Advances in Oxide Electronics
Research:
From High k Gate Dielectrics to
Diluted Magnetic Oxides
Prof. J. Raynien Kwo 郭瑞年
Part I. High k Gate Dielectrics
The Development of Oxide Electronics
in Two Decades
High Tc
Superconductors
Memory
Devices
DRAM,FRAM
Dielectrics
High k
Ferroelectrics
Sensors
Optoelectronics
Transparent
Conductors
Oxide
Electronics
Field Effect
Devices
Gate oxide
Epitaxial Oxides
Materials
Integration
Magnetic Oxides
For Spintronics
近來大力推動奈米科技的背景
來自微電子學可能遭遇瓶頸的考慮
Moore‘s Law : 摩爾定律
A 30% decrease in the size of
printed dimensions every two years.
矽晶上電子原件數每兩年會增加一倍
Intel Transistor Scaling and Research Roadmap
High Performance Logic
Low Operating Power Logic
CMOS scaling , When do we stop ?
Reliability:
25 22 18 16 Å
processing and yield issue
Tunneling : 15 Å
Design Issue: chosen for 1A/cm2 leakage
Ion/Ioff >> 1 at 12 Å
In 1997, a gate oxide
was 25 silicon atoms thick.
.
.
.
In 2005, a gate oxide
will be 5 silicon atoms
thick, if we still use SiO2
Bonding:
Fundamental Issues--• how many atoms do we need to get
bulk-like properties?
EELS -- Minimal 4 atomic layers !!
• Is the interface electronically abrupt?
• Can we control roughness?
and at least
2 of those 5
atoms will
be at the
interfaces.
Basic Characteristics of Binary Oxide Dielectrics
Al2O3
Y2O3
HfO2
Ta2O5
ZrO2
La2O3
TiO2
3.9
9.0
18
20
25
27
30
80
Band gap (eV)
Band offset (eV)
9.0
3.2
8.8
2.5
5.5
2.3
5.7
1.5
4.5
1.0
7.8
1.4
4.3
2.3
3.0
1.2
Free energy of formation
MOx+Si2
M+ SiO2
@727C, Kcal/mole of MOx
-
63.4
116.8
47.6
-52.5
42.3
98.5
7.5
Stability of amorphous
phase
High
High
High
Low
Low
Low
High
High
Silicide formation ?
-
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Hydroxide formation ?
-
Some
Yes
Some
Some
Some
Yes
Some
Oxygen diffusivity
@950C (cm2/sec)
2x 10-
5x 10-25
?
?
?
10-12
?
10-13
Dielectrics
Dielectric constant
SiO2
14
Electrical characterization / optimization
-- Good News !!
 Low electrical leakage is common .
 Have Attained an EOT under 1.0-1.4 nm .
-- Major problems are :
 High interfacial state density
 Large trapped charge
 Low channel mobility
 Electrical stability and reliability
Research Programs
 Low defect high k ultrathin films
--Interface engineering
--Electrical characterization and optimization
 Identify new material candidates for metal gate
--Metal gate/high k integration
 Integration of high k, and metal gate with Si- Ge
strained layer, or with strained Si
 High k dielectrics for high mobility semiconductors
like Ge and III-V semiconductors
Multi-chamber MBE Within-situ Analysis System
Wafer in
In-situ
SPM
oxide
MBE
annealing
& metal
chamber
III-V
MBE
in-situ
XPS
metal
chamber
oxide
& metal
MBE
Functional
chamber
Si-Ge
MBE
Major Research Topics
• Novel MBE template approach for ALD
growth
• High k dielectrics on III-V semiconductors
• Enhancement of k in the new phase
through epitaxy
• Fundamental study by IETS for detections
of phonons and defects in high k dielectrics
• Diluted magnetic oxide Co-doped HfO2
Novel MBE template for ALD growth
Can you make an excellent HfO2 Film
With low EOT ?
Interface Engineering !
Structural properties of MBE-grown HfO2
HRTEM
HRTEM
HfO2(3.6nm)
No IL !!
Deposited by MBE
IL (1.4 nm)
RHEED
Deposited by ALD
atomically order
Si(100) surface
amorphous
HfO2 surface
The MBE Template for ALD Growth
MBE and ALD composite film deposition procedure
MBE template film growth
ALD bulk film growth
MBE HfO2
ALD HfO2
MBE Al2O3
ALD Al2O3
Case 1
Case 2
ALD
ALD Al
HfO
2O23
MBE Al
HfO
MBE
2O23
Silicon
• Low pressure ( <1x 10-8 torr ) maintained during MBE growth
• ALD precursors : TMA, H2O, Tsubstrate : 300℃
The structure of ALD Al2O3 with a MBE Al2O3 template
AR-XPS
Normalized Intensity (a. u.)
TEM
Si 2p
MBE + ALD Al2O3
MBE Al2O3
108
106
104
102
100
98
Binding Energy (eV)
ALD Al2O3
MBE Al2O3
Silicon
No peak formed at 103.4eV
No SiO2 formed at interface for
both MBE and MBE+ALD Al2O3
96
ALD Al2O3 (1.9nm) with MBE (1.4nm) Al2O3 template
Ti-Au / MBE+ALD Al2O3 (3.3nm)/ p-Si / Au
Capacitance(pF)
160
Ti-Au / MBE+ALD Al2O3 (3.3nm)/ p-Si / Au
As-deposited
120
80
ALD Al2O3
MBE Al2O3
Silicon
10k
50k
100k
500k
40
2
Capacitance (uF/cm )
2.0
As-deposited
1.5
Experiment
CVC model [1]
1.0
0.5
0.0
-2
-1
0
0
Voltage
(V)
1
2
-8
-1
0
1
2
V (voltage)
Consistent with two
parallel capacitor plates
In series
k= 8.9
EOT=1.41nm
Vfb= -0.6V
Vg= -0.3V
Vg= -0.4V
Vg= -0.45V
(By conductance method)
2
G /  (S/rad/cm )
p
-2
Dit :2.0x10
2.2 E11 cm-2eV-1
-8
1.5x10
-8
1.0x10
-9
5.0x10
0.0
[1]. CVC model provide by NCSU, Dr. John Hauser
3
10
4
10
Frequency (Hz)
5
10
• High k dielectrics on III-V semiconductors
• Enhancement of k in the new phase through
epitaxy
Can you make k even higher ?
Basic Characteristics of Binary Oxide Dielectrics
Dielectrics
SiO2
Al2O3
GGG
Dielectric constant
3.9
9.0
12
Band gap (eV)
9.0
8.8
Breakdown field
(MV/cm)
10
Hydroxide
formation
Recrytallization
temperature
Thermal stability in
contact with GaAs at
high T
WSi2 gate
compatibilty
Gd2O3
Sc2O3
HfO2
14
13
15 - 20
5.4
5.4
6.5
5.7
5-8
3-5
3.5
?
5
Slight
Some
High
High
Likely
Some
> 1100C
~ 750C
> 850C
> 900C
?
~500C
?
~700C
~ 850C
~ 850C
?
?
?
?
?
Reacted at
530C
?
?
HRTEM of Low Temp Growth of
HfO2 on GaAs
Amorphous HfO2
on GaAs (100)
HfO2
56 Å
A very abrupt
transition from
GaAs to HfO2 over
one atomic layer
thickness was
observed.
GaAs
High Resolution TEM Images of Pure HfO2
on GaAs (001)
An abrupt transition from GaAs to
HfO2 and no interfacial layer
Coexistence of four monoclinic
domains rotated by 90º.
The Structure of Pure HfO2 Grown on
GaAs(001)
L
Thickness is about 43A
Monoclinic phase
a=5.116A, b=5.172A, c=5.295A, b=99.180
Coexistence of 4 domains rotated 90º
from each other
KB
HA
H
9.2º
HB
KA K
Can you make k even higher ?
Phase transition engineering !
Crystal structures of HfO2
and the corresponding k
Hf
O
29
26.17
Stable phase temperature >2700oC
Dielectric constants
70
20
>1750oC
16 *
**
<1750oC
The dielectric constant increases when
HfO2 structure is changed from monoclinic to other symmetry
* Xinyuan Zhao and David Vanderbrit, P.R.B, VOL. 65, 233106, (2002).
** G-M. Rignanese and X. Gonze, P.R.B, VOL. 69, 184301, (2004).
Structure of HfO2 doped with Y2O3
Grown on GaAs(001)
I det /I mon
15
10
14
10
13
10
12
10
11
10
10
10
9
10
8
10
7
10
6
10
5
10
4
10
3
10
2
10
1
10
0
10
-1
10
-2
10
-3
10
-4
10
3.4
GaAs(004)
l
YDH(004) L-scan
YDH(400) H-scan
YDH(040) K-scan
(004)
1x
1.1x
YDH(004)
(040)
GaAs(400)
YDH(400)
h
GaAs(040)
YDH(040)
3.6
3.8
4.0
4.2
4.4
4.6
4.8
k
(400)
Cubic phase
a=5.126A, b=5.126A, c=5.126A
α=90,β=90, γ=90
L/H/K(r.l.u_GaAs)
YDH(040)/(400) phi scans
FWHM~3.091(degree)
0.01
I det /I mon
Find many peaks:
• (022)(400)(200)(311)(31-1)(113)(420)(133)(20-2)
• All peaks of the film match the JCPDS of cubic
phase HfO2
• Use the d-spacing formula to fit the lattice
parameters
HfO2 doped with Y2O3 grown on GaAs(001) is
CUBIC
1E-3
0
90
180
270
Phi(degree)
360
Comparison between Monoclinic phase
and Cubic phase of HfO2
Without doping Y2O3
With doping Y2O3
200 L scan
200 & 220 L scan
220 Lscan
D
C
C
y
x
A
B
B
B
D
A
D
C
A
B
Monoclinic phase
D
Cubic phase
T=110A
k=32
Y2O3+HfO2
GaAs(001)
-1
100k
50k
30k
10k
10
1
10
-1
10
-3
10
-5
10
-7
10
-9
2
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
J (A/cm )
2
C(F/cm )
The Electrical Property of HfO2 doped with
Y2O3 Grown on GaAs(001)
-6
YDH/GaAs
HfO2/GaAs
-4
-2
0
2
4
E (MV/cm)
0
1
2
V (volts)
3
4
6
Electrical properties - MBE-HfO2 (0.8) /Y2O3 (0.2)
MBE-HfO2 (0.8) /Y2O3 (0.2)
Annealing Temperature effects
0
1
2
3
TiN/MBE-HfO2 (0.8)/Y2O3 (0.2)(~10nm)/P-Si
175
175
o
FG 400 C 30min
150
2.4
2
150
Cox (pF)
-3
2.8
200
125
125
100
10 nm
75
k ~ 17.9
2
3
4
5
0
2.0
-2
1.6
10
10
k ~ 10
1.2
-6
10
0.8
-8
10
25
0
1
V (voltage)
3
-4
0.4
-1
2
1x10
50
0
1
10
0.8
1k
10k
100k
-2
1.6
0
MOS diodes of TiN/HfO2(0.8)/Y2O3(0.2)/p-Si(100) 2.8
Different RTP conditions
2.4
2
k ~ 30.1
2.0
75
P-Si(100)
25
-1
1.2
MBE-HfO 2 +Y 2O 3
50
0
100
TiN
-2
2
-1
Jg (A/cm )
-2
C (uF/cm )
200
0.0
As-deposited
o
RTP 800 -3
C 20s N-22
o
RTP 900 C 20s N2
-3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.00.4
E (MV/cm)
0.0
-1
0
1
2
V(voltage)
3
4
Increase of k from 15 to over 30 !
5
Cross Sectional HRTEM Study of
The Y-doped HfO2 Films in Cubic Phase

Interfaces of YDH(100)/GaAs, and YDH(111)/Si are atomically sharp
11nm
⊙=[110]YDH
YDH
[001]YDH
[1-10]YDH
[001]GaAs
5nm
GaAs ⊙=[110]
GaAs
HRTEM image of yttrium-doped HfO2
films 11 nm thick on GaAs (001).
HRTEM image of yitturm-doped HfO2
films 7.5 nm thick on Si (111).
The Enhancement of k through
“Phase Transition Engineering”
k= (1 + 8 am/ 3Vm) / (1 -4am/ 3Vm)
Clausius-Mossotti Relation
Change of molar volume of Y doping HfO2
 The high k materials, such as HfO2, ZrO2, TiO2, or Ta2O5, commonly
have a high temperature phase with a high dielectric constant.
 We plan to achieve the enhancement through phase transition
engineering by additions of dopants such as lower valence cations,
followed by proper post high temperature anneals.
IETS Study to Detect
Phonons and Defects in High k Dielectrics
But what is inside of high k ?
Degradation of Mobility in High k Gate Stack
Phonons may
have reduced
mobility
seriously !
Fechetti et al, JAP
90, 4587, (2001).
Correction from the charge trapping effect
Inelastic Electron Tunneling Spectroscopy
 Elastic tunnel current increases linearly with the potential difference
biased on both electrodes; while, inelastic electron tunneling occurs at
the characteristic energy of vibrational levels.
 (b) The I-V curve, conductance-V curve, and the IETS spectrum would
result from both elastic processes and the inelastic tunneling channels.
Trap-Related Signatures in IETS
I
Background I-V
Trap Assisted
Tunneling
Charge Trapping
V
2
d I
dV 2
Trap Assisted
Tunneling
Charge Trapping
Wei He and T.P.Ma, APL 83,2605, (2003) ; APL 83, 5461, (2003)
V
Interactions Detectable by IETS





Substrate Silicon Phonons
Gate Electrode Phonons
Dielectric Phonons
Chemical Bonding (Impurities)
Defects (Trap States)
IETS of Si MOS Devices made of
High k Gate Dielectrics

Al/HfO2(40 A)/n-type Si

Electrical stress induced traps in
Al/HfO2/Si structure

Determination of trap energy and
location in staked HfO2/Y2O3/Si structure
IET spectrum of Al/HfO2/Si MOS structure
IET spectra of the MBE-grown HfO2(~25Å)
annealed at 600oC in vacuum for 3 minutes
-1
0
Wave Number (cm )
400
600
800
1000
200
Si LO
Si TO
d I/dV (a.u.)
Si TA
Al
Hf-O
Hf-O-Si
1200
Si-O
Hf-O
Hf-O
Hf-O
2
Al-O
2
Al-O-Hf
(a) Substrate Injection
Hf-O-Si
Si-O
(b) Gate Injection
0
25
50
75
100
V
(mV)
bias
125
150
Electrical stress induced traps in Al/HfO2/Si MOS
structure annealed at 600oC in vacuum
•The IETS spectra of HfO2 change slightly after electrical stress.
•The trap assisted tunneling are observed in IETS spectra after
Initial
After 200 sec
400 sec
800 sec
-1.2 V, DC stress1
400 sec DC stress at -1.2V.
d2I/dV2 (Arbitrary Units)
Forward-bias (gate positive)
The features of trapassisted tunneling
0
0
25
25
50
50
75
75
100
125
100
125
Voltage
Voltage(mV)
(mV)
150
150
175
175
200
200
Determination of Physical Locations and Energy Levels
of Trap in Stacked HfO2/Y2O3/Si Structure
post-deposition annealing at
o
400 C
o
600 C
in N2 ambiance for 3 min
21
4 3
Si
HfO2~ 2nm
Y2O3~ 2nm
12
Si (100)
Substrate
3 4
2
2
d I/dV (a.u.)
Charge trapping
6
98
5
5 6
7
-200 -150 -100 -50
0
50
Vbias (mV)
9
HfO2 ~2nm dt
●● ●
d0
Y2O3 ~2nm
Substrate Si(100)
Vt = V f Vr / (V f + Vr )
d t = d 0V f / (V f + Vr )
Vf and Vr are the voltages
where the charge trapping
features occur in forward
bias and reverse bias.
8
7
100 150
M. Wang et al,
200APL. 86, 192113 (2005)
APL, 90, 053502 (2007)
Major Accomplishments
First demonstration of atomically abrupt high k HfO2/Si
interface by the MBE process, and employed as a MBE template
for ALD.

 With a novel MBE template for ALD growth, we have achieved
excellent electrical properties (ID of 240 mA/mm and gm of 120
mS/mm) in HfO2 MOSFETs, compared very favorably to the
results reported by other major groups.
 Passivation of the GaAs surface by high k dielectrics HfO2 in
both amorphous and epitaxial films with sharp interfaces.
 Stablization of the cubic HfO2 phase with Y doping by epitaxy
at 600C on GaAs (100) as well as on Si (111) to achieve an
enhanced k over 33.
Major Accomplishments
• IETS has detected phonons, defects (traps), and interfacial
structures in ultra-thin HfO2/Si, and stacked HfO2/Y2O3/Si
bilayer.
• IETS is capable of monitoring the defect formation with
electrical stress applied to dielectrics.
• By using a simple mode, we are able to estimate the energy and
physical location of traps in dielectrics.
• MBE grown GGG on n-Ge(100) demonstrated abrupt interface
between oxide and substrate. The GeO2 was notably suppressed
in the MBE growth process.
• For 600°C annealing, GGG remains good dielectric property,
but HfO2 becomes leaky. The better thermal stability of GGG
makes it promising for GGG based devices.
Part II: Diluted Magnetic Oxides
Can you make HfO2 magnetic ?
---Observation of Room Temperature
Ferromagnetic Behavior in
Cluster Free, Co doped HfO2 Films
Introduction and motivation
 Both injection and transport of spin-polarized carriers are necessary
for the spintronic devices. Using diluted magnetic semiconductor
(DMS) as the ferromagnetic contact is one way to achieve this goal.
 Several models such as Zener’s model, bound magnetic polaron,
F-center theory and impurity band exchange model were used to
describe the ferromagnetism.
 The potential usage of HfO2 as alternative
high-k gate dielectrics in replacing SiO2 for
nano CMOS.
 Giant magnetic moment in Co doped HfO2
was reported recently.
F-center
Fe3+
Fe3+
Oxygen
vacancy
HR-TEM Images of
the High-T and Low-T Grown Films
High-T (700°C) grown film
Low-T (100°C) grown film
Characterization of
High T (700ºC) Grown Films
Magnitude of Fourier Transform
50 20%
40
4%
30
2%
20
10
1%
0
2
4
6
8
10
20%
4%
 EXAFS of the high-T grown samples
2%
1%
0
2
4
6
R(A)
1st O shell + 2nd Co shell
2.04A
2.49A
8
10
showed a progressive formation of Co
clusters in the film (Co = 1-20 at.%).
 Superparramagnetic temperature
dependence.
 Saturation moment increases with
increasing Co doping concentration.
Co:HfO2
YSZ
1000Å
40-100℃
Magnitude of Fourier Transform
Structural Characterization of Low-T Grown Films
Co-O, N=3.8, R=1.97A
250
Co-Co, N=6, R=2.5A
200
film grown at high T
(700oC)
150
film grown at low T
(40~100oC)
100
50
0
0
2
4
6
8
R(A)
 RHEED pattern shows that Co doped HfO2 film grown at low-T is a
poly-crystalline structure.
 No sign of Co clusters was detected by EXAFS.
 Co is likely to be at the interstitial site with 2+ sate.
10
Magnetic Property of low-T grown films
Para +
Ferromagnetic
Temp. depende
 At 10K, the low T-grown Hf0.953Co0.047O2 thin film showed a Ms of 0.47 μB
/Co and a Hc of 170 Oe.
 At 300 K, the Ms and Hc decrease to 0.43 μB /Co and 50 Oe, respectively.
Magnetic Property of low-T grown films
 Ferromagnetic behavior was observed
at both 10K and 300K.
Substrate
Temperature (℃)
40~100 40~100 40~100
Doping
concentration
(at.%)
4.7
5.5
10
Ms at 10K (μB/Co)
0.47
0.36
0.13
Ms at 300K(μB/Co)
0.43
0.29
0.1
 The magnetic properties are
stable after annealing in O2 at 350oC.
The magnetic moment decreases with
 Correlation between saturation
increasing Co doping due to enhanced
magnetization with the
dopant dopant associations.
concentration of oxygen vacancies
F-center Exchange Mechanism:
--An electron orbital created by an oxygen vacancy with trapped
electrons is expected to correlate with magnetic spins dispersed inside
the oxides.
Impurity-band Exchange Model :
--The hydrogenic orbital formed by donor defect (like oxygen
vacancy) associated with an electron overlaps to create delocalized
impurity bands.
-- If the donor concentration is large enough and interacts with the
magnetic cations with their 3d orbitals to form bound magnetic
polarons leading to ferromagnetism.
γ3δp ≈ 4.3, where γ = ε(m/m*)
δp : polaron percolation threshold
Xp : cation percolation threshold
 H : hydrogenic radius
Theoretical Analysis
Using the Impurity Band Exchange Model
 p
Material ε
m*/m γ
γH
(nm)
δP
(10-6)
ZnO
4
0.28
14
0.76
1500
TiO2
9
1
9
0.48
5900
SnO2
3.9
0.24
16
0.86
1000
HfO2
15
0.1
150 7.95
1.27
Al2O3
9
0.23
39
72
2.07
δp : polaron percolation threshold
Xp : cation percolation threshold
 H : hydrogenic orbital radius
and xp are two landmarks on the magnetic phase diagram.
 Ferromagnetism occurs when δ > δp and x < xp.
 The δp of HfO2 based DMO was deduced approximately from 1.27×10-6 to
8.15×10-5
 The appearance of ferromagnetic insulator behavior in the TM doped HfO2
is more likely than the widely studied TM doped ZnO, TiO and SnO .
Major Accomplishment
 Uniformly doped Co in HfO2 thin films without clusters or second
phases were achieved by low-T MBE growth as confirmed by
EXAFS, and are stable up to 350℃ anneals under most ambient.
 Ferromagnetic behavior in B-H loops and temperature dependence
was observed at both 10K and 300K, and magnetic moment of Mn
decreases with increasing Mn concentration (4-10 at.%).
 Have ruled out the reported “giant magnetic moment effect”.
 Found qualitative correlations between the values of saturation
magnetization with the concentration of oxygen vacancies.
 Consistent with a recently proposed F-center exchange mechanism,
or a impurity band exchange model to account for the apparent
ferromagnetism.
 The appearance of ferromagnetic insulator behavior in the Co
doped HfO2 is more likely than Co doped ZnO, TiO2, and SnO2
systems for doping concentrations under cation percolation
threshold.
Acknowledgement
NTHU---Prof. Minghwei Hong, Prof. T. B. Wu
Prof. Y. L. Soo
NSRRC---Dr. Chia-hong Hsu, Dr. H. Y. Lee
CCMS---Dr. M. W. Chu, Dr. S. C. Liou,
Dr. C. H. Chen
NTU---Prof. C. W. Liu
NCTU---Prof. A. Chin
A.S.---Dr. S. F. Lee
And All of Our Group Students
Group Members: 何信佳博士、J. P. Mannaerts、吳彥達、李
昆育、 李威 縉、朱其新、李毅君、潘欽寒、徐雅玲、邱亦欣、
黃建中、杭孟琦、張宇行、洪宏宜 、游璇龍 、謝政義
Acknowledgement :
Thanks to the Support of the NSC National Nano Project
UST/CNST Project , and ITRI/ERSO Project.