INSTITUTE OF PHYSICS PUBLISHING EUROPEAN JOURNAL OF PHYSICS

Eur. J. Phys. 28 (2007) 1–7 doi:10.1088/0143-0807/28/1/001

The magnetron method for the

determination of e/m for electrons:
revisited
A A Azooz
Department of Physics, College of Science, Mosul University, Mosul, Iraq

Received 16 July 2006, in final form 6 September 2006

Published 30 October 2006
Online at stacks.iop.org/EJP/28/1

Abstract
Additional information concerning the energy distribution function of electrons
in a magnetron diode valve can be extracted. This distribution function is a
manifestation of the effect of space charge at the anode. The electron energy
distribution function in the magnetron is obtained from studying the variation
of the anode current with the magnetic field in the region when the current
starts to fall with increasing magnetic field. This region has not received much
interest so far.

1. Introduction

The magnetron method for the determination of the charge-to-mass ratio e/m for electrons is
regarded as a simple and informative experiment employed in many undergraduate physics
laboratories [1–3]. The basic theoretical concept employed in this experiment is the charged
particle motion in crossed electric and magnetic fields. This concept is fully described in
most textbooks on electromagnetic theory [4, 5]. The result of the e/m value obtained from
this experiment can be regarded as of adequate accuracy if the simplicity of the equipment
and procedure is taken into consideration. Furthermore, discrepancies between students’
experimental results and the standard table value of e/m form by themselves a good field
of discussion and assessment of possible experimental and theoretical errors, biases and
approximations.
Many laboratory equipment suppliers have been offering complete experimental kits for
this experiment. The basic element of the kit is the thermionic diode vacuum tube. This
usually consists of a concentric cylindrical cathode–anode configuration surrounded by an
external solenoid. The basic physical concept used in this experiment is the motion of a
charged particle in crossed electric and magnetic fields. Electrons accelerated from the inner
cylindrical cathode towards the outer anode cylinder are forced to follow a curved path under
the action of the perpendicular magnetic field produced by the solenoid. If the magnetic field
is strong enough, electrons will follow a closed path within the cathode–anode spacing. This
0143-0807/07/010001+07$30.00

c 2007 IOP Publishing Ltd Printed in the UK 1
2 A A Azooz

closed path will prevent electrons from reaching the anode. In such a case, the anode current
is supposed to drop suddenly to zero if the electrons are monoenergetic. In practice, however,
a smoother drop in current is observed on most occasions and the student is asked to take the
magnetic field value needed to cause an anode current to drop to a certain fraction of its initial
value, as the cut-off field used in the calculation of e/m.
The electron trajectory in this experiment is usually considered to be circular for simplicity.
This assumption gives
8V
e/m = , (1)
B 2 (b − a)2
where V is the anode voltage, B is the magnetic field, b and a are the anode and cathode radii,
respectively, and e/m is the electronic charge-to-mass ratio.
Equation (1) is derived on the assumption that the electric field within the cathode–anode
spacing is uniform. This is not the case indeed. However, more elaborate evaluation, which
takes into account the radial variation of the electric field, gives a similar result for e/m even
when the path is not circular. This only differs from the result of equation (1) by a numerical
factor. In either case, estimation of the best value of the cut-off magnetic field is of crucial
importance. However, the problem is not trivial. One way to increase the steepness of the
fall curve relating the anode current Ia to the magnetic field B is by use of lower filament
current values in conjunction with elevated anode voltages in the anode current saturation
region. This will reduce the space charge formation within the tube active space. Even so, a
sharp vertical fall of Ia at cut-off is never achieved. The main reason for this is that thermionic
electrons emitted from the filament are never monoenergetic. This will result in that not all
electrons reach the anode with the same velocity (energy). The energy values will follow
some distribution function. The smooth fall of the anode current with increasing magnetic
field is nothing but a manifestation of this energy distribution function f (E). This function
is of physical importance. This importance is related to the fact that it is a representation
of the effects of both the space charge collisions effect and the nature of the thermionic
emission effect. Thus, one can argue that the study of this distribution function may compete
in importance with the legacy of evaluating e/m.
It has been almost a universal practice in physics lab experiments over the past century to
try to plot measured quantities in such a way as to give straight lines. This of course helps us
to extract other physical results from the values of the slope and/or the intercept of this line.
However, this philosophy is changing during this computer era. Comparisons of experimental
data with more involved mathematical equations are becoming easier. Fits of equations with
both linear and nonlinear parameters can now be carried out without much effort. Such fitting
techniques using (Matlab) programing are used here to demonstrate how one can obtain the
electron energy distribution function in the magnetron e/m experiment.

2. Mathematical modelling

A typical experimental relation between the magnetic field B and the anode current Ia in a
magnetron is shown in figure 1. Point x is where electrons start to miss reaching the anode.
Point y represents the magnetic field which prevents all electrons from reaching the anode.
Thus, the region xy is the main region of interest. The value of the magnetic field B in this
region is related to the electron kinetic energy E.
We have
mv 2
Bev = , (2)
r
The magnetron method for the determination of e/m for electrons: revisited 3

Figure 1. Typical anode current behaviour versus magnetic field.

where r is the radius of curvature. At cut-off, r = (b − a)/2, where a and b are the cathode
and anode radii, respectively.
This leads to
1 2 r 2 e2 B 2
E= mv = = CB 2 . (3)
2 2m
The reduction in the anode current −
Ia is caused by the reduction of the number of
electrons reaching the anode (−
n). This is caused by the increase of the magnetic field by

B.
n is the number of electrons per unit time which have values of kinetic energy between
E and E +
E. Thus,

n

n =
E = f (E)
E, (4)
C
B 2
where f (E) is the electron energy distribution function (EEDF).
In the limit when
B 2 → 0, this gives
1 dn −e dIa −2m dIa
f (E) = 2
= 2
= 2 (5)
C dB C dB r e dB 2
or
−4m 1 dIa
f (E) = . (6)
e(b − a)2 B dB
The energy distribution function can be calculated from the anode current fall region
using equation (6) provided that the quantity dIa /dB is properly evaluated from experimental
data. For such an evaluation, one can use numerical differentiation techniques [6] or choose
a suitable empirical representation of the experimental data of Ia versus B, that can be
analytically differentiated. One shape representative empirical relation may have the form
Ia = a1 tanh(a2 B + a3 ) + a4 , (7)
where a1 , a2 , a3 and a4 are free fitting parameters. a1 and a2 are the two main shape
determining factors. a3 and a4 are the coordinates of the half current cut-off point marked by
an arrow in figure 1. If a reasonable fit using this equation can be found, the fitted mathematical
equation can then be analytically differentiated to obtain f (E) through substitution in
equation (6).
4 A A Azooz

Figure 2. The electronic circuit used.

3. Experiment

The experimental set-up used is shown in figure 2. It represents a typical set-up used in the
standard vacuum tube diode experiment. A variable 300 V dc power supply is used for the
anode circuit while the filament current is drawn from a 6.3 V ac source. The only modification
here is the use of a Variac transformer in order to change the filament voltage and thus the
filament temperature. The diode is externally surrounded by a solenoid coil that produces
approximately 5.5 mT A−1 along the tube axis.

4. Results and discussion

Although this experiment is usually used in undergraduate physics laboratories to teach

students how the value of e/m for electrons can be measured through the interaction of
charged particles with electric and magnetic fields, our aim here is different. The purpose here
is to study the behaviour of the electron energy distribution function. The main factor that
affects this function is the space charge. This in turn is affected by two factors, which are the
rate of electron emission from the cathode and the electric field of the anode. At equilibrium,
space charge electrons will possess some form of an energy distribution. Equations (6) and (7)
are used to study this distribution function. The first set of measurements is concerned with
the study of the anode voltage effect on the EEDF. The filament voltage is set to 6.3 V, which
is the manufacturer’s setting. Four sets of measurements are carried out at anode voltages
of 100, 150, 200 and 270 V. The anode current is recorded against the current in the coil in
each case. A typical curve similar to that in figure 1 is always obtained. The results for the
200 V case are shown in figure 3. The dotted line represents the experimental data. The solid
line represents the fit to equation (7). Results at other voltage values are much similar. All
fits produced over 95% confidence level, which is the default convergence condition of the
MATLAB computer program used.
In order to investigate the above two effects, two sets of measurements are carried out.
The first set is concerned with the effect of the accelerating anode voltage on the EEDF.
Results of measurements at 100, 150, 200 and 270 V are shown in figure 4. For all these
measurements, the filament voltage is set to 6.3 V ac, which is the manufacture’s standard
rating. The EEDF for each case is calculated using equations (6) and (7). All distribution
functions are normalized such that they have the same area under the curve in order to make
easier comparisons. Two interesting conclusions can be drawn from this figure. The first is
that the most probable energies are not equal to the accelerating voltage but they are at about
half the value of the anode voltage. This can be explained by the fact that electron collisions
within the space charge will result in overall average energies well below the maximum anode
The magnetron method for the determination of e/m for electrons: revisited 5

Figure 3. Anode current versus coil current for an anode voltage of 200 V and a filament voltage
of 6.3 V. The dotted line represents experimental data. The solid line represents fit to equation (7).

Figure 4. The effect of the anode voltage on the EEDF. The distribution is more Maxwellian-like
at low voltages changing to Gaussian at higher anode voltages.

potential. More interesting is the shape of the EEDF. It is clear that the EEDF at the two lower
accelerating voltages is more or less of a Maxwellian shape. However, as the anode voltage
is increased, this shape starts to approach the Gaussian shape. This result is consistent with
the fact that the thermionic process is a Maxwellian one. At low anode voltages, the process
is dominated by thermionic effect at the cathode. As the anode voltage is increased, random
collision space charge effects will dominate and the distribution changes to the Gaussian
shape.
This argument is further supported by the fact that the full width at half maximum
increases with increasing filament voltages in figure 5. This figure shows larger dispersions
in energy values associated with higher filament voltages. These are caused by larger number
of collisions as a result of the increase in the space charge electron number density due to
6 A A Azooz

Figure 5. The EEDF of electrons for different filament voltages. The FWHM increases with
increasing filament temperature.

the elevated filament temperatures. However, in all cases, the end of the upper tail of the
distribution function always takes the value of the accelerating anode voltage. It is thus
advisable to consider the magnetic field which almost completely cuts off the anode current
in the calculation of e/m using equation (1).
Advanced students are encouraged to push this type of analysis further. One suggestion
here is to attempt finding empirical fits to the EEDF obtained under various conditions. One
such attempt is based on the assumption that the EEDF represents a mixture of two groups
of electrons. The first group of electrons is those which are not much affected by collisions
after being emitted from the cathode. The EEDF for these electrons is assumed to have a
Maxwellian distribution function of the form


E
f (E) = E 1/2 exp − . (8)
kT
The second group is those which are highly affected by collisions. These collisions tend
to produce a Gaussian-shaped EEDF. For those mathematically talented students, the exercise
can be trying to use numerical differentiation techniques instead of data fitting. Larger number
of experimental data points is a necessity in this case.

5. Conclusions

The magnetron method experiment for the determination of e/m for electrons in undergraduate
physics laboratories can be further modified to serve the purpose of studying the electron
energy distribution function within the space charge. This can deepen students’ understanding
of the statistical mechanics concept of energy distribution function, Maxwellian distribution
and Gaussian distribution, FWHM and collision effects. This experiment helps to improve
students’ computing abilities especially as far as data fitting procedures are concerned. No
extra equipment is needed to do this experiment. As a matter of fact, the e/m experiment can
be carried out by students during the first semester, while the EEDF experiment is done during
the second semester.
The magnetron method for the determination of e/m for electrons: revisited 7

References
[1] Shamos M H 1959 Great Experiments in Physics (New York: Holt, Rinehart and Winston)
[2] Edmonds D S 1983 Cioffari’s Experiments in College Physics (Lexington, MA: Heath and Co.) p 373
[3] American Institute of Physics 1964 Common Apparatus AAPT Novel Experiments in Physics pp 237–41
[4] Reitz J R and Milford F J 1969 Foundations of Electromagnetic Theory (Reading, MA: Addison-Wesley)
pp 156–7
[5] Tipler P A 1991 Physics for Scientists and Engineers 3rd edn (New York: Worth) chapter 24
[6] Andrei H, Covlea V, Covlea V V and Barna E 2003 The smoothing and the digital processing of Langmuir probe
Rom. Rep. Phys. 55 51–6