Advanced Diagnostics for Full Characterization of ECR-MW Ion Sources

Advanced Diagnostics for Full Characterization of ECR-MW Ion Sources Iulian Teliban∗ , Thomas Trottenberg, Viktor Schneider, Holger Kersten Institut f¨ur Experimentelle und Angewandte Physik, Christian-Albrechts-Universit¨at Kiel, D-24098 Kiel, Germany Horst Neumann, Michael Tartz, Frank Scholze Leibniz-Institut f¨ur Oberfl¨achenmodifizierung, D-04318 Leipzig, Germany April 30, 2008 Abstract. The performance of an ECR-MW broad beam ion source using different diagnostics is investigated. For this purpose tools used in plasma diagnostics are improved and adapted. The ECR plasma in the source and the secondary plasma in the target chamber are analyzed using Langmuir probes. Ion current density profiles in the beam are measured using a Faraday-cup array. The energy of the beam ions is carefully investigated using a three-grid retarding-field analyzer. Correlations between axial and radial profiles of the ion beam and source plasma are investigated. 1 Introduction Ion beam sources are widely used in material processing, especially for ion beam etching and ion beam assisted deposition [1, 2, 3, 4]. Using of ion beams for electric spacecraft propulsion is already applied and continuously improved [5, 6, 7], e.g. SMART-1, DEEP SPACE1. Knowledge about the beam properties, e.g. ion-energy distribution, ion-current density and their profiles, are necessary in all applications. Therefore, plenty of work has been spent for improvement of ion sources [9, 14] and development of beam diagnostics. In this paper we characterize the performance ∗email: [email protected] of an electron cyclotron resonance (ECR) microwave (MW) broad beam ion source [2] using different diagnostics. This source was designed to overcome different disadvantages of common broad beam ion sources, i.e filament-less microwave plasma combined with the availability of an adapted grid system. The beam properties such as density, energy and the profile are investigated using Faraday-cups and retardingfield analyzers. Axial and radial profiles of beam energy and density as function on source parameters are measured. The ECR plasma in the source is investigated using cylindrical Langmuir probes. Plasma profile is carefully determined by five radially distributed probes. The relatively high working pressure (about 5×10−2Pa) leads to a homogeneous surrounding plasma in the beam chamber. The plasma in the target chamber is analyzed using planar Langmuir probes. The results from different diagnostics are checked for consistency and discussed in the summary section. 2 Experimental setup A 50 l high vacuum (HV) stainless steel chamber is used for investigation of a broad beam ion source (IS, figure 1). It is a microwave, ECR source developed by the Institut f¨ur Oberfl¨achenmodifizierung Leipzig. Details 1 about this MW ion source has been already discussed in other paper [2]. The ion source and the chamber are vertically orientated to compensate the gravitational force by the ion beam for the later experiments which imply falling microparticles [8, 9]. The vacuum chamber was pumped down to a pressure of about 10−5Pa by means of a turbomolecular pump backed by a scroll pump. The operation pressure is set to about 5 ×10−2Pa by means of the process gas (argon) flow into the ion source. Due to the high transparency of the ion source grid system, the pressure in the ion source is only slightly higher than the chamber pressure. IS RFA,Planar LangmuirProbe Cylindrical LangmuirProbes Fraday-cuparray, PlanarLangmuirProbe MWAntenna Ar 410mm Fig. 1: Schematic of the experimental set-up (VIBEX) Electrical (Langmuir) probes Spatially resolved measurements of plasma parameters inside and outside discharge chamber of the ion source are performed using Langmuir probes. Inside radial profiles of plasma parameters are measured using five cylindrical probes inserted through the acceleration and screen grid orientated parallel with the magnetic field. The probe tips (d=100 µm and l=4 mm) are free ends of tungsten wires whose other parts are isolated by ceramic tubes. Outside plasma parameters are measured using a planar probe. It consists of a stainless steel planar plate of 15 mm diameter. The current −voltage probe characteristics are acquired and evaluated using a computer-controlled system. Plasma potential and electron-energy distribution are evaluated from the second derivative of the probe characteristic [10]. The plasma ion and electron densities are calculated from saturation currents. Retarding-Field Analyzer The energy distribution function of resulting ions is measured using a retarding-field analyzer (RFA). Basically, the RFA has three plane square-mesh grids with transparency of 70%. The grids are parallel to each other and stacked on the top of a plane collector. The ensemble is exposed to the flux of charged particles and neutrals coming from the ion source. + - - -5V -30V CG1 G2 - - -30V G3 l1l2l2 z + Fig. 2: Schematic diagram of the potential distribution in the retarding-field analyzer. Figure 2 is a sketch of the analyzer as seen from a plane along the analyzer z axis. All conductive analyzer components are made of stainless steel. The first grid facing to the plasma is spot welded on a diaphragm with an aperture 15 mm in diameter. The grid to grid distances are l1=1 mm between collector and the third grid and l2=2 mm for all other grids. The grid distances are higher than in typical RFAs for plasma diagnostics, where the ion energies is much smaller [11, 12] in order to avoid discharges between grids caused by the high voltage applied on the retarding grid. The collector is maintained at a constant potential of −15 V. Grids No. 1 and 3 are held at the same constant potential of −30 V. The retarding grid No. 2 is swept with 100 V around the expected ion en2 ergy. The measurement of ion energy distribution function first consists of recording the collector current as a function on the retarding grid potential. The conversion from the obtained I-V characteristic into the ion energy distribution function is done by first derivative of the collector current [13]. Faraday-Cup Array The ion-beam density profile is measured using a linear Faraday-cup array consisting of nine analyzers. They are radially distributed from center to the wall in order to cover half of the vacuum chamber. The distance between array cups is about 1.5 cm. On this way the linear array is able to measure half of radial ion density profile in one measurement. The Faraday cup, unlike the RFAs, have only one (electron repelling) grid and the collector. The grid is negatively biased in order to repel the plasma electrons. The collector is connected through a resistor to ground, which allows for the determination of the flowing ion current by means of a measurement of the voltage drop over the resistor. 3 Ion source parameters The electric grid potentials and grid currents at the source grids are deciding parameters for the beam density and ion energy [14]. For a better understanding of the ion source operation we analyze the current characteristic at the grids. Figure 3 shows a sketch of ion source and electrical circuits of the grids. An internal anode (beam) positively biased (Ub) is used to control the beam energy. The outer grid (accelerator) is negatively biased (Ua) and controls the beam density. The internal grid is used only for screening and is kept floating. Argon operation gas and molybdenum grid system are used. The microwave power is set to 240W for all measurements. Details about ion source operation can be found elsewhere [2]. Figure 4 shows the effect of total voltage between grids (|Ua|+Ub) on the current characteristic. These results show that almost simUa Ub Fig. 3: Schematic drawing of the IS discharge vessel and electrical circuits [14]. ilar current values has been achieved for different combinations of beam (Ub) and accelerator (Ua) voltage. Therefore the current depend only 200 400 600 800 1000 1200 1400 1600 1800 0 50 100 150 200 grid voltage (V) current (mA) Beam Accelerator Fig. 4: Beam and accelerator current in argon at different grid voltage. Operation gas pressure is p= 5 ×10−2P a. The anode current is reduced by the losses to the outer grid. on the total voltage applied between grids, and is nearly independent if the voltage is achieved by the beam or accelerator voltage setting, respectively. At low potential values the sheath formed in front of the screen grid moves into the grid holes and results a high acceleration current (red line). For high grid voltages the sheath is squeezed out of the grid holes and formed in front of the screen grid and the accelerator current decrease. A minimum value of 3 the accelerator current is required for a normal operation of the ion source, i.e. the grids are less sputtered. The potential disturbance caused by the grids is screened within plasma sheath, but a residual disturbance remains across the discharge plasma in the pre-sheath [15]. The potential gradient into pre-sheath increases with the grid bias and causes an additional drift of charged particles to the ambipolar diffusion current and leads to a higher beam current to the anode (blue line). 4 Ion beam diagnostic Up to here we have discussed only the processes occurring at and in the ion source. But the final ion beam profile characteristics are the most important issues of an ion source. The influence of the source parameters on the beam characteristics are further analyzed in this section by means of the measured ion current density and ion-energy distribution function using Faradaycups and retarding-field analyzers. First 1D ion beam profiles of the beam density are mapped using the Faraday-cup array at different source parameters. Furthermore, by varying the axial distance from the grids, 2D maps of the broad beam are estimated. Figure 5 shows 2D profiles of the beam for different settings of the beam energy (Ub). The accelerator grid is grounded for no beam profile and kept at -100 V for other profiles. At the first glance these results show presence of a low density ion beam (no beam) even for grounded accelerator grid. This is caused by the ECR plasma potential (inside of the ion source) which may be higher than 0 V. This setting is not common for the standard operation of the present ion source. It will be analyzed only for a better understanding of the plasma inside and outside the source. For all profiles a rapid decay of the ion density with axial distance, i.e. with about 50% in 10 cm, is observed. This is caused by the relatively high neutral gas pressure. The mean free path for Ar+is about 15 cm and is slightly increasing with the ion energy. On the other Fig. 5: Ion density profiles for different beam energies. The white dotted line represents the source center. By the colorbar the current density scale is denoted. Please note the large difference in the scales. side the beam width is similar to the source diameter (125 cm) close to the grids and slightly 4 broader far from the source, i.e. no significant space charge in the beam. In the radial direction the beam has a humped profile for high ion energies (600 eV, 800 eV) and presents a maximum at about 4 cm from the center. The profile is symmetric in azimuthal direction and is slightly shifted from center of the source. Therefore, presence of the MW antenna (alumina cup) in the center of ECR plasma (figure 3) influence the beam profile for high ion energies. The shift between the profile center and the source center (white dashed line) indicates that the focusing grids are not perfect parallel. Further more, the energy of the beam ions is carefully estimated using a three-grid retardingfield analyzer at about 25 cm above the ion source. The analyzer carrier (see figure 1) is moved to measure radial energy and density profiles. Figure 6 shows the ion-energy distriFig. 6: IEDF for different beam energies. The x axis is normalized to the anode potential. Line color denotes the set ion energy as shown by the legend. bution function (IEDF) for different beam energies. They are directly estimated from the first derivative of I-V analyzer characteristic. These profiles are acquired in the area of maximum current density, i.e. about 3 cm from the center (see figure 5). The energy axis is normalized to the anode potential setting. At the first view, IEDF show a maximum with about 50 eV above the set energy, i.e. anode potential. The shift indicate that ECR plasma potential is about (50-60) V above anode potential but these details will be discussed in the plasma diagnostic section. On the other side the width of IEDF is slightly broader for low beam energy. It decrease from 45 eV at low beam energy (Ub=200 V) to 30 eV for energetic ions (Ub=800 V on the anode). Figure 7 shows radial profiles of IEDF for different anode potentials. By the color filled profiles the IEDF is denoted. They show a similar width for all radial positions. The wrapFig. 7: IEDF radial profiles at different beam energies. The external read line is extrapolated from IEDF maximum for every radial position of the analyzer. per of IEDF profiles (red line) shows the radial density profiles at about 25 cm above the ion source. The density units are arbitrary because the RFA transparency is not known and difficult to estimate, i.e. repelling grids are not perfect aligned. Density profile shape shows a good agreement with Faraday-cup array measurements (figure 5). Ion energy and density dependence on accelerator grid potential is further analyzed. Figure 8 shows IEDF profiles for different anode and accelerator potentials. For a better view the yaxis is reversed. This dependence shows a clear influence of the accelerator potential on the beam density. A gain in beam density of about 30% is obtain by increasing the accelerator potential with 200 V. On the other hand the ion energy is not influenced by the accelerator voltage. These results are in good agreement with other previous measurements [14] 5 - - - - - Fig. 8: IEDF dependence on the accelerator potential. The beam energy is coded by the profile color and xaxis. 5 Plasma diagnostic The shape and the thickness of the sheath formed in front of the screen grid determine the energy and density of the resulting beam. On the other hand the thickness and shape is affected by the plasma density (MW power) and the voltage applied on the grids. Therefore, for a better understanding of the beam energy and density the ECR plasma inside the ion source has to be investigated. This is done using five cylindrical Langmuir probes radially distributed near the top of the MW antenna. Furthermore, the surrounding plasma in the target chamber is investigated using a planar probe. On this way a correlation between plasma parameters inside and outside of the source and the grid voltage is obtained. Plasma and floating potential First, the potential distribution across the ion source and the vacuum chamber is sketched as a function of set beam energy (figure 9). In figure 9 the red color line shows is potential distribution with accelerator potential setting to 0 V. The anode is floating at about 70 V which is similar with the floating potential measured by the probe. The blue and green lines are representations for anode potential setting to 200 eV and 400 eV, respectively. For both values the accelerator grid is set to -100 V. The screen grid is kept floating for all measurements. In this U =470V plasma U =400V anode U =386V screen U =270V plasma U =200V anode U =181V screen U =125V plasma U =70V anode U =45V screen Anode Potential U =-100V acc Distance ECRPlasma Screen Accelerator Targetplasmachamber Fig. 9: Potential distribution across the ion source and the vacuum chamber sketch all distances and the shape of the lines are taken only for a qualitative representation of the potential distribution. The real measured values are denoted by the markers. They are determined by plasma potential estimation from the I−Vprobe characteristic and by the measured potential at the grids. These results shows that plasma potential inside of the ion source is always about 70 V higher than the anode potential. The outside surrounding plasma measurements indicates a plasma potential of about 15 V for no beam situation and 3-4 V for beam setting to 200 V and 400 V, respectively. Furthermore, radial electric fields caused by the space charge in the beam have been not identified across the entire beam width. Axial and radial profiles have been measured and they indicate almost constant values of the plasma parameters. A clear difference between ECR plasma and external plasma, produced by the beam ions can be already seen from potential measurements. 6 Fig. 10: Plasma and floating potential profiles measured inside of the ion source. By markers, the measured values are denoted (five Langmuir probes). Yellow marked area shows the alumina cup where the MW antenna is localized. The radial profiles of the floating and plasma potential inside the source are represented in figure 10. These profiles are measured at fixed hight about 8 mm from the screen green. They show constant value of the plasma potential along radial direction. Only a slight decay of the floating potential in radial direction is observed. The small energy shift seen in the IEDF profiles (figure 6) can be now justified by the ECR plasma potential which is always 60-70 V higher than the anode potential. Therefore, for the beam energy setting, i.e. by the anode potential, one has to consider this shift caused by the ECR plasma potential. Plasma electron density The plasma electron density profiles are also estimated from the probe characteristics and are shown in figure 11. They show higher values close to the MW antenna and another maximum at about 40mm. These regions, where the electron plasma density is higher, denote the ECR resonance areas as it was sketched in figure 3. At a microwave frequency of 2.45 GHz the resonance is fulfilled in two toroidal regions surrounding the antenna where the magnetic field reaches 87.5 mT. These regions shows a slightly lower density for high beam energy setting because of the ECR plasma quasi-neutrality, i.e. more ions leave the ion source at higher anode voltage. Small decay in density can be seen in Resonanceregion Fig. 11: Plasma electron density radial profiles of ECR plasma. The colored vertical bands indicates the ECR resonance areas. They have a torus shape surrounding the MW antenna. radial direction due to the lower energy transfer from MW to plasma in the outside resonance torus. Fig. 12: Electron temperature profile measured by Langmuir probes. They are estimated directly from I−Vprobe characteristic. Plasma electron temperature Simulations [16] and experimental investigations [17, 18] on electrons of ECR-IS plasmas have shown presence of three electron populations: hot (∼10-20 keV), warm (∼keV) and cold (∼tens of eV). In out investigation the bulk cold electrons are dominant and are measured by means of Langmuir probes. The electron temperature profiles are shown in figure 12. They shows bulk electrons with about 18 eV 7 temperature for all anode potentials. A slightly higher temperature in the resonance areas is observed. Outside measurements has shown a 100 times lower plasma electron density (5 ×1014 m−3) which slightly decay with the distance from the grids. Furthermore, ion density estimation indicates a quasineutral plasma above the ion source. Presence of two kinds of electrons as function on the grid potential is observed. For no beam situation, when the anode is kept floating, electrons with about 3eV temperature are measured across all vacuum chamber. They are produced outside of ion source by collision of ECR electrons (∼18 eV), which escape from the source with neutrals. For more energetic beams (>200 eV) secondary slow electrons (∼0.5 eV) produced at the chamber wall by the energetic ions are dominant in the surrounding plasma. 6 Summary In this paper we have characterized the performance of a ECR/MW broad beam ion source according to beam density, ion energy distribution function and beam profile using electrical tools such as Faraday cups and retarding-field analyzers. 2D maps of the resulting beam profile has been measured using a Faraday-cup array. The beam profile generated by superposition of multiple beamlets, shows a hunch shape slightly sifted according to the source center for high energetic beams. The beam diameter is similar to the source width even far from the grids. The radial electric caused by the space charge in the beam does not play a significant role in our experiment. The ion energy was investigated using a three grid retarding-field analyzer. They show a 50 eV above the set energy, i.e. by the anode potential, caused in principal by the ECR plasma potential. Dependence of beam energy and density on the source parameters has been carefully analyzed. They confirm that the anode potential sets the beam energy and the grid voltage the beam-current density. The ECR plasma inside the source and surrounding outside plasma are simultaneous investigated by means of Langmuir probes. Using cylindrical probes the ECR resonance areas are identified and localized in the ion source. Probe results confirm a high density plasma (∼1016 m−3) produced by microwave excitation as a reliable source for ion beam sources. Measurements of plasma potential indicates always value of (60-70) V above the anode potential. After this measurements the shift in energy of the resulting beam has been better understood. Therefore, the ECR plasma potential determine directly the ion beam energy although the plasma is controlled by the anode potential. A planar probe has been used for plasma diagnostic in the target chamber. The relatively high neutral gas pressure and secondary electrons emitted from the chamber wall leads to a surrounding lower plasma density (∼1014 m−3). In many application of surface processing the neutralization of the beam current on the target is necessary and additional electron beam tools are used. In this investigations probe measurements have indicated a quasineutral plasma and very low values of the plasma and floating potential even close to the grids (3-4 V) in the target chamber. These investigations on beam energy, density and profile complete the previous experimental [9, 19] and theoretical [20, 7, 21] works. Furthermore, it is proved that relatively cheap tools can be used for diagnostic of broad ion beams. 7 Acknowledgment The investigation on broad beam ion source is founded by the DLR project no. 50JR0643. References [1] B. Wolf. Handbook of Ion Sources. CRC, Boca Raton, FL, 1995. [2] M. Zeuner, F. Scholze, H. Neumann, T. Chasse, G. Otto, D.Roth, A. Hellmich, and B. Ocker. Surface &Coatings Technology, 142:11–20, 2001. 8 [3] F. Scholze, H. Neumann, M. Tartz, J. Dienelt, and H. Schlemm. Rev. Sci. Instrum., 77(03):107, 2006. [4] N. Sakudo. Rev. Sci. Instrum., 71(2):1016– 1022, 2000. [5] J. E. Pollard, and M. Martinez-Sanchez. Journal of Prop. Power, 14(5):688, 1998. [6] H. Neumann, M. Tartz, F. Scholze, T. Chass, H. Kersten, and H. Leiter. Contrib. Plasma Phys., 47(7):487–497, 2007. [7] M. Tartz, E. Hartmann, R. Deltschew and H. Neumann. Rev. Sci. Instrum., 71(2):732–734, 2000. [8] H. Kersten, R. Wiese, H. Neumann, and R Hippler. Plasma Phys. Controlled Fusion, 48:B105–B113, 2006. [9] H. Kersten, R. Wiese, M. Hannemann, A. Kapitov, F. Scholze, H. Neumann, and R. Hippler. Surface &Coatings Technology, 200:809–813, 2005. [10] I. H. Hutchinson. Principles of Plasma Diagnostics. Cambridge University Press, 2002. [11] J. A. Simpson. Rev. Sci. Instrum., 32(12):1283–1293, 1961. [12] U. Kortshagen, and M. Zethoff. Plasma Sources Sci. Technol., 4:541–550, 1995. [13] C. Bohm, and J. Perrin. Rev. Sci. Instrum., 64(1):31–44, 1992. [14] M. Zeuner, H. Neumann, F. Scholze, D. Flamm, M. Tartz, and F. Bigl Plasma Sources Sci. Technol., 7:252–268, 1998. [15] K. U. Riemann. J. Phys. D: Appl. Phys., 24:493–518, 1991. [16] V. D. Dougar-Jabon, A. M. Umnov and V. B. Kutner. Rev. Sci. Instrum., 67(3):1152–1154, 1995. [17] D. Hitz, G. Melin, and A. Girard. Rev. Sci. Instrum., 71(2):839–845, 2000. [18] A. Girard, D. Hitz, G. Melin, and K. Serebrennikov. Rev. Sci. Instrum., 75(5):1381– 1388, 2004. [19] M. Zeuner, J. Meichsner, H. Neumann, F. Scholze, and F. Bigl. J. Appl. Phys., 80(2):611–622, 1996. [20] M. Tartz, E. Hartmann, F. Scholze, and H. Neumann. Rev. Sci. Instrum., 69(2):1147–1149, 1998. [21] J.H. Peters, M. Tartz, and H. Neumann. Rev. Sci. Instrum., 77:03B, 2006. 9