30 cm Ion Thruster Discharge Cathode Erosion *,

Transcription

1 3 cm Ion Thruster Discharge Cathode Erosion *, G. J. Williams, Jr., T. B. Smith, and A. D. Gallimore The Plasmadynamics and Electric Propulsion Laboratory The University of Michigan 1919 Green Rd., Rm. B17 Ann Arbor, MI 4815 (734) IEPC-1-36 Relative erosion rates and impingement ion production mechanisms have been identified for the discharge cathode of a 3 cm ion engine using laser-induced fluorescence (LIF). Mo and W erosion products as well as neutral and singly ionized xenon were interrogated. The erosion increased with both discharge current and voltage and spatially resolved measurements agreed with observed erosion patters. Ion velocity mapping identified backflowing ions near the regions of erosion with energies potentially sufficient to generate the level of observed erosion. Ion production regions downstream of the cathode were indicated and were suggested as possible sources of the erosion causing ions. Nomenclature A ij Einstein A coefficient (s -1 ) c Speed of light ( m/s) E I Kinetic energy of an ion (ev) E ij Transition energy (ev) e Electron charge ( C) h Plank s Constant ( Js) f Oscillator strength g i Degeneracy of state i g(ν) Gaussian line shape (s) G Gaunt factor I Intensity (W/m 2 ) I SAT Saturation Intensity (W/m 2 ) k Boltzman constant ( J/K) l(ν) Power broadened line shape (s) m Mass (kg) n Number density (m -3 ) R CD Collisional depopulation rate (m 3 s -1 ) T e Electron temperature (ev) V D Discharge voltage (V)) v Velocity (m/s) γ Homogeneous relaxation rate (Hz) λ Wavelength (m) ν Frequency (Hz) σ Cross-section (m 2 ) Introduction Ion thrusters are being scaled to different powers and operating conditions for space flight applications. A baseline for this scaling is the NASA Solar Electric Propulsion Technology Readiness (NSTAR) 3 cm ion thruster. Several wear-tests on NSTAR-like hardware have been conducted to demonstrate long duration operation and life-limiting phenomena. One of the potential failure mechanisms identified during these wear tests was erosion of the discharge cathode assembly. Severe erosion of the outer edge of the orifice plate, W at 145 µm/khr; of the tip of the cathode tube, Mo at 28 µm/khr; and of the heater coil outer sheath, Ta, were observed in the 2 hr development wear-test. 1 A keeper electrode was introduced as an engineering solution, and, after a subsequent 1 hr wear-test 2, and an 82 hr life demonstration test (LDT), 3 the cathode erosion was largely eliminated while the keeper orifice plate (Mo) eroded in a distributed pattern with a maximum rate of roughly 63 µm/khr near the mid-radius point. 3 It has been suggested that higher powers and longer periods of operation can be accommodated by thickening the orifice plates. 4 However, * Presented as Paper IEPC-1-36 at the International Electric Propulsion conference, Pasadena, CA, October, 21. Copyright 21 by George J. Williams, Jr. Published by the Electric Rocket Propulsion Society with permission.

2 recent developments in cathode assembly design may not make this a feasible option. 5,6 The source of the high-energy ions causing the erosion has remained largely unknown. Ray-tracing after the 2 hr test indicated the source of the ions was located between 1 and 11 mm downstream of the orifice plate. The cathode to anode potential was between 25 and 27 V for the above tests, and this voltage is slightly below the predicted sputtering threshold of Mo by Xe II (singly charged) ions. 7 A potential-hill near the exit of the cathodes, 8 sheath effects, charge-exchange collisions, 9 and plasma oscillations might yield high-energy Xe II ions required for the erosion. 1,11 Because Xe III (doubly charged) ions may not require these anomalous phenomena to explain the source of the erosion, they are often assumed to be the primary cause of the erosion. 12 However, recent investigations have shown back-flowing Xe II ions near the face of the discharge cathode with energies sufficient to erode the cathode or keeper. 1,11,13, This paper primarily discusses laser-induced fluorescence (LIF) erosion measurements and Xe II velocimetry near the discharge cathode assembly (DCA) exit plane. A physical mechanism through which Xe II ions might lead to the erosion is then suggested. The contribution of Xe III ions is discussed in more detail in parallel investigations. 12,14 Results of LIF measurements of Xe II velocity distributions and W and Mo relative concentrations near the exit of the discharge cathode are used to support the discussions. 1,11,13 These investigations were conducted at the Plasmadynamics and Electric Propulsion Laboratory (PEPL) at the University of Michigan. Apparatus and Procedure Thruster A functional model thruster (FMT) provided by NASA GRC was used in this investigation. The FMT s were the immediate predecessor to the NSTAR engineering model thrusters (EMT s) which preceded the NSTAR flight thrusters. The FMT s differed principally from the EMT s in their soft aluminum construction. The magnetic field, discharge cathode assembly (DCA), and geometry of the discharge chamber were identical to EMT-1 s. The quartz windows replaced roughly twenty percent of the anode surface as shown in Fig. 1 and have had a negligible impact on discharge chamber and thruster performance. 11 The thruster has been operated over the entire NSTAR power throttling range at both GRC and at PEPL using a SKIT-Pac provided by NASA GRC. Overall performance of the thruster was comparable to the engineering model thrusters Measurements were made with and without the presence of a keeper electrode. Thruster operating conditions are summarized in Table 1 and correspondence to points in the NSTAR throttling matrix is noted. 15 Detailed dimensions of the components of the DCA are not provided as per an agreement with NASA GRC. Vacuum Facility This investigation was performed in the 6 m x 9 m large vacuum test facility (LVTF) at PEPL. Four CVI Model TM-12 Re-Entrant Cryopumps provided a combined pumping speed of 14, l/s on xenon with a base pressure of less than Torr. The back pressure during 2.3 kw operation was roughly Torr, corrected for xenon. Xenon flow was controlled to the thruster using a dedicated propellant feed system provided by NASA GRC. The flow rates were periodically calibrated using a bubble flow meter. No significant variation was observed. The FMT was mounted on a two-axis positioning system with roughly.25 cm resolution for both stages. A 2 m by 2.5 m louvered graphite panel-beam dump protected windows downstream of the thruster and suppressed back sputtering from the chamber. The panels were located 4 m downstream of the thruster. Laser and Optics An argon-ion pumped dye laser (Coherent model model) was used in single and in multiple-beam techniques. 16 Four-beam LIF (shown schematically in Fig. 2) measured three velocity components simultaneously; a second axial beam was used in the fourbeam technique to increase the resolution of the axial velocity. Angles associated with the measurement of various components are given in Table 2. Both single and 4-beam techniques used a Hamamatsu reference cell to provide a zero velocity datum. The laser was typically scanned over a.1 nm interval in.61 pm increments. The beams were delivered to the thruster in a manner identical to that used in previous Hall and ion thruster 2

3 interrogations. 13,16 Alignment was facilitated by a wire crosshair on the side of the FMT plasma screen. Wavelengths associated with interrogation of the various species are given in Table 3. Data were collected using Spex 5 M and Spex H1 monochromators fitted with Hamamatsu 928 PMT's. Both fluorescence signals were recorded via computer. The Spex 5 M monochromator s slits were set to 5 µm. Because of a 2x magnification, the effective spot size at the focus of the collection lens was 25 µm. The bi-conical sample volume in the cathode plume was roughly.5 cm long and.1 cm in diameter at its ends. Measurements taken far downstream of the cathode at low power indicate that the ambient plasma outside of the cathode plume contributes less than five percent to the natural fluorescence signal. The laser delivery and signal collection optics are shown schematically in Fig. 2. Several rapid (~ 1 s) scans were taken at each data point. The rapid scans prevented long lock-in time constants from artificially shifting the fluorescence spectra to lower frequencies which would falsely indicate lower velocities. These scans were then Chauvenet filtered and averaged. 17 Theory Laser-Induced Fluorescence The absorbing neutral xenon, Xe I, or singly ionized xenon, Xe II, will see the wavelength of the incoming laser photons shifted by the relative motion of the particle in the direction of the photon. v ν = ν i c (1) The bulk velocity components from the various laser beams are extracted by a straight forward geometrical regression. 16 The energy distributions were calculated directly from the velocity distributions: E i [V] = mv 2 i 2e (2) Lineshape Models A detailed lineshape model was used to determine the temperature and Doppler shifts of the fluorescence signals. 16 Only Gaussian broadening was considered since the magnetic fields, Stark broadening, and natural line widths are negligibly small. However, because detailed spectral data was unavailable for the Xe I, Mo, and W transitions, a simpler model was employed. 18 Since the monochromator s acceptance line width was orders of magnitude greater than the laser line width (.1 nm versus.1 nm), instrument broadening was also neglected. Neutral density filters were periodically placed in the laser beam path to check for saturation. There was no indication of saturation during 4-beam operation. However, during 1-beam interrogation of Mo, the fluorescence FWHM varied significantly. Saturation intensity can be approximated by 19 I SAT = hc 2σλ A ji, (3) i i < j where the cross-section was approximated from tabulated spectral data: 2 σ = A 21 λ 2 8π g(ν o ). The power-broadened fluorescence lineshape takes the form 19 l(ν) = γ 2 ( ν ν o ) 2 + ( γ 2) 2 1+ I ISAT The homogeneous relaxation rate is a combination of natural and collisional relaxation rates. These are roughly of the same order of magnitude in this investigation: A ji γ/2. The power-broadened lineshape was convolved with a Gaussian lineshape, g(ν) = c ν (4) m 1 2 (ν - ν ) 2 exp 4ln(2) 2πkT 2 (5) ν D to fit the Mo data yielding a more accurate estimate of the temperature. Erosion Measurements The relative populations of the sputtered Mo and W atoms ground states and the lower states of the pump transitions associated with the LIF may vary with 3

4 operating condition. Due to the low plasma density near the discharge cathode, assuming local thermodynamic equilibrium (LTE) does not appear justified. One alternative to the LTE model is the steady-state corona equilibrium (CE) model. 21 Appropriate for low-density plasmas, this model assumes that radiative processes dominate the excited state relaxation, while the excitation is dominated by collisions. Because the number densities predicted just downstream of the DCA are on the order of 1 12 cm -3, this model can be used to relate the density of the excited state to that of the ground state. 21 For coronal equilibrium, the rate of collisional excitation and radiative relaxation balance: n o n e R oj = n j A oj. Solving for the ratio of the excited state to the ground state yields n o = g n o n e f oj G 1 j g j E oj T 2 exp E oj e T e 1 A oj The ratio of the populations of an excited state for two different temperatures is strictly a function of T e : n i (T e2 ) n i (T e1 ) = T 1 6 e 2 exp E ij + E ij (6) kt e2 T e1 kt e1 Unfortunately, each of these models, LTE and CE, is a function of T e which cannot be measured via LIF. In situ measurement of T e near the DCA by electrostatic probes such as Langmuir probes was beyond the scope of this investigation. However, T e can be approximated using an empirical fit to 3 cm ion thruster data 22,23 as a function of discharge voltage, V D, 24 T e = 1.9 ln V D. (7) 2. Equation 7 predicts T e to vary between 2. and 5.3 ev for discharge voltages between 24 and 32 V. No prediction in the variation in T e across the plume of the DCA was attempted without probe data. Equations 6 and 7 were used to correct the LIF signal obtained at different discharge voltages for variations in the population of the excited state which might otherwise be interpreted as variations in the overall population of sputtered material. Results Typical Xe I, Xe II, Mo and W LIF spectra are given in a previous publications. 1,13,16 Reference cell LIF data were used to assure proper laser tuning and to provide zero velocity references for the velocimetry. Erosion Un-keepered Configuration Figure 3 shows Mo and W LIF signals as a function of radial position as indicated by a picture of the exit plane of the un-keepered DCA. Note that the signals peak at roughly the same radial position (.3 cm) and indicate erosion peaks near the transition from the cathode orifice plate (W) to the cathode tube (Mo). The resolution of the LIF technique (roughly.5 cm) did not allow a more detailed mapping. Mo data were only taken over half of the diameter. The locations of peak Mo and W LIF signals are in very good agreement with the observed regions of erosion. Subsequent Mo data were taken at the.3 cm location both to insure a signal at other operating conditions and because the Mo surface subject to impinging ions is limited to this region. W data were taken at roughly the same point which corresponds to the region of greatest W erosion in the 2 hr wear test. Figure 4 shows the Mo LIF signal as a function of discharge current for V D = 25 and 27 V. Note that there is a noticeable increase in signal with current. There is a non-negligible signal at J D = 6. A. Data taken at V D = 27 V show that the Mo LIF signal also increases with discharge voltage. Keepered Configuration Figure 5 shows the variation in Mo LIF signal across the face of the DCA keeper. The data were taken.5 cm downstream of the keeper orifice plate. The signal maximum is at roughly.5 cm. The width of the window in the discharge chamber prevented measurements beyond 1 cm off centerline. Subsequent Mo data were taken at the.5 cm point. No W data were collected for keepered operation. Figure 6 shows the Mo LIF signal corrected for T e as a function of discharge 4

5 current for V D = 25 V and 27 V. Dependence son current and voltage are again evident, and the signal is non-negligible at J D = 8.2 A Assuming that the erosion-rates measured in the 2 hr wear-test were similar to those achieved for similar DCA/FMT operating conditions in this investigation, the orifice plate erosion-rates are given in Fig. 7 as a function of operating condition. Note that the Mo data show a greater dependence on operating condition which may be an artifact of the much smaller (and noisier) W LIF signal. The Mo data show average sensitivities of 34 µm/khr/a and 52 µm/khr/v for a benchmark of 28 µm/khr at 12.1 A, 26.7 V indicating a stronger dependence on discharge voltage than discharge current. Figure 7 also shows the keeper erosion-rate as a function of operating condition assuming the 13 A, 25 V case produces a rate similar to that measured in the 82 hr life-test. The average sensitivities for keepered operation are roughly 14 µm/khr/a and 25 µm/khr/v for a benchmark of 63 µm/khr at 13.1 A, 25 V. Thus, though the rates are smaller, the relative dependence on current and voltage is similar. Ion Velocities Four-beam LIF yielded Xe II velocities and temperatures in the region downstream of the DCA. Figure 8 maps the ion velocities downstream of the un-keepered DCA for the condition corresponding to that of the 2 hr wear-test, i.e. 12 A, 27 V. Note the back-flowing ions near the.3 cm radial position and the high off-axis velocities along the face of the DCA. The region of transition from low to high velocity appears as a quiescent region between.3 and.6 cm downstream. Figure 9 maps the ion velocities downstream of the keepered DCA for the 13 A, 25 V condition (which corresponds to that at the beginning of the LDT). Note the back-flowing ions near the.6 cm radial position and the high off-axis velocities along the centerline. The ions are back-flowing both axially and radially. There is a region of low velocity between.1 and.2 cm downstream of the keeper exit plane. Directed energy distributions along the face of the un-keepered cathode for 12 A, 27 V operation are given in Figure 1. Note that magnitudes of backflowing ion energies peak near the location of maximum erosion,.4 to.6 cm. In this region, a significant fraction of the ions were flowing back towards the cathode. These data agree with previous investigations at higher discharge voltages. 13,18 Directed energy distributions along the face of the keeper are given in Figure 11 for 13 A,25 V operation. Note that magnitudes of back-flowing ion energies peak near the location of maximum erosion, and that at the.6 cm location a significant fraction of the ions are flowing back towards the cathode. Table 4 lists the velocities and energies of the Xe II ions along the faces of the DCA for different operating conditions and configurations. Note the high velocities at.2-.3 cm off centerline positions in the un-keepered cases. Ion densities The magnitude of the LIF signal provided a rough estimate of the relative Xe II densities downstream of the DCA. Figure 12 shows an extended region of highdensity downstream of the orifice of the un-keepered cathode. The keepered data shown in Fig. 13 indicates a more localized peak just downstream of the keeper followed by an extended region of moderate density. Both sets of plots are for 13 A, 25 V operation to isolate, to the degree possible, the effect of the keeper electrode on the plasma structure. The LIF signal strength significantly decreased radially outwards along the face of the cathode in both cases. From the x-z views (Figs. 12c and 13c) the region of high density is more localized along the centerline of the plume in the un-keepered case. Error in both plots is estimated to be on the order of 1 percent. The ion density at the exit of the DCA has been measured to be on the order of cm -3 for 6 A, 18 V HCA operation. Conservatively estimating the density to linearly increase with current, the centerline density for 13 A operation would be on the order of cm -3. A model of the discharge keeper erosion is being developed which incorporates the variation in plasma density indicated by these measurements. 14 Discussion Erosion The throttling of NSTAR and post-nstar 3 cm ion thrusters is nominally done at a constant discharge voltage. 15 Therefore, the trends evident in Figs. 4, 6, and 7 are of value despite uncertainties in absolute values. Note that both keepered and un-keepered data 5

6 show a gradual rise in predicted erosion rate with discharge current (Fig. 7). Modest changes in J D below 12 A should moderately reduce the erosion. These trends indicate that the reduced erosion observed in the 1 hr wear-test and in the LDT resulted in part from counter acting contributions of a decrease in discharge voltage and an increase in discharge current. However, since the erosion rates and the ion distributions varied only slightly with the variation of these parameters, this investigation suggests that this alone would result in only a slight decrease in overall erosion. A fundamental change in the source of the ions generating the erosion is suggested. Figure 14 summarizes the suggested explanations for the erosion observed during the 2 hr wear test. Of particular significance was the measurement of Xe II ions with high radial velocities along the face of the cathode which probably sheared the heater coil. The measurement of a second, backflowing population of ions which probably resulted in the erosion around the edge of the orifice plate was also important. Table 5 extends the explanation to the keepered configuration. Plasma Properties The keeper electrode appeared to significantly modify the structure of the plasma downstream of the DCA. In particular, the ions appeared to be created in a more collimated region in the unkeepered cases. This appeared to influence ion paths near the DCA surfaces. Plasma Structure Localized increases in plasma potential can arise in a plasma in regions of relatively high ion density. 25 The more mobile electrons tend to leave the region more quickly than the more massive ions. As a result, the plasma potential increases to hold in the electrons and maintain quasineutrality. This phenomenon has been observed in the measurement of a negative anode fall in NSTAR-like ion thrusters. 26 It is also strongly suggested by internal plasma potential measurements made in Hg ion thrusters. 27 Ions in these regions of high-density would be expected to have small velocities. As the ions leave these regions they would fall through a potential and be accelerated. The presence of a potential-hill is suggested by the decrease in axial velocities in the keepered case (Fig. 9). Note that in both keepered and un-keepered configurations the DCA run near plume-mode. In such a mode, the region of ionization required to draw electrons out of the cathode extends significantly downstream of the cathode/keeper orifice. 5 Such operation is conducive to the formation of a potential-hill. 8 The extended region of increasing ion velocities downstream of the unkeepered DCA (Fig. 8) also suggests a potential-hill of a few volts. 13 Note, however, Fig. 9 shows that a significant fraction of the ions impacting the DCA surface in the keepered configuration are actually moving towards the cathode centerline. This suggests that a different mechanism of ion acceleration is in part responsible for the erosion. The structure of the Xe II density plots (Figs. 12 and 13) is consistent with the potential-hill descriptions of the ion accelerating mechanisms. More detailed plasma density measurements would be useful in resolving the plasma structure. High-speed internal probe measurements analogous to those made on a Hall thruster might provide the greatest insight. 28 Figures 15 and 16 suggest partial explanations for the different distributions of ion velocities, and densities and erosion rates observed in the keepered and un-keepered DCA configurations. These figures give a rough indication of how the velocimetry indicated the plasma downstream of the DCA differed between the two operating conditions and do not attempt to provide a final and complete description of the structure of the plasma. Note that there are two regions of primary ionization in Fig. 16. The region of ionization in the keepered case appeared to be farther downstream than in the unkeepered case. This may be due to the primary electrons being more focused as is illustrated in the figures and has been found experimentally. 29 The regions of varying ion acceleration are labeled in the figures. The contours indicate variations in ion density (and hence plasma potential) with the highest density towards the centerline. The boundaries of the secondary regions of ionization are illustrative and would require detailed probe measurements to determine quantitatively. The vectors associated with ion velocities indicate ions which have been accelerated after leaving the regions of 6

7 high ion density acquiring their velocity through a drop in potential. Ion Energies It might be argued that the above discussion is academic in that the energies even in the tails of the distributions in Figs. 1 and 11 are below the nominal threshold of sputter erosion for Xe on Mo (27 to 31 ev). 7 However, as mentioned above, two effects which might enable Xe II ions to be partially responsible for the observed erosion are acceleration through the DCA sheath and the angle of incidence of the impacting ions. Figure 17 compares the energies of Xe II and Xe III ions accelerated through a 2 V sheath. The Xe II energies correspond to the total energies (not axial energies as given in Fig. 11) measured for 13 A, 25 V keepered operation. The Xe III ions are assumed to have twice the energy of the Xe II ions entering the sheath but the same temperature. Hence, it would be inappropriate to simply multiply each velocity in the measured Xe II velocity distribution by a factor of 2. The Xe III velocity distribution is calculated assuming a Maxwellian distribution about a bulk velocity 2 that measured for Xe II. Note the narrowing of the profiles as they are accelerated to higher energies. Note also that a portion the tail of the Xe II ion energy distribution at the wall is above 3 ev in the 6 mm data while all of the Xe III energies at the wall are above 3 ev. Thus, Xe III would be expected to dominate the erosion through 4 mm. However at 6 mm the relative contribution of Xe II to the erosion process would depend on the ratios of yields and of number densities: both of which could vary by a couple orders of magnitude. Table 4 compares the pre-sheath and post sheath bulk energies for various operating conditions and configurations. Note the change in angle between the entering ions and those impacting the walls. Recall that the sputtering threshold decreases (by up to a factor of 2) with increasing angle of incidence up to roughly 5 degrees. 3 A portion of these Xe II ions therefore have energies sufficient to contribute to the erosion. A more detailed examination of the relative effects of singly and doubly charged ions is provided in a parallel investigation. 14 Sources of Experimental Error Since the LIF data were time averaged over tens of seconds, erosion induced by fluctuations in the plasma would also have been indistinguishable from that caused by stead-state ion impingement. Equally, the ion velocities would be time-averaged and may or may not include contributions from these effects. While it is know that fluctuations in the cathode voltage are present, 13 the resolution of ion velocities and erosion measurements on time scales of µs was beyond the scope of this investigation. The combination of small angles between interrogating beams and of fluctuations in the reference cell datum resulted in significant uncertainty in the ion velocity measurements. While the reference cell was stabilized, the angles could not be increased without significantly modifying the FMT. While technically possible, such modifications (e.g. providing optical access into the discharge chamber from just upstream of the ion grids) were beyond the scope of this investigation. However, doubling or tripling the angles of interrogation would increase the velocity resolution significantly and would provide a more rigorous measurement of the potentialhill. Conclusions The use of LIF to measure real-time internal erosion rates has been demonstrated. W, Mo, and Xe II species were interrogated to provide a clearer picture of the discharge cathode erosion process. Spatial variations in LIF signal strengths agree with measured erosion patterns. The Mo, and to a lesser degree W, LIF signals were proportional discharge current and voltage. The data indicate that operating at lower currents and lower voltages will reduce erosion. The erosion and ion velocity data support a general explanation of the DCA erosion process: erosion is caused by back-flowing ions created just downstream of the DCA exit plane. The presence of the keeper electrode appears to change the density distribution but has only a marginal affect on the velocity distribution. Significant radial velocities and back-flowing ions were observed with and without the keeper. One source of the back-flowing ions appears to be associated with the primary region of ionization near the exit of the DCA. The magnitude of the increase in local plasma potential (a potential hill) is proportional to 7

8 discharge voltage and also to discharge current. Ions move radially from their region of creation at high velocities. The more significant the potential-hill, the greater the back-flowing component of these ions. The presence of back-flowing ions with radial velocities towards the centerline indicates the presence of an additional mechanism. Singly ionized xenon should not be neglected in the determination of discharge cathode erosion rates or processes unless it is shown that the densities of Xe II and Xe III warrant such neglect. In many cases, the energy of Xe II through the sheath may be sufficient to erode cathode potential surfaces. In addition, Xe II ions, which are relatively easy to detect, provide an indication of the energy and direction of Xe III ions. This investigation also indicates that the keeper voltage may play a significant role in the mitigation of discharge cathode assembly erosion. Modeling and future experimental investigations of the sensitivity of the erosion to keeper voltage may be of value. Acknowledgements This work was made possible by the continuing support of NASA GRC and the personnel associated with the On-Board Propulsion Branch, especially M. Patterson. The research has been conducted under NASA grants NAG and NAG monitored by J. Sovey. The authors would like to thank the Department s technicians and the other students in the PEPL group for their assistance and support. References [1] Patterson, M. J., et al., 2.3 kw Ion Thruster Wear Test, AIAA , 31 st Joint Propulsion Conference (July, 1995). [2] Polk, J. E., et al., A 1-Hour Wear Test of the NASA NSTAR Ion Thruster, AIAA , 32 nd Joint Propulsion Conference (July, 1996). [3] Polk, J. E., An Overview of the Results from an 82 Hour Wear Test of the NSTAR Ion Thruster, AIAA , 35 th Joint Propulsion Conference (June, 1999). [4] Brophy, J. R., et al., The Ion Propulsion System on NASA s Space Technology 4/Champollion Comet Rendezvous Mission, AIAA , 35 th Joint Propulsion Conference (June, 1999). [5] Domonkos, M. T., Evaluation of Low-Current Orificed Hollow Cathodes, Ph.D. Thesis, The University of Michigan, October, 1999, pp [6] Katz, I., and Patterson, M. J., Optimizing Plasma Contactors for Electrodynamic Tether Missions, Presented at Tether Technology Interchange, September 9, 1997, Huntsville, AL. [7] Mantenieks, M. A., Sputtering Threshold Energies of Heavy Ions, IEPC , 25 th International Electric Propulsion Conference, [8] Kameyama, I., and P. J. Wilbur, Potential-Hill Model of High-Energy Ion Production Near High- Current Hollow Cathodes, ISTS-98-Aa2-17, 21 st International Symposium on Space Technology and Science, (May, 1998). [9] Crofton, M. W., The Feasibility of Hollow Cathode Ion Thrusters: A Preliminary Characterization, AIAA , 36 th Joint Propulsion Conference (July, 2). [1] Williams, G. J., et al., FMT-2 Discharge Cathode Erosion Rate Measurements via Laser-Induced Fluorescence, AIAA , 36 th Joint Propulsion Conference, 2. [11] Williams, G. J., The Use of Laser-Induced Fluorescence to Characterize Discharge Cathode Erosion in a 3 cm Ring-Cusp Ion Thruster, Ph.D. Dissertation, The University of Michigan, 2. [12] Domonkos, M. T., et al., Investigation of Keeper Erosion in the NSTAR Ion Thruster, IEPC-1-38, 27 th International Electric Propulsion Conference (October, 21). [13] Williams, G. J., et al., Characterization of the FMT-2 Discharge Cathode Plume, IEPC-99-14, 26 th International Electric Propulsion Conference (October, 1999). 8

9 [14] Williams, G. J., et al., Measurement of Doubly Charged Ions in Ion Thruster Plumes, IEPC-1-31, 27 th International Electric Propulsion Conference (October, 21). [15] Rawlin, V. K., et al., NSTAR Flight Thruster Qualification Testing, AIAA , 34 th Joint Propulsion Conference (July, 1998). [16] Williams, G. J., et al., Laser Induced Fluorescence Measurement of the Ion Velocity Distribution in the Plume of a Hall Thruster, AIAA , 35 th Joint Propulsion Conference (June, 1999). [17] Smith, T. B., et al., Deconvolution of 2-D Velocity Distributions from Hall Thruster LIF Spectra, IEPC-21-19, 27 th International Electric Propulsion Conference (October, 21). [18] Williams, G. J., et al., Laser Induced Fluorescence Characterization of Ions Emitted from Hollow Cathodes, AIAA (June, 1999). [19] Miles, R., Lasers and Optics, Course notes, Princeton University, [26] Foster, J. E., et al., Plume and Discharge Plasma Measurements of an NSTAR-type Ion Thruster, AIAA , 36 th Joint Propulsion Conference (July, 2). [27] Beattie, J. R., and J. N. Matossian, Mercury Ion Thruster Technology, NASA CR , (March, 1989). [28] Haas, J. M., and A. D. Gallimore An Investigation of Internal Ion Number Density and electron Temperature Profiles in a Laboratory Model Hall Thruster, AIAA--3422, 36 th Joint Propulsion Conference (July, 2)., [29] Jack, T. M., et al., The Effect of the Keeper Electrode on Hollow Cathode Characteristics, AIAA , 36 th Joint Propulsion conference (July, 2). [3] Duchemin, O. B., et al., A Review of Low-Energy Sputtering Theory and Experiments, IEPC-97-68, 25 th International Electric Propulsion Conference (October, 1999). [2] Verdeyn, J. T., Laser Electronics, 3 rd ed., Prentice-Hall, 1995, p 216. [21] Huddleston, R. H., and S. L. Leonard, Plasma Diagnostic Techniques, Academic Press, New York, 1965, pp [22] Poeschel, R. L., and J. R. Beattie, Primary Electric Propulsion Technology Study, NASA CR , [23] Poeschel, R. L., 2.5 kw Advanced Technology Ion Thrusters, NASA CR-13576, [24] Rock, B. A., Development of an Optical Emission Model of the Determination of Sputtering Rates in Ion Thruster Systems, Ph.D. Dissertation, Arizona State University, 1984, pp [25] Chen, F. F., Introduction to Plasma Physics and Controlled Fusion, Volume 1, Plenum Press, New York, 1984, pp

10 Table 1: DCA/FMT Operating Conditions Designation J D (A) V D (V) P (kw) m C m M TH (sccm) (sccm) 4 A A A A, 25 V A, 27 V hr wear-test 12 A, 32 V A A, 25 V A, 27 V A Where: J D = discharge current, V D = discharge voltage, P = Thruster power m C = discharge cathode flow rate m M = main flow rate TH = NSTAR/DS1 throttling point Table 2: LIF Interrogation angles. α (deg) β (deg) γ (deg) Un-keepered data 1.51± ± ±.42 Keepered data 1.57± ± ±.23 Table 3: LIF Transitions Species λ Laser (nm) Term Symbols {Energy levels (ev)} λ Fluorescence (nm) Term Symbols [Energy levels (ev)] I SAT (mw/cm 2 ) Xe I s [1/2] o 5f[1 1/2] s [1/2] o 5f[1 1/2] 1 { } { } Xe II d 4 D 6p 4 P o s 4 P 6p 4 P o 3 { } { } Mo { } { } 1 W { } { } 5

11 Table 4: A comparison of ion energies before and after passing through the sheath. Normal incidence corresponds to deg, parallel to 9 deg, and directly away from surface to 18 deg. Radial position (cm) Velocity (m/s) Before sheath Energy (ev) Angle (degree) Trough sheath (V D to cathode, 2 V to keeper) Velocity Energy Angle (m/s) (ev) (degree) Un-keepered 12 A, 27 V Un-keepered 13 A, 25 V Keepered 12 A, 27 V Keepered 13 A, 25 V

12 Table 5: Explanations for observed DCA erosion Observed Erosion Test Explanation Heater coil eroded flush with 2 hr High-velocity ions radially the orifice plate emanating from the DCA Electron beam weld eroded flush with the orifice plate Orifice plate eroded near its outer edge Cathode tube eroded with a curvature towards its outer edge Keeper orifice eroded to a chamfer Keeper orifice plate eroded across its surface Keeper orifice plate edge beveled plume 2 hr A combination of radial and back-flowing ions passing through the plasma sheath 2 hr Back-flowing ions originating in the plume 2 hr A combination of backflowing ions originating in the plume and in the ambient plasma 82 hr Back-flowing ions from the ion plume (potentially coupled to impinging ions generated in the cathode-to-keeper gap 82 hr Back-flowing ions originating in the ambient plasma 82 hr Back-flowing ions from the ambient plasma Windows Figure 1 Photograph and schematic of the FMT 3 cm ion thruster. Note the location of the windows on the discharge chamber wall and the ground screen. 12

13 inne r axial beam vertical beam, z outer axial beam lateral beam α β y, west north, x Figure 2 Schematic of laser beam delivery. Note the location of the beams on the lens. 1. Normalized LIF Signal W Mo. Figure 3 Radial distribution of Mo and W LIF signals which have been self-normalized. Note that no Mo data were taken right of centerline. Data shown are for 12 A, 27 V. 13

14 4 LIF Signal (arb units) V 27 V Discharge Current (A) Figure 4 Mo LIF signal as a function of discharge current for 25 and 27 V un-keepered operation. Solid lines are curve fits through the data Normalized Intensity c L Figure 5 Mo LIF signal across the face of the keeper electrode for 13 A, 25 V operation. Note that the signal corresponds to the erosion pattern. (Keeper picture taken from Polk. 3 ) 14

15 LIF Signal (arb units) V 27 V Discharge Current (A) Figure 6 Mo LIF signal as a function of discharge current for 25 V and 27 V keepered operation. Solid lines are curve fits through the data. Erosion-rate (µm/khr) Un-keepered, Mo, 25 V Un-keepered, Mo, 27 V Un-keepered, W, 25 V Un-keepered, W, 27 V Keepered, Mo, 25 V Keepered, Mo, 27 V Discharge Current (A) Figure 7 Erosion rates predicted as a function of operating condition assuming the erosion rates observed in the 2 hr and 82 hr wear tests were matched at the same operating conditions in this test. 15

16 Figure 8 Xe II velocity map for 12 A, 27 V un-keepered operation. Note the regions of back-flowing ions along the cathode face (x=) and the region of small velocities along the centerline between.3 and.6 cm downstream. Figure 9 Xe II velocity map for 13 A, 25 V keepered operation. Note the regions of back-flowing ions along the cathode face (x=) and the region of small velocities along the centerline between.3 and.6 cm downstream. 16

17 Centerline.1 cm.2 cm.3 cm cm cm Figure 1 Normalized axial Xe II energy distributions as a function of radial distribution along the face of the un-keepered DCA. 17

18 Figure 11 Energy distributions across the keeper face. Note the significant energies 4 and.6 cm downstream (roughly in the center of the keeper). 18

19 QGa QGa a. Viewing in the x-y plane b. Isometric view QGa QGa c. Viewing in the y-z plane d. Viewing in the x-z plane Figure 12 Three-dimensional plot of the variation in Xe II LIF signal strength for 13 A, 25 V un-keepered operation. Note the second peak 4 mm downstream of the cathode and the quick decrease in density downstream of that peak. 19

20 QGa QGa a. Viewing in the x-y plane. b. Isometric view. QGa QGa c. Viewing in the y-z plane. d. Viewing in the x-z plane Figure 13 Three-dimensional plot of the variation in Xe II LIF signal strength for 13 A, 25 V keepered operation. Note the smaller second peak 4 mm downstream of the keeper and the third smaller peak near 1 mm. 2

21 Cathode Orifice Plate (W) Erosion near outer diameter 16 µm/kr E-beam weld now flush Only surface-contouring near the orifice chamfer Back-flowing ions have the highest energies near the outer diameter Cathode Tube (Mo) Downstream edge eroded 275 µm/khr Edge curved Back-flowing ions have high energies in this region Inward-flowing ions may have resulted in the curvature Cathode Heater (Ta) Before test, extended downstream of orifice plate After test, flush with orifice plate High radial velocities could have sheared anything downstream of the orifice plate at cathode potential Figure 14 Explanations for the erosion observed during the 2 hr wear test. 21

22 Figure 15 A schematic of the proposed Xe II (and Xe III) acceleration mechanisms for un-keepered operation. The figure is not drawn to scale. 22

23 Figure 16 A schematic of the proposed Se II (and Xe III) acceleration mechanisms for keepered operation. The figure is not drawn to scale. 23

24 5x1-3 Distribution Xe II, Presheath Xe II, At wall Xe III, Presheath Xe III, At wall 1 6 mm 5x1-3 4 Distribution mm 14x Distribution mm 14x Distribution mm 2 4 Ion Energy (ev) Figure 17 Directed energy profiles of the singly and doubly charged ions before entering the sheath and at the wall. Xe III ions are assumed to have twice the energy of the Xe II ions that enter the sheath