Post on 29-Jul-2019
11
Electroless Capping and Diffusion Barriers
For Copper Metallization – Material properties
Prof. Yosi Shacham – Diamand ,
22
Co alloy barriersCo alloy barriers
CoReP – V. DUbin, 1993
NiReP – N. Petrov et al., 2001
CoWP – Y. Shacham, 1996
CoWB- T. Osaka et al, 2002
CoMoP – Y. Shacham, 1999
(Approximated date of 1st publication or patent )
33
• Low solubility of Cu in Co and no phase formation, • Cu solubility is about 0.1% at 400C• P - Low solubility in Co -
→ enrichment of grain boundaries?→ Affects microstructure, reducing grain size → Froms amorphous structure at high concentration (> 12%)
• W: Low solubility in Co -→ Stuff the grain boundaries of the Cobalt
Co - P
Electroless Co alloys - Co(1-x-y) WxPy
Negligible solid solubilitysolubility of P in fcc Co is less than 0.47 at. %Ishida and Nishizawa, Bull. Alloy Phase Diag. 11, 555 (1990)
Negligible solid solubilitysolubility of W in fcc and E Co is less than 1 at. %
Co - W
44
2. Electroless deposition of conformal ultra-thin Co1-x-yWxPy films
Top FieldBottom Field / Sidewall
Sidewall
Comparison to current industrial technologyIonized Metal Plasma PVD
Case study: Co0.9W0.02P0.08
A. Kohn, M. Eizenberg,
and Y. Shacham-Diamand,
Appl. Surf. Sci., to be published
BF CS TEM micrographs
55
400 600 800 1000 1200 140010-1
100
101
T (oC)
W s
olub
ility
in fc
c C
o (a
t. %
)
Sykes Magneli et al. Larikov et al. Takayama et al. (XRD) Takayama et al. (EPMA)
400 600 800 1000 120010-2
10-1
100
101
T (oC)
Cu
solu
bilit
y (a
t. %
)
Hasebe and Nishizawa Bruni and Christian Old and Haworth Hasebe and Nishizawa
Proposed system for ULSI Cu metallization
• Cu: Low solubility in Co and no phase formation• P, W: Low solubility in Co
→ enrichment of grain boundaries?• P: Affects microstructure, reducing grain size
→ amorphous structure?• W: Proposal
→ introduction of a refractory alloying element may improve barrier efficiency?
Co - P
Co - W
Co alloys - Co(1-x-y) WxPy
Co - Cu
Theoretical
calculation
Negligible solid solubilitysolubility of P in fcc Co is less than 0.47 at. %Ishida and Nishizawa, Bull. Alloy Phase Diag. 11, 555 (1990)
66Nishizawa et al. , Bull. Alloy Phase Diag. 5, 161 (1984)
Co - Cu
The CoWP system for ULSI Cu metallization
Nagender Naidu et al., “Phase Diagrams of Binary Tungsten Alloys”Indian Institute of Metals 60 (1991)
• Cu: Low solubility in Co and no phase formation• P, W: Low solubility in Co
→ enrichment of grain boundaries?• P: Affects microstructure, reducing grain size
→ amorphous structure?• W: Proposal
→ introduction of a refractory alloying element may improve barrier efficiency?
Ishida et al. , Bull. Alloy Phase Diag. 11, 555 (1990)
Co - P
Co - W
Co alloys - Co(1-x-y) WxPy
77
hcp Co
fcc Co
Orthorhombic Co2P
0.4 0.6 0.8 1.0 1.2 1.4o
Rad
ial I
nten
sity
(a.u
.)
600oC
400oC
as-dep
s (A-1)
Evolution of microstructure with thermal annealCo0.9W0.02P0.08
Radial intensity of the SAEDas a function of the scattering vector
0 5 10 150
100
200
Num
ber o
f gra
ins
(-)
Grain size (nm)
0 5 10 15 200
100
200
Num
ber o
f gra
ins
(-)
Grain size (nm)
0 40 80 1200
100
N
umbe
r of g
rain
s (-)
Grain size (nm)
Dark field plan view TEM micrographs, SAED, apparent grain size histograms
as-dep
400°C
600°C
88
As-deposited structure:
Hexagonal close-packed cobalt nanocrystallites (d ~ 3-5 nm), with a preferred basal plane orientation embedded in an amorphous Co(W,P) matrix.
Evolution of structure with thermal anneal:
T ~ 300°C: hcp Co + amorphous Co(W,P) → hcp Co ; Ea = 1.6 ± 0.1 eV, constant nucleation rate, diffusion limited
T ~ 420°C: hcp Co → hcp Co + orthorhombic Co2P ; Ea = 4.7 ± 0.1 eV
T > 500°C: Delayed hcp Co → fcc Co transformation relative to bulk Co
P bonding shifts to covalent bonding at T > 600°C
Structure during failure of barrier:
At T ~ 450°C, the microstructure is hcp Co nanocrsytallites (d ~ 15 nm, 1 hour anneal), and small amounts of orthorhombic Co2P .
→ Failure mechanism : grain boundaries diffusion
Summary: Evolution of microstructure with thermal anneal
99
CoMoPCoMoP and and CoWPCoWP were deposited on sputtered seed were deposited on sputtered seed layers:layers:
Ti/Cu or Ti/Co on Silicon oxide.Ti/Cu or Ti/Co on Silicon oxide.
Ti improves the adhesion to the oxide.Ti improves the adhesion to the oxide.
Cu or Co are the seed layer.Cu or Co are the seed layer.
The samples were cleaned prior to the depositionThe samples were cleaned prior to the deposition
1010
Basic properties of Co(Mo,P)
30 30 –– 60 60 µΩµΩ.cm.cm60 60 –– 180 180 µΩµΩ.cm.cmResistivityResistivity
CoWPCoWPCoMoPCoMoP
1. The resistivity depends on the composition, thickness and seed type
2. Under similar conditions, e.g. same thickness, composition and seed type, the CoMoP layers has higher resistivity than CoWP
1111
Effect of Effect of CoWPCoWP and and CoMoPCoMoPcapping layers on Cu capping layers on Cu oxidation preventionoxidation prevention
1414
CoWPCoWP and and CoMoPCoMoP basic deposition Solutionbasic deposition Solution
“Ingredient” “Role” “Concentration”
CoSO4·7H2O Cobalt source 23 gr/l
HB3O3 Buffer 31 gr/l
3Na-citrate Cobalt complexing 130 gr/l
NaH2PO2 Reducing agent and Phosphor source 21 gr/l
RE610 Surfactant gr/l 0.05
KOH pH set 8.9-9
Na2MoO4 Mo source gr/l 0.1
Na2WO4 W source gr/l 10
1515
CoMoPCoMoP and and CoWPCoWP basic propertiesbasic properties
Properties CoMoP CoWP Mixed
potential -789 mV –702mV
Resistively 60-180 µΩ⋅cm 48 µΩ⋅cm
1818
Experiment procedureExperiment procedure
““SandwichSandwich”” samples of Cu between barrier layers samples of Cu between barrier layers
on SiOon SiO22 were made and subjected to oxidation were made and subjected to oxidation
condition by heating in open air furnace.condition by heating in open air furnace.
Surface resistance measurement were taken by Surface resistance measurement were taken by
4pp during the process.4pp during the process.
XPS profiling and spectrum was used to see the XPS profiling and spectrum was used to see the
the changes in profile and Cu oxidation state.the changes in profile and Cu oxidation state.
1919
Heat treatmentHeat treatment
For #1,#6 and 7#:For #1,#6 and 7#:300 min at 200C.300 min at 200C.170 min at 300C.170 min at 300C.200 min at 350C.200 min at 350C.
For #8:For #8:300 min at 200C.300 min at 200C.60 min at 300C.60 min at 300C.
2020
Surface resistance measurementSurface resistance measurement
R/Ro Vs Oxidation time
0
0.5
1
1.5
2
2.5
3
3.5
0 200 400 600 800
Time (min)
R/R
o
CoMoP/Cu/CoMoPon Co seadCoWP/Cu/CoWP onCo seadCoWP/Cu/CoWP onCu seadCoMoP/Cu on Cusead
2727
DiscussionDiscussionThe The CoWPCoWP and and CoMoPCoMoP oxidizing protection properties are oxidizing protection properties are close, with small advantage for close, with small advantage for CoMoPCoMoP..It seems that Co diffusion to the surface controls the barrier It seems that Co diffusion to the surface controls the barrier layer oxidizing leading to barrier splitting to oxidized Co and layer oxidizing leading to barrier splitting to oxidized Co and Co depleted layers. Co depleted layers. Until the oxidizing of the barrier is well advanced, the barrierUntil the oxidizing of the barrier is well advanced, the barriercontinues to defend the Cu from oxidizing.continues to defend the Cu from oxidizing.Barrier advanced oxidizing leads to itBarrier advanced oxidizing leads to it’’s failure as barrier and s failure as barrier and Cu diffuse to the surface.Cu diffuse to the surface.
2929
Experiment procedureExperiment procedure
““SandwichSandwich”” samples of Cu between samples of Cu between
barrier layers on SiObarrier layers on SiO22 were made and were made and
subjected to different thermal stress by subjected to different thermal stress by
annealing in a vacuum furnace .annealing in a vacuum furnace .
XPS profiling was used to see the the XPS profiling was used to see the the
changes in profile and Cu penetration.changes in profile and Cu penetration.
3131
Normalized profile for Cu penetration and Normalized profile for Cu penetration and the effective diffusion coefficient the effective diffusion coefficient
CoWP(innerCoWP(inner) ) CoWP(outerCoWP(outer) ) CoMoPCoMoP~0.26~0.26nmnm22/sec /sec ~ 0.21~ 0.21nmnm22/sec/sec ~ 0.1~ 0.1nmnm22/sec/sec
3232
SummerySummery
CoMoPCoMoP -- a novel electroless deposited film barrier option was presenteda novel electroless deposited film barrier option was presented
and compared to and compared to CoWPCoWP..
Both films shows similar barrier behavior with slight advantage Both films shows similar barrier behavior with slight advantage for for CoMoPCoMoP, ,
which has however higher electrical resistance.which has however higher electrical resistance.
Both films seems to prevent Cu oxidation to temperature of 300C,Both films seems to prevent Cu oxidation to temperature of 300C, and to and to
prevent Cu diffusion to temperature of 500C (materials profilingprevent Cu diffusion to temperature of 500C (materials profiling).).
The two films are interesting options for future capping layer fThe two films are interesting options for future capping layer for Cu or Cu
metallization. metallization.
3333
0 200 400 600 800
3.0
2.0
#2 – 225 nm
1.0
Time [min]
R(t)/R(0)
#4 - 115 nm
#3 - 210 nm
#1 – 215 nm
200C 300C 350C
Capping layer integrity – annealing in air & measuring resistivity insitu
3434
Effect of Effect of CoWPCoWP liner on liner on the reliability of Cu Dual the reliability of Cu Dual
damascene interconnectsdamascene interconnects
3535
CoWPCoWP capping layer (IBM) (1)capping layer (IBM) (1)
Cross section views of electromigration test structures:
3 level metal
2 level metal
3636
CoWPCoWP capping layer (IBM) (2)capping layer (IBM) (2)
TEM cross-sectional image of a Cu interconnect coated with CoWP
2 level metal
3737
CoWPCoWP capping layer (IBM) (3)capping layer (IBM) (3)
Elements concentration (by EDS). The electron probe moved from the top surface of a Cu damascene line, through the CoWP and amorphous SiCxHy coating layers and ended in the SiLK dielectric.
3838
CoWPCoWP capping layer (IBM) (4)capping layer (IBM) (4)
The resistance of a damascene Cu conductor, with and without a thin metal film on the top surface, vs time.
3939
CoWPCoWP capping capping layer(IBMlayer(IBM) (5)) (5)
FIB images of EM tested lines for uncoated and CoWP coated samples with current density of 3.6x106 A/cm2.
280 °C for 2.8 h
280 °C for 1100h
4040
CoWPCoWP capping layer (IBM) (6)capping layer (IBM) (6)
Conclusions
The results of this testing further support the hypothesis that the uncoated surfaces, or interfaces of Cu with dielectric, are the major sources of electromigration and thus reliability degradation.
In summary an investigation of Cu electromigration in Cu damascene interconnections with and without thin CoWP, CoSnP, and Pd coatings showed that electromigration failure lifetimes can be drastically changed. The migration of Cu at the top surface of a Cu damascene line was greatly reduced in the samples with 10–20 nm thick caps so that the Cu electromigration lifetime was markedly improved.
4242
Barrier Analysis & monitoringBarrier Analysis & monitoring
Materials science techniques:Materials science techniques:AES, SIMS, RBS, SEMAES, SIMS, RBS, SEM
Electrical characterization:Electrical characterization:II--VVCC--V & CV & C--tt
4343
Example: Example:
Testing of Testing of CoWPCoWP barrier layersbarrier layers
- AES - Auger Electron Spectroscopy- The Transient Capacitance Method
4444
Copper profiles as measured by AES. The Copper profiles as measured by AES. The Example:Example:sputtering rate was: 12A/min for sputtering rate was: 12A/min for Co(W,PCo(W,P) on Cu, 25 ) on Cu, 25
A/min for Cu, 10A/min for A/min for Cu, 10A/min for Co(W,PCo(W,P) on Co, 8A/min for ) on Co, 8A/min for the sputtered Co.the sputtered Co.
0 10 20 30 40 50 60 70 80 90 100 110 120 130 1400
5
10
15
20
25
30
35
40
Con
cent
ratio
n (A
rb.)
Sputtering Time (min)
Co(W,P) Cu Co(W,P) Co SiO2 Si
600C, 4hr.
As deposited
520C, 2hr
4545
400 600 800 1000 1200 140010-1
100
101
T (oC)
W s
olub
ility
in fc
c C
o (a
t. %
)
Sykes Magneli et al. Larikov et al. Takayama et al. (XRD) Takayama et al. (EPMA)
Proposed system for ULSI Cu metallization
• Cu: Low solubility in Co, no phase formation
• P: Affects microstructure, reducing grain size → amorphous structure?
• P, W: Low solubility in Co → enrichment of grain boundaries?
• W: Proposal → refractory alloying element may improve barrier efficiency?
Co - P Co - W
Co alloys - Co(1-x-y) WxPy
Negligible solid solubilitysolubility of P in fcc Co is less than 0.47 at. %Ishida and Nishizawa, Bull. Alloy Phase Diag. 11, 555 (1990)
4646
Aqueous solution, pH: 8.0 – 9.0, T = 85° - 95°C Alkali ions can be replaced by (NH4)+
Component Aim
CoSO4⋅7H2O Metal ion source
Na2WO4⋅2H2O Metal ion source (induced co-deposition)
NaH2PO2⋅2H2O Reducing agent
C6H5Na3O7⋅2H2O Complexing agent – reducing the electrochemical potential difference
H3BO3 Buffer – fixing the pH
KOH pH adjustment: electrochemical potential, rate, and mechanism
Surfactant RE-610 Reducing surface tension, extracting H2
Reducing agent: Borane – dimethylamine complex C2H10BN Buffer: NH4OH + CH3COOH
Electroless deposition of Co alloys
Co(1-x-y)WxPy
Co(1-y)Py
Co(1-x)Wx
BufferNH4OH + CH3COOH pH~10.5CoSO4 – 13.2 gr/lNa2WO4 – 3.3 gr/l
Reducing agent DMAB – 4 gr/l
4747
1. Obtain electroless Co alloyswith largest possible content of P and / or W
Determined by electron probe micro-analysisCompared to Rutherford backscattering spectroscopy, Auger electron spectroscopy
Details regarding the deposition process:
A. Kohn, M. Eizenberg, Y. Shacham-Diamand, and Y. Sverdlov,
Mater. Sci. Eng. A 302, 18 (2001)
Alloy P concentration (at. %)
W concentration(at. %)
Co1-x-yWxPy 8±2 2±1
Co1-yPy 10±2 -
Co1-xWx - 4±1
4848
2. Electroless deposition of conformal ultra-thin Co1-x-yWxPy filmsCase study: Co0.9W0.02P0.08
A. Kohn, M. Eizenberg,
and Y. Shacham-Diamand,
Appl. Surf. Sci., to be published
Bright field, cross-sectional TEM micrographs
Comparison to current industrial technologyIonized Metal Plasma PVD
4949
2. Electroless deposition of conformal ultra-thin Co1-x-yWxPy films
Bottom Field / Sidewall
A. Kohn, M. Eizenberg,
and Y. Shacham-Diamand,
Appl. Surf. Sci., to be published
Case study: Co0.9W0.02P0.08Comparison to current industrial technologyIonized Metal Plasma PVD
Bright field, cross-sectional TEM micrographs
5050
2. Electroless deposition of conformal ultra-thin Co1-x-yWxPy films
Top FieldBottom Field / Sidewall
Comparison to current industrial technologyIonized Metal Plasma PVD
Case study: Co0.9W0.02P0.08
A. Kohn, M. Eizenberg,
and Y. Shacham-Diamand,
Appl. Surf. Sci., to be published
Sidewall
Co
0.9 W0.02 P
0.08
Co0.9 W
0.02 P0.08
Bright field, cross-sectional TEM micrographs
5151
1 – Evaluated by C-t2 – Evaluated by C-V
Evaluation of diffusion barrier quality Results
A. Kohn, M. Eizenberg, Y. Shacham-Diamand,
B. Israel, and Y. Sverdlov
Microelectronic Eng. 55, 297 (2001)
Electroless Co0.9P0.1 , 30 nm thick, are stable barriers at 450°C during approximately 10 hours
Current allowed total thermal budget: Thermal cycles equivalent to 400°C, 10 – 60 minutes
Barrier
30 nm thick
Thermal Stress1 Bias and Thermal Stress2
1 MV/cm at 300ºC / 30 min
Co0.96W0.04 Fails after 450ºC / 1hr Fails
Co0.9P0.1 Fails after 450ºC / 10 hr Stable
Co0.9W0.02P0.08 400ºC/30 min: Stable
500ºC/30 min: Fails
N/A
5252
4. Investigate the structure of electroless Co alloys and its evolution as a result of heat treatments
Questions:1. What is the as-deposited structure?
2. How does the structure change with thermal anneal?
3. What is the structure during failure of the diffusion barrier?
Case study: Co0.9W0.02P0.08Electroless Co0.9W0.02P0.08 10 – 100 nm
Sputtered Co or Cu 2 – 20 nm
(Sputtered Ti) (5 nm)
SiO2 100 nm
Si wafer
5353
30 35 40 45 50 55
Siλ/2 (004)
εCo(01·0)
εCo(01·1) εCo
(00·2)
Difference
Co seed
Co0.9W0.02P0.08 as-dep.
Inte
nsity
(arb
. uni
ts)
2θ (°)
As-deposited structure: ?300°C: hcp Co400°C: hcp Co (main) + orthorhombic Co2P (minor)600°C-700°C: hcp → fcc transition (delayed)Same results on Co and Cu seed layer
Structure of Co0.9W0.02P0.08 and its evolution with thermal treatments
30 35 40 45 50 55
TiP2?
Co2P (201)
εCo(01·0)
αCo(111) εCo
(01·1)
αCo(200)
εCo(00·2)
Siλ/2 (004)
2θ (°)
Inte
nsity
(arb
. uni
ts)
700°C
600°C
500°C400°C300°C
as-dep.150°C
Cobalt seed
Powder XRD, Bragg-Brentano geometry Thermal anneal: 1 hour, Vacuum ≤ 10-6 Torr
5454
Radial intensity of the selected area electron diffraction (SAED)as a function of the scattering vector
hcp Co
fcc Co
0.3 0.4 0.5 0.6
_Si [216]
Si [103]
Si [100]
Si [011]
Rad
ial I
nten
sity
(a.u
.)o
s (A-1)
(01⋅0)(00⋅2)
(01⋅1)
(111)
(002)
DC s =∆ ; C ~ 1 → D ~ 3 – 5 nm
0.4 0.5 0.6o
Rad
ial I
nten
sity
(a.u
.)
s (A-1)
Gaussian fit
• Hexagonal close-packed cobalt• Grain size: several nm• Preferred basal plane orientation
)sin2 s(λ
θ=
As-deposited structureAnalysis of selected area electron diffractions
5555
hcp Co
fcc Co
Orthorhombic Co2P
0.4 0.6 0.8 1.0 1.2 1.4o
Rad
ial I
nten
sity
(a.u
.)
600oC
400oC
as-dep
s (A-1)
Evolution of microstructure with thermal annealCo0.9W0.02P0.08
Radial intensity of the SAEDas a function of the scattering vector
0 5 10 150
100
200
Num
ber o
f gra
ins
(-)
Grain size (nm)
0 5 10 15 200
100
200
Num
ber o
f gra
ins
(-)
Grain size (nm)
0 40 80 1200
100
N
umbe
r of g
rain
s (-)
Grain size (nm)
Dark field plan view TEM micrographs, SAED, apparent grain size histograms
as-dep
400°C
600°C
5656
Co0.9W0.02P0.08
0 200 400 600
60
80
100
120
hcp Co + Co2Phcp Cohcp Co + amorph. Co
Cooling 1st Run
1st Run
2nd Run
Res
istiv
ity (µ
Ω·c
m)
Temperature (oC)
Heating rate ~ 2.8°C/min ; Vacuum ~ 5 × 10-6 Torr
Tracking structural changesIn-situ resistivity as a function of temperature
After anneal:Significant temperature independent
contribution to electron scattering
(imperfections including impurities)Co BulkFilm dT
d dTd ρ
=ρ
BulkFilm ρ>ρ
575713.6 14.0 14.4
-14
-13
-12
-11
-10 460 440 420
104/Tc (K-1)
ln[(d
T/dt
)/Tc2 ] (
-)
Tc (oC)
Ea – apparent activation energydT/dt – heating rateTc ≡ dρ / dT is minimal
(maximum rate of crystallization)
Constant kTE
T)dt
dT( ln
C
a2
C
+−=
240 280 320 36090
100
110
120
(e)(c)(d)(b)(a)
Res
istiv
ity (µ
Ωcm
)
Temperature (oC)
(a) 1.4 oC/min(b) 2.8 oC/min(c) 5.6 oC/min(d) 11.2 oC/min(e) 16.8 oC/min
Crystallization and Co2P phase formation processes
16.4 16.8 17.2 17.6 18.0-13
-12
-11
-10
340 320 300Tc (oC)
ln[(d
T/dt
)/Tc2 ] (
-)
104/Tc (K-1)
Ea = 1.6 ± 0.1 eV Ea = 4.7 ± 0.1 eV
hcp Co + amorphous Co(W,P) → hcp Co hcp Co → hcp Co + orthorhombic Co2P
Kissinger analysis:
5858
[ ] )t-(tk(T)- exp - 1 )t(f n0⋅=
)0(tf(t)-1
)(tf(t) 1
=ρ+
∞→ρ=
ρ
n = a + b⋅ca = 1 ; Constant nucleation rate (N ~ t )b = 3 ; 3D growthc = 0.5 ; Diffusion controlled growth
0 10 20 30
100
110
120n2 n3n1
Res
istiv
ity (µ
Ω·c
m)
T ~ 260oC
Time (X103 sec)
f(t) – fraction crystallized
k (T) – apparent rate coefficientt0 – incubation timen – J-M-A index
Fraction crystallized as a function of time at a constant temperature:
Crystallization processJ-M-A analysis:
103 104
-4
-2
0
2
n1 = 1.5 ± 0.2
n2 = 2.6 ± 0.2
n3 = 1.3 ± 0.2
ln(-l
n(1-
f)) (-
)
Time (sec)
5959
Chemical binding statesCo, W – no observable change in chemical binding (XPS, EELS)P – significant changes
X-ray photoelecton spectroscopy:The 2p core level can not be fitted by a single doublet p3/2, p1/2
Fitting is obtained by two doublets marked as P1 and P2
As-deposited: amorphous matrix 600°C: covalent binding
as-dep Data Fitting
P2
P2
P1P1
400oC
P1 P1P2
P2
P1 P1P2
P2
600oC
132 130 128 126
P1P1P2
P2
Co2P
Binding Energy (eV)
Inte
nsity
(arb
. uni
ts)
Sample
Area ratio P1/P2 (-)
±10%
Co0.9W0.02P0.08 as-dep 1:1.1 Co0.9W0.02P0.08 400oC 1:2 Co0.9W0.02P0.08 600oC 1:4.9 Co2P Reference 1:5
6060
As-deposited structure: • hcp Co nanocrystallites (d ~ 3-5 nm), preferred basal plane orientation• amorphous CoWP matrix.
Evolution of structure with thermal anneal: • T ~ 300°C: hcp Co + amorphous Co(W,P) → hcp Co ; Ea = 1.6 ± 0.1 eV,
constant nucleation rate, diffusion limited• T ~ 420°C: hcp Co → hcp Co + orthorhombic Co2P ; Ea = 4.7 ± 0.1 eV• T > 500°C: Delayed hcp Co → fcc Co transformation relative to bulk Co• P Chemical binding shifts to covalent at T > 400°C
Structure during failure of barrier: At T ~ 450°C, the microstructure is hcp Co nanocrystallites
and small amounts of orthorhombic Co2P
→ Failure mechanism : grain boundaries diffusion
Summary: Evolution of microstructure with thermal anneal
6161
5. Kinetics of Cu diffusion via electroless Co alloysand a comparison to PVD Co
Questions:1. What is the type of kinetics of Cu diffusion through the electroless Co alloy films?
2. What are the kinetic parameters of this diffusion?
3. Why are Co1-x-yWxPy alloy thin films, predominately Co (1 – x – y ≈ 0.9),
effective barriers, while polycrystalline pure Co films are ineffective?
6262
Special case of ultrafine-grained polycrystals
For Co0.9W0.02P0.08 :
d – nanometer scale
s >>1 - low solubility of Cu in Co
d << sδ/2
d – grain size
s – segregation coefficient
δ – grain boundary width
Cu lattice diffusion in Co:Ea ≈ 2.86 eV ; D0 ≈ 1 cm2/sec
C' Kinetics T/Tm ~ 0.3
L = 2(Dlt)1/2 << d
A0 Kinetics T/Tm ~ 0.5 or long anneal times
L = 2(Dlt)1/2 ≥ d
Landolt-Börnstein Numerical Data and Fundamental Relationships in Science and Technology,edited by O.Madelung and H.Mehrer(Springer,Berlin,1999),Group III,Vol.26,p.52.
6363
Analysis of type-C Cu grain boundary diffusivity
What are the kinetic parameters of the grain boundary diffusion?
Z
z = L
6464
Assuming a 1D case:2
gb2
gbgb
z)t,z(C
D t
)t,z(C∂
∂=
∂∂
Copper grain boundary diffusivity at processing temperatures : 0.3 – 0.4 Tm
6565
Boundary conditions:
Assuming a 1D case:2
gb2
gbgb
z)t,z(C
D t
)t,z(C∂
∂=
∂∂
0 z
C
Lz
gb =∂
∂
=
(2)
impuritygb C )0t,z(C == ; 0<z≤L(3)
Copper grain boundary diffusivity at processing temperatures : 0.3 – 0.4 Tm
geometricfactormax,gbgb C
dq )t,0z(C ⋅
δ== q = 2 ; d ≥ L
q = 3 – d/L ; d < L(1)
6666
⎩⎨⎧ π+
⋅⎭⎬⎫π+
+−
π=
===
∑∞
= L2z)1n2(cos
L4tD)1n2(
- exp )1n2(
)1(4 - 1 )0t,z(C - )t,0z(C
)0t,z(C - )t,z(C2
gb22
0n
n
gbgb
gbgb
Solution:
Boundary conditions:
Assuming a 1D case:2
gb2
gbgb
z)t,z(C
D t
)t,z(C∂
∂=
∂∂
0 z
C
Lz
gb =∂
∂
=
(2)
impuritygb C)0t,z(C == ; 0<z≤L(3)
2 2
2 2 2n 0
gb(2 1)8 1 1 - exp -(2 1) 4
Dt n tn L
MM
ππ
∞
=∞
⎧ ⎫+⎪= ⎨ ⎬+ ⎪ ⎭⎩∑ L)t,0z(C M gb ⋅==∞;
Copper grain boundary diffusivity at processing temperatures : 0.3 – 0.4 Tm
max,gbgb Cd
q )t,0z(C ⋅δ
==(1)geometricfactor
q = 2 ; d ≥ Lq = 3 – d/L ; d < L
6767
10-6
10-4
10-2
300oC 350oC 400oC 450oC
10-6
10-4
10-2
0 40 80 12010-6
10-4
10-2
Sputter depth (nm)
I Cu/
Co
(-)
Co0.9W0.02P0.08
Co0.9P0.1
Sputtered Co
Sputtered Co
Co0.9W0.02P0.08
SIMS depth profiles(a)300°C 8 hr(b)450°C 8 hrBright field cross sectional TEM micrographs
8 hour anneal
6868
14 15 16 17 1810-18
10-16
10-14
440 400 360 320
This study Pellerin et al.J. Appl. Phys. 75, 5052 (1994)
Dgb
(cm
2 /sec
)
104/T (K-1)
T (oC)
Ea = 1.8 ± 0.2 eVD0 = 10(-1.2 ± 0.8) cm2/secδ = 0.5 nm, q = 2
Cu grain boundary diffusion parameters
Sputtered Co
6969
Ea = 1.8 ± 0.2 eVD0 = 10(-1.2 ± 0.8) cm2/secδ = 0.5 nm, q = 2
Cu grain boundary diffusion parameters
13 14 15 16 17 1810-20
10-18
10-16 480 440 400 360 320
104/T (K-1)
Dgb
(cm
2 /sec
) Electroless Co0.9W0.02P0.08 Electroless Co0.9P0.1
T (oC)
Ea = 1.25 ± 0.20 eVD0 = 10(-8.3 ± 1.0) cm2/secδ = 0.5 nm, q = 3 – d/T
Ea ≈ 1.3 eVD0 ≈ 10-7.2 cm2/secδ = 0.5 nm, q = 3 – d/T
Electroless Co0.9W 0.02P 0.08
Electroless Co0.9P 0.1
Sputtered Co
14 15 16 17 1810-18
10-16
10-14
440 400 360 320
This study Pellerin et al.J. Appl. Phys. 75, 5052 (1994)
Dgb
(cm
2 /sec
)
104/T (K-1)
T (oC)
7070
gb
2
fail D4L t ≡
For a 30 nm thick Electroless Co0.9P 0.1 barrier:
According to Diffusion kinetics measurements:
At 450°C: Dgb ~ 4×10-17 cm2/sec → tfail ~ 16 hours
According to Electrical measurements:
At 450°C: → tfail ~ 8-12 hours
Estimated lifetime for a 10 nm thick Co0.9W0.02P0.08 barrier at 400°C
→ tfail ~ 35 hours
Failure criterion
tfail – time to failure of the diffusion barrierL – thickness of the diffusion barrierDgb – grain boundary diffusivity
7171
A0 Kinetics T/Tm ~ 0.5 or long anneal times
L = 2(Dlt)1/2 ≥ d
Why do we see a significant reductionin the pre-exponential kinetic diffusion parameter?
Analysis of steady-state type-A Cu grain boundary diffusivity
7272
Steady-state type-A Cu grain boundary diffusivityAnneal: ~0.5Tm → A0 kinetics 8 hours → steady statePlateau level is given by:
solC)dq-(1 gbC
dq C ⋅
δ+⋅
δ=
With increase of temperature: 1st term ↓ (d ↑)2nd term ↑ (Csol ↑ ; d ↑)
- plateau concentrationCgb – grain boundary concentrationCsol – solubility of Cu in Coq – geometrical factorδ - grain boundary widthd – grain size
C
10-4
10-3
10-2
10-1
as-dep 550oC 600oC 650oC 700oC
10-4
10-3
10-2
0 40 80 120
10-4
10-3
10-2
Depth (nm)
Co0.9W0.02P0.08
Co0.9P0.1
PVD Co
I Cu/
Co
(-)
SIMS depth profiles
7373
Assumptions:
5. Exponential dependence of the solubilityCsol = Cs0 ⋅exp (-Es/kT)
e.g.: for bulk Co, 900°C-1100°C:ES ~ 0.7 – 0.8 eV ; Cs0 ~ 4.5 ⋅ 103 at. %
4. q δ ⋅ Cgb : weak temperature dependence
2. δ ~ 0.5 nm regardless of the temperatureMishin and Herzig, Mat. Sci. Eng. A 260, 55 (1999)
1. For Co0.9W0.02P0.08 : gbCdq
⋅δ << solC)
dq-(1 ⋅
δ
3. Grain boundaries’ volume fraction:
dqδ
< 0.05
kTE - Cln ) (Iln s0s
Cu/Co α≅
(α ~ 3.3 : Calibration coefficient for SIMS measurements: Csol = α ⋅ ICu/Co)
8 10 12 14
0.1
1
10
1000 800 600 400T (oC)
hcp Cofcc Co
Cu
solu
bilit
y (a
t. %
)
104/T (K-1)
Hasebe et al. Bruni et al. Old et al. Hasebe et al.
Grain boundaries’ passivation
7474
2. Solubility of Cu in Co
for 550° – 700°C regime:
Es = 0.52 ± 0.15 eV
Cs0 = (6 ± 1) × 102 at. %
Results:1. For Co0.9W0.02P0.08 : Cgb is negligible
Passivation of the grain boundaries
8 10 12 14
0.1
1
10
1000 800 600 400T (oC)
hcp Cofcc Co
Cu
solu
bilit
y (a
t. %
)
104/T (K-1)
Hasebe et al. Bruni et al. Old et al. Hasebe et al. This study
10.0 11.0 12.0
10-3
700 650 600 550 T (oC)
I Cu/
Co (
-)
104/T (K-1)
A. Kohn, M. Eizenberg, and Y. Shacham-Diamand, J. Appl. Phys. 92, 5508 (2002)
Mechanism of improved barrier properties:Grain boundaries’ passivation
kTE - Cln ) (Iln s0s
Cu/Co α≅
7575
1.
2. Tungsten increases the passivation of the grain boundaries
What is the contribution of the tungsten alloying?
d1 ) - C ( Cq ) - I( I CoWP
gbCoPgb
CoWPCu/Co
CoPCu/Co ⋅
αδ
=
% at. 3 C - C CoWPgb
CoPgb ≈
Results: 0.025 0.030 0.035 0.040 0.045
5
10
15
20
25
400 600
20
40
d [n
m]
T [oC]
∆I
Cu/
Co X
10-5 (-
)
d-1 (nm-1)
7676
Classification of Cu transport kinetics via the Co alloy films: T/Tm ~ 0.3 : Predominately via the grain boundaries (type-C)T/Tm ~ 0.5 or long anneal times: Effective diffusivity (type-A)
Kinetic parameters of Cu diffusion:
(δ = 0.5 nm)
Why are electroless Co-alloy films effective barriers? Passivation of the grain boundaries
Summary: Kinetics of Cu diffusion via electroless Co alloysand a comparison to PVD Co
PVD Co 1.8 ± 0.2 eV 10(-1.2 ± 0.8) cm2/secCo0.9W0.02P0.08 1.25 ± 0.20 eV 10(-8.3 ± 1.0) cm2/secCo0.9P0.1 ≈ 1.3 eV ≈ 10-7.2 cm2/sec
7777
SummaryI. Electroless deposition of Co(W,P) alloys
→ Potential process for depositing barriers and encapsulation layers in ULSI Cu metallization
II. Co0.9P0.1, Co0.9W0.02P0.08 effective barriers at 450°C→ Relevant barrier for ULSI Cu metallization
III. Study of the structure of the electroless deposited Co(W,P) films→ Failure mechanism of the barriers is grain boundaries diffusion
IV. Kinetic parameters of Cu diffusion in PVD Co and electroless Co alloys→ Diffusivity is 2-3 orders of magnitude lower
Significant difference in the pre-exponential factor
V. Proposed explanation for effectiveness of electroless Co-alloy barriers→ Passivation of the grain boundaries
7878
30 35 40 45 50 55 60
2
4
6
8
Co seed layeras-deposited
Siλ/2 (004)
Inte
nsity
(a.u
.)
2θ (°)
Co0.9P0.1 Co0.9P0.08W0.02
The influence of W on the as-deposited structure?
7979Nishizawa et al. , Bull. Alloy Phase Diag. 5, 161 (1984)
Co - Cu
Proposed system for ULSI Cu metallization
Nagender Naidu et al., “Phase Diagrams of Binary Tungsten Alloys”Indian Institute of Metals 60 (1991)
• Cu: Low solubility in Co and no phase formation• P, W: Low solubility in Co
→ enrichment of grain boundaries?• P: Affects microstructure, reducing grain size
→ amorphous structure?• W: Proposal
→ introduction of a refractory alloying element may improve barrier efficiency?
Ishida et al. , Bull. Alloy Phase Diag. 11, 555 (1990)
Co - P
Co - W
Co alloys - Co(1-x-y) WxPy
8080
[001]
[010]
[100]
[111]
Space group = 62 Pnmaa = 0.565 nm, b = 0.351 nm, c = 0.66 nm; ρ = 7.56 gr/cm3
Orthorhombic Co2P
Cell viewed in various directions
Blue – Co ; Red – P
8181
Major structural changes occur after a short timecompared to the MOS failure time. Failure can not be attributed to recrystallization and recovery.
35 40 45 500.0
2.0x102
4.0x102
6.0x102
2θ (°)
As dep.
Inte
nsity
(a.u
.)
450oC / 2 hr 450oC / 5 hr 450oC / 10 hr
εCo (01-10)
Co2P (201) εCo
(01-11)
εCo(0002)Si
λ/2 (004)
Phase evolution (XRD)Co0.9P0.1 / Co
0 2 4 6 8 100.0
0.2
0.4
0.6
0.8
1.0
Co0.9P0.1 Co0.96W0.04
Thermal treatment (450oC) time (hr)
Rs/R
0 (-)
Relative sheet resistance
Failure mechanism of the d.b.
8282
Electroless deposition of Co alloysProcess:
Co(II)Cit + e- → Co(I)CitadsCo(I)Citads + e- → Co(s) + Cit
Reduction of Co ions (complexed with the citrate)
H2PO2- → HPO2
-ads + Hads
Reducing agent is hypophosphite:H ?extraction? on the catalytic seed layer
Reaction with OH- ions:
HPO2-ads + OH- → H2PO3
- + e-
Reaction of H atoms is dependant on the catalytic seed layer:
Hads + OH- → H2O + e- Catalytic seed layer: Pd, Pt, Rh2Hads → H2 Catalytic seed layer: Cu, Au, Ag
In parallel, a competing reaction of hypophosphitedeposits P:
H2PO2- + 2H+ + e- → P + H2O
W deposition? Induced co-deposition: iron group ion + refractory metalProposed Mechanism: Podlaha et al., J. Electrochem. Soc. 144 (1997) 1672
Induced co-deposition of MoO42- and ion M (Fe2+, Co2+, Ni2+) complexed with a ligand L
MoO42- + M(II)L + 2H2O + 2e- → [M(II)LMoO2]ads + 4OH-
[M(II)LMoO2]ads + 2H2O + 4e- → Mo(s) + M(II)L + 4OH-
8383
Periodicity in P depth profileAES depth profiles
SiSiSiOSiO22
TiTiCoCo
CoCo0.90.9WW0.020.02PP0.080.08
as-deposited
3000C
5000C
7000C
4000C
~12 nm
8484
Chemical binding statesCo, W – no observable change in chemical binding (XPS, EELS)P – significant changes
X-ray photoelecton spectroscopySuggested fitting of 2p3/2 and 2p1/2 P states :P1 and P2 2p binding states1. p3/2 and p1/2 ∆E = 0.84 eV Goodman et al., Phys. Rev B 27, 7440 (1983)2. p3/2 and p1/2 area ratio = 2:13. FWHM, G/L ratio equal for p3/2 and p1/2 for P1, P24. 0.2 ≤ G/L ≤ 0.8Results:1. P1 and P2 – equal B.E., FWHM, G/L ratio for all samples
→ validates assumptions2. Area ratio of P1:P2 → ratio of binding states
as-dep Data Fitting
P2
P2
P1P1
400oC
P1 P1P2
P2
P1 P1P2
P2
600oC
132 130 128 126
P1P1P2
P2
Co2P
Binding Energy (eV)
Inte
nsity
(arb
. uni
ts)
Sample
Area ratio P1/P2 (-)
±10%Co0.9W0.02P0.08 as-dep 1:1.1 Co0.9W0.02P0.08 400oC 1:2 Co0.9W0.02P0.08 600oC 1:4.9 Co2P Reference 1:5
8585
Same results on Co and Cu seed layer
No phase reaction between Cu and the electroless film up to 700°C
Structure of Co0.9W0.02P0.08 and its evolution with thermal treatmentsCu : Influence of seed layer and / or phase interaction?
Powder XRD, Bragg-Brentano geometry
Thermal anneal:1 hourVacuum ≤ 10-6 Torr
Copper seed
30 35 40 45 50 55
5 x 1 0 2
2 x 1 0 3
3 x 1 0 3
4 x 1 0 3
5 x 1 0 3
2θ (°)
as depInte
nsity
(a.u
.)
300oC
500oC
700oC
Cu (111)
αCo(200)
εCo(01·1)
Co2P (201)
Siλ/2 (004) αCo
(111)
εCo (00·2)
εCo (01·0)
8686200 400 600
2.025
2.030
2.035
2.040
2.045
2.050
o
fcc Co (111)
hcp Co (00·2)
d-sp
acin
g (A
)
Annealing temperature (oC)
Pseudo-Voigt, single-line fit of the (00·2) hcp / (111) fcc Co plane Th. H. De Keijser et al., J. Appl. Cryst. 16, 309 (1983)
Domain size, microstrain
Interplanar spacing
Structure of Co0.9W0.02P0.08 and its’evolution with thermal treatments
Calibrated using the λ/2 Si (004) reflection
• As-deposited: fcc?• 300°C: hcp• Above 400°C: transition to fcc
Analysis of XRD data
0 200 400 6000
20
40
60
80
0.0
0.5
1.0
1.5
2.0
2.5
Mic
rost
rain
, ∆d/
d ·1
0-2 (-
)
Dom
ain
size
, D (n
m)
Annealing temperature (oC)
8787
file: “S150n13500KleftisSiO2”
30 35 40 45 50 55
Siλ/2 (004)
εCo(01·0)
εCo(01·1) εCo
(00·2)
Difference
Co seed
Co0.9W0.02P0.08 as-dep.
Inte
nsity
(arb
. uni
ts)
2θ (°)
As-deposited structure?
Powder XRD, Bragg-Brentano geometry
I. Electroless Co0.9W0.02P0.08
II. Sputtered Co (2 nm thick)III. SiO2
Cross sectional phase contrast TEM image – as deposited film
Z.A. [00⋅1]
(10⋅0)
(12⋅0)–5 nm
I.
II.
III.
Fast Fourier transform
8888
0 5 10 15 200.0
0.2
0.4
0.6
0.8
1.0
400oC300oC
l (nm)
α (-
)
8 at. % 10 at. % )
r1)
C1 - 1 (
r1(V
l4 )
rr()
C1 - 1 - (1
1 2
pP2Co
hcp2
p
Co
P
⋅++
⋅=α
α - Fraction coverage of the grain boundaryrCo – Atomic radius of CorP – Covalent radius of PCP – Atomic concentration of P in the filmVhcp – Volume of hcp unit cell
Why does Co2P nucleate at ~ 420°C?
Estimation of grain boundary coverage by Passuming 1 ML of P enveloping hcp Co grains with a side length or diameter, l
Co - PEnrichment of P at the grain boundaries