Comparison of

MOST and Bipolar transistor

models
Willy Sansen
KULeuven, ESAT-MICAS
Leuven, Belgium

[email protected]

Willy Sansen 10-05 011

Table of contents

• Models of MOST transistors

• Models of Bipolar transistors
• Comparison of MOSTs and Bipolars

Ref.: W. Sansen : Analog Design Essentials, Springer 2006

Willy Sansen 10-05 012

From Bipolar to MOST transistors

100 GHz GaAs

10~40 GHz 10~40 GHz
GaAs
BICMOS
2~10 GHz
BIP
1~2 GHz 1~2 GHz
BICMOS
200 MHz CMOS
CMOS

1994 2000
Ref.Toshiba

Willy Sansen 10-05 013

The SIA roadmap

Year Lmin Bits/chip Trans/chip Clock Wiring

µm Gb/chip millions/chip MHz

1995 0.35 0.064 4 300 4-5

1998 0.25 0.256 7 450 5
2001 0.18 1 13 600 5-6
2004 0.13 4 25 800 6
2007 0.09 16 50 1000 6-7
2010 0.065 64 90 1100 7-8
2003

Semiconductor Industry Association

Willy Sansen 10-05 014

The law of Moore
10

L-Micron

Year
.1
1980 1985 1990 1995 2000 2005

Willy Sansen 10-05 015

ISSCC 2005 paper distribution

40
Analog/RF
35

30

25
Digital
20

15

10

0
350 250 180 130 90 65 nm Lmin

Willy Sansen 10-05 016

Price MPW silicon for different L (nm)

10000

Cost MPW $/sq.mm

1000

100
1000 nm 100

Willy Sansen 10-05 017

MOST layout
B S G D S G+ D+

p+ n+ n+ n+ n+
p

p
WM

LM

Willy Sansen 10-05 018

MOST layout : Cox and CD

G+
S D+
tox Cox εsi
CD =
tsi
n+ n+
CD
εox
Cox =
tox
p tsi
B CD
=n-1
Cox
Willy Sansen 10-05 019
MOST layout : Cox and CD values
εsi = 1 pF/cm
εsi 2εsi(φ - VBD)
CD = tsi = εox = 0.34 pF/cm
tsi qNB
φ ≈ 0.6 V
q = 1.6 10-19 C
Example : L = 0.35 µm; W/L = 8
NB ≈ 4 1017 cm-3
VBD = -3.3 V : tsi = 0.1 µm
CD ≈ 10 -7 F/cm2
Lmin
tox = tox = 7 nm
50 CD
Cox ≈ 5 10 -7 F/cm2 = n - 1 ≈ 0.2
Cox
Willy Sansen 10-05 0110
N-well CMOS technology

Gate oxyde

Polysilicon gate

Willy Sansen 10-05 0111

MOST IDS versus VGS and VDS

VDS = ct
IDS IDS VGS
linear
saturation
VGS

VGS

VGS

0 VT VGS 0 VDS
Saturation because high VDS
Willy Sansen 10-05 0112
Table of contents
• Models of MOST transistors
• MOST as a resistor
• MOST as an amplifier in strong inversion
• Transition weak inversion-strong inversion
• Transition strong inversion-velocity saturation
• Capacitances and fT
• Models of Bipolar transistors
• Comparison of MOSTs & Bipolar transistors

Willy Sansen 10-05 0113

MOST IDS versus VDS

IDS
linear
VDS > VGS-VT
saturation
Linear resistor VGS-VT
Ron = n
1
β (VGS-VT) 0 VGS-VT VDS
W
β = KP L
Willy Sansen 10-05 0114
MOST parameters β , KP , Cox , ...

KPn ≈ 300 µA/V2

W
β = KP L Cox ≈ 5 10 -7 F/cm2

KP = µ Cox εox = 0.34 pF/cm

εox εsi = 1 pF/cm
Cox =
tox tox = 7 nm
Lmin Lmin = 0.35 µm
tox =
50 µ p ≈ 250 cm2/Vs
µ n ≈ 600 cm2/Vs

Willy Sansen 10-05 0115

 Example : Analog switch on CL

VDD We want to switch 0.6 V to a

load capacitance CL of 4 pF.
We want to do this fast,
vIN vOUT
with time constant 0.5 ns.
CL Supply voltage VDD = 2.5 V
VT = 0.5 V
Use standard 0.35 µm CMOS.

Choose
minimum channel length and
find an average VGS !

Willy Sansen 10-05 0116

 Example : Analog switch on RL

VDD We want to switch 0.6 V to a

load resistor RL of 5 kΩ.
vIN vOUT W/L = 8
RL Supply voltage VDD = 2.5 V
0.35 µm CMOS: VT = 0.5 V
vOUT ? Ron ?

Choose
Ron minimum channel length !
vIN vOUT
RL

Willy Sansen 10-05 0117

Body effect - Parasitic JFET

VT = VT0 + γ [ |2ΦF| + VBS - |2ΦF| ]

γ CD
n= = 1+ |2ΦF| ≈ 0.6 V
|2ΦF| + VBS Cox
n ≈ 1.2 … 1.5

γ ≈ 0.5 …0.8 V1/2

Reverse vBS increases |VT| and decreases |iDS | !!!

n = 1/κ subthreshold gate coupling coeff. Tsividis

Willy Sansen 10-05 0118

 Ex. : Analog switch with nonzero VBS

VDD Switch 0.6 V to a

load capacitance CL of 4 pF
vIN vOUT or a load resistor RL of 5 kΩ.
CL W/L = 8 (Ron = 125 Ω @ VBS = 0)
Supply voltage VDD = 2.5 V
0.35 µm CMOS: VT = 0.5 V
VDD vOUT ? for γ = 0.5 V-1

vIN vOUT
RL Start with VBS = 0.

Willy Sansen 10-05 0119

Table of contents
• Models of MOST transistors
• MOST as a resistor
• MOST as an amplifier in strong inversion
• Transition weak inversion-strong inversion
• Transition strong inversion-velocity saturation
• Capacitances and fT
• Models of Bipolar transistors
• Comparison of MOSTs & Bipolar transistors

Willy Sansen 10-05 0120

 MOST IDS versus VGS

IDS vs IDS ~ (VGS-VT)

si IDS = K’n W (VGS-VT)2

wi L
VGS
log IDS K’ =
KP
2n
n= ??
vs
K’n ≈ 100 µA/V2
K’p ≈ 40 µA/V2
VGS
Slope q/nkT VGS IDS ~ exp nkT/q

Willy Sansen 10-05 0121

MOST small-signal model : gm & rDS

D iDS G D
+
G
+ vGS rDS
vGS gmvGS
- -
S=B S diDS S
gm =
dvGS

W W 2 IDS
gm = 2K’ n L (VGS-VT) = 2 K’ n IDS =
L VGS-VT

Willy Sansen 10-05 0122

The transconductance gm

Is gm ~ IDS

or ~ IDS ?
Willy Sansen 10-05 0123
MOST small-signal model : rDS
IDS
linear VEL
rDS = ro =
IDS
saturation
1
λ =
VEL

VEn = 4 V/µmL

0 VGS-VT = VDSsat VDS L = 1 µm

IDS = 100 µA
’ W
IDS = K n (VGS-VT) 2 (1 + λVDS)
L ro = 40 kΩ

Willy Sansen 10-05 0124

MOST single-transistor gain Av

IB 2 VE L
Av = gmrDS =
+ VGS-VT
vout
vin
-
Av ≈ 100

If VEL ≈ 10 V
and VGS-VT ≈ 0.2 V

Willy Sansen 10-05 0125

Design for high gain :

High gain High speed

VGS-VT Low (0.2 V)

L High

VGS-VT sets the ratio gm/IDS !

Willy Sansen 10-05 0126

Example: single-transistor amplifier

We want to realize a three-stage amplifier

with a total gain of 10.000.
We use three single-transistor stages in series.
What minimum lengths do we have to use in
an advanced 65 nm CMOS technology
with VE = 4 V/µm ?

Choose
VGS-VT = 0.2 V !

Willy Sansen 10-05 0127

pMOST small-signal model

D iDS G D
+
G
+ vGS rDS
vGS gmvGS
- -
S=B S S
S=B
+ S S
vGS +
- vGS rDS
G gmvGS
D iDS -
G D
Willy Sansen 10-05 0128
 MOST small-signal model: gm & gmb

G D B
iDS + +
+ + vGS rDS vBS
vGS vBS
- - - -
S gmvGS gmbvBS
diDS diDS
gm = gmb =
dvGS dvBS
gmb CD
= =n-1
gm Cox

Willy Sansen 10-05 0129

Table of contents
• Models of MOST transistors
• MOST as a resistor
• MOST as an amplifier in strong inversion
• Transition weak inversion-strong inversion
• Transition strong inversion-velocity saturation
• Capacitances and fT
• Models of Bipolar transistors
• Comparison of MOSTs & Bipolar transistors

Willy Sansen 10-05 0130

IDS & gm versus VGS : weak inversion

IDS vs wi : weak inversion

si VGS
W
wi VGS IDSwi = ID0 exp nkT/q
L

gm vs
Subthreshold slope :
nkT/q ln(10)
si
IDSwi
wi gmwi =
VGS nkT/q

Willy Sansen 10-05 0131

Transconductance gm versus VGS

gm
mS
gmsat
6 VT

4 si vs

2
wi
0.2 0.5 VGS-VT
0
0 0.5 1 1.5 V VGS

Willy Sansen 10-05 0132

Transition voltage VGSt between wi & si
VGS
W IDS = K’n W (VGS-VT) 2
IDSwi = ID0 exp nkT/q
L L
IDSwi 2 IDS
gmwi = gm =
nkT/q VGS-VT
gmwi 1 gm 2
= =
IDSwi nkT/q IDS VGS-VT
kT
(VGSt-VT)t = 2n
q
Willy Sansen 10-05 0133
Transition Voltage VGSt for different L

kT W kT 2
(VGSt-VT)t = 2n IDSt ≈ K’ (2n )
q n L q

Is independent of channel length L

Is still true in … years !

kT W
(VGSt-VT)t = 2n ≈ 70 mV IDSt ≈ 2 µA for = 10
q L
for nMOST

Willy Sansen 10-05 0134

Transition wi - si

log IDS si
wi
IDSt ~ (VGS-VT) 2
VGS
~ nkT/q

VGS
VT VT + 70mV

Willy Sansen 10-05 0135

Ratio gm/IDS at the transition wi - si

gm
1
IDS kT/q
4x

1
20 gm 2
nkT/q =
IDS VGS-VT
10

IDS
0
0.01 0.1 1 10 100
IDSt

Willy Sansen 10-05 0136

EKV model for smooth wi-si transition

W KP
IDS = K’ VGSTt2 [ ln (1 + ev ) ]2 VGST = VGS-VT K’ =
L 2n
VGST kT
v= VGSTt = (VGS-VT) t = 2n
VGSTt q
≈ 70 mV
Small v : ln (1 + ev )≈ ev

W W VGS-VT
IDS = K’ VGSTt2 e2v = K’ VGSTt2 exp ( )
L L n kT/q
IDSt
Large v : ln (1 + ev )≈v
Enz, AICSP ‘95,
W W 83-114
IDS = K’ VGSTt2 v2 = K’ (VGS-VT) 2 Cunha, JSSC Oct.98
L L 1510-1519

Willy Sansen 10-05 0137

Transition current IDSt between wi & si

W
IDSt = IDS = K’ VGSTt2 IDSt = 2 µA for W/L = 10
L
V=1
i=1 IDS
i= = [ ln (1 + ev ) ]2 inversion
IDSt coefficient

v = ln ( e i - 1)

VGS-VT = VGSTt ln ( e i - 1)

kT
VGSTt = 2n ≈ 70 mV
q

Willy Sansen 10-05 0138

Relation VGS-VT and inversion coefficient i
VGS-VT (mV)
800
VGS-VT =
700 weak inv. moderate inv. strong inv.
600 VGSTt ln ( e i - 1)
500 mV
500

400
kT
VGSTt = 2n
300
q
200 mV
200

100
80 mV
0
0 mV
-100
0.5 2 8 40
-200
IDS
0.01 0.1 1 10 100 i i=
IDSt
Willy Sansen 10-05 0139
Transconductance gm between wi & si
IDS
i= = [ ln (1 + ev ) ]2 gm ≈ ….
IDSt

gm nkT 1-e- i 1
GM = = Large i : GM =
IDS q i i
i
Small i : GM = 1 -
2
Alternative approximation :
1
GM = Large i : GM = 1
1+ 0.5 i + i
i
i
Small i : GM = 1 -
4
Willy Sansen 10-05 0140
GM versus inversion coefficient i
GM
1.0 gm nkT
GM =
0.9 sqrt IDS q
0.8

exp GM =
0.7

0.6
1-e- i
0.5
i
0.4

0.3 1
0.2 1+ 0.5 i + i
0.1
8 IDS
0.0 i=
0.01 0.1 1 10 100 i IDSt

Willy Sansen 10-05 0141

Table of contents
• Models of MOST transistors
• MOST as a resistor
• MOST as an amplifier in strong inversion
• Transition weak inversion-strong inversion
• Transition strong inversion-velocity saturation
• Capacitances and fT
• Models of Bipolar transistors
• Comparison of MOSTs & Bipolar transistors

Willy Sansen 10-05 0142

IDS & gm vs VGS : velocity saturation

IDS vs vs : velocity saturation

si IDSvs = WCoxvsat (VGS-VT)
wi VGS vsat ≈ 107 cm/s

gm vs
gmsat = WCoxvsat
si
wi is absolute max. !
VGS

Willy Sansen 10-05 0143

The saturation region and velocity saturation

IDS saturation
linear

(VGS-VT )vs

Saturation region
Square-law region
(VGS-VT )t

0 VGS-VT = VDSsat VDS

Willy Sansen 10-05 0144

Transconductance gm versus VGS

gm gmsat = WCoxvsat
mS vsat = 107 cm/s
gmsat
6 VT

4 si vs

2
wi
0.2 0.5 VGS-VT
0
0 0.5 1 1.5 V VGS

Willy Sansen 10-05 0145

Velocity saturation : vsat & θ
[large VGS]
K’n W (VGS-VT) 2
L K’n W
IDS = ≈ (VGS-VT)
1 + θ (VGS-VT) θ L
θ
1 + (VGS-VT) K’n W
’ W 2
gmsat ≈ 2K n (VGS-VT) 2 ≈
L [1 + θ (VGS-VT)] 2 θ L

= WCoxvsat
µ 1 1
θL= =
2n vsat Ec Ec is the vertical critical field !

θ L ≈ 0.2 µm/V : For L = 0.13 µm θ ≈ 1.6 V-1

Willy Sansen 10-05 0146
Velocity saturation : θ & RS & vsat

K’n W (VGS-VT) 2
IDS = L
1 + θ (VGS-VT) [large VGS]
K’n W gm 1
gmsat ≈ gmRs = ≈
θ L 1 + gm RS RS

θ µ 1 1
RS = RS ≈ ≈
K’n W/L 2n W K’nvsat W Coxvsat

Willy Sansen 10-05 0147

Transition Voltage VGSTvs between si and vs

K’n W (VGS-VT) 2
L IDSsat = WCoxvsat (VGS-VT)
IDS =
1 + θ (VGS-VT)

K’n W
gmsat = WCoxvsat ≈
θ L

1 vsat Is proportional to
(VGS-VT)vs = ≈ 2nL
θ µ channel length L !!!

≈ 5 L ≈ 0.62 V if L = 0.13 µm
Willy Sansen 10-05 0148
Transition Current IDSvs between si and vs

2n vsat vsat2
IDSvs ≈ K’ WL ( )2 ≈ 100 n εox W
µ µ
IDSvs µCox
≈ 10 A/cm K’ =
W 2n
εox L
Cox = tox =
tox 50

W = 2.6 µm & L = 0.13 µm : vsat = 107 cm/s

IDSvs ≈ 2.6 mA n = 1.4
µ= 500 cm2/Vs

Willy Sansen 10-05 0149

Transconductance gm between si and vs
gmsat = W Cox vsat gmsat ≈ 17 10-5 W/L S/cm

W
gmK’ = 2K’ (VGS-VT) gmK’ ≈ 1.2 10-9 VGST W/L2 S/cm
L
VGST

1 1 1 W 17 10-5
= + gm ≈ L
gm gmK’ gmsat L 1 + 2.8 104 L / VGST in cm

If VGST = 0.2 V, vsat takes over for L < 65 nm (If 0.5 V for L < 0.15 µm)

Willy Sansen 10-05 0150

 Now in velocity saturation ?
VGS-VT ≈ 0.2 V
gm
(0.5 V)
W
mS/µm
gm 2K’
= (VGS-VT)
10 W L
vsat gm 0.06 VGST
≈
W L2 in µm
1
K’ gm
= Cox vsat
W
0.1 0.2 V 0.5 V
10 nm 0.1 µm L 1 gm 0.17
≈
W L
in µm
Willy Sansen 10-05 0151
Range of VGS-VT values for si vs time
10
VGSTvs
VGSTvs/3
VGS-VT (Volts) v.sat
3VGSTwi
si VGSTwi

clear si

.1 si
wi

Year
.01
1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008

Willy Sansen 10-05 0152

MOST operating region in si

gm VGS-VT ≈ 0.5 V
mS VGS-VT ≈ 0.2 V
gmsat
6

4 si vs
VT
2 wi
0.2 0.5 VGS-VT
0
0 0.5 1 1.5 V VGS

Willy Sansen 10-05 0153

gm vs VGS for different L (same tox)
Exercise :

gm gm
L>>Lmin

Lmin ≈ 50 tox

L>>Lmin
Lmin ≈ 50 tox

VGS-VT VGS-VT
W
W = ct = ct
L
Willy Sansen 10-05 0154
gm vs VGS for different tox (≈ Lmin/50)
Exercise :

gm gm

Lmin small

Lmin small

Lmin large Lmin large

VGS-VT VGS-VT
W
W = ct = ct
L
Willy Sansen 10-05 0155
Table : MOST IDS , gm & gm/IDS
Summary :

Ref.: Laker, Sansen : Design of analog …, MacGrawHill 1994; Table 1-4

Willy Sansen 10-05 0156

Gate current

For 0.1 µm CMOS :

tox ≈ 2 nm

JG ≈ 4 10-2 A/cm2

For 10 x 0.5 µm
IG ≈ 2 nA

JG (A/cm2)
L
≈ 4.5 105 exp( - )
6.5
L in nm

Ref. Koh, Tr ED 2001, 259-

Annema, JSSC Jan.05, 135.
Willy Sansen 10-05 0157
Table of contents
• Models of MOST transistors
• MOST as a resistor
• MOST as an amplifier in strong inversion
• Transition weak inversion-strong inversion
• Transition strong inversion-velocity saturation
• Capacitances and fT
• Models of Bipolar transistors
• Comparison of MOSTs & Bipolar transistors

Willy Sansen 10-05 0158

MOST capacitances

Willy Sansen 10-05 0159

MOST capacitances CGS & CGD

rG CGD CDB
G D B

+ +
CGS vGS vBS CSB
- rDS
-
gmvGS gmbvBS
S
2
CGS ≈ 3 WLCox ≈ 2W fF/µm for Lmin
εox
LminCox ≈ Lmin ≈ 50 εox ≈ 2 fF/µm
tox
CGD = WCgdo
Willy Sansen 10-05 0160
MOST fT where iDS = iGS
iGS
G D
iGS = vGS CGS s
+
CGS iDS0
vGS
- iDS = gm vGS
gmvGS rDS

S
2 ’ W µCox
CGS = 3 WLCox gm = 2K L (VGS-VT) K’ =
2n
gm 1 3 µ vsat
fT = = (VGS-VT) or ≈
2π CGS 2π 2n L2
2π L
fmax ≈ fT / 8π rGCGD
Willy Sansen 10-05 0161
Design for high speed :

High gain High speed

VGS-VT Low (0.2 V) High (0.5 V)

L High Low

VGS-VT sets the ratio gm/IDS !

Willy Sansen 10-05 0162

 Maximum fT values versus channel length L
100
f (GHz) fm
fT µ
fTsat fT = (VGS-VT)
2πL2
0.2 … 1 V
vsat
fTsat =
10 2πL
fT
VGS-VT = 0.2 0.5 1V fm =
1 + αBD
CBD
αBD ≈
1 Cox
.1 1 L (micron) 10

Processors
Willy Sansen 10-05 0163
fT model in si and velocity saturation

gm CGS = kW k = 2 fF/µm = 2 10 -11 F/cm

fT =
2π CGS
W 17 10-5
gm =
L 1 + 2.8 104 L / VGST L in cm

1 13.5
fT = GHz
L 1 + 2.8 L / VGST
L in µm

If VGST = 0.2 V, vsat takes over for L < 65 nm

If VGST = 0.5 V for L < 0.15 µm

Willy Sansen 10-05 0164

fT model in si and weak inversion
gm gm nkT 1-e- i
fT = GM = =
2π CGS IDS q i

IDS 1-e- i
gm = but IDS = i IDSt
nkT/q i

IDSt
gm = i (1 - e - i)
nkT/q

fT IDSt K’ VGSTt2 W/L

= i (1 - e - i)
fTH = =
fTH 2π CGS nkT/q 2π WL Cox nkT/q

= 2 µ kT/q
4 K’ nkT/q
≈ i for small i ! =
2π Cox L2 2π L2
Willy Sansen 10-05 0165
 fT versus inversion coefficient i
fT
=
1 fTH
GM
fT
i (1 - e - i)
fTH

0.5 GM =

1-e- i

0 IDS
0.01 0.1 1 10 100 i=
IDSt

Willy Sansen 10-05 0166

Exercise: MOST fT or not fT ? all L= Lmin

1 µ K’IDS IDS
fT = (VGS-VT) = =
2π L2 π Cox WL3 πWLCox(VGS-VT)

IDS VGST W=ct VGST

fT fT fT

IDS VGST=ct W VGST=ct

W=ct > IDS IDS =ct VGST

W= > IDS =ct W VGST

VGST = VGS-VT W IDS

Willy Sansen 10-05 0167

MOST capacitances CSB & CDB

rG CGD CDB
G D B

+ +
CGS vGS vBS CSB
- rDS
-
gmvGS gmbvBS
S
CjSB0
CSB = φjS ≈ φjD ≈ 0.5 … 0.7 V
1 + VSB/φjS
CjDB0
CDB =
1 + VDB/φjD

Willy Sansen 10-05 0168

RF MOST model

CG /5

CGD
G D

RG /3 +
vGS rDS
CGS -
gmvGS
S

CG = CGS + CGD
Ref. Sansen, etal, ACD, XDSL,
Ref. Tin, Tr. CAD, April 1998, 372 RFMOS models, Kluwer 1999

Willy Sansen 10-05 0169

Single-page MOST model

IDS = K’ W (V -V ) 2 VGS-VT ≈ 0.2 V K’n ≈ 100 µA/V2

n GS T K’p ≈ 40 µA/V2
L

W W 2 IDS
gm = 2K’ n L (VGS-VT) = 2 K’ n IDS =
L VGS-VT
VEL
rDS = ro = VEn ≈ 5 V/µmL VEp ≈ 8 V/µmL
IDS
vsat = 107 cm/s

3 µ
1 vsat
fT = (VGS-VT) or now ≈
2π 2n L2
2π L
Willy Sansen 10-05 0170
Growing number of parameters !

BSIM4 : http://www-device.eecs.berkeley.edu/bsim/bsim_ent.html
Model 11 : http://www.semiconductors.philips.com/Philips_Models/mos_models
EKV : http://legwww.epfl.ch/ekv/model.html /model11/index.html

Willy Sansen 10-05 0171

Benchmark tests

1. Weak inversion transition for IDS and gm/IDS ratio

2. Velocity saturation transition for IDS and gm/IDS ratio

3. Output conductance around VDSsat

4. Continuity of currents and caps around zero VDS

5. Thermal and 1/f noise
6. High frequency input impedance (s11) and
transimpedance (s21)

Willy Sansen 10-05 0172

Table of contents

• Models of MOST transistors

• Models of Bipolar transistors
• Comparison of MOSTs & Bipolar transistors

Willy Sansen 10-05 0173

Bipolar transistors
WB WB

Lateral PNP transistor

NPN transistor

Willy Sansen 10-05 0174

 Bipolar transistor ICE versus VBE

C iCE
iBE iCE
B
+ 0.7 V : not more
vBE
not less
-
E vBE
0 0.5 1V
VBE ICE
kT/q IBE =
ICE = IS exp β
is leakage current
IS ≈ 10 -15 A kT/q = 26 mV at 300 K
β ≈ 10 … 1000

Willy Sansen 10-05 0175

Bipolar transistor small-signal model : gm & ro
rB
C iCE B’ B C
+
B vBE
+ rπ ro
vBE gmvBE
-
-
E E E

diCE ICE gm 1
gm = = = ≈ 40 V-1
dvBE kT/q ICE kT/q
dvBE dvBE β VE
rπ = = β = ro = VEn ≈ 20 V
diBE diCE gm ICE VEp ≈ 10 V

Willy Sansen 10-05 0176

Bipolar transistor capacitance Cπ

rB Cµ C
B’ B

+
CCS
rπ Cπ - vBE ro
gmvBE
E
S
CjBE0
Cπ = CjBE + CD CjBE = φjE ≈ 0.7 V
1 + VBE/φjE

CD is the diffusion capacitance

Willy Sansen 10-05 0177

Diffusion capacitance CD

rB B
Cµ C
B’

+
CCS
rπ Cπ - vBE ro
gmvBE
E S

QB diCE ICE
CD = = τF = τFgm = τF
vBE dvBE kT/q
WB2 WB
Base transit time τF = or now ≈
2Dn vsat
≈ 10 … 200 ps
Willy Sansen 10-05 0178
Bipolar transistor capacitances Cµ & CCS

rB B
Cµ C
B’

+
CCS
rπ Cπ - vBE ro
gmvBE
E
S
CjBC0
Cµ = CjBC CjBC =
1 + VBC/φjC
CjCS0
CCS = CjCS CjCS = φjC ≈ φjS ≈ 0.5 V
1 + VCS/φjS

Willy Sansen 10-05 0179

Bipolar transistor fT
rB B
Cµ C
B’

+
CCS
rπ Cπ - vBE ro
gmvBE
E=S

gm 1 1 vsat
fT = = or ≈
2π Cπ 2π 2π WB
τF + CjBE+Cµ
gm

For a current drive ! fmax ≈ fT / 8π rBCµ

Willy Sansen 10-05 0180

Bipolar transistor fT versus ICE
1
fT
2πτF 1 1
fT =
2π
τF + CjBE+Cµ
gm

ICE
1
2πfT 1 kT 1
slope
= τF + (CjBE + Cµ)
2πfT q ICE
τF
1/ICE slope

Willy Sansen 10-05 0181

Single-page Bipolar transistor model

VBE
kT/q
ICE = IS exp IS ≈ 10 -15 A kT/q = 26 mV at 300 K

ICE VE
gm = ro = VEn ≈ 20 V VEp ≈ 10 V
kT/q ICE

1 1 vsat
fT = or ≈
2π Cje+Cjc 2π WB
τF +
gm

Willy Sansen 10-05 0182

Table of contents

• Models of MOST transistors

• Models of Bipolar transistors
• Comparison of MOSTs and Bipolars

Willy Sansen 10-05 0183

Comparison MOST - Bipolar

β?

CD
n=1+
Cox

4… 6 x

Ref. Laker Sansen Table 2-8

Willy Sansen 10-05 0184

Comparison MOST - Bipolar : minimum VDS

IDS
VDSsat ≈ VGS-VT

IDS
VGS-VT ≈
K’n W
L

VCEsat ≈ kT/q’s
0 VCEsat VDSsat VDS
Ref. Laker - Sansen Table 2-8

Willy Sansen 10-05 0185

Comparison MOST - Bipolar : gm/IDS ratio

gm
1
IDS kT/q

4 X
1
V-1 20
nkT/q

10

IDS ICE
0
0.01 0.1 1 10 100 i

Willy Sansen 10-05 0186

Design plan for gm :

IDS = K’n W (VGS-VT) 2

L
W W 2 IDS
gm = 2K’ n L (VGS-VT) = 2 K’ n IDS =
L VGS-VT

4 variables with 2 equations >> 2 free variables

Choose VGS-VT and L !

Willy Sansen 10-05 0187

Comparison MOST - Bipolar

dvi2

vsat ≈ 107 cm/s

Ref. Laker Sansen Table 2-8

Willy Sansen 10-05 0188

Table of contents

• Models of MOST transistors

• Models of Bipolar transistors
• Comparison of MOSTs and Bipolars

Ref.: W. Sansen : Analog Design Essentials, Springer 2006

Willy Sansen 10-05 0189

Reference books on Transistor models
T. Fjeldly, T. Ytterdal, M. Shur, “Introduction to Device Modeling and
Circuit Simulation”, Wiley 1998.
D. Foty, “MOSFET Modeling with SPICE, Prentice Hall
K. Laker, W.Sansen, "Design of Analog Integrated Circuits
andSystems", MacGrawHill. NY., Febr.1994.
A. Sedra, K.Smith, "Microelectronic Circuits", CBS College
Publishing, 2004.
Y. Taur, T. Ning, “Fundamentals of Modern VLSI Devices”
Cambridge Univ. Press, 1998.
Y. Tsividis, “Operation and modeling of the MOS transistor”,
McGraw-Hill, 2004.
A. Vladimirescu “The SPICE book”, Wiley, 1994

Willy Sansen 10-05 0190

References on Analog Design
P.Allen, D.Holberg, "CMOS Analog Circuit Design", Holt, Rinehart and Winston. 1987,
Oxford Press 2002
P.Gray, P.Hurst, S. Lewis, R.Meyer, "Analysis and Design of Analog Integrated
Circuits", Wiley, 1977/84/93/01
R.Gregorian, G.Temes, "Analog MOS Int. Circuits for Signal Processing", Wiley, 1986.
Huijsing, Van de Plassche, Sansen, "Analog Circuit Design", Kluwer Ac.Publ.
1993/4/5….
D.Johns, K.Martin, "Analog integrated circuit design", Wiley 1997.
K.Laker, W.Sansen, "Design of Analog Integrated Circuits and Systems",
McGraw Hill. NY., Febr.1994.
H.W.Ott, "Noise reduction techniques in Electronic Systems", Wiley, 1988.
B. Razavi, “Design of analog CMOS integrated circuits”, McGraw Hill. NY., 2000.
A.Sedra, K.Smith, "Microelectronic Circuits", CBS College Publishing, 1987.

Willy Sansen 10-05 0191