TFE 06 - ASICS FOR MEMS

1
TFE 06 - ASICS FOR MEMS

• CMOS MEMS - Present Future
• Integrated Smart Sensor Calibration

2
Outline

• MEMS Present Future
• 0) Introduction
• 1) Present Technology overview
• 1.1) Pre-CMOS approach
• 1.2) Intra-CMOS approach
• 1.3) Post-CMOS approach
• 1.4) Mass Sensitive Chemical Sensor
• 1.5) Force Sensor Array
• 2) Future
• 2.1) CMOS MEMS based products
• 2.2) CMOS NEMS
• 2.3) Biotronics
• 2.4) Siliconless CMOS MEMS

3
Outline

• Integrated Sensor Calibration
• 0) Introduction
• 1) 1-Dimentional calibration
• 1.1) Calibration principle
• 1.2) Mathematics
• 1.3) Implementation
• 2) 2-Dimentional Calibration
• 2.1) Calibration principle
• 2.2) Mathematics
• 2.3) Implementation
• Conclusion

4
TFE 06 - ASICS FOR MEMS

• CMOS MEMS - Present Future
• Integrated Smart Sensor Calibration

5
0) Introduction

• All the IC nowadays and since 15 years
• CMOS technology
• High process yield, high reliability
• Well established fabrications technologies
• ?
• MEMS processes only drive for a while by
  universities and defense labs
• Due to unstandardized processes and low volumes
  ordered

6
0) Introduction

• In the mid-1990s, development of
• Accelerometers for airbag deployment
• DMD (Digital Micro mirror Devices) for image
  projection
• Why?
• Integration of the MEMS
  structure into a basic CMOS
  process
• Not only lowers manufacturing costs but also
  allows tighter integration of MEMS with ICs
  

7
0) Introduction

• Result
• Reduced chip size
• Improvement device performance
• Furthermore, high volume applications (inertial
  and pressure sensors) and
  applications requiring sensor arrays (DMDs) need
  both CMOS-MEMS processes.
• Thus, CMOS-based MEMS is more and more used
  within CMOS processes to release devices.
• An increasing number of Microsystem can be formed
  within the regular CMOS process sequence
  Above all magnetic, optical and temperatures
  sensors.

8
1) Technology overview

• Process
• Based on depositing thin films of metal or
  crystalline
  material on a substrate
• Then , applying patterned masks by
  photolithographic imaging
• Finally, etching the films to the mask.
• Wet etching a liquid solution will dissolve the
  material
• Dry etching reactive ion etching (Released of
  beams and cantilever), sputter, and vapor phase
  etching
• In the other hand, several devices (new ones)
  must be produced with additional micromachining
  and thin film deposition steps
• 1) Pre-CMOS approach
• 2) Intermediate-CMOS Approach
• 3) Post-CMOS Approach

9
1.1)Pre-CMOS Approach

• Structures of the MEMS are formed
  before the regular CMOS process sequence.
• Avoid thermal budget constraints during the MEMS
  fabrication typically, structures are buried
  and sealed.
• And pre-processed wafer are used as starting
  materials for the subsequent CMOS process.

10
1.2) Intra-CMOS approach

• Additional fabrication steps are
  performed in-between the regular
  CMOS steps.
• More highly integrated process Better
  performances Why? Because the micro structures
  and electronics part are closer together.
• And because inserting before the back end
  interconnect metallization ensures process
  compatibility with polysilicon.
• Ex Pressure sensors from Infineon Technologies
  and Freescale.

11
1.3) Post-CMOS approach

• Two possibilities
• MEMS structures are built from the
  CMOS layers
  (pressure, inertial, flow, chemical
  sensors).
• Or built by additional layers deposited on top of
  the CMOS wafer (DMD by TI
  or Honeywells thermal imagers).
• These two approaches are interesting because they
  can be entirely outsourced and so processed at
  any CMOS foundry.
• Main Drawback Stringent thermal budget, limiting
  temperatures to about 400C

12
1.4) Mass sensitive Microsensor

• CMOS based sensor for detection
  of volatile organic
  compounds in air
• Based on cantilever beam vibrations and
  electro-thermal excitation
  detected by 4 Piezoresistors in Wheastone bridge
  and heating resistors.
• Cantilever coated with chemical sensitive layer
  upon absorption of analyte molecules, cantilever
  mass increases and its fundamental resonance
  frequency decrease.
• This is recorded by an on-chip amplifying
  feedback circuit
• Realized using a 0.8 µm CMOS technology of AMS

13
1.5) Force sensor array

• Sensor used for recording force images
  and force distance curves
  in Atomic Force Microscopy (AFM)
• Consisting in 10 cantilevers with integrated
  thermal actuators, piezoresistive sensors and
  driving and signal conditioning circuitry
• External controller applies heating power to the
  thermal actuators
• Maintain constant cantilever defections
• Piezoresistive output signal is amplified on-chip
  and applied to the external controller.
• Operation time per cantilever 100µs
• Images area 1.1 x 110 µm
• Vertical resolution 3 nm

14
2) Future

• 1) CMOS MEMS based products
• New products beyond the currently dominating
  pressure sensor and
  accelerometers
• Based on the trends
• Systematic process development of laboratory CMOS
  MEMS for industrial mass production
• Co-integration of digital interfaces and
  µcontrollers with microstructures
• less expensive and more powerful
• Development of packaging to protect the
  vulnerable CMOS chip from environmental impact

15
2) Future

• 2) CMOS NEMS as platform for Nano
• High sensitivity to differents effects
• They have to deliver more than a basic device
  function some
  challenges have to be faced
• Interconnect challenge Human is capable to
  handle mm sized objects and only few
  electrons generate signals
• The 300K question Problem of influence of
  the external conditions and thus find the right
  optimization
• Everything nano? Miniaturization is not a value
  by itself it has a cost, and we have to focus on
  the crucial nano part and do the rest with
  µtechnology
• But IC technology can bring the vast fabrication
  experience gained over the last decades, and it
  is to be noticed that gates thicknessesve
  reached the nano-level

16
2) Future

• 3) Micro Biotronics

TRONICS
BIO

• Miniaturized Hardware
• Micro electronics
• Mechatronics

• Biology
• Biochemistry
• Biomedical science

Biophysical and Biomedical microdevices MEMS for
surgical sewing, Neuro MEMS, Biochemical sensors,
Bio-mimetic devices
17
2) Future

• 3) Micro Biotronics
• Challenges are
• Requires operations in water abhorred by
  microelectronics
• Living cells have to be kept alive by a
  mirofluidic supply system
• Stability between biological and electronic
  materials
• Limited lifetime of enzymes and antibodies

18
2) Future

• 4) Siliconless CMOS MEMS
• Silicon might be expensive for MEMS requiring
  small parts of it.
• Maybe make a whole µsystem out of polymer
  (plastic, glass) would be cost-effective.
• Recent thin film transistors and organic circuits
  on flexible polymeric substrate have been
  demonstrate that was feasible.

19
TFE 06 - ASICS FOR MEMS

• CMOS MEMS - Present Future
• Integrated Smart Sensor Calibration

20
0) Introduction

• Why calibration?
• Sensor should provide correct transfert
  from the f signal to the
  electrical output signal
• Increase performance reliability and so boost
  the MEMS market
• BUT increase MEMS production costs
• It takes time and attention per individual sensor
• Need several reference measurements and
  correction
• Solution Include at the sensor a programmable
  calibration facility (implemented as a digitally
  circuit integrated)
• It does NOT eliminate the need to do
  calibration!!

21
1) 1-Dimentional calibration

• Calibration principle
• Normally collecting a set of measurements
  data
  and then compute (complicated one!)
  correction formula
• Here Each measurement is directly used to
  compute one programmable coefficient in
  a correction function
• And the next measurement makes use of this to
  compute another coefficient but without
  affecting the previous calibration
• By instance
• 1) Offset
• 2) Gain
• 3) Linearity
• 4)

22
1) 1-Dimentional calibration

• Mathematics
• Yref reference signal (a1 is a dimensionless
  number)

23
1) 1-Dimentional calibration
24
1) 1-Dimentional calibration

• Implementation
• Analog Signal Processor for Polynomial
  Sensor Calibration
• Flow Diagram

25
1) 1-Dimentional calibration

• Implementation
• Easy to implement on µcontroller (repetitive
  character)
• Or
• With Harware digital as well as
  analog
• Following example analog implementation with
  current signals
• They can easily be added, substracted or
  multiplied (Kirschoff)
• All the currents are represented by differential
  current and carried by common bias current

26
1) 1-Dimentional calibration

• Implementation (block diagram)

Caption Sensor Signal Calibration
points Addition Substraction
Correction coef
27
1) 1-Dimentional calibration
28
1) 1-Dimentional calibration
29
2) 2-Dimensional Polynomial

• Why 2 dimensions?
• By instance, a pressure sensor not only
  affected by pressure but also
  by temperature
• Calibration principle
• Same method as before with a dimension more
• Select the error you want to correct (1 offset,
  2 gain)
• Instead of correcting it once and pass through to
  the next error, make a set of corrections for a
  predefined number of temperatures values
• So that the calibration function is equal to the
  desired function in all temperature calibration
  points
• The results are better with that method (speed
  and error reduction)

30
2) 2-Dimensional Polynomial

• Mathematics

31
2) 2-Dimensional Polynomial
Transfer error surface before calibration
32
2) 2-Dimensional Polynomial
Transfer error surface after calibration
33
2) 2-Dimensional Polynomial

• To prevent escalation of the
  polynomial factors in
  the formulas
• Make first a correction of the cross sensitivity
  (TC
  in our case) before linearizing sensitivity of
  the input
  variable (pressure)
• But complicated in our case temperature is swept
  through the whole temperature range several times
• And pressure cycles require less time for
  measurement than TC
• So the solution is
• Keep the order of corrected output values
  computation
• But change the order of corrections coefficients
  computation by
• At T t1, applying pressure p1 to calculate a11
  and apply t2 to calculate a21
• Why can we make that? Because at the given T,
  all the calibrations functions are equals!!
• Its the definition of this method!

34
2) 2-Dimensional Polynomial

• Implementation
• Microcontroller-based pressure sensor system with
  digital
  Calibration

35
2) 2-Dimensional Polynomial

• Implementation
• Multiplexer makes possible to read out the T
  sensor and the
  pressure sensor with only one ?S AD converter
• Calibration program and coefficients are stored
  in the µcontroller memory
• Software is written in C language and is composed
  of 2 parts
• Calibration part Compute calibration
  coefficients measure output
• Measurement part Compute corrected values of
  pressure with equations
• Functioning
• User have to enter reference datas for the whole
  calibration process
• Once the number of cal. Points reference
  signals are in the µcontroller, the system is
  totally autonomous and compute his own
  calibration coefficients and then compute the
  real measure of pressure.

36
Conclusion

• MEMS in the future will be more and
  more used since they
  are needed in a huge range of
  applications and since the IC processes can be
  used to produce some at a lower cost and
  with a best reliability
• Calibration is a very important part but costly
  of the MEMS process
• If the cost of such a process can be decreased
  and calibration made easier for the use, it would
  improve significantly the MEMS market
• Weve seen a method which is going in that sense

37
Questions?