V. Bellido-González 2, J. O Brien 3, P. Kelly 1 D. Arnell 1, D. Monaghan 2

Transcription

1 Application of Fourier Transform Analysis to Pulsed Magnetr on Sputter ing Technology V. Bellido-González 2, J. O Brien 3, P. Kelly 1 D. Arnell 1, D. Monaghan 2 1 Salford University, IMR (UK) 2 Gencoa Ltd (UK) 3 Pilkington PLC, R&D Group (UK) 1

2 Structure of the presentation The Fourier Transform Our problem Magnetron sputtering + pulse + power supply... Phenomena involving magnetic fields and light Initial development Experiments Results Applications Conclusions 2

3 The Fourier Transform The Fourier transform, in essence, decomposes or separates a waveform or function into sinusoids of different frequency which sum to the original waveform. It identifies or distinguishes the different frequency sinusoids and their respective amplitudes f t f(t)=a 0 /2 + {a k cos(k 2 t) + b k sin(k 2 t) } 3

4 The Fourier Transform COSINE DECOMPOSITION 0.8 Value a1*cos(x) a2*cos(2x) a3*cos(3x) Variable f(t)=a 0 /2 + {a k cos(k 2 t) + b k sin(k 2 t) } 4

5 The Fourier Transform { (a k2 + b k2 )} vs k SPECTRUM k = k Intensity,a.u Freq., KHz 5

6 The Fourier Transform from the SPECTRUM Original function Periodic pulse (power) 4000 Power (watts) Reconvolved Original Time (µs) 6

7 Our Problem In what way our system responds to different frequencies? What s the effectiveness of a pulse wave package? What are the mechanisms involved in the response to a pulsed wave? 7

8 Magnetron Sputtering + Pulse + Power Supply +... Fibre Optic Cable OEM Sight Glass Variable Magnetron Monochromator Power Supply Gas Controller Piezo Valve O 2 Gas Supply Channel A Advanced Energy Pinnacle Plus Power Supply Unit Substrate Holder Channel B 8

9 Magnetron Sputtering + Pulse + Power Supply +... Different approaches Electronics: Capacitors + Inductors +... Physical:... dark shield space, acceleration, deceleration, energy distribution of particles... Physico-chemical: Absorption of energy at particular wavelengths for a system under the influence of a magnetic field. 9

10 Magnetron Sputtering + Pulse + Power Supply +... A difficult question How much of all this pulse is information of the power supply? How much from the plasma? 10

11 Phenomena involving magnetic fields and light Two typical absorption phenomena involving matter under irradiation when in an external magnetic field: EPR: Electron Paramagnetic Resonance NMR: Nuclear Magnetic Resonance Due to the Spin: E= h = ( h B h B 11

12 Phenomena involving magnetic fields and light Frequency range depends on magnetic field. Magnetron Sputtering T EPR: Electron ( = x Hz/T For a field 874 G Freq= 2.45 GHz NMR: Nuclei ( ~ Hz/T For a field 300 G Freq~ 100 KHz-10MHz h B 12

13 Initial Development NMR absorbers 13

14 Initial Development Magnetic field on target surface (VTech 0,0) Magnetic Flux density (Tesla) Ti 47, Ti 49 # abund. 13% = x 10 7 rad/(t s) Distance from cetre, mm Resonance Frequencies for target surface of Ti and Al on a VTech magnetron Al 27 # abund. 100 % = x 10 7 rad/(t s) Freq., khz freq.-ti freq-al distance from centre, mm 14

15 Initial Development The fundamental frequency will be part of the effective pulse. Now if we take the first 4 cosine harmonics for Ti Pulse for Ti (to 4-th order freq) for Vtech in (0,0) Signal KHz 80% Time, µs 20%... then a pulse and a duty cycle are created!! Total 15

16 Initial Development Taken the first 4 cosine harmonics for Al (on a VTech 0,0) Pulse for Aluminium (4-th order) for Vtech in (0,0) Signal KHz % 74% Time, µs Total The resonance freq. is around 544 KHz (3rd harmonic) 16

17 Initial Development Maybe pulse package should be changed to accommodate an appropriate duty cycle : New Ti package wave for maintaining main resonance frequency (118 KHz) and 80:20 group pulse Voltage Time, µs 17

18 Initial Development If the NMR contributions are significant we should see: 1.- for Ti, changes on the main 118 KHz when compared to other frequencies. 2.- for Al, changes on 3rd harmonic 544 KHz when comparing with other frequencies. 3.- As in NMR spectroscopy, changes in the chemistry (e.g. reactive sputtering) should shift the main resonances therefore the absorption of energy should be affected NMR is quantitative, i.e. No. of absorbers/emitters. We would be able to see linear or monotonous curves. 18

19 Experiments Dual (coplanar)& Single Magnetron sputtering (VTech 0,0) Experiments for Ti and Al targets For Ti: 120 KHz, 50:50 dual # 50:50 single & 80:20 single For Al: 180 KHz, 50:50 dual # 50:50 single & 74:26 single Power for Ti: 0.1, 0.5 & 2 KW per target average. Power for Al: KW per target average Gas: P= 1.2 mtorr # Ar+O 2 (OEM control) 19

20 Experiments I-V waveforms collected with a digital oscilloscope Deconvolution performed on the following wave forms: Voltage Current Power= V x I ***(mainly)*** 20

21 Results Power spectrum: Ti dual Ti-GEN-03 P spectrum, 120 KHz, 2 KW # 50:50 dual Ti-GEN-03 # 120 KHz 2 KW # dual Power, Intensity,a.u. watts Freq., KHz Time, µs Intensity Power (watts) 21

22 Results Comparative intensities for Ti dual# 120KHz#50:50 Linearity of FT coeff. with power for Ti dual 120 KHz, 50: Intensity coeff I-120 I-240 I-360 Linear (I-120) Linear (I-240) Linear (I-360) Power, KW 22

23 Results Ti single 80:20 pulse # Intensity linearity Linearity Ti# 120 KHz of some 80:20 frequency s # Power intensities intensity for Ti harmonics 120KHz, 3-10 for 80:20 different power. Intensity Power Intensity, at frequency a.u Linear (I-960) Power, Fredq., KW KHz I-720 I 0.1 KW I-840 I 0.5 KW I-960 I 2 KW Linear (I-840) 23

24 Results Al comparative Al# Al 180 # KHz :50, KHz # Power Power Intensity Intensity Coeff. for Vs first power 3 harmonics Power Power Intens. Intensity,a.u. Coeff., a.u Power, Freq., KW KHz KHz 361 KHz 541 KHz I d#50:50 Linear (180.5 I s#50:50 KHz) Linear (361 I s#75:25 KHz) Linear (541 KHz) 24

25 Results Al in reactive mode (O 2 ) Power Intensity FT for Al coeff. target at for diff. diff. freq. O2 for level Al + Ar + O2 Pow er e r Intensity Int. coe ff., coeff., a.u. a.u O2 input (red. Frequency, of intensity KHz at 396 nm) I@2888 KHz I@3069 KHz 20% O2 I@10830 KHz 70% O2 I@541 KHz I@5054 KHz 25

26 Results Al in reactive mode (O 2 ) Freq. 180 & 360 KHz are not monotonous. So not possible to control reactive gas input by it them. Freq. 541 KHz is the NMR freq. and it is monotonous. CONTROL POSSIBLE. Frequency group KHz could offer a good control also. 26

27 Where are these frequencies coming from? Ti-GEN-02 P spectrum, 120 KHz, 0.5 KW # 50:50 dual Intensity,a.u Intensity Freq., KHz? Electr onic (outside NMR and not EPR)? Plasma resonance of electron circulation (ExB modes?) 27

28 Circulation frequencies? S N N N J.W. Bradley (UMIST / UK) measured circulation frequencies ~ 1 MHz on a 150 mm diameter magnetron and suggested that this could play a role in the pulse energy absorption. 28

29 Resonance relaxation Resonance relaxation reported by Günter Mark (MELEC) and Jones Alami (Linköping University) 1000 V per square Fluctuations in the voltage signal Target voltage Target power 100KW per square Courtesy of Jones Alami Linköping University Sweden 50 A per square Corresponding fluctuations for the power Time scale: 50 us per square 29

30 Applications FT could be applied to reactive sputtering control as the shape of the pulse changes so the intensity coefficients. N. Malkomes, et al., Comparison of Different Concepts for Stabilization of Reactive Sputtering Reported for Nb 2 O 5 and TiO 2. SVC, Lake Buena Vista, April (Fraunhofer Institute for Surface Eng. & Thin Films # IST) Nb 93 has an abundance of 100% and NMR sensitive Ta 181 abundance ~ 100% and NMR sensitive. FT could be applied in order to select a particular pulse frequency and a particular pulse package form. 30

31 Conclusions Our research hasn t been conclusive on wether or not NMR plays a role in pulse plasma MS. However FT applied to a I-V (or P) wave gives the system spectrum. Variations of the spectrum are observed with power, target material and reactive gas. Linearity of freq. coeff. with power for some freqs. Correlation with reactive gas input (optical emission) could offer contr ollability. A reactive control process has been devised by other authors. 31

32 Conclusions A group of freq. around 3-10 MHz shows up which respond in different way to power and reactive gas concentration. At this moment it hasn t been possible to separate the effects of power supply and plasma system. Further experimental effort will be needed in order to understand FT information from pulsed plasma process waves. THANK YOU 32

33 This document was created with Win2PDF available at The unregistered version of Win2PDF is for evaluation or non-commercial use only.