Novel Microelectromechanical Systems (MEMS) for the Study of Thin Film Properties and Measurement of Temperatures During Thermal Processing Haruna Tada M.S. Thesis Defense July 21, 1999 Committee Members: Peter Wong and Ioannis Miaoulis, Tufts University Paul Zavracky, Northeastern Univ. / MicroOptical Corp.
56
Embed
Novel Microelectromechanical Systems (MEMS) for the Study of Thin Film Properties and Measurement of Temperatures During Thermal Processing Haruna Tada.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Novel Microelectromechanical Systems (MEMS) for the Study of Thin
Film Properties and Measurement of Temperatures During Thermal
Processing
Haruna Tada
M.S. Thesis Defense
July 21, 1999
Committee Members:
Peter Wong and Ioannis Miaoulis, Tufts University
Paul Zavracky, Northeastern Univ. / MicroOptical Corp.
Tufts University
Overview Introduction
• background & motivation• what are T-MEMS?
Thin film properties• experimental setup• numerical model• results
Heat transfer model• T-MEMS radiative properties• steady state temperature distribution
Evaluation• temperature range & resolution• proposed modifications• effects of high temperature & adhesion
Conclusions
Tufts University
Rapid Thermal Processing (RTP) RTP in Microelectronics Industry
• single wafer processing with radiant heat source• high temperatures (up to ~1000 °C)• high heating rates (100 °C/sec)• short processing times (~seconds)
Thermal requirement forecast for the year 2000• uniformity (± 2 °C) over 12" wafer• accuracy (± 3 °C)
Challenge• accurate temperature measurement techniques are
needed to meet the requirements
Tufts University
Temperature Measurements in RTP
Thermocouples• highly intrusive• delicate & difficult to handle• contact resistance between
changes with temperature and film deposition Alternative methods needed to meet thermal requirements
of the microelectronics industry
Thermocouple wafer(Sensarray)
Tufts University
MEMS Temperature Sensors Microelectromechanical Temperature Sensors (T-MEMS)
• small temperature sensors based on MEMS technology• ex-situ measurement of maximum process temperature• based on differences in thermal expansion coefficients
SEM micrograph of T-MEMSTop view by optical microscope
Tufts University
Design & Modeling Behavior of T-MEMS depend on thin film properties
• Young's modulus, E(T)• thermal expansion coefficient, (T)• functions of temperature
Previous study of thin film properties• Young's modulus of thin films
– resonance structures– tensile testing of micromachined specimen– mostly done at room temperature
• lack in information on thermal expansion coefficient at elevated temperatures
Tufts University
Approach New technique for determining thin film properties of
poly-Si and SiO2
• use T-MEMS as test structures to find (T) Evaluate T-MEMS design
• effect on wafer temperature– numerical models for radiative property and
temperature distribution• performance
– temperature range & resolution Refine T-MEMS design
• model beam curvature based on properties found
Tufts University
Study of Thin Film Properties
T-MEMS design Experimental setup Numerical model
Results
Tufts University
T-MEMS Design Bending T-MEMS
• array of multilayered cantilevers over Si substrate6 m gap by design, ~23 m in actual sample
• deflect down at high temperature due to difference in thermal expansion coefficients of layers
• adhere to substrate at contact
0.19 m SiO2
0.54 m poly-Si1.03 m SiO2
LPCVD SiO2
LPCVD poly-Si
thermal SiO2
Si substrate
Tufts University
T-MEMS Design
3” wafer
beams are initially curved up due to residual stress
changing widths
dec
reas
ing
len
gth
s
. . .1 2 3 . . . 14
100 m99 m
.
.
.
50 m
die size ~ 4 mm × 4 mm
..
.
..
.
..
.
Tufts University
Microscale Curvature Measurement
lamp housing
CCD camera with telescopic lens
fiber opticbundle
collimating lens
cube beamsplitter
quartz plate withAl foil reflector
Al reflector
thermocoupleembeddedin Si wafer
W-halogenlamp T-MEMS sample
quartz rods (support)
Si wafer (support)
output to
computer
output to
computer
light source
Tufts University
Microscale Curvature Measurement
Imaging System• collimated light source
illuminating curved sample only flat portion of beam is seen by the camera
Curvature Measurement• analyze CCD image to find
"apparent length"• curvature found through
geometric relation between beam curvature and apparent length
substrate beam
image of beam on camera
apparent length
Tufts University
Numerical model developed by Townsend (1987) Discretize beam layers into small sub-layers
• assume no stress gradient within each sub-layer
Solve for curvature:• constrain interface• Force = 0• Moment = 0
Thermally Induced Curvature
SiO2
poly-Si
SiO2
t
z
ti
i
n
.
.
.
0
Tufts University
Thermal strain:Thermal strain:
Curvature Equation
-1 for j < iij = 0 for j = i
1 for j > i
Neutral plane:
(Townsend, 1987)
Curvature:
Tufts University
Reduction of Variables Curvature at temperture T* is dependent on 4 variables:
• ESi,ESiO2 at T*
• Si, SiO2 variation from initial temperature to T*
• E and appear as a product• need to know three before finding the final property
Reduction of variables• parametric study to find the effect of each variable• for T-MEMS, E(T) found to have little influence on K
use literature values as approximation, then find (T)• other film structures can be designed to isolate the
effects of E
Tufts University
Piecewise-Linear Approximation of (T)
Low temperature range (< 300 °C)
• SiO2 is constant in general, of silica glass materials do not vary significantly at temperatures below 300 °C
• Si increases linearly up to 300 °C
High temperature range (300 ~ 1000 °C)
• Si is proportional to specific heat of Si based on physicsprinciple, verified for bulk crystalline Si
• SiO2 increases linearly up to 1000 °C
S
i (10
-6 °
C-1)
2
3
4
5
0 300 600 900 1200
temperature (°C)
for bulk, crystalline Si
Tufts University
Strategy for Low Temperature Range
0 100 200 300
25°C
300°C
linear fit
Si
(°C-1)
1 2 3
4
5
temperature (°C)
Tufts University
Results: Curvature Measurements
-0.0015
-0.001
-0.0005
0
0.0005
0.001
0.0015
0 200 400 600 800 1000
-7.5
-5
-2.5
0
2.5
5
7.5
temperature (°C)
K (m
-1)
tip d
eflection
for 100
m b
eam (
m)
initial upward curvature due to residual stress
minimum curvature
limit of system
Tufts University
temperature (°C)
S
i (10
-6 °
C-1)
2
3
4
5
6
50 100 150 200 250 300
Results: Si(T) at Low Temperatures
Si(T) approximated to be linear up to 300 °C
Tufts University
01234567
0 200 400 600 800 1000
temperature (°C)
(1
0-6 °C
-1)
poly-Si film
bulk crystalline Si
SiO2 film
Results:(T) at High Temperatures
Si(T) assumed to be proportional to specific heat
SiO2(T) approximated as linear between 300 ~ 1000 °C
Tufts University
Results: Numerical Fit
temperature (°C)
K
(m
-1)
-0.003
-0.002
-0.001
0
0 200 400 600 800 1000
Tufts University
Heat Transfer Model
Thermal requirements Radiative properties of T-MEMS Steady-state heat transfer model Wafer temperature distributions
Tufts University
Thermal Requirements of T-MEMS
Requirement of a non-intrusive temperature sensor:must not affect the heating of wafer• temperature of the wafer is same w/ or w/o the sensor
Requirement of an accurate temperature sensor:temperature indicated by the sensor is the same as actual wafer temperature• local temperature distribution surrounding the sensor is
uniform
Radiative effects on T-MEMS structures may affect the temperature of the wafer numerical model was developed to predict the effects
Tufts University
Radiative Effects on a Wafer Properties of silicon wafer
• varies dramatically with temperature• partial transparency at low temperatures• wafer becomes opaque at temperatures above 700 °C
Thin films (< microns) • thin film interference effects at wafer surface
Thick films (> microns) • incoherent effects; analyzed by raytracing
Large 2-D surface patterns• averaging by area fill factors
average = Fii Fi= Ai / Atotalwhere area fill factor is:
(Abramson, 1998)
Tufts University
Experimental Verification Si wafer at high temperatures
• partial transparency• increase in absorption at high temperatures
Single SiO2 films at high temperatures
• thin film interference Simple patterns (stripes) at high temperatures
• average area method for 2-D patterns Multilayered film at room temperature
• thin film interference for multilayered film• verify thickness measurement of T-MEMS films
Tufts University
Five T-MEMS Regions
T-MEMS Radiative Properties
Find net property of T-MEMS die by averaging
Si partialtransparency
thin filminterference
incoherenteffects
Si substrate 3-films 1-film
3-films & air 1-film & air
2
1
3
4
5
Tufts University
Total Radiative Properties of T-MEMS
00.1
0.20.30.4
0.50.6
0.70.8
0 200 400 600 800 1000
00.10.20.30.4
0.50.60.70.8
0 200 400 600 800 1000temperature (°C) temperature (°C)
Si substrate 3-films 3-films & air
T-MEMS average 1-film 1-film & air
tota
l no
rmal
ab
sorp
tivi
ty
tota
l no
rmal
em
issi
vity
Tufts University
Simulates a patterned wafer heated radiatively Heat transfer terms:
• conduction through wafer• radiation from lamp• radiative heat loss from wafer
• steady state: q = 0 Parameters:
• heat source: 2200 °C, = 0.3
• flampwafer = 0.1; constant
• use and of wafer at 800 °C
• kwafer = 30 W/mK
• 1/4 of wafer modeled due to symmetry • no convective term: assumes vacuum
Evaluation of original design Proposed design modification
Effect of high temperature Comment on adhesion
Tufts University
Performance of Original Design Original Design
• beam length: 50 ~ 100 m• width ratios: 0.2 ~ 0.85• 6 m between Si and beam• total of 714 beams on a die
Theoretical temperature range• 460 to over 2000 °C• thermal processing rarely exceeds 1100 °C
large portion of beams will not be used Theoretical resolution
• varies between 0.1 °C and 9.7 °C in 900 - 1100 °C temperature range
Tufts University
Modified Design Compile a "Wish List"
• temperature range: 900 ~ 1100 °C• resolution: < 0.5 °C• die size: as small as possible
Beam selection• 50 ~ 100 m in length• 0.2 - 1.0 width ratios• 6 m gap• total of 867 beams tested• selected 97 beams having contact
temperature between 900 °C ~ 1100 °C
Tufts University
Evaluation of Modified Design Modified design
• 0.2 ~ 1.0 width ratios• 62 ~ 73 m in length• 6 m gap depth• 97 beams, fits on ~1.3 mm square area
Resolution• vary between 0.1 °C to 9 °C
need to fill in "gaps" in temperature
900 950 1000 1050 1100
Tufts University
Improving Resolution Customized beam designs with specific target temperature are
needed to fill in gaps in resolution
Proposed design: varying bottom layer length• adjusting the bottom layer length will give full control of
contact temperature• can be modeled by simple geometry
Tufts University
Effects of High Temperature Effect of long-time exposure to high temperatures (~850°C)
• room-temperature tip deflection decrease with time Possible reason: thermal oxide growth on top layer T-MEMS may be annealed to have zero initial curvature
0
0.0005
0.001
0.0015
0.002
0 15 30 45 60
-12
-10
-8
-6
-4
-2
0
K (m
-1)
tip d
eflection
(m
)
total time (min)
Tufts University
Adhesion
Adhesion between bottom layer (SiO2) and substrate(Si) is a necessity for T-MEMS
Preliminary testing with loose beams on Si wafer• beams on plain Si wafer, heated to ~ 600 °C• test adhesion strength
– lightly rubbed by cotton swab after cooling• adhesion was confirmed under microscope
– adhesion stregth at room temperature is stronger than fracture strength of beams
Tufts University
Conclusions Thin Film Properties
• T-MEMS used as testing structures for finding properties• developed experimental apparatus for measuring
microscale curvature at very high temperatures
• thermal expansion coefficient of poly-Si and SiO2 found for high temperatures
T-MEMS as Temperature Sensors• theoretical evaluation of original design• design modification to target specific temperature ranges• thermally non-intrusive when used on Si wafer • beam adhesion confirmed in preliminary study
Tufts University
Future Work: Thin Film Properties
Modify beam design to target other properties
Extend study to other materials
• SiNx (silicon nitride) on SiO2 beams
Modify experimental setup
• view larger curvatures
• reduce uncertainty
Verify results with alternative methods
• resonance method for E(T)
• wafer curvature measurement for the product E
SEM micrograph ofSiNx-on-SiO2 beams
Tufts University
Future Work: Temperature Sensors
Finalize design modifications
• define target temperature range
• temperature resolution
• optimize die size
Fabrication, testing & calibration of modified design
• experimental testing with thermocouples
Verify adhesion using 6-m gap
Model temperature gradient during transient state
Tufts University
Acknowledgements Committee Members:
• Professors Peter Wong & Ioannis Miaoulis, Tufts Univ.
• Professor Paul Zavracky, Northeastern Univ. / MicroOptical Corp.