Title: Plasma-Etch Limits: Molecular Dynamics Simulations and Vacuum Beam Measurements

1
FLCC Seminar
Your Photo Here

• Title Plasma-Etch Limits Molecular Dynamics
  Simulations and Vacuum Beam Measurements
• Faculty David B. Graves
• Department Chemical Engineering
• University UC Berkeley

2
Current and Future Challenges in Plasma Etch
Following Scaling and Beyond...
R. Chau, Intel, 2005
3
Photoresist Roughness Challenges
4
Etch Roughness Challenges CD Control
Following Ma, Levinson and Wallow, AMD 2007
5
Etch Fundamentals Control Performance Metrics
CD/Anisotropy
Etch Fundamentals Ion impact energy/angle Radical
impact Electron impact Passivation
layers Surface transport Etc.
Roughness
Selectivity
Rate Uniformity
Following Tom Wallow, AMD, 2007
6
How to Attack Current and Future Plasma Etch
Challenges?

• 1. Conduct studies of fundamentals of plasmas and
  plasma-surface interactions to develop intuition
  and insight into dominant mechanisms, usually on
  model systems.
• 2. Use fundamental studies to improve and extend
  empirical and statistical studies to address
  real, practical systems in a way that can
  directly affect process development.

7
Ultimate Goals of Research

• Develop theory about how plasmas alter/etch
  surfaces at atomic scale
• - what are important factors (species energies
  angles surface temperature types of mask
  relevant length and time scales, etc.)
• 2. How do these (and other?) factors govern
  fidelity of mask-to-film pattern transfer?
• 3. Combine simulation and experiment to develop
  methods to usefully simulate nanofeature
  profile evolution given information about plasma
  conditions (i.e. given factors above)
• 4. Use these simulations/experiments to help
  develop and control practical plasma etch tools
  and processes

e.g. Professor Jane Chang, UCLA profile
simulation, FLCC
8
Multi-Scale Plasma Etch
Pressure 5 - 500 mTorr Gas temperature 600
- 1000 K
Electron temperature 2- 8 eV Ion energy
20-1000 eV
Ar/C4F8/CHF3/...
Etch Gas in
plasma sheath d 1 mm
plasma
Dielectric film
300 mm
SiF4 COF2
Gas Flow Out
VRF V01sin (w1t) V02sin (w2t)
e.g. Professor Mike Lieberman, FLCC collaborator
9
Mechanisms of Plasma Etching Passivation or
Modification Layers

• All surfaces exposed to plasma are MODIFIED
• All surface processes occur through, and are
  affected by, a layer of surface modification
• Plasma-induced surface modification is a FIRST
  order effect and must be included in any serious
  model of plasma-surface interaction.

Top surface modified layer
Resist
Film
Feature sidewall modified layer
Feature bottom modified layer
Substrate
10
Mechanisms of Plasma Etching Passivation or
Modification Layers

• Surface modification typically nm thick
• Modification strongly influenced by ion
  bombardment-induced energy deposition, bond
  breaking and mixing
• Neutral species (typically radical fragments)
  play important roles as both etch precursors
  (e.g. F) and depositing precursors (e.g. CFx)
• Very few details understood, however

11
How to Model/Simulate Plasma-Surface Interactions?
Molecular dynamics simulations - classical,
semi-empirical potentials - resolves
vibrational timescales O(10-15 s)
1. Ion impact - crucially important energy input
10-13-10-12 s collision cascade - MD time
and length scales match physics of
interactions - weakly bound species after
collision cascade removed simple TST for
thermal desorption with Eb ? 0.8 eV. 2.
Radical-surface chemistry - accuracy of
interatomic potentials?? (cf. ab-initio) - time
and length scales adequate?? (cf. experiment)
12
MD Cell and Assumptions for Etch Simulation
Impact events followed for 1 ps excess energy
removed statistics collected new impact point
chosen repeat sequence 103 times for steady
state surface.
Surface composition and structure must reach
steady state.
Bottom boundary fixed new Si added here
2 nm
13
Molecular Dynamics Characteristics

• Accessible time- and length-scales match part of
  the plasma-surface interaction problem
• Energetic impacting species dissipate their
  energy within a picosecond among 102 103
  atoms
• Tersoff-Brenner style REBO potentials for Si-C-F
  and C-H-F systems (Tanaka, Abrams, Humbird and
  Graves)
• Potentials are short-ranged, designed to simulate
  covalent bonds

C.F. Abrams and D.B. Graves, J. Appl. Phys., 86,
5938, (1999) J. Tanaka, C.F. Abrams and D.B.
Graves, JVST A 18(3), 938 , (2000) D. Humbird
and D.B. Graves, J. Chemical Physics, 120 (5),
2405-2412, 2004.
14
Molecular Dynamics (MD) Simulation
Interatomic Potential
Interatomic Forces
typical MD time step
update positions
evaluate forces
Ions assumed to neutralize before impact fast
neutral interacting with surface
update velocities
is assumed to model all reactive and non-reactive
interactions
15
What are Mechanisms of Fluorocarbon Plasma
Etching?

1. Known that etching generally takes place through
   a film of fluorocarbon material (various F/C
   ratios)
2. Generally assume that film reduces the rate of
   etching on the substrate
3. Substrates that react with C (e.g. O, but also N)
   will usually result in thinner steady state
   films, all things being equal
4. F atoms known to greatly reduce selectivity to
   PR, Si or nitride
5. Details very murky/ no self-consistent picture
   (descriptions, not predictive models)

16
Surface Transport and Chemistry Fluorocarbon
Plasma Etching

• Steady state etching requires
• FC film material deposited and removed
• at equal rates.
• Etchant (F) transported to substrate
• interface.
• Etchant reacts with substrate to form etch
  product
• Etch product transported to film surface.
• Etch product leaves surface.

FC film
All processes must occur simultaneously!
Substrate material
17
Si Etch Yield vs. Average FC Film Thickness
Typical Experimental Results
18
Model Study of Fluorocarbon Plasma Etching (Si)

• Si etch analogous to other non-O containing films
  (e.g. silicon nitride, photoresist)
• Role of FC film in etch similar for all materials
• Popular chemistry F-deficient (e.g. C4F8 C4F6
  C5F8, etc.), heavily diluted in Ar
• Model chemistries
• xCF2 yF Ar (200 eV)
• xC4F4 yF Ar (100, 200 eV)
• xCF yF Ar (100, 200 eV)

19
Sticky Precursors Yield Desired Result (Ar
200 eV Normal incidence)
Flux ratio (CxFy/F/Ar)
Etch Yield (Si/Ar)
Film Thickness (nm)
Case
Neutral specie (avg. KE)
i C4F4 (20 eV) F (300K) 551 0.053 1.67
ii C4F4 (20 eV) F (300K) 371 0.09 1.03
iii C4F4 (20 eV) F (300K) 641 0.039 2.21
iv C4F4 (20 eV) F (300K) 731 0.036 2.53
v C4F4 (20 eV) F (300K) 821 0.0 gt5.5
vi CF (5 eV) F (300K) 2051 0.087 1.16
vii CF (300K) F (300K) 9091 0.105 1.08
viii CF (300K) F (300K) 80191 0.199 0.63
20
Si Etch Yield vs. Average FC Film Thickness
Varying C4F4/F/Ar or CF/F/Ar ratios
ExperimentalResults
MD Simulation Results
21
Typical Snapshots Showing Fluctuations
Note layering of near-surface region, fluctuating
FC layer surface modification 4-5 nm.
22
Relatively Large Products Leave Surface
Role of FC clusters in plasma, emitted
by surface? Re-deposition of clusters/heavy
species?
23
FC Films are NOT All Alike!
Comparison of films deposited by CF on initially
bare Si with CF at (a) 300K, (b) 5eV, and (c)
100eV.
24
Conclusions FC Film Formed During Etch

• Stickiness of FC precursor important
• precursor C/F1 creates porous, fluffy, open
  FC film
• weakly cross-linked and low density FC film can
  be sputtered in clusters causes film thickness
  fluctuations
• F transports to substrate and products are
  removed through pores and film thickness
  fluctuations
• FC film thickness fluctuates as impacting ions
  occasionally remove clusters of FC assists etch
  product removal and enhances overall transport
• Ion impact and ion mixing still play central role
  in FC plasma etch with FC film present

25
Questions Inspired by Simulations

• Do surface fluctuations represent a future
  fundamental limitation to plasma etch pattern
  transfer fidelity?
• Random, nanometer-scale fluctuations can lead to
  differences between otherwise identical (even
  adjacent) features
• Electrostatic charging of fluctuating surface
  topography/composition might amplify fluctations,
  altering ion trajectories over larger distances.

Suggested by R.A. Gottscho, Lam Research
26
Thoughts Inspired by Simulations, continued

• Are surface fluctuations from point to point on
  the surface related to roughness?
• Similar ideas came up with polymer/organic film
  etch simulations
• No thermal diffusion of any species included in
  simulation could this be important for
  time-scales and length-scales of importance in
  etching? (Very likely...)

27
Plasma-Organic Film Interaction Mechanisms

• E.g. photoresist etch roughening mechanisms
• Organic masking layers for novel pattern transfer
• Nano-imprint lithography
• Block co-polymers
• Other applications involving organic films

28
How Do Organic Polymers Resist Etching?

• Organic polymers are soft and easily sputtered
• not obvious how they act as etch masks!
• Plasma dramatically modifies top surface layer
• First step in understanding etch/roughening
  mechanisms for organic films is to understand
  near-surface modifications due to plasma

MD study begun with simulated beam experiments
polystyrene/Ar then polystyrene/Ar/F
29
Experimental Technique

• UHV Chamber
• Base Pressure 5x10-8 Torr pumped with a 2000
  Ls-1 turbo pump
• PHI Model 04-191 Ion Gun
• Chamber pressure rises to
• 1x10-6 Torr
• Ions He, Ar and Xe
• Energies 0.2 - 1 keV
• Beam Size 0.5 cm
• Substrate temperature control
• Neutralizing filament to prevent
• surface charging

30
Evolution of sputter yield with fluence Ar
Rohm and Haas 193 nm photoresist
Mass Loss
Etch yield from slope
31
MD Simulation of Model Polystyrene Impacted with
100 eV Ar
dehydrogenated surface layer
ion penetration depth (nm)
transition region large changes in materials
properties over a very small thickness

undisturbed polymer
32
Comparison of steady-state sputtering yields

• Empirical formula etch yield is proportional to
  Ohnishi parameter

N total number of atoms in monomer NC number of
carbon atoms in monomer NO number of oxygen
atoms in monomer
33
Surface roughening of 193 nm photoresistion
energy and substrate temperature

• Xe bombardment ion energy and substrate
  temperature effect
• (fluence 1.3x1017 ionscm-2 for all samples)

1 mm
34
Surface roughening of 193 nm photoresistcomparis
on of Xe, Ar, and He bombardment

• (fluence 1.3x1017 ionscm-2 for all samples)

1 mm
35
MD 100 eV Ion Penetration and KE Deposition
2 nm
Ar
Ar (S.S.)
Xe
He
Polystyrene layers. 500 impact trajectories on
virgin PS surface. Trajectories shifted to the
same initial lateral location each trajectory is
color coded to the KE remaining in the ion at a
given position. Ar (S.S.) shows impacts on the
steady state (dehydrogenated) surface, indicating
greater scattering and shallower penetration for
a given ion.
36
Summary sputtering of polymers by rare gas ions,
normal incidence

• Polymer sputtering characterized by an initial
  high yield. A lower steady-state yield, similar
  to pure carbon, is reached after fluences of
  5x1016 ionscm-2.
• Steady-state yields of Ar bombardment follow the
  empirical Ohnishi parameter, taking into account
  inherent chemical effects of the polymer.
  Ohnishi parameter does not necessarily hold true
  in the presence of more complex etch chemistry.
• The amount of material removed prior to reaching
  steady-state is polymer dependent.
• more mass removed prior to reaching steady-state
  for 193 nm photoresist compared to 248 nm
  photoresist
• Ion beam etch yields consistent with argon plasma
  experiments.

37
Summary roughening of polymers by rare gas ions,
normal incidence

• Ion energy effect small increase in roughening
  from 200 eV to 1000 eV
• Ion mass effect on roughening Xe gt Ar gt He
• Substrate temperature effect increased roughness
  with increased substrate temperature (45C gt
  20C)

38
Current Vision Competing Mechanisms in Sputtering
Incoming Ions
CxHy Products
Ion Scattering
Dehydrogenation

• Transient periodcompetition between
  dehydrogenation/crosslinking and HC removal
• Large HC cluster ejection can remove components
  from the initial crosslink
• Once significant crosslinking/dehydrogenation
  occurs, large cluster ejection is hampered,
  dehydrogenation dominates

Crosslinking
Modified Layer
Undisturbed Polymer
39
Summary and Concluding Remarks Organic Film
Sputtering and Roughening

1. Virgin organic films are soft and sputter
   readily.
2. Rare gas ion bombardment (e.g. Ar) can create
   protective C-rich skin at surface, greatly
   reducing etch yield.
3. Plasma-generated reactive radicals (e.g. F) can
   attack and thin or remove skin, resulting in
   great increase in etch yield.
4. Scissioning polymer behaves differently in
   transient than cross-linking polymer ion
   bombardment decomposes scissioning polymer into
   monomer more than cross-linking polymer.
5. Relationship with observed greater roughening for
   scissioning polymer still speculative greater
   cluster ejection due to monomer decomposition?
   Greater MD cell-to-cell variation linked with
   more roughness? Cluster desorption related to
   observed greater roughness at higher surface
   temperature? Why do higher mass ions result in
   rougher sputtered films? Why does lower ion
   energy (between 1000 eV and 100 eV) result in
   rougher sputtered films?

40
Challenge to Connect Length Scales

• Experiments show roughness and structure on 10s
  nm 100s nm length scales.
• MD suggests smoothing on nm length scales.
• Need to extend theories to energies, materials
  and chemistries of interest to low temperature
  plasma processing studies, such as plasma
  etching.
• Nm-scales becoming important for plasma
  processing of semiconductor devices and related
  thin film products very helpful to improve
  understanding of fundamental phenomena.
• Couple nm-scale phenomena to feature scale
  through feature scale simulations

41
Acknowledgements

• Joe Vegh (PhD student, UC Berkeley)
• Dave Humbird (currently at Lam Research)
• Erwine Pargon (currently at LETI, Grenoble)
• Dustin Nest (PhD student, UC Berkeley)
• Harold Winters, John Coburn and Dave Fraser, UCB
• G. Oehrlein, et al. University of Maryland
• SRC CAIST
• NSF GOALI (DMR 0406120 )
• NSF NIRT (CTS-0506988)
• FLCC UC Discovery Grant from the
  Industry-University Cooperative Research Program
  (IUCRP)