Bakir Chapter5
Bakir Chapter5
Bakir Chapter5
MOS Structure
MOSFET Basics & Device Physics
MOSFET Circuit Models
SPICE Analysis
Literature:
MOS Capacitor
N-Type
vacuum level
metal work function (fundamental property of particular metal)
= 4.28 eV for Al
= 4.33 eV for Ti
= 4.50 eV for Cr
= 4.55 eV for W
= 5.10 eV for Au
= 5.65 eV for Pt
semiconductor work function (depends on material & doping)
electron affinity = E0 EC (fundamental property of semiconductor)
= 4.03 eV for Si
= 4.07 eV for GaAs
=4.00 eV for Ge
MOS Capacitor
Equilibrium Energy Band Diagram
N-Type
No band bending
under equilibrium,
because M = S!
MOS Capacitor
Biased Band Diagram
N-Type Semiconductor
Note:
EF is constant inside the
semiconductor because there is
no dc carrier flow through the
insulator
No band bending inside the
metal because it is considered an
equipotential region
Constant energy slope inside
insulator because there are no
charges inside it and thus the
electric field is constant across it
Pierret, Fig. 16.5
Depletion
small Vg < 0
ps = ni e
= ni e
i,bulk
i,bulk
$
%
(Ei,surface EF )/kT
(EF Ei,bulk )/kT
= nbulk = ND
Inversion
Vg < VT
P-Type
1
(x) = #$Ei (x) Ei,bulk %&
q
Surface potential s
s =
1#
%
$Ei,bulk Ei,surface &
q
Fermi potential F
(
1$
& )> 0 for p type
E
E
%
F'
q i,bulk
*< 0 for n type
( kT $N &
-+ ln + A , for p type
- q % ni '
F = )
- kT $ND &
- q ln + n , for n type
% i'
*
F
s = 2 F at onset of inversion
s > 2 F in inversion
0 < s < 2 F in depletion
qNA
d
dx K s0
K s0
qNA
(x) =
(W x)
K s0
qNA
= (x) =
(W x)2
2K s0
qNA
s =
W2
2K s0
Width of Depletion Region W and Maximum Width at Onset
of Inversion WT:
2K s0
W=
s
qNA
2K s0
4K s0kT #NA &
WT =
2F =
ln % (
2
qNA
q NA
$ ni '
How does s depend on Vg in depletion mode?
EFm
EFs
No charges inside the oxide with thickness x0, i.e. oxide = 0, yields
doxide oxide
=
=0
dx
K o 0
oxide =
oxide dx = x0 oxide
x 0
The electric field inside the oxide can be obtained from the
boundary condition stating that the dielectric displacement D
at the oxide-semiconductor interface must be continuous:
Doxide = Dsemi,x=0
oxide = K s semi
Ko
Ks = 11.8 for Si
Ko = 3.9 for SiO2
Ks
x0
Ko
Vg = s +
Ks
2qNA
x0
s
Ko
K s 0
semi (0)
0 s 2 F
s 0
qNA
qNA 2K
semi (0) =
W=
K s 0
K s 0 qNA s
=
2qNA
s
K s 0
MOS Capacitor
Gate Voltage Vg vs.
Surface Potential s
P-Type
Surface potential is
rather rapidly varying
with Vg in the depletion
region, but not in the
accumulation or
inversion region
In the accumulation and
inversion regions,
changes in the applied
voltage are dropped
almost totally across the
oxide
Accum.
Depletion
Inversion
MS = qMS = M S 0
Before
Contact
After
Contact
Sze, Fig. 6.9
VFB
1
=
Co
% 1 xo
(
'
x (x) dx*
x
'& 0 0
*)
VFB
Q + Qm + Qot
= MS f
Co
1
with e.g. Qm =
x0
xo
x m (x) dx
N-Channel MOSFET
Device Operation
Vg < VT: Depletion (or Accumulation)
No conductive path between
source S and drain D
Drain current ID = 0 for all VD > 0
Vg VT: Inversion
Conductive channel (n-type)
between source S and drain D
Drain current ID > 0 for all VD > 0
Open Question:
ID(VD) Characteristic?
Jaeger, Blalock, Fig. 4.5
5.2.1 MOSFET
Qualitative Theory
ID(VD) Characteristic
ID = R1
VD
channel
For a given Vg, ID increases linearly
with VD
Rchannel decreases with increasing
Vg, thus the slope of ID vs. VD
increases with increasing Vg
= 0 with VD,sat = Vg VT
D =VD,sat
Carrier mobility in
channel n n(bulk)
GND
because of
coordinate system
+Jn in y direction
Jn Jny q nn y
Pierret, Fig. 17.6
=
d
= +q nn
dy
ID =
ny
dx dz = Z
x (y)
c
d
Jny dx = Z
q n (x,y) n(x,y) dx
dy
0
!####
"####$
n (y) Qn (y)
( )
Assumption:
does not depend on x
ID = Z n (y) Qn (y)
Effective
mobility
d
dy
Assumption:
n (y) = n
dy = Z
Charge in inversion
layer at y [C/cm2]
VD
(y) Q (y) d = Z Q
n
VD
d
0
!
#"#
$
n
=?
Qgate = Qsemi = Qn
Qgate Co Vg =
!"#
[C/cm2 ]
K
Co (Vg VT ) = o 0 (Vg VT )
$
xo
Co / A
K o0
Qn =
(Vg VT )
xo
VD
Z nC o
ID =
L
Z n
0 Qn d = L
VD
K o0
0 x (Vg VT ) d
o
%
VD2 ( for 0 VD VD,sat and Vg VT
'(Vg VT )VD
*
2 *)
'&
VD,sat = Vg VT
Square-Law Theory
ID,sat =
ZnCo
2L
(Vg VT )2
MOSFET IV Characteristic
Square-Law Theory
Z nC o "
VD2 %
ID =
$(Vg VT )VD '
L $#
2 '&
MOSFET IV Characteristic
Square-Law Theory
Linear/Non-Linear Regime (Triode Region)
0 VD VD,sat = Vg VT
Z nC o "
VD2 %
ID =
$(Vg VT )VD '
L $#
2 '&
Z nC o "
ID
(Vg VT )VD %&
#
L
for VD Vg VT
Z nC o "
2$
(V
V
)
T %
2L # g
MOSFET IV Characteristic
Square-Law Theory
MOSFET IV Characteristic
Linear Regime
NMOS device (n-channel MOSFET) in common-source
configuration operated in linear regime
Example: VT = VTN = 1 V
K n = CO = 250 A / V 2
Jaeger, Blalock,
Fig. 4.7
Z nC o #
(Vg VT )VD %&
$
L
1
1
#
%
#Z C
%
I
* = ) n o (V V )* = R (V )
Ron = ) D
g
T
on
g
) VD V 0 *
L
$
&
$
&
D
ID
Voltage-Controlled Resistor
MOSFET IV Characteristic
Limitations of Square-Law Theory
Use Charge-Sheet or
Exact-Charge theories
qNA WT
Q
2F B
Co
Co
Threshold Voltage
Work Function Difference MS
In reality M S, i.e. a voltage
MS = q-1 (M S) has to be
applied to the gate to achieve
flat band condition
As a result, the threshold
voltage is shifted by the work
function difference:
VT = MS + 2F
QB
Co
n-channel
p-channel
>0
<0
MS
<0
<0
QB
<0
>0
Threshold Voltage
Work Function Difference MS
Threshold Voltage
Oxide Charges
With the types of oxide charges
and their charge density
(i)
(ii)
(iii)
(iv)
Co Co
Co
Co
Co
with 0 cm, cot 1; the charge density Qit of the interface-trapped
charges depends on the surface potential s
Threshold Voltage
Adjustment by Implantation
In modern device processing (e.g. n-channel and p-channel
MOSFETs in CMOS processes), the threshold voltage is
adjusted to the desired value by ion implantation, implanting
a controlled number of either boron or phosphorous ions into the
near-surface region of the semiconductor (in the channel region)
Under inversion bias, the additional dopant adds to the charge
near the Si-SiO2 interface (with charge density QI) and, thus,
results in a VT shift of the structure:
VT = QI / Co
Depending on the dopant type, both positive (for acceptor atoms,
B) and negative (for donor atoms, P) VT shifts are possible
Example: QI = 5 1011 cm-3 boron, xo = 10 nm yields VT = +0.23 V
MOS Transistor
Circuit Symbols
5.3.1 DC Models
NMOS & Square-Law Theory
Long Channel
Approximation
Short Channel
Approximation
Saturation Region
VGS > VT; VDS > VDsat
ID =
Triode Region
VGS > VT; VDS VDsat
2 %
2 &
"
#
VDS
VDS
ID = K n $(VGS VT )VDS
' ID = K n %(VGS VT )VDS
(
2 '&
2 ('
$#
%$
Cut-Off Region
VGS < VT
ID = 0
Kn
(VGS VT )2
2
ID =
Kn
(VGS VT )2 (1+ VDS )
2
ID = 0
K n = nCox (W / L)
Kn
(VGS VT )2 = 50 A
2
VDS = 5 V
R1
V =4V
R1 + R2 DD
VEQ
VDD R2
R1R2
= 600 k
R1 + R2
IGREQ
!
= 0,because IG = 0
+ VGS +
(ID + IG )
"#$
= ID ,because IG = 0
RS
V
)
#
2
EQ
GS
GS
T
2
$
VEQ = VGS + IDRS %
ID =
=ID ,because IG =0
Saturation
Kn
(VGS VT )2
2
Current Sources
Current sources are widely
used to establish transistor
working points in integrated
circuits
Ideal current source delivers
constant current independent
of voltage across the source
How can we use MOSFETs
as effective current sources?
MOS transistors operated in
saturation region (pinched-off
region) deliver constant
current for VDS > VDsat
Kn
(VGS VT )2 (1+ VDS1)
2
(1+ VDS2 )
Kn
(VGS VT )2 (1+ VDS2 ) = IREF
I
2
(1+ VDS1) REF
Kn
(VGSN VTn )2
2
K
= ID1 = IREF = p (VGSP VTp )2
2
ID1 =
ID4
274 k
v DS = VDS + v ds
iG = IG + ig
iD = ID + id
ig = y11v gs + y12 v ds
id = y21v gs + y22 v ds
!i $ ! y
y12 $ ! v gs $
g
11
# & = ##
&& ## &&
#i & y
" d % " 21 y22 % " v ds %
iG
v GS
iD
v GS
Qpoint
Qpoint
IG
VGS
ID
VGS
y12 =
Qpoint
y22 =
Qpoint
iG
v DS
iD
v DS
Qpoint
Qpoint
IG
VDS
Qpoint
ID
VDS
Qpoint
Long - Channel ( = 0)
=0
Short - Channel ( 0)
=0
y12
=0
=0
y21 = gm
= K n (VGS VT )
y22 =
1
ro
=0
2 ID
VGS VT
Kn
ID
(VGS VT )2 =
2
1+ VDS
iG = 0
ID
VGS VT
2
y22
ID
ID
iD = ID + id =
(v GS VT )2
Kn
K
2 $
(VGS + v gs VT )2 = n "#(VGS VT )2 + 2(VGS VT )v gs + v gs
%
2
2
By comparison we find
id =
Kn "
2 $
2(VGS VT )v gs + v gs
#
%
2
or
v gs 2(VGS VT )
PMOS
BJT
IC
(kT/q)
Transconductance gm
Relation
MOSFET
2ID
2K nID
VGS VT
1 + VDS
1
ID
ID
Input Resistance r
o o (kT/q)
=
gm
IC
Output Resistance ro
VA + VCE VA
IC
IC
VA + VCE
VA
(kT/q)
(kT/q)
2( 1 + VDS )
VGS VT
Small-Signal Requirement
v be (kT/q)
v gs 2(VGS VT )
IC = IS e
qVBE /kT
" V %
1 $1+ CE '
VA &
#
ID =
Kn
V VT
2 GS
) (1+ V )
DS
DC Analysis:
Find equivalent circuit by
replacing all capacitors with
open circuits and inductors
with short circuits
Find the Q-point from the DC
equivalent circuit using
appropriate large-signal model
for the transistor
AC Analysis:
Find AC equivalent circuit by replacing all capacitors by short circuits and
all inductors by open circuits, replacing DC voltage sources by ground
connections and replacing DC current sources by open circuits
Replace FET by its small-signal model
Simplify the resulting AC circuit as much as possible
Common-Source Amplifier
DC Analysis
R1
V = 5.21 V
R1 + R2 DD
REQ =
R1R2
= 243.2 k
R1 + R2
Common-Source Amplifier
AC Analysis
Additional Parameters:
Ri = 1 k, R3 = 100 k
RG = R1 || R2 = 243.2 k
Small signal parameters:
gm = y21 =
2 ID
= 1.23 mS
VGS VT
1
1 + VDS
ro =
=
= 54.9 k
y22
ID
RL = ro || RD || R3= 3835
Calculate vgs, vo and vo/vi:
RG
v gs =
v = 0.996 v i
Ri + RG i
v o = RLgmv gs = 4.72 v gs
A v = v o / v i = RLgm
RG
= 4.7
Ri + RG
Common-Source Amplifier
AC Analysis
A v gmRL
Common-Source Amplifier
Input/Output Resistance
Rout
ro RD
vx
=
= ro ! RD RD = 4.3 k
ix
Input Resistance
Output Resistance
Saturation Region
iD =
KP W
v GS VT
2 L
) (1+ LAMBDA v )
DS
Threshold Voltage
VT = VTO + GAMMA
v SB + PHI PHI
Junction Capacitances
CJ =
CJO
MJ
! v $
#1+ R &
" PB %
and CJSW =
CJSWO
MJSW
! v $
#1+ R &
" PB %
Name
Default
KP
20 A/V2
VTO
1V
Threshold Voltage
VT
Surface Potential 2 F
PHI
0.6 V
Body Effect
GAMMA
LAMBDA
UO
600 cm2/Vs
CGDO
CGSO
CGBO
CJ
Grading Coefficient
MJ
0.5 V0.5
Transconductance
Zero-Bias Threshold Voltage
Mobility
Comment
calculated
Name
Default
Sidewall Capacitance
CJSW
MJSW
0.5 V0.5
TOX
100 nm
IS
10 fA
Built-In Potential
PB
0.8 V
RD
RS
Oxide Thickness
Comment
Name
Default
KP
20 A/V2
VTO
1V
LAMBDA
Transconductance
KP W
iD =
v VT
2 L GS
) (1+ LAMBDA v )
DS
v DS %
W"
iD = KP
$ v GS VT
' v DS 1+ LAMBDA v DS
L #
2 &
VT = VTO
CJ = CJSW = 0
ID =
Kn
VGS VT
2
) (1+ V )
DS
VT = VT0 +
VSB + 2F 2F
Jaeger, Blalock,
Fig. 4.13&14
5.4.2c Capacitances
Internal capacitances limit high-frequency performance of
electronic devices (including MOSFET)
In logic applications, capacitances limit switching speed
In amplifiers, capacitances limit frequency at which useful
amplification can be achieved
Capacitances of NMOS transistor operated in triode region:
CGC
WL
+ CGSO W = C!!ox
+ CGSO W
2
2
C
WL
= GC + CGDO W = C!!ox
+ CGDO W
2
2
CGS =
CGD
Gate-Drain Capacitance:
CGSO and CGDO are overlap capacitances originating from
overlap of gate with source/drain; they are specified as
capacitance per unit width with units [F/m]
In addition, source-bulk and drain-bulk capacitances CSB and
CDB exist between source/drain and bulk substrate (body); these
capacitances are associated with junction capacitances of reverse
biased pn-junctions between source/drain and substrate; each
capacitance consists of a component (CJ in [F/m2]) proportional to
the bottom area of source/drain and a sidewall component (CJSW
in [F/m]) proportional to the perimeter length
Saturation and cut-off region: see Jaeger, Blalock, page 206-207
SPICE Model
Q-Point of Common-Source Amplifier
SPICE results:
VGS = 3.340 V
VDS = 3.935 V
ID = 1.440 mA
SPICE Analysis
AC Sweep: Frequency Transfer
AC sweep analysis provides frequency transfer plot as output
Analysis of choice to investigate frequency dependencies
SPICE Analysis
AC Sweep: Frequency Transfer
Vin = 1 mVp
Why do we
see a low
frequency
but no high
frequency
cut-off?
SPICE Analysis
Transient Analysis
Transient signal analysis provides waveform (at a given signal
frequency) as output signal
Analysis of choice to investigate phase relations and signal distortions
SPICE Analysis
Transient Analysis: Small Amplitude
Vin = 0.1 Vp
Av = 4.64
SPICE Analysis
Transient Analysis: Large Amplitude
Vin = 1 Vp
Gain: |A|
What is the
origin of the
distortion?
4.70