Transcript ECE444: Theory and Fabrication of Integrated Circuits

ECE444:
Theory and Fabrication of Integrated Circuits
Where students create integrated circuits from scratch…
Electrical and Computer Engineering Department
University of Illinois Urbana-Champaign
The ECE444 laboratory teaches core concepts for the
production of integrated circuits:
•
•
•
•
•
•
Thermal Oxidation
Photolithography
Etching
Dopant Diffusion
Metal Evaporation
Electrical Testing
Thermal Oxidation
Silicon is the dominant semiconductor used in integrated circuit processing,
in large part due to its ability to form a robust (tough) native oxide.
This oxide is used for multiple purposes in the fabrication of ICs:
– Diffusion barrier for selectively doping (adding impurities to) silicon
– Dielectric (insulator) for MOS devices
– Passivation and protection of the silicon surface
Of particular importance in semiconductor processing is cleanliness. For
oxidation, cleanliness must be targeted to the molecular level. A
specialized process called the RCA Clean is implemented before
oxidation to remove
– organic contaminants (oils)
– trace metals
– alkali ions (sodium).
Thermal Oxidation
After cleaning with the RCA clean, silicon wafers
are placed into a high temperature furnace
(900ºC < T < 1200ºC) in the presence of
oxygen or water where the following reaction
occurs:
Si + O2 → SiO2
or
Si + 2H2O → SiO2 +H2
By controlling temperature and oxidation time
precisely, oxide thickness can be predicted and
controlled accurately.
ECE444 students performing an
oxidation process in a high
temperature furnace.
Photolithography and Etching
After oxidation, the silicon wafer is completely covered with silicon dioxide. This
oxide will prevent dopants from reaching the underlying silicon wafer.
In order to create integrated circuits, the silicon wafer must be doped with
impurities (boron and phosphorus are the most common) selectively – this is
accomplished by removing the oxide in specific areas so the dopants are
allowed to diffuse (movement due to high temperature) into the exposed
silicon.
Selective removal of the oxide is performed using
– Ultraviolet light (UV) sensitive photoresist (PR) to coat the wafer
– An alignment/illumination system (mask aligner or stepper) to expose
the PR
– A mask or reticle with desired circuit pattern that only allows the UV light
to be transmitted in the shape of the circuit pattern
– Acid to etch the oxide through openings created in the PR mask by
photolithography
Photolithography and Etching
Photolithography is very much like taking a picture:
– PR coated wafer = film
– Mask aligner or stepper = camera
– Mask or reticle = subject of the picture
But photolithography is binary – either the film is exposed or not exposed; there
are no shades of gray.
Photolithography
Mask or Reticle:
The mask is tranparent plate of fused silica
(high purity glass) with an optically
opaque film applied to one surface. A
detailed layout of the circuit is created
using computer aided design (CAD)
software, and this pattern is etched into
the opaque film. The etched film on the
mask creates the hard copy of the circuit
pattern.
CAD full adder layout
The mask/reticle is then used to transmit
UV light in the pattern of the circuit.
Isometric detail view of the ECE444
CAD layout
Photolithography
PR Application:
Photoresist (an organic polymer sensitive to
UV light and resistant to attack by acids)
is applied to the oxidized wafer using a
photoresist spinner. This process uses
centrifugal force from high speed rotation
of the wafer.
The PR is applied as a small puddle in the
center of the wafer. When the wafer
spins, the PR spreads out over the wafer
due to centrifugal force. After spinning, a
uniform layer of PR remains on the
surface.
ECE444 student dispensing photoresist
onto an oxidized silicon wafer. Note the
yellow cast to the picture – short
wavelength light (green, blue, violet, and
ultraviolet) exposes PR, so it has been
filtered out of the room light, leaving only
red, orange, and yellow to see with!
Photolithography
Alignment and Exposure:
The PR coated wafer is placed into a
system (mask aligner or ‘stepper’)
which allows the mask to be aligned
to the wafer. After alignment, the
system opens a shutter to allow UV
light to illuminate the PR through the
mask for a controlled period of time.
The PR which is exposed to UV light
undergoes a photochemical reaction
to make the PR more acidic (indene
carboxylic acid is produced).
ECE444 student loading PR coated wafer into an
Ultratech 1000WF Step and Repeat Projection
Alignment system (also known as a ‘stepper’).
Photolithography
Development:
After the wafer is exposed to UV light through the mask, the acidic regions of
PR are removed by dipping the wafer into an alkaline (base) developing
solution. The acidic PR reacts chemically with the basic developer to
form water soluble salts that dissolve in the developer.
At this point the mask image can be seen in the PR (remember that the PR
was illuminated with UV light through the mask, so only light in the shape
of the circuit reaches the PR – the rest of the PR did not change!).
Note: the image from the mask has only been transferred to the PR. The
PR will be used as a mask for etching the underlying oxide in an acid
bath.
Etching
The previous steps produced a pattern in the PR
layer coating the oxidized wafer. This patterned
PR will now be used for selectively etching the
oxide areas that are exposed.
The patterned, PR coated wafer is placed into a
hydrofluoric (HF) acid bath to remove the
exposed oxide. HF will react chemically with
the oxide to form water soluble products that
dissolve in the water used to dilute the acid.
When the oxide is etched away, the silicon beneath
the oxide can be seen. Fortunately, HF does
not react with silicon (this is ideal – the HF is
selective with regards to the two materials
present on the wafer).
The PR is then removed, leaving a permanent
pattern etched into the oxide.
Oxidized and etched 100mm diameter
wafer fabricated in the ECE444
laboratory. Mask level 1 used for
photolithography. Purple areas are
silicon dioxide, silver areas are exposed
silicon.
Dopant Diffusion – physics
Silicon is a column IV element – this means there are four electrons in the
outermost shell of the atom. It is these electrons that are used when
bonding to other atoms. In a wafer, each silicon atom bonds to four other
silicon atoms (each Si-Si bond shares one electron). So in an intrinsic
(pure) silicon wafer, all the electrons in the outer shell are part of a bond –
they are ‘stuck’ between the bonded silicon atoms.
In order for current to flow in a material there must be ‘loose’ electrons. But all
the electrons in silicon are working at holding the atoms together, which
means it is not a good conductor of current.
So what can be done to allow the silicon to conduct current more easily? Free
‘carriers’ of current must be added. The goal is to find an element about the
same size as a silicon atom so that it fits together well with the silicon, but
with more electrons in its outer shell.
Dopant Diffusion – physics
In the periodic table, the closer elements are to each other, the more similar
they are. So the best candidates would come from column V (which have
five outer shell electrons). The element closest to silicon in column V is
phosphorus.
If phosphorus is inserted into the silicon wafer in a certain way, it will take the
place of a silicon atom and bond with its four neighbor silicon atoms. After
bonding, phosphorus has an electron left over that is not bonded to a silicon
atom. It turns out this extra electron is not strongly held by the phosphorus
atom any more, so it can be removed easily. This electron then becomes a
‘carrier’ for current – it is free to move around the wafer. So the conductivity
of the silicon wafer increases. This type of silicon ‘doped’ with phosphorus
is called an n-type semiconductor.
Dopant Diffusion – physics
Extending this idea of inserting an element with a different number of valence
electrons, a column III element (such as boron) could be added to the
silicon wafer. In this case, the boron will try to bond with four silicon
atoms, but it only has three electrons to bond with. This means there is
an incomplete bond with one of the silicon atoms – a ‘hole’ where an
electron would normally be. This ‘hole’ behaves much like an electron
and can move around the wafer, but with an opposite charge (+). So a
different type of current carrier is present in the wafer that increases the
wafer’s conductivity. This type of silicon ‘doped’ with boron is called a ptype semiconductor.
By adding impurities to silicon, the conductivity increases. This conductivity
can be adjusted by the amount of impurity added.
Dopant Diffusion – physics
Now for the interesting part - when n-type silicon comes into contact with p-type
silicon. A built-in potential (voltage) develops that must be overcome before
current can flow from the n-type to p-type regions.
Think of carriers as being able to only move across a flat surface or down a
slope. The built in potential is a hill that the carrier can not go up. So in
order for the carrier to keep moving, the low part must be pushed up to be
level or higher than the top of the hill. In the case of an n-type / p-type
junction, the energy to push up the low side comes in the form of a voltage
applied to the wafer. The voltage is used to ‘push up’ the ‘ground’ on the low
side of the hill before current flows from n-type to p-type regions.
But if the voltage is reversed, the energy is used to push the low side lower
while keeping the high side at the same height! That means the carrier
probably won’t ever make it up the higher hill, so it is stuck (no current
flows).
Dopant Diffusion – physics
So when n-type silicon is brought into contact with p-type silicon (a pn junction),
current can flow in only one direction. This is the fundamental
semiconductor device – a pn junction diode – a one way switch for current.
The devices used in integrated circuits are specialized combinations of pn
junctions. The junctions are formed by the addition of impurity atoms from
columns III and V of the periodic table into the silicon wafer through
diffusion.
Dopant Diffusion - Predep
The goal of the dopant predeposition diffusion is to
move dopant atoms from a source to the wafer,
and then allow the dopants to diffuse into the
wafer.
The source of dopant can be in several forms – solid
(boron nitride and phosphorus oxide ceramic
discs), liquid (boron tribromide and POCl3), or gas
(diborane or phosphine).
In order for the dopants to move into the silicon, they
must be given energy, usually in the form of heat.
In order for the diffusion to occur in a reasonable
time, the temperature must be very high (900ºC
<T<1200º).
At this temperature the dopant (in the form of an oxide)
reacts with the exposed silicon surface to form a
highly doped glass. It is from this glass that the
dopants can then diffuse into the wafer.
The first dopant diffusion in the ECE444 process is a
boron diffusion (the wafer is originally doped at a
low level with phosphorus). This diffusion forms
the first pn junctions selectively on the wafer
through the openings in the oxide defined by
photolithography.
ECE444 Diffusion furnace
Dopant Diffusion - Drive
After the predeposition diffusion the dopants are situated
close to the surface of the wafer. However, they must
diffuse even farther to lower the overall concentration
in order for some of the devices to work properly.
The first diffusion (predeposition) introduces dopants into
the wafer.
The second diffusion (drive) redistributes the dopants and
allow the dopants to diffuse into the wafer more
deeply (up to ~3 micrometers)
In addition, oxygen and water vapor are introduced during
the drive diffusion to grow a new oxide over the areas
which were exposed to bare silicon during the
photolithography process. This new oxide can be
patterned again so that other selective diffusion
processes can be performed to create other types of
devices.
100mm diameter wafer fabricated in
the ECE444 laboratory following
boron predeposition, boron drive,
and re-oxidation.
Repeat the process…
At this point, the process of oxidation-photolithography-etching-diffusion can be repeated to produce
the various types of electronic devices required for a circuit.
Some modern processes may require more than 20 iterations of this sequence!
Oxidation
Photolithography
Etching
Diffusion (Ion
Implantation)
The following slides show the rest of the processes
performed in the ECE444 lab.
ECE444 Process
Phosphorus Diffusion
• Photolithography Mask 2
• Etch
• Phosphorus Predeposition
Gate Oxide Formation
• Photolithography Mask 3
• Etch
• Gate oxidation
ECE444 Process
Electrical contact vias (holes) to silicon
• Photolithography Mask 4
• Etch
Metal definition
• Photolithography Mask 5
Metallization
After all diffusion and oxidation steps are
completed, metal is deposited onto
the surface of the wafer. This metal is
used to ‘wire’ the devices fabricated in
the silicon together.
The wafers are put into a large chamber
and the air is pumped out of the
system. A piece of aluminum located
on a tungsten ‘boat’ in the system is
heated to high temperature, causing
the aluminum to melt and evaporate.
The evaporated aluminum will solidify
into a thin film when it touches the
silicon wafer.
Thermal evaporation vacuum system used
in the ECE444 laboratory. This tool was
designed and built for an independent
study project.
Metallization
After metallization, the wafer is completely covered by the aluminum.
It must be patterned and etched to form the actual ‘wires’ connecting
individual devices into a circuit.
Wafer after aluminum evaporation
Completed wafer
Steps to create ECE444 wafer:
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
Initial oxidation
Photolithography Mask 1
Oxide etch
Boron predep
Boron drive and re-oxidation
Photolithography Mask 2
Oxide etch
Phosphorus predeposition
Photolithography Mask 3
Gate oxidation
Photolithography Mask 4
Etch
Photolithography Mask 5
Metal evaporation
Metal definition
Time to see if it works…
Testing
After completion of the wafer, it must be
tested to verify operation.
The devices fabricated are extremely
small (some dimensions are as small
a 1micrometer!), so specialized
probes are used to make electrical
contact.
ECE444 Probe station
Once contact is made, several different
instruments are used to measure
electrical performance.
Example of electrical data from a device
fabricated in the ECE444 laboratory
ECE444 - notes
The ECE444 laboratory was started in the mid-1960s with the goal of teaching
future engineers the processes required to fabricate integrated circuits.
Over time the lab has gained recognition as one of the premier courses in
this demanding field.
This recognition is due largely to the reinforcement of theory learned in lecture
with practical hands-on experience in the lab (each student processes their
own wafer from beginning to end).
Industry provides strong support through past and continuing donations of
actual industrial processing equipment.
Wafer Photo credits: Mark Hart
ECE444 - equipment
Ellipsometer
Diffusion/oxidation furnace
Ultratech stepper
Prometrix FT650
Photoresist spin station
Thermal evaporator
ECE444 - equipment
Probe station
Four point probe
Optical microscopes
Photoresist process station
Electrical test station
Plasma asher
For more information…
See http://fabweb.ece.uiuc.edu