T=300K# x 2.2 4 O 1.2 100 µm DD S D - nanoHUBWeek13HW... · ECE305# # Spring2015# ECE/305# # 3#...

11
ECE 305 Spring 2015 ECE305 Spring 2015 1 ECE 305 Homework SOLUTIONS: Week 13 Mark Lundstrom Purdue University 1) Measured IV data for an nchannel MOSFET are shown below. Relevant parameters for this MOSFET are: T = 300 K Oxide thickness: x 0 = 2.2 nm Relative dielectric constant: K O = 4 Power supply voltage: V DD = 1.2 V Series resistances: R S = R D = 100 Ω− μm Actual channel length = 85 nm Assume that the MOSFET is 1 micrometer wide, W = 10 4 cm .

Transcript of T=300K# x 2.2 4 O 1.2 100 µm DD S D - nanoHUBWeek13HW... · ECE305# # Spring2015# ECE/305# # 3#...

Page 1: T=300K# x 2.2 4 O 1.2 100 µm DD S D - nanoHUBWeek13HW... · ECE305# # Spring2015# ECE/305# # 3# Spring2015# HW#Week13#Solutions(continued)# # R CH = 1 Wµ n C ox L V GS −V T) µ

ECE  305     Spring  2015  

ECE-­‐305     Spring  2015  1  

ECE  305  Homework  SOLUTIONS:  Week  13    

Mark  Lundstrom  Purdue  University  

   1)   Measured  IV  data  for  an  n-­‐channel  MOSFET  are  shown  below.      Relevant  parameters  

for  this  MOSFET  are:    

T  =  300  K  Oxide  thickness:     x0 = 2.2  nm    Relative  dielectric  constant:     KO = 4  Power  supply  voltage:   VDD = 1.2  V   Series  resistances:     RS = RD = 100 Ω− µm  Actual  channel  length  =  85  nm    Assume  that  the  MOSFET  is  1  micrometer  wide,   W = 10−4 cm .    

 

 

Page 2: T=300K# x 2.2 4 O 1.2 100 µm DD S D - nanoHUBWeek13HW... · ECE305# # Spring2015# ECE/305# # 3# Spring2015# HW#Week13#Solutions(continued)# # R CH = 1 Wµ n C ox L V GS −V T) µ

ECE  305     Spring  2015  

ECE-­‐305     Spring  2015  2  

HW  Week  13  Solutions  (continued)    

Answer  the  following  questions:    1a)   From  the  so-­‐called  “on-­‐current”,   ID VGS =VDD ,VDS =VDD( ) = 1.2 V ,  estimate  the  

average  velocity  of  electrons  in  the  channel  at  the  source  end  of  the  channel.    (Note:    You  will  need  to  correct  for  the  effect  of  the  series  resistance  as  discussed  in  HW12.)  

 Solution:  

 

Qn(0) = Cox VGS − ID RS −VT( )  

Cox =

KOεo

xo

= 1.6×10−6 F cm2  

Qn(0) = 1.6×10−6( ) 1.2−1120×10−6 ×100− 0.26( ) = 1.32×10−6

  Qn(0) q 8.3×1012 cm-2

 

υ(0) =

ION

WQn(0)= 1120×10−6

10−4 ×1.32×10−6 = 8.5×106 cm/s  

 

υ(0) = 8.5×106 cm/s  

 1b)   From  the  linear  region  of  operation,  estimate  the  effective  mobility  of  this  

MOSFET.    

Solution:  At  the  highest  gate  voltage:  

RTOT = RCH + RS + RD = 340Ω− µm  

RCH = RTOT − RS + RD = 340− 200( )Ω− µm   RCH = 140Ω− µm    Because  the  width  is  given  as  1  micrometer,   RCH = 140 Ω− µm .  

 

   Assume  very  small     VDS  with  a  small   ID ,  so  the  intrinsic  gate  voltage  is  about  the  applied  gate  voltage.    

υ(0) =

ION

WQn(0)

ID =

WµnCox

L′VGS −VT( ) ′VDS

Page 3: T=300K# x 2.2 4 O 1.2 100 µm DD S D - nanoHUBWeek13HW... · ECE305# # Spring2015# ECE/305# # 3# Spring2015# HW#Week13#Solutions(continued)# # R CH = 1 Wµ n C ox L V GS −V T) µ

ECE  305     Spring  2015  

ECE-­‐305     Spring  2015  3  

HW  Week  13  Solutions  (continued)    

RCH = 1WµnCox

LVGS −VT( )

 

µn =

LRCHWCox VGS −VT( ) =

LRCHW Qn

 

 

Qn = −Cox VGS −VT( ) = −1.6×10−6 1.2− 0.28( ) = 1.47 ×10−6

   

µn =

LRCHW Qn

= 8.5×10−7

140×10−4 ×1.47 ×10−6 = 412  cm2/V-­‐s  

 

µn = 412 cm2 /V-s

     

 2)   Assume  the  square  law  theory  of  the  MOSFET:    

ID =

WµnCox

LVG −VT( )VD −

VD2

2⎡

⎣⎢

⎦⎥     0 ≤VD <VDsat     VG ≥VT    

ID =

WµnCox

2LVG −VT( )2

    VD ≥VDsat   VG ≥VT  

 2a)   Derive  an  expression  for  the  electric  field  vs.  position  along  the  channel  in  the  

linear  region  of  operation.    

Solution:  The  drain  current  is  a  drift  current:  

ID y( ) =WµnQn y( )E y y( )  ,  where  in  the  square  law  theory:  

Qn y( ) = −Cox VG −VT −V y( )⎡⎣ ⎤⎦ .  Equate  this  to  the  expression  for  the  “linear  region”  current:  

ID y( ) =WµnQn y( )E y y( ) = ID =

WµnCox

LVG −VT( )VD −

VD2

2⎡

⎣⎢

⎦⎥  

E y y( ) = − 1

L

VG −VT( )VD −VD

2

2⎡

⎣⎢

⎦⎥

VG −VT −V y( )⎡⎣ ⎤⎦    

Page 4: T=300K# x 2.2 4 O 1.2 100 µm DD S D - nanoHUBWeek13HW... · ECE305# # Spring2015# ECE/305# # 3# Spring2015# HW#Week13#Solutions(continued)# # R CH = 1 Wµ n C ox L V GS −V T) µ

ECE  305     Spring  2015  

ECE-­‐305     Spring  2015  4  

HW  Week  13  Solutions  (continued)  For  very  small  drain  voltages,  we  can  ignore   VD

2 2 ,    and   Qn y( ) ≈ −Cox VG −VT( ) ,  so    

E y y( ) ≈ −

VD

LVD <<VG −VT  which  is  just  what  we  would  expect  for  a  resistor.  

 2b)   Derive  an  expression  for  the  electric  field  vs.  position  along  the  channel  at  

VD =VDsat .    

Solution:  Begin  as  in  2a).    The  drain  current  is  a  drift  current:  

ID y( ) =WµnQn y( )E y y( )  ,  where  in  the  square  law  theory:  

Qn y( ) = −Cox VG −VT −V y( )⎡⎣ ⎤⎦  Equate  this  to  the  expression  for  the  saturation  region  current:  

ID y( ) =WµnQn y( )E y y( ) = ID =

WµnCox

2LVG −VT( )2

 

VG −VT −V y( )⎡⎣ ⎤⎦E y y( ) = − 1

2LVG −VT( )2

   electric  field  is  gradient  of  voltage,  so:  

− VG −VT( ) dV

dy+V dV

dy= − 1

2LVG −VT( )2

 

− VG −VT( )dV

0

V y( )

∫ + V dV0

V y( )

∫ = − 12L

VG −VT( )2dy

0

y

∫  

 which  can  be  integrated  to  find:  

− VG −VT( )V y( ) + V 2 y( )

2= − 1

2VG −VT( )2 y

L⎛⎝⎜

⎞⎠⎟  

or  

V 2 y( )− 2 VG −VT( )V y( ) + VG −VT( )2 y

L⎛⎝⎜

⎞⎠⎟= 0

   Solve  the  quadratic  equation  for  V(y):  

V y( ) = VG −VT( ) ± VG −VT( ) 1− y / L    Choose  the  minus  sign  because   V 0( ) = 0    and   V L( ) =VG −VT  (drain  end  at  pinch-­‐off).  

V y( ) = VG −VT( ) 1− 1− y / L⎡

⎣⎤⎦  

Page 5: T=300K# x 2.2 4 O 1.2 100 µm DD S D - nanoHUBWeek13HW... · ECE305# # Spring2015# ECE/305# # 3# Spring2015# HW#Week13#Solutions(continued)# # R CH = 1 Wµ n C ox L V GS −V T) µ

ECE  305     Spring  2015  

ECE-­‐305     Spring  2015  5  

HW  Week  13  Solutions  (continued)    

Now  we  can  find  the  electric  field  in  the  channel:  

E y( ) = − dV

dy= −

VG −VT( )2L

1− y / L( )−1/2

 

E y( ) = −

VG −VT( )2L

1− y / L( )−1/2

 Examine  answer:    At  y  =  0,  we  get    

E 0( ) = −

VG −VT( )2L  

As   y → L ,   E y → L( )→−∞  At  the  pinch-­‐off  point,  the  electric  field  approaches  infinity  and  the  inversion  layer  charge  approaches  zero.    The  product  of  the  two  is  finite.    This  is  a  limitation  of  the  simple  theory  that  the  velocity  saturation  model  tries  to  fix.  

   3) There  is  often  a  sheet  of  “fixed  charge”  at  the  oxide-­‐semiconductor  interface.  (By  fixed  

charge,  we  mean  a  charge  that  does  not  vary  with  surface  potential,   φS .)    Assume  a  1.5  nm  thick  oxide  with  a  relative  dielectric  constant  of  4.0.    Answer  the  questions  below.    3a)   Assume  that  a  fixed  charge  of   QF q = 1011 cm-2  is  present.    Explain  how  the  fixed  

charge  shifts  the  threshold  voltage.  What  is  the  direction  of  this  shift  (e.g.  is  it  a  positive  or  negative  voltage  shift?).  

 Solution:  The  voltage  drop  across  the  oxide  is  changed  by  an  amount:  

ΔV = −

QF

Cox

   

Cox =

KOε0

xo

= 4.0×8.854×10−14

1.5×10−7 = 2.36×10−6    

ΔV = −

QF

Cox

= −1.6×10−19 ×1011

2.36×10−6 = −6.8×10−3 V  

ΔV = −6.8×10−3 V  

This  change  in  voltage  drop  acriss  the  oxide  shifts   VT .    The  shift  is  negative.    If  we  are  dealing  with  an  N-­‐MOSFET,  then  the  threshold  voltage  will  be  a  little  smaller  (less  positive).  If  we  are  dealing  with  a  P-­‐MOSFET,  the  threshold  voltage  is  also  a  little  smaller,  which  means  that  the  threshold  voltage  will  be  a  little  more  negative.    Charge  at  the  oxide-­‐Si  interface  reduces  the  magnitude  of   VT  for  N_MOSFETs  and  increases  the  magnitude  of   VT  for  P-­‐MOSFETs.  

Page 6: T=300K# x 2.2 4 O 1.2 100 µm DD S D - nanoHUBWeek13HW... · ECE305# # Spring2015# ECE/305# # 3# Spring2015# HW#Week13#Solutions(continued)# # R CH = 1 Wµ n C ox L V GS −V T) µ

ECE  305     Spring  2015  

ECE-­‐305     Spring  2015  6  

HW  Week  13  Solutions  (continued)    Note  that  this  is  a  fairly  large  charge  by  today’s  standards.    High  quality  gate  oxides  can  have  10  times  less  fixed  charge.    In  the  late  1970’s,  however,  oxides  were  100  nm  thick  and  the  corresponding  shifts  in  voltage  were  much  larger  (~0.45  V)  for  this   QF .    (Do  you  see  why?)  

 3b)   In  HW  1,  we  saw  that  the  number  of  atoms  per  cm2  on  a  (100)  Si  surface  is  

NS = 6.81×1014 cm-2 .    Assume  each  atom  has  one  charged  dangling  bond,  so  a  

fixed  charge  of   QF q = 6.81×1014 cm-2  is  present.    How  much  is  the  voltage  shift  now?  

 Solution:  The  voltage  drop  across  the  oxide  is  changed  by  an  amount:  

ΔV = −

QF

Cox

= 1.6×10−19 × 6.81×1014

2.36×10−6 = −46 V    

ΔV = −46 V  This  is  an  enormous  shift  in  voltage.  MOSFET  technology  would  not  be  possible  without  the  remarkable  ability  of  SiO2  to  passivate  (e.g.  neutralize)  more  than  99.9%  of  the  dangling  bonds.  

   4) The  gate  electrode  of  an  MOS  capacitor  is  often  a  heavily  doped  layer  of  

polycrystalline  silicon  with  the  Fermi  level  located  at   EF ≈ EC  for  an  n+  polysilicon  gate  and   EF ≈ EV for  a  p+  polysilicon  gate.    These  heavily  doped  semiconductors  act  almost  like  metals.    Sketch  the  following  four  equilibrium  energy  band  diagrams:  

 4a)    An  n-­‐type  Si  substrate  ( N D = 1017 cm-3 )  with  an  n+  polysilicon  gate.    Solution:    (first  separated,  then  together)  

   

Page 7: T=300K# x 2.2 4 O 1.2 100 µm DD S D - nanoHUBWeek13HW... · ECE305# # Spring2015# ECE/305# # 3# Spring2015# HW#Week13#Solutions(continued)# # R CH = 1 Wµ n C ox L V GS −V T) µ

ECE  305     Spring  2015  

ECE-­‐305     Spring  2015  7  

HW  Week  13  Solutions  (continued)    A  small  electron  transfer  will  occur  from  the  gate  to  the  semiconductor.  The  result  is:  

 Note:    The  semiconductor  is  accumulated.  There  is  a  little  depletion  of  electrons  in  the  polysilicon  gate.  This  does  not  happen  for  a  metal  because  there  are  so  many      4b)    An  n-­‐type  Si  substrate  ( N D = 1017 cm-3 )  with  a  p+  polysilicon  gate.    

Solution:  (first  separated,  then  together)    

   Electrons  will  transfer  from  the  semiconductor  to  the  gate.                        

Page 8: T=300K# x 2.2 4 O 1.2 100 µm DD S D - nanoHUBWeek13HW... · ECE305# # Spring2015# ECE/305# # 3# Spring2015# HW#Week13#Solutions(continued)# # R CH = 1 Wµ n C ox L V GS −V T) µ

ECE  305     Spring  2015  

ECE-­‐305     Spring  2015  8  

HW  Week  13  Solutions  (continued)    

   

Note:    The  semiconductor  is  depleted.    We  again  see  some  depletion  in  the  polysilicon  gate.  

 4c)    A  p-­‐type  Si  substrate  ( N A = 1017 cm-3 )  with  an  n+  polysilicon  gate.    

Solution:  (first  separated,  then  together)  

 Electrons  will  transfer  from  the  gate  to  the  semiconductor.    

   

Page 9: T=300K# x 2.2 4 O 1.2 100 µm DD S D - nanoHUBWeek13HW... · ECE305# # Spring2015# ECE/305# # 3# Spring2015# HW#Week13#Solutions(continued)# # R CH = 1 Wµ n C ox L V GS −V T) µ

ECE  305     Spring  2015  

ECE-­‐305     Spring  2015  9  

HW  Week  13  Solutions  (continued)    Note:    The  semiconductor  is  depleted.  The  polysilicon  gate  shows  some  depletion  here  too.  

 4d)    A  p-­‐type  Si  substrate  ( N A = 1017 cm-3 )  with  a  p+  polysilicon  gate.    

Solution:  (first  separated,  then  together)    

   

A  small  electron  transfer  will  occur  from  the  semiconductor  to  the  gate.    

   Note:    The  semiconductor  is  accumulated.    The  polysilicon  gate  also  depleted  a  little.  

   5)    Compute  the  “metal”-­‐semiconductor  workfunction  difference  for  each  of  the  cases  

above.    

5a)    An  n-­‐type  Si  substrate  with  an  n+  polysilicon  gate.    

Solution:  We  must  find  the  location  of  the  Fermi  level  in  the  semiconductor.    Begin  with  

n0 = N D = NCe EFS−EC( ) kBT ,  which  can  be  solved  for  

Page 10: T=300K# x 2.2 4 O 1.2 100 µm DD S D - nanoHUBWeek13HW... · ECE305# # Spring2015# ECE/305# # 3# Spring2015# HW#Week13#Solutions(continued)# # R CH = 1 Wµ n C ox L V GS −V T) µ

ECE  305     Spring  2015  

ECE-­‐305     Spring  2015  10  

HW  Week  13  Solutions  (continued)    

EC − EFS( )q

=kBTq

lnNC

N D

⎛⎝⎜

⎞⎠⎟   NC = 3.23×1019 cm-3  

 

EC − EFS( )q

=kBTq

lnNC

N D

⎝⎜⎞

⎠⎟= 0.026ln 323( ) = 0.150 eV  

 

ΦMS = χ − χ + EC − EFS( ){ } = − EC − EFS( ) = −0.150 eV    

ΦMS = −0.150 eV  

 The  work  function  difference  is  small,  as  can  be  seen  in  the  energy  band  diagram  of  4a).  

 5b)    An  n-­‐type  Si  substrate  with  a  p+  polysilicon  gate.    

Solution  

We  have  the  same  Si  substrate,  so  

EC − EFS( )q

= 0.150 eV  

ΦMS = χ + EG − χ + EC − EFS( ){ } = 1.12 − 0.15 eV =0.97 eV    

ΦMS = 0.97 eV  

 The  work  function  difference  is  close  to  the  bandgap  of  Si,  as  can  be  seen  in  the  energy  band  diagram  of  4b).    Note:    Since  the  left  contact  is  a  p-­‐type  semiconductor  and  the  right  contact  is  an  n-­‐type  semiconductor,  we  can  show  that  the  magnitude  of  the  built-­‐in  potential  is  given  by  the  PN  junction  equation:    

ΦMS q =

kBTq

lnN AN D

ni2

⎝⎜⎞

⎠⎟    

               

Page 11: T=300K# x 2.2 4 O 1.2 100 µm DD S D - nanoHUBWeek13HW... · ECE305# # Spring2015# ECE/305# # 3# Spring2015# HW#Week13#Solutions(continued)# # R CH = 1 Wµ n C ox L V GS −V T) µ

ECE  305     Spring  2015  

ECE-­‐305     Spring  2015  11  

HW  Week  13  Solutions  (continued)    

5c)    A  p-­‐type  Si  substrate  with  an  n+  polysilicon  gate.    

Solution:  

p0 = N A = NV e EV −EFS( ) kBT ,  which  can  be  solved  for    

EFS − EV( )q

=kBTq

lnNV

N A

⎛⎝⎜

⎞⎠⎟   NV = 1.83×1019 cm-3  

 

EFS − EV( )q

=kBTq

lnNV

N A

⎝⎜⎞

⎠⎟= 0.026ln 183( ) = 0.135 eV  

ΦMS = χ − χ + EG − EFS − EV( ){ } = −EG + EFS − EV( ) = −0.985 eV    

ΦMS = −0.985 eV  

 The  work  function  difference  is  very  large  in  magnitude,  as  can  be  seen  I  the  energy  band  diagram  of  4c).    

 5d)    A  p-­‐type  Si  substrate  with  a  p+  polysilicon  gate.    

Solution:  The  same  p-­‐type  substrate,  so:  

EFS − EV( )q

= 0.135 eV  

ΦMS = χ + EG − χ + EG − EFS − EV( ){ } = EFS − EV( ) = 0.135 eV    

ΦMS = 0.135 eV  

 The  work  function  difference  is  very  small  in  magnitude,  as  can  be  seen  in  the  energy  band  diagram  of  4d).