Adjustment of electron temperature in ECR microwave plasma

Vacuum (3) 53 Adjustment of electron temperature in ECR microwave plasma Ru-Juan Zhan a, Xiaohui Wen a,b, *, Xiaodong Zhu a,b, Aidi zhao a,b a Structure Research Laboratory, University of Science and Technology of China, Academia Sinica, Hefei, Anhui 3, People s Republic of China b Department of Modern Physics, University of Science and Technology of China, Hefei, Anhui 3, People s Republic of China Received 3 April ; received in revised form August ; accepted October Abstract In glowdischarge plasma used for processing of materials, the electron temperature affects directly reactive species determining results of plasma processing of materials. In this paper, the experiment results of adjustment of electron temperature (T e ) in an Electron Cyclotron Resonance (ECR) device in H and H +CH plasmas are given by two methods. One method to adjust T e is the grid method, by installing a floated or biased grid in the plasma; the T e can be increased or decreased. Another method is noble gases doped into the discharge gases; increasing noble gas concentration can decrease T e in a certain scale. The changes of species in plasma are inspected by the optical emission spectroscope and mass spectroscope when T e is dropped. r 3 Elsevier Science Ltd. All rights reserved. Keywords: Adjustment of electron temperature; Ecr microwave plasma; Grid; Doping noble gas 1. Introduction Glowdischarge plasma had been widely applied in materials processing. In this kind of discharge plasma, the electron temperature (T e ) is a very important parameter. Because T e in plasma is much higher than temperature of neutral gas, the electron energy is enough to break chemical bonds and then active particles (molecules, atoms, ions, cluster) can be produced. The electron temperature determines the generation and the variation of reactive species in plasma [1] and affects the results of materials processing. In plasma processing of *Corresponding author. Department of Modern Physics USTC, Hefei, Anhui 3, People s Republic of China. Tel.: +-551-31-1; fax: +-551-33-3. E-mail address: rjzhan@ustc.edu.cn (R.-J. Zhan). materials, sometimes, the plasma having higher electron temperature is a need such as for deposition process; sometimes, the soft plasma with low electron and ion temperature is needed in plasma etching in order to achieve higher selectivity and less damaging []. Since the ion motion is not significant in producing or sustaining plasma for a range of RF or microwave frequency, control of electron temperature is of great importance for the plasma processing of materials. There are some reports on electron temperature control. The electron temperature is decreased by using emitting cold electrons [3], or by changing length of pins installed in a large hollowcathode []. Recently Sato et al. reported a newsimple negatively biased grid method to decrease electron temperature [5,]. In their experiment, noble gas was only used to produce plasma, and negatively -X/3/$ - see front matter r 3 Elsevier Science Ltd. All rights reserved. PII: S -X()1-

5 R.-J. Zhan et al. / Vacuum (3) 53 biased voltage was only investigated. Though there are also attempts to achieve higher selectivity by downstream region where electrons are not so energetic in inductively coupled plasma [,], under the same discharge power and operation pressure and to keep no lowplasma density, the study of adjusting T e is necessary. Here we report some experiment results of adjustment of electron temperature by using two methods in ECR microwave H and H +CH reaction gases plasma. The first method is using a floated or biased (negatively or positively) grid. This method is similar to that described in Ref. [5], but here positive biased grid and non-noble gas are investigated. The second method is doping noble gases in discharge gases, though doping noble gas in discharge gas is commonly used in many devices, adopting it as a method for T e control is seldom discussed. Changes of electron temperature and electron density under different conditions are given.. Experiments and results The experiments are carried out in the ECR microwave plasma device, the schematic of which is shown in Fig. 1. In this device, a vacuum bend waveguide is taken to link the microwave pressure window and the reaction chamber to avoid destruction of the ceramic pressure window caused by energetic particles impinging directly on to it. In order to avoid plasma generation in the vacuum bend waveguide, a pure iron yoke was used to reduce the magnetic field strength in the vacuum waveguide. The ceramic pressure window and vacuum waveguide are relatively free of contamination in this structure. The H gas or H +CH is used to produce plasma. Discharge gas pressure is 1 Pa, and input microwave power is 3 W with the frequency of.5 GHz. A grid made of stainless-steel wires is placed in the deposition chamber above substrate (dotted line in Fig. 1). The grid separates plasma into two parts. The grid is insulated from the chamber wall by PTFE frameworks. Five different mesh sizes are employed in the experiment. A Langmuir double probe placed about 1 cm belowthe grid is used to diagnose T e and N e of plasma passing the grid. Fig. shows the changes of T e and N e in plasma passing grids with different mesh sizes; no bias potential is applied to the grids. The T e and N e in plasma passing grid are all decreased obviously and the density N e will be decreased more while mesh size dropped. Fig. 3 shows changes of T e and N e when negative bias (relative wall of discharge chamber) grid is applied. The T e of plasma under the grid is decreasing with negative bias increasing, and the change of N e is little. Both the floating grid and negative biased grid will provide a retarding potential difference. That 1 no grid grid added 1 1 Te 3 1 3.3 1. 1...5 -- mesh size(mm) ( cm -3 ) Fig. 1. Schematic of ECR device. Fig.. Variations of N e, T e using different mesh size grids in H plasma, pressure 1 Pa, microwave power 3 W.

R.-J. Zhan et al. / Vacuum (3) 53 51 Te. - -1 - -3-5 -5-3 - -1 bias voltage(v) Fig. 3. N e, T e under negative grid bias in H plasma, other parameters are same, as in Fig... 1.5 1..5 ( cm -3 ) adjusting bias applied on grid placed into plasma. However, the grid method is a little complex in practice, here, we present another very simple method to decrease the electron temperature of plasma. We can take method of doping noble gases in discharge gases to decrease T e. In noble gas (such as argon and helium), there is metastable state. The accumulative ionization due to collisions between metastable atoms and electrons is important. The electrons after collision and newly produced electrons by ionization belong to low energy electrons. Fig. 5 shows the variations of T e in plasma with different concentration of doped noble gases (argon and helium) in H gas. The is, there is a negative potential barrier. So the electrons, which can pass the grid, are only higher energy electrons in the upper region of the grid. These electrons will slow down when they are passing the grid, and cannot be accelerated under the grid, because there is no external electric field being used for acceleration. Beyond resonance region, microwave energy cannot be transformed into electron kinetic energy. Therefore, the electrons in the region belowgrid possess lower energy than that in the upper region of grid so that T e in plasma belowgrid decreases. Moreover, the height of the negative potential barrier increases with negative bias applied on grid increasing, so the T e in plasma belowgrid is decreasing with negative bias increasing. When a positive bias relative wall of discharge chamber is used on grid, the T e of plasma under the grid will be increased with bias increasing and the change of N e is little (see Fig. ). In this situation the electrons in the upper region of the grid will all be accelerated when they are passing the grid, the electron temperature under the grid will be increased. This result is useful for deposition process, because the higher T e is favorable for excitation and ionization of gas molecules and atoms. These experiments results indicate that the grid method is effective for adjusting T e in H and H +CH reaction gases plasmas; the electron temperature of plasma can be controlled by using 1 Te. 1 3 5 5 3 1 bias voltage (v) Fig.. N e, T e under positive grid bias in H plasma, other parameters are same as Fig.. 5 3 Concentrate(%) Fig. 5. T e under different noble gas concentration, parameters are same as in Fig.. He Ar 3..5. 1.5 1..5 ( cm -3 )

5 R.-J. Zhan et al. / Vacuum (3) 53 (x cm -3 ) 3 3 3 He Ar 1 1 1 3 Concentration(%) Fig.. N e under different noble gas concentration, parameters are same as in Fig.. Relative intensity 5 3 NO GRID CH,CH/H=1:1,L3nm no.113-1 - - - - Grid Voltage(V) Fig.. OES results of CH concentration under different grid voltage, in H +CH (1:1) plasma. relative intensity 5 15 5 Hβ Hα..1..3..5 Ar concentration (%) Fig.. Intensity of H a and H b spectra under different Ar concentration. Relative abundance 1 1 1 1 15 1 13 5 3 NO GRID CH CH CH C3H3 - - - - Grid Voltage(V) Fig.. Mass spectroscopy of carbon hydrogen species under different grid voltage, in H +CH (1:1) plasma. changes of electron density N e in mixed gas discharge plasma are given in Fig.. Here the decrease of T e and increase of density N e are ascribed to cumulative ionization due to collisions between electrons and metastable atoms. We can see that doping Ar is more effective than doping He in decreasing T e and increasing N e. The changes of species in plasma are inspected by using optical emission spectroscopy (OES) and mass spectroscopy (MS) when T e is adjusted. Fig. gives variations of H a and H b spectra intensity under different Ar concentration in an H plasma. In CH plasma, it is observed from Figs. and that CH, CH are decreased obviously and C H, C H, C 3 H 3 are increased after the grid is placed into the plasma, but variation of these species is not obvious when a negative bias applied on the grid is increasing. The phenomena may be related with the amount of T e drop. The variation of the reactive species in plasma when T e is adjusted will be studied further. 3. Conclusions We have presented experiment results using two methods to adjust the electron temperature in an ECR microwave H and H +CH reaction gases glow discharge plasma when discharge power and

R.-J. Zhan et al. / Vacuum (3) 53 53 operation pressure are constant. The electron temperature in processing plasma can be dropped and increased by installing a floated or biased (negatively or positively) grid to separate processing region from the ECR resonance region. In this condition, when T e of plasma is deceased, the electron density of processing plasma has a small drop. If doping a little noble gas in discharge gases, the electron temperature in processing plasma will also be dropped effectively and the plasma density will rise. The second method is more convenient for some needing low T e case such as plasma etching. The reactive species in plasma also varies when the electron temperature of plasma is adjusted. It is observed that carbon hydrogen species tended to contain more carbon atoms while T e of plasma is dropped in the range from about 15 ev to a fewev. The reactive species determine results of processing of materials. So it is possible to control results of processing of materials by control of electron temperature of plasma. Acknowledgements This work was supported by the National Nature Science Foundation of China. Under Grant Nos. 1553 and 1353. References [1] Pierre B, Jean Claude Boettner, Lionel Vandenbulcke. Jpn J Appl Phys 1;31:155 13. [] Keizo Suzuki, Naoshi Itabashi. Pure Appl Chem 1; : 5. [3] Alexeff I, Jones WD. Appl Phys Lett 1;:. [] Sato N, Iizuka S, Koizumi T, Takada T. Appl Phys Lett 13;:5. [5] Kohgi Kato, Satoru Iizuka, Noriyoshi Sato. Appl Phys Lett 1;5:1. [] Kohgi Kato, Tetsuji Shimizu, Satoru Iizuka, Noriyoshi Sato. Appl Phys Lett ;:5. [] Fukasawa T, Nahamura A, Shindo H, Horike Y. Jpn J Appl Phys 1;33:13. [] Jiwari N, Fukasawa T, Nakamura A, Kubota K. Jpn J Appl Phys 1;33:5.