Results of a 2000-Hour Wear Test of the NEXIS Ion Engine 

IEPC-2005-281   

 

Presented at the 29

th

 International Electric Propulsion Conference, Princeton University,  

October 31 – November 4, 2005 

 

John Steven Snyder,

*

 Dan M. Goebel,

†

 James E. Polk,

†

 Analyn C. Schneider,

*

 and Anita Sengupta

*

Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, 91109 

The Nuclear Electric Xenon Ion System (NEXIS) ion thruster was developed for 

potential outer planet robotic missions under NASA’s Prometheus program.  This engine 
was designed to operate at power levels ranging from 16 to over 20 kWe at specific impulses 
of 6000 to 7500 s for burn times of up to 10 years, satisfying the requirements of nuclear 
electric propulsion systems such as that on the proposed Prometheus 1 mission to explore the 
icy moons of Jupiter.  State-of-the-art performance and life assessment tools were used to 
design the thruster.  Following the successful performance validation of a Laboratory Model 
thruster, Development Model hardware was fabricated and subjected to vibration and wear 
testing.  The results of a 2000-hour wear test are reported herein.  Thruster performance 
achieved the target requirements and was steady for the duration of the test.  Ion optics 
performance was similarly stable.  Discharge loss increases of 6 eV/ion were observed in the 
first 500 hours of the test and were attributed to primary electron energy decreases due to 
cathode insert conditioning.  Relatively high recycle rates were observed and were identified 
to be high-voltage-to-ground arcs in the back of the thruster caused by wire insulation 
outgassing and electron penetration through the plasma screen.  Field emission of electrons 
between the accelerator and screen grids was observed and attributed to evolution of field 
emitter sites at accelerator grid aperture edges caused by ion bombardment.  Preliminary 
modeling and analysis indicates that the NEXIS engine can meet mission performance 
requirements over the required lifetime.  Finally, successful validation of the NEXIS design 
methodology, design tools, and technologies with the results of the wear test and companion 
performance and vibration tests presents significant applicability of the NEXIS development 
effort to missions of near-term as well as long-term interest for NASA. 

I.

 

Introduction 

nterest in science objectives at the outer solar system, specifically at the moons of Jupiter, has recently spurred the 
development of high-power electric propulsion systems.  Such missions require high-power, high-Isp thruster 

operation and long life that represent major increases over the capabilities of state-of-the-art ion engines.  For 
example, preliminary requirements for the proposed Jupiter Icy Moons Orbiter mission are for thrusters that operate 
at specific impulses of 6000-9000 sec and powers of 20-50 kW with throughputs greater than 2000 kg.

1

  As a part of 

the Nuclear Electric Xenon Ion System (NEXIS) project,

2

 a JPL-led team has developed a thruster designed to meet 

the life and performance goals and has demonstrated the required performance in laboratory tests.

3

    Two 

development-model NEXIS thrusters have been fabricated and have been demonstrated to meet the performance 
objectives

4

 and to survive launch loads.

9

I 

 

The NEXIS thruster was originally proposed in response to a recent NASA Research Announcement (NRA) 

which solicited proposals to identify and develop thruster technologies that enable nuclear electric propulsion 
missions to the outer planets.  Under this NRA, the JPL-led team proposed and was awarded funding to develop a 
thruster with a single-operating-point design at a nominal power of 22 kW, providing a specific impulse of 7500 sec 

 

The 29

th

 International Electric Propulsion Conference, Princeton University,  

October 31 – November 4, 2005 

 
 

1

                                                           

*

 Member of the Technical Staff, Advanced Propulsion Technology Group, Member AIAA. 

†

 Section Staff, Propulsion and Materials Engineering Section, Member AIAA. 

at 78% total efficiency and a throughput of 2000 kg.  Later, the NASA Prometheus program was established to 
develop nuclear power and electric propulsion for exploration.  The NEXIS project became part of Prometheus and, 
along with continued technology development, supported mission planning for the proposed Jupiter Icy Moons 
Orbiter mission.  Performance requirements for NEXIS were transitioned from the original NRA requirements to 
those for the Prometheus program. 

 

The NEXIS Laboratory Model (LM) thruster was designed to meet the NRA requirements as well as the 

additional objectives of operation at Isp’s in the range of 6500 to 8500 sec with high efficiencies at powers of 15 to 
25 kW.

3

  The LM thruster design is based on the heritage ring-cusp design successfully used in the NSTAR

5

 and 

XIPS

6

 ion thrusters, wherein a hollow-cathode discharge is produced in a cylindrical-conical discharge chamber 

with magnetic multipole confinement of charged particles, and the ion beam is extracted with an electrode set to 
produce thrust at high Isp.  The NEXIS design departs from NSTAR and XIPS with the use of a graphite keeper and 
carbon-carbon grids to provide the required life.  The discharge chamber and magnetic field circuit were designed 
with physics-based models, validated by test, to provide a high efficiency and flat beam profile.  The NEXIS LM 
thruster design met all of its performance goals for the NRA and the JIMO mission without re-design using 
experimental iteration.  It achieved over 78% efficiency at 7500 sec and 25 kW of power with a beam flatness 
parameter of 0.82, validating the design tools and methodology and providing useful performance data for mission 
planners.

3

 

Following the success of the LM thruster design and test, the design was transitioned to Development Model 

(DM) hardware, i.e. hardware that is designed to pass dynamic and thermal environmental testing while meeting all 
performance and lifetime criteria.  Since the LM hardware met all performance objectives, there were no changes to 
the electromagnetic or fluid designs.  In order to meet dynamic requirements, the flat carbon-carbon ion optics were 
replaced with a dished set of optics specifically designed for vibration tolerance.  Performance-based design criteria 
were handed off to industry, where the mechanical design of the NEXIS DM thruster was completed.  Two DM 
thrusters were fabricated; DM1a was used for performance and wear testing,

4

 and DM1b was used for vibration 

testing.

9

  The DM1b thruster was also subjected to post-vibration functional testing and demonstrated performance 

similar to or exceeding DM1a. 

 

The focus of this paper is the results of a 2000-hour wear test of the NEXIS DM11 ion engine.  The test had 

three main objectives:  (1) to establish test conditions that provide reasonable assurance of obtaining representative 
results for final flight conditions and hardware; (2) to characterize the operation of the thruster over the duration of 
the test, including any performance degradation; and (3) to characterize the wear of the engine and its components, 
including the wear rates of known wear modes and identification of unanticipated wear phenomena.  An ancillary 
goal is to demonstrate adequate thruster design maturity for the project to progress to the next development step 
toward flight hardware.   

An overview of the NEXIS project, including a discussion of the design and fabrication of the DM, is provided 

in Ref. 2.  Discussion of the carbon-carbon ion optics may be found in Ref. 7.  Structural design and analysis of the 
engine is discussed in Ref. 8.  Comparison of performance data from the DM1a engine to the models used in engine 
design are provided in Ref. 4.  Results from the vibration test 
of the DM1b engine are discussed in Ref. 9. 

 

The 29

th

 International Electric Propulsion Conference, Princeton University,  

October 31 – November 4, 2005 

 
 

2

 

II.

 

Test Setup 

A.

 

Test Article 

The NEXIS Development Model 1a (DM1a), shown in 

Fig. 1, served as the test article for the wear test.  In brief, the 
thruster consisted of a 65-cm-dia cylindrical-conical 
discharge chamber with six magnet rings forming the ring-
cusp magnetic field which closed the 60 G contour along the 
entire anode surface.  A plasma-spray coating of the same 
stainless-steel material as the chamber construction was 
applied to the interior of the chamber to provide for flake 
retention.  The physical and electromagnetic designs of the 
chamber were accomplished using state-of-the-art ion 
thruster design tools developed at JPL.  The discharge 

Figure 1.  NEXIS DM1a Ion Engine. 

chamber and propellant feed lines, including isolators, were welded as assemblies.  A conventional laboratory model 
cathode with a graphite keeper was used for the wear test.  The DM thrusters were fabricated using standard JPL 
flight hardware processes and documentation, yielding a complete set of Assembly Inspection Data Sheets (AIDS). 

The NEXIS carbon-carbon ion optics represent the culmination of development efforts led by JPL over the last 

several years [REF JPC05 NEXIS grids].  The optics were designed for engine performance and life based on JPL’s 
state-of-the-art 2D and 3D plasma simulation codes, and for structural robustness based on design tools developed 
and validated under the CBIO project [REF].  Grid thicknesses and dish depth were chosen as compromises between 
performance objectives and structural requirements.  The structural design of the optics was validated in the DM1b 
thruster vibration test [REF].  An intragrid electric field strength of 2.2 kV/mm was chosen for the optics design 
based on a detailed study of the voltage standoff capability and surface sensitivity to arc damage [REF Dan HV arc].  
The active beam diameter of the NEXIS ion engine is 57 cm, slightly smaller than the discharge chamber radius to 
provide a good plasma profile and small peak-to-edge ion current density to the grids.  Bipod flextures were chosen 
as the interface between the carbon-carbon ion optics and the metal discharge chamber to absorb the difference in 
thermal expansion of the two assemblies. 

The thruster ground screen was originally designed to be 

similar to the design used in NSTAR-class thrusters, where 
the majority of the screen is a fine mesh material. 

 

Preliminary performance testing before the wear test, 
however, identified the significant mesh area in back of the 
thruster as a penetration point for plasma electrons which 
could be accelerated to high energies and create plasma 
discharges within the rear ground-screen-to-thruster cavity.  
In order to prevent such discharges during the wear test the 
majority of the plasma screen was covered with solid metal 
masking. 

 

The 29

th

 International Electric Propulsion Conference, Princeton University,  

October 31 – November 4, 2005 

 
 

3

Finally, neutralizer requirements for the NEXIS ion 

engine were similar to that of the NEXT ion thruster 
developed by the NASA Glenn Research Center (GRC) 
[REF], so this design was baselined for the NEXIS thruster.  
A brazed NEXT neutralizer assembly was provided by GRC 
for the DM1a wear test. 

B.

 

Test Facility and Support Equipment 

NEXIS DM thruster testing was performed in a JPL 

facility converted from use for spacecraft thermal-vacuum 
testing into an ion thruster test facility.  The vacuum tank is 4 m in diameter and 12 m long, with its cylindrical axis 
oriented vertically and the ~2-3m of the tank extending above the roof of the enclosing building.  The walls and 
bottom end of the facility are lined with water-cooled shrouds for management of the thermal load caused by 
thruster operation at 20 kW.  Additionally, the chamber is lined with graphite panels to reduce metallic backsputter 
onto the thruster.  The shrouds and panel reduce the effective size of the chamber to 2.4-m dia. with the thruster 
located 7.6-m from the water-cooled, graphite chevron beam dump.  A photograph of the DM1a thruster in the test 
facility is shown in Fig. 2.  The facility has a total of ten cryopumps; six nude CVI pumps at the tank ceiling level 
above the thruster, two low-pumping-speed cryotubs mounted to wall originally used for spacecraft thermal vacuum 
testing, and four cryosails mounted at the bottom of the chamber behind the chevroned beam dump.  The total 
pumping speed on xenon is initially 250 kL/sec and falls to 200 kL/sec as the chamber reaches thermal equilibrium, 
which typically takes about 24 hours of thruster operation at 20 kW.  The facility backsputter rate was measured 
with a water-cooled quartz crystal micrograph mounted near the thruster to be 7.5 µm/khr during performance 
testing prior to initiation of the wear test. 

 

Power supplies for thruster operation are located on the upper level of the test facility building, close to the 

engine location.  A Diversified Technologies, Inc., 8-kW, 4.5-A supply is used for the beam supply.  This power 
supply was specially designed with a fast-acting switch that opens the load circuit within about 10 µs, thereby 
limiting the charge transfer in a high-voltage arc to 1 mC.  This charge transfer limit was derived from testing of the 
carbon-carbon optics materials and limits surface roughening which can cause reduced voltage standoff [REF Dan 
2005 HV Breakdown JPC].  The accelerator grid power supply is a standard Glassman 1.5-kV, 2-A supply which 
was factory-modified to reduce the stored energy.  The thruster discharge is run with a 55 V, 55 A Sorenson.  A 25-

 

Figure 2.  NEXIS DM1a Ion Engine in Wear 
Test Facility. 

 

The 29

th

 International Electric Propulsion Conference, Princeton University,  

October 31 – November 4, 2005 

 
 

4

V relayed supply was incorporated into the thruster circuit for performing screen grid ion transparency 
measurements.  NIST-traceable calibrations were performed for all voltage- and current-measuring shunt resistors. 

 

The engine recycle circuit was developed using a dedicated National Instruments PXI chassis and a Labview-

based Real Time software module.  The recycle hardware detects a screen undervoltage condition, produced when 
the beam supply fast-acting switch opens, and initiates a recycle.  A pulse counter is used to record the recycle rate.  
All recycle parameters are software programmable for flexible engine test conditions.  The data acquisition and 
control system (DACS) was derived from that successfully for the NSTAR Extended Life Test (ELT).  The 
Labview-based software controls all power supplies and monitors and records thruster and facility telemetry.  The 
engine is shut down in the event of any specified out-of-bounds condition (e.g. tank pressure or discharge voltage).  
Beam current and mass flow rates are auto-controlled by the DACS system to set values.  The PC-based system is 
located remotely from the vacuum tank, away from the power supplies, etc., for operator safety and comfort. 

 

Xenon propellant is supplied from a high-purity feed system copied from the NSTAR ELT test facility.  

Three UNIT UFM-1661 meters monitor flow rates with total uncertainties of 2.5% of the full-scale of 10 sccm for 
cathode flows and 100 sccm for the main flow.  All propellant line connections where line internal pressures are less 
than atmospheric are welded or under vacuum to preclude cathode contamination from fitting leaks.  The flow 
system has integral ports for calibration, pump-out, and nitrogen purges.  Flow rates were calibrated with a NIST-
traceable system before testing. 

C.

 

Test Plan 

The NEXIS wear test setup and plan was governed by the objectives of the test.  The first goal of the test was to 

establish test conditions that provide reasonable assurance of obtaining representative results for final flight 
conditions and hardware, i.e. those necessary for the Herakles thruster development [REF] in support of proposed 
Prometheus missions.  After test conditions are chosen, the performance of the thruster at those conditions must be 
characterized over the length of the test.  At the conclusion of the test, inspection and analysis are used to measure 
the wear rates of the known wear phenomena, identify unknown wear-related issues or wear phenomena, and 
identify thruster design issues related to wear, life, or performance degradation (if observed).  The final goal is to 
demonstrate adequate thruster design maturity for the project to proceed to the next development step toward flight 
hardware.  The NEXIS DM has already taken an important step in that path with the conclusion of vibration testing. 

Selection of the nominal operating point for the wear test was based on system-level requirements for the 

Herakles thruster [REF] dictated by proposed Prometheus missions.  Although the NEXIS thruster was originally 
designed for operation at a specific impulse (Isp) of 7500 sec, the Herakles nominal Isp of 7000 sec was selected for 
the wear test.   The Herakles discharge propellant utilization design point of 92% was also used, and the beam 
current was chosen to match the average beam current density of the Herakles design.  In addition to the nominal 
design point, several alternate operating points were selected for periodic performance measurements.  The 
Prometheus project requirements call for ±2% throttling in power and in thrust for peak-power tracking and 
trajectory variability, respectively.  Together with the nominal condition these requirements added eight additional 
operating points to the test plan.  The thruster was also characterized at the nominal NRA design point of 7500 sec 
and at a lower Isp of 6000 sec that was desirable from a Prometheus mission perspective for high-thrust operations. 

Performance testing of the engine and optics was performed at regularly scheduled intervals.  At the nominal 

wear test condition, discharge chamber performance curves (i.e. discharge losses vs. discharge propellant utilization 
efficiency) were obtained during the test at the nominal discharge voltage, and at the beginning and end of the test at 
off-nominal discharge voltages.  Optics performance data (i.e. perveance and electron backstreaming limits and 
screen transparency) and neutralizer plume mode margin were also investigated.  At the two off-nominal Isp 
conditions both engine and optics performance data were acquired; at the eight Herakles off-nominal points only 
performance data were taken.  

Ion optics performance was measured using standard procedures.  Perveance measurements were made by 

holding the beam current and accelerator grid voltage constant while varying the screen grid voltage and recording 
the accelerator grid current.  The discharge current was adjusted in this case to maintain constant beam current.  The 
perveance limit was defined as the point at which the rate-of-change of accelerator grid current was 0.02 mA/V.  
Electron backstreaming (EBS) onset was determined by reducing the magnitude of the accelerator grid voltage at 
constant discharge current and monitoring the beam current, a 1% change in which defined the EBS limit.  The 
screen grid transparency to ions was measured by biasing the screen grid negative of the cathode by twenty-five 
volts and recording the bias current.  The ratio of the screen power supply current to the total current (i.e. bias 
current plus screen supply current) yielded the transparency. 

A full suite of pre-test inspections and characterizations was performed in order to facilitate accomplishment of 

the test objectives.  The ion optics dimensions were measured and the optics assembly gap and alignment were 
measured.  Grid gap uniformity was better than 5% across the entire span of the 57-cm-dia. beam area.  The 
discharge cathode condition was documented, including cathode and keeper orifice plate diameters and 
profilometry.  Magnetic field mapping and plasma spray adhesion tests on the discharge chamber surfaces were 
performed.  The thruster was completely photodocumented at the component and assembly levels.  The total mass of 
the wear test article was 29.1 kg. 

III.

 

Results and Discussion 

The NEXIS DM1a thruster was operated for a total of 2020 hours during the wear test.  During this period it 

processed 47.3 kg of xenon and experienced neither significant changes nor adverse trends in performance.  The 
carbon backsputter rate from thruster-induced facility sputtering was 7 µm/khr measured near the thruster exit plane.  
The facility cryopumps were turned off three times during the test, once because frequent pressure spikes were 
observed which suggested xenon buildup on cryosail surfaces, and twice for repair of facility support equipment (i.e. 
liquid nitrogen valve and water manifold).  During these times the tank pressure reached ~ 1 Torr. 

A.

 

Thruster Run Time Parameters 

Controlled parameters for the nominal wear test condition are summarized in Table 1.  These values were 

typically regulated to within ±0.5%.  Active closed-loop control of the beam current was achieved through variation 
of the discharge current to achieve the target value.  Data for the dependent parameters and engine performance 
displayed in the following figures are shown as real data acquired by the DACS at 
twenty-minute intervals; there is no averaging or removal of any data points in the 
figures.  Data points which were acquired during recycle events or performance 
testing are evident in the figures.  Indicator lines have been drawn on the figures to 
designate performance test and cyropump regeneration times to facilitate 
interpretation. 

Table 1.  Wear Test 
Controlled Parameters. 

Vs 4760 

V 

Va -500 

V 

Jb 4.08 

A 

Jk 1.0 

A 

Jnk 3.0 

A 

mdotm 54.7 

mdotc 6.1 
mdotn 4.6 

 

Discharge current and voltage over the course of the wear test are shown in Fig. 3.  

Engine operating conditions were changed slightly during the first 50 hours of the test 
to achieve the desired operating conditions as the thruster and facility came into 
thermal equilibrium.  Overall, an increase in discharge voltage of approximately 0.3 V 
was observed during the test.  Discharge current varied to maintain constant beam 
current in the face of small variations in discharge voltage, but as will be shown later 
the discharge losses did not increase after the first 500 hours of the test.  Diurnal 
temperature variations in the facility induced minor fluctuations in uncontrolled 

 

The 29

th

 International Electric Propulsion Conference, Princeton University,  

October 31 – November 4, 2005 

 
 

5

34

32

30

28

26

24

D

is

c

ha

rg

e C

u

rr

e

n

t 

(A

)

2000

1500

1000

500

0

Time (hours)

30

28

26

24

22

20

D

is

c

ha
rge
 V
ol

tag
e 

(V

)

Voltage

Current

 

Figure 3.  Discharge Current and Voltage.  Dark bars represent cyropump regeneration events and grey 
bars represent when performance measurements occurred. 

parameters as is observed near the 1200 hour mark, for example.  Relatively frequent thruster restarts are visible in 
the discharge current data near the end of the test that resulted from electrical noise penetration into the DACS 
system which was corrected before the end of the test. 

Performance of the neutralizer, namely the keeper and coupling voltage, is shown in Fig. 4.  The coupling 

voltage was steady throughout the test at  -16 V, slightly higher than nominal for other extended tests because of 
larger distance between the thruster and neutralizer to reduce interaction of its plume with stray thruster magnetic 
fields.  Neutralizer keeper voltage decreased by nearly a volt over the course of the test.  A step change in the keeper 
voltage occurred at 1870 hours when the neutralizer flow rate was increased by 0.4 sccm (neutralizer keeper 
discharge noise was observed in the DACS system at this time, related to the previously mentioned thruster restarts). 

Total thruster power and tank pressure are shown in Fig. 5.  Several hours to the better part of a day were 

required for the thruster and chamber to come into thermal equilibrium and this is reflected in tank pressure 
variations after thruster restarts.  Pumping capacity of the cryosails in the bottom endbell was slightly reduced as the 
beam dump heated to operational temperature.  Diurnal pressure variations are also observed.  The nominal tank 

18

16

14

12

10

8

N

e

ut

ral

iz

e

r K

e

ep

er

 V

o

lta

ge

 (

V

)

2000

1500

1000

500

0

Time (hours)

-18

-16

-14

-12

-10

-8

C

ou
pl

ing
 V

o

ltag
e (
V

)

Keeper

Coupling

 

Figure 4.  Neutralizer Performance.  Dark bars represent cyropump regeneration events and grey bars 
represent when performance measurements occurred. 

9x10

-6

8

7

6

5

4

3

T

ank

 P

res

s

ur

e (

T

or

r 

X

e)

2000

1500

1000

500

0

Time (hours)

21.5x10

3

21.0

20.5

20.0

19.5

19.0

18.5

To
ta

l P

o

w

e

r (

W

)

Power

Pressure

Figure 5.  Tank Pressure and Total Power.  Dark bars represent cyropump regeneration events and 
grey bars represent when performance measurements occurred. 

 

The 29

th

 International Electric Propulsion Conference, Princeton University,  

October 31 – November 4, 2005 

 
 

6

pressure rose slightly during the test from 4.6

×

10

-6

  Torr  to  5.3

×

10

-6

 Torr, likely related to hotter ambient 

temperatures coupled with carbon deposition on cryosail surfaces.  Total thruster power was relatively stable at the 
20.4 kW level. 

Accelerator grid current as a function of time is shown in Fig. 6.  Increases in current observed at about the 500 

hour mark prompted a test for field emission which was observed as a current between the screen and accelerator 
grids of 200 µA at the nominal wear test voltages with the thruster plasma discharges off.  When plotted in the 
standard Fowler-Nordheim format it was verified that the current was field emission.  There was no effect of xenon 
flow rate or thruster temperature on the levels of the observed vacuum current between the grids.  Accelerator grid 
current of the operational thruster during the course of the test varied with time as seen in the figure.  Large step 
changes in current levels were observed at the same times as some of the performance tests, i.e. at 1046, 1261, and 
1605 hrs, as well as a very large change at 1953 hrs.  Field emission currents were measured with the thruster off at 
several times during the test and the values obtained at the nominal wear test voltages are plotted in Fig. 6 also.  The 
levels of vacuum field emission that were measured did not account for the difference in operational accelerator grid 
that were observed, e.g. at 1415 hrs the field emission current was 15 mA but the accelerator grid current was 
approximately 50 mA greater than the baseline value of 40 mA at the beginning of the wear test.  The reason for the 
difference is not presently understood.  At times when the accelerator grid current was near the baseline value there 
was no detectable field emission current at the nominal grid voltages (e.g. 1050 hrs). 

Additional understanding of the field emission behavior with time can be achieved when inspecting the current-

voltage curves as displayed in Fig. 7.  It can be seen that the threshold electric field for field emission is varying 
with time, likely the result of evolution of field emitter sites with time.  Surface morphology is altered by ion 
bombardment but also by localized heating and blowoff of the field emitter tips via emission current because of their 
relatively high resistance.  It is also known that the controlled charge transfer of 1 mC during recycle events 
removes field emitter sites by arc conditioning.  Thresholds of 1.8 to 2.1 kV/mm were observed when field emission 
current was present at the nominal 2.2 kV/mm.  Threshold were greater than 2.2 kV/mm, of course, at times when 
field emission current was not present at the nominal voltages.  These results indicate that if the NEXIS ion optics 
had been operated at more moderate electric field strengths then the field emission currents would not have been 
observed. 

120

100

80

60

40

20

0

Curre

nt (m

A)

2000

1500

1000

500

0

Time (hours)

 Accelerator Grid Current (Thruster Operating)
 Field Emission Current (Thruster Off)

 

Figure 6.  Accelerator Grid Current.  Dark bars represent cyropump regeneration events and grey bars 
represent when performance measurements occurred. 

Before the test had concluded and the post-test inspection performed, it was postulated that excessive charge 

transfer during recycle events had roughened the accelerator grid surface, thus creating field emitter sites [REF].  
Although this hypothesis was supported at the time with the test data available, post-test inspection of the 
accelerator grid revealed a complete lack of physical evidence for such events as will be discussed later.  The field 
emission currents observed in Figs. 6 and 7 were not caused by excessive charge transfer during recycle events. 

 

The 29

th

 International Electric Propulsion Conference, Princeton University,  

October 31 – November 4, 2005 

 
 

7

16

14

12

10

8

6

4

2

0

Field

 Em

is

s

ion

 Curre

nt (m

A)

3000

2500

2000

1500

1000

Intra-grid Electric Field (V/mm)

 556 hours
 1408 hours
 1414 hours
 1506 hours
 2020 hours

NEXIS
Operating Point

Observed Threshold
for Field Emission

No Field Emission or
Enhanced Breakdown
In This Range

Figure 7.  Field Emission Threshold. 

B.

 

Engine and Optics Performance 

Overall engine performance of the NEXIS engine is shown in Fig. 8.  Although diurnal variations are present as 

well as Isp and efficiency decreases associated with neutralizer flow increase near the end of the test, engine 
performance was stable during the test.  No significant changes nor long-term trends are apparent in the data.  This 
is expected because of the stability of the run-time parameters already discussed. 

Discharge losses, an important performance parameter and indicator of cathode health, are shown in Fig. 9.  

Losses were initially 177 eV/ion, increasing to 181 eV/ion over the first 200 hours, to 184 eV/ion by 500 hours, then 
were stable with slight variations from 183 to 185 eV/ion for the duration of the test.  Full discharge performance 
curves, shown in Fig. 10, confirm that the discharge performance was stable after 500 hours.  The discharge loss 
increases in the first 500 hours are directly related to internal discharge cathode plasma potential increases that have 
been measured in the NEXIS laboratory model discharge cathode [REF AIAA 2004-3430].  The internal plasma 
potential increases early in life as the cathode insert surface conditions and evolves, which reduces the primary 
energy of electrons exiting to the thruster discharge chamber.  Discharge modeling [REF JPC05 Herakles] shows 

7400

7200

7000

6800

6600

Specific

 Impulse (sec)

2000

1500

1000

500

0

Time (hours)

500

480

460

440

420

Thrus

t (m
N)

0.80

0.75

0.70

0.65

0.60

Total Efficie

ncy

Efficiency

Thrust

Isp

 

Figure 8.  Engine Overall Performance.  Dark bars represent cyropump regeneration events and grey bars 
represent when performance measurements occurred. 

 

The 29

th

 International Electric Propulsion Conference, Princeton University,  

October 31 – November 4, 2005 

 
 

8

that the measured 0.7-V cathode plasma potential increase fully accounts for the observed change in discharge 
performance shown in Figs. 9 and 10.   

200

190

180

170

160

D

ischarge L

osses (eV/ion)

2000

1500

1000

500

0

Time (hours)

 

Figure 9.  Discharge Losses.  Dark bars represent cyropump regeneration events and grey bars represent 
when performance measurements occurred. 

Optics performance measured at 6000, 7000, and 7500 sec, shown in Figs. 11-13, was largely unchanged over 

the course of the test.  No detectable changes in screen grid ion transparency (Fig. 11) were measured.  Some scatter 
is observed in the perveance limit data at the nominal 7000-sec condition which is related to changes in field-
emission currents during perveance testing.  There appears to be a slight decrease in the perveance limit at 6000 sec 
(Fig. 12), but this increase in perveance margin is similar to the data scatter in the 7000 sec data.  EBS limits (Fig. 
13) were also steady over the course of the test, although a laboratory circuit limited collection of EBS data in the 
first half of the test. 

Neutralizer performance, shown in Fig. 14, was stable near the flow rates used in the wear test and showed no 

change in plume mode margin.  Large increases in coupling voltage, however, were observed at low flow rates near 
the end of the test.  It is not clear what caused these changes, although they may be related to backsputtered carbon 
deposition on the neutralizer keeper which had degraded the keeper-to-ground electrical isolation by the end of the 
test. 

0.75

0.70

0.65

0.60

0.55

Scre

en G

rid Io

n Transparency

2000

1500

1000

500

0

Time (hours)

 Isp = 6000 sec
 Isp = 7000 sec
 Isp = 7500 sec

Pre-

Test

Figure 11.  Screen Grid Transparency. 

 

210

200

190

180

170

160

150

D

is

c

ha

rge

 Lo

s

se

s

 (

e

V

/io

n

)

0.95

0.90

0.85

0.80

0.75

Discharge Propellant Utilization Efficiency

 Pre-Test
 500 hours
 1000 hours
 1600 hours
 2000 hours

V

D

 = 26.5 V

 

 

Figure 10.  Discharge Chamber Performance Curves. 

 

The 29

th

 International Electric Propulsion Conference, Princeton University,  

October 31 – November 4, 2005 

 
 

9

-24

-22

-20

-18

-16

-14

C

oupling

 Voltage

 (V)

7

6

5

4

3

2

Neutralizer Flow Rate (sccm)

 0 hours
 1016 hours
 1600 hours
 2000 hours

Figure 14.  Neutralizer Performance. 

3800

3700

3600

3500

3400

3300

Impingement-Limited Tota

l Voltage

 (V)

2000

1500

1000

500

0

Time (hours)

 Isp = 6000 sec
 Isp = 7000 sec
 Isp = 7500 sec

Pre-

Test

Figure 12.  Perveance Limit. 

Engine performance at the 6000, 7000, and 7500 sec 

operating points is compared at various times throughout 
the test in Table 2.  Following the trend of stability 
throughout the test for the nominal wear test condition, 
variations of only a few to several tenths of a percent are 
present in the performance data for the off-nominal Isp test 
conditions.  As would be expected, data for the eight 
alternate Herakles throttling test conditions (i.e. ±2% in 
thrust and power), not shown here, were also very stable 
during the course of the test. 

C.

 

Post Test Inspection and Analysis 

Of primary interest during the post-test disassembly 

and inspection of the thruster was the identification of the 
source of field emission current observed during the test 
and also the relatively high recycle rate, shown in Fig. 15.  
Recycle rate trends varied throughout the test, initially 
having “bursts” of recycles interspersed between quiet 
periods, then an extended period with low recycle rates of 

-500

-450

-400

-350

-300

-250

-200

Electr

on Backstreaming

 Limit (V

)

2000

1500

1000

500

0

Time (hours)

 Isp = 6000 sec
 Isp = 7000 sec
 Isp = 7500 sec

Pre-

Test

Figure 13.  Electron Backstreaming. 

 

The 29

th

 International Electric Propulsion Conference, Princeton University,  

October 31 – November 4, 2005 

 
 

10

 

 

Table 2.  Comparison of Engine Performance Over Time for Three Main Test Conditions. 

Time 

(hours)

Total 

Power 

(kW)

Discharge 

Prop. Eff.

Discharge 

Losses 

(eV/ion)

Isp 

(sec)

Total 

Efficiency

Thrust 

(mN)

250

20.4

0.918

181.9

7023

0.757

446.0

550

20.4

0.920

183.0

7038

0.756

446.0

795

20.4

0.920

184.9

7030

0.758

446.3

1016

20.5

0.920

184.3

7036

0.756

446.3

1600

20.4

0.920

184.5

7021

0.755

446.0

2000

20.4

0.920

184.4

7031

0.758

446.5

250

16.2

0.919

199.4

6044

0.745

406.2

550

16.2

0.920

202.0

6034

0.742

406.6

1016

16.2

0.917

201.0

6035

0.745

406.3

1600

16.2

0.918

203.7

6037

0.744

407.3

2000

16.2

0.919

203.5

6043

0.745

407.1

250

22.8

0.919

172.0

7517

0.762

471.6

550

23.0

0.918

172.0

7535

0.756

472.0

1016

23.0

0.916

172.0

7531

0.759

472.9

1600

23.3

0.913

171.6

7502

0.750

473.4

2000

23.1

0.911

172.0

7504

0.754

475.6

No

minal Wear 

Test

 P

o

in

t

6000 sec

7500 sec

 

0-3 per hour, and finally sustained recycling at rates of 
~15/hr and occasionally higher.  The total number of 
recycles recorded during the test was 33,000.  Recall, 
before conclusion of the test it had been postulated 
[JPC05 NEXIS] that excessive charge transfer to the 
accelerator grid during recycle events caused surface 
roughening and formation of field emitter sites.  After 
removal and disassembly of the ion optics, the accelerator 
grid was thoroughly examined for physical evidence of 
arc damage.  Arcing of material coupons at charge 
transfers of 5 mC is readily visible, and at 10 mC difficult 
to miss with the unaided eye [REF Dan HV arcing].  
Inspection revealed that the upstream side of the 
accelerator grid was in pristine condition, however, as 
shown in Fig. 16.  There was a complete lack of physical 
evidence for arc damage or obvious locations for field 
emission on the accelerator grid surface.  Inspection of 
the downstream surface of the screen grid similarly had 
no physical evidence of arc damage, but did show 
localized discoloration on the edges of many apertures 

which is attributed to heating by field-emission electrons accelerated through the intra-grid electric field.   

Inspection of the thruster body after removal of the plasma screen revealed that the insulation on the wire 

harness had suffered significant decomposition.  This was most evident near the discharge cathode as seen in 
Fig. 17, where the outer Teflon sleeving has remained intact but the insulation surrounding the electrical conductor 
has been almost completely removed.  Wire insulation degradation including liquid decomposition products was 
observed up to 50 cm outside of the thruster plasma screen.  Thruster design and assembly documents specified 
Teflon-coated wire sheathed in Teflon sleeving for electrical harness, but it was evident than non-Teflon-insulated 
wire had been inadvertently used.  Analysis of intact insulation revealed that the insulation was PVC-based, known 
to decompose at temperatures well below the 240 

°

C measured during thruster performance testing.  Heavy 

discoloration on the thruster external anode surface was observed near the wire location, especially on the magnet 
cusp locations.  Corresponding discoloration   was observed on the nearby locations of the plasma screen, especially 
along the magnetic field cusp line.  None of these surfaces showed evidence of arcing or pitting. 

The physical evidence strongly supports the presence of plasma discharges, not simply high-voltage arcs, 

between the anode and plasma screen in the rear portion of the thruster.  The three elements required for electrical 

100

80

60

40

20

0

Arc R

ate (#

 of arcs in 1 hour)

2000

1500

1000

500

0

Time (hours)

 

Figure 15.  Engine Recycle Rate.  Rate is uncorrected for recycles occurring during performance testing 
and operator-induced diagnostic testing.  Dark bars represent cyropump regeneration events and grey bars 
represent when performance measurements occurred. 

Figure 16.  Post-Test Inspection Photograph of 
Upstream Surface of Accelerator Grid.  Note 
pristine, un-arced condition of surface. 

 

The 29

th

 International Electric Propulsion Conference, Princeton University,  

October 31 – November 4, 2005 

 
 

11

breakdown were present:  electrons from the ambient plasma 
which were able to penetrate the small mesh area in the plasma 
screen as observed before the wear test, a source of neutral gas 
(i.e. decomposing wire insulation), and high voltage.  Localization 
of the events to the magnetic cusp further indicates that they were 
plasma discharges.  The frequent breakdowns were observed in 
the telemetry as the fast-acting beam supply switch caused the 
high-voltage-to-ground discharge to be interrupted.  It is 
concluded from the inspection of the thruster and grids that the 
majority of the recycles observed during the wear test were caused 
by the wire insulation decomposition. 

Prior to the disassembly and physical inspection, the grid 

assembly was inspected for grid gap and alignment. The grid gap 
was determined by measuring the distance between the upstream 
surfaces of the screen and accelerator grids and subtracting the 
measured thickness of the screen grid.  Post-test gap 
measurements across the entire active grid area are shown in Fig. 
18.  Apart from a single outlier data point that was likely a 
measurement artifact, the uniform control of the gap within 5% 
was unchanged; the measured normalized grid gap was 1.02 ± 
0.03 before and after the wear test.  Likewise, no measureable 
changes in grid alignment were detected.  Laser 
profilometry of the accelerator grid surface has also 
been performed to characterize the grid erosion 
experienced during the test, and the final results will 
be used for further refinement and validation of grid 
life models [REF CEX3D] which have already shown 
excellent predictive capability [REF JPC05 CBIO]. 

 

The 29

th

 International Electric Propulsion Conference, Princeton University,  

October 31 – November 4, 2005 

 
 

12

No problems were observed in the discharge 

cathode assembly when it was disassembled and 
inspected.  The cathode heater heat shielding were 
intact and showed no damage and the cathode orifice 
plate showed the typical appearance of slight 
sputtering.  No tungsten crystal formations nor foreign 
material depositions were observed in the cathode 
insert.  The graphite keeper had loosely bound soot 
deposition on the outer diameter and in a thin ring 
surrounding the orifice as seen in Fig. 19.  Apart from 
the latter deposition the keeper orifice plate was clean 
suggesting uniform ion bombardment.  Laser profilometry of the plate 
indicated no detectable erosion (small machining grooves evident in the 
pre-test scans were still present in the post-test scans).  This is in contrast 
to the significant keeper erosion observed in the 8000-hour NSTAR wear 
test [REF].  Keeper life modeling [REF Katz JPC05] in addition to 
fundamental plasma measurements of the cathode plasma discharge [Dan 
and Tina IEPC05] and the careful selection of cathode operating 
parameters suggest that the NEXIS cathode design will have a keeper life 
in excess of 100,000 hours. 

D.

 

Discussion of Field Emission Behavior 

Post-test inspections of the thruster and ion optics have ruled out 

damage to the accelerator grid from excessive arc charge transfer as the 
cause of the observed electron field emission and frequent recycle rate.  
Although accelerator grid currents observed during recycle events were 
initially believed to be accel-grid-to-ground arcs [REF JPC05 NEXIS], it 

1.10

1.05

1.00

0.95

0.90

N

o

rmalize

d

 G

rid

 G

a

p

-30

-20

-10

0

10

20

30

Distance from Centerline (cm)

 0° - 180°
 60° - 240°
 120° - 300°

Design Gap

Figure 18.  Results of Post-Test Grid Gap Inspection.   

Figure 17.  Post-Test Inspection 
Photograph of Wire Harness Near 
Discharge Cathode Assembly.  Note 
decomposition of wire insulation inside the 
intact Teflon sleeving. 

 

Figure 19.  Post-Test Photograph of 
Discharge Cathode Assembly 
Keeper Orifice Plate.   

was concluded that the current to the accelerator grid was instead ions pulled from the discharge chamber after the 
beam supply fast-acting switch terminated the plasma discharge in the rear of the thruster. 

Electron field emission from the accelerator grid appears to have originated at the upstream edges of the 

accelerator grid apertures.  Numerous indications of localized heating by collection of accelerated electrons are 
evident on the upstream edges of screen grid apertures across the grid surface.  The sharp corners of the aperture 
edges enhance the local electric fields, promoting field emission from the negatively-biased accelerator grid and 
tending to focus collection of electrons on the positively-biased screen grid in those regions.  Thruster test data also 
support accelerator grid aperture edges as the location of field emission current.  The data in Figure 20, acquired 
during optics perveance testing at 1260 hours, show the effects of direct ion beam impingement on the nominal 
accelerator grid current.  Before perveance testing the accelerator grid current had been steady at 50 mA.  During 
perveance testing, as the screen voltage is decreased at constant beam current, direct impingement of accelerator 
grid surfaces occurred.  After the perveance test and all nominal test conditions had been to restored to their values 
prior to perveance testing (i.e. 2pm in the figure), the accelerator grid current was steady at the increased value of 65 
mA.  These same effects were observed at other times during the wear test.  This is a direct indication of the 
evolution of field emitter sites at the accelerator grid apertures caused by ion bombardment.  Such phenomena have 
also been demonstrated in testing of carbon-based gridlets [REF CSU IEPC]. 

The time dependence of the field emission current observed during the wear test (Fig. 6) is likely related to the 

competing effects of field emitter site evolution by ion bombardment and the documented “self-healing” of the 
surface by 1-mC arc charge transfer [REF DAN HV, CSU IEPC].  Optics performance testing, as well as the ion 
bombardment of the accelerator grid during the numerous recycle events cause by wire outgassing, evolved field 
emitter sites across the accelerator surface.  Field emission was mostly likely at the aperture edges where the electric 
field was enhanced by the sharp edges.  Field emitter sites were then removed by conditioning of the accelerator grid 
surface through 1-mC charge transfers.  Net field emission from the accelerator grid likely increased as the evolution 
of field emitter sites dominated over their removal by conditioning.  Net conditioning of field emitter surfaces has 
been demonstrated in gridlet testing [CSU IEPC] and was also observed at the 1950 hour mark of the wear test, 
where within a period of 2-3 hours there were very frequent conditioning arcs which caused the operational 
accelerator grid current to fall rapidly from a 95-mA peak to the beginning-of-test value of 40 mA.  During this time 
period the vacuum field emission was occasionally measured and it was observed to fall from 30 mA to undetectable 
at the nominal wear test voltages. 

Investigation of the voltage standoff of the carbon-carbon material used in the NEXIS grids was performed with 

small coupons in a ball-and-plate test where the threshold for field emission was arbitrarily defined as 1 µA [REF 
Dan HV arc] in a specific region of the sample.  It was determined from those tests that fields of 4 kV/mm could 
reliably be held off for non-apertured, undamaged materials.  Introduction of apertures and of surface modification 
by arcing was found to reduce the threshold 
voltage for field emission.  Evolution of field 
emitter sites over the much larger area of the 
NEXIS grids by ion bombardment could be 
expected to show larger currents at similar electric 
fields because of the greater number of field 
emitter sites.  Gridlet experiments have 
demonstrated that the apertured carbon-carbon 
can be conditioned to hold off fields of up to 
10 kV/mm with the application of several hundred 
conditioning arcs [REF CSU IEPC].  The physical 
area of the NEXIS ion optics is roughly fifty 
times larger than the gridlet area in those tests, 
suggesting some tens of thousands of 
conditioning arcs required to fully condition grids 
of the NEXIS size.  The majority of the 33,000 
recycles in the wear test occurred in the rear of 
the thruster instead of between the grids, and thus 
the grids were not able to be fully conditioned in 
the face of accelerator grid surface evolution by 
ion bombardment. 

80

70

60

50

40

Ac

c

e

l Gr

id C

u

rr

ent (

m

A)

2:00 PM

6/23/2005

2:30 PM

3:00 PM

3:30 PM

5000

4000

3000

2000

1000

Screen Voltage (V)

Screen Voltage

Accel Current

Figure 20.  Effect of Accelerator Grid Ion Beam 
Impingement on Field Emission Current.  Direct ion 
bombardment causes evolution of field emitter sites. 

 

The 29

th

 International Electric Propulsion Conference, Princeton University,  

October 31 – November 4, 2005 

 
 

13

The results of the NEXIS wear test, combined 

 

The 29

th

 International Electric Propulsion Conference, Princeton University,  

October 31 – November 4, 2005 

 
 

14

with the ball-and-plate materials tests and the gridlet tests, suggest that field emission behavior from carbon-based 
grids is complex interaction of the evolution of field emitter sites by ion bombardment, removal of those sites by 
conditioning of grid surfaces by controlled arc charge transfer, and that there may be an appreciable effect of grid 
surface area on the net behavior of grid surfaces.  These phenomena should be characterized and well understood for 
future implementation of larger-sized carbon-based grids. 

IV.

 

Conclusion 

The NEXIS Development Model thruster successfully completed a 2000-hour wear test intended to characterize 

thruster operation and wear phenomena at high-Isp, high-power operating points in support of the Prometheus 
project.  Engine operation was stable throughout the test and showed no measurable performance degradation after 
initial conditioning of the discharge cathode insert.  Discharge performance was steady after an initial 500-hour 
period in which cathode conditioning caused an increase of ~6 eV/ion.  Stability of discharge losses is essential for 
demonstrating the ability to meet challenging mission performance requirements for extremely long life times.  
Engine thrust, specific impulse, and efficiency reached the target values and remained steady.  There were no 
significant measurable changes in ion optics performance.  Thruster performance at off-nominal operating points 
followed this trend of stability for the duration of the test.   

Relatively high recycle rates were observed during the test and were concluded to result from plasma discharges 

in the rear of the thruster between the anode and plasma screen, supported by neutral gas produced by wire 
insulation decomposition.  The post-test condition of the accelerator grid surface was pristine, ruling out arc damage 
to that surface as the cause of observed field emission current between the screen and accelerator grids.  Test data, 
post-test inspection, and the results of fundamental materials experiments indicate that ion bombardment of the 
accelerator grid during recycle events and optics performance testing evolved field emitter sites with time, causing 
the time-variable field emission.  These effects were counteracted by self-cleaning of the grids through application 
of 1-mC conditioning arcs.  The remaining issues from the test are then elimination of high-voltage breakdowns in 
the back of the thruster and control of ion bombardment of the accelerator grid surface. 

The many successes of the NEXIS DM development, vibration test, and wear test are applicable to near-term 

NASA missions in addition to the targeted high-power, high-Isp Prometheus project goals.  The design methodology 
was fully validated by test; discharge chamber and ion optics models agreed very well with measured performance.  
Engine performance was excellent, meeting all target requirements for NEP missions to the outer planets with 
performance stability during the 2000-hour wear test.  Initial evaluations of component wear and performance 
suggest that the NEXIS DM thruster would perform to mission specifications for the required lifetime.  Structural 
design models produced hardware which demonstrated the ability to survive vibration testing.  Significant advances 
in grid manufacturing technology were achieved with a doubling of carbon-carbon grid size while improving grid 
gap and hole pattern control.  Finally, the NEXIS design approach very rapidly matured hardware with significant 
infusion of new technologies, resulting in a final configuration that with few modifications could readily proceed to 
Engineering Model hardware. 

Acknowledgments 

The authors wish to acknowledge the significant contributions of John Beatty, Ryan Downey, John Anderson, 

Ray Swindlehurst, and Allison Owens to the work contained herein.  We also thank Mike Patterson of the NASA 
Glenn Research Center for providing a NEXT neutralizer for the wear test.  The research described in this paper was 
carried out by the Jet Propulsion Laboratory, California Institute of Technology, under a contract with the National 
Aeronautics and Space Administration in support of Project Prometheus. 

References 

1.

 

S.R. Oleson, “Electric Propulsion Technology Development for the Jupiter Icy Moons Orbiter Project,” AIAA 2004-
5908, Space 2004 Conference and Exhibit, San Diego, CA, Sept. 2004. 

 
2.

 

T.M. Randolph and J.E.Polk, “An Overview of the Nuclear Electric Xenon Ion System (NEXIS) Activity,” AIAA 
2004-5909, Space 2004 Conference and Exhibit, San Diego, CA, Sept. 2004. 

 

3.

 

D.M. Goebel, J.E. Polk, and A. Sengupta, “Discharge Chamber Performance of the NEXIS Ion Thruster,” AIAA 
2004-3813, 40

th

 Joint Propulsion Conference, Fort Lauderdale, FL, July 2004. 

 

 

The 29

th

 International Electric Propulsion Conference, Princeton University,  

October 31 – November 4, 2005 

 
 

15

4.

 

J.E. Polk, D.M. Goebel, J.S. Snyder, A.C. Schneider, and A. Sengupta, , “NEXIS Ion Engine Performance and Wear 
Test Results,” AIAA-2005-4393, to be presented at the 41

st

 Joint Propulsion Conference, Tucson, AZ, July 2005. 

 

5.

 

J.R. Brophy,  “NASA’s Deep Space 1 Ion Engine,” Review of Scientific Instruments, Vol. 73, No. 2, Feb. 2002, pps. 
1071-1078. 

 

6.

 

J.R. Beattie, “XIPS Keeps Satellites on Track”, The Industrial Physicist, June 1998. 

 

7.

 

J.S. Beatty, J.S. Snyder, and W. Shih, “Manufacturing of 57cm Carbon-Carbon Composite Ion Optics for the NEXIS 
Ion Engine,” AIAA 2005-4411, to be presented at the 41

st

 Joint Propulsion Conference, Tucson, AZ, July 2005. 

 

8.

 

J. Monheiser, J.E. Polk, and T.M. Randolph, “Conceptual Design of the Nuclear Electric Xenon Ion System,” AIAA 
2004-3624, 40

th

 Joint Propulsion Conference, Fort Lauderdale, FL, July 2004. 

 

9.

 

NEXIS Vibe JPC05.