Introduction to Plasma Etching

Dr. Steve Sirard

Technical Director
Lam Research Corporation

Lam Research Corp. 1

Day 1 Review – Plasma Fundamentals

► Plasmas consists of
electrons, ions,
neutrals, radiation
▪ ne ~ ni << ng (weakly
ionized)
Collisional Processes
-Ionization
+ e- -Dissociation ► Collisional
-Excitation processes sustain
e- the plasma and
h
create radicals
(etchant)
+ * + ▪ Electrons are very
e- *
hot
e-
+ + ► Sheaths form at the
walls/substrate to
confine electrons
and directionally
accelerate ions

Lam Research Corp. 2

Day 1 Review – Anisotropic Plasma Etching

+ + 0.015

Flux
0.010

+
Sheath (V ~ 10 – 1000V)

IonIED
0.005

0.000
0 20 40 60 80 100

Ion Energy (eV)

Ion Energy (eV)

+
SiO2(s) + CxFy + I+(Ei)  SiF4(g) + CO(g)

Vertical, anisotropic etch

Electrode

Synergy!

Lam Research Corp. 3

Day 2 - Outline

► Primary etching variables available to process engineers

► Common pattern transfer issues

► Advanced etch strategies: Pulsing strategies, Atomic layer etching

► Within-wafer etch uniformity control

► Plasma & surface diagnostics

Lam Research Corp. 4

What “knobs” are available to tune etch processes?
► Etching in general is very complex!

► Advanced plasma etch chambers are equipped with a lot of “knobs” for
controlling the etch process
▪ Wafer temperature
▪ Upper electrode temperature
▪ Temperature gradients
▪ Chamber pressure
▪ Gas chemistry (~20 gases on a chamber to choose from)
▪ Gas ratios (gas partial pressures)
▪ Gas flow rate (residence time)
▪ Total RF power
▪ Multiple RF excitation frequencies (up to 3 generators)
▪ Pulsing of RF powers (duty cycle, frequency)
▪ Pulsing of gases (duty cycle, frequency)
▪ Etch time
▪ Multiple uniformity knobs

► Overall, a tremendously large process space  long development cycles!

Lam Research Corp. 5
Key etch variables: Gas Chemistry

► Etchant gases (e.g., CF4, Cl2, C4F8)

▪ Provide reactants to etch materials of interest
▪ May provide polymer precursors for anisotropy, selectivity

► Oxidants (e.g., O2)

▪ Used to increase etchant concentrations
▪ Suppress polymer formation

► Radical-Scavengers (e.g., H2)

▪ Increase polymer formation, selectivity
▪ Reduce etchant concentration

► Inhibitor-Former (e.g., CH3F, CH4)

▪ Induce anisotropy
▪ Improve selectivity

► Inert Gases (e.g., Ar, He)

▪ Dilute etchant
▪ Stabilize plasma
▪ Enhance anisotropy
▪ Improve heat transfer
Lam Research Corp. 6
Key etch variables: Chamber Pressure

► Pressure directly influences major

phenomena that control plasma
etching Physical Sputtering

1. Energy of ions bombarding surfaces

Ion Energy
2. Ion-to-neutral ratio/flux Ion-assisted etching

3. Polymerization potential
Chemical plasma
4. Electron energies etching

5. Surface coverage by physisorption

0.001 0.01 0.1 1.0 10
Pressure (Torr)
6. Chemical kinetics

7. Relative rates of mass transport

processes

8. Etch rate uniformity across wafer

Lam Research Corp. 7

Key etch variables: RF Power/Frequency

► RF Power/Excitation Frequency
can influence:

1. Spatial distributions of species

across the plasma (Uniformity)

2. Ion energy distribution

3. Plasma density

4. Effects plasma chemistry

▪ For example, one frequency may
promote polymer deposition, while
another frequency may etch polymer

5. Selectivity

Lam Research Corp. 8

Key etch variables: Surface Temperatures

► Surface temperatures can effect:

1. Surface morphology

2. Polymer deposition

3. Selectivity

4. Resist flowing/wiggling/roughness 3-Frequency

5. Product volatility/Etch rate

6. Temperature gradients  Thermophoresis effects  Selectivity/Uniformity

impact
Tetch ~ Tg Tetch << Tg

Lam Research Corp. 9

Common Pattern Transfer Issues

Lam Research Corp. 10

Process Issues  Loading Effects: Macro-loading

► Macro-loading
▪ Etch rate for a given process becomes slower with more exposed etch area
— Due to overall depletion of reactant with more exposed area to etch

► Examples of macro-loading issues in industry

▪ Process shift on two different wafers with same features but different etchable
area
▪ Chamber etch rate drift due to build up of polymer on chamber parts during
etch processing

Lam Research Corp. 11

Macro-loading Example: Depletion of reactant with larger
exposed area
► Photoresist etch with oxidizing plasma chemistry
▪ When exposed area of resist is large, etch rate is reduced everywhere, but even more so
in the wafer center
▪ When exposed area of resist is small, etch rate is higher and more uniform
▪ Typical edge fast etch rate of resist is due to loading of plasma by large area of resist

PR coupons
3500

3000
PR etch rate (A/min)

2500
Si
2000

1500
PR coupons
1000

500 PR coupons on PR wafer

PR coupons on Si wafer
0
PR -150 -100 -50 0 50 100 150
Wafer position (mm)

Lam Research Corp. 12

Macro-loading Characteristics

► Etch rate is limited by the arrival of neutrals (neutral-limited regime)

► Macro-loading is a function of total exposed area reacting with gas phase

species

► Center-to-edge uniformity variations can be a result of macro-loading

► Compensation Strategies
▪ General process fine-tuning and uniformity compensation
▪ Increase etchant flux to make less neutral-limited (e.g., pressure, gas ratio change,
higher RF power to increase dissociation)

Lam Research Corp. 13

Loading effects: Micro-loading

► Macro-loading– etch rate becomes slower due to overall depletion of

reactant with more exposed etch area

► Micro-loading - etch difference between a given feature located in an

area of high density compared to the same feature in an area of low
density (isolated) on the same chip (assumes same nominal aspect
ratio)
▪ Due to local depletion of reactant

Isolated trench etches faster than

dense area
Lam Research Corp. 14
Is Micro-loading real??

► Unequal consumption across the chip results in lateral concentration

gradients
▪ However, significant gradients may require 100s of microns at low pressures

► Compensation
▪ Lower pressure  fewer collisions  higher diffusion rate  less micro-loading
▪ Increase etchant flux (e.g., Change gas ratio, increase source power)

Lam Research Corp. 15

Aspect Ratio Dependent Etching (ARDE)

► ARDE – etch rate becomes slower with higher aspect ratio or smaller
critical dimensions
▪ Sometimes the phenomena is also called “RIE Lag”

Aspect Ratio = Depth / Width

Lam Research Corp. 16

Primary ARDE Mechanisms

► Classical-ARDE: Higher aspect ratio features generally etch slower than

smaller aspect ratio features

► Four primary mechanisms used to explain ARDE:

1. Neutral shadowing

2. Ion shadowing

3. Differential charging

4. Knudsen transport

Lam Research Corp. 17

ARDE Mechanisms: Neutral Shadowing

► The neutral angular distribution is

isotropic

► Neutrals incident at a large angle to

the normal will hit the top or sidewall
of the feature, and can be lost (e.g.,
recombination) before reaching the
bottom of the feature
▪ They will be “shadowed” by the walls of
the feature
▪ Ion/neutral flux ratio is aspect ratio
dependent due to different angular
distributions
▪ High aspect ratio features may become
“starved” for neutrals (ER slows down in
HAR feature)

Lam Research Corp. 18

ARDE Mechanisms: Ion Shadowing

► The ion angular distribution is

generally anisotropic

► Higher pressures can cause ion

scattering in the sheath, causing
spread in ion energies/angular
distributions

► Ions incident at larger angles to the

normal hit the top or sidewall of the
feature, but not the bottom (i.e, they
are shadowed)

► Similar to neutral shadowing, ion

shadowing shifts the ion/neutral flux
ratio in different AR features

Lam Research Corp. 19

ARDE Mechanisms: Differential Charging

► Electron flux to the wafer

periodically occurs as the sheath
oscillates, and has a much less
anisotropic angular distribution

► Differential charging can result in

potentials large enough to deflect
ions (alters flux to feature bottom)

► Others have reported ion energy

drops of ~30% for AR ~ 3

Lam Research Corp. 20

ARDE Mechanisms: Knudsen transport

► Knudsen transport – Neutral reactants travel to the bottom of the feature by

being reflected diffusively from the sidewalls without reacting

► From Coburn & Winters, APL, (1989)

▪ JbSb = Jt – (1-)Jt – (1-Sb)Jb (describes gas fluxes into and out of feature)
— Jb, Jt = flux to the bottom or top of feature
— Sb = reaction probability at the bottom of the feature
—  = transmission probability (decreases with increasing aspect ratio)

Etch Rate Ratio (bottom/top) vs Aspect Ratio

Lam Research Corp. 21

What about Inverse-ARDE (or Reverse-RIE Lag)?

Classical ARDE Inverse ARDE

Lam Research Corp. 22

Inverse-ARDE (or Reverse-RIE Lag)

► Mechanism  Polymer-precursor shadowing

▪ Less polymer forms in high aspect ratio feature, thus higher etch rate

More polymer
Shadowing gives deposition here
less polymer
deposition here

Lam Research Corp. 23

How do we fix ARDE??

► Multiple mechanisms can lead to ARDE in plasma etching

▪ Neutral shadowing
▪ Ion shadowing
▪ Differential charging
▪ Knudsen transport

► Solution to ARDE issues can depend on which mechanism(s) is/are responsible

► In previous dielectric etch study, we observed that differential charging was a

primary mechanism for classical ARDE
▪ To mitigate, low pressures improve (as long as not too low  neutral limited)
▪ Higher RF powers (higher ion energies) improve
▪ Balancing the etchant/deposition flux can also compensate (gas chemistry, gas ratios)
▪ RF pulsing/atomic layer etching improves

► Inverse-ARDE is related to shadowing of polymer precursors in narrow

features
▪ Shift etch chemistry to less polymerizing condition (increase F/C ratio, add O 2, etc)

Lam Research Corp. 24

Other issues: Faceting and Necking Strasser, JVSTB, 2007
Lee, JECS, 2010

► Faceting ► Necking
▪ Generally due to increased yield per ion ▪ Can be due to heavy polymer deposition
at a corner (Dependent on ion at the top of the contact or from re-
energy/flux) deposition of polymer precursors,
forward scattered from photoresist
▪ Position of neck may be dependent on
the angle in the resist

Lam Research Corp. 25

Bowing

► Bowingof the feature sidewall can

have several root causes
▪ Ion scattering from the resist mask
(dependent on facet angle)
▪ Ion scattering in the sheath (lower
pressure may help)
▪ Too much oxygen in the process (less
sidewall polymer protection, leads to
more isotropic etch)
▪ Exacerbated by polymers that deposit
well at the top of features, but deposit
poorly deeper down the feature

Lam Research Corp. 26

 Clarycon
Microtrenching Schaepkens, APL, (1998)
Bogart, JVSTA, (2000)

► Microtrenching- Localized higher etch rate at

bottom corners of trench

► Potential Mechanisms
1. Ion scattering from sloped trench sidewalls
2. Ion deflection due to differential charging of
microstructures

Lam Research Corp. 27

Striations

► Extended roughness on the

sidewalls of etched features
▪ Seen on both holes and trenches

► Due to roughness/striation
formation in the resist being
transferred to underlayers

► Exacerbated with 193nm resists

► Likely related to how plasma

modifies the resists at different
length scales
▪ Ions impact ~2nm of surface, causing
graphitic-densified region
▪ VUV radiation may either chain scission
or cross-link at deeper depths (~100nm)
▪ Different mechanical properties of
modified layers can lead to resist
buckling or roughening

Lam Research Corp. 28

Advanced Etch Strategies

Lam Research Corp. 29

RF Pulsing Can Lower Ratio of High:Low Ion Energy Flux

CW 2 MHz Bias

CW 60 MHz Source  Pulsing reduces the ratio of high to low

energy ions (constant peak power)
– [Ghi/Glo]pulsed < [Ghi/Glo]cw

Pulsed 2 MHz Bias

CW 60 MHz Source

 RF pulsing can be used to access energy

distributions not available in continuous
operation

RF pulsing can access new process regimes to break existing tradeoffs

Lam Research Corp. 30

RF Pulsing Controls Neutral:Ion Flux

Different decay time constants allows to vary Ion/Neutral flux over larger range

Pulsing improves selectivity and controls ARDE

Slide - 31 Lam Research Confidential

Bias pulsing can improve line edge roughness

On just turned off Off

Neutral
Passivant
Ion
Passivants
Ions sputter
build-up, less
(can facet, lower repeat
LER, better
selectivity, worse Neutrals
selectivity
LER ) saturated,
High bias for reduces intra-
increased etch cell loading
rate, less bow Reactant still
etching

3x Improvement on LER observed

Lam Research Corp. 32

Atomic Layer Etching
How can we achieve a more precise etch??

Continuous Wave Plasma Processing

3 – 5 nm
Time

► Atthe atomic scale, continuous plasmas “surface precision” degrades

as we etch for longer times
▪ Ions are damaging surface
▪ Mixing occurring in the reaction layer (~3 – 5 nm thick)

Lam Research Corp. 33

Directional “Atomic Layer” Processing: The Penultimate
Means of Reducing Variability

Start Reaction A Switch Steps Reaction B End/Repeat

Ar+
Etch Cl2
Si Surface

Si ALE

Chlorination Ion Bombardment

Si Source
Deposition O Radical
SiO2 Added

SiO2 ALD

Adsorption Adsorption

► Use separate, self-limiting steps for atomic layer removal

Lam Research Corp. 34
Avoid Use of Energetic Reactive Ions to Achieve “Atomic”
Layer Precision

Cl+ 50 eV Vs. Ar+ 50 eV

~0.5 nm
Efficient in Inert ions create
breaking bonds, disordered regions near

~3 nm
creating more surface and also re-
mixing and crystallize these
disorder regions
Si Si
Source: Humbird and Graves, 2004

• Roughening effect, inhomogeneity • Smoothing effect

• Used in conventional etching: • Used in directional ALE schemes:

10 nm Source: 10 nm
Si Lam, 2010 Si

Source: Agarwal &

1 nm
Kushner, 2007

Lam Research Corp. 35

ALE Process Window

Ion Energy too low Ideal Regime Excessive ion

energy/bombardment

Incomplete Removal ALE Window Sputtering

1. Saturate the surface with

reactant (without ion
bombardment)

2. Bombard surface with ions at the

proper ion energy, in the
absence of excess reactant

Lam Research Corp. 36

Profile benefits of using separated and self-limiting steps

Conventional ALD
Deposition

Conventional Etch ALE

Lam Research Corp. 37
Other benefits from separated and self-limiting steps

Lam Research Corp. 38

Atomic layer etch process Limitations
“But I am not afraid to consider the final question as to whether, ultimately---in the great
future---we can arrange the atoms the way we want; the very atoms, all the way down!
-Richard P. Feynman, Dec 29th 1959 at the annual meeting of American Physical Society

ALD/ALE: Feynman’s dream…

► Limitations from the process:
▪ Photon-induced etching and
damage
▪ Ion-induced mixing and damage
even with inert ions at low energies
▪ Steric hindrance preventing fully
saturated coverage
ALD/ALE: Penultimate reality
▪ Adsorbed reactants on chamber
walls compromise separation of
dosing and activation steps
▪ Trade-off between atomic layer
processing benefits and throughput

Lam Research Corp. 39

Etch Uniformity Control

Lam Research Corp. 40

Etch uniformity control

► Leading edge wafer fabrication processes require very tight uniformity control
across the wafer (Out to <3mm from the edge of the wafer; Require specs ~
<1 nm 3-sigma)
▪ Uniformity in etch depths
▪ Uniformity in critical dimensions

► Center-to-edge uniformity in plasma processes are difficult due to

discontinuities that occur at the wafer edge
▪ Impact on local chemistry and ion flux
▪ Temperature gradients Plasma

Wafer

► Advanced etch tools use a variety of compensation strategies to achieve

uniformity across the wafer

Lam Research Corp. 41

Adjusting the gap distance between upper and lower
electrodes tunes the plasma uniformity across the wafer

Lam Research Corp. 42

 For ion-limited etch regimes, etch rates can be flipped from
center-fast to edge-fast by changing gap distance
Pilot™ wafer measurement results
CF4/Ar process – Ar/CF4 process
150 28 At wafer center
148
Etch rate changes by ~2.1%
146

144 27 Ion flux changes by ~2.2%

TOX ER (nm/min.)

142

140

Isat (uA)
138
26
136

134 At wafer edge

132
25
130 Etch rate changes by ~-8%
128
Ion flux changes by ~-7%
126
24
-150 -100 -50 0 50 100 150
1.54 gap
Wafer radius (mm)
1.2 gap
0.85 gap
Etch rate
Isat

(v3)
Slide - 43 Lam Research Confidential
Gap changes can also impact etch uniformity of neutral-
limited processes

Lam Research Corp. 44

 Effective Tuning Knobs for ER and CD Uniformity for In situ Mask +
Dielectric Etching

PR
Typical Operation Mode

ARC / Si-SOC
High gap (>1.07) TEM

Organic + tuning gas

Low K
Low gap (0.85)
+ tuning gas

SiCN
Medium gap (1.07)
+ tuning gas

Sub-Si

Enables all in one etching with optimized uniformity in each step

(v3)
Slide - 45 Lam Research Confidential
 PR
Case study: Effect of gap and tuning gas on CD BARC
SiON
uniformity for in-situ mask + dielectric etching
Case study: CD uniformity improvement with adjustable
ACLgap
and tuning gas Oxide
Example: Line CD Uniformity Data on Patterned Wafers
W
70 7
60 6
50 5
CD (nm)

3s (nm)
40 4
30 3
20 2
10 1
0 0
BARC(1.07gap)/ACL(1.

BARC(1.07gap)/ACL(0.

BARC(1.07gap)/ACL(0.
BARC 1.07gap
07)/PEOX(1.07)

85)/PEOX(1.07)

85)/PEOX(0.98)
1.07gap 1.07gap

ACL 1.07gap 0.85gap 0.85gap

OXIDE 1.07gap 1.07gap 0.98gap

While maintaining the average CD values, uniformity is improved by

optimizing the gap height for each step.

(v3)
Slide - 46 Lam Research Confidential
Increased temperature control across the wafer

Lam Research Corp. 47

Multi-zone electrostatic chuck (ESC) temperature control
allows precise etch rate tuning

Lam Research Corp. 48

Plasma & Surface Diagnostics

Lam Research Corp. 49

Plasma & Surface Diagnostics

► Because of the complexity of the chemical and physical environments

in a plasma, a large array of techniques are required to characterize
them

► Parameters that are of normal interest

▪ Electron and ion densities
▪ Neutral densities (both species created in the plasma and byproducts of etch
process)
▪ Respective temperatures of these species
▪ Energy distributions
▪ Characterization of electric fields
▪ Surface modification
▪ Relative importance of different species within plasma

Lam Research Corp. 50

Plasma Diagnostics

► Langmuir Probe – Conducting wire

placed in the plasma with a variable
bias, V, applied. The current (I) is
measured as a function of V.
▪ What can you measure?
—Floating potential
—Plasma potential
—Plasma density
—Ion current density
—EEDF/Electron temperature

Lam Research Corp. 51

Plasma Diagnostics

► Residual Gas Analyzers – uses mass spec to

analyze gases present in vacuum environment
▪ What can you measure?
— Gas species analysis/leak detection
— Insight into reaction mechanisms

Snapshot OES of N2/H2 Plasma

► Optical
Emission Spectroscopy (OES) –
examines photon emissions from the plasma 60
N2/H2 process

▪ Characteristic set of wavelengths emitted for 50

a given species 40

OES (arb. units)

▪ What can you measure? 30

— Monitor species in the plasma 20

— End-pointing etch process 10

300 400 500 600 700 800

Wavelength (nm)

Lam Research Corp. 52

Plasma Diagnostics – Optical Emissions

► OES – continued
▪ Endpointing – measure emission from a key species as a function of time to
determine when to stop etch
▪ Troubleshoot chamber drifts

Endpoint Detection Endpoint Time Trend Charts

Endpoint Chamber excursion?
Chamber drifting?
Bad wafer?

Lam Research Corp. 53

Plasma Diagnostics

► Actinometry using OES – add small/known amount of noble gas (e.g., Ar) to a
reactive plasma and concurrently monitor the emissions of the noble gas and
the reactive species (e.g., F).
▪ Infer densities of certain species (e.g., [F])
▪ Many assumptions have to be met, so be careful
— Excited state emission energy of noble gas is similar to emission in reactive species
— Same group of electrons responsible for excitation of both levels
— Excitation efficiencies of these levels will have similar dependence on plasma parameters

► Absorption Spectroscopy – use stable, well-defined light source to excite

radicals from ground electronic state
▪ What can we measure?
— Measure absolute densities of plasma species

Lam Research Corp. 54

Plasma/Surface diagnostics – Shaded structures

► Shadedstructures – structures or windows that are often placed on

the substrate to “shade” or block out targeted species from the
plasma
▪ What can be measured?
— Separate out the effects of radicals, radiation, and ions

Impact of Radicals, UV/VUV, and

Impact of Radicals (block ions, radiation) Radical+UV/VUV

neutrals ions

Si roof
0.6 mm gap
Si wafer
Measure neutral-driven
deposition in ion-shaded
region

Lam Research Corp. 55

Surface Diagnostics

► Spectroscopic ellipsometry
▪ Measure optical constants and film thickness
▪ Used ex-situ and in-situ to study plasma effects/kinetics

► IR absorption
▪ Provide chemical identification

► XPS
▪ Surface elemental analysis
▪ Chemical identification (bonding info, etc)
▪ Adsorbate coverage

► Spinning Wall (Donnelly group at UH)

▪ Elemental analysis and chemical identification (with proper diagnostic)
▪ Adsorbate coverages
▪ Recombination probabilities

► Plasmas causes a lot of damage to near surface region, so often will see
significant oxidation when performing ex situ surface analysis
Lam Research Corp. 56
Check us out for career and internship opportunities!

Bay Area, CA (Headquarters) Portland, OR

Villach, Austria

www.lamresearch.com

Stephen.Sirard@lamresearch.com

Lam Research Corp. 57