Oxidation/Diffusion/Ion Implantation

Thermal Oxidation
Formation of SiO 2

Silicon Oxidation
• Silicon easily forms an oxide
– Native oxide present upon exposure to air at R.T. (1.5-2.0 nm)
• Thermal Deposition or Chemical Vapor Deposition (CVD)
• Thermal: Dry (O 2 ) or Wet (O 2 /H 2 O)
• Dry Oxidation described by Deal-Grove Model
Dry Oxidation: Si + O 2 → SiO 2
Wet Oxidation: Si + 2H 2 O → SiO 2 + 2H 2
*Notice Silicon is required for the reaction so once an oxide starts to
form, oxygen has to move towards the Si-SiO 2 interface to
combine with silicon – Silicon is consumed in this reaction (44%
of the thickness of the oxide is consumed). The interface has not
seen the atmosphere.

Oxide thickness and Rate Coefficients
t 2 + At = B(t +! )
ox ox
Approx. for
Thin Oxides
Approx. for
Thick Oxides
t ox
! B A
(t +! )
t 2 ! B(t +! )
ox
• B/A is the linear rate
coefficient
• B is the parabolic rate
coefficient
Note:
• Oxidation rate slows as the oxide thickens.
• τ is a factor that accounts for any oxide present when
another oxidation begins.
! = t o
2 + At o
B

B and B/A coefficients for wet/dry oxidation as a function of
HCl concentration for (111) and (100) surfaces
Chlorine cleans the gas ambient of impurities

Model is not good at predicting oxidation rate in
the initial oxidation regime
Oxidation rate is
slightly higher than
the model predicts

SiO 2 is amorphous but has some short range order
A Si atom is located at the center of a tetrahedron
Randomly oriented
network joined by
combination of bridging
(oxygen atoms shared in
tetrahedra) & nonbridging
oxygen atoms

Oxide Characterization
Primary metric of importance is t ox
Color Chart - colors due to interference between incident and
reflected light. Colors repeat so need to have some idea of t ox
Surface profilometry – mechanical scan with sharp stylus;
you do need a step in the film for reference or baseline
Ellipsometry – polarized light is used to illuminate wafer at
an angle to surface; beam is reflected from both oxide and silicon
surface, difference in polarization leads to calculated t ox ; need n f
Nanospec – interference technique that uses light at nearly
normal incidence; reflected light is measured as a function of λ and
t ox can be calculated.
Electrical techniques – Form MOS capacitor and obtain C-V
curves to find: breakdown voltage, mobile/trapped charge, t ox from
C ox (know Area (A) and ε ox ). Used to determine quality of oxides as
well as t ox .

Ellipsometer
Nanospec
Profilometer
C-V measurements

Shifting and Slope of C-V Curve
Fixed charge shifts the
curve while the
presence of Si/SiO 2
interface states affects
slope of C-V curve
Post oxidation Anneal (POA) in an inert (N 2 or Argon) will lower
fixed charge density at the Si-SiO 2 interface.

Oxidation Furnaces

Diffusion
Diffusion - the movement of a
chemical species from an area
of high concentration to an area
of lower concentration. The
controlled diffusion of dopants
into silicon is the foundation of
forming a p-n junction to
fabricate devices.

The purpose of diffusion is to introduce dopants
(substitutional point defects or intentional impurities)
into a silicon wafer in a controlled manner. A designer
will dictate a concentration level – lightly doped wafers
may be just a few ppm (10 17 atoms/cm 3 ).
Fick’s Diffusion Law – basic equation to describe a
movement of particles from a high concentration region.
J = !D "C(x,t)
"x
Leads to an equation where 2 boundary
conditions are needed to solve.
J is net flux of material (units are number per
unit time per unit area);
D is the diffusion coefficient,
C is the impurity concentration
(-) indicates movement in direction of
decreasing concentration

Solid Solubility
Maximum concentration of a material that can be dissolved in
another material.
Solid solubility of common silicon impurities: note that concentration increases to
the left (all rights reserved, reprinted with permission, © 1960 AT & T).

Impurity Profile

Analytic Solutions of Fick’s Law
Two Sets of Boundary Conditions
1) Predeposition diffusion: dopant source is fixed at the surface for all times > 0
2) Drive-in diffusion: an initial amount of dopant is introduced and then diffused
into the wafer

Predeposition and Drive-In
Predep – constant source
Drive-in – limited source
C(z, 0) = 0
C(z, 0) = 0
C(0,t) = C S
C(!,t) = 0
C(!,t) = 0
Q = constant
! z $
C(z,t) =
C(z,t) = C S
erfc#
&, t > 0
" 2 Dt %
C S is the fixed surface concentration, erfc is
tabulated;
Dt is the characteristic diffusion length.
The dose, Q, (atoms/cm 2 ) from predep is given
by:
Q T
(t) = 2 ! C(0,t) Dt
C S
= C(0,t) =
z ! 0
Q T
!Dt e"z2 4Dt ,t > 0
Q T
!Dt

Junction depth equations
When diffusing a dopant into a wafer of the opposite type, the
junction depth will be the distance from the wafer surface where
the surface concentration, C S , is equal to the background
concentration, C B
Predep – constant source
"
x j
= 2 Dterfc !1
$
#
C B
C S
%
'
&
Drive-in
x j
=
"
4Dt ln$
#
C B
Q T
!Dt
%
'
&

Calculation example
A wafer is heated to 1100°C and exposed to phosphorus. After 5
minutes, the dopant source is removed. The furnace temperature is
raised to 1200°C and diffusion is carried out for 2 hours.
Assume:
• Diffusion coefficients for P (determined from the Arrhenius
equation: temp dependence of the reaction rate)
D P = 1.43 × 10 -13 cm 2 /sec at 1100°C
D P = 1.17 × 10 -12 cm 2 /sec at 1200°C
• Wafer is p-type with a substrate concentration of 10 16 cm -3
Find:
a) The dose, Q
b) The junction depth, x j , after predep
c) The junction depth, x j , after drive-in

Analysis of Diffused Profiles
After diffusion, sheet resistance is measured in units of Ω/square by
either a 4 point probe or by the Van der Pauw method.

Diffusion Systems
Components:
• High T Furnaces, multi-zone
• Gas/liquid lines or Solid sources fit into quartz boat
• Quartz boats, slotted to hold wafer
• Mechanism for pushing/pulling boats
Slotted boats

Ion Implantation

Ion Implantation
An alternative to diffusion as a way to introduce dopants: ionized
impurity atoms accelerated through an electrostatic field strike the
surface of the wafer.
The dose is controlled by the ion current.
Typical ion energies 1-200 keV.
Advantages: Low temperature, tailor dopant profile
Disadvantages: Expensive process, low throughput, damage to the
lattice

Three components: Source, acceleration tube, and end station
• Starts with feed gas
containing implant
species: BF 3 , AsH 3 , PH 3 ;
gas is ionized
• Ion stream is accelerated
in high vacuum tube
(

Both horizontal and vertical
pairs of deflection plates are
used; beam is rastered back and
forth and up and down,
writing uniformly across the
wafer
Or beam is rastered in only one
direction; wafers are placed on
the perimeter of a spinning disc
for higher throughput
Dose (Q or ϕ) determined by
measurement of current:
Q = tI
qA
Where t is time, q is
electronic charge, I is
current, and A is wafer area
Units (ions/cm 2 )

Vertical Projected Range (R p ) and Straggle (ΔR p )
When an energetic ion enters a solid, it will begin to lose energy. The
distance that the ion travels in the semiconductor is its range, R. For a
uniform beam, the quantity of interest is not the total distance traveled, but the
average depth, the projected range, R p .
Model the range of depths:
N(x) =
!
2!"R p
e #(x#R p ) 2 /2"R p
2
Where R p is the projected
range, ΔR p is the standard
deviation (or straggle), ϕ is the
dose

Calculation example
A 30-keV implant of Boron is performed into n-type silicon. The
dose is 10 12 cm -2 .
Find:
a) The depth of the peak of the implanted profile (R p ).
b) The concentration at that depth.
c) The concentration at a depth of 3000 Å (0.3 µm).

Channeling can occur
when ion velocity is
parallel to a major crystal
orientation. Some ions
may travel considerable
distances with little
energy loss.
Ion implantation causes lattice damage that
can be repaired by high temperature
annealing to recrystallize; there is a threshold
dose above which the damage is complete (no
evidence of long range order, surface is
amorphous).
(A) Models of the diamond structure
along a major crystal axis 〈110〉 and
along a random direction. (reprinted
by permission, Academic Press, after
Mayer et al.).