A 10 nN resolution thrust-stand for micro-propulsion devices · The 10 nN resolution obtained from this thrust-stand matches that from state-of-the-art nano-Newton thrust-stands,

A 10 nN resolution thrust-stand for micro-propulsion devices

Subha Chakraborty Daniel G Courtney and Herbert Shea

EPFL

Copyright (2015) American Institute of Physics This article may be downloaded for personal use only Any other use requires prior permission of the author and the American Institute of Physics

The following article appeared in Review of Scientific Instruments (Vol86 Issue 11 p 115109) and may be found at httpdxdoiorg10106314935471orathttpscitationaiporgcontentaipjournalrsi861110106314935471

REVIEW OF SCIENTIFIC INSTRUMENTS 86 115109 (2015)

A 10 nN resolution thrust-stand for micro-propulsion devicesSubha Chakraborty Daniel G Courtney and Herbert Sheaa)

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

(Received 16 September 2015 accepted 26 October 2015 published online 16 November 2015)

We report on the development of a nano-Newton thrust-stand that can measure up to 100 microN thrustfrom different types of microthrusters with 10 nN resolution The compact thrust-stand measuresthe impingement force of the particles emitted from a microthruster onto a suspended plate of size45 mm times 45 mm and with a natural frequency over 50 Hz Using a homodyne (lock-in) readoutprovides strong immunity to facility vibrations which historically has been a major challenge fornano-Newton thrust-stands A cold-gas thruster generating up to 50 microN thrust in air was first used tovalidate the thrust-stand Better than 10 nN resolution and a minimum detectable thrust of 10 nNwere achieved Thrust from a miniature electrospray propulsion system generating up to 3 microNof thrust was measured with our thrust-stand in vacuum and the thrust was compared with thatcomputed from beam diagnostics obtaining agreement within 50 nN to 150 nN The 10 nN resolutionobtained from this thrust-stand matches that from state-of-the-art nano-Newton thrust-stands whichmeasure thrust directly from the thruster by mounting it on a moving arm (but whose naturalfrequency is well below 1 Hz) The thrust-stand is the first of its kind to demonstrate less than3 microN resolution by measuring the impingement force making it capable of measuring thrust fromdifferent types of microthrusters with the potential of easy upscaling for thrust measurement at muchhigher levels simply by replacing the force sensor with other force sensors C 2015 AIP PublishingLLC [httpdxdoiorg10106314935471]

I INTRODUCTION

Interest in microthrusters has grown significantly in thelast two decades for possible applications in small satellitesand for deep-space missions12 Several miniature thrustertechnologies such as cold-gas thrusters13 electrospray pro-pulsion using liquid metals4 or ionic liquids5ndash11 micro re-sisojets1213 and micro pulsed plasma thrusters13 are underinvestigation as prospective candidates targeting stable thrustsin the range of a few microN to several hundred microN Severalthrust-stands have been reported over the last few years formeasuring thrust from these microthrusters with sub-microN reso-lutions14ndash24 Most of these thrust-stands operate by mountingthe microthruster and a counter-mass on a pivoted torsion armand measuring deflection under applied thrust with a displace-ment sensor such as a capacitive sensor18 a laser displace-ment sensor (LDS)19 a linear voltage differential transformer(LVDT)20 or an optical interferrometer21 The natural frequen-cies of the reported thrust-stands are typically in the sub-Hz ranges and an additional damping mechanism such asviscous damping2022 damping coil17 electrostatic damping19

or Eddy current damping23 is used to reach steady deflectionof the arms more quickly and to minimize coupling to low-frequency facility vibrations which is a major source of noisefor thrust-stands with low natural frequencies Different typesof calibration sources are used for accuracy and resolutionestimation and several of the thrust-stands have reported sub-100 nN resolution161820ndash22 and as low as 10 nN1923 The main

a)Email herbertsheaepflch

challenges for such thrust-stands are facility vibrations drifteffects of electrical and fluidic connectors to the thrusters andthe difficulty in operating with microthrusters with differentthrust to mass ratios

Another class of thrust-stand operates by firing thethruster at a target plate and measuring the force on theplate25ndash31 The present work describes one such thrust-standThe method allows for easier adaptation to different typesof thrusters by completely separating the thruster fromthe thrust measuring unit thereby removing sensitivity tothe thruster mass and interconnects used for control orpropellant supply The best thrust resolution reported withthis technique is 3 microN using an optical sensor27 Howeverthe recorded force is not a direct measurement of thrustand measurements are complicated by unknown coefficientsof restitution amongst impacting particles Accordingly aprior knowledge of the relation between the impingementforce on the plate and the thrust on the thruster is requiredto determine thrust from the measured impingement forceThis requirement also complicates the applicability of direct(eg mechanical or electrostatic) force based calibrationsWu et al26 measured the impingement force from a nitrogencold-gas thruster at atmospheric and lower pressure and foundit to be within a few percent of computed thrust Grubišicand Gabriel27 reported a similar method and they obtained thesimilar relation between the computed thrust and measuredimpingement force from cold xenon and argon emissionLongmier et al29 measured thrust from a 5 kW xenon HallEffect thruster and the results agree within a few percentfrom that measured directly with an inverted pendulum thrust-stand The impingement force can be measured with different

0034-6748201586(11)1151099$3000 86 115109-1 copy 2015 AIP Publishing LLC This article is copyrighted as indicated in the article Reuse of AIP content is subject to the terms at httpscitationnewaiporgtermsconditions Downloaded to IP

128179254202 On Tue 17 Nov 2015 202951

115109-2 Chakraborty Courtney and Shea Rev Sci Instrum 86 115109 (2015)

techniques such as pressure tap26 optical displacementsensor27 or strain guage29 To date this technique has beenused for thrust measurement in the range of several mN oreven higher

In this article we report on an impingement type thrust-stand with a thrust range of 100 microN with better than 10 nNresolution up to around 30 microN total thrust and better than20 nN resolution above 30 microN The thrust-stand measuresthe impingement force on a conductive low-mass plate us-ing a commercially sourced micro-machined capacitive forcesensor and is suitable for measuring thrust from many differenttypes of microthrusters operating in the microN ranges Comparedwith existing implementations2627 the presented approach pri-marily differs in its use of a homodyne detection scheme Herethe impinging beam is interrupted by means dependent on theplume type at a defined frequency f ref which is greater thanlow-frequency facility vibrations yet below the natural fre-quency of the target plate Accordingly the natural frequencyof the device is designed to be gt50 Hz compared with thefew Hz or less typical of both existing impingement basedstands27 and direct-thrust torsional balances1923 By suppress-ing all external influences outside a narrow bandwidth nearthe excitation frequency this approach enables high resolutionand low noise measurements although frequency dependentthruster characterizations are inherently limited to slow vari-ations see Section III This dynamic detection method alsoobviates the need for extensive design and characterization ofa damping mechanism often necessary for a steady state lowthrust measurement by either (direct1923 or impingement27-based) approach

A description of the thrust-stand is provided in Section IIThrust measurement up to 50 microN with a noise-floor lowerthan 10 nN has been demonstrated in Section III from a sub-sonic cold-nitrogen source operating in atmospheric conditionThe total impingement force on the plate has been comparedwith a direct measurement of thrust on the capillary andthe results agree within 10 The noise-floor of the thrustmeasurements taken inside a vacuum chamber is lower than10 nN between 5 Hz le f ref le 15 Hz and thrust measurementsup to 3 microN have been demonstrated in Section IV foran electrospray device emitting high velocity charged ionsand droplets inside a vacuum chamber using an electro-static gate electrode to periodically (10 Hz) modulate thethrust

II DESCRIPTION OF THE THRUST-STAND

The thrust-stand operates by measuring the impingementforce of the jet emitted by the thruster on a sensing plateThis technique allows the sensing element to be very low-massand hence to have a natural frequency well above buildingvibrations To measure the force on the sensing plate we use aMEMS-based capacitive force probe against which the plateis pushed by the thruster beam A schematic of the thrust-standand its operating principle are shown in Figures 1(a) and 1(b)

The force sensor mounted on a translation stage is firstbrought in contact with the suspended plate producing a forceFOFF on the sensor The microthruster aligned on the centralaxis of the plate is then fired at the other side of the platewhich intercepts all the emitted particles This produces anadditional force F on the plate related to the thrust Th fromthe microthruster which is what we seek to determine It canbe shown that32

F =(1 +

kp

kf

)(Fmeas minus FOFF) (1)

where Fmeas is the force sensor reading kp is the spring constantof the beams of the suspended plate for displacement of theplate normal to its plane (z-direction) and kf is the springconstant of the force sensor The impingement force F can bedetermined by measuring the increase in force sensor output(Fmeas minus FOFF) when the microthruster is operated The thruston the microthruster Th is related to the impingement forceF by the generalized relation25

T h =1

(1 + K)F (2)

where 0 le K le 1 is an effective coefficient of restitution givenby the ratio of the linear momentum of the particles normal tothe plane of the plate after and before impinging the plate Kdepends on the type of impact of the emitted particles with theplate and is discussed in the context of sub-sonic cold-nitrogenand electrospray emission in Sections III and IV respectively

To decrease noise and to minimize the effect of possiblemechanical drift on force measurements a lock-in measure-ment is performed with the impingement force F appliedperiodically on the plate (either by pulsing the thruster or byinterrupting the impinging plume periodically) at a frequencyf ref in square-wave shape The force sensor output thus alter-nates periodically between FOFF and Fmeas The rms value

FIG 1 (a) Schematic of key elements of the thrust-stand The emitted beam from the microthruster is intercepted by a sensing plate which pushes against theforce sensor whose output is related to the thrust as described in the text (b) Operating stages of the thrust-stand The force sensor is first brought in contactwith the plate using the translation stage Then the thruster is turned on and the complete emitted beam from the microthruster is intercepted by the plate Thethrust is delivered in pulses to enable a homodyne measurement for lower noise The force sensor measures the impingement force on the plate

This article is copyrighted as indicated in the article Reuse of AIP content is subject to the terms at httpscitationnewaiporgtermsconditions Downloaded to IP

128179254202 On Tue 17 Nov 2015 202951


Rout of the first harmonic of the periodic force sensor outputvoltage is directly measured with a lock-in amplifier (ModelSR850 from SRS) operating at the same reference frequencyf ref and the impingement force F is obtained as

F =πradic

2times

(1 +

kp

kf

)SFRout (3)

where SF is the sensitivity of the force sensor ie force per unitchange of output voltage and the factor π

radic2 is the ratio of the

peak-to-peak value of a square wave signal to the rms valueof its first harmonic The impingement force can be measuredwith a resolution Fres and up to a range Fmax given by

Fres =

(1 +

kp

kf

)F∆ (4)

Fmax =

(1 +

kp

kf

)(Frange minus FOFF) (5)

where F∆ and Frange are the resolution and maximum measur-able force of the force sensor The force sensing element inthe thrust-stand is a model FT-S100 force sensor from Femto-Tools which can measure up to Frange = 100 microN force with aresolution F∆ = 5 nN having spring constant kf = 50 Nm andsensitivity typically SF = 50 microNV but individually calibratedfor each device To measure higher thrusts (10rsquos of mN) otherforce sensors such as FT-S1000 FT-S10000 or FT-S10000with higher ranges from the same manufacturer can simplyreplace the FT-S100 force sensor The force sensor is mountedon a vacuum compatible piezo-motor driven translation stage(Model AG-LS25-V6 from Newport 50 nm minimum incre-mental motion)

The sensing plate consists of a central square region tointercept the beam and four springs one at each corner Theplate is designed to meet the following criteria32

1 To ensure Fres le 10 nN one requires kp le kf Since kf= 50 Nm the suspended plate is designed with kple 50 Nm

2 The platersquos natural frequency fp is chosen to be between50 Hz and 100 Hz Building vibrations occur mostly atsub-Hz frequencies In a vacuum chamber the vibrationsfrom the pumps can be at several 100rsquos of Hz For ahomodyne measurement f ref should be chosen where theelectrical noise and facility vibrations are minimal Ourdesign targets a natural frequency far from both low-frequency and high frequency bands of facility vibrationf ref is then chosen to be a few Hz which is well belowfp and yet clear of the low and high frequency vibrationbands

3 The platersquos spring constant kp should be independent ofboth the location where the particle plume impinges onthe plate and of the diameter of the impinging beam Toensure this the particle intercepting region of the sus-pended plate shown in Figure 2(a) is designed to be atleast 100 times stiffer than the spring region of springconstant kp le 50 Nm such that the particle interceptingregion remains flat when the beams deflect normal to theplane of the plate under applied force with less than 1error associated with the spot-size and location on theparticle intercepting region

4 The size of the particle-intercepting region should be largeenough to capture the entire plume emitted over a wideemission angle For example the emission half-angle ofelectrospray devices can be as high as 40 (Ref 33) Toensure that the present design is compatible with suchsources the particle intercepting area is therefore chosento be 45 mm times 45 mm such that it can capture all emittedparticles from a 3 cm distance with an emission half-angleas high as 40

5 The particle capturing area has to be electrically conduc-tive for compatibility with charged beam sources (such aselectrospray microthrusters)

To reduce the mass of the sensing plate enough to keepfp ge 50 Hz the sensing region consists of an aluminum framecovered by aluminized mylar Aluminum is chosen for its lowYoungrsquos modulus E and relatively high Eρ for manufactur-ing the springs and plate frame where ρ is the density Forsimplicity of fabrication all beams and frame elements werelaser cut from 130 microm thick Al sheet To increase the secondmoment of inertia and hence the stiffness of the arms ofthe frame in the particle intercepting region the arms areplastically deformed out of plane of the plate using a punchin the form of a semi-circular indent about the axis of the armsas shown in Figure 2(b) The inner diameter of the indent is setto ten times the thickness of the frame so that it can be bent asdesired using a punch To cover the entire particle interceptingarea a 2 microm thick pre-stretched mylar membrane metallizedwith 20 nm of aluminum is attached on the frame using 25 micromthick ARclearreg 8154 adhesive as shown schematically inFigure 2(b) The metallization side of the membrane faces to-wards the microthruster (to collect a charged beam) In order toachieve electrical contact between the metallization side of themembrane and the aluminum frame a 2 mm diameter circularaperture in the membrane close to the center of the frame isfilled with EPOTEKreg H20S two-part conductive epoxy Thestructure is modeled using finite element methods within theCOMSOL multiphysics software package and the dimensionsmarked in Figures 2(a) and 2(b) have been optimized for thedesired range of spring constant and natural frequency Froma few possible sets of dimensions that can provide the desiredspring constant and natural frequency the chosen dimensionsare Ls = 10 mm Lt = 6 mm b = bt = bi = 05 mm thicknessh = 130 microm and di = 13 mm the simulated spring constantis kp = 502 Nm (slightly above kf) and natural frequencyfp = 818 Hz The stiffness of the particle intercepting region iscomputed to be nearly 124 times higher than that of the springregion

Figures 2(c) and 2(d) show an image of a manufacturedplate and a microscopic image of the semi-circular indent onone of the frame-arms respectively The natural frequenciesof the manufactured plates are measured by analyzing theirstep responses with a laser displacement sensor (Keyencemodel LK-H022) The spring constants are measured witha load cell (Futek model LRF-400) and a mechanical stageThe spring constant and natural frequency of plate used forthrust measurement from the cold-nitrogen ejecting capillaryare 482 Nm and 582 Hz respectively with less than 1uncertainty32


128179254202 On Tue 17 Nov 2015 202951


FIG 2 (a) Schematic of the particle intercepting plate The particle intercepting region is a frame covered with metallized mylar membrane using ARclear8154 adhesive for attachment (b) Cross section schematic of a frame arm showing the out-of-plane indent stamped before laser-cutting beams to increasestiffness (c) Image of a manufactured plate (d) Microscopic image of the semi-circular indent of the frames The thickness of the plate is h = 130 microm and thesemi-circular indent is intended to have an inner diameter 10h The supporting springs are not indented to maintain low spring stiffness

These metrics are approximately in agreement withrequirements 1 and 2 stated above a spring constant equal to50 Nm and a natural frequency between 50 and 100 Hz Whilethe spring constant agrees with that simulated to within 4the discrepancy in natural frequency is indicative of furtherdevelopment required to improve the simulation fidelity whenevolving the design

III ATMOSPHERIC CHARACTERIZATIONWITH A COLD-NITROGEN EMITTING CAPILLARY

The purpose of the atmospheric characterization setup isto experimentally verify the noise-floor resolution and mea-surement accuracy within a 50 microN thrust range The thrust-stand with a cold-nitrogen thruster is mounted inside a plasticbox on a vibration isolation table A silica capillary of innerdiameter 150 microm and length 18 mm at a distance 8 mm fromthe plate is used to emit a sub-sonic stream of nitrogen Acomplete schematic of the lock-in thrust measurement schemeis given in Figure 3(a) with a picture of the setup in Figure 3(b)

A three-way electrically controllable pressure switch (SMCmodel V100) is used to periodically pulse the inlet pressurebetween the ambient pressure P0 and Pin set by a pressurecontroller (Fluigent model MFCS-8C better than 1 mbar reso-lution) having a 1 mm diameter soft-tube outlet A transistor-transistor logic (TTL) signal at the reference frequency f refinternally generated from the lock-in amplifier (SRS modelSR850) is used to control the switch Since the hydraulicimpedance of the capillary is much larger than the hydraulicimpedance of the soft tube the Mach number at the inlet of thecapillary is Minlet ≪ 1 and the flow of the nitrogen gas throughthe capillary can be assumed nearly incompressible and vis-cosity dominated34 Under these conditions it can be shownthat the thrust on the capillary is proportional to (∆P)2 where∆P = (Pin minus P0) is the pressure difference across the capillaryExperimental evidence on thrust from sub-sonic cold-gas25ndash27

suggests that K asymp 0 Under these conditions the measuredimpingement force on the plate and thrust on the capillary canbe written as32

F = T h = α(∆P)2 (6) This article is copyrighted as indicated in the article Reuse of AIP content is subject to the terms at httpscitationnewaiporgtermsconditions Downloaded to IP

128179254202 On Tue 17 Nov 2015 202951


FIG 3 (a) Schematic of the atmospheric characterization setup of the thrust-stand using a nitrogen gas emitting capillary The pressure at the inlet of thecapillary is pulsed using a three-way pressure switch The shaded region is inside a plastic box to stop air currents (b) Image of the atmospheric characterizationsetup for the thrust-stand The entire setup is inside a plastic box and on a vibration isolation table

where α is a constant that depends on the inner dimensionsof the capillary and on the viscosity of cold nitrogen gasThe output of the force sensor is monitored with the lock-inamplifier operating at f ref and with time constant Tc = 3 s and24 dBoctave roll-off for its phase sensitive detector (PSD)The output of the force sensor is also observed directly with anoscilloscope (Lecroy wavesurfer 424) triggered at the risingedge of the TTL reference signal The thrust noise-floor atdifferent frequencies is determined by averaging the lock-inamplifier output with ∆P = 0 mbar for 100 s for f ref valuesbetween 1 Hz and 10 Hz One obtains a noise-floor less than10 nN between 3 Hz le f ref le 10 Hz (equivalent to less than60 nN

radicHz noise spectral density at Tc = 3 s with 24 dBoctave

roll-off) with vibration isolation enabled as well as disabledindicating that a minimum thrust of around 10 nN can be poten-tially detected with the thrust-stand There is a few nN offsetpartly contributed by the random facility vibrations in the pass-band of the PSD and partly due to coupling of vibrations ofthe pressure-switch operating at f ref The low noise levelsachieved through this technique and these lock-in settingscome at the expense of an ability to observe thruster stabilityover short periods At the configured 3 s time constant changesin the thruster output for example through fluctuations in thefeed pressure manifest in the output subjected to a settlingtime of sim12ndash15 s Variations in thruster output at higher rateswill be suppressed

For thrust measurement ∆P is increased in steps and ateach step the lock-in amplifier output is monitored for 100 s

In Figure 4(a) the pulsating force output traces recorded withthe oscilloscope and averaged over 32 traces are shown atf ref = 1 Hz and f ref = 4 Hz for the same pressure difference of400 mbar across the capillary High frequency fluctuations aresuperimposed on the otherwise square-wave pulse shapes ofthe force sensor output However the lock-in amplifier selectsthe first harmonic of the signal at f ref and its time constant andsharp roll-off decouple these high frequency fluctuations driftand sub-Hz facility vibrations in the output providing highsignal to noise ratio and hence reliable thrust measurement atlower thrust levels than possible using only the oscilloscope

The Fourier transform of the traces shown in Figure 4(a)at two different f ref are shown in Figure 4(b) The amplitudeof the first odd harmonic which is directly measured by thelock-in amplifier decreases as the reference frequency f ref isincreased from 1 Hz to 4 Hz This is due to finite slope at therising and falling edge of the square-wave force sensor outputsarising from the response time typically a few ms of theunder-damped plate This finite slope distorts the square-waveshape of the pulses more so for higher reference frequenciesleading to a reduction in the first harmonic amplitude at f refand evolution of even harmonics at 2 f ref which are absent ina perfect square wave signal Consequently the conversion ofthe lock-in amplifier output Rout to the peak-to-peak forceF using Eq (3) needs to be corrected to take into account thereduction of the amplitude of the first harmonic In Figure 4(c)a comparison is made between the average lock-in amplifieroutput and that obtained from the Fourier transform of the

FIG 4 Thrust measurements using a cold-gas thruster at 400 mbar (a) Oscilloscope traces of the raw force sensor output operating at fref = 1 Hz and 4 Hz(b) Fourier transform of the oscilloscope traces at fref = 1 Hz and 4 Hz showing the dependence of the first harmonic amplitude on fref (c) Comparison ofthe thrust measured from (i) the first harmonic of the oscilloscope trace (ii) the lock-in amplifier and (iii) the peak-to-peak amplitude measurement of theoscilloscope traces


128179254202 On Tue 17 Nov 2015 202951


oscilloscope traces as f ref is increased from 1 Hz to 10 Hz Boththese outputs are converted to the corresponding peak-to-peakvalue and normalized by the peak-to-peak force measuredfrom the oscilloscope traces Both outputs are smaller thanthe peak-to-peak signal and decrease with increasing f ref Thetypically 1ndash2 difference between the first harmonics fromoscilloscope readings and the lock-in amplifier reading is dueto the vertical gain inaccuracy of the oscilloscope At lower f refthe difference between actual peak-to-peak and that obtainedfrom the first harmonic measurement decreases however thenoise-floor increases above 10 nN at below 3 Hz Thereforef ref = 4 Hz is selected for thrust measurement which leads toa multiplication factor of 105 to be incorporated in Eq (3)

In Figure 5(a) average thrust output measured by the lock-in amplifier averaged over 100 s is plotted as a function of thepressure difference ∆P varied 0 mbar to 400 mbar in 10 mbarsteps (blue line labeled FT-S100) The standard deviation ofthe measured thrust at each pressure difference is shown aserror-bars and also in the inset The measured thrust follows aparabolic variation with pressure as expected from Eq (6) Thestandard deviation of the measured thrust is below 10 nN forup to around 30 microN thrust levels and typically between 10 nNand 20 nN above 30 microN thrust levels

To verify that the total thrust is indeed the thrust measuredon the sensing plate we compare the thrust measured with thethrust-stand with that measured directly on the capillary Todo this one and three identical capillaries are mounted on aload cell (Futek model LRF-400) and the thrust per capillaryis directly measured The load cell has a 100 mN force rangeand can only resolve forces to a few microN resolution The averageand standard deviation of three successive measurements ofthrust per capillary from one and three capillaries with the loadcell are also plotted on Figure 5(a) (red and black lines) Itcan be seen that the directly measured thrust on the capillarywith the load cell follows the thrust measured with the thrust-stand using the sensing plate very closely indicating validityof F asymp Th but with around 10 uncertainty due to relativelylarge standard deviation of the load cell measurements

In order to estimate the minimum resolvable thrust fromthe thrust-stand measurements are repeated from pressuredifference of 0 mbar to 40 mbar in steps of 1 mbar with all other

the settings identical to that described above In Figure 5(b)the average and standard deviation of 100 s of thrust measure-ment at each pressure difference are plotted A minimum thrustwell-below 10 nN can be detected and thrust differences lessthan 10 nN can be resolved in this sub-microN thrust range

Cold-gas thrust measurements were repeated at varieddistances between the capillary and the plate from 8 mm to24 mm The measured thrust changed by less than 1 despiteincrease of the spot-size by a factor of three on the plate This issimilar to the observations by Wu et al26 and indicates that theimpingement force measurement is independent of the spot-size of the particle plume on the plate A 02 uncertainty be-tween repeated measurements with the same plate and around2 uncertainty using different manufactured plates has beenobserved By laterally offsetting the capillary from the centralaxis of the plates by up to 10 mm a plate dependent 3ndash7maximum reduction in measured force is observed reflectingthe most significant source of uncertainty of the measuredimpingement force

IV THRUST MEASUREMENT FROMAN ELECTROSPRAY DEVICE

We operated our thrust-stand in a vacuum chamber tomeasure the thrust from an electrospray device emitting abeam composed principally of ions The ionic-liquid electro-spray device used for thrust measurement is a porous borosil-icate emitter strip array emitting from ionic liquid EMI-BF4and capable of up to 10rsquos of microA beam current details of whichcan be found elsewhere6 For electrospray devices indirectthrust measurement by time-of-flight mass spectrometry (ToF-MS) has been the most widely used method but it requiresa good knowledge of the beam composition6ndash9 Direct mea-surement of thrust has also been performed for some electro-spray propulsion devices6819 If an electrospray emitter emitsa current I of particles with uniform kinetic energy qV em thenormal thrust on the device can be written as6

T h = β0γ0I

2m0Vem

q (7)

FIG 5 (a) Measured thrust per capillary vs applied nitrogen pressure difference Blue data with error bars smaller than data points form the lock-in outputof the force sensor reading the force on the sensing plate (ie thrust-stand) Inset standard deviation of the thrust measurement with the thrust-stand Red andblack data points are direct thrust measurements taken with a load cell showing much higher noise but confirming agreement between direct thrust and thrustintercepted by sensing plate (b) Measured thrust from thrust-stand vs applied pressure difference up to 40 mbar to determine detectable thrust and resolutionThrusts values well below 10 nN can be resolved with the thrust-stand


128179254202 On Tue 17 Nov 2015 202951


FIG 6 (a) Schematic of the thrust-stand for thrust measurement from an electrospray device An electrostatic gate is used for pulse the thrust (the source emitscontinuously but the beam is periodically electrostatically deflected) The shaded region is inside the vacuum chamber (b) Image of the thrust-stand with thegate electrode assembly mounted on a flange ready to install inside the vacuum chamber

where m0 and q are mass and charge of a monomer emittedfrom the ionic liquid γ0 is a factor that takes into account theemission of different charge species together and is obtainedfrom ToF traces of the emission6 β0 is a factor that correctsfor the angular spread of the emission into the thrust normalto the axis of emission and is obtained from measurementof current density as a function of angle from the axis ofemission6 The values of γ0 and β0 are measured before thethrust measurements and are γ0 = (145 plusmn 003) and (163plusmn 007) and β0 = (093 plusmn 001) and (093 plusmn 001) in positiveand negative polarities of emission respectively32

The thrust-stand is mounted on a flange of a vacuumchamber and installed inside the vacuum chamber with thesensing plate placed 28 mm away from the emitter A sche-matic of the setup is shown in Figure 6(a) and a picture inFigure 6(b) For the homodyne detection we interrupt theimpinging beam using an electrostatic gate The gate consistsof a stainless steel three-electrode grid assembly the middlegrid periodically set to the gate voltage and the other twofixed at minus40 V to reduce secondary electron emission fromthe plate under charge bombardment This assembly is placedbetween the sensing plate and the electrospray emitter Eachelectrode has a 50 mm diameter aperture with stainless steel

grid of 81 optical transparency The emitted beam currentI is obtained by subtracting the extractor current Iex betweenthe extractor electrode and ground (ie beam intercepted bythe extractor electrode) from the total emitter current Iemsupplied by the high voltage emitter power supply The gatesignal pulsated at the reference frequency f ref between 0 Vand Vgate = 3000 V or minus3000 V depending on emission po-larity is generated using a high voltage pulse generator (DEImodel PVX-4140) controlled by a signal generator and thesame control signal is used as reference input for the lock-inamplifier The noise-floor of the thrust-stand output is typi-cally around 10 nN for f ref between 5 Hz and 15 Hz with Tc= 1 s and 24 dBoctave roll-off (equivalent to noise spectraldensity less than 40 nN

radicHz) Again controlled changes in

the thruster output will manifest slowly at a rise time greaterthan 4 s for these lock-in settings Thrust measurement hasbeen performed at f ref = 10 Hz The vibrations of the turbo-molecular pump and primary pump do not excite the plate atthis frequency

In Figures 7(a) and 7(b) the thrust measured by the thrust-stand and that obtained indirectly from Eq (7) by measuringI have been plotted for positive and negative polarities ofemission Here a multiplication factor of 1(081)3 = 188 is

FIG 7 Measured thrust from the porous borosilicate electrospray emitter (colloid thruster) Plot of thrust measured at fref = 10 Hz using the thrust-stand andindirectly computed thrust from beam diagnostics for (a) positive and (b) negative polarities of emission


128179254202 On Tue 17 Nov 2015 202951


included in Eq (2) to take into account the transparency of thethree-electrode assembly between the emitter and the plateThe error-bars on the measured thrust represent the standarddeviation of thrust measurements for 38 s and the error-bars on the indirectly calculated thrust using Eq (7) repre-sents the propagated standard deviations of γ0 β0 and I Theaverage thrust values are very close to each other typically50 nNndash150 nN less than that obtained from ToF measurementup to the maximum measured thrust This discrepancy couldbe partly due to uncertainty in the values of γ0 β0 and K usedto calculate thrust along with deficits in the kinetic energyof emitted charged particles from qV em Furthermore themeasured thrust being lower than that calculated is consistentwith some population of emitted particles becoming neutralprior to the electrostatic gate a phenomenon known to occurthrough fragmentation of solvated ions in flight35 The min-imum thrust measured with the thrust-stand is below 50 nNThe use of a sensing plate and the low noise levels and vibra-tion immunity it affords is thus validated by the agreement be-tween the thrust measurements using two different techniques

The thrust measurement apparatus lifetime is limitedusing electrospray thrusters as the energetic charged beamsputters the thin aluminum from the mylar membrane Thisindicates a non-zero value of K in Eq (4) Sputteringyield simulation36 suggest that for |Vem| between 1 kV and2 kV the value of K can be between 004 and 00732 inpositive and negative polarities of emission which has beenincorporated in calculating thrust Th from directly measuredF in Figures 7(a) and 7(b) however more work needs tobe done for accurate estimation of K for electrospray thrustmeasurement and to make a more robust plate less susceptibleto sputtering (for example using graphite layer insteadof aluminum metallization29) so that higher thrusts fromelectrospray devices and other electric propulsion devices canalso be measured for a long measurement duration withoutcompletely eroding the aluminum metallization layer

V CONCLUSION

In this article a nano-Newton thrust-stand has been re-ported that can measure thrust with 10 nN resolution bymeasuring the impingement force of the beam emitted fromthe microthruster using a plate to intercept the beam Byphysically isolating the thruster this thrust-stand alleviateschallenges associated with measuring low thrust-to-mass de-vices and reduces disturbances due to physical connectionsto the thruster The particle intercepting plate has a naturalfrequency over 50 Hz and a homodyne thrust measurementtechnique has been demonstrated to measure thrust from acold-gas ejecting capillary and from an electrospray emitterdemonstrating that thrust measurement is possible for differenttypes of microthrusters The reference frequency and timeconstant for measurement are chosen to ensure a noise-floorwell-below 10 nN making thrust measurement as low as 10 nNpossible Measurement from a cold-gas ejecting capillary at4 Hz has demonstrated that thrust up to 30 microN with a resolutionbetter than 10 nN and thrust above 30 microN with resolutionbetter than 20 nN can be measured The impingement forcemeasurement has been compared with direct measurement of

thrust on the capillary and the results agree to within theirrespective uncertainties Thrust measurement up to 3 microN hasbeen performed from an electrospray device with minimumdetectable thrust of less than 50 nN This compact thrust-standthus provides state-of-the-art resolution and noise-floor Themethodology could be scaled up to measure thrust in the 10rsquosof mN ranges simply by replacing the FT-S100 force sensorwith force sensors with higher range

Limitations and necessary improvements to the thrust-stand have been discussed Thrust accuracy is limited in partby a required correlation between thrust on the microthrusterand impingement force on the plate Furthermore althoughthe implemented homodyne detection scheme enabled a high-resolution and low noise measurement the slow settling timeof the lock-in amplifier at the chosen settings limits measure-ment of thrust variations to sub-Hz ranges The measurementspeed can be improved by reducing the time constant of thelock-in amplifier but at the expense of a higher noise-floorWhen applied specifically to ion thrusters two limitationsare identified First electrostatic gating is not effective onneutral particles Hence a reduced accuracy is anticipatedwhen measuring thruster plumes which include significantpopulations of high speed neutral particles Finally ablation ofthe sub-microm thick aluminum coating on the impingement platepresently limits the lifetime of the apparatus

ACKNOWLEDGMENTS

The authors are grateful to Dr Caglar Ataman forimportant discussions This work was partially supportedby the Swiss National Science Foundation under GrantNo 200021_146365 and European Space Agency Net-workPartnering Initiative No 400010906313NLPA

1J Mueller R Hofer and J Ziemer Survey of Propulsion TechnologiesApplicable to CubeSats (Jet Propulsion Laboratory Colorado USA 2010)

2D Spence E Ehrbar N Rosenblad N Demmons T Roy S HoffmanD Williams V Hruby and C Tocci ldquoElectrospray propulsion systems forsmall satellitesrdquo in Proceedings of the 27th Small Satellite ConferencePaper SSC13-VII-5 2013

3U Kvell M Puusepp F Kaminski J-E Past K Palmer T-A Gronlandand M Noorma Proc Est Acad Sci 63(2S) 279ndash285 (2014)

4F Ceccanti L Paita U Cesari M De Tata N Giusti P Balducci MDel Pistoia D Nicolini and L Di Napoli in 31st International ElectricPropulsion Conference Paper IEPC-2009-170 Ann Arbor MichiganUSA 20ndash24 September 2009

5S Dandavino C Ataman C N Ryan S Chakraborty D Courtney J P WStark and H Shea J Micromech Microeng 24 075011 (2014)

6D G Courtney S Dandavino and H Shea ldquoComparing Direct and Indi-rect Thrust Measurements from Passively Fed and Highly Ionic Electro-spray Thrustersrdquo J Propul Power (in press) preprint available at httpinfoscienceepflchrecord212755

7K W Stark and A Sherman ldquoResearch and development in needle andslit colloid thrustersrdquo NASA TN D-5305 Goddard Space Flight CentreWashington DC USA 1970

8R S Legge Jr and P C Lozano J Propul Power 27(2) 485ndash495 (2011)9G Lenguito and A Gomez J Appl Phys 114 154901 (2013)

10J K Ziemer T M Randolph G W Franklin V Hruby D Spence NDemmons T Roy E Ehrbar J Zwahlen R Martin and W Connolly inProceedings of the IEEE Aerospace Conference Big Sky MT USA 6-13March 2010

11L F Velaacutesquez-Garciacutea A I Akinwande and M MartiacutenezndashSaacutenchez JMicroelectromech Syst 15(5) 1272ndash1280 (2006)

12R H Lee A M Bauer M D Killingsworth T C Lilly J A Duncan andA D Ketsdever AIAA Paper 2007-5185 2007


128179254202 On Tue 17 Nov 2015 202951


13See httpwww busekcomcubesatprop__e mainhtm for informationabout different types of micro-propulsion devices

14S Rocca C Menon and D Nicolini Meas Sci Technol 17 711ndash718(2006)

15E Canuto and A Rolino ISA Trans 43 169ndash187 (2004)16J P Lake G Cavallaro G Spanjers P B Adkison and M J Dulligan

ldquoResonant operation of a micro-Newton thrust standrdquo AFRL-PR-ED-TP-2002ndash308 Air Force Research Laboratory Edwards AFB CA USA 2003

17J K Zeimer in Proceedings of the 27th International Electric PropulsionConference Paper IEPC-01ndash238 Pasadena CA USA 15-19 October2001

18D Packan J Jarrige P Thobois C Blanchard and P-Q Elias inProceedings of the 33rd International Electric Propulsion ConferencePaper IEPC-2013-418 Washington DC USA 6-10 October 2013

19M Gamero-Castantildeo Rev Sci Instrum 74(10) 4509ndash4514 (2003)20A J Jamison A D Ketsdever and E P Muntz Rev Sci Instrum 73(10)

3629ndash3637 (2002)21C Phipps J Luke T Lippert M Hauer and A Wokaun J Propul Power

20(6) 1000ndash1011 (2004)22Y X Yang L C Tu S Q Yang and J Luo Rev Sci Instrum 83 015105

(2012)23J Soni and S Roy Rev Sci Instrum 84 095103 (2013)24J P Luna C H Edwards J G del Amo and B Hughes in Proceedings

of the 32nd International Electric Propulsion Conference Paper IEPC-2011-011 Weisbaden Germany 11-15 September 2011

25S Hoffman and J J Janos ldquoForce due to air and helium jets impingingnormal to a flat plate for near-vacuum and sea-level ambient pressuresrdquoNASA TN D-7002 NASA Langley Research Centre Washington DCUSA 1971

26C K Wu H X Wang X Meng and W X Pan Acta Mech Sin 27(2)152ndash163 (2011)

27A N Grubišic and S B Gabriel Meas Sci Technol 21 105101 (2010)28B-x Du Y Zhao W Yao and X-q Chen J Meas Sci Instrum 4(2)

103ndash110 (2013)29B W Longmier B M Reid A D Gallimore F R Chang-Diaz J P Squire

T W Glover G Chavers and E A Bering J Propul Power 25(3) 746ndash752(2009)

30M D West C Charles and R W Boswell Rev Sci Instrum 80 053509(2009)

31Y Takao K Eriguchi and K Ono J Appl Phys 101 123307 (2007)32S Chakraborty ldquoAn electrostatic ion-guide and a high-resolution thrust-

stand for characterization of micro-propulsion devicesrdquo PhD dissertation(EPFL Switzerland 2015)

33S Chakraborty C Ataman S Dandavino and H Shea in Proceedings ofPowerMEMS Atlanta USA 2-5 December 2012 pp 528ndash531

34P Hill and C Peterson Mechanics and Thermodynamics of Propulsion 2nded (Addison-Wesley 1992) p 70

35P C Lozano ldquoEnergy properties of an EMI-Im ionic liquid ion sourcerdquo JPhys D Appl Phys 39(1) 126ndash134 (2006)

36See httpwwwsrimorg for information about SRIM simulator


128179254202 On Tue 17 Nov 2015 202951

REVIEW OF SCIENTIFIC INSTRUMENTS 86 115109 (2015)

A 10 nN resolution thrust-stand for micro-propulsion devicesSubha Chakraborty Daniel G Courtney and Herbert Sheaa)

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

(Received 16 September 2015 accepted 26 October 2015 published online 16 November 2015)

We report on the development of a nano-Newton thrust-stand that can measure up to 100 microN thrustfrom different types of microthrusters with 10 nN resolution The compact thrust-stand measuresthe impingement force of the particles emitted from a microthruster onto a suspended plate of size45 mm times 45 mm and with a natural frequency over 50 Hz Using a homodyne (lock-in) readoutprovides strong immunity to facility vibrations which historically has been a major challenge fornano-Newton thrust-stands A cold-gas thruster generating up to 50 microN thrust in air was first used tovalidate the thrust-stand Better than 10 nN resolution and a minimum detectable thrust of 10 nNwere achieved Thrust from a miniature electrospray propulsion system generating up to 3 microNof thrust was measured with our thrust-stand in vacuum and the thrust was compared with thatcomputed from beam diagnostics obtaining agreement within 50 nN to 150 nN The 10 nN resolutionobtained from this thrust-stand matches that from state-of-the-art nano-Newton thrust-stands whichmeasure thrust directly from the thruster by mounting it on a moving arm (but whose naturalfrequency is well below 1 Hz) The thrust-stand is the first of its kind to demonstrate less than3 microN resolution by measuring the impingement force making it capable of measuring thrust fromdifferent types of microthrusters with the potential of easy upscaling for thrust measurement at muchhigher levels simply by replacing the force sensor with other force sensors C 2015 AIP PublishingLLC [httpdxdoiorg10106314935471]

I INTRODUCTION

Interest in microthrusters has grown significantly in thelast two decades for possible applications in small satellitesand for deep-space missions12 Several miniature thrustertechnologies such as cold-gas thrusters13 electrospray pro-pulsion using liquid metals4 or ionic liquids5ndash11 micro re-sisojets1213 and micro pulsed plasma thrusters13 are underinvestigation as prospective candidates targeting stable thrustsin the range of a few microN to several hundred microN Severalthrust-stands have been reported over the last few years formeasuring thrust from these microthrusters with sub-microN reso-lutions14ndash24 Most of these thrust-stands operate by mountingthe microthruster and a counter-mass on a pivoted torsion armand measuring deflection under applied thrust with a displace-ment sensor such as a capacitive sensor18 a laser displace-ment sensor (LDS)19 a linear voltage differential transformer(LVDT)20 or an optical interferrometer21 The natural frequen-cies of the reported thrust-stands are typically in the sub-Hz ranges and an additional damping mechanism such asviscous damping2022 damping coil17 electrostatic damping19

or Eddy current damping23 is used to reach steady deflectionof the arms more quickly and to minimize coupling to low-frequency facility vibrations which is a major source of noisefor thrust-stands with low natural frequencies Different typesof calibration sources are used for accuracy and resolutionestimation and several of the thrust-stands have reported sub-100 nN resolution161820ndash22 and as low as 10 nN1923 The main

a)Email herbertsheaepflch

challenges for such thrust-stands are facility vibrations drifteffects of electrical and fluidic connectors to the thrusters andthe difficulty in operating with microthrusters with differentthrust to mass ratios

Another class of thrust-stand operates by firing thethruster at a target plate and measuring the force on theplate25ndash31 The present work describes one such thrust-standThe method allows for easier adaptation to different typesof thrusters by completely separating the thruster fromthe thrust measuring unit thereby removing sensitivity tothe thruster mass and interconnects used for control orpropellant supply The best thrust resolution reported withthis technique is 3 microN using an optical sensor27 Howeverthe recorded force is not a direct measurement of thrustand measurements are complicated by unknown coefficientsof restitution amongst impacting particles Accordingly aprior knowledge of the relation between the impingementforce on the plate and the thrust on the thruster is requiredto determine thrust from the measured impingement forceThis requirement also complicates the applicability of direct(eg mechanical or electrostatic) force based calibrationsWu et al26 measured the impingement force from a nitrogencold-gas thruster at atmospheric and lower pressure and foundit to be within a few percent of computed thrust Grubišicand Gabriel27 reported a similar method and they obtained thesimilar relation between the computed thrust and measuredimpingement force from cold xenon and argon emissionLongmier et al29 measured thrust from a 5 kW xenon HallEffect thruster and the results agree within a few percentfrom that measured directly with an inverted pendulum thrust-stand The impingement force can be measured with different

0034-6748201586(11)1151099$3000 86 115109-1 copy 2015 AIP Publishing LLC This article is copyrighted as indicated in the article Reuse of AIP content is subject to the terms at httpscitationnewaiporgtermsconditions Downloaded to IP

128179254202 On Tue 17 Nov 2015 202951








F =(1 +

kp

kf



T h =1

(1 + K)F (2)





128179254202 On Tue 17 Nov 2015 202951



F =πradic

2times

(1 +

kp

kf

)SFRout (3)




Fres =

(1 +

kp

kf

)F∆ (4)

Fmax =

(1 +

kp

kf












128179254202 On Tue 17 Nov 2015 202951









128179254202 On Tue 17 Nov 2015 202951











128179254202 On Tue 17 Nov 2015 202951










T h = β0γ0I

2m0Vem

q (7)



128179254202 On Tue 17 Nov 2015 202951











128179254202 On Tue 17 Nov 2015 202951




V CONCLUSION




ACKNOWLEDGMENTS














128179254202 On Tue 17 Nov 2015 202951


























128179254202 On Tue 17 Nov 2015 202951








F =(1 +

kp

kf



T h =1

(1 + K)F (2)





128179254202 On Tue 17 Nov 2015 202951



F =πradic

2times

(1 +

kp

kf

)SFRout (3)




Fres =

(1 +

kp

kf

)F∆ (4)

Fmax =

(1 +

kp

kf












128179254202 On Tue 17 Nov 2015 202951









128179254202 On Tue 17 Nov 2015 202951











128179254202 On Tue 17 Nov 2015 202951










T h = β0γ0I

2m0Vem

q (7)



128179254202 On Tue 17 Nov 2015 202951











128179254202 On Tue 17 Nov 2015 202951




V CONCLUSION




ACKNOWLEDGMENTS














128179254202 On Tue 17 Nov 2015 202951


























128179254202 On Tue 17 Nov 2015 202951



F =πradic

2times

(1 +

kp

kf

)SFRout (3)




Fres =

(1 +

kp

kf

)F∆ (4)

Fmax =

(1 +

kp

kf












128179254202 On Tue 17 Nov 2015 202951









128179254202 On Tue 17 Nov 2015 202951











128179254202 On Tue 17 Nov 2015 202951










T h = β0γ0I

2m0Vem

q (7)



128179254202 On Tue 17 Nov 2015 202951











128179254202 On Tue 17 Nov 2015 202951




V CONCLUSION




ACKNOWLEDGMENTS














128179254202 On Tue 17 Nov 2015 202951


























128179254202 On Tue 17 Nov 2015 202951









128179254202 On Tue 17 Nov 2015 202951











128179254202 On Tue 17 Nov 2015 202951










T h = β0γ0I

2m0Vem

q (7)



128179254202 On Tue 17 Nov 2015 202951











128179254202 On Tue 17 Nov 2015 202951




V CONCLUSION




ACKNOWLEDGMENTS














128179254202 On Tue 17 Nov 2015 202951


























128179254202 On Tue 17 Nov 2015 202951











128179254202 On Tue 17 Nov 2015 202951










T h = β0γ0I

2m0Vem

q (7)



128179254202 On Tue 17 Nov 2015 202951











128179254202 On Tue 17 Nov 2015 202951




V CONCLUSION




ACKNOWLEDGMENTS














128179254202 On Tue 17 Nov 2015 202951


























128179254202 On Tue 17 Nov 2015 202951










T h = β0γ0I

2m0Vem

q (7)



128179254202 On Tue 17 Nov 2015 202951











128179254202 On Tue 17 Nov 2015 202951




V CONCLUSION




ACKNOWLEDGMENTS














128179254202 On Tue 17 Nov 2015 202951


























128179254202 On Tue 17 Nov 2015 202951











128179254202 On Tue 17 Nov 2015 202951




V CONCLUSION




ACKNOWLEDGMENTS














128179254202 On Tue 17 Nov 2015 202951


























128179254202 On Tue 17 Nov 2015 202951




V CONCLUSION




ACKNOWLEDGMENTS














128179254202 On Tue 17 Nov 2015 202951


























128179254202 On Tue 17 Nov 2015 202951


























128179254202 On Tue 17 Nov 2015 202951

Documents

A 10 nN resolution thrust-stand for micro-propulsion devices · The 10 nN resolution obtained from this thrust-stand matches that from state-of-the-art nano-Newton thrust-stands,