advantages of hard x-ray photoelectron spectroscopy ... · pdf filespectroscopy (haxpes) in...
TRANSCRIPT
Advantages of Hard X-ray Photoelectron Spectroscopy (HAXPES) in the Characterization of
Advanced Semiconductor films.
Giuseppina ContiApplied Materials Inc., Santa Clara, CA
Christian Papp, Chuck Fadley,LBNL, Berkely, CA
2
G_Conti BNL May 21,2009
Outline:
– High-k/Metal gate: Entire film stack physical characterization (thickness, composition, depth profile and chemical state).
– Soft x-ray XPS fails to provide depth profile and chemical state information of the entire film stack.
– Hard-X ray XPS using a synchrotron beam at Sping8 provides non-destructive depth profile and chemical state.
Can the entire gate stack layers be investigated without sample manipulation ?
3
G_Conti BNL May 21,2009
Metal Oxide Silicon (MOS) Transistor
Gate
Source Drain
Base
Oxide
carrierschannel
Provides control for on/off state
Insulating gate electrode from channel
Provides carriers for channel conduction
Drains away channel current
Allows S/D connection
Provides substrate bias
20ÅHfOx/SiO2
100Å TiN
Poly-SiT>500Å
4
G_Conti BNL May 21,2009
Experiment: Sample description
~100 Å
Si wafer: substrate
SiO2
HfO2
TiN
Cap layer: poly-Si
Si wafer: substrate
SiO2
HfO2
TiN
Cap layer: poly-Si
Si wafer: substrate
SiO2
HfO2
TiN
Cap layer: poly-Si~50 Å
~20 Å
~8 Å
TiN/highKinterfaceinfluences the Vt shift
Highk/SiO2 interface influences the mobility & leakage current
This sample was annealed at T=1000°C
5
G_Conti BNL May 21,2009
Soft X-ray XPS depth profile : Ar sputtering
TiNHfO2
Hf4f at the TiN – HfO2 interface Hf4f at the HfO2 – SiO2 interface
Is this Hf0
HfN, or an artifacts??
HfO2
SiO2/Si
O-Hf
Problems: 1) Is Ar sputtering changing the material chemical state?2) Is Ar sputtering increasing the surface roughness?
Problems: 1) Is Ar sputtering changing the material chemical state?2) Is Ar sputtering increasing the surface roughness?
fHf-O
012 14 16 18 20 22 24
BE(eV)
ALD HfO2 before TiN deposition
Hf4f
6
G_Conti BNL May 21,2009
Soft X-ray : AR-XPS after Ar sputtering
SiO2HfO2
TiN about 30-40A
Si wafer sub.
Log[(IXi (θ0 )/IXi (θn)]]
surface
Qualitative depth atom distribution
substrate
SiO2HfO2
TiN
Si wafer sub.
High Roughness
SiO2 Film thickness
7
G_Conti BNL May 21,2009
1st Conclusion:
The combination of Ar sputtering and AR-XPS analysis fails to provide a chemical characterization of the entire stack layer of about 150A.
The use of Ar sputtering introduces artifacts such as ion mixing and differential sputtering.
The surface after partial Ar sputtering is so rough ( roughness by AFM data increases of more than 1 order of magnitude) that AR-XPS is compromised.
So…..
8
G_Conti BNL May 21,2009
Hard X-ray photoemission
AR-XPS Data at ∼6 keV were collected atSPring8
Feasibility Study of HK-MG stack by
9
G_Conti BNL May 21,2009
350 450 550 650BE(eV)
Hf4p3
N1s
Hf4p2
Ti2pO1s
Ti2s
350 450 550 650BE(eV)
Hf4p3
N1s
Hf4p2
Ti2pO1s
Ti2s
00 100 200 300BE(eV)
coun
ts
Hf4f
Si2p
Si2s
Hf4d
C1s
Ti3s
Ti3p
00 100 200 300BE(eV)
coun
ts
Hf4f
Si2p
Si2s
Hf4d
C1s
Ti3s
Ti3p
Hard X-ray ( X-ray Energy=6000eV) : survey spectrum
☺Hf is detected without any sample manipulation
or sample sputtering
10
G_Conti BNL May 21,2009
Angle Resolved Hard X-ray data
n
θ
0
3000
6000
9000
0 50 100 150 200 250
BE(eV)
AU
theta_10_cos Theta_30-cos Hf_theta50_cosTheta 70 cos Theta 88 cos
Si2s
Si2p
Ti
Ti
Hf4f
hard-Xray
0
1000
2000
3000
4000
5000
6000
7000
8000
200 300 400 500 600BE(ev)
Cou
nts
Hf_11_88_costHf_12_70_costHf_13_50_costHf_14_30_costHf_10_10_cost
C1s Hf2p3
N1s
Ti2pO1s
Ti2s
hard-Xray
0
1000
2000
3000
4000
5000
6000
7000
8000
200 300 400 500 600BE(ev)
Cou
nts
Hf_11_88_costHf_12_70_costHf_13_50_costHf_14_30_costHf_10_10_cost
C1s Hf2p3
N1s
Ti2pO1s
Ti2s
C1s Hf2p3
N1s
Ti2pO1s
Ti2s
sample
Θ = 10°(top film surface)Θ = 88°(bulk film)
11
G_Conti BNL May 21,2009
Sample:1 Si sub/8A SiO2/20A HfO2/100A/TiN/50A/ poly-Si
SESSA
Si wafer: substrate
SiO2
HfO2
TiN
Cap layer: poly-Si~50Å
~100Å
~20Å
~8 Å
12
G_Conti BNL May 21,2009
Experimental data of angle resolved XPS hard-X-ray are fitted by Sessa Simulation
Hf 4f
010 20 30 40 50 60 70 80 90
take-off angles
AU
Hf4f_data Sessasim
Hf4f
film surface Film bulk
HfO2
Ti
2010 20 30 40 50 60 70 80 90
take off angle
AU
Ti2S dataSessa Sim. Int.
Si wafer: substrate
SiO2
HfO2
TiN
Cap layer: poly-Si
◊ Experimental data : Area of Hf4f and Ti2s
Sessa simulation: Area of Hf4f and Ti2s of the simulated spectra
13
G_Conti BNL May 21,2009
Conclusions:
•Hard X-ray AR-XPS shows the following advantages in comparison with traditional AR-XPS and Ar sputtering XPS :
•Surface and bulk information are obtained without sample manipulation
•The interface at TiN/HfO2 layer is observed and the dielectric layer consists of HfOx.
( from previous presentation at LBNL October workshop on Hard X-ray)
14
G_Conti BNL May 21,2009
Recent data analysis Chemical State Information as a function of depth
-4
1
Si2s C1s O1s N1s Ti2p Hf4p3
Relative depth plot experimental data Top surface
Bulk
))
(
)(
ln(
8830
°°
xI
xI
The intensity of SiO2 and Si° at the bottom of the film stack are too weak to be discriminated from the top Si layer
HfTiNCOSix
where
;;;;;≡
15
G_Conti BNL May 21,2009
X-ray = 6KeVData analysis : Si2pSi2p spin/orbit splitting.
400
600
800
1000
1200
5830 5840 5850
Cou
nts
/ s (
Res
id. ×
1)
Kinetic Energy (eV)
angle=70
Si2p3
Si2p1
So2pO
1000
1500
2000
2500
3000
5835 5840 5845 5850
Cou
nts
/ s (
Res
id. ×
1)
Kinetic Energy (eV)
theta=50
Si2p3
Si2p1Si2pO
500
1000
1500
5835 5840 5845 5850
Cou
nts
/ s (
Res
id. ×
1)
Kinetic Energy (eV)
angle=30
Si2p3
Si2 p1Si2pO
Spectral resolution ~0.47eV for all the take-off angles.
16
G_Conti BNL May 21,2009
Experimental data : O1s
0
1000
2000
3000
4000
5000
6000
7000
525528531534537540
BE(eV)
coun
ts/s
ec
t=88 T=70 t=50 t=30
O1s
O-Si O-Me
Realtive % O-Me
0
5
10
15
20
25
30
35
40
45
50
29 34 39 44 49 54 59 64 69 74 79 84 89
angles
%O
-Me
O1s
0102030405060708090
100
29 39 49 59 69 79 89
angles
rela
tive
%
O1s-Si O1s-Met
O-Si
O-Me
Topsurface
Si-O TiN O HfOx
3.00E+04
4.00E+04
5.00E+04
6.00E+04
7.00E+04
5408 5410 5412 5414 5416 5418 5420
Cou
nts
/ s (
Res
id. ×
2)
Kinetic Energy (eV)
angle=88
O1s-Me
O1s-Si
17
G_Conti BNL May 21,2009
0
2000
4000
6000
8000
10000
12000
450455460465
BE(eV)
coun
ts/s
ec
t=88 ( /10) T=70 t=50 t=30
0
2000
4000
6000
8000
10000
12000
14000
16000
394396398400402
BE(eV)
coun
ts/s
ec
t=88 ( /10) T=70 t=50 t=30
Ti2pN1s
Experimental data: Ti2p and N1s
N1s BE is constant for all the angle .
0.00E+00
2.00E+04
4.00E+04
6.00E+04
8.00E+04
5475 5480 5485 5490 5495 5500
Cou
nts
/ s (
Res
id. ×
2)
Kinetic Energy (eV)
angle=88angle=88 A
angle=88 B angle=88 C
angle=88 D
an gle=8 8 E
18
G_Conti BNL May 21,2009
Ti2p comparison with reference samples:1) TiN; TiOx
1000450 455 460 465 470
TiN TiO Angle=88
TiN_ref.
TiOx_ ref
Sample at θ=88
TiN
TiOx x>1
TiO
XPS International data base
19
G_Conti BNL May 21,2009
Experimental data: Hf4f
7012 14 16 18 20 22 24
BE(eV)
angle=88 angle=70 angle=50 angle=30
013 14 15 16 17 18 19 20 21 22 23
BE(eV)
AU
HfON HXray XPS HfO2
Comparison: references: HfO2 and HfON
HfO2
Angle=88
HfO2 ref.
HfON ref.
20
G_Conti BNL May 21,2009
Hf4f peak fitting
200
400
600
800
1000
1200
1400
1600
1800
5922 5924 5926 5928 5930 5932 5934
Cou
nts
/ s (
Res
id. ×
1)
Kinetic Energy (eV)
angle=88
Hf4f7N
Hf4fNO
Hf4f7O
Hf4f5O
Hf4f5N
))
(
)(
ln(
8830
°°
xI
xI
HfN HfNO HfO2
TiN interface
Substrate
21
G_Conti BNL May 21,2009
SiO2 15 Å
TiNO 20 Å
TiN 77Å
10 Å diffusion
Si 25.5 Å
Si substrate
MoSi2 4.5 Å
MoSi2 4.5 Å
Si 25.5 Å
2 Å diffusion
69 X
MoSi2 4.5 ÅSi 18 Å
2.15
2.10
2.05
2.00
1.95
1.90
Inci
denc
e A
ngle
[deg
.]
470 465 460 455 450Binding Energy [eV]
Inte
nsity
[arb
. u.]
534 532 530 528
Binding Energy [eV]
O 1s oxygen 1 oxygen 2
Inte
nsity
[arb
. u.]
470 465 460 455 450Binding Energy [eV]
Ti 2p Ti oxydized Ti and Ti satellite
2.20
2.15
2.10
2.05
2.00
1.95
1.90
Inci
denc
e A
ngle
[deg
.]
534 532 530 528Binding Energy [eV]
Standing wave : 100A TiN/ native SiO2/ MoSix mirror
Christian Papp at al.;See poster
TiN:• without cap layer• not annealed
22
G_Conti BNL May 21,2009
Suggested stack layer structure by H-AR-XPS and by standing wave
analysis
Summary
HfON 19Å
SiO2 16 Å
Poly-Si ~30Å
HfO2Si substrate
TiNO 80Å
Ti
2010 20 30 40 50 60 70 80 90
take off angle
AU
Ti2S dataSessa Sim. Int.
TiN-Hf interface
TEM
Si substrate
HfOx
TiN
Film stack was annealed at T=1000C for mimicking Gate First device
23
G_Conti BNL May 21,2009
Conclusions:
•The chemical characterization of Si, O, Ti and Hf shows that our film does not have sharp interfaces between each layer.
•The presence on O in the TiN layer could explain the higher than expected resistivity.
•The intermixing of TiN with HfO2 explains the large leakage current of this sample.
•For the first time the diffusion of N into the HfO2 layer has been demonstrated.
•The intermixing layer of N-Hf-O is about 19Å
•This work shows that HAXPES, including standing wave excitation, is a powerful technique providing useful information for advanced semiconductor film development.
•Collaborations between industry and universities may greatly help in developing advanced devices.
24
G_Conti BNL May 21,2009
Gate
Source Drain
Base
Oxide
carrierschannel
20ÅHfOx/SiO2
100Å TiN
Poly-SiT>500Å
1. Metal gate characterization: New films under investigations.
2. High K dielectric with K > HfO2 and EOT < HfO2 .
3. Ultra shallow junction As and B dopants distribution and chemical state
Outlook : HAXPES applications
Si based- Photovoltaic cells: 1. AZO characterization: Zn-O atomic distribution. Where is Al?