EECS 105 Spring 2005, Lecture 27

Lecture 27:
PN Junctions

Prof. Niknejad

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Diffusion
 Diffusion occurs when there exists a concentration
gradient
 In the figure below, imagine that we fill the left
chamber with a gas at temperate T
 If we suddenly remove the divider, what happens?
 The gas will fill the entire volume of the new
chamber. How does this occur?

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Diffusion (cont)
 The net motion of gas molecules to the right
chamber was due to the concentration gradient
 If each particle moves on average left or right then
eventually half will be in the right chamber
 If the molecules were charged (or electrons), then
there would be a net current flow
 The diffusion current flows from high
concentration to low concentration:

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Diffusion Equations
 Assume that the mean free path is λ
 Find flux of carriers crossing x=0 plane
n ( ) 1
n(0) F  vth  n( )  n( ) 
2
n(  )
1  dn   dn  
F  vth  n(0)     n(0)    
2  dx   dx  
1 1
n( )vth n( )vth dn
2 2 F  vth 
dx
 0  dn
J  qF  qvth 
dx

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Einstein Relation
 The thermal velocity is given by kT
1
2 mn*vth2  12 kT
Mean Free Time
  vth c
 c kT q c
vth   v   kT * 
2
th c
mn q mn*

dn  kT  dn
J  qvth   q  n 
dx  q  dx
 kT 
Dn     n
 q 

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Total Current and Boundary Conditions

 When both drift and diffusion are present, the total
current is given by the sum:
dn
J  J drift  J diff  q n nE  qDn
dx
 In resistors, the carrier is approximately uniform
and the second term is nearly zero
 For currents flowing uniformly through an interface
(no charge accumulation), the field is discontinous
J1 ( 1 ) J1  J 2
 1 E1   2 E2
E1  2
J 2 ( 2 ) 
Department of EECS
E2  1 University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Carrier Concentration and Potential

 In thermal equilibrium, there are no external fields
and we thus expect the electron and hole current
densities to be zero:
dno
J n  0  qn0  n E0  qDn
dx
 n  d
no E0   no 0
dno q
 
dx  Dn   kT  dx

 kT  dno dn0
d0     Vth
 q  n0 n0

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Carrier Concentration and Potential (2)

 We have an equation relating the potential to the
carrier concentration
 kT  dno dn
d0     Vth 0
 q  n0 n0
 If we integrate the above equation we have
n0 ( x)
0 ( x)  0 ( x0 )  Vth ln
n0 ( x0 )

 We define the potential reference to be intrinsic Si:

0 ( x0 )  0 n0 ( x0 )  ni

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Carrier Concentration Versus Potential

 The carrier concentration is thus a function of
potential
n0 ( x)  ni e0 ( x ) / Vth
 Check that for zero potential, we have intrinsic
carrier concentration (reference).
 If we do a similar calculation for holes, we arrive at
a similar equation
p0 ( x)  ni e 0 ( x ) / Vth
 Note that the law of mass action is upheld
n0 ( x) p0 ( x)  ni2e 0 ( x ) / Vth e0 ( x ) /Vth  ni2

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

The Doping Changes Potential

 Due to the log nature of the potential, the potential changes
linearly for exponential increase in doping:
n0 ( x) n0 ( x) n0 ( x)
0 ( x)  Vth ln  26mV ln  26mV ln 10 log 10
ni ( x0 ) ni ( x0 ) 10
n0 ( x)
0 ( x)  60mV log 10
10
p0 ( x )
0 ( x)  60mV log 10
10
 Quick calculation aid: For a p-type concentration of 1016
cm-3, the potential is -360 mV
 N-type materials have a positive potential with respect to
intrinsic Si

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

PN Junctions: Overview
 The most important device is a junction p-type
between a p-type region and an n-type region NA
 When the junction is first formed, due to the
concentration gradient, mobile charges
transfer near junction
 Electrons leave n-type region and holes leave
p-type region −+−+−+−+−+− −
−−−−−− V
 These mobile carriers become minority −+−+−+−+−+−
carriers in new region (can’t penetrate far due +++++ +
to recombination)
 Due to charge transfer, a voltage difference
occurs between regions
 This creates a field at the junction that causes ND
drift currents to oppose the diffusion current n-type
 In thermal equilibrium, drift current and
diffusion must balance
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

PN Junction Currents
 Consider the PN junction in thermal equilibrium
 Again, the currents have to be zero, so we have
dno
J n  0  qn0  n E0  qDn
dx
dno
qn0  n E0   qDn
dx
dno
 Dn
E0  dx   kT 1 dn0
n0  n q n0 dx
dpo
Dp
dx kT 1 dp0
E0  
n0  p q p0 dx
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

PN Junction Fields
p-type n-type
NA ND

p0  N a p0 ( x )

J diff ni2
E0 p0 
Nd

 x p0 xn 0
n0  N d

ni2 J diff
n0  E0
Na

– –++

Transition Region
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Total Charge in Transition Region

 To solve for the electric fields, we need to write
down the charge density in the transition region:
 0 ( x)  q( p0  n0  N d  N a )

 In the p-side of the junction, there are very few

electrons and only acceptors:
 0 ( x )  q ( p0  N a )  x p0  x  0
 Since the hole concentration is decreasing on the p-
side, the net charge is negative:
N a  p0  0 ( x)  0

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Charge on N-Side
 Analogous to the p-side, the charge on the n-side is
given by:
 0 ( x)  q( n0  N d ) 0  x  xn 0

 The net charge here is positive since:

N d  n0  0 ( x)  0
n0  N d

ni2 J diff
n0  E0
Na

– –++

Department of EECS Transition Region University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

“Exact” Solution for Fields

 Given the above approximations, we now have an
expression for the charge density
q (ni e 0 ( x ) / Vth  N a )  x po  x  0
 0 ( x)   0 ( x ) / Vth
 q ( N d  n i e ) 0  x  xn 0
 We also have the following result from
electrostatics
dE0 d 2  0 ( x)
 2 
dx dx s
 Notice that the potential appears on both sides of
the equation… difficult problem to solve
 A much simpler way to solve the problem…

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Depletion Approximation
 Let’s assume that the transition region is
completely depleted of free carriers (only immobile
dopants exist)
 Then the charge density is given by
 qN a  x po  x  0
 0 ( x)  
  qN d 0  x  xn 0
 The solution for electric field is now easy
dE0  0 ( x)
 Field zero outside
dx s transition region
x  0 ( x' )
E0 ( x )   dx '  E0 ( x p 0 )
xp0 s

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Depletion Approximation (2)

 Since charge density is a constant
x  0 ( x' ) qN
E0 ( x )   dx'   a ( x  x po )
xp0 s s
 If we start from the n-side we get the following
result
xn 0  0 ( x' ) qN d
E0 ( x n 0 )   dx'  E0 ( x)  ( x n 0  x )  E0 ( x )
x s s

Field zero outside qN d

E0 ( x )   ( xn 0  x )
transition region s

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Plot of Fields In Depletion Region

p-type ––––– +++++ n-type
––––– +++++

NA ––––– +++++ ND
––––– +++++

Depletion
Region

qN a qN d
E0 ( x )   ( x  x po ) E0 ( x )   ( xn 0  x )
s s

 E-Field zero outside of depletion region

 Note the asymmetrical depletion widths
 Which region has higher doping?
 Slope of E-Field larger in n-region. Why?
 Peak E-Field at junction. Why continuous?
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Continuity of E-Field Across Junction

 Recall that E-Field diverges on charge. For a sheet
charge at the interface, the E-field could be
discontinuous
 In our case, the depletion region is only populated
by a background density of fixed charges so the E-
Field is continuous
 What does this imply?
qN a qN d
E 0 ( x  0)  
n
x po   xno  E 0p ( x  0)
s s
qN a x po  qN d xno
 Total fixed charge in n-region equals fixed charge
in p-region! Somewhat obvious result.
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Potential Across Junction

 From our earlier calculation we know that the
potential in the n-region is higher than p-region
 The potential has to smoothly transition form high
to low in crossing the junction
 Physically, the potential difference is due to the
charge transfer that occurs due to the concentration
gradient
 Let’s integrate the field to get the potential:
x qN a
 ( x)   ( x po )   ( x' x po )dx'
xp0 
s
x
qN a  x' 2

 ( x)   p    x' x po 
s  2  x
p0
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Potential Across Junction

 We arrive at potential on p-side (parabolic)
qN a
 ( x)   p 
o
p
( x  x p0 )2
2 s
 Do integral on n-side
qN d
n ( x )  n  ( x  xn 0 ) 2
2 s
 Potential must be continuous at interface (field
finite at interface)
qN d 2 qN a 2
n (0)  n  xn 0   p  x p 0   p (0)
2 s 2 s

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Solve for Depletion Lengths

 We have two equations and two unknowns. We are
finally in a position to solve for the depletion
depths qN qN
n  d
xn20   p  a
x 2p 0 (1)
2 s 2 s

qN a x po  qN d xno (2)

2 sbi  Na  2 sbi  Nd 
xno    x po   
qN d  Na  Nd  qN a  Nd  Na 

bi  n   p  0

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Sanity Check
 Does the above equation make sense?
 Let’s say we dope one side very highly. Then
physically we expect the depletion region width for
the heavily doped side to approach zero:

2 sbi Nd
xn 0  lim
N d  qN d N d  N a
0 
2 sbi  Nd  2 sbi
x p 0  lim   
N d  qN a  Nd  Na  qN a
 Entire depletion width dropped across p-region

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Total Depletion Width

 The sum of the depletion widths is the “space
charge region”
2 sbi  1 1 
Xd0  x p 0  xn 0    
q  Na Nd 
 This region is essentially depleted of all mobile
charge
 Due to high electric field, carriers move across
region at velocity saturated speed
2 sbi  1  1V V
Xd0   15   1μ E pn   10 4
q  10  1μ cm

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Have we invented a battery?

 Can we harness the PN junction and turn it into a
battery?
 ND NA  NDN A

bi  n   p  Vth  ln  ln 
  Vth ln 2
 ni ni  ni

 Numerical example:
ND N A 10151015
bi  26mV ln 2
 60mV  log 20
 600mV
ni 10

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Contact Potential
 The contact between a PN junction creates a
potential difference
 Likewise, the contact between two dissimilar
metals creates a potential difference (proportional
to the difference between the work functions)
 When a metal semiconductor junction is formed, a
contact potential forms as well
 If we short a PN junction, the sum of the voltages
around the loop must be zero:
0  bi   pm  mn
+ mn n
bi p bi  ( pm  mn )
−
 pm
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

PN Junction Capacitor
 Under thermal equilibrium, the PN junction does
not draw any (much) current
 But notice that a PN junction stores charge in the
space charge region (transition region)
 Since the device is storing charge, it’s acting like a
capacitor
 Positive charge is stored in the n-region, and
negative charge is in the p-region:

qN a x po  qN d xno

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Reverse Biased PN Junction

 What happens if we “reverse-bias” the PN
junction?
+
 bi  VD VD VD  0
−
 Since no current is flowing, the entire reverse
biased potential is dropped across the transition
region
 To accommodate the extra potential, the charge in
these regions must increase
 If no current is flowing, the only way for the charge
to increase is to grow (shrink) the depletion regions
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Voltage Dependence of Depletion Width

 Can redo the math but in the end we realize that the
equations are the same except we replace the built-
in potential with the effective reverse bias:
2 s (bi  VD )  N a  V
xn (VD )     xn 0 1  D
qN d  Na  Nd  bi

2 s (bi  VD )  N d  VD
x p (VD )     x p 0 1 
qN a  Na  Nd  bi

2 s (bi  VD )  1 1 
X d (VD )  x p (VD )  xn (VD )    
q  Na Nd 
VD
X d (VD )  X d 0 1 
bi
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Charge Versus Bias

 As we increase the reverse bias, the depletion
region grows to accommodate more charge
VD
QJ (VD )  qN a x p (VD )   qN a 1
bi
 Charge is not a linear function of voltage
 This is a non-linear capacitor
 We can define a small signal capacitance for small
signals by breaking up the charge into two terms

QJ (VD  vD )  QJ (VD )  q(vD )

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Derivation of Small Signal Capacitance

 From last lecture we found
dQD
QJ (VD  vD )  QJ (VD )  vD  
dV VD

dQ j d  V 
C j  C j (VD )     qN a x p 0 1  
dV dV   
V VD  bi V VR
qN a x p 0 C j0
Cj  
VD VD
2bi 1 1
bi bi
 Notice that
qN a x p 0 qN a  2 sbi  N d  q s N a N d
C j0      
2bi 2bi  qN a  N a  N d  2bi N a  N d

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Physical Interpretation of Depletion Cap

q s N a N d
C j0 
2bi N a  N d
 Notice that the expression on the right-hand-side is
just the depletion width in thermal equilibrium
1
q  1 1  s
C j0   s    
2 sbi  Na Nd  Xd0

 This looks like a parallel plate capacitor!

s
C j (VD ) 
X d (VD )

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

A Variable Capacitor (Varactor)

 Capacitance varies versus bias: Cj
C j0

 Application: Radio Tuner

Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27

Part II: Currents in PN Junctions

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Diode under Thermal Equilibrium

Minority Carrier Close to Junction
Thermal - +
Generation p-type − - + n-type
- + J n ,diff
- +
− - + J p ,diff
- +
- +
−
J p ,drift - - ++
+
J n , drift - - ++
ND - + NA
- +
E0
+ qbi
Carrier with energy
Recombination
below barrier height
−
 Diffusion small since few carriers have enough energy to penetrate barrier
 Drift current is small since minority carriers are few and far between: Only
minority carriers generated within a diffusion length can contribute current
 Important Point: Minority drift current independent of barrier!
 Diffusion current strong (exponential) function of barrier
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Reverse Bias
 Reverse Bias causes an increases barrier to
diffusion
 Diffusion current is reduced exponentially

p-type - +
- + n-type
- +
- +
- +
ND -
- +
+
NA
+
−
q (bi  VR )

 Drift current does not change

 Net result: Small reverse current
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Forward Bias
 Forward bias causes an exponential increase in
the number of carriers with sufficient energy to
penetrate barrier
 Diffusion current increases exponentially
p-type - +
- + n-type
- +
- +
- +
ND -
- +
+
NA
+
−
q (bi  VR )

 Drift current does not change

 Net result: Large forward current
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Diode I-V Curve

Id
Is

I d (Vd  )   I S  qVkTd 

I d  I S  e  1
 

qVd
1 kT

 Diode IV relation is an exponential function

 This exponential is due to the Boltzmann distribution of carriers versus energy
 For reverse bias the current saturations to the drift current due to minority
carriers

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Minority Carriers at Junction Edges

Minority carrier concentration at boundaries of

depletion region increase as barrier lowers …
the function is
pn ( x  xn ) (minority) hole conc. on n-side of barrier

p p ( x   x p ) (majority) hole conc. on p-side of barrier

( Barrier Energy ) / kT
e
pn ( x  xn )
 e q ( B VD ) / kT
NA
(Boltzmann’s Law)

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

“Law of the Junction”

Minority carrier concentrations at the edges of the

depletion region are given by:

pn ( x  xn )  N Ae q ( B VD ) / kT
 q ( B VD ) / kT
n p ( x   x p )  N De

Note 1: NA and ND are the majority carrier concentrations on

the other side of the junction
Note 2: we can reduce these equations further by
substituting
VD = 0 V (thermal equilibrium)
Note 3:of EECS
Department assumption that p << N and n << N University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Minority Carrier Concentration

qVA
pn 0 e kT

p side n side
qVA
x
n p 0e kT  qVkTA   Lp
pn ( x)  pn 0  pn 0  e  1 e
 

Minority Carrier
pn 0 Diffusion Length
np0
-Wp -xp xn Wn

The minority carrier concentration in the bulk region for

forward bias is a decaying exponential due to recombination

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Steady-State Concentrations
Assume that none of the diffusing holes and
electrons recombine  get straight lines …
qVA
pn 0 e kT

p side n side
qVA
n p0e kT

pn 0
np0
-Wp -xp xn Wn

This also happens if the minority carrier

Ln , p  Wn , p
diffusion lengths are much larger than Wn,p
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Diode Current Densities

qVA
pn 0 e kT
qVA

p side n side dn p n p 0e kT
 np0
qVA
( x) 
n p 0e  x p  (Wp )
kT
dx
pn 0
ni2
np0 np0 
Na
-Wp -xp xn Wn

dn p D  qVA

J n  qDn
diff
q n
n p 0  e  1
kT
dx x  x Wp  
p

dp D p  qVA

J p   qD p
diff n
 q pn 0 1  e kT
dx x  xn Wn  

 Dp Dn   qVA 
J diff  qni 
2
   e  1
kT
 N dWn N aW p
  
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Diode Small Signal Model

 The I-V relation of a diode can be linearized
 q (VkT
d  vd )
 qVd qvd
I D  iD  I S  e  1  I S e kT e kT
 
2 3
x x
e x  1  x   
2! 3!

 q(Vd  vd ) 
I D  iD  I D 1   
 kT 
qvd
iD   g d vd
kT

Department of EECS University of California, Berkeley

EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Diode Capacitance
 We have already seen that a reverse biased diode
acts like a capacitor since the depletion region
grows and shrinks in response to the applied field.
the capacitance in forward bias is given by
S
Cj  A  1.4C j 0
X dep
 But another charge storage mechanism comes into
play in forward bias
 Minority carriers injected into p and n regions
“stay” in each region for a while
 On average additional charge is stored in diode
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Charge Storage
q (Vd  vd )
pn 0 e kT

p side q (Vd  vd )
n side
n p 0e kT

pn 0
np0
-Wp -xp xn Wn

 Increasing forward bias increases minority charge density

 By charge neutrality, the source voltage must supply equal
and opposite charge
1 qI d
 A detailed analysis yields: Cd  
2 kT
Time to cross junction
(or minority carrier lifetime)
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Forward Bias Equivalent Circuit

 Equivalent circuit has three non-linear elements: forward

conductance, junction cap, and diffusion cap.
 Diff cap and conductance proportional to DC current
flowing through diode.
 Junction cap proportional to junction voltage.
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Fabrication of IC Diodes

cathode annode

p+ p n+
n-well
p-type

p-type

 Start with p-type substrate

 Create n-well to house diode
 p and n+ diffusion regions are the cathode and annode
 N-well must be reverse biased from substrate
 Parasitic resistance due to well resistance
Department of EECS University of California, Berkeley
EECS 105 Fall 2003, Lecture 27 Prof. A. Niknejad

Diode Circuits
 Rectifier (AC to DC conversion)
 Average value circuit
 Peak detector (AM demodulator)
 DC restorer
 Voltage doubler / quadrupler /…

Department of EECS University of California, Berkeley