Chapter

Electronics I - PN Junction

p++ n

NA >> ND

A K

Fall 2017

talarico@gonzaga.edu 1
 PN Junction (Diode)

• When N-type and P-type dopants are introduced side-by-

side in a semiconductor, a PN junction or a diode is
formed.
JUNCTION
source: Razavi

talarico@gonzaga.edu 2
 Diode I/V characteristics
• Our goal is to find out the current-voltage characteristics of the
device D
“smoke”

ΦB D

D
Equilibrium

• Diode operation regions:

– Equilibrium = open circuit terminals = no bias
– Forward bias
– Reverse bias
– (Breakdown)
talarico@gonzaga.edu 3
PN Junction in Equilibrium (no Bias applied)
• Because each side of the junction contains an excess of holes
or electrons compared to the other side, there exists a large
concentration gradient. Therefore, a diffusion current flows
across the junction from each side.
ND < NA +
nn0 pp0
nn0 ≈ ND
pn0 ≈ ni2/ND
pn0 np0
pp0 ≈ NA
np0 ≈ ni2/NA

source: Razavi talarico@gonzaga.edu 4

PN junction in Equilibrium (no bias applied)
• Diffusion across the junction causes a lot of electrons and holes to recombine leaving a
depletion region in proximity of the junction
• The fixed ions left behind create an electric field
• The electric field opposes the diffusion of holes in the N region and free electrons in the P
region
• Once the electric field is strong enough to stop the diffusion currents the junction reaches
equilibrium
– A “barrier” opposing diffusion forms
Depletion Region
ND < NA
N P
− − − + + + + + +
−
− − − + + + + + +
− − − + + − + + + + + +
− − − + + − + + + + + +
− − − + + + + + +
− − − − + + + + + +
Free Positive Donor Negative Free
Electrons Ions Acceptor Ions Holes 5
 Depletion approximation
• What happens inside the PN junction in equilibrium is not quite as black and white as depicted

ρ (x) = q ( p0 (x) − n0 (x) + N D − N A )

• On the N-side of the transition region (0 < x < xn0),

since there are no acceptors (NA = 0) and the hole
concentration is negligible (p0 ≈ 0)

ρ (x) = q ( p0 (x) − n0 (x) + N D − N A ) ≈ q ( −n0 (x) + N D )

• Since n0(x) < ND on the N-side of the transition (and even

more so on the P-side of the transition) there is a roll-off
in electron concentration

source: Howe & Sodini

talarico@gonzaga.edu 6
 Depletion approximation
• On the P-side of the transition region (−xp < x < 0),
since there are no donors (ND = 0) and the electrons
concentration is negligible (n0 ≈ 0)

ρ (x) = q ( p0 (x) − n0 (x) + N D − N A ) ≈ q ( p0 (x) − N A )

• Since p0(x) < NA on the P-side of the transition region

(and even more so on the N-side of the transition)
there is a roll-off in hole concentration

source: Howe & Sodini

• The depletion approximation assumes that the mobile carriers (n and p) are small in number compared to
the donor and acceptor ion concentrations (ND and NA) in the depletion region (SCR), and the device is
charge-neutral elsewhere (QNR)

• Depletion approximation

– NA >> pp hence ρ ≈ −qNA for −xp ≤ x ≤ 0 That is, there are no free mobile
– ND >> nn hence ρ ≈ qND for 0 ≤ x ≤ xn charges in depletion region
– The charge density is negligible in the bulk regions; that is ρ ≈ 0 for x > xn and x < − xp

talarico@gonzaga.edu 7
source: Sodini

One more time …

P-type N-type
• aa

NOTE: for the example

in this picture NA < ND

= 0 ≈ 0 (minority carriers)
(P-side)

Na >> p0 (depletion approximation)

= 0
(N-side)

≈ 0 (minority carriers) Nd >> n0 (depletion approximation)

talarico@gonzaga.edu 8
 Depletion approximation

Depletion approximation means

we assume the charge density
in each transition region to be
uniform

Charge density

qND
+ x

−
−qNA
source: G.W. Neudeck

talarico@gonzaga.edu 9
PN junction in Equilibrium (no bias applied)
ND < NA • Since the circuit is open no currents exists
Depletion Region
• Therefore in equilibrium the drift currents
N P resulting from the electric field generated
by the fixed ions in the depletion region
− − − + + + + + + must exactly cancel the diffusion currents
−
− − − + + + + + + • This must hold for both types of charge
− − − + + − + + + + + + carriers: Jn = Jp = 0
− − − + + − + + + + + + dp
− − − + + + + + + J p = 0 = pqµ p Ex − qD p
− dx
− − − + + + + + + dn
J n = 0 = nqµn Ex + qDn
Free Positive Donor Negative Free dx
Electrons Ions Acceptor Ions Holes
• Let’s take one of the two equations and
pp0 try to solve it:
nn0 dp
pn0 pqµ p Ex = qD p
dx
np0 dV
Ex = −
dx
−xn0 JUNCTION x
p0 dV dp
x − pµ p = Dp
dx dx 10
0
 Built-in voltage (equilibrium)
depletion region
dV dp
− pµ p = Dp
− pp0 dx dx
nn0
+ + − Dp dp
pn0
V ( xp0) p( xp0)

− −∫ dV = ∫
+ + np0 V (− xn0)
µp p(− xn0)
p
−
p p0
V (−xn0 ) −V (x p0 ) = VT ln
N pn0
P
−xn0 0 xp0 NA
x ΦB = VT ln
ni2 / N D
Ex
!
+ − NA ND
ΦB ΦB = VT ln
ni2

Example
Find built in voltage of a silicon PN ΦB ≅ 26mV × ln
10181017
≅ 26mV × ln(10) × Log10
10181017
≅
junction at room temperature 10 20 10 20
and with NA ≈ 1018 and ND ≈ 1017 ≅ 60mV ×15 ≅ 900mV

talarico@gonzaga.edu 11
 PN Junction in equilibrium (no bias applied)
Depletion Region
• Some of the thermally generated holes
N P
in the N region (minority carrier) moves
− − − + + + + + + + toward the junction and reach the edge
− − − − + + + + + + of the depletion region. There, they
+ + + + + − − −
− − − + + + + + + experience the strong electric force
+ + + − − −
− − − + + + + + + + + present in depletion region, and get
+ + + − −
− − − − + + + + + + swept all the way into the P region
− − − − + + + + + + (where they become majority carriers).
Ex VD • The same happen for the thermally
generated minority carriers (free
Idrift N P ID electrons) in the P region.
Idiff
+ ΦB −
• At equilibrium the flow of carriers swept by the electric field (drift) and the flow
of carriers due the gradient in carriers concentration (diffusion) cancel out
ID = Idiff – Idrift =0
talarico@gonzaga.edu 12
Width of the Depletion region in equilibrium
Idiff
Idrift
metal contact
N P metal contact

− − − − + + + + + +
− − − + + − + + + + + +
K
− − − + + − + + + + + + A
− − − − + + + + + +
−xn0 0 +xp0
E
concentration [cm−3]
pp0=NA
W=Xd0
nn0=ND
pn0≈ni2/ND np0≈ni2/NA
−xn0 +xp0 x
ρ = charge density [Cb/cm−3]

+qND

+xp0
−xn0 x
|Q +|= qNDAxn0 |Q −|= qNAAxp0

−qNA 13
Width of the Depletion region in equilibrium
ρ = charge density [Cb/cm−3] Charge conservation:
+qND | Q+ |=| Q− | ! QJ 0
+xp0
x "
−xn0

|Q +|= qNDAxn0 |Q −|= qNAAxp0 AqN D xn0 = AqN A x p0

"
−qNA

E(x)
N D xn0 = N A x p0
−xn0 +xp0
x
Poisson Equation:
dE ρ −qN A x p0 −qN D xn0
= E(0) = = ! Emax
dx ε si ε si ε si
!

dE =
ρ
dx
As long as we know how to compute
ε si the area of a rectangle we are fine
Permittivity of crystalline silicon:
with the math !!!
εsi ≈ 11.7×8.85×10−12 F/m
εr ε0 talarico@gonzaga.edu 14
Width of the Depletion region in equilibrium

• Let’s go from the electric field to the voltage

− E(x) dΦ 0
E(x) = − ⇔ − E(x)⋅ dx = dΦ 0
qN A x p0 qN D xn0 dx
=
ε si ε si As we move in the x direction (from –xp0 to xn)
− + +
the area under –E(x) keeps growing
− x NOTE: for convenience I flipped the orientation
−xp0 0 +xn0 of the p-side and the n-side with respect to the x
E direction so the name of the
− ΦB
+ depletion region’s edge at the p-side becomes
–xp0 and the name of the depletion region’s
Φ0(x)
edge at the n-side becomes +xn0
source: Howe & Sodini
parabola Area “blue” triangle
Area “red” triangle P ΦB=Φ0 (xn0)−Φ0 (−xp0)

x
parabola N
2
1 qN A x p0
Area “red” triangle =
2 ε si
2
1 qN A x p0 1 qN D xn0 2
ΦB = +
2 ε si 2 ε si
1 qN D xn0 2
Area “blue” triangle =
2 ε si
talarico@gonzaga.edu 15
 Metal-Semiconductor Contact Potential
source: Howe & Sodini

• At first glance would seem strange that the potential drops at the contacts would have to add up to the
built in potential of the junction. Were this is not the case, KVL would imply that current would flow in
thermal equilibrium when a resistor is connected across the metal contacts to the p and n region. This
conclusion violate our basic understanding of the meaning of equilibrium: namely the net current must be
zero in equilibrium

talarico@gonzaga.edu 16
 Width of the Depletion region in equilibrium
⎛ N ⎞ ⎛ N + ND ⎞ ⎛ NA ⎞
W = xn0 ⎜ 1+ D ⎟ = xn0 ⎜ A xn0 = W ⎜
⎝ NA ⎠ ⎝ N A ⎟⎠ ⎝ N A + N D ⎟⎠
W = x p0 + xn0
Charge neutrality: ⎛ N ⎞ ⎛ N + ND ⎞ ⎛ ND ⎞
W = x p0 ⎜ 1+ A ⎟ = x p0 ⎜ A x p0 = W ⎜
N A x p0 = N D xn0 ⎝ ND ⎠ ⎝ N D ⎟⎠ ⎝ N A + N D ⎟⎠
2 2
qN A qN D 2 qN A 2 ⎛ N D ⎞ qN D 2 ⎛ N A ⎞
ΦB = x p0 2 + xn0 ΦB = W ⎜ + W ⎜
2ε si 2ε si 2ε si ⎝ N A + N D ⎟⎠ 2ε si ⎝ N A + N D ⎟⎠

ΦB =
q
W 2 NAND + NDNA
2 2
W =
2 2ε si
ΦB
( NA + ND )
2

=
2ε si
ΦB
( NA + ND )
2

2ε si ( N A + N D )2 q N A N D2 + N D N A2 q N A N D (N D + N A )

2ε si N + ND
xn0 =
2ε si
ΦB
( NA + ND ) N A2
=
2ε si
Φ
NA
W= ΦB A ! Xdo
N A N D (N A + N D )2 N D (N A + N D )
B
q NAND q q

2ε si ND
x p0 = ΦB
q N A (N A + N D )

talarico@gonzaga.edu 17
 Charge on either side of the depletion region
(in equilibrium)
ρ = charge density [Cb/cm−3]
QJ 0 !| Q+ |=| Q− |= AqN D xn0 Junction Charge
+qND
= xn0
−xp0
NA N N xn0 x
QJ 0 = AqN DW = Aq A D W
NA + ND NA + ND |Q +|= qNDAxn0

−qNA
⎛ NA ⎞
xn0 = W ⎜
⎝ N A + N D ⎟⎠ |Q −|= qNAAxp0

⎛ ND ⎞
x p0 = W ⎜
⎝ N A + N D ⎟⎠

NAND 2ε si N + ND N N
QJ 0 = Aq ΦB A = A 2ε si q A D Φ B
NA + ND q NAND NA + ND

= W=Xd0

talarico@gonzaga.edu 18
 Effect of bias voltage on the PN Junction
(a) Open circuit (b) Forward Bias (c) Reverse Bias
ID = Idiff − Idrift (Equilibrium)
+ VD= 0 − VF=VD VR=−VD
source: + VD − + VD −
Streetman ID = 0 ID > 0 ID < 0

The drift current ΦB ΦJ

has the same ΦJ
Idiff Idiff Idiff
direction of the Idrift Idrift Idrift
electric field
(and it is proportional
Φn ΦJ=ΦB−VF
to it)
J p,drift = pqµ p E ΦB ΦJ=ΦB+VR
J n,drift = nqµn E Φp

talarico@gonzaga.edu 19
source:
PN Junction under Reverse Bias
Millman & Halkias

ID
• The external voltage VR is directed from
− ΦB + the N side to the P side (same as ΦB),
− ΦB + therefore the electric field (ER) due to the
bias VR enhances the built in field E. The
depletion region W is widened
VR = −VD VR = −VD
source: • Since the “barrier” potential rises higher
Razavi E than in equilibrium it gets harder for the
+ ΦB − majority carriers to cross the junction and
Equilibrium
(open circuit) easier for the minority carrier to be
swept (drifted) across the junction.

+ ΦJ − • The current carried under reverse bias is

Reverse Bias ΦJ = ΦB + VR negligible the minority carriers are
too few !!

ER talarico@gonzaga.edu 20
 PN junction under reverse bias

KVL: VD + Φmn + ΦJ + Φmn = 0

ΦJ = −VD − Φmn − Φmp = ΦB−VD

potential
= ΦB
potential
barrier in source: Howe & Sodini
barrier
non-equilibrium
in equilibrium talarico@gonzaga.edu 21
PN junction under reverse bias
(a) charge density (b) electric field (c) potential

Example:

PN junction under reverse biases

VD=−2.4 V and VD=−6.4V

Assuming the potential barrier in equilibrium

is ΦB = 0.8V

As the PN junction gets more reverse biased

• the depletion region Xd = xp + xn becomes
wider
• the potential barrier becomes stronger

source: Howe & Sodini 22

 Application of Reverse Biased PN Junction:
Voltage Dependent capacitor
p

Varactor

Qualitatively
we see that if
VR goes up then
CJ goes down
source: Razavi

• A reverse biased PN junction can be used as a capacitor. By varying VR, the

depletion width changes, and so does the capacitance associated.
Therefore a reverse biased PN junction can be seen as a voltage-
dependent capacitor.

talarico@gonzaga.edu 23
 Capacitance of Reverse Biased Diode

2ε si NA + ND 2ε si ⎛ VD ⎞ N A + N D VD
Xd @V = ( B D) N N
Φ − V = Φ B⎜ 1−
Φ ⎟⎠ N N = X 1−
ΦB
⎝
d0
D
q A D q B A D

width of
depletion
region in
equilibrium
careful: cap per unit area

ε si ⋅ A qε si N A N D 1 qε si N A N D 1 1
C J @V = = = =
D
Xd @V 2 ( N A + N D ) ( Φ B − VD ) 2 ( N A + N D ) Φ B (1− VD / Φ B )
D

CJ 0
=
1− VD / Φ B ε si A
= CJ 0 =
Xdo

CJ0 = junction capacitance at zero bias (in equilibrium)

talarico@gonzaga.edu 24
Junction capacitance of reverse biased diode
source: Streetman CJ
CJ 0 CJ 0
C J (VD ) = M = M
The junction ⎛ VD ⎞ ⎛ VR ⎞
⎜⎝ 1− Φ ⎟⎠ ⎜⎝ 1+ Φ ⎟⎠
capacitance B B
is non linear
• M=0.5 for the abrupt PN junction we
CJ0 modeled
• Typical values for M range from 0.3 to
0.5
• SPICE uses VJ (or PB) to denote the
built in voltage in equilibrium (ΦB)

VR=−VD VD= VF
Reverse Voltage Forward Voltage

talarico@gonzaga.edu 25
 Junction capacitance of reverse biased diode
NAND NAND ⎛ VD ⎞ VD
QJ @V = A 2ε si q
NA + ND
( Φ − VD) = A 2 ε q
NA + ND
Φ B⎜ 1− ⎟ = Q 1− =
⎝ ΦB ⎠ ΦB
D
B si J0

charge stored in
V = ς
= QJ 0 1+ R either side of the
ΦB depletion region
in equilibrium

• The relation between charge QJ and voltage (VR=−VD) across the diode is not linear
• It is difficult to define a capacitance that changes whenever VR changes
/A

NAND
QJ = A 2ε si q
NA + ND
( Φ B + VR ) = ς Φ B + VR
P
1 𝑑𝑄dQ ς /A
"
𝐶" |@%& C=J @V = . J
𝐴 𝑑𝑉-dV/R
P
==
2 d
3 dV
(
ς Φ B + VR ) =
2 Φ B + VR
=
0V1/
R =V&
P
R @VP @VP
square
root ς /A CJO
law = =
⎛ V ⎞ V
1+ R
2 Φ B ⎜ 1+ R ⎟
VP ⎝ ΦB ⎠ ΦB
source: Sedra & Smith @VP @VP
talarico@gonzaga.edu 26
 Junction capacitance of reverse biased diode
Important result to remember
when using non linear devices

dQJ QJ (VR )
C J @V = ≠
P
dVR VR =VP
VR V
R =VP
/A

VP
source: Sedra & Smith talarico@gonzaga.edu 27
Application of a reverse biased diode
source: Razavi

1 1
f res =
2p LC

talarico@gonzaga.edu 28
source:
Millman & Halkias
PN Junction in Forward Bias
ID p-n
ID
• The electric field (EF) due to the
− ΦB + − ΦB + bias (VF) is opposite to the built
in field E. The depletion region
W is shortened
VD=VF VD=VF
• Since the “barrier” potential
source:
Razavi
+ ΦB − becomes lower than in
equilibrium it gets easier for
Equilibrium the majority carriers to diffuse
(open circuit) through the junction and
harder for the minority carrier
to be swept (drifted) across the
junction.
+ ΦJ − ΦJ = ΦB − VF
Forward Bias
• The majority carriers are a lot,
so we expect a considerable
current flow:
ID = Idiff − Idrift
EF E

talarico@gonzaga.edu 29
 Diode’s Current/Voltage Characteristic
source: Razavi
VD
VT
I D = I S (e −1)
VD
very steep !
ID

ID

⎛ qVD ⎞
≈ −IS I D = I(drift) ⎜ e KT −1⎟
⎝ ⎠

−I(drift)
NOTE: VD

Consider |VD/VT| ≈ 2.3 exp(2.3) ≈ 9.97 ≈ 10

exp(−2.3) ≈ 0.1
source: Streetman
At room temperature VD = 2.3 × VT ≈ 59.8 mV≈ 60 mV

talarico@gonzaga.edu 30
 Diode’s Current/Voltage Characteristic
• How can we quantify the I/V behavior of the diode ?
• We need to look at how electrons and holes move under the effect of the bias
voltage
holes at −xp0
• In equilibrium: holes at xn0
−xp0 and xn0 are
p p0 p p0 the edges of the
− Φ B /VT electrons at xn0
Φ B = VT ln ⇔ pn0 = p p0 ⋅ e ⇔ pn0 = Φ B /VT depletion region
pn0 e in equilibrium
n
electrons at −xp0 n p0 = Φ Bn0/VT
e
• With an external bias voltage applied the situation is similar but now the potential
barrier is ΦB – VD instead of only ΦB quasi-neutral
holes at −xp P-side N-side region
holes at xn
eVD /VT
pn = p p ⋅ Φ B /VT A − K
Law of the e VD + + GND
−
electrons at −xp
Junction
eVD /VT quasi-neutral
n p = nn ⋅ Φ B /VT x
−xp and xn are e region
the edges of the electrons at xn −Wp −xp 0 xn Wn
depletion region The Law of the Junction relates the minority carrier
with non zero bias concentration on one side of the junction to the majority WP WN
carrier concentration on the other side of the junction 31
 Diode’s Current/Voltage Characteristic
• When we apply a forward bias VD the potential barrier is reduced from ΦB to ΦB−VD.
This reduces the drift field and upsets the balance between diffusion and drift that
exists at zero bias.

• Electrons can now diffuse from the N side (where they are majority carriers) to the P
side (where they become minority carriers). Similarly, holes diffuse from the P side into
the N side. This migration of carriers from one side to the other is called injection.

• As a result of injection, more electrons are now present at –xp and more holes appear at
xn than when the barrier was higher

– There is an excess of minority carriers at the edges of the depletion region

• Let’s make some simplifying assumptions to help quantifying pn(xn), pp(−xp), nn(xn)
and np(xp)

– There is electric field only in the space charge region (that is between –xp and xn).
The regions from –Wp to –xp (a.k.a. P bulk region) and from xn to Wn (a.k.a. N bulk
region) are quasi-neutral (it is like they were perfect conductors and in perfect
conductors there is no electric field inside)
talarico@gonzaga.edu 32
 Diode’s Current/Voltage Characteristic
• The requirement that the bulk regions on either side remain charge-neutral
even under applied bias (remember: E(x) = ρ(x)/ε, so no field means no charge
density) implies that any increases in the minority carriers due to injection
must be counter balanced by in an increase of majority carriers

– On the P side ρ(x=−xp) = 0 implies that:

p p (−x p ) = N A + n p (−x p ) ≈ p p0 + n p (−x p )

– On the N side ρ(x=xn) = 0 implies that:

nn (xn ) = N D + pn (xn ) ≈ nn0 + pn (xn )

• Assuming the level of injection is low we can neglect the slight increase in
majority carrier concentration due to carrier injection and write:
p p (−x p ) ≈ p p0 ≈ N A nn (xn ) ≈ nn0 ≈ N D
n p (−x p ) << p p0 pn (xn ) << nn0
talarico@gonzaga.edu 33
Minority and Majority carriers concentrations in a
forward biased diode
source:
Quasi-neutrality demands that at every

• Concentrations Howe & Sodini

Howe
&
point in the quasi-neutral regions:
excess minority carrier concentration =
excess majority carriers concentration
Sodini

pn0
np0
x
talarico@gonzaga.edu 34
Minority and Majority carriers concentrations in
a forward biased diode
Quasi-neutrality demands that at every point in the quasi-neutral regions:
excess minority carrier concentration = excess majority carriers concentration

p-QNR p
NOTE:
at this point we did
p(−xp) not put much thinking
qPp in the profile of the
Na minority carriers
diffusing in the QNRs,
however if we
n assume the diffusion
occurs over a short
distance we can
ni 2 approximate the
Na n(−xp) curves with straight
qNp lines !
x
−Wp −xp 0

source: Sodini
talarico@gonzaga.edu 35
 Diode’s Current/Voltage Characteristic
• Under the low injection condition we can approximate the Law of the
Junction as follows:
p p (−x p ) VD /VT p p0 VD /VT
pn (xn ) = Φ B /VT ⋅ e ≈ Φ B /VT ⋅ e = pn0 ⋅ eVD /VT The concentration of minority carriers
e e at the edges of the space charge region
n (x ) n are an exponential function of the
n p (−x p ) = nΦ B /VnT ⋅ eVD /VT ≈ Φ Bn0/VT ⋅ eVD /VT = n p0 ⋅ eVD /VT applied bias
e e
source: G.W. Neudeck

ni 2 VD /VT
pn (xn ) ≈ pn0 ⋅ e VD /VT
≈ ⋅e
ND
ni 2 VD /VT
n p (−x p ) ≈ n p0 ⋅ e VD /VT
≈ ⋅e
NA

talarico@gonzaga.edu 36
 Minority carriers in a forward biased PN junction
• The excess concentration of minority carriers at the edges is:
Δn p ! n p0 ⋅ eVD /VT − n p0 = n p0 (1− eVD /VT ) Δpn ! pn0 ⋅ eVD /VT − pn0 = pn0 (1− eVD /VT )

• As the injected carriers diffuse through the quasi-neutral regions they recombine with the
majority carriers and eventually disappear. The excess minority carriers decay exponentially
with distance, and the decay constant is called diffusion length L.
source:
Sedra & Smith p+ n

It can be shown through the

current continuity equation
that under the low injection
condition the concentration’s
profile of the minority
carriers diffusing in the quasi −( x−xn )

pn (x) = Δpn e
Lp
neutral regions is exponential
( x+x p )

n p (x) = Δn p e Ln

talarico@gonzaga.edu 37
Minority carrier concentrations
Equilibrium vs. Forward Bias
source: G.W. Neudeck

talarico@gonzaga.edu 38
 Aside (a simple trick to simplify the math)
x '' ! 0'' ≡ −x p
• Change the axis selection for the bulk regions as follows:
x ' ! 0' ≡ xn

source: G.W. Neudeck

x ''
−
Ln
n p (x '') = Δn p e
• This way we can write the minority carriers concentrations as follows: −
x'
Lp
source: G.W. Neudeck pn (x ') = Δpn e

Ln = Dnτ n
L p = D pτ p
Ln = Diffusion length for electrons [m]
Lp = Diffusion length for holes [m]
Dn = Diffusion coefficient for electron [m2/s]
Dp = Diffusion coefficient for holes [m2/s]
VA = voltage applied ≡ VD
τn = mean lifetime for electrons [s]
τp = mean lifetime for holes [s]
talarico@gonzaga.edu 39
 Physical interpretation of L, D, and τ
• Physically Lp and Ln represents the average distance minority carriers can
diffuse into a “sea” of majority carriers before being annihilated
∞ ∞

∫ xΔp (x)dx
n ∫ xΔn (x)dx
p

Lp ! 0
∞
Ln ! 0
∞

∫ Δp (x)dx
n ∫ Δn (x)dx
0
p
0

• Physically τp and τn represents the average time minority carriers can

diffuse into a “sea” of majority carriers before being annihilated

• Physically the Dp and Dn represents the ease with which holes and
electrons diffuse
KT KT
Dp ! µ p Dn ! µn
q q

talarico@gonzaga.edu 40
 PN Junction current
• The current that flows through any “transverse” section x must be the
same and is formed by both holes and electrons
J(x) = J n (x) + J p (x)
J n (x) = J n,drift (x) + J n,diff (x)
J p (x) = J p,drift (x) + J p,diff (x)

• If we look at the current through a section x in the quasi-neutral regions,

since in the quasi-neutral regions there is no electric field there will be no
drift injected electrons diffusing on P-side (minority carriers)

= q n n p0 ( eVD /VT − 1) e p p for − W p ≤ x ≤ −x p

dn p (x) D
J n (x) = J n,diff (x) = qDn
( x+x
+(x+x)/L
)/L p n

dx Ln

= q p pn0 ( eVD /VT − 1) e

dp (x) D −( x−xn )/L p
J p (x) = J p,diff (x) = −qD p n for xn ≤ x ≤ Wn
dx Lp
injected holes diffusing on N-side (minority carriers)
talarico@gonzaga.edu 41
 PN Junction currents
• The easier sections for evaluating the diffusion currents are the edges of the space
charge region (x=−xp and x=xn)
n p0 ( eVD /VT − 1)
Dn
J n (−x p ) = J n,diff (−x p ) = q
Ln

pn0 ( eVD /VT − 1)

Dp
J p (xn ) = J p,diff (xn ) = q
Lp
• The total current through a section of the diode can be found as follows:
We do not know have this value (yet) !
J = J n (−x p ) + J p (−x p ) = J n (xn ) + J p (xn )
We do not know have this value (yet) !
• If we make the simplifying assumption that there is no recombination-generation
in the space-charge region, this means that the electrons/holes current at
x=−xp and at x=xn do not change in the space-charge region and therefore we can
write:
J n (−x p ) = J n (xn )
J p (xn ) = J p (−x p )
talarico@gonzaga.edu 42
 PN Junction current
• Therefore we can add together the diffusion currents at x=−xp and x=xn

( ) ( )
D D
J = J n (−x p ) + J p (xn ) = q n n p0 eVD /VT − 1 + q p pn0 eVD /VT − 1
Ln Lp

⎛ Dn ⎞ V /V
pn0 ⎟ ( e D T − 1)
Dp
J = ⎜q n p0 + q
⎝ Ln Lp ⎠

⎛ ⎞ V /V
pn0 ⎟ ( e D T − 1)
Dn Dp
I D = J × A = I n,diff + I p,diff = ⎜ Aq n p0 + Aq
⎝ Ln Lp ⎠
scale current (a.k.a. saturation current)

⎛ Dn ni 2 D p ni 2 ⎞ V /V
I D = Aq ⎜ +
⎝ Ln N A L p N D ⎠
⎟ ( e D T
− 1) = I S ( eVD /VT
− 1)

43
Minority concentrations for a forward biased
short base diode
• If the diode is short the exponential decays can be approximated with
straight lines
dn p n p (−x p ) − n(−Wp ) n p0 eVD /VT − n p0
≈ =
dx WidthP WidthP

dpn pn (xn ) − pn (Wn ) pn0 eVD /VT − pn0

≈ =−
dx −WidthN WidthN

VD A K GND
pn0

np0

−Wp −xp 0 xn Wn
WidthP WidthN
x

⎛ D Dp ⎞ V /V
I D = Aq ⎜ n
n p0 + pn0 ⎟ ( e D T −1) = I S ( eVD /VT −1)
⎝ WidthP WidthN ⎠
talarico@gonzaga.edu 44
Current densities in the forward bias case
No matter which section x we consider Jp
and Jn must add to the same total value J
source: G.W. Neudeck

talarico@gonzaga.edu 45
 Assumptions used to derive the
Ideal diode equation
• Depletion approximation (in depletion region: NA >> pp0(x), ND >> nn0(x))
• Negligible fields in the quasi neutral regions (QNR). This means that the
applied voltage VD is dropped entirely across the space charge region
(SCR)
• Quasi-neutral regions that are so long that the injected carriers all
recombine before reaching the end contacts
• Low levels of injection
• No recombination-generation in the space-charge region. This means
that the electron current and the hole current are constant in the space-
charge region ( that is Jn(−xp)=Jn(xn) and Jp(xn)=Jp(−xp) )

• Real diodes may deviate considerably from the ideal equation.

VD
n = emission coefficient (a.k.a. ideality factor)
* nVT
I D = I (e
S −1) Typically the ideality factor n varies from 1 to 2 46
 Ideal diode
⎛ VD ⎞
V V
y = log10 ⎜⎜ I S ⋅ e VT ⎟⎟ = log10 ( I S ) + log10 (e)⋅ D ≈ log10 ( I S ) + 0.43 D
⎝ ⎠ VT VT
source: Sodini
x = VD

dy 0.43
ID log10|ID| dx
≈
VT

at room temperature (300K ) :
IS VT
≈
26mV 60mV
≈
0.43 0.43 10
every 60 mV the
current changes
VD VD by a decade
IS

talarico@gonzaga.edu 47
 Deviations from the ideal diode

Ideal with IS

VDIODE (V)

source: Hu talarico@gonzaga.edu 48
 Ideal diode equation
• Although we derived the ideal diode equation by subjecting the diode to a
forward bias, the steps followed can also be applied for the case of reverse
bias (the only difference is that now VD < 0).
• The equation works for both forward and reverse region.

I D = I S ( eVD /VT − 1)
ID

source: Streetman ≈ I S eVD /VT

VD

≈ −I S
talarico@gonzaga.edu 49
 Minority concentrations
for reverse biased diode
• In forward bias, minority carriers are injected into quasi-neutral regions
• In reverse bias, minority carriers are extracted (swept away) from the quasi-
neutral regions
• In reverse bias there is a very small amount of carriers available for extraction
– ID saturates to small value
Minority-carrier density
distribution as a function
of the distance from the
junction.
(a) Forward biased junction
(b) Reverse biased junction
(The depletion region is
assumed so small relative
to the diffusion length that
is not indicated in the figure)

source: Millman & Halkias talarico@gonzaga.edu 50

Minority carrier concentrations
Equilibrium vs. Reverse Bias
source: G.W. Neudeck

talarico@gonzaga.edu 51
 Small-signal equivalent model for diode
• Examine the effect of a small signal overlapping the bias:
vD (t) ≡ VD + vd (t)
iD is a function of vD:
⎛ vD ⎞ ⎛ VD +vd ⎞ ≡ gd
Q ≡ (I D ,VD ) iD = I S ⎜⎜ e −1⎟⎟ = I S ⎜⎜ e VT −1⎟⎟
VT

⎝ ⎠ ⎝ ⎠
diD diD
iD ≅* iD V + (vD −VD ) = I D + vd
D
dvD VD
dvD VD

iD − I D ≅ gd vd ⇔ id ≅ gd vd
VD VD

d ⎡⎢ ⎛ VT ⎞⎤⎥
vD
diD I S e VT I S e VT + I S − I S
= I S ⎜⎜ e −1⎟⎟ = = =
dvD VD
dvD ⎢⎣ ⎝ ⎥
⎠⎦VD V T VT

ID + IS
=
VT
* Applied Taylor
source: Sedra & Smith Figure 4.14 Development of the diode small-signal model.
talarico@gonzaga.edu 52
 Small-signal equivalent model for diode
• From a small signal point of view the diode behaves as a (non
linear) resistance of value:
1 VT
rdiode = = rdiode = 1/gd
gd I D + I S

ID + IS
• rdiode depends on bias: gd =
VT

ID
• In forward bias: In reverse bias:
gd ≈
VT
gd ≈ 0

• Under the assumption that the signal overlapping the bias is

small the i-v relationship of the diode can be linearized:
vD = VD + vd iD ≅ I D + gd vd = I D + id

source: Sedra & Smith Figure 4.14 Development of the diode small-signal model.
talarico@gonzaga.edu 53
 Small-signal equivalent model for diode
• Besides having current flowing through it (resistive behavior) the diode
also exhibit charge displacement mechanisms that can be modeled as a
capacitive behavior

• The charge response to the voltage applied across the diode is non-linear

• As long as the voltage signal superimposed to the bias voltage is small, the
relationship between charge and voltage can be linearized.

vD = VD + vd
≡ Cdiode
dqD
qD ≈ QD V + (vD −VD ) = QD V + Cdiode ⋅ vd
D
dvD VD
D

• Complete small signal model for diode rdiode Cdiode

talarico@gonzaga.edu 54
 Capacitive effects in the PN junction
• Diodes have two capacitive effects

• But … please careful:

there are way too many letters “D”:

– depletion capacitance = junction capacitance = CJ

– diffusion capacitance = storage capacitance = CS A A

– Cdiode = CJ + CS Cdiode CJ CS
K K

• We already know quite a bit about the junction cap. that develops when
the diode is in reverse bias (… but what about forward bias?)

• So far we did not even think about the existence of storage (diffusion)
capacitance

• We need to dig more into the capacitive effects …

talarico@gonzaga.edu 55
 Junction Capacitance
source: Gray, Hurst, Lewis and Meyer

For Hand (Chawla-Gummel)

Calculation
SPICE
≈ 2 × CJ 0

ΦB ≡ ΦB
≡
2

Behavior of junction capacitance as a function of bias voltage VD

talarico@gonzaga.edu 56
 Junction Capacitance
• Earlier, we derived the capacitance associated with depletion region
(junction capacitance) for a reverse biased diode as:
CJO ε si q N A N D 1
Cj V = CJO = zero-biasjunction capacitance = A ⋅ ⋅
D
V 2 N A + N D ΦB
1− D
ΦB

• Under significant forward bias VD > ΦB/2 the approximation that

there are no mobile carriers in the depletion region becomes invalid
(this was the basic assumption behind the derivation !!)
• Therefore, under forward bias a common approximation is to take
the value of CJ at VD = ΦB/2 (or being slightly more pessimistic)
CJO
Cj V ΦB = = 2 × CJO ≈ 2 × CJO
D=
2 1−1 / 2

talarico@gonzaga.edu 57
 Diffusion capacitance
• The diffusion capacitance is due to charge carrier storage
occurring in QNRs (to maintain quasi-neutrality)
VD VD
p-QNR p p(xn ) =
ni2 VT
e n(−x p ) =
ni2 VT
e
Nd Na
Majority (−) minority carriers charge in QNRs?
Majority (+)
p(−xp)
Let’s compute charge in minority
qPp
Na carriers “triangles”:
qA ⎛ n 2 VD n 2 ⎞
qPn = (Wn − xn ) ⎜⎜ i e VT − i ⎟⎟ = −qNn
2 ⎝ Nd Nd ⎠
n
qA ⎛ n 2 VD n 2 ⎞
Minority (+) qNp = − (Wp − x p ) ⎜⎜ i e VT − i ⎟⎟ = −qPp
2 ⎝ Na Na ⎠

ni 2 Minority (−) What happen to the majority

n(−xp)
Na carriers in the QNRs ?
qNp
x
−Wp −xp 0

Quasi-neutrality demands that at every point in the quasi-neutral regions:

excess minority carrier concentration = excess majority carriers concentration
talarico@gonzaga.edu 58
 Diffusion capacitance
• Let’s examine what happen when we apply a small increase in
voltage (vD=VD+vd=VD+ΔvD)
Small increase in VD
n
n(xn) [for VD+vd] Small increase in qPn
n(xn) [for VD]

Small increase in qNn

Behaves as a capacitor of capacitance:

p VD
p(xn) [for VD+vd] dq qA ni2 1 VT
p(xn) [for VD]
Cdn = Pn = (Wn − xn ) e
dvD vD =VD
2 N d VT

talarico@gonzaga.edu 59
 Diffusion capacitance
• If we recall the equation of the short-base diode:
⎛ D ni2 Dp ni2 ⎞ VD /VT
I D = Aq ⎜⎜ n
+ ⎟⎟ ( e −1) = I D,n (−x p ) + I D, p (xn )
⎝ WP − x p N a WN − xn N d ⎠

• using some math manipulation we can rewrite the diffusion capacitance of

the n-QSR region as a function of the portion of diode current due to the
holes in n-QNR (ID,p) :
V 2 V 2
ni2 1 VT (Wn − xn ) Dp 1 (Wn − xn ) (Wn − xn ) ni2 VDT 1 (Wn − xn )
D
qA
Cdn = (Wn − xn ) e = qA e ≅ I D, p
2 N d VT (Wn − xn ) Dp VT 2Dp Dp Nd VT 2Dp

• We define transit time of the holes through the n-QNR the average time
for a hole to diffuse through N-QNR:
2

τ Tp ≡
(Wn − xn )
2Dp
talarico@gonzaga.edu 60
 Diffusion capacitance
I D, p
• Then in summary: Cdn ≅ τ Tp
VT
• Aside:
– we already had a name for the average time it takes a hole to diffuse
through an n-region. It is the hole’s lifetime τp
L p2
τp =
Dp

– In fact, if we compute the amount of excess charge given by the holes

in n-QSR accurately (see slide 39):
x' x'
− −
∞ ∞ Lp ∞ Lp
qPn = qA ∫ pn (x ') ⋅ dx ' = qA ∫ Δpn e ⋅ dx ' = qA ∫ pn0 ( e VD /VT
−1) e dx ' =
0 0 0
x'
−
∞ Lp
= qApn0 ( eVD /VT −1) ∫ e dx ' = qApn0 ( eVD /VT −1) L p
0

talarico@gonzaga.edu 61
 Diffusion capacitance τp ≡
L p2
Dp

– Therefore: Note: The increment in the

2 stored hole charge comes
dq ni2 L p VD /VT L p 1 ni2 Dp VD /VT I from injected holes from the
Cdn = Pn = qA e = qA e ≈ τ p D, p P-SIDE (anode) to the N-side
dvD vD =VD
N d VT Dp VT N d Lp VT (cathode)

– Similarly for the excess charge and capacitance in the p-QNR

x ''
∞ ∞ −
qNp = −qA ∫ n p (x '') ⋅ dx '' = −qA ∫ Δn p e Ln
⋅ dx '' = −qAn p0 ( eVD /VT −1) Ln Note: The increment in stored
0 0
electron charge is negative,
but results from electrons that
−dqNp dqNp Ln2 1 ni2 Dn VD /VT I are injected from the cathode
Cdp = = = qA e ≈ τ n D,n
dvD dvD Dn VT N a Ln VT (N-side) to the anode (P-SIDE)
vD =VD vD =VD

– The total excess charge stored and the associated total capacitance is:
qS = qPn + qNp = τ p I D, p + τ n I D,n

dqS I D.n I
CS = = Cdn + Cdp ≈ τ n + τ p D. p
dvD vD =VD
VT VT

talarico@gonzaga.edu 62
 Diffusion capacitance
p-QNR p n
n(xn) [for VD+vd]
p(−xp) for VD+vd n(xn) [for VD]
+dqPp
p(−xp) for VD −dqNn

Na

p
n
p(xn) [for VD+vd]
p(xn) [for VD] +dqPn
ni 2 −dqNp n(−xp) for VD+vd
n(−xp) for VD
Na

x
−Wp −xp 0

talarico@gonzaga.edu 63
 Diffusion capacitance: physical intuition
• If we had looked more carefully at the expressions of the charge stored in
the QNRs rather than immediately starting to crunch equations we would
have noticed that:
⎛ VD ⎞
qPn = −qNn ∝ ⎜⎜ e VT −1⎟⎟
⎝ ⎠
⎛ VD ⎞
−qNp = qPp ∝ ⎜⎜ e VT −1⎟⎟
⎝ ⎠
• Therefore, the stored charge qS, is proportional to the diode’s current ID,
which is also proportional to exp(VD/VT)−1
⎛ VD ⎞
qS = qPn + qNp ∝ ⎜⎜ e VT −1⎟⎟ ∝ I D ⇔ qS = τ T I D
⎝ ⎠
• The constant of proportionality is called charge-storage time or transit
time
ID τ
CS ≅ τ T ≅ τ T gd = T
VT rdiode
talarico@gonzaga.edu 64
 Diffusion capacitance: physical intuition
• There is a very simple explanation to this proportionality, ID is the rate of
minority charge injection into the diode.

• In steady state. This rate must be equal to the rate of charge recombination,
which is qS/τT

• In a one-sided junction the transit time τT is the recombination lifetime on the

lighter-doping side, where most of the charge injection and recombination
take place

• In general, τT is an average of the recombination lifetimes on the N-side and P-

side

• In any event, ID and qS are simply linked through a charge storage time τT
(a.k.a. τS)

talarico@gonzaga.edu 65
 Diffusion capacitance: physical intuition
Example

• NA >> ND
– N-side is more lightly doped than the P-side
– Diffusion of minority carriers on N-side dominates
VD VD
L p2
Dp VT Dp n2
VT
τ p ≡ hole lifetime ≡
I D ≈ I D, p (xn ) ≈ qA pn0 e = qA i
e Dp
Lp Lp N d

– The excess charge stored is: qS ≈ qPn ≈ τ p I D, p (xn ) ≈ τ p I D

τS ≈ τ p
ID
– And finally:
CS ≈ τ p ≈ τ p × gD
VT

talarico@gonzaga.edu 66
 Bias Dependence of CJ and CS
Cdiode • CJ (associated with
charge modulation
in depletion region)
dominates in reverse
bias and small
CS forward bias
1
Cdiode CJ
CJ ∝
ΦB −VD

• CS (associated with
charge storage in
QNRs to maintain
quasi-neutrality)
dominates in strong
Vdiode forward bias
VD
VT
source: Sodini CS ∝ e
talarico@gonzaga.edu 67
Summary: B = 5×1015 cm−3K−3/2

Eg = 1.12 eV

ni = 1×1010 cm−3

source:
Sedra & Smith
Summary:

source:
Sedra & Smith
 The Five Semiconductor Equations
d 2Φ ρ q dΦ
• Poisson’s Equation: dx 2
= −
ε
= −
ε
( p − n + ND − NA ) E(x) = −
dx
Gauss ' Law

d 2Φ dE ρ
= − = −
dx 2 dx ε
• Electron current density: J n = qµn E + qDn
dn
dx

dp
• Hole current density: J p = qµ p E − qD p
dx

∂n 1 ∂J n
• Continuity of electrons: = + ( Gn − Rn )
∂t q ∂x

∂ p 1 ∂J p
• Continuity of holes: =
∂t q ∂x
(
+ G p − Rp )

talarico@gonzaga.edu 70
 SPICE Diode model
source: Antognetti & Massobrio

τT
CJ(0)
ΦB

talarico@gonzaga.edu 71
 SPICE characteristic of diode

General Sxntax:
D[name] <anode_node> <cathode_node> <modelname>
.model <modelname> D (<paramX>=<valueX> <paramY>=<valueY> ...)

Example:
D1 1 2 diode_1N4004
.model diode_1N4004 D (IS=18.8n RS=0 BV=400
+ IBV=5.00u CJO=30 M=0.333 N=2)

D2 3 4 idealmod_d
.model idealmod_d D
+ IS=10f
+ n = 0.01
+ IBV=0.1n
* By default BV = inf

source: Antognetti & Massobrio SPICE

talarico@gonzaga.edu 72