EE5036: VLSI Fabrication Principles

Dopant Diffusion
 Dopant diffusion
• Conductivity of semiconductors can be changed by
orders of magnitude via doping
• Formation of junctions of differently doped regions
key to operation of many semiconductor devices
• Requirement of shallow junctions with higher
doping as MOSFET dimensions continue to shrink
• Solid/gas phase diffusion & ion implantation
processes
• Electrical activation of the dopants important
• Diffusion is migration of dopants from regions of
higher concentration towards regions of lower
concentrations
Silicon VLSI Technology by Plummer
• Two step process: predeposition & drive-in
• Predeposition introduces a controlled no. of
impurity atoms
• Drive-in thermally diffuses the dopants to the
desired junction depth

Silicon VLSI Technology by Plummer.

• Alloy junction/mesa junction methods
• Solid phase diffusion from layers deposited on the
wafer surface
• High temperature gas phase diffusion in a furnace
(B2H6, PH3, AsH3)
• Issues like safety, reproducibility led to ion implantation

Silicon VLSI Technology by Plummer.

 Solid Solubility
• The maximum concentration that can be dissolved
under equilibrium conditions without forming a
separate phase
• Retrograde nature, maximum concentration below the
melting point

Silicon VLSI Technology by Plummer

• Electrically active concentration maybe much less than
solid solubility
• Formation of neutral clusters with point defects, no
contribution of free electrons/holes

• As atoms, threefold coordinated with Si lattice while

retaining two electrons in a dangling bond
Silicon VLSI Technology by Plummer.
 Macroscopic Diffusion
• Fick’s first law: Diffusion flux proportional to the
concentration gradient
C
F = −D
x

• Fick’s second law: Change in concentration in a

volume determined by the fluxes into/out of the
volume
C Fin − Fout F  2C
= =− =D 2
t x x x

Silicon VLSI Technology by Plummer

 Solution for steady state
• Simplest solution under steady state
C  2C
= D 2 =0
t x
C = a + bx
• Diffusion of oxidant in SiO2 during oxidation of Si
•

C  C0 − CI 
F2 = − D = D 
x  x0 

Silicon VLSI Technology by Plummer

 Fixed limited dose in the middle
• Spike/delta function of dopant in the middle of the
lightly doped region
C → 0 as t → 0 for x  0
C →  as t → 0 for x = 0


 C( x,t )dx = Q
−
 x2   x2 
−  − 
Q
• The solution is C( x,t ) = = C (0,t ) e
 4 Dt   4 Dt 
e
2  Dt

• Peak concentration goes as 1 t

• Surface concentration falls by 1 e at x = 2 Dt , also
called as the diffusion length
Silicon VLSI Technology by Plummer
• Time evolution of gaussian profiles

• Solution remains gaussian when more time is added

Silicon VLSI Technology by Plummer.

 Fixed limited dose at the surface
• Dopant source Q introduced at the surface

• The solution is
 x2   x2 
−  − 
Q
C( x,t ) = = C (0,t ) e
 4 Dt   4 Dt 
e
 Dt
with Q
C ( 0,t ) =
 Dt Silicon VLSI Technology by Plummer
 Infinite source
• Diffusion from an infinite source

C = 0 as t = 0 for x  0
C = C as t = 0 for x  0

• Infinite source made up of small slices each of width

Δx diffusing as a gaussian profile
 ( )   ( x −  )2 
2
C n x − x C
0
C ( x,t ) = 
2  Dt i =1
xi exp 

−
4Dt
i
=
 2  Dt − exp  − 4Dt d

   

x −
=
C
  ( )
exp − 2 d with  =
2 Dt
x 2 Dt Silicon VLSI Technology by Plummer
 z
• Error function solutions erf ( z )=
2
  exp (
− 2
)
d
0
C  x 
C ( x,t )= 1 − erf  
2  2 Dt 

• Complimentary error function: erfc ( x )=1-erf ( x )

C  x 
C ( x,t )=  erfc  
2  2 Dt 

Time evolution of erfc profiles

Silicon VLSI Technology by Plummer.

 Constant Surface Concentration
• Symmetry C ( x,t )=CS erfc  x 

  2 Dt 

• Introduced dose

  x  2CS
Q= CS erfc  dx = Dt
0   2 Dt  
Silicon VLSI Technology by Plummer
 Intrinsic dopant diffusion coefficients
• Diffusion coefficient increases exponentially with
temperature D=D0 exp  − EA 
 kT 

• ni large at high processing

temperatures so
‘intrinsic’ applies to many
situations
• Slow diffusers (As, Sb) &
fast diffusers (P, B)

Silicon VLSI Technology by Plummer

 Successive diffusion steps
• Multiple diffusion steps
• Constant temperature: ( Dt )eff = D1 ( t1 + t2 + ...) = D1t1 + D1t2 + ...
• Dt product the thermal budget
• Different temperatures: ( Dt )eff = D1t1 + D2t2 + .......
• Highest temperature step dominates thermal budget
• Same thermal budget: D1t1 = D2t2
• BJT, emitter by predep & base by implant + drive-in

Silicon VLSI Technology by Plummer

 Design & evaluation of diffused layers
• Sheet resistance
𝐿 𝐿
𝑅=𝜌 =𝜌
𝐴 𝑊𝑡ℎ

𝜌 𝐿 𝐿
𝑅= = 𝜌𝑠
𝑡ℎ 𝑊 𝑊

• ρs actual unit Ω, typically used is Ω/square or Ω/□

• Useful for characterizing thin films & doping processes
• Conductivity  = qp p
• Non-uniform doping
1 1
s = =
 xj xj

q   p ( x ) − CB    p ( x )  dx
0 Silicon VLSI Technology by Plummer
• Numerical integration of this equation, Irvine curves

Gaussian profile (p type)

Silicon VLSI Technology by Plummer

berkeley.edu
• Design of B diffusion process with ρs= 300 Ω/□, xj=3μm
& CB=1015cm-3
1
= 3.7 ( cm )
−1
• Average conductivity  =
s x j
xj

 x j = q   p ( x ) − CB    p ( x )  dx
0

 x2 
p ( x ) = CS exp  − 
 4Dt 
 x 2j 
CB = CS exp  −
 4Dt 
 
• From Irvine’s curve: CS=4 Χ 1017cm-3 & Dt=3.7 Χ 10-9cm2
• Can find out dose also Q = CS  Dt
Silicon VLSI Technology by Plummer.
 Methods & Equipment
• High temperature system heating wafers to 800-11000C
• Inert ambient of N2 or Ar
• Use of capping layer like oxide to prevent evaporation
of the dopants
• Proper loading/unloading of the wafers, temperature
ramp up/down very important to avoid thermal shock
• Thermal budget important consideration for shallow
junctions
• Anomalous Transient Enhanced Diffusion (TED) caused
by ion implantation of dopants

Silicon VLSI Technology by Plummer

 Rapid thermal annealing
• Rapid Thermal Annealing (RTA) systems with ramp rates
of 1000C/s very useful
• Optical energy transfer between the radiating lamp &
the wafer

Silicon VLSI Technology by Plummer

• Heat loss at the edges, radial temperature gradients
• Multiple lamps with independent power control or
proper design of chamber reflections to balance
primary heat flux & emitted radiation, pyrometers to
measure temperature
• Flash annealing- ramp to intermediate temperature &
then use of ms flash
Furnace RTA
Batch Single wafer
Long time Short time
Slow ramp Fast ramp
Long stabilization Short steady state
Excellent temperature Accurate temperature
control control difficult
Silicon VLSI Technology by Plummer
 Numerical solutions
• Hopping of atoms from one plane to another in lattice

• Vibration of atoms about their equilibrium position at

Debye frequency (vd ≈ 1013 Hz)
• The hopping frequency vb =vd exp ( − Eb kT )
• N(vb/2) atoms jumping right/left per unit time
• Net flux
−vb −vb −vb 2 C C
F= ( N 2 − N1 )=  x ( C2 − C1 ) =  x = −D
2 2 2 x x
Silicon VLSI Technology by Plummer
• Atoms jumping to & from plane i only
vb
N = N i + t ( N i −1 − 2N i + N i +1 )
i
+

2
vb
Ci = Ci + t ( Ci −1 − 2Ci + Ci +1 )
+

2
Dt
Ci = Ci + 2 ( Ci −1 − 2Ci + Ci +1 )
+

x
+
Ci − Ci Ci −1 − 2Ci + Ci +1
=D
t x2
Ci  2Ci
=D 2
t x

Silicon VLSI Technology by Plummer

 Electric field effects
• Significant electric field effects if doping concentration
higher than ni
• Field induced by higher mobility of electrons & holes
compared to dopant atoms enhances diffusion

C
Ftotal = −D + Cv
x

C   C  C
= D −v
t x  x  x

Silicon VLSI Technology by Plummer

• Velocity of the particles v=

 kT  n 
• The electric field is  =− with = ln  
x q  ni 

• D & μ related by Einstein’s relation D  = kT q

C  n    n
• Total flux Ftotal = −D − DC ln   = − DC ln  C 
x x  ni  x  ni 

• Charge neutrality N D+ +p=N A− + n

• Law of mass action np=ni2

C C
• Total flux Ftotal = −hD with h  1 +
x C 2 + 4ni2
Silicon VLSI Technology by Plummer
• Species with different concentration, significant
changes in the diffusivity of low concentration dopant

PN junction simulation with electric

field effects

MOSFET simulation with

electric field effects

Silicon VLSI Technology by Plummer

 Concentration dependent diffusion
• Fick’s law: flux proportional to the concentration
gradient
• Actual diffusion profiles boxlike, diffusion faster in
higher concentration regions

C   eff C 
=  DA 
t x  x 
𝑛 𝑛2
𝐷∝ or 𝐷 ∝ 2
𝑛𝑖 𝑛𝑖

Silicon VLSI Technology by Plummer

• Isoconcentration experiments, profile of one isotope
introduced into high concentration background doping
of another
• Boron, two isotopes B10 & B11
2
− n = n 
DA = D +D
eff 0
+D  
ni  ni 
• Intrinsic conditions D*A = D0 +D − +D =
• Individual diffusivities follow Arrhenius equation
 D.E 
D=D.0 exp  − 
 kT 
   ( i)

2
• Effective diffusivity D = D  1 + n n + n n
eff * i

− =
A A
 1+  + 
D D  
 = 0 & = 0
D D Silicon VLSI Technology by Plummer
• D.0 & D.E values for typical dopants

• Interaction of dopants with neutral & charged point

defects

Silicon VLSI Technology by Plummer

 Segregation effects
• Different solubility in different materials

• Segregation coefficient- ratio of equilibrium doping

concentration on each side of the interface

⎯⎯
k1
X A ⎯⎯ → XB
k2

FAB = k1C A − k2CB 0.3 for B 

10 for As 
CB k1 CSi  
Steady state FAB =0 = = k0 k0 =  
C A k2 CSiO2 10 for Sb 
 CB  10 for P 
 F = k1  C A − 
 k0 
Silicon VLSI Technology by Plummer
• B segregates into the oxide layer depleting the
concentration in Si near the interface

B, As & P profiles after oxidation Interfacial dopant loss

• Interfacial pileup & dose loss can also have significant

effect
Silicon VLSI Technology by Plummer
 Atomic Scale Diffusion
• Point defects-important role in dopant diffusion

• Vacancy/interstitial assisted diffusion mechanisms

Vacancy assisted diffusion Kick-out & interstitialcy assisted diffusion

Silicon VLSI Technology by Plummer

 Oxidation enhanced/retarded diffusion
• Oxidation of Si enhances the diffusion of P or B &
retards the diffusion of Sb

Simulation results showing OED for Experimental results showing OED

B & ORD for Sb for As & ORD for Sb

• OED correlated with growth of stacking faults

Silicon VLSI Technology by Plummer
• Volume expansion with oxidation, consumption of
vacancies & generation of interstitials
(1 + 2 ) Si + 2OI + 2 V  SiO2 + 2 I + stress

• B prefers to diffuse with interstitials & Sb prefers to

diffuse with vacancies
• Physical size & mismatch with Si Silicon VLSI Technology by Plummer
 Effect of Nitridation
• Nitridation of Si in NH3 ambient reverse effect
• Diffusion of B, P retarded & that of Sb enhanced
• Stacking faults also shrink

• Injection of vacancies by nitridation

Silicon VLSI Technology by Plummer
Fahey et al. APL 46, 784 (1985)
 Mathematical formulation
• Diffusion with a fraction fI by interstitial type
mechanism & a fraction fV (=1-fI) by vacancy type
mechanism
 CI CV 
D eff
A = D  f I * + fV * 
*
A
 CI CV 

• Activation energies for both mechanisms comparable

fI & fV values for some common dopants in Si

Silicon VLSI Technology by Plummer
Fahey et al. APL 46, 784 (1985)
• Effective diffusivity D eff =D0 exp  − E A 
A A  
 kT 
• Sum of diffusivity of each component DAeff =DAI + DAV
• Breaking further into diffusivity of the mobile species &
the fraction of the mobile dopant atoms
C AI C AV
eff
D =d AI
A + d AV
CA CA
• Interaction of dopant atoms with point defects- origin
of mobile species
• Dopant A interacting with interstitial I A + I  AI
• AI the actual interstitial assisted mobile species (the
substitutional dopant is immobile unless it interacts
with a point defect)
Silicon VLSI Technology by Plummer
• An interstitial can combine with vacancy giving rise to
an atom on the lattice site V + I  Si S
• Interstitial supersaturation drive dopant atoms to
mobile state enhancing dopant diffusivity
• OED- interstitial supersaturation during oxidation
• Nitridation- introduce vacancies which recombine with
interstitials & reduce their concentration
• Dopant diffusion from surface creates AI species, which
in turn pump interstitials in the interior
• Diffusion pumping of point defects leads to point
defects supersaturation, which in turn feed back &
enhance the diffusion itself

Silicon VLSI Technology by Plummer

• Important implications for actual devices
• Emitter push effect in BJT- enhanced base width than
expected

Simulation results for chemical pumping Emitter push effect in BJT

& enhanced tail distribution of P

Silicon VLSI Technology by Plummer

• Coupling between interstitials & mobile dopants
 
( CI + C AI ) = − ( FI + FAI )
t x
• Quasi-steady state, time derivatives small

( FI + FAI ) = 0 → FI + FAI = 0 or FI = − FAI
x
• Dopant flux balanced by reverse flux of interstitials
• Interstitial flux F = −d CI → CI = FAI
x x
I I
dI
• Introducing equilibrium interstitial concentration
(
 CI CI*
)=
FAI
x d I C*I
Silicon VLSI Technology by Plummer
• Dopant flux is equivalent to the flux of mobile species
eff C A C AI
DA =d AI  DAeff C A = d AI C AI
x x
• Role of “DC” product in dopant diffusion
• Rewriting  CI CI( *

=
)
FAI
=
d AI C AI DAeff C A
=
x *
d I CI *
d I CI d I C*I
• Low temperatures & highly mobile species

Simulation results for B profile Simulation results for P & As profile

Silicon VLSI Technology by Plummer
 Chemical Equilibrium formulation
• Interstitial mediated reaction A + I  AI
• Law of mass action C AI = kC ACI
• Fick’s law to the mobile species
C AI  C A CI 
FAI = −d AI = −d AI  kCI + kC A 
x  x x 

• Substitutional dopant immobile FA=0

C A FAI   C A CI 
=− = d AI  kCI + kC A 
t x x  x x 
  C AI C A C AI CI     
= d AI  +    d C ln ( A I )
C C
x  C A x CI x  x  x
AI AI


Silicon VLSI Technology by Plummer

Fahey et al. APL 46, 784 (1985)
• Original equation
C A  eff  C A 
= DA  
t x  x 
• Compare & neglect CI x
eff C AI
DA =d AI
CA
• For interstitial type diffuser only,
* CI CI
DA = DA *  d AI C AI = DAC A *
eff *

CI CI
• Using this & including electric field effects
CI    CI n  
FAI = − D C A *
*
 ln  C A *  
 x  CI ni  
A
CI
Silicon VLSI Technology by Plummer