RESEARCH ARTICLE | JUNE 16 2004

Electrodynamic Containment of Charged Particles

R. F. Wuerker; H. Shelton; R. V. Langmuir

Journal of Applied Physics 30, 342–349 (1959)

https://doi.org/10.1063/1.1735165

CrossMark

 
View Export
Online Citation

Articles You May Be Interested In

To enhance the purity and crystallinity of magnetite nanoparticles prepared by surfactant-free

electrochemical method by imposing higher voltage

Downloaded from http://pubs.aip.org/aip/jap/article-pdf/30/3/342/7926269/342_1_online.pdf

AIP Conference Proceedings (February 2014)

Characterization of nanocrystalline ZnO:Al films by sol-gel spin coating method

AIP Conference Proceedings (September 2014)

Formation and Investigation of Corona Charged Films from Polylactic Acid

AIP Conference Proceedings (January 2010)
JOURNAL OF APPLIED PHYSICS VOLUME 30, NUMBER 3 MARCH, 1959

Electrodynamic Containment of Charged Particles

R. F. WUERKER, H. SHELTON, AND R. V. LANGMUIR*
Ramo-Wooldridge Research Laboratory, Los Angeles 45, Californiat
(Received August 1, 1958; revised September 2,1958)

Electrically charged iron and aluminum particles having diameters of a few microns have been contained
in a confined region of space by means of alternating and static electric fields. The theory is essentially that
of alternating gradient focusing; here the motion is governed by Mathieu's equation. Under certain circum-
stances when many particles are confined the three dimensional focusing force and the Coulomb repulsion
results in a "crystaline" array which can be "melted" and reformed.

STUDIES in this laboratory on the electrical charging

in vacuum of small dust particles resulted in the
The negative 2: 1 ratio between the vanatlOn with
distance of fields in the two directions shows that when
development of a method of containing these particles the electric field is focusing toward the origin in the z
in dynamic equilibrium by alternating electric fields. 1 ,2 direction then it must be defocusing in the r direction
This new technique is based upon the strong focusing and vice versa.
principle which has as its analogue the classical problem The differential equations of a motion of particle of

Downloaded from http://pubs.aip.org/aip/jap/article-pdf/30/3/342/7926269/342_1_online.pdf

of the upside-down pendulum. Essentially this device charge-to-mass ratio elm in this potential field are
can be thought of as a closed form of W. Paul's and
M. Raether's electric mass filter.3 The new variation can
be employed to suspend any charged particle (e.g.,
rPz
-=- (e) (2VdC) z+ (e)
- --- (2Vac)
- - zcosQt, (4)
df m Z02 m zo2
charged dust, ions, electrons, etc.) in dynamic
equilibrium. 4 and
The necessary condition for the proper operation of
this type of suspension system depends upon finding dr=+ (e- ) (VdC)
-.
2

- r- (e- )(Vac)
- rcosQt. (5)
electrode configurations which give sinusoidally time dt- m Z02 m zo2
varying forces whose strengths are proportional to the
distance from a central origin. Under this condition, The equations of motion of a single particle in the two
the differential equation of particle motion is a special directions of space are seen to be identical except for the
case of the Mathieu differential equation. A three negative 2: 1 ratio between the constants. Equation (4)
dimensional electrical configuration which satisfies this is a function of z only while (5) is a function of r only.
requirement is the circularly symmetric potential The motions in z aild r are therefore mutually in-
dis tribu tion dependent. Each of the above equations is thus a
special case of the Mathieu differential equation which
in its general form is usually written
(1)
rPu (6)
-+(a-2q cos2x)u=O
where Vac is the peak value of the alternating signal of dx?-
angular frequency Q applied in series with the constant
voltage V dc. Differentiation shows that the field in which u may represent either z or r. The dimension-
intensities have the required space dependence and are less constants in the above equation are related to those
of the present physical problem through the trans-
z formation equations
E z =2(- Vdc +Vac cosW)- (2)
x=QtI2, (7)
Z02
and
r
Er= - (- V dc + Vac cosQt)-. (3) (8)
Z02
and
* California Institute of Technology. Consultant to Ramo-
Wooldridge. (9)
t A division of Thompson-Ramo-Wooldridge Company.
1 Shelton, Wuerker, and Langmuir, Bull. Am. Phys. Soc. Ser.
II, 2, 375 (1957).
2 H. C. Corben, Bull. Am. Phys. Soc. Ser. II, 2, 375 (1957). Clearly motion with negative q is the same as with
3 W. Paul and M. Raether, Z. Physik 140, 262-273 (1955). positive q except for a change of phase of the drive.
4 It is understood that Professor W. Paul has independently
been conrlucting similar experiments on the containment of Physically the Mathieu differential equation demon-
atomic particles (unpublished). strates that when the alternating driving force 2q
342
 E LEe T ROD Y N A M ICC 0 N T A I N MEN T 0 F C H A R G EPA R TIC L E S 343

vanishes the particle will oscillate with a natural 0.2

01 02 03 04 05 06 07 08 09 10 II 12 13
harmonic motion of frequency equal to (a)t.
O. I
flJ·o .,ot>'C
The Mathieu Eq. (6) is solvable in an infinite series ,..,.o.t>-"f.-o'>' \
0
00 00 I~ .....•.... '.9.~~
u=Ae''' L C2nEi2n"+BE-I''' L
_00
C2nE-i2n" (10) -0
.It. ~No
"'~~~ ~;<9. ~
-0.2 ~
{~
n=-OO

~-,..~~
~
in which the quantities f.L and C 2n are functions of the -0

parameters a and q. The exponent f.L in the above

equation is all important in that it determines the two
-0.4

-0. 5
.
a-
~
...
a
"'~R: ~"-
,,~ ~'ll
-6' . ·0 ~
different types of solutions. Thus if f.L is either a real or "jo 'b. 0 )
complex number the amplitude builds up exponentially
-0.6 f- r-qz=2qr'4(£-)(~) I~2--
I I I
'\~.'
~,

and the particle is not stably bound. If on the other -0.7

hand f.L=i{j the motion is bounded and the particle FIG. 1. Normalized stability diagram for the axially symmetric
remains in dynamic equilibrium. For stable contain- electrodynamic suspension system. The solid line passes through
the theoretical values. The three curves within the region of
ment the quantity {jx is related to the frequency of the stability are the experimentally determined loci along which the
dominant term in the series solution, and, as pointed resultant motions are, respectively, in the following ratios
out earlier, it is equal to (a)t when q is zero. Thus, for fJ./fJr=2, 1, and ;. The probable error is indicated by the size of

Downloaded from http://pubs.aip.org/aip/jap/article-pdf/30/3/342/7926269/342_1_online.pdf

the circles.
the present physical problem {jx is related to the funda-
mental frequency of particle motion in the two orthog- region of stable operation represent experimentally
onal directions w.={j.xlt and wr={jrxlt. Using the determined values which will be fully described later.
transformation Eq. (7) between the normalized in- Operation of the present system is specified in a-q
dependent variable (x) and time (t) gives the funda-
space by a straight line intercepting the origin and of
mental frequencies in the two directions as a function slope
of their respective {j's and the driving frequency Q, (13)
al q= 2Vdcl Vac.
i.e., w.={j.Q/2 and W r={jrQ/2. Accordingly fJ or its
implied w will be referred to as the "resultant frequency That is to say, for given applied voltages Vdc and Vac
of motion." this operational line determines the range of driVIng
It can be shown that the particle will be stably frequency Q through which a particle of given elm
bound if the values of a and q are within the region will be stably bound. Conversely if the frequency is
bounded by the curves also held constant the al q operational line specifies the
range of elm values which will be accepted (i.e.,
higher elm particles correspond to higher q values).
Thus it can be seen that by properly adjusting the
ratio of the two voltages so that the operational line
(corresponding to the special case of (j=0), and just passes one of the edges of the stability curve the
elm acceptance of the chamber can be made quite
1 1 1 11 narrow. It is with a similar operation that Professor
a=l-q--q2+-t--q4---qfi+ (12) W. Paul made his channel device into a sensitive mass
8 64 1536 35864 spectrometer.3
When no direct voltage is added in series with the
(corresponding to {j= 1).6
drive the operation of the system is specified along the
The question of the range of a-q values necessary for
abscissa (a=O) of Fig. 1. Solution of Eq. (12) for a=O
the stable confinement of a particle in three dimensions
shows that the maximum value of q for stability is
can be represented by plotting Eqs. (11) and (12) for
qma" = 0.908; hence the minimum applied driving fre-
the ,. and z direction of the cylindrical coordinate
quency according to Eq. (9) is
system on a single Cartesian graph. Figure 1 shows the
resulting "necktie" stability plot for the axially sym-
metric potential distribution of Eq. (1). The stable
a., ar, q., and qr values are contained within the solid
boundary curves (corresponding to (jr=O, f3r= 1,
Q. min = 1.484[ (~) (2:200) r (14)

The evaluation of this expression for the specific case

fJ.=O, and fJ.= 1) which have been drawn through the of 2Vaolz02= 1000 v/cm2 is presented in Table I in order
theoretical values. 6 The points and curves within the to illustrate range of frequencies encountered with
'N. W. McLachlan, Theory and Applications of Mathieu particles of different charge-to-mass ratios. The theory
Functions (Oxford Universitr Press, New York, 1947). further predicts the existence of other narrow regions
S The values for Eqs. (11) and (12) were taken from published of bounded operation, corresponding to values of f3> 1,
tables. Computation Laboratory, U. S. Bureau of Standards,
Tables Relating to Mathieu Function (Columbia University Press, but to date the existence of these higher modes has not
New York, 1951). been experimentally verified.
344 WUERKER, SHELTON, AND LANGMUIR

TABLE 1. ing the il.1stantaneous acceleration over a cycle of the

drive
(2 V.d.,' = 1000 ./em')
d Z = d z =~f" d zdx= _ (a+ q2)Z.
2 2 2
elm (19)
coulombs/ i=flmi,,/2 ..
Particle kilogram CY/sec dx2 dx2 7r 0 dx2 2
Electron l.76X1011 313 Mc Under the assumed approximations, the apparent
Proton 9.6X10 7 7.31 Mc motion has been reduced to a differential equation of
Uranium 238 4.0SX10 5 478 Kc
harmonic motion, and our approximate theory shows
that the particle will vibrate stably with a resultant
20-1' diam aluminum particle carrying dimensionless frequency of motion of
0.35 million units of elementary
charge, and thus charged to 51 v 0.005 S3 cy/sec
(20)
The preceding paragraphs have summarized some
of the mathematical methods and salient results In the special case of a= 0 the particle will vibrate in
obtained from Mathieu equation theory. As with many the potential field of Eq. (1) with resultant frequency
such nonlinear problems, physical intuition is easily of motion of

Downloaded from http://pubs.aip.org/aip/jap/article-pdf/30/3/342/7926269/342_1_online.pdf

lost in the mathematical manipulations. A simplified
approximate solution when {3 and q are small has been
found which helps one more easily to comprehend the (me) (Va")
wz ={3zfJ/2=-./'l - - (l)radians
---
n second
Z02
(21)
physical principles governing the operation of the and
electrodynamic suspension system. Referring again to
Eq. (6) we assume that the particle position as repre- w r ={3,n/2=-
1 (~)- - (V00) (l)radians
- --. (22)
sented by z can be separated into two components
-./'l m Z02 n second
z=Z+o (15) These approximations are valid when q;S0.4. The 2: 1
ratio between the two frequencies of resultant motion
such that 0, the smaller component, is a displacement correspond to the 2: 1 ratio between the electric field
of small amplitude governed by the periodically strengths in the two orthogonal directions. In the
varying applied driving force 2q, while Z describes the absence of series dc voltage and for small values of q
average value of the displacement over a complete the particle is expected to vibrate in a characteristic
period of the drive. Z will change much more slowly, 2: 1 Lissajous pattern.
but with a larger amplitude than 0, and its values define The addition of direct voltage in series with the drive
the resultant motion. Thus the solution, when stable, strengthens the effective binding in one direction at the
will consist of a slow vibration upon which is super- expense of the other with the result that the resultant
imposed a smaller ripple due to the drive. Substitution vibrational frequencies will be correspondingly altered.
of Eq. (15) into the Mathieu equation enables one to For example, the application of a series voltage which
effect a simplification under the assumption that results in a force toward the axis along the r direction
0<<Z, but do/dt»dZ/dt acts to make up for the inherent geometrical weakness
rPo in this direction. Thus it is expected that the proper
-= - (a-2q cos2x)Z. (16) addition of a series voltage can cause the particle to
dx 2 vibrate with equal resultant motions in both directions
(i.e., wz=w r ), and the trajectory will have the over-all
If a«q the above equation is easily integrated under the appearance of a 1: 1 or circular Lissajous pattern.
assumed relative constancy of Z The approximate theory shows that this condition will
occur when a z= - qz2/4. A further increase in r focusing
qZ
0= --(cos2x) (17) will increase the resultant frequency of motion in the r
2 direction while decreasing the resultant frequency in
the z direction, and one finds a condition in which
demonstrating that when the particle is stably bound the particle will vibrate on the average twice as fast
the micromotion is 180 0 out of phase with the drive. in tht:. r direction as in the z direction (i.e., wr = 2w z ).
Substitution of Eqs. (17) and (15) back into the original The particle trajectory then will have the appearance
differential equation gives of a 1: 2 Lissajous pattern. The approximate theory
shows that this condition will occur when az = -5qlj12.
d2z aqZ Further increase in r focusing will eventually cause the
- = -aZ+- cos2x+2qZ cOS2X-q2Z cos22x. (18)
dx2 2 static field to exactly cancel out the binding effect of
the driving frequency in the z direction (i.e., w.=O).
The resultant acceleration, d2Z/dx2, is found by averag- On setting {3=0 in Eq. (20) it is seen, referring back
 E LEe T ROD Y N A M ICC 0 N T A I N MEN T 0 F C H A R G EPA R TIC L E S 345

to Eq. (11), that the approximate analysis has given

the first term in the expression for the lower boundary
of stability.
We have seen that the electrodynamic suspension
system is able to compete against anharmonic forces
(corresponding to negative values of a) and still
maintain a particle in equilibrium. The case of uniform
forces such as gravity or constant electric fields will
now be considered. Such forces modify the original
Mathieu differential equation giving
d2z
-+(a-2q cos2x)z=A (23)
dx2
where A, the normalized constant force, is related to
the physical problem by
(24)

Downloaded from http://pubs.aip.org/aip/jap/article-pdf/30/3/342/7926269/342_1_online.pdf

the physical force being represented by F. The complete
solution of this differential equation is the sum of the
particular integral corresponding to A and the comple-
mentary function [Eq. (10)]. However, by using the FIG. 3. Photograph of the experimental electrodynamic
suspension system.
approximate solution outlined in the previous para-
graphs one is able to obtain an approximate expression
for the particular integral in closed form; namely physical problem this displacement is
A=F/tnw z2 • (27)
d2Z ( q2)
-+ a+- Z=A. (25) If this displacement is equal to the dimensions of the
dx2 2
apparatus establishing the potential field, the particle
is lost. For example if the force is due to gravity the
The normalized solution of this equation is
particle can "fall out" when the resultant frequency of
motion becomes too small. Gravity will accordingly
(26) slightly alter the appearance of the lower stability
curves of Fig. 1.

It may be seen that the uniform force displaces the EXPERIMENTAL ELECTRODYNAMIC SUSPENSION
center of motion by an amount proportional to its APPARATUS
magnitude and inversely proportional to the square A small experimental chamber which gave the
of the resultant frequency of motion. In the present required potential distribution was machined out of
aluminum. This, in turn, was mounted above an electric
powder injector, and the whole apparatus was mounted
within a vacuum envelope. Figure 2 shows schematically
the complete apparatus with the electrical circuit
through which the driving voltage (Vae ), the series
voltage (Vde), and the "uniform" voltage V g are applied
to the trapping chamber. As can be seen, the driving
signal is applied between the end caps and the annular
ring by a variable dc voltage and ac audiogenerator.
In addition, means were provided for applying an
alternating voltage V/l across the end caps for the
purpose of experimentally measuring the resultant
frequency of motion in the z direction. As shown, ports
were drilled in the caps and the ring electrode for the
POWDER
INJECTION
purpose of microscopically examining the interior,
PULSE introducing the dust particles, the particle charging
FIG. 2. Schematic diagram of the electrodynamic current, and the carbon arc illumination. Figure 3
suspension system. shows a photograph of the apparatus as it is mounted
346 WUERKER, SHELTON, AND LANGMUIR

injected into the interior by pulsing the bucket of the

injector to ground potential for a few milliseconds
through a thyratron circuit. Observation perpendicular
to the axis of the chamber through the horizontal
microscope enables one to observe visually a contained
particle which has been charged by the beam and is
bobbing around the interior of the chamber under the
confining influence of the alternating electric fields.
Figure 5 shows a photomicrograph of a charged piece of
aluminum dust (around 20 J.ldiam) when a/q=0.7 It
will be noted that the trajectory is a 2: 1 Lissajous
pattern upon which is superimposed the driving
frequency.
As expected from the theoretical analysis, the
addition of voltage (V g) across the end caps displaces
the particles either up or down depending upon the
polarity and sign of charge of the particle. By properly

Downloaded from http://pubs.aip.org/aip/jap/article-pdf/30/3/342/7926269/342_1_online.pdf

I. 0

o. 9
FIG. 4. Photograph of the complete apparatus for studying the
electrodynamic suspension of charged dust particles. o. 8 1
upon the vacuum face plate. These components are
o.7 I
contained in turn within a brass vacuum envelope o. 6
(/
J .,-
"
which mounts the microscope tubes, the illumination
port, and the charging gun. (See Fig. 4.) 5
p'"/
.J- /"

The experimental operation of the suspension system .,-

is as follows: first the chamber is evacuated to a .4 L
W"
pressure less than 20 J.l of Hg, the driving signal is ~
applied between the ring and end caps (initially the
driving frequency is around 150 cy/sec with an ampli- o.2 /
tude of Vac/v'1 = 500 v rms), the carbon arc illuminator .c/
V
o.I
is struck, and the charging current is turned on. The
chamber is then ready to accept the powder which is "/
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

QZ'4(-M( ~:~)(~Z)-
FIG. 6. The normalized resultant frequency of motion
fJ as function of q when a=O.

adjusting this voltage gravity can be canceled out, the

sign of the charge of the particle can be determined and
the elm ratio can be estimated [using Eq. (27)]. When
gravity is neutralized the driving frequency can be
extended to very high values without having the
particle fall out of the chamber. By applying a small
alternating voltage (VIl ) of approximately 0.1 v rms
across the end caps using a second oscillator the
resultant frequency of motion in the z direction can be
measured by visually observing when the particle
motion is in resonance with the applied signal. A small
FIG. 5. Retouched microphotograph of the 2: 1 Lissajous difference between the two frequencies results in an
trajectory in the T - z plane of a single charged particle of alumi- easily observed beat frequency which disappears when
num dust contained by the electrodynamic suspension system.
V",,=SOO v rms, Vdc=O, n/2,..=200 cy/sec, and ",./2,..=16.3 7 The photomicrographs were taken with a Leitz MIKAS
cy /sec. One calculates from these experimentally measured 3S-mm microscope camera using Kodak Tri-X film. The speed of
values: e/m=0.OOS3 coulombs/kg, a.=O, q.=0.232, and fJ. film was increased by developing the negatives with FR X-SOO
=0.163. Developer. Exposure times were typically between iol sec.
 E LEe T ROD Y N A M ICC 0 N T A IN MEN T 0 F C H A R G EPA R TIC L E S 347

the two are exactly equal. At resonance the orbit elongates

by an amount determined by the background vacuum FIG. 8. Retouched
microphotograph of a
pressure. Figure 6 shows the experimental measure- 1: 2 Lissajous trajectory
ments of the normalized resultant frequency of z viewed in the y-z plane.
motion as a function of the parameter q when Vdc=O Vac=500 v rms, 2Vdc
= -144 volts, 0= 148
(i.e., a/ q= 0). These data were obtained by varying cy /sec, and elm
both the driving frequency and driving voltage. =0.00625 coulombs/kg
Inspection of this graph shows that the approximate gram. One calculates
from these experimental
theory begins to fail by more than 1% when q;GOA. values that qz=0.502,
The product of the resultant frequency of motion and az= -0.102, and az/qz
= -0.204.
drive frequency is directly proportional to the product
of the charge to mass ratio and the gradient of the
driving field intensity in the z direction in accordance
with Eq. (20). This simplified relationship enables one the verge of no longer being stable. Besides experi-
to measure accurately the charge to mass ratio of the mentally tracing out the region of stable operation,
particle knowing the driving voltage and chamber one can further determine the loci of points where the
dimensions. As the driving frequency is progressively resultant motions in the two directions are in 2: 1, 1: 1,
decreased, the resultant motion speeds up, the ripple and 1: 2 ratios as seen in Fig. 1. The determination of

Downloaded from http://pubs.aip.org/aip/jap/article-pdf/30/3/342/7926269/342_1_online.pdf

due to drive becomes more prominent, and finally the these curves becomes a little difficult for large values
of q.
In essence, using single particles, the above apparatus
can be employed as an analogue computer of the
Mathieu differential equation. It should be mentioned
that once a particle is trapped within the chamber it
stays. Some have been held in suspension for as long
as a weekend.

Containment of Many Particles

Besides verifying the theory concerning the con-
tainment of a single particle the electrodynamic
suspension system also showed that it could compete
against interparticle Coulomb forces. Experimentally
it was found that when the powder was initially
FIG. 7. Retouched microphotograph of a 1: 1 Lissajous Trajec- injected and charged, instead of accepting a single
tory in the r-z plane. Vac=500 v rms, 2Vdc= -90.2 v, 0= 148
cy/sec, and e/m=0.OO625 coulombs/kg. One calculates from
particle the chamber filled up with many. This behavior
these experimental values that qz=0.502, az= -0.0643, and in fact was the more usual case. At first the motions
az/qz= -0.128. were quite violent and mixed up; however, on dissi-
pating the initial kinetic energies by increasing the
motion builds up in the z (vertical) direction and the
background pressure to several microns the particles
particle is thrown out of the chamber.
could be made to take up stable arrays. Arrays of more
Measurement of the frequency of the "upper region
of stability" gives a third way of estimating the charge particles than could easily be counted have been seen.
to mass ratio of the particle using Eq. (14).
The addition of voltage (V dc ) in series with the drive
Vac enables one to explore experimentally the full
region of stable operation of this suspension system FIG. 9 Retouched micro-
as mapped in Fig. 1. Figure 7 shows a microphotograph photograph of a single
of a single particle trajectory which is like a 1: 1 particle trajectory viewed
in the y-z plane deep
Lissajous pattern. Figure 8 shows the 1: 2 Lissajous within the region of stable
pattern obtained by further increasing the series operation. Vac=500 v rms
2Vdc= -269 volts, 0=92.5
voltage (V dc ). Figure 9 shows the "perverse" trajectory cycles/second, and elm
down deep within the region of stable operation. =0.00623 coulomb/kg. The
values of the normalized
The experimental points on the boundary curves of parameters are q,= 1.29,
the stability diagram are plotted in Fig. 1. It was az= -0.49, and az/q,
= -0.38.
possible to measure a point without loss of the charged
particles by careful manipulation of the voltages and
frequency and by recognizing when the motion was on
348 WUERKER, SHELTON, AND LANGMUIR

In Fig. 10 the effect, as seen in the r- z plane, of the

series voltage (Vde) on a typically stable array of
positively charged aluminum particles can be seen.
Each line corresponds to a single particle which is
vibrating about its stable position under the influence
of the alternating field. One notes that on proceeding
away from the center the amplitude of vibration
increases. The orientation of the individual particle
vibrations serves to map out the electric field pattern
within the chamber. Most interesting is the effect of
varying either the driving frequency or the driving
voltage. If the frequency is increased (or conversely
Vae decreased) the "cloud" of particles expands to a
new equilibrium. If on the other hand the frequency
(a)
is decreased the array is compressed. As the drive is
further decreased the array is progressively compressed
and the individual particle motions get larger. At some
point (which is a function of the vacuum pressure)

Downloaded from http://pubs.aip.org/aip/jap/article-pdf/30/3/342/7926269/342_1_online.pdf

the static array "melts" and the particles again move
around in a random fashion in large orbits. If the
frequency is increased the particles "recrystallize."
The process of "melting and recrystallization" can
be repeated over and over again. The time of "re-
(b)
crystallization" is a function of the number of particles,
background pressure, and the amplitude and frequency
of the drive. If the pressure is quite high (around
S-10,u) and the frequency also large (q<0.2) the
recrystallization time will be of the order of seconds.
(e)
If, conversely, the pressure is lowered to around 0.1 Jl
then the static array may take many minutes to re-
FIG. lO. Many body suspension viewed in the r-z plane. establish itself.
Experimental values: V.c =500 v rms, D=210 cy/sec. (a) Top
picture: 2Vdc=-30 v, w z =21.7 cy/second; (b) Middle picture: The array of Fig. 10 is unique due to its symmetry
2Vdc=0, w z =26.0 cy/sec; (c) Bottom picture: 2Vdc= +24.7 V, (gravity in this case was canceled out by the application
w,= 29.4 cy/sec. The average charge to mass ratio of a single
particle was e/m=0.OO765 coulombs/kg. of V g =20 v). In most cases a wide range of elm is

TABLE II.

n~
--
B
2,,- Number of particles
cy/sec ~IOO 32 5 3 q

2lO 0.248 0.229 0.220 0.220 0.221 0.221-0.307

200 0.272 0.257 0.247 0.248 0.246-0.339
190 0.305 0.284 0.275 0.273 0.273 0.273-0.376
180 0.319 0.309 0.309 0.304-0.418
170 0.390 0.362 0.347 0.347 0.345 0.347-0.469
160 0.415 0.400 0.400 0.400 0.396-0.530
150 0.530 0.481 0.465 0.465 0.456-0.602
145 0.503 0.504 0.645
140 melts
at 143
cy/sec 0.579 0.555 0.551 0.547 0.545-0.691
13S 0.645 0.610 0.609 0.603-0.744
130 melts
at 131
cy/sec 0.692 0.690 0.692 0.680-0.801
125 melts
at 125
cy/sec 0.806 0.799 0.785-0.866
melts
at 122 unstable--o.907
cy/sec 121 cy /sec
 E LEe T ROD Y N A M ICC 0 N T A I N MEN T 0 F C H A R G EPA R TIC L E S 349

accepted and the static pattern has the appearance of FIG. 12. Suspen-
a stalactite with the particles of high elm on top and sion of five positively
those of progressively lower elm ratios dangling below. charged particles
viewed in the r-8
The high elm particles can be rejected by decreasing plane. V'C=500 v
the frequency until they are unstable or conversely rms,2Vde=O,n=210
the low elm particles dropped out by increasing the cy/sec, and w,=23.1
cy /sec. The charge
frequency. By using the series voltage (Vdc ) as described to mass ratio of a
in the theory only particles within a narrow band can single particle was
later found to be elm
be kept. =0.00765 coulomb/
Using the second oscillator the resonance frequency kg.
of the cloud can be investigated by observing when the
particles absorb energy. Starting with the uniform
a top view of the five particles for which data were taken
cloud of Fig. 10 the frequency was measured as a
in Table II. Note that each particle lies at apex of a
function of the drive-also the "melting frequency" was
regular pentagon. In the three particle case the particles
noted. After a run was completed some particles were
were bound in an equilateral triangle in y-{} plane.
Besides containing particles of only one sign we have
seen the simultaneops containment of particles of both

Downloaded from http://pubs.aip.org/aip/jap/article-pdf/30/3/342/7926269/342_1_online.pdf

sign. In this case when V g is added across the top caps
the particles are caused to move vertically in opposite
directions.
HIGH-FREQUENCY-EXCITATION
The audio source used in the previously described
work was replaced by a 300 megacycle source which
theoretically would confine electrons. Experiment
demonstrated that excitation by the high frequency
resulted in the production of a visible glow inside the
chamber. The vacuum was such that an electron mean-
free path was about 1000 t.imes the chamber dimensions.
The glow could be extinguished by addition of dc
voltage in series with the driving signal. Details of this
FIG. 11. Suspension of 32 positively charged particles viewed
work will be reported at a later date.t
in the r-8 plane. Vae=500 v rms, 2V'=O,n= 135 cy/sec,w.=43.6
cy/sec. The average charge to mass ratio of a single particle was ACKNOWLEDGMENTS
e/m=O.00765 coulomb/kg. We wish to thank Dr. David B. Langmuir and Dr.
H. C. Corben for their many helpful discussions and
thrown out of the chamber and a new run taken by encouragement. Also we want to thank Mr. Richard L.
changing the frequency. This process was repeated Young for his assistance in the construction of some
progressively until only one particle was left. Table II of the equipment and making some of the measure-
shows the experimental results. It is seen that for a ments, and Mrs. Virginia Gannon for her assistance in
given drive "{3" is a function of the number of particles. the preparation of the manuscript.
The values q listed on the right were computed from
the one particle data. t Note added in proof.-The following references have recently
come to our attention: (a) M. L. Good, Univ. Calif. Rad. Lab.
Figure 11 shows a microphotographic view in the Report No. 4146 (1953). Declassified (1956). (b) Fisher, Osberg-
y - {} plane of the 32 particles for which data were taken haus, and Paul, Forschungsber. Wirtsch. Ministeriums Nordrhein
Westfalen, No. 415 (1958). (c) Paul, Reinhard, and von Zahn, Z.
in Table II. This top view is typical of the observed Physik 152, 143-182 (1958). (d) E. Fisher, thesis, University of
"crystalline" arrays of many particles. Figure 12 shows Bonn (unpublished) (1958).