Dimensional Model

Two-dimensional modelling of a low current electron gun
dry neutralizer for ion engines
IEPC-2024-404
Presented at the 38th International Electric Propulsion Conference
Pierre Baudis Convention Center • Toulouse, France
June 23-28, 2024
Klevis Gasa 1 and Stephen B. Gabriel 2

University of Southampton, Southampton, United Kingdom
Alexander Daykin-Iliopoulos 3 and Francesco Guarducci 4

Mars Space Ltd, Southampton, United Kingdom
Abstract: Dry, or propellantless, dispenser cathodes represent an attractive alternative

to the conventional neutralization system for miniaturized ion thrusters due to the specific
impulse savings and propellant system simplification. An electron gun configuration,
consisting of an extraction electrode and a focus electrode, is able of achieving an electron
cathode current in the 0-100 mA range. The emitted cathode current depends on the optics
geometry and the operational conditions. Thus, a 2D-axisymmetric electron optics model was
created by using COMSOL Multiphysics to simulate and support the development of this kind
of neutralizer. By combining the electrostatic module and the charged particle tracing, the
model simulates the electron current emission in space-charge limitation and how the external
fields affect the particles. Simulations at different operational modes were performed on an
optics configuration and the electron beam's behavior was obtained and discussed in detail. A
comparison was carried out between the simulation results and experimental findings,
showing a good agreement between them. Although the model tends to overestimate the
emitted cathode current up to a maximum of 40%, it provides a good estimation of the cathode
emission current trends and partition on the electrodes and the anode, suggesting that the
model could be used as a support in the design of an electron gun neutralizer.
I. Introduction
Miniaturised electric propulsion systems have been recognised as candidates for missions where low thrust levels
and high specific impulse are required. Radio-Frequency ion thrusters, such as the RIT3.5, has been shown to be an
ideal candidate for stand-alone deep space CubeSat system for low-cost science and exploration missions. However,
due to a historical lack of interest in very low power thrusters, challenges remain open for what concerns the propulsion
systems needed for the new generation of satellites. Critical aspects are represented by the limited power budget and
space availability. Radiofrequency ion thrusters do not require a discharge cathode; however, they still require a
neutralization system. For small low power thrusters as the RIT3.5, the currents needed for the neutralization are
typically in the 0-100 mA [1], depending on the operating conditions. Hollow cathodes and field emission cathodes
are inefficient for this range as they are not optimized for it. The former would require propellant flow to keep the
self-heating process, and consequently an associated fluidic system, that would result in a loss of specific impulse
[2,3]. The latter would require a large surface area, area, around 250 cm2 for 100 mA of emission [4]. For further
1
PhD Candidate, Faculty of Engineering and Physical Sciences, [email protected].
2
Professor, Faculty of Engineering and Physical Sciences, [email protected].
3
Senior Research Engineer, Mars Space Ltd, [email protected]
4
Director, Mars Space Ltd, [email protected].
1
The 38th International Electric Propulsion Conference, P. Baudis Convention Center, Toulouse, France June 23-28, 2024
Copyright 2024 by the Electric Rocket Propulsion Society. All rights reserved.
details a review of these neutralization techniques in this emission range is presented in Ref [2]. Dry dispenser
cathodes, i.e., propellantless cathodes operating without the need of a gas flow, have been adopted for space
applications in configurations able to produce 1-10 mA [5,6]. An extraction electrode, or gate electrode, is biased
respect to the cathode to extract the electrons in a space charged limited emission. To reduce the current collected by
the gate electrode, in magnetrons, travelling wave tubes and cathode-ray tubes, an electron gun configuration typically
adopts a focus electrode [7]. This electrode is used to focus the beam by reducing radial electric field components and
by counteracting the coulomb repulsion between electrons in the beam [8]. This results in an increase in magnitude of
the extracted electron beam, as less current is collected by the gate electrode. Careful considerations are required in
designing these two electrodes to avoid adverse effects caused by the space-charged effects. The magnitude of the
space-charge effects on the electron beam is expressed by the perveance. A high perveance results in a strong space-
charge that reduces the focus effect and thus lowers efficiency. However, a high perveance results in a higher emission.
[9,10]. The perveance is expressed as:
𝐼𝐼𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏
𝑃𝑃 = 3/2
𝑉𝑉𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 (1)
where 𝐼𝐼𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 is the electron beam emitted by the cathode and 𝑉𝑉𝑔𝑔𝑔𝑔𝑔𝑔𝑔𝑔 is the extraction gate electrode voltage.
Because of on the requirements to operate with a low volage on the extraction electrode [2], a high perveance
electron gun configuration is required, i.e., P > 1 µPerv [10]. This type of configuration has a large extraction aperture
that strongly perturb the electric field. Validated analytical models can be found in literature for the design of electron
guns [11]. However, for high perveance devices, numerical methods are essential due to the complexity of the fields
and the particle behavior [10]. Thus, a simulation model was still required in the first phase of the development.
Among several commercially available software [12–14], COMSOL Multiphysics has been selected since it excels in
the integration of multiple physical phenomena, and it has been already used for simulating hollow cathodes for space
application [15–17].
To design a propellantless neutralizer configuration that would allow to achieve the discharge current required by
ion thrusters as the RIT3.5, a 2D-axisymmetric electron optics model based on an electron gun configuration has been
developed. This work represents a second iteration of the model presented in Ref. [2]. Section II aims to show the
COMSOL model by presenting the main assumptions and equations, domains and boundaries employed in its
development. Section III presents some of the results of the electron optics model with emphasis on a comparison
with the experimental results done with a breadboard neutralizer. In the final section, conclusions are drawn on the
suitability of such a model to predict the electron beam behavior.
II. Modeling Procedure
A. Assumptions and Main Equations
The 2D-axisymmetric electron optics model consists of the combination of the Electrostatics interface with
Charged Particle Tracing interface. The first defines the electric field and potential distributions created by the
electrode's interaction, while the second is used to evaluate the particles' trajectories that can be affected by particle-
boundary interactions and particle-domain interactions. The charged particles streams are influenced by the
electrostatic of the domain and contemporarily influence it because of their space charge density. An iterative
procedure that alternates between them until a convergence is adopted to fully account for both influences. Space
charge is evaluated and assigned to each mesh element that contains the particles and this value is then used to re-
evaluate the electric potential of the domain. It is defined as follow:
𝑁𝑁𝑠𝑠
𝑑𝑑𝜌𝜌𝑠𝑠 1 (2)
= � 𝑓𝑓𝑟𝑟𝑟𝑟𝑟𝑟 𝑞𝑞𝑖𝑖
𝑑𝑑𝑡𝑡 𝑉𝑉𝑠𝑠
𝑖𝑖=1
Where 𝜌𝜌𝑠𝑠 is the average space charge density of a mesh element s with volume 𝑉𝑉𝑠𝑠 , while 𝑓𝑓𝑟𝑟𝑟𝑟𝑟𝑟 is the effective
frequency of release and represents the number of real particles per unit time represented by each particle in the model.
For a more robust and consistent statistical convergence, the space charge density contribution was computed by
ramping-up the cumulative average over a number 𝛽𝛽 of iterations of the solver sequence. The value is updated with
each iteration by scaling the new computed charge density contribution by a factor, according to the following:
𝑖𝑖𝑡𝑡𝑡𝑡𝑡𝑡
𝜌𝜌̅𝑠𝑠,𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 + 𝜌𝜌
𝛽𝛽 𝑠𝑠 (3)
𝜌𝜌̅𝑠𝑠 =
2
Where 𝜌𝜌̅𝑠𝑠,𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 is the stored value of the cumulative space charge density from the previous iteration, 𝜌𝜌𝑠𝑠 is the
contribution of the particles to the space charge density in the current iteration, 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 is the iteration number and 𝛽𝛽
maximum number of ramping-up iterations. Once the ramping-up iterations are completed, the cumulative space
charge density at each successive iteration, 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 > 𝛽𝛽, is derived as cumulative averaging from:
−1
𝑖𝑖 𝑖𝑖−1
𝜌𝜌̅𝑠𝑠 = �� 𝑤𝑤𝑗𝑗 � �� 𝑤𝑤𝑗𝑗 � 𝜌𝜌̅𝑠𝑠,𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 + 𝜔𝜔𝑖𝑖 𝜌𝜌𝑠𝑠 � (4)

𝑗𝑗=1 𝑗𝑗=1
where 𝑤𝑤𝑗𝑗 is the weight of each iteration, equal to 1 in case of uniform condition, with 𝑖𝑖 = 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 − 𝛽𝛽 + 1.
The electrostatic field is defined by the Poisson’s equation:
𝛻𝛻 ⋅ (𝜀𝜀0 𝜀𝜀𝑟𝑟 𝑬𝑬) = 𝜌𝜌𝑣𝑣 (5)
where 𝜀𝜀0 is the vacuum permittivity, 𝜀𝜀𝑟𝑟 is the relative permittivity, 𝑬𝑬 is the electric field and 𝜌𝜌𝑣𝑣 is the charge density.
Under static conditions, the electric potential V is defined as:
𝑬𝑬 = −𝛻𝛻 𝑉𝑉 (6)
The equation of motion for the electron is the Newtonian formulation. It defines a set of ordinary differential
equations based on Newton’s law of motion:
𝑑𝑑 𝑑𝑑𝒒𝒒
�𝑚𝑚𝑝𝑝 � = 𝑭𝑭 (7)
𝑑𝑑𝑡𝑡 𝑑𝑑𝑡𝑡
where 𝒒𝒒 is the particle position, 𝑚𝑚𝑝𝑝 is the particle mass, and 𝑭𝑭 is the force on the particles.
Corrections on the relativistic effects of the particle mass due to very high speed are neglected to minimize the
computational cost, since the maximum particles speed is of one order of magnitude lower than the speed of light.
Moreover, for non-relativistic particles, the contribution of the particles to the magnetic force is assumed to be
negligibly small when compared to the electric force. Thus, the self-magnetic field is also neglected.
The electron current emitted by the cathode is defined by the space-charge limited emission from a surface. The
released electrons create a cloud on the surface that limits further emission from the surface. This is estimated with
the Child’s Law for parallel planes in diode mode in vacuum:
4𝜀𝜀0 2𝑒𝑒 𝑉𝑉 3/2

𝐽𝐽 = � (8)
9 𝑚𝑚𝑒𝑒 𝑑𝑑 2
Where 𝐽𝐽 is the current density, 𝑉𝑉 is the potential difference across the diode, 𝑚𝑚𝑒𝑒 is the electron mass and 𝑑𝑑 is the gap
thickness. On the emitter surface, the space charge limited current can be computed by creating a virtual cathode
surface from where the particles are released, located a short distance away from the real cathode. The particle velocity
at the cathode emission surface is nonzero and can be computed by solving Poisson’s Equation in the thin gap between
the two surfaces. Further details on the modules’ capabilities are available in the relative documentation [18,19].
B. Geometry and Mesh
Figure 1 shows the components of the neutralizer. The cathode and the electrodes represent boundaries of the
model. The cathode emitter surface is the region where the electrons are emitted, the focusing electrode and the gate
extraction electrode are used to create the electric field to extract the electrons. A casing element was added to enclose
the extraction gate electrode. An anode, not shown in the figure, was placed downstream to simulate the presence of
an ion plume and to collect the electron beam.
Casing
Gate Electrode
Focus Electrode
Emitter
Figure 1. Neutralizer components and geometry
A sensitivity analysis on a selected configuration mesh size was performed by varying it from coarse to extremely
fine. The resulting emitted cathode current, along with the required time for the solution convergence, are exposed in
Table 1. The cathode current converges to a stable solution starting from the normal mesh, however the solution time
for all the mesh sizes is very similar, ~3 min. It was decided to choose the extra fine mesh since the difference in time
is low.
Table 1. Sensitivity analysis on the mesh size
Mesh Size
Extremely
Parameter Coarse Normal Fine Extra Fine
Fine
Cathode Current [mA] 20.65 21.13 21.13 21.09 21.06
Time [min: s] 3:11 3:18 3:19 3:23 3:52
The main factor that influences the solution time is represented by the number of iterations for the solution
presented in Eq. (4). A sensitivity analysis with an extra fine mesh size is shown in Table 2. The cathode current
converges to a stable solution after a minimum of 35 iterations.
Table 2. Sensitivity analysis on the iteration number
Iterations
Parameter 20 25 30 35 40 45 50 70
Cathode
17.75 15.88 21.99 21.1 21.09 21.1 21.11 21.11
Current [mA]
Time [min: s] 1:34 1:56 2:28 2:47 3:16 3:42 4:11 6:01
C. Boundary Conditions
Boundary conditions are required for both the electrostatic and the particle tracing modules. Each solid wall
boundary has an electric potential, V, defined as a parametric input. The cathode and the focusing electrode were set
at cathode potential 𝑉𝑉𝐶𝐶 , the gate extraction electrode potential is defined by 𝑉𝑉𝐺𝐺 , the anode potential is defined by 𝑉𝑉𝐴𝐴 .
A zero-charge condition is used for the free space boundaries, so that:
𝒏𝒏 ⋅ (𝜀𝜀0 𝜀𝜀𝑟𝑟 𝑬𝑬 ) = 0 (9)
Where 𝒏𝒏 is the surface normal and 𝑬𝑬 is the electric field. A dielectric material was used as a boundary condition
between the electrodes to simulate a ceramic element used in the breadboard testing. This condition is defined as a
thin layer with thickness 𝑑𝑑𝑠𝑠 and a bulk relative permittivity on the surface 𝜀𝜀𝑟𝑟𝑟𝑟 that shields the electric field:
𝒏𝒏 ⋅ 𝑫𝑫 = – 𝛻𝛻𝑡𝑡 ⋅ 𝜀𝜀0 𝜀𝜀𝑟𝑟𝑟𝑟 𝑑𝑑𝑠𝑠 𝛻𝛻𝑡𝑡 𝑉𝑉 (10)
Where 𝛻𝛻𝑡𝑡 represents a tangential derivative.

For all the wall boundaries, a freeze condition is applied. The particle sticks on the boundary keeping the position
and velocity at the instant the wall was hit so that velocity or energy of the particle can be recovered, 𝒒𝒒′ = 𝒒𝒒𝒄𝒄 and
𝒗𝒗′ = 𝒗𝒗𝑐𝑐 at time steps t > 𝒕𝒕𝑐𝑐 . The symmetry axis boundary condition has a specular reflection condition. The particle
hitting the boundary conserves the kinetic energy. The particle position after the bounce 𝒒𝒒′ is the same as before, 𝒒𝒒′ =
𝒒𝒒, while the velocity after the bounce is expressed as 𝒗𝒗′ = 𝒗𝒗 – 2(𝒏𝒏 ⋅ 𝒗𝒗)𝒏𝒏. These boundary conditions are
summarized in the following Table 3.
Table 3. Boundary conditions summary for the COMSOL model

Boundary Condition
Element Electrostatic Particle Tracking
Symmetry wall − Bounce
Cathode 𝑉𝑉𝐶𝐶 Freeze
Gate electrode 𝑉𝑉𝐺𝐺 Freeze
Focusing Electrode 𝑉𝑉𝐶𝐶 Freeze
Anode 𝑉𝑉𝐴𝐴 Freeze
Dielectric 𝒏𝒏 ⋅ 𝑫𝑫 = – 𝛻𝛻𝑡𝑡 ⋅ 𝜀𝜀0 𝜀𝜀𝑟𝑟𝑟𝑟 𝑑𝑑𝑠𝑠 𝛻𝛻𝑡𝑡 𝑉𝑉 Freeze
Free Space 𝒏𝒏 ⋅ (𝜀𝜀0 𝜀𝜀𝑟𝑟 𝑬𝑬 ) = 0 Freeze
III. Numerical Results and Discussion
Simulations were carried out to assess the total electron current emitted by the cathode, 𝐼𝐼c, and the currents
collected on the electrodes. The cathode electron current could end up on the gate extraction electrode, expressed by
𝐼𝐼G, or leave the neutralizer orifice, expressed by the bias current 𝐼𝐼B. This current could be collected by the anode
boundary or by the casing. These relations are summarized in Eq. (11) and (12), and illustrated in Figure 2.
𝐼𝐼𝑐𝑐 = 𝐼𝐼𝐺𝐺 + 𝐼𝐼𝐵𝐵 (11)
𝐼𝐼𝐵𝐵 = 𝐼𝐼𝐴𝐴 + 𝐼𝐼𝑐𝑐𝑐𝑐𝑐𝑐 (12)
Figure 2. Emitted cathode current distribution
A. Simulation Results
The simulations were carried out for different electric optics configurations at different operating conditions. Here,
the results of one of these configurations are explored to analyze the electron beam current behavior. Figure 3 shows
the electric potential distribution and the electric field distribution in the electron gun neutralizer region. The extraction
gate electrode is biased with respect to the cathode with 600 V. The anode, positioned downstream and not visible in
the figure, is set at 30 V. The equipotential lines reveal the effect of the two electrodes. The gate electrode creates the
potential difference required to extract the electrons, while the focusing electrode reduces the radial component of the
electric field. This creates equipotential lines parallel to the emitter surface, and consequently axial electric field lines.
The radial component would lead to a defocus of the beam that would result in a higher gate current, and consequently
an efficiency loss. The electric potential on the axis peaks in correspondence of the gate electrode to then slowly
decrease until the anode boundary is reached.
Figure 3. Distribution of electric potential in the model with equipotential lines (left); electric potential
distribution with constant electric field lines (right)
Figure 4 shows the electric potential along the axis at different gate to cathode voltage bias. As expected, at
increased gate bias voltage corresponds an increase in the maximum potential reached. However, downstream the
neutralizer, this effect is reversed. The increase in total current emission due to the increasing bias results in an increase
of space-charge density, and thus, a decrease of the potential.
Cathode Anode
Gate
Focusing
electrode
Casing
Figure 4. Electric potential distribution along the axis at different gate voltage
A comparison of the electron density distribution for two operating conditions is shown in Figure 5. The gate bias is
set to 300 V on the left and 600 V on the right side of the figure. In both cases the maximum electron density is close
to the emitter surface. The density at higher voltage bias is overall higher and the density appears to be higher in the
region close to the gate electrode instead of the axis. This characteristic is typical of an electron gun in a high-
perveance configuration. Table 4 summarizes the perveance number of this configuration at different operating
conditions. For all cases, the electron gun has a high perveance, thus the gate aperture affects the electric potential
distribution, and the beam is not focused on the center.
Table 4. Perveance estimation at different operating conditions

Gate to Cathode Voltage [V]
Parameter 100 200 300 400 500 600
Current [mA] 4.1 11.5 21.1 32.3 45.2 59.4
Perveance [ µPerv] 4.1 4.1 4.0 4.0 4.0 4.0
Figure 5. Electron density distribution at different gate voltage, 300V (left); at 600 V (right)
Figure 6 shows the normalized electric potential variation along the normalized gate electrode aperture radius. The
data refers to the cross-section in correspondence of the end of the gate electrode channel. Because of the size of the
gate aperture, the electric potential on the axis is around 60% for most of the operating conditions. This suggests that
the effective extraction bias voltage is lower than the set bias, resulting in an efficiency loss. Reducing the gate aperture
should increase the effective voltage and consequently the emission, but more current would be collected by the gate
electrode. The potential increases closer to the gate electrode, and thus also the electron density, as seen in the previous
figure.
Figure 6. Electric potential distribution ratio along the gate electrode channel radius
Particle tracing allows to examine the electrons trajectories and their properties. Figure 8 shows the comparison of
the particle trajectories for two gate voltages as a function of their energy. The electrons are accelerated until the end
of the extraction electrode orifice channel, and they are then decelerated by the adverse electric potential distribution
until the anode is reached. Because of the deceleration, an accumulation of space-charge is created downstream the
neutralizer that forces the beam to diverge. When the gate voltage is set to 300 V, figure on the left, the electrons are
accelerated to ~250 eV by the gate. Most of the electron beam reaches the anode while a small percentage of particles
8
is redirected to the gate, reaching an energy of 300 eV when hitting the electrode. In the 600 V case, the electrons are
accelerated to ~450 eV. In this second case the space-charge density downstream of the neutralizer is higher, as shown
previously in Figure 5, and thus the electrons divergence is increased. Part of the particles lines on the side boundary
would reach the anode if the region prolongated. Figure 9 presents the distribution of the velocities for the same cases.
At 300 V of bias, the electrons are accelerated up to a velocity of 1x107 m/s, while at 600 V the velocity exceeds
1.4x107 m/s. As shown previously in Figure 4, there is a local minimum on the potential between the casing and the
anode. Thus, for both cases in Figure 9, the velocities present a local minimum in that area. After the deceleration
phase the electrons are re-accelerated by the anode. This suggests that a lower voltage of the anode, or a higher distance
would increase the space-charge accumulation downstream of the neutralizer and consequently increase the electron
current being deflected back on the casing.
Figure 7. Electron particle tracing as a function of the electron energy at 300VG (left); 600VG (right)
Figure 8. Electron particle tracing as a function of the electron velocity at 300VG (left); 600VG (right)
B. Validation Against Experimental Data
An experimental test campaign has been performed with a breadboard with the same optics presented here. Details
on the setup and methodology adopted for the test are given in Ref. [20]. The gate voltage was limited to 580 V for
hardware limitations.
Figure 10 shows a comparison between the simulation currents and the experimental results at different operating
gate bias conditions. There is a good qualitative agreement between the model and the experimental results in terms
of currents trends. The model overestimates the current emitted by the cathode, especially at high gate voltages. The
maximum cathode current emission was 40 mA, while the model estimates 56 mA. This overestimation is then
reflected on the electrodes’ currents. The anode current measured experimentally reaches a maximum of 38 mA, while
the simulation predicts 52 mA. The casing current and the gate current are low in both cases, respectively 1.3 mA and
4.5 mA. This suggests that the optics are good enough to focus the beam into the gate orifice. Downstream the gate
orifice, electron beam diverges, but most of the current is collected by the anode ultimately. The discrepancy between
the model predictions and the measurements could be explained by the inherent difficulty in ensuring the setup
configuration being identical to the model. A small variation in the distances between the electrodes, the thermal
expansions of the different components, or the misalignments between the electrodes are sources of deviation.
60
50
40
Current (mA)
30
20
10
0
100 200 300 400 500 600
Gate Voltage (V)
I Cathode - M I Anode - M I Gate - M I Casing - M
I Cathode - T I Anode - T I Gate - T I Casing - T
Figure 9. Comparison between the currents predicted by the model vs current measured
experimentally
Figure 11 presents the relative error, 𝛿𝛿𝑟𝑟𝑟𝑟 , between the model and the experimental results. It is expressed as the
absolute difference divided by the measured data:
�𝐼𝐼𝑀𝑀𝑀𝑀𝑀𝑀 − 𝐼𝐼𝐸𝐸𝐸𝐸𝐸𝐸 �
𝛿𝛿𝑟𝑟𝑟𝑟 = (13)
𝐼𝐼𝐸𝐸𝐸𝐸𝐸𝐸
Where 𝐼𝐼𝑀𝑀𝑀𝑀𝑀𝑀 and 𝐼𝐼𝐸𝐸𝐸𝐸𝐸𝐸 are respectively the model current and the experimentally measured current. The simulation’s
cathode and anode currents differ from the experimental result by 25% up to 400 V on the gate, while the gate and
casing currents present a broader variation. The gate current error is 0.5 times the measured value up to 400 V but
increases at higher voltages up to 2.5. The overestimation of the emitted cathode current results in a higher defocus
effect and thus in a higher gate current. On the other hand, the model predicts no current collected by the casing up to
500 V, while experimentally, this current varies from 0.2 mA to 1.3 mA.
10
0.5 3
2.5
0.4
Relatvive Error
Relatvive Error
2
0.3
1.5
0.2
1
0.1 0.5
0 0
100 200 300 400 500 600 100 200 300 400 500 600
Gate Voltage (V) Gate Voltage (V)
I Cathode I Anode I Gate I Casing
Figure 10. Relative error of the predicted currents with respect to the experimental measurements;
cathode and anode current error (left); gate and casing current error (right)
In addition to the inherent geometric differences, another explanation for the casing current difference relies on
the neutrals in the vacuum chamber that interacts with the electron beam. Their interaction results in the ionisation of
neutrals and the creation of low energy electrons that effect the space-charge. The ions would reduce the space-charge
and consequently help with the beam focus, while the new electrons would be attracted by the closest electrode. The
model assumes only electrons as species, thus no interaction with the neutrals is simulated.
Figure 12 shows the ratio between each current and the relative emitted cathode current as a function of the gate
bias for both the model and the experimental test. There is a good agreement between the simulation and the
experimental results on how the emitted cathode current is distributed among the electrodes. In both cases, the anode
current represents more than 90% of the total current, while the remaining 10% of the cathode emission is lost on the
electrodes. The casing current is at most 3% of the cathode current, while the gate electrode collects between 5% and
8% of the cathode current.
1 0.1
0.95
Current Ratio
Current Ratio
0.9 0.05
0.85
0.8 0
100 200 300 400 500 600 100 200 300 400 500 600
Gate Voltage (V) Gate Voltage (V)
I Anode - M I Anode - T I Gate - M I Casing - M
I Gate - T I Casing - T
Figure 11. Ratio between electrodes’ currents and emitted cathode currents in the model and test
IV.Conclusion
A 2D-axisymmetric electron optics model was developed in COMSOL Multiphysics to support the development
of a dry neutralizer in an electron gun configuration. The electron optics model was required due to the complexity of
11
the fields created by the electrodes and to investigate the particle tracing behavior. An anode was located downstream
the neutralizer to simulate a diode mode emission in space-charge emission current. Simulations at different
operational modes were performed on an optics configuration to obtain electron beam's behavior. Results show the
creation of an electron beam focused through the gate electrode by the focus electrode. Downstream the gate electrode,
the electron beam is decelerated by the adverse electric field leading to an accumulation of space-charge that diverges
the beam. The beam divergence increases with the applied gate voltage as the space-charge increases.
Comparison with the experimental results has shown good agreement with the predicted currents. Although the
model tends to overestimate the emitted cathode current up to a maximum of 40%, it still provides a good estimation
of the cathode current trends and the current partition on the neutralizer electrodes and the anode. Both in the model
and the tests the anode current represents more than 90% of the total current, while the remaining 10% of the cathode
emission is lost on the electrodes. These results have allowed the validation of the COMSOL simulation model as a
supporting tool in the design of the breadboard model DN3.4 and the engineering model EM-DN3.4.
Future work should consist in investigating the effect of the cathode misalignment and the electrodes’ relative
distances. Other studies should address the influence the neutrals in the electron beam divergence.
Acknowledgments
This work has been conducted under the ESA Activities: AO/1-10538/20/NL/MG (Activity No. 1000030122) and
AO/1-11410/22/NL/SD (Activity No. 1000034872).
References
[1] Guarducci, F., Marangone, D., Clark, S., Lewis, R., Gabriel, S., Smirnova, M., and Mingo, A., “Development and
Industrialization of the RIT 3.5,” 37th International Electric Propulsion Conference, 2022, pp. 1–17.
[2] Gasa, K., Rogers, E., Gabriel, S. B., Daykin-Iliopoulos, A., Bosi, F., and Guarducci, F., “Development of a Low Current
Dry Neutralizer in a Pierce Gun Configuration,” The 37th International Electric Propulsion Conference, 2022, pp. 1–12.
[3] Lev, D. R., Mikellides, I. G., Pedrini, D., Goebel, D. M., Jorns, B. A., and McDonald, M. S., “Recent Progress in Research
and Development of Hollow Cathodes for Electric Propulsion,” Springer Singapore, 2019. https://doi.org/10.1007/s41614-
019-0026-0
[4] Bock, D., and Tajmar, M., “Highly Miniaturized FEEP Propulsion System (NanoFEEP) for Attitude and Orbit Control of
CubeSats,” Acta Astronautica, Vol. 144, No. January, 2018, pp. 422–428. https://doi.org/10.1016/j.actaastro.2018.01.012
[5] Capacci, M., Matticari, G., Materassi, M., Nicolini, D., and Frigot, P.-E., “Neutraliser for FEEPs : Review on the
Development Status,” 2009.
[6] Seifert, B., and Krejci, D., “Integrated Electric Propulsion Systems for Small Satellites,” No. May, 2014.
[7] A. S. Gilmour, J., “Microwave and Millimeter-Wave Vacuum Electron Devices: Inductive Output Tubes, Klystrons,
Traveling-Wave Tubes, Magnetrons, Crossed-Field Amplifiers, and Gyrotrons,” Artech, 2020.
[8] Pierce, J. R., “Rectilinear Electron Flow in Beams,” Journal of Applied Physics, Vol. 11, No. 8, 1940, pp. 548–554.
https://doi.org/10.1063/1.1712815
[9] Spataro, B., Behtouei, M., Di Paolo, F., and Leggieri, A., “A Low-Perveance Electron Gun for a High-Efficiency Ka-Band
Klystron,” European Physical Journal Plus, Vol. 137, No. 7, 2022, pp. 1–11. https://doi.org/10.1140/epjp/s13360-022-
02987-y
[10] Humphries, S., “Charged Particle Beams - S. Humphries.Pdf,” 2002.
[11] Vaughan, J. R. M., “Synthesis of the Pierce Gun,” IEEE Transactions on Electron Devices, Vol. 28, No. 1, 1981, pp. 37–
41. https://doi.org/10.1109/T-ED.1981.20279
[12] Lurtmarrasfeldt, W. B., “Egun-an Electron Optics and Gun Design Program,” Vol. 28, 1988.
[13] Humphries, S., “TRAK Charged Particle Tracking in Electric and Magnetic Fields,” Vol. 604, 2008, pp. 597–604.
https://doi.org/10.1063/1.45318
[14] “OPERA Simulation Software.” Retrieved 10 May 2023. https://www.3ds.com/products/simulia/opera
[15] Liu, H., Li, M., Ning, Z., Ren, J., Tang, H., Yu, D., Demidov, E. V., Eliseev, S. I., and Kudryavtsev, A. A., “2-D Modeling
of Orificed Hollow Cathodes of Stationary Plasma Thrusters SPT-100,” IEEE Transactions on Plasma Science, Vol. 43,
No. 12, 2015, pp. 4024–4033. https://doi.org/10.1109/TPS.2015.2457677
[16] Gabriel, S. B., “Validation of a Drift Diffusion Model for a Hollow Cathode,” 2019, pp. 1–12.
[17] Gabriel, S. B., “Two-Dimensional Modelling of Hollow Cathodes with Krypton Propellant,” International Electric
Propulsion Conference, No. 2, 2022, pp. 1–8.
[18] COMSOL Multiphysics, “AC / DC Module User ’ s Guide,” 2019.
[19] COMSOL Multiphysics, “Particle Tracing Module,” 2011.
[20] Gasa, K., Daykin-iliopoulos, A., Guarducci, F., and Gabriel, S. B., “Experimental Characterization of a Low Current Dry
Neutralizer,” The 38th International Electric Propulsion Conference, 2024, pp. 1–13.
12

Dimensional Model

Uploaded by

Copyright:

Available Formats

Dimensional Model

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Dimensional Model

Uploaded by

Copyright:

Available Formats

Two-dimensional modelling of a low current electron gun

dry neutralizer for ion engines

Klevis Gasa 1 and Stephen B. Gabriel 2

Alexander Daykin-Iliopoulos 3 and Francesco Guarducci 4

Abstract: Dry, or propellantless, dispenser cathodes represent an attractive alternative

II. Modeling Procedure

A. Assumptions and Main Equations

𝜌𝜌̅𝑠𝑠 = �� 𝑤𝑤𝑗𝑗 � ��� 𝑤𝑤𝑗𝑗 � 𝜌𝜌̅𝑠𝑠,𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 + 𝜔𝜔𝑖𝑖 𝜌𝜌𝑠𝑠 � (4)

4𝜀𝜀0 2𝑒𝑒 𝑉𝑉 3/2

B. Geometry and Mesh

Figure 1. Neutralizer components and geometry

Table 2. Sensitivity analysis on the iteration number

𝒏𝒏 ⋅ (𝜀𝜀0 𝜀𝜀𝑟𝑟 𝑬𝑬 ) = 0 (9)

𝒏𝒏 ⋅ 𝑫𝑫 = – 𝛻𝛻𝑡𝑡 ⋅ 𝜀𝜀0 𝜀𝜀𝑟𝑟𝑟𝑟 𝑑𝑑𝑠𝑠 𝛻𝛻𝑡𝑡 𝑉𝑉 (10)

Where 𝛻𝛻𝑡𝑡 represents a tangential derivative.

Table 3. Boundary conditions summary for the COMSOL model

III. Numerical Results and Discussion

𝐼𝐼𝑐𝑐 = 𝐼𝐼𝐺𝐺 + 𝐼𝐼𝐵𝐵 (11)

𝐼𝐼𝐵𝐵 = 𝐼𝐼𝐴𝐴 + 𝐼𝐼𝑐𝑐𝑐𝑐𝑐𝑐 (12)

Table 4. Perveance estimation at different operating conditions

I Cathode I Anode I Gate I Casing

You might also like

𝜌𝜌̅𝑠𝑠 = �� 𝑤𝑤𝑗𝑗 � �� 𝑤𝑤𝑗𝑗 � 𝜌𝜌̅𝑠𝑠,𝑝𝑝𝑝𝑝𝑝𝑝𝑝𝑝 + 𝜔𝜔𝑖𝑖 𝜌𝜌𝑠𝑠 � (4)