MECH 466

Microelectromechanical Systems
University of Victoria
Dept. of Mechanical Engineering

Lecture 2:
Scaling Laws & Microfabrication

© N. Dechev, University of Victoria 1

Overview

Scaling Laws
Microfabrication of Chips
Overview of MUMPs Microfabrication

© N. Dechev, University of Victoria 2

 Scaling Laws

Scaling laws can be used to answer the question of “Why go small ?”

They allow us to determine whether physical phenomena will scale
more favorably or will scale poorly.
Generally, smaller things are less effected by volume dependent
phenomena such as mass and inertia, and are more effected by
surface area dependent phenomena such as contact forces or heat
transfer.
Friction > Inertia
Heat Dissipation > Heat Storage
Electrostatic Force > Magnetic Force

© N. Dechev, University of Victoria 3

Scaling Laws

Example 1: A cube of material:

a a

2
Surface Area for a cube = 6a
3
Volume for a cube = a
As we scale down, the value for volume will decrease more rapidly
than the value for surface area. In other words, the volume/surface
area ratio will decrease.

© N. Dechev, University of Victoria 4

 Scaling Laws

Assume a = 10 units
a
For a = 10:
S.A. (surface area) a a
V (volume)

Therefore, the ratio of

Assume a = 1 unit:
S.A. (surface area)
V (volume)

Therefore, the ratio

© N. Dechev, University of Victoria 5

Scaling Laws

Assume a = 0.1 unit:

a
S.A. (surface area)
V (volume)
a a

Therefore, the ratio

-What is the significance of this?

-Consider heat storage vs. heat dissipation.

-heat storage is proportional to volume
-heat dissipation is proportional to surface area

© N. Dechev, University of Victoria 6

 Scaling Laws

-Given three cubes of different sizes

a
a = 10, 1 and 0.1, with all other properties the same,
consider the ‘rate of cooling’.
a a

-Assume that all cubes start at the same high temperature,

and are allowed to cool to the same low temperature.

-The total heat contained in the smallest cube

is 1,000,000 times less than the large cube.

-The total surface area (heat dissipation area)

of the smallest cube is only 10,000 times less than
the large cube.

-Therefore, there is 100 times more heat dissipation, per unit volume,
in the smallest cube. In other words, it will cool at a much faster rate,
even though all other conditions are the same for all cubes.
© N. Dechev, University of Victoria 7

Scaling Laws

Example 2: What is the effect of scale, on the spring constant k of a

Beam?

Step 1: Derive expression for parameter of interest

- where: k = spring constant, E = young’s modulus, w = beam

width, t = beam thickness, l = beam length
Step 2: Identify all scale (i.e. length) related parameters.
- this includes w, t and l. Note that E is not a function of length.

© N. Dechev, University of Victoria 8

 Scaling Laws

Step 3: Re-define all scale related parameters in terms of a single

scale variable, such as L.
- Let: l = L, w = aL, and t = bL,
Step 4: Re-write expression in terms of the scale variable L:

Step 5: Simplify expression to derive relation between parameter of

interest and the scale variable:

- where C = constant of proportionality, in this case:

© N. Dechev, University of Victoria 9

Scaling Laws

By simplifying, we find k is proportional to L.

In other words, as L decreases, k decreases.
Therefore, the smaller the beam, the smaller the spring constant k, or
the more flexible it is.

© N. Dechev, University of Victoria 10

 Scaling Laws

Example 3: What is the effect of scale, on the buoyancy force fb of a

solid sphere, submersed in a liquid?
- Note: fb = (volume displaced) x (density of fluid) x g
- Volume of a sphere

-Therefore:

-Therefore:

© N. Dechev, University of Victoria 11

Scaling Laws

Example 4: Stress in a rod connected to a mass experiencing a

constant acceleration. How is the stress related to the scale?

Rod, of radius r
acceleration, a

Mass h

l
w

© N. Dechev, University of Victoria 12

 Scaling Laws
a r
Solution:
Step 1: Derive governing equation.

(Tensile Stress)
Mass h

where σ = stress, F = force, A = area l

w

(Newton’s Second Law)

where m = mass, a = acceleration

-assume mass of rod is negligible.

total mass m = (lwh)ρ

Therefore, total force acting on rod is: F = lwhρa

© N. Dechev, University of Victoria 13

Scaling Laws
a r
Total cross-sectional area of rod:

Therefore, the governing equation is: Mass h

l
w

Step 2: Identify all parameters related to length.

-these are: l, w, h, r

Step 3: Redefine length related parameters.

l = L, w = eL, h = fL, r = gL

© N. Dechev, University of Victoria 14

 Scaling Laws
a r
Step 4: Re-write Expression.

Mass h

l
w

Therefore, σ ∝ L.

- as L increases, σ increases proportionally.

- as L decreases, σ decreases proportionally.

Question: If the dimensions of this system were all increased by a

factor of 10x, (l, w, h, and r), is there a greater risk of rod failure?

Answer: Yes. Internal rod stress is 10x greater than before.

© N. Dechev, University of Victoria 15

Microfabrication

MEMS are fabricated with a set of technologies closely related to

IC (integrated circuit) fabrication.

Silicon is the base substrate for many microelectronic

technologies due to its semiconducting properties.
A semiconductor is a class of material where the conductivity of
the material can be controlled by various means such as doping,
temperature, light, electric fields, and other means.
Microelectronic microfabrication has been extensively developed
over the past 40 years, and the technology and infrastructure is
well established to work with silicon.

© N. Dechev, University of Victoria 16

 Microfabrication
From Sand to Chips
Natural occurring minerals containing silicon are usually silicon-
oxides, such as sand.

To make it industrially useful, it must be purified into pure silicon.

Silicon dioxide (Sand)

[http://levelx.me/technology/from-sand-to-
silicon-how-a-intel-cpu-is-built/]

Purified Silicon Chunks, [http://periodictable.com]

© N. Dechev, University of Victoria 17

Microfabrication
From Sand to Chips
Single-crystal silicon wafers are used as the base substrate of most
microelectronic devices.

Silicon crystals of bulk size do not exist naturally, but are man-
made. A common method used is called Czochralski (CZ) which
involves single silicon crystal growth from a molten mass.

Silicon Crystal Fabricator Diagram of Crystal Growth

Vision 300, [Kayex] Chamber, [CAESER]
© N. Dechev, University of Victoria 18
 Microfabrication
From Sand to Chips
The material used for growing a single crystal silicon ingot must be
electronic grade silicon (EGS), which must have 99.999999999% purity.

Begin with Crystal Seed [CAESER] Rotate the Seed and Pull Upwards [CAESER]

Silicon Ingot,
[Images from Kayex]

Crystal is pulled up[CAESER]

© N. Dechev, University of Victoria 19

Microfabrication
From Sand to Chips
The silicon “ingot” is next:
(a) sliced into wafers, and (b) wafers are lapped (mechanical grinding)

(a) Ingots are Sliced with Multi-Wire Sawing (MWS) (b) Wafers are Lapped to Remove Roughness
[Image from WaferNet Inc.] [Image from WaferNet Inc.]

© N. Dechev, University of Victoria 20

 Microfabrication
From Sand to Chips
The cut wafers are then:
(c) etched, (d) polished, and (e) stored for transport.

(c) Wafers are Placed in Acid Baths

(d) Wafers are Polished (e) Finished Raw Wafers
[Image from WaferNet Inc.]
[Image from WaferNet Inc.] [www.products.cvdequipment.com]

© N. Dechev, University of Victoria 21

Microfabrication
From Sand to Chips
PROCESSING:

Wafers are processed by growing layers of “materials” on top. This

may involve Oxide layer growth, or CVD (chemical vapour
deposition) to grow thin layers of more exotic materials like: silicon
nitride, polysilicon, or silicon oxide, etc... (right image)

Raw/Clean wafers are transported from

Manufacturer to processing facility
using “Wafer Boats” [Entegris] Step A: Oxide layer growth machine
[Prof. Dr.-Ing. H. Ryssel]
© N. Dechev, University of Victoria 22
 Microfabrication
From Sand to Chips
PROCESSING:

“Spin Coating” is used to deposit thin layers of polymers onto wafers.

Spin Coating Process

Step B: [www.phonon.com] Step B1: Hot Plate for thermal cure OR
thermal drying of polymers on
wafers

© N. Dechev, University of Victoria 23

Microfabrication
From Sand to Chips
PROCESSING:

“Chrome Masks” with UV exposure are used to selectively “cure” the

polymer and change its properties in specific areas.

Chrome Mask
[http://semimd.com]

Step C: Mask Aligner and UV Exposure system

[http://cnst.nist.gov/nanofab]
© N. Dechev, University of Victoria 24
 Microfabrication
From Sand to Chips
PROCESSING:

The regions/areas of the polymer that were not exposed to UV light

must be ‘etched/removed’ with a solvent wash.

Step D: Solvent Rinse to “Etch” away polymer that

was exposed to UV light (previous step)
Step F: [www.daitron.com]

© N. Dechev, University of Victoria 25

Microfabrication
From Sand to Chips
PROCESSING:

The regions/areas of the polymer that were not exposed to UV light

must be ‘etched/removed’ with a solvent wash.

Reactive Ion Etching (Diagram)

Step E: [Michael Huff, J. of Sensor Review] Reactive Ion Etching Machine
[http://en.wikipedia.org/wiki/Reactive-ion_etching]

© N. Dechev, University of Victoria 26

 Microfabrication
From Sand to Chips
Surface micromachining Steps A to F are
repeated for “each layer” on a chip.

Multiple layers are created for the

electrical or mechanical layers that make
the circuit, transistors, resistors, structure,
etc... For example, the CMOS wafer with
dozens of chip patterns, shown on the left.

The MEMS-based MUMPs process

(described on pp. 33-38) uses 100 mm
wafers, with 10x10 mm chip patterns. Complete processed wafer with
CMOS, etc... [Siemens]
A typical ‘chip pattern’ may be up to 10
mm x 10 mm in size. Hence the pattern is
duplicated many times onto a round silicon
wafer, to best utilize the area.
© N. Dechev, University of Victoria 27

Microfabrication
From Sand to Chips
For microelectronics, the ‘chip patterns’ must
first be inspected before the wafer is diced
(cut up into rectangular pieces).

A multi-tip probe (consisting of 30+ tips) with

an automatic machine is used to do this
inspection procedure. The probes are pressed
onto each chip pattern and electrical tests are
performed to determine: function within
allowed operating tolerances, maximum
speed, etc...

Even on a single wafer, it is possible for

Individual chips (regions) on wafers
different chip patterns to exhibit different are inspected and tested.
electrical characteristics. This is micro- [Dr. Rer.Nat. L. Frey]
fabrication variability results in ‘grading
chips’ to different quality.
© N. Dechev, University of Victoria 28
 Microfabrication
From Sand to Chips
The wafer is diced using diamond
tipped circular saw.

The diced chips are sorted based on

their previous inspection results.

They are now referred to as ‘loose die’

but due to their ‘micro-sized patterns’ Wafer is sliced into individual chip die
[Majelac Technologies]
they cannot connect to the outside
world.

For most chips, the option exists to

receive the loose die directly. For our
laboratory, we will use our own probe
tip to electrically interface with the
chips.
Loose die [N. Dechev]
© N. Dechev, University of Victoria 29

Microfabrication
From Sand to Chips
The loose die are packaged in some way
to protect them from the environment.
Microelectronic packaging is a large
industry in and of itself!

Various chip carriers (enclosures) are

available, from cheap black epoxy, to
metal cans, to high quality ceramic+gold
Die pads are wire bonded to
packages (shown below). chip carrier leads [image from S. He]

All microelectronic chips will have

‘bonding pads’ along their edge, to
which tiny gold wires (25-50 um) are
connected. The other end of these wires
are connected to the leads on the chip
carrier. Die are packaged into chip carriers or other
packaging systems [Corwil Tech. Corp.]
© N. Dechev, University of Victoria 30
 Microfabrication
From Sand to Chips
The result is the ‘chip’ shown in its
various forms below:

Epoxy packaged 44-pin surface mount chip

[curiousinventor.com]

Epoxy packaged 16-pin

DIP (dual in-line pin) chip

Ceramic packaged DIP chip

BGA (ball grid array) chip packaging systems
© N. Dechev, University of Victoria [Digi-Key, Texas Instruments] 31

MEMS Microfabrication

There are two main microfabrication methods for MEMS:

•Bulk Micromachining, which is based on the etching and
bonding of thick sheets of material such as silicon oxides and
crystalline silicon.
•Wet Etching (Anisotropic, or Isotropic)
•Dry Etching (Plasma Etching, Reactive Ion Etching)
•Surface Micromachining, which is based on the successive
deposition and etching of thin films of material such as silicon
nitride, polysilicon, silicon oxide and gold.
•Multi-User MEMS Processes (MUMPs)
•Surface Micromachining with Planarization,(MEMX,
formerly SUMMiT)

© N. Dechev, University of Victoria 32

 Surface Micromachining
Multi-User MEMS Processes (MUMPs)
MUMPs fabrication process to create a microgripper tip.

Oxide 1 2.0 um thick PSG, by LPCVD

Nitride Poly 0 0.5 um thick, LPCVD
0.6 um thick, by CVD
Substrate

© N. Dechev, University of Victoria 33

Surface Micromachining
Multi-User MEMS Processes (MUMPs)
MUMPs fabrication process to create a microgripper tip.

Poly 1 2.0 um thick, by LPCVD

Oxide 1
Nitride Poly 0

Substrate

© N. Dechev, University of Victoria 34

 Surface Micromachining
Multi-User MEMS Processes (MUMPs)
MUMPs fabrication process to create a microgripper tip.

Poly 2 1.5 um thick, by LPCVD

Oxide 2 0.75 um thick PSG, by LPCVD
Poly 1
Oxide 1
Nitride Poly 0

Substrate

© N. Dechev, University of Victoria 35

Surface Micromachining
Multi-User MEMS Processes (MUMPs)
MUMPs fabrication process to create a microgripper tip.

0.5 um thick Gold, by

‘lift-off patterning’ (requires no etch)
Metal
Poly 2
Oxide 2
Poly 1
Oxide 1
Nitride Poly 0

Substrate

© N. Dechev, University of Victoria 36

 Surface Micromachining
Multi-User MEMS Processes (MUMPs)
MUMPs fabrication process to create a microgripper tip.
Oxide Release process to ‘free microstructures’ from substrate.
Etch away Oxide 1 and Oxide 2 with HF.
Metal
Poly 2
Oxide 2
Poly 1
Oxide 1
Nitride Poly 0

Substrate

© N. Dechev, University of Victoria 37

Surface Micromachining
Multi-User MEMS Processes (MUMPs)
MUMPs fabrication process to create a microgripper tip.
Oxide Release process to ‘free microstructures’ from substrate.
Released Structure!
Metal
Poly 2

Poly 1

Nitride Poly 0

Substrate

© N. Dechev, University of Victoria 38

 Surface Micromachining
Multi-User MEMS Processes (MUMPs)
MUMPs fabrication process to create a microgripper tip.

Oxide 2
Poly 1 Lift Structure

Poly 2

Poly 1

Substrate
Oxide 1 Anchor 1

Poly 1
Poly 2
Poly 2 Gripper Tip
Gripper Tip Upper Level

2 μm
Section A Poly 2 Poly 1 Lift
Gripper Tip Structure
2.75 μm
Substrate

Microgripper tip fabricated with MUMPs

[N. Dechev]
© N. Dechev, University of Victoria 39