View
223
Download
4
Category
Preview:
Citation preview
8:30 – 9:00 Research and Educational Objectives / Spanos
9:00 – 9:50 Plasma, Diffusion / 9:00 – 9:50 Plasma, Diffusion / Graves, Lieberman, Cheung, HallerGraves, Lieberman, Cheung, Haller
9:50 – 10:10 break10:10 – 11:00 Lithography / Spanos, Neureuther, Bokor 11:00 – 11:50 Sensors & Metrology / Aydil, Poolla, Smith, Dunn
12:00 – 1:00 lunch 1:00 – 1:50 CMP / Dornfeld, Talbot, Spanos
1:50 – 2:40 Integration and Control / Poolla, Spanos
2:40 – 4:30 Poster Session and Discussion, 411, 611, 651 Soda 3:30 – 4:30 Steering Committee Meeting in room 373 Soda 4:30 – 5:30 Feedback Session
3rd Annual SFR Workshop, November 8, 2000
11/08/2000
2
UC SMART Major Program AwardD. Graves, E. Haller, M. Lieberman, and N. Cheung
SFR Workshop11/8/00
Plasma and Diffusion Studies
11/08/2000
3
Outline
• Plasma-Surface Interactions and Modeling (Graves)• Diffusion in Isotopically Controlled Silicon (Haller)• Plasma Sources for SFR (Lieberman)• Plasma-Assisted Surface Modification: Low
Temperature Bonding and Layer Transfer (Cheung)
11/08/2000
4
Plasma-Surface Interactions and Plasma Modeling
David Graves, Frank Greer, Dave Humbird, John Coburn, and Dave Fraser
University of California Berkeley
SFR WorkshopNovember 8, 2000
Berkeley, CA
11/08/2000
5
Radical Enhanced Atomic Layer Chemical Vapor Deposition
(REALCVD)
Frank Greer, John Coburn, David Fraser, David Graves
SFR WorkshopNovember 8, 2000
11/08/2000
6
What is A.L.D.?• Atomic Layer Deposition
– A.K.A. – Sequential Deposition, Digital Deposition, …
– Usually a two step process• Chemisorption of a metallic precursor
• Reactive ligand stripping by reactive stable molecule
– Distinguishing feature Each step is self-limiting
– Film thickness controlled by number of cycles
M
~1 mL of M-Lx
~1 mL of M-Y
L
L
L
L
YS
S
S
S
L SL S
11/08/2000
7
Likely A.L.D. Applications
• High-k Gate Dielectrics– Highly uniform deposition
– Atomic layer control of interface
• Nitride Diffusion Barriers– Highly conformal deposition
• Theoretically 100% step coverage
11/08/2000
8
Barrier Layer Deposition by A.L.D.
Key Issues:Barrier Properties
ConductivityConformality
Thermal budget**
Current barrier/seed films deposited via i-PVDchallenged by high aspect ratios
Nitride diffusionbarrier
Low k ILD
11/08/2000
9
Radical Enhanced Atomic Layer CVD (REAL CVD)1
• Uses a volatile precursor and a radical
source to deposit a film
• Reactants introduced in separate steps to
achieve atomic layer control
• Products are nitride film and HCl
• Radical flux instead of heated substrate
catalyzes precursor decomposition
– Potentially lower processing temperatures
1A. Sherman U.S. Patent 1999.
Wafer
Heated (?) Pedestal
Precursor/RadicalInlet
Step 2: TiClx (ab) + XH*(g) + N*
(g) TiN (s) + HCl(g)
Step 1: TiCl4 (g)
11/08/2000
10
Surface Reaction Products
Atom Source
Schematic of the Beam Apparatus in Cross-section (Top View)
Quadrupole MassSpectrometer (QMS)
Main Chamber
Analysis SectionRotatable Carousel
+Experimental Diagnostics1. QCM - Measures mass change of film2. QMS
- Measures products formed on film- Characterize beams
Silicon-coatedQuartz Crystal
Microbalance (QCM)
Ion Source+
Chopper
X
TiCl4
Precursor Doser
X = O, N , D
Load Lock
Future in-vacuo Auger
Analysis
11/08/2000
11
Experimental Procedure
• Conventional ALD-like sequence– Each step in process monitored in-situ with QCM
– Cycle used multiple times to grow a film • Surface temp varied (32, 80, 135 oC)
TiCl4
~1 mL of TiClx
1. Adsorption
2. Pump down
D
~1 mL of Ti-Dx
3. Dechlorination
4. Pump down
5. Nitridation
N
~1 mL of Ti-Nx
11/08/2000
12
0
2E+14
4E+14
6E+14
8E+14
1E+15
1.2E+15
0 5000 10000 15000 20000
Precursor Adsorption QCM Results
• Two distinct uptake regimes– Initial rapid ads. followed by
significantly slower ads.
• Adsorption may not be confined to a single monolayer
• Data fairly well represented by assuming precursor adsorbs two monolayers of TiCl2 *
• Model fit allows estimation of:– Sticking probabilities
– Active site (*) densities*
TiClx
*
* Suntola, T. et al. 1996
Uptake Upon Exposure to TiCl4
TiC
l 2 M
olec
ules
/cm
2
TiCl4 Exposure (L)
TiCl4
Adsorption
1 L = 1015 Molecules
11/08/2000
13
Adsorption Model Results
• initial sticking ~ 100x self sticking– contributes to uniformity
• self sticking decreases with inc. T– coverage is more monolayer-like at higher
temperatures
TiCl4
5x10146x10-58x10-3135
5x10149x10-58x10-380
5x10141x10-48x10-332
Site Density (#/cm2)
s(TiClx)
self
sticking
s(TiN)
initial
sticking
Temperature
(oC)
0
2E+14
4E+14
6E+14
8E+14
1E+15
1.2E+15
0 2000 4000 6000 8000 10000
Exposure (L)
TiC
l2 m
ole
cule
s/cm
2 Raw Adsorption DataModel Fit to Data
11/08/2000
14
0.00E+00
4.00E+14
8.00E+14
1.20E+15
1.60E+15
2.00E+15
0 2000 4000 6000 8000 10000 12000
Raw DataModel Fit
Dechlorination Results
• Cl loss due to abstraction by D atoms• ex-situ XPS analysis shows ave. Cl conc. = 2.2% (0.6%-3.8%)*
D
Cl
Cl Lost Upon Exposure to DeuteriumQCM
Dechlorination
• Exponential decay in Cl/cm2
fit assuming replacement of Cl
with D
• kD+ClDCl = 3x10-4
• Fairly insensitive to T
• Calculated removed Cl
density within a few% of ads.
assumptions
• Suggests TiCl2 is appropriate
surface species
Cl A
tom
s/cm
2
D atom Exposure (L) * Process not optimized for chlorine removal
11/08/2000
15
0
2E+15
4E+15
6E+15
8E+15
1E+16
0 500 1000 1500
Raw DataModel Fit
Nitrogenation ResultsN
D
Uptake Upon Exposure to N atoms
QCM
Nitrogenation
N A
tom
s/cm
2
Nitrogen Atom Exposure (L)
• Nitrogen uptake does not
saturate!
• ex-situ XPS shows N/Ti > 3
• Initial uptake of nitrogen fit to
linear model
• sN = 8x10-3
• fairly insensitive to T
• uptake slows over time
suggesting diffusional
limitations into the film
11/08/2000
16
Summary and Implications
1. TiCl4 has two adsorption regimes at these temperatures:• Initial rapid uptake
(s1=8x10-3)• Slower continuing uptake
(s2~8x10-5)• Change likely coincides
with 1st monolayer coverage
2. D atoms remove Cl from surface:
= 3 x10-4
• ex-situ XPS shows not all Cl removed from surface
Summary of Results• Cl content of films
– TiCl4 overexposure
– Insufficient D exposure
0
2E+14
4E+14
6E+14
8E+14
1E+15
1.2E+15
0 5000 10000
Exposure (L)
TiC
l2 m
ole
cule
s/cm
2
Raw Adsorption DataModel Fit to Data
11/08/2000
17
Summary and Implications
3. N atoms react with surfaceInitial Reaction
Probability:s = 8x10-3
• Rate slows over time, but does not stop
• Likely becomes diffusion limited
Summary of Results
H N
• Low rxn. probabilities for all species• Correct comparison for processing is rate of transport of radicals to feature bottom with rate of
reaction on feature sides• Transport rate > Reaction rate for aspect ratio ~ 5
– uniform rxn. rates throughoutfeatures
• As aspect ratio increases, transport rate will slow– could lead to non-uniform deposition
11/08/2000
18
Milestones
• 2001
– Design and test precursor doser, conduct preliminary film deposition studies
– Test self-limiting film formation kinetics
• 2002
– study effects of surface type, preparation and temperature on film properties
– measure radical-surface kinetics
• 2003
– apply method for patterned barrier film application
– Explore DRAM capacitor applications
11/08/2000
19
Damage Profiles of Low-Energy Ion Bombardment:
Application to Ion Milling
Dave Humbird, David GravesSFR Workshop
November 8, 2000Berkeley, CA
11/08/2000
20
Motivation
• Theory suggests that ion energy deposition is closely related to surface roughening and rippling: possible LER impact
• MD studies show energy deposited by low-energy (eg. 200 eV) ion impact is localized to a very narrow region near the surface
• Motivates question: how small a feature can be etched or milled by very tightly focused ion beam?
11/08/2000
21
Molecular Dynamics (MD) Simulation
Interatomic Potential Interatomic Forces
rjk
typical MD time step:
initialconfiguration
update positions
update velocities
evaluateforces• Simulation is non-
equilibrium!
• Kinetic + potential energy is always conserved
11/08/2000
22
Ion-Induced Displacement Simulations• Simulation cell size is 20 x 20 x 20 ų
(512 atoms)
• Introduce energetic Ar with velocity normal to surface
• Follow motion of all Si atoms until collision cascade subsides
• Repeat x 2000 for statistics
• Formulate a distribution (in r and z, relative to the point of impact) of atoms’ displacements 1
Expense: 100 impacts = 40 min (PIII-700)
1 in the spirit of Winterbon, Sigmund, and Sanders, Mat. Fys. Medd. Dan. Vid. Selsk. 37 (14) (1970)
11/08/2000
23
Ion Energy Deposition (Displacement) ProfileAtomic Displacement Distribution
200 eV Ar Impacts, Normal Incidence• Si atoms are displaced from their
origins by a distance proportional to the color temperature. (red → ~6 Å, blue → 0 Å)
ion trajectory
Relative coordinate system:
Ar
r
z
projected pointof impact
vacuumsolid
axisymmetric
20Å
11/08/2000
24
Ion “Milling” Simulation
• Simulation cell is substantially larger (60 x 60 x 40 ų, 9200 atoms)
• Introduce energetic Ar ion above surface, confined to a “beam” of radius 12 Å about the center
• Follow collision cascade
• Delete sputtered Si atoms
• Cool cell to room temperature
• Accumulate impacts on the same cell
Expense: 100 impacts = 20 hours (PIII-700)
11/08/2000
25
Feature EvolutionRepresentations of the surface after 330 impacts (200 eV Ar, normal incidence)
Stereograph top view
Tissue-paper plot
• side view, +y direction25 Å
11/08/2000
26
Milestones2001 Milestone
Write and test MD code to study energy deposition profiles. Conduct preliminary sputtering studies. Compare to
phenomenological theory.
2002 Milestone
Write and test code to study ion-induced surface diffusion. Compare results to phenomenological theory and experiment.
2003 Milestone
Simulate energy deposition and surface diffusion andcompare to LER experiments.
11/08/2000
27
Diffusion in Isotopically Controlled Silicon
Eugene E. Haller
University of California at Berkeley
and
Lawrence Berkeley National Laboratory
SFR-UCB SMART, November 8, 2000
11/08/2000
28
Collaborators
• C. B. Nelson, GSRA in MSE Dept., UCB
• Dr. H. Bracht, Univ. of Münster, Germany
• Dr. S. Burden, ISONICS Corp., Golden, CO
11/08/2000
29
Diffusion, Native Defects and Impurities
– Major experimental observation: Substitutional
impurities diffuse assisted by native defects
– The major native defects are vacancies and self-
interstitials in monatomic crystals
– Many native defects exist in more than one charge state:
Fermi level effects
11/08/2000
30
11/08/2000
31
Dopant and Self-Diffusion in Crystalline Solids
interstitialcy- vacancy - Exchange (ring) mechanism
I V
DSD = VCVeqDV + ICI
eqDI + Dex
11/08/2000
32
Self-Diffusion in Crystalline Solids
• Self-diffusion studied with radioactive isotopes:
– often limited by low activity and short half-life of the radiotracer (e.g. T1/2 of 31Si is 2.6 hrs)
– deposition alters the near-surface crystal quality of the material under study
– special safety requirements
• Self-diffusion studied with stable isotopes:
– at least two stable isotopes must exist (28Si, 29Si, 30Si)
– growth of epitaxial layers by VPE or MBE
– self-diffusion can be observed at several buried interfaces
11/08/2000
33
Isotope Heterostructures
The Isotope Superlattice
1017
1018
1019
1020
1021
1022
0 0.2 0.4 0.6 0.8 1 1.2 1.4
natSi -28Si-Superlattice: as-grown
30S
i-C
ON
CE
NT
RA
TIO
N (
cm-3
)
DEPTH (microns)
28Si
30Si
11/08/2000
34
Undoped Silicon Isotope Heterostructure
1017
1021
1019
1018
1020
1022
Con
cen
trat
ion
(cm
-3)
29Si
30Si
Depth (micron)
1 20 3 4 5 6
11/08/2000
35
Self-Diffusion in Pure Silicon
1017
1018
1019
1020
1021
1022
4 63 5
1322C; 30 min
1045C; 12 d
x (m)
30S
i con
cent
rati
on (
cm-3
)
Profiles:
H. Bracht, E. E. Haller, and R. Clark-Phelps, Phys. Rev. Lett 81 (1998) 393
as-grown
11/08/2000
36
Self-Diffusion in Pure Silicon
10-19
10-17
10-15
10-13
10-11
0.6 0.7 0.8 0.9
1400 1200 1000 800
103/T (K
-1)
DS
i (cm
2 s-1)
T (oC)
Temperature dependence:
120 scm530
eV75.4
D
Q
H. Bracht, E. E. Haller, and R. Clark-Phelps, Phys. Rev. Lett 81 (1998) 393
11/08/2000
37
Thermodynamic Properties
defect H f+H m S f+S m H f S f H m S m
I 4.95 eV 13.2 k 3.2 to 3.5 4 to 6 (1.4 to 1.8) (7 to 9)
V 4.24 eV 6.6 k 2.0 to 4.0 1 to 16 (0.2 to 2.2) (-9 to 6)
exp exp ,,,,,,
kT
HH
k
SSDC
mVI
fVI
mVI
fVI
VIeq
VI
• I-related data consistent with theoretical results
• reliable data for V are still lacking (controversy: low T high T)
Results:
Transport coefficients:
11/08/2000
38
Boron and Si Self-Diffusion
• Boron diffusion mechanism
• Considering charge states:
and
IS BIB
IBB SI0
hIBB SI 00
11/08/2000
39
Boron and Si Self-Diffusion• Boron and Si self-diffusion are coupled via the Si
interstitial population
• System of four coupled differential equations has to be solved numerically
kf,b = forward/backward rate constants
.0
0
0
0
0
0
0
2
2
x
CD
xt
C
CCkCCkx
p
p
C
x
C
xD
t
C
CCkCCkx
CD
t
C
CCkCCkt
C
SiSi
Si
IBboBfII
II
IBboBfB
B
B
IBboBfB
si
si
I
I
I
sI
S
11/08/2000
40
Boron and Si Self-Diffusion
• Modification of the Si self-diffusion coefficient through the presence of B
• Extraction of as function of the temperature and EF
• Determination of donor level position of I0/+
• In case CI ~ Cboron, compensation of boron acceptors will accrue, allowing the determination of CI
oeqVVV
eqV
eqIII
eqII CCCDCfCCDCfSiD V /]//[
IeqI DC
11/08/2000
41
Boron and Si Self-Diffusion
B in Si Superlattice Simulation
1.0E+17
1.0E+18
1.0E+19
1.0E+20
1.0E+21
1.0E+22
0 400 800 1200 1600
Depth (nm)
Co
nc
en
tra
tio
n (
cm
-3)
30Si
BAs-Grown
11/08/2000
42
Conclusions and Outlook
• Boron doped Si isotope heterostructures hold promise for the first determination of the interstitial transport coefficient CIVI as a function of EF and temperature.
• In case CI ~ CB , we can determine CI
• Accurate knowledge of CIVI is crucial for modeling B diffusion
• Analogous experiments will be conducted with P. High P concentrations will induce the formation of vacancies
11/08/2000
43
SFR WorkshopNovember 8, 2000
Berkeley, CAM. A. Lieberman, A. J. Lichtenberg, A. M. Marakhtanov,
K. Takechi, and P. Chabert Berkeley, CA
Plasma Sources for Small Feature Reproducibility
11/08/2000
44
Milestones• September 30, 2001 - Etch resist in LAPS to examine uniformity. - Characterize instability using OES/actinometry and planar probe.
- Install the Z-scan sensor and explore the spectral RF signature of plasma instabilities.
• September 30, 2002 - Characterize plasma instability using V-I-phase probe. - Model for reduced electron temperature and density.
• September 30, 2003 - Develop and test instabilities control. Reduce electron
temperature and density in discharges.
11/08/2000 Y. Moon, K. Bevans, and D. A. Dornfeld
Experimental LAPS Setup
A
Xm(0.5uH)
X1
X2
Rs
Vrf
Plasma
900 pF(fixed)
12-1000 pF
12-500 pF
Bm1
Voltage sensorD
C B
Bm2
Antenna coil embedded in the plasma
Eight quartz tubes threaded by copper
antenna tubes
36.0 cm x 46.5 cm processing area
New series-parallel
system configuration
11/08/2000
46
Oxygen Discharge Model
0exp8
1
2
2
k iz
kO
a
izii
rRn
D
Knn
r1 r2 r8…..
Substrate holder metal
Side metal
Calculation area
RQuartz-5
5
-15 15 (cm)
B.C. Vn=0
Chamber geometry used in the model
(Center of the chamber)(cm)
222
OidissOO nnKnD
Particle balance :
Power balance :
walls iTabs dSEeP iecT EEEE
11/08/2000
47
0 5 10 15 20 25 3010
13
1014
1015
1016
Pabs
=500 W
O2 gas pressure
1 mTorr 5 mTorr 20 mTorr 50 mTorr 100 mTorr
Flu
x of
ions
inci
dent
on
subs
trat
e ho
lder
(cm
-2s-1
)
Distance from the center (cm)
Ion flux profiles at substrate surface for various gas pressures (simulation results)
11/08/2000
48
Photoresist Etch Model
O
ref
biasiref
BpkVk
jAkRE
1)(
1
..
1
E.R. : Etch rateji : Ion current density at substrate surfacepO : O-atom pressure at substrate surfacekref : Ion-induced desorption rate constant at
Vbias= 0 V
1 10 1000
50
100
150
200
Data (500 W) Fit (500 W) Data (1000 W) Fit (1000 W) Data (1500 W) Fit (1500 W)
Etc
h R
ate
(nm
/min
)
Gas Pressure (mTorr)
Plots of etch rate vs.
pressure for various rf
powers. The fits are from
the above equation
with simulation results.
11/08/2000
49
Etch rate profiles at a gas pressure of 5 mTorr at Vbias= 0 V and Vbias= -80 V
-30 -20 -10 0 10 20 30
100
40
200
5 mTorr, Vbias=0 V 5 mTorr, Vbias=0 V 5 mTorr, Vbias=-80 V 5 mTorr, Vbias=-80 V
Etc
h R
ate
(nm
/min
)
Vertical Position (cm)
Applying a Vbias leads to
increase in both etch rate
and uniformity over the
processing area.
Effect of applying a substrate bias on
etch rate and uniformity
11/08/2000 Y. Moon, K. Bevans, and D. A. Dornfeld
Experimental TCP Setup
Mass Spectrometer
PlasmaPlasmaInductive CoilInductive Coil
Langmuir ProbeLangmuir ProbeOES (PMT)OES (PMT)
30
cm
19 cm
Planar Probe
Monochromator
11/08/2000
51
Instability Windows
0 20 40 60 80 1000
200
400
600
800
SF6/Ar (1:1)
Pressure (mTorr)
0 20 40 60 80 1000
200
400
600
800
INSTABILITY
Pure SF6
Pressure (mTorr)
Po
we
r (W
)
• Unstable in most of discharge conditions used in etching
• Instability linked to average attachment rate
Capacitive
Capacitive
Inductive
11/08/2000
52
Effect of Matching Network
0,0 0,4 0,8 1,2 1,6 2,00,0
0,2
0,4
0,6
0,8
Lig
ht in
ten
sity
(a.u
.)
time (ms)
• In Ar/SF6 discharges the frequency depends on the settings of the matching
network
Ar / SF6 (1:1)
p = 5 mTorr
Prf = 500 W
f1 = 1.0 kHz
f2 = 8.5 kHz
SETTING 1 (LOW FREQ.) SETTING 2 (HIGH FREQ.)
11/08/2000
53
Determination of time varying densities from planar and cylindrical probes currents
The negative ion fraction a can be deduced from the saturation current ratio:
en
n fI
VIR
sat
p
time (ms)
, where
Ar / SF6 (1:1)
p = 5 mTorr
Prf = 550 W
f = 0.8 kHz
11/08/2000
54
Results in (1:1) Ar/SF6 discharge
Ar / SF6 (1:1)
p = 5 mTorr
Prf = 500 W
f = 0.8 kHz
• The electron density changes by a factor of 20 with fast rise and decay times
• Positive and negative ion densities decay at the same rate
11/08/2000
55
Results in pure SF6 discharge
SF6
p = 5 mTorr
Prf = 500 W
f = 12.9 kHz
5
400
• The electron density exhibit sharp peaks; the ion density is slightly modulated
• The discharge spends a long time in low-density regime with 400
11/08/2000
56
Future Work
• Modeling silane deposition processes in LAPS
• Modeling reduced electron temperature and density processing plasmas in LAPS or TCP
• Upgrading TCP with higher power, Z-scan, and new matching network
• Instability measurements in upgraded TCP
11/08/2000
57
Plasma Assisted Surface Modification:Low Temperature Bonding
SFR WorkshopNovember 8, 2000
Yonah Cho, Adam Wengrow, Changhan Yun, Nathan Cheung
UC-Berkeley, CA
2001 GOAL: Establish plasma recipes and bonding data for Si, oxide and metal surfaces
11/08/2000
58
Motivation : Heterogeneous Integration by Bonding
SiO2
SOI
Si
Si
SOI
Photonics on Si Silicon-On-Insulators
Microfludic Transport
200 m
Dual Gate MOS
11/08/2000
59
Bonding Methods & Temperature Dependence
• Anodic BondingAnodic Bonding
• Direct BondingDirect Bonding• Metal BondingMetal Bonding
• Adhesive BondingAdhesive Bonding
Choice of bonding methoddepends on thermal budget & materials systems
Hydrophobic Si
Hydrophilic Si
3000
2500
2000
1500
1000
500
0100 200 300 400 500 600 700 800 900
Bon
din
g E
ner
gy (
mJ/
m2 )
Annealing Temperature (oC)
Bonding Energy vs. Temperature
11/08/2000
60
*Plasma Treatment*Gases Used:
O2, He, N2, Ar
Pressure 200mTorr
Power50W-300W
Time15-30 seconds
+
O2
Thermal Annealing
Room Temp Bonding
Chemical Cleaning:Piranha + RCA
-
Plasma Assisted Bonding
11/08/2000
61
Bonding Kinetics
2.0 2.5 3.0 3.5
10
100
1000
B
ondi
ng E
nerg
y (m
J/m
2 )
Temperature (oC)
50100200
1000/Temperature (K)
Si-SiO2 Plasma-treated
Si-Si
Chemically cleanedSi-Si
Chemically cleaned SiO2-SiO2
Ea ~ 0.07 eV
Ea ~ 0.2 eV
Ea > 0. 7 eV
Plasma treated Si surfaces achieve covalent bonding as low as 105oC
O2 Plasma-treated Si-SiO2
11/08/2000
62
Surface Chemistry: XPS
90 95 100 105 110 115100
200
300
400
500
600
In
tena
sity
(A
.U.)
90 95 100 105 110 115
90 95 100 105 110 115100
200
300
400
500
600
Inte
nasi
ty (
A.U
.)
Binding energy (eV)90 95 100 105 110 115
Binding energy (eV)
SiO2Si
HydrophobicSi
HydrophilicSi
O2 plasma treated
Si
Thermal SiO2
SiO2Si
11/08/2000
63
Plasma Effect Over Time
0 1 2 3 4
100
2000
Room Temperature
Annealed at 167 oC
Bo
nd
ing
Str
en
gth
(m
J/c
m2)
Time between plasma treatment and bonding (days)
11/08/2000
64
Surface Morphology
Chemically cleaned SiO2 plasma
150W-300W30 seconds
No Change in Surface Roughness
RMS < 0.5 nm
RMS < 0.5 nm
11/08/2000
65
Surface Modification by Concomitant/Sequential Plasma Treatment
• Mechanical and chemical properties of surface can be effected.
• Modified surface layer can be from substrate or grown / deposited / transferred layer.
• Physical and/or chemical changes of surface possible by altering chemical composition of plasma and ion energies.
Surface Layer
Si Substrate
PlasmaTreatment
ModifiedSurface
*By simply varying the potential of the PIII reactor, it can serve as surface modifier to implanter.
11/08/2000
66
2002 and 2003 Goals
•Demonstrate concomitant plasma treated deposition layer as effective diffusion layer.
•Demonstrate polymer surface modification with plasma implantation
11/08/2000
67
Layer Transfer for System Integration
SFR WorkshopNovember 8, 2000
Adam Wengrow ,Changhan Yun,, Yonah Cho,and Nathan Cheung
Berkeley, CA
2001 GOAL: Establish new layer transfer methods for integration of photonics ,IC, and MEMS
11/08/2000
68
Motivation
CPU
UV-Blue laser
Micro mirrors
1m
Micro pump
100 m
200 m
Micro-fluidic channels
x-y Detector Arrays
LED display
Paste-and-Cut Approach for Integration of Si IC, Photonics, and MEMSPaste-and-Cut Approach for Integration of Si IC, Photonics, and MEMS
11/08/2000
69
Substrate A
Substrate AProcessing A
Substrate A
MaterialGrowth A
Substrate A
Cut
Receptor Wafer
Substrate B
Substrate BProcessing B
Substrate B
MaterialGrowth B
Substrate BCut
Paste Paste
Paste-and-Cut for Materials IntegrationPaste-and-Cut for Materials Integration* Bypass material compatibility limitations* Flexible process flow for integration
11/08/2000
70
Mechanical Transfer of Pattern-Implanted Si
Implantation Masks
Si donor wafer
Si handle wafer
H+
SiO2
DeviceRegion
Hydrogenpeak
TransferredSi Layer
1. H+ Implantation through mask pattern
2. Wafer Bonding at low temperature
3. Mechanical pulling for layer transfer
Localized can exceed the fracture strength of silicon, overcoming the limitation of the thermal cut method
11/08/2000
71
Thermal-cut versus Mechanical-cut
0 20 36 50 75 100100
101
102
103
104
105
FIA(%)
t(se
c)
No
laye
r tr
ansf
er
>1day
550C
600C650C
700C
FIA =Total area
Implanted area
Hydrogen implantation: 51016 H+/cm2 at 150keV
No thermal-cut for FIA20%
1m
Mechanical-cut shown possiblefor FIA~20%
11/08/2000
72
Poly-Silicon Layer Transfer Using Ion-cutH+
Si donor
Handle wafer
Bondinginterface
Hydrogenpeak SiO2
Transferred poly-Si layer
Poly-Si
Poly dep. & H+ implant
Low-temperature wafer bonding Thermal or mechanical layer cleavage
Si donor
Si donor
Oxygen plasma
Si donor Handle wafer
CMP poly-Si
Handle wafer
Plasma surface activation for bonding
Rough surface for CVD Si.
Smooth after CMP
11/08/2000
73
Poly-Silicon CMP for Direct Bonding
2m
2m
0nm76nm
0m
0m
2m
7.6nm
0m 0m
2m
0nm
As-deposited:RMS roughness ~10.2nmNOT BONDABLE
After CMP:RMS roughness ~0.93nmBONDABLE
10 times smoother after CMP!
11/08/2000
74
Poly-Si Layer Transfer Time vs. Temperature
13 14 15 16 17 18 19 20
-8
-7
-6
-5
-4
-3
-2
-1
ln[1
/t(s
ec)]
1/kT(eV-1)
Ea~1.3eV
Ea~0.17eV
400500600 300
5 sec.10 sec.
20 sec.
1 min.2 min.
5 min.10 min.
20 min.
50 min.
T(C)
t
Poly-Sic-Si
Poly-Si has 10 times smaller Ea thansingle-crystal Si !
RMS roughness of cut surface ~ 4.2nm, same as single-crystal cleavage
11/08/2000
75
2002 and 2003 Goals
Demonstrate integration vehicle with microfluidic flows, photonic source and photonic detection
Establish low-temperature layer transfer methodologies for Si-based, III-V based, and polymer- based thin-film devices
Recommended