Lecture 3. Understanding Transistors:
Technology Characterization Technology Characterization
Jaeha Kim
Mixed-Signal IC and System Group (MICS) Mixed Signal IC and System Group (MICS) Seoul National University
jaeha@ieee.org
j @ g
Motivations
Good circuit design starts with understanding transistors that you build circuits with
that you build circuits with
In a sense, you are asking the following questions:
Wh t I th t i t f (i t t )?
What can I use these transistors for (intent; usage)?
For each intent, what are the metrics that describe its quality?
Acknowledgements:
Prof. Boris Murmann at Stanford
Ref: Willy M.C. Sansen, “Analog Design Essentials”, Ch. 1.
Every Device Has a Purpose
When building a circuit, the designer utilize a different property for each device (it is the “design intent”)
property for each device (it is the design intent )
What are the possible ways to use transistors?
Ways to Utilize MOS Transistors
Current source
Resistor
VCCS
VCCS
Switch
MOS as a Resistor
MOS transistor operating in linear region acts as a resistor
Linear region: VGS > Vth and VDS < VDS,sat,
According to the long-channel model:
And if VDS << VGS-Vth (=VDS,sat)
R
onvs. Technology
For constant gate overdrive (Vov = VGS-Vth), the on- resistance per square decreases with technology resistance per square decreases with technology
But, the maximum available VGS-Vth has been scaling
d ( V L f 0 35 d t 90 )
down (e.g. VDD Lmin from 0.35um down to 90nm)
As a result, the minimum Ron stayed roughly constant Then what about below 65nm?
Then what about below 65nm?
Exercise: Analog Switch on C
L Required Ron < 0.5ns / 4pF = 125-ohms
Ron varies VGS-Vth ( avg. of 2.0V and 1.4V)
Calculate the minimum W/L required
Actual Ron is found higher than predicted – can you guess why?
Body Effect
As bulk voltage (VBS) drops the increased drops, the increased reverse-bias increases the depletion charge to the depletion charge to be inverted
Vth increases!
Vth increases!
The parameter is technology dependent
Check out how body effect is scaling with technologyy g gy
MOS as a VCCS
+
gmvgs vgs
-
MOS transistor operating in saturation region can be approximated as a VCCS
approximated as a VCCS
Long-channel model (square-law model) states:
Closer Look at G
m: I
DSvs. V
GS Weak inversion:
Strong inversion:
Velocity saturation:y
Closer Look at G
m: G
mvs. V
GS Weak inversion:
Strong inversion:
Velocity saturation:y
Transition Between Weak & Strong Inversion
By equating the gm-expressions for weak and strong
inversion regions we can find where the transition occurs inversion regions, we can find where the transition occurs
Transition point is: p
It means, to operate transistors in strong-inversion, gate-
It means, to operate transistors in strong inversion, gate overdrive must be at least 70mV
Note this is independent of the channel length L
MOSFET Small-Signal Model
! !
gd gb
gs
gg C C C
C ! Cdd ! Cdb Cgd
Note: body effect (gmb) term is not included Note: body effect (gmb) term is not included
Output Resistance (r
o= 1/g
ds)
Non-zero gds (= dIDS/dVDS) is caused by two main effects
Channel length modulation (CLM)
Channel length modulation (CLM)
Threshold voltage variation (DIBL)
Typically modeled as Iyp y DSDS (1+V( DSDS) or (1+V) ( DSDS/VEEL))
CLM: the effective channel length decreases as VDS
Leffeff = L - L L= (V ( DD-VDsatDsat))
DIBL: drain voltage can influence the field at the source of short-channel devices and therefore change Vth
of short channel devices and therefore change Vth
VTH = VTH0 – VDS
g is not a good parameter for your designs to rely on g is not a good parameter for your designs to rely on
MOS Capacitance Model
C - gate capacitance
Cg gate capacitance
Cjc – depletion layer
C C junction caps
Csb, Cdb junction caps
Col "overlap capacitance"
Gate Capacitance vs. V
GSJunction Capacitance at Source/Drain
Junction capacitances are nonlinear, too
C is a function of junction bias
CJ is a function of junction bias
1.0
0.8 0.9
bitrary units]
0.6 0.7
Capacitance [arb
N+ junction area
0.4 0.5
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8
N d lt (V)
N+ junction perimeter P+ junction area P+ junction perimeter
Node voltage (V)
Basic Figures of Merit
• Current efficiency
Square Law
• Current efficiency
– Want large gm, for as little
current as possible D
m
I g
VOV
2
• Transit frequency gm 3 VOV
– Want large gm, without large Cgg I t i i i
Cgg 2 L2
• Intrinsic gain
– Want large gm, but no gds
ds m
g g
VOV
2
Device Characterization
* gmid.sp
* NMOS characterization, L=0.18um
.param gs=0.7 .param dd=1.8
vds d 0 dc 'dd/2' vgs g 0 dc 'gs'
mn d g 0 0 nch L=0.18um W=5um
.op
.dc gs 0.2V 1V 10mV DD
.probe ov = par('gs-vth(mn)') .probe gm_id = par('gmo(mn)/i(mn)')
* For BSIM4, use cggbm in the following line .probe ft = par('1/6.28*gmo(mn)/cggbo(mn)')
b d (' ( )/ d ( )')
.probe gm_gds = par('gmo(mn)/gdso(mn)')
.options post brief dccap .inc cmos018.sp
end .end
g
m/I
DPlot
40
30
35 0.18um NMOS
2/VOV
BJT (q/kT)
20 25
m/I D [S/A]
5 10 15
g m
-0.20 -0.1 0 0.1 0.2 0.3 0.4 0.5
5
Transit Frequency Plot
50
0.18um NMOS 40
0.18um NMOS Square Law Model
20 30
f T [GHz]
gm
f 1
10
f 20
gg
T C
f 2
-0.20 -0.1 0 0.1 0.2 0.3 0.4 0.5
V [V]
V [V]
Intrinsic Gain Plot
80
0.18um NMOS
60
70 Long Channel Model, =0.3
30 40 50
g m/g ds
Short Channel Device
10 20 30
-0.20 -0.1 0 0.1 0.2 0.3 0.4 0.5
10
Assignment – Technology Characterization
Plot the Id-Vds curves for W/L=20/2 nMOS & pMOS
Characterize Ron (from linear region) for each VGS
Characterize Ron (from linear region) for each VGS
Characterize gm (from saturation region) for each VGS
Plot Id-Vgs curves for Vds=Vdd/2
Plot Id-Vgs curves for Vds=Vdd/2
Also plot as gm and gds as function Vgs
Characterize the capacitance components
Characterize the capacitance components
The components listed in slide 13 vs. Vds or Vgs
Measure second order effects such as:
Measure second-order effects such as:
Body effect: Vth vs. VBS
Short-channel effect: Vth vs. L
Assignment – cont’d
Plot the following parameters for a reasonable range of g /ID and channel lengths for various technologies
gm/ID and channel lengths for various technologies
Transit frequency (fT)
Intrinsic gain (gm/gds)
Current density (ID/W)
In addition, tabulate relative estimates of extrinsic capacitances
Cgd/Cgg and Cdd/Cgg
Tip: try to automate the procedure as much as you can
Using mulan/simba script
Transit Frequency Plot
NMOS, 0.18...0.5um (step=20nm), V
DS=0.9V
20 25
L=0 18um
15 20
T [GHz]
L=0.18um
5 10
f T
5 10 15 20
5
g /I [S/A]
L=0.5um
g /I [S/A]
Sweet Spot
NMOS, 0.18...0.5um, step=20nm
L=0 18um 200
S/A]
L=0.18um
• gm/ID ~ 10..12 S/A can be a good
150
*f T [GHz*S can be a good
choice for designs in which power
d d
50 100
g m/I D* and speed are
equally important
5 10 15 20
50
Intrinsic Gain Plot
NMOS, 0.18...0.5um (step=20nm), V
DS=0.9V L 0 5
90
100 L=0.5um
70 80
g m/g ds
40 50
g 60
5 10 15 20
30 40
g /I [S/A]
L=0.18um
g /I [S/A]
Current Density Plot (Sizing Chart)
NMOS, 0.18...0.5um (step=20nm), V
DS=0.9V
101
W [A/m] L=0.18umI D/W
L=0.5um
5 10 15 20
V
DSDependence
102
NMOS, L=0.18um
VDSDS=0.9V • V dependence
VDS=0.4V VDS=1.4V
• VDS dependence is relatively
weak
D/W [A/m]
• Typically OK to work with plots
t d f
101
I D generated for
VDD/2
5 10 15 20
g /I [S/A]
gm/I
D [S/A]
Extrinsic Capacitances (1)
1
NMOS, L=0.18um
Cdd/C
Again, usually
0.8
Cdd/C
gg
Cgd/C
0.70 gg
OK to work with estimates taken at V /2
0 4 0.6
taken at VDD/2
0.2 0.4
0.24
0 0.5 1 1.5
0
Extrinsic Capacitances (2)
0.8
NMOS, g
m/I
D=10S/A, V
DS=0.9V
C /C
0.6 0.7
Cgd/C
gg
Cdd/C
gg
0.4 0.5
0.2 0.3
0.2 0.25 0.3 0.35 0.4 0.45 0.5 0
0.1
L [m]