FIB and the sample

FIB and the sample
Hugo Bender
Chris Drijbooms, Patricia Van Marcke,
Paola Favia, Olivier Richard, Jef Geypen, Koen Marrant,
Eveline Verleysen
Leuven, Belgium
4. FIB-Workshop
Focused Ion Beams in Research, Science and Technology
Halle, 29./30. Juni 2009
Outline
• Introduction : FIB and application
• Ion beam
– image quality
– curtaining
– redeposition
– discharges
• Ion beam interactions with
– Si
– metals
– Al
– Ni
• Conclusions
Hugo Bender - PT/MCA/SA
© imec 2009
2
FIB/SEM configuration
• “Dual beam”
• “CrossBeam”
• “MultiBeam”
Figures ©FEI
Hugo Bender - PT/MCA/SA
© imec 2009
3
FIB applications – Cross-section imaging
contacts still buried
Cu
gate
Si
small structures
(10-20 nm dimensions)
© Anabela Veloso
Hugo Bender - PT/MCA/SA
© imec 2009
4
FIB applications – Cross-section imaging
huge structures
(50 µm dimensions)
Si
Cu
4 to 10 h milling !
Hugo Bender - PT/MCA/SA
© imec 2009
5
FIB applications – TEM preparation
3 mm Cu grid
metal
needle
metal
needle
2 : lift-out
3 : attach to Cu grid
1 : mill “chunk” free
4 : attached to Cu grid
metal
needle
5 : final thinning
Cu grid
3 mm Cu grid
Hugo Bender - PT/MCA/SA
© imec 2009
6
FIB applications – TEM preparation
AFM
cantilever
Pt weld
Cu
grid
cut
Pt weld
omniprobe
omniprobe
diamond tip is on the backside
Cu
grid
TEM
viewing
direction
Hugo Bender - PT/MCA/SA
© imec 2009
7
FIB applications - Other
• Atom probe tips
• Back-contacting for SSRM
• Marking
• Device modification
• Micro-machining
• Biological
© Sebastian Koelling
Hugo Bender - PT/MCA/SA
© imec 2009
8
Interactions with the substrate
40-30-5-2 keV
Hugo Bender - PT/MCA/SA
© imec 2009
9
30 keV 0.92nA
Ion beam image quality
5 keV 1.0 nA
2 keV 0.77nA
particle on the surface
Hugo Bender - PT/MCA/SA
© imec 2009
10
Ion beam image quality
sample edges before “clean” milling
30 kV
5 kV
2 kV
sample edges after “clean” milling
more rounding on the milled edge
Hugo Bender - PT/MCA/SA
© imec 2009
12
Redeposition
30 keV
staircase
5 keV clean
ion-Pt
e-Pt
e-Pt
“no” redeposited
material on the surface !
redeposited
material
30 keV a-Si
Hugo Bender - PT/MCA/SA
© imec 2009
13
Redeposition
FIB staircase
SEM
10 µm BCB
via etched from the
backside
through 50 µm Si
redeposited
material
FIB
major redeposition occurs strongest
on top side of open structures
Hugo Bender - PT/MCA/SA
© imec 2009
14
Redeposition during Pt deposition
10 µm
redeposition around
2x2x0.5 µm Pt pads
Pt:
30 keV 100 pA
Pt:
5 keV 75 pA
Pt deposition = competition
deposition and milling !
© Jay Mody
Hugo Bender - PT/MCA/SA
© imec 2009
15
Curtaining
25 µm
Cu TSV
trenches with barrier/Cu seed layer only
• related to beam tails : worse at lower keV
• induced by fast/slower milling materials or topography
• can be “avoided” for TEM preparation : backside milling
Hugo Bender - PT/MCA/SA
© imec 2009
16
Curtaining : low-k/Cu
Cu SiOC
k
7
4
11
N2/O2
lowest
in situ O2 medium
N2/O2
highest
EFTEM
AFM
low k/Cu
relative
thickness
%
step from step below the lines
Cu/lowk in oxide
in Si
nm
nm
nm
40
7.6
4.9
3.4
60
5.2
4.0
2.6
80
6.4
4.4
3.5
FIB-induced steps are similar, i.e. low-k density varies due to processing
conditions, scales with k-values
S. Hens et al, Phys. Conf. Ser. 169, 415-418 (2001).
Hugo Bender - PT/MCA/SA
© imec 2009
17
Curtaining : 50µm open TSV – TEM preparation
• Deep structures : mounting orthogonal
• Very high aspect ratio structures are difficult to fill with Pt
H-bar TEM specimen : SEM
TEM
64
Cu
12
TaN
Hugo Bender - PT/MCA/SA
© imec 2009
18
Curtaining : 50µm open TSV – TEM preparation
cleaved
standard lift-out
wafer
surface
curtaining
problem
avoided
50 um
chunks ready for lift-out
specimen landed on top of the TEM grid
Cu grid
Hugo Bender - PT/MCA/SA
© imec 2009
19
Discharge damage in Kelvin structure
top view
tilted
cross-section
through damage
Hugo Bender - PT/MCA/SA
© imec 2009
20
Breakdown charge
V=
QFIB
Cox
QFIB= (E× t ox ) × (εo × εr × S/tox )
= E × εo × εr × S
• Kelvin structure : area S ~ 20000 µm2
• Breakdown field for CVD oxide ~ 10
MV/cm
• breakdown charge ~ 0.7 nC
imaging
1-4 pA :
700-175 s
crater with 2700 pA :
0.26 s
i.e contacting in advance
necessary !
Hugo Bender - PT/MCA/SA
© imec 2009
21
Ion beam interactions : Si
• many materials are completely amorphized by the
ion beam
– Si, Ge, III-V, …
– silicides
– many oxides
FIB
• indication for amorphisation :
absence of channeling
contrast
Hugo Bender - PT/MCA/SA
© imec 2009
22
Damage layer in Si 10 vs 30 keV : normal incidence
Pt
10
keV
30
keV
Hugo Bender - PT/MCA/SA
© imec 2009
23
Interaction layer 30 keV Pt depostion
Auger spectroscopy
depth profile
wedge FIB prepared
Hugo Bender - PT/MCA/SA
© imec 2009
24
Surface protective layers
• properties
– > 150 nm
– not reacting with the top layer
– contrast in TEM with top layer, preferably amorphous, light
elements
– not planarising the topography
– stress free
• options
– wafer process line : a-Si, poly-Si, stress-free nitride, …
– low-T CVD glass
– sputtered glass
– sputtered/evaporated Al or Ni
– e-beam Pt or W
Hugo Bender - PT/MCA/SA
© imec 2009
26
Au and glue protective layers
Au
Au on
glue
Hugo Bender - PT/MCA/SA
© imec 2009
27
e-Pt capping
e-Pt
• shrinkage of the low-k
and collapse of the
barrier
low-k
• Pt diffusion in a-Si
SiO2
e-Pt
e-Pt
Cu
low-k
TaN
Hugo Bender - PT/MCA/SA
© imec 2009
28
Si sidewall damage and slope
“simulated” specimen
in-situ ion-Pt covered
70 pA
150 pA
150 pA
1000 pA
+/-0.4º
Hugo Bender - PT/MCA/SA
© imec 2009
29
Si sidewall damage : 30 keV Ga
60-65 nm
20-23 nm
Hugo Bender - PT/MCA/SA
© imec 2009
30
Reducing sidewall damage : 5 keV sputtering
top
6 µm
deep
150 pA +/-0.4º
150 pA +/-0.4º
5 keV Ga 15º
Hugo Bender - PT/MCA/SA
© imec 2009
31
Reducing sidewall damage : XeF2 etching
150 pA 0º tilt
XeF2 30 s - no beam
amorphous layer etched
XeF2 : very efficient but useless due to the high roughness
also creates steep topography at interfaces
Hugo Bender - PT/MCA/SA
© imec 2009
32
Reducing sidewall damage : low-keV Ar milling
Ar thinning on both sides after 30 keV Ga thinning :
- Ar 1 keV for 6 min., angle of incidence 75º
- Ar final thinning 300 eV for 15 min.
- reduction of amorphisation to 3-4 nm
Hugo Bender - PT/MCA/SA
© imec 2009
34
Ion beam interactions : metals
• channeling contrast occurs in all freshly milled
metal, e.g. Al, Cu, Ni, W, Au, TiN, … indicating
that full amorphisation does not occur
• no channeling in TiAl3
Hugo Bender - PT/MCA/SA
© imec 2009
36
Ion beam interactions : metals
A comparison of EDS microanalysis in FIB-prepared and
electropolished TEM thin foils, C.R. Hutchinson, R.E. Hackenberg,
G.J. Shiflet, Ultramicroscopy 94, 37–48 (2003)
α Cu -β Cu4Ti
comparison of electropolished and FIB prepared specimens of Cu /
Cu4Ti :
• large increase in dislocation density attributed to the collapse and
coalescence of point defects created during milling + high defect
mobilities in metals
• no Ga is detected in the EDS spectra
Hugo Bender - PT/MCA/SA
© imec 2009
37
Ion beam interactions : metals
30 keV ion-Pt deposition
original
surface
Pt
Pt
Pt
Al
Hugo Bender - PT/MCA/SA
© imec 2009
Cu
38
Interaction layer ion-Pt on TiN
TEM
HAADF-STEM
ion-Pt :
complimentary
Pt+Ga vs C
Pt-rich
C-rich
Pt-Ga-TiN
Ga tail
wedge FIB prepared
measurements Ann De Veirman, NXP
Hugo Bender - PT/MCA/SA
© imec 2009
39
Ion beam interactions : Al on Si
• DC/magnetron sputtered Al / native oxide / Si (200 mm
wafer)
• idem covered with 150 nm low-T CVD SiO2 layer
• idem with TiN/Ti barrier, after anneal : TiN/TiAl3/Al
• in-situ lift-out in a dual-beam FIB/SEM :
– standard specimens finished 30 keV Ga
– “cross-specimen” finished 30 / 5 / 2 keV Ga
Hugo Bender - PT/MCA/SA
© imec 2009
40
Al / thin oxide / Si : TEM - HREM
30 keV Ga
Al lattice is continuous in
the dark layer
Al
Al
Al
dark contrast layer in the
Al near Al/SiO2/Si
interface
Hugo Bender - PT/MCA/SA
© imec 2009
41
Al / thin oxide / Si : HAADF-STEM
30 keV Ga
HAADF-STEM
Al
bright contrast layer :
at Al-Al grain boundaries
in the Al near Al/SiO2/Si
interface
Si
50 nm
P. Favia, EMC2008
Hugo Bender - PT/MCA/SA
© imec 2009
42
Al / thin oxide / Si : HAADF-STEM – EDS/EELS
30 keV Ga
EDS/EELS :
- accumulation of Ga in Al near the
interface
- width of the Ga profile is much larger
than on the images and depends on the
sense of the linescan
Al
Si
EDS – EELS (O)
100
50 nm
Si
O
O
100
Counts (a.u.)
Counts (a.u.)
80
Al
Ga Al
60
40
Ga
Si
50
20
0.0
0.0
5
10
15
Position (nm)
P. Favia, EMC2008
20
0.0
5
10
15
20
Position (nm)
Hugo Bender - PT/MCA/SA
© imec 2009
43
Al-Al grain boundary : HAADF-STEM – EDS
ion-Pt
EDS
80
30 keV Ga
60
CVD-SiO2
Counts
Al
40
Al
Ga
20
0.0
100 nm
accumulation
of Ga at the Al
grain
boundaries
10
20
Position (nm)
30
and
EDS – EELS (O)
60
ion-Pt
40
Al
at the CVDSiO2/Al
Counts (a.u.)
O
CVD-SiO2
40
Si
20
Ga
Al
200 nm
0.0
10
20
Position (nm)
P. Favia, EMC2008
30
40
Hugo Bender - PT/MCA/SA
© imec 2009
44
TiN / TiAl3 /Al : HAADF-STEM – EDS
30 keV Ga
100
SiGe
TiN
TiAl3
Al
80
SiGe
Ge
Al
TiN
TiAl3
Counts
60
Ti
40
Ga
Al
20
Si
TiAl3
50
500
500
nm
nm
500
500
nm
nm nm
500
100
150
Position (nm)
accumulation of Ga at TiN/TiAl3 and TiAl3/Al interfaces
P. Favia, EMC2008
Hugo Bender - PT/MCA/SA
© imec 2009
45
Al / ion-Pt – trench sidewall : TEM / HAADF-STEM
staircase + clean without or with extra tilt
“immediately” ion-Pt or e-Pt deposition
30 keV Ga
HAADF-STEM
(no tilt during clean : ~ 2º slope)
TEM
16 nm
gray contrast
ion-Pt
rounded
at the a-Si
Al
Al
ion-Pt
Si
43 nm
Si
20 nm
22 nm
a-Si
Hugo Bender - PT/MCA/SA
© imec 2009
47
Al / ion-Pt – trench sidewall : EDS/EELS
150
30 keV Ga
EDS – EELS (O)
O
ion-Pt
Pt
100
Al
Si
Counts (a.u.)
Al
50
Ga
20 nm
0.0
10
20
30
40
Position (nm)
Al / Pt-Al / Al-O / Pt
Ga profile in the bright region
Hugo Bender - PT/MCA/SA
© imec 2009
48
Al / ion-Pt – trench sidewall 30 vs 5 keV
30 keV
5 keV
gray contrast
ion-Pt
rounded
at the a-Si
Al
43 nm
Si
22 nm
6 nm
a-Si
Hugo Bender - PT/MCA/SA
© imec 2009
49
Ion beam interactions : Ni on Si
• evaporated Ni on Si
• silicide formed at the interface during the
deposition
• specimens
– chunk / plan parallel specimen : finished with 30 keV Ga and tilted to
compensate the slope
– chunk / needle specimen as for atom probe : finished with 2 keV Ga,
no tilt possible
Hugo Bender - PT/MCA/SA
© imec 2009
50
Ni on Si – 30 keV FIB lift-out - HAADF-STEM
increased
contrast/brightness
ion-Pt
Ni
Ni
95 nm
Ni
Si
Si
Ni-silicide layer
50
50 nm
nm
extra layer :
~10%Ni/90%Si
(EDS)
Si
10 nm
nm
10
E. Verleysen / S. Koellilng, tbp
55 nm
nm
Hugo Bender - PT/MCA/SA
© imec 2009
51
Ni on Si – needle finished 2 keV - TEM
slope ~11º
(SiGe multilayer)
E. Verleysen / S. Koellilng, tbp
Hugo Bender - PT/MCA/SA
© imec 2009
52
Ni on Si – needle finished 2 keV - TEM
Ni
amorphous
Ni-Si
crystalline
overhang
4 nm
a-Si
Si
E. Verleysen / S. Koellilng, tbp
Hugo Bender - PT/MCA/SA
© imec 2009
53
Ni on Si – needle finished 2 keV – HAADF-STEM
Ni-rich overhang
Ni-silicide
3.6 nm
8.7 nm
50
50 nm
nm
10% Ni / 90% Si (EDS)
E. Verleysen / S. Koellilng, tbp
Hugo Bender - PT/MCA/SA
© imec 2009
10
10 nm
nm
10
10 nm
nm
54
Ni on Si – sloped Ni thickness
e-Pt deposited
afterwards
Ga beam
70 nm
Ni-Si mixing under the Ni
of the sloped Ni edge
E. Verleysen tbp
Hugo Bender - PT/MCA/SA
© imec 2009
55
Ni reaction
• Ni reacts with the amorphized Si, forming a Nisilicide layer on the outsides of the TEM specimen
(a ring in case of needle sample)
• the “10% Ni” layer thickness :
– 30 kV
3.5 nm
– 2 kV
8.7 nm
thickness difference likely related to different
slope :
“0º” for the chunk vs “11º” for the needle
• under thin Ni also Ni-Si forms due to the Ga beam
• compare :
– Si-Cr reaction with 30 keV Ga in Si/Cr stack : Barna et al., J. Appl. Phys.
102, 053513 (2007)
Hugo Bender - PT/MCA/SA
© imec 2009
56
Conclusions - outlook
• the ion beam is always interacting with your
sample
• needs
– better low keV image quality
– faster milling systems
(plasma-FIB, higher energy, higher currents)
Hugo Bender - PT/MCA/SA
© imec 2009
57
Hugo Bender - PT/MCA/SA
© imec 2009
58
EFUG2009 : 5 October 2009, Arcachon
http://www.imec.be/efug/
Hugo Bender - PT/MCA/SA
© imec 2009
59