Metallization
The process of metallization is one of the most vital and lengthy processes in creating
an integrated circuit. The metals laid down during this process provide the connections
that circuit needs in order to work. Without metallization, the electrons would be unable
to move. Metallization begins after the formation of the n channel in the wafer and does
not end until the wafer is ready for a protective buffer layer coating. During the
metallization process, the wafer moves to and from different areas in the manufacturing
facility for processes such as photoresist application and ion implantation. This section
will explain the purpose of the different metals and detail the process of applying each
layer. While reviewing the process the reader may find it helpful to use the links
provided to learn about the processes that occur during the metallization process.
Metallization makes use of four primary processes. These four processes involved in
metallization are: Sputtering Evaporation
Electro-Plating Liftoff
![]()
![]()
Move the mouse over each picture to start videos of getting the wafers
into the liftoff machine and washing the wafers.
Sputtering, evaporation, and electro-plating are all different ways of applying the
various metals while liftoff is the process of removing the excess metal. Below is
informative overview of the metallization process.
The metallization process begins after the formation of n channels in the wafer. The
first metal applied is the gate metal, titanium tungsten nitride (TiWN). TiWN is applied
using the sputtering process to coat the entire wafer as
shown below. The benefit of using TiWN is that TiWN is essentially a refractory metal (W).
Because of this property it can withstand the 900oC temperatures used later in
the ion-implantation anneal.
(Source: ITT GaAsTEK)
Once the TiWN is down, the wafer is ready for a layer of photoresist. The photoresist,
applied by photolithography, covers all of the TiWN
except for where the n-channels (the gates) are. Then, using evaporation,
the wafer is coated with a thin layer of nickel (Ni).
(Source: ITT GaAsTEK)
The liftoff process then removes the nickel coating on the photoresist. This removes
all of the nickel, except for that which covers the TiWN over the n-channel. The
photoresist is also removed.
(Source: ITT GaAsTEK)
The TiWN then undergoes a reactive ion etch that removes all of the TiWN, except for
that which is under the Ni. This results in the Ni and TiWN forming a T-shape. The purpose
of this T-shape is to keep the n+ implant, generally Si+, from doping the TiWN during the ion-implantation process.
(Source: ITT GaAsTEK)
After the removal of the protective nickel barrier, the wafer undergoes a series of
photolithography and ion-implantation processes. These processes place more implants into
the wafer and form boundaries between the transistor and other elements of the circuit.
Once these steps are completed, another photoresist is applied. This one covers the SiON
layer applied during the photolithography and ion-implantation processes. This resist
covers everything except for the n+ regions of the transistor. The n+ regions are where
source and drain of the transistor reside.
(Source: ITT GaAsTEK)
Nickel (Ni), Germanium (Ge), and then Gold (Au) are then evaporated onto the wafer to
form the ohmic metal layer. The nickel acts as an adhesion layer while the gold and
germanium make up an eutectic mixture that consists of approximately 30% Ge and 70% Au.
This mixture has the advantage of having a low melting point of approximately 350oC
to 400oC. Without this mixture, it would take a temperature of around 600oC
just to cause the gold to melt. Once the mixture melts, the Ge diffuses into the uppermost
portion of the n+ layer, increasing the n+ dopant level. Because of the high level of
dopants, the electrons are able to tunnel through the Schottky barrier formed by the gold
over the n+ doped Ge. The quantity of electrons travelling by the tunneling process is
fairly linear for I-V mixtures. This is, in fact, the nature of the ohmic contact.
Normally the electrons would have to travel overtop of the barrier, which is an
exponential process for I-V compounds.
(Source: ITT GaAsTEK)
The undesired metal covering the resist undergoes liftoff and a different photoresist
replaces the resist used in the previous process. The next coating, applied by evaporation, is metal 1. Metal 1 consists of three metals
applied in the following order: titanium (Ti), palladium (Pd), and gold (Au). Titanium
provides adhesion and gold provides the conductivity. The layer of palladium, placed
between the two, provides a barrier to prevent the Ti and Au from chemically interacting.
If there was no barrier between the two, the two would chemically interact and the Au
could possibly seep into the GaAs wafer, which would be disastrous. It is also important
to note that a barrier is not needed between the Au of the ohmic metal and the Ti of metal
1 because the layers are sufficiently thin so that no chemical interaction takes place.
(Source: ITT GaAsTEK)
Metal 1 also forms some of the metal interconnects in the circuits. It is not a global
metal though in that it cannot cross ohmic metals. Metal 1 also serves as the bottom plate
in the capacitors and it connects the inside of the inductors to the outside circuit.
Through photolithography processes, a layer of SiN, used as the dielectric in the
capacitor, is deposited onto the entire circuit. Via’s are then etched through the
SiN.
(Source: ITT GaAsTEK)
The next two layers of metal placed on the circuit are TiWN, deposited by sputtering, and Ti/Au, deposited by evaporation.
(Source: ITT GaAsTEK)
Next, a thick layer of photoresist, placed by photolithography, covers everything but
the sources and drains of the transistors, and any connection made between them and other
devices in the circuit. A layer of electro-plated gold fills
these voids. This thick layer provides the majority of the wiring by connecting the source
and drain of the transistors to the rest of the circuit.
(Source: ITT GaAsTEK)
Finally, the metallization process is complete and after the removal of the
photoresist, the circuit is ready for a buffer layer coating.
(Source: ITT GaAsTEK)
Evaporation is a physical vapor deposition (PVD) process for depositing a thin film of
metal on the surface of a wafer. This process works by heating a metal in a vacuum chamber
to a hot enough temperature so that the vapor pressure of the metal is significant. The
wafers in the chamber are left at room temperature. Once enough of the metal has
evaporated, it will begin to condense on the surface of the wafer just as steam condenses
on a bathroom mirror after someone takes a hot shower. By controlling the temperature and
composition of the metal being evaporated, the amount of time evaporation takes place, and
the temperature of the wafer, the final amount, structure, and composition of the metal on
the wafer can be controlled.
There are several different methods for heating the metal to evaporate it. For metals
with a low melting point (around 1000 K), the metal is placed in a crucible and a current
is sent through it. The current heats the metal, but not the crucible, so the only thing
that evaporates is the metal. This will not work for metals with a high melting point,
however, because higher temperatures will cause the crucible material to evaporate as well
and contaminate the film on the wafer. To heat these materials an electron beam is focused
on a very small area of the metal to be evaporated. The electrons heat the metal as they
collide with it and melt a small portion of the metal. This way, the only material
touching the melted metal is more of the same metal. The crucible never heats to the point
that it contributes significantly to the vapor pressure.
Once the metal starts to evaporate, it acts as a point source of metal atoms. The atoms
can be thought of to shoot straight out of the metal in all directions. The wafers are
held in place on a semi-spherical surface above the metal. As the metal melts, the wafers
are rotated about the metal to ensure a uniform layer of metal deposits on each wafer (see
Figure 1).
Figure 1, Evaporation Diagram
(Source: ITT GaAsTEK)
Evaporation works well for many elements, but not so easily for alloys and compounds.
The problem arises from the fact that different parts of the alloy or compound may have
different vapor pressures for a given temperature. In a material made of two metals, the
metal with the lower melting point will have a higher vapor pressure than the second
metal. Therefore, more of the metal with the lower melting point will be deposited on the
wafer. This can be corrected by altering the composition of the original metal to
compensate, a process known as overloading. If the original metal is overloaded with the
metal that has a lower vapor pressure, less of that metal evaporates, but there is more of
it, so the two effects counter each other. This is only practical for metals with melting
points that are close to each other and it is still difficult to control the composition
of the metals on the wafer. For metals with melting points that differ significantly,
other techniques, such as sputtering, must be used.
How ITT Uses Evaporation
At ITT, most metal deposition is done by evaporation. Metals that are evaporated
include nickel, germanium, gold, titanium, and palladium. The machines that ITT uses
(pictured in Figures 2 and 3) work by accelerating electrons at the metal target. The
electron source (pictured in Figure 4) works by thermionic emission, the same principle
that a television uses. The element is heated by a current and emits electrons. These
electrons are accelerated towards the metal target in the evaporation chamber using
magnetic fields. Since they evaporate different metals at different stages in the process,
the evaporation machine holds several different metals and allows the operator to select
which one to use (see video below).
CHA Industries Mark 50 Evaporator
 Another
Evaporator Used at ITT
|
 Electron
Source
|
![]()
Move the mouse over the picture to play a video of metals in the
evaporation machine.
ITT also uses titanium-tungsten-nitride (TiWN) for gate metallization, but this
compound cannot be evaporated because the melting point of tungsten is so high (see
above). To learn more about how ITT deposits this compound, go to the sputtering page.
Electroplating is the deposition of a coating on an object by an electric current. It
is an electrolytic process in which the electrolyte is a solution containing ions of the
metal to be deposited. In an electrolytic process, ions from the anode are attracted to
the cathode by applying an electric voltage between them. The anode is the metal to be
deposited and the cathode is the surface to be plated. As metal ions move from the anode
to the cathode, they are reduced and deposited on the surface forming a thick layer of
metal. At the same time, the anode metal dissolves, providing more metal ions in the
electrolyte. When a non-dissolving anode is used, metal ions are replenished by adding
fresh electrolyte. Electroplating requires that plating current be conducted to all
portions of the wafer to produce a uniform thickness.
Basic Electroplating Process (Source: ITT GaAsTEK)
Electroplating is usually the last major step in the fabrication process of a GaAs
wafer and is often used to deposit metal on the backplane. It is usually saved for the
last step because perfoming subsequent lithographic operations on the thick plated metal
would be very difficult.
Electroplating is used when it is necessary to deposit a thick layer of gold. The
plating is commonly between 4.5 to 5 microns thick. Because plating is a much more
efficient process than evaporation and sputtering, it is the only economical means for
depositing thick layers of gold. Evaporation and sputtering would use large amounts of
gold, most of which would be deposited on places other than the wafer, and would have to
be reclaimed. In addition to economical and recovery problems, such thick films would be
very hard to pattern using etching or liftoff processes.
Plating on GaAs devices is almost always done in gold because of gold’s many
favorable properties for electronic components.
Gold’s Advantageous Properties
- high electrical conductivity
- resistant to oxidation
- easily soldered or welded
- extremely ductile
- plates well
- doesn’t react with oxygen, nitrogen, sulfur, selenium, or carbon at any temperature
- hydrogen is essentially insoluble in gold
- resistant to most acids
As with all metallization processes, good electrical contact and a properly cleaned
surface are necessary. A common practice for good electrical contact is to form a small
strip of metal near one edge of the wafer to provide an area for electrical contact that
is well-defined. The surface must be free of any exposed oxide which would be very
harmful. Usually oxides are not a problem because most GaAs processing methods consist of
plating gold on top of sputtered or evaporated gold.
Two measures of the effectiveness of the electroplating process are cathode efficiency
and throwing power. Williams defines these two terms as
follows: "Throwing power is the ability of the plating bath to produce
deposits of uniform thickness on cathodes having irregular surfaces. Cathode
efficiency (or current efficiency) is the ratio of the weight of the metal actually
being deposited to that which would be deposited if all the current has been utilized in
depositing the metal. In general, a portion of plating current may be involved in
electrochemical reactions other than metal deposition, such as depositing gases or
reducing ionic species. The type of baths used to plate pure gold for microelectronics
purposes usually have cathode efficiencies near 100% (cathode efficiencies can be much
lower in other plating processes)" (311).
Many gold plating baths use a gold cyanide complex, usually buffered acidic because the
gold cyanide complex remains stable in acidic solutions, down to a pH of about 3. These
baths operate at nearly 100% cathode efficiency and are meant to produce pure gold
(>99.99%). Platinum is used for the anode, although this is economic reason rather than
a material reason. Other anodes, such as platinized titanium or carbon anodes can be used
but these would be much more costly for plating the quantities and size needed (Williams, 312).
Step1 (GaAsTEK)
The surface is cleaned with plasma. It has a thin layer of TiWN sputtered onto the surface
and then Ti/Au is evaporated on top.
Step2 (GaAsTEK)
A thick photoresist is developed on top of the TiWN & Ti/AU mixture through
lithographic processes.
Step3 (GaAsTEK)
Clean surface with O2 plasma and electroplate thick layer of gold unto
openings.
Step4 (GaAsTEK)
Remove photoresist and inspect surface.
All four images above from ITT GaAsTEK
Copyright © ITT GaAsTEK and Virginia Tech
metallization/metallization.htm
