Fermi level and Fermi-Dirac distribution: Question bank

Before we start, a reminder of the only symbols used below, so nothing is unearned:

$f (E)$ — probability that a state at energy $E$ is occupied, a number between $0$ and $1$ .
$E_{F}$ — the Fermi level, the energy where $f = \frac{1}{2}$ .
$k_{B} T$ — the thermal energy scale (about $0.0259$ eV at room temperature); it sets how "fuzzy" the filling is.
$g (E)$ — the Density of states, how many allowed states sit near energy $E$ .

True or false — justify

The Fermi level is always the highest energy an electron occupies.

False. That is only true at exactly

T = 0

; for

T > 0

some states above

E_{F}

are occupied, so the highest occupied energy drifts above

E_{F}

E_{F}

is defined by

f = \frac{1}{2}

, not by "topmost electron".

f (E_{F}) = \frac{1}{2}

, then there is definitely an electron sitting at

E_{F}

False.

f

is only the probability of occupation if a state exists there. In an intrinsic semiconductor

E_{F}

lies in the band gap where

g (E) = 0

, so no electron can be there at all.

f (E)

tells you the number of electrons at energy

E

False.

f (E)

is a dimensionless probability in

[0, 1]

. The actual count needs

\int g (E) f (E) d E

— you must weight by the Density of states.

T = 0

K the Fermi–Dirac distribution is a perfect step:

1

below

E_{F}

0

above.

True. As

T \to 0

the exponent

(E - E_{F}) / k_{B} T

blows to

- \infty

below and

+ \infty

above, forcing

f = 1

then

f = 0

with a jump at

E_{F}

The curve

f (E)

is symmetric (mirror image) about the point

(E_{F}, \frac{1}{2})

True in the antisymmetric sense:

f (E_{F} + Δ) + f (E_{F} - Δ) = 1

, so the "filled above" mirrors the "empty below". It's point-symmetry, not axis-symmetry.

Raising the temperature raises the Fermi level.

False in general. Heating smears the curve around

E_{F}

but does not by itself shift it;

E_{F}

's position is fixed by requiring total electron count to be conserved (see Chemical potential and equilibrium), and can move slightly either way depending on the Density of states shape.

For a semiconductor,

E_{F} = μ

(the chemical potential) exactly.

Essentially true.

E_{F}

is the electron chemical potential; the "Fermi level" name is conventional. At

T = 0

they coincide, and at device temperatures the tiny difference is usually ignored.

The Boltzmann approximation

f \approx e^{- (E - E_{F}) / k_{B} T}

is valid everywhere.

False. It only holds when

E - E_{F} ≫ k_{B} T

(states well above

E_{F}

). Near or below

E_{F}

it wrongly predicts

f > 1

, which is impossible for a probability.

In equilibrium the Fermi level is the same everywhere in a connected system.

True. A flat

E_{F}

is the condition for equilibrium; any tilt would drive a net electron flow until it flattens (Chemical potential and equilibrium).

Spot the error

"Since

f (E_{F}) = \frac{1}{2}

, half the electrons in the material sit at

E_{F}

Error: confuses a per-state probability with a population fraction.

f = \frac{1}{2}

means each state at

E_{F}

is filled half the time; how many electrons are actually near

E_{F}

depends on

g (E)

there, which can even be zero.

"To get the number of holes, use

f (E)

directly in the valence band."

Error: a hole is a missing electron, so its probability is

1 - f (E)

, not

f (E)

. The hole distribution mirrors the electron one about

E_{F}

"Electrons obey Fermi–Dirac because they are heavy; light particles use Boltzmann."

Error: mass is irrelevant. The reason is the Pauli exclusion principle — at most one electron per state — which caps occupation at

1

. Classical Boltzmann has no such cap.

"At

E = E_{F} + 3 k_{B} T

f \approx e^{- 3} \approx 0.05

, so Boltzmann is exact there."

Error: it's an approximation, not exact. True

f = 1/ (1 + e^{3}) = 0.047

vs Boltzmann

0.050

; close but not equal because the

+ 1

in the denominator is small yet nonzero.

"An intrinsic semiconductor has

E_{F}

exactly in the middle of the gap, always."

Error:

E_{F}

sits near midgap but shifts if the conduction and valence Density of states differ (unequal effective masses), tilting the balance point. Doping moves it much further (see Intrinsic and extrinsic semiconductors).

f (E)

can exceed

1

if the temperature is high enough."

Error:

f = 1/ (1 + e^{x})

with a positive denominator strictly larger than

1

's... actually the denominator

\geq

a positive number, so

f \in (0, 1)

always. Occupation of a single state can never exceed one electron — that's Pauli again.

Why questions

Why does the denominator have a "

+ 1

" instead of just

e^{(E - E_{F}) / k_{B} T}

The

+ 1

comes from the

n = 0

(empty-state) Gibbs weight in the derivation. It is exactly what caps

f

1

and enforces Pauli exclusion; dropping it gives Boltzmann, valid only far above

E_{F}

Why do we care about states just a few

k_{B} T

around

E_{F}

and not the whole band?

Because

f

transitions from

1

0

only over a window of a few

k_{B} T

; deep below

E_{F}

everything is full (nothing can change) and far above everything is empty. All the action — conduction, thermal excitation — happens in that thin fuzzy band.

Why does

E_{F}

being flat mean "equilibrium"?

A spatial gradient in

E_{F}

is a gradient in electron chemical potential, which is a driving force for diffusion/drift of electrons. Only when

E_{F}

is flat is there zero net force and hence no net current — the definition of equilibrium.

Why can we write

μ \to E_{F}

T = 0

but keep the name at

T \neq = 0

T = 0

the chemical potential equals the top of the filled sea, which is the

T = 0

Fermi level. At higher

T

μ

shifts only slightly in normal semiconductors, so keeping the label

E_{F}

is a harmless convention.

Why is the hole probability

1 - f

symmetric to the electron probability?

Because

f (E_{F} + Δ) + f (E_{F} - Δ) = 1

: the chance a state

Δ

above

E_{F}

is filled equals the chance a state

Δ

below is empty. Electrons above and holes below are mirror images about

E_{F}

Why does the Boltzmann tail make carrier concentration calculations easy?

It replaces the messy

1/ (1 + e^{x})

with a plain exponential

e^{- (E - E_{F}) / k_{B} T}

, which integrates cleanly against

g (E)

to give a simple closed form for

n

and

p

in non-degenerate semiconductors.

Edge cases

What is

f (E)

for a state exactly at

E_{F}

T \to 0

It stays

\frac{1}{2}

— the single point

E = E_{F}

gives

e^{0} = 1

f = 1/2

at every temperature, even as everything around it snaps to a step.

What happens to

f

deep in the valence band (

E ≪ E_{F}

) at room temperature?

f \to 1

(states essentially all filled), so the electron probability of emptiness

1 - f \to 0

— very few holes deep down; holes cluster near the valence band top, closest to

E_{F}

If the Density of states

g (E) = 0

at some energy, what does

f (E)

there mean?

Nothing physical for counting —

f

still returns a probability, but with no states the contribution

g f

is zero. This is the intrinsic-semiconductor midgap situation.

Degenerate case: what if

E_{F}

lies inside the conduction band (heavy doping)?

Then states near the band edge have

f

near

1

, the Boltzmann approximation fails, and you must use full Fermi–Dirac. Such a "degenerate" semiconductor behaves more metal-like (see Intrinsic and extrinsic semiconductors).

Limiting behaviour: what does

f

approach as

T \to \infty

for any fixed

E

The exponent

(E - E_{F}) / k_{B} T \to 0

, so

f \to \frac{1}{2}

for every state — infinite thermal energy makes occupation completely indifferent to energy.

What is

1 - f (E)

physically, and when is it the quantity you actually want?

It is the probability a state is empty — the hole occupation. You want it whenever counting holes in the valence band or asking "how likely is this filled level to be vacated?".

Zero-input check: at

E - E_{F} = 0

what is

f

, and does temperature change it?

f = 1/ (1 + e^{0}) = 1/2

, independent of

T

. This single fixed point is why "Fermi equals fifty" — the

\frac{1}{2}

crossing is temperature-proof.

Connections

Pauli exclusion principle — the one-per-state rule behind the whole distribution.
Density of states — pair with $f (E)$ to get real electron counts.
Carrier concentration n and p — where the Boltzmann shortcut pays off.
Chemical potential and equilibrium — why a flat $E_{F}$ means equilibrium.
Intrinsic and extrinsic semiconductors — where $E_{F}$ sits and how doping moves it.