Chapter Eight Semiconductor crystals

Elements : Group IV
C(graphite) 1S22S22P2
Si 1S22S22P63S23P2
Ge 1S22S22P63S23P63d104S24P2

Compounds :
IV-IV : SiC
III-V : GaAs, InSb, GaP, ..
Ga 1S22S22P63S23P63d104S24P1
As 1S22S22P63S23P63d104S24P3
II-VI : ZnS, CdSe, …
1
Eg
determines intrinsic conductivity and intrinsic carrier concentration
k BT
(Eg can be obtained by optical absorption)

2
Two types of semiconductors

Direct gap Indirect gap

ε ε conduction band
conduction band

Eg Evertical Eg

valence band valence band

k k
0 0

Band edges (extremes) Band edges of valence

and conduction bans
at same k
at different ks
(most compounds)
( Ge[111], Si[100], … )

3
4
Direct absorption process Indirect absorption process

energy of emitted phonon

hωg=Eg hω=Eg+ hΩ

5
Optical absorption of InSb
Eg=0.23eV

A sharp threshold

6
 Why is your computer chip made of Si, but the laser in
your CD player is made of GaAs (GaN in the future)?

Comparison of absorption
Weak absorption GaAs
Si and emission

Absorption
Absorption

red light
1.5eV
1.1eV

Energy of light photon Energy of light photon

Light emission is related

– very high efficiency in GaAs for excited electron to emit light
– very low efficiency in Si 7
Why is GaN interesting?
GaN
ultraviolet light
Absorption red light
1.5eV 3.4eV

Energy of light (photon)

After decades of efforts, finally it is possible to make

blue light emitter and laser.

Shorter wavelength light focuses to smaller spot

implies higher density of information on a CD.
8
Calculated band structure for Ge
FCC lattice w/. lattice constant a
In reciprocal space

L 4π
a
X

Octahedron
Origin point Γ (0,0,0)
Point X (2π/a, 0, 0)
Point L (π/a, π/a, π/a)

9
L Γ X
 FCC Lattice along ΓL FCC Lattice along ΓX
8 6
(101)
(110)
7 (110) (011) 5
(011)
(100) (101)
6 (010) (111)
(001) (101) (110)
(110) (101)
(111) 4
5 (100)
h /2ma2)

(001)

h /ma2)
(010)
(011) (101)
4 (110) 3

2 2
2 2

ε (2π
(001) (010)
ε (3π

(100)(010)(001)
3 (111) (100)
2
2
(111) (011)
1
1
(000) (000)
0 0
-1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.5 1.0
L kx ky kz(π/a) Γ kx (2π/a)
X
[111] [100]
10
 Free electron model Nearly free electron model
(Sommerfeld) (Bloch theorem)
n,k
k
( n is the band index,
(hk is the momentum)
hk is the crystal momentum)
For a given band n,
h 2k 2 no simple explicit form.
Energy ε(k) = General
2m r property
r :r
ε n (k + G) = ε n (k)
The mean velocity for e
r in band n with k
r hk 1 r r
Velocity v(k) = = ∇ kε (k ) r 1 r r
m h v n (k) = ∇ k ε n (k )
h

r r r r ikr • rr
e ik • r ψ nkr ( r ) = u nkr ( r )e
Wave function ψ (r) =
r r r r
k
V u nk ( r + T ) = u nk ( r )
r r
11
 General remarks about Bloch’s theorem

1. Bloch’s theorem introduces a wavevector k which turns out to play the

same fundamental role in the general problem of motion in a periodic
potential as that in free electron model.
But k is not proportional to the electronic momentum.

p̂ψ nkr
h h
(
= ∇ψ nkr = ∇ u nkr ( r )e
i i
r ikr • rr
) In general, ψnk is not an eigenstate
r r
ik • r h r of momentum operator.
= hkψ nk + e
r ∇ u nk ( r )
r
i
Dynamical significance of k can be acquired when we consider the
motions of Bloch electrons to externally electromagnetic fields.

2. k in Bloch’s theorem can always be confined to the first Brillouin zone.

Any k’ out of zone can be back to zone by shifting a displacement
G (reciprocal lattice vector). k’=k+G

12
3. For a given n, the eigenstates and eigenvalues are periodic functions of k
in reciprocal lattice.
r r
ψ ( r ) = ψ n,k ( r )
r r
n, k + G
r

ε = ε n,kr
r r
n, k + G

4. An electron in a level ( band index n and wave number k) has a

non-vanishing mean velocity.
r r 1 r r
v n (k) = ∇ k ε n (k) Remarkable!
h
It asserts that there are stationary levels for Bloch electrons, in spite of
the electron with fixed lattice of ions.
Against Drude’s naive picture of electron-ion scattering.

13
 r r 1 r r
v(k) = ∇ k ε(k)
h
In an external electric field E,
The energy gained by the electron in a time interval ∆t
r r r r r 1 r
∆ε = F • ∆ l = (−eE) • (v∆t) = (−eE) • ∇ kr ε(k)∆∆
hr
( )
r
= ∇ kr ε(k) • ∆k
setting ∆t→0r r
r − eE dk r r
dk = dt h = −eE = F
h dt

In general, equation of motion for an Bloch electron under Lorentz forces

r
dk r 1 r r  r
h = Fext = q E +  ∇ k ε(k)  × B  SI unit
dt  h  

Weak external forces such that band structure still holds. 14

 Real momentum
h h
k p̂ k = k ∇ k = ∑ i(k + G) C k + G
2
k = ψ k (r ) = ∑ C k + G e i(k + G)•r
i i G where G
r r
k k = ∑ Ck +G = 1
2 2
= hk + ∑ hG C k + G
G G

Under a weak external force F,

Impulse = the change of momentum of the crystal
r r r r r
J = ∫ Fdt = ∆Ptotal = ∆PB−electron + ∆Plattice

∆PB−electron
r r r r
( 2 r
= (∇ kr PB−electron ) • ∆k = h∆k + ∑ hG ∇ kr C kr + Gr • ∆k
G
)
r r
( 2 r
∆Plattice = − ∑ hG ∇ kr C kr + Gr • ∆k )
G
r
r r dk r same as for
Therefore,
J = h∆k h =F free electrons
dt 15
Holes in semiconductors
„ In a completely filled band (valence band), no current can flow since
electrons are Fermions and obey the Pauli exclusion principle.
„ The empty states in the valence band are called “holes”.
„ The electrons can “move” if there is an empty state (a hole) available.
„ A hole acts under the external forces as if it has a positive charge +e.
Missing electron = producing hole

16
 ( )
r r r
In a full band : all pairs of states k, − k are filled and ∑ k = 0 .
r r
If an electron of wavevector ke is missing, ∑ k = − k e . Alternatively speaking,
r r r
a hole of wavevector k h is produced and k h = − k e .

Setting the energy of the top of valence band is zero,

ε
the lower in the band the missing electron lies,
the higher the energy of the system.

k The band is symmetric :

r r r
ε e (k e ) = ε e (−k e ) = −ε h (k h )
1 r 1
( )
v h = ∇ k ε h = − ∇ kr (− ε e ) = v e
h h h e

The group velocity of the hole is the same as

that of the electron.
17
 How does a hole move?
t1 t2 t3

r
dk e r 1 r r r
h = (−e) E + ∇ k ε e (k e ) × B  the equation of a motion for an electron
dt  h e 
Applying to a missing electron : creation of a hole

( )
r r
d − kh  r 1 r r r  dk h r 1 r r r
h = (−e) E + ∇ k ε h (k h ) × B  h = (+e) E + ∇ k ε h (k h ) × B 
dt  h h
 dt  h h 
the equation of a motion for a hole
18
exactly the equation of motion for a particle of positive charge
 1 1 d 2 ε(k)
Effective mass (band mass) ∗ = 2
m h dk 2
◙ For a free electron ε(k) = h2k2/2m → m*=m
◙ For electrons in a band, their masses depend on band curvature.
~2 2
h K  2λ 
ε K~ ≈ (λ ± U ) + 1 ±  ~
K≡k−
G
distance to the zone boundary
2m  U 2
ε U>0
near the lower edge of the 2nd band
+
2nd band ~ h2 ~ 2
λ+U ε(K) = ε c + K where εc=λ+U
2m e
n
λ ctro
e le
λ-U rf e
e
-
near the top edge of the 1st band
1st band 2
~
k ε(K) = ε −
h ~ 2 where ε =λ-U
0 π/a v K v
2m h 19
=G/2
 h 2 ~ 2 h 2 ~ 2  2λ  me 1
K = K 1 +  =
2m e 2m  U m 2λ /U + 1

h2 ~ 2 h 2 ~ 2  2λ  mh 1
K =− K 1 −  =
2m h 2m  U m 2λ /U − 1

r r 1 r r r
v(k) = ∇ k ε(k)
h
r r r
F
r
dv(k) 1 d∇ k ε(k) 1 d 2 ε(k) dk 1 d 2 ε(k)  dk 
= = = 2 h 
dt h dt h dk dt h dk  dt 
2 2 

1 1 d 2 ε(k) Definition of the

From Newton’s 2nd law ∗
= 2
m h dk 2 effective mass
Considering an anisotropic energy surface

 1  1 d 2 ε(k)
 ∗ = 2 where µ and ν are Cartesian coordinates.
 m  µν h dk µ dkν 20
reciprocal effective mass tensor (3x3)
 In three (two) dimensions, constant energy surfaces (lines) are not
necessarily spherical (circular), and the effective mass is a tensor:
 1  1 d 2 ε(k)
 ∗ = 2
 m  µν h dk µ dkν

In two dimensions, free electrons ε(k x , k y ) =

h2
2m e
(
k 2x + k 2y )
m 0   1  1 d 2 ε(k) 1
m* =  e Q ∗  = 2 = δ µν
 0 m e   m  µν h dk µ dkν m e

The effective mass depends on the curvature of the bands;

The flat bands have large effective masses
The curved bands have small effective masses

Near the bottom of a band, m* is positive

Near the top of a band, m* is negative
21
 2
∗
m = h
d 2 ε(k)
ε dk 2 ε

m* <0 k k

ε ε

m* >0 k
k

m* can be determined by cyclotron resonance measurements.

22
Effective mass in semiconductors
Cyclotron resonance energy surfaces of the conduction and
eB valence bands near the band edge
ωc = ∗
m where m* is the cyclotron effective mass
Electron Heavy hole Light hole
Crystal
(me/m) (mhh/m) (mlh/m)
GaAs 0.066 0.5 0.082 2
m∗ = h
InAs 0.026 0.39 0.025 d 2 ε(k)
dk 2
Cu2O 0.99 -- 0.58
ε h2k2
direct conduction band ε = Eg +
2me
Eg
k
h2k2
Heavy hole band ε = − 23
Light hole band 2mh
 Bẑ
Let B in the z direction and E in the x-y plane
r r r
dP  dP   dP  E
=  + 
dt  dt  field  dt  scattering Boltzmann transport,
r r r  mvr 
(
= q E + v×B + −) τ
 where τ is the relaxation time of momentum
 
u = v x + iv y du u q
= (E x + v y B)
dv x v x q
+ + = (Ε − iuB)
dt τ m Ε = E x + iE y dt τ m
dv y v y q
+ = (E y − v x B) Ε = Eo eiωt = Eo cos(ω t )x̂ + Eo sin(ω t )ŷ
dt τ m
u = uo eiωt = uo cos(ω t )x̂ + uo sin(ω t )ŷ
circularly polarized
1  q
 + iω  o
u = (Εo − iuo B) The maximum absorption of
τ  m electromagnetic energy by
ωcτ >> 1 semiconductor occurs at the
− i q Εo
uo =
m(ω + ωc − i / τ ) qB .
cyclotron frequency
ωc = ∗ 24
m
 ω=2.4×1010 Hz
T=4K

Silicon Germanium

G.Dresselhaus, A.F. Kip, and C. Kittel, Phys. Rev. 98, 368 (1955) 25
 Si Energy contours in k-space near the
conduction band minimum.
There are six equivalent pockets.
ellipsoids of revolution
Considering the electrons situated close to (0,0, ko)
dispersion relation can be written as,
r
h 2  k x + k y (k z − k o )  t : transverse
2 2 2
dk r r ε=  +
h = −ev × B 2  mt ml  l : longitudinal
dt 
dk x ehBk y kz B
h =− cos θ
dt mt
dk y ehBk x ehB(k z − k o )
h = cos θ − sin θ ko θ
dt mt ml
dk ehBk y kx
h z = sin θ 1/ 2
dt mt  sin 2θ cos 2θ  eB
ωc = eB + 
2 
=
 m l m t m t  mc
Let kx=kxoeiωt, ky=kxoeiωt, kz=ko+kzoeiωt,
depending on the orientation of B
when θ=0, mc=mt 26
when θ=900, mc=(mlmt)1/2
The perturbation theory of band edges suggests that for a direct gap crystal,
the electron effective mass should be proportional to the bandgap Eg
me
m ≈ 0.5 ~ 0.6 (eV) −1
Eg

The smaller the bandgap, the smaller the effective mass.

27
Ge 1S22S22P63S23P63d104S24P2

sp3 hybrid : a mixture of the s- and p- levels

Tetrahedral bonding

The valence band edge at k=0

P3/2 states : fourfold degenerate
mJ : 3/2, 1/2, -1/2, -3/2
Energy difference = Δ
P1/2 states : doubly degenerate
mJ : 1/2, -1/2

Δ: an energy corresponding to
spin-orbit interaction

28
 ε = Ak 2 ± B2 k 4 + C 2 (k 2x k 2y + k 2y k 2z + k 2z k 2x ) A, B, C [h 2 / 2m]

L Γ X UK L L Γ X UK L
[111] [100]
crystal A B C Δ(eV) ml(m) mt(m)

Si -4.29 0.68 4.87 0.044 0.92 0.19 [100] axis

Ge -13.4 8.48 13.2 0.29 1.59 0.82 [111] axis

GaAs -6.98 -4.5 6.2 0.341 29

 Intrinsic carrier concentration
In contrast to metallic conductivity, the conductivity of semiconductor
is strongly temperature dependent.
-- “free” charges must be thermally excited and overcome Eg
charge carrier concentrations n & p have a strong dependence on T.
Semiconductor are called “intrinsic”,
when “free” electrons and holes can be created only by electronic
excitations from the valence band to the conduction band.
∞
n = ∫ D c (ε )f e (ε , T)dε Electron concentration in the conduction band
E c

Ev
p = ∫ D v (ε )f h (ε , T)dε Hole concentration in the valence band
−∞

1  ε −µ
f e (ε ) = ≈ exp − 
Fermi-Dirac distribution exp[(ε − µ ) / k BT] + 1  k BT 
μ-ε>> kBT 1 ε − µ 
f h (ε ) = 1 − f e (ε ) = ≈ exp 
1 + exp[−(ε − µ ) / k BT] k
 B T
30
In the parabolic approximation (for simplicity),
h 2k 2
The energy of an electron in the conduction band, ε k = E c + 2m
e
3/2
Density of states, D(ε ) = 1  2m e  ε − Ec
2 2 
2π  h 
h 2k 2
The energy of a hole in the valence band, ε k = EV −
3/2
2m h
1  2m 
Density of states, D(ε ) = 2  2 h  Ev − ε
2π  h 
3/2
∞ ∞ 1  2m e   µ −ε 
n = ∫ D c (ε )f e (ε , T)dε = ∫ 2 2 
ε − E c exp dε
Ec E c 2π  h   k BT 
3/2
1  2m e   µ ∞  −ε 
= 2 2  exp  ∫ ε − E c exp dε
2π  h   k BT Ec  k BT 
∞  −ε   − Ec 
∫ ε exp dε * exp 
3/ 2
 µ − Ec   k BT   k BT 
 m e k BT 
0

n = 2 2 
exp 
 2πh   k BT 
31
 3/2
Ev
1  2m h 
Ev
ε − µ 
p = ∫ D v (ε )f h (ε , T)dε = ∫ 2 2 
E v − ε exp dε
−∞ −∞ 2π  h   k BT 
3/2
1  2m h   − µ  Ev  ε 
= 2 2  exp  ∫ E v − ε exp dε
2π  h   k B T  −∞  k BT 
∞  −ε   E 
∫ ε exp dε * exp v 
3/ 2
 Ev − µ   k BT   k BT 
 m h k BT 
0

p = 2 2 
exp 
 2πh   k BT 
3/ 2 3/ 2
 m e k BT   µ − E c   m h k BT   Ev − µ 
np = 2 2 
exp 2 2 
exp 
 2πh   k BT   2πh   k BT 
 − Eg 
3
 k BT 
= 4 2 
(m e m h ) exp
3/ 2
 =constant depends on
 2πh   k BT  material and temperature
32
The law of mass action. Eg=Ec-Ev Independent of EF, (µ)
For an intrinsic semiconductor n=p,

 − Eg 
3/ 2
 k BT 
n i = p i = 2 2 
(m e m h )
3/ 4
exp 
 2πh   2k BT 
Taking this value back to n(µ) or p(µ), we can obtain µ

1 3  mh 
Fermi level µ = E g + k B T ln 
2 4  me 

At T=0, µ lies half-way between the conduction and valence bands.

As T increases, µ moves toward the band with smaller effective mass
µ does not go far from mid-gap when mh≈me

33
Equal densities of states in the Different densities of states in the
conduction and valence bands conduction and valence bands

Intrinsic semiconductor : # of holes = # of electrons (n=p)

34
 When me=mh, then µ=0.5Eg Fermi level is in the middle of the gap.

vd
Intrinsic mobility µ =
E
ne 2τ e pe 2τ h
Electrical conductivity σ = neµ e + peµ h = +
me mh
crystal µe(cm2/Vs) µh(cm2/Vs) The hole mobilities are typically
GaAs 8000 300 smaller than the electron
mobilities because of the
InAs 30000 450 occurrence of band degeneracy
at the valence band edge at the
Diamond 1800 1200 zone center, thereby making
Si 1350 480 possible interband scattering
processes that reduce the
Ge 3600 1800 mobility.

PbS 550 600

T=300K 35
Impurity conductivity
Doping : addition of impurities to the crystal
(1) Donors – Group of V such as P, As, Sb N-type
substitutional impurity for semiconductor
each dopant atom contribute an electron
(2) Acceptors – Group of III such as Al, Ga, In P-type

attract electrons from valence band of semiconductor

create a hole per atom

Where do electrons / holes of the dopants go? free or bound

Low T : bound
High T : free kBT > Ed (electron), Ea (hole)
Donor Acceptor
activated activated
energy energy
36
 Activated energy – From Bohr model
Hydrogen atom e2 mv 2
F= = and 2π r = nλ
4πε o r 2
r
4πε o h 2
rn = 2
n = a on
me
e4m 1 13.6eV
En = − = − Ionization energy
32π ε o h n
2 2 2 2
n 2
13.6eV

N-doped Silicon P-doped Silicon

37
 The fifth valence electron of P atom is not required for bounding and
is thus, only weakly bound. The binding energy can be estimated by
treating the system as a hydrogen atom embedded in a dielectric.

Donor e2 me v2
F= = and 2π r = nλ ε = κε o
4πε r 2
r
Dielectric constant
4πκε o h 2
 m
ad = 2
n = a o  κ  ~ao(10)(10) m e instead of m
mee  me 
e4me 1 13.6eV  m e 
Ed = =   ~ 13.6eV(10-2)(10-1)
32π κ ε o h n
2 2 2 2 2
κ 2
 m
Electron energy

Ec
Ed
ED κ P As Sb
Donor level
11.7 Si 45 54 43
Ev 15.8 Ge 13 14 10
Ionization energies Ed [meV]
38
 The valence –three Boron (B) accepts an electron from the Si lattice.
The hole that is thereby created in the valence band orbits around the
negatively charged impurity.

The Bohr model applies qualitatively for holes just as for electrons,
but the degeneracy at the top of the valence band complicates the
effective mass problem.
Electron energy

Ec
B Al Ga In
Si 45 57 65 157
Acceptor level
EA
Ea Ge 10.4 10.2 10.8 11.2
Ev
Ionization energies Ea [meV]

39
 In a doped semiconductor,
an electron in the conduction band can originate either from
the valence band or from the ionization of a donor;
a hoe in a valence band may correspond either to
the electron in the conduction band or to the negatively charged acceptor.
_ _ n
Electron energy E

(neutral)
Density of doped donor Nd = Ndo + Nd+
N d+ (ionized) donor
(neutral)
N a-
Density of doped acceptor Na = Nao + Na-
(ionized) acceptor
+ + p
1
N od = N d
Neutrality condition n + Na- = p + Nd+ exp[(E D − E F ) / k BT ] + 1
1
N oa = N a
exp[(E F − E A ) / k BT ] + 1
40
 For pure N-type semiconductor : only donors are available
 EF − Ec   k B Tm e 
3/ 2
 k B Tm h 
3/ 2

n = n o exp  where n o = 2 2 
and p o = 2 2 
 k BT   2 π h   2 π h 
And
N d = N od + N d+ n = Nd+ + p

For the simple case Nd+>> ni k BT << E g therefore, n ~ Nd+ = Nd - Ndo

 1   1 
n ≈ N d − N = N d 1 −
o
 = N d  
 exp[(E D − E F ) / k BT ] + 1   1 + exp[− (E D − E F ) / k BT ] 
d

n  Ec   EF   

exp 
 = exp   
no  k BT   k BT  1
n ≈ Nd  
 n 
 1 + n exp[E c /k BT ]exp[− E D / k BT ] 
 o 

 n  Ed   1  Ed  2

n 1 + exp    ≈ N d → exp  n + n − Nd = 0
 no  k BT   no  k BT  41
 no  E d  Nd  Ed  
solution n = exp −  − 1 + 1 + 4 exp  
2  k BT   no  k BT  

Nd  Ed 
At low temperatures, such that 4 exp  >> 1
no  k BT 
 Ed 
n ≈ n o N d exp −  Freeze-out range
 2k T
B 

A sufficiently large number of donors still retain their valence electrons,

i.e. are not ionized.
Nd  E 
4 exp d  << 1
At the intermediate temperatures, such that no  k BT 
n ≈ N d = constant Saturation range
The concentration of donor electrons in the conduction band has reached
the maximum possible value, equal to the concentration of donor.
All donors are ionized.
42
 At high temperatures, such that k B T ≈ E g
 Eg 
n ≈ N d + n i ≈ N d + n o exp −  Intrinsic range
 2k BT 

The concentration of electrons excited from the valence band across Eg

increases and eventually outweighs the electron density due to donors.
It behaves as an intrinsic semiconductor.

N-doped Ge with a P concentration Nd=1×1013cm-3 : Ed~0.012eV

Intrinsic Ge : ni=2.4×1013cm-3 (300K) : Eg~0.67eV
N  E 
4 d exp d *  = 1 T*= K
no  k BT 
k BT ** ≈ E g T**=7800K

“Law of mass action” is also valid for the doped semiconductor

 − Eg   − Eg 
3
 k BT 
np = 4 2 
(m e m h ) exp
3/ 2
 = n o p o exp 
 2πh 
43
 k BT   k BT 
 A semiconductor doped with Nd donor electrons

kBT<Ed
Dopant carriers are thermally
excited to conduction band
 Ed 
n ~ n o N d exp − ,
 2k BT 
3/ 2
m k T
where n o = 2 e B2 
 2πh 
Eg>>kBT>Ed
All carriers are excited (Saturation)
n~Nd
Eg~kBT
Intrinsic carriers are excited from
valence band
 Eg 
n ≈ N d + n i ≈ N d + n o exp − 
 2k BT  44
A semiconductor doped with Na acceptor holes
Same results
Low Temperatures,  Ea 
kBT<Ea p~ p o N a exp − ,
 2k BT 
3/ 2
m k T
where p o = 2 h B2 
 2πh 
Intermediate temperatures,
Eg>>kBT>Ea
p=Na

High Temperatures,
Eg~kBT
 Eg 
p ≈ N a + p i ≈ N a + p o exp − 
 2k BT 
45
Fermi level
Assume all Nd electrons are excited into conduction band
n - p = Nd
Mass action law np = nipi = ni2 =pi2 p= ni2/n
N-type semiconductors in saturation regime, n~Nd and p=ni2/n
n i2 n 2 − N d n − n i2 = 0
n = p + Nd = + Nd
n n=
1
2
(N d + N d2 + 4n i2 )
n, p
n~Nd  µ − Eg 
n(N d ) = n o exp 
 k BT 
offset E v = 0
intrinsic
ni=pi
p~1/Nd
Nd
 n(N d )  saturation  no 
µ = E g + k B T ln   µ = E g − k B T ln   46
n
 o   Nd 
 at low temperatures at high temperatures

Electron energy
Electron energy Ec Ec
µ Ed
ED ED
µ
0.5Eg
Ev Ev

N-type semiconductor
Electron energy

Electron energy
Ec Ec
0.5Eg
µ
EA EA
µ Ea
Ev Ev

P-type semiconductor 47
Saturation range (Ed, Ea < kBT < Eg)

N-type : n ≈ Nd >> p dominated by electrons 0

electrical conductivityσ ≈ N d e µe + p e µ h >0
Hall coefficient 1 1
RH ≈ − ≈−
ne Nd e

P-type : p ≈ Na >> n dominated by holes 0

electrical conductivity σ ≈ N a e µh + n e µe >0
Hall coefficient 1 1
RH ≈ ≈
p e Na e

48
 E.M. Conwell, Proc. IRE 40, 1327 (1952).

phonon

intrinsic

Data was obtained using the Hall effect. 49

E.M. Conwell, Proc. IRE 40, 1327 (1952).

50