EE203 Semiconductor Devices

L10: Charge Carrier Density - 4

Arun Tej M.
Nondegenerate Semiconductors

𝑛𝑜 = 𝑁𝐶 𝑒 − 𝐸𝐶 −𝐸𝐹 /𝑘𝑇
What were the assumptions made in deriving these equations?
− 𝐸𝐹 −𝐸𝑉 /𝑘𝑇
𝑝𝑜 = 𝑁𝑉 𝑒

a. Use Fermi-Dirac distribution function at 𝐸𝐶 (𝐸𝑉 ) under Boltzmann approximation

b. All the available states in the CB (VB) can be represented by an effective density of states 𝑁𝐶
(𝑁𝑉 ) located at the CB (VB) edge 𝐸𝐶 (𝐸𝑉 )

Implicit:
 Dopant density is at least an order smaller than semiconductor atom density
 Dopant atoms are far apart – donor 𝑒 − s (acceptor ℎ+ s) do not interact with each other (isolated
atoms in a sea of Silicon)
 Dopant atoms introduce energy states that are non-interacting and discrete
Degenerate Semiconductors
What will happen if the donor concentration is increased to higher values?
• Distance between donor atoms reduces. At some point, the donor atoms begin to interact with
each other. Single discrete donor energy level will then split into a band of energies
• Further increase? Donor band widens. When donor density reaches close to 𝑁𝐶 , donor band
overlaps with the CB.
• When 𝑒 − concentration exceeds 𝑁𝐶 , the Fermi level lies inside the CB – called n-type
degenerate semiconductor. States between 𝐸𝐹 and 𝐸𝐶 will be filled in these.
• Similar case for acceptor atoms: Fermi level lies inside VB. States between 𝐸𝑉 and 𝐸𝐹 will be
empty in these p-type degenerate semiconductors.

𝐸𝐹
Impurity-Induced Band Gap Narrowing
• ND ↑ → donor states overlap in energy with each other and also with the CB of the host Si
• Bottom of CB actually gets lowered from its intrinsic value to the bottom of the overlapping dopant
band
• Band gap is a function of doping

𝐸𝐶𝑜
𝐸𝐷 ∆𝐸𝐺 𝐸𝐶

𝐸𝐺𝑜 𝐸𝐺
 −1/4
0.5 −4 0.25 −4
For n-type Si, ∆𝐸𝐺 = 𝐶1 𝑁𝐷 + 𝐶2 𝑁𝐷 𝐶1 = 4.37 × 10−11 𝐶2 = 1.27 × 10−6

Can be neglected for doping concentration up to ~1018 𝑐𝑚−3

0.20

0.15

∆𝐸𝐺
0.10
(𝑒𝑉)

0.05

0.00
1015 1016 1017 1018 1019 1020 1021
𝑁𝐷 𝑐𝑚−3
Dopant Ionization

If the donor concentration is 𝑁𝐷 , how many donors will be ionized at room temperature?

Usual assumption: All of them are ionized. Hence, 𝑒 − concentration in CB 𝑛𝑜 = 𝑁𝐷 .

How accurate is this assumption or under what conditions is this assumption valid?

𝐸𝐶 𝐸𝐶
~45 meV
𝐸𝐷 𝐸𝐷

𝐸𝑉 𝐸𝑉
𝑇 = 0K 𝑇 = 300K
Probability of occupation for CB: Fermi-Dirac Function
𝐸𝐶
~45 meV
1 𝐸𝐷
𝑓(𝐸𝐶 ) = 𝑛𝑜 = 𝑁𝐶 𝑒 − 𝐸𝐶 −𝐸𝐹 /𝑘𝑇
𝐸 − 𝐸𝐹 𝐸𝐹
1 + exp 𝐶
𝑘𝑇 (𝑁𝐶 = 2.8 × 1019 cm−3 )
𝐸𝑉
𝑛𝑜 ≤ 𝑁𝐷 = 1016 cm−3 (𝑠𝑎𝑦) 𝑇 = 300K 𝑘𝑇 = 26 meV

𝑁𝐶 If we assume Fermi-Dirac function for probability of

𝐸𝐶 − 𝐸𝐹 = 𝑘𝑇𝑙𝑛 ≥ 0.206 eV
𝑛𝑜 occupation for 𝐸𝐷 as well,

1 1
𝑓 𝐸𝐶 = = 3.62 × 10−4 𝑓 𝐸𝐷 =? = 2.04 × 10−3
0.206 0.206 − 0.045
1 + exp 1 + exp
0.026 0.026

𝑓 𝐸𝐶
~0.177 Self-contradictory?
𝑓 𝐸𝐷
 ?
𝐸𝐶
Analogy: Water in a glass
𝐸𝐷
𝐸𝐹

𝐸𝑉
𝑇 = 300K

No. of 𝑒 − s in 𝐸𝐷 𝑁𝐷 × 𝑓 𝐸𝐷
=
−
No. of 𝑒 s in CB 𝑁𝐶 × 𝑓 𝐸𝐶

1016 1 Higher 𝑁𝐷 → higher the ratio

= × ~0.002
2.8 × 1019 0.177

Ionization of a donor atom depends on

(a) how many donor atoms are present and (b) how many states are available in CB
Probability Statistics of Dopants

Si Si Si Si Si Si As Si Si Si

Si Si Si Si Si Si Si Si Si Si
+
As
Si Si Si Si Si Si Si Si Si

𝐴𝑠 0 ↔ 𝐴𝑠 + + 𝑒 − 1 𝑁𝐷0 𝑁𝐷+ = 𝑁𝐷 − 𝑁𝐷0 = 𝑁𝐷 1 − 𝑓(𝐸𝐷 )

𝑓(𝐸𝐷 ) = =
1 𝐸 − 𝐸𝐹 𝑁𝐷
1 + 2 exp 𝐷
𝑁𝐷 = 𝑁𝐷+ + 𝑁𝐷0 𝑘𝑇 𝑁𝐷
𝑛𝑑 =
1 𝐸 − 𝐸𝐹
𝑛𝑑 = 𝑁𝐷 − 𝑁𝐷+ No. of filled donor states 1 + exp 𝐷
= 2 𝑘𝑇
Total no. of donor states
 1
𝑓(𝐸) =
𝐸 − 𝐸𝐹
1 + exp
𝑘𝑇

𝑓(𝐸) 𝐸𝐹 − 𝐸 What if the no. of ways in which we can fill a state is not
= exp
1 − 𝑓(𝐸) 𝑘𝑇 same as the no. of ways in which it can be emptied?

𝑓(𝐸) 𝐸𝐹 − 𝐸 𝑁𝐹
= exp ×
1 − 𝑓(𝐸) 𝑘𝑇 𝑁𝐸
𝐸𝐶
𝑓(𝐸𝐷 ) 𝐸𝐹 − 𝐸𝐷 2
= exp ×
𝐸𝐷 1 − 𝑓(𝐸𝐷 ) 𝑘𝑇 1

𝐸𝑉
Probability Statistics of Dopants
𝐸𝐶
• Fermi-Dirac function was used for finding the 𝐸𝐷
probability of filling energy states in a band 𝐸𝐹
• Pauli exclusion principle was taken into
account and each energy level could be 𝐸𝑉
populated by two 𝑒 − s: spin = + 1Τ2 and spin =
− 1Τ2 𝑁𝐷
𝑛𝑑 =
• Even though each donor level has two spin 𝐸 − 𝐸𝐹
1 + g. exp 𝐷
states, here an ionized arsenic atom can 𝑘𝑇
receive only one 𝑒 −
𝑛𝑑 : Density of electrons occupying donor level
• Insertion of one 𝑒− is sufficient to fill the
vacancy requirement and the addition of 𝑁𝐷 : Density of donor atoms
second 𝑒 − is not possible
1
Degeneracy factor for donor = g = 1/2
• A correction factor is, hence, introduced in
2
Fermi-Dirac statistics for donor atoms. Degeneracy factor for acceptor = g = 1/4
 𝑁𝐷 +
𝑁𝐷
𝑛𝑑 = For a large fraction of donor atoms to be ionized, should be large
1 𝐸 − 𝐸𝐹 𝑁𝐷
1 + exp 𝐷
2 𝑘𝑇
𝑁𝐷 𝐸𝐹 − 𝐸𝐷
If 𝐸𝐷 − 𝐸𝐹 ≫ 𝑘𝑇, we can write 𝑛𝑑 ≈ 1 𝐸 − 𝐸𝐹
= 2𝑁𝐷 exp
exp 𝐷 𝑘𝑇
2 𝑘𝑇
𝐸𝐹 − 𝐸𝐶
If 𝐸𝐷 − 𝐸𝐹 ≫ 𝑘𝑇, obviously 𝐸𝐶 − 𝐸𝐹 ≫ 𝑘𝑇. So, we can use the equation 𝑛0 = 𝑁𝐶 exp
𝑘𝑇

Ratio of 𝑒 − s in donor state to the total no. of 𝑒 − s will then be

𝐸𝐶
𝐸𝐹 − 𝐸𝐷 𝐸𝐷
𝑛𝑑 2𝑁𝐷 exp
𝑘𝑇
=
𝑛𝑑 + 𝑛𝑜 2𝑁 exp 𝐸𝐹 − 𝐸𝐷 + 𝑁 exp 𝐸𝐹 − 𝐸𝐶 𝐸𝐹
𝐷 𝑘𝑇 𝐶 𝑘𝑇

𝑛𝑑 1
=
𝑛𝑑 + 𝑛𝑜 1 + 𝑁𝐶 exp 𝐸𝐷 − 𝐸𝐶 𝐸𝑉
2𝑁𝐷 𝑘𝑇
 Exercise

Consider Phosphorus doping of Si where donor level is 45 meV below CB. The doping concentration
is 1016 cm−3 . Determine the fraction of total 𝑒 − s still in the donor states at 𝑇 = 300K.

𝑛𝑑 1
= = 0.0041 = 0.41%
𝑛𝑑 + 𝑛𝑜 1 + 𝑁𝐶 exp 𝐸𝐷 − 𝐸𝐶
2𝑁𝐷 𝑘𝑇

 Almost completely ionized. Temperature, 𝑁𝐶 , 𝑁𝐷 ,and donor level dependent.

 Solubility analogy.

Homework: For the above exercise problem, calculate the temperature at which 90% of donor atoms
are ionized (you may use trial and error to get an approximate value).
Fermi Energy With Doping

0 𝐸𝐶
𝑁𝐶 𝐸𝐷 /2
𝑛𝑜 = 𝑁𝐶 𝑒 − 𝐸𝐶 −𝐸𝐹 /𝑘𝑇
𝐸𝐶 − 𝐸𝐹 = 𝑘𝑇ln
𝑛0 −50 𝐸𝐷

𝑁𝐷 = 1016 𝑐𝑚−3
As temperature is lowered, more and more 𝐸𝐹 (meV)
donors become neutral i.e., occupied by
electrons. Fermi level get closer to donor
level.

When Fermi level merges with donor level, −250

ignoring degeneracy, how many donor 0 400
levels will be occupied? 200
𝑇(K)
 With Doping

no
(log scale) Not controlled by doping; Equal e-s and h+s

Intrinsic
Useful region for device operation

Extrinsic
ND Smaller range for Ge than for Si
Partial Ionization/
Carrier Freeze-out
− 𝐸𝐺 Τ2𝑘

1/T

T
Compensation 𝑁𝐷+ = 𝑁𝐷 − 𝑛𝑑
𝑁𝐴− = 𝑁𝐴 − 𝑛𝑎
• Only donors: 𝑛𝑜 ≈ 𝑁𝐷 Only acceptors: 𝑝𝑜 ≈ 𝑁𝐴
• Both donors and acceptors Thermal Donor
e-s e-s
• Compensation 𝑁𝐷 − 𝑁𝐴
𝑛𝑜
−− −−−−−−−
Fully compensated + + ++ ++ + 𝐸𝐷
𝑁𝐷 = 𝑁𝐴 Un-ionized Ionized
donors 𝑛𝑑 𝑁𝐷+
donors
⇒ No donated e-s in CB and h+s in VB
𝑛𝑎 Ionized 𝑁 −
⇒ 𝑛𝑜 = 𝑛𝑖 = 𝑝𝑜 Un-ionized 𝐴
acceptors
acceptors − − −− −− −
𝐸𝐴
++ ++ ++ ++ +
𝑝𝑜
Thermal Acceptor
h +s h+s
Space Charge Neutrality

Assume uniformly doped semiconductor at equilibrium, with all dopants ionized 𝑁𝐷+ = 𝑁𝐷 𝑁𝐴− = 𝑁𝐴
For the semiconductor to be electrostatically neutral, No. of +ve charges = No. of -ve charges
𝑝𝑜 + 𝑁𝐷 = 𝑛𝑜 + 𝑁𝐴 𝑝𝑜 − 𝑛𝑜 + 𝑁𝐷 − 𝑁𝐴 = 0

𝑛𝑜 𝑝𝑜 = 𝑛𝑖2
1/2
𝑛𝑖2 𝑁𝐷 − 𝑁𝐴 𝑁𝐷 − 𝑁𝐴
2
− 𝑛𝑜 + 𝑁𝐷 − 𝑁𝐴 = 0 𝑛𝑖2
𝑛𝑜 𝑛𝑜 = + + 𝑛𝑖2 𝑝𝑜 =
2 2 𝑛𝑜
For 𝑁𝐷 − 𝑁𝐴 ≫ 𝑛𝑖 , 𝑛𝑜 ≈ 𝑁𝐷 − 𝑁𝐴
Similarly, for a p-type semiconductor

2 1/2
𝑁𝐴 − 𝑁𝐷 𝑁𝐴 − 𝑁𝐷 𝑛𝑖2
𝑝𝑜 = + + 𝑛𝑖2 𝑛𝑜 =
2 2 𝑝𝑜