RF Signal Processing using MEMS: Challenges and Prospects

1
RF Signal Processing using MEMS Challenges and
Prospects

• Roger T. Howe
• Depts. of EECS and ME
• Berkeley Sensor Actuator Center
• University of California at Berkeley

2
Outline

• What is the killer app for MEMS/NEMS mechanical
  filters?
• Challenges in fabricating and integrating RF
  micro/nano resonators
• BSAC/BWRC Integrated Microwatt Transceiver
  project 2001-2005

3
Motivation for RF MEMS

• From S. Chou, MEMS RD at Intel, April 25, 2001

4
RF Transceiver Block Diagram

• Eliminating off-chip filters Intel, Discera
  (Michigan spin-off)

5
FBARs

• Agilent (followed by Lucent, TDK, and Samsung)
• are nearing production of these 1-5 GHz
  resonators

• Use FBARs fortrying out newradio
  architec-tures (BWRC)
• Advanced RF packaging ismandatory

R. Ruby, et al, (Agilent Technologies), ISSCC
2001.
6
What about MEMS Resonators?

• Features high Q resonance with low power, but
  with impedance matching, interconnect, and
  power handling issues
• Not a drop-in replacement of a ceramic or SAW
  filter in cell phones!

7
Outline

• What is the killer app for MEMS/NEMS mechanical
  filters?
• Challenges in fabricating and integrating RF
  micro/nano resonators
• BSAC/BWRC Integrated Microwatt Transceiver
  project 2001-2005

8
Progress in MEMS Resonators

• Structures approaching fo 1 GHz
• Harold Craighead (Cornell) Si at 350 MHz
• Clark Nguyen (Michigan) poly-Si at gt150 MHz
• Michael Roukes (Caltech) SiC at 635 MHz
• Q, noise, transduction, and dynamic range issues
• John Vig (ARL), Michael Roukes, Tom Kenny
  (Stanford)

9
Resonator Design Choices

• High stiffness or low stiffness?
• Bulk modes are stiff
• Bending or tornsional modes are soft
• Transduction (drive and sense)
• Electrostatic
• Piezoelectric (FBARs)
• Materials and processes

10
First-Run Bulk Longitudinal Resonator

• Stiff mode
• High frequency with micro rather than
  nano
• Poisson effect useful coupling on sides
• Deep sub-?m gaps
• Compare another stiff mode by Michigan at
  IEDM-00

• B. Bircumshaw, O. OReilly, and A. P. Pisano,
  BSAC

11
Bulk Longitudinal Resonator

• B. Bircumshaw, A. P. Pisano, O. OReilly

12
Nano Tuning Fork

• S. Bhave, L. Chang, T.-J. King, and R. T. Howe

13
Circuit Model for MEMS Resonator

• Electrostatic drive, capacitive sense

14
Motional Resistance Scaling

• Symmetrical plate drive and sense
• Shrinking the electrode gap g (4th power) or
• Raising the DC bias VP (2nd power)

15
Demanding Requirements for Resonators

• small structural dimensions (mms)
• even smaller gaps (lt 100 nm)
• low material damping, low anchor losses
• very tight control on structural dimensions and
  materials

16
Implications of Req Scaling

• Lowering Req ? deep submicron gaps (sidewall
  structures for lateral excitation, J. Clark, et
  al, IEDM 2000)
• DC electric fields (VP/g) ? push to field
  emission limit
• Bottom line Req in the 10 k? range appears
  feasible at 1 GHz

17
Further Application Issues

• Dynamic range
• High end nonlinearity in drive and sense
• Low end noise (see J. Vig., IEEE Trans. UFFC,
  Nov. 1999)
• Frequency reproducibility
• Fabrication variations
• Temperature variations

18
Extrinsic Circuit Elements

• Interconnect to and from IC adds resistances Rint
  and capacitances Cint
• Feedthrough capacitance Cf

19
Applications?

• Not a drop-in replacement of a ceramic or SAW
  filter!
• Features high Q resonance with ultra-low
  power, but with impedance matching and
  interconnect issues
• Candidate application
• LO-less sensor node transceivers

20
Outline

• What is the killer app for MEMS/NEMS mechanical
  filters?
• Challenges in fabricating and integrating RF
  micro/nano resonators
• BSAC/BWRC Integrated Microwatt Transceiver
  project 2001-2005

21
Goals

• Reduce power by 100 x over state-of-the art
  sensor node transceivers using CMOS off-chip
  components
• Can autonomy be achieved?
• Less than 100 ??W average power use ambient
  energy scavenging
• Applications CITRIS project

22
Analog OFDM Subsamping Transceiver using RF NM
Filters

• B. Otis and Prof. J. M. Rabaey, EECS Dept. and
  BWRC

23
NM Filter Specs

• Motional resistance 25 k?
• 10 resonators per channel ? 2.5 k?
• Linearity, dynamic range challenging for
  electrostatic transduction
• NM resonator technologies

24
Transmitter Architecture

• B. Otis and Prof. J. M. Rabaey, EECS Dept. and
  BWRC

25
Options for Integration

• Integrate after deep submicron CMOS
• Parasitic elements degrade performance if MEMS
  resonators are on a separate chip or even
  fabricated adjacent to CMOS
• Parallel assembly processes to integrate
  MEMS/NEMS resonators into microsystem

26
Polysilicon MEMS-CMOS Integration

• 1.02 MHz tuning fork with Pierce amplifier
• 3000 Å-thick polysilicon interconnects (RC
  low-pass filters)
• T. A. Roessig, et al, Hilton Head 1998. (BSAC
  design in Sandia IMEMS)

200 ?m
27
Polycrystalline Silicon Germanium

• Semiconductor alloy
• Compatible with CMOS
• Conventional LPCVD furnace
• SiH4 or Si2H6, GeH4, B2H6
• Leverage IC industry research
• Heterojunction BJTs (in production)
• CMOS gates

28
Poly-SiGe MEMS after CMOS

• UC Berkeley baseline CMOS with Al-2 Si
• Post-CMOS temperature lt 450ºC
• 90oC H2O2 release maskless!

29
Stacked Resonator on Amplifier
25 ?m

• Andrea Franke, et al, HH 2000.

30
Schematic Cross-Section
31
Poly-SiC Liftoff Process

• Exploits poornucleation on SiO2

32
Poly-SiC Test Structures
10 mm
patterned Si substrate
33
Parallel Microassembly Processes
K. Böhringer, et al, ICRA, Leuven, Belgium, May
1998
34
NM Resonator Metrology

• Imaging a NM resonator is a critical capability
  (for all NMASP projects)

Scanning Acoustic Tunneling Microscope
Prof. Jeff Bokor
35
Taxonomy of Microassembly

• Parallel microassembly
• Multiple parts assembled simultaneously
• Deterministic pre-determined destination for
  parts
• Stochastic random process determines part
  destinations
• Serial microassembly
• Pick and place on a microscale

36
Stochastic Microassembly

• Pattern complementary hydrophobic shapes onto
  parts and substrates using SAMs.
• no shape constraints on parts
• no bulk micromachining of substrate
• submicron, orientational alignment

• U. Srinivasan, Ph.D. ChemEng, May 2001

37
Mirrors onto Microactuators

• Self-assemble mirrors onto microactuator arrays
• Si (100) mirrors
• Nickel-polySi bimorph actuators

38
Mirrors on Microactuators
39
Commercial Stochastic Self-Assembly

• Alien Technology
• (Prof. J. Stephen Smith, UC Berkeley EECS
  Dept.)
• Gravitational energy well
• 11,000 elements/min
• 99.99 yield
• 1 µm alignment

Alien Technology
40
Transceiver Integration
Micropart orientation by complementary binding
sites
RF passive
NEMS filter bank
Dense vertical feedthroughs
J. S. Smith and R. T. Howe
CMOS transceiver
41
Conclusions

• Micro/nano resonators
• Ultra-low power is one application space
• Fabrication technology challenges
• Many, but no show-stoppers
• FBARs are coming (or here )
• stepping stone for system designers

42
Project Personnel

• Faculty Investigators
• Roger Howe and Jan Rabaey, co-PIs
• Jeff Bokor (metrology)
• Tsu-Jae King (poly-SiGe)
• Roya Maboudian (poly-SiC)
• Al Pisano (NM resonator design)
• Steve Smith (integration by assembly)

43
Acknowledgements

• DARPA MEMS Program
• Modular SiGe-RF MEMS Project
• Profs. T.-J. King, A. P. Pisano,R. Maboudian, J.
  M. Rabaey, O. OReilly and J. S. Smith, UC
  Berkeley
• Graduate students Brian Bircumshaw (ME) Brian
  Otis (EECS) and Sunil Bhave (EECS)