Slide 1

DOPING: DEPOSITING IMPURITIES

INTO SI IN A CONTROLLED MANNER

In the last chapter diffusion, we have seen that the dopants such as B, P could be
added to the Si wafer via diffusion process.

We discussed about two cases (1) constant source and (2) Limited source and , 2 step
diffusion (pre-deposition and drive in deposition) is employed.
In constant source, the concentran profile follows erfc and in limited source, the
concentration profile follows Gaussian.

We discussed other topics such as gettering, sheet resistance and dopant redistribution
as well.

In this chapter, we discuss about ion-implantation , the other method available for
doping.
Slide 2

ION IMPLANTATION

One of the methods for modifying Si wafer.

Why Ion Implantation? Why not diffusion which we

discussed in the last chapter?

Although diffusion is a relatively less expensive process, , the depth of doping and the
quantity of doping is not precisely controllable. Hence the transistor dimensions would
not be very precise. When the transistor dimensions were 10 micron (i.e.10,000 nm)or
so, if the variability in doping level is 100 nm, it was acceptable because it was only 1%.
When the transistor size itself is 100 nm, this much variability is not acceptable.

For the ICs with small transistors, ion implantation is used. In ion implantation, the
phosphorous (or boron) must be first ionized. Then they must be accelerated and made
to impinge on the wafers. Then they will penetrate into the wafer. This is similar to firing
bullets into a solid wooden door. The bullets will get embedded into the wood.

In this chapter, we will discuss about ion implantation in detail.

Slide 3

ION IMPLANTATION

Creation of defects

How impinging ion comes to rest?

Nuclear stopping
Electronic stopping

In ion implantation, the dopants must be first ionized. Then they must be accelerated
and made to impinge on the wafers.

As the doping ion comes with high energy and impinges on the substrate (Si), it creates
defects in the substrate (like vacancies, interstitials). Though it is an disadvantage, it will
also help the process in other way which we will discuss later.

The doping ion comes with certain velocity and after colliding with atoms present in the
substrates, it will come to rest i.e. ion movement stopped. This is occurring via two
process: (1) Nuclear stopping (ions colliding with nucleus of an atom and (2)
electronic stopping (ions colliding with electronic cloud of an atom)
Slide 4

ION IMPLANTATION

-dE/dX = N {SN (E) + SE (E)}

As the energy is decreasing with distance travelled by the ion in the substrate, the
negative sign is given in LHS. N = number of target atoms (Si). Energy is decreasing
when doping ion interacts with target atoms. So “N” comes in to picture here. S N(E) is
the energy term representing nuclear stopping and SE(E) represents electronic
stopping. Both are functions of energy i.e. ion energy. How? We will see later. Right now
presume that depending on ion energy whether it is high or low , one or the other
mechanism going to dominate.

The ion hits the target as shown in the diagram and it comes to rest after traversing for
certain distance. As it colliding with target atoms, it may not travel in the same direction
of incident ion. It gets deflected and go in any direction as shown by the curve in the
diagram. So the distance travelled by an ion before comes to rest is called as Range.

The expression for range is given in the slide. We are getting this expression by
substituting equation 2 in equation 1

Eo is the initial energy and hence taken in the lower limit of the integral. The final energy
is zero (as it comes to rest) and hence taken in the upper limit.
In the next step, we swapped these limits to get rid of negative sign.
Slide 5

DOPING PROFILE

But what we are interested is the maximum distance (i.e distance in the direction of
incident ion) traversed by an ion. Because it gives the penetration depth i.e. the
maximum depth that the dopant ion can penetrate through the target atoms. Since we
have many ions hits the surface, we are reporting the average value. It is defined as
projected range (RP). It is related to Range. As we are reporting average value, we
have to mention stand deviation also. In semiconductor jargon, it is called as straggle.

So, these are the two parameters that characterize the ion implantation process

Next we are interested in “ doping profile”. How it is going to look like?

If we assume the target material is amorphous, then the profile is given by simple
Gaussian function.
One question arise here. But our target material is silicon which is crystalline. How we
are addressing this? We will discuss later. Now assume, the concentration profile (N vs
x ) is given by Gaussian function and the equation is shown in the above slide. Note that
N(Rp) is the peak concentration obtained in the direction of incident beam.

The dose (Q) or the total amount of dopant in the semiconductor is given by the area
under the curve and the equations are shown in the slide.
Slide 6

DOPING PROFILE (CONTD..)

The equation in the slide clearly shows that peak concentration could by controlled by
dose. The dose in turn controlled by controlling ion beam current in the equipment.

The value of Rp and delta(Rp) depend on particular ion-target combination i.e. the
molecular masses of ion and target strongly will influence these values. The equation
for the same is given in slide no.8

If you look at the plot (concentration (N) vs. distance (x) from top of the wafer), the
surface concentration is not maximum. So, here in ion implantation , you could control
where do you want the concentration to be maximum.

Also, from the plot, you could see, as the dose (Q) increases, the N(Rp) increases.
Besides, the profile approach more towards Gaussian.

As already told, the above equations are hold true only when the target is amorphous.
But Si is crystalline.

This is achieved by the following means:

• Misalign the beam by 7-10 degrees i.e. the inclined ion beam should not be directed
along a particular crystallographic axis. Otherwise ions moving at a considerable
distance w/o colliding the target atoms. This is called channeling
• Forcing more number of ions in the target i.e. increasing the dose. This would induce
more disorder in the system (see the N vs. x curve in the slide: when Q increases,
the curve approaches Gaussian)
• Depositing thin oxide on the surface as oxide is amorphous
Slide 7

STOPPING PROCESS

We were discussing about Sn(E) and Se(E). That is both nuclear and electronic
stoppings are function of energy. Let see the model now.

The plot is drawn b/w energy loss and energy of the incoming ion.

In the plot you can see, when the energy is E0 (initial energy), the energy loss is
minimum. But after losing some energy, the energy transfer becomes more efficient and
hence the energy loss reaches to peak position. Then it falls to zero (it is obvious as the
energy of the incoming ion is lowered, the energy available for transfer is also lowered).
However, this is difficult to model. Hence, it is assumed that energy loss is constant with
energy (given by the dotted line in the plot) – this is taken to be Sn(0).

For electronic stopping, the de/dx is decreasing with energy. It is proportional to E^1/2.
Now the Sn(E) and Sn(0) is given in the same plot. It is intersecting at one point. That is
critical energy. We have two zones now (1) E> EC (electronic stopping predominant)
and (2) E<Ec (nuclear stopping predominant).

The equation for range given in slide no . 4 is modified based on the models proposed
for nuclear and electronic stopping.
Slide 8

We discussed already that Rp is related to R. Now Rp is related to R by the equation

shown in the slide.

In ion implantation also, lateral diffusion occurs. This is called transverse straggle.

The equation and profile are shown in the slide. But comparably less when compared to
diffusion process.
Slide 9

EQUIPMENT

Now, we will see what is the equipment used for ion implantation and how it will
be operated.

Equipment and operation: Ion implantation will be approximately 10 ft long, 10 ft wide

and 10 ft high
We can divide the equipment into four sections as explained below.
Ionizing chamber
Ion selection chamber
Acceleration chamber
Doping chamber

In this slide, we will discuss about Ionizing chamber: For doping phosphorous,
phosphorous pentoxide (P2O5) is used. For doping boron, boron trifluoride (BF 3) is
used. Phosphrous pentoxide is a solid. It has to be heated and evaporated. BF3 is a
gas at normal conditions. Any material used for ion implantation must be brought to
gaseous state. Let us assume that BF3 is used for doping. The atoms in BF3 must be
converted to ions. In the ionizing chamber, a low pressure is maintained and a tungsten
filament is heated using electricity .When it is very hot, electrons are emitted from the
filament in a process called thermionic emission. Some of the electrons hit the BF3
molecules and create ions such as BF3+, BF2+, BF+ and B+. Thus ions are created
from the gas.
Slide 10

EQUIPMENT

Ion selection chamber: The next section is the selecting chamber. We would like to
dope the wafer with boron only and not with fluorine. Hence only B+ must be selected
from the mixture of ions and added to the wafer. In order to select the required ion (B+),
electric field and magnetic field are used. If we send the stream of ions between two
electric plates, which have positive and negative connection, the ions will be attracted to
the negative plate and move towards the negative plate. If a magnetic field is also
applied at the same time, the direction of the movement will change. The speed and the
direction of the movement will depend on the charge and the mass of the ions. A plate
with a hole in the appropriate place can be used to block all the ions except the desired
one. Thus B+ ion can be separated from the mixture of ions.
Slide 11

EQUIPMENT

Accelerating chamber; The B+ ions will come with a moderate speed. In order to
ensure that they penetrate the wafer, they must be accelerated. For this, a set of ring
electrodes are used If negative charge is applied on the plates, the ions will get
accelerated considerably. They form what is called as “ion beam”. The ionizing chamber
must be kept under high vacuum. If any neutral molecules are there, the boron ion (B+)
may hit them and take up an electron and become neutral boron atom (B). We should
not dope the wafer with boron atom, but should use only the boron ion. We will see the
reason shortly.

Doping chamber:
In order to ensure that only boron ions are used for doping, just before the beam enters
the doping chamber, a magnetic field is used to change the direction of the ion beam.
The magnetic field will change the direction of only the ions. The neutral atoms will not
be affected by the magnetic field. Thus only the ions will fall on the wafer. The next
section is called wafer chamber. The beam size is approximately 1 cm2. The wafer size
can be 200 mm or 300 mm. So the beam has to be scanned over the entire wafer.
There are two ways of scanning: one is called electrical scanning and the other is
mechanical scanning. In the electrical scanning, using electrodes, the beam is moved
by applying suitable positive and negative voltages. In the mechanical scanning the
wafer is moved, while the beam is held stationary. Sometimes, a combination of beam
scan and rotation of wafer is also used to ensure uniformity.
When the boron ion is added to the wafer, the positive charge has to be neutralized.
This is done by supplying electrons from the back of the wafer. We can measure the
total current used for this and from this calculate the amount of boron deposited. By this
method, the exact quantity of boron added can be measured very precisely.

Why are the neutral atoms removed from the beam just before deposition? The total
quantity of impurities added is measured using the current used for neutralizing the
charge. If neutral atoms fall on the wafer, they will not need any electron. Thus the
quantity of doping will be calculated incorrectly. Hence the neutral atoms are removed
just before doping. The ions falling on the wafer will collide with the silicon atoms and
lose momentum. After many collisions, they will come to rest. Depending on the speed
of the incoming ions, the location where they stop is determined.
Slide 12

CHANNELING
1. Hold the wafer
at an angle (~ 8 degree)
==> increase transverse
straggle(called undercut)
BLOCK

Also causes
“shadow”

==> Too much angle is

also a problem

Shadow Undercut

As I discussed already , Misalign the beam by 7-10 degrees i.e. the inclined ion beam
should not be directed along a particular crystallographic axis. Otherwise ions moving at
a considerable distance w/o colliding the target atoms. This is called channeling.

But you cannot misalign the beam to a greater extent. Because it will increase the
lateral diffusion (transverse straggle) and causes shadow as shown in the figure. There
wont be any doping in the shadow region.
Slide 13

RAPID THERMAL ANNELAING

DIFFUSION VS ION IMPLANTATION

Rapid thermal annealing : After ion implantation, the Si crystal will have a lot of
defects and will have stress. The stress must be relieved and the defects reduced. This
process is called ‘repair’. If the wafer is heated and cooled, then these goals can be
achieved. However, the dopants will diffuse at high temperature and hence the profile
will change significantly.
If the wafer is heated quickly, the dopant diffusion is minimized, but the defect and
stress levels will reduce. This process is called Rapid Thermal Annealing (RTA). The
typical activation energy for repair (5 eV) is more than the typical activation energy for
diffusion (3 or 4 eV). Hence, high temperatures favor repair over diffusion.
The rapid increase in temperature is achieved by using lamps or lasers, but furnace can
be used for relatively low temperatures (<500 ˚C). Heating rates in the range of 100 to
1000 ˚C/s can be achieved. The high temperatures are measured using pyrometers. In
case of lamps, tungsten halogen bulbs or arc lamps are used. These operations are
conducted in high vacuum. Otherwise, the silicon wafer surface will get oxidized quickly.
Comparison of diffusion and ion implantation: When we compare diffusion versus
ion implantation, we should note the following: the deposition plus diffusion process
depends on the chemical nature of the species and the solubility. We can never implant
more than the solubility limit of the dopant by the diffusion process. In implantation, the
doping depends on the energy of the ion beam. We can dope more than the solubility
limit. Implantation is more expensive than diffusion. We can control the junction depth,
and profile much better than implantation; we can also see or ensure that the maximum
concentration of the dopant occurs inside in a specific location, whereas in diffusion, the
maximum concentration will always occur on the surface. In the diffusion, the side
diffusion is significant in ion implantation even though there is transverse struggle – the
movement to the side is relatively less.