)
with gb as the volume fraction of bubbles not too close to 1 or 0. The average density in the bubbly liquid is
Pm -= P L ( 1 - %). (25)
With a typical value for gb of 0.2 at the sonotrode in water, the sound speed, density and wave resistance are:
c m ~ 20 -m
- ' Pm ~ 800 -kg
- pmCm 7_ 1.6"104 kg (26)
S m3 ' m2s "
These values differ significantly from that of the pure water (pc=l.5*106 kg.m2.s 1) and from those of air
(ca~=344 m.sl).
Penetrating greater distances the volume fraction rapidly decreases and reaches zero when the cavitation threshold
Pr is approached.
The cavitation zone in the reactor can be characterized by three sections:
Zone I: Heavily cavitating region above the sonotrode, length around one centimetre, bubble volume fraction
of up to 0.2, wave resistance around 2. l04 kg.m2-sq, sound speed of 20 m.s q.
Zone II: Cavitating transition region with a quick increase in wave resistance and sound speed
to values of 2.105 kg.mLs ~ and 200 m.s 1.
Zone III: Non cavitating zone with sound pressures below threshold. Sound speed and wave resistance
are the values for pure water: pc=l.5* 106 kg.m2.s ~ and c=1500 m.st.

EXPERIMENTAL
We choose the Grignard reaction
C4H9C1(1) + Mg(s) solvent) C4H9MgCI(1 )
to evaluate the characteristics of the sonochemical reactor. This
reaction is known to show long induction periods, slow reaction Table 1: Measured kinetic parameters for the Grignard
rates and a high exothermic behaviour. We used tetrahydrofuran reaction of 2-¢hlorobutane.
as solvent because of its higher boiling point compared with
parameter value
diethylether. 2-chlorobutane (R-C1) as reactant exhibits no side
reactions like Wurtz coupling. The measured kinetic model dm 3
without ultrasound is quite simple in the experimental ko 5.5"106
concentration range ofc[C4HgC1] and c[C4HgMgCI] < 1 mol/dm3: kg. s
kJ
r' = k0 exp[-Z~-]cR_c, I m°l l. (27, 50--
Lkg.sJ EA mol
The homogeneous reaction rate in the reactor is determined by: kJ
300 --
I mol] (28) (-ZXH): mol
r = P~e~eMgr' Ldm 3 .sJ'
The measured reaction parameters are shown in Table 1.
1842 C. HORST el al.

If we define an ultrasonic enhancement factor Eus:

k 0,measuredwithultrasound
Eus - (29)
ko
all experimental results can be explained with this value.

INFLUENCE OF TEMPERATURE
All experiments were carricd out at
Temperatures of 20 or 0 °C. Fig. 9 shows 20°C 0 °C
the measured reaction rate constants as a
function of temperature in an Arrhenius
-3.0 I I
plot. All reaction rate constants under
sonication are greater than without
ultrasound. This is due to the fact that the
sonicated magnesium is free of
passivating oxide films. Furthermore
In(k)
-4,0

-5,0
! I- °1
activated metal regions exist by the
impact of liquid jets and shock waves
which show different reaction kinetics ~.0
than the untreated surface regions.
A comparison of the gradients in the
Arrhenius plot makes clear that the -7,0
activation energies for the reaction with !
or without ultrasound are nearly the same T
in the observed temperature range.
According to Heinicke (1984) one might Fig. 9. Arrheniusplot for various reaction rate constants
say that the obviously mechano chemical with (filled circle) and without (empty circle) ultrasound.
reaction under ultrasound is mainly
temperature dominated. Only longer lasting surface effects cause the observed rate enhancement. At far lower
temperatures a significant influence of the high temperatures in the pitting zone should become evident.

INFLUENCE OF STA TIC PRESSURE, USED SONOTRODE AND BUBBLED GAS

To evaluate the characteristics of the constructed
reactor, experiments were carried out varying the Table 2 Measured enhancement factors for different static pressures, gas
static pressure, the sonotrode geometry and the flow stream velocities and sonotrodes.
velocity of inert gas.
static pressure sonotrode gas stream Eus
The influence on the enhancemcnt factor Eus is
[MPa] number {dm3/hl
shown in Table 2.
The used intensity was 25 W/cm2 with both 0.2 0 1.8
sonotrodes. The difference between the concentrator 0.2 1 1 2.7
horns was the applied power generator with an 0.2 2 0 3.7
electrical power consumption of 70 W (sonotrode 1) 0.2 2 1 3.0
or 200 W (sonotrode 2). It is easy to see that the 0.3 1 0 3.0
influence of bubbled gas is relatively low for the high 0.3 1 1 1.3
power sonotrode (2). The enhancement factors lie 0.3 2 0 4.0
between 3.0 and 4.0 and are greater at higher static 0.3 2 4.0
pressure. The low power sonotrode has very unstable
operational modes. Enhancement factors vary
between 1.3 and 3.0 and depend strongly on the bubbled gas amount and the hydrostatic pressure.
An increase in the applied static pressure causes smaller but more violently collapsing bubbles. These bubbles
cause more severe damage on the magnesium surface and reaction rates are accelerated.
A donation of inert gas to the reaction mixture showed only very little influence on the reaction rate. A possible
reason for this behaviour is the fact that the solvent stays gas saturated over the entire reaction time at the elevated
pressure. Excess gas is just rinsed through the reactor and does not alter the total gas content in the liquid or in the
oscillating bubbles.

INFLUENCE OF APPLIED I N T E N S I T Y
Fig. 10 depicts the influence of the applied ultrasonic intensity on the enhancement factor. With higher intensity
comes a higher sound pressure and therefore greater cavitation bubbles.
The energy stored in a cavitating bubble of radius R~,,=before collapse can be written as
4n 3
Wcav -- - ~ - PQoR max' (30)
 A novel sonochemical reactor for heterogeneous reactions 1843
(see Kristensen,1978). The maximum radius of
bubbles is nearly independent of their initial
size (Young, 1989) and increases with higher
sound pressure.
As long as the Rayleigh collapse time of the , j
bubble is smaller than half the oscillation time J
an increase in R , ~ causes greater effects.

PULSED ULTRASOUND
2 us = (1 +0.23*intensity) ° s
A way for saving energy in ultrasonic
applications is the use of pulsed ultrasound.
The generator is switched on and off for a given
period. This duration of sonication is expressed
0 2Jo
as a part of one second for the used generator 40I ,
6~I 880
100
system. The total sonication time becomes only intensity [W/cm~J

a fraction of the reaction time. Fig. 11 shows Fig. 10. Enhancement factor Eus as a function of applied intensity.
the influence of pulsation time on the observed
enhancement factor. It is evident that even at the lowest cycle time nearly full enhancement is present. This is
obviously due to the fact that the reaction time constant, the pulsation time and the oscillation period are extremely
different:
Co(R-Cl)
~,=otion -~ lO 5 [s], (31) i
ro
• p~sa,,o° ~ 1 [s],
"to~ill=io. = 5-10 -5 [s]. (33)
(32)
::3] I

During one pulsation cycle of 0.1 s the sound

field exhibits 2000 oscillations. Every oscillation t

causes a lot of cavitating bubbles cleaning and

treating the metal surface. Therefore slow
reactions like the Grignard reaction will never be
able to take full advantage of the clean and I
reactive surface production with ultrasound. The
optimum surface treating of Grignard oI i i
0 :~ ,0 ~o 8'0 ,~o
magnesium with lowest energy consumption can pulse duration [% of 1 second]
be achieved with high intensity ultrasound Fig. 11. hdluence of pulse duration on the enhancement factor Eus.
radiated as short pulses with long reaction times
without sonication.

MODELING OF SOLID FRAGMENTATION AND REACTION

Collapsing bubbles are heated to very high temperatures and are capable to produce shock waves and high velocity
solvent microjets which, on interaction with the solid surface, can lead to considerable changes in surface
morphology and fragmentation of solid. Other possible mechanisms of solid fragmentation are the results of
collision associated with the rapidly alternating pressures in the liquid medium which may cause particles to reach
very high accelerations. With such high acceleration a particle can collide with an object in the reactor such as
other particles, the tip of the sonifier horn or the reactor wall. With consideration of particle comminution in
heterogeneous reactions we employ a general model to describe particle size distribution with chemical reaction.

MA THEMA TICAL FORMULA TION

A vessel (reactor) with sonifier source is continuously fed and depleted by a stream of particles which are recycled
in the feed stream (Fig. 12). Each particle, during the time it is in the reactor, shrinks and is cleaved according to a
prescribed differential law. To formulate this question for a spherical particle mathematically, let
Xrn~
__~(x,t) + O(Rsu(x,t)) = 0uo(x,t) 0u(x,t) a(x)u(x,t)+ -|a(y)b(x,y)u(y,t)dy,
& c3x I+R I+R 3
(34a,b)

where R, is the rate at which a particle of size x shrinks, u(x,t) is the distribution density of particle size x, which
refers to the reactor volume. Uo(X,t) is the distribution density of particle size x in the feed. 0 is the space velocity
and R the recycle ratio, c is the concentration of liquid reactant. Co is the feed concentration, p, and M. are the
density and molar weight of the particle respectively. The second term on the left-hand side in equation (34a) is the
1844 C. HORSTet al.

rate at which particles in the reactor are shrinking past

size x at time t. The first two terms on the right-hand Uo, Q, Co
side are the feed and deplete. The third term represents I Vessel U, Q, c
the loss of particles of size x because of their breaking up ii
u, 1%, c, a , b
into smaller particles, where a(x) is the rate at which a
particle of size x breaks. The last term represents the
increase of particle size x because of the break-up of
larger ones, where b(x,y) is the distribution of products R,u,c
from a particle of size y breaking.
Equation 34b represents the concentration change of
liquid component, which is coupled with the size Fig. 12. Feed, take-offand recycle arrangements.
distribution. Equations 34a,b are the desired differential
equations in the reactor distribution function u. The initial conditions are
u(x,0) = uo(x) and c(0) = c o. (35)
We assume the initial particle size distribution and concentration to equal the size distribution and concentration in
the feed. This implicates that the size distribution and the feed concentration is independent of time. These initial
conditions determine the size distribution and the concentration in the reactor in terms of the prescribed well-
known functions P~, uo, a(x) and b(x,y) and the parameters 0 and R. In the function 1% implicates the influence of
the chemical kinetic. To describe the shrinking rate of a particle the shrinking-core-model will be used. Because of
the cleaning effect of ultrasound the resistance of ash and product-layers will be neglected and the shrinking rate of
the particle size x is
dx k¢ff (x)cM s
R s = -- = 2 (36)
dt 9
with
kefr (x) -t = kf I + k -l ' (37)
kfx
- Sh = ash + bshSC ~ Re ~ (38)
D
where l%rrrepresents the effective mass transport coefficient, kf is the mass transport coefficient, k is the reaction
rate constant. The evolution of the breakage functions is described by several authors (Ziff, 1991). Since b(x,y) is
the rate of production of particle size x from those of size y it must be normalized so that mass is conserved:
Y
' b ( x , y ) d x = 1. (39)
0
There exists a function 13(x) to satisfy
lY(x)
b(x,y) = - - (40)
~(y) '
The function [3(x) must be any differentiable function with the initial condition 13(0) = 0. We consider the ease
where the breakage rate of particle size x has the form a(x) = eax ~, and where the distribution of product
fragmentation has the form
,,
b(x,y) = - - - - . (41)
yc~+l
From this consideration the relation between rate of breakage and production of product fragmentation,
furthermore the relation between rate of breakage and the new function ~(x) can be easily derived, then we have
f3(x) = c,~(a(x)) ~ where y=(ot+l)/)~, (42)

APPLICATIONS
If a particle functions in heterogeneous reaction as a catalyst, i.e. the particle does not shrink, the second term in
equation 34a might be deleted. Then the equations 34a and 34b can be solved separately (the kinetic of the
catalytic reaction has to be applied). To compute the concentration, the particle size distribution shall be first
calculated, which is a pure problem of comminution. The degradation of polymers can be dealed with the same
equation. The case that either the chemical reaction or the particle fragmentation are neglectable is studied, too.
We assume that the concentration of the liquid component in the reactor is constant and equal to 0.5 mol/dm 3. In
addition, the normalized particle size is used.

RESUL TS O F SIMULATION AND DISCUSSION

Case 1: steady state solution for particle fragmentation
The influence of the reflux ratio R on the steady state size distribution with given functions of breakage are shown
in Fig. 13. The curves move to the left side (smaller particle size region) with increasing reflux ratio. By a high
 A novel sonochemical reactor for heterogeneous reactions 1845
reflux ratio at which batch reactor behaviour is observed, the particles will be cleaved into very small sizes. An
impulse curve is obtained with infinite values of the recycle ratio. The distribution in that region is very narrow
and the value is very large. Narrow size distributions are observed by low and high reflux ratios. Wide size
distributions with two peaks represent the passage from bigger particles to smaller ones. This represents that one
can choose the suitable reflux ratio for a desired size distribution.
Case 2: unsteady state solution for particle fragmentation
In Fig. 14 the progress of the size distribution in time for a reflux ratio equal to 5 is presented. The change of the
size distribution is observed at different times. The bigger particles (> 0.7) are cleaved and decrease fast. The
medium ones (~ 0.65) remain their position because the increasing rate from the break-up of larger particles is
approximately equal to the break-up of themselves. This equilibrium can be observed at the beginning stage. After
150 seconds the steady state is almost reached. Two peaks in the distribution curve are also observed.
4 ~ - ~ n 1 r r r"
1] . r . . .

'+'' +
+..,,++,+- q] i .++-

O+4 O6 ILl
~ - ~ ' ~ ~ [-1
F i g ~ a m ' c l e fragmentation. Fig. 14. Unsteadystatesolutionfor particlefragmentation.
Case :3: solution for chemical reaction without fragmentation
In this case the particles remain in the reactor and the liquid concentration is constant (c=0.5 mol/dm3). Particles
take part in the chemical reaction and are consumed. The change of the particle size distribution is demonstrated
in Fig. 15. The bell-shape of the distribution curve is gradually wider because the radius change of smaller
particles during shrinking is faster than that of bigger ones.
Case 4: solution for chemical reaction with simultaneous particle fragmentation
In this case a limit of particle breakage is introduced. We assume that particles smaller than 0.2 do not break.
Another kinetic of breakage is also introduced. It will be diffcmnt if particles participate in a chemical reaction and
break simultaneously. A wider bell-shape, faster decrease of large particles and an approximate equilibrium
particle size at the starting stage are expected. All these phenomena are shown in Fig. 16. The distribution depends
on the kinetic of breakage and the chemical kinetic. The bigger particles decrease more due to fragmentation than
by chemical reaction. The bell-shape moves slightly to the left (to smaller particles). As shown in Fig. 16 an
equilibrium particle size lies at ~_ 0.7. The flat curve profile in the smaller particle region indicates that the
influence of the chemical reaction is relatively strong because smaller particles undergo the greatest size changes.

imnundw t
Ip0,~00,1000.1500,2000,,t000,4000.5000

J
.+

0.2 0.4 O.e ~8 O.2 O,4 01D 0J

Nom',alJzed Sh'e [-] mmnuu~.,a,, H
Fig. 15. Solution for chemical reaction Fig. 16. Solutionfor chemicalreaction
without particle fragmentation. withsimultaneousparticlefragmentation.
1846 C. HORSTet al.

SUMMARY
The constructed sonochemical reactor with a conical funnel geometry shows good sound field and reaction
management qualities. A fundamental theoretical investigation of the main energy conversion paths in the reactor:
electrical power, mechanical energy, radiated energy, energy consumption in the liquid and finally usable energy
from cavitating bubbles can be carried out. Ultrasound may become applicable in technical scale reactors when
every effectiveness factor in the energy transformation cascade is optimized for the final goal to achieve maximum
cavitation with m i n i m u m energy input. The extraordinary results with the chosen Grignard reaction give a taste of
the possible benefits using ultrasound in chemical reactions. Further insight will be gained when the coupling of
ultrasonic parameters with surface activation and particle fragmentation is done. A design of desired particle size
distributions can be done by varying the recycle ratio or the applied intensity in the reactor system. The coupled
effects of fragmentation and reaction in heterogeneous reactions may be modelled with the given theory and a
precise measurement of the ultrasonic influences may lead to a better understanding of the cause and the effects of
ultrasound.

NOTATION
A = Displacement constant m S = Surface mz
B = Displacement constant m T = Stress Pa
a = Breakage rate t = Time s
b = Damping constant s~ u = Size Distribution kg.m 4
b = Breakage function W = Energy J
CL = Sound speed (liquid) m.s l V = Voltage V
c = Concentration mol.dm 3 V = Volume m3
c~t = Elastic constant Pa x = Reactor co-ordinate m
d = Diameter, thickness m Z_ = Wave impedance kg.m2.s ~
Eus = Enhancement factor - ('AH)r = Reaction enthalpy J.mol ~
f = Frequency s1 tx = Absorption coetiicient dB.cm t
hH = Stiffness constant V.m t e = Volume fraction
I = Current A q~ = Phase shift o
K = Compressibility Pa l L = Opening angle o
k = Wave number m "t rl = Viscosity kg.mLs ~
ko = Reaction rate constant dm3.kg l . s 1 • = Permittivity F.m ~
1%~, kf = Mass transport coefficient m.s ~ ~.L = Wave length m
v-=0~/c3t = Displacement velocity m.s ~ PL = Density (liquid) kg.m 3
P® = Static pressure Pa ~ = Surface Charge Density C.m 2
la = Sound pressure Pa ~ = Surface tension N.m -1
R,r = Radius m O = Space velocity s "~
Rs = Shrinking rate x = Time constant s
R = Recycle ratio - o = Angular frequency s~
r = Reaction rate mol.dm'3.s ~ ~ = Displacement m
s = Constant

REFERENCES
Beyer, T.R. and Letcher, S.V., 1969, Physical Ultrasonics, Academic Press New York
Heinicke, G., 1984, Tribochemistry, Akademie Verlag Berlin
Horst, C., Chen, Y.-S., Kunz, U. and Hoffmann, U., 1995, Design ofa sonoreactorfor heterogeneous reactions,
Congress Proceeding 1995 World Congress on Ultrasonics Berlin (to be published)
Kristensen, J.K., Morch, K.A. and Hansson, I., 1978, A simple model for cavitation erosion of metals,
J. Phys.D: Appl. Phys. 11 899
Martin, P.D. 1993, Sonochemistry in industry: progress and prospects, Chemistry and Industry 5 233
Meyer, E. and Neumann, E.G., 1967, Physikalische und Technische A kustik, Vieweg & Sohn Braunschweig
Prosperetti, A., 1977, Thermal effects and damping mechanisms for the forced radial oscillation of gas bubbles in
liquids, J. Acoust. Soc. Am. 61 17
Suslick, K.S., 1988, Ultrasound." its chemical, physical and biological effects, VCH Publishers New York
Young, F.R., 1989, Cavitation, McGraw-Hill Book Company London
Ziff, R.M., 1991, Phys.A. Math.Gen. 27, 2821