Simulation of a Hyperbolic Field Energy Analyzer arXiv ...

Simulation of a Hyperbolic Field

Energy Analyzer

Angel Gonzalez-Lizardo, Ernesto Ulloa

Abstract

Energy analyzers are important plasma diagnostictools with applications in a broad range of dis-ciplines including molecular spectroscopy, electronmicroscopy, basic plasma physics, plasma etching,plasma processing, and ion sputtering technology.The Hyperbolic Field Energy Analyzer (HFEA) isa novel device able to determine ion and electron en-ergy spectra and temperatures. The HFEA is wellsuited for ion temperature and density diagnosticsat those situations where ions are scarce. A simula-tion of the capacities of the HFEA to discriminateparticles of a particular energy level, as well as to de-termine temperature and density is performed in thiswork. The electric field due the combination of theconical elements, collimator lens, and Faraday cupapplied voltage was computed in a well suited three-dimensional grid. The field is later used to computethe trajectory of a set of particles with a predeter-mined energy distribution. The results include theobservation of the particle trajectories inside the sen-sor, the comparison of the input energy distributionto the energy distribution of the particles capturedby the Faraday cup, and the IV characteristic at theFaraday cup, using the voltage sweep at the conicalelements as the abscissa.

1 Introduction

Energy analyzers are important plasma diagnostictools with applications in a broad range of dis-ciplines including molecular spectroscopy, electronmicroscopy, basic plasma physics, plasma etching,

plasma processing, and ion sputtering technology.Retarding Potential Analyzers (RPAs) are diagnos-tic tools generally using a series of electrostatic gridsto selectively repel some particles in order to detectthe ion energy distribution. Typical construction ofan RPA comprises three electrostatically-biased meshgrids and an optional fourth grid. Grids are alignedinside a housing with a detector placed behind them.In this work, a simulation of a grid-less energy ana-lyzer is presented. Devices similar to the one pre-sented in this work have been used for detectionof low energy particles [1], focusing of space-charge-dominated electron beams [2], ionospheric inducedperturbations caused by the space shuttle [3], andother applications. Additionally, single probes whichcan measure multiple parameters simultaneously andcan operate at relatively high temperatures have usesin the analysis of high temperature plasmas generatedin fusion research devices. The Hyperbolic Field En-ergy Analyzer (HFEA) is a device able to determineion and electron energy spectra and temperatures. [4]At low plasma densities, Langmuir probes are wellsuited for measuring electron temperatures, but iontemperatures may be a challenge [5]. The HFEA iswell suited for ion temperature and density diagnos-tics at those situations where ions are scarce. TheHFEA consists of three main components: a planeparallel disc with interchangeable diameter orificesfor rejection of electrons and collimation of the ionbeam, a dual conical lens for particle energy selec-tion and beam focusing, and a Faraday Cup as theparticle collector.

Figure 1: HFEA Components

An exploded view of the HFEA parts is shown inFigure 1. According to Leal-Quiros [4], the conicallens constitute a hyperbolic cross-section lens dueto the equipotential surfaces in the vicinity of a di-aphragm, which are hyperboloids of revolution. A

1

arX

iv:1

610.

0060

4v1

[ph

ysic

s.pl

asm

-ph]

3 O

ct 2

016

saddle point crossed by straight field lines forms inthe center of the device. An angle of 54◦ − 44’ be-tween the straight lines and the axis at the saddlepoint was used for designing the conical lenses. Ifa potential VR is imposed on the saddle point, onlyparaxial ions with energy greater than qVR can passthrough the lens into the Faraday Cup.

It is known that small systems of a few thousandparticles can simulate accurately the collective be-havior of real plasmas [6]. With this in mind, a simu-lation of the capacities of the HFEA to discriminateparticles of a particular energy, as well as to deter-mine temperature and density is performed in thiswork. The electric field created by the combinationof the conical elements, collimator lens, and Fara-day cup applied voltage was computed in a three-dimensional grid and used to compute the trajectoryof a set of particles with a predetermined energy dis-tribution. The results include particle trajectoriesinside the sensor, the comparison of the input energydistribution to the energy distribution of the parti-cles captured by the Faraday cup, and the Current-Voltage (IV) characteristic at the Faraday cup, usingthe voltage sweep at the conical elements as the ab-scissa.

2 Methodology

A matlab simulation of the electric field near the con-ical lens is performed and analyzed. The geometry ofthe device was created in matlab (Figure 2) usingsurfaces to represent the boundary between the dif-ferent elements of the analyzer. Figure 2 shows thetwo-cone electrode and the outer cylinder containingthe assembly. Initial voltage conditions were given toeach of the elements in order to find the solution ofequation 1, with Φ the electric potential matrix atthe 3D space in the vicinity of the conical lens.

∇2Φ = 0, (1)

This equation was numerically solved for a rect-angular grid of n × n × n with Dirichlet boundarycondition. The potential at the surfaces of the de-vice under study were fixed. The 3D electric field is

Figure 2: HFEA geometry

found by and computing the gradient of the potentialat each point of the grid, by

−→E = −∇Φ (2)

The solution for the electric field was used later onto compute the forces exerted on particles enteringthe device in a kinematic simulation of their trajec-tories. This trajectory simulation was performed byassuming the entering particles had a Maxwellian en-ergy distribution. The distribution of velocities of theparticles reaching the Faraday cup was obtained andcompared with the initial distribution.

Figure 3 shows a graphical representation of theequipotential surfaces in the plane yz of the de-vice. The colored graph shows the lower potential inblueish colors, while the highest potential are closerto the red. The faraday cup voltage can be observedas a yellow oval in top of the z axis.

Figure 4 shows the electric field lines obtained for aparticular negative voltage applied to the two-coneselectrode.

2

Figure 3: Equipotential surfaces for VC=-143 V

Figure 4: Electric Field for VC=-143 V

3 Simulations

The methodology used in this work is quite simple.The objective was to produce a scaled version of theIV characteristic obtained in [4] for electrons andions. A simulation using the two species in equalnumber was performed to achieve the goal.

To simulate the effect of applying a ramp voltagefrom -200 V to 200 V to the two-cones electrode inthe device, the 3D electric field produced by all thevoltages in the device ~E was computed for the rampvoltage applied to the electrode, in steps of one volt.The electric field is computed at discrete points of thespace inside the device in a n × n × n grid. The 3Delectric field grid for each level of voltage is stored ina file to be used in the subsequent particle trajectorysimulation. Figure 4 shows a profile of the electricfield at the central plane of the device for a partic-ular level of voltage. Figure 3 shows a profile of theequipotential surfaces at the same plane.

When the electric field data for all 401 levels ofvoltage was obtained, a simulation of particles be-ing subject to the electric field for each step of volt-age was performed. A number of particles np weredefined as entering the device at a random positionwithin the collimator lens input hole. The velocity ofthe entering particles was defined as obeying a nor-mal distribution on the z axis and zero velocity inthe other two directions (this condition can be easilychanged). For the case of simulation of ions and elec-trons and ions together, only hydrogen ions that lostone electron were included in the simulation for sim-plicity. The following classical mechanics equationsof motion were used in this simulation:

~ap =~E

m(3)

~∆sp = ~vpts + ~at2s (4)

sp = sp + ~∆sp (5)

~vp = ~vp + ~ats (6)

Where ~ap is the particle acceleration, ~E is thethree-dimensional magnetic field at the closest gridposition from the particle, ~∆sp is the incrementalchange in particle position, ~vp is the particle velocity,

3

ts is the time step selected for calculation, and sp isthe current position of the particle.

The electric field used for the acceleration calcu-lation was approximated to the electric field at thenearest point of the grid from the position of the par-ticle ~Eγ . Hence for a particle i, with i = 1 . . . np withnp the total number of particles in the simulation,its acceleration at the position (xki , y

ki , z

ki ), with k

representing the kth step given by the particle in itstrajectory, is computed as [6]:

~aki =qi ~Eγmi

(7)

The displacement of the particle i in step k is com-puted using

∆ski = ~v(k−1)i ts + ~aki

t2s2

(8)

Thus, the absolute position of the particle i in stepk is calculated as

sk+1i = ski + ∆ski (9)

and the velocity vector is computed as

~vkp = ~vk−1p + ~aki ts (10)

The position of each particle in the simulation iscompared to the coordinates of all the surfaces in thedevice to detect the end of its trajectory. When theparticle touches any surface, no further calculation ofposition, velocity and acceleration is performed. Theparticles that hit the Faraday cup input are countedto compute both the current in the Faraday cup andthe energy probability distribution of particles at theFaraday cup.

4 Results

One of the first results being pursued by the simu-lation was the verification of the formation of a sad-dle point in the potential generated by the voltageapplied to the cone electrodes in the center of thesegment connecting the tips of the conical electrode.Figure 5 shows the saddle point formed by the volt-age magnitude in the center plane for an arbitrary

voltage applied to the cones VC and a small voltageapplied to the Faraday cup. It can be observed thatthe voltage barrier is minimum at the center betweenthe two conical electrodes.

Figure 5: Saddle point in the center plane of the volt-age

Figure 6: Saddle point voltage vs. VC for Faradaycup Voltage of 67 V, and collimator voltage of ±10V, switching voltages polarities to the opposite of theramp

Figure 6 shows the saddle point voltage as a func-

4

tion of the VC for the particular of a ramp from -200V to 200 V, ±67 V at the Faraday cup, and a colli-mator voltage of ±10 V. It most be pointed out thatthe polarity of the Faraday cup and collimator lensvoltages were opposite to the VC sign to obtain thisparticular plot, explaining the discontinuity at zero.

Figure 7 shows the graphic output for a particlesimulation using 1,000 particles with a normal energydistribution on the z axis of the simulation, shown inFigure 8 shows the energy distribution impressed tothe particles entering the device. The same numberof ions and electrons were included in the simulationto keep the neutrality of the total charge. It maybe observed that particles tend to be concentrated inthe plane xz of the device, perpendicular to the axisof the conical electrodes. The electric field formedby the conical electrodes and the Faraday cup appar-ently pushes the particles into that plane. It can bealso observed that only a fraction of the particles iscaptured by the Faraday cup (a simplified version ofthe Faraday cup was used in this model).

Figure 7: Simulation graphical output for 1000 par-ticles, -200 to 200 V, 67 V and 10 V

Figure 9 shows the current collected at the FaradayCup for a the distributions shown in Figure 8. It may

Figure 8: Particle Energy Distribution at Collimatorlens and at the Faraday Cup.

be observed that the form of the IV characteristic forions (left side of the plot) is very similar to the onesobtained in [4]. The right side of the plot shows theelectron current, which is also similar to the behaviorreported in [4], which is shown in Figure 10. Noticethat from -50 V to about 100 V the IV characteristicshows a behavior very similar to the one shown atFigure 10. From -150 V to -50 V, the shape of theIV characteristic shows the same behavior as the ioncurrent collected by Leal in [4], shown in Figure 12.

5 Conclusions

A simulation of the Hyperbolic Field Energy Ana-lyzer is performed with satisfactory results, achiev-ing to determine and plot the dependency betweenthe saddle point potential and the retarding potentialapplied to the conical elements that form the hyper-bolic potential lens at the center of the device.

The current at the collector (Faraday Cup) is alsoplotted against the retarding potential applied to theconical lens, exhibiting results similar to the ones ob-tained experimentally [7, 8, 4].

The energy distribution obtained in the Faradaycup remarkably resembles the energy distribution atthe input port (L0) (Figure 8). It is clear that themeasurements taken with the sensor will be done on apopulation which is representative of the input pop-ulation. This simulation, however, only takes intoaccount the particles already inside the sensor, after

5

Figure 9: IFaraday vs. VC for |VFaraday| = 67 V andVL0

= 10 V.

Figure 10: IV characteristic for only electrons, for|VFaraday| = 67 V and VL0 = 10 V

collimated by the lens L0. A more thorough simula-tion must be done to take into account the discrimi-nation performed for the collimator lens.

Figure 11: IFaraday vs. VC for |VFaraday| = 67 V andVL0 = 10 V.

Figure 12: IV characteristic for only ions, for|VFaraday| = 67 V and VL0 = 10 V

6

Future Work

Determining the relationship between the voltage ap-plied to the Faraday cup and ability of the sensor tofocus particles into it is the next step in the simula-tion of the HFEA. Also, establishing the performanceof the sensor at higher magnitudes of E will be inves-tigated, new geometries for the electrodes, and sim-ulation of the sensor inside a plasma simulation aretasks in the plans of the authors. These tasks willrequire higher computational resources and probablysoftware other than only matlab.

References

[1] W. E. H. P. B. Shyn, T. W.; Sharp, “Gridlessretarding potential analyzer for use in very lowenergy charged particle detection,” Review of Sci-entific Instruments, vol. 47, no. 9, pp. 1005–1015,September 1976.

[2] Y. Cui, Y. Zou, A. Valfells, M. Reiser, M. Wal-ter, I. Haber, R. A. Kishek, S. Bernal, and P. G.OShea, “Design and operation of a retarding fieldenergy analyzer with variable focusing for space-charge-dominated electron beams”, journal = Re-view of Scientific Instruments, year = 2004, vol-ume = 75, number = 8, pages = 2736-2745,.”

[3] D. L. Reasoner, S. D. Shawhan, and G. Mur-phy, “Plasma diagnostics package measure-ments of ionospheric ions and shuttle-inducedperturbations,” Journal of Geophysical Re-search: Space Physics, vol. 91, no. A12,

pp. 13 463–13 471, 1986. [Online]. Available:http://dx.doi.org/10.1029/JA091iA12p13463

[4] E. Leal-Quiros, “Novel probes and analyzers forRF heated plasmas and microwave heated plas-mas for controlled fusion research: The hyper-bolic energy analyzer, the magnetic moment an-alyzer, the double energy analyzer, and the vari-able energy analyzer,” Ph.D. dissertation, Uni-versity of Missouri-Columbia, 1989.

[5] I. H. Hutchinson, Principles of plasma diagnos-tics. Cambridge (U.K.), New York, Melbourne:

Cambridge University Press, 2002, autre tirage :2005.

[6] C. K. Birdsall and A. B. Langdon, Plasma physicsvia computer simulation, ser. Series in plasmaphysics. New York: Taylor & Francis, 2005, orig-inally published: New York ; London : McGraw-Hill, 1985.

[7] M. A. P. E. Leal-Quiros and E. Garcıa-Otero,“The hyperbolic energy analyzer: A novel diag-nostic - ion probe,” in Proceedings of the 10thInternational Workshop on ECR Ion Sources. .Conf., F. Meyer and M. Kirkpatrick, Eds. ORNLand US-DOE, January 1991, pp. 97–119.

[8] E. Leal-Quiros. (2005, June) Basic plasma diag-nostic; probes & analyzers. [Online]. Available:http://www.pupr.edu/plasma

7