Effects of Magnetic Field on a Hollow Cathode Helium Discharge

Dennis Boyle and Jeff Parker

Abstract
Cylindrical and planar Langmuir probes are used to investigate the properties of a
hollow cathode helium discharge. Visual observations and properties of I-V
characteristics are described. Effects of neutral pressure, magnetic field, and
planar probe orientation are reported. Increased confinement due to the magnetic
field leads to more ionizing collisions and up to two orders of magnitude higher
plasma density. The orientation of the planar probe with respect to the magnetic
field changed the electron saturation current by a factor of three, also due to the
confining effect of the magnetic field.

Introduction
Langmuir probes are one of the simplest diagnostics for measuring low temperature
plasma properties. With a single I-V characteristic, three important plasma parameters can be
determined: electron temperature, density, and plasma potential. That Langmuir probes are still
used today, over eight decades after their introduction, demonstrates their importance and
usefulness.
We consider the locally planar probe theory, where the Debye sheath does not extend out
much from the probe, or λD << d, where λD is the Debye length and d is the probe size.
Additionally, we assume the ions are singly charged and that the sheath is collisionless. In what
follows, we use the convention that electron currents are defined to be positive.
There are three main features of the ideal Langmuir probe I-V characteristic: the flat
region at low potentials, a sharply rising region at intermediate potentials, and a flat region at
high potentials (Fig. 1). At low potentials far below the plasma potential, virtually all of the
electrons are repelled and the current collected is ion current. Results from sheath theory tell us
that the ions must be accelerated to the Bohm velocity as they enter the sheath. In the ion
saturation region, the current collected is independent of probe voltage and for a thermal
equilibrium electron distribution is given by [Hutchinson]
Iis = −neAuB exp( 12 ) (1)
Te
uB = (2)
mi
where A is the collecting area. The ion saturation current provides a means of determining the
plasma density if the electron temperature is known. As ions are typically much colder than
electrons, the ion current is essentially constant until it quickly decreases to zero as the probe
voltage reaches the plasma potential.
As the voltage is raised into the intermediate region, fewer electrons are repelled and the
collected current rises rapidly. The slope of the characteristic is related to the bulk plasma
electron distribution function by
dI e 2 A 2e(V p − V ) ⎛⎜ 2e(V p − V ) ⎞⎟
= fe (3)
dV me me ⎜ me ⎟
⎝ ⎠
where fe is the electron distribution function, Vp is the plasma potential and V is the probe
potential. For a thermal equilibrium electron distribution, this becomes
 dI e 2 A 2πTe ⎛ e(V − V p ) ⎞
= n exp⎜ ⎟ (4)
dV me me ⎝ Te ⎠
Te ⎛ e(V − V p ) ⎞
Ie = neA exp⎜ ⎟ (5)
2πme ⎝ Te ⎠
where n is the bulk plasma density. This region of the I-V characteristic provides a means of
determining the electron distribution function, or equivalently the temperature of a thermal
distribution.
For potentials above the plasma potential, the electron current saturates as all electrons
are collected. The electron saturation current is given by
Te
I es = neA
2πme . (6)

The “knee” in the characteristic between the regions above and below the plasma potentials
provides a means of determining the plasma potential. Additionally, in principle the electron
saturation current allows another way to calculate the density, though in practice it is not often
used, for reasons discussed below.
The potential which yields zero net current is called the floating potential. It is the
potential at which an unbiased probe would reach equilibrium and is easily determined from the
I-V characteristic. For thermal electron distributions, there is an alternate method of determining
the plasma potential when the electron temperature and floating potential are known. By setting
Iis + Ie = 0 at V = Vf, we obtain
T m T
V p = V f + e (ln( i ) + 1) ≈ V f + 4 e (7)
2e 2πme e
Real Langmuir probe data is much more difficult to interpret due to several possible
effects [Hershkowitz]. At low and high voltages, there is usually no perfect saturation. The rising
current region of the characteristic is often non-exponential, and it is incorrect to assume that the
shape directly corresponds to the distribution function. Sheath expansion and surface effects can
distort the shape of the characteristic away from the ideal curve. Collisional effects which have
been neglected here can also play a role in the interpretation of a characteristic. Because of such
effects, parameters as measured by Langmuir probes are generally considered to be accurate to
no better than 20 – 30% [Hutchinson].
Magnetic fields greatly complicate the interpretation of probe characteristics
[Hutchinson]. Certain limits provide simplification. In the limit of magnetized ions and electrons,
the particles are stuck to field lines, and the correct collecting area to use is the projected area of
the probe onto the magnetic field. There may also be a depletion of electrons in the flux tube
reaching the plasma, with electron current controlled by diffusion [Koo]. If a planar probe's
normal is perpendicular to the magnetic field, the particle collection is controlled by diffusion.

Apparatus
The experimental apparatus consists of a vacuum system and electrical systems to sustain
a plasma discharge, operate the Langmuir probes, and create a magnetic field. The vacuum
vessel is made of Pyrex glass and connected through a gate valve to a turbomolecular pump,
which is backed by a mechanical pump. Helium gas from a cylinder is introduced into the vessel
through a shutoff valve, flow meter, and precision needle valve. Pressure is controlled by the
needle valve and gate valve, and is measured in the main vessel by a Convectron thermocouple
gauge with a range of 10-4 – 103 T and uncertainty ~10%. The gauge is calibrated for N2 and
helium pressure is ~ 125% of the indicated value over the experimental range.
The discharge electrical system uses a 3 kV, 400 mA Hippotronics dc power supply. The
current and voltage indicators are of the rotating needle galvanometer type and are so sensitive to
static electricity that the needle can be moved to display any value by waving a finger nearby;
hence the measurements are highly unreliable. The high voltage terminal is connected through a
~2.5 kΩ ballast resistor to the cathode, which is a ~15 cm diameter hollow stainless steel
cylinder at the left of the vacuum vessel. The anode is grounded, and consists of a stainless steel
plate mounted at the right of the vessel.
The Langmuir probes consist of stainless steel probe tips mounted on ceramic rods which
pass through Wilson seals at the top and bottom of the vessel, allowing them to be rotated and
retracted. Four spherical probes are located at the bottom, while on top there is a planar probe, a
double cylindrical probe, and a cylindrical probe on a bent arm that has been melted to more
resemble a spherical probe. A Kepco amplifier boosts the signal from a Wavetek function
generator and biases the probes. The voltage across a 101Ω shunt resistor between the amplifier
and the probe is input to a Preston isolation amplifier with a bandwidth of 100 kHz. A LeCroy
digital oscilloscope measures the current to the probe from the output of the isolation amplifier
and the bias voltage across a 20:1 voltage divider.
The magnetic field is powered by a 500 A, 10 V power supply and consists of two water
cooled copper coils with a total resistance of 46.6 mΩ, inner diameter of 25 cm, outer diameter
of 50 cm, and a separation of 18 cm. On the magnetic axis, at the location of the cylindrical
probe, the field is ~1.2 G/A.

Procedure
Langmuir probe I-V characteristics are measured using the cylindrical and planar probes
for the hollow cathode discharge at discharge voltages Vd = 250 – 3200 V, pressures p = 50 – 250
mT, and magnetic fields B = 0 – 235 G. The planar probe is rotated to different orientations with
respect to the magnetic field. The perpendicular and parallel orientations are determined by the
angles where the electron saturation current reaches a maximum and minimum, respectively. The
cylindrical probe is located near the magnetic axis, whereas the planar probe is somewhat away
from the magnetic axis. The function generator is set so the entire characteristic is shown, with
care taken to avoid drawing electron saturation current >15 mA. Visual observations used a
frequency of 1 – 10 Hz while oscilloscope measurements typically used a frequency of ~100 Hz.

Data Analysis
Analysis of the I-V characteristics is performed in MATLAB. First, the floating potential
Vf is determined by finding the zero of a smoothing spline fit to the I-V data. Next, a linear fit to
the data in the ion current region is performed and subtracted from the measured current to yield
the electron current, Ie = I – Ii. The linear fit to the ion current is evaluated at Vf to give the ion
saturation current used in the density calculation. The linear region of a semilog plot of V-Ie sets
the boundaries for an exponential fit of the form Ie = a exp(bV), which yields Te/e = 1/b for a
Maxwellian distribution function. The plasma potential Vp is determined using the knee method
by finding the intersection of two linear fits to the I-V data in the regions just before and after
electron saturation; the electron saturation current Ies is found by evaluating the smoothing spline
at Vp. Finally, an alternate calculation of Vp is performed by finding the maximum of the
derivative of the smoothing spline.
Observations
Before conducting measurements of the discharges with the Langmuir probe, visual
observations are made. At low pressures p = 50 – 75 mT, a very high discharge voltage Vd ~3000
V is necessary to sustain the plasma, which appears as a uniform green glow that maintains the
cylindrical shape of the cathode section of the vessel even in the open probe section of the vessel.
With the magnetic field B = 0, the discharge current Id is ~2 mA and the glow is barely visible.
As B is increased to 235 G, Id rose to ~30 mA and the glow becomes much brighter. At higher p =
250 mT, a plasma can be sustained at Vd ~250 V, and now appears as a blue-purple glow with
dark striations. Also, the glow no longer maintains the cylindrical shape of the cathode and blue-
purple sheaths appear around the grounded metal probe bases. Now, addition of the magnetic
field causes the glow to dim, the sheaths on the probe bases to recede, and Id to decrease to the
point Vd must be increased to sustain the plasma. As B is increased further, the discharge reverts
to the uniform green column. With Vd = 400 V, the discharge changes from the uniform green
column to the striated blue-purple glow at B = 170 G.
When biased negatively, the Langmuir probes have little to no visible effect on the
plasmas except a small sheath around the probe. When biased strongly positive to collect
electron saturation current, the sheath emits a bright white light. In the uniform green column
regime, the flux tube passing through the probe gets noticeably dimmer, with little to no decrease
in brightness in the rest of the glow. In the striated blue-purple glow regime, the sheaths around
the probe bases often disappear, and with high enough bias the entire glow disappears.

Results
The I-V characteristics measured by the Langmuir probes deviate significantly from the
simple theory, though the main attributes do appear in modified form (Fig. 2, 3). At no point is
the electron saturation region flat, but is instead increasing approximately linearly. The same is
true for ion saturation using the cylindrical probe, though the planar probe does yield a flat ion
saturation current. Over the range of experimental parameters, Vf = –37 – +0.7 V and Iis = 0.01 –
0.5 mA, with uncertainties in Iis much larger on the cylindrical probe due to the increasing ion
current. In general, the oscilloscope measurements from the cylindrical probe were made with a
voltage resolution too poor to determine a meaningful Te. On the planar probe, exponential
regions appear in all of the I-V characteristics, though it is uncertain if the plasma actually has a
Maxwellian distribution function. With a magnetic field, the knee does not occur immediately
after the exponential region, but rather the exponential region leads smoothly into a linear region
before the knee. Without a magnetic field, it appears that there are two exponential regions. With
or without a magnetic field, it is unclear where, if at all, the temperature should be calculated.
With a magnetic field, using only the exponential region for the fits yields Te = 0.15 – 0.5 eV for
the different values of B and orientations of the probe. Without a magnetic field, fitting the first
exponential region yields Te = 1 – 8 eV.
Due to the large amounts of uncertainty in most of the calculated values, direct
comparisons are only useful where the I-V characteristics are qualitatively similar and the values
can be calculated consistently as the parameters are changed. Merely rotating the planar probe
should not have any effect on the plasma itself; indeed the shape of the characteristic and Vp vary
little. The effect of rotating the probe on Ies is shown in Fig. 4. In determining the effect of B on
ne, I-V curves within the same dataset can be analyzed consistently to show its relative effect,
though the absolute values of ne and comparisons between different data sets may not be
accurate. Figure 5 shows that increasing the magnetic field increases ne, except at high p.
Discussion
The transition from the uniform green column to the striated purple discharge probably
occurs when collisions with neutrals dominate and cause greater radial diffusion. At low neutral
pressure, there are not enough neutral collisions to cause the plasma to diffuse radially and it
maintains its cylindrical shape. At higher pressure diffusion does occur, but a magnetic field
provides radial confinement and restores the green cylindrical column. Due to increased
confinement, electrons undergo more ionizing collisions with neutrals before escaping. This
leads to plasma density increasing with magnetic field strength, as seen in Fig. 5.
Without a magnetic field, the I-V curves appeared to contain two exponential regions. It is
possible that the first knee corresponds to Vp and the probe strongly perturbs the plasma at higher
potential, causing strange results. However, the same behavior is seen in all the zero field
measurements, and could also correspond to the presence of high-energy electrons in addition to
the bulk plasma. In this interpretation, the high energy electrons leave the plasma, charging it
positively until a large enough potential difference is attained. Here, the fit to the first
exponential region corresponds to the spread in energies of the high-energy electrons rather than
Te. The observation that this behavior did not occur in the presence of a magnetic field could be a
result of greater collisional energy losses due to increased confinement. It is possible that more
neutral collisions also leads to reduced Te, but temperature measurements are not considered
reliable enough to say conclusively.
One would expect that for magnetized particles, saturation currents would be higher when
the planar probe is perpendicular to the magnetic field than when parallel to it. When the probe
is parallel to the field, magnetized particles have to reach the probe through diffusion, rather than
free streaming along the field lines. This was seen experimentally for both electrons and ions,
which were expected to be magnetized, though for ions the effect was somewhat smaller. As the
probe orientation varied from perpendicular to parallel to the magnetic field, the general trend
was a decrease in electron saturation current, as seen in Fig. 4.

Conclusions
While the I-V traces from the Langmuir probes were difficult to interpret, we were able to
conclude that increased confinement due to the magnetic field led to more ionizing collisions and
higher plasma density. Additionally, it was observed that the orientation of the planar probe with
respect to the magnetic field had a significant effect on electron saturation current, also due to the
confining effect of the magnetic field.

References
Hershkowitz, N. How Langmuir Probes Work.
Hutchinson, I. Plasma Diagnostics
Koo, B. et. al. Langmuir probe in low temperature, magnetized plasmas: Theory and
experimental verification. Journal of Applied Physics, Vol. 86, No. 3, 1999.
Owens, D.K. Notes on Langmuir Probes
Current

Probe Bias
Figure 1:x Ideal
10 Langmuir probe I-V characteristic
Raw Data
Smoothing spline
6 V
f
Ii

4 I
if
I + C exp(eV/T )
i e
Vf + 4Te
I (mA)

Ies
0

−2

−4
−120 −100 −80 −60 −40 −20 0 20
V (V)
Figure 2: I-V characteristic on cylindrical probe, B = 220 G, p = 63 mT
 Raw Data
12 Smoothing Spline
V
f
10 Ii
Iif
8 I + C exp(eV/T )
i e
I (mA)

Vf + 4Te
6
Ies
4

0
−100 −80 −60 −40 −20 0 20 40 60 80 100
V (V)

Figure 3: I-V characteristic on planar probe, B = 0, p = 125 mT,

−3
x 10
3.5

2.5
(A)

2
e,sat
I

1.5

0.5
0 50 100
Angle with B (degrees)
Figure 4: Electron saturation current as planar probe is rotated
 16
10

15
10
Density (m−3)

P=63mT Vd=3000V Cylindrical

P=125mT Vd=1400V Cylindrical
P=125mT Vd=2900V Cylindrical
14
10 P=250mT Vd=480V Cylindrical
P=125mT Vd=1450V Planar 90
P=125mT Vd=1450V Planar 0
P=75mT Vd=3000V Planar 90
P=75mT Vd=3000V Planar 0
13
10
0 0.005 0.01 0.015 0.02 0.025
B (Tesla)
Figure 5: Plasma density versus magnetic field strength over wide range of parameters
1. The two most common diagnostics for a laboratory plasma are a voltage meter and a current
meter. These two instruments are commonly on the power supply which creates the discharge.

2. Neglecting convection due to low neutral pressure and flow velocity, we can find the
minimum power required to sustain a temperature of 500 °C with heat loss only due to radiation.
The radiated power for an ideal blackbody is
P = A σT 4 ,
where A is the area, σ = 5.67 * 10-8 W / m2 K4 is the Stefan-Boltzmann constant, and T is the
temperature. A probe with a total surface area of 2 cm2 will radiate a power of 4 W. For a 100V
potential drop, a 40 mA current to the probe would be required.

3. We did not rotate the cylindrical probe.

4. Ion saturation current is fairly flat for the planar probe but not flat for the cylindrical probe,
likely due to sheath expansion effects. The electron saturation current is usually not flat but
steadily rising with higher voltage. According to the theory, the ratio of saturation currents is
given by
I es A mi A
= 0.66 e ~ 56 e ,
I is Ai me Ai
where Ae and Ai are the effective collecting areas when collecting electrons and ions,
respectively. We measured a wide distribution of ratios Ies / Iis, ranging from 3-300, with an
average ~20. There were glows around the probe when it was in the electron saturation regime,
and could become very bright when large amounts of current were drawn.

5. The plasma potential was estimated in three ways: from the “knee”, from Vf + 4Te, and from
the maximum of the first derivative of the I-V characteristic. The plasma potential determined
from the knee and from Vf + 4Te were often similar for plasmas with Maxwellian temperature
profiles generally within a quarter of a volt. As the derivative was performed on a smoothed
curve, we expect this method to round out the bending of the characteristic and somewhat
underestimate the plasma potential. This was seen in practice as it usually gave estimates
somewhat less than the other two methods.

6. See discussion

7. We did not use the spherical probes.