Chapter: 5

Ion Implantation
 Advantages of ion implantation
• Very precise dose control (Faradays cup).
The ion implanter forms a simple electrical circuit. By monitoring the current in the circuit (or by a
monitoring circuit with Faraday cups), significant accuracy in the implanted dose can be maintained.
Assuming a current sensitivity of nA, and a minimum required implantation time of 10 seconds, it can
be shown that doses as low as 1011cm-2, can be measured.
On the contrary, in chemical source pre-deposits, dose values less than 51013/cm2 are not achievable.
• Dose not limited to solubility: High dose introduction is not limited to solid solubility limit
values.
• Uniformity: Excellent doping uniformity is achieved across the wafer (< 1% variation across
12” wafer) and from wafer to wafer.
• Less dopant lateral diffusion, good for small device
(short channel).
• Less contamination: Done in high vacuum, it is a
very clean process step.
• Profile control: Besides precise dose control, one
can also control the profile (peak depth and spread
range) better than diffusion (peak concentration
always near surface).
(top) Doping by diffusion and “drive-in”.
(bottom) Doping by ion implantation with or without “drive-in”.
 Advantages/disadvantage of ion implantation
Advantage:
• Low-temperature process (can use photoresist as mask)
• Wide selection of masking materials, e.g. photoresist, oxide, poly-Si, metal
• Very fast (6" wafer can take as little as 6 seconds for a moderate dose)
• Complex profiles can be achieved by multi-energy implants.

Disadvantage:
• Very expensive equipment ( $1M or more).
• At high dose values, throughput is less than diffusion (chemical source pre-
deposition on surface).
• Ions damage the semiconductor lattice. Not all the damage can be corrected by
annealing.
• Very shallow and very deep doping are difficult or impossible.
• Masking materials can be “knocked” into the wafer creating unwanted impurities,
or even destroying the quality of the interface.
Implantation equipment
 Schematic of an ion implanter

• Ion source: operates at a high voltage(25kV) and convert the electrically neutral dopant
atoms in the gas phase into plasma ions and undesired species. Some sources: Arsine,
Phosphine, Diborane, … Solid can be sputtered in special ion sources.
• Mass spectrometer: a magnet bend the ion beam through right angle, and select the
desired impurity ion and purge undesired species. Selected ion passes through an aperture.
• Accelerator: add energy to beam up to 5MeV. (contained, to shield possible x-ray).
• Scanning system: x and y axis deflection plates are used to scan the beam across the wafer
to produce uniform implantation of desired dose. The beam is bended to prevent the
neutral particles from hitting the target .
Schematic of an ion implanter
 Plasma ion source and ion extraction

Variable extraction voltage Positive ions are attracted to the exit side of
(typically 30KV ) the source chamber, which is biased at a large
negative potential with respect to the filament.

• Filament emits electrons, which are accelerated to gain enough energy.

• The electrons collide with the molecules or atoms, and ionize them.
• The ions are extracted, rough-focused, then travel toward the magnetic analyzer.
Nielsen Type gaseous sources
Plasma ion source and ion extraction
Example
Plasma ion source and ion extraction
Example
Ion Selection(Analyzing Magnet)
 Select implantation species
• Dopant gas containing desired species, BF3,
BCl3, B2H6, PH3, AsH3, AsF5.
• If using solid or liquid source, they need to
be heated to vaporize .
• Plasma provides positive ions, (B11)+, BF2+,
(P31)+, (P31)++.

Only one mass will have exactly the

correct radius of curvature to exit the
source through the slit.

BF3 gas spectrum

 Analyzing magnet
An analyzer magnet bends the ion beam through a right angle to select the desired species.
The equation of forces: -----(1) ---------(2)
The radius R: solving for ‘r’ in 1 and substituting v from 2 we have :

Mass to charge ratio of the selected ions: M/q = R2 B2 / (2 Vext)

The displacement D: (should be 2/R, not 1/2R)

Most commercial mass filters have radius of By adjusting the magnetic field, only selected
meter or less ions will enter the accelerating column.
 Ion acceleration

Final kinetic energy of the ion =

Q ( Vext + Vacc)
Example: Vext = 30KV, Vacc = 70KV
Energy of the ion = 100 keV

This shows 14 equal acceleration

plates.
For example, if the desired
acceleration is 70keV, each section
would contribute 5k volts.
 Neutral trap
The beam at this point is primarily composed
of ions. Some neutrals may have reappeared.
Commonly, the neutrals are ions that
combined with thermal electrons:

• There may be ions that collided with other ions in the beam, undergoing a charge exchange.
• Neutrals are highly undesirable since they will not be deflected in electrostatic end-station
scanning mechanisms.
• To avoid this problem, most systems are equipped with a bend, i.e. parallel plates of an
electrostatic deflection system.
• Neutrals are not deflected and so do not follow the beam, but instead strike a beam stop.
• The ions are sufficiently deflected by the plates to continue to travel down the tube.
 Beam scanning

Electrostatic scanning: low/medium

beam current implanters, I < 1mA.

This type of implanter is suitable for low dose implants. The focused ion beam is
The beam current is adjusted to result in t > 10 sec/wafer. scanned over the wafer in a
With scan frequencies in the 100 Hz range, good implant highly controlled manner, in
uniformity is achieved with reasonable throughput. order, to achieve uniform
doping. Either the wafer or
the beam could be stationary.
 Mechanical scanning
• Excellent wafer cooling needed.
• Substantial load/unload time.
• 15 - 25 wafers /disc.
• Excellent throughput for high dose implants.
• High current (20mA)
• Scan speed adjustment to ensure uniform dose.

Eaton HE3 High Energy Implanter, showing the

ion beam hitting the 300mm wafer end-station.

Multi-wafer, mechanical
scanning end-station

Rotating disk, 1000rpm (rotations per minute).

Stationary ion beam, beam diameter 5cm.
 Implant dose
• To control the dose the wafer is placed in the faradays cup in the end station.
• The cup is simply a cage that captures all of the charge that enters it. The ion current into
the wafer is measured directly by connecting an ammeter between the Faraday cup and
ground. The dose is just the time integral of the current divided by the wafer area.
 Practical implantation dosimetry (dose measurement)
Dose
Complete System
 Coulomb scattering

The incident atom will strike the target atom and due to columbic interaction, the incident and
target atom scatters conserving both the momentum and energy. The energy lost by the
incident atom depends on the scattering angle which in turn depend on scattering parameter b
Ion Implantation Energy Loss Mechanism
 Electron Stopping
Ions are embedded into the wafer and are scattered at random angles. The ions loose kinetic
energy, thus, slowing to a stop, by 2 mechanisms:
1. Non‐Local Electronic Stopping 2. Local Electronic Stopping

Drag force caused by charged ion in "sea"

of electrons (non‐local electronic stopping). Collisions with electrons around atoms
transfers momentum and results in local
electronic stopping.
Due to the large numbers and small individual effects involved, these discrete interactions can
be approximated by a continuum mechanism. That is, the effect of the electrons on the ion is
very much like a particle moving through a fluid (This interaction slows the ions by “viscous
friction” similar to a rock thrown into water. Note: many electrons are freed from the lattice
atoms creating an “ocean of electrons” that the ions must pass through). A viscosity due to the
electrons can be assigned to the crystal media. In that model, there is a viscous drag force
assumed to be proportional to the velocity for energies of interest to typical microelectronic
applications:
To a first order

Thus Energy loss per unit length or the drag force

 Nuclear Stopping
• An incident ion scatters off the core charge on an atomic nucleus,
modeled to first order by a screened Coulomb scattering potential.

• Function of b:The energy transfer during any collision will be a

sensitive function of the impact parameter. The lower the impact
parameter, the greater the energy loss.

• Ratio of masses: The average energy loss will also be

a function of the ratio of the mass of the ion to that
of the target atom. The smaller this ratio, the larger
the average energy loss per collision will be

• Energy: Finally, the average energy loss will be a

function of the energy itself. The solid atom is
chemically bonded. It is sitting in an approximately
parabolic potential minimum. At ion low energies the
average collision does not transfer enough energy to
break this bond. The result is an almost elastic collision. The ion may change direction, but it

will not lose much energy. Thus, Sn is expected to increase with ion energy at low energy.
 Nuclear Stopping

As ion enters wafer with high E, the energy

loss rate is low. Rate increases to max as it
slows down, and then decreases to 0
because at much lower energies the
collisions are mostly elastic i.e the target
atom continues to sit in parabolic well and
only scatters the ion elastically.

The maximum value of Sn can be approximated by the relation

where
 Total Energy Loss

The total energy loss during an ion trajectory is given by the

sum of nuclear and electronic losses (these can be
treated independently).

• Electronic collisions dominates at higher energy

• Nuclear collisions dominate at lower energy
 Nuclear and electronic components of S(E) for
several common silicon dopants as a function of energy

The critical energy Ec when the

nuclear and electronic stopping
are equal is
B: ≈ 10keV
P: ≈ 150keV
As : > 500keV

Thus, at high energies, electronic stopping

dominates; at low energy, nuclear stopping
dominates.
 Dopant (impurity) concentration profile
• Gaussian distribution for first order
The impurity is shown implanted completely
approximation.
below the wafer surface (x=0).
• Rp= projected range, is a function of ion
energy and mass, and atomic number
of impurity as well as target material.
• Rp = straggle = standard deviation.
• Np = peak concentration at x=Rp.
• Dose Q=N(x)dx=(2)1/2Np Rp.
 Projected Range and straggle

𝑆𝑖𝑛𝑐𝑒 𝑆=
𝑑𝐸 𝑑𝐸
𝑑𝑥
=
𝑑𝑥( ) 𝑛𝑢𝑐𝑙𝑒𝑎𝑟
+ ( )
𝑑𝐸
𝑑𝑥 𝑒𝑙𝑒𝑐𝑡𝑟𝑜𝑛𝑖𝑐
=𝑆𝑛 + 𝑆𝑒

𝑅𝑃 0 0
𝑑𝐸 𝑑𝐸
𝑅 𝑝=∫ 𝑑𝑥=∫ =∫
0 𝐸0
𝑑𝐸 𝐸
𝑆 𝑛 +𝑆 𝑒
0

𝑑𝑥

To avoid these complications, the projected

range and its standard deviation are often
obtained from LSS (Lindhard, Scharff, and
SchiOtt) tables
 Projected Range and straggle

Projected range (solid lines and left axis) and standard deviation (dashed lines and right
axis) for (A) n-type, (B) p-type,
 Actual Profiles are not Gaussian
This approximation to the profile is good for lower energies, but is less correct of higher
energies where the profile is “skewed”

1.) Negative skewness (shown above) means the distribution is shifted toward the surface,
possibly due to back scattering: Common for light elements at high energy
2.) Positive skewness means the distribution is shifted away from the surface, possibly due to
channeling (will discuss shortly): Common for heavy elements at low energy
Asymmetry is usually expressed in terms of the skewness moment

• Light ions such as boron, are more effectively scattered backwards

more ions to come to rest on the surface-side of Rp
• Heavy Ions, such as Arsenic, scatter more in forward direction
the amount of dopant on the deeper side will be higher .
Moments description

𝑚1
Range: 𝑅 𝑝=
𝑄𝑇

Std. Dev:
√
𝜎 𝑃 =∆ 𝑅 𝑃 =
𝑚2
𝑄𝑇
𝑚3
Skewness:𝛾 = 3
𝑄𝑇 ∆ 𝑅 𝑃

𝑚4
Kurtosis: 𝛽= 4
𝑄𝑇 ∆ 𝑅𝑃
 Channeling and tailing
• The above theory is fine for amorphous target
• If ions hit along major crystal orientation --> channeling results
• ion moves further through lattice
• Range depends on crystal axis, beam to crystal alignment, dose (damage along channel)

Avoid channeling -> misalign 7-10º

Tail due to:

(1) Some ions in channels due to
statistical distribution and diffusion
after implantation
(2) Interstitial diffusion until find
vacancy trap site
 Critical angle for channeling
Channeling can occur when the ion velocity is parallel to a major crystal orientation. In this
situation, some ions may travel considerable distances with little energy loss (Figure), since
nuclear stopping is not very effective and the electron density in a channel is low. Once in a
channel, the ion will continue in that direction, making many glancing internal collisions that
are nearly elastic until it comes to rest or finally de-channels. The latter may be the result of a
crystal defect or impurity. Channeling is characterized by a critical angle
Critical Angle: defined as the maximum angle between the ion
and the channel for a glancing collision to occur.
where E is the incident energy in kilo-electron-volts and d is the atomic spacing along the ion
direction in angstroms.
 Positive skewness due to Channeling
The channeling direction need not be close to the initial
ion velocity. A scattering event inside the target can
redirect the incident ion along a crystallographic
orientation, but the probability of this occurring is small

Along certain crystal directions, “tubular holes”

exist that allow ions to travel long distances
without undergoing a high angle collision. The
ions “skip” down these tubes undergoing only
glancing collisions. This extends the effective
range resulting in a positive skewness.
 Preventing channeling

Channeling does not occur if there is significant implant damage that turns the implanted
layer into an amorphous one. Heavy ions like P31 and As75 at large doses do not show
channeling.
Light ions and/or low dose implants are prone to channeling. In such instances,
channeling can be prevented by:
• Tilting and twisting the wafer to close crystal openness as seen by ion beam
• Implanting through a thin amorphous layer (e.g oxide)
• Implanting heavy, but electrically inactive species like Si or Ar prior to actual dopant implant.
The pre-implant turns the wafer surface into an amorphous layer
Preventing channeling
 How thick mask is needed?
In order to act as an efficient mask, the thickness of Mask Si
the mask should be large enough that the tail of
the implant profile in the silicon should not
significantly alter the doping concentration (CB).


C * ( xm ) C P*  exp 
xm  R *
P
  C
2

2 B
 2R *P 
(“*” means for value in the mask layer)

 C P* 
Mask thickness xm xm RP*  RP* 2 ln   RP*  mRP*
 CB 
2
Q  
 *  
1  x  RP  Q  xm  RP* 
   
Dose penetrating the mask: QP  2 R * exp  2  R *  dx  2 erfc  2 R * 
  p    p 
p xm
 
Rule of thumb: good masking thickness
Xm = Rp + 4.3Rp, C(x=Xm)/C(x=Rp)  10-4.
 [ ] 𝑑𝑥= 𝑄 𝑒𝑟𝑓𝑐
[√ ]
∗ 2
1 𝑥− 𝑅 𝑝
𝑄
∞ − 𝑥𝑚 − 𝑅 𝑝∗
∗∫
2 ∆ 𝑅𝑝 ∗
𝑇𝑜 𝑠h𝑜𝑤 𝑄 𝑃 = 𝑒
√2 𝜋 ∆ 𝑅𝑝 𝑥 𝑚
2 2 ∆ 𝑅𝑝 ∗
∞
2
∫
2
−𝑧
𝑟𝑒𝑐𝑎𝑙𝑙 ( 𝑑𝑒𝑓 ) ; erfc 𝑢= 𝑒 𝑑𝑧 … … … … (1)
√𝜋 𝑢
[ ] 𝑑𝑥
∗ 2
∞ 1 𝑥 − 𝑅𝑝
𝑄 −

∗∫
2 ∗
∆ 𝑅𝑝
𝑄𝑃 = 𝑒
√2 𝜋 ∆ 𝑅 𝑝 𝑥 𝑚

[ ]
∗
1 𝑥−𝑅 𝑝
𝑡h𝑒𝑛𝑑𝑥=√ 2∆ 𝑅𝑝 𝑑𝑧
𝐿𝑒𝑡 𝑧 = ∗
√ 2 ∆ 𝑅𝑝∗
∗
𝑥 −𝑅𝑝
𝑧= 𝑚 𝑤h𝑒𝑛 𝑥 → ∞ ; 𝑡h𝑒𝑛𝑧 → ∞
√2 ∆ 𝑅𝑝∗

[ ]
∞ ∞
𝑄 𝑄 2
∫ ∫
2 2

𝑇h𝑒𝑟𝑒𝑓𝑜𝑟𝑒 , 𝑄 𝑃 = 𝑒 −𝑧
√ 2 ∆ 𝑅𝑝 𝑑 𝑧 = ∗ −𝑧
𝑒 𝑑𝑧
√ 2 𝜋 ∆ 𝑅 𝑝∗ 2 √𝜋
[ ] [ ]
∗ ∗
1 𝑥 − 𝑅𝑝 1 𝑥𝑚− 𝑅 𝑝
√2 ∆ 𝑅𝑝
∗
√2 ∆ 𝑅𝑝
∗

𝐶𝑜𝑚𝑝𝑎𝑟𝑖𝑛𝑔 𝑤𝑖𝑡h 𝑒𝑞𝑢1

[ ]
∗
𝑄 𝑥𝑚 − 𝑅 𝑝
𝑄 𝑃 = 𝑒𝑟𝑓𝑐
2 √2 ∆ 𝑅𝑝 ∗
Preventing channeling
Preventing channeling
 Lateral scattering

 The implanted ion also scatter laterally

1 y  
2
C ( x) around the impact point, which can also
C ( x, y )  exp     
2 R  2  R   be approximated by Gaussian distribution
with transverse straggle ΔRT.
 A 2-D formulation of implantation profile
Monte Carlo simulation of many ions into Si
Implantation Damage
 Damage process
• Main Component of Damage: One component of energy transfer when a high energy
ion enters a wafer is collision with lattice nuclei.
• Transfer of energy from the incident: Implanted ions transfer energy to lattice atoms
which can either eject them from surface or break them free.
• Chain reaction: Freed atoms collide with other lattice atoms to create more free
atoms
– Damage continues until all freed atoms stop
• One energetic ion can cause thousands of displacements of lattice atoms
 Damage repair and dopant activation
• Damage + Inactive dopants: The implantation process produces considerable substrate
damage that must be repaired during subsequent processing. Furthermore, if the
implanted species is intended to act as a dopant, it must occupy lattice sites.
• Activation: This process of moving a large fraction of the implanted impurities onto lattice
sites is known as impurity activation.
• Annealing: Both damage repair and implant activation are normally done by heating the
wafer (annealing) after implant. Often the anneal accomplishes both tasks simultaneously.
For that reason, these two processes will be treated simultaneously
 Thermal Annealing

Lattice atoms Dopant atoms

 Damage and the critical dose
• Critical Dose: There exists a threshold dose above which the damage is complete. That is,
after the implant no evidence of long-range order exists, and the surface of the substrate is
rendered amorphous
• Dependence: The critical dose depends on implant energy, implant species, target material,
and substrate temperature during implantation
• Figure shows the critical dose for several impurities in silicon as functions of the wafer
temperature. At high temperatures, the substrate self-anneals, and the threshold dose
becomes very large. Threshold doses for light ions are also large since a greater fraction of
the energy loss is electronic.

The maximum dose required to create

amorphous material is called the critical dose.
Because the material can anneal out damage
during the implant, the critical dose increases
with increasing temperature.
 Primary and secondary defects
• Primary defects: As an ion passes through the crystal, point defects consisting of interstitials
and vacancies are created by direct interaction or through collisions with recoiled target
atoms. The defects created by the implantation process are called primary defects. For
example it is estimated that the interstitial silicon concentration is 7 X1016 cm-2, for Si wafer
implanted with 2 X1015 cm-2 boron at 200 keV, roughly 35 times the implant dose.
• Secondary Defects: Secondary defects occur when an implanted wafer is annealed. As
previously mentioned, point defects have a high energy in the crystal. This energy can be
reduced by recombination or by agglomeration into extended defects. Typically, these defects
take the form of small point defect clusters such as divacancies or condense into higher
dimensional defects like dislocation loops.
• Heavier atoms have greater critical interstitial density: For the secondary defect formation
the Critical interstitial concentration for Boron is> 2X 1016 cm-2.For P it is > 5X 1016 cm -2 and
critical interstitial concentration for a heavy ion like Sb exceeds 1017 cm-2. Heavy ions,
therefore, tend to amorphize the substrate before secondary damage is generated
• Extended and isolated secondary defects. It has been suggested that implant light atoms like
B form isolated defects, while heavier ions form more extended defects
 Annealing the secondary defects
• Annealing processes that minimize secondary defects are of considerable technological
interest. In these applications, high temperatures are required to ensure that all of the
dopant is activated and few residual extended defects remain
• Isochronal annealing curves display the active carrier concentration normalized to the dose
as a function of the annealing temperature for a fixed annealing time. Typical times are 30 or
60 min. Anneals are done in nitrogen.
Result of an isochronal anneal experiment for boron

Region I: At low temperature the carrier concentration is

dominated by point defects. As the anneal temperature is raised,
point defect repair begins, with some vacancies capturing
nearby interstitials. This reduces the net trap concentration in
the substrate, raising the free carrier concentration
Region II: In the 500 to 600C range, the diffusivity of the defects
is raised sufficiently to cause agglomeration and the formation
of extended defects. This is particularly true at higher boron
fluences, where the damage is large.
Region III: Finally, at high temperature, these extended defects
are annealed out and the active carrier concentration
approaches the implant dose. This requires temperatures of
850C to over 1000C, with high fluence implants requiring the
highest anneal temperatures.
Fraction of implanted boron activated in silicon for several
isochronal anneals
 Solid phase Epitaxy
SPE
SPE: Solid-phase epitaxy (SPE) is a transition between the amorphous and crystalline phases
of a material. It is usually done by first depositing a film of amorphous material on a
crystalline substrate. The substrate is then heated to crystallize the film. The single crystal
substrate serves as a template for crystal growth.
SPE Annealing: The annealing step used to recrystallize or heal silicon layers amorphized
during ion implantation is also considered one type of Solid Phase Epitaxy. The Impurity
segregation and redistribution at the growing crystal-amorphous layer interface during this
process is used to incorporate low-solubility dopants in metals and Silico
 SPE and annealing high energy Implant
In theory, the crystal reforms, using the underlying undamaged substrate as a template. This
causes most of the impurities to be incorporated into the growing lattice on nearly equal
footing with the displaced substrate material
Buried amorphous layer due to high energy implants: one must first recognize that
amorphous layers need not extend to the surface. Obviously, the degree of damage near the
surface is much less than that further into the substrate. Buried amorphous layers are
therefore possible by using high energy implants
SPE regrowth begin both sides sandwiching the amorphous region: SPE regrowth fronts will
begin at both sides of the amorphous region. The plane where these fronts meet contains
defects that can degrade device performance
 Annealing temperature
Anneal temperatures: This regrowth can be done at
temperatures as low as 600C, since the regrowth
velocity of silicon at this temperature is greater than
300 A/min for 100 and about one order of
magnitude less for 111
Thus, a 30-min anneal at 600C will regrow almost 1
m of material, which is well in excess of the
amorphization depth of any reasonable implant.
The activation of impurities occurs at much lower
temperatures because of SPE. Figure 5.17 shows
isochronal curves for the activation of phosphorus
in single crystal and amorphous materials. A large
percentage of the implant is activated even at 550C.
 High temperature anneal to repair damage
Amorphous layers include not only simple 1-D defects like dislocations, but also 2-D and 3-D
defects like twins and stacking faults. Defects will be formed when these disparate growth
fronts meet. To reduce the concentration of the defects to an acceptable level, high
temperature anneals are usually required (of order 1000C); however, even these high
temperature anneals may not be sufficient to remove all of the damage. Figure 5.18 shows the
annealing characteristics of high-dose arsenic implants .
Even after 1000C anneals, residual damage remains. The
damage is concentrated in the regions near the edge of the
implant. High temperature anneals also allow the activation
of impurities in the implant tail regions, outside the
amorphous layer. To avoid excessive diffusion during the
high temperature anneal, the point defect concentration is
first lowered by a low temperature SPE step.
 Diffusion during subsequent anneals
During high temperature steps after implant (most commonly an activation anneal),
the implanted impurities will begin to diffuse, broadening the implantation profile.
For implantations far away from the surface and for reasonable short characteristic
diffusion lengths, the new profile can be approximated by: