Semiconductor Devices: Lectures 1&2

I. Carrier Transport in Semiconductors

To begin with we need to review some of the basic principles which govern charge transport
due to the presence of (i) an applied electric field, (ii) a charge density gradient.

(1) Carrier drift in applied field (2) Carrier Diffusion

The electron drift velocity is: When there is a spatial variation in the
qτ electron density, there is a diffusion
vn = E = μnE
m*n current. In one dimension:
dn
where τ is the mean time between J n = −qDn
scattering events, and μn is the electron dx
mobility. There is a similar relation for where Dn is the electron diffusivity. Dn
holes. The mobility depends on field at and μn are related via the Einstein relation:
high field due to inelastic scattering k T k Tτ
Dn = B μ n = B *
processes, and the drift velocity q mn
saturates; the maximum value of vn is The hole mobility and diffusivity are
~105ms-1 in Si. The net current is: usually much smaller than the electron
( )
J = J n + J p = q n0 μ n + p 0 μ p E values because of its heavier mass.

Mobility and diffusivity data

for Si at 300K as a function of
dopant density. The decrease
at high density is due to
scattering from impurities.

1E14 1E16 1E18 1E20 (cm-3)

(3) Carrier Recombination

The recombination rate R is proportional to the product of the number of
electrons in the conduction band and the number of holes in the valence band:
R = βnp
In thermal equilibrium the recombination rate is equal to the generation rate:
G th = Rth = βn0 p0
If there is an excess of carriers of a particular type, e.g. caused by illumination
of a doped semiconductor, the excess carriers will recombine. If Δp holes are
injected into n-type material:
dp
= − βn0 p ⇒ p( t ) = p0 + Δpe − βn0 t
dt
(βn0)-1 is the minority carrier lifetime τp.

Semiconductor Devices R.A. Taylor TT 2006 1

 (4) Continuity
In general, carrier populations vary in time and space due to the combined effects of
diffusion and recombination. The continuity equations for electrons and holes are:
∂n 1 ∂J n ∂p 1 ∂J p
= − Rn =− − Rp
∂t q ∂x ∂t q ∂x
Under steady state conditions the minority carrier distribution is given by:
∂p ∂ 2 p p − p0
= 0 = Dp −
∂t ∂x 2 τp
Using the boundary condition p(x→∞) = p0 yields the solution:
− x
p ( x ) = p 0 + [ p( 0 ) − p 0 ] e
Lp

where Lp =√Dp.τp is the diffusion length, and (p(0)-p0) is the excess hole population at x=0.

II. The PN-Junction

Now consider a semiconductor that

contains both p-type and n-type
regions. The large carrier
concentration gradients at the
junction between n-type and p-type
regions cause carrier diffusion. This
leaves positively charged donor ions
on the n-side and negatively charged
acceptor ions on the p-side.
Consequently there is a negative
space charge region (SCR) on the p-
side and a positive SCR on the n-
side of the junction.
The SCR creates an electric field
which points from n-to-p. The hole
drift current is clearly n-to-p
whereas the hole diffusion current is
p-to-n. The electron diffusion
current is p-to-n, and the drift
current is n-to-p. At equilibrium
there is no net current across the
junction, and the Fermi level is the
same on both sides of the junction.

Semiconductor Devices R.A. Taylor TT 2006 2

 (5) P-N Junction at Equilibrium
np
The electrostatics of the junction at equilibrium are
described by the Poisson equation:
nn
d 2V dE ρ q
=− =− =− ( N D − N A + p − n)
dx 2 dx ε 0ε r ε 0ε r
where we have assumed that all donors and acceptors
are ionized. Far away from the junction we have: pp
d 2V
= 0 so that ( N D − N A + p − n) = 0
dx 2
Assume the junction is abrupt. In the depletion region
the free carriers are totally depleted, so that :
d 2V qN A
= −dA ≤ x < 0
dx 2 ε 0ε r -dA dD
2
d V qN D
=− 0 < x ≤ dD
dx 2 ε 0ε r
The electric field is then:
qN A ( x + d D )
E( x ) = − −dA ≤ x < 0
ε 0ε r
qN D x
E( x ) = -E M 0 < x ≤ dD
ε 0ε r This figure illustrates the variation of the energy
qN D d D qN A d A bands across the junction together with the field
where EM = = E, potential V and charge density ρ.
ε 0ε r ε 0ε r

(6) Depletion Layer at a p-n junction

The potential difference across the junction is:

dD dD
0
qN A ( x + d A ) qN D ( x − d D )
VC = − ∫ E ( x )dx = ∫ dx − ∫ dx
−d A −d A
ε 0ε r 0
ε 0ε r
q 1
= ( N A d D2 + N D d D2 ) =
E M W (i.e. the area under E )
2ε 0 ε r 2
where W is the width of the junction:
2ε 0ε r ( N A + N D )VC
W = dA + dD =
qN A N D
VC can be obtained from the condition that the drift current balances the diffusion current:
dV dn( x )
− n( x )qμ e + qDe =0
dx dx
Integrate this from -dA to dD to obtain:
k T n
VC = Vn − V p = B ln( n )
q np
where nn is the electron density in the n-region, and np is the electron density in the p-region.
N.B. n p = ni2 / N A , where ni is the intrinsic carrier density.

Semiconductor Devices R.A. Taylor TT 2006 3

III. P-N Junction under bias

A voltage applied to a p-n junction

changes the balance between the
drift and diffusion currents. Under
forward bias the applied voltage
reduces the electrostatic potential
across the junction; the drift current
is reduced in comparison with the
diffusion current, giving minority
carrier injection – holes into the n-
region and electrons into the p-
region. Under reverse bias the
electrostatic potential is increased
which dramatically reduces the
diffusion currents, leaving a small
reverse current.

(7) Voltage-Current Relation for a PN Diode

The ideal diode is assumed to operate under the following conditions:

(i) the depletion layer is abrupt, and there is charge neutrality outside of the layer,
(ii) the charge densities at the boundaries are given by the electrostatic potential,
(iii) the minority carrier injection is weak, much less than the majority densities,
(iv) there are no generation or recombination currents in the depletion layer.
The relation for VC can be re-written:
nn = n p e qVC / k B T and p p = pn e qVC / k B T
We expect these relations to hold for non-equilibrium carrier densities when a bias V is applied:
n̂n = n̂ p e q( VC −V ) / k B T and p̂ p = p̂n e q( VC −V ) / k B T
Since n̂n ≈ nn we then have :
n̂ p − n p = n p ( e qV / k B T − 1 ) at x = − d A : p̂n − pn = pn ( e qV / k B T − 1 ) x = d D
In the neutral n- and p-regions there is no electric field, so the continuity equations yield:
d 2 p̂ n p̂ n − p n
= → p̂n − p n = p n ( e qV / k B T − 1 )e − ( x − d D ) / L D
dx 2 D pτ p

d 2 n̂n n̂ p − n p
= → n̂ p − n p = n p ( e qV / k B T − 1 ) e ( x + d A ) / L A
dx 2 Dnτ n
The total current flowing is:
⎡ dp̂ ⎤ ⎡ dn̂ p ⎤
J = J p ( d D ) + J ( − d A ) = ⎢ − qD p n ⎥ + ⎢qDn ⎥
⎣ dx ⎦ d ⎣ dx ⎦ − d
D A

qD p pn qDn n p ⎛ Dp Dn ⎞⎟
= J S ( e qV / k BT − 1 ), where JS = + = qni2 ⎜ +
Lp Le ⎜ L p N D Le N A ⎟
⎝ ⎠
Thus the current increases exponentially under forward bias, but saturates at − J S for negative bias.

Semiconductor Devices R.A. Taylor TT 2006 4

 The model described above applies more or less to
Ge p-n diodes, but not for Si or GaAs diodes. The
reason for this is that Ge has a relatively high
intrinsic carrier density because of its small
bandgap:
ni(Ge) = 2.5x1013cm-3
Junction ni(Si) = 1.45x1010cm-3
breakdown ni(GaAs) = 1.79x106cm-3
and at room temperature the diffusion current
dominates. However, if ni is small, the generation
Forward Si
current in the depletion region can dominate.

The large asymmetry in the current-voltage

Reverse Si
characteristics of a p-n junction makes it an ideal
device for rectifying ac currents. There are many
other applications, e.g. photodiodes, solar cells,
light-emitting diodes and diode lasers.
Junction breakdown occurs at large reverse bias
voltage by two different mechanisms:
Theory forward (Ge) (a) tunnelling: a valence electron in the p-region
makes a transition to the conduction band of the
n-region. This requires a field of >108Vm-1,
Theory reverse (Ge)
which requires high dopant density ~5x1023m-3.
(b) avalanche: a thermally generated electron gains
enough kinetic energy in the electric field
(>qEG ) to cause impact ionisation in which an
electron-hole pair are generated. Multiplication
factors ~100 can be achieved with several
qV/kBT hundred volts of reverse bias.

IV. Photodiodes
Photodiodes are convenient devices for measuring optical signals. There are two modes of
operation:
(i) Photovoltaic mode: When a p-n junction is illuminated by above-bandgap light electron-
hole pairs are generated which immediately come under the effect of the junction electric
field: electrons are swept to the positively charged n-region and holes to the negatively
charged p-region, thus reducing the voltage below its equilibrium value. The current
qηP
generated by a beam of light with power P and photon energy hν is: i P = , where η is
hν
the fraction of photons that generate e-h pairs. The new drift current is i0 + i P , and so the
diffusion current must also increase in order to bring the junction back to equilibrium. If the
qΔ V J
change in junction voltage is ΔVJ the diffusion current becomes: i D = i0 exp( ).
k BT
qΔ V J k T ⎛ip ⎞
Consequently, i0 + i P = i0 exp( ) and ΔV J = B ln⎜ + 1 ⎟ . Thus the open circuit
k BT q ⎜i ⎟
⎝ 0 ⎠
voltage increases logarithmically with power at high power.

Semiconductor Devices R.A. Taylor TT 2006 5

ΔVJ

iP/i0

(ii) Photoconductive mode: In this case a reverse bias is applied to the junction, which has the
effect of reducing the drift and diffusion currents to very low values, and increasing the width
of the depletion layer. If photons are now absorbed in the depletion layer a "generation"
current is produced which is proportional to the absorbed power. This is the normal mode of
operation of a photodiode.
The sensitivity of the device (responsivity RO) is the photocurrent per unit power, i.e.
ip qη ηλ
RO = = = amps watt - 1
P hν 1.24
The quantum efficiency of a typical Si photodiode is ~80%, so that RO~0.55AW-1. The
responsivity shows a broad peak centred near the bandgap energy: it falls off at longer
wavelengths due to reduced optical absorption in the depletion layer, and at shorter
wavelengths as light is absorbed in the surface layer. The responsivity can be improved by
incorporating a relatively thick intrinsic semiconductor layer between the two doped regions
(to form a p-i-n diode) so that more of the incident light is absorbed – although this will
depend on the wavelength.

Semiconductor Devices R.A. Taylor TT 2006 6

The speed of a photodiode detector depends on various factors:
a) the time of drift of carriers across the depletion layer: given an electron saturation
velocity vs ~ 5x104ms-1, depletion layer width W~2.5μm gives a response time ~50ps, and
so a bandwith of ~20GHz.
b) diffusion through the p- and n-regions: since diffusion is slower than drift, these layers
must be much thinner than the depletion layer, and 0.5μm is typical.
c) the RC time constant of the circuit: to utilise the intrinsic speed of the diode, the load
resistance for a typical junction capacitance of 25pF must be ~0.3Ω. For incident power
P~1μW, iP ~10-7A, and so the voltage is ~30nV; this is rather low, and noise would be a
severe problem. An impedance matching 50Ω load would be a better solution; the
bandwidth would drop to <1GHz, but this is adequate for most purposes.

V. Solar Cells

Solar cells use p-n junctions to generate

power from solar radiation. which
provides about 1kWm-2 at the Earth’s
surface.

Solar cell equivalent circuit.

The current through the cell is given by:

I = I S (exp( qV / k B T ) − 1 ) − I L
where IL is the light-induced current, and V is the voltage across the diode.
The maximum power that can be extracted is close to the value of the product of the short
circuit current ISC times the open circuit voltage VOC, and depends on the choice of load
resistor. When I = 0
qV
k T I
VOC = B ln( L + 1 ) and the output power is : P = IV = I S V(e k BT − 1 ) − I LV
q IS
The maximum output power is:

Semiconductor Devices R.A. Taylor TT 2006 7

 ⎡ k T qV M k T⎤
PM ≅ I L ⎢VOC − B ln( 1 + )− B ⎥
⎣ q k bT q ⎦

ISC
IM

Power

VM VOC

When there is a series resistance RS - arising

for example at the electrical contacts to the
diode - the output voltage is: Vout = V-IRS,
and the power available is significantly
lower. The area of the contacts can be
increased in order to minimise RS, but this
reduces the effective optical apperture and
so IL. So once again we see that device
design requires some degree of compromise.

Semiconductor Devices R.A. Taylor TT 2006 8