Mark

 Lundstrom     2/10/2013  

SOLUTIONS:    ECE  606  Homework  Week  5    

Mark  Lundstrom  
Purdue  University  
(corrected  3/26/13)  
 
Some  of  the  problems  below  are  taken/adapted  from  Chapter  4  in  Advanced  
Semiconductor  Fundamentals,  2nd.  Ed.  By  R.F.  Pierret.  
 
 
1) The  conduction  band  minima  in  GaP  occurs  at  the  boundary  of  the  first  Brillouin  zone  
along  [100]  directions.    The  constant  energy  surfaces  are  ellipsoidal  with   m* = 1.12m0  
and   mt* = 0.22m0 .  Determine  the  density-­‐of-­‐states  effective  mass  for  electrons  in  GaP.  
 
Solution:  
The  situation  here  is  much  like  silicon  –  except  that  the  minima  occur  at  the  BZ  
boundary.    The  result  is  that  we  can  only  include  one-­‐half  of  each  ellipsoid  inside  the  
BZ.      The  “6”  in  enq.  (4.25a)  of  ASF  (p.  95)  is  replaced  by  3.    We  conclude:  
 
) ( )
1/3
mn* = 32/3 m*mt*2    
DOS
 
mn* ) DOS
= 0.79m0  
 
 
2) Assuming  that  the  Si  valence  bands  are  spherical  with   mhh *
= 0.537m0  and   mlh* = 0.153m0 ,  
determine  the  fraction  of  holes  that  are  in  the  heavy  hole  band.    You  should  assume  
non-­‐degenerate  conditions  at   T = 300 K.    Also  assume  that  the  effective  masses  quoted  
above,  which  were  measured  by  cyclotron  resonance  at  4K,  can  be  used  at  room  
temperature.  
 
Solution:  
 
Begin  with:  
 
3/2
1 ⎛ 2m*k BT ⎞
p = NV F 1/2 (η F )    m-­‐3     NV = ⎜ m-­‐3    where:     η F = ( EV − E F ) k BT    
4 ⎝ π  2 ⎟⎠
 
This  expression  assumes  a  single,  parabolic  energy  band  (i.e.  valley  degeneracy  is  1).  
 
For  the  heavy  and  light  hole  bands:  
 
phh = NVhhF 1/2 (η F ) plh = NVlhF 1/2 (η F )
   

ECE-­‐606     1   Spring  2013  

Mark  Lundstrom     2/10/2013  

 
The  fraction  of  holes  in  the  heavy  hole  band  becomes:  
 
( )
3/2
*
phh phh NVhh mhh
= = =    
phh + plh NVhh + NVlh
( ) ( )
3/2 3/2
p *
mhh + mlh*
 
Inserting  numbers,  we  find:  
 
phh 0.537 3/2
=  
p ( 0.537 )3/2 + ( 0.153)3/2
 
phh
= 0.868  
p
 
Most  of  the  holes  are  in  the  heavy  hole  band.    This  should  be  expected,  because  the  
heavy  hole  band  has  a  much  higher  density  of  states.  
 
( ) ( )
3/2 3/2
NVhh ∝ mhh
*
>> NVlh mlh*  
 
Note  that  we  did  NOT  need  to  assume  MB  (non-­‐degenerate)  statistics.  
 
 
3) In  a  Si  MOSFET,  the  source  and  drain  regions  are  heavily  doped.    Assuming  that  the  
dopants  are  fully  ionized  and  that   N D = 1020  cm-­‐3,  compute  the  location  of  the  Fermi  
level.    Compare  the  answer  you  get  assuming  Maxwell-­‐Boltzmann  statistics  with  the  
answer  from  Fermi-­‐Dirac  statistics.    You  may  assume  T  =  300K  and  that  the  effective  
density  of  states  is  
3/2
1 ⎛ 2m* k T ⎞
N C = ⎜ D 2B ⎟ = 3.23× 1019 cm -3  
4 ⎝ π ⎠
 
You  will  need  to  compute  an  inverse  Fermi-­‐Dirac  integral.    A  FD  integral  tool  is  
available  on  nanoHUB.org  at:    nanohub.org/resources/fdical  
 
You  can  also  find  an  iPhone  app  that  does  Fermi-­‐Dirac  integrals.  
 
Solution:  
 
( )
n = N CF 1/2 (η F ) = 3.23× 1019 F 1/2 (η F ) = 1020
 

ECE-­‐606     2   Spring  2013  

Mark  Lundstrom     2/10/2013  

1020
F 1/2 (η F ) = = 3.10
3.23× 1019  
η F = ⎡⎣F 1/2 ( 3.10 ) ⎤⎦
−1

 
 
Using  the  nanoHUB  tool  or  the  iPhone  app,  we  find:  
 
E − EC
η F = ⎡⎣F 1/2 ( 3.10 ) ⎤⎦ = 2.184 = F
−1

k BT  
 
So  the  Fermi  level  is  2.184  kBT  above  EC  for  FD  statistics.  
 
Now  repeat  for  Maxwell  Boltzmann  statistics:  
 
F 1/2 (η F ) = 3.10 ⇒ eηF
 
 
E − EC
η F = ln ( 3.10 ) = 1.13 = F
k BTL  
 
So  the  Fermi  level  is  1.13  kBT  above  EC  for  MB  statistics.  
 
In  general,  we  find  that  the  Fermi  level  is  higher  when  we  use  FD  statistics.    This  
occurs  because  as  states  are  filled,  we  need  to  go  higher  in  the  conduction  band  to  find  
empty  state.    The  implicit  assumption  for  MB  statistics  is  that  all  states  are  nearly  
empty.  
 
 
4) In  3D,  the  distribution  of  electrons  in  the  conduction  band,   n ( E ) ,  is  peaked  at  an  
energy,   Ê  that  is  very  near  the  bottom  of  the  conduction  band,   EC .    Assume  Maxwell-­‐
Boltzmann  carrier  statistics  and  derive  an  expression  for   Ê  for  the  following  cases:  
4a)   3D  semiconductor  with  parabolic  energy  bands  
4b)   2D  semiconductor  with  parabolic  energy  bands  
 
Solution:  
 
4a)  
n( E ) = D ( E ) f ( E ) ≈ D ( E ) e ( F )
− E− E k BT
   
 
 
dn ( E ) dD ( E ) −( E− EF ) k BT D ( E ) −( E− EF ) k BT
= e − e  
dE dE k BT

ECE-­‐606     3   Spring  2013  

Mark  Lundstrom     2/10/2013  

 
( ) = 0 → k T dD ( E )
dn Ê
= D Ê   ( )
B
dE dE
Ê

in  3D,   D ( E ) = K ( E − EC )
1/2
(K  is  a  constant),  so  
   
 
k BT
( ) ( )
−1/2 1/2
K Ê − EC = K Ê − EC  
2
 
k BT
Ê = EC +  
2
 
4b)  
 
Begin  with:  
 
dD ( E )
k BT
dE
= D Ê   ( )
Ê
 
in  2D,   D2 D ( E ) = K ,  so  
 
k BT × 0 = K    nonsense.    There  is  no  peak  to   n ( E ) .  
 
 
5) Sometimes  we  need  to  know  the  average  kinetic  energy  for  electrons  in  the  
conduction  band  (or  for  holes  in  the  valence  band).    Answer  the  follow  questions:  
 
5a)   Derive  a  general  expressions  for  the  average  kinetic  energy  per  electron  in  the  
conduction  band.  
 
Solution:  
∞

E
∫ ( E − E ) D ( E ) f ( E ) dE
EC
C

u= = ∞
   
n
∫ D ( E ) f ( E ) dE
EC

 
5b)   Assume  parabolic  energy  bands  and  evaluate  the  general  expression  for  a  1D  
semiconductor.    Simplify  your  answer  for  Maxwell-­‐Boltzmann  carrier  
statistics.  
 

ECE-­‐606     4   Spring  2013  

Mark  Lundstrom     2/10/2013  

Solution:  
∞

E
∫ ( E − E ) D ( E ) f ( E ) dE
EC
C 1D

u= = ∞
 
nL
∫ D ( E ) f ( E ) dE
EC
1D

 
g 2m*D
D1D (E) = V
π E − EC
 
 
You  should  be  able  to  show:  
∞

nL = ∫ D ( E ) f ( E ) dE = N
EC
1D
F −1/2 (η F )    (*),  
1D

 
where:     N1D = 2m*D k BT π 
.      
 
Now,  lets  work  on  the  numerator:  
 
⎧⎪ 1 2m*D ⎫⎪ ( E − EC ) dE  
∞ ∞ 1/2
1 1
E = ∫ ( E − EC ) ⎨ ⎬ dE = 2m*D ∫
π  E − EC ⎪ 1+ e( EF )E− k BT
π EC 1+ e
( E− EF ) k BT
EC ⎩⎪ ⎭
 
η ≡ ( E − EC ) k BT     dE = k BT η  
 
η F ≡ ( E F − EC ) k BT  
 
( k Tη ) k BTdη  
∞ 1/2
1
E = 2m*D ∫ B
π EC 1+ eη −ηF
 
 
∞
2m*D k BT η1/2 dη
E = k BT ∫  
π 0 1+ e
η −η F

 
∞
η1/2 dη π
∫0 1+ eη−ηF = 2 F 1/2 (η F )  (See  “Notes  on  FD  Integrals”)  
2m*D k BT π ⎛ k BT ⎞ ⎛k T⎞
E = ⎜ ⎟ F 1/2 (η F ) = N1D ⎜ B ⎟ F 1/2 (η F )  (**)  
 ⎝ 2 ⎠ ⎝ 2 ⎠

ECE-­‐606     5   Spring  2013  

Mark  Lundstrom     2/10/2013  

Combine  (*)  and  (**)  to  find:  

 
E ⎛ k T ⎞ F (η )
u1D = = ⎜ B ⎟ 1/2 F  
nL ⎝ 2 ⎠ F −1/2 (η F )
 
 
For  a  nondegenerate  semiconductor,  all  FD  integrals  reduce  to  exponentials,  so  we  
find:  
 
⎛k T⎞
u1D = ⎜ B ⎟  
⎝ 2 ⎠
 
 
5c)   Assume  circular,  parabolic    energy  bands  and  evaluate  the  general  expression  
for  a  2D  semiconductor.    Simplify  your  answer  for  Maxwell-­‐Boltzmann  carrier  
statistics.  
 
Solution:  
 
We  begin  the  same  way…  
 
∞

E
∫ ( E − E ) D ( E ) f ( E ) dE
EC
C 2D

u2 D = = ∞
 
nS
∫ D ( E ) f ( E ) dE
EC
2D

 
Use  the  2D  DOS:  
m*
D2 D (E) = D2
π   
 
and  eventually  find:  
 
E F 1 (η F )
u2 D = = ( k BT )  
nS F 0 (η F )
 
For  a  nondegenerate  semiconductor,  all  FD  integrals  reduce  to  exponentials  and  we  
find:  
 
u2 D = k BT  

ECE-­‐606     6   Spring  2013  

Mark  Lundstrom     2/10/2013  

 
 
5d)   Assume  spherical,  parabolic  energy  bands  and  evaluate  the  general  
expression  for  a  3D  semiconductor.    Simplify  your  answer  for  Maxwell-­‐
Boltzmann  carrier  statistics.  
 
Solution:  
 
We  begin  the  same  way…  
 
∞

E
∫ ( E − E ) D ( E ) f ( E ) dE
EC
C 3D

u3D = = ∞
 
n
∫ D ( E ) f ( E ) dE
EC
3D

 
Use  the  3D  DOS:  
 

D3D ( E) =
( m ) 2( E − E )
3/2
D C

π  2 3
 
 
and  eventually  find:  
 
E ⎛ 3k T ⎞ F (η )
u3D = = ⎜ B ⎟ 3/2 F  
n ⎝ 2 ⎠ F 1/2 (η F )
 
For  a  nondegenerate  semiconductor,  all  FD  integrals  reduce  to  exponentials  and  we  
find:  
 
3k BT
u3D =  
2
 
6) Assume  Si  at  T  =  300  K,  doped  with  arsenic  at   N D = 1016  cm-­‐3.    Make  reasonable  
assumptions  and  answer  the  following  questions.  
 
6a)    Compute  the  density  of  electrons  in  the  conduction  band.  
6b)    Compute  the  location  of  the  Fermi  level.  
6c)      Compute  the  fraction  of  the  dopants  that  are  ionized.  
6d)    Compute  the  density  of  holes  in  the  valence  band.  
 
 
Solution:  

ECE-­‐606     7   Spring  2013  

Mark  Lundstrom     2/10/2013  

 
For  problems  6)  -­‐  10),  let’s  assume    at  T  =  300  K  (from  Pierret,  ASF,  2nd  Ed.,  p.  113)  
 
3/2
1 ⎛ 2mn* k BT ⎞
NC = ⎜ = 3.23× 1019 cm -3
4 ⎝ π  2 ⎟⎠
 
3/2
1 ⎛ 2m k T ⎞
*
p B
NV = ⎜ ⎟ = 1.83× 1019 cm -3
4 ⎝ π ⎠ 2
 
It  should  be  understood  that  the  effective  masses  in  these  expression  are  the  
density-­‐of-­‐states  effective  masses.  
 
For  problems  6)  –  10),  the  starting  point  is  space  charge  neutrality:  
 
p − n + N D+ − N A− = 0
   
⎛ E − EF ⎞ ⎛ E − EC ⎞ ND NA
NV F 1/2 ⎜ V ⎟ − N C F 1/2 ⎜ F ⎟ + ( EF −ED ) kBT
− =0
⎝ k BT ⎠ ⎝ k BT ⎠ 1+ gD e 1+ gA e( EA −EF ) kBT
 
 
 
For  this  problem,  we  can  assume  that   N A = 0 .    Since  the  doping  is  moderate,  we  
assume  non-­‐degenerate  carrier  statistics.    Because  Arsenic  is  a  shallow  dopant  and  
we  expect  the  Fermi  level  to  be  well  below  ED,  we  can  assume  that  the  dopants  are  
fully  ionized.    Because  the  semiconductor  is  extrinsic,   n >> p .    We  conclude  that  
n = N D  to  a  very  good  approximation.    All  these  assumptions  can  be  checked  as  we  
proceed.  
 
6a)     n = N D = 1016 cm -3    
 
6b)    Determine  EF  from  the  electron  density:  
 
n = N C e( → E F − EC = k BT ln ( n N C )    
E F − EC ) k BT

 
(
E F = EC + k BT ln 1016 3.23× 1019 = EC − 8.1k BT   ) E F = EC − 8.1k BT  
 
So  the  assumptions  that  the  semiconductor  is  non-­‐degenerate  is  good.  
 
6c)    Begin  with:  
 

ECE-­‐606     8   Spring  2013  

Mark  Lundstrom     2/10/2013  

N D+ 1 1
= ( EF −ED ) kBT
= ( EF −EC +EC −ED ) kBT
 
N D 1+ gD e 1+ gD e
 
According  to  Fig.  4.14,  p.  111  of  Pierret,  ASF:     EC − E D = 0.054 eV = 2.08k BT    and  
we  also  know  that  for  donors,   g D = 2 .  
 
N D+ 1 1 N D+
= ( −8.1kBT +2.08 kBT ) kBT
= = 0.995 = 0.995
N D 1+ 2e 1+ 2e−6.02 ND
   
 
So  our  assumptions  that  the  dopants  are  fully  ionized  is  good.  
 
6d)    Use  the  Law  of  Mass  action: np = ni2    
 
( )
2
ni2 1010
p= = 16
= 104       p = 104 cm -3  
n 10
 
So  our  assumption  that   n >> p  is  good.  
 
 
 
7) Assume  Si  at  T  =  640  K,  doped  with  arsenic  at   N D = 1016  cm-­‐3.    Make  reasonable  
assumptions  and  answer  the  following  questions.  
 
7a)    Compute  the  density  of  electrons  in  the  conduction  band.  
7b)    Compute  the  location  of  the  Fermi  level.  
7c)      Compute  the  fraction  of  the  dopants  that  are  ionized.  
7d)    Compute  the  density  of  holes  in  the  valence  band.  
 
 
 
Solution:  
 
Dopants  should  be  fully  ionized  and  semiconductor  should  be  nondegenerate,  but  
ni  may  be  large,  which  means  that  p  may  not  be  negligible.  
 
From  Fig.  4.17,  p.  117  of  Pierret,  ASF  we  find:  
 
ni ( 640 K ) ≈ 1016 cm -3  
 
The  space  charge  neutrality  condition  becomes:  

ECE-­‐606     9   Spring  2013  

Mark  Lundstrom     2/10/2013  

 
p − n + N D+ − N A− = 0 = p − n + N D = ni2 n − n + N D = 0  
 
which  can  be  solved  for    
 
1/2
N ⎡⎛ N ⎞ 2 ⎤
n = D + ⎢⎜ D ⎟ + ni2 ⎥  
2 ⎣⎝ 2 ⎠ ⎦
 
7a)    Solve  for  n:  
 
1/2
N D ⎡⎛ N D ⎞ ⎤
2

= 5 × 1015 + ⎡( 5 × 1015 ) + (1016 ) ⎤  

2 2 1/2
n= + ⎢⎜ ⎟ + ni2 ⎥
2 ⎣⎝ 2 ⎠ ⎣ ⎦
⎦
 
n = 5 × 1015 + 1.12 × 1016     n = 1.62 × 1016 cm -3  
 
7b)    Determine  EF  from  the  electron  density:  
 
n = N C e( → E F − EC = k BT ln ( n N C )  but  NC  is  temperature  dependent.  
E F − EC ) k BT

 
3/2 3/2 3/2
1 ⎛ 2mn* k BT ⎞ ⎛ T ⎞ ⎛ 640 ⎞
NC = ⎜ = N C ( 300K ) ⎜ = 3.23× 10 ⎜ 19
= 1.01× 1020 cm -3
4 ⎝ π  2 ⎟⎠ ⎝ 300 ⎟⎠ ⎝ 300 ⎟⎠
 
 
(
E F − EC = k BT ln 1.62 × 1016 1.01× 1020 = −8.74k BT )  
 
E F = EC − 8.74k BT
   (but  realize  that  kBT  is  bigger  now).  
 
7c)    Proceeding  as  in  6c),  we  find:  
 
EC − E D = 0.054 eV = 0.98k BT    and
 
 
 
N D+ 1 1 N D+
= ( −8.74 kBT +0.98 kBT ) kBT
= = 0.999 = 0.999
N D 1+ 2e 1+ 2e−6.02 ND
   
 
Dopants  are  fully  ionized  as  assumed.  
 
7d)    Use  the  Law  of  Mass  Action: np = ni2    
 

ECE-­‐606     10   Spring  2013  

Mark  Lundstrom     2/10/2013  

( )
2
ni2 1016
p= = = 6.2 × 1015       p = 6.2 × 1015 cm -3  
n 1.62 × 1016
 
So  in  this  case,  the  hole  concentration  is  not  negligible.    The  semiconductor  is  
nearly  intrinsic.  
 
 
8) Assume  Si  at  T  =  77  K,  doped  with  arsenic  at   N D = 1016  cm-­‐3.    Make  reasonable  
assumptions  and  answer  the  following  questions.  
 
8a)    Compute  the  density  of  electrons  in  the  conduction  band.  
8b)    Compute  the  location  of  the  Fermi  level.  
8c)      Compute  the  fraction  of  the  dopants  that  are  ionized.  
8d)    Compute  the  density  of  holes  in  the  valence  band.  
 
Solution:  
In  this  case,  we  worry  that  there  may  not  be  enough  thermal  energy  to  ionize  the  
donors.    The  intrinsic  concentration  should  be  very  small,  so  holes  can  be  ignored.    
The  space  charge  neutrality  condition  becomes:  
 
p − n + N D+ − N A− = 0 = −n + N D+  
⎛ E − EC ⎞ ND
−N C F 1/2 ⎜ F ⎟ + =0
⎝ k BT ⎠ 1+ gD e( EF −ED ) kBT
 
Assume  non-­‐degenerate  carrier  statistics  (check  later):  
 
⎛ E − EC ⎞ ND
−N C exp ⎜ F ⎟ + =0
⎝ k BT ⎠ 1+ gD e( EF −ED ) kBT
 
 
This  equation  can  be  solved.    The  answer  is  eqn.  (4.88)  in  Pierret,  ASF  
 

{(
EF = EC + kBT ln Nζ 2N C ⎡⎢ 1+ 4N D Nζ
⎣ )( )
1/2
}  
− 1⎤⎥
⎦
where  
 
Nζ = ( N C g D ) e ( C D )
− E −E k BT

   
 
Now  lets  put  numbers  in:  
 

ECE-­‐606     11   Spring  2013  

Mark  Lundstrom     2/10/2013  

3/2 3/2
⎛ T ⎞ ⎛ 77 ⎞
N C = N C ( 300K ) ⎜ = 3.23× 10 ⎜ 19
= 4.20 × 1018 cm -3
⎝ 300 ⎟⎠ ⎝ 300 ⎟⎠
 
 
k BT = 6.64 × 10−3 eV    
 
Nζ = ( N C g D ) e ( C D )
− E −E k BT
(
= 4.20 × 1018 2 e ( ) )
− 54 6.64
= 6.17 × 1014  
 

{
EF = EC + kBT ln ( 6.17 × 1014 2 × 4.20 × 1018 ) ⎡(1+ 4 × 1016 6.17 × 1014 ) − 1⎤  
⎣
1/2

⎦ }
 
{
EF = EC + k BT ln ( 7.35 × 10 −5 ) ⎡⎣( 65.8 ) − 1⎤⎦  
1/2
} E F = EC − 7.65k BT    
 
 
Note  that   E D = EC − 8.13k BT    
 
So  the  Fermi  level  is  above  the  donor  level,  but  far  enough  below  EC  that  we  can  
use  non-­‐degenerate  carrier  statistics.  
 
8a)    Since  we  know  the  Fermi  level,  we  can  get  the  electron  density:  
 
⎛ E − EC ⎞
n = N C exp ⎜ F ⎟ = 4.20 × 1018 exp ( −7.65 ) n = 2.00 × 1015 cm -3  
⎝ kBT ⎠
   
8b)    We  already  solved  this  problem.  
 
E F = EC − 7.65k BT  
 
8c)   Since   n = N D+ = 2.00 × 1015 ,  we  deduce  
 
N D+
  = 0.2    
ND
We  could  also  get  this  directly  from  
N D+ 1
=  
N D 1+ gD e EF −ED ) kBT
(

 
8d)    The  number  of  holes  is  essentially  zero.    We  could  get  this  from  the  law  of  mass  
action,  but  then  we  need  to  figure  out  ni  at  77  K.    Another  way  is:  
 

ECE-­‐606     12   Spring  2013  

Mark  Lundstrom     2/10/2013  

⎛ E − EF ⎞ ⎛ E − EC + EC − EF ⎞ ⎛ E ⎞ ⎛ E − EF ⎞
p = NV exp ⎜ V ⎟ = NV exp ⎜ V ⎟ = NV exp ⎜ − G ⎟ exp ⎜ C
⎝ kBT ⎠ ⎝ kBT ⎠ ⎝ kBT ⎠ ⎝ kBT ⎟⎠
 
3/2
1 ⎛ 2m k T ⎞
* 3/2 3/2
⎛ T ⎞ ⎛ 77 ⎞
⎟ = NV ( 300K ) ⎜
p B
NV = ⎜ ⎟ = 1.83× 1019 ⎜ = 2.38 × 1018 cm -3
4 ⎝ π ⎠ 2
⎝ 300 ⎠ ⎝ 300 ⎟⎠
 
 
Also  need  the  bandgap  of  Si  at  77  K.  From  pp.  82-­‐83  of  Pierret,  ASF,  we  have  
 
αT 2
EG (T ) = EG ( 0 ) −    
T +β
 
For  Si:  
 
4.730 × 10−4 ( 77 )
2

EG (T ) = 1.170 − = 1.166 eV  
77 + 636
 
so  
 
⎛ 1166 ⎞
p = 2.38 × 1018 exp ⎜ − ⎟ exp ( 7.65 ) = 2.38 × 1018 ( 5.46 × 10 77 ) ( 2.10 × 10 3 )  
⎝ 6.64 ⎠
 
p = 2.73 × 10 −55 cm -3  
 
This  is  essentially  zero…  hardly  worth  the  effort  of  computing  it.  
 
Since  we  have  n  and  p,  we  can  deduce   ni = 2.34 × 10 −20 cm -3  
 
 
 
 
9) Assume  Si  at  T  =  300  K,  doped  with  arsenic  at   N D = 1018  cm-­‐3.    Make  reasonable  
assumptions  and  answer  the  following  questions.  
 
9a)    Compute  the  density  of  electrons  in  the  conduction  band.  
9b)    Compute  the  location  of  the  Fermi  level.  
9c)      Compute  the  fraction  of  the  dopants  that  are  ionized.  
9d)    Compute  the  density  of  holes  in  the  valence  band.  
 
 
 
 
Solution:  

ECE-­‐606     13   Spring  2013  

Mark  Lundstrom     2/10/2013  

 
This  requires  a  little  thought…  the  temperature  is  high  enough  so  the  donors  may  
be  fully  ionized,  but  the  number  of  dopants  is  high,  so  the  Fermi  level  might  be  too  
close  to  the  donor  level  to  assume  complete  ionization.    It  might  also  be  too  close  to  
the  conduction  band  to  assume  non-­‐degenerate  carrier  statistics.    Let  assume  full  
ionization  and  nondegenerate  carrier  statistics  and  see  where  it  leads.  
 
 
9a)    Assume  complete  ionization:     n = N D = 1018 cm -3    
 
9b)    Determine  EF  from  the  electron  density:  
 
n = N C e( → E F − EC = k BT ln ( n N C )    
E F − EC ) k BT

 
( )
E F = EC + k BT ln 1018 3.23× 1019 = EC − 3.48k BT   E F = EC − 3.48k BT  
 
So  the  assumptions  that  the  semiconductor  is  non-­‐degenerate  is  OK.  
 
9c)    As  in  problem  6c):  
 
N D+ 1 1 N D+
= ( −3.48 kBT +2.08 kBT ) kBT
= = 0.67 = 0.67
N D 1+ 2e 1+ 2e−1.40 ND
   
 
So  our  assumptions  that  the  dopants  are  fully  ionized  is  not  justified.    We  need  to  
start  over  and  solve  the  problem  using  the  procedure  of  problem  8).  
 
The  rest  of  this  problem  is  left  as  an  exercise  for  the  interested  student!  
 
 
10)    Assume  Si  at  T  =  300  K,  doped  with  arsenic  at   N D = 1020  cm-­‐3.    Make  reasonable  
assumptions  and  answer  the  following  questions.    (In  this  case,  you  should  assume  
that  “heavy  doping  effects”  cause  the  dopants  to  be  fully  ionized  and  the  bandgap  to  
shrink  by  0.1  eV.)  
 
10a)    Compute  the  density  of  electrons  in  the  conduction  band.  
10b)    Compute  the  location  of  the  Fermi  level.  
10c)    Compute  the  fraction  of  the  dopants  that  are  ionized.  
10d)    Compute  the  density  of  holes  in  the  valence  band.    
 
 
 
Solution:  

ECE-­‐606     14   Spring  2013  

Mark  Lundstrom     2/10/2013  

 
We  expect  the  Fermi  level  to  be  inside  the  conduction  band.    We  can  assume  full  ionization  
and  ignore  holes,  but  we  cannot  use  nondegenerate  statistics.  
 
n = N D  to  a  very  good  approximation.  
 
10a)     n = N D = 1020 cm -3    
 
10b)    Determine  EF  from  the  electron  density:  
 
⎛ E − EC ⎞
n = N C F 1/2 ⎜ F ⎟     N C = 3.23× 1019 cm -3  
⎝ k BT ⎠
 
−1 −1
EF − EC ⎡ ⎛ n ⎞⎤ ⎡ ⎛ 10 20 ⎞ ⎤
= ⎡⎣F 1/2 ( 3.10 ) ⎤⎦ = 2.184
−1
= ⎢F 1/2 ⎜ ⎟ ⎥ = F
⎢ 1/2 ⎜ 19 ⎟ ⎥
kBT ⎣ ⎝ NC ⎠ ⎦ ⎣ ⎝ 3.23 × 10 ⎠ ⎦
 
 
(from  Fermi  Dirac  Integral  Calculator  on  nanoHUB.org    
https://nanohub.org/tools/fdical)  
 
E F = EC + 2.184k BT  
 
So  the  semiconductor  is,  indeed  degenerate  with  EF  inside  the  conduction  band.  
 
 
10c)    By  assumption,  we  are  above  the  metal-­‐insulator  transition  and  
 
N D+
= 1  
ND
 
10d)    Use  the  Law  of  Mass  Action: np = ni2      But  bandgap  narrowing  increases  ni.  
 
( )
ni2 = 1020 exp ΔEG / k BT = 1020 exp (100 / 26 ) = 4.68 × 1021  
 
ni = 6.84 × 1010  

p=
ni2
=
(
4.68 × 1021 )
= 4.68 × 103       p = 4.68 × 103 cm -3  
18
n 10
 
So  our  assumption  that   n >> p  is  good.  

ECE-­‐606     15   Spring  2013  

Mark  Lundstrom     2/10/2013  

 
11)    For  the  energy  band  sketched  below,  provide  sketches  of  the  following:  
 

 
 
11a)    the  carrier  densities,  n(x),  and  p(x)  vs.  position.  
 
Solution:  
 

     
 
 
11b)    the  electrostatic  potential,   ψ ( x ) ,  vs.  position.  
 
Solution:  

ECE-­‐606     16   Spring  2013  

Mark  Lundstrom     2/10/2013  

     
 
11c)    the  electric  field  E  vs.  position.  
 

       
 
11d)    the  space  charge  density,   ρ ( x )    vs.  position  
 
Solution:  

ECE-­‐606     17   Spring  2013