Professional Documents
Culture Documents
Week2HW Solutions26
Week2HW Solutions26
nS = ∫ D2 D ( E ) f0 ( E ) dE = nS = ∫ D ( E )(1) dE
2D
(T = 0 K)
EC EC
EF
m* m*
nS = ∫ gV
π !2
dE = 2 ( E − EC )
π !2 F
EC
m* 2m*
nS = gV ( E − EC ) = π !2 ( EF − EC )
π !2 F
Case ii): graphene: DOS:
E 2E
D2 D ( E ) = gV = 2 2 ( E > 0) (valley degeneracy is 2 for graphene)
π υF π υF
2 2
Aside: Note that if we define the effective mass of graphene by E ≡ m*υ F2 then
we could use the parabolic band DOS and get the right DOS for graphene!
∞ EF
nS = ∫ D2 D ( E ) f0 ( E ) dE = nS = ∫ D ( E )(1) dE
2D
(T = 0 K)
0 0
ECE 656 Homework 2: (Week 2) (continued)
EF
2E E F2
nS = ∫
0 π 2υ F2
dE =
π 2υ F2
E F2
nS = 2 2
π υF
1b) In terms of the Fermi wave vector, kF :
Case i): parabolic energy bands: DOS:
!2k 2
= E − EC ( E > EC )
2m*
! 2 k F2
= E F − EC = E F ( E > EC )
2m*
2 2
2m* 2m* ! k F k F2
nS = ( E − EC ) → nS = π !2 × 2m* = gV × 2π
π !2 F
k F2 k F2
nS = gV =
2π π
Case ii): graphene: DOS:
E = υ F k ( E > 0 )
E F = υ F k F ( E > 0)
F
( υ k )
2
E2 k F2
nS = 2F 2 → F F
=
π υF π υ 2 2
F
π
k F2
nS = (same as for parabolic energy bands)
π
Aside: Why are the two expressions the same?
At T = 0 K, all of the states with k < k F are occupied and all for k > k F are empty.
The area of occupied k-space is π k F2
In 2D, each state occupies an area in k-space of ( 2π )
2
A
So the number of states occupied is:
where the factor of 2 is for spin degeneracy and gV is the valley degeneracy. The sheet
carrier density is
N k F2 k F2
nS = = × gV = (because valley degeneracy is 2 for both cases consider here.)
A 2π π
Working in k-space, we get the same answer for the two different band structures,
because the DOS in k-space does not depend on band structure.
2) Assume a finite temperature and work out the sheet carrier densities, nS , for:
2a) Electrons in the conduction band of a 2D parabolic band semiconductor
2b) Electrons in the conduction band (E > 0) of graphene.
Your answers to these questions should be in terms of material parameters and the Fermi
level.
Solution:
2a) parabolic energy bands
∞ ∞
⎛ m* ⎞ 1
nS = ∫ D2 D ( E ) f0 ( E ) dE = ∫ ⎜ gV 2⎟ ( E− E F ) k B T
dE
EC EC ⎝ π ⎠ 1+ e
⎛ m* ⎞
∞
1 ⎛ m* ⎞
∞
1
2⎟ ∫ 2⎟ ∫
nS = ⎜ gV ( E− E F ) k B T
dE = ⎜ gV ( E− EC + EC − E F ) k B T
dE
⎝ π ⎠ EC 1+ e ⎝ π ⎠ EC 1+ e
define:
E − EC E − EC dE
ηF = F η= dη = dE = k BTdη
k BT k BT k BT
with this change of variables, we find:
⎛ m* ⎞ k BTdη
∞
2 ∫
nS = ⎜ gV
⎝ π ⎟⎠ 0 1+ e F
η −η
the integral can be done analytically:
∫ 1+ e
dη
η −η F (
= ln 1+ eηF )
0
but we also recognize it as a Fermi-Dirac integral of order 0:
(
F 0 (η F ) = ln 1+ eηF )
so the answer is:
nS = N 2 DF 0 (η F )
m*k BT
N2D = gv
π 2
2b) graphene
∞ ∞
⎛ 2E ⎞ 1
nS = ∫E 2 D ( ) 0 ( )
D E f E dE = ∫0 ⎜⎝ π 2υ 2 ⎟⎠ 1+ e( E− EF ) kB T
dE
C F
define:
EF E dE
ηF = η= E = k BT η dη = dE = k BTdη
k BT k BT k BT
2
⎛ 2 ⎞∞ E ⎛ 2 ⎞ ∞ k BT η 2⎛ k T⎞
∞
η dη
nS = ⎜ 2 2 ⎟ ∫ ( E− E F ) k B T
dE = ⎜ 2 2 ⎟ ∫ k Tdη = ⎜ B ⎟ ∫ 1+ e
⎝ π υ F ⎠ 0 1+ e ⎝ π υ F ⎠ 0 1+ e
η −η F B
π ⎝ υ F ⎠ 0
η −η F
The integral is recognized as a Fermi-Dirac integral of order 1:
2 2
2⎛ k T⎞ 2⎛ k T⎞
∞
η dη
nS = ⎜ B ⎟
π ⎝ υ F ⎠ ∫0 1+ eη−ηF = π ⎜⎝ υB ⎟⎠ F 1 (η F )
F
nS = N 2 DF 1 (η F )
2
2 ⎛ k BT ⎞
N2D = ⎜
π ⎝ υ F ⎟⎠
∑ x 0
k x >0,k y
υ k f()
( E ) ∫0 −π∫/2 k dk dθ υ x f0 num
υx =
+
= ∞ + π /2 = (DOS in k-space cancels from num
∑ f0 ( E ) den
k x >0,k y ∫ ∫ k dk dθ f0
0 − π /2
and denom)
∞ + π /2 ∞ + π /2
k
num = ∫ ∫ k dk dθ υ x f0 = ∫ ∫ k dk dθ cosθ f0
0 − π /2 0 − π /2 m*
f0 = 1 k < k F ( E < E F )
⎛ k F3 ⎞
k F + π /2 kF + π /2
k
num = ∫ ∫ k dk dθ * cosθ = * ∫ k dk
2
∫ cosθ dθ = ×2
0 − π /2 m m 0 − π /2 m* ⎜⎝ 3 ⎟⎠
⎛ 2 ⎞ ⎛ k F3 ⎞
num = ⎜ ⎟ ⎜ * ⎟ (*)
⎝ 3⎠ ⎝ m ⎠
∞ + π /2 k F + π /2 kF + π /2
k F2
denom = ∫ ∫ k dk dθ f0 = ∫ ∫ k dk dθ = ∫ k dk ∫ dθ = × π (**)
0 − π /2 0 − π /2 0 − π /2
2
Using (*) and (**)
⎛ 2 ⎞ ⎛ k F3 ⎞
⎜⎝ 3 ⎟⎠ ⎜ m* ⎟
num ⎝ ⎠ 4 k F 4
υ x+ = = = = υF
denom π kF 2
2
3π m *
3π
k x >0,k y
υ k f()
∑ x 0 ( ) ∫ ∫ k dk dθ υ x f0
E
num
υx =
+
= 0 ∞− π+/2π /2 =
∑ f0 ( E ) den
k x >0,k y ∫ ∫ k dk dθ f0 0 − π /2
∞ + π /2 ∞ + π /2
num = ∫ ∫ k dk dθ υ x f0 = ∫ ∫ k dk dθ υ F cosθ f0
0 − π /2 0 − π /2
f0 = 1 k < k F ( E < E F )
k F + π /2
⎛ k F2 ⎞
kF + π /2
4b) For this bandstructure, derive an expression for the velocity, υ x ( kx ) as a function of
kx .
Solution:
Velocity is related to bandstructure by:
1 dE
υx =
dk x
kx 1
υx =
m ( 0 ) (1+ 2α E )
*
alternatively, we could define an energy dependent effective mass by:
m* ( E ) = m* ( 0 ) (1+ 2α E )
and write the velocity as
k
υx = * x
m (E)
5) For parabolic energy bands, the 2D density of states is
m*
D2 D ( E ) = g v Θ ( E − ε1 ) .
π !2
Assume a non-parabolic band described by the so-called Kane dispersion,
2k 2
() ()
E k ⎡⎣1 + α E k ⎤⎦ =
2m* 0 ( ) ,
and derive the density of states.
Note that N ( k ) dk x dk y is the number of states per unit area. Now map these states
onto energy:
1 1
D2 D ( E ) dE = g v N ( k ) dk = kdk
A π
Note that by convention, D2 D ( E ) dE is the number of states per unit area. Solve for
D2 D ( E ) :
1 dk 1 dk
D2 D ( E ) = g v N (k ) = gv k (*)
A dE π dE
The bandstructure is:
2k 2
E + α E2 =
2m* ( 0 )
Differentiate with respect to k:
(
d E + α E2
=
dE )(1+ 2α E )
dk dk
from which we find
2 dk
1+ 2α E = * k
m ( 0 ) dE
dk m ( 0 )
*
k
dE
=
2
(1+ 2α E ) (**)
Now insert (**) into (*) to find:
m* ( 0 )
D2 D ( E ) = g v(1+ 2α E )
π !2
So the 2D DOS increases linearly with energy when we take conduction band non-
parabolicity into account.
which looks like a simple, circular band in 2D. But we must realize that states are
spaced ( 2π L ) in k , but they are spaced ( 2π L ) × m m in k . Accordingly, we
x x x 0 xx x
find
A mxx myy A mxx myy mxx myy
N ( k ) dk x dk y = ×2× dkx dky = k = A
2π kd k
kd
( 2π ) 2π 2 π
2
m0 m0 m0
1 1 mxx myy
D2 D ( E ) dE = N ( k ) dk x dk y = k dk
A π m0
or, solving for the DOS:
1 mxx myy dk
D2 D ( E ) = k (**)
π m0 dE
Now use (*) to find:
dk m0
k =
dE 2
and insert this in (**) to find
m mxx myy mxx myy
D2 D ( E ) = gV 02 = gV
π m0 π 2
For the primed valleys of Si, mxx* = m* is the longitudinal effective mass and m*yy = mt* is
the transverse effective mass. The valley degeneracy of the primed valleys is g v = 4 .
Accordingly, for the primed valleys (of (100) Si), we find
mxx myy m*mt*
D2′ D ( E ) = gV =4 ,
π 2 π 2
specific numbers: unprimed valleys:
2 n 2π 2 ( 0.042 ) 2
εn = = * n (eV)
2m2t Si2 m m0
ε1 = 0.046 ε 2 = 0.185 ε 3 = 0.415 ε 4 = 0.739
mt* m
D2 D = gv 2 = 0.38 02
π π
m0
= 4.2 × 1014 (eV-cm2)-1
π 2
specific numbers primed valleys:
2 m 2π 2 ( 0.042 ) 2
ε m′ = = * m (eV)
2mt2t Si2 mt m0
ε1′ = 0.221 ε 2′ = 0.884 ε 3′ = 1.989 ε 4′ = 3.54
Note that these are unreasonably large energies (especially for m > 1) – due to
the assumption of infinite barriers and parabolic energy bands.
mt*m* m
D2 D = gv = 1.66 02
π 2
π