Frequency Agile Microwave Oven
Bonding System (FAMOBS)
From feasibility study to production prototype
Prof. Marc DesmulliezHeriot-Watt University
Edinburgh, UK
1/28
• Microwave heating – principles and applicators
• Funding schemes and partners
• Project overview
• Open end oven prototypes
• Results - Curing
• Conclusions and future work
Contents
2/28
FAMOBS funding schemes
FAMOBS (EU)
• Project funded by the European Commission within FP7
• 3 year project starting in November 2008
• 5 industrial collaborators, 4 RTDperformers and 4 SME associations
FAMOBS (UK)
• 18 month research project (feasibility and project deepening)
• Funded by the IeMRC• 3 research partners + 4 industrial
partners
3/28
Microwave heating
• Generates heat at the molecular level by forced oscillation ofpolar bonds
• Materials ability to absorb microwaves depends on:
• Complex permittivity, conductivity, frequency & temperature
• Heating is volumetric as compared to thermal transfer fromconvection heating
Convection – heat penetrates from surface Microwave – uniform heat generation
4/28
http://www.2450mhz.com/PDF/Doc/900065.pdf
Microwave Applicators Semiconductor manufacturing
GERLING APPLIED ENGINEERING, INC.
Lambda TechnologiesTM
(http://www.microcure.com/) 5/28
Project overview - FAMOBS
• To develop an open-ended ‘microwave oven’ for in-situ curing and bonding within microelectronics processing applications.
• Research has been on several fronts and consists of
• Design of open ended oven
• Pulsing and VFM implementation
• RF modelling
• Multiphysics modelling
• System integration
• Characterisation of cured specimens
6/28
Open ended oven - A gain in productivity
Traditional Assembly Process
FAMOBS Manufacturing Process
(Source: Prof. Chris Bailey, University of Greenwich)
• In addition, microwave processing is more efficient than a convection oven(increase in heating ‘rate’ by a factor of 10 depending on the material)
• For the rapid curing of both conductive and non conductive based polymerdielectrics in semiconductor manufacturing and packaging
7/28
Challenges – Multiphysics modelling
8/28
Open ended oven integration
Microwave oven on die placement arm
Die, Adhesives, Encapsulants, Underfills, etc
FAMOBS oven prototype system
9/28
FAMOBS System Stakeholder Map
FAMOBS
System
Research
Foundry
Test Q&R
Services
FablessIDM
PCB
Packaging
GE Aviation
10/28
Other applications
• Petroleum compounds extraction
• Ceramic composites sintering and annealing
• Sintering of carbon composites
• Bonding applications for micro-fluidic, microelectronics andMEMS
11/28
Manufacturing capability
readiness levels
12/28
9
10
7
6
5
4
3
2
1
8
Fully production capable process qualified on full range of parts over extended period
Fully production capable process qualified on full range of parts over significant run lengths
Capability and rate confirmed via economic run lengths on production
Process optimised for capability and rate using production equipment
Basic capability demonstrated using production equipment
Process validated in laboratory using representative equipment
Experimental proof of concept completed.
Applicability and validity of concept described and examined or demonstrated
Process concept proposed with scientific foundation
Process concept proposed. Process unreported in literature, potential for generating IP.
Manufacturing technology proven and assessment in laboratory environmental
Pre-Production in relevant environmental
Production Implementation
Phase 3
Phase 2
Phase 1
Ph
ase
of
Dev
elo
pm
ent
MCRL State of Development
Continuous improvement to fully capable production processProduction Improvement
Phase 4
(source : Nabil Gindy, University of Nottingham)
MCRL1
MCRL2
MCRL3
MCRL4
MCRL5
MCRL6
MCRL7
MCRL8
MCRL9
• The electrical length of the cavity can be controlledthrough adjustment of additional conducting rods.
• Disadvantages:• Mechanical design for control of length of conducting
rods• Substantial heat generated within dielectric
• Waveguide Cavity Resonator• Open end – access to cavity volume • 2 waveguide sections
• Dielectric filled propagating section• ‘Air’ filled cut-off section• fc
(dielectric) < fc(air)
• Evanescent field used for heating
1st generation prototype – PT1
PT1
PT2
PT6
PT5
PT4
PT3
13/28
MCRL1
MCRL3
MCRL5
MCRL6
MCRL7
MCRL8
McRL9
• Excessive loss tangent (tanδ) of dielectric insert
• Substantial heat generated within dielectric
- curing occurred through thermal conduction
• Decrease in cavity Q Factor
- quasi degeneracy disturbs modal field pattern within cavity
Single Feed, Multiple Heating Spots @ ≈ 10.19GHz
• The fringing field depth of the evanescent field can becontrolled through proper choice of bulk dielectricpermittivity
• Disadvantages:
2nd generation prototype – PT2
MCRL4
MCRL2
PT1
PT2
PT6
PT5
PT4
PT3
14/28
MCRL1
MCRL3
McRL5
McRL6
McRL7
McRL8
McRL9
• Higher dielectric constant, er = 30; Q = 10,000
• 11mm2 x 110mm ceramic dielectric
• Testing did not produce measurable results
• Cross-section was overly small compared to ‘X-Band’ connectors (SMA, SMB)
• Purchase of MMCX connectors suitable up to5GHz
• Very poor coupling from feed line into cavity.
• Drilling of connector hole (dia. 0.8mm) causeda failure within the ceramic.
• Results suggest lower permittivity (er < 10)cavity would be more suitable.
3rd generation prototype – PT3
MCRL2
MCRL4
PT1
PT2
PT6
PT5
PT4
PT3
15/28
MCRL1
MCRL3
MCRL4
MCRL5
MCRL6
MCRL7
MCRL8
MCRL9
• Optimised design with dielectric insert for improved heating rate
• Bulk Dielectric er = 2.1 Length = 90mm
• Dielectric Insert er = 6 Optimised length = 3.5mm
4th generation prototype – PT4
MCRL2
Temperatures recorded with PT4, PT2 and a convection oven
PT1
PT2
PT6
PT5
PT4
PT3
16/28
Temperatures recorded with the optimised, normal open end oven and a convection oven
4th generation prototype – PT4
MCRL1
MCRL3
MCRL4
MCRL5
MCRL6
MCRL7
MCRL8
MCRL9
MCRL2
Measurement temperature results for a 150oC curing cycle by pulsing the
source.
Electric field distribution comparing shift in maxima due to inclusion of an optimised dielectric insert: a) TM11, 1st mode b) TM11, 2nd mode
• Disadvantages:• Temperature measurement• Couldn’t be integrated with the placement machine
PT1
PT2
PT6
PT5
PT4
PT3
17/28
MCRL1
MCRL3
MCRL4
MCRL5
McRL6
MCRL7
MCRL8
MCRL9
Pyrometer
Open end /
load section
• New design for integration with the pick and placement machine
• Integrated IR pyrometer for temperature sensing
5th generation prototype – PT5
MCRL2
PT1
PT2
PT6
PT5
PT4
PT3
18/28
5th generation prototype – PT5
MCRL1
MCRL3
MCRL4
MCRL5
MCRL6
MCRL7
MCRL8
McRL9
MCRL2
• Results of the IR pyrometer measured temperaturecurves for a 150oC curing cycle by pulsing the source.
• The pulsing method has resulted in a better controlof programmed temperature with a temperatureaccquisition rate of 1 Hz.
• Disadvantages:• Dielectric losses in the bulk dielectrics and insert• Difficulty in the integration of pyrometer
PT1
PT2
PT6
PT5
PT4
PT3
19/28
Advantages
Easier integration of pyrometer
No losses due to dielectric filling and no need of replacing dielectrics
Could be easier for tuning (impedance matching) with extra tuners (screws)
No need of precise control of the robotic arm as material can be moved inside the cavity
PT1
PT2
PT6
PT5
PT4
PT3
MCRL1
MCRL2
MCRL3
MCRL4
MCRL5
MCRL6
MCRL7
MCRL8
MCRL9
• Different air cross-sections
• Designed and modelled
• Advantages• Easier integration of pyrometer• No losses due to dielectric
filling and no need of replacing dielectrics
• Could be easier for tuning (impedance matching) with extra tuners (screws)
• No need of precise control of the robotic arm as material can be moved inside the cavity
6th generation prototype – PT6
20/28
P1 P2 P5
PT1 PT2 PT3PT4
PT5 PT6
FAMOBS oven prototypes
Different air cross-sections – reduced dielectric losses and easier pyrometer integration
Higher dielectric constant dielectric Very poor coupling from feed line into cavity.
21/28
Curing investigation with PT2, PT4 and PT5 ovens
(a) (b)
Open chip Chip after
cavity-fill
Chip after
curing
(c)
• Outcome• An LM2940C-12 voltage regulator chip in a QFN package successfully tested for
functionality after curing• Observations suggested that ‘hardening’ of the encapsulant paste occurred with no
apparent damage to the chip package• EO1080 encapsulant samples cured at different temperature profiles to evaluate the
degree of cure
• High power continuous wave source not suitable for sensitive packages• SFM – Single Frequency Microwave (pulsed) with feedback control• VFM – Variable Frequency Microwave with feedback control
22/28
00:00:10.00 00:2:30.00
00:00:10.00
With out control loop, ~ 15 W full power
Optimised open ended oven with control loop feedback
1 Hz temperature acquisition rate
• Peak temperature recorded by the thermal imaging camera versus heating
time
Infrared temperature profiles
With PT2 With PT4
With PT5
23/28
Temperature (oC) Measurement Temperature sampling rate
(Hz)
Degree of cure Time
(s)
150 Fibre optic thermometer
(PT4)
12.5 and 5.5 98% 270
180 Fibre optic thermometer
(PT4)
12.5 and 5.5 81% 180
150 Pyrometer(PT5) 1 100%
Tg is 115 oC
270
Two step: 115 and 150 Pyrometer(PT5) 1 Closer to 100%
Tg is 100 oC
270
Comparison of different curing cycles
Curing investigation - Results
• Measurement methods and the cure cycles are compared by degree of cure.• Complete curing is achieved with lower pulse rate but without any reduction
of overall cure time of 270 seconds.
24/28
System integration - solid state components
• Possibility for compact power delivery modules for integration with the pick and placement machine
Control
(Lab view)
RF source, Pulsed Power
amplifier, Isolator
(Dimensions 11.25×20×1
cm)
FAMOBS oven integrated
with a pyrometer
25/28
Conclusions
• A multidisciplinary project requiring different classes of experts
• Design for system integration
• Research not necessarily follow a linear pattern
• A new generation oven with a pyrometer attached has been designed, fabricated and RF tested
• Successful curing has been achieved with the new integrated oven
• Pulsing technique used tends to control the temperature of the curing material quite accurately
• Hotspots and thermal runaway problems avoided with controlled curing
26/28
Future work
• Multi-physics model needed to ling data to the degree of cure
• Further evaluation of cure should be under taken with commercial packages
• System design• Further tests in integration of device into precision
placement machine• Efficient coupling of power from source to cavity
27/28
Contact Details
Heriot-Watt University
Riccarton,
Edinburgh,
EH14 4AS, UK.
http://www.hw.ac.uk
Prof. Marc Desmulliez,
Tel: +44 (0)131 451 3340
Fax: +44 (0)131 451 4155
Email: [email protected]
Contact
28/28
Website: www.famobs.eu