The MOS Capacitor: N Polysilicon Gate Gate Oxide 3.9 0
The MOS Capacitor: N Polysilicon Gate Gate Oxide 3.9 0
The MOS Capacitor: N Polysilicon Gate Gate Oxide 3.9 0
M
O
S
n+ polysilicon gate
gate oxide
ox = 3.9 o
0
p-type
s = 11.7 o
Oxide = SiO2 ... a near-perfect insulator. We assume zero charge in the oxide for
this course --> electric field is constant and potential is linear in the oxide.
n+ polysilicon has a potential which is the maximum possible in silicon:
n+ = 550 mV
p-type substrate has potential which is p = - 60 mV log (Na / 1010)
,,
,,,,,,
,,,,,,
metal
interconnect to gate
n+ polysilicon gate
Vox,o +
_
VBo
_
+++++++++++++++++++
gate oxide
,,,,,,,,,,
,,,,,,,,,,
metal interconnect to bulk
o(x)
charge
density
0
-tox
Eo(x)
electric
field
-tox
o(x)
n+
potential
-tox
Oxide doesnt permit any steady-state current between the n+ poly gate and the
substrate. Therefore, if we wait long enough for transient currents to die out, the
electron and hole currents are zero -Jn = 0
and
Jp = 0
Even though the structure isnt in equilibrium, the absence of current implies
that we can relate potential to carrier concentration in the silicon substrate (since
thats all we assumed in deriving the 60 mV rule.)
,,
,,,,,
,,,,,
metal
interconnect to gate
+
_ mn+
gate oxide
ox = 3.9 o
n+ polysilicon gate
+
_
VGB
+V
_ ox
0 +
_VB
,,,,,,,,,
,,,,,,,,,
p-type
s = 11.7 o
+ pm
_
When VGB = VFB, the gate is shifted from its thermal equilibrium potential (n+)
to a new value of VFB + n+ = -(n+ - p) + n+ = p, which is the same potential
as the p-type bulk. Therefore, there is no potential drop across the MOS structure
in flatband
(x)
charge
density
-tox
E(x)
electric
eld
-tox
x
(x)
250mV
potential
-tox
-250mV
VGB + n+
-500mV
-750mV
-1.0V
If we continue to make the gate-bulk voltage more negative, the gate will take on
a negative charge QG < 0. The substrate has a positive charge, which comes from
holes that are attracted by the negative gate charge
(x)
- QG (accumulated holes)
charge
density
-tox
0
QG
E(x)
-tox
electric
field
x
+
QG
Eox =
ox
(x)
-tox
potential
0
x
-250mV
VGB + n+
VGB - VFB
-500mV
-750mV
-1.0V
-1.25V
Now we make VGB > VFB. Note that thermal equilibrium falls into this range of
applied bias.
(x)
QG
charge
density
Xd
0
-tox
-qNa
E (x)
electric
eld
0
-tox
Xd
(x)
1V
VGB + n+
potential
VGB
500 mV
-tox
s = 185 mV
Xd
x
-500 mV
Keep increasing VGB --> surface potential keeps increasing. At some point, the
surface is n-type (i.e., we say that it is inverted) and the electron charge makes a
signicant contribution to the charge density.
How do we model this phenomenon? We approximate that onset of inversion as
the point where the electron concentration ns at the surface is the same as the
hole concentration Na in the bulk. (In other words, the surface is as n-type as
the bulk is p-type.)
The gate-bulk potential at the onset of inversion is called the threshold voltage,
VTn. To nd the threshold voltage, we need to consider the electrostatics in
depletion (no electrons at the surface at the onset of inversion) -- with the surface
potential equal to the opposite of the bulk potential:
s, max = p
(x)
1.5 V
VTn + n+
1V
s,max = - p = 420 mV
Vox
500 mV
VTn - VFB
Xd,max
-tox
0
-500 mV
VB,max
- p
The bulk charge in inversion is found from the depletion width Xd,max
2 p
Q B, max = qN a X d, max = qN a --------------------------------------- = 2q s N a ( 2 p )
( ( 1 2 )qN a ) s
where the relationship between the depletion width Xd,max and the drop across
the depletion region s,max - (p) = -p - p = -2p can be found from Poissons
Equation.
The threshold voltage is the sum of the atband voltage (which cancels the builtin potential drop from gate to bulk), the drop across the oxide at the onset of
inversion, and the maximum potential drop across the depletion region
1
V Tn = V FB 2 p + --------- 2q s N ( 2 p )
a
C
ox