PULSATION AND VIBRATION

4. For practical application, it must be possible to produce or Conclusions

distribute the required sonic energy in all parts of the bed, which
may be large. This specification, combined with the well-known
rapid absorption of sonic energy in porous materials, would
suggest multiple small sources, rather than a single large source.
Of the various known methods of producing sonic energy,
only three seem a t all likely to be practical, namely: sirens,
whistles, and vibrating diaphragms.
Both sirens and whistles can be designed to produce high
pressure sonic energy at low frequency, but they suffer from the
serious disadvantage of requiring very large flow rates of gas.
I n many cases, this would make these devices impractical for
promoting fluidization without special arrangements to separate
sound-producing gas from the fluidization gas while simultane-
ously transmitting vibrations to the boiling bed. This complica-
a tion becomes particularly serious if multiple sound-producing
units are to be used in large beds. Furthermore, high power
sirens and whistles are so bulky that their use would be difficult.
+
Vibrating diaphragm units seem much more likely to be suc-
cessful for this application. Those presently on the market
in the form of loud-speakers are unsuitable because of low power
rating and difficulty of isolating from reaction conditions. How-
ever, the principles of design of vibrating diaphragm devices
are sufficiently well understood that special units can be built,
perhaps embodying as power sources high power, low frequency
vibrators of the types used for vibrating bins and conveying dry
powders; such vibrators are made both for air actuation and
for electrical powering. Such units could be designed with
their actuating elements enclosed between double vibrating dia-
phragms, thereby exposing only the diaphragms, cables, and pres-
sure-balancing line to the materials in the bed. It seems reason-
able to believe that such units could even be made to withstand
moderate temperatures.
This work shows results to be expected in the application of
sonic energy t o the promotion of fluidization. The mechanism
by which sonic energy unlocks the particles and enables them to
be fluidized vi11 be an interesting field for further research.
Sonic energy will cause nonfluent materials of small particle
size to flow sufficiently well that good fluidization is possible
without channeling and stagnation. With materials which are
normally fluent, the effect, if any, on fluidization is too small to
be observed visually, or found with the capacitometer. The
sound pressure used must be above some minimum value peculiar
to the material but certainly sufficient to secure approximately
110 decibels after absorption in beds about 1 foot deep. The
sound frequency in the range from 40 to a t least 400 cycles per
second is not important as long as the sound pressure is adequate.
Through resonance with the natural periods of the apparatus,
the use of a particular frequency will permit higher sound pres-
sure to be secured from a given sound energy source.
Vibration of the column is also effective in promoting the flow
of nonfluent materials of small particle size, but its effectiveness
may be of limited range because of the poor mechanical coupling
between walls and particles, and the consequent vibration damp-
ing in the intermediate layers of material. If the vibration can
be transferred t o the fluid, it becomes sonic energy.
literature Cited
(1) Lamb, H., “Dynamical Theory of Sound,” 1st ed., Article 360a,
Arnold Publishers, London, 1910.
(2) Morse, R. D., and Ballou, C . O., Chem. Eng. Progr., 47, 199-204
(1951).
(3) Morse, R.D., and Von Wettberg, E. F., Jr., U. S. Patent 2,667,-
706 (Feb. 2, 1954).
for review December 20, 1954.
RECEIVED March 19, 1955.
ACCEPTED

Effect of Vibration on Natural

Convective Heat Transfer
ROBERT LEMLICH
Deparfmenf o f Chemical Engineering,
Universify o f Cincinnafi, Cincinnafi 2 I , O h i o

In order to study the effects of vibration on heat transfer, an experimental investigation was
carried out involving natural convection from electrically heated wires of three different diam-
eters subjected. to transverse vibration in air. Marked improvement in the coefficient of heat
transfer was obtained b y using vibration in the range of 39 to 122 cycles per second, even to
the extent of quadrupling the film coefficient. An increase in coefficient was observed for an
increase in amplitude or an increase in frequency. No effect was observed for a change in
the direction of vibration. In an effort to account for this latter observation, the concept of a
stretched film surrounding the entire path of vibration was proposed. In addition, a pair of
dimensionless correlations based on a vibrational Reynolds number were presented for vibration
in air or other diatomic gases. An approximate extension to other fluids was also proposed.

0 F THE many variables that affect the rate of heat transfer,

the effects associated with vibration are among the least
known. Patents on the subject ( 1 , 12, 14-16, 18) offer little
quantitative data. However, the meager information that is
available indicates that considerable improvement in the rate of
heat transfer is possible u-here vibration is employed. West
and Taylor ( 1 7 ) obtained improvements to 70y0in heat transfer
coefficient by applying 100 pulses per minute t o water in a
double pipe exchanger. Martinelli and Boelter (10, 1 1 ) reported
increases in coefficient to fivefold by vibrating a 0.75-inch cyl-
inder in natural convection under water. However, Boelter
( 8 ) indicated that results obtained from a later check on some
of the experimental data by Mason (IS)did not, agree with the
original.
Accordingly, the present study is an investigation of the effect
of vibration on heat transfer in a simple experimental system in

June 1955 INDUSTRIAL AND ENGINEERING CHEMISTRY 1175

 ENGINEERING, DESIGN, AND PROCESS DEVELOPMENT
which all the variables may be independently measured and
their effects incorporated in an over-all dimensionless correlation.
Tabulation of data and results together with a more extended
discussion of this investigation is available in the dissertation (7).
Experimental
The system selected was that of natural convection from a
horizontal, electrically heated Nichrome wire, undergoing trans-
verse vibration in room air. As shown in Figure 1, the wire was
stretched between a pair of bridges 36.3 inches apart and allowed
to vibrate. Vibration was induced by connecting the armature
of an electric interrupter-type buzzer t o the wire a t a point about
9 inches t o the left pf the right hand bridge. By temporarily
allowing the center of the wire to vibrate a t right angles to a
uniform magnetic field, picking up the induced electromotive
force, amplifying it, and projecting it on an oscilloscope screen,
the vibrational wave form was shown t o be essentially sinusoidal.
POTENTIOMETER- n Results
AM METE R h between extreme positions of the center line over the course of
I
U ranging from 39 to 122 cycles per second; and direction of vi-
IISV. AC
Figure 1. Schematic diagram
The amplitude of vibration, which was controlled by varying
the dry cell voltage to the buzzer or by changing the linkage
between armature and wire, was measured with a calibrated low
power microscope, the precision of which was =!=0.0003 inch.
The frequency of vibration, which was varied by altering the
' tension on the wire, was determined with a stroboscopic light
calibrated to an accuracy of =k1%. The 60-cycle electrical power
input to the wire (and hence the rate of heat dissipation) was
measured with a voltmeter and ammeter; their accuracies were
&2%.
Temperature Measurement. The temperature a t the middle
of the Nichrome wire was measured by a calibrated 0.005-inch-
diameter Chromel-Alumel thermocouple connected to a potenti-
ometer with an accuracy of & l o F. The cold junction was at
room te,mperature so that the temperature difference between
wire surface and room air could be measured directly. Attempts
a t welding a hot junction to the Nichrome wire failed under
sustained vibration. Other techniques such as the use of a
plated junction or the setting of the thermocouple wires in a
groove would also have proved unsatisfactory since they would
have resulted in a local resistance irregularity and, therefore,
a local temperature irregularity a t the very point where the
temperature was being sensed. Accordingly, each wire a t the
hot end of the thermocouple was wrapped around the main
wire so as to give a split hot junction but with a separation of
less than 0.3 inch.
The technique of utilizing the electrical resistance of the main
wire as a measure of the temperature (requiring a wire composi-
tion with a high temperature coefficient of electrical resistance
rather than Nichrome) would also have proved unsatisfactory.
The frequent changes in wire tension required for changing the
frequency would have repeatedly altered the resistance of the
wire, necessitating continual recalibration.
Previous investigators in natural convection (6, 6 ) found that
the error incurred in measuring surface temperature by securing
1176 INDUSTRIAL AND ENGINEERING CHEMISTRY
the thermocouple to the surface in question is small. For the
present work this fact was further checked by temporarily re-
placing the Nichrome wire with a length of electrically heated
hypodermic tubing through which an electrically insulated ther-
mocouple was threaded and centered longitudinally. A split
junction thermocouple of the type described above was then
wrapped around the outside of the tube and also centered longi-
tudinally. For various levels of heat flux, whether the tube was
vibrated or not, the temperature indicated by the outer thermo-
couple was lower than that of the inner by 2 to 3y0 of At, the
difference between surface temperature and ambient temper-
ature. This corresponds to a difference of only 2 to 3% in the
heat transfer coefficient, h. Thus the error is not only small
but roughly constant, so that i t tends to cancel in the final
correlations.
The vibration itself generated no appreciable heat as shown by
the absence of any measurable At when current to the vibrating
wire or tube was shut off.
The effects on h of 5 major variables were investigated in a
total of 100 runs carried out with vibration. These variables
included wire diameter, D,in three different sizes, namely 0.0253,
0.0396, and 0.0810 inch; At ranging from 7" t o 365' F.; amplitude
of vibration a t the center of the wire, H (defined as the distance
of one cycle) ranging from 0.055 to 0.231 inch; frequency, F,
bration, that is, whether the vibration was horizontal or vertical.
In addition, 37 control runs were conducted without vibration.
These latter agreed with McAdams (9) generalized correlation
for horizontal cylinders with an average deviation of about 2 ~ 6 %
for the points themselves and about 3% for the trend.
The value8 of h which were determined under vibration are for
the central portion of the wire where the thermocouple is located
and not for the ?ire as a whole, since the amplitude of vibratory
displacement necessarily varies along the length of the wire.
However, because the central portion of the vibrating wire rep-
resents the top of a sine wave and as such is quite straight (flat)
even a t the crest of a vibration, these values of h which are cal-
culated and later correlated are essentially for vibrating horizon-
tal cylinders.
Despite the fact that vibration increases h and so cools the
middle portion of the wire more than the ends, calculation based
on the low temperature coefficient of electrical resistance for
Nichrome showed that no appreciable error was introduced by
using the average heat dissipation for the entire wire in computing
h. Similarly, numerical analysis, reinforced by auxiliary evidence
of a direct experimental nature, indicated the heat conducted
along the wire was negligible. I n addition, unsteady-state
calculation, as well as considefations based on heatup time,
showed the rapid 60 cycles per second alternations in current
induced no appreciable alternation in wire temperature. Finally,
calculation for the smooth, shiny, Nichrome surface yielded only
a negligible rate of heat transfer by radiation, the effect of which
was reduced even further by cancellation in the ratio of coefficients
appearing in the final correlations. Thus, no spurious heat
effects interfered with the study.
Effects of Vibrational Variables. The effects of amplitude and
frequency are illustrated in Figures 2 and 3 for the 0.0396-inch-
diameter wire. Holding other variables constant, h increases
with both amplitude and frequency of vibration. This is in
accord with the film concept and follows from the increased mag-
nitude and frequency of the disturbance a t the film, which in
turn decreases its thickness. Furthermore, the effect of fre-
quency on h is greater a t a larger amplitude of vibration than a t
a smaller amplitude. This also follows, since increasing the
number of times per second that a large disturbance occurs
Vol. 47, No. 6
-- PULSATION A N D VIBRATION

for vertical vibration would be less than that for hori-

zontal vibration. Now, no difference could be dis-
cerned between the effects of horizontal and vertical
40 vibration. Therefore, it is unlikely that the wire could
3 20 have carried its film back and forth in the manner
t;; indicated in Figure 4,A, but apparently stretched it,
rK as indicated in Figure 4,B.
2. The shape within the stretched film (shown
W
L shaded) would Drobablv have much the same rate of
naturaf conveciive heat transfer in a horizontal posi-
tion as in a vertical position. While the author knows
of no data for such a configuration, correlations ( 4 ) are
available for plane surfaces (Figure 4,D) in air a t .
I - moderate conditions of temperature and ressure
which indicate that for fixed dimensions a n f a t , the
- total natural convective heat loss from both surfaces
-zw Oiometr 0.0396 inch
- of a vertical plate is only 4% higher than that from
both surfaces of a horizontal plate.
2
Y - Frequency 90 vibmtions/ ucand
W
LL 3. A simple experiment was conducted wherein
:: 2
I I I1IIll I I I I, a lighted cigarette was held under one of the wires in
6 810 20 40 60 80 100 eo0 400 horizontal vibration. The smoke appeared to curl up
TEMPERATURE MFFERWGE At IN OF. and around the vibrating path, as illustrated in Fig-
ure 4,E. This would seem to indicate the existence
Figure 2. Effect of amplitude of a stretched film.
These arguments, of course, do not represent
results in a greater over-all disturbance a t the film than would rigorous proof of this stretched film concept. However, they do
result from correspondingly increasing the frequency of a small constitute evidence in its favor.
disturbance.
Correlation
These figures also show that a given increase in amplitude or
frequency produces a smaller increase in h a t higher values of At Attempts were made to correlate the data by utilizing appro-
tJhan a t lower values of At. I n other words, the effect of ampli- priate modifications of a relationship proposed by Martinelli
tude and frequency diminishes as At increases. This can be and Boelter for their work, but without success. Other attempts,
explained in terms of the increased disturbance a t the film due to which included endeavoring t o account for h in terms of a known
natural convection currents produced by a high At. Under these natural convective effect added to known forced convective
coiiditions the additional disturbance caused by an increase effect, also failed.
in amplitude or frequency represents a smaller proportion of the Generalized Correlation. The only method of attack which
total disturbance than it does for the situation with a low At did prove successful was that of correlating dimensionless groups.
wild feebler natural convection currents. I n the extreme a By dimensional analysis these groups are Nusselt number, Nu;
iwgative slope appears a t high amplitude in Figure 2. Grashof number, Gr; Prandtl number, Pr; vibrational Reynolds
Since the data were gathered for an over-all correlation, little number, Re, H I D , and @At where 0. is the thermal coefficient of
a,ttempt was made a t deliberately keeping the parameter of am- volumetric expansion. The Reynolds number is defined in
plitude or frequency constant from run to run. The scatter in terms of the average velocity 0 = 2HF by the expression
Figures 2 and 3 is thus largely due to lack of complete constancy Dtip
Re = - (1)
of parameters from one run to the next, rather than to experi- c1
mental error. where p is density and p is dynamic viscosity. A somewhatj
The direction of vibration has no effect within the limits of similar term was utilized by Martinelli and Boelter.
experimental error. This can best be seen
from Fiaures 5 and 6. where the solid
Y

points represent vertical vibration and 6 30 I I I I I I l l I I 1 I I 1 1 1 1 I I I 1

the hollow points represent horizontal *I-Y.
- -
vibration. No differencein trend between
the t n o types of points can be discerned.
a20
I -
A
Physical Picture
With regard to a physical picture, the
experimental data seem to indicate, at
least as an approximation, that within -
the limits of the present range oE vari- -
ables the vibrating wire does not carry 8 4 - -
the film back and forth with it, as repre-
sented in Figure 4 4 , but rather that the B
t i - Diameter 0.0396 inch -
E
film surrounds the entire vibrating path, LL
I I 1 I 1 1 1 1 I I I I I l l l l I I I I
w 2
as represented in Figure 4,B. The 8 2 4 6 810 20 40 6 0 80 100 200 400 600
TEMPERATURE DIFFERENCE AT IN O F .
evidence to support this contention is
threefold. Figure 3. Effect of frequency
1. If the wire did carry its film back
and forth, there would be something of a
vector summation of velocities a t right angles between the horizon- Combining the groups with considerations based on the data
tally moving wire and the rising convection current, as shown in itself
Figure 4, C. This vector sum would never be less than the ver-
tical current itself, and except for the two instants a t either crest h
F-l='p (2)
of the cycle, would always be greater. On the other hand, the
vector summation for a vertically moving wire would result in a
net addition half the time and in a canceling subtraction during where h' is the heat transfer coefficient for the same diameter
the other half. Thus the average effective velocity over a cycle wire without vibration but a t the same At as with vibration;

lune 1955 INDUSTRIAL AND E N G I N E E R I N G CHEMISTRY 1177

 ENGINEERING, DESIGN, AND PROCESS DEVELOPMENT

- t
8. STRETCHED F I L M

-
A. MOVING FILM
E. SMOKE PATTERN

C. INSTANTANEOUS
VECTOR SUMMATION

Figure 4. Film considerations

that is, a t the same Gr X Pr, since the effect of small changes in the Reynolds number and the generalized correlations. Since
room temperature on the properties of air is slight. Pr for air the room temperature varied only very moderately, an increase
is virtually constant, and the inclusion of H I D was unnecessary. in At corresponded to an increase in average film temperature.
All fluid properties are evaluated a t the arithmetic mean film This in turn decreased density and increased viscosity, so that
temperature. for a fixed amplitude and frequency the Reynolds number de-
Figure 5 shows a plot of Equation 2, the generalized correla- creased. Now, as is evident from the logarithmic nature of the
tion for air. It should also be applicable to other diatomic gases correlations, a t high Reynolds numbers (corresponding to high
as well, since they all have virtually the same Pr. The quantity parametric values in Figure 2 ) even a small decrease in Re or
h/h' - 1 is the fractional increase in coefficient resulting from Re, induces a considerable decrease in h. Thus for the highest
vibration, and is given directly. This improvement reached amplitude in Figure 2 the increase in At actually reduced h
over 3009.'0, which corresponds to a quadrupling of the coefficient. despite the increase in Gr.
Another advantage to having the correlation in terms of the Extension to Other Fluids. I n an effort to extend the correla-
ratio of coefficients in that any errors in their determination tion to include other fluids, the data of Martinelli and Boelter on
tend to cancel each other. I n addition, due to the generality water were employed. Pr was taken into account by multiplying
inherent in the expression h/h' - 1 which involves only the the abscissa of Figure 5 by the quantity 0.80/PrO.'6. For air and
improvement in coefficient rather than the coefficient itself, the other diatomic gases, this reduces to unity (8)leaving the correla-
form of the correlation may also be applicable, at least in a general tion unchanged.
way, to bodies of noncylindrical shape.
The average deviation for the correlation is about &13%,
where % refers to percentage improvement in coefficient. For I I I I I I I l l I I 1 1 1 1 -
a given h' the average error in h itself is about &8%. For im- 4- -
provements in coefficient of over loyo,which covers most of the - WIRE
SYMBOL DIAMETER
VIBRATION
DIRECTION -
experimental range, the curve is closely approximated by 0.0253 INCH VERTICAL
0 0.0'253 HORIZONTAL
h
0.75 + 0.0031 Re2.05 (pAt)0.33
(Gr)o.41 0
0 0.0396
0.0396
VERTICAL
HORIZONTAL
A 0.0810 VERTICAL
The quantity h' may be estimated from the McAdams corre- A 0.0810 HORIZONTAL

lation for stationary horizontal cylinders ( 9 ) for which Bosworth

0.8
( 3 ) has fitted an equation.
An alternative correlation based on the stretched film concept
is also possible by making use of D +
H , the longer linear dimen-
sion within this film, rather than the shorter dimension, D. A
modified Reynolds number may then be defined for the stretched
film by

A plot of h/h' - 1against Re, is shown in Figure 6. The resulting

correlation is simpler than that of Figure 5 , but the scatter is
greater. By fitting a straight line to the results, this alternate
correlation may be expressed by

--1 = 0.00265 Ree2.'3 (5)

h'

Attempts t o utilize some other characteristic stretched film

dimension in the Reynolds number, such as the hydraulic radius,
correlated no better. Nevertheless, the greater simplicity and
partial success of the alternate correlation represent further sup-
port in favor of the stretched film concept.
Referring back to the earlier discussion of Figure 2, the nega-
tive slope in this plot now may be readily explained in terms of Figure 5. Correlation for vibration in air
1173 INDUSTRIAL AND ENGINEERING CHEMISTRY Vol. 41, No. 6
 PULSATION A N D VIBRATION

6
- I I I I I I I l l I I I
-
I l l

WIRE
SYMBOL DIAMETER
VIBRATION
DIRECTION
0.0253INCH VERTICAL "J-
/* 1
1
4 -

7 DATA OF PRESENT AUTHOR

o DATA OF MARTINELLI AND BOELTER & 0
-
7

OM53 HORIZONTAL
0.0396 VERTICAL 2 -
0.0396 HORIZONTAL
0.0810 VERTICAL
0.0810 HORIZONTAL

0.8
-
I -
-
-
0.6 -
-
0.4 -
PRESENT AUTHOR
I
-

L- -I

:I
0.2I 0.2 -
=I==
-

:f
0.1
-
-
0.08
-
0.06 -
-
0.04 0.04 - 1 . 0
I.
-
I I I I
i
I 1 1 1 1 I I I I l l
0.02
2 4 6 810 20 40 60 80

Res
Figure 6. Correlation for vibration in air based o n Figure 7. Comparison with d a t a of Martinelli a n d
stretched film concept Boelter

Figure 7 shows the final plot for both the water data of Marti-
nelli and Boelter and the air data of the present investigator. The
general trend of the two sets of results are in agreement with each
other. A curve based on the combined results and a curve based
on those of the present author alone are shown. The latter curve
is basically the same as that of Figure 5 , but includes the Pr
modification.
With regard to these two curves shown in Figure 7 , if a single
correlation is to be selected for all fluids, both liquid and gaseous,
the curve based on the author's data alone is believed to be the
more reliable. This contention is supported by the greater
scatter among the plotted results based on the work of Martinelli
and Boelter, and the previously mentioned doubt cast on their
validity. Therefore, modifying Equation 3 to include the effect
of Pr and combining with Bosworth's equation yields the follow-
ing for improvements in excess of IO'%:
~
in terms of a vibrational Reynolds number defined as twice the
product of amplitude] frequency, linear dimension] and density,
divided by viscosity.
5 . The improvement in heat transfer coefficient for the central
portions of the vibrating wires which are essentially horizontal
vibrating cylinders in air is dimensionlessly correlated by Figure
5 . For improvements in excess of loyo,this correlation is fitted
by Equation 3. As an alternative, a simpler but less precise
dimensionless correlation based on the stretched film concept is
presented by Figure 6 or Equation 5 .
6. For an approximate extension to other fluids (as well as
for diatotnic gases such as air) a modified dimensionless correla-
tion is presented in Figure 7 . For improvements in heat transfer
coefficient greater than lo%, this modified correlation is fitted
by Equation 6.
Acknowledgment
The author is indebted to William Licht for reviewing the man-
uscript and for his helpful encouragement. I n addition, the
author wishes to thank Erica M. Houser for assisting with some
of the tedious calculations and Roger Keith for preparing the

[0.63 + 0.35(Gr X P r ) o . 1 7 ] 2(6) figures. The experimental work was performed by the author
a t the Polytechnic Institute of Brooklyn.
Thus the solid curve of Figure 7 , or alternately Equation 6 makes
possible an estimation of the effect of vibration on natural con-
vective heat transfer in other fluids as well as in diatomic gases. Nomenclature
D = diameter, f t . or inch
Summary of Conclusions F = frequency of vibration, sec.-1
Gr = Grashof number, dimensionless
1. Vibration can materially increase the coefficient of heat h = coefficient of heat transfer, B.t.u./hr. ft., F. O

transfer, a quadrupling of the coefficient having been obtained. h' = coefficient of heat transfer for same diameter a t same
2 . Holding other variables constant in turn, the heat transfer Pr X Gr as occurs without vibration
coefficient increased with both amplitude and frequency. The H = amplitude of vibration, it. or inch
direction of vibration was without significant effect. Nu = Nusselt number, dimensionless
3 . The film about each wire was apparently stretched to Pr = Prandtl number, dimensionless
cover the entire path of vibration rather than carried back and Re = vibrational Reynolds number, dimensionless
forth by the wire. Re, = vibrational Reynolds number for stretched film, dimen-
4. The effect of the vibrational variables may be correlated sionless

lune 1955 INDUSTRIAL AND ENGINEERING CHEMISTRY 1179

ENGINEERING, DESIGN, A N D PROCESS DEVELOPMENT
At = temperature difference, O F.
V = av. velocity over one cycle, ft./sec.
p = thermal coefficient of volumetric expansion, O F.-1
p = dynamic viscosity, lb./ft. sec.
p = density, lb./cu. f t .
(p = function
Literature Cited
(1) Andreas, A., Ger. Patent 717,766 (Feb. 5, 1942).
(2) Boelter, L. &I. K., personal communication, 1953.
(3) Bosworth, R. C. L., “Heat Transfer Phenomena,” p. 101,
Wiley, New York, 1952.
(4) Ibid., p. 104.
(5) Colburn, A. P., and Hougen, 0. A,, IXD. ENG.CHEM.,22, 522
(1930).
(6) Fa;ber,‘E. A., and Scorah, R. L., Trans. Am. SOC.Mech. Engrs.,
70, 369 (1948).
(7) Lemlich, R., dissertation for Ph.D., University of Cincinnati,
June 1954.
ULTRASONIC CHEMICAL EFFECTS
VIRGINIA GRlFFlNG
The Cafholic Universify o f America, Washington, D. C .
U LTRASONIC chemical effects have been studied to deter-
mine why any chemical reactions occur when a system is
irradiated with a high intensity ultrasonic beam rather than to
concentrate on which reactions go and by what chemical kinetics.
The experimental fact exists t h a t no observable chemical reac-
tions take place unless there is a permanent gas dissolved in the
solution and the sound intensity is sufficiently high to make the
liquid cavitate.
Thus it is assumed that a gas dissolved in the solution will
form minute bubbles in the sound field. The liquid serves as
an efficient transducer which carries the energy of the-high in-
tensity sound wave from the source to the interfaces between the
small bubbles and the liquid. I n the adiabatic compression of a
sound wave, the production of a high temperature is dependent
on the existence of a large difference between the adiabatic and
isothermal compressibility. As the sound wave of 5 to 10 watts
per sq. cm. travels through the liquid, the temperature difference
due to the adiabatic compression is of the order of a few degrees
centigrade while a t these intensities a sniall bubble of gas may be
compressed to half its volume, producing a temperature varia-
tion of several hundred degrees. A temperature gradient in
the gas bubble is then set up, which produces periodic tempera-
ture variation in the liquid adjoining the gas bubble.
If one assumes the model there are two possibilities for explain-
ing the chemical reactions to be effected. The chemical reac-
tions could be gas phase thermal reactions taking place in the gas
bubble or the chemical reactions might take place a t the liquid-
gas interface, but still be due to the high gas temperature in the
bubbles. These same thermal processes are responsible for
the loss of acoustic energy.
According to this idea, two thermal properties of the gas con-
tent of the bubble are important. First, y will determine the
temperature reached in the compressed bubble. As the y of the
gas approaches one the temperature inside the bubble becomes
much lower. This explains why vapors such as ether inhibit these
1180 INDUSTRIAL AND ENGINEERING CHEMISTRY
(8) Mcddams, W. H., “Heat Transmimion,” 2nd ed., p. 242.
McGraw-Hill, Kew York, 1942.
(9) Ibid., p. 243.
(10) Martinelli, R. C., M.S.thesis in mechanical engineering, Univer-
sity of California, 1938.
(11) Slartinelli, R. C., and Boelter, L. M. K., PTOC.6th I n h . Congr.
A p p l . Mech., 578 (1938).
(12) Maschinenfabrik Oerliken, Brit. Patent 532,144 (Jan, 17, 1941).
(13) Mason, W. E., personal communication, 1954.
(14) Ormell, E. A. I., Ger. Patent 736,883 (May 20, 1943).
(15) Robinson, R. S., U. S. Patent 2,514,797(July 11, 1950).
(16) Schrey, A., French Patent 806,030 (Dec. 5, 1936).
(17) West, F. B., and Taylor, A. T., Chem.. Eng. Progr., 48,39 (1952).
(18) Worn, G. A , , and Rubin, F. L., U. S. Patent 2,664,274(Decem-
ber 29, 1953).
RECEIVED for review December 20, 1954. ACCBPTEDApril 1 1 , 1955.
From a dissertation presented b y R. Lemlioh, in partial fulfillment of the
requirements for the doctor of philosophy degree at the University of Cin-
cinnati, June 1954.
reactions even if the solution is saturated with air. Thus one
would expect the rate of a chemical reaction to increase as the y
of the dissolved gas increases whether the reaction occurs in the
gas bubbles or in the liquid phase. Secondly, the thermal con-
ductivity of the gas will determine how long the high tempera-
ture is maintained inside the bubble. If the reaction takes place
inside the bubble the gas with the lower thermal conductivity
will give a higher chemical yield than one with high conductivity.
On the other hand, for any thermal reaction in the liquid phase,
the thermal conductivity of the gas should have the opposite
effect. A series of experiments was conducted to answer these
questions.
The chemical effects directly produced by ultrasonics are ther-
mal gas phase reactions; they take place inside the gas bubbles.
In water solutions, the reaction is the production of radicals by
the thermal decomposition of water-probably OH radicals
.4 similar reaction has not been obtained when water is not
present even though a wide variety of organic substances has been
tried. If another reactive gas or vapor is present in the bubble,
the OH radical attacks the vapor forming other products within
the bubble-e.g., carbon tetrachloride. These products then
diffuse in the solution; various secondary reactions between these
products and the species dissolved in the solution have been
observed. Another secondary effect of the chemical reactions
is luminescence ( 1 ) ; the intensity of luminescence shows the same
general behavior as the chemical yields.
The degradation of polymers, although due to cavitation, is
not due to “hot spots” developed in and around the gas bubbles
It is postulated that polymer degradation is caused by the same
type of force that is so effective in mechanical dispersion and
scrubbing action a t a low frequency of the order of 20 kc.
Although the experiments have conclusively proved that the
chemical reactions take place in the gas phase, it is impossible to
rule out completely that the mechanism may be gaseous dis-
charge inside the bubble rather than temperature Presentlv
the data are being re-examined and new euperiments under-
taken to settle this question.
literature Cited
(1) Griffing, V., and Sette, D., J . Chem. P h y s . , 23, 503 (1955).
Vol. 47, No. 6