Professional Documents
Culture Documents
MODULE 4
Syllabus
48
Dept of ECE,VVIET
MODULE 4 MEMS (15EC831)
• Volume and surface are two physical quantities that are frequently involved in micro device
design.
• Volume relates to the mass and weight of device components.
• Ex. thermal inertia is related to the heat capacity of the solid which is a measure of how fast we
can heat or cool a solid. This is used to design a thermally actuated device.
• Surface properties are related to pressure and the buoyant forces in fluid mechanics and heat
absorption or dissipation by a solid in convective heat transfer.
• To minimize a physical quantity, one must weigh the magnitudes of the possible consequences
of the reduction on both the volume and surface of the particular device.
• Equal reduction of volume and surface of an object is not normally achievable in a scale down
process.
• Consider the example of a solid of rectangular geometry having 3 sides a>b>c.
49
Dept of ECE,VVIET
MODULE 4 MEMS (15EC831)
• Elephant has and S/V ratio of 10-4 /mm and dragonfly has10-1 /mm
• Hence dragon fly requires less energy and power and so low consumption of food and water,
while elephant needs more food to generate more energy.
• From the scaling formula, a reduction in size by 10 times(l=0.1) means 103 =1000 times
reduction in volume , but 102 =100 times reduction in surface area.
• A reduction of volume by 1000 means 1000 times reduction in weight.
• Micro mirrors are essential parts of micro switches used in fiber optic networks in
telecommunication.
• These mirrors rotate to a tightly controlled range at high rates.
• Angular momentum is an important factor in both rotation control and rate of rotation.
• Ex. Estimate the reduction of torque required in turning a micro mirror with a reduction of 50
percent in the dimensions.
50
Dept of ECE,VVIET
MODULE 4 MEMS (15EC831)
Solution: the torque required to turn the mirror about the y-y axis is related to mass moment of inertia
of the mirror Iyy, expressed as
The mass moment of inertial of the mirror with a 50 % reduction of the size becomes
A reduction of a factor of 32 is achieved in mass moment of inertia, giving 50% reduction of dimension.
51
Dept of ECE,VVIET
MODULE 4 MEMS (15EC831)
• Inertia decides the amount of force required to move a part and how fast movements can be
achieved and stopped.
• The inertia of a solid is related to its mass and the acceleration required to initiate or stop the
motion of a component.
• When minimized, the effect of reduction in the size on the power P, force F and pressure p and
the time t required to deliver the motion has to be seen
• If a solid moves from one position to another, the distance that the solid travels, S, is
S α l, where l stands for the linear scale.
The velocity v =s/t
• From particle kinematics, wkt
52
Dept of ECE,VVIET
MODULE 4 MEMS (15EC831)
Accleration a:
• Consider F=Ma in which a = F/M
• Scaling is obtained as
•
Time t:
• Consider
•
• The transient time is expressed as
•
•
• Power density is related to force scaling vector as
53
Dept of ECE,VVIET
MODULE 4 MEMS (15EC831)
•
• Using this the scaling laws for rigid body dynamics is established
•
• This table is useful in scaling down devices in a design process
Ex.
54
Dept of ECE,VVIET
MODULE 4 MEMS (15EC831)
•
•
• The above breakdown voltage V varies with the gap between the two plates as per paschen
effect.
• Fig 6.5 shows that the breakdown voltage V drops drastically with increase in d for d<5µm.
• It decreases after the gap widens from d>5µm.
• Voltage variation reverses at d ͌ 10µm.
• The breakdown voltage increases linearly as gap increases.
• We can say that applied voltage Vαd or in scaling Vαl1 for the working range of d˃ 10 µm
• The scaling of is neutral, so α
• The electrostatic potential is expressed
•
•
55
Dept of ECE,VVIET
MODULE 4 MEMS (15EC831)
•
• A factor of 10 decrease in linear dimensions(W,L,d) will decrease the potential energy by
a factor of 103 =1000
• Wkt, the electrostatic forces can be produced in 3 dimensions in parallel plate arrangements
as
56
Dept of ECE,VVIET
MODULE 4 MEMS (15EC831)
57
Dept of ECE,VVIET
MODULE 4 MEMS (15EC831)
• This shear deformation is considered possible by a relative motion of a pair of plates placed at
the top and bottom of the bulk fluid
ER
• The relative motion of the plates represents a shear force which causes the fluid flow.
• The associate shear stress τ is considered proportional to the rate of change of induced shear
strain θ given by
58
Dept of ECE,VVIET
MODULE 4 MEMS (15EC831)
•
• Figure 6.8 shows a fluid flow through a small circular conduit of length L and radius a.
59
Dept of ECE,VVIET
MODULE 4 MEMS (15EC831)
• The pressure drop over the length L is computed using Hagen-Poiseuille law as
•
• The pressure drop for a section of a capillary tube of length L is computed as
•
• The scaling laws for fluid flows in capillary tubes are derived as for volumetric flow and
for pressure drop per unit length, whre a is the radius of the tube.
•
60
Dept of ECE,VVIET
MODULE 4 MEMS (15EC831)
Ex.
Also pressure drop per unit length is so this increases by 102 = 100 times.
• When the radius of the conduit s very small, capillary effect appears.
• This is due to surface tension of the fluids.
• The pressure required to overcome the surface tension is
• So, the pressure drop per unit length of a liquid is
•
•
Ex. What will happen to the pressure drop in the fluid in above ex. If the tube radius is microscale?
Soln: the pressure drop per unit length of the tube, from the scaling law in equation increases
1000 times with a 10 times reduction of the tube radius.
The situation is one order of magnitude more severe than the case in meso or macroscale.
• Due to adverse effect in scaling down, special pumping techniques such as piezoelectric,
electro-osmotic, electrohydrodynamic pumping are used
• These are based on surface pumping forces. Ex. Piezoelectric pump.
• Here the forces generated on tube wall is used to drive the fluid flow is used.
• The surface force F, proportional to the surface area of the inner wall of the tube scales
favorably.
• This surface area of the inner wall of the tube is S= 2πaL and the volume of the fluid is V= πa2L
• The surface area to volume ration S/V = 2/a.
• Scaling down the tube radius will result in the increase of the surface force available for
pumping the unit volume of the fluid
61
Dept of ECE,VVIET
MODULE 4 MEMS (15EC831)
• Scaling laws are done in two scales, one being the meso and microscale and the other in sub
micrometer scale(<1µm).
• With this sub micrometer designs thermos physical properties vary with the size of MEMS.
6.6.1 Scaling in Heat Conduction
Basics
Heat conduction in a solid slab:
• In the figure, a solid slab with the temperature at the left side wall maintained a Ta and
the right side wall maintained at Tb with Ta > Tb
• The temperature difference causes the heat to flow from left side to right side of the slab.
• The total amount of heat flow through the slab, Q, is proportional to the cross-sectional
area A, the temp difference between the two faces, the time t allowed for the heat to
flow.
• The heat flow is inversely proportional to the distance that the heat has to travel, i.e.,
thickness d of the slab
62
Dept of ECE,VVIET
MODULE 4 MEMS (15EC831)
•
• Where qx is the heat flux along x coordinate, k is the thermal conductivity of the solid
and T(x,y,z,t)is the temperature field in the solid in a Cartesian coordinate system at
time t.
• The heat conduction in a solid is
•
• From the above equation, the scaling law for heat conduction for solids in meso and
microscales is Qα l1 (A is replaced by α)
• So the reduction in size leads to the decrease of total heat flow in a solid
6.6.1.2 Scaling in Termal Conductivity in Submicrometer Regime
• Thermal conductivity k in solids is estimated by
• Here λ α 1/ρ, where ρ is the mass density of the solid with an order similar to volume
i.e., l3
• The scaling of heat flow in a solid in the sub micrometer regime is obtained by
combining the above two equations.
•
• So, a reduction in size of 10 would lead to a reduction of total heat flow by 100
6.6.1.3 Scaling in Effect of Heat Conduction in Solids of Meso and Micro scales
• F0 is the Fourier number with has no dimension
• It is used to determine the time increments in a transient heat conduction analysis
63
Dept of ECE,VVIET
MODULE 4 MEMS (15EC831)
•
• Here α is the thermal diffusivity of the material
• t is the time for heat to flow across the characteristic length L
• Fourier number is the ratio of the rate of heat transfer by conduction to the rate of energy
storage in the system.
• From the above equation, the scaling in time for heat conduction in a solid is
•
• Where both F0 and α are constants
Ex. Estimate the variation of the heat flow and the time required to transmit heat in a solid with a
reduction of size by a factor of 10
Soln: from the equations
and
the total heat flow and the time required for heat transmission are both reduced by 102 =100 times with
a reduction of size by a factor of 10
• There exists the boundary layers at the interfaces of solids and fluids.
• Heat transfer in fluids is in the mode of convection governed by Newton’s cooling law
expressed as
•
• Heat transfer h depends on the velocity of the fluid
• This does not play a significant role in scaling the heat flow.
• The total heat therefore depends on the cross-sectional area A, of the order of l2
• Therefore the scaling of heat transfer in convection is for fluids in meso and micro
regimes.
• Consider only the convective heat transfer of gases in the regime
64
Dept of ECE,VVIET
MODULE 4 MEMS (15EC831)
• When the gases pass in narrow channels at sub micrometer scale, the heat transfer break down.
• The convective heat transfer has become conduction of heat among the gas molecules due to
boundary layer effect
• Consider the figure given
• The passage is represented by a gap H <7λ, where λ is the mean free path of the gas
• The transfer of heat between the two plates is predominantly by conduction through the gas
between the plates rather than convection.
• The mean free path is proportional to the reciprocal of mass density of gas as
• C is the specific heat of the gas at constant volume and the mean velocity V is obtained from
•
• Here T is the mean temperature of the gas and m is the molecular weight of the gas
65
Dept of ECE,VVIET
Module 5 Mems and MicroSystems(15EC831)
Module 5
OVERVIEW OF MICRO
MANUFACTURING
Syllabus
5.1 Introduction
66
Dept of ECE, VVIET
Module 5 Mems and MicroSystems(15EC831)
5.1 Introduction
67
Dept of ECE, VVIET
Module 5 Mems and MicroSystems(15EC831)
• The part of the substrate that is not covered by the protective mask is dissolved in the etchants and
removed.
• The etching can undercut the part under the protective mask also.
• It may also damage the protective mask itself.
Basics
Why Silicon is chosen as an ideal substrate for MEMS?
• It is mechanically stable and it can be integrated into electronics substrate
• Si is an ideal structural material. It has Young’s modulus as steel, but is light as aluminum.
• Materials with high young’s modulus gives linear relationship between applied load and the
induced deformations
• It has a melting point at 1400 degree centigrade, hence it is stable.
• Its thermal coefficient of expansion is 8 times smaller than steel and 10 times smaller than
aluminum.
• It has no mechanical hysteresis and has greater flexibility
Crystal structure of silicon
• Silicon has a face centered cubic(FCC) unit cell as in figure
• The unit cell has atoms situated in fixed locations defined by lattice(imaginary lines) of dimension
b called lattice constant.
• In FCC, atoms are situated at 8 corners of cubic structure and at the center of each of six faces
68
Dept of ECE, VVIET
Module 5 Mems and MicroSystems(15EC831)
• The 4 additional atoms form a sub cubic cell of the diamond lattice.
• A Si unit cell has 18 atoms with 8 atoms at the corners and 6 atoms on the faces and another 4
interior atoms.
5.2.2 Isotropic and Anisotropic Etching
• For substrates made of homogeneous and isotropic materials, the chemical etchants will attack the
material uniformly in all directions- called isotropic etching
• This is hardly desirable in micro manufacturing due to lack of control of the finished geometry of
the work piece.
• Most substrate materials are not isotropic in their crystalline structures.
• Ex. Silicon has a diamond cubic crystal structure.
• Hence some parts of crystal are stronger and more resistant to etching than others.
• Three planes of silicon crystals are of interest in micro machining.
• They are (100), (110) and (111) planes as shown
• Two most common orientations used in the IC industry are (100) and (111) orientations. And in
micromachining (110) is the favored orientation.
• In this orientation, the wafer breaks or cleaves more cleanly.
• The (111) plane is the toughest plane to treat, hard to etch and is least used in micro machining.
• The plane(111) intersects the plane(100) at an angle of 54.740.
• In this case we can expect different etching rates in different orientations.
•
• Disadvantages of anisotropic etching is
o It is slower than isotropic etching
o The etching rate is temperature sensitive
o It usually requires an elevated temperature around 1000c in the process
69
Dept of ECE, VVIET
Module 5 Mems and MicroSystems(15EC831)
• Alkaline chemicals with pH > 12, are used for anisotropic etching.
• Popular anisotropic etchants for silicon include potassium hydroxide (KOH), ethylene—diamine
and pyrocatecol (EDP), tetra-methyl ammonium hydroxide (TMAH), and hydrazine.
• Most etchants based on the above chemicals are diluted with water, normally 1:1 by weight
• The selectivity ratio of a material is defined as the ratio of the etching rate of silicon to the etching
rate of another material using the same etchant.
• For example, silicon dioxide has a selectivity ratio of 103 , meaning this material has an etching
rate that is 103 times slower than the etching rate for silicon.
• The higher the selectivity ratio of the material, the better the masking material it is.
• The timing of etching and the flow patterns of etchants over the substrate surfaces need to be
carefully controlled in order to avoid serious under-etching and undercutting
• Use an SiO2 layer as mask for the silicon substrate in KOH etchants for trenches of modest depth.
• It is inexpensive and etching is a slow process, the SiO2 mask itself can be attacked by the etchants
if the system is left in the etchant for a long period of time, as in deep etching
• In such cases, silicon nitride should be used as the mask instead
70
Dept of ECE, VVIET
Module 5 Mems and MicroSystems(15EC831)
71
Dept of ECE, VVIET
Module 5 Mems and MicroSystems(15EC831)
Plasma Etching
• Plasma is a neutral ionized gas carrying a large number of free electrons and positively charged ions.
• A common source of energy for generating plasma is a radio-frequency (RF) source.
• The process involves adding a chemically reactive gas such as CCl2F2 (DichloroDifluoromethane)
to the plasma
• It contains ions and has its own carrier gas (inert gas such as argon gas).
• In the Figure, the reactive gas produces reactive neutrals when ionized in the plasma.
• The reactive neutrals bombard the target on both the sidewalls as well as the normal surface, whereas
the charged ions bombard only the normal surface of the substrate.
• Etching of the substrate materials is accomplished by the high-energy ions in the plasma bombarding
the substrate surface
• Chemical reactions occur between the reactive neutral ions and the substrate material.
• This reaction causes local evaporation, and results in the removal of the substrate material.
• The etching front moves more rapidly in the depth direction than in the direction of the sidewalls.
• This is due to the larger number of high-energy particles(both the neutral ions and the charged ions)
bombarding the normal surface, while the sidewalls are bombarded by neutral ions only.
72
Dept of ECE, VVIET
Module 5 Mems and MicroSystems(15EC831)
• It is highly desirable that the angle θ be kept at a minimum in deep-etched trenches that separate the
plate electrodes.
• Obtaining deep trenches with vertical walls have been a major impediment of bulk manufacturing
for a long while.
• The bulk manufacturing technique is regarded as suitable only for MEMS with low aspect ratios,
with tapered cavity walls."
•
• Deep reactive ion etching (DRIE) is a process that can overcome this problem.
• The DRIE process has extended the use of the bulk manufacturing technique to the production of MEMS
of high aspect ratio with virtually vertical walls., i.e., θ ͌ 0.
• The DRIE process differs from dry plasma etching in that it produces thin protective films of a few
micrometers on the sidewalls during the etching processes.
• It uses a high-density plasma source.
• It allows plasma (ion) etching of the substrate material and the deposition of etching-protective material
on the sidewalls as in Figure 9.7b.
• Etching-protective materials (shown in black in the figure) are materials of high selectivity ratio, ex. silicon
dioxide, polymers.
• Polymeric materials such as photoresists are produced by polymerization during the plasma etching
process.
• Reactant gases could be used in DRIE. Ex. fluoropolymers (nCF2) in the plasma of Argon gas ions.
• This reactant can produce a polymer protective layer on the sidewalls while etching takes place.
•
5.2.6 Comparison of Wet versus Dry Etching
73
Dept of ECE, VVIET
Module 5 Mems and MicroSystems(15EC831)
74
Dept of ECE, VVIET
Module 5 Mems and MicroSystems(15EC831)
•
• Surface micromachining saves material and also eliminates the need for a die attach.
75
Dept of ECE, VVIET
Module 5 Mems and MicroSystems(15EC831)
• A mask (mask 1) is made in step 2 to cover the surface of the PSG layer for the subsequent etching.
• Another mask (mask 2) is made for the deposition of polysilicon microstructural material in step 4.
• The PSG that remains in step 5 is subsequently etched away to produce the desired cantilever beam
as shown in step 6.
• The etchant used in the last step for the sacrificial PSG layer is 1:1 HF.
• After etching, the structure is rinsed in deionized water thoroughly followed by drying under infrared
lamps.
•
5.3.3 Mechanical Problems Associated with Surface micromachining
• There are three major problems that result from surface micro-machining.
• These are (1) adhesion of layers, (2) interfacial stresses, and (3) stiction.
Adhesion of Layers
• Whenever two layers of materials, whether similar or dis-similar, are bonded together, a possibility
of delamination exists.
• A bilayer structure can delaminate at the interface either by
o peeling of one layer from the other or
o by shear that causes the severing of the interfaces locally along the interface.
o Figure 9.11 illustrates both these failures.
76
Dept of ECE, VVIET
Module 5 Mems and MicroSystems(15EC831)
•
• Excessive tensile residual stress in Si02 layer can cause multiple cracks in the layer.
• The third type of stress that could be introduced in thin-film structures is the intrinsic stress
• This is due to the local change of atomic structure during microfabrication processes.
• Excessive doping, for ex. could introduce substantial residual stresses.
Stiction
• In surface micromachining, the phenomenon of two separated pieces sticking together is called
stiction.
• For ex. difficulty in separating two transparencies after the thin dividing paper is pulled out.
• Stiction is the most serious problem to deal with in surface micro-machining.
• It occurs when the sacrificial layer is removed from the layers of the material that it once separated.
• The thin structure that was once supported by the sacrificial layer may collapse on the other material.
• for ex, the production of a thin beam in Figure 9.13
• Stiction could happen with the thin polysilicon beam dropping onto the top surface of the silicon substrate
after the removal of the sacrificial PSG layer (Fig. 9.13b).
• The two materials would then stick together after the joint.
• Considerable mechanical forces are required to separate the two stuck layers again
• These excessive forces can break the delicate microstructure.
• Stiction is the main cause for the large amount of scraps in surface micromachining.
•
•
• Stiction occurs due to hydrogen bonding of surfaces during rinsing of the interface after the etching of the
PSG sacrificial layer, or by forces such as the van der Waals forces
• Remedial actions are temporary spacers using polysilicon and sacrificial polymer columns that can be
removed by etching with oxygen plasma afterward.
•
77
Dept of ECE, VVIET
Module 5 Mems and MicroSystems(15EC831)
78
Dept of ECE, VVIET
Module 5 Mems and MicroSystems(15EC831)
• Masks used in the x-ray lithography are transparent to x-rays, so a thin film of gold is applied to the
area that will block x-ray transmission.
•
• The thin mask used for this purpose is silicon nitride.
• The deep x-ray lithography will cause the exposed area to be dissolved in the subsequent
development of the resist material(Fig. 9.15b).
• The PMMA photoresist after the development will have the outline of the product.
• Next is electroplating of the PMMA photoresist with a desired metal, usually nickel, to produce the
tubular product of the required wall thickness (Fig. 9.15c)
• The desired tubular product is produced after the removal of the photoresist materials by oxygen
plasma or chemical solvents.
79
Dept of ECE, VVIET
Module 5 Mems and MicroSystems(15EC831)
Photoresist Materials
Basic requirements for photoresist materials for the LIGA process include the following: -
➢ It must be sensitive to x-ray radiation.
➢ It must have high resolution as well as high resistance to dry and wet etching.
➢ It must have thermal stability up to 140°C.
➢ The unexposed resist must be absolutely insoluble during development.
➢ It must exhibit very good adhesion to the substrate during electroplating.
• PMMA is an optimal choice of photoresist material for the LIGA
• But, its low lithographic sensitivity makes the lithographic process extremely slow.
• Another shortcoming of PMMA is its vulnerability to crack due to stress.
5.4.3 Electroplating
• Electroplating is an important step in the LIGA process.
• Electroforming of metal films onto the surface of the cavities in the photoresist after x-ray
lithography has been performed as illustrated in Figure 9.15c.
• Nickel is the common metal to be electroplated on the photoresist walls.
• Other metals that could be used for electroplating include Cu, Au, NiFe and NiW.
• The conductive substrate and the carrying photoresist structure form the cathode in an electroplating
process is illustrated in Figure 9.16.
•
• Electroplating works on the principle that the nickel ions Ni 2+ from the nickel chloride (NiC12)
solution react with the electrons at the cathode to yield nickel as shown in the process below:
Ni 2+ + 2e - → Ni
• The H2 gas is the product of the H2+ ions produced by the electrolysis of the solution.
• This is present on the surface of the cathode and may cause non-uniform Ni plating.
• The chemical reaction of H2+ ions to H2 gas at the cathode is presented below:
2H + + 2e - → H2
• To avoid damage of the plated surfaces by hydrogen bubbles, we need to control the pH of the
solution, the temperature, and the current density in the electrolysis.
80
Dept of ECE, VVIET
Module 5 Mems and MicroSystems(15EC831)
Bulk Micromanufacturing
• Straightforward, involving well-documented fabrication processes.
• Less expensive in the process, but material loss is high.
• Suitable for simple geometry, e.g., micropressure sensors dies and some actuating elements.
• Limited to low-aspect ratio in geometry; i.e., the surface dimensions are much greater than that
of the depth. This is because the overall height of the microstructure is limited by the thickness of
commercially available silicon wafers.
Surface Micromachining
• Requires the building of layers of materials on the substrate.
• Complex masking design and productions.
• Etching of sacrificial layers is necessary.
• The process is tedious and more expensive.
• There are serious engineering problems such as interfacial stresses and stiction.
• Major advantages: (1) not constrained by the thickness of silicon wafers; (2) wide choices of
thin film materials to be used; (3) suitable for complex geometries such as micro valves and
actuators.
81
Dept of ECE, VVIET