Intuition The one core idea
A MOSFET is a switch whose gate is supposed to be the sole boss of a current-carrying
channel between source and drain. Every "short-channel effect" is just one story: when the
channel gets short, the drain and the junction depletion regions start stealing that
control — so before you can understand the effects, you must know exactly what each symbol
(charge, field, depletion width, length) is and what picture it names .
This page assumes you have seen nothing . We build every letter from the ground, in an order
where each one only uses letters already defined. Whenever you want the bigger picture of the
device itself, see MOSFET operation and regions .
Definition Our reference device and sign convention (fix this before any symbol)
Everything below describes an n-channel MOSFET (nMOS) unless stated otherwise. That means:
The body/bulk is p-type silicon; the source and drain are n-type islands in it.
The mobile carriers in the channel are electrons (negative charge).
We take voltages as positive when the named terminal is above the source : V GS > 0 and
V D S > 0 turn an nMOS on. The body is normally tied to the source or to the most negative
supply.
Conventional current I D flows into the drain and out of the source (drain → source
inside the device). The electrons physically move the opposite way , source → drain —
conventional current always points against electron flow. When this page says "carriers walk
source → drain" it means the electrons ; the current I D counts as drain → source.
pMOS is the mirror image: n-type body, p-type source/drain, holes carry current, and
every sign flips (V GS , V D S , V T all negative). Every formula below holds for pMOS if you
read all voltages and charges with their signs reversed. See MOSFET operation and regions .
Before any symbol, look at the object. A MOSFET is four things stacked and side-by-side.
Figure 1 — cross-section of the nMOS: the four terminals, the channel, and the length L .
Definition The four terminals (plain words)
Source (S) — where the mobile electrons come from .
Drain (D) — where they go to .
Gate (G) — a metal plate on top that, through voltage, pulls electrons into a thin sheet
just below it. It never touches the silicon — a glassy insulator sits between.
Body / Bulk (B) — the big block of p-type silicon underneath everything.
The channel is the thin sheet of mobile electrons the gate pulls up, directly under the gate,
bridging source to drain (see Figure 1 ). No channel → no current → switch is OFF. Channel
present → switch ON.
Definition Channel length
L
The distance from the source edge to the drain edge , measured along the direction the
current flows . In Figure 1 it is the horizontal gap the carriers must cross. Units: metres
(nanometres in modern parts). Picture: the length of the bridge the carriers walk across.
W
The dimension perpendicular to current flow — how wide the bridge is (into the page). Units:
metres. Picture: a wider bridge lets more lanes of carriers flow in parallel, so current
scales with W .
Why the topic needs L : every short-channel effect is defined by L getting small . It is
the single knob the whole chapter turns.
L and W are different directions
L is along the current (source→drain). W is across it. Current ∝ W / L : wide and
short = lots of current. Never swap them.
A voltage is an electrical "push" measured in volts (V) . We always measure it between two
terminals — it is a difference. Two matter most:
Definition Body/source-bias voltage
V S B (the body-effect knob)
V S B = voltage of the source above the body . In an nMOS the source–body junction must
stay reverse-biased (V S B ≥ 0 ), never forward-biased, or the junction conducts and the
transistor stops behaving as a switch. Making V S B larger reverse-biases the junction
harder , which widens the bulk depletion region and raises V T — this is the body
effect . See Threshold voltage and body effect .
Why the topic needs both pushes: short-channel misbehaviour is exactly the drain (V D S )
doing part of the gate's (V GS ) job. You cannot state that without naming both.
Q
The amount of electricity, measured in coulombs (C). In this topic Q usually means charge
per unit area (C/m 2 — coulombs spread over a square metre of the device).
Picture: how densely the dots of charge are packed on a sheet.
Definition Electron charge
q
The magnitude of one electron's charge, q = 1.6 × 1 0 − 19 C (a positive number; the
electron itself carries − q ). We multiply it by how many carriers to get total charge.
ε — and its two flavours
How easily a material lets an electric field pass through it. Units: farads per metre (F/m) .
It splits into two pieces:
ε = ε 0 ε r
ε 0 = 8.85 × 1 0 − 12 F/m — the permittivity of free space (a fixed
universal constant).
ε r — the relative permittivity (a pure number, no units): how many times more
permissive this material is than vacuum. For silicon ε r ≈ 11.7
(ε s i ≈ 1.04 × 1 0 − 10 F/m ); for silicon dioxide
ε r ≈ 3.9 (ε o x ≈ 3.45 × 1 0 − 11 F/m ).
Picture: ε r is the material's "springiness" score; multiply by the universal
ε 0 to get the absolute value used in formulas.
Now the single most important derived symbol — the oxide capacitance.
Definition Oxide capacitance per area
C o x
C o x = t o x ε o x [ F/m 2 ]
where t o x = oxide thickness (how thick the glass under the gate is, in metres).
Plain words: how much channel charge the gate builds up per volt it applies.
Picture: the gate + channel form a parallel-plate capacitor; a thinner glass (t o x
small) plates sit closer → more charge per volt → bigger C o x .
C o x is the currency-converter of the whole topic
Every threshold formula has a Q / C o x in it. That is because Q is measured in charge but
V T is measured in volts — and C o x (charge-per-volt) is exactly the exchange rate that
converts one into the other. Whenever you see Q / C o x , read it as "how many volts does this
pile of charge cost the gate?" (Units check: F/m 2 C/m 2 = C/F = V . ✓)
This is the concept the entire chapter hinges on. Build it slowly.
Definition Depletion region
Around any junction (a boundary between differently-doped silicon) the mobile carriers get
swept away, leaving a zone of fixed, immobile dopant charge with no free carriers . It is
"depleted" of things that can move, and it widens under reverse bias (pull the two sides
apart electrically → moat gets wider). Picture: a dead moat around the junction where nothing
flows. See Depletion region physics of pn junctions for the full build.
There are two directions a depletion region can grow, and confusing them is the classic error.
Figure 2 — the two perpendicular depletion directions: sideways (x d S , x d D ) versus
straight down (x d m ).
Definition Vertical gate depletion depth
x d m
How deep the gate's depletion pushes straight down into the bulk (see Figure 2 ). Units:
metres. Picture: the gate's influence sinking downward like roots.
Definition Lateral source/drain depletion widths
x d S , x d D
How far the source and drain junctions push their depletion moats sideways into the channel
(the two wedges in Figure 2 ). These grow as their junctions are reverse-biased harder —
x d D in particular widens as V D S rises. Picture: two triangular wedges creeping in
from the left and right edges of the bridge.
Definition Junction depth
x j
How deep the source and drain diffusions go into the silicon. Picture: the thickness of the
"posts" at each end of the bridge — sets how big the triangular wedges can be.
x d " symbols are perpendicular
x d m points down (gate into bulk). x d S , x d D point sideways (junctions into
channel). The parent note warns of this — never substitute one for the other.
Intuition Why depletion widths
are the short-channel story
When L is large, the two side-wedges are a tiny slice of the bridge → the gate owns almost all
the charge. As L shrinks toward x d S + x d D , the wedges eat the whole bridge → the gate
owns almost nothing. That single geometric fact is the root of threshold roll-off, DIBL, and
punch-through. See Drain-Induced Barrier Lowering (DIBL) .
Definition Threshold voltage
V T
The value of V GS at which the channel just forms — the switch's turn-on voltage. Units:
volts; positive for nMOS , negative for pMOS. Picture: the exact height you must lift the
gate hand before the bridge appears.
Definition Fermi potential
ϕ F and why the surface term is 2 ϕ F
ϕ F (volts) measures how heavily the bulk is doped — the built-in voltage gap between the
doped bulk and intrinsic (undoped) silicon:
ϕ F = q k T ln n i N A
where N A = acceptor doping, n i = intrinsic carrier density, k T / q ≈ 0.026 V
at room temperature. Why the channel needs exactly 2 ϕ F of surface bending: the bulk
surface starts ϕ F below intrinsic (it is p-type). To build an n-channel we must bend it
the same ϕ F above intrinsic — so the electron density at the surface equals the hole
density deep in the bulk (the definition of strong inversion ). Going from − ϕ F to
+ ϕ F is a total swing of 2 ϕ F . That is the whole reason the factor is 2.
Definition Bulk depletion charge
Q B
The fixed dopant charge (per unit area) that the gate must "support" to hold the surface bent by
2 ϕ F . Units: C/m 2 ; it is negative in an nMOS (ionised acceptors), but by
convention we write its magnitude and account the sign in the formula. Its value follows from
depletion-region physics:
Q B = 2 q ε s i N A ( 2 ϕ F + V S B )
Reading it: more doping N A or a larger reverse body bias V S B → more fixed charge to
support → larger Q B . In a short device the side-wedges already hold part of Q B , so the
gate holds less → needs less voltage → V T falls. See Threshold voltage and body effect .
Definition Flat-band voltage
V F B
A fixed offset voltage (volts) from the gate/semiconductor work-function difference and built-in
oxide charge — the "starting-line" voltage before any bending happens.
Putting the long-channel threshold together (every symbol now earned), with V S B = 0 :
V T 0 = starting offset V F B + bend to invert 2 ϕ F + volts to hold Q B C o x Q B
Each term is a voltage the gate must spend : an offset, the bending, and the depletion-charge
cost converted to volts by C o x .
Definition Electric field
E
Voltage spread over distance: E = V / distance , in volts per metre (V/m) . It is the
force per charge pushing carriers along. Picture: the steepness of a hill; steeper = harder
push.
Intuition Why field explodes in a short channel
The lateral field across the channel is E = V D S / L . Keep V D S fixed and shrink L → the
same push over a tiny distance → a huge field. This is why velocity saturation is a
short -channel effect. See Velocity saturation and carrier transport .
μ
How easily carriers move for a given push: velocity v = μ E at low field. Units:
m 2 / ( V⋅s ) . Picture: how slippery the road is. Rougher road (surface
scattering) → lower μ . See Surface scattering and effective mobility .
Definition Saturation velocity
v s a t and critical field E cr i t
Carriers cannot speed up forever; above a critical field E cr i t they hit a top speed
v s a t (m/s) because collisions with the lattice steal the extra energy. Picture: a car
with a speed governor — press the pedal harder past a point and speed stops rising
(Figure 3 ).
Figure 3 — carrier velocity versus field: linear at low field, flattening to v s a t .
I D
The current flowing through the channel — the thing the switch either passes or blocks. Units:
amperes (A). Direction: by convention I D is positive flowing into the drain and out of
the source (drain → source inside the nMOS); the electrons physically drift the other way
(source → drain). For pMOS the sign of I D reverses. The ideal long-channel saturation law is
I D , s a t = 2 1 μ C o x L W ( V GS − V T ) 2 ,
and every short-channel effect is a correction to it.
Definition The three slope parameters — sign, size, and where they enter
Each is a positive fitting number measuring how strongly one leak grows; all three grow as
L shrinks :
CLM parameter λ ( > 0 , units V − 1 , typ. 0.01 –0.3 V − 1 ) :
enters as I D = I D , s a t ( 1 + λ V D S ) — makes I D creep up with V D S .
DIBL coefficient η ( > 0 , units mV/V , typ. tens–hundreds mV/V) : enters as
V T ( V D S ) = V T ( 0 ) − η V D S — the minus sign says more drain voltage lowers V T .
Mobility factor θ ( > 0 , units V − 1 , typ. 0.1 –1 V − 1 ) :
enters as μ e f f = μ 0 / ( 1 + θ ( V GS − V T )) — more overdrive lowers mobility.
Picture: each is a knob measuring "how bad" one leak is; larger value = worse device.
Definition Subthreshold slope
S
How many millivolts of V GS you must remove to cut the leakage current by a factor of 10
(mV/decade). Smaller S = sharper OFF switch. See Subthreshold conduction and leakage .
Each node below is one Layer from this page; follow the arrows to see which symbols feed which
effect.
Layer 2 voltages VGS VDS VSB
Layer 3 Cox from eps over tox
Layer 4 depletion xdS xdD xdm xj
Charge sharing DIBL punchthrough
Layer 6 field E equals VDS over L
velocity saturation vsat Ecrit
Layer 6 mobility mu and theta
Layer 7 current ID and slopes lambda eta theta
The scaling story (Scaling theory (Dennard scaling) ) is why L keeps shrinking; everything
else here is what breaks when it does.
Cover the right side and test yourself — you are ready for the parent topic only if each reveals cleanly.
Is this page's reference device nMOS or pMOS, and which carriers flow? nMOS — p-type body, n-type source/drain, electrons carry the current.
Which way does conventional I D point versus electron flow in an nMOS? I D is positive drain → source inside the device; electrons drift the opposite way, source → drain.
What is channel length L , and in which direction is it measured? The source-to-drain distance, measured along the direction of current flow.
What does C o x = ε o x / t o x physically mean, and its units? Channel charge the gate builds per volt applied, in F/m²; thinner oxide → larger C o x .
What is absolute vs relative permittivity? ε = ε 0 ε r ; ε 0 = 8.85 × 1 0 − 12 F/m is fixed, ε r is a unitless material multiplier (Si ≈ 11.7, SiO₂ ≈ 3.9).
Why is the surface term in V T 0 exactly 2 ϕ F ? The surface must bend from − ϕ F (bulk p-type) to + ϕ F (strong inversion), a total swing of 2 ϕ F .
What is Q B , its units, and its formula? The bulk depletion charge per area (C/m²),
Q B = 2 q ε s i N A ( 2 ϕ F + V S B ) .
What is the difference between x d m and x d S ? x d m is vertical (gate depletion into the bulk); x d S is lateral (source depletion into the channel).
Why does reverse body bias V S B raise V T ? It widens the bulk depletion region, increasing Q B , so the gate must spend more volts.
What are the signs and roles of λ , η , θ ? All positive; I D ( 1 + λ V D S ) , V T − η V D S , μ 0 / ( 1 + θ ( V GS − V T )) — each grows as L shrinks.
Why does the lateral field E = V D S / L blow up in a short channel? Same V D S across a tiny L gives a huge volts-per-metre push.
What does v s a t mean and why does it exist? A top carrier speed reached above E cr i t ; collisions with the lattice steal any extra energy so velocity stops rising.
What does subthreshold slope S measure? The V GS change (mV) needed to change leakage current by one factor of 10.