EE143  

– Fall  2016
Microfabrication  Technologies

Lecture  8:  Diffusion

Reading:  Jaeger  Chapter  4

Prof.  Ming  C.  Wu  

[email protected]
511  Sutardja Dai  Hall  (SDH)

Surface  Diffusion:  Dopant  Sources

(a)  Gas  Source:  AsH3,  PH3,  B2H6
ceramic  wafer  of  boron  nitride

(b)  Solid  Source BN Si BN Si

(c)  Spin-­on-­glass SiO2+dopant  oxide

(d)  Liquid  Source.

1
 Fick’s  First  Law  of  Diffusion

¶N
J = - D
¶x
D = diffusion coefficient

Fick’s  Second  Law  of  Diffusion

Continuity Equation for Particle Flux :

Rate of increase of concentration is equal to the
negative of the divergence of the particle flux
¶N ¶J
= -
¶t ¶x
(in one dimension)

Fick' s Second Law of Diffusion :

Combine First Law with Continuity Eqn.
¶N ¶ 2 N
=D 2
¶t ¶x
Assumes D is concentration - independent,
which isn' t true in many situations in modern
devices

2
Diffusion  Coefficients  of  Impurities  in  Si

- E A

Substitutional D = DO e kT

Diffusers Interstitial
Diffusers

Cu

Au
B,P

As

Diffusion  Coefficients
æ E ö
D = DO expç - A ÷ Arrhenius Relationship
è kT ø
E A = activation energy
k = Boltzmann's constant = 1.38 x10-23 J/K
T = absolute temperature

3
 Diffusion  Mechanisms  in  Si    
(a)  Interstitial  Diffusion

Example: Cu, Fe, Li, H Fast Diffusion

Cu

10-6 cm2/sec
Au

Diffusion  Mechanisms  in  Si    

(b)  Substitutional  Diffusion   (c)  Interstitialcy Diffusion

Example: Dopants in Si ( e.g. B, P,As,Sb)

4
 Constant  Source  Diffusion
Complementary  Error  Function  Profiles

æ x ö
Concentration : N (x, t ) = N 0 erfcç ÷
è 2 Dt ø
¥
Dt
Total Dose : Q = ò N (x, t )dx = 2 N 0
0
p
N 0 = Surface Concentration
D = Diffusion Coefficient
erfc = Complementary Error Function

Limited  Source  Diffusion

Gaussian  Profiles
Concentration :
é æ x ö 2 ù Q é æ x ö 2 ù
N ( x, t ) = N 0 expê- ç ÷ ú = expê- ç ÷ ú
êë è 2 Dt ø úû pDt êë è 2 Dt ø úû
Q
N 0 = Surface Concentration N 0 =
pDt
D = Diffusion Coefficient
Gaussian Profile

10

5
 Two  Step  Dopant  Diffusion  
(1)  Predeposition dopant  gas

dose  control SiO2 SiO2

Doped  Si  region
Si

(2)  Drive-­in   Turn  off  dopant  gas

or  seal  surface  with  oxide

profile  control SiO SiO2

SiO2
(junction  depth;; 2

concentration) Si

Note:  Predeposition by  diffusion  can  be  

replaced  by  a  shallow  implantation  step.
11

Normalized  Concentration  versus  

Depth
Predeposition Drive-­in

12

6
 Diffusion  of  Gaussian  Implantation  
Profile

13

Successive  Diffusions:  Thermal  Budget

Example: Dttotal for Well drive-in and S/D annealing

Temp
Temp  

Thermal Budget
( Dt ) effective = å ( Dt )
step i
i
well
drive-­in S/D
step Anneal
step
time

For  a  complete  process  flow,  only  those  steps  with  high  

Dt  values  are  important
14

7
 Solid  Solubility  Limits
• There  is  a  limit  to  the  
amount  of  a  given  impurity  
that  can  be  “dissolved”  in  
silicon  (the  Solid  Solubility  
Limit)

• At  high  concentrations,  all  

of  the  impurities  introduced  
into  silicon  will  not  be  
electrically  active

15

High  Concentration  Diffusion  Effects

1) E-­Field  Enhanced  Diffusion
2) Charged  point  defects  enhanced  diffusion

Log C(x)
High conc. profile:
D gets larger Log C(x)
when C(x)
is large

J large
Low conc profile:
Erfc or gaussian J small
x x

* C(x) looks “flatter”

at high conc. regions

16

8
 Electric-­field  Enhancement
Example:  Acceptor  Diffusion

Na(x) p(x)

Na(x)=Na-­(x) hole  gradient

Hole  diffusion  
tendency
E  build-­in
x
Complete  acceptor At  thermal  equilibrium,  hole  
ionization  at  diffusion   current  =0
temperature Hole  gradient  creates  build-­in
electric  field  to  counteract  the  
hole  diffusion  tendency

17

Electric  Field  Enhancement

holes  tend  to  move

+[p] away  due  to  hole
concentration  gradient

Ebuild-­in
B-­
B-­ acceptors
experience
an  additional
drift  force

Enhanced  
Diffusion  for  B-­ acceptor  atoms  

18

9
 Electric  Field  Enhancement  – Substrate  
Perturbation

As
diffusion
caused  by  As
conc  gradient

Uniform  B  
conc  in  
substrate

B-­

19

Example  :  High  Concentration  Arsenic  

diffusion  profile  becomes  “box-­like”

20

10
 Summary  of  High-­Concentration  
Diffusion
• If  doping  conc <  ni:
– Use  constant  diffusivity  solutions
– (profile  is  erfc or  half-­gaussian)

• If  doping  conc >  ni:

– Solution  requires  numerical  techniques

21

Transient  Enhanced  Diffusion  (TED)

Dopant  Implantation C(x)
no  annealing

900oC,  several  minutes

(After  excess  point  defects  
Si Implantation recombine.  slower    diffusion)
induced  excess
substrate point  defects

Implantation  creates x
large  number  of  excess  Si
interstitials  and  vacancies. 900oC,  several  sec  (TED)
(1000X  than  thermal  process). Extremely  rapid  diffusion  due  
After  several  seconds  of   to  excess  point  defects
annealing,  the  excess  point  
defects    recombine.

22

11
Diffusion:  p-­n  Junction  Formation

x j = Metallurgical Junction Depth

P - n junction occurs at the point x j where the net impurity
concentration is zero
(i. e. p - type doping cancels out n - type doping)
æ N ö
Gaussian Profile : x j = 2 Dt lnçç 0 ÷÷
è N B ø
æ N ö
Error Function profile : x j = 2 Dt erfc-1 çç B ÷÷
è N 0 ø

23

Sheet  Resistance

A =W •t
æ r öæ L ö æ L ö
R = ç ÷ç ÷ = RS ç ÷
è t øè W ø è W ø
r
RS = = Sheet Resistance [Ohms per Square]
t

24

12
 Resistivity  vs.  Doping

[
r = s -1 = q(µn n + µ p p)]
-1

n - type : r @ [qµn (N D - N A )]
-1

p - type : r @ [qµ p (N A - N D )]
-1

25

Resistivity  Measurement:  Four-­

Point  Probe

V
r = 2ps [W - m] for t >> s
I
pt V
r = [W - m] for s >> t
ln 2 I
r p V V
RS = = @ 4.53 [W/square]
t ln 2 I I
Four Terminal Resistance Measurement
26

13
 Impurity  Profiling  with
Secondary  Ion  Mass  Spectroscopy  (SIMS)

27

14