Charge Neutralization and Direct Thrust Measurements from Bipolar Pairs of Ionic-Electrospray Thrusters

Daniel G. Courtney* and Herbert Shea

Microsystems for Space Technologies Laboratory, Institute of Microengineering Neuchâtel campus, School of

Engineering, École Polytechnique Fédérale de Lausanne, Rue de la Maladière 71b, 2000 Neuchâtel, Switzerland

*dcourtney@alum.mit.edu

Kathe Dannenmayer and Alexandra Bulit

ESA-ESTEC, Noordwijk, Netherlands

Accepted for publication in the AIAA Journal of Spacecraft and Rockets, 2017

This is an author generated pre-print of the final (accepted) draft manuscript.

Charge Neutralization and Direct Thrust Measurements from Bipolar Pairsof Ionic-Electrospray Thrusters

Daniel G. Courtney∗ and Herbert Shea†

Microsystems for Space Technologies Laboratory (LMTS),Ecole Polytechnique Federale de Lausanne (EPFL), Neuchatel, CH-2002, Switzerland

Kathe Dannenmayer‡ and Alexandra Bulit§

ESA-ESTEC, Noordwijk Netherlands

Electrospray (ES) thrusters are unique in their ability to emit both positive and negative beams,each contributing similarly to thrust. Spacecraft charging could be prevented without a dedicatedneutralizer if currents of simultaneously emitted opposing polarity beams were matched; despitea lack of implicit coupling between them. We discuss and evaluate experimentally both an activecurrent-balance control circuit and a passive self-balancing configuration. Our highly ionic ES sourceincludes a pair of machined porous glass ES emitter arrays in a single holder. In the passive config-uration, unbalanced charge collection on a common floating extractor yields rapid charging towardsartificially imposed 200 V limits without means for charge suppression, showing the unsuitabilityof that architecture. Active balancing of emitted currents was fully effective at charge neutraliza-tion during stable periods of emission, yet insufficient during potential alternations. We presentmethods to improve this approach, including demonstrating beam current modulation at up to 450Hz. Direct thrust measurements when simultaneously emitting opposing polarity beams of the ionicliquid 1-ethyl-3-methylimidazolium tetrauoroborate are presented. Direct thrust measurements upto 35 µN were in excellent agreement with commanded levels and computations based on measuredcurrents (up to 190 µA) and voltages (up to 2100 V).

Nomenclature

α± = positive or negative thrust coefficient (NA−1V −1/2)∆V = thruster ground to common extractor potential (V)γ = potential alternation duty cycleRiso = thruster ground to facility ground resistance (Ω)Csys = thruster ground to facility ground capacitance (F)fExi = fraction of module i beam impinged on extractor gridIi = emitted current, module i (A)Iexi = extractor current, module i (A)TGND = thruster ground to facility ground potential (V)

T±i = positive or negative thrust of module i (N)

TiAvg = average thrust over one polarity alternation period (N)VExA = module A extractor potential (V)V+ = positive beam voltage (V)V− = negative beam voltage (V)

1. INTRODUCTION

Electrospray thrusters configured as Ionic LiquidIon Sources (ILIS)[1], wherein beams of primarily ionsand small clusters are emitted from Ionic Liquid (IL)propellants, have potential applications to both small

∗Postdoctoral Researcher, EPFL-IMT-LMTS, AIAA Mem-ber, dcourtney@alum.mit.edu, Present address: Busek Co.,Natick MA 01701 USA†Associate Professor, EPFL-IMT-LMTS.‡Electric Propulsion Engineer ESA-ESTEC-TEC-MPE.§Electric Propulsion Engineer ESA-ESTEC-TEC-MPE.

satellites and as distributed thrusters on larger space-craft. Compact thruster prototypes with 100’s of emis-sion sites distributed over a few cm2 frontal area havebeen shown to deliver 10’s of µN of thrust with lessthan 1 W of input power and with a specific impulsein the range of 1000 to 3000s[2, 3]. The inherently lowpropellant flow rates are conducive to passive regula-tion of the IL propellant supply by capillarity; therebyobviating the need for active propellant control[1]. Theability to emit high performance beams of both posi-tively and negatively charged particles has been shownpreviously, see for example [1–6]. The possibility ofachieving charge neutralization without a dedicatedneutralizer has, correspondingly, been envisioned[7, 8].Removing the resources required for an electron emit-ting neutralizer would further simplify the technologyand therefore suitability for small spacecraft.

Figure 1 presents a schematic overview of a charge-neutralized bipolar ILIS thruster pair. Equal beamsof positive and negative ion beams are emitted simul-taneously from adjacent sources due to enforced beam(VA, VB) and extractor (VExA, VExB) voltages. Peri-odic potential alternation can be employed to suppresselectrochemical degradation within the IL propellantand at contact electrodes[9, 10]. Here we discuss theviability and some subtleties of simultaneous bipolaremission and present charge neutralization and directthrust measurements from demonstration devices, sim-ilar to those presented in Ref. [2].

2

Thruster Module A Thruster Module B

+ ++++ + ++ ++ ++++ + ++ +

+ ++++ + ++ +

- ---- - -- -- ---- - -- -

- ---- - -- -

+ ++++ + ++ ++ ++++ + ++ +

+ ++++ + ++ +

++

+

++

+

- ---- - -- -- ---- - -- -

- ---- - -- -

--

-

--

-

+ ++++ + ++ ++ ++++ + ++ +

+ ++++ + ++ +

++

+

++

+- ---- - -- -

- ---- - -- -- ---- - -- -

--

-

--

-

Periodic polarity

alternation

Positive

particles

Negative

particles

Thruster /

SC ground

Zero total current

VA

VExA

VB

VExB

TGND

(V)

Space /

facility ground

FIG. 1: Electrospray sources configured to emit zero total beam current through simultaneous bipolar emissions.

1.1. ILIS Charge Neutralization

Electrospray and liquid metal Field Effect Electro-spray (FEEP) thrusters do not emit electrons inher-ently. They are often (necessarily in the case of FEEP)operated at only positive polarity alongside a neutral-izing electron source[11–15]. As a particularly rele-vant example, the NASA-JPL Drag Reduction Sys-tem (DRS) recently flown aboard the ESA led LISAPathfinder mission demonstrated colloidal electrospraythrusters. This system[13], the first demonstrationof electrospray propulsion in space to be reported indetail[14], includes a carbon nanotube field effect cath-ode as neutralizer.

Here, we primarily consider charge neutralization asopposed to beam neutralization. Charge neutraliza-tion requires the total emitted and collected currentsfrom a spacecraft or electrically isolated thruster mod-ule to sum to zero to prevent charging. Without neu-tralization, a spacecraft’s potential relative to its envi-ronment could quickly charge, leading to issues includ-ing damaging arcs or discharges, corruption of scien-tific measurements and/or a disruption of the emittedbeam[16, 17]. Detrimental effects may occur due toboth differential charging between electrically isolatedcomponents[18] or charging of the spacecraft and as awhole with respect to the ambient space plasma[16].When ion beams are emitted, charging to a potentialapproaching the beam energy could stall the emissionand attract particles back towards the spacecraft[19].In the dual ion beam configuration at hand it is note-worthy that while this would lead to an attraction ofions at one polarity, the opposite polarity would be fur-ther accelerated away from the spacecraft[20].

The degree to which a spacecraft may float at highpotential, and the effects of such potentials, are heav-ily dependent on the local environment and spacecraftdesign. Environment induced current sources may in-clude the local plasma and/or photo-emission fromspacecraft surfaces. ILIS and colloidal electrospraythrusters are highly scalable (see for example Refs.[3, 4, 6, 21]) yet can also be configured to provide ex-

treme levels of precision[13]. Accordingly these systemsare applicable to a myriad of applications both at LowEarth Orbit (LEO) and in higher orbits including GEOor interplanetary space.

An immediate complication inherent to ion-ioncharge neutralization using ILIS is evident consideringapplications at high orbits. Here, the plasma densityis low and therefore the Debye length λ will be muchlarger than a small satellite such that electron/ion col-lection from ambient may be negligible. When λ is rela-tively large, the capacitance of an approximately spher-ical spacecraft of radius Rsc is shown by Hutchinson[22]to be Csc ≈ 4πϵ0Rsc (Rsc/λ+ 1), reducing to sim-ply the vacuum level 4πϵ0Rsc when Rsc << λ. Inthis limit, a 30 cm radius small satellite would havecapacitance Csc ≈30 pF and a current imbalance ofonly 10 µA would charge a spacecraft to 1 kV inonly 3 ms. Thus, practical beam current equalizationmust occur quickly and to a high degree of accuracy.High voltage charging in GEO may not necessarily becatastrophic, see for example the discussions of theSCATHA spacecraft in Refs. [17, 19, 23]. However;beam stalling complications such as sputter damage byreturned ions or failures due to differential chargingmay be incurred[19].

In the relatively dense LEO plasma, small Debyelengths augment the effective capacitance of the space-craft and a relatively large level of ambient electronflux may be accessible from the local plasma with min-imal potential rise. However; the level of collected cur-rent is highly variable. The reviews of Hastings[23] andGarrett[18] in particular detail conditions whereby sig-nificant spacecraft charging may occur in LEO; withspecific emphasis on the drastic reductions in localplasma density which can occur in eclipse or in near-polar orbits. A particular example, highlighted byboth reviews is evidence that the Defense MeterologicalSatellite Program (DMSP) satellites charged to morethan -1kV in their sun-synchronous orbits, see Ref.[24] for details. Suppression of negative charging isparticularly challenging[16, 23] as the ambient ion flux,necessary to suppress negative charge, is much lowerthan electron flux and predominantly only collected by

3

ram facing surfaces. In addition to governing collectedion flux, spacecraft attitude may also influence electroncollection due to wake effects[23] which limit currentcollection to conducting surfaces in LEO. For example,the space shuttle charged positively to 5kV by ejectionof a (5kV ) electron beam when the shuttle’s primarycurrent collection surfaces (engine bells) were in thespacecraft’s wake during STS-9[25].

Hence, while some circumstances may exist whererapid swings in spacecraft charge may be acceptableor be suppressed by the local environment, a bipo-lar ILIS system should, in general, be capable ofrapid charge neutralization to maximize applicability.Plasma thrusters that emit a bipolar beam naturallyor use plasma based cathode neutralizers, such as hol-low cathodes, are inherently forms of plasma contac-tors which automatically provide a passive means ofcharge neutralization[23]. Here, the current of low en-ergy electrons within the emitted plasma is not directlycontrolled but rather self regulates due to the chang-ing potential structure within the beam and/or down-stream of emission. The result is an automatic chargebalance facilitated by a coupling voltage between thecathode, thruster plume and local space plasma of afew 10’s of volts[26]. This scenario differs from bipolaroperation of an ILIS thruster where positive and nega-tive high particle mass, high energy beams are emittedwithout any low energy electron population. Currentsof these two beam are dictated by the applied elec-trostatic fields in the vicinity of each source alone (inpassively fed thrusters), with no implicit coupling be-tween those fields. If beam currents were well balancedas part of the thruster design, spacecraft charging couldbe held close to the natural level reached by environ-mentally induced collection/emission of charged parti-cles from spacecraft surfaces.

Two methods of enforcing automatic charge neutral-ization are experimentally evaluated herein: i) a passivemethod previously proposed and tested in Refs. [3, 20]and ii) a closed loop active current-control method.

1.2. Beam Neutralization

Beam neutralization refers to the neutralization ofspace-charge within the emitted beam(s). Beam neu-tralization can contribute to performance through,for example, influencing the aforementioned cathodeto plume coupling voltage in plasma-based electro-static thrusters, or contribute to accelerating low en-ergy (e.g. charge-exchanged) particles within a plumeback towards the spacecraft. The latter may leadto sputtering or contamination damage on spacecraftsurfaces[23, 26]. In ILIS, the extremely low vapourpressure of ILs limit emissions to species ejected in theelectrospray. Accordingly nearly all emitted particlestravel at high speed, suppressing the degree to whichcharge-exchange ions may be readily attracted back toweakly charged spacecraft surfaces. Furthermore, duethe low plume density, high particle mass and high en-ergy of bipolar ILIS beams, interactions due to poor

beam neutralization are not expected to significantlyalter thruster performance if charge neutralization isachieved. In section 3.3 we present thrust measure-ments from a grounded device emitting adjacent ac-tively balanced bipolar beams and compare those mea-surements with a sum of calculated thrusts based onpreviously published correlations.

2. EXPERIMENT APPARATUS ANDMETHODS

2.1. Sources

A demonstration thruster comprising two of thesources described in Ref. [2] within a single 57 mmx 27 mm x 10 mm package was used in all bipolarmeasurements presented, see Figure 2. As in the citedwork, each emission source (module) contains a 1 cmdiameter, 3 mm thick porous borosilicate substrate fea-turing nine 7.5 mm long triangular prisms each termi-nated by a sharp (few 10’s of µm radius) edge. Anextracting grid comprising 350µm wide laser cut slitsin 100 µm thick molybdenum is aligned with the emit-ting edges and positioned roughly 100 µm above them.A ∼ ±2 kV voltage applied between the grid and ILsoaked within the porous glass initiates numerous emis-sion sites along the edges, facilitating ion beams of sev-eral hundred µA. Unlike in Ref. [2], contact with theIL is made via a stainless steel, rather than aluminum,flange. The same conventional CNC milling techniquewas used to form the emitters. A second porous borosil-icate disc (Duran Group P3 grade) was mounted be-low each emitting layer and served as the IL reservoir.Since the IL itself is polarized, independent reservoirsfor each module were required. The interfacial Laplacepressure induced by the reservoir has been shown tobe critical to the emission characteristics[27]; the P3reservoir grade used in all experiments here matchesRef. [2].

All demonstration thrusters were wet with the IL1-ethyl-3-methylimidazolium tetrafluoroborate (EMI-BF4) which had been previously degassed in high vac-uum for at least 12 hours. Liquid has been applied tothe upper surface of each emitter and allowed to fillboth the emitter and reservoir under vacuum. The ex-tractor grid was aligned and fixed into position afterwetting.

2.2. Test Facilities

The charging and current control experiments pre-sented in section 3.1 and 3.2 have been performed pri-marily at the Microsystems for Space Technologies Lab-oratory (LMTS) at the Ecole Polytechnique Federale deLausanne, Switzerland. Additionally, thrust measure-ments were made at the European Space Agency (ESA)Propulsion Laboratory (EPL) at the European Space

4

Module A

Alternating polarity+ Beam - Beam

Module BCommon

Extractor grid

Vem

Reservoir layer

Emitter layer

Emission site

Particle beams

Thruster com.

a) b) c)

FIG. 2: a)Photograph of a dual thruster module device. b) Microscopic photograph of the porous glass emittersapplied here. c) Cross-sectional overview of the critical components of each thruster module.

Research and Technology Centre (ESTEC), Noordwijk,Netherlands.

At the EPFL, the facility comprised a six-wayISO160 cross pumped by a Varian V70 turbomolecu-lar pump. The measured pressure typically rose from abase pressure less than 2 x 10−6 mbar to, at most, 1 x10−5 mbar during emission. At ESTEC, measurementswere performed within the Galileo vacuum facility com-prising a 1.0 m long x 1.2 m diameter chamber pumpedby both a turbomolecular pump and cryo-heads yield-ing base pressures below 1 x 10−7 mbar and operationalpressures up to ∼ 5 x 10−6 mbar during emission.

In all presented data the emission polarity of eachmodule has been alternated using a custom-madeswitching circuit driven by a 1 Hz square wave inputwith 50% duty cycle. This circuit utilized Voltage Mul-tiplier OC100G optodiodes to effect optically-isolatedsymmetric switching of HV signals in less than 400 µs.In all tests one positive and one negative high volt-age supply were used to power either one or two activemodules. At the EPFL, EMCO H30NR and H30PRDC-DC converters were used to supply floating emis-sion voltages during charge neutralization testing whileStanford Research Systems P350 high-voltage supplieswere used during current following demonstrations. AtESTEC, FUG models HCN 140-12500 and HCN 35M-20000 high voltage supplies were used as positive andnegative supplies respectively.

Measurements of the emitter and extractor currentsfor both modules were made with custom made opti-cally isolated current monitors. The accuracy of thesemonitors, verified with a Keithley 6487 picoammeterwas within ± 2 µA at the currents of interest. As dis-cussed in the introduction a few µA of unbalanced cur-rent may be sufficient to induce relatively severe space-craft charging and, as will be shown in section 3, wassufficient to induce charging on the isolated facility dur-ing neutralization tests. As a result, this level of accu-racy presents some limit to quantitative interpretationof current sums when near zero. The current monitorinternal circuitry includes a 1 kHz low pass filter.

During charge neutralization measurements, all in-put and output signals were isolated via opto-isolators

or disconnected. Two zener diodes were installed be-tween the local reference ground and facility groundto limit charging of the equipment to within ± 200V .The effective capacitance and resistance of the localreference to facility isolation were measured to be lessthan 1 nF and approximately 500 MΩ with an AgilentE4980A LCR meter and a Keithley 6487 electrometerrespectively. The thruster reference to facility groundpotential TGND was measured using a non-invertingamplifier a gain of 100:1. The available facilities didnot support active measurement of true beam voltageoutputs during floating operation due to a lack of iso-lated feedback. Instead, characterizing current versusvoltage measurements were made prior to tests withthe facility and local reference grounds connected.

It should be noted that facility effects may havecontributed to the measured data. For example, therelatively small vacuum chamber at EPFL was heldat facility ground and would have emitted secondaryelectrons upon bombardment from the emitted elec-trospray beams. These secondary electrons could havesuppressed charging, particularly to positive potential.However; on numerous occasions (see section 3) thethruster floating potential rose to large positive values.

Direct thrust measurements were made at the ES-TEC facility using a Mettler-Toledo AX504 balancemodified for use in vacuum. The balance output wasverified using calibrated masses and had a resolutionof approximately 1µN . An electromagnetic force actu-ator was used to verify consistent balance output pe-riodically as the thruster was installed in the facilityand electrical interconnects were made. Specifically,the relative response to consistent steps in magnetiza-tion current was measured with the thruster unloaded,with electrical connections made and, finally, with amoderate potential below the emission voltage applied.Thrust measurements were made after confirming rel-ative responses in the final case matched the unloadedconfiguration.

5

V+

V-

1 Hz

IexB

IexA

IB

IA

Module A Module B

Thruster ground

Riso

Csys

++

+++ +

--

-

-- -

TGND +-

N/C

-- --+

+

++ Impingement

Current

monitor

FIG. 3: Configuration for automatic charge neutraliza-tion with thruster supplies floated with respect to ex-tractor electrodes at a common potential either floatingor tied to facility ground.

2.3. Passive Current Balancing

Automatic current balancing has been proposed andrecently demonstrated in Ref. [20] whereby the emittersupplies are floated with respect to a common extrac-tor electrode. This configuration is shown in Figure 3.When in operation any excess charge accumulation atthe thruster common due to a current imbalance leadsto an increase or decrease in potential with respect tothe common extractor. These variations in potentialnaturally tend to suppress or augment emissions in amanner which suppresses charge accumulation. Thatis, the extractor is intended naturally to float, relativeto the thruster common, at a potential which enforces acurrent balance. Some degree of beam interception onthe extractor is expected in any electrospray thruster;hence the impact of this charge accumulation was ofparticular interest here. Accordingly, two versions ofthis configuration were investigated in the work pre-sented here. First, the common extractor was tied tofacility ground. Second, the extractor has been isolatedfrom the facility ground to emulate spacecraft condi-tions where, for example, the extractor may be tied tothe spacecraft and have no active means for dissipatingcharge. Referring to Figure 3 these two configurationsare denoted by a selectable extractor contact.

2.4. Closed Loop Current Control

An active feed back mechanism was experimentallyevaluated in this study as an alternative to the passivebalance method described above. Referring to Figure4, the potential between the IL and the extractor gridof module (A) was modulated by altering the extrac-tor potential relative to the thruster ground (VExA).The control circuitry permits VExA, and thereby the

V+

V-

1 Hz

IexA

IexB

IB

IA

Thruster ground

Riso

Csys

TGND +-

+ +

VExA

Module A

++

+++ +

++

++

Module B

--

-

-- -

-- --

+

-

Impingement

Current

monitor

FIG. 4: Configuration for closed loop charge neutral-ization where the sum of emitted currents has beenused to modulate the module A emitter to extractorpotential.

emitter to extractor potential, to vary by up to the ±300 V limits of an EMCO G06CTR bipolar DC to DCconverter. As will be shown in section 3, this rangeof extracting potential variation corresponds to many10’s of µA of current control.

The extractor voltage control circuitry is shown inFigure 5. Vishay LH1500AT solid state relays (SSRs)are controlled via a comparison of drive and follower in-puts. The output is fed through a low pass filter (LPFin the Figure) with cut off frequency of approximately300 Hz. Although relatively slow compared with thecharge rates described in the introduction, this cut-offfrequency was selected to be compatible with the fewms rise time characteristics of the SSRs. When usedto balance current output, the comparator inputs com-prise the module B current and the inverse of the mod-ule A current. The circuit drives the extractor voltageto be more positive when the comparator input is pos-itive and negative otherwise. Hence when the drivesignal (e.g. module B emitted current) is larger thanthe follower (e.g. inverted module A emitted current)the extracting potential at module A, VA − VExA isreduced, promoting either enhanced negative emissionor reduced positive emission; always tending towardsa current that balances the drive signal. The followersignal could be amplified or reduced by a small amountprior to comparison with the drive via a configurablegain in a nominally unity-gain inverting amplifier. Thismanual tuning function has been included to accommo-date for slight differences in the current monitor gainsand, in practice, differences in the transmission frac-tions between devices. This component is not shown inFigure 5.

The extractor voltage, rather than that of the emit-ter, has been modulated for two largely practical rea-sons. First, electronic components and circuitry aresimplified when controlled signals are at several hun-dred rather than several thousand volts. In ILIS, the

6

+

-

-300V

+300V

10M

10M

LPF

2K

2K

VexA

TGND

LH1500AT

Drive

Follower

5V

FIG. 5: Schematic overview of the balance driver cir-cuit used here.

beam potential may be many kV in some applicationsyet the operating voltage range is typically within sev-eral hundred volts of a start-up potential; below whichno emission occurs (see the current-voltage measure-ments in section 3). Furthermore, as increasingly scaledup ILIS thrusters are envisioned, complications mayarise when attempting to modulate the primary cur-rent carrying HV signals to the emitters. Under nom-inal conditions, the extractor electrode intercepts onlya small portion of the emitted beam current makingit further suited to modulation via low current compo-nents. Second, the emitter to ground plane potentials,VA and VB in Fig. 1, establish the total beam ener-gies and therefore specific impulse and thrust-per-unit-current for constant species charge-to-mass ratio andbeam shape. Through only modulating the extractorpotential, the total beam energy and therefore theseperformance metrics are virtually unchanged. Nomi-nally a second acceleration or deceleration grid alignedwith and closely position downstream of each extractororifice would ensure efficient acceleration or decelera-tion to the enforced outer potentials[6] and provide adegree of performance-enhancing focusing when accel-eration is applied. However, when in close proximityto the beam this electrode would present an additionalsource of beam current interception to be accountedfor by a practical implementation of closed-loop total-current control circuitry. In this demonstration, theextractor grid of module B was tied to thruster groundthrough a monitor and a simple ground plane on theconnection PCB well beyond the plume serves to, albeitweakly, define the downstream ground plane of moduleA. This ground plane is identified as the ’Common’electrode in Fig. 2.

Current following data from an individual emittermodule, emitting EMI-BF4, were acquired to validateand characterize the circuitry. In this test, a singlesource identical to that presented in Ref. [27] was uti-lized. This source again mirrors the configuration ofRef. [2] yet includes only a single 7.5 mm long emitteredge. Referring to Figure 6(a), two synchronized Agi-lent 33220A arbitrary function generators were used tosupply a drive signal comprising a sinusoid modulat-ing a 1 Hz square wave. The same 1 Hz square wave

was used to alternate the emitter polarity. This artifi-cial drive was input to the controller described abovealong with the emitted current feedback signal as the’follower’, see Figure 6(a). The frequency of the con-trol signal was then varied while recording the emissioncurrent. Figure 6(b) provides sample data at 20 Hz, 50Hz and 100 Hz using a modulation amplitude of ∼ 5µA about ±25 µA. At these relatively low frequenciesreasonable current following was achieved. However,in Figure 6(c) data acquired at 150 Hz, 250 Hz and400 Hz indicate the onset of both a phase lag and gainreduction. A ∼ 2 µA high frequency noise signal waspresent on all data and intrinsic to the current mea-surement circuit used in these circuit validation tests.

While short of a detailed spectrum analysis thesedata confirm that rapid modulation of the extractorgrid provides a simple means for modulating the totalemitted current at relatively high speed. However; theresponse rate fell below the ms time scales discussed asbeing potentially required in the introduction. Despitetheir relatively slow speed, SSRs were selected in lieuof, for example, the OC100G optocouplers used in theswitching circuit due to their low power consumption.The selected SSRs consume roughly 1 mW of powerwhile in operation, while the OC100G optocouplers re-quire in excess of 300 mW (100 mA at 3.0 V totalforward voltage) when active. Compared with the fewhundreds of mW input to each thruster module, thelatter is clearly significant hence the low power SRRoption was favoured.

This SSR based circuit was utilized in all charge neu-tralization tests presented here. However; an additionalsingle emitter current following test was performed us-ing OC100G optocouplers to effect switching. Thattest, presented in section 4.2 indicated a cut-off fre-quency of 450 Hz. The frequency limitations in Figure6 may therefore have been predominantly limited bythe drive circuitry rather than emission processes.

2.5. Total Thrust Output from BipolarSources

In earlier work[2] we demonstrated that thrust fromthe sources used here can be described by equation 1for each polarity of emission. Equation 2 can then beapplied to determine an average over one period of po-tential alternation.

T±i = α±

[|IBeam|

√|V |

]i

(1)

TiAvg = γT+i + (1− γ)T−

i (2)

Here, T+i and T−

i are the thrust output from ILISmodule i (A or B) in positive and negative polarities re-spectively. IBeam is the beam current, taken as the dif-ference between emitted current and that interceptedby the extractor grid, while V is the applied voltageand γ is the alternation duty cycle. The terms α+ andα− are aggregate coefficients accounting for the charge-to-mass ratio of emitted species, beam divergence and

7

V+

V-

IexA

IA

++ V

ExA

Signal Gen.

+

-

1 Hz

(a)

0

20

20Hz

0

20

50Hz

0 10 20 30 40 50 60 700

20

100Hz

Time (ms)

CommandEmitted

Cu

rre

nt (µ

A)

40

(b)

0

20

40

150Hz

0

20

250Hz

0 5 10 15 200

20

400Hz

Time (ms)

CommandEmitted

Cu

rre

nt (µ

A)

(c)

FIG. 6: Response of the current follower circuit to in-put sinusoidal waves at the indicated frequency emit-ting EMI-BF4.

energy losses. In Ref. [2] these were found to be α+ =

0.00201 NA−1V −1/2 and α− = 0.00197 NA−1V −1/2

when emitting EMI-BF4 from sources equivalent tothose used here.

The form of equation 1 differs from that previouslyderived and reported for droplet dominated colloidalelectrosprays[13] through the power of the current de-pendence. In colloidal devices, the charge-to-mass ratioand current are both strongly tied to an enforced massflow rate[28], leading to charge-to-mass ratio whichvaries with current to the one-half power and ultimatelya current to the three-halves dependence in expressionsfor thrust. However; in the presented devices, charge

to mass ratio has been shown to be relatively constantover large ranges of beam current[2], leading to propor-tional contribution due to current in the thrust expres-sion. Similarly, no clear trends in beam divergence orfractional energy loss over voltage have been observed

3. RESULTS

Current versus voltage measurements from the twinmodule device utilized in all charge neutralization re-sults are shown in Figure 7. The relationships aresimilar to previously presented data from this type ofemitter[2] and are characteristic of the emission fromdevices used for thrust measurements at ESTEC, pre-sented at the end of this section. To highlight themany 10’s of µA discrepancies between current magni-tudes without balance measures in place, the module Acurrent data have been inverted and plotted alongsidemodule B currents of the opposite polarity. Specifically,Figure 7(a) presents positive module B emission cur-rents and (inverted) negative module A emission cur-rent. Figure 7(b) presents negative module B emissioncurrent and (inverted) positive module A current. Af-ter a startup region between roughly 1500 V and 1750V the currents are seen to rise at rates of, roughly,0.6-0.7 µA/V . Hence the ± 300 V range of extractionvoltage adjustment provided by the closed loop balancecircuitry equates to up to ±200 µA of current control.Beam interception by the extractor grid was typicallya few to ∼ 10 µA, equivalent to up to ∼5 % of theemitted current.

Figure 8 presents unbalanced measurements of emit-ted current, the thruster common potential with re-spect to facility ground (TGND) and the summation ofall recorded currents. Here the thruster was configuredas in Figure 4, yet with the balance circuit disabledsuch that the module A extractor was tied to TGND

through a current monitor. Sample data is providedover a 2s period, yet the measurements are representa-tive of the steady state in this configuration. As evidentfrom Figs. 8a) and 8c), total emitted currents were gen-erally non-zero in this configuration. Specifically Fig.8c) provides the summation of emitted and interceptedcurrents from both modules, the filtered trace has beensmoothed with 10ms moving average. Referring to Fig.8b), when the total current was negative, the thrustercommon quickly charged positively until the +200Vzener-diode enforced limit was reached. Similarly, dur-ing regions of net positive emission, the common poten-tial charged negatively to the opposite limit. It is note-worthy that during the latter periods some couplingbetween modules may have been present as evidencedby the short period of relatively (but not completely)matched currents around 0s and 1 s on the plot. Theexact cause of this coupling was not deduced with cer-tainty.

8

1400 1500 1600 1700 1800 1900 2000 21000

20

40

60

80

100

120

140

160

180

200

IEmB

−IEmA

IExB

−IExA

Cu

rre

nt (μ

A)

Voltage (V)

(a)Positive emission current from module B and(inverted) negative current from module A.

1400 1500 1600 1700 1800 1900 2000 2100

−200

−180

−160

−140

−120

−100

−80

−60

−40

−20

0

IEmB

−IEmA

IExB

−IExA

Voltage (V)

Cu

rre

nt (μ

A)

(b)Negative emission current from module B and(inverted) positive current from module A.

FIG. 7: Current versus voltage measurements. Themodule A current has been inverted to indicate typicallevels of current magnitude mismatch during simulta-neous ± emissions.

3.1. Passive Charge Neutralization

The same device was subsequently arranged as inFigure 3 to evaluate automatic beam current balanc-ing techniques. First, in Figure 9, the thruster powersupplies and TGND reference were floated with respectto the extractors. Both extractors were then tied to fa-cility ground through their respective isolated currentmonitors. These data, presented in a format identicalto Fig. 8, confirm that currents from the two emitterscan be well balanced by this method as evidenced bythe overlayed module B and negative module A cur-rents in Fig. 9a). In Fig 9b) the thruster potential wasobserved to vary slightly, by roughly 20 V above or be-low the extractor potential (facility ground) in responseto the emitted currents. Referring to Fig. 9c), thesummation of current including extractor currents wasclose to zero considering the accumulated uncertainty

0 0.5 1 1.5 2

−100

0

100

Em

itte

d C

urr

en

t (μ

A)

0 0.5 1 1.5 2

−200

0

200

TG

ND P

ote

ntia

l (V

)

0 0.5 1 1.5 2−40

−20

0

20

40

Cu

rre

nt S

um

(μ

A)

Time (s)

Raw

Filtered

(IEmA

-IExA

)+(IEmB

-IExB

)

+200V Limit

-200V Limit

IEmA

IEmB

−IEmA

a)

b)

c)

FIG. 8: Current, thruster common potential and cur-rent summation measurements under conditions of un-balanced emission.

of the monitors (± 2 µA per channel) yet a non zerotrend remained; at roughly +4 µA and -5 µA duringpositive and negative emitter A current cycles respec-tively. When only the emitter current sum is considered(not shown) current summation magnitudes reduce toroughly +0.5µA and -3 µA during positive and negativeemitter A current cycles respectively; with propagateduncertainties of approximately ±3 µA.

In Figure 10, the configuration again took the form ofFigure 3 yet the common extractor electrode was left tofloat separately from the thruster power supplies. Thisconfiguration was intended to more accurately simulatea spacecraft environment through removing the pathto facility through in the extractors. As evident fromthe sample data, the configuration is no longer stable.Note that in Fig. 10c) the vertical axis scale has beenincreased significantly compared with Figs. 8 and 9. Inresponse to a small but non-zero current on the float-ing extractor electrodes the thruster common potentialwas observed to float up or down while maintaining aninitial balance of currents similar to Fig. 9. Once theenforced ± 200V limits were reached, and accordingly apath for current discharge from TGND was established,the current outputs were no longer balanced in reactionto the charged extractor. It should be noted that theextractor currents, included in the current summation,lose some meaning here as there is initially no clearpath to dissipate collected charge via the monitors.

9

0 0.5 1 1.5 2

0 0.5 1 1.5 2

0 0.5 1 1.5 2Time (s)

−100

0

100E

mitte

d C

urr

en

t (μ

A)

−200

0

200

TG

ND P

ote

ntia

l (V

)

−40

−20

0

20

40

Cu

rre

nt S

um

(μ

A)

Raw

Filtered

(IEmA

-IExA

)+(IEmB

-IExB

)

+200V Limit

-200V Limit

IEmA

IEmB

−IEmA

a)

b)

c)

FIG. 9: Passively balanced sources with a commongrounded extractor exhibited a low, but non-zero,thruster common potential and current summation.

0 0.5 1 1.5 2

0 0.5 1 1.5 2

0 0.5 1 1.5 2Time (s)

−100

0

100

Em

itte

d C

urr

en

t (μ

A)

−200

0

200

TG

ND P

ote

ntia

l (V

)

-100

-50

0

50

100

Cu

rre

nt S

um

(μ

A)

Raw

Filtered

(IEmA

-IExA

)+(IEmB

-IExB

)

+200V Limit

-200V Limit

IEmA

IEmB

−IEmA

a)

b)

c)

Extractor

charging

Balanced

FIG. 10: Passively balanced sources with a commonfloating extractor exhibited current balancing brieflywhile the thruster common potential rose towards ar-tificial ± 200V limits.

3.2. Closed Loop Charge Neutralization

The same device used to obtain the data presented inFigures 8 through 10 was configured for active chargebalancing as in Figure 4. Sample data are presentedin Figures 11 and 12. The thruster common poten-tial TGND is again indicated in Figs. 11b) and 12b) inaddition to the potential applied to the extractor viathe balancing circuit. Recall that the current balancingcircuit employed in this study only considered the emit-ted current, not that the extractor grids. The manualtuning capability enabled some compensation for thisomission through augmenting or decreasing the drivesignal to the module A extractor electrode. In practicefor a particular tuning setting, evidence of suppressedcharging was only achieved in either the positive or neg-ative emission mode from module B see Figs. 11 and12 respectively. For example, in Fig. 11 total currentsare largely nulled during positive beam emission frommodule B. In response, the TGND potential rises onlygradually before rapidly railing against the -200V limitupon the next polarity alternation cycle. In Fig. 12,the scenario is reversed in that the TGND potential risesgradually off the -200V limit during periods of negativeemission from module B. In these sample data currentsare more globally matched compared with Fig. 11;however high speed transients during switching, poorlyresolved by the 1 kHz data acquisition, may have con-tributed to constraining the floating potential near the-200V limit.

0 0.5 1 1.5 2

0 0.5 1 1.5 2

0 0.5 1 1.5 2Time (s)

Raw

Filtered

(IEmA

-IExA

)+(IEmB

-IExB

)

+200V Limit

-200V Limit

−100

0

100

Em

itte

d C

urr

en

t (μ

A)

−200

0

200

Vo

lta

ge

(V

)

−40

−20

0

20

40

Cu

rre

nt S

um

(μ

A)

TGND

VExA

IEmA

IEmB

−IEmA

a)

b)

c)

FIG. 11: Actively balanced currents tuned to suppresscharging during positive B module emission but poorlytuned during the opposite half period of pulsation.

10

0 0.5 1 1.5 2

0 0.5 1 1.5 2

0 0.5 1 1.5 2Time (s)

TGND

VExA

+200V Limit

-200V Limit

(IEmA

-IExA

)+(IEmB

-IExB

)

IEmA

IEmB

−IEmA

−100

100E

mitte

d C

urr

en

t (μ

A)

−200

200

−40

−20

20

40

Cu

rre

nt S

um

(μ

A)

a)

b)

c)

Vo

lta

ge

(V

)

FIG. 12: Actively balanced currents tuned to suppresscharging during negative B module emission but poorlytuned during the opposite half period of pulsation.

3.3. Commanded Bipolar Thrust

Thrust measurements were made during dual beamoperation and with the current balancing circuit active.These data were acquired at the ESA ESTEC facilitywith a device that was nominally identical to that usedto acquire the charging data presented in the previoussections. As in all previously presented measurements,polarities were alternated at 1 Hz.

The software based closed loop control routine ad-justed the power supply voltage settings in reaction tothe error between desired thrust and the sum of thatcalculated using equation 2 for each module. In prac-tice only the positive voltage was modulated throughthe proportional-integral controller while the negativesupply magnitude was set to be 1% lower. This valuewas established empirically as suitable to minimize thedifference in calculated thrust between periods of po-larity alternation and thereby assist in reducing thespread of data in the directly measured thrust.

Figure 13 provides representative calculated and di-rectly measured thrusts values over a sequence of com-manded thrust measurements from 15 to 35 µN in 5µN steps. Per-module and total thrusts were calcu-lated using equations 1 and 2 respectively. At peakthrust, of ∼35 µN , each module was emitting approxi-mately ±190 µA of current at ±2100 V ; correspondingto 0.8W of total input power. The zero level of directthrust measurements has been adjusted for each pulseto account for drift in the balance output. Specifically,the zero level has been established from the average ofdata acquired both between 45 s and 15 s before and

7 s and 45 s after each pulse. The measured thrusttrace has been broken between each pulse to reflectthis continuously adaptive zero reference. The calcu-lated thrust data have been smoothed using a movingaverage filter time constant equivalent to 1 s. Higherthrust levels were prohibited by practical current limitsobserved using OC100G optocoupler based polarity al-ternation circuitry. While recordings slightly below thecommanded thrust level are evidence of deficiencies inthe simplistic controller, good agreement between thecalculated and directly measured thrust levels was ob-served over the full range tested.

4. DISCUSSION

4.1. Passive Charge Neutralization

Referring to section 3.1, we have identified chargecollection on common extractor electrodes as a poten-tial hindrance to passively achieving automatic chargeneutralization on bipolar ILIS thrusters. Specifically, inFigure 9, balanced positive and negative currents wereemitted in a stable manner from two modules whentheir extractor electrodes were at a common potentialand tied to the facility ground. In that test, the floatingpotential of the source supplies deviated only slightlyas needed to maintain balanced emitted currents. How-ever; in Figure 10, when the tie to ground was severed,automatic current balancing was no longer stable andthe supply common potential quickly railed against theimposed ±200 V limits.

Referring to equation 3 and Fig. 3, this methodof charge balancing relies on rapid adjustments to thethruster ground potential to common extractor poten-tial difference, ∆V = TGND − VEx due to net chargeaccumulation. Specifically, if a net positive emissionoccurs ∆V will charge negatively thereby suppressingpositive emission and enhancing negative emission un-til an equilibrium between module A and B emittedcurrents (IA and IB) is quickly achieved. However; thecharge collected at the extractor electrode must also beconsidered. For this electrode to remain neutral equa-tion 4 must also be satisfied. Here fExA and fExB

are the fractions of emitted current intercepted by theextractor grid. These interception fractions are uncon-trolled characteristics of the particular thruster mod-ules and hence ∆V is the only free variable in general.Thus a simultaneous solution to both Eqns. 3 and 4at a single value of ∆V would be required in a stableconfiguration yet is generally not feasible. An excep-tion exists if the intercepted current fractions are equalfExA = fExB . However; such a requirement imposes asevere constraint on device fabrication and operationalconsistency over time. This constraint was not achievedhere. Hence the observed loss of control when the ex-tractor current was floated in section 3.1 is consistentwith an inability to simultaneously balance currents atthe thruster common and extractor. Some mitigationcould be imposed through adding capacitance between

11

0 50 100 150 200 250 300 350 400 450 500

0

5

10

15

20

25

30

35

40

45

Time (s)

Th

rust (µ

N)

Module A Module B Total

Measured

Mod. A

Mod. B

Total

Direct thrust

measurement

Calculated

FIG. 13: Comparison of calculated and directly measured thrusts after repeated measurements.

thruster common and the extractor; however this wouldtend to suppress balancing rate using this method andagain provide no explicit means to ensure overall chargeneutralization is achieved.

IA(VA −∆V ) + IB(VB −∆V ) = 0 (3)

IExA(VA −∆V ) + IExB(VB −∆V ) = 0 (4)

fExAIA(VA −∆V ) + fExBIB(VB −∆V ) = 0 (5)

Ultimately, charge neutralization necessitates nulli-fying the sum of Eqns. 3 and 4; however as emissioncurrents are only dictated by the local emitter to ex-tractor potential differences, no feedback mechanismwith respect to the local environment is implicit to en-force this constraint. If, for example, the extractor weretied to spacecraft chassis, this scenario could lead to arapid accumulation of charge that may only be miti-gated by either emitted particle return or environmentdependent ambient particle fluxes or photo-emissions.While elegant in its simplicity, our finding has beenthat this passive method does not implicity guaranteetotal currents from the spacecraft sum to zero.

This result is consistent with that reported by Mier-Hicks and Lozano[20]. There it was shown, theoret-ically and experimentally, that the return of, rela-tively, low energy particles in the emitted beam cansuppress charging. Particles with energy well belowthe total beam potential are known to be presentdue to ion or ion-cluster fragmentation downstream ofemission[29, 30]. While effective at maintaining chargeneutralization to within several tens of percent of thebeam voltage (positive or negative), those authors alsodemonstrated that acceleration of these heavy particlestowards the thruster can induce damage. A potentialrisk to both the thruster and other spacecraft subcom-ponents.

4.2. Active Charge Neutralization

Closed loop current control is potentially less attrac-tive in comparison to passive methods. However; thepresented results in section 3.2, lend support to its ap-plicability. In Figures 11 and 12 rapid potential swingsfrom the imposed -200 V limit were suppressed usingclosed loop control. However; manual tuning was re-quired to achieve the states presented and no tuningsetting could be achieved where the TGND referencewas steadily maintained near facility ground. Thus thesense and balance circuitry applied here did not com-pletely null total currents over full periods of potentialalternation. More sophisticated implementations of thecircuit should include extractor currents and, if appli-cable, downstream accelerator grid currents in the sum-mation to calculate and null total beam currents out ofthe spacecraft/thruster as accurately as possible.

The ∼100 Hz maximum operating frequency of thebalance circuitry may not be sufficiently fast to sup-pressms timescale high voltage charging in some space-craft environments, as discussed in the introduction.In Fig. 12, currents were largely matched save forpossible brief periods during each potential alternationevent. A higher bandwidth modulation circuit may bemore appropriate to suppress such events. To considerif faster modulation rates may be feasible, an addi-tional experiment was performed. Here the previouslydescribed current modulation circuit was modified toswap the Vishay LH1500AT SSRs for Voltage Multi-plier OC100G optocouplers. The extractor modulationoutput low pass filter cut off frequency was also in-creased to approximately 1 kHz. Another single linearstrip emitter configured as in Fig. 6(a) was used to eval-uate this modification. This source was wet with the IL1-ethyl-3-methylimadazolium thiocyanate (EMI-SCN).

Figure 14 presents AC modulation data with the

12

10 20 30 400

10

20

10 20 30 400

10

10 20 30 400

10

10 20 30 400

10

0 10 20 30 40 500

10

Cu

rre

nt (µ

A)

50Hz

100Hz

250Hz

350Hz

450Hz

Command

Emitted

Time (ms)

FIG. 14: AC drive tests with EMI-SCN, included toreinforce current timing limitations are, in part, due toelectronics.

modified circuit and single emitter source. The drivesignal was modulated between roughly 5 and 15 µA ofequivalent emission at the indicated frequencies. Cur-rent following was achieved at up to 450 Hz withoutloss of phase or attenuation, although an overshoot atthe modulation maxima of a few µA did persist at andabove 250Hz. Compared with the SSR based circuitry,Fig. 6, these data demonstrate a more than two foldincrease in supported frequency. If executed in a powerefficient manner, modulation at these speeds may en-able beam current matching at the ms time scales nec-essary to suppress fast charging events on orbit. Suchan implementation would also require accurate currentsensing at a similar bandwidth. Circuitry capacitancescould complicate this requirement through inducingcurrent signals which are not associated with net chargeloss, particularly during fast potential alternation tran-sitions. Means to slow these transitions without losingthe ability for rapid current balancing may be required.For example, transitions could comprise a slow modula-tion of the commanded beam voltages based on closedloop feedback of one current polarity while separatelya fast current balance circuit, similar to that used here,acts to null total currents through modulating the op-posing polarity’s extraction voltage.

A physical time scale may limit modulation speedand, in particular, could compromise current matchingduring potential alternations. Lozano and Martinez-

Sanchez[31] measured onset delays of a few to 10’s ofms on individual externally wetted ILIS. Those authorspostulated that onset times were likely dominated byviscous forces inherent in the specific emitter geome-try and wetting state, rather than charge relaxationfor example. Compared to the thin liquid meniscus ofexternally wetted emitters used in the reference, liquidin the the porous glass emitters used here was locatedin and channeled through multiple pore orifices nearthe emitter apex. Viscous effects may therefore havebeen suppressed, and the liquid may have been moreimmediately subjected to high electric fields (owningto the absence of a shielding effects due to a metal-lic emitter). Onset times would therefore have beenshorter; however, this hypothesis requires verification.Furthermore, in the devices presented here, 100’s ofemission sites are likely present over the emitter ar-rays each making a small contribution towards the to-tal beam current and characterized by an onset field(/voltage) dependent on local conditions. At a sin-gle polarity and emitting 100’s of µA, individual emis-sion site onset and collapse is likely to occur duringbeam current amplitude modulation yet at sufficientlydistributed potentials to avoid large, discreet, steps incurrent[32]. Regardless, minimum onset times may dif-fer between polarities (as observed in Ref. [31]) andemission sites; complicating current following duringpotential alternations. Recall in Figs. 11 and 12 thefloating ground potential was always driven to the im-posed limits upon polarity alternations. Thus in ad-dition to improvements in current following circuitry,further study of onset times directed at the specificelectro-hydrodynamic conditions of a particular emis-sion source should be pursued to completely assess thefeasibility and optimal implementation of charge neu-tralization in this manner.

The introduced method may never lead to exact cur-rent balancing necessitating some reliance on ambientconditions. For example a constant bias error on theorder of 1nA in the onboard current summation, equiv-alent to roughly five parts per million of the beam cur-rent in the present case, would lead a spacecraft with30 pF capacitance to charge at approximately 33 V/s.Thus although the µA level accuracy of monitors usedhere enabled total-current control at a level two ordersof magnitude lower than the beam currents, bias er-rors may have prevented long-term steady-state chargebalancing; even if extractor currents had been activelyaccounted for. However, active control lends itself torelatively simple customization to suit an environmentor mission. A trait which may suppress the degree towhich improved current measurement accuracies are re-quired. For example, in an electron-rich LEO environ-ment, the target level for current balance may be ad-justed to favour a slightly negative current bias at alltimes. Positive charging may then be suppressed bycollection of an ambient electron current much lowerthan that of the beam. This is in contrast to a passivesystem where the spacecraft charge may periodicallyswing to high negative levels not easily nulled by rel-atively immobile ambient ions. Conversely in GEO or

13

deep-space, a slight positive current balance may en-able photo-emission to null resultant negative charging.

It is noteworthy that while numerous electrospraythrusters similar to those shown here could be com-bined to yield multi-Watt high thrust systems, thetolerable level of net current uncertainty to preventcharging would remain unchanged. The required levelof relative current sensing resolution would thereforeincrease accordingly. At very high power levels ap-proaching kilowatts, consideration of dedicated neu-tralizers as in electrostatic plasma thrusters may be-come favourable to offset associated control complex-ity.

4.3. Compliance with Electrochemical ChargeBalancing

When operated as a simultaneous bipolar thruster,total currents at any instant must be matched to en-sure charge neutralization. Electrochemical balancingthrough potential alternation seeks to suppress reac-tions through, similarly, preventing a net charge trans-fer at the IL to conductor interface[9]. These two bal-ances must be considered in parallel. In doing so itis important to clarify that achieving electrochemicalneutralization at a propellant tank does not, inherentlyensure beam charge neutrality, and vice-versa. Con-sider the simple, two source, configuration presentedhere. Two tanks, and therefore IL to metal electricalcontacts, are required to achieve simultaneous bipo-lar emission. Suppression of electrochemical degra-dations (of the propellant or electrode) via potentialalternation[9, 10] is achieved by switching polarity at arate faster than the time required for the double layerpotential to exceed the electrochemical window of theIL-interface. This interfacial potential increases due tothe collection of charges of one polarity or another. Aspreviously identified[33, 34], the currents emitted dur-ing each half period of operation must match to ensurezero total charge transfer on average. If this conditionis not met, charge would gradually accrue over mul-tiple periods that after some time must be dissipatedthrough an electrochemical reaction or other discharge.Such dissipations may occur at a lower rate than if noalternation were employed; however, after the 1000’s ofhours of operation envisioned for ILIS thrusters, longterm propellant stability may necessitate suppressionof these integral imbalances.

Thus two constraints exist in a bipolar ILIS propul-sion system: i) total beam currents must instanta-neously sum to zero to neutralize charge and ii) theaverage current over potential alternation periods, fromeach propellant tank, should be nulled. Duty cyclemodulations have been employed to satisfy the laterrequirement for ILs with a propensity to emit highercurrents in one polarity compared with another[33].However; the practical benefits of such a solution arediminished in a bipolar configuration when satisfyingconstraint i). Thus it is apparent that, at least fora thruster pair, satisfaction of both constraints may

be best achieved if both sources emit the same mag-nitude of current at all times. This subtlety placesan additional emphasis on the benefits of high speed,closed loop, current control in bipolar ILIS. Specifi-cally, a relatively slow, current control loop on a singlethruster module be used to suppress electrochemicalcharge buildup over time, while a fast charge neutral-ization circuit effects accurate mirroring by one or moreother modules. Even with such an arrangement cur-rents intercepted by the extractor or other electrodeswould compromise the degree to which both of theseconstraints could be completely satisfied at all times.

4.4. Multiplexed Bipolar Thrust

The results of section 3.3, demonstrated that the to-tal thrust output from bipolar modules emitting op-posite polarity beams in close proximity was consis-tent with the sum total of thrust calculated from eachmodule individually. The thrust coefficients α+ andα− in equation 2 were generated using different single-polarity devices of the same fundamental design as eachmodule in the bipolar devices used here. To emphasizethis agreement, Figure 15, presents numerous thrustmeasurements and calculated thrust levels versus thecommanded thrust for two devices. Data in Figure15(a) were acquired using the same device utilized inmeasuring the data of Fig. 13 while data in Fig. 15(b)were acquired with another, again nominally identical,dual module device operating with EMI-BF4. Directlymeasured thrusts exhibited greater variability in thelatter case yet were often, again, within a few µN ofcalculated total thrust.

As downstream beam neutralization is not funda-mentally required for ILIS operation, as evidenced bytheir typical unipolar operation, the lack of a signif-icant detraction from calculated thrusts was not un-expected. Had the thruster electronics been floatedand allowed to charge to many hundreds of volts dur-ing thrust measurements, deviations in thrust outputwould have been expected due downstream accelera-tion or deceleration of the emitted beams in response tothe resultant non-zero electrostatic field at the thrusterexit[20]. Although no clear effects were observed here,more thorough investigations of beam neutralizationin bipolar ILIS are required to fully comprehend theimplications of beam interactions. For example, de-spite charge neutralization via current balancing, vir-tual anodes (or cathodes) formed due to poor beamneutralization could lead to accelerated return of lowenergy species and/or charge-exchange susceptible neu-trals born from fragmentation events. Future investi-gations could include, for example, emissive probes asused to measure colloid and FEEP thrusters in Refs.[35] and [36]. However numerical modelling as per-formed in Refs. [37, 38] would likely have significantcontributions to interpreting experimental data due tothe relatively unique nature of the plume with a negli-gible free electron population.

14

0 5 10 15 20 25 30 35 400

5

10

15

20

25

30

35

40

Commanded Thrust (µN)

Th

rust O

utp

ut (µ

N)

Commanded

Calculated

Measured

(a)Device 1, used to acquire data in Fig. 13.

0

5

10

15

20

25

30

35

40

Th

rust O

utp

ut (µ

N)

0 5 10 15 20 25 30 35 40Commanded Thrust (µN)

Commanded

Calculated

Measured

(b)Device 2.

FIG. 15: Compiled directly measured and calculatedthrusts from two devices plotted against commandedthrust.

5. CONCLUSIONS

Charge neutralization necessitates all currents to andfrom a spacecraft sum to zero. The degree to whichelectrostatic ion thruster beams must explicitly com-prise zero net current, and the time scales inherent, aredependent on the local space environment. A thrusterthat is benign to its environment should target chargeneutralization by design. Ion-ion charge neutralizationusing ionic electrospray sources has been discussed andconsidered experimentally. Two methods for effecting

charge neutralization were evaluated experimentally, anatural/passive means and a closed-loop current con-trol circuit.

With no explicit coupling between the emission cur-rent governing emitter-to-extractor potential differenceand the spacecraft to environment potential, the effec-tiveness of passive automatic current balancing via acommon extractor floated with respect to the thrustercommon was demonstrated to be limited. Specifically

under typical scenarios where extractor currents areneither identically balanced nor negligible, current col-lection on the extractor grids was shown to lead torapid charging of thruster common surfaces with re-spect to an isolated facility ground.

Alternatively, a simple circuit was used to match themagnitudes of positive and negative beam currents at∼10 ms time scales. This method was shown to beeffective at suppressing charge accumulation in somecircumstances. However; rapid charging occurred uponbeam polarity alternations and differences in extractorcurrent interception fractions between polarities neces-sitated a polarity-specific manual tuning. In futuredevelopments, the former may be partially mitigatedthrough utilizing faster circuitry; modulation up to 450Hz was shown to be feasible here. Similarly, a circuitwhich includes extractor currents in the to-be-nulledcurrent summation may suppress the latter complica-tion.

Two thruster modules, operated in close proximity,were shown to yield a sum total of thrust consistentwith previously reported thrust coefficients. Specifi-cally, thrust levels up to 35 µN were commanded, andmeasured directly, at up to 0.8 W of total input power.No detrimental or performance limiting effects due tobeam interactions or a lack of neutralization were ob-served. However; beam neutralization of ionic electro-spray beams requires further study.

Acknowledgments

This work has been supported through EuropeanSpace Agency Networking/Partnership Initiative con-tract #4000109063/13/NL/PA. The authors would liketo thank Dr. Shu T. Lai for his helpful responses to ourqueries.

References

[1] Lozano, P. and Martınez-Sanchez, M., “Ionic liq-uid ion sources: characterization of externally wet-ted emitters,” Journal of Colloid and InterfaceSciences, Vol. 282, No. 2, 2005, pp. 415.

[2] Courtney, D. G., Dandavino, S., and Shea, H.,

“Comparing Direct and Indirect Thrust Mea-surements from Passively Fed Ionic ElectrosprayThrusters,” Journal of Propulsion and Power ,Vol. 32, No. 2, 2015, pp. 392–407.

[3] Krejci, D., Mier-Hicks, F., Fucetola, C., Lozano,

15

P., Schouten, A. H., and Martel, F., “Design andCharacterization of a Scalable Ion ElectrosprayPropulsion System,” 34th International ElectricPropulsion Conference, Hyogo-Kobe, Japan, July2015.

[4] Gassend, B., Velasquez-Garcia, L. G., M., A. I. A.,and Martinez-Sanchez, “A microfabricated planarelectrospray array ionic liquid ion source with in-tegrated extractor,” Microelectromechanical Sys-tems, Journal of , Vol. 18, No. 3, 2009, pp. 679–694.

[5] Legge, R. and Lozano, P., “Performance of HeavyIonic Liquids with Porous Metal ElectropsrayEmitters,” 44th Joint Propulsion Conference &Exhibit, AIAA2008–5002 , Hartford, CT, July2008.

[6] Dandavino, S., Ataman, C., Ryan, C. N.,Chakraborty, S., Courtney, D. G., Stark, J. P. W.,and Shea, H., “Microfabricated electrospray emit-ter arrays with integrated extractor and acceler-ator electrodes for the propulsion of small space-craft,” Jounal of Micromechanics and Microengi-neering , Vol. 24, No. 7, 2014, pp. 075011.

[7] Perel, J., Mahoney, J. F., Moore, R. D., andYahiku, A. Y., “Research and Development of aCharged-Particle Bipolar Thruster,” AIAA Jour-nal , Vol. 7, No. 3, 1969, pp. 507–511.

[8] Lozano, P. and Martınez-Sanchez, M., “Stud-ies on the Ion-Droplet Mixed Regime in ColloidThrusters,” Ph.D Thesis, Massachusetts Instituteof Technology, Cambridge, MA, 2003.

[9] Lozano, P. and Martınez-Sanchez, M., “Ionic liq-uid ion sources: suppression of electrochemicalreactions using voltage alternation,” Journal ofColloid and Interface Sciences, Vol. 280, 2004,pp. 149–154.

[10] Brikner, N. and Lozano, P. C., “The role of up-stream distal electrodes in mitigating electrochem-ical degradation of ionic liquid ion sources,” Ap-plied Physics Letters, Vol. 101, No. 19, 2012,pp. 193504.

[11] Roussel, J.-F., Tondu, T., Mateo-Velez, J.-C.,Chesta, E., D’Escrivan, S., and Perraud, L., “Mod-eling of FEEP Plume Effects on MICROSCOPESpacecraft,” IEEE Transactions on Plasma Sci-ence, Vol. 36, No. 5, 2008, pp. 2378–2386.

[12] Zafran, S., Beynon, J. C., Kidd, P. W., Shel-ton, H., and Jackson, F. A., “One-mlb ColloidThruster System Development,” Journal of Space-craft , Vol. 10, No. 8, 1973, pp. 531–533.

[13] Ziemer, J. K., Randolph, T. M., Franklin, G. W.,Hruby, V., Spence, D., Demmons, N., Roy, T.,Ehrbar, E., Zwahlen, J., Martin, R., and Connolly,W., “Colloid micro-newton thrusters for the spacetechnology 7 mission,” IEEE Aerospace Confer-ence 2010 , Big Sky, MT, March 2010, also AERO–2010–5446760.

[14] Demmons, N., Hruby, V., LaMarre, N., Metivier,E., Ziemer, J., and Parker, M., “Electros-rpay Thruster Propellant Feed System forGravity Wave Observation Missions,” 52nd

AIAA/SAE/ASEE Joint Propulsion Conference,Propulsion and Energy Forum, Salt Lake City, UT,July 2016, also AIAA–2016–4739.

[15] Reissner, A., Buldrini, N., Seifert, B., Horbe, T.,Plesescu, F., and Scharlemann, C., “The IFMThruster - Introducing very high ∆v Capabilitiesfor Nanosats and Cubesats,” 52nd AIAA JointPropulsion Conference, Salt Lake City, UT, 2016,also AIAA-2016-5044.

[16] Lai, S. T., Spacecraft Charging , AIAA AmericanInstitute of Aeronautics and Astronautics, 2011.

[17] Garrett, H. B., “The Charging of Spacecraft Sur-faces,” Reviews of Geophysics and Space Physics,Vol. 19, No. 4, 1981, pp. 577–616.

[18] Garrett, H. B. and Whittlesey, A. C., “Space-craft Charging, An Update,” IEEE Transactionsof Plasma Science, Vol. 28, No. 6, 2000, pp. 2017–2028.

[19] Lai, S., “An Overview of Electron and Ion BeamEffects in Charging and Discharging of Space-craft,” IEEE Transactions on Nuclear Science,Vol. 36, No. 6, 1989, pp. 2027–2031.

[20] Mier-Hicks, F. and Lozano, P. C., “Spacecraft-Charging Characteristics Induced by the Op-eration of Electrospray Thrusters,” Journal ofPropulsion and Power , 2016.

[21] Courtney, D. G., Li, H. Q., and Lozano, P., “Emis-sion Measurements from Planar Arrays of PorousIonic Liquid Ion Sources,” IOP Journal of PhysicsD , Vol. 45, No. 48, 2012, pp. 485203.

[22] Hutchinson, I. H., “Ion Collection by a Spherein a Flowing Plasma: 3. Floating Potential andDrag Force,” Plasma Physics and Controlled Fu-sion, Vol. 47, No. 1, 2004.

[23] Hastings, D. E., “A Review of Plasma Interactionswith Spacecraft in Low Earth Orbit,” Journal ofGeophysical Research, Vol. 100, 1995, pp. 14457–14483.

[24] Frooninckx, T. B. and Sojka, J. J., “Solar CycleDependence of Spacecraft Charging in Low EarthOrbit,” Journal of Geophysical Research, Vol. 93,1992, pp. 2985–2996.

[25] Sasaki, S., Kawashima, N., Kuriki, K., Yanag-isawa, M., and Obayashi, T., “Vehicle chargingobserved in the SEPAC Spacelab-1 experiment,”Journal of Spacecraft and Rockets, Vol. 23, No. 2,1986, pp. 194–199.

[26] Goebel, D. M. and Katz, I., Fundamentals of Elec-tric Propulsion: Ion and Hall Thrusters, John Wi-ley and Sons, Inc., Hoboken, New Jersey, 2008.

[27] Courtney, D. G. and Shea, H., “Influencesof porous reservoir Laplace pressure on emis-sions from passively fed ionic liquid electrospraysources,” Applied Physics Letters, Vol. 107, No. 10,2015, pp. 103504.

[28] de la Mora, J. F., “The Fluid Dynamics of TaylorCones,” The Annual Review of Fluid Mechanics,Vol. 39, 2007, pp. 217–243.

[29] Courtney, D. G. and Shea, H., “Fragmentationin Time-of-Flight Spectrometry-Based Calcula-tions of Ionic Electrospray Thruster Performance,”

16

Journal of Propulsion and Power , Vol. 31, No. 5,2015, pp. 1500–1504.

[30] Lozano, P., “Energy Properties of an EMI-Imionic liquid ion source,” Journal of Physics D: Ap-plied Physics, Vol. 39, 2006, pp. 126–134.

[31] Lozano, P. and Martınez-Sanchez, M., “On thedynamic response of externally wetted ionic liq-uid ion sources,” Journal of Physics D: AppliedPhysics, Vol. 38, No. 14, 2005, pp. 2371.

[32] Guerra-Garcia, C., Krejci, D., and Lozano, P.,“Spatial uniformity of the current emitted by anarray of passively fed electrospray porous emit-ters,” Journal of Physics D: Applied Physics,Vol. 49, No. 11, 2016, pp. 115503.

[33] Coffman, C., Courtney, D. G., Hicks, F., Jamil, S.,Li, H., and Lozano, P., “Progress Toward a Vari-able Specific Impulse Electrospray Propulsion Sys-tem,” 47th Joint Propulsion Conference & Exhibit ,San Diego, CA, July 2011, AIAA–2011–5591.

[34] Castro, S. and de la Mora, J. F., “Effect of tipcurvature on ionic emission from Taylor conesof ionic liquids from externally wetted tungstentips,” Journal of Applied Physics, Vol. 105, 2009,

pp. 034903.[35] Marrese-Reading, C., Polk, J., Mueller, J.,

Owens, A., Tajmar, M., Spindt, C., and Fink,R., “In-FEEP Thruster Ion Beam Neutralizationwith Thermionic and Field Emission Cathodes,”27th International Electric Propulsion Conference,2001, also IEPC–01–290.

[36] Marrese-Reading, C., Ziemer, J. K., Gamero-Castano, M., Bame, D., Demmons, N., and Hruby,V., “Plasma Potential Measurments in the Plumeof a Colloid Micro-Newton Thruster,” 42nd AIAAJoint Propulsion Conference, 2006, also AIAA–2006–4042.

[37] Wheelock, A., Cooke, D. L., and Gatsonis, N. A.,“Electron-Ion Beam Coupling Through CollectiveInteractions,” 42nd AIAA Joint Propulsion Con-ference, Sacramento, CA, July 2006, AIAA-06-5024.

[38] Tajmar, M. and Wang, J., “Three-DimensionalNumerical Simulation of Field-Emission-Electric-Propulsion Neutralization,” Journal of Propulsionof Power , Vol. 16, No. 3, 2000, pp. 536–544.