The controlling quantity is VSB (for an nMOS, source-to-bulk). The parameter that
sets how stronglyVT responds is the body-effect coefficient γ (gamma).
The gate charge QG must balance two things inside the semiconductor:
the inversion chargeQinv — the actual channel electrons we want, and
the depletion chargeQdep — the immobile ionised acceptor ions in the depleted
region below the channel.
QG=Qinv+Qdep
To turn on the transistor we need enough gate voltage to (a) create the depletion region and
(b) then invert the surface. So threshold is roughly
VT=VFB+2ϕF+CoxQdep
where VFB is flat-band voltage, 2ϕF is the surface potential needed for strong
inversion, and Qdep/Cox is the gate voltage "used up" supporting the depletion charge.
Why this step? Poisson's equation dx2d2ψ=−ερ with a
constant charge density ρ=−qNA integrates twice to a parabola; evaluating from the edge to
the surface pins ψ∝W2. Inverting for W and multiplying by qNA gives the square-root
law — depletion charge grows only as ψ, not linearly.
Now the band bending needed at threshold. With no back-bias the surface must bend by 2ϕF.
With a reverse source-body bias VSB, the total potential the depletion region must sustain is
2ϕF+VSB:
Qdep(VSB)=2qεsiNA(2ϕF+VSB)
Plug into VT and split the increase from the VSB=0 baseline:
Why the subtraction of 2ϕF? Because VT0 (threshold at VSB=0) already
includes the depletion charge for band bending 2ϕF. We only add the extra charge caused
by the back-bias, so we subtract the baseline term. Otherwise we'd double-count.
Imagine the channel is a tiny water pipe and the gate is a hand pressing to open it. Below the
pipe there is soft mud (mobile carriers). If you suck the source higher than the ground
(that's the back-bias), you pull the soft mud away and expose hard, dry, stuck rocks
(the fixed ions). Now your hand (the gate) has to push harder just to clear the rocks before it
can even start opening the pipe. So the "minimum push to turn on" — the threshold — goes up.
And because clearing rocks gets easier per extra scoop, the extra push grows like a square root,
not a straight line.
Dekho, MOSFET ka threshold voltage VT koi fixed constant nahi hai. Jab hum source ko body
(substrate) ke comparison mein upar utha dete hain — matlab VSB>0 (reverse bias) — tab
channel ke neeche jo depletion region hoti hai wo wide ho jaati hai. Zyada depletion region
ka matlab zyada fixed ionised acceptor ions expose ho gaye. Ab gate ko pehle in ions ko
"support" karna padta hai, tabhi channel invert hoga. Isliye transistor ko ON karne ke liye
zyada gate voltage chahiye — yaani VT badh jaata hai. Isi ko body effect kehte hain.
Formula yaad rakho: VT=VT0+γ(2ϕF+VSB−2ϕF). Yahan
γ (gamma) body-effect coefficient hai, aur wo depend karta hai channel doping NA aur
oxide capacitance Cox par: γ=2qεsiNA/Cox. High doping ya
thick oxide ho to γ bada — matlab body effect strong. Sabse important baat: yeh relation
linear nahi, square-root hai. Poisson equation se depletion charge ψ ke saath
badhta hai, isliye shuru mein VT tezi se badhta hai phir dheere.
Ye matter kyun karta hai? Real circuits mein, jaise NAND gate ke andar series-connected
transistors, upar wale transistor ka source ground se upar chala jaata hai, to uska VSB>0
ho jaata hai. Body effect uska VT badha deta hai, overdrive (VGS−VT) kam ho jaata hai,
current girta hai, gate slow ho jaata hai. Isiliye VLSI designers body effect ko kabhi ignore
nahi kar sakte. Bas mantra yaad rakho: "Source up, threshold up — by a root."