Title: Tutorials on Systems Miniaturization Luiz Ot
1Tutorials on Systems Miniaturization Luiz Otávio
S. Ferreira - LNLSNovember 28, 2001
2Outline
- Introduction to Systems Miniaturization
- Microfabrication Technologies
- Microsystems Development and Packaging
- Microfabrication in Brazil
3Introduction to Systems Miniaturization
- Microsystems
- Sets of microdevices capable of integrated
sensing, analysis and actuation. - Microdevices
- Microstructures capable of actuation, or signal
transduction, or chemical reaction, etc.
Ivo M. Raimundo Jr. IQ/UNICAMPMUSA2000
Nobuo OkiUNESP Ilha SolteiraMUSA2000
Luiz O.S. Ferreira LNLS
4Why Miniaturization?
- Reduction on mass and size.
- Integration with electronics.
- Exploitation of new effects due to small size.
- Cost/performance advantages.
- Improved reproducibility, accuracy and
reliability. - Redundancy and arrays.
- Low power consumption.
- Less material used for manufacturing.
- Avoiding of rare or aggressive to environment
material. - Easy disposal.
5Vocabulary
- USA
- MEMS
- Microelectromechanical Systems
- Europe
- Micro SystemsMicro Systems Technology
- Asia
- Mechatronics
- Micromanufacturing
- Other names
- Micromechanics
- Nanotechnology
- Microtechnology
- Meso Systems
6Market Demands on Miniature Systems
- Environment
- Medicin Technology
- Information Technology
- Biotechnology
- Automotive
- Consumer electronics
- Projected sales for 200332 B (US)
Source Solid State Technology, July 1999, pp.
63-65.
7Technological Possibilites - 1
- Microtechnology for electronics
- Technologies developed or improved on last 20
years - Silicon crystal production.
- Thin film technology.
- Lithography and etching.
- Modeling.
- Characterization.
- Non electronic interations
- Springs, membranes, piezoresistive effect,
heaters, etc. - Well developed material and technology low cost
if large scale production. - Systems integration.
8Technological Possibilities - 2
- Full system approach
- Technologies for
- Assembly
- Interconnection
- Housing
- System integration
- Bonding and joining
- SMD, COB, TAB, DCA, Wire bonding, Flip Chip
- Analysis of the interactions
- Reliability
- Performance and cost
- Volume
9Technology Adaptation
- Old technologies, from micro-electronics and from
mechanics, are adapted for use on micro-systems
integration. - Some new steps must be developed.
- Old materials are used on new ways different
properties. - Only a whole system approach leads to effective
systems. - Numerical analysis of interdependencies.
10Why Integration?
- Better shielding of weak electric signals from
sensors. - Individual sensor calibration on factory. Lower
calibration cost. - On board intelligence.
- Reduction of connection cables.
- Standard communication protocols.
- Save cost on extra electronics housing.
11How to Integrate?
- Monolithic Integration
- Very difficult and expensive.
- Very large scale of production.
- Large number of interconnects.
- Number of masks.
- Time of development.
- Yield.
- MCM
- Hybrid Integration
- In 1997, 8 of the pressure sensors and 12 of
the accelerometers where monolithically
integrated.
12Product Oriented Approach
- Problem and product oriented approach YES!
- Technology oriented approach NO!
- Technology manufacturability.
- Important technologies are not silicon based
- Mechanical micromachining.
- High aspect ratio microstructuring (LIGA).
- Replication methods
- Electroplating,
- Injection molding.
- Hot embossing.
13Availability of Production
- Many prototypes of sensors.
- Small number on the market.
- Prototyping labs are not equipped to make 100,000
devices batchs. - Moving the prototype to a foundry implies on
starting again from the scratch. - Orders of less than 250,000 devices are not
attractive to silicon foundries. - Multi-User prototyping approach (The MUSA
Project).
14COSTS
- CMOS foundry for monolithically integrated
sensors US30 Millions. - Micromechanical parts line (if the ion
implantation is made externally) US4 Million. - Hybrid integration (assembly and thick film
line) US1 Million. - CMOS processed silicon US 2.5 to 8. Cent per
mm2 US750 to 2100 for a processed 20cm waver. - Sensor process US0.35 per mm2 for batch of more
than 50,000 chips. - Surface micromachining US1.80 per mm2 for
10,000 Chips batch, and 30 cents per mm2 for
500,000 chips batch. - Less than 1 Million chips per year is a risk.
- Bellow 10,000 chips a year a big problem.
15People Demand
- 1996 total 48,000
- USA Japan Europe
- 29,000 13,000 6,000
- 2002 projected total100,000