Non-invasive measurements of plasma parameters via optical emission spectroscopy Amy Wendt Dept. of Electrical and Computer Engineering University of Wisconsin - Madison Michigan Institute for Plasma Science and Engineering Ann Arbor, MI, December 10, 2014 Wednesday, December 10, 14 Acknowledgments • Collaborators and students . Dr. John Boffard, UW Physics Dept. Prof. Chun C. Lin, UW Physics Dept. Shicong Wang, UW ECE Dept. Cody Culver, UW Materials Science Program . Nathaniel Ly, UW ECE Dept. Dr. Ryan Jung, UW Physics Dept. (now at IBM) . Lauren Aneskavich, UW ECE Dept. (now at Rockwell Automation) . Dr. Svetlana Radovanov, Applied Materials Corp. Dr. Harold Persing, Applied Materials Corp. • Support . National Science Foundation . Applied Materials . UW Graduate School/Wisconsin Alumni Research Foundation 26 Wednesday, December 10, 14 Low-temperature plasma applications • applications are diverse • manipulating plasma is a means to control technological outcomes • diagnostic tools are needed to improve plasma science understanding 3 Wednesday, December 10, 14 LTPs are non-equilibrium systems - Plasma e + A ⟹ e- (1-10 eV) collisions with high EM power input ------------------ energy e- drive: ions (cool) excitation (glow) neutrals (cool) ionization (sustains plasma) high-T chemistry in low-T gas • energetic electrons are key link in process outcomes 4 Wednesday, December 10, 14 Electron energy distribution functions • Maxwellian EEPF appears as straight line - . corresponds to thermodynamic equilibrium within electron population . characterized by single parameter - Te • RF inductively coupled plasmas: EEDFs with ‘depleted’ high energy ‘tails’ . attributed to: inelastic collisions escape to walls Godyak et al., PSST 11, 525 (2002) 5 Wednesday, December 10, 14 Presentation Outline: Optical diagnostics for plasma characterization • Optical emission spectroscopy of argon-containing plasmas . emission model for determination of EEDF, Teff metastable and resonance level densities electron density - ne Example: non-Maxwellian EEDFs in inductively coupled plasma • Real-time plasma monitor . metastable and resonance level densities - nm and nr . effective electron temperature - Teff . electron density - ne • VUV detection . resonance level densities - nr - as a proxy for Ar VUV flux 6 Wednesday, December 10, 14 Each plasma emits distinctive spectrum • http://wapedia.mobi/en/Gas_filled_tube - different rare gases • Also, plasmas with the same gas changes colors with conditions 7 Wednesday, December 10, 14 Partial Ar energy level diagram • Allowed transitions resulting in photon emission 2p1 (J=0) 13.48 eV 2p (J=1) 13.33 eV 2 2p (J=2) 13.30 eV 3 2p (J=1) 13.28 eV 4 2p (J=0) 13.27 eV 5 2p (J=2) 13.17 eV 6 2p7 (J=1) 13.15 eV 2p (J=2) 13.09 eV 8 2p9 (J=3) 13.08 eV 2p (J=1) 12.91 eV 10 826.45 nm 826.45 nm 772.42 nm 727.29 nm 696.54 nm 852.14 nm 794.82 nm 747.12 nm 714.70 nm 935.42 nm 866.79 nm 810.37 nm 772.38 nm 1148.8 nm 1047.0 nm 965.78 nm 912.30 750.39 nm 750.39 nm 667.73 nm 840.82 nm 738.40 nm 706.72 nm 857.81 nm 751.47 nm 922.45 nm 800.62 nm 763.51 nm 978.45 nm 842.47 nm 801.48 811.53 nm 811.53 11.83 eV 1s2 (J=1) 11.72 eV 1s3 (J=0) 11.62 eV 1s4 (J=1) 11.55 eV 1s5 (J=2) 8 Wednesday, December 10, 14 Argon emission spectrum • Ultimate Goal: use emission spectra to extract encoded information about plasma conditions 600 W, 25 mTorr ICP 103 102 101 100 -1 OES Signal (arb. units) (arb. Signal OES 10 10-2 350 400 450 500 550 600 650 700 750 800 850 900 950 Wavelength (nm) 9 Wednesday, December 10, 14 Inductively coupled plasma (ICP) • Operates in Ar and Ar mixtures at 13.56 MHz • Electron energy distribution functions (EEDF) . Langmuir probes . optical diagnostics 10 Wednesday, December 10, 14 Determining plasma properties from spectrum: emission model • Goal: . predict photon emission rate Φij for i⟶j • Approach: . account for mechanisms that i populate emitting state - level i ij j l 11 Wednesday, December 10, 14 Emission model • Goal: . predict photon emission rate Φij for i⟶j • Include: . electron excitation from level l excitation probability is a i function of electron energy ij measured “cross sections” j used to characterize excitation probabilities e- excitation l 12 Wednesday, December 10, 14 Emission model - excitation cross sections • cross section vs. electron energy . measure of electron impact excitation probability . key link between emission intensity ) 2 800 and EEDF cm 2p (J=0) -20 1 . measured by Prof. Chun Lin’s 600 750.39 nm group: excitation from ground state 400 • Phys. Rev. A 57, 267 (1998) • Phys. Rev. A 68, 032719 (2003) 200 • Ar. Data Nucl. Data Tables 93, 831 (2007) 0 Emission Cross Section (10 050100150200250 excitation from metastable levels Electron Energy (eV) • Phys. Rev. Lett. 81, 309 (1998) • Phys. Rev. A 59, 2749 (1999) • Phys. Rev. A 75, 052707 (2007) 13 Wednesday, December 10, 14 Emission model - reabsorption • Goal: . predict photon emission rate Φij for i⟶j 2p • Include mechanisms for i x populating level i : ij . electron excitation from level l . radiation trapping: excitation of j metastables by photon 1sy absorption radiation trapping e- excitation l 14 Wednesday, December 10, 14 Emission model - cascades • Goal: . predict photon emission rate Φij for i⟶j cascades • Include mechanisms for populating level i : 2p i x . indirect electron excitation from level l ij . cascades from higher lying states j . measured cross sections include 1sy cascades radiation trapping direct e- excitation l 15 Wednesday, December 10, 14 Emission model - electron excitation of metastables • Goal: . predict photon emission rate Φij for i⟶j • Include mechanisms for 2px populating level i : i . electron excitation: ij from ground state j from metastable/resonance 1sy 1sy levels radiation . photon reabsorption/trapping trapping from metastable/resonance levels e- excitation g.s. l 16 Wednesday, December 10, 14 Excitation from resonance levels 1s2 & 1s4 • 1s3 and 1s5 are metastables: . no radiative losses . high concentrations possible • 1s2 and 1s4 populations significant ) 250 -3 when radiation is trapped: 1s4 resonance level cm 200 8 . 1s2 and 1s4 resonance levels: have radiative decay channel to 150 ground state 100 generally lower concentrations 50 exception: reabsorption when 0 ground state density is high (10 Density Number 0510152025 . Electron impact excitation from Pressure (mTorr) resonance levels significant cross sections not measured estimates used Adv. At. Mol. Opt. Phys. 54, 319 (2006) 17 Wednesday, December 10, 14 Cross sections: variations in threshold and energy dependence used in model • 419.8 and 420.1 nm - 1.0 mTorr 35 420.07 nm 3p 420.07 nm close in wavelength 9 30 (gs3p9) ) 419.83 nm 3p5 2 25 cm -20 20 419.83 nm Teff 5.9 eV (gs3p ) • difference between exc. from gs (10 15 5 Opt ground state and Q 10 metastable cross 5 excitation from ground state 0 419.5 420.0 420.5 05101520253035 sections Energy (eV) 25 mTorr 1400 420.07 nm (1s 3p ) • large difference in peak 1200 5 9 ) 2 intensities at 1 mTorr vs. 1000 excitation from metastable/ cm -20 800 T 2.8 eV resonance state 25 mTorr eff 5 (10 600 exc. from 3p 4s 419.83 nm (1s43p5) Opt . higher ground state Q 400 419.83 nm (1s 3p ) contribution at higher Te 200 5 5 0 419.5 420.0 420.5 036912 Energy (eV) J. Phys. D. 45:045201 (2012) 18 Wednesday, December 10, 14 Emission model: summary • Model goal: . predict photon emission rate Φij ⟶ for i j cascades • Effects included: . electron excitation from level l 2px cascades i metastable contribution ij . radiation trapping j 1s • Needed inputs y radiation . atomic data (Aij, cross sections,...) trapping . electron energy distribution function direct e- excitation . number densities (ground state AND 1sy for argon) l 19 Wednesday, December 10, 14 Emission model: intensity of i→j transition l Φij = KneRij nlkij ⇥ l ⇤ • sum over initial level l (ground state, 1s2, 1s3, 1s4, 1s5) l • k ij is the electron impact excitation rate . depends on cross sections, EEDF 2 ∞ kl = Ql (E)f (E)√EdE ij m ij e ⇥ e 0 • reabsorption correction factor Rij is function of 1sy concentrations • use ratios to standard line to eliminate K, ne l Φij Rij l nlkij = l Φi j Ri j nlk l ij • calculate line ratios for multiple trial EEDFs • seek EEDF giving best match to observed line ratios 20 Wednesday, December 10, 14 Trials for Maxwellian EEDF • minimum χ2 gives best fit electron temperature 14 12 2 10 15 mTorr 8 6 2.5 mTorr Reduced 4 2 0 12345678 OES Te (eV) OES OES Te =2.4 eV Te =5.5 eV 21 Wednesday, December 10, 14 Emission model predictions 100 • contributions from ground 80 state vs. 1sy 2.5 mTorr 60 Teff = 5.4 eV levels n /n = 1.4x10-3 m 0 40 % Contribution % 20 0 2p1 2p2 2p3 2p4 2p5 2p6 2p7 2p8 2p9 2p10 ground state 100 1s3 & 1s5 metastable 1s2 & 1s4 resonance 80 15 mTorr 60 Teff =2.7 eV n /n = 3.2x10-4 m 0 40 % Contribution % 20 0 22 2p1 2p2 2p3 2p4 2p5 2p6 2p7 2p8 2p9 2p10 Wednesday, December 10, 14 Application: Non-maxwellian EEDFs in ICPs Wednesday, December 10, 14 Inductively coupled plasma source • 2.5 turn flat coil antenna • 13.56 MHz • 50 cm diameter chamber • measure emission model inputs • compare EEDFs with independent measurements gas variable pressure Power ne Te (mTorr) (W) (cm-3) (eV) Ar pressure 1-50 600 1010-1012 2-6 Ar power 2.5, 15 20-1000 109-1012 5, 3 10 11 Ar/N2 0-86% N2 2.5, 15 600 10 -10 5, 3 Ar/Ne 1-40% Ar 10 600 1010-1011 3-6 Ne pressure 5-25 600 1-3 x1010 5-7 plasma inductively-coupled 13.56 MHz plasma 10 cm 2.5-turn antenna 24 Wednesday, December 10, 14 Model input: ground state density • n0 = p/kTgas • gas temperature determined with laser absorption spectroscopy • Doppler width of 3p54s-3p54p transition line 1.0 0.8 0.0026 nm 0.6 810 K 0.4 0.2 0.0 801.47 801.48 801.49 Norm.