Académique Documents
Professionnel Documents
Culture Documents
Content
1. CMOS inverter: device perspective ........................................................................................................................5
1.1. Fabrication steps ...............................................................................................................................................6
1.1.1. Photolithography or Lithography ...............................................................................................................6
1.1.2. Oxidation..................................................................................................................................................10
1.1.3. Poly-Si deposition ....................................................................................................................................12
1.1.4. Ion-implantation .......................................................................................................................................12
1.1.5. Metallization ............................................................................................................................................13
1.1.6. Forming a transistor (NMOSFET) ...........................................................................................................26
1.1.7. Isolation methods .....................................................................................................................................32
1.1.8. Shallow Trench Isolation (STI) process ...................................................................................................33
1.2. MOS Capacitor ...............................................................................................................................................39
1.2.1. MOS capacitor with gate-voltage .............................................................................................................41
2. CMOS inverter: circuit perspective.......................................................................................................................47
2.1. Static input-output transfer curve characterization .........................................................................................48
2.1.1. Noise-margin and noise-immunity ...........................................................................................................51
Last updated: 10 Feb 2010 (working document)
1
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
2
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
3
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Gate oxide
Poly-Si Gate
W
p substrate
Bulk or Body
5
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Photo-mask / mask:
o High-quality glass containing opaque (chromium) patterns
o Masks having patterns are created at the very end of the design phase using CAD tool mask
generation / ‘tape-out’
o Each reticle (die-reticle) of mask (which is meant for single instance of IC) repeated two-
dimensionally across the entire Si wafer surface
Photoresist / resist:
o Polymer sensitive to UV light
o Positive photresist dissolves when exposed to UV light
o Negative photoresist hardens when exposed to UV light
Etching:
o Reactive Ion Etching: Gaseous mixture to remove material which are not covered by the photoresist
6
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 1-2. (a) SiO2 is grown on Si substrate by oxidation process and (b) SiO2 layer is covered with
photoreist or resist
7
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
8
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
9
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
1.1.2. Oxidation
10
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
11
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
1.1.4. Ion-implantation
12
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Accelerated n-type (Arsenic, As or Phosphorus, P) or p-type (Boron, B) ions are impinged on Si substrate
o Ions are stopped by Si crystal structure damage of the Si structure
o Thermal annealing diffuses the implanted ions and repairs the Substrate Si structure (the regular
geometric arrangement of atoms) by substituting Si atoms with implanted ions at the implanted sites.
1.1.5. Metallization
Interconnect layers:
o Depending on the fabrication process, 4 to 8 metal interconnect layers
o Al interconnect (easier to deposit at relatively lower temperature and adheres well on SiO2)
13
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
14
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
15
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 1-8. Metal-1 etching for forming all Metal-1 layer interconnects
16
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 1-9. Inter-layer SiO2 deposition for isolating Metal-1 and Metal-2 layers
17
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 1-10. Via etching for connecting Metal-1 interconnects to Metal-2 interconnects at required locations
18
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
19
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 1-12. Metal-2 etching to form all interconnects in Metal-2 layer (Note: Metal-2 interconnect routing
is kept in perpendicular direction to w.r.t. Metal-1 interconnect routing for reducing inter-layer stray cap.)
20
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 1-13. Inter-layer SiO2 deposition for isolating Metal-2 layer and the higher layer
21
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 1-14. Via etching for connecting Metal-2 interconnects to higher-layer interconnects at required
locations
22
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
23
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 1-16. higher-layer interconnect etching (in perpendicular direction to adjacent metal layers)
24
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 1-17. Top dielectric and passivation layer and die-pad opening (showing 3-Metal process as example)
25
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 1-18. Field-Oxide isolation areas are formed first, within which exposed islands are available for
forming the NMOSFETs
Figure 1-19. Forming thin gate-oxide (SiO2) and the poly-Si gate: step - 1
26
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
- Poly-Si
- SiO2 (Thin-OX)
- SiO2 (Field Oxide)
- Si substrate (p-type)
Figure 1-20. Forming thin gate-oxide (SiO2) and the poly-Si gate: step - 2
Poly-Si (Gate)
- SiO2 (Thin-OX)
- SiO2 (Field Oxide)
- Si substrate (p-type)
Figure 1-21. Forming thin gate-oxide (SiO2) and the poly-Si gate: step - 3
27
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 1-22. Forming thin gate-oxide (SiO2) and the poly-Si gate: step - 4
28
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
29
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 1-24. Insulating the Poly-Si Gate and poly-Si interconnects before metal contacts to active areas can
be formed
30
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
31
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 1-27. NMOS and PMOS FETs using Field-Oxide (FOX) isolation in LOCOS process
32
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 1-28. NMOS and PMOS FETs using trench isolation in STI process
Silicon wafer
o Crystalline Si
o With current technology, 200mm (8-inch) wafers are common
o To prevent CMOS latch up, heavy doping of the substrate
o Epitaxy layer is lightly doped (helps forming buried layer in BJT / BiCMOS process)
33
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
34
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
35
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
36
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 1-34. Top-view and channel-length side-view and channel-width side-view of NMOS FET
37
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
38
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Si + O2 SiO2
o SiO2 : insulator
o Resistivity = 1012 ohm-cm
o Die-electric constant, = 3.97 0, where 0 = 8.854 F/cm2
39
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
40
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
+VG
V(x)
poly-Si Gate VG
VOX + xOX
- X=0
+ S= QS/COX
S
- p (NA) Si
E(x) = dV(x)/dx
x
Surface charge, QS
41
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
V(x)
poly-Si Gate VG
VOX + xOX
- X=0
+ - - - - - - S
S
- p (NA) Si
E(x) = -dV(x)/dx
x
Ionized acceptors
Bulk charge, QB S= QS/COX QB/COX
42
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
43
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
44
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
45
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
46
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
V DD
S
G
D
V IN V O UT
D
G
S
V SS
47
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
VOUT
Gain = ∞
RIN = ∞
ROUT = 0
NMH = VDD/2 = NML
VIN
48
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
49
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
50
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Static noise-margin
o NMH = VOH – VIH
o NML = VIL - VOL
VID = VIH - VIL (single-ended i/p for CMOS)
VOD = VOH - VOL (single-ended o/p for CMOS)
o Noise-margin is resistive/current load dependant
51
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Noise sources: supply noise, cross talk noise, offset voltage, etc
Cross-talk: capacitive coupling and inductive coupling between two neighbouring wires
52
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
v(t)
i(t)
Supply-noise:
VDD
53
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Offset-error example:
VDD
VDD
Figure 2-8. Offset-error due to additional resistance in current return path (GND)
For good noise immunity, the signal swing (= VOH - VOL) and the noise margins (NMH and NML) have to be
greater than the combined impact of fixed sources of noise
54
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
55
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
v0 v1 v2 v3 v4 v5 v6
3
V (v)
v0
1
v1 v2
-1
0 2 4 6 8 10
t (ns)
56
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
v0 v1 v2 v3 v4 v5 v6
v1 = f(v0) v1 = finv(v2)
v3 f(v) finv(v)
v1
sqrt(
v3
finv(v) f(v)
v2 v0 v0 v2
Requirement to be regenerative:
o Transfer-curve must have a transient region with a gain (absolute value) > 1
o Gain > 1 region is surrounded by two valid regions, where the gain < 1.
57
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
2.1.5. Fan-in
58
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
2.1.6. Fan-out
Fan-out of a gate: the number of receiver/driven gates connected to the output of the driving gate
o The larger the fan-out of the gate, the slower the output switching
To improve the current drive capability
o Transistor may have to be up-sized
o One or more buffer / inverter may have to be inserted between the driver-gate and the driven-gate.
(Recall regenerative property) to improve (1) voltage-swing (2) rise-time and fall-time
Example of large fan-out path: Clock-tree within an IC
59
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
2.1.7. Directivity
Ideally, changes in an output level should not affect unchanging inputs of the same gate: unidirectional
Capacitive coupling between inputs and outputs exits in real circuit
60
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 2-12. CMOS inverter dynamic characterization bench using programmable switched current-load
61
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
62
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Input sensitivity for switching (in saturated region), VID = VIH - VIL
For noise-immunity in ac-coupled switching operation, VOD > VID
VOUT (V)
VOD VOD (V)
ILOAD (A)
VIN (V)
VID (V)
2.2.2.1. Definitions
63
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
V in
50%
tpHL tpLH
V out
90%
50%
10% t
tf tr
64
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
R
vIN
vOUT
CL
Figure 2-16. RC modeling of the inverter output loading for tPD estimation and tR, tF estimation
RC modeling:
o vOUT(t) = VOD (1 – exp(–t/)), where = RC
o t = – ln(1 – x), where x = vOUT(t) / VOD, 0 ≤ x ≤ 1
o t2 – t1 = – ln( (1 – x2) / (1– x1) )
o Example:
for 10% - 90% time, x2 = 90% and x1 = 10%
for 0% - 50% time, x2 = 50% and x1 = 0%
65
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
= ln (2) = 0.69 RC
Rise-time, tR
= Time taken to reach 90% of the of peak-to-peak o/p swing from the 10%
= ln(9) = 2.2 RC
Fall-time, tF
= Time taken to reach 10% of the of peak-to-peak o/p swing from the 90%
= ln(9) = 2.2 RC
66
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
v0 v1 v2 v3 v4 v5
v0 v1 v5
2.2.3. Power-dissipation
67
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
R
vIN
vOUT
CL
68
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
o Assuming VOH is able to reach VDD after charging, the energy stored by CL after the charging of CL is
independent of the R
o Half the energy supplied from the source is lost within the resistor during charging
T 2 T 2
dvout 1 1
v R t iR t dt
DD out L
2 2
ECL V v C dt C V V
L DD OH C V
L OH C LVDD
0 0 dt 2 2
o Upon application of 0V at the input, (1/2) CL VDD2 stored in the resistor gets dissipated in the resistor
Average power:
1 T Vsupply
isupply t dt
T
T 0
Pavg p (t ) dt
0
T
Used for determining battery-life for portable power-source
Used for determining heat dissipation requirement
69
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
70
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
71
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
72
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
The electron inversion layer mobile charge, Qn(y) [charge per unit area under the gate at y, 0 y L]
= - [(VGSn – VTn) – V(y)] COX, where VGSn > VTn
Electric field, E(y) = -dV(y)/dy
Two conditions necessary to establish a current flow:
o VGSn > VTn Excess charge (inversion layer charge formation)
o VDSn > 0 Creation of an electric field across the mobile-charge (inversion) layer
A drift current caused by the electron movement through the channel
Channel acting as an nonlinear resistor
73
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
74
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
ID = – vn(y) Qn(y) W
where,
o vn(y) = the drift velocity of electrons under the field E(x) established along the length (average value
of electric field = –VDS / L) [ L / s ]
vn(y) = – n E(y) = – n dV/dy
n = surface electron mobility approx. half of bulk electron mobility
o W = width of the channel [ m ]
o Qn(y) = charge underneath the gate-oxide per unit length [ C / m ]
Qn(y) = COX (VGS – VT – V(y))
ID = [n dV/dy ] [ COX (VGS – VT – V(y)) ] W
IDn dy (where, y=0 to y=L) = (n COX) W [(VGSn – VTn) – V(y)]) dV (where, V(y=0)=0 to V(y=L)=VDSn)
IDn = (n COX) (W/L) [(VGSn – VTn) – VDSn/2]) VDS
o Process transconductance, k’n = n COX [in A/V2]
o Device transconductance, n = k’n (W/L) [in A/V2]
75
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
76
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 3-4. For a fixed value of VGS, IDn vs. VDSn curve
77
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
78
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
79
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
80
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
81
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 3-8. I-V curve for NMOS FET with channel-length modulation w.r.t. VDSn
82
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
sqrt(IDn)
(sub-threshold / (active)
VGSn
cut-off) VTn
On-set of strong inversion when surface potential induced 2|F| = 2 * (kT/q) ln (NA / ni)
83
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
84
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
85
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 3-10. Velocity saturation of electron drifting under applied electric field along the channel
Active-region operation:
IDn = (n / [ 1 + VDS / (Ecrit L)]) COX (W/L) [(VGSn – VTn) – VDSn/2]) VDS
For large L (long-channel device), [ 1 + VDS / (Ecrit L) ] approaches to 1
86
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Saturation-region operation:
IDn,SAT = (n / [ 1 + VDSn,SAT / (Ecrit L) ]) COX (W/L) [(VGSn – VTn) – VDSn,SAT/2] VDSn,SAT
where VDSn,SAT = (VGSn – VTn) / [ 1 + (VGSn – VTn) / (Ecrit L) ]
IDn
Identical VGSn
Long-channel
Short-channel
Figure 3-11. Velocity saturation effect on drain current for NMOS device
IDn,SAT vs VGSn turns into a linear curve from quadratic when VGSn >> VTn
87
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
In saturated region, MOS device behaving like a current source, where value of the current is governed by
the gate-bias (VGSn)
Change in the current-value of this current source w.r.t. the change in the gate-bias transconductance
Transconductance, gm = IDn / VGSn
= [(n/2) (VGSn – VTn)2] / VGSn
= n (VGSn – VTn)
88
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
89
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 3-12. Linear operation in deep non-saturation (‘triode’) region for the square-law model
90
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
91
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Leff
Mask L
FOX
encroachment
Mask W
Weff
LD Gate
92
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
93
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Gate-to-channel capacitance, C2
= Weff (L – 2 LD) COX = Weff (Leff) COX
Gate-to-Drain capacitance and Gate-to-Source capacitance, C1 = C3
COX (LD) Weff
= (CGSO) Weff or (CGDO) Weff
94
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Gate-to-Bulk capacitance, C5
= CGBO (Leff)
Bulk-to-Channel depletion capacitance across the reverse-bias junction, C4
95
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
96
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Figure 3-17. Drain / Source geometry: the bottom capacitance and the sidewall capacitance
97
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
98
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
If V1 = 0V, V2 = VDD
K0.5(0V, VDD) = (2 0 / VDD) [(1 + VDD/0) 1/2 - 1]
K0.33(0V, VDD) = ((3/2) 0 / VDD)[ (1 + VDD/0) 2/3 - 1]
CBS (0V, VDD) = Cj0 W X K0.5(0V, VDD) + Cjsw 2(W + X) K0.33(0V, VDD)
CBD(0V, VDD) = Cj0 W X K0.5(0V, VDD) + Cjsw 2(W + X) K0.33(0V, VDD)
99
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
100
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
101
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
102
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
VIL calculation:
o IDn (sat) = IDp (non-sat)
(n/2) (VGSn – VTn)2 = (p/2) [2(VSGp– |VTp|) – VSDp]) VSDp
o (n/2) (VIN – VTn)2 = (p/2) [2((VDD - VIN) – |VTp|) – (VDD – VOUT)]) (VDD – VOUT)
o IDn(VIN) = IDp(VIN, VOUT)
Taking differentials,
(dIDn/dVIN) dVIN = (IDp/VIN) dVIN + (IDp/VOUT) dVOUT
o dVOUT/ dVIN = [ (dIDn/dVIN) - (IDp/VIN) ] / (IDp/VOUT)
dVOUT/ dVIN = -1, at VIN = VIH
o VIN ( 1 + n / p) = 2 VOUT – VDD - |VTp| + (n / p) VTn
VIL is obtained solving the above (highlighted) simultaneous equations in terms of VIN and VOUT,
where the first equation is of quadratic form and the second equation is of linear form
VIH calculation:
o (n/2) [ 2(VIN – VTn) - VOUT ] VOUT = (p/2) (VDD - VIN – |VTp|)2
o IDp(VIN) = IDn(VIN, VOUT)
o dVOUT/ dVIN = [ (dIDp/dVIN) - (IDn/VIN) ] / (IDn/VOUT)
dVOUT/ dVIN = -1, at VIN = VIH
103
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
104
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
105
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
o Assumptions:
Step-input voltage, VIN = VDD
Linear time-invariant (LTI) model for COUT
At the beginning portion of the discharge (t 0)
o VGSn = VDD
o VOUT(t=0) VDD NMOS is saturated
dVOUT n
o I Dn COUT VDD VTn 2
dt 2
A linear decay in VOUT(t) until VOUT(t) = (VDD - VTn)
The smaller the COUT, the faster the decay
The higher the n, the faster the decay
dVOUT
dt COUT
n
VDD VTn 2
2
Integrating from t=0, and using the lower boundary condition VOUT(t=0) = VDD
VOUT t VDD
t COUT
n
VDD VTn 2
2
106
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
n
VDD VTn 2
2
VOUT (t ) VDD t , where 0 t t0
COUT
Let t = t0, so that VOUT(t = t0) = (VDD - VTn) Entry into non-saturation from saturation
o VOUT(t = t0) = (VDD - VTn) = VDD – [ n/2 (VDD – VTn)2 / COUT ] t0
COUT
t0 VTn
n 2
2 VDD VTn
o
107
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
ln
o
t t0
VOUT VDD VTn C OUT
o
VOUT 1 e t t0 n 2VDD VTn
108
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
2e t t 0 / n
VOUT (t ) VDD VTn
o
1 e t t 0 / n
Combining saturation and non-saturation discharging,
High-to-low switching time = time taken from V1 = (0.9 VDD) to V0 = (0.1 VDD),
v VDD VTn
dVOUT (t )
v V0
dVOUT (t )
t HL COUT COUT
v V1
I Dn , SAT v VDD VTn Dn , Non SAT
I
VDD VTn
VOUT t V1 VDD VTn
v VDD VTn
dVOUT (t )
n
n
v V1
I Dn , SAT VDD VTn
2
VDD VTn 2
2 V1 2
109
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
v V0
dVOUT (t ) 1 V0 dVOUT dVOUT
v VDD VTn
I Dn , Non SAT n VDD VTn VDD VTn VOUT 2VDD VTn VOUT
t HL s n n
V V0 2VDD VTn
sn 2 Tn ln 1 , where V = 0.1 V and V = V - V = 0.9 V
VDD VTn
0 DD 1 DD 0 DD
V0
110
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
111
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
dVOUT p
o I Dp COUT
dt
2
VDD VTp 2
p
2
V DD VTp
2
o VOUT (t ) t
COUT
A linear charging in VOUT(t) until VOUT(t) = (VDD - |VTp|)
The smaller the COUT, the faster the decay
The higher the p, the faster the decay
o Let t = t1, so that VOUT(t = t1) = |VTp| Entry into non-saturation from saturation
o VOUT(t = t1) = [ p/2 (VDD – |VTp|)2 / COUT ] t1 = |VTp|
o t1 = |VTp| COUT / [ p/2 (VDD – |VTp|)2 ]
o VOUT(t) = [ p/2 (VDD – |VTp|)2 / COUT ] t , where 0 t t1
Toward the end of the charge (t t1)
o IDp = - COUT dVOUT/dt = (p/2) [2(VDD – |VTp|) – VDSp]) VDSp
t t1 / p
VOUT (t ) VDD VDD VTp 2e
t t1 / p
o
1 e
112
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
t LH s p p
VTp V0
s p 2
ln
2 VDD VTp
1
VDD VTp V0 , with V0 = 0.1 VDD and V1 = 0.9 VDD
113
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
tp = (tpHL + tpLH) / 2
114
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
v VDD VTn
dVOUT (t )
v VDD 2
dVOUT (t )
t pHL COUT COUT
v VDD
I Dn , SAT I
v VDD VTn Dn , Non SAT
VDD VTn
VOUT t
v VDD VTn
dVOUT (t ) VTn
n
n
o v VDD
I Dn , SAT VDD VTn
2
VDD VTn 2
2 VDD 2
v VDD 2
dVOUT (t ) 1 VDD 2 dVOUT dVOUT
v VDD VTn
I Dn , Non SAT n VDD VTn VDD VTn VOUT 2VDD VTn VOUT
VDD 2
1 VOUT 2VDD VTn
ln
n VDD VTn VOUT VDD VTn
VDD
2VDD VTn
o
1
ln 2
VDD VTn
n VDD VTn VDD VDD VTn 2VDD VTn
2
1 4VDD VTn
ln 1
n VDD VTn VDD
Last updated: 10 Feb 2010 (working document)
115
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
COUT
n RnCOUT
n VDD VTn
COUT
p R p COUT
p VDD VTp
'
sp
2 VTp
ln
4 VDD VTp
1
VDD VTp VDD
116
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
4.4.5. RC modelling
117
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
118
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
Average value (during charging/discharging between VOH and VOL) of the depletion capacitance, CDB
CBD (VOL, VOH) = Cj0 W X K0.5(VOL, VOH) + Cjsw 2(W + X) K0.33(VOL, VOH)
where,
Km(VOL, VOH) = 0 / [(1-m) (VOH - VOL)] [(1 + VOH /0) (1-m) - (1 + VOL/0) (1-m)]
W and X are the geometry of the Drain
Assuming VOL = 0V and VOH = VDD
CBD (0, VDD) = (2 0 / VDD) [ (1 + VDD / 0)0.5 – 1] + (3 0 / 2 VDD) [ (1 + VDD / 0sw)0.67 – 1]
119
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
VI = VDD/2
If VTn |VTp|
o n p
(W/L)p = (kn’ / kp’) (W/L)n
o VIL = (1/4) [VTn + (3/4) VDD]
120
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
n / p = kn’ / kp’
VI = [VDD - |VTp| + sqrt(kn’ / kp’) VTn ] / (1 + sqrt(kn’ / kp’))
Last updated: 10 Feb 2010 (working document)
121
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
tHL = (sn nINT) + (sn nL) = (sn Rn) CINT + (sn Rn) CL
tLH = (sp pINT) + (sp pL) = (sp Rp) CINT + (sp Rp) CL
122
EC31004 VLSI Engineering (Spring 2009-10) Spring 2009-10
o CL and the required tHL and tLH are the design specifications
o Wn, Ln, Wp, Rp are the design variables
123