Etching Applications and Discoveries Made Possible by Advanced Ion Energy Control

Transcription

1 Etching Applications and Discoveries Made Possible by Advanced Ion Energy Control Possible by Advanced Ion Energy Control V. M. Donnelly Department of Chemical and Biomolecular Engineering University of Houston Acknowledgements: DoE Plasma Science Center, NSF 1

2 Outline Parallel nano-patterning with a monoenergetic, low- energy ion beam. Using the monoenergetic, low-energy ion beam to do etching in the plasma. Details on generation e the monoenergetic, oee low-energy e ion flux with pulsed plasmas and synchronous bias. Discovery of an important new cause of anisotropic etching: indigenous photo-assisted etching (PAE). 2

3 Pantograph Mechanical device used to copy drawings. Can be on a different scale 3

4 Nanopantography Side View: Broad area ion beam V b TopView: metal, V m dielectric conducting film or substrate, V s Essential components: Parallel, monoenergetic ion beam and electrostatic lenses built on the wafer Sub-10nm patterning simultaneously; Self-alignment, immune to vibration and thermal expansion; Etching (Ar+), deposition (M+). 4

5 Nanopantography: write any desired nanopatterns simultaneously Whatever is written once on the imaginary plane here is reproduced at the bottoms of all the lenses here. Side view Top view 5

6 Setup of Nanopantography for Si etching Matching Network Ion source Ar pulsed plasma Collimated, monoenergetic Ar + beam Etching Chamber Lens array voltage Cl 2 effusive beam Sample with lens array 6

7 Details of Sample Tilting Stage External gear In-vacuum gear Sample Grounded mesh Lens sample Loadlock External motor In-vacuum motor Motorized stage To pump 7

8 Extract Ion Beams with Desired Narrow Energy Spread ON 0 V Pulsed Plasma +100 V OFF Pulsed plasma Ring acceleration voltage Extract ions at low T e and T i in afterglow decreases energy spread and increases the directionality of ions Ring acceleration voltage (high-voltage pulse): Set the desired ion beam energy. 8

9 Nearly Monoenergetic Ar Ion Beam The energy spread was 3.4 ev (FWHM) for a peak ion beam energy of ev. Normal lized IED dc bias: 30 V ; 50 V 70 V; 100 V 30.5 ev 51.0 ev 71.5 ev (b) ev Ion energy (ev) IED (A.U.) Better RFA resolution. V peak =99 ev FWHM= 22eV Ion energy (ev) Lin Xu, Demetre J. Economou, Vincent M. Donnelly, and Paul Ruchhoeft, Appl. Phys. Lett. 87, (2005). 9

10 Etching results with Cl 2 and monoenergetic Ar + beams (ion incidence angle: normal and 5 degree) SEM Top view of lens array L. Xu et al. Nano Letters, 5, 2563 (2005). Cr layer (top electrode) Lens bottom surface TEM HAADF-STEM, dots b) ~110 nm 950 nm 20 nm 10 nm 50 nm Size reduction factor of ~95X Focused point can be displaced along the substrate surface 10

11 Setup of nanopantography for Ni deposition Ni target Plasma source Ar pulsed plasma Ni coil Pulsed Ni lamp Optical diagnostics (OAS and OES) Pumping Drift tube Pumping Collimated Ni + and Ar + ions with energy spread 1.2 ev for 100 ev beam L. Xu, et al. APL. 87, (2005). Deposition chamber Lens array voltage Sample with lens (ion landing energy: 20eV) 11

12 SEM images (top view) of sample after Ni deposition Normal Incidence Ni + Dark: bottom surface of lens 10nm 700 nm Bright: Metal top electrode The focusing ration is 70X for the Ni nanodots. Height = 5-10 nm by AFM. 12

13 Etching of Si with Cl2 and Ar+ beams: X-tilting Etching of Si with Cl2 and Ar+ beams: X, Y-tilting 13

14 Another Motivation for Precise Control of IED Chemical Physical Sputtering E th (I) Low energy ions : No reaction occurs except for pure chemical etching, if any. Sputtering E th (II) High energy ions: chemical sputtering, sub-surface ion penetration, physical sputtering not well-controlled etching, substrate damage. (I) (III) (II) (III) Medium energy ions: Only chemical sputtering. More confined to surface. Etching slows or stops when adsorbed d etchant reacts away. minimum damage, layer-by-layer etching. Layer-by-layer etching, high selectivity, and no substrate damage are/will be critical requirements. To achieve these, one needs to control the IED. Accurate control of peak ion energy. Narrow width of IED. Work with medium energy ions (region III). 14

15 Apparatus for Controlled Ion Energy Faraday-shielded ICP. Plasma power as well as voltage on Boundary Electrode can be pulsed. Diagnostics : Langmuir Probe, Ion Energy Analyzer, OES. H. Shin et al., Plasma Sources Sci. Technol. 20 (2011)

16 Normalized IEDs in cw Ar ICP with different continuous DC bias voltages applied to the boundary electrode Normaliz zed IED V -4V Ground +4V +8V +12V V p measured by Langmuir probe W, 14 mtorr Ar Energy (ev) 16

17 (ev) T e Electron Temperatures and Number Densities in Different Rare Gas Pulsed ICPs ON 11 OFF Experiment Model Ar Ar Kr Kr Xe Xe Time ( s) n e (cm -3 ) 1x x x x x x x x x x10 11 (a) ON OFF Ar Kr Xe time ( s) T e decays more slowly in heavier gas due to slower e-diffusion rate. n e larger in heavier gas due to faster ionization and slower ion diffusion rates. n e (measured at the edge of the plasma) hardly decays in 80 s. 17

18 Comparison of Measured and Modeled Decay of Electron Density in Afterglow of Pulsed Ar ICP 1.4x10 11 ne 1.2x x10 11 Density (cm -3 ) 8.0x x x10 10 n Ar+ model 2.0x Time ( s) Model in good agreement with experiment. The near-constant plasma density ~80 s into the afterglow is due to diffusion of positive ions form the high density region near the center of the ICP to the edge. 18

19 Ion Energy Distributions (IED) in Different Rare Gas Pulsed ICPs 0.5 Ar Kr 0.4 Xe Vdc s after power off (a) IED Energy (ev) V p = V p_i -V p_f = 2.6, 2.2,1.8 V for Xe, Kr, Ar Flux order Xe > Kr > Ar because of higher n e and afterglow T e in heavier gas. IED width Xe > Kr > Ar because of higher afterglow T e and V p in heavier gas. 19

20 Normalized IEDs in pulsed Ar ICP with different continuous DC bias voltages applied to the boundary electrode Norma alized IED eV 4eV +4V +8V +12V Ground -4V -8V 4eV 4eV Energy (ev) 112 W average power, MHz, 10kHz, 20% duty cycle, 14 mtorr Ar 20

21 Normalized IEDs in pulsed Ar ICP with different synchronized DC bias voltages applied to the boundary electrode Distributi ion Time-Avera raged Ion Energy Ar press. (mtorr): Energy (ev) 21

22 Pulsed Plasma, Monoenergetic IEDs for Si Etching Gas inlet Boundary Electrode Faraday Shield RF pulsing system Time-Averaged Io on Energy Distribu ution Ar pressures 7 mtorr 14 mtorr 28 mtorr 50 mtorr Energy (ev) Matching network Periscope for OES Monochromator substrate for etching Faraday-shielded ICP. Plasma power as well as voltage on Boundary Electrode can be pulsed. Diagnostics : Langmuir Probe, Ion Energy Analyzer, OES. H. Shin et al., Plasma Sources Sci. Technol. 20 (2011)

23 Ion Energy Dependence of Si Etching in Chlorine-Containing Plasmas Chlorine Containing Plasmas Poly-Si Kinetic study of low energy argon ion-enhanced plasma etching of polysilicon with atomic/molecular chlorine, J. P. Chang, J. C. Arnold, G. C. H. Zau, H-S. Shin, and H. H. Sawin, J. Vac. Sci. Technol. A, 15, 1853 (1997). Characteristics of polycrystalline Si etching with Cl and Ar + beams in high vacuum: Etching exhibits a threshold ion energy of 16 ev. Above this threshold, etching rate increases with the square root of ion energy. Yield (Si atoms etched per io on) E th 16 ev Sqrt. ion energy (ev 1/2 ) 23

24 What We Expect in a Pulsed Plasma for Si Etching Time e-averaged Ion Energy Distr ribution 0.05 E th ~16 ev 50 mtorr no etching etching Relative Et tching Rate mTorr 60mTorr E th ~16 ev ER (Å/min mtorr Energy (ev) E 1/2 (ev 1/2 ) Ar/Cl 2 (few %) pulsed plasma. Above threshold h etching looks like what we 20 s ON 80 s OFF. expect. Synchronous DC bias in the afterglow at s. Big question (and big problem): What /cm 3 p-type Si causes sub-threshold h etching? 24

25 What Causes Sub-threshold Etching? Spontaneous chemical etching by Cl atoms? - It has been reported that this does not happen for p-type or i-si (only heavily doped n-type). - This is confirmed in our cross section SEMs: no undercutting of the mask. SiO 2 mask Si Therefore spontaneous chemical etching by Cl atoms is not the reason. 25

26 What Causes Sub-threshold Etching? Ar metastable-assisted etching? - Ar metastables energies (11.55, ev for 3 P 2 and 3 P 0 ) sufficient to cause desorption of SiCl x. - Never been reported. Unlikely, given the nearly unit efficient quenching of rare gas metastables on surfaces. Also, Cl 2(g) quenches Ar metastables. - We find pure Cl 2 plasmas have a constant, non-zero etching rate below ~12 ev, suggesting similar sub-threshold etching in Cl 2 and Cl 2 /Ar plasmas. Therefore Ar metastables-induced etching is not the cause of 1.0 sub-threshold etching. 0.5 Relative Etc ching Rate Boundary Voltage 1/2 (V 1/2 ) 26

27 What Causes Sub-threshold Etching? Neutralization of low energy ions? -Ar + releases up to the ionization energy (15.76 ev) minus the work function when neutralized at a surface, sufficient to cause desorption of SiCl x. - Could an Auger electron (or hole created in the valence band), cause a reaction in the chlorinated layer that leads to etching. - If we positively bias the sample to reject low-energy Ar +, sub-threshold etching should stop. Photoassisted t etching (PAE) induced d by light produced d in the plasma? PAE of Si with lamps and lasers in the presence of a high pressure of Cl 2 has been reported. - If we positively bias the sample to reject low-energy Ar +,sub-threshold etching should not stop. Electron-assisted etching? Reported to be very weak effect. Should be much less important in Cl 2 vs dilute Cl 2 /Ar plasmas. - If we negatively bias the sample to reject electrons, sub-threshold etching should stop. 27

28 Grids / Substrate Bias Experiments energetic positive ions positive ions Cl, Cl 2, photons Cl, Cl 2, photons Cl 2 /Ar Plasma Current (ma) Si -5V Voltage (V) te lative Etching Rat Rel V 0.1 L/0.1S 0V Conclusion: Sub-threshold etching is due to photo-assisted etching induced by plasma light. 28

29 Wavelength Dependence for Photo- Assisted Etching 1000 No mask or window tching Rate (A/min n) Si Et 100 Beneath Si mask Beneath quartz window 10 Si mask blocks most light from reaching the sample. Quartz transmits >170 nm. Seems ~90% of PAE due to <170 nm. Both Cl and Ar known to have intense VUV lines between 100 and 140 nm. 29

30 30

31 PHOTO-ASSISTED Si ETCHING: HALOGEN DEPENDENCE ing rate (nm m/min) Etch % Cl 2 / 50% Ar 50% Br 2 / 50% Ar % HBr / 50% Ar 25% Cl 2 / 25% Br 2 / 50% Ar 25% Cl 2 / 25% HBr / 50% Ar 400 Photo-assisted etching: - fastest in Cl 2 /HBr - slowest in Br 2 Isotropic etching: - HBr-containing plasmas - probably H atoms 200 photo- assisted etching rates = arrow lengths isotropic etching E 1/2 (ev 1/2 ) 31

32 PHOTO-ASSISTED Si ETCHING: HALOGEN DEPENDENCE u.) min) (XPS a.u es (nm/m overage ching rate Cl + Br c Etc Ar 7504 C IA Cl 2 Br 2 HBr Br 2 /Cl 2 HBr/Cl 2 Measure halogen by XPS Measure Ar 7504 A emission intensity. (Proportional to VUV intensity?) Photo-assisted etching rate scales as the product of light intensity times halogen coverage. 32

33 Summary Mono-energetic ion fluxes with selectable energy can be formed in the afterglow of a pulsed plasma with synchronous bias of a boundary electrode. This can be used to extract an ion beam through a grid for remote processing such as nano-pantography. The same approach could be used to improve etching processes (nearthreshold energies, better selectivities and less physical damage), but... At low ion energies, photo-assisted etching, mainly from VUV light generated by the plasma, becomes dominant. This seems to be a serious problem for future etching processes. 33

34 Co-PI: D. J. Economou Postdoc: H. Shin Students: L. Xu, W. Zhu, S. Sridhar, L. Liu 34