Doping: Depositing impurities into Si in a controlled manner
Download
Report
Transcript Doping: Depositing impurities into Si in a controlled manner
Doping: Depositing
impurities into Si in
a controlled manner
Overview
Diffusion vs Implantation
Mechanism,Models
Steps
Equipment
Goal:
Controlled Junction Depth
Controlled dopant concentration and profile
P+
P+
Source
N “well”
Wafer (Substrate): P Type
Drain
Preferred
location of
maximum
concentration
need not be the
surface
Ion Implantation
Diffusion & Ion
Implanatation
Bombardment of
ions
SOURCE
Electric
Field
OXIDE
Ions
Wafer (Substrate)`
BLOCK
Junction is where
N=P
Can also be used
when doping N in N
Diffusion & Ion
Implantation
Diffusion
Solid-in-solid
high temperatures (1000 C)
Distances covered are in um or nm
Diffusion
OXIDE
Wafer (Substrate)`
BLOCK
Mechanism , Models
Substitutional (10-12 cm2/s)
Interstitial replacement (10-6 cm2/s)
Interstitial movement
Substitutional preferred (better control)
Au, Cu diffuse by interstitial mechanism
B, P etc by substitutional mechanism
Two ideal cases
Constant source, limited source
Using Fick’s First & second law
J = Flux
D - Diffusivity of A in B
N- Concentration
x - distance
J D
N
x
N
2 N
D 2
t
x
Models
Constant Source
Concentration at x=0 is
N (0, t ) NNo o
x
N (, t ) 0
N ( x, t ) N o erfc(
2 Dt
Limited source
Dose Q = constant
Approx by Delta Fn
)
N ( x, 0) 0
Complementary Error Function
Q N ( x, t )dx 2 N 0
0
0
x 0,t
N ( x, t ) dx Q
0
N (, t ) 0
N ( x, 0) 0
Total Dose Q
N
x
Dt
Q
N ( x, t )
e
Dt
x2
4 Dt
Models
Constant Source
Concentration at x=0 is
No
Important Parameter : Dt
Impurity
Concentration
N0
species, temp and time
3
2
1
Distance from Surface
Models
Limited Source
Dose Q
Important Parameter : Dt
Impurity
Concentration
N0
Area under the curve is
constant
1
3
2
Distance from Surface
If you normalize, erfc
drops faster than Gaussian
Diffusivity
E
kT
D D0 e
Diffusivity
Follows Arrhenius behavior
Wafer goes through heating cycles many times in the
process
Effective Diffusivity * time = sum (Diffusivity * time)
Concept of thermal budget
Dt total Diti
D
i
1000
T
Diffusion
Max absorption (at a given temp)
Usually quite high
Good for emitter and collector, but not for base
Not all dopant can contribute to electron/hole near
solubility limit
Solubility limit in the range of 10 20/cm3 at 1000o C
Diffusion into silicon
Faster on grain boundaries
10 times in poly silicon
Diffusivity in SiO2 usually very low (Segregation
occurs)
Junction Formation
N
Carrier
Conc
Impurity
Conc
Jn
P
Distance
from surface
Diffusion: Drive In: Dopant re
distribution
Deposited dopant must be pushed into Si
Re-distribution of dopant
Oxidation of exposed Si to protect
OXIDATION
Dopant Diffusion
*Dopant profile changes due to diffusion
* Also due to preference for Oxide/Silicon: N-type piles
up in Si, P-type depletes in Si
Diffusion: Steps
OXIDE
Dep
Diffusion
BLOCK
1.Pre Clean
To remove particles
Thin oxide grows
2.HF Etch
To remove oxide
Not too much!
3.Deposit (pre dep)
Deposit enough to be higher
than the solubility limit
4.Drive In
High temp to enable diffusion
inside Si
Also forms SiO2 (with high
5.Deglaze (HF Etch)
dopant concentration)
Oxide may act as dopant source in future
2-STEP diffusion (usual)
steps
Removing highly doped oxide may be
problem (for dry etch)
Diffusion: Dep:
schematic
Wafers are Horizontal
Gas
Flow
Better Uniformity
Less wafers per batch
Vertical
Poor Uniformity
More wafers per batch (or
can have smaller chamber)
Gas
Flow
Dummy wafers placed in the beginning & end
Doping: Gas phase
Dopant can be in Gas/Liquid/Solid state, but is typically carried
using N2 in gaseous form
Chamber
Carrier Gas (N2) +
Source
Reaction gas
*Carrier gas may be
bubbled through liquid
source
*Carrier gas may pass
over heated solid source
* inert gas can provide
volume to maintain laminar
flow
Doping: Gas phase
Reaction/Diffusion Limited
Phosphorus oxy chloride
4 POCl3 3 O2 2P2O5 6 Cl2
Phosphine
2 PH 3 4 O2 P2O5 3 H 2O
2 P2O5 5 Si 4 P 5 SiO2
Arsenic Oxide
2 As2O3 3 Si 3SiO2 4 As
Diborane
B2 H 6 3 O2 B2O3 3 H 2O
300o C
B2 H 6 6 CO2
B2O3 3 H 2O 6CO
Boron Tribromide
4 BBr3 3O2 2 B2O3 6 Br2 2 B O 3 Si 4 B 3 SiO
2 3
2
Solid phase
Solid Source
Slugs between wafers
Lower through put
Cleaning is issue (slugs can break)
Safer to handle(no toxic vapor at room temp)
Spin coating (with solvents)
Similar to photo resist coating
Cost of extra spin/bake steps
thickness variations
Doping: Solid phase
Phosphorous pentoxide
2 P2O5 5 Si 4 P 5 SiO2
Arsenic Oxide 2 As O 3 Si 3SiO 4 As
2 3
2
Antimony Tri Oxide
2 Sb2O3 3Si 3SiO2 4 Sb
Boron Trioxide
2 B2O3 3 Si 4 B 3 SiO2
Tri Methyl Borate (TMB)
2(CH 3O )3 B 9O2 B2O3 6 CO2 9 H 2O
900o C
Issues
Side diffusion
Increases with temperature/time
Limits the space between devices
Maximum dopant concentration is near surface
==> majority of current near surface
(Surface tends to have max defects)
==> less control
Dislocation generation (thermal drive in)
Surface contamination (dep)
Low dopant concentration and thin junction (small junction
depth) are difficult
At 0.18 um , junction depth is ~ 40 nm
At 0.09 um, junction depth may be 20 nm
Issues: Side diffusion
Side diffusion (Lateral Diffusion)
BLOCK
Wafer (Substrate)`
Diffusion
OXIDE
BLOCK
Example of Real systems :
*Hitachi-Vertron V
*1m x 3.5m x 3.3m
*200 mm wafer
*150 wafers at a time
* higher thermal budget,
* good control, uniformity
* high throughput
*Hitachi-Zestone VII
*2m x 3m x 3m
*300 mm wafer
*one wafer at a time
* lower thermal budget,
* better control, uniformity
* low throughput
Example of Real systems :
Protemp
Gettering
To remove unwanted impurities
Try to get them to the back of
wafer
Defects
Ar implant
Dep SiN/SiO2 (stress)
Oxygen during crystal growth
(intrinsic)
High Conc P on back of wafer
Measurement
Sheet Resistance (average)
Four point probe, VDP (Van der Pauw)
Bevel
Interference
Dye
SIMS
Diffusion: Summary
Diffusion
Temp, Time, Thermal budget
Doping (more important for older nodes)
Relevant for all nodes
2 step (constant source, limited source)
Solid/Liq/Gas
Ion Implantation
“Somewhat similar” to Sputtering
Dopant goes inside the silicon
sputtering deposits on the surface
Used for controlled doping
concentration
profile (depth)
Equipment
Mechanism
Issues
Summary
Equipment
Neutral
Beam Trap
and Beam
Gate
900
Analyzing
Magnet
Focus
Acceleration
Tube
Beam Trap
and Gate
Plate
Y-Axis
Scanner
wafer in
wafer
Process
chamber
Ion
Source
© Peter van Zant
1. Ion Source
Gas or solid source (no liquid source)
Solid heated to obtain vapor (P2O5)
effectively gas source
Mass flow meters (to control the flow better)
Gas usually Fluorine based
AsF5 , BF3 , SbF3 , PF3 , PF5
Ionization chamber
low pressure (milli/ micro torr) to ionize and
minimize contamination
heated filament (thermionic emission)
positively charged ions created
2. Analyzing
Selection, analyzing, mass analyzing, ion separation
Similar to Mass Spectroscope
Usually the second stage (before acceleration)
Magnetic field to control the path
Charge to Mass Ratio
Some of the species from BF3 source B , BF , BF2
Selection of B+
B
BF
BF2
3. Acceleration
Acceleration needed for implantation
Positive ions accelerated with ring anodes
Energy range: 5 keV for low, 2 MeV for high
High energy ==> high
throughput
few seconds per wafer
Beam Current
Medium current : 1 mA
High current: 10 mA
Current ~ Dose
Beam Focus (magnetic/electric)
SOI
High
Current
Oxygen
100 mA
10 mA
Low Energy
1 mA
High
Current
Low
Current
keV
High Energy
MeV
4. Scanning
Beam size ~ 1 sqr cm
Wafer size 200 mm or 300 mm
Issues:
neutral atoms need to be removed because...
dose calculated by current integrator
Electrical (beam) scanning & Mechanical (wafer) scanning
Beam Scan:(medium current)
beam moves outside the wafer for turn
controlling XY plates may be destroyed by discharge
Rotate wafer for uniformity
Wafer scan: (high current)
Beam shuttering: (electrical/mechanical) turn beam off
when not on wafer
5. Target chamber
End chamber
low particle, high vacuum
Wafer held on
clamp (more particles) OR ESC (less particles)
Anti-static devices on the chamber
Integrate the current to measure dose
For 2+ ions, divide by 2 and so on...
Wafer charging:
minimize by connecting wafer to ground (with a charge
counter)
dielectrics may get damaged
use flood gun to provide electron (and count it in
measurement)
Mechanism
Electrons attract the +vely charged ions
Nuclei repel the +vely charged ions
Inelastic collision:
Electron (ionization)
Nuclear (nuclear reactions)
Elastic collision
Electron
Nuclear (atom substitution)
At low energy Nuclear collisions predominant
At high energy electronic collisions predominant
Variation in ‘stopping cross section’
Gaussian profile expected (projected range Rp)
Implantation
Mask with Photoresist or oxide
resist for medium and low energy, moderate dose
high energy/high dose: increase in temp
Resist re-flow
Cross link (for organics)
less soluble (stripping an issue)
Faraday Cage
Retain secondary electron from wafer
Otherwise, wafer under dosed
-Ve Bias
e-
Issue: Transverse Straggle
implant
OXIDE
Gaussian
BLOCK
Transverse Straggle
(Diffraction)
Even in implantation, dopants present in lateral direction
Channeling
Some ions
will move
through
“channels”
without
experiencing
nuclear
or electron
collision for
a “long” time
==> No Gaussian Profile
Channeling
1. Hold the wafer
at an angle (~ 8 degree)
BLOCK
==> increase transverse
straggle(called undercut)
Also causes
“shadow”
==> Too much angle is
also a problem
Shadow
Undercut
Channeling
2. Dep amorphous
material on the top
implant
OXIDE
It has to be very thin
and not stop ions
BLOCK
3. Damage top
of wafer and make it
amorphous (eg high
energy silicon implant)
Channeling
4. Increase temperature
==> reduce channel cross section
Channeling critical angle ~ (Z/E) 1/2
==> Low energy implants more likely to channel
TED
Transient Enhanced Diffusion
Damage during implantation
==> point defects (vacancies)
interstitial silicon atoms
reduced during anneal
Channel dopant diffuse to surface
==> VT modification
©Solid State Technology
RTA
Anneal to heal the damage
Diffusion during anneal an issue
High temp repair is faster than anneal
Repair energy barrier 5 eV, diffusion barrier 3 or 4 eV
1. Adiabatic (laser, heats surface , < micro sec)
profile control difficult (not used)
2. Thermal flux ( micro to 1 sec)
laser, ebeam, flash lamp
surface+bulk heating
rapid cooling ==> point defects
3. Iso thermal (W-Halogen lamp)
30 sec (1100 C)
Diffusion vs Ion Implantation
Dep+Diffusion: depends on chemical nature and solubility
Implantation: on energy of ion beam
Expensive
Better Control of junction depth, dose, profile
Less ‘transverse straggle’