ELECTRONIC DEVICES

EQUILIBRIUM CARRIER CONCENTRATION

Lecture - 6
 Physical Interpretation of formula for ni
!
!ԥ#$ %& " (" )(*
ni= 2 ! exp -( )
' %&

NC fB(EC, T)

• The ni represent the occupied conduction band states.

• fB(EC,T) is the Boltzmann approximation to the Fermi Dirac
function at E = Ec and temperature T.
• That means this term represents the fraction of states occupied
for energy Ec within an interval dE at any temperature.
• So, we could effectively interpret NCas the number of available
states at Ec.

2
 Effective Band Picture
ET

EC Ec NC/cm3
(Effective density of
EV
states at conductive
band edge)
Ev

EB

• Fraction of electrons present at Ec can be find out by multiplying

the states with the fraction given by the Boltzmann approximation.

3
• Now all of us know that the equation is
!
!ԥ#$ %& " (" )(*
ni = 2 '!
exp -( %&
)

• But from the above equation how do we know that the

Bol@mann approximation is valid???

• Hence we have to derive a formula for Ef and check whether

indeed Ef is below 3 times KT.

• Now we have to locate the fermi level in an intrinsic

semiconductor in the next slide.

4
 Location of Ef by Simple Reasoning
• We all know that the Fermi energy level is that level above which no
level is occupied and below which all available states are occupied at
T=0K.
Ef Ec
Ec Eg
Eg
Ev
Ev Ef

• Supposing Ef was in conduction • Now If we assume the Ef to be in

band, then what it means is that the valance band, then all states
below Ef whatever states are there above Ef should be unoccupied,
which are allowed they must be but we know that at T = 0 the entire
occupied. valence band is occupied.
• This means there must be
conduction electrons at T = 0 which
is not true.
5
• Hence Ef should be somewhere between Ec & Ev at T=0K.
• Now at T at higher temperature the Fermi energy is away from the
valance band, but close to the conduction band.

By use of this Fermi

Ec
Dirac function we can
Ef find out the electron
concentration using the
equation.
Ev

• All the electrons have to come from the valence band, which means the
number of unoccupied states in the valence band should be equal to the
number of occupied states in the conduction band because the electrons
which have come to the conduction band for higher temperatures have
come from the valence band.
• Hence as Ef is close to Ec and far from Ev, hence it has been concluded
that the no. of unoccupied states in valance band will be less than the no.
of occupied states in conduction band or we can mention it as ni > pi.

6
• Unoccupied states in the valence band are the holes; this should be
equal to occupied states in the conduction band in a pure
semiconductor at any temperature. If this two turn out to be not
equal for a certain choice of Ef then that choice of Ef is not correct.
• The number of holes is given by : pi=∫ N(E) dE [1-f(E,T)].
• 1-f(E,T) represents unoccupied states.

N(E)
Area=ni Ec
f(E)

Ef ni > pi

1-f(E)
Area=pi Ev
N(E)

So it is very clear that if Ef is close to the conduction band edge then area ni
will be less than area pi.
• So if the two constants in the density of states functions for the
conduction and valence bands are exactly same, then it is clear that
area pi =area ni , that is Ef is exactly in the middle of the energy gap.
• If Ef is exactly in middle of valance band and conduction band then it has
been stated that ni = pi. But it is not exactly at the middle of the energy
gap because the effective mass of electrons and holes is not the same.
• Which is reflected from different shape of N(E).
• Hence, the parabolic density of states functions for conduction band and
valence band are slightly different :constants are slightly different.
• However, in most cases it is considered to be at the middle of band gap.
• Now, If Ef is very close to the middle of the energy gap then what is
the distance between Ef and Ec?
• The energy gap in silicon is 1.1 electron volt. Half of that would be 0.55.
• Let us take Eg= 1.1ev ,then ½ of it would be 0.55 electron volt;
that is Ec − Ef.
• Now KT =0.26 eV so 0.26(3kT) is 0.078 eV. If we compare 0.55 eV and 0.078
eV we will always find that (Ec − Ef )in a pure semiconductor like silicon
and in many other semiconductors will always be more than 3KT at any
temperature. So, Boltzmann approximation is valid.
8
 • Now, from the equation we have to find out Ef, because until know the value
of Ef, we could not find out the value for ni.
! !
!ԥ#$ %& " (" )(* !ԥ#+ %& " (*)(,
ni= 2 exp -( ) pi= 2 exp -( )
'! %& '! %&

• Use BolDmann approximation to the Fermi Dirac function and then use
parabolic density of states approximation and we replace the boHom of the
valence band by negative infinity. And then our result will be exactly
analogous to this. So we can write down the result for pi simply from the
result of ni.
• The formula for pi should be the same as the formula for ni with the
replacements of mn replaced by mp and Ec– Efreplaced by Ef – Ev
• Fermi Dirac fraction should be <1, therefore the negative sign is kept outside
and the term inside the bracket should be positive.
• So, Ec – Ef or Ef –Ev should be positive.
• Now we can eliminate the Ef between these two results by ni× pi.
!ԥ#$ & !
'"(') ('*
n ip i = 4 𝑚𝑛𝑚 𝑝 " exp - -> ni = 𝑁𝐶𝑁𝑉 exp
%! #$ !#$

ni2 (-
NC NV exp - %&
9
 Properties of Silicon & GaAs
Parameter Symbol (Units) Si GaAs

Ɛg (ev)
Band gap 1.12 1.42

Electron affinity χ (ev) 4.05 4.07

Intrinsic
ni (cm-3) 1.5x1010 1.8x106
Concentration

Effective NC (cm-3) 2.8 x 1019 4.7 x 1017

density of states NV (cm-3) 1 x 1019 7 x 1018

Dielectric
Ɛr 11.7 12.9
Constant

• These values that are being shown or valid at room temperature,

300k.
10
• We can use these values of ni at T = 300K and then you can use the
formula to find out the value at any other temperature.
• The expression of ni as a function of temperature :
𝟑
#𝑬𝒈
ni= B 𝑻 exp
𝟐
𝟐𝑲𝑻
• Now we know the value at 300K; so using that value to get the
constant ‘B’ which is independent of temperature, and then using
that constant and we can find ni at any other temperature.
• Now,
𝟑
ni(T) 𝑻 𝟐 #𝑬𝒈 𝟏 𝟏
= exp −
ni(300K) 𝟑𝟎𝟎 𝟐𝑲 𝑻 𝟑𝟎𝟎

• This is how we can estimate ni at any other temperature.

11
Effective Mass of Silicon & GaAs
Parameter Symbol(Units) Si GaAs

mn/m0
0.26 0.09
Conductivity
effective mass
mp/m0
0.38 -------

mn/m0
1.18 0.068
Electron affinity
mp/m0
0.81 0.56

12
• Now, in the previous slide two types of effective masses has been
shown: Conductivity effective mass and Density of states effective
mass.
• The effective mass used varies with the physical situation. This is
because it is an average value.
• In case of “Density of states effective mass” the value should be used
to estimate Nc or Nv effective density of states. In case of silicon, the
electron is heavier than the hole.
• Whereas in case of “Conductivity effective mass” the effective mass
of electron is less than that of holes. So electron is lighter than the
hole for conductivity situation.
• It can also be stated that the movement of the hole is actually
movement of bound electrons and therefore that movement is going
to be slow.
• So Density of states effective mass is not involved in the concept
of movement, this is some weighted average.

13
Carrier Concentration in Intrinsic Semiconductor
500K 300K
#,-
ni = 𝑁𝐶𝑁𝑉 exp ./0
1018 Si
Ge %
#,-
1014 ni= B 𝑇 exp
&
./0
GaAs
ni(cm-3)

1010
1 ,- 2
106 log ni = log B + log T – ×
. ./ 0
0.5 1 1.5 2 2.5 3 3.5 4
1000/T(K-1)
• Now this log of T is dependent on temperature but this variation with
temperature is very small because it is logarithmic dependence on
temperature whereas it depends on 1/T varies much more rapidly.
• Hence, Log ni versus 1/T term would be a straight line.
• In Log ni vs 1/T graph slope of the straight line is Eg/2K. Thus
depends on the energy gap and Boltzmann constant K.
• So we can estimate band gap from this plot.
• That is why Y axis is in log scale. 14
 ni (log scale) Slope = -Eg/2K

/ (- 0
log ni = log B + log T – ×
! !% &

1/T

• The slope can be used to get the energy gap of the

semiconductor.
• Now from n i and pi , Ef has been calculated.

,' 3,4 5'

Ef = .
- KT ln 5(

• The second term depends on temperature. But this dependence

."
on temperature is negligible because quantity is small.
.#

• So Ef is approximately equal to Ec + Ev/2.

15
 Questions
Ø In case of intrinsic semiconductor Fermi level is a function of
temperature, but the figure shows that Fermi-Dirac function at
various temperature appears to have a Fermi-level independent of
temperature?
E
T1 T2 > T1
T2
Ef1

Ef2

0 1/2 1
f (E,T)

,' 3,4 5'

Ef = .
- KT ln 5(
16
• The y axis is E – Ef and not E. We are plotting the energy as with
reference to the Fermi-level and not the absolute energy. So when we
take E – Ef on the y axis then any variations in Ef need not be shown on
the graph. If however we plot the function as shown in second diagram
where we are plotting the electronic energy on the y axis, and the
Fermi-Dirac function as a function of the electronic energy.
• Here variation of Fermi-level with temperature will be seen. Now,
Green line which represents the Fermi-Dirac function for the
temperature T1. For a higher temperature T2 shown by Red line.
• Notice that the red line showing the Fermi-level for T2 is lower than the
green line shown for Fermi-level corresponding to T1. So Ef2 is below Ef1
for T2 > T1. This is because according to the formula we derived we can
see that as temperature increases because of the second term the Fermi-
level moves down.
• So when we plot the Fermi-Dirac function as a function of absolute
energy the variation in Fermi-level with temperature will be seen
whereas if we plot the Fermi-Dirac function with respect to E – Ef , in
that case the variation in Fermi-level with temperature will not be seen.

17
Ø Is there any analogy to explain the fact that electronic energy can only
vary in discrete intervals and not continuously?

• Suppose there is a stair case and there is a ball. Now the question is,
at what heights can the ball exist on this stair case?

• The ball can either exist at this particular stair case that is the region
of either 1 or 2 or 3 or 4 (Green). It cannot exist, in the vertical
portions (red) of the stair case, So, you can see here that the ball can
only exist at discrete heights.

18
 d

1 2 3 4 5 6
• Suppose a street has signals at regular intervals; let us say the distance between two
consecutive signals is d; the signals turn green at interval of ‘T’ minutes.
• Imagine that there is a car which wants to pass through this street uninterrupted even
the signals turning red or green.
• What are the speeds in which the car can move so that it will not be interrupted by
the signals?
• The speed of the car is given by d/nT where n is an integer. The speed of the can only
vary in discrete steps. So the car is only allowed in these particular speeds.
• Let us take n = 1. Now the car is moving and the signal at 1 has turned green, it has
passed signal 1. After a time T minutes it is at 2, when this signal is turning green. So
it has crossed a distance d in time T. Since the signal has just turned green it can move
uninterrupted and go to the next signal and this will continue.
• So in the speed of d/T that is n = 1 the car moves uninterrupted. Now another speed
that is allowed is n = 2 that is the speed = d/2T.
• The speed is half of the speed for n = 1. That is, the car has passed signal 1 when it is
green, and next time the signals turns green the car is exactly half way. When it comes
at 2 the signal turns green for the second time, and then again the car moves
uninterrupted. So the allowed speeds given by d/T, d/2T, d/3T and so on.
• So the car’s speed can only change in discrete intervals, discrete steps in this particular
situation.
19
Any Question?