processing and defect control in advanced ge technologies · processing and defect control in...
TRANSCRIPT
Processing and Defect Control in Advanced Ge Technologies
C. Claeys*
IMEC, Kapeldreef 75, 3001 Leuven, Belgium*Also E.E. Dept, KU Leuven, Kasteelpark Arenberg 10,
3001 Leuven, Belgium
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 2
Outline
– Introduction/Motivation
– Ge Fabrication Techniques
– Processing Aspects
Wet cleaning
Surface passivation
Shallow junction formationGermanidation
– Conclusions
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 3
Introduction
Interest in defects due to their impact on: physical processes (e.g. diffusion)electrical device performance yield
Interest since the early days of semiconductors, but now there is physical insight
No longer trial and error but ENGINEERING
Origin of defects can be
Grown-in (dislocations, vacancies, interstitials, swirls, COPs……)
Process-induced (dislocations, precipitates, metals, twinning,…)
Scaling is putting stringent requirements on the resolution of the analytical techniques
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 4
Introduction
• DownscalingChannel doping levels: UTSOI
High-κ dielectrics
FUSI/Metal gates
carrier mobility control
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 5
Scaling Aspects: Mobility Boosting
Use of high-κ dielectrics + metal gate + strained Si
S. Datta et al, IEDM 2003, p. 653
35% increase for strained Si/SiGe
n-MOS
SiON/Control Si
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 6
Why are we interested in Ge ?
• Ge has been shown to be a good material to make transistors….
Ge
Si
GaAs (electrons)
… and it has a high (low field) electron and hole mobility
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 7
Ge used in first transistor radios
• On October 18, 1954, the U.S. company I.D.E.A. announced Regency TR1.– Used 4 germanium transistors.
• Sony's first radio, TR-55, in 1955– Used 5 germanium transistors.
Regency TR1
Sony TR-55
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 8
Early results: pMOS and nMOS on 200mm GeOI
Partnership with
Ge Active area
Spacers0.15 μmGate
oxide
0
100
200
300
400
500
600
700
-2 -1 0 1
L=0.5μm L=0.25μm L=0.18μm
I sour
ce(μ
A/μ
m)
Vgate
[V]
W = 2μm
pMOS
0
20
40
60
80
100
0 1 2 3 4
I sour
ce(μ
A/μ
m)
Vgate
[V]
W = 2μm
0.5μm 0.15μm0.18μm
nMOS
pMOS• Good drive current>500mA/μm for
L=0.18μm
nMOS• Low drive current• Ge/high-k interface quality is
one of the main limiting factors
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 9
Experimental Results - Mobility
10
100
1000
0.1 1Effective Vertical Field (MV/cm)
Hol
e M
obili
ty (c
m2 /V
s)
Si/SiO2 universal hole mobility
Ge/Si/SiO2/HfO2/TaN/TiN
L = 10 µm, no VTA, no halo
Best peak mobilityachieved so far ~358 cm2/Vs
Mobility as a function of vertical field for L = 10 µm Ge pMOSFETs
following 350°C 20 min H2 PMA
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 10
Yoshiki Kamata Toshiba Corporation (2008)Materials Today, Jan/Feb 2008, vol. 11, no.1-2
Ge brought the microelectronics industry two Nobel prizes. The first transistor was based on Ge, as was the first integrated circuit. So why did we give it up?
Though semiconductor makers quickly switched to Si as the prime semiconducting material, Ge is now on the microelectronics research agenda again, and is being discussed as a possible replacement for Si.
It’s back to the future.
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 11
Motivation/Objective
Diffusion limited leakage current for Si diodes: 1 nA/cm2
Ge diodes: 10-4 – 10-7 A/cm2 theoretical limit at 298 K
ni 103 times higher: 103 (generation) or 106 (diffusion) higher current
Leakage current depends on the presence of defects
Impact defects on other parameters: lifetime, low frequency noise, reliability, …
Systematic study of the origin of grown-in and processing-induced defects in Ge
Defect Engineering
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 12
Czochralski Crystal Growth
Tilted 35° away [001] pulling axis
2 inch to 300 mm
Ge crystal
Higher Ge material cost ($600/kg) vs Si ($54/kg)
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 13
Defects in Ge
Vanhellemont et al., in Defects and Diffusion in Semiconductors – An Annual Retrospective VII, Trans. Tech. Publ. Inc., 230, 149 (2004)
Defect in as-grown Ge (row of dislocations)
* High-res. Ge, H-atm.crystal growth
* 30°, 60° and 90° disl.* Disl. sink for [V]
No V2-H complexesTilted 35° away [001] pulling axis
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 14
Crystal Growth Aspects
⇒ 30°, 60° and 90° dislocations: sinks for [V]
Reduction of V2-H complexes and COPs
⇒ Dopant and self-diffusion controlled by [I]
⇒ Reduced Oi content (IR: Ge-856 cm-1 Si- 1107 cm-1)Oxygen precipitation?
⇒ High quality Ge substrates for μE applications
GeOI, epitaxial Ge on Si, Ge condensation, …
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 15
Thick GeOI Process
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 16
Smart Cut Process for Ge
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 17
Defects in GeOI
K.K. Bourdelle, APL, 86 (2005) 181910
{311} defect
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 18
(Si)GeOI formation by Ge condensation
• Estimated defect density: < 1. 106 cm-2
Needs further improvement
Epitaxial deposition of SiGe on SOI
Si
BOXSi
SiGe
SOI
Ge
High T° oxidation (wet/dry)
Si
BOX
SiGe
SiO2
Ge is condensed between oxides
Si
BOX10<Ge<30 nm
SiO2
High temperature (>1000°C) oxidation of SiGe layers on SOI
50 nm
Si
SiO2(BOX)
SiGe
SiO2
Oxidation at 1150oC
Ge concentration ~ 40%
50 nm
2
1Scanning direction:
1 to 2
SiO2
SiGe
SiO2
Position (nm)
Cou
nts
100806040200
1500
1000
500
0
2 Ge
• Uniform Ge profile in SiGeCould allow to make ultra-thin (Si)GeOI substrates for fully depleted devices
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 19
Substrates – Epi Ge in Si STI
• Epi-Ge in Si STI – Utilizes Si STI module for device isolation.
– Enables monolithic integration of Ge and Si devices
Si substrate Si substrate
GeSiO2Start with Si STISelective etch of SiGrowth of epi GeAnneal & CMP
SiO2
Si substrate
Epi-Ge
SiO2STI
Si substrate
Epi-GeSiO2STI
Proof of Concept• 250 nm epi Ge• 4×108 TD / cm2
Note: TEM is without CMP.
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 20
Selective growth of Ge on Si
Selective growth of Geafter formation of STI
could reduce the active area leakage
Pure Ge on Si SiGe buffer layers
• Reduction in dislocations density needed
100 μm
Top view nomarski
• Nomarski indicates uniform deposition and smooth surface
10μm100nm thick 50nm thick
RMS:0.82nmRMS:0.97nm• AFM shows low RMS roughness for 50
and 100nm films
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 21
0.0E+00
5.0E+06
1.0E+07
1.5E+07
2.0E+07
2.5E+07
700 750 800 850 900 950Anneal Temperature (°C)
Dis
loca
tion
Den
sity
(cm-2
)
1 min3 min20 min
Substrates – Epi Ge on Si
• 200 mm of epi-Ge on Si used for most IMEC lots.– 2 μm thick, relaxed epi-Ge wafer from ASM
– As-grown: ~1×108 threading dislocations/cm2
– Dislocation density reduced by ~10x with >800°C anneal.
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 22
Crystal Growth Aspects
⇒ Same bulk minority lifetime as in Si
Less Auger recombination for high doping
⇒ More prone to metallic contamination
Fe, Ni, Co dominant, less impact Cun-type lower lifetime than p-type
Mid gap acceptors increase the leakage current
⇒ Gettering is difficult in Ge
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 23
Metals in Ge
10-8
10-7
10-6
10-5
10-4
10-3
10-2
10-1
1011 1012 1013 1014 1015 1016 1017
Germanium at 300 K
Leak
age
Cur
rent
Den
sity
per
μm
Dep
letio
n(A
./cm
2 )
Metal Concentration (cm-3)
Ni;Fe
CuMn
CoSolubility Ni at 700oC
Leakage current per μm depletion width at 300 K for differenttransition metals in function of their concentration
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 24
Ge Processing Issues
• Is Ge manufacturable in a Si processing line ?– Same toolset can be used, but processes have to be modified
• Ge surface and Ge/high-k interface quality– Passivation of interface states – EOT scaling– Cleaning is critical (severe etching during wet processing)
• Dopant activation and re-crystalization– Especially n-type activation– Leakage current of junctions
• Quality and availability of substrates– Defect density of Ge layer ?
• Main question: Ion(Ge transistor) > Ion(sSi transistor)?– Ion/Ioff ratio
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 25
Etching of Ge during wet processing
Etch rate calculated from weight loss, all treatments at Room Temperature
• HF does not etch Ge• all other solutions (incl. DI-water): measurable etch rate
Etch solution etch rate (nm/min)
H2O (dissolved oxygen) 0.005-0.00610% HF <0.001BOE <0.0011M HCl 0.010.5%HF/1M HCl 0.0061M NH4OH 0.03SC1(0.001/0.01/5) 17-27SC1(1/1/7) 200-280
Etch solution etch rate (nm/min)
H2O (dissolved oxygen) 0.005-0.00610% HF <0.001BOE <0.0011M HCl 0.010.5%HF/1M HCl 0.0061M NH4OH 0.03SC1(0.001/0.01/5) 17-27SC1(1/1/7) 200-280
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 26
Transistor Structure
Relaxed Ge on Si
Isolation
NiGe
TaN/TiN gate
HDD
Extension
n wellp+ p+
Relaxed Ge on Si
Isolation
NiGe
TaN/TiN gate
HDD
Extension
n wellp+ p+
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 27
Typical 200 mm Ge pMOS Process Flow
TiN
2 μm epi-Ge
NiGe
TaNExtensionSi/SiO2/HfO2
n well
p+ HDD
Metallization
Halo: P 25°
Dielectric: 4 nm HfO2
Sinter: 350°C 20 min H2
Metallization: TiN / Ti / Al / TiN
NiGe: 10 nm Ni, RTA, Etch, RTA
Anneal: 500°C 5 min
HDD: Ge PAI, B HDD
Spacer: 90 nm width
Extension: BF2
Gate: 10 nm TaN / 100 nm TiN
Passivation: ~6 ML epi-Si / SiO2
Isolation: ~200 nm SiO2, 45°
Wells: P 90, 180, & 570 keV
Start: 2 μm epi Ge on Si
SiO2
SiO2
SiO2
Halos
Si substrate
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 28
Ge surface preparation
• HF-dip → 0.3 nm GeOx
• HBr-dip → O-free
• HF-dip + NH3 anneal → 0.6 nm GeOxNy
• Out of the box → 1.68 nm GeOx
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
Inte
nsity
(arb
. un.
)
40 38 36 34 32 30 28 26Binding Energy (eV)
Out-of-the-box
GeOxNy
GeOx
O-free
Ge
GeO2XPS:Ge3d region
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 29
Surface Passivation
Si
HfO 2
Ge
SiO x layer
S i
HfO 2
Ge
Si
HfO 2
Ge
SiO x layer
CeO2 good passivation : Nit 1x1012 (p) 6x1011(n) eV-1cm-2
S under investigation
Good results with Si
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 30
CV curve shows a passivatedsurface and a reduced Nit
Si passivation of Ge surface
• CV curve on Si passivated Ge surface has low NitH2 post treatment further improves C-V curves
- effect is similar to the passivation of silicon with H2
Si
HfO2
Ge
0
5 10-7
1 10-6
1.5 10-6
2 10-6
-2 -1 0 1 2 3
C (F
.cm
-2)
VGate
(V)
after H2 annealHf Hf HfO OO O O OHfO2 Hf Hf HfO OO O O OHfO2
Ge
Ge
Ge
Ge
Ge
Ge
Ge
Ge
Ge
Ge
Ge
GeGe
1 MLSiSi Si Si Si Si Si
Si Si Si Si Si Si1 MLSi
Si Si Si Si Si Si1 MLSi
Si Si Si Si Si Si
Si Si Si Si Si SiSi O Si Si SiSiO O O O
O
O O O O OOthinSiO2
SiO
Si O Si Si SiSiO O O O
O
O O O O OOthinSiO2
SiO
thickness control is essential
• Si layer too thickstress-induced interface defectsC-V curve shows high Dit
• Si layer too thinGe will oxidize and defective layer with poor interface is formed
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 31
Shallow Junction Formation
Ge: lower anneal temperatures for activation (400-600°C)
Higher I/I damage in Ge, but anneal at lower temperatures
Sheet resistivity: max solubility, dopant out-diffusionactivation degree
Thermal budget: balance between dopant activation, defect annealing and junction depth
Boron shallow junction can be obtained by Ge pre-amorphizationand RTA
More difficult to control n-type dopants like P, As an Sb
Co-implantation with non-doping impurities like N and C gives promising results for P – less dose loss and lower P in-diffusion
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 32
Boron Doping – Sheet Resistivity
Annealing temperature dependence at 60 s
0
20
40
60
80
100
120
300 400 500 600 700
Annealing T (oC)
Rs
(Ohm
/sq)
1e16 at/cm21e15 at/cm2
Sheet resistance versus RTA temperature (60 s), corresponding with a 4.5 keV B 1015 or 1016 at/cm2. A Ge pre-amorphization implantation was done at 120 keV & 1015 at/cm2. A junction depth at a carrier concentration of 1018 cm-3 of ~70 nm
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 33
As-implantedDepth of amorphous layer: 100 nm
Annealed at 400C, 60s, N2
• Fully re-grown• No visible residual defects at the EOR
Re-crystallization of pre-amorphized, B-doped Ge
2
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 34
Sb Doping – Void Formation
SEM image (x150 000) of a 10 nm SiO2/Ge sample implanted with 1015 at/cm2 Sbat 70 keV, showing voids extending up to 100 nm below the surface
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 35
P (as-)implanted and re-grown Ge: microstructure
After re-growth at 400C, 60 s, N2
• P-doped Ge fully re-grows at 400C (60 s) • No evidence of P clusters or residual damage after re-growth at low T• Inactive P may be in small P-V clusters or interstitials
60 nm
as-implanted P 25 keV 3e15 at/cm2
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 36
Control P doping in Ge - Precipitation
0100200300400500600700800
1E+13 1E+14 1E+15 1E+16Dose (at/cm2)
Rs
(Ohm
/sq)
P, 500C-60sP, 500C-1sP, 600C-1s
Above SS implant
15 keV P in Ge; no SiO2 cap
15 keV 5x1015 cm-2 P, 60 s at 500°C
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 37
Impact NiGe on ID
10-12
10-10
10-8
10-6
10-4
10-2
-1-0.8-0.6-0.4-0.20
Impact of NiGe
NiGe junction
No NiGe junction
Forw
ard
curr
ent (
A)
Forward voltage (V)
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 38
Impact NiGe on IR
10-10
10-9
10-8
10-7
10-6
10-5
10-4
00.20.40.60.81
Impact of NiGe
NiGe junction
No NiGe junction
Rev
erse
cur
rent
(A)
Reverse voltage (V)
10-12
10-11
10-10
10-9
10-8
10-7
10-6
00.20.40.60.81
Impact of NiGe
No NiGe junction
NiGe junction
Rev
erse
cur
rent
(A)
Reverse voltage (V)
10-10
10-9
10-8
10-7
10-6
10-5
10-4
00.20.40.60.81
Impact of NiGe
NiGe junction
No NiGe junction
Rev
erse
cur
rent
(A)
Reverse voltage (V)
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 39
Junction Diode: Theoretical Analysis
Geometrical componentsbulk – surface – corner
IR = AJA + PJP +NCJC + Ipar
3 diodes structures with different P/A ratios should be measured
Physical componentsdiffusion - generation
⎟⎟⎠
⎞⎜⎜⎝
⎛+=
Dp
p
An
nidA NL
DNL
DqnJ 2
geff
AigA
WqnJτ
=
n+
pWA
+-
e
h
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 40
Junction Diode: Graphical Technique
0
0.5
1
1.5
2
2.5
3
0 0.5 1 1.5 2 2.5
I/A= JA+ P/A * JP
y = 0.17422 + 1.2777x R= 0.98903
I/A (
A/ m
2 )P/A (1/um)
For a Fixed VFor a Fixed VRR
PA JAPJ
AI
+=
Intercept!! Slope!!
PApar
CC
PAR J
APJ
AI
JA
NJAPJ
AI
+≈+++=
If there isn’t corner generation component
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 41
Impact NiGe and PMA
S. Sonde, E. Simoen, C. Claeys, A. Satta, B. De Jaeger, G. Nicholas and M. Meuris, in Avanced Gate Stack, Source/Drain and Channel Engineering for Si-Based CMOS 3: New Materials, Processes and Equipment, Electrochem. Soc. Trans., vol. 6, no. 1, 2007, pp. 31-39
10-3
10-2
350 400 450 500 550 600
Area component: Algebraic method
NiGe junction/ 0.5V
No NiGe junction/ 0.5V
Ja (A
/cm
2 )
PMA temperature (oC)
10-7
10-6
10-5
350 400 450 500 550 600
Perimeter component: Algebraic method
NiGe junction/ 0.5V
No NiGe junction/ 0.5V
Jp (A
/cm
)
PMA temperature (oC)
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 42
Experimental : Impact NiGe
void
Nickel
NiGe on Ge
SiO2 isolation
Small defects
270°C / Selective Etch / 330°C
1 μm
Germanide overgrowthon SiO2 isolation
NiGe on Ge
Massive voiding
330°C / Selective Etch
1 μm
D. Brunco et al.
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 43
Impact NiGe on LF noise in p+n Ge junctions
Low-frequency noise spectra for a germanided and a non-germanidedp+-n junction at a forward bias of 0.5 V. The area is 15000 mm2 and the perimeter P=10300 mm.
10-23
10-22
10-21
10-20
10-19
10-18
10-17
100 101 102 103 104 105
Noi
se S
pect
ral D
ensi
ty (A
2 /Hz)
Frequency (Hz)
no NiGe
NiGeV
F=0.5 V
A: 15000 μm2
P: 10300 μm
R.M. Todi, S. Sonde, E. Simoen, C. Claeys and K.B. Sundaram, Appl. Phys. Lett., vol. 90, no. 4, pp. 043501-1/3, 2007.
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 44
Germanidation: Impact on LF Noise
SI at 1 Hz and 100 mA for NiGe germanided and non-germanided p+-n junctions versus area.
10-18
10-17
103 104 105 106
Noi
se S
pect
ral D
ensi
ty (A
2 /Hz)
Junction Area (μm2)
~A0.39
NiGe
no NiGe
f=1 HzIF=100 μA
R.M. Todi, S. Sonde, E. Simoen, C. Claeys and K.B. Sundaram, Appl. Phys. Lett., vol. 90, no. 4, pp. 043501-1/3, 2007
SI is proportional with A0.39
Defect-assisted GR noise origin
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 45
Germanidation : Ni
DLT-spectra at tw=51.2 ms, tp=1 ms and a pulse from -4 to -1 V for a 4 mm diameter NiGe Schottky barrier on n-type Ge and corresponding with different RTA temperatures: 300, 350, 400 and 450oC
0 100
2 10-3
4 10-3
6 10-3
50 100 150 200 250
n-type Ge - Ni: 10 nm
DLT
S (p
F)
Temperature (K)
-4-->-1 Vtw=51.2 mstp=1 ms
300oC350oC
450oC
400oC
Low temperatures used
Phase = f(T)
300-350ºC: no deep electron traps
400-450ºC: double acceptor at Ec-03 eV – substitutional Ni
Low defect concentration
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 46
Germanidation : Pd and Pt
DLT-spectra corresponding with a Ni-, Pt- and Pd-germanided n-type Ge Schottky barrierfor an anneal temperature of 400oC. A bias pulse from -4 -1 V was applied for 1 ms, witha repetition time constant tw=51.2 ms. The Ge: Ni and Ge: Pd spectra are offset in they direction for reasons of clarity
30 nm metal deposited by Arsputtering
400ºC: V-related traps
Sputter damage
Low defect concentration
-1 10-2
-5 10-3
0 100
5 10-3
1 10-2
50 100 150 200 250 300
Tanneal
= 400oC
DLT
S (p
F)
Temperature (K)
Ge: Ni
Nis-/2-
Ge: Pt
Ge: PdSb-V ?
-4-->0 V
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 47
Germanidation : Pd and Pt
I-T curves for the Pt-germanided n-type Ge Schottky barriers and for the pure Pt Schottkybarrier indicated by as-deposited
300ºC –Pt2Ge400ºC: PtGe450ºC: PtGe2
200ºC: Sputter damage reduces
10-12
10-10
10-8
10-6
10-4
0 50 100 150 200 250 300
30 nm Pt on n-Ge
Rev
erse
Cur
rent
(A)
Temperature (K)
VR=-4 V
as-deposited
Tanneal
=300 oC400 oC
500 oC
E. Simoen, K. Opsomer, C. Claeys, K. Maex, C. Detavernier, R.L. Van Meirhaegheand P. Clauws, J. Electrochem. Soc. Vol. 154, no. 10, 2007, pp. H857-H861
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 48
Conclusions
Good performing p-channel transistors have been fabricated
mobility up to 350 cm2/Vs for L =0.12 μm
NiGe reduces the series resistance but increases the perimeter leakage current and lowers the LF noise
voids in the silicon substrate
PMA reduces the leakage current Optimum 450-500°C
Good understanding Ge processing
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 49
Ge pMOS vs. ITRS roadmap
• Ion and Ioff vs. gate length for (unstrained) Ge pMOSITRS specifications for Si pMOS are used
Ion: High Performance ITRS spec seems obtainable with Ge pMOS
Ioff: High Performance ITRS spec seems obtainable with GeOItechnology
On-state currents Off-state currents
10-710-610-510-410-310-210-1100101
0.01 0.1 1
Ge, ISOURCE
Ge, IDRAINO
ff-c
urre
nt (μ
A/μ
m)
Gate length (μm)
ITRS HP Spec
VDD
=1V
101
102
103
104
0.01 0.1 1
Ge, VDD
=1VGe, V
DD=0.6V
On-
curr
ent (μA
/μm
)
Gate length (μm)
ITRS HP Spec
Ge pMOS could fulfill ITRS specs, for GeOI substrates
G. Eneman et al.
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 50
What about Ge nMOS?
• Low mobility/drive current measured for n-type MOSFETs (also in literature) – Si IL cannot passivate Ge nMOS. Accurate Dit analysis only at 80K [1]
– Other potential problem: poor dopant activation for n-type dopants(activation limit for P ~ 6 × 1019 cm-3)
0
50
100
150
200
250
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Mob
ility
(cm
2 /Vs)
Vg (V)
pMOS
nMOS
[1] K. Martens et al., EDL 27(5), 405 (2006)
0
1 1013
2 1013
3 1013
4 1013
0 0.2 0.4 0.6 0.8In
terf
ace
stat
e de
nsity
(1/c
m2 /e
V)
Valence band offset (eV)
80Kn-type capacitor
Val
ence
ban
d ed
ge
Con
duct
. ban
d ed
ge
Ge nMOS and pMOS mobility Interface state density
Consider other options for nMOS. nFET with GaAs?
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 51
Ge and III/V transistors
CMOS: Work towards combination of Si, Ge and GaAs or other III/V materials on one substrate
Si CMOS
Si wafer
strained Ge pMOS
Strained Ge selectively grown on Si or GeOI substrate
GaAs or other III/V nMOS
GaAs or other III/V selectively grown on Ge
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 52
Acknowledgements
The author wish to acknowledge D. Brunco,P. Clauws, B. De Jaeger, G. Eneman, W. Guo, M. Meuris, G. Nicholas, K. Opsomer, R. Todi, A. Satta, E. Simoen, S. Sonde, J. Vanhellemont for useful discussions and the use of co-authored results.
Explore team at IMEC
IMEC Core Partners:
IEEE ED Colloquium/Orlando/Feb. 21, 2008C. Claeys 53
Conclusions