Application Notes : kSA MOS – Measurement of Stress Evolution
Version: 1.0
- Introduction
- Stress in Thin Films
- Measure thin film stress
- Multibeam approach (MOS)
- Interpreting measurements
- MOS Implementation
- Stress Evolution
- Coalescence
- Ag on SiO2
- High mobility material (Sn)
- Interrupt & Regrowth
- Stress measurements
- Sputter Deposition
- Whisker Growth
- Summary
- Acknowledgments
Measurement of stress evolution in thin films using
real-time in situ wafer curvature (k-Space MOS)
Eric Chason
Brown University, School of Engineering
- Intro to k-Space MOS (multi-beam optical sensor)
- theory
- capabilities
- analysis
- Examples
- polycrystalline films
- steady-state stress
- Scaling with D/RL
- stress vs thickness
- steady-state stress
- sputtering
- tin whisker formation
- battery materials
- polycrystalline films
Stress in thin films is a generic problem
Leads to decreased performance, deformation, failure
Want to:
understand stress,
control stress,
predict stress
→
First need to
measure it
Measure thin film stress via wafer curvature
Curvature measures product
of average stress x thickness
Multibeam approach (MOS): easy to implement/robust
- System requirements:
- Ports to measure specular reflection
- Reflective surface (backside ok)
- Measurement technique
- Etalon produces array of parallel beams
- camera measures change in beam spacing (δd/d):
Multi-beam technique reduces sensitivity to vibration
Interpreting curvature measurements
How does curvature relate to evolving stress distribution?
Stress distribution:
Study change of curvature with time
(1) Stress in new layers
at the surface, σ(hf ,t)
- Incremental stress proportional to slope of κ vs h
- But only if stress not changing in rest of film
Stress distribution:
Study change of curvature with time
(2) Change in stress of
existing layers
MOS Implementation
MOS can be implemented on many platforms
Deposition techniques:
- CVD
- sputtering
- PVD
- MBE
- PLD
- electrodeposition
MOS on GaN rotating
disk CVD reactor
(Hearne et al)
Materials systems:
- heteroepitaxy
- SiGe/Si, InGaAs/GaAs
- optoelectronics
- GaN, AlGaN, GaSb
- polycrystalline metals
- hard coatings
- DLC, a-C
- oxides
- TiO2 , CeO2
Examples from stress evolution studies
1. Residual stress in polycrystalline films
- Electrodeposition/evaporation
- Dependence on growth conditions, material
- Evolution with film thickness
2. Sputter deposition
- Effect of processing parameters (surface roughness)
3. Mechanical properties of Sn films
- stress leads to whiskers
- enhance stress relaxation
4. Strain in battery materials
- large volume changes
- associated with phase changes
Features of stress evolution in polycrystalline films
Stages of film/stress evolution:
- Nucleation
- Compressive or no stress
- Coalescence
- Tensile rise
- Continuous film
- Steady-state compressive (for high atomic mobility)
Stress depends on kinetics (temperature, material, deposition rate)
- Ag on SiO2
(Chason, Hearne, JAP 2013) - Fe on MgF2
(Thurner and Abermann, TSF, 1990) - Lower T, same growth rate:
- → more tensile
- At 30°C:
- Fe: tensile,
Ag: compressive
- Fe: tensile,
- Tensile → grain boundary formation
- Compressive → insert atoms into grain boundary (driven by surface supersaturation)
- Mediated by kinetic processes on surface:
- Growth rate R, diffusivity D, grain size L
Write equations for evolution of stress
Master equation for stress evolution at triple junction:
hgb = Rate of growth of grain boundary
Steady state stress: dependence on growth rate
Electrodeposited Ni on Au, Hearne et al, JAP 97 (2005)
- Stress reaches steady-state (constant slope)
- Different σSS for each growth rate
- Model prediction:
→Determines growth rate for
stress-free films
Stress vs thickness: effect of coalescence of islands
Grain boundary growth rate changes as film grows
Stress changes with grain boundary velocity
Model fits Ag on SiO2 data
Change atomic mobility (D) at constant R, L
- Islands are cylindrical caps, contact angle ~68 deg,
- Fitting parameters: σc, σT , τ
- Use same σT (442 MPa) and σc (-359 MPa) for all temperatures
- τ different for each T (proportional to 1/D)
Role of grain boundary in high mobility material (Sn)
Monitor stress during electrochemical deposition
- Evaporate seed layer of Sn (1 µm)
- Electrodeposit Sn film at constant voltage
Stress behavior during interrupt & regrowth
Stress measurements in electrodeposited Sn
Equivalence between growth and etching:
- negative chemical potential on surface induces tensile stress in film
- confirms role of surface chemical potential in stress evolution
Stress evolution during sputter deposition
Additional parameters: ion energy, gas pressure
Stress evolution during sputter deposition (LLNL)
Be targets for NIF: need films with low stress (thick > 100 µm)
Higher growth rate (power) → more tensile
Higher T → more compressive
Be sputtering results
Lower pressure → more compressive initially
BUT: Incremental stress changes from compressive to tensile
as layer gets thicker → kept same temperature, growth rate
Reason: Stress change correlated with rougher surface morphology
Greater roughness → Turns off compressive stress generation
Film becomes tensile
Sn whisker growth: driven by stress from IMC (intermetallic) formation
Whiskers form in Pb-free Sn coatings on Cu – cause systems failure (satellites, pacemakers)
IMC forms at Cu-Sn interface
Measure stress evolution with MOS
Wafer curvature measures total force exerted by film.
Remove Sn layer – change in curvature gives stress in Sn
Reduce whiskering by enhancing stress relaxation
- Measure mechanical properties of layers:
- Sn and Sn alloys
- Find coatings that have low stress even after IMC grows
- These results agree with conclusions from whisker studies
- More relaxation with
- larger grain size
- thickness
- horizontal grain boundaries
Stress evolution during charging/discharging of batteries
(lithiation of Sn anode)
Measure stress associated with phase changes
Need to know layer thicknesses to interpret MOS data
(Chen, Guduru 2013)
Summary
- Multi-beam wafer curvature (MOS) enables stress evolution to be monitored in real-time
- useable on wide variety of platforms
- sensitive, robust, easy to interpret
- Stress dynamics provide more information than single stress measurement
- Key for
- modeling
- understanding sources of stress
- controlling stress (optimizing processing conditions
- Key for
- Frontiers
- Understanding multi-component materials
- Energetic particle effects
Take home point:
In situ monitoring useful for understanding stress evolution
Acknowledgments
Brown:
Jae Wook Shin (LAM)
Chun-Hao (Charly) Chen
Alison Engwal
Julie Zaskorski (REU)
Brittni Thomas (REU)
Nitin Jadhav
Tanmay Bandakhar
Allan Bower
Ben Freund
Brian Sheldon
Huajian Gao
Pradeep Guduru
K.S. Kim
Sandia:
Jerry Floro (UVa)
Sean Hearne (Sandia)
LLNL:
George Gilmer
Luis Zepeda-Ruiz
Alex Hamza
Chris Walton
Morris Wang
NIST:
Gery Stafford
k-Space:
Darryl Barlett
Chuck Taylor
Eric Friedmann
Roy Clarke