Transcript Document

novel dielectrics for advanced
semiconductor devices
Cristiano Krug and Gerry Lucovsky
Department of Physics
North Carolina State University
outline
band edge states - nanocrystalline HfO2 and ZrO2
theory and experiment
inherent limitations
engineering solutions
band edge states - non-crystalline Zr and Hf silicate alloys
theory and experiment
inherent limitations
engineering solutions
novel device structure
experimental result
proposed device structures
research plan
band edge states
nanocrystalline HfO2 and ZrO2
theory and experiment
inherent limitations
engineering solutions
theory -- crystal field and Jahn-Teller term-splittings
model calculation - ZrO2 band edge d-states
2
two issues
realtive energy (eV)
1.5
T2g
C-F
C-F
1 cubic
0.5
spherically
symmetric
0
Eg
-0.5
J-T
tetra.
J-T - orthorhombic
J-T
mono.monoclinic
mono.
z-gb
x-gb
can XAS detect
mixtures of
tetragonal and
monoclinic nanocrystallites?
and
can mixtures
account for range of
defect state energies
in electrical
measurements ?
HfO2
O K1 edge
RPECVD
6
4
2
0
-2
530
Eg (2)
532
536
ZrO2
O K1 edge
reactive
evaporation
0.3
0.2
0.1
0
-0.1
T2g (3)
534
0.4
derivative absorption (arb. units)
derivative absorption (arb. units)
nano-crystallite grains - different for different processing
Stefan Zollner’s results at Freescale - XRD and SE
Eg (2)
538
photon energy (eV)
as-deposited MO-RPECVD
films by IR are monoclinic
similar result for ZrO2
-0.2
531
532
533
534
T2g (4+)
535
536
537
538
539
photon energy (eV)
multiple features in T2g region
are indicative of mixture of
monoclinic and tetragonal
by XRD
theory -- crystal field and Jahn-Teller term-splittings
model calculation - ZrO2 band edge d-states
2
realtive energy (eV)
1.5
T2g
C-F
C-F
1 cubic
0.5
spherically
symmetric
0
Eg
-0.5
J-T
tetra.
J-T - orthorhombic
J-T
mono.monoclinic
mono.
z-gb
x-gb
can XAS detect
mixtures of
tetragonal and
monoclinic nanocrystallites?
YES
model predicts at
least 4 features in
T2g band
observed for
reactive
evaporation, but not
for MO-RPECVD
p-bonded d*-states/defects at conduction band edge
in absorption constant (e2) and conductivity (PC)
onset of strong optical absorption - lowest Eg state nc-ZrO2
B. Rogers,
S. Zollner
absorption constant (cm-1)
106
T2g
(1 of
3 features)
Eg
(2 features)
105
band edge
"defect state"
optical
band gap
104
5
5.5
6
6.5
7
7.5
photon energy (eV)
8
8.5
9
[photoconductivity]1/2 (arb. units)
4000
ZrO2
V.V. Afanase'v
A. Stesmans
3000
3500
Eg band edge
feature
2500
2000
1500
band edge
"defect state"
1000
optical
band gap
500
0
4.5
5
5.5
6
photon energy (eV)
model calculations indicate band edge defect state is
associated with a Jahn-Teller distortion at internal grain
boundary and is intrinsic to nano-crystalline thin films
6.5
localized band edge J-T
d*-states
inherent asymmetry in
transport and trapping
including BTI’s
trapping/FrenkelPoole transport
tunneling but
not F-P x'port
trap depth
0.5-0.8 eV, same state
PC and band edge abs.
Z . Yu et al., APL 80, 1975
(2002),
in Chap 3.4 - High-K
dielectrics, M. Houssa (ed),
IOP, 2004.
crystal field and Jahn-Teller term-splittings
model calculation using Zr and O atomic states
2
realtive energy (eV)
1.5
T2g
C-F
C-F
1 cubic
0.5
J-T - orthorhombic
J-T
mono.monoclinic
mono.
z-gb
x-gb
spherically
symmetric
0
Eg
-0.5
J-T
tetra.
can mixtures
account for range of
defect state energies
in electrical
measurements ?
YES
3x energy
scale
~ 0.5 - 0.8 eV
engineering solutions
NEC solution
limit applied bias so that injection into
band edge defect states is not possible
modify band tail states by alloying with
divalent (MgO) or trivalent oxides (Y2O3)
e.g. Y2O3 in cubic zirconia introduces vacancies random
distribution gives cubic structure and eliminates J-T term
splittings,
but
evidence for absorption associated with excitations
to/from midgap state
issue: is this state electrically active ?
study has just been undertaken
bulk x'tal-90.5% ZrO 2
9.5% Y2O3
6
5
term-spitting removed - but new
absorption band at ~4.1 eV
4
3
6.3 eV
2
200
1
5.3 eV
0
5
5.5
6
6.5
7
7.5
8
8.5
9
imaginary part of dielectric constant (e2)
photon energy (eV)
8
nanocrystalline
nc-ZrO2
ZrO2
7.0 eV
6
Eg
(2 features)
4
5.7 eV
0
6.5
100
50
4
4.1
eV
4.5
photon energy (eV)
2
6
150
0
3.5
6.2 eV
5.5
2
ZrO2-9.5%Y
ZrO
2 -9.5%Y O23O3
"cubic zirconia"
cubic zirconia
8.5 eV
T2g
(1st feature)
5
, absorption constant (cm-1)
imaginary part of dielectric constant (e2)
VUV spectroscopic
ellipsometry and UV-VIS
transmission
7
7
7.5
photon energy
(eV)
A
8
8.5
9
sub-band-gap absorption - O vacancies
Jahn-Teller term-split d-states of nc-ZrO2
not in Y-Zr-O, but edge broadened
outline
band edge states
non-crystalline Zr and Hf silicate alloys
theory and experiment
inherent limitations
engineering solutions
IR results - GB Rayner - PhD thesis, NCSU
450
810
900°C
800°C
700°C
Absorbance (a.u.)
Absorbance (a.u.)
x = 0.23
1066
x = 0.10
700°C
as dep.
1000
800
600
1600
400
grows with
increasing x in
as-films deposited
800°C
as dep.
1200
810
900°C
600°C
1400
Si-O-1 group
shoulder ~ 950 cm-1
450
600°C
1600
(a)
1068
1400
1200
1000
800
600
400
Wavenumbers (cm-1)
Wavenumbers (cm-1)
(b)
900°C
Counts (a.u.)
Counts (a.u.)
900°C
800°C
700°C
10
18
26
34
800°C
700°C
600°C
600°C
as dep.
as dep.
42
2 (Degrees)
50
58
66
10
18
26
34
42
50
58
66
2 (Degrees)
chemical phase separation
“non-crystalline” by XRD, but,
x=0.23 nano-crystalline by TEM and EXAFS
changes
continuously with
annealing in inert
ambient, Ar
SiO2 features at
1068, 810 and 450
cm-1 sharpen up
with increasing Tann
Figure 3.7 The (a) IR absorption spectra and (b) XRD results of x = 0.23 alloy as a
function of annealing temperature up to 900C.
comparison of extended x-ray absorption fine structure
and x-ray diffraction
Zr–O
Zr–O
(a)
(b)
900°C
As Dep.
1.0
2.0
3.0
4.0
5.0
6.0
7.0
x ~ 0.55
Fourier Transform (a.u.)
Fourier Transform (a.u.)
x ~ 0.25
Zr–Zr
0.0
Zr–Zr
8.0
900°C
As Dep.
0.0
1.0
2.0
Radius (Angstroms)
3.0
4.0
(a)
x = 0.5
6.0
7.0
8.0
Counts (a.u.)
800o C
o
700 C
600o C
(b)
tetragonal
ZrO 2
900o C
Counts (a.u.)
5.0
Radius (Angstroms)
x = 0.23
from
HRTEM
o
900 C
o
800 C
x~0.25, ~3 nm
x~0.5, ~10 nm
700oC
o
600 C
As Dep.
10
18
26
34
42
2 (Degrees)
As Dep.
50
crystallite
size
difference
for x ~ 0.25
and
x ~ 0.5
58
66
10
18
26
34
42
2 (Degrees)
50
58
66
chemical phase separation (CPS) in Zr silicate and ZrSiON
alloys after 900°C annealing
18
absorption (arb. units)
16
dE~0.5 eV
phase-separated
4
absorption (arb. units)
Zr silicate alloys
x = 60%
ZrO2
14
12
non-crystalline
n-c
10
x = 35% ZrO2
non-crystalline
8
6
530
532
534
536
photon energy (eV)
Zr, Si oxynitride
~20% ZrO2
900oC anneal
3
2
Si 3s*
(Si-O)
Zr 4d*
SiO2
1
derivative
absorption
CPS
0
538
540
530
532
534
536
538
540
photon energy (eV)
doubly degenerate Eg feature in non-crystalline Zr silicate alloys
independent of alloy composition
after 900°C anneal, chemical phase separation and crystallization
Eg narrowed/shifted 0.5 eV in Zr silicate, asymmetric in ZrSiON
statistical/mean field disruption of SiO2 network
1:1 representation of silicate alloys
(NaO1/2)x(SiO2)1-x
cs corners shared/Si atom
4
(CaO)x(SiO2)1-x
3
2
(ZrO2)x(SiO2)1-x
1
xo
(YO3/2)x(SiO2)1-x
cs 0= 0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
i) metal ions, Na1+, Ca2+, Y3+,
Zr4+, etc.. disrupt network
converting bridging Si-O-Si to
terminal Si-O1- group
ii) number of terminal groups
valence of metal ion, 1 for Na,
2 for Ca, 3 for Y and 4 for Zr
iii) connectedness of network
defined by shared corners,
Cs between SiO4/2 units
iv) Cs = 4 perfect 3 D network,
Cs= 1,0 completely disrupted
mixture of SinOm molecular
ions and metal ions
alloy composition, x
rate of network disruption increases with valence of metal ion
when normalized on a per/atom basis
for x > xo for Cs = 0, silicate is “inverted” and SinOm are minority species
pseudo-ternary (SiO2)1-x-y(Si3N4)y(ZrO2)x alloys
remote plasma enhanced chemical vapor deposition
(SiO2)0.4(Si3N4)0.25(ZrO2)0.35
[b]
1000C
signal level (a.u.)
absorption (a.u.)
[a]
1100C
1000C
900C
900C
as dep.
as dep.
1500
1200
900
600
-1
wave number (cm )
300
538
536
534
532
530
528
O 1s core state energy (eV)
as-deposited amorphous alloy – significant Si oxynitride bonding
after anneal at 1000°C – chemical phase separation into
SiO2, nanocrystalline ZrO2 with N-bonding
pseudo-ternary (SiO2)1-x-y(Si3N4)y(ZrO2)x alloys
remote plasma enhanced chemical vapor deposition
(SiO2)0.3(Si3N4)0.4(ZrO2)0.3
[d]
1000C
signal strength (a.u.)
absorption (a.u.)
[c]
1100C
1000C
900C
900C
as dep.
as dep.
1500
1200
900
600
300
538
536
534
532
530
528
-1
wave number (cm )
O 1s core level energy (eV)
as-deposited amorphous alloy – significant Si oxynitride bonding
after anneal at 1000°C – no chemical phase separation and
self-organization encapsulating ZrSiO4 bonding groups
viable engineering solution, k~9-10, EOT to 0.7-0.8 nm
novel device structures
(one example)
Ge – direct deposition of SiO2 with
& without pre-oxidation, 0.5-0.6 nm
experimental results for Ge-SiO2
same as RPAO step for GaN
no preoxidation C-V is as good
as the best discussed by
Saraswat of Stanford Univ. at
Workshop on Future Electronics
2005
~Vfb
-Qf
two approaches
Dit
0.4 W-cm n-type - Al
pre-oxidation of Ge leads to an
increase in Dit, but a decrease in
negative fixed charge – next step
interface nitridation!
i) 15 oxidation followed by
plasma nitridation
ii) grow 3-5 atomic layers of
pseudo-morphic Si on Ge and
oxidize surgically to prevent
Ge-O bond formation use on-line
AES
this worked in mid-late 80's, but
was not followed-up
research plans
device testing - ZrO2-Y2O3 and
atomically engineered ZrSiON alloys
nitrided Ge interfaces - two approaches
nano-scale vertical p-n junctions
(~25 nm diameter!)
a precursor to vertical MOS devices
(SRC patent application in process)