Upload
vuhuong
View
214
Download
0
Embed Size (px)
Citation preview
Bruker Corporate Profile
Bruker: The leading provider of high-performance scientific instruments and solutions for life science, diagnostic and applied markets. Founded in 1960
$1.79B revenue
>6,600 employees
>80% of revenues outside of the US
Direct sales organizations in 33 countries
1,000 R&D professionals with strong track record of innovation
KEY FIGURES
Bruker BioSpin
BrukerCALID
Bruker
Nano
03/05/2017 Bruker Confidential 2
We have distinct capabilities and leadership positions in many product lines
03/05/2017
Bruker
BioSpinBruker
CALIDBruker
NANO
Industries Served
1. Academic
2. Government
3. Pharma Biotech
Industries Served
1. Pharma/Biotech
2. Industrial
3. Academic
4. Diagnostics
Industries Served
1. Industrial
2. Semiconductor
3. Academic
Indicative: not representative of all products sold and industries served
Bruker Confidential 3
Bruker NANO Group Profile
Bruker Advance X-ray
Solutions Division
(AXS)
Bruker Nano Surfaces
Division
(BNS)
Bruker Nano Analytical
Division
(BNA)
Over $500M Revenue
~60% Industrial
~40% Academic/Gov’t
~24% N Amer; 32% Europe; 44%
APAC/Japan
~Over 280 R&D professionals
BNANO GROUP KEY FIGURES
Bruker Semiconductor
Division
(BSD)
New
Bruker Semiconductor Division
Formed January 2015 to focus on the
needs of Semiconductor customers
November 2015: Acquired Jordan
Valley Semiconductors
Bruker Nano Group
03/05/2017 Bruker Confidential 4
What can HRXRD give us, who uses it and for what?
• High-resolution X-ray diffraction (HRXRD) provides a wealth of information about epitaxial materials
• Crystal lattice misfit/strain, tilt and defectivity/quality…
• Composition and thickness
• Arrangement, shape and lattice distortion in arrays of patterned structures
• It is first-principles (no calibration) and non-destructive characterization and metrology technique
• Does not require material/process dependent optical constants
• Accurate and precise with very few assumptions
• Has been used for 30+ years in the compound semiconductor industry for a wide range of materials (III-V, III-nitride, II-VI…) and devices (LEDs, lasers, CPV, detectors…)
03/05/2017 Bruker Confidential 5
History of Production HRXRD
2000
-3000 -2000 -1000 0 100010
-6
10-5
10-4
10-3
10-2
experiment
simulation
Refl
ecti
vit
y
Delta Theta [ arcsec ]
QC200
First 200mm system with automated fitting
1986 1990 2002 2004 2008 2015
310
410
-250 -200 -150 -100 -50 0 50 100 150
Inte
nsit
y (c
ps)
AXIS 2 (arcsec)
QC1 / QC2
First PC Diffractometers
JV QC3
Production tool for GaN LEDS
010
110
210
310
410
510
-6000 -5000 -4000 -3000 -2000 -1000 0 1000 2000 3000
Standardized Chi Squared: 0.039 R Factor: 17.19
Inte
ns
ity
Seconds
Reference Comparison 1 (Active)
BedeMetrix family
First fully automated 300mm system
JVX7300LSI
Fully automated fast RSMs
Fab200
First fully automated 200mm system
CS to Si HBT
CS developments
Si developments
03/05/2017 Bruker Confidential 6
Silicon semiconductor metrology tools
• Tools: JVX7300 series
• Channels: XRF, XRR and HRXRD Applications:
• 7300HR: SiGe & Si:C on bulk or (FD)SOI, various ALD films, HKMG stacks, silicides…
• 7300LSI: Ge and III/V on Si for sub-10 nm, HKMG, FinFETs, GaN-on-Si, MEMS
• 7300F(R): Metal / magnetic films, WLP
• In-line tools for silicon semiconductor device manufacturers for process control of product wafers
JVX7300LSI
03/05/2017 Bruker Confidential 7
HRXRD in Si Logic
• Introduced into the Si industry with strain engineering for sub-100 nm logic devices
• Epitaxial SiGe S/D stressors forPMOS mobility enhancement
• Also Si:C / Si:P S/D stressors for NMOS, but less widespread
• Used for R&D, CVD chamber qual., process diagnostics / ramp and in-line metrology
• Solid metrology pads less relevant / not available
• Transition from planar to 3D (FinFET) devices
• Novel channel materials,e.g. SiGe, Ge and III-V for sub 10 nm nodes
Source: A. Steegen, “Logic Scaling Beyond 10 nm”, IMEC Technology Forum US (July 2013)
03/05/2017 Bruker Confidential 8
Principle of X-ray diffraction based stress/strain analysis
• X-ray diffraction uses the crystal lattice as a “strain gauge”
• The relation between the lattice parameter and diffraction angle is defined by Bragg’s law, 2𝑑 sin 𝜃𝐵 = 𝑛λ
• Most sensitive stress/strain analysis method for semi. (ITRS 2011)
• Broadening of peaks gives dislocation densities
03/05/2017 Bruker Confidential 9
What do we mean by “high-resolution”?
• Epitaxial films and structures have a high degree of crystalline perfection
• The features (peaks and interference fringes) in the diffracted X-ray intensity from epilayer-substrate material systems are very closely spaced
• angular range of a few degrees at most
• High-resolution is needed to resolve these features
∆𝑑
𝑑=∆λ
λ+
∆𝜃
tan 𝜃𝐵
• High-resolution usually means highly collimated and monochromatic X-ray beams and precise goniometry
03/05/2017 Bruker Confidential 10
High resolution XRD setups
• Most common setup uses a parallel beam and point (0D) detector
• Source and detector angles scanned using a motorized goniometer
• Large (mm) and small (50 um) spot configurations are available
• JV also developed an innovative small-spot, focused beam HRXRD approach for fast in-line measurements on planar pads
X-ray tube
Conditioningcrystal
Slits
Sample
Mirror
Slits
Detector
2θω
(a)
03/05/2017 Bruker Confidential 11
Common Bragg diffraction geometries and anatomy of a HRXRD curve
• Symmetric Bragg geometry is sensitive to lattice parameter perpendicular to the surface
• Can determine SiGe composition from ∆𝜔 assuming fully strained layer
• Asymmetric geometries are also sensitive to the lattice parameters both parallel and perpendicular to the surface
• Can use symmetric / asymmetric combination for unique composition / relaxation determination
Symmetric geometry
Asymmetric geometries
Glancingincidence
Glancingexit
(a) (b)
10
110
210
310
410
510
610
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0
Inte
nsi
ty(a
rb. u
nit
s)ω (deg)
Δω
Substrate peak
Layer peak
Fringes
03/05/2017 Bruker Confidential 12
Example: Fully strained SiGe epilayer
• Composition / strain determined from measured lattice misfit
• Misfit normal to surface from layer peak position
• ∆𝑑/𝑑 = −∆𝜔 cot θ𝐵
• Thickness from interference fringe period
• 𝑡 = λ/(2∆𝜔𝑓 cos θ𝐵)
• No dependence on uncertain materials parametersSymmetric 004 reflection from 22.5 nm
epitaxial film of Si1-xGex with x = 49% on a Si(001) substrate
MeasurementSimulation
03/05/2017 Bruker Confidential 13
More complex SiGe/Si examples
• More complex stacks give rise to more complex interference effects
• Measured data can be automatically fit to dynamical X-ray diffraction theory by refining the parameters of a structural model
• Obtain individual layer thickness and composition for fully strained structures
Periodic superlattice structure: [8.2nm Si0.9Ge0.1/ 22.4nm Si]×5 on a Si(001) substrate
NIST SRM2000 standard:23.7nm Si/ 48.2nm Si0.84Ge0.16 on Si(001)
03/05/2017 Bruker Confidential 14
Lattice deformation in epitaxial thin-films
• If the lattice mismatch to the substrate and/or thickness is small, then an epilayer can be strained so that the in-plane lattice parameter is equal to that of the substrate (fully strained)
• Tetragonal distortion of the unit cell
• For large mismatch (high composition) or thickness, it may become energetically favorable to relax
• Creation of dislocations and/or roughening
Misfit dislocations
03/05/2017 Bruker Confidential 15
Comparison of HRXRD data from strained and relaxed SiGe epilayers
• Degradation of device performance and yield loss
• Relaxed material has about 50% less strain than a pseudomorphic layer
• Relaxed material will contain dislocations at the interface and in the layer - increased leakage
• HRXRD provides a unique, automated solution for strain metrology and assessment of lattice defectivity
010
110
210
310
410
510
-3000 -2500 -2000 -1500 -1000 -500 0 500 1000
Inte
nsi
ty (
cps)
Omega (sec)
Fully strained layer
t = 492 Å, x = 19.6%
Interference fringes
010
110
210
310
410
510
-7000 -6000 -5000 -4000 -3000 -2000 -1000 0 1000
Inte
nsi
ty (
cps)
OMEGA-2THETA (arcsec)
Partially relaxed layer
t ~ 350 Å (XRR), x > 50%
No interference fringes
“Triangular” layer peak
No interference fringes“Triangular” layer peak
03/05/2017 Bruker Confidential 16
Pseudomorphic SiGe layer
• Fully automated alignment, measurement and analysis of relaxation was developed for monitoring SiGe epi for 65nm / 90nm process
• Needs 2 measurements, no information about defectivity
• State of the art analysis in 2000s
210
310
410
510
-2000 -1500 -1000 -500 0
Inte
nsity
(cps
)
OM EGA-2TH ETA (arcsec)
0
100
200
300
400
500
600
700
-10 0 10 20 30 40 50 60 70 80 90 100 110
Inten
sity
(cps
)
RE L A X A T IO N %
(004) scan relaxation scan
0
250
500
750
1000
1250
1500
1750
2000
0 25 50 75 100 125
Inte
nsity
(cps
)
(arcsec)
210
310
410
510
-1500 -1250 -1000 -750 -500 -250 0 250
Inte
nsity
(cps
)
OM EGA-2TH ETA (arcsec)
Relaxation = 100%
Composition = 20%
Relaxation = 0%
Composition = 12%
03/05/2017 Bruker Confidential 17
Fast reciprocal space mapping - FastRSMs
• Linear (1D) detector replaces analyzer crystal / slits and point (0D) detector and allows routine RSMs to be measured in the fab
• Simultaneously intensity acquisition over a large range of 2θ angles
• x10-100 faster than conventional approach (minutes not hours)
• Provides more information than available by single HRXRD curves
• Automated RSM analysis for epi. process development and control of thin-films and patterned nanostructures
X-ray tube
Conditioningcrystal
Slits
Sample
Mirror Linear detector
03/05/2017 Bruker Confidential 18
RSMs from fully strained epitaxial thin-films
• Layer peak position gives the lattice parameters
Τ∆𝑎 𝑎 = Τ∆𝐿 𝐿
Τ∆𝑎|| 𝑎 = Τ∆𝐻 𝐻
• Composition and relaxation can be obtained
• Tilt of bonded layers also available
• Peak in the asymmetric RSM is located at H = 1 indicating the layer is fully strained (𝑎𝑆𝑖𝐺𝑒,|| = 𝑎𝑆𝑖)
• Thickness fringes are visible in the L direction
𝑡 ∝ 1/∆𝐿𝑓𝑟𝑖𝑛𝑔𝑒𝑠
004 113ge
Si
SiGe
37.6 nm Si0.81Ge0.19 on Si(001)
03/05/2017 Bruker Confidential 19
RSMs from relaxed epitaxial thin-films
• SiGe peak is shifted away from H=2 in the asymmetric 224ge map indicating relaxation (𝑎𝑆𝑖𝐺𝑒,|| > 𝑎𝑆𝑖)
• Peak is broadened due to dislocations, 𝑤(𝐻)/𝐻 ∝ 1/√𝜌
03/05/2017
004 224ge Substratepeak
Layerpeak
r = 1
r = 0
0 < r < 1
H
LH
~35 nm Si0.33Ge0.67 (R = 70%) on Si(001)
Si
SiGe
Bruker Confidential 20
RSMs from strained thin-films on strain relaxed buffers (SRBs)
• Ge peak is shifted away from H < 2 in the asymmetric 224ge map indicating relaxation (𝑎𝐺𝑒,|| > 𝑎𝑆𝑖)
• GeSn peak is not shifted in H wrt to Ge peak (𝑎𝐺𝑒𝑆𝑛,|| = 𝑎𝑆𝑖)
• Composition and relaxation of each layer can be obtained
004 224geGe0.89Sn0.11 / Ge on Si(001)
Si
Ge
GeSn
03/05/2017 Bruker Confidential 21
Patterned epitaxial nanostructures
• In blanket epitaxy you have simple biaxial stress
• Blanket pads are less relevant and / or no longer available
• In epitaxial nanostructures you have
• Micro-loading effects in selective growth
• Stress-state is far more complex, i.e. elastic relaxation of the epi and distortion of the substrate lattice
03/05/2017 Bruker Confidential 22
RSM from Si fins made using spacer double patterning (SDP) lithography
• Fins act as a diffraction grating, 𝑃 ∝ 1/∆𝐻𝐺𝑇𝑅 = 42.2 ± 0.5 nm
• X-pattern is characteristic of trapezoidal features, 𝛼 = 9 ± 1°
• Evidence of significant pitch walking error from SDP lithography
• Strong half-order GTR peaks (corresponds to 2 x pitch), ∆𝑃 = 5 ± 0.5 nm
x1
x2
x3
Si fins SiGe fins
Thin-film Planar
(a) (b)
(c) (d)
2α
0 +1 +2-2 -1-1.5 -0.5 +0.5 +1.5
GR order+2.5-2.5
(a)
(b)
03/05/2017 Bruker Confidential 23
x1
x2
x3
Si fins SiGe fins
Thin-film Planar
(a) (b)
(c) (d)
x1
x2
x3
Si fins SiGe fins
Thin-film Planar
(a) (b)
(c) (d)
Si
SiGe
RSMs from epitaxial SiGe finsAsymmetric 113ge reflection
• SiGe in a uniaxial stress state, cf. biaxial stress state for thin-films
• Elastic relaxation perpendicular to the line direction
• Other theory remains the same as planar films
• Composition and thickness determined from fitting
• x = 25%, t = 39.4 nm
03/05/2017 Bruker Confidential 24
Defectivity from RSMs
• Sharp peaks are from high quality material
• Broad peak is due to defectivity
• Ratio of these can be used to monitor defectivity
03/05/2017 Bruker Confidential 25
Next Generation Epi: Nanowires
• Complex nanowire films (before etch) can be measured and analysed for
• Individual layer thickness and composition
• Pitch and pitch walking
03/05/2017 Bruker Confidential 26
Andreas Schultze et al
RSMs from III-Vs deposited in narrow (60 nm) trenches etched into thick SiO2 layer on Si
• Interest as a high mobility channel materials in sub-10 nm nodes
• GaAs and InGaAs peaks are very broad due to the high density of threading dislocations, 𝜌~1010cm-2 despite the ART structures
004 115Si
GaAs
InGaAs
03/05/2017 Bruker Confidential 27
The Future
• Key improvements are being made in X-ray source and detector technology for fab suitable systems
• Brightness and power improvements increase on an on-going basis
• Detector sizes and efficiencies increase for different energy lines
• Software developments
• Analysis of strain and composition for 3D structures
• Analysis of pitch walking from more complex patterning schemes
• Automated defect analysis of epi films in 3D structures
• Is there more shape information available?
03/05/2017 Bruker Confidential 29
X-ray based CD Metrology (XCD)
• HRXRD will give shape and size information but has major influences due to strain effects
• Small angle scattering of X-rays eliminates these issues giving the ability for shape profiling using X-rays
• Key requirements for production system are
• High brightness source for transmission geometry in small (50µm) area
• High efficiency detector for 1D data collection
• X-ray source compatible with small tool footprint and high availability
• Analysis using modelling rather than library approach
• Data being collected on 3D NAND and dense metal arrays for logic samples
03/05/2017 Bruker Confidential 30
Summary
• HRXRD has been used for many years for epi characterisation
• Started with simple AlGaAs / GaAs type films
• Developed through InP and into Si substrates
• Still demand for HRXRD in GaN LED and power
• Logic films have become more complex, with added complexity of 3D structures to measure
• HRXRD particularly suited as can measure films without being sensitive to the 3D nature
• Can add additional information about 2D and 3D nature using asymmetric RSMs
• Automated analysis and reporting is now possible
• Thanks to collaborators / colleagues
• Bruker: Matthew Wormington, Kevin Matney, John Wall and the rest of the apps teams
• Andreas Schultze (imec)
03/05/2017 Bruker Confidential 31