biro.ferenc@ttk.mta.hu
Optimisation of Low Dissipation Micro-hotplates:
Thermo-mechanical Design and Characterisation
Ferenc Bíró1,2
Csaba Dücső1, Zoltán Hajnal1, Andrea Edit Pap1, István Bársony1
1 Institute of Technical Physics and Matrials Science - MFA ,
2University of Pannonia, Hungary
7/2 Nanocatalytic gas sensors
Aims:
• Development of integrable micro-hotplate
• Low power consumption, (<30mW at >500°C)
• Thermo-mechanical stability over extended lifetime ( >5 years)
• CMOS compatible process
Optimisation of Low Dissipation Micro-hotplates Thermo-mechanical Design and Characterisation 10/2
Means:
• FEM modeling
• Selection of appropriate structural materials
• Functional testing and verification
Materials & Methods:
• n-type Si, <100> wafer
• Standard CMOS processes
• TiOx/Pt: DC magnetron sputtering
• Filament patterned by lift-off
• Cavity etching from back side by Deep Reactive Ion Etching (DRIE)
• Porous alumina (catalyst support) by electrochemical anodisation of Al layer on top of the hotplate
Layer structures Layer thicknesses are given in nm.
Type A B C D E
SiO2 Thermally
grown
100 100 100 1000 1000
SiO2
CVD 100 100 100 - -
Si3N4
LPCVD 200 200 - - -
TiOx/ Pt
15 / 270
15/
270 15/
270 15/
270 15/
270 Si3N4
LPCVD - - 200 200 150
SiO2
CVD 300 300 300 300 300
Porous alumina
1400 Plate Diam.
100 µm 1400 Bridge Structure of alternative
membranes investigated
7/4 Nanocatalytic gas sensors
Thermo-mechanical FEM Simulation
Optical view of two hotplates for the targeted pellistor.
Membrane diameter: 300 µm Chip size: 1×1 mm2. Optimisation of Low Dissipation Micro-hotplates Thermo-mechanical Design and Characterisation 10/4
• Structures analyzed by COMSOL Multiphysics®
• Considered physical effects & parameters
• Bottom surface of membrane:
Natural convective cooling
in horizontal back plan, parameterized by Cooling area/perimeter
• Top surface of membrane:
Natural convective cooling in horizontal plan, parameterized by Cooling area/perimeter.
• Filament:
Embedded in SiO2-Si3N4
Joule heating
Temperature dependent TCR
•Cavity:
Filled with air
• Side surfaces:
Kept at 25 °C
400×400 µm area were considered
• Multilayer:
Temperature dependent thermal and electrical properties
Heat transfer between layers and Si cube
Mechanical stress
Simplified geometry of the full- membrane sensor models.
biro.ferenc@ttk.mta.hu
Characterisation of Micro-heaters
Applied structures:
• 10×10mm2square membranes and wafers.
Average residual stress calculated from wafers curvature using Stoney equation.
• Wafer curvature measured by Makyoh topography.
1. Residual Mechanical Stress Measurement
Layer structure
Parameters of
Si wafer Parameters of layer structures Es
[GPa] υs [s.d.] ds
[µm] df [nm] R
[m] σf [MPa]
C 130 0,28 370 700 -30,6 -192
D 130 0,28 370 1200 25,6 133
E 130 0,28 370 1150 21,5 166
+
⋅ ⋅
−
⋅
= ⋅
pre post f s
s s
s d R R
d
E 1 1
) 1 ( 6
2
σ υ Results
• Residual stress below 200MPa
• Membrane:
Type C compressive, flat Type D, E, tensile, bumpy
• C selected for operational device SiO2Thermal
100nm
SiO2CVD 100nm
TiOx/Pt 15/270nm
Si3N4LPCVD 200nm
SiO2CVD 300nm
Porous-Al2O3
1400nm
7/6 Nanocatalytic gas sensors
Characterisation vs. Modelling of Micro-heaters
2. Thermal characterisation
▻ C.
Hot Resistance Temperature Method (HRTM) Filament temperature calculated by measured resistanceand TCR of Pt filament. 0
0
1 R T R T
H
HP +
−
=
α
Optimisation of Low Dissipation Micro-hotplates Thermo-mechanical Design and Characterisation 10/6
▻ B.
Micro Melting Point Measurement(MMPM)• Compounds of definite melting points selected to cover temperature range between 212 and 884°C
• Sharp solid-liquid phase transformation
• Small grains dropped on the hotplate by glycerol based suspension
• Phase transformation observed by optical microscopy
▻ A.
FEM simulation (current ramp)biro.ferenc@ttk.mta.hu membrane
0 5 10 15 20 25 30 35 40 45 50 55 0
200 400 600 800 1000 1200 1400
AgNO3 KNO3
Pb(NO3)2
CuCl2+2H2O V2O5 Na2SO4
Temperature /O C
Power / mW
Temperature measured by melting point of salts.
Average temperature of Pt, calculated by COMSOL.
Maximum temperature of Pt, calculated by COMSOL.
Average temperature of Pt, calculated from the resistivity.
• FEM calculated average temperature
Temperature measured by MMPM
• FEM model is correct for design
• MMPM is effective for calibration.
• HRTM method gives lower average temperature
• Huge differences between the average temperature and the hottest point of the filament as calculated by FEM⇨
to be verified by high resolution IR camera
Temperature vs. Power relationship determined by different methods and calculated by FEM
2. Thermal characterisation
coincidence Results
7/8 Nanocatalytic gas sensors
Charactrisation of Micro-heaters
Method:
• Hotplates powered by square wave current.
• Voltage-, current waveforms measured by DSO.
2 3 4 5 6 7
0 100 200 300 400 500 600
Temperature /O C
Time / ms Rise time of micropellistors:
A : t
90%=2.3ms B : t90%=2.2ms C : t90%=2.3ms
3. Dynamic properties: Response time
Optimisation of Low Dissipation Micro-hotplates Thermo-mechanical Design and Characterisation 10/8
Results:
• Quick response time (~2.3ms,t90), for all types investigated.
• No significant difference between various structures.
Rise time of microheaters, driven by square wave current. Input power is 23mW in all cases.
biro.ferenc@ttk.mta.hu
Charactrisation of Micro-heaters
Test conditions:
• Devices powered by square wave current pulses.
Input power amplitude is constant 30mW, D=20%, T=1s.
4. Dynamic properties: Accelerated Life Time Test
Results:
• No membrane crack was observed.
• Resistance changing in time.
• Failure: electromigration in the filament.
Optical micrograph of degraded Pt filament after 8.7×105 temperature cycles.
Resistance of micro-heaters of hotplate vs. time.
0 200000 400000 600000 800000 1000000 350
355 360 365 370 375 380
Device close to the end of its life.
Operation conditions:
Heating power = 30 mW Period time = 1 sec Duty cycle = 20 %
Resistance / Ohm
Time / sec Device1, type A Device2, type A 30mW
0 Pin
t D=20%
1 sec
7/10 Nanocatalytic gas sensors
Conclusions
• Thermo-mechanical properties of alternative membrane designs were modeled by COMSOL code and
• Tested experimentally in terms of temperature vs. power, response time and long-term stability. FEM results are verified.
• The selected full-membrane type structure reaches 700oC at 30mW.
23.3oC/mW !
• The full membrane is stable, can stand cycling between room temperature and 700oC for minimum 8×105periods.
• Quick response time with porous catalyst support layer.
• Improvement of filament characteristics is necessary
Acknowledgements
This work was partially supported by
Hungarian National Development Agency grant TÁMOP-4.2.2/B-10/1-2010-0025.
and
Tateyama Kagaku Industries Co. Ltd., Japan.
Optimisation of Low Dissipation Micro-hotplates Thermo-mechanical Design and Characterisation 10/10
biro.ferenc@ttk.mta.hu