critical dimension control -...
TRANSCRIPT
10/6/2003 FLCC Research Seminar 1
Critical Dimension Control
FLCC Research Seminar 10/06/03Costas J. Spanos
10/6/2003 FLCC Research Seminar 2
Outline• What is CD and why it matters• Anatomy of CDU • Present Control Status• The Post Exposure Bake factor• Plasma Etch CDU Opportunities• A combined litho/plasma controller framework?• What is next
10/6/2003 FLCC Research Seminar 3
What is the Critical Dimension?CD
Gate is typically made out of Polysilicon
10/6/2003 FLCC Research Seminar 4
Basic CD Economics
(D. Gerold et al, Sematech AEC/APC, Sept 97, Lake Tahoe, NV)
Today’s data suggests a “price” of more than $10/nm…Today’s data suggests a “price” of more than $10/nm…
10/6/2003 FLCC Research Seminar 5
Why was this Improvement Important?
600k APC investment, recovered in two days...
10/6/2003 FLCC Research Seminar 7
Projected Parametric Variation in Future Nodes
(assuming fixed overall % in CD variation)Delay and Energy variation caused by lint variation -
8 Bit Mirror Adder
0
10
20
30
40
50
60
70
80
0 50 100 150 200Technology Node (nm)
% V
aria
tion
caus
ed b
y lin
t
Linear (% Delay variation caused by lint variation )Linear (% Energy variation caused by lint variation )
10/6/2003 FLCC Research Seminar 8
Outline• What is CD and why it matters• Anatomy of CDU• Present Control Status• The Post Exposure Bake factor• Plasma Etch CDU Opportunities• A combined litho/plasma controller framework?• What is next
10/6/2003 FLCC Research Seminar 9
Statistical Metrology for CDs
Spatial
Random/Deterministic
Within Die
Among Die
Among Wafers
Causal
Equipment Recipe
?Other
Response from Equipment A
Random/Deterministic
10/6/2003 FLCC Research Seminar 10
Nested Variance – A Simple Model
CDijkl = µ + fi + wj + lk + Ll + σTrue mean
Across-field
Across-wafer
Across-lot
Lot-to-lot
Random
CD variation can be thought of as nested systematic variations about a true mean:
Field
Wafer
Lot
Jason Cain, SPIE 2003
10/6/2003 FLCC Research Seminar 11
Across-Wafer Systematic Variation
- -
=
Average Wafer Scaled Mask Errors Across-Field Systematic Variation
Across-Wafer Systematic Variation
Separation of Across-Field and Across-Wafer Systematic Variation:σ2 = (3.28 nm)2 σ2 = (1.77 nm)2 σ2 = (1.23 nm)2
σ2 = (2.46 nm)2
10/6/2003 FLCC Research Seminar 12
Across-Wafer Systematic VariationThe remaining across-wafer systematic variation was removed using a polynomial model:
CD(Xw,Yw) = aXw2 + bYw
2 + cXw + dYw + eXwYw
- =
Across-Wafer Systematic Variation with bivariate
Gaussian removed
Polynomial Model Remaining Across-Wafer Systematic Variation
σ2 = (2.39 nm)2 σ2 = (1.39 nm)2 σ2 = (1.95 nm)2
10/6/2003 FLCC Research Seminar 13
Systematic Variation Model Summary• Across-field systematic variation:
– Mask errors (magnified scaling factor) removed– Polynomial terms in X and Y across the field
• Across-wafer systematic variation:– Across-field systematic variation removed– Bivariate Gaussian for spot effect (likely linked to develop effect)– Polynomial terms in X and Y across the wafer
Field WaferMetrology?Process replication?
10/6/2003 FLCC Research Seminar 14
CD Variability Summary• Lot to Lot: controlled by “standard” EWMA
controllers. – Accounts for less than ~1/3 of CDU.
• Within Lot: not actively controlled.– Rapid equipment drift. Exposure to PEB timing. PEB
Bake Oven match. Etch chamber match.• Across Wafer: not actively controlled.
– Stepper stage, PEB plate uniformity, Develop uniformity, etch chamber uniformity.
• Across Field: not actively controlled.– Mask manufacture, optics, mask error factor, feature-
level compensation (OPC).
10/6/2003 FLCC Research Seminar 15
Outline• What is CD and why it matters• Anatomy of CDU • Present Control Status• The Post Exposure Bake factor• Plasma Etch CDU Opportunities• A combined litho/plasma controller framework?• What is next
10/6/2003 FLCC Research Seminar 16
The Workcell ControllerMost process steps are so interrelated that must be controlled together using feed forward and feedback loops.Crucial pieces of equipment must by controlled by SPC throughoutthis operation.
EquipmentModel
Step
EquipmentModel
Step
AdaptiveControl
AdaptiveControl
test test test
10/6/2003 FLCC Research Seminar 19
Results for 27k-entry Library with external PEB sensors
Foc. FittingR2 = 0.9955
-0.5
-0.3
-0.1
0.1
0.3
0.5
-0.5 -0.3 -0.1 0.1 0.3 0.5
Actual Foc.
Estim
ated
Foc
.
Foc. FittingR2 = 0.9955
-0.5
-0.3
-0.1
0.1
0.3
0.5
-0.5 -0.3 -0.1 0.1 0.3 0.5
Actual Foc.
Estim
ated
Foc
.
PEB Tim e Fitting R2 = 0.9733
57585960616263
57 58 59 60 61 62 63
Actual Time
Estim
ated
Tim
e
PEB Tim e Fitting R2 = 0.9733
57585960616263
57 58 59 60 61 62 63
Actual Time
Estim
ated
Tim
e
PEB Tem p Fitting R2 = 0.9868
128
129
130
131
132
128 129 130 131 132
Actual Tem p
Estim
ated
Tem
p
PEB Tem p Fitting R2 = 0.9868
128
129
130
131
132
128 129 130 131 132
Actual Tem p
Estim
ated
Tem
p
Exposure FittingR2 = 0.9849
28.0
28.5
29.0
29.5
30.0
28.0 28.5 29.0 29.5 30.0
Actual Exp.
Estim
ated
Exp
.
Exposure FittingR2 = 0.9849
28.0
28.5
29.0
29.5
30.0
28.0 28.5 29.0 29.5 30.0
Actual Exp.
Estim
ated
Exp
.
10/6/2003 FLCC Research Seminar 20
A Profile Inversion Scheme?
Spin Coat, PAB
PEB, Develop SSSExposure
Profile Inversion Engine
10/6/2003 FLCC Research Seminar 21
Outline• What is CD and why it matters• Anatomy of CDU • Present Control Status• The Post Exposure Bake factor• Plasma Etch CDU Opportunities• A combined litho/plasma controller framework?• What is next
10/6/2003 FLCC Research Seminar 22
PEB Evolved from a Single Zone to Multi-Zone Control System Why?
Multi-Zone ControlSingle Zone Control
8 Years of Product
Evolution
PEB Track Equipment EvolutionComplexity is Increasing
10/6/2003 FLCC Research Seminar 23
Monitoring Plate Drift
Spread 0.272 °C One week later Spread 0.384 °C
Same wafer, same plate
No appreciable drift in mean
10/6/2003 FLCC Research Seminar 24
Auto-Calibration ResultsSteady State Range Optimization
Settings in Use After OnWafer Auto-Calibration
0.506 °C 0.277 °CTwo
OnWafer “missions”
10/6/2003 FLCC Research Seminar 25
Across plate CD map improvement Example
95 96 97 98 99 100
95 96 97 98 99 100
3-21 10nmC
-150
-100
-50
0
50
100
150
y
-150 -100 -50 0 50 100 150
x
Contour Plot
-150
-100
-50
0
50
100
150
y
-150 -100 -50 0 50 100 150
x
Contour Plot
Target
before
after
CD Map
CD Map
10/6/2003 FLCC Research Seminar 26
90 91 92 93 94 95 96 97 98 99 100
90 91 92 93 94 95 96 97 98 99 100
10nm Corr
Aggregate Improvement on several plates(achieved ~60% sigma reduction)
Target
before
after
10/6/2003 FLCC Research Seminar 27
But how about the transient PEB impact?
P6P6P5P5P4P4P3P3P2P2P1P1
Six plates used for 90nm CD lines (193nm Lithography)Six plates used for 90nm CD lines (193nm Lithography)
10/6/2003 FLCC Research Seminar 28
CD Spread vs “Overshoot” Spreadcorrected for “Cooling Rate Spread”.
2
4
6
8
10
12
Thre
e-S
igm
a Le
vera
ge R
esid
uals
1.50 1.75 2.00 2.25 2.50Overshoot Ranges Leverage, P<.0001
45
6
1
3
2
10/6/2003 FLCC Research Seminar 30
Auto-Calibration ResultsTransient Range Optimization
1.859 °C 0.715 °C
Settings in Use After OnWafer Auto-Calibration
Two OnWafer
“missions”
10/6/2003 FLCC Research Seminar 31
Outline• What is CD and why it matters• Anatomy of CDU • Present Control Status• The Post Exposure Bake factor• Plasma Etch CDU Opportunities• A combined litho/plasma controller framework?• What is next
10/6/2003 FLCC Research Seminar 32
The CDU opportunities in plasma…
107 105
105105
Note how the three runs on 105 have the same sloppy shape, while the one run on 107 is flatter…
10/6/2003 FLCC Research Seminar 33
CDU opportunities in plasma
“before” data is hotter, further, the pre-etch step is significantly less uniform…“before” data is hotter, further, the pre-etch step is significantly less uniform…
10/6/2003 FLCC Research Seminar 34
CDU Opportunities in Plasma:balancing the He zones
beforebefore afterafter
10/6/2003 FLCC Research Seminar 35
CDU Opportunities in Plasma Chamber A vs Chamber C
Main Etch A Main Etch C
Post Etch CPost Etch A
10/6/2003 FLCC Research Seminar 36
Outline
• What is CD and why it matters• Anatomy of CDU • Present Control Status• The Post Exposure Bake factor• Plasma Etch CDU Opportunities• A combined litho/plasma controller framework?• What is next
10/6/2003 FLCC Research Seminar 37
Present Status of “Active” CD Control
Spin
HMDS
PA Bake
Exposure
PEB
Develop
PD Bake
Photoresist Removal
ELMELM
Poly Etch System
ADIADI AEIAEI
Etch
Etch
Etch
10/6/2003 FLCC Research Seminar 38
On-wafer and in-situ metrology in pattern transfer
Spin
HMDS
PA Bake
Exposure
PEB
Develop
PD Bake
Photoresist Removal
ELMELM
Poly Etch SystemEtch
Etch
Etch
T (t, x, y)T (t, x, y)
T (t, x, y)V (t, x, y)E (t, x, y)…
T (t, x, y)V (t, x, y)E (t, x, y)…
I (x, y)I (x, y)
OCDOCD
Thin FilmThin Film
OCDOCD OCDOCD
10/6/2003 FLCC Research Seminar 39
New style CDU control has to incorporate many strategies
Spin
HMDS
PA Bake
Exposure
PEB
Develop
PD Bake
Photoresist Removal
ELMELM
Poly Etch SystemEtch
Etch
Etch
T (t, x, y)FF control
T (t, x, y)FF control
T (t, x, y)V (t, x, y)E (t, x, y)FF/FB Control, chuck diagnostics
T (t, x, y)V (t, x, y)E (t, x, y)FF/FB Control, chuck diagnostics
I (x, y)Optimal Pattern Design
I (x, y)Optimal Pattern Design
OCDProfile Inversion
FB Control
OCDProfile Inversion
FB Control
Thin FilmFB/FF Control
Thin FilmFB/FF Control
OCDFB Control
OCDFB Control
OCDFB/FF Control
OCDFB/FF Control
10/6/2003 FLCC Research Seminar 40
What is the ultimate CDU?
• With OCD, and PEB FF control, across wafer 3σ of 2nm (300mm wafers, 90nm lines, 193nm litho) has been demonstrated.
• Through transient control and OCD, we expect to remove another 0.5nm from the 3σ value.
• State of the art 3σ of 1.0~1.5nm CDU is possible if plasma chamber matching is included in the mix.
• All of these techniques are scalable, and will be amortized (andneeded!) for EUV.
10/6/2003 FLCC Research Seminar 41
Outline• What is CD and why it matters• Anatomy of CDU • Present Control Status• The Post Exposure Bake factor• Plasma Etch CDU Opportunities• A combined litho/plasma controller framework?• What is next
10/6/2003 FLCC Research Seminar 42
What remains to be seen• Will the addition of more autonomous sensors help?
– On-wafer aerial image sensor– On-wafer multi factor plasma sensor
• What will be the impact on mask fabrication uniformity?• Will “profile” inversion needs finally bring first principle
modeling into routine manufacturing operation?• Will these ideas be amortized beyond pattern transfer?