IV.

Metal-Oxide Field-Effect Transistors (MOSFET)

4.1 Device Operation

Consider a hypothetical semiconductor as is shown. It is constructed

Metal
similar to a parallel-plate capacitor, i.e., a layer of insulator is sand- insulator
− − − −
wiched between a metal plate and a p-type semiconductor. The p-type p−type − − − −
semiconductor

semiconductor includes mobile holes (majority carriers) and stationary, − − − −

negatively-charged dopant ions.

If voltage, v1 , is applied to this device with the positive terminal at- v1

+
−
tached to the metal plate, electric charges, Q = Cv1 , appear on both
+ + + +
terminals of this device. Because the metal plate is positively charged Depletion
region
− − − −
and the electric field is strongest in the vicinity of the insulator/p- − − − −
− − − −
material interface, mobile holes in this regions are repelled, forming a
“depletion” region.
v1
If v1 is increased above a threshold value, Vt , electric field becomes

+
−
strong enough to pull electrons (minority carriers) to the insulator/p- + + + +
material interface. As the holes had been repelled from this region, an Inversion − − − −
layer
“inversion layer” is formed which contains electrons in the conduction Depletion
region
− − − −
− − − −
band. This layer or “channel” is a “virtual” n-type material.

A simplified structure of a metal-oxide field-effect transistor

(MOSFET or MOS for short) is shown. The device is fabri- Source Gate Drain

cated on a p-type substrate (or Body). Two heavily doped SiO2 Metal

n-type regions (Source and Drain) are created in the sub- 111111111111
000000000000
n+ Channel region n+
strate. A thin (fraction of micron) layer of SiO2 , which is L
an excellent electrical insulator, is deposited on the p-type
material in between the source and drain regions. Metal is p−type substrate

deposited on the insulator to form the Gate of the device

(thus, metal-oxide semiconductor). Metal contacts are also Body

made to the source, drain, and body region.

The length of the channel region, L, is the the smallest feature of the transistor on the chip
surface. The drain and source regions are typically much wider (picture is zoomed on the
channel region). The width of the device W (dimension into the page) is also much larger
than L.
The region between the source and the drain (p-type body, the oxide insulator layer and
the gate metal) resembles the hypothetical parallel-plate capacitor above. As discussed, if
a voltage greater than some threshold value is applied between the gate and the body, an

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-1

inversion layer will forms The inversion layer is a virtual n-type “channel” connecting the
source to the drain. Both source and drain regions are heavily doped to provide a source
of free electrons and assist in the formation of this inversion layer. This device is called a
“n-channel” enhancement MOS or NMOS for short.
Let’s first consider the case when no channel is formed. If we apply a voltage between the
drain and the source NO current flows as n/p/n regions between the drain and the source
look like two diodes back to back. This case is called the MOS cut-off mode.
Note that we need to ensure that drain/body and source/body diodes are reverse biased so
that no current can flow from the drain to the body or from the source to the body. For
now, we assume that the source is connected to the body and vDS ≥ 0 for this purpose. We
will examine implications of this assumption later.
We denote the threshold voltage for the formation of the channel as Vt . We apply a voltage
vGS to the NMOS (vGS = vGB ). As long as vGS < Vt , no channel is formed, the MOS is in
cut-off and no current can flow from drain to source (iD = 0).
When, vGS ≥ VT , the virtual n-type channel is formed. As both drain and source region are
also n-type, formation of this channel provide a n-type path for the current to flow from the
drain to the source. The total charge in the channel is |Q| = CVOV . VOV = vGS − Vt is called
the “overdrive” voltage. C is the capacitance of the gate/body structure:
ǫox
|Q| = C VOV with C= W L = Cox W L
tox
where Cox is the capacitance per unit area. ǫox and tox are, respectively, the permittivity
and the thickness of the insulator region.
v v
GS + DS iD
+
Now, if we apply a voltage vDS to the device an electric S
−
G
−
D
field of E = vDS /L appears along the channel. Electrons
in the channel move from the source to drain with a drift 111111111111
000000000000
000000000000
111111111111
n+ n+
velocity of
Induced channel

vDS
vdrif t = µn |E| = µn
L p−type substrate

where µn is the mobility of electrons. The resulting cur-

B
rent, iD is:
dQ dQ dx |Q| |Q| vDS
iD = = × = × vdrif t = µn
dt dx dt L L L
W
iD = µn Cox VOV vDS
L
iD = gDS vDS and gDS = µn Cox (W/L)VOV

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-2

Effectively, the device acts like a resistor. Its conduc- iD

tance, gDS , is controlled by VOV . For this case plot of

v GS = Vt + VOV3
iD versus vDS is a straight line with a slope of gDS . For
VOV < 0 no channel exists (zero conductance). v GS = Vt + VOV2

The above description is correct only for small values of v GS = Vt + VOV1

vDS as in that case, vGD = vGS − vDS ≈ vGS and the in-
vDS
duced channel is fairly uniform (i.e., has the same width v GS < Vt

near the drain as it has near the source). If we increase

vDS (keeping vGS the same), vGD = vGS − vDS becomes
v
smaller than vGS . As such, the size of the channel near v
GS + DS
+
iD
− −
the drain becomes smaller than its size near the source, S G D

as is shown. This is called the “Triode” mode. 111111111111

000000000000
000000000000
111111111111
In this configuration, the charge per unit length near the n+ n+
Induced channel
source is Cox W VOV (similar to that of a straight channel).
The charge per unit length near the drain, however, is p−type substrate

reduced to Cox W (VOV − vDS ). The average charge per

unit length in the channel is:
B

|Q|
= 0.5[Cox W VOV + Cox W (VOV − vDS )] = 0.5Cox W (2VOV − vDS )
L
|Q| vDS W 2
iD = µn = µn Cox (2VOV vDS − vDS )
L L L

Note that for small vDS , the above equation for iD reduces to the one for a straight channel.

When vDS = VOV , the depth of the channel at the drain

becomes effectively zero. It is said that the channel is Source
Channel
Drain
“pinched” off. At this point (vDS = VOV ):

W 2 W 2
iD = 0.5µn Cox (2VOV VOV − VOV )VOV = 0.5µn Cox VOV
L L

If vDS is increased further, the channel pinched-off point remains very close to the drain and
iD remains approximately constant, given by the expression above. This case of vDS ≥ Vt is
called the “Saturation” mode.

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-3

The circuit symbol for an NMOS is shown on the right. The NMOS D
iD

physical structure is symmetric, i.e., drain and source positions can be G B

exchanged without any change in device properties. Similarly, the source

and the drain are indistinguishable on the circuit symbol and the arrow S

is placed on the body (pointing inward as it is p-type). For most ap- D

iD
plications, the body is connected to the source, leading to a 3-terminal
G
element. In this case, the source and the drain are not interchangeable. A
simplified circuit symbol for this configuration is usually used as is shown
S
below (with the arrow on the source, pointing outward as it is n-type).
Directions of currents and voltages are chosen such that they are all ≥ 0
D iD
An NMOS is a 3-terminal device with six parameters (3 voltages i G= 0 +
and 3 currents). Similar to a BJT, two parameters (iS and vGD ) vDS
G
can be written in terms of the other four. Since iG = 0 (iS = +
vGS −
−
iD ), an NMOS has three parameters and its iv characteristics i S = iD
S
equation is in the form of iD = f (vDS , vGS ).

Note that iD = f (vDS , vGS ) is a surface in the 3D-space of (iD , vGS , vDS ) as is shown below
(left). Projection of this surface on the iD -vDS plane (with vGS pointing into the paper) is
shown on the right. The three modes of operation of a MOS are identified on this figure:
1) Cut-off mode in which no channel exists (VOV < 0 for NMOS) and iD = 0 for any vDS .
2) Triode mode in which the channel is formed but not pinched off (VOV ≥ 0 and vDS ≤ VOV ).
3) Saturation mode in which the channel is formed and pinched off (VOV ≥ 0 and vDS ≥ VOV ).

Cut-off: VOV < 0, iD = 0

W 2
Triode: VOV ≥ 0 & vDS ≤ VOV iD = 0.5µn Cox [2VOV vDS − vDS ]
L
W 2
Saturation: VOV ≥ 0 & vDS ≥ VOV iD = 0.5µn Cox V (1 + λvDS )
L OV

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-4

Channel-width Modulation: In saturation, the channel is pinched off and the pinch-
off point remains “approximately” at the drain. If the pinch-off point remains exactly at
the drain, iD would be constant. In reality, iD increases slightly when vDS is increased.
This effect is called “channel-width modulation.” If we extrapolate NMOS iv lines in the
saturation mode, they will all intersect the horizontal axis at −VA . (similar to the Early
effect in BJTs). Channel-width modulation effect is included in the saturation mode iD
equation through the (1 + λvDS ) term with channel-modulation parameter, λ ≡ 1/VA . Note
that VA is much larger than the threshold voltage.
Body Effect: Recall that the drain-body and source-body pn junctions should be reverse
biased. In deriving NMOS iv characteristics, we had assumed that the body and the source
are connected. It is not straight forward to connect the body of every MOS to its source in
an IC chip. The common practice is to attach the body of the chip to the smallest voltage
available from the power supply (zero or negative). In this manner, drain-body and source-
body junctions of all NMOS would be reverse biased. The impact of this approach is to
lower the threshold voltage for the MOS devices slightly. This is called the body effect.
Body effect can degrade device performance. For analysis here, we will assume that body
effect is negligible.
p-Channel Enhancement-Type MOSFET (PMOS)
D
iD
The physical structure of a PMOS is identical to an NMOS except that
G B
the semiconductor types are interchanged, i.e., body & gate are made of
n-type material, source & drain are made of p-type material, and a p-
S
channel is formed. As the sign of charge carriers is reversed, all voltages
and currents in a PMOS have opposite signs compared to those of an D
iD
NMOS. In order to get positive values for voltages and currents in a PMOS, G

the convention is that the drain current is flowing out of the drain and
subscripts for voltages are reversed (as is shown). S

Since the threshold voltage in a PMOS is negative, the over-

D iD
drive voltage is defined as VOV = vSG − |Vtp |. The channel- i G= 0 −
width modulation coefficient is also positive and defined as G vSD
λp = 1/VAp . With these conventions, the iv equations for a −
vSG +
+
PMOS are: S
i S = iD

Cut-off: VOV < 0, iD = 0

W 2
Triode: VOV ≥ 0 & vSD ≤ VOV iD = 0.5µn Cox (2VOV vSD − vSD )
L
W 2
Saturation: VOV ≥ 0 & vSD ≥ VOV iD = 0.5µn Cox V (1 + λvSD )
L OV

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-5

4.2 Solving MOS circuits in DC

A MOS is very similar to a BJT in that the applied voltage vGS controls iD flowing through
the drain-source circuit. As such, we can solve a MOS circuit utilizing a method similar to
that used for BJTs: Write down GS-KVL and DS-KVL, assume the MOS is in a particular
state, solve the circuit with the corresponding MOS equation and validate the assumption.
There are some differences, however:
1) A MOS is controlled with vGS and iG = 0. As such, typically no resistor is necessary in
the gate circuit (as opposed to an RB which was necessary for a BJT).
2) When MOS is in cut-off, iD = 0. However, iD = 0, does not mean that MOS is in
cut-off. This is a very important property that is utilized in the design of CMOS logic gates
(discussed in the next section).
To see this, assume that a MOS is in triode. Condition of iD = 0 gives:

W 2
iD = 0.5µn Cox [2VOV vDS − vDS ]=0 → vDS = 0
L

Thus, iD can be zero if a MOS is in the triode mode and vDS = 0.

We can also reach this conclusion by noting that when a MOS is in cut-off no channel exists
and thus iD = 0. However, if MOS is ON and a channel exists, iD would be zero if no voltage
is applied between the drain and the source terminals to drive iD . Because vDS = 0 < VOV ,
this MOS has to be in triode.
3) MOS iv equations are quadratic. Often, we find two roots in solving a MOS circuit with
one being unphysical. Furthermore, it is usually easier to check for the saturation mode than
the triode mode (as we see below).

Example: In the circuit below, RD = 1 k and VDD = 12 V. Compute vo for vi = 0, 6, and

12 V (µn Cox (W/L) = 0.5 mA/V2 , Vt = 2 V, and λ = 0).
VDD
Part A: vi = 0 V.
RD
GS-KVL: vGS = vi = 0 < Vt → NMOS in cut-off iD = 0 iD vo
DS-KVL: VDD = RD iD + vDS → vo = vDS = VDD = 12 V vi

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-6

Part B: vi = 6 V.

GS-KVL: vGS = vi = 6 → VOV = vGS − Vt = 6 − 2 = 4 V

Since Vt > 0, MOS is ON. Assume the NMOS is in saturation (with λ = 0):

W 2
iD = 0.5µn Cox VOV = 0.5 × 0.5 × 10−3 × (4)2 = 4.0 mA
L
DS-KVL: VDD = RD iD + vDS
12 = 103 × 4 × 10−3 + vDS → vo = vDS = 8 V

Since vDS = 8 > VOV = 4 V, out assumption of MOS in saturation is justified.

Part C: vi = 12 V.

GS-KVL: vGS = vi = 12 → VOV = vGS − Vt = 12 − 2 = 10 V

Since Vt > 0, the NMOS is ON. Assume the NMOS is in saturation (with λ = 0):

W 2
iD = 0.5µn Cox V = 0.5 × 0.5 × 10−3 × (10)2 = 25.0 mA
L OV
DS-KVL: VDD = RD iD + vDS
2 = 103 × 24 × 10−3 + vDS → vo = vDS = −13 V

Since vDS = −13 < VOV = 10 V, our assumption of saturation mode is NOT justified.
Assume the NMOS is in triode:
W 2 2
iD = 0.5µn Cox [2VOV vSD − vSD ] = 0.5 × 0.5 × 10−3 × (20vSD − vSD )
L
DS-KVL: 12 = 103 iD + vDS

The above two equations in two unknown can be solved by substituting for iD into DS-KVL:

12 = 103 × 0.25 × 10−3 × (20vSD − vSD

2
) + vDS
2
vDS − 24vDS + 48 = 0 → vDS = 21.8 V and vDS = 2.2 V

The first root vDS = 21.8 > VOV = 10 V is not acceptable for a MOS in triode. The second
root is the correct one as vDS = 2.2 < VOV = 10 V.
Thus vo = vDS = 2.2 V.

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-7

4.3 MOS transfer function

We will find the transfer function of a MOS by examining the circuit below.
VDD

GS-KVL: vGS = vi RD

DS-KVL: VDD = RD iD + vDS iD vo

vi

Starting at vi = 0, we find vGS = 0 < VT and the NMOS is in

cut-off with vo = VDD . The NMOS remains in cut-off as long as
vi = vGS < Vt .

When vi = vGS = Vt (point A in the figure), the

NMOS turns ON (i.e., a channel is formed). If we
increase vi slightly beyond this point, VOV = vGS −
Vt = vi − Vt is positive but small. Thus, iD would be
small leading to vDS = VDD − RD iD be close to VDD .
As such to the right of point A, vDS > VOV and the
NMOS is in saturation.
As we increase vi (further to the right of point A), VOV increases leading to a larger iD and
a smaller vDS . At some point vDS = VOV and the NMOS enters the triode mode (point B).
For larger vi (and VOV ), iD increases (and vo = vDS decreases) but not as fast as it was in
the saturation mode as is shown.
We can observe this behavior by considering the 3-D iv characteristic surface of the MOS
(figure below left). The DS-KVL, VDD = RD iD + vDS , represents a “plane” in this 3-D space
which is parallel to the vGS axis. The intersection of the DS-KVL plane and the NMOS iv
surface is the operating path of the NMOS (shown below with points A, B, and C identified).
The figure below (right) is the projection of this 3-D structure onto the iD vDS plane with
vGS pointing into the paper. The projection of the KVL plane and the operating path of the
NMOS overlaps and is the load line in this figure.

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-8

Note that if we look at the 3-D picture from the top with iD axis pointing out of the paper,
we should see the transfer function of the NMOS.
The NMOS transfer function provides an insight into MOS functional circuits. In the sat-
uration mode, vo has an approximately linear dependence on vi (similar to a BJT in the
active mode). A MOS in saturation can be utilized in amplifier circuits (discussed later).
MOS switches and RTL logic gates can be constructed by using the transition from cut-off
to triode (similar to BJT in cut-off/saturation modes). There are, however, some differences
with similar BJT circuits:
An NMOS remains in cut-off for vi < Vt and
value of Vt can be set through the manufacturing
process (as opposed to vI < VD0 = 0.7 V for a
BJT).
When a BJT is in saturation, vo = Vsat and is
independent of the transistor and/or circuit pa-
rameters. When a MOS is in triode, vo and iD
depend on MOS parameters as well as the value
of RD . As can be seen from the figure, a larger
RD leads to a lower vo in the low state and a
“rapid” transition from cut-off to triode.

While it may appear the a large RD is require to get “good” transfer function for a logic
gate, large RD reduces gate switching times considerably. NMOS digital gates are not used
today because of the CMOS technology offers a far superior circuit.

4.4 Complementary MOS (CMOS)

Complementary MOS technology employs pairs of NMOS and PMOS. CMOS devices are
more difficult to fabricate than NMOS ones. However, most of MOS circuits are based on
CMOS technology today because of the excellent properties of CMOS circuits (both analog
and digital).
The basic idea is to replace RD in the NMOS inverter circuit above with a PMOS. Recall
that in the NMOS inverter circuit, the transfer function would resembles an ideal inverter
for large RD (low vo in triode mode, rapid transistor from high to low state). The PMOS is
configured such that when vi is high (vo is low), it would act as an effectively infinite resistor.
Furthermore, When vi is low (vo is high), the PMOS acts as a small resistor thus allowing a
high switching speed.

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-9

Two fundamental properties of CMOS logic gates are:

• A Low state of 0 and a High state of VDD , thereby allowing for a maximum voltage
swing. These states are independent of MOS device parameters. As a result, the gate
has a wide noise margin.

• Zero static power consumption. i.e., iD = 0 when the gate is in one the two states. iD
is non-zero only during transition from state to another.

CMOS Inverter
We consider the response of the inverter to vi = 0 (low) and vi = VDD (high). Note that
vo = vDS1 = VDD − vDS2
VDD

GS1-KVL: vi = vGS1 → vGS1 = vi Q2

GS2-KVL: vDD = vSG2 + vi → vSG2 = VDD − vi i D2

vi vo
DS-KVL: VDD = vSD2 + vDS1 i D1

KCL: iD1 = iD2 Q1

Case 1: vi = 0
Since vGS1 = vi = 0 < Vtn , NMOS will be in cut-off. Therefore, iD1 = 0. Since vSG2 =
VDD − vi = VDD > |Vtp |, PMOS will be ON.

vGS1 = v1 = 0 < Vtn → Q1 is OFF → iD1 = 0

vSG2 = VDD − vi = VDD > |Vtp | → Q2 is ON iD2 = 0

where we used KCL and iD1 = 0 to get iD2 = 0.

Since PMOS (Q2) is ON and iD2 = 0, PMOS should be in triode and vSD2 = 0 (see discussion
of page 4-6)

vGS1 = v1 = 0 < Vtn → Q1 is OFF → iD1 = 0

vGS2 = VDD − vi = VDD > |Vtp | → Q2 is ON iD2 = 0 → vSD2 = 0

Then, vo = VDD − vSD2 = VDD .

The above analysis is valid as long as vi < Vtn (Q1 OFF & Q2 ON) and vo remains at
vo = VDD .

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-10

Case 2: vi = VDD
Since vGS1 = vi = VDD > Vtn , NMOS will be ON. Since vSG2 = VDD − vi = 0 < |Vtp |, PMOS
will be in cut-off and iD2 = 0.

vGS1 = v1 = VDD > Vtn → Q1 is ON

vGS2 = vi − VDD = 0 < |Vtp | → Q2 is ON → iD2 = 0

By KCL, iD2 = 0 implies iD1 = 0. Now, NMOS (Q1) is ON and iD1 = 0. Thus, NMOS
should be in triode with vDS1 = 0.

vGS1 = v1 = VDD > Vtn → Q1 is ON iD1 = 0 → vDS1 = 0

vGS2 = vi − VDD = 0 < |Vtp | → Q2 is ON → iD2 = 0

Then vo = vDS1 = 0.
The above analysis is valid as long as vi > VDD − |Vtp | (Q1 ON & Q2 OFF) vo remains at
vo = 0.
In sum, when vi = 0, vo = VDD and when vi = VDD , vo = 0. Therefore, this is an inverter (or
a NOT gate). Note that 1) the low and high states were NOT set by transistor parameters,
and 2) iD1 = iD2 = 0 and the gate has zero power consumption when in either state.
The transfer function of a CMOS inverter is shown below. From our analysis, when vi < Vtn ,
NMOS is OFF and PMOS is in triode. When vi becomes larger than Vtn , NMOS moves
from cut-off to saturation (PMOS still in triode). Increasing vi further leads to both MOS
to be in saturation. The, NMOS transistor moves to triode. Finally, when vi > VDD − |Vtp |
NMOS is in triode and PMOS is in cut-off.
When Vtn < vi < VDD − |Vtp | both MOS are ON (and one is in saturation), a current iD
would flow during the transition between high and low states (as shown below).
NMOS OFF
vo PMOS Triode
NMOS Saturation
PMOS Triode

NMOS Saturation
PMOS Saturation

NMOS Triode
PMOS Saturation
NMOS Triode
PMOS OFF

VDD − |Vtp | vi
Vtn

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-11

For a “matched” pair of MOS (i.e., Vtn = |Vtp |, µn (W/L)n = µp (W/L)p ) the transfer function
would be symmetric about vi = 0.5VDD , vo = 0.5VDD point. In this case, the maximum value
of iD that flows during the transition occurs when vi = 0.5VDD and vo = 0.5VDD .

CMOS NAND Gate

As we saw in the CMOS inverter analysis, transistors in a CMOS logic circuit would be
either in triode with iD = 0 and vDS = 0 or would be in cut-off. As such, in the solution
below: 1) we will use GS-KVLs to find which transistors are OFF (and their iD = 0), 2)
we will use KCL to show that all iD are zero, 3) we will look for any MOS that is ON with
iD = 0. These transistors should have vDS = 0. 4) From the vDS values, we will compute vo .
GS1-KVL: vi = vGS1 → vGS1 = vi VDD

GS2-KVL: v2 = vGS2 + vDS1 → vGS2 = v2 − vDS1

Q3 Q4
GS3-KVL: vDD = vSG3 + v1 → vSG3 = VDD − v1 i D3 i D4
vo
GS3-KVL: vDD = vSG4 + v2 → vSG4 = VDD − v2 i D2
v2
Q2
DS-KVL: VDD = vSD4 + vDS2 + vDS1
i D1
v1
DS-KVL: vSD3 = vSD4 Q1

KCL: iD1 = iD2 = iD3 + iD4

vo = vDS1 + vDS2 = VDD − vSD3 = VDD − vSD4

v1 = 0, v2 = 0 We first find vGS and state of all transistors:

vGS1 = v1 = 0 < Vtn → Q1 is OFF → iD1 = 0

vGS2 = v2 − vDS1 = −vDS1 → Q2 is ?
vSG3 = VDD − v1 = VDD > |Vtp | → Q3 is ON
vSG4 = VDD − v2 = −VDD > |Vtp | → Q4 is ON

While vDS1 can be mathematically negative and still satisfy DS-KVL above, NMOS operation
requires vDS ≥ 0. Thus, Q2 should be OFF (however, this is not used in analysis below).
By KCL iD2 = iD1 = 0 and iD3 + iD4 = iD1 = 0. Since iD ≥ 0 for both PMOS and NMOS,
the last equation can be only satisfied if iD3 = iD4 = 0. We add values of iD to the table
above and look for transistors that are ON and have iD = 0. These transistors (Q3 and Q4)
have to be in triode mode with vSD3 = vSD4 = 0.

vGS1 = v1 = 0 < Vtn → Q1 is OFF → iD1 =0

vGS2 = v2 − vDS1 = −vDS1 → Q2 is ? iD2 =0
vSG3 = VDD − v1 = VDD > |Vtp | → Q3 is ON iD3 = 0 → vSD3 = 0
vSG4 = VDD − v2 = −VDD > |Vtp | → Q4 is ON iD4 = 0 → vSD4 = 0

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-12

Finally, we find vo = VDD − vSD3 = VDD . So, when both inputs are low, the output is HIGH.
We can check and see if our agreement for Q2 OFF is correct. Assume Q2 is ON. This
requires vGS2 > Vtn . Since iD2 = 0 and Q2 is ON, vDS2 = 0 (Q2 in triode). But vo =
vDS1 + vDS2 = VDD leading to vDS1 = VDD and vGS2 = v2 − vDS1 = 0 − VDD = −VDD < Vtn ,
a contradiction of of Q2 being ON. Therefore, Q2 should be OFF.
v1 = 0, v2 = VDD

vGS1 = v1 = 0 < Vtn → Q1 is OFF → iD1 = 0

vGS2 = v2 − vDS1 = VDD − vDS1 → Q2 is ?
vSG3 = VDD − v1 = VDD > |Vtp | → Q3 is ON
vSG4 = VDD − v2 = 0 < |Vtp | → Q4 is OFF → iD4 = 0

In this case, the state of Q2 is unknown. By KCL iD2 = iD1 = 0. Also, iD3 + iD4 = iD1 = 0
leading to iD3 = 0. We add values of iD to the table above and look for transistors that are
ON and have iD = 0. This transistor (Q3) have to be in triode mode with vSD3 = 0.

vGS1 = v1 = 0 < Vtn → Q1 is OFF → iD1 =0

vGS2 = v2 − vDS1 = VDD − vDS1 → Q2 is ? iD2 =0
vSG3 = VDD − v1 = VDD > |Vtp | → Q3 is ON iD3 = 0 → vSD3 = 0
vSG4 = VDD − v2 = 0 < |Vtp | → Q4 is OFF → iD4 =0

Finally, we find vo = VDD − vSD3 = VDD . So, when v1 is LOW and v2 is HIGH, the output
is HIGH.
We can go back and find the state of Q2. We will find Q2 to be OFF (left as an exercise).
v1 = VDD , v2 = 0

vGS1 = v1 = VDD > Vtn → Q1 is ON

vGS2 = v2 − vDS1 = −vDS1 < Vtn → Q2 is OFF → iD2 = 0
vSG3 = VDD − v1 = 0 < |Vtp | → Q3 is OFF → iD3 = 0
vSG4 = VDD − v2 = VDD > |Vtp | → Q4 is ON

We used vDS1 ≥ 0 to find the state of Q2. By KCL iD1 = iD2 = 0. Also, iD3 + iD4 = iD2 = 0
leading to iD4 = 0. We add values of iD to the table above and look for transistors that
are ON and have iD = 0. These transistors (Q1 and Q4) have to be in triode mode with
vDS1 = vSD4 = 0.

vGS1 = v1 = VDD > Vtn → Q1 is ON iD1 = 0 → vDS1 = 0

vGS2 = v2 − vDS1 = −vDS1 < Vtn → Q2 is OFF → iD2 =0
vSG3 = VDD − v1 = 0 < |Vtp | → Q3 is OFF → iD3 =0
vSG4 = VDD − v2 = VDD > |Vtp | → Q4 is ON iD4 = 0 → vSD4 = 0

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-13

Finally, vo = VDD − vSD4 = VDD . So, when v1 is HIGH and v2 is LOW, the output is HIGH.
v1 = VDD , v2 = VDD

vGS1 = v1 = VDD > Vtn → Q1 is ON

vGS2 = v2 − vDS1 = VDD − vDS1 → Q2 is ?
vSG3 = VDD − v1 = 0 < |Vtp | → Q3 is OFF → iD3 = 0
vSG4 = VDD − v2 = 0 < |Vtp | → Q4 is OFF → iD4 = 0

By KCL iD2 = iD1 = iD3 + iD4 = 0. We add values of iD to the table above and look for
transistors that are ON and have iD = 0. As can be seen in equations below, Q1 is ON and
iD1 = 0. Thus, vDS1 = 0. We use this value to find that Q2 should also be ON.

vGS1 = v1 = VDD > Vtn → Q1 is ON iD1 = 0 → vDS1 = 0

vGS2 = v2 − vDS1 = VDD − vDS1 = VDD > Vtn → Q2 is ON iD2 = 0 → vDS2 = 0
vSG3 = VDD − v1 = 0 < |Vtp | → Q3 is OFF → iD3 =0
vSG4 = VDD − v2 = 0 < |Vtp | → Q4 is OFF → iD4 =0

Finally, vo = vDS1 + vDS1 = 0. So, when v1 is HIGH and v2 is HIGH, the output is LOW.
From the “truth table,” the output of this gate is LOW only if both input states are HIGH.
Therefore, this is a NAND gate.

CMOS NOR Gate

VDD

Exercise: Show that this is a NOR gate. v1

Q4
i D4
v2
Q3
i D3 vo

i D1 i D2

Q1 Q2

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-14

4.5 Exercise Problems

Problems 1 to 4. Find iD (µn Cox (W/L)n = µp Cox (W/L)p = 0.4 mA/V2 , λ = 0, Vtn = 3 V
and Vtp = −3 V).

+ − + +
5V 5V
+ −
10V 1V
+ + + +
4V 1V 6V 4V
− − − − − −

Problem 1 Problem 2 Problem 3 Problem 4

Problem 5. Find VS (µn Cox (W/L)n = 0.5 mA/V2 , λ = 0, and Vtn = 0.8 V).
Problem 6. Consider this PMOS with µp Cox (W/L)p = 0.6 mA/V2 , λ = 0, and Vtp = −1 V.
A) For what values of VG , PMOS will be ON?, B) Find the range of VD for which PMOS is
in triode (answer in terms of VG ). C) Find the range of VD for which PMOS is in saturation
(answer in terms of VG ). D) If PMOS is in saturation with iD = 75 µA, find VOV , VG and
the corresponding range of VD .
Problem 7. Find iD and vDS (µCox (W/L) = 0.4 mA/V2 , λ = 0, and Vt = 3 V).
Problem 8. Find vSG , iD , and vSD (µp Cox (W/L)p = 0.4 mA/V2 , λ = 0, and Vtp = −3 V).

20V

1M 1k

7V 10V

7V

VS 5V
VG 1M 1k
1k
10 µ A

− VSS VD

Problem 5 Problem 6 Problem 7 Problem 8

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-15

Problem 9. Find vGS , iD , and vDS (µn Cox (W/L)n = 0.4 mA/V2 , λ = 0, and Vtn = 3 V).
Problem 10. Find vo for vi = 12 V (µp Cox (W/L)p = 0.5 mA/V2 , λ = 0, and Vtp = −2 V).
Problem 11. Find R such that PMOS is in saturation with VOV = 0.6 V (µp Cox = 1 mA/V2 ,
(W/L) = 10/0.18, λ = 0, and Vtp = −0.4 V).
Problem 12. Design the circuit below (i.e., find RS and RD ) in order to get iD = 0.25 mA
and VD = 0. (µn Cox = 60 µA/V2 , W/L = 100/3, λ = 0, and Vt = 1 V).

15V 2.5V

1k 12V 1.8V RD

vi VD
R
vo

1M 2k
1k RS

− 15V − 2.5V

Problem 9 Problem 10 Problem 11 Problem 12

Problem 13. Find VD (µn Cox (W/L) = 0.5 mA/V2 , λ = 0, and Vt = 0.8 V).
Problem 14. Find V1 and V2 (µn Cox (W/L) = 5 mA/V2 , λ = 0, and Vt = 1 V).
Problem 15. Find iD and vDS for A) ignoring channel-width modulation (λ = 0) and B)
including channel-width modulation (λ = 0.05 V−1 ). NMOS has µn Cox (W/L) = 1 mA/V2
and Vt = 0.4 V.
Problem 16. Show that this circuit is a NOR gate with a LOW state of 0.2 V and a HIGH
state of 12 V (µn Cox (W/L)n = 0.5 mA/V2 , Vt = 1 V).

2.5V

Q2
12V
V1
5V 1.8V
10k
Q1
1k 5k
i1
VD V2 vo
v2 v1
1V
1k
Q2 Q1
− 2.5V

Problem 13 Problem 14 Problem 15 Problem 16

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-16

Problem 17. Show that this is a three-input NOR gate.
Problem 18. Show that this is a three-input NAND gate.
VDD VDD

v1
Q6
Q6 Q5 Q4
v2
Q5 vo

v3 v3
Q4 Q3
vo
v2
Q2

Q1 Q2 Q3
v1
Q1

Problem 17 Problem 18

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-17

4.6 Solution to Selected Exercise Problems

Problem 1. Find iD (µn Cox (W/L)n = µp Cox (W/L)p = 0.4 mA/V2 , λ = 0, Vtn = 3 V and
Vtp = −3 V).
+
This is a NMOS transistor with vGS = 4 V and vDS = 10 V.
10V
+
VOV = VGS − Vtn = 1 V VOV > 0 → MOS is ON 4V
− −

vDS = 10 > VOV = 1 V → MOS is in saturation

2
iD = 0.5µn Cox (W/L)n VOV = 0.5 × 0.4 × 10−3 (1)2 = 0.2 mA

Problem 2. Find iD (µn Cox (W/L)n = µp Cox (W/L)p = 0.4 mA/V2 , λ = 0, Vtn = 3 V and
Vtp = −3 V).
−
This is a PMOS transistor with vSG = −1 V and vGD = 5 V. 5V
+

+
VOV = VSG − |Vtp | = −4 V 1V
−

VOV < 0 → MOS is OFF → iD = 0

Note vSD = vSG + vGD = −1 + 5 = +4 V.

Problem 3. Find iD (µn Cox (W/L)n = µp Cox (W/L)p = 0.4 mA/V2 , λ = 0, Vtn = 3 V and
Vtp = −3 V).
+
This is a PMOS transistor with vSG = 5 V and vGD = 6 V. 5V
−

+
VOV = VSG − |Vtp | = 2 V VOV > 0 → MOS is ON 6V
−

vSD = vSG + vGD = 5 − (−6) = 11 V

vSD = 11 > VOV = 2 V → MOS is in saturation
2
iD = 0.5µp Cox (W/L)p VOV = 0.5 × 0.4 × 10−3 (2)2 = 0.8 mA

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-18

Problem 4. Find iD (µn Cox (W/L)n = µp Cox (W/L)p = 0.4 mA/V2 , λ = 0, Vtn = 3 V and
Vtp = −3 V).
+
This is a NMOS transistor with vGS = 4 V and vDS = 1 V.
1V
+
VOV = VSG − Vtn = 1 V VOV > 0 → MOS is ON 4V
− −

Since vDS = 1 = VOV = 1 V, NMOS is at the boundary of saturation and triode modes
(“edge of saturation”). We can use either formulas for iD . Saturation is simpler:

2
iD = 0.5µn Cox (W/L)n VOV = 0.5 × 0.4 × 10−3 (1)2 = 0.2 mA

Problem 5. Find VS (µn Cox (W/L)n = 0.5 mA/V2 , λ = 0, and Vtn = 0.8 V).
7V
Since iD = 10 µA, NMOS should be ON. Assume NMOS in saturation:

2
iD = 0.5µn Cox (W/L)n VOV VS

2
10 × 10−6 = 0.5 × 0.5 × 10−3 VOV → VOV = 0.2 V 10 µ A

− VSS
vGS = VOV + Vtn = 1 V
vGS = VG − VS = 0 − VS → VS = −1 V
vDS = VD − VS = 7 − (−1) = 8 V

Since vDS = 8 > VOV = 0.2 V, assume of MOS in saturation is correct and VS = −1 V.

Problem 6. Consider this PMOS with µp Cox (W/L)p = 0.6 mA/V2 , λ = 0, and Vtp = −1 V.
A) For what values of VG , PMOS will be ON?, B) Find the range of VD for which PMOS is
in triode (answer in terms of VG ). C) Find the range of VD for which PMOS is in saturation
(answer in terms of VG ). D) If PMOS is in saturation with iD = 75 µA, find VOV , VG and
the corresponding range of VD .
5V
Note: vSG = 5 − VG , vSD = 5 − VD , and VOV = vSG − |Vtp | = 4 − VG :
VG

Part A: For PMOS to be ON:

VD

VOV ≥ 0 → 4 − VG ≥ 0 → VG ≤ 4 V

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-19

Part B: For PMOS to be in triode:

vSD ≤ VOV
5 − VD ≤ 4 − VG → VD ≥ VG + 1

Part C: For PMOS to be in saturation:

vSD ≥ VOV
5 − VD ≥ 4 − VG → VD ≤ VG + 1

Part D:

2 2
75 × 10−6 = iD = 0.5µp Cox (W/L)p VOV = 0.5 × 0.6 × 10−3 VOV → VOV = 0.5 V
vSG = VOV + |Vtp | = 1.5 V
vSG = 5 − VG = 1.5 V → VG = 3.5 V

and from the result of part C, we get VD ≤ VG + 1 = 4.5 V.

Problem 7. Find iD and vDS (µCox (W/L) = 0.4 mA/V2 , λ = 0, and Vt = 3 V).
10V
3 3 3
GS-KVL: 7 = vGS + 10 iD = VOV + Vt + 10 iD → 4 = VOV + 10 iD 7V

DS-KVL: 10 = vDS + 103 iD

1k
Obviously, NMOS cannot be in cut-off because if iD = 0, GS-KVL gives
VOV = 4 > 0. Assume NMOS in saturation:
2
iD = 0.5µn Cox (W/L)n VOV
GS-KVL: 4 = VOV + 103 × 0.5 × 0.4 × 10−3 VOV
2

2
0.2VOV + VOV − 4 = 0 → VOV = −7.62 V and VOV = 2.62 V

Negative root is unphysical (we need VOV > 0). Thus, VOV = 2.62 V (vGS = 5.62 V) and

GS-KVL: 4 = VOV + 103 iD → iD = 1.38 mA

DS-KVL: 10 = vDS + 103 iD → vDS = 8.62 V

Since vDS = 8.62 > VOV = 2.62 V, our assumption of NMOS in saturation is justified.

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-20

Problem 8. Find vSG , iD , and vSD (µp Cox (W/L)p = 0.4 mA/V2 , λ = 0, and Vtp = −3 V).
20V
Since iG = 0, the two 1 MΩ resistors form a voltage divider and
VG = 20 × (106 )/(106 + 106 ) = 10 V.
1M 1k

SG-KVL: 20 = 103 iD + vSG + VG = 103 iD + VOV + |Vtp | + VG

→ 20 − 3 − 10 = 7 = VOV + 103 iD
SD-KVL: 20 = 103 iD + vSD + 103 iD = vSD + 2 × 103 iD 1M 1k

Obviously, PMOS cannot be in cut-off because if iD = 0, SG-KVL gives

VOV = 7 > 0. Assume PMOS in saturation:
2
iD = 0.5µp Cox (W/L)p VOV
GS-KVL: 7 = VOV + 103 × 0.5 × 0.4 × 10−3 VOV
2

2
0.2VOV + VOV − 7 = 0 → VOV = −8.92 V and VOV = 3.92 V

Negative root is unphysical (we need VOV > 0). Thus, VOV = 3.92 V (vSG = 6.92 V) and

SG-KVL: 7 = VOV + 103 iD → iD = 3.08 mA

SD-KVL: 20 = vSD + 2 × 103 iD → vSD = 13.8 V

Sine vSD = 13.8 > VOV = 3.92 V, our assumption of PMOS in saturation is justified.

Problem 9. Find vGS , iD , and vDS (µn Cox (W/L)n = 0.4 mA/V2 , λ = 0, and Vtn = 3 V).
15V

GS-KVL: 15 = 103 iD + vGS + 106 iG = 103 iD + VOV + Vtn + 0

1k
→ 12 = VOV + 103 iD
DS-KVL: 15 = 103 iD + vDS + 103 iD − 15
→ 30 = 2 × 103 iD + vDS 1M
1k

NMOS cannot be in cut-off because if iD = 0, GS-KVL gives

− 15V
VOV = 12 > 0. Assume NMOS in saturation:
2
iD = 0.5µn Cox (W/L)n VOV
GS-KVL: 12 = VOV + 103 × 0.5 × 0.4 × 10−3 VOV
2

2
0.2VOV + VOV − 12 = 0 → VOV = −10.64 V and VOV = 5.64 V

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-21

Negative root is unphysical (we need VOV > 0). Thus, VOV = 5.64 V (vGS = 8.64 V) and

GS-KVL: 12 = VOV + 103 iD → iD = 6.36 mA

DS-KVL: 30 = vDS + 2 × 103 iD → vDS = 17.28 V

Sine vDS = 17.28 > VOV = 5.64, our assumption of NMOS in saturation is justified.

Problem 10. Find vo for vi = 12 V (µp Cox (W/L)p = 0.5 mA/V2 , λ = 0, and Vtp = −2 V).
12V
vSG = 12 − 0 = 12 V vi

VOV = vSG − |Vtp | = 12 − 2 = 10 V vo

2k
Since VOV > 0, assume PMOS in saturation:

2
iD = 0.5µp Cox (W/L)p VOV = 0.5 × 0.5 × 10−3 × (10)2 = 25 mA
DS-KVL: 12 = vSD + 2 × 103 iD → vSD = −38 V

Since vSD = −38 < VOV , PMOS is NOT in saturation. Assume PMOS in triode:

2
iD = 0.5µp Cox (W/L)p (2vSD VOV − vSD )
DS-KVL: 12 = vSD + 2 × 103 iD = vSD + 2 × 103 × 0.5 × 0.5 × 10−3 × (20vSD − vSD
2
)
2
→ 0.5vSD − 11vSD + 12 = 0 → vSD = 20.8 V and vSD = 1.15 V

vSD = 20.8 V is unphysical (vSD ≥ VOV = 10 V). Thus, vo = vSD = 1.15 V.

Problem 11. Find R such that PMOS is in saturation with VOV = 0.6 V (µp Cox =
0.1 mA/V2 , (W/L) = 10/0.18, λ = 0, and Vtp = −0.4 V).
1.8V
2
iD = 0.5µp Cox (W/L)VOV = 0.5 × 10−3 × (10/0.18) × (0.6)2 = 10 mA
R

SG-KVL: 1.8 = RiD + vSG = 10 × 10−3 R + VOV + |Vtp |

1.8 = 0.01R + 0.6 + 0.4 → R = 800 Ω
SD-KVL: 1.8 = RiD + vSD → vSD = 1 > VOV = 0.6 V, in saturation

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-22

Problem 12. Design the circuit below (i.e., find RS and RD ) in order to get iD = 0.25 mA
and VD = 0. (µn Cox = 60 µA/V2 , W/L = 100/3, λ = 0, and Vt = 1 V).
2.5V
Since iD > 0, NMOS is ON. Assume saturation:
RD
2
iD = 0.5µn Cox (W/L)VOV VD
VG
2
0.25 × 10−3 = 0.5 × 60 × 10−6 (100/3)VOV → VOV = 0.5 V
VS
vGS = VOV + Vt = 0.5 + 1 = 1.5 V
RS

vGS = VG − VS = 0 − VS → VS = −1.5 V
− 2.5V
vDS = VD − VS = 0 − (−1.5) = 1.5 V

Since vDS = 1.5 > Vt = 1 V, assumption of NMOS in saturation is justified.

GS-KVL: 0 = vGS + RS iD − 2.5 = 1.5 + 0.25 × 10−3 RS − 2.5 → RS = 4.0 k

DS-KVL: 2.5 = RD iD + VD = 0.25 × 10−3 RD + 0 → RD = 10 k

Problem 13. Find VD (µn Cox (W/L) = 0.5 mA/V2 , λ = 0, and Vt = 0.8 V).
5V
Since the gate is connected to the drain, vDS = vGS . This leads to vDS =
vGS > vGS − Vt = VOV . Thus, this transistor is always in saturation. This 1k

configuration is called a diode-connected transistor. VD

Proceed to the solution with vDS = vGS = VOV + Vt :

2
iD = 0.5µn Cox (W/L)VOV
DS-KVL 5 = 103 iD + vDS = 103 × 0.5 × 0.5 × 10−3 VOV
2
+ VOV + 0.8
2
0.25VOV + VOV − 4.2 = 0 → VOV = −6.56 V and VOV = 2.56 V

Negative root is unphysical (we need VOV > 0). Thus, VOV = 2.56 V and VD = vDS = vGS =
VOV + Vt = 3.36 V.

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-23

Problem 14. Find V1 and V2 (µn Cox (W/L) = 5 mA/V2 , λ = 0, and Vt = 1 V).
2.5V

By KCL: iD1 = iD2 = iD . Assume both in saturation

Q2

V1
2
iD = iD1 = 0.5µn Cox (W/L)VOV 1 Q1
3
GS1-KVL: 0 = VGS1 + 10 iD − 2.5 V2

0 = VOV 1 + 1 + 103 (0.5 × 5 × 10−3 VOV

2
1 ) − 2.5
1k

2 − 2.5V
2.5VOV 1 + VOV 1 − 1.5 = 0

→ VOV 1 = −1.0 V and VOV = 0.60 V

Negative root is unphysical (we need VOV > 0). Thus, VOV 1 = 0.6 V (vGS1 = 1.6 V).
Since iD1 = iD2 , transistors have the same µn Cox (W/L), and are both in saturation, VOV 2 =
VOV 1 = 0.6 V and vGS2 = vGS1 = 1.6 V

vGS1 = VG1 − VS1 = 0 − VS1 → V2 = VS1 = −vGS1 = −1.6 V

vGS2 = VG2 − VS2 = 2.5 − VS2 → V1 = VS2 = 2.5 − vGS2 = 0.90 V
vDS1 = VD1 − VS1 = V1 − V2 = 0.90 − (−1.6) = 2.5 V
vDS2 = VD2 − VS2 = 2.5 − V1 = 2.5 − 0.90 = 1.6 V

Since VOV 1 = VOV 2 = 0.6 > 0 and VDS1 = 2.5 > VOV 1 = 0.6 as well as VDS2 = 1.6 > VOV 2 =
0.6, our assumption of both MOS in saturation is justified,.

Problem 15. Find iD and vDS for A) ignoring channel-width modulation (λ = 0) and B)
including channel-width modulation (λ = 0.05 V−1 ). NMOS has µn Cox (W/L) = 1 mA/V2
and Vt = 0.4 V.
1.8V
Part A: λ = 0
5k

2 2
iD = 0.5µn Cox (W/L)n VOV 1 = 0.5 × 10 (0.6) = 0.180 mA
−3
1V

DS-KVL: 1.8 = 5 × 103 iD + vDS → vDS = 0.90 V

Since vDS = 0.9 > VOV = 0.6 V, assumption of NMOS in saturation is justified.

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-24

Part B: λ = 0.05 V−1

2
iD = 0.5µn Cox (W/L)n VOV 1 (1 + λvDS ) = 0.180 × 10 (1 + 0.05vDS )
−3

DS-KVL: 1.8 = 5 × 103 iD + vDS

1.8 = 0.9(1 + 0.05vDS ) + vDS → vDS = 0.86 V
iD = 0.180 × 10−3 (1 + 0.05vDS ) = 0.188 mA

Note that by ignoring channel width modulation, the relative error in iD and vDS is ≈ λvDS
(for this problem, λvDS = 0.05 × 0.90 = 4.5%).

Problem 16. Show that this circuit is a NOR gate with a LOW state of 0.2 V and a HIGH
state of 12 V (µn Cox (W/L)n = 0.5 mA/V2 , Vt = 1 V).
12V

GS-KVL vGS1 = v1 , vGS2 = v2 10k

4 i1
DS-KVL 12 = 10 i1 + vo vo
v2 v1
KCL i1 = iD1 + iD2
vo = vDS1 = vDS2 Q2 Q1

Case 1: v1 = v2 = 0.2. Since vGS1 = 0.2 < Vt = 1 and vGS2 = 0.2 < Vt = 1, both transistors
are in cut-off: iD1 = iD2 = 0. Then, i1 = iD1 + iD2 = 0 and from DS-KVL, vo = 12 V.
So, When v1 = 0.2 (LOW) and v2 = 0.2 (LOW), Q1 and Q2 are OFF and vo = 12 V (HIGH).

Case 2: v1 = 0.2, v2 = 12 V. Since vGS1 = 0.2 < Vt = 1, Q1 is in cut-off and iD1 = 0. Since
vGS2 = 12 > Vt = 1, Q2 is not in cut-off. Assume Q2 is in saturation (VOV 2 = vGS2 − Vt =
11 V). Then:

2 2
iD2 = 0.5µn Cox (W/L)n VOV 2 = 0.5 × 0.5 × 10 (11) = 30 mA
−3

vDS2 = vo = 12 − 104 (iD2 + iD1 ) = −288 V

Since vDS2 = −288 < vOV 2 = 11 V, Q2 is not in saturation. Assume Q2 is in triode:

2 2
iD2 = 0.5µn Cox (W/L)n (2vDS2 VOV 2 − vDS2 ) = 0.25 × 10−3 (22vDS2 − vDS2 )
12 = 104 iD2 + vDS2 → 2
12 = 2.5[22vDS2 − vDS2 ] + vDS2
2
−2.5vDS2 + 56vDS2 − 12 = 0 → VDS2 = 22.2 V and VDS2 = 0.22 V

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-25

First root is not physical (vDS2 = 22.2 > VOV 2 = 11 V). vDS2 = 0.2 V is correct as
vDS2 = 0.22 < VOV = 11 V (our assumption of Q2 in triode is justified).

So, when v1 = 0.2 (LOW) and v2 = 12 V (HIGH), Q1 is OFF, Q2 is in triode and vo = 0.2 V
(LOW).

Case 3: v1 = 12, v2 = 0.2 V. This is similar to case 2. By symmetry:

When v1 = 12 V (HIGH) and v2 = 0.2 (LOW), Q1 is in triode, Q2 is OFF, and vo = 0.2 V
(LOW).

Case 4: v1 = 12, v2 = 12 V. Since vGS1 = 12 > Vt = 1, and vGS2 = 12 > Vt = 1, both

transistors are ON (VOV 1 = VOV 2 = 11 V). Since the two transistors are identical, vGS1 = vGS2
and vDS1 = vDS2 , both are in the same state and iD1 = iD2 = 0.5i1 (we only need to analyze
one of them). Assume both transistors are in triode:

2 2
iD1 = iD2 = 0.5µn Cox (W/L)n [2vDS2 VOV 2 − vDS2 ] = 0.25 × 10−3 [22vDS2 − vDS2 ]
12 = 104 (2iD2 ) + vDS2 → 2
12 = 5[22vDS2 − vDS2 ] + vDS2
2
−5vDS2 + 121vDS2 − 12 = 0 → VDS2 = 22.1 V and VDS2 = 0.11 V

First root is not physical (vDS2 = 22.1 > VOV 2 = 11). vDS2 = 0.11 V is correct as vDS2 =
0.11 < vGS2 − Vt = 12 − 1 = 11 (our assumption of Q2 in triode is justified). By KVL:
vo = vDS1 = vDS2 = 0.11 V.
So, When v1 = 12 V (HIGH) and v2 = 12 V (HIGH), Q1 and Q2 are in triode and vo = 0.1 V
(LOW).
Since the output is HIGH only when both inputs are LOW, this is NOR gate.

ECE65 Lecture Notes (F. Najmabadi), Winter 2012 4-26