Mark

 Lundstrom     09/11/2014  

SOLUTIONS:    ECE  305  Homework:  Week  3  

 
Mark  Lundstrom  
Purdue  University  
(9/17/14)  
 
 
1) Assume  Silicon  (bandgap  1.12  eV)  at  room  temperature  (300  K)  with  the  Fermi  level  
located  exactly  in  the  middle  of  the  bandgap.    Answer  the  following  questions.  
 
a) What  is  the  probability  that  a  state  located  at  the  bottom  of  the  conduction  band  is  
filled?  
 
b) What  is  the  probability  that  a  state  located  at  the  top  of  the  valence  band  is  empty?  
 
Solution:  
1
1a)    Begin  with  the  Fermi  function:   f ( E ) = ( E− EF ) k BT    
1+ e
For  our  problem:   E = EC    
1
f ( EC ) = ≈ e ( C F)
− E −E k BT
( EC − EF ) k BT
 
1+ e
 
The  approximation  is  extremely  good,  because  the  bottom  of  the  conduction  band  is  
very  far  above  the  Fermi  level.  The  semiconductor  is  said  to  be  nondegenerate.  
 
E + EV
EF = C    (The  Fermi  level  is  exactly  in  the  middle  of  the  bandgap.)  
2
E + EV EC − EV EG
EC − E F = EC − C = =  
2 2 2
 
so  the  probability  is  
f ( EC ) ≈ e ( C F )
− E −E k BT − EG 2k BT
=e  
EG = 1.12 eV     k BT = 0.026 eV   EG 2k BT = 21.5  

f ( EC ) = e− EG 2k BT
= e−21.5 = 4.43× 10−10  
 
What  if  we  had  not  made  the  non-­‐degenerate  approximation?    Then  
1 1 1
f ( EC ) = =
( EC − EF ) k BT 1+ e G B = = 4.43× 10−10  
1+ e
E 2k T
1+ 2.26 × 10 9

 
Same  answer  as  with  the  non-­‐degenerate  assumption.  
 

ECE-­‐305     1   Fall  2014  

Mark  Lundstrom     09/11/2014  

HW3  Solutions  (continued):  

1
1b)  Begin  again  with  the  Fermi  function:   f ( E ) = ( E− EF ) k BT    
1+ e
For  this  part:     E = EV    
1
f ( EV ) = ( EV − EF ) k BT
 
1+ e
 
The  Fermi  function  gives  the  probability  that  a  state  is  full.    We  want  the  probability  
that  the  state  is  empty,  which  is    
 
e(
EV − E F ) k BT
1 1
1− f ( EV ) = 1− = =  
1+ e(
EV − E F ) k BT
1+ e(
EV − E F ) k BT
e ( V F)
− E −E k BT
+1
1
1− f ( EV ) = ≈ e ( F V)
− E −E k BT
( EF − EV ) k BT
 
1+ e
 
The  approximation  is  extremely  good,  because  the  Fermi  level  is  very  far  above  the  
top  of  the  valence  band.    The  nondegenerate  approximation  can  be  used  for  the  
valence  band  too.  
 
E + EV
EF = C    (The  Fermi  level  is  exactly  in  the  middle  of  the  bandgap.)  
2
E + EV E − EV EG
E F − EV = C − EV = C =  
2 2 2
so  the  probability  is  
1− f ( EV ) = e ( F V )
− E −E
= e− EG
k BT k BT
 

1− f ( EV ) = e− EG 2k BT
= 4.43× 10−10  
 
 
2) For  Si  at  room  temperature,  calculate  the  following  quantities.  (Numerical  answers  
required,  and  don’t  forget  to  include  units  with  your  answer).  
 
a) The  density  of  states  in  the  conduction  band, gC ( E ) ,  at  an  energy  26  meV  above   EC .  
b) The  density  of  states  in  the  valence  band,   gV ( E ) ,  at  an  energy  26  meV  below   EV .  
c) The  effective  density  of  conduction  band  states. N C .  
d) The  effective  density  of  valence  band  states,   NV .  
e) Compute  the  ratio  of  the  effective  density  of  conduction  band  states  to  the  atomic  
density  of  Si.  
 
 

ECE-­‐305     2   Fall  2014  

Mark  Lundstrom     09/11/2014  

HW3  Solutions  (continued):  

 
Solution:  
 
2a)    Begin  with  the  DOS  expression  (SDF,  eqn.(2.6a),  p.  41)  
 
mn* 2mn* ( E − EC )
gC ( E ) =     E > EC    
π 2!3
 
3/2
m 2m ⎛ m* ⎞
gC ( E ) = 0 2 3 0 ⎜ n ⎟ E − EC  
π ! ⎝ m0 ⎠
 
3/2
9.11× 10−31 2 × 9.11× 10−31 ⎛ mn* ⎞
gC ( E ) = ⎜m ⎟ E − EC    (everything  in  MKS  (SI)  units)  
( )
3
3.142 × 1.06 × 10−34 ⎝ 0⎠
 
3/2
⎛ m* ⎞
gC ( E ) = 1.05 × 10 ⎜ n ⎟
56
E − EC  
⎝ m0 ⎠
 
According  to  SDF,  p.  34,   mn* m0 = 1.18  
gC ( E ) = 1.05 × 1056 (1.18) ( )
3/2 −1
E − EC = 1.35 × 1056 E − EC J-m 3  
 
Note  the  units.    This  is  the  number  of  states  per  unit  energy  (in  Joules)  per  unit  
volume  (in  cubic  meters).  
 
We  are  asked  for  the  density  of  states  26  meV  (milli  electron  volts)  above  the  
bottom  of  the  conduction  band.    In  Joules:  
E − EC = 0.026 × 1.60 × 10−19 = 4.16 × 10−21 J    
   
1 1
gC ( 0.026 eV + EC ) = 1.35 × 1056 4.16 × 10−21 3
= 8.71× 1045  
J-m J-m 3
The  answer  is  correct,  but  the  units  are  inconvenient.    Let’s  express  the  answer  per  
eV  per  cubic  cm,  which  is  the  more  typical  way  of  doing  things  for  semiconductors.      
 
3
1 1.6 × 10−19 J 1 ⎛ 10−2 m ⎞
gC ( 0.026 eV + EC ) = 8.71× 10 × ×
45
 
J eV m 3 ⎜⎝ cm ⎟⎠
1
gC ( 0.026 eV + EC ) = 1.39 × 1021  
eV- cm 3

ECE-­‐305     3   Fall  2014  

Mark  Lundstrom     09/11/2014  

HW3  Solutions  (continued):  

Note:    The  total  number  of  states  in  a  range  of  energy  of   k BT = 0.026 eV  is  
approximately  
1
gC ( 0.026 eV + EC ) × 0.026 = 3.62 × 1019 ,  
cm 3
 
which  is  a  reasonable  number,  as  we  shall  see  below.    This  answer  is  approximate,  
because  to  get  the  total  number  of  states  in  the  bottom   k BT  of  the  conduction  band,  
we  should  do  the  integral  
 
gC ( E ) dE  .  
EC +k BT
N TOT = ∫
EC

 
A  note  about  units.    Calculations  should  be  done  in  the  MKS  (also  called  the  SI)  
system  of  units.    All  of  the  fundamental  constants  (e.g.  Plank’s  constant,  Boltzmann’s  
constant,  rest  mass  of  an  electron,  etc.)  are  in  these  units.    After  we  have  an  answer,  
it  is  often  convenient  to  convert  to  different  units.    For  example,  in  semiconductor  
work,  we  like  to  quote  carrier  and  doping  densities  per  cubic  cm,  not  per  cubic  
meter,  which  is  the  proper  MKS  unit.  
 
Be  careful  about  units!  
 
 
2b)  Begin  with  the  DOS  expression  (SDF,  eqn.(2.6b),  p.  41)  
 
m*p 2m*p ( EV − E )
gV ( E ) =     E < EV    
π 2!3
 
3/2
m0 2m0 ⎛ m*p ⎞
gV ( E ) = ⎜ ⎟ EV − E  
π 2 ! 3 ⎝ m0 ⎠
 
3/2
9.11× 10−31 2 × 9.11× 10−31 ⎛ mp ⎞
*

gV ( E ) = ⎜ ⎟ EV − E    (everything  in  MKS  (SI)  units)  

( )
3
3.142 × 1.06 × 10−34 ⎝ m0 ⎠
 
3/2
⎛ m*p ⎞
gV ( E ) = 1.05 × 10 ⎜ ⎟ 56
EV − E  
⎝ m0 ⎠
 
According  to  SDF,  p.  34,   m*p m0 = 0.81
 
 

ECE-­‐305     4   Fall  2014  

Mark  Lundstrom     09/11/2014  

HW3  Solutions  (continued):  

 
gV ( E ) = 1.05 × 1056 ( 0.81) ( )
3/2 −1
EV − E = 7.65 × 1055 EV − E J-m 3  
 
Note  that  the  DOS  in  the  valence  band  is  a  little  smaller  than  in  the  conduction  band  
because  the  hole  effective  mass  is  smaller  than  the  electron  effective  mass.  
 
We  are  asked  for  the  density  of  states  26  meV  (milli-­‐electron  volts)  below  the  top  of  
the  valence  band.    In  Joules  
 
EV − E = 0.026 × 1.60 × 10−19 = 4.16 × 10−21 J    
 
1 1
gC ( EV − 0.026 eV ) = 7.65 × 1055 4.16 × 10−21 3
= 4.94 × 1045  
J-m J-m 3
 
The  answer  is  correct,  but  the  units  are  inconvenient.    Let  express  the  answer  per  eV  
per  cubic  cm,  which  is  the  more  typical  way  of  doing  things  for  semiconductors.      
3
1 1.6 × 10−19 J 1 ⎛ 10−2 m ⎞
gV ( EV − 0.026 eV ) = 4.94 × 10 × 45
×  
J eV m 3 ⎜⎝ cm ⎟⎠
1
gV ( EV − 0.026 eV ) = 7.90 × 1020  
eV- cm 3
 
Note:    The  total  number  of  states  in  a  range  of  energy  of   k BT = 0.026 eV  is  
approximately  
1
gV ( EV − 0.026 eV ) × 0.026 = 2.06 × 1019  
cm 3
 
which  is  a  reasonable  number,  as  we  shall  see  below.  
 
 
2c)   This  is  a  problem  about  the  “effective  density  of  states”  which  is  different  from  the  
density-­‐of-­‐states.    The  effective  density  of  states,  is  roughly  the  total  number  of  states  
in  an  energy  range  of   k BT  near  the  bottom  of  the  conduction  band  or  top  of  the  
valence  band.    It  is  derived  by  integration  as  discussed  in  the  text.    The  result  is  given  
in  SDF  on  p.  51  
 
3/2
⎛ m* ⎞ 1
N C = 2.510 × 10 ⎜ n ⎟ 19
 
⎝ m0 ⎠ cm 3
Using   mn* = 1.18m0 ,  we  find  
 

ECE-­‐305     5   Fall  2014  

Mark  Lundstrom     09/11/2014  

HW3  Solutions  (continued):  

 
N C = 2.510 × 1019 (1.18)
3/2
cm −3 = 3.22 × 1019 cm −3  
N C = 3.22 × 1019 cm −3  
 
Note  that  this  is  close  to  the  value  we  estimated  in  part  2a):  
 
1
gC ( 0.026 eV + EC ) × 0.026 = 3.62 × 1019  
cm 3
 
2d)    Repeat  2c)  but  for  the  valence  band.  
3/2
⎛ m*p ⎞ 1
NV = 2.510 × 10 ⎜ ⎟
19
 
⎝ m0 ⎠ cm 3
Using   mn* = 0.81m0 ,  we  find  
NV = 2.510 × 1019 ( 0.81)
3/2
cm −3 = 1.83× 1019 cm −3  
NV = 1.83× 1019 cm −3  
 
Note  that  this  is  close  to  the  value  we  estimated  in  part  2b):  
1
gV ( EV − 0.026 eV ) × 0.026 = 2.06 × 1019  
cm 3
 
2e)    Recall  the  atomic  density  of  Si  is   N Si = 5.00 × 1022 cm -3  so  the  ratio  
 
N C 3.22 × 1019
= = 6.44 × 10−4  
N Si 5.00 × 10 22

 
NC
= 0.06 %  
N Si
 
The  number  of  states  near  the  bottom  of  the  conduction  band  that  can  be  occupied  
is  a  small  fraction  of  the  density  of  silicon  atoms.      
 
 
 
 
 
 
 

ECE-­‐305     6   Fall  2014  

Mark  Lundstrom     09/11/2014  

HW3  Solutions  (continued):  

 
3) Consider  a  region  of  Si  at  room  temperature.    For  each  of  the  following  cases,  calculate  
the  equilibrium  electron  and  hole  concentrations  (n  and  p).    Assume  that  the  dopants  
are  fully  ionized.  
 
a) Intrinsic  material  ( N D = N A = 0 )  
b) N D = 1.00 × 1013 cm -3 N A = 0  
c) N D = 1.00 × 1017 cm -3 N A = 0  
d) N D = 0 N A = 1.00 × 1017 cm -3  
e) N D = 1.00 × 1017 cm -3 N A = 3.00 × 1017 cm -3  
 

Solution:  
 
3a)    For  intrinsic  material,  we  know  that   n = p = ni .    According  to  SDF,  p.  54,    
 
ni = 1.00 × 1010 cm -3  so  
n = p = ni = 1.00 × 1010 cm -3  
 
 

3b)   First,  let’s  ask  what  to  expect.    The  donor  density  is  1000  times  the  intrinsic  density.    
All  of  the  donor  electrons  will  go  in  the  conduction  band,  overwhelming  the  number  
of  intrinsic  carriers  that  were  there.    So  we  expect  
 
n ≈ N D = 1.00 × 1013 cm -3  
We  would  determine  the  hole  density  from   np = ni2 ,  so  
ni2 ni2 1020
p= ≈ = 13 = 1.00 × 107 cm -3  
n N D 10
Note  that  doping  the  semiconductor  n-­‐type  means  that  we  now  have  more  electrons  
in  the  conduction  band,  and  we  have  fewer  holes  in  the  valence  band  than  for  the  
intrinsic  semiconductor.  
 
How  would  we  do  this  problem  more  accurately?  
 
Begin  by  assuming  that  the  semiconductor  is  neutral:  
p − n + N D+ − N A− = 0    
Assume  that  the  dopants  are  fully  ionized  and  use   p = ni2 n    
ni2
− n + N D − N A = 0  
n

ECE-­‐305     7   Fall  2014  

Mark  Lundstrom     09/11/2014  

HW3  Solutions  (continued):  

 
Solve  the  quadratic  equation  for  n  
2
N − NA ⎛ N − NA ⎞
n= D + ⎜ D ⎟⎠ + ni    
2

2 ⎝ 2
Plug  in  numbers:  
 
2
1013 − 0 ⎛ 1013 ⎞
n= + ⎜ ⎟ + 1020 = 0.5 × 1013 + 0.25 × 1026 + 1020
2 ⎝ 2 ⎠
 
= 0.5 × 1013 + 0.25 × 1026 + 0.000001× 1026
 
n = 0.5 × 1013 + 0.5 × 1013 = 1.0 × 1013    
 
n = 1.00 × 1013 cm -3  
 
 
Just  as  expected.    We  find  the  hole  density  from  
 
n2 n2 1020
p = i ≈ i = 13 = 1.00 × 107 cm -3  
n N D 10
 
p = 1.00 × 107 cm -3  
 
3c)   First,  let’s  ask  what  to  expect.    The  donor  density  is   107  times  the  intrinsic  density.    All  
of  the  donor  electrons  will  go  in  the  conduction  band,  overwhelming  the  number  of  
intrinsic  carriers  that  were  there.    So  we  expect  
 
n ≈ N D = 1.00 × 1017 cm -3  
We  would  determine  the  hole  density  from   np = ni2  ,  so  
ni2 ni2 1020
p= ≈ = 17 = 1.00 × 103 cm -3  
n N D 10
 
Note  that  doping  the  semiconductor  n-­‐type  means  that  we  now  have  more  electrons  
and  we  have  fewer  holes  than  the  intrinsic  semiconductor.  
 
Now  do  this  problem  more  accurately.  
 
Use  the  quadratic  equation  for  n:      
 
 

ECE-­‐305     8   Fall  2014  

Mark  Lundstrom     09/11/2014  

HW3  Solutions  (continued):  

 
2
N − NA ⎛ N − NA ⎞
n= D + ⎜ D ⎟ + ni2    
2 ⎝ 2 ⎠
 
Plug  in  numbers:  
 
2
1017 − 0 ⎛ 1017 ⎞
n= + ⎜ ⎟ + 1020 = 0.5 × 1017 + 0.25 × 1034 + 1020
2 ⎝ 2 ⎠  

= 0.5 × 1017 + 0.25 × 1034 + 0

 
n = 0.50 × 1017 + 0.50 × 1017 = 1.00 × 1017  
 
n = 1.00 × 1017 cm -3  
 
Just  as  expected.    Next,  we  find  the  hole  density  from  
 
n2 n2 1020
p = i ≈ i = 17 = 1.00 × 103 cm -3  
n N D 10
 
p = 1.00 × 103 cm -3  
 
3d)   Again,  let’s  ask  what  to  expect.    The  acceptor  density  is   107  times  the  intrinsic  density.    
All  of  the  acceptor  holes  will  go  in  the  valence  band,  overwhelming  the  number  of  
intrinsic  carriers  that  were  there.    So  we  expect  
 
p ≈ N A = 1.00 × 1017 cm -3  
We  would  determine  the  electron  density  from   np = ni2  ,  so  
ni2 ni2 1020
n= ≈ = 17 = 1.00 × 103 cm -3  
p N A 10
 
Note  that  doping  the  semiconductor  p-­‐type  means  that  we  now  have  more  holes  in  
the  valence  band,  and  we  have  fewer  electrons  in  the  conduction  band  than  the  
intrinsic  semiconductor.  
 
Now  do  this  problem  more  accurately.  
 
Use  the  quadratic  equation  for  p:  
 

ECE-­‐305     9   Fall  2014  

Mark  Lundstrom     09/11/2014  

HW3  Solutions  (continued):  

 
2
N − ND ⎛ N − ND ⎞
p= A + ⎜ A ⎟ + ni2    
2 ⎝ 2 ⎠
Plug  in  numbers:  
 
2
1017 − 0 ⎛ 1017 ⎞
p= + ⎜ ⎟ + 1020 = 0.5 × 1017 + 0.25 × 1034 + 1020
2 ⎝ 2 ⎠
 
= 0.5 × 1017 + 0.25 × 1034 + 0
 
p = 0.50 × 1017 + 0.50 × 1017 = 1.00 × 1017  
 
p = 1.00 × 1013 cm -3  
 
Just  as  expected.    We  find  the  electron  density  from  
ni2 ni2 1020
n= ≈ = = 1.00 × 103 cm -3  
p N A 1017
 
n = 1.00 × 103 cm -3  
 
3e)   Since  the  net  p-­‐type  doping  is   N A − N D = 3× 1017 − 1× 1017 = 2 × 1017  is  much  greater  
than  the  intrinsic  density,  we  expect:   p = 2 × 1017    and   n = ni2 p = 0.5 × 103 cm -3  .    This  
is  correct,  as  we  can  see  by  solving  the  problem  properly.  
 
2 2
N − ND ⎛ N − ND ⎞ 3× 1017 − 1× 1017 ⎛ 2 × 1017 ⎞
p= A + ⎜ A ⎟⎠ + n 2
= + ⎜⎝ 2 ⎟⎠ + 10  
20

2 ⎝ 2 i
2
 
p = 1× 1017 + 2 × 1034 + 1020 = 2 × 1017  
 
p = 2 × 1017 cm -3  
 
ni2 ni2 1020
n= ≈ = = 5.00 × 102 cm -3  
p 2 × 1017 2 × 1017
 
n = 5.00 × 102 cm -3  

ECE-­‐305     10   Fall  2014  

Mark  Lundstrom     09/11/2014  

 
HW3  Solutions  (continued):  
 
One  might  conclude  from  these  examples,  that  there  is  no  need  to  use  space-­‐charge  
neutrality  and  the  quadratic  equation.    That  would  be  wrong,  as  seen  in  the  next  
problem.    It  only  happened  in  this  problem  because  in  each  case  (except  for  case  a)),  the  
net  doping  was  much  larger  than  the  intrinsic  carrier  density.  
 
 
4) Assuming  silicon  with  completely  ionized  dopants,  compute  n  and  p  for  the  following  
case.  
 
N D = 5.00 × 1016 cm -3    
N A = 0  
T = 700 K  
 
Solution:  
 
We  expect  the  intrinsic  carrier  density  to  be  MUCH  larger  at  higher  temperatures.    
Using  the  information  on  p.  57  of  SDF,  we  have  
 
( )
ni 700 K = 2.865 × 1016 cm -3  
 
This  is  close  to  the  doping  density,  so  we  need  to  use  the  quadratic  equation.  
2
N − NA ⎛ N − NA ⎞
n= D + ⎜ D ⎟ + ni2  
2 ⎝ 2 ⎠
 

( 2.5 × 10 ) + ( 2.865 × 10 )
2 2
n = 2.5 × 1016 + 16 16
 
 
n = 2.5 × 1016 + 6.25 × 1032 + 8.21× 1032 = 6.3× 1016  
 
n = 6.3× 1016 cm -3  
 
( )
2
ni2 2.865 × 1016 8.21× 1032
p= = = = 1.30 × 1016 cm -3  
n 6.3× 1016 6.3× 1016

 
p = 1.30 × 1016 cm -3  
 
 

ECE-­‐305     11   Fall  2014  

Mark  Lundstrom     09/11/2014  

 
HW3  Solutions  (continued):  
 
5) For  each  of  the  cases  (a-­‐e)  in  Prob.  3,  calculate  the  Fermi  level  position,  with  respect  to  
the  intrinsic  level   ( E F − Ei ) .    Note  that  you  need  to  consider  sign.    
 
Solution:  
 
In  problem  3),  we  computed  n  and  p.    For  this  problem,  we  need  an  expression  that  
relates  n  and  p  to  the  Fermi  level  and  intrinsic  level.    The  expressions  are  on  p.  53  of  
SDF:  
 
n = ni e(
E F − Ei ) k BT
   
p = ni e(
Ei − E F ) k BT
 
Solving  for   ( E F − Ei )  from  the  first  equation,  we  have  
⎛ n⎞
(E − Ei ) = k BT ln ⎜ ⎟  
F
⎝ ni ⎠
or  in  electron  volts  
 
( EF − Ei ) = k BT ln ⎛ n ⎞ = 0.026ln ⎛ n ⎞  
q q ⎜⎝ n ⎟⎠ ⎜⎝ 1010 ⎟⎠
i

 
 
3a)       n = p = ni = 1.00 × 1010 cm -3
 
 
( EF − Ei ) = 0.026ln (1) = 0  
q
 
(E F
− Ei )
= 0  or  we  could  say:     ( E F − Ei ) = 0 eV  
q
 
3b)   n = 1.00 × 1013 cm -3
 
(E F
− Ei ) ⎛ 1013 ⎞
= 0.026ln ⎜ 10 ⎟ = 0.180  
q ⎝ 10 ⎠
 
(E F
− Ei )
= 0.180  
q
 

ECE-­‐305     12   Fall  2014  

Mark  Lundstrom     09/11/2014  

HW3  Solutions  (continued):  

 
3c)   n = 1.00 × 1017 cm -3  
 
( EF − Ei ) = 0.026ln ⎛ 1017 ⎞ = 0.026ln 107 = 0.419  
q ⎜⎝ 1010 ⎟⎠ ( )
 
(E F
− Ei )
= 0.419  
q
 
Note:    The  larger  the  n-­‐type  doping,  the  higher  the  Fermi  level  is  above  the  intrinsic  
level.  
 
 
3d)   n = 1.00 × 103 cm -3  
 
( EF − Ei ) = 0.026ln ⎛ 103 ⎞ = 0.026ln 10−7 = − 0.419  
q ⎜⎝ 1010 ⎟⎠ ( )
 
(E F
− Ei )
= − 0.419  
q
 
Note  the  sign  –  the  Fermi  level  is  now  below  the  intrinsic  level  because  this  is  a  p-­‐
type  semiconductor.  
 
3e)     n = 5.00 × 102 cm -3  
 
( EF − Ei ) = 0.026ln ⎛ 5 × 102 ⎞ = 0.026ln 5 × 10−8 = − 0.437  
q ⎜⎝ 1010 ⎟⎠ ( )
 
(E F
− Ei )
= − 0.437  
q
 
Also  a  p-­‐type  semiconductor,  so  the  Fermi  level  is  below  the  intrinsic  level.  
 
 
 
 
 

ECE-­‐305     13   Fall  2014  

Mark  Lundstrom     09/11/2014  

HW3  Solutions  (continued):  

 
6) For  each  of  the  cases  (a-­‐e)  in  Prob.  3,  calculate  the  Fermi  level  position,  with  respect  to  
the  bottom  of  the  conduction  band  (EF  -­‐  EC).    Note  that  you  need  to  consider  sign.    
 
 
Solution:  
 
This  is  basically  the  same  problem  as  5)  –  we  just  want  to  know  the  position  of  the  
Fermi  level  with  respect  to  another  energy,  bottom  of  the  conduction  band  instead  
of  the  intrinsic  level.      
 
To  solve  this  problem,  we  need  an  expression  that  relates  n  to  the  Fermi  level  and  
bottom  of  the  conduction  band.    The  expression  is  on  p.  52  of  SDF:  
 
n = N C e(
E F − EC ) k BT
   
Solving  for   ( E F − EC )  we  have  
⎛ n ⎞
(E F
− E C ) = k B
T ln ⎜⎝ N ⎟⎠  
C

or  in  electron  volts  

 
( EF − EC ) = k BT ln ⎛ n ⎞ = 0.026ln ⎛ n ⎞  
q q ⎜⎝ N ⎟⎠ ⎜⎝ N ⎟⎠
C C

 
We  computed   N C    in  prob.  2c:   N C = 3.22 × 1019 cm −3  
 
( EF − EC ) = 0.026ln ⎛ n ⎞  
q ⎜⎝ 3.22 × 1019 ⎟⎠
Note  that  this  quantity  will  be  negative  for  a  non-­‐degenerate  semiconductor,  
because  the  Fermi  level  is  always  well  below  the  conduction  band.  
 
For  case  3a)     n = p = ni = 1.00 × 1010 cm -3
 
 
( EF − EC ) = 0.026ln ⎛ 1010 ⎞ = 0.569  
q ⎜⎝ 3.22 × 1019 ⎟⎠
 
(E F
− EC )
= −0.569  or  we  could  say:     ( E F − EC ) = −0.569 eV  
q
 
 

ECE-­‐305     14   Fall  2014  

Mark  Lundstrom     09/11/2014  

HW3  Solutions  (continued):  

 
Case    3b)   n = 1.00 × 1013 cm -3  
 
( EF − EC ) ⎛ 10
= 0.026ln ⎜
⎞ 13
= −0.390    
(E F
− EC )
= −0.390  
19 ⎟
q ⎝ 3.22 × 10 ⎠ q
 
Case  3c)   n = 1.00 × 1017 cm -3  
(E F
− EC ) ⎛ 1017
= 0.026ln ⎜
⎞
= −0.150    
(E F
− EC )
= −0.150  
19 ⎟
q ⎝ 3.22 × 10 ⎠ q
 
Note:    The  larger  the  n-­‐type  doping,  the  closer  the  Fermi  level  is  to  the  conduction  
band.  
 
3d)     n = 1.00 × 103 cm -3  
(E F
− EC ) ⎛
= 0.026ln ⎜
103 ⎞
= −0.988    
(E F
− EC )
= − 0.988  
19 ⎟
q ⎝ 3.22 × 10 ⎠ q
 
Note:    This  value  is  getting  close  to  the  bandgap  of  Si.    The  Fermi  level  is  getting  
close  to  the  top  of  the  valence  band.  
 
3e)     n = 5.00 × 102 cm -3  
(E F
− EC ) ⎛ 5 × 102 ⎞
= 0.026ln ⎜ = −1.01    
(E F
− EC )
= −1.01  
⎝ 3.22 × 10 ⎟⎠
19
q q
 
Also  a  p-­‐type  semiconductor,  so  the  Fermi  level  is  close  to  the  top  of  the  valence  
band.  
 
 
7) For  the  case  of  problem  4),  calculate  the  Fermi  level  position,  with  respect  to  the  
intrinsic  level   ( E F − Ei ) .    Note  that  you  need  to  consider  sign.    
 
Solution:  
 
The  electron  density  was   n = 6.3× 1016 cm -3  
and  the  intrinsic  carrier  concentration  was   ni ( 700 K ) = 2.865 × 1016 cm -3  
(E F
− Ei )
=
k BT ⎛ n ⎞
ln ⎜ ⎟
q q ⎝ ni ⎠  
 

ECE-­‐305     15   Fall  2014  

Mark  Lundstrom     09/11/2014  

HW3  Solutions  (continued):  

 
k BT 1.38 × 10−23 × 700
at  T  =  700  K:     = = 0.0604 eV    
q 1.6 × 10−19
 
(E F
− Ei ) ⎛ 6.3× 1016 ⎞
= 0.060ln ⎜ = 0.048
⎝ 2.865 × 10 ⎟⎠
16
q
 
 
(E F
− Ei )
= 0.048
q
 
 
The  Fermi  level  is  almost  in  the  middle  of  the  gap  –  very  close  to  the  intrinsic  level.    
This  happens  because   n = 6.3× 1016 cm -3  is  only  a  little  bigger  than   ni ( 700 K ) .  
 
 
8) Your  textbook  defines  a  non-­‐degenerate  semiconductor  as  one  for  which  the  Fermi  
level  is  at  least   3k BT  below  the  conduction  band  edge  and  at  least   3k BT above  the  
valence  band  edge.    What  is  the  largest  electron  density  for  non-­‐degenerate  Si?  
 
Solution:  
Require:     ( EC − E F ) ≥ 3k BT    
n = N C e(
E F − EC ) k BT
 
 
n ≤ N C e(
−3k BT ) k BT
= N C e−3  
 
( )
n ≤ 3.22 × 1019 × e−3 = 1.60 × 1018  
 
n = 1.60 × 1018 cm -3  is  the  largest  electron  density  in  Si  at  room  temperature  for  
which  we  can  use  non-­‐generate  expressions.  
 
Note:    Your  book  is  fairly  conservative  on  this.    Many  people  would  use  non-­‐
degenerate  equations  when  
 
n ≤ N C = 3.22 × 1019 cm -3  
 
Some  might  even  use  the  assumption  for  higher  doping  densities,  but  they  understand  
what  they  are  doing  and  what  the  errors  are.  
 

ECE-­‐305     16   Fall  2014  

Mark  Lundstrom     09/11/2014  

HW3  Solutions  (continued):  

 
9) For  each  case  below,  explain  physically  (i.e.  in  words  in  no  equations)  whether  the  
intrinsic  level  is  above  or  below  the  middle  of  the  gap,  then  compute  the  location  of  the  
intrinsic  level  relative  to  the  middle  of  the  bandgap,   Emid = ( EC + EV ) 2  .  
 
a) Silicon  at  T  =  300K  with   mn* = 1.18m0  and   m *p = 0.81m0 .  
b) GaAs  at  T  =  300K  with   mn* = 0.067m0  and   m *p = 0.524m0 .  
 
Solution:  
 
a)  Silicon:  
 
The  carrier  density  is  basically  the  number  of  states  times  the  probability  that  the  
states  are  occupied.    The  density  of  states  varies  as  effective  mass  to  the  3/2  power,  
so  the  conduction  band  density  of  states  is  larger  than  the  valance  band  density  of  
states  for  Si.    To  get  equal  numbers  of  electrons  and  holes,  the  Fermi  level  
(intrinsic  level)  will  need  to  be  a  little  below  the  middle  of  the  bandgap.    This  
will  make  the  probability  that  conduction  band  statesare  occupied  a  little  smaller  
than  the  probability  that  valence  band  states  are  empty,  and  compensate  for  the  
larger  DOS  in  the  conduction  band.  
 
To  compute   Ei − Emid  we  use  eqn.  (2.36)  in  p.  62  in  SDF  and  find  
 
Ei − Emid 3 k BT ⎛ mp ⎞
*
⎛ 0.81⎞
= ln ⎜ * ⎟ = 0.75 × 0.026 × ln ⎜ = −0.007 eV  
q 4 q ⎝ mn ⎠ ⎝ 1.18 ⎟⎠
 
Ei − Emid
= −0.007 eV  
q
 
The  intrinsic  level  is  7  meV  below  the  middle  of  the  gap  in  Si.  
 
b)  GaAs:  
 
For  GaAs,  the  conduction  band  density  of  states  is  smaller  than  the  valance  band  
density  of  states  (just  the  opposite  of  Si).    Using  the  same  reasoning  as  in  9a),  to  get  
equal  numbers  of  electrons  and  holes,  the  Fermi  level  (intrinsic  level)  will  need  
to  be  a  little  above  the  middle  of  the  bandgap.    This  will  make  the  probability  
that  conduction  band  states  are  occupied  a  little  largerthan  the  probability  that  
valence  band  states  are  empty,  and  compensate  for  the  smaller  DOS  in  the  
conduction  band,  
 

ECE-­‐305     17   Fall  2014  

Mark  Lundstrom     09/11/2014  

 
HW3  Solutions  (continued):  
 
To  compute   Ei − Emid  we  use  eqn.  (2.36)  in  p.  62  in  SDF  and  find  
 
Ei − Emid 3 k BT ⎛ mp ⎞
*
⎛ 0.524 ⎞
= ln ⎜ * ⎟ = 0.75 × 0.026 × ln ⎜ = +0.040 eV  
q 4 q ⎝ mn ⎠ ⎝ 0.067 ⎟⎠
 
Ei − Emid
= +0.040 eV  
q
 
The  intrinsic  level  is  40  meV  above  the  middle  of  the  gap  in  GaAs.  
 
This  example  shows  that  the  intrinsic  level  is  always  close  to  the  middle  of  the  
bandgap  -­‐-­‐  even  when  the  effective  masses  of  electrons  and  holes  are  quite  different.  
 
 
10) A  semiconductor  (not  necessarily  Si)    is  doped  at   N D = 2 × 1014 cm -3 and  
N A = 1.2 × 1014 cm -3 .    The  thermal  equilibrium  electron  concentration  is  found  to  be  
n = 1.1 × 1014 cm -3 .    Assume  complete  ionization  of  dopants,  and  compute  the  intrinsic  
carrier  concentration  and  the  equilibrium  hole  concentration.  
 
Solution:  
 
Begin  with  the  general  expression:  
2
N − NA ⎛ N − NA ⎞
n= D + ⎜ D ⎟⎠ + ni  
2

2 ⎝ 2
 
2
2.0 × 1014 − 1.2 × 1014 ⎛ 0.8 × 1014 ⎞
1.1× 10 = 14
+ ⎜ ⎟ + ni2 = 0.4 × 1014 + 0.16 × 1028 + ni2
2 ⎝ 2 ⎠
 
 
Solve  for   ni2  :  
 
0.16 × 1028 + ni2 = 0.7 × 1014 0.16 × 1028 + ni2 = 0.49 × 1028 ni2 = 0.33× 1028
     
 
ni = 5.7 × 1013 cm -3
   

ECE-­‐305     18   Fall  2014  

Mark  Lundstrom     09/11/2014  

 
HW3  Solutions  (continued):  
 
ni2 0.33× 1028
np = ni2 → p = = = 3.00 × 1013
n 1.1× 1014
   
 
p = 3.00 × 1013 cm -3
 
 
 
11) Answer  the  following  questions  about  resistivity  at  T  =  300K.  
 
a)   Compute  the  resistivity  of  intrinsic  Si,  Ge,  and  GaAs.  
 
b)   Compute  the  resistivity  of  n-­‐type  Si,  Ge,  and  GaAs  doped  at   N D = 1019 cm -3 .    
Assume  complete  ionization  of  dopants.  
 
Solution:  
 
According  to  eqn.  (3.7)  on  p.  85  of  SDF  
1
ρ= Ω-cm    
nqµ n + pqµ p
Units  (MKS  (SI)):    What  are  the  units  of   1 nqµ n  or   1 pqµ p ?  
1 1 1 1 V V
→ -3 × × 2 = − m = − m = Ω-m    MKS  or  SI  
nqµ m C m /(V-s) ( C/s ) A
but  we  are  usually  given  carrier  densities  per  cm3  and  mobility  in  cm2/(V-­‐s)  
 
1 1 1 1 V V
→ × × = − cm = − cm = Ω-cm  
nqµ cm -3
C cm /(V-s) ( C/s )
2
A
 
so  then  the  answer  comes  out  in   Ω - cm .  
 
11a)    We  need  the  carrier  densities  and  mobilities:  
 
n = p = ni    
 
From  Fig.  3.5,  p.  80  of  SDF  for  silicon:  
cm 2 cm 2
µ n ≈ 1400     µ p ≈ 460  
V-s V-s
 

ECE-­‐305     19   Fall  2014  

Mark  Lundstrom     09/11/2014  

 
HW3  Solutions  (continued):  
 
1 1
ρ= = Ω-cm
nqµ n + pqµ p ni q µ n + µ p ( )  
1
ρ= = 3.4 × 105 Ω-cm  
10 × 1.6 × 10
10 −19
× (1400 + 460 )
 
ρ = 3.4 × 105 Ω-cm    (silicon)  
 
This  is  a  very  large  resistivity  –  not  as  large  as  an  insulator,  but  very  large  for  a  
semiconductor.  
 
For  Ge:      
1
ρ= Ω-cm  
(
ni q µ n + µ p )
 
From  Fig.  3.5,  p.  80  of  SDF  for  Ge:  
cm 2 cm 2
µ n ≈ 4000     µ p ≈ 2000  
V-s V-s
 
From  Fig.  2.20,  p.  54  of  SDF  for  Ge:  
 
ni ( 300 K ) = 2 × 1013 cm -3    
 
1
ρ= = 5.2 × 101 Ω-cm
2 × 10 × 1.6 × 10
13 −19
× ( 4000 + 2000 )
 
 
ρ = 5.2 × 101 Ω-cm    (Ge)  
Intrinsic  Ge  is  fairly  conductive!    This  happens  because  the  mobilities  are  higher,  but  
mostly  because  the  bandgap  is  much  lower,  so  the  intrinsic  carrier  concentration  is  
much  larger.  
 
For  GaAs:      
 
1
ρ= Ω-cm  
(
ni q µ n + µ p )
From  Fig.  3.5,  p.  80  of  SDF  for  GaAs:  

ECE-­‐305     20   Fall  2014  

Mark  Lundstrom     09/11/2014  

cm 2 cm 2
µ n ≈ 8500     µ p ≈ 430  
V-s V-s
HW3  Solutions  (continued):  
 
From  Fig.  2.20,  p.  54  of  SDF  for  Ge:  
 
ni ( 300 K ) = 2.25 × 106 cm -3    
 
1
ρ= = 3.1× 108 Ω-cm
2.25 × 10 × 1.6 × 10−19 × (8500 + 430 )
6
 
 
ρ = 3.1× 108 Ω-cm    (GaAs)  
 
Intrinsic  GaAs  has  a  resistivity  that  is  orders  of  magnitude  larger  than  Si!    This  
happens  even  though  the  electron  mobility  is  much  larger  than  Si  because  the  
bandgap  is  much  larger,  so  the  intrinsic  carrier  concentration  is  orders  of  magnitude  
smaller.      
 
11b)    Since  we  are  heavily  doped  n-­‐type,  we  can  ignore  the  contribution  from  holes.  
 
1
ρ= Ω-cm  
nqµ n
 
For  Si:      
 
From  Fig.  3.5,  p.  80  of  SDF  for  Si:  
cm 2
µ n ≈ 110  
V-s
1
ρ= −19
= 5.7 × 10−3 Ω-cm
1× 10 × 1.6 × 10 × 110
19
 
 
ρ = 5.7 × 10−3 Ω-cm  (Si)  
 
This  is  about  eight  orders  of  magnitude  lower  than  the  intrinsic  resistivity  of  Si.    The  
ability  to  control  the  resistivity  over  orders  of  magnitude  is  what  makes  
semiconductors  so  useful.  
 
For  Ge:      
 
From  Fig.  3.5,  p.  80  of  SDF  for  Si:  

ECE-­‐305     21   Fall  2014  

Mark  Lundstrom     09/11/2014  

cm 2
µ n ≈ 900    
V-s
HW3  Solutions  (continued):  
1
ρ= −19
= 6.9 × 10−4 Ω-cm
1× 10 × 1.6 × 10 × 900
19
 
 
ρ = 6.9 × 10−4 Ω-cm  (Ge)  
 
Even  lower  than  Si  because  the  mobility  is  higher.  
 
For  GaAs:      
 
From  Fig.  3.5,  p.  80  of  SDF  for  GaAs:  
cm 2
µ n ≈ 3200    
V-s
1
ρ= −19
= 2.0 × 10−4 Ω-cm
1× 10 × 1.6 × 10 × 3200
19
 
 
ρ = 2.0 × 10−4 Ω-cm  (GaAs)  
 
Even  lower  than  Ge  because  the  mobility  is  higher.  
 
 
Comments:    For  Si,  we  found  that  the  range  of  possible  resistivities  is  
 
3.0 × 105 Ω-cm < ρ < 6 × 10−3 Ω-cm  
 
about  eight  orders  of  magnitude.    In  practice,  we  could  never  have  pure,  intrinsic  Si,  there  
will  always  be  some  unintentional  dopants  no  matter  how  pure  we  try  to  make  Si,  so  the  
resistivities  would  not  be  as  high  as  indicted  here,  but  it’s  common  to  find  resistivities  in  
the  1000’s.    Also,  one  can  dope  Si  to   1020 cm -3 ,  so  resistivities  10  times  lower  than  indicated  
above  can  be  obtained.  
 
How  do  these  limits  compare  to  insulators  and  metals?    A  Google  search  shows  
 
Resistivity  of  copper:     ρ Cu = 1.68 × 10−8 Ω-cm
   
Resistivity  of  diamond:     ρ diamond = 10 − 10 Ω-cm  
13 20

ECE-­‐305     22   Fall  2014  

Mark  Lundstrom     09/11/2014  

So  semiconductors  are  not  great  metals  and  not  great  insulators  –  their  usefulness  comes  
from  the  ability  to  vary  their  resistivity  controllably  with  doping.  
 
 
HW3  Solutions  (continued):  
 
12)    You  are  given  a  10  Ohm-­‐cm  silicon  wafer  at  300  K.    
 
12a)    If  it  is  n-­‐type,  What  is  the  electron  density?  
 
12b)    If  it  is  p-­‐type,  what  is  the  hole  density?  
 
Solution:  
 
From  Fig.  3.8  on  on  p.  86  of  SDF,  we  can  read  off  the  doping  density  for  a  given  
resistivity.      
 
12a)    10  Ohm-­‐cm  n-­‐type  corresponds  to   N D ≈ 5 × 1014 cm -3  .    Under  these  conditions:  
 
n ≈ N D ≈ 5 × 1014 cm -3  
 
12b)    10  Ohm-­‐cm  p-­‐type  corresponds  to   N A ≈ 1.3× 1015 cm -3  .    Under  these  
conditions:  
 
p ≈ N A ≈ 1.3× 1015 cm -3  
 
 
13)   Assume  a  hypothetical  semiconductor  with  the  following  properties:  
 
EG = 1 eV N C = NV = 1 × 1019 cm -3  
EC − EF = 0.4 eV  
T = 600 K  
 
Answer  the  following  questions.  
 
13a)   What  is  the  equilibrium  electron  density?  
13b)   What  is  the  equilibrium  hole  density?  
13c)   What  is  the  intrinsic  carrier  concentration,   ni ?  
13d)   What  is  the  net  doping  density,   N D − N A ,  assuming  that  the  dopants  are  fully  
ionized?  

ECE-­‐305     23   Fall  2014  

Mark  Lundstrom     09/11/2014  

13e)   Where  is  the  intrinsic  level  located  with  respect  to  the  conduction  band?  
(Note:    You  can  simply  write  down  the  answer  without  any  work,  if  you  
explain  your  reasoning.  
 
 
 
HW3  Solutions  (continued):  
 
Solutions:  
13a)   What  is  the  equilibrium  electron  density?  
 
n = N C e(
E F − EC ) k BT
   
1.38 × 10−23 × 600
k BT ( eV ) = = 0.052    
1.6 × 10−19
n = 1019 e ( )
− 0.4 0.052
= 4.56 × 1015 cm -3     n = 4.56 × 1015 cm -3  
 
13b)   What  is  the  equilibrium  hole  density?  
 
p = NV e(
EV − E F ) k BT
 
EV − E F = ( EV − EC ) + ( EC − E F ) = − EG + 0.4 = −1.0 + 0.4 = −0.6  
p = NV e(
EV − E F ) k BT
= 1019 e−0.6/0.052 = 9.75 × 1013 cm -3   p = 9.75 × 1013 cm -3  
 
13c)   What  is  the  intrinsic  carrier  concentration,   ni ?  
 
np = 4.56 × 1015 × 9.75 × 1013 = 4.45 × 1029 = ni2  

ni = ni2 = 6.67 × 1014 cm -3  

 
 
13d)   What  is  the  net  doping  density,   N D − N A ,  assuming  that  the  dopants  are  fully  
ionized?  
2
N − NA ⎛ N − NA ⎞
n= D + ⎜ D ⎟ + ni2
2 ⎝ 2 ⎠
 
Solve  for   N D − N A    
ni2
ND − NA = n −
n
 
4.45 × 1029
N D − N A = 4.56 × 1015 − = 4.66 × 1015
4.56 × 1015
 

ECE-­‐305     24   Fall  2014  

Mark  Lundstrom     09/11/2014  

 
N D − N A = 4.66 × 1015 cm -3
 
 
 
 
HW3  Solutions  (continued):  
 
13e)   Where  is  the  intrinsic  level  located  with  respect  to  the  conduction  band?  
(Note:    You  can  simply  write  down  the  answer  without  any  work,  if  you  
explain  your  reasoning.  
 
Since   N C = NV  putting  the  Fermi  level  exactly  at  the  middle  of  the  bandgap  
will  produce  equal  numbers  of  electrons  and  holes.  
 
Exactly  at  mid-­‐gap.  
 
 
14)   For  the  energy  band  diagram  sketched  below,  answer  the  following  questions.  
 
14a)    Sketch  the  electrostatic  potential,   V ( x ) ,  vs.  position,  x.  
14b)  Sketch  the  electric  field,  E ( x ) ,  vs.  position,  x.  
14c)    Sketch  the  electron  density,   n ( x )  vs.  position,  x.  
14d)    Sketch  the  hole  density,   p ( x ) ,    vs.  position,  x.  
 

 
 
 
 

ECE-­‐305     25   Fall  2014  

Mark  Lundstrom     09/11/2014  

 
 
 
 
 
 
 
HW3  Solutions  (continued):  
 
Solution:  
14a)    Sketch  the  electrostatic  potential,   V ( x ) ,  vs.  position,  x.  
 

 
 
14b)  Sketch  the  electric  field,  E ( x ) ,  vs.  position,  x.  
 
 

 
 

ECE-­‐305     26   Fall  2014  

Mark  Lundstrom     09/11/2014  

 
 
 
 
 
 
 
HW3  Solutions  (continued):  
 
14c)    Sketch  the  electron  density,   n ( x )  vs.  position,  x.  

 
 
14d)    Sketch  the  hole  density,   p ( x ) ,    vs.  position,  x.  
 

 
 

ECE-­‐305     27   Fall  2014