Generation by Impact Ionization

ℰԦ
𝐸𝐶 Consider a high electric field region
Accelerating 𝑒 − s (ℎ+ s) – transfer of energy to lattice
𝐸𝑉 Generate EHPs – Increased current
Can result in avalanche multiplication (Later: Breakdown)

cm−2 s −1
cm−3 s −1
𝐽𝑁 𝐽𝑃
𝐺 = 𝛼𝑁 + 𝛼𝑃
𝑞 𝑞

Ionization coefficients Units: cm−1

Ionization coeff.: No. of EHPs generated per unit distance travelled by a carrier

More Details: http://www.iue.tuwien.ac.at/phd/triebl/node20.html

ℰ: 𝑖𝑛 𝑉𝑐𝑚−1
6
− 1.231×10 ൗ|ℰ|
𝛼𝑁 = 7.03 × 105 × 𝑒

1.75 × 105 ≤ |ℰ| ≤ 6.0 × 105

6
− 2.036×10 ൗ|ℰ|
𝛼𝑃 = 1.582 × 106 × 𝑒

1.75 × 105 ≤ |ℰ| ≤ 4.0 × 105

6
− 1.693×10 ൗ|ℰ|
𝛼𝑃 = 6.71 × 105 × 𝑒

4.0 × 105 ≤ |ℰ| ≤ 6.0 × 105

What is the field required to get an ionization coefficient of 104 cm-1 for 𝑒 − s in Si?

6
− 1.231×10 ൗ|ℰ| (1 𝜇m-1)
𝛼𝑁 = 7.03 × 105 × 𝑒

ℰ = 2.9 × 105 𝑉𝑐𝑚−1

At this field, if an electron travels 1 𝜇m, it produces 1 EHP.

Considered as a critical field for onset of impact ionization

Thickness required to operate at 3 kV for Si? → Large resistance in the forward bias
For some material with a critical field of 30 × 105 𝑉 𝑐𝑚−1 ?

Si GaAs SiC Diamond Larger 𝐸𝐺 ⇒ Larger field required

Critical Field Higher impact for higher fields
3 4 25 100
× 105 𝑉 𝑐𝑚−1
(after a critical field is reached)
𝐸𝐺 (𝑒𝑉) 1.12 1.42 2.9 5.5
EE203 Semiconductor Devices

L15: Excess Carriers – Recombination

Arun Tej M.
Recombination

1. Direct band-to-band recombination (also called radiative recombination)

2. Auger recombination
3. Indirect recombination through defect states
4. Surface recombination
Direct Band-to-Band Recombination

• 𝑒 − from CB falls into an empty state (ℎ+ ) in VB

𝐺𝑡ℎ 𝑅𝑡ℎ
• No additional momentum → Only energy change is involved in the process
• Energy given to a photon → also called radiative recombination
• High chance for direct band gap materials
• Direct bandgap materials are excellent for light emitting devices
• Spontaneous process (probability that an EHP recombines is constant in time)
• Rate of recombination will be proportional to no. of 𝑒 − s in CB and no. of ℎ+ s in VB
𝐺𝑡ℎ = 𝑅𝑡ℎ = 𝛽𝑛𝑝
• If the system is in equilibrium, there will be zero net recombination (𝑛𝑝 = 𝑛𝑖2 )
Consider an 𝑛 −type semiconductor, with equilibrium concentrations 𝑛𝑛𝑜 and 𝑝𝑛𝑜

 𝑛𝑛𝑜 ≫ 𝑝𝑛𝑜

Let excess EHPs be generated by shining light

𝑛𝑛 = (𝑛𝑛𝑜 + ∆𝑛) and 𝑝𝑛 = (𝑝𝑛𝑜 + ∆𝑝)

ℎ𝜈 𝐺𝐿 𝐺𝑡ℎ 𝑅
𝑅 = 𝛽 𝑛𝑛𝑜 + ∆𝑛 𝑝𝑛𝑜 + ∆𝑝

𝐺 = 𝐺𝐿 + 𝐺𝑡ℎ 𝐺𝑡ℎ = 𝑅𝑡ℎ = 𝛽𝑛𝑛𝑜 𝑝𝑛𝑜

To maintain overall charge neutrality ⇒ ∆𝑛 = ∆𝑝

𝑑𝑝𝑛 𝑑𝑝𝑛
Net rate of change in hole concentration is given by =𝐺−𝑅 ⇒ = 𝐺𝐿 + 𝐺𝑡ℎ − 𝑅
𝑑𝑡 𝑑𝑡
𝑑𝑝𝑛
Under steady-state, =0 ⇒ 𝐺𝐿 + 𝐺𝑡ℎ − 𝑅 = 0 ⇒ 𝐺𝐿 = 𝑅 − 𝐺𝑡ℎ ≡ 𝑅𝑛𝑒𝑡
𝑑𝑡

𝑅𝑛𝑒𝑡 = 𝛽 𝑛𝑛𝑜 + ∆𝑛 𝑝𝑛𝑜 + ∆𝑝 − 𝛽𝑛𝑛𝑜 𝑝𝑛𝑜 = 𝛽 𝑛𝑛𝑜 + 𝑝𝑛𝑜 + ∆𝑝 ∆𝑝

𝑅𝑛𝑒𝑡 = 𝛽 𝑛𝑛𝑜 + 𝑝𝑛𝑜 + ∆𝑝 ∆𝑝

Assume low-level injection conditions: ∆𝑛 and ∆𝑝 are small compared to 𝑛𝑛𝑜

Also, neglect the term with (∆𝑝. ∆𝑝) relative to that with (𝑛𝑛𝑜 . ∆𝑝)

⇒ 𝑅𝑛𝑒𝑡 ≈ 𝛽𝑛𝑛𝑜 ∆𝑝

1
𝜏𝑝 ≡ Defined as the recombination lifetime of the excess minority carriers
𝛽𝑛𝑛𝑜

Recombination life time is inversely proportional to the doping concentration

∆𝑝 𝐸𝑥𝑐𝑒𝑠𝑠 𝐶𝑎𝑟𝑟𝑖𝑒𝑟 𝐶𝑜𝑛𝑐𝑒𝑛𝑡𝑟𝑎𝑡𝑖𝑜𝑛

𝑅𝑛𝑒𝑡 ≅
𝜏𝑝 𝑅𝑒𝑐𝑜𝑚𝑏𝑖𝑛𝑎𝑡𝑖𝑜𝑛 𝐿𝑖𝑓𝑒 𝑇𝑖𝑚𝑒
 𝑝𝑛 (𝑡)
Uniformly illuminated
semiconductor 𝑝𝑛 𝑝𝑛 (𝑡 = 0)
ℎ𝜈
𝜏𝑝 𝐺𝐿
𝑝𝑛𝑜

𝑡
0 𝜏𝑝
EHPs generated uniformly
at a rate 𝐺𝐿 Light suddenly turned off at 𝑡 = 0 ⇒ 𝐺𝐿 = 0
∆𝑝 𝑝𝑛 − 𝑝𝑛𝑜 ⇒ 𝑝𝑛 (𝑡 = 0) = 𝑝𝑛𝑜 + 𝜏𝑝 𝐺𝐿
𝐺𝐿 = 𝑅𝑛𝑒𝑡 = = ⇒ 𝑝𝑛 = 𝑝𝑛𝑜 + 𝜏𝑝 𝐺𝐿
𝜏𝑝 𝜏𝑝
𝑝𝑛 (𝑡 → ∞) = 𝑝𝑛𝑜
𝑑𝑝𝑛
⇒ = 𝐺𝐿 + 𝐺𝑡ℎ − 𝑅 = 𝐺𝑡ℎ − 𝑅 = −𝑅𝑛𝑒𝑡
𝑑𝑡
𝑑𝑝𝑛 𝑝𝑛 − 𝑝𝑛𝑜 −𝑡/𝜏𝑝 Minority carriers recombine with majority carriers
⇒ =− 𝑝𝑛 𝑡 = 𝑝𝑛𝑜 + 𝜏𝑝 𝐺𝐿 𝑒
𝑑𝑡 𝜏𝑝 and decay exponentially with a time constant 𝜏𝑝
Exercise

• A semiconductor sample with 𝑛𝑛𝑜 = 1014 𝑐𝑚−3 is illuminated with light and 1013 EHPs/𝑐𝑚3
are generated every microsecond. The light is turned off suddenly at 𝑡 = 0. If 𝜏𝑛 = 𝜏𝑝 = 2 𝜇𝑠,
find the minority carrier concentration 1 𝜇𝑠 after the light is turned off. Given: 𝑛𝑖 = 109 𝑐𝑚−3 .
Indirect Recombination via Localized States

• Important for silicon

• Probability of direct band-to-band recombination is very low in indirect band gap materials
• Almost universally, assistance of a third state is involved for recombination
• Third state – From defects or impurities
• Traps, Recombination Centers

Most probable event

 Si

0 EC
Li+(0.033) P+(0.044) +(0.049) Sb+(0.039)
0.1 As
0.2 S+(0.18)
0.3 S2+(0.37)
Ni2-(0.35)
0.4 Fig. 4-9 from Book by
0.5 Mn+(0.53) Au-(0.54) Streetman & Banerjee
Zn2-(0.55)
0.5
Cu-(0.49)
0.4 Au+(0.35)
Zn-(0.31)
0.3 Cu+(0.24)
0.2 Ni-(0.22)
0.1 B-(0.045) Ga-(0.065) Al-(0.057)
0 EV
Shockley – Read – Hall (SRH) Recombination

Four basic processes for which transition rates are required

1. Electron Capture (e- from CB to ER)
2. Electron Release (e- from ER to CB) (Equivalently, e- from ER to VB)
(Equivalently, e- from VB to ER)
3. Hole Capture (h+ from VB to ER)
4. Hole Release (h+ from ER to VB)

(Topic not properly covered in class)

Rate of e- capture will be proportional to:
(a) No. of e- s in CB 1
𝑅𝑐𝑛 ∝ 𝑛 × 𝑁𝑡 (1 − 𝑓𝑡 ) 𝑓𝑡 = 𝐸𝑡 −𝐸𝐹 /𝑘𝑇
(b) No. of unoccupied recombination center states 1+𝑒

The proportionality constant is called capture coefficient: 𝑐𝑛 = 𝜎𝑛 𝑣𝑡ℎ𝑛

Capture cross section of the e- s for the defect state X Thermal velocity of the e- s
If the defect center lies within this volume, the e- gets captured by it
𝐸𝑡
𝑅𝑐𝑛 = 𝑐𝑛 𝑛𝑁𝑡 (1 − 𝑓𝑡 ) (Derivation not covered in class)

Rate of e- release ∝ No. of e- s in the defect level

Many empty states are present in CB and, thus, it is not the rate limiting factor

𝑅𝑒𝑛 = 𝑒𝑛 𝑁𝑡 𝑓𝑡 𝑒𝑛 ≡ emission coefficient for e-s

 𝑅𝑐𝑛 = 𝑐𝑛 𝑛𝑁𝑡 (1 − 𝑓𝑡 ) 𝑅𝑒𝑛 = 𝑒𝑛 𝑁𝑡 𝑓𝑡 𝑅𝑐𝑝 = 𝑐𝑝 𝑝𝑁𝑡 𝑓𝑡 𝑅𝑒𝑝 = 𝑒𝑝 𝑁𝑡 (1 − 𝑓𝑡 )
(Derivation not covered in class)
At thermal equilibrium, detailed balance demands 𝑅𝑐𝑛 = 𝑅𝑒𝑛 𝑅𝑐𝑝 = 𝑅𝑒𝑝

(Otherwise, there will be a net build-up or decay of electrons in the defect level)

1 − 𝑓𝑡 1 1 − 𝑓𝑡 𝐸𝑡 −𝐸𝐹 /𝑘𝑇
𝑅𝑐𝑛 = 𝑅𝑒𝑛 ⇒ 𝑒𝑛 = 𝑐𝑛 𝑛𝑜 𝑓𝑡 = 𝐸𝑡 −𝐸𝐹 /𝑘𝑇
⇒ =𝑒 ⇒ 𝑒𝑛 = 𝑐𝑛 𝑛1
𝑓𝑡 1+𝑒 𝑓𝑡
𝑓𝑡 𝑛1 = 𝑛𝑜 𝑒 𝐸𝑡 −𝐸𝐹 /𝑘𝑇
𝑅𝑐𝑝 = 𝑅𝑒𝑝 ⇒ 𝑒𝑝 = 𝑐𝑝 𝑝𝑜 𝑒𝑝 = 𝑐𝑝 𝑝1
1 − 𝑓𝑡
𝑝1 = 𝑝𝑜 𝑒 𝐸𝐹 −𝐸𝑡 /𝑘𝑇 𝑛1 𝑝1 = 𝑛𝑜 𝑝𝑜 = 𝑛𝑖2
 𝑅𝑐𝑛 = 𝑐𝑛 𝑛𝑁𝑡 (1 − 𝑓𝑡 ) 𝑅𝑒𝑛 = 𝑐𝑛 𝑛1 𝑁𝑡 𝑓𝑡 𝑅𝑐𝑝 = 𝑐𝑝 𝑝𝑁𝑡 𝑓𝑡 𝑅𝑒𝑝 = 𝑐𝑝 𝑝1 𝑁𝑡 (1 − 𝑓𝑡 )

Non-equilibrium steady state: (Derivation not covered in class)

Let the semiconductor be illuminated uniformly to give a generation rate 𝐺𝐿 . Under steady state,
the electrons entering CB and leaving CB will be equal. Same for holes in VB.
𝑑𝑛𝑛 𝑑𝑝𝑛
= 𝐺𝐿 − 𝑅𝑐𝑛 − 𝑅𝑒𝑛 = 0 = 𝐺𝐿 − 𝑅𝑐𝑝 − 𝑅𝑒𝑝 = 0 ∆𝑛
𝑑𝑡 𝑑𝑡 𝜏𝑛 ≡
𝑅𝑛
Net rate of e- capture, 𝑅𝑛 = 𝑅𝑐𝑛 − 𝑅𝑒𝑛 = 𝑐𝑛 𝑁𝑡 𝑛 1 − 𝑓𝑡 − 𝑛1 𝑓𝑡 Define
∆𝑝
𝜏𝑝 ≡
Net rate of h+ capture, 𝑅𝑝 = 𝑅𝑐𝑝 − 𝑅𝑒𝑝 = 𝑐𝑝 𝑁𝑡 𝑝𝑓𝑡 − 𝑝1 1 − 𝑓𝑡 𝑅𝑝

𝑐𝑛 𝑛 + 𝑐𝑝 𝑝1
𝑅𝑛 = 𝑅𝑝 ⇒ 𝑐𝑛 𝑁𝑡 𝑛 1 − 𝑓𝑡 − 𝑛1 𝑓𝑡 = 𝑐𝑝 𝑁𝑡 𝑝𝑓𝑡 − 𝑝1 1 − 𝑓𝑡 ⇒ 𝑓𝑡 =
𝑐𝑛 𝑛 + 𝑛1 + 𝑐𝑝 (𝑝 + 𝑝1 )
 𝑐𝑛 𝑛 + 𝑐𝑝 𝑝1
𝑅 = 𝑅𝑛 = 𝑅𝑝 = 𝑐𝑛 𝑁𝑡 𝑛 1 − 𝑓𝑡 − 𝑛1 𝑓𝑡 𝑛1 𝑝1 = 𝑛𝑜 𝑝𝑜 = 𝑛𝑖2 𝑓𝑡 =
𝑐𝑛 𝑛 + 𝑛1 + 𝑐𝑝 (𝑝 + 𝑝1 )

𝑛𝑝 − 𝑛𝑖2 1 1
𝑅 = 𝑅𝑛 = 𝑅𝑝 = 𝜏𝑝𝑜 ≡ 𝜏𝑛𝑜 ≡ ∆𝑛 = ∆𝑝
𝜏𝑝𝑜 𝑛 + 𝑛1 + 𝜏𝑛𝑜 (𝑝 + 𝑝1 ) 𝑐𝑝 𝑁𝑡 𝑐𝑛 𝑁𝑡

(𝑛𝑜 + ∆𝑛)(𝑝𝑜 + ∆𝑝) − 𝑛𝑖2 𝑝𝑜 ∆𝑛 + 𝑛𝑜 ∆𝑛 + ∆𝑛∆𝑛

𝑅 = 𝑅𝑛 = 𝑅𝑝 = =
𝜏𝑝𝑜 𝑛𝑜 + ∆𝑛 + 𝑛1 + 𝜏𝑛𝑜 (𝑝𝑜 + ∆𝑝 + 𝑝1 ) 𝜏𝑝𝑜 𝑛𝑜 + ∆𝑛 + 𝑛1 + 𝜏𝑛𝑜 (𝑝𝑜 + ∆𝑛 + 𝑝1 )

𝑅𝑛 𝑝𝑜 + 𝑛𝑜 + ∆𝑛 𝑅𝑝 ∆𝑝 ∆𝑛
= = 𝜏𝑝 ≡ 𝜏𝑛 ≡
∆𝑛 𝜏𝑝𝑜 𝑛𝑜 + ∆𝑛 + 𝑛1 + 𝜏𝑛𝑜 (𝑝𝑜 + ∆𝑛 + 𝑝1 ) ∆𝑝 𝑅𝑝 𝑅𝑛

𝑛𝑜 + ∆𝑛 + 𝑛1 𝑝𝑜 + ∆𝑛 + 𝑝1
𝜏𝑛 = 𝜏𝑝 = 𝜏 = 𝜏𝑝𝑜 + 𝜏𝑛𝑜
𝑝𝑜 + 𝑛𝑜 + ∆𝑛 𝑝𝑜 + 𝑛𝑜 + ∆𝑛

𝑛𝑜 + 𝑛1 𝑝𝑜 + 𝑝1
𝜏 = 𝜏𝑝𝑜 + 𝜏𝑛𝑜 For low-level injection
𝑝𝑜 + 𝑛𝑜 𝑝𝑜 + 𝑛𝑜

(Derivation not covered in class)

Consider a p-type semiconductor under low-level injection 𝑝 = 𝑝𝑜 ≫ 𝑛𝑜 , 𝑛1 , 𝑝1

𝑛𝑜 + 𝑛1 𝑝𝑜 + 𝑝1
𝜏 = 𝜏𝑝𝑜 + 𝜏𝑛𝑜 ⇒ 𝜏 ≈ 𝜏𝑛𝑜
𝑝𝑜 + 𝑛𝑜 𝑝𝑜 + 𝑛𝑜

Similarly, for an n-type semiconductor under low-level injection 𝜏 ≈ 𝜏𝑝𝑜

Excess carrier lifetime in an extrinsic semiconductor is dominated by the minority carrier lifetime

For high-level injection (𝑖. 𝑒. , ∆𝑛 = ∆𝑝 ≫ 𝑛𝑜 , 𝑝𝑜 )

𝑛𝑜 + ∆𝑛 + 𝑛1 𝑝𝑜 + ∆𝑛 + 𝑝1
𝜏𝑛 = 𝜏𝑝 = 𝜏 = 𝜏𝑝𝑜 + 𝜏𝑛𝑜 ⇒ 𝜏 ≈ 𝜏𝑝𝑜 + 𝜏𝑛𝑜
𝑝𝑜 + 𝑛𝑜 + ∆𝑛 𝑝𝑜 + 𝑛𝑜 + ∆𝑛

Under high level injection, excess carrier lifetime reaches maximum value and becomes
independent of injection

(Derivation not covered in class)

 𝑛𝑝 − 𝑛𝑖2
𝑅 = 𝑅𝑛 = 𝑅𝑝 =
𝜏𝑝𝑜 𝑛 + 𝑛1 + 𝜏𝑛𝑜 (𝑝 + 𝑝1 )

𝑛𝑝 − 𝑛𝑖2
For 𝜏𝑛𝑜 = 𝜏𝑝𝑜 = 𝜏 𝑅 = 𝑅𝑛 = 𝑅𝑝 =
𝜏 𝑛 + 𝑛𝑜 𝑒 𝐸𝑡 −𝐸𝐹 /𝑘𝑇 + 𝜏 (𝑝 + 𝑝 𝑒 𝐸𝐹 −𝐸𝑡 /𝑘𝑇 )
𝑜

𝑛𝑝 − 𝑛𝑖2
𝑛𝑜 = 𝑛 𝑖 𝑒 − 𝐸𝑖 −𝐸𝐹 /𝑘𝑇 ⇒ = 𝐸𝑖 −𝐸𝑡 𝐸𝑡 −𝐸𝑖
𝜏 𝑛+𝑝+ 𝑛𝑖 𝑒 𝑘𝑇 + 𝑛𝑖 𝑒 𝑘𝑇

𝑛𝑝 − 𝑛𝑖2
=
𝐸𝑡 − 𝐸𝑖
(Topic not covered properly in class) 𝜏 𝑛 + 𝑝 + 2𝑛𝑖 𝑐𝑜𝑠ℎ
𝑘𝑇

𝑛𝑝 − 𝑛𝑖2 𝐸𝑡 − 𝐸𝑖
= 𝑥=
𝜏 𝑛 + 𝑝 + 2𝑛𝑖 𝑐𝑜𝑠ℎ𝑥 𝑘𝑇
 𝑛𝑝 − 𝑛𝑖2
𝑅=
𝐸𝑡 − 𝐸𝑖
𝜏 𝑛 + 𝑝 + 2𝑛𝑖 𝑐𝑜𝑠ℎ
𝑘𝑇

𝑥 ↓ ⇒ 𝑐𝑜𝑠ℎ𝑥 ↓ ⇒ 𝑅 ↑ 𝐸𝑡 − 𝐸𝑖 ↓ ⇒ 𝑅 ↑

Recombination rate is high when Et is close to Ei

Defect levels that are away from intrinsic level are

𝐸𝑡 − 𝐸𝑖 less likely to act as ‘efficient’ recombination centers
𝑥=
𝑘𝑇

(Topic not properly covered in class)

Auger Recombination (pronounced as oh-zhay)
• EHP recombines → energy released is given to another carrier
• Three carriers involved: Also known as 𝑒 − 𝑒 − ℎ or ℎ − ℎ − 𝑒 process
• Two 𝑒 − s (in 𝑛-type) and two ℎ+ s (in 𝑝-type) are involved
• Highly unlikely except for heavily doped materials 𝑅
• E.g. 𝑛+ emitter in BJTs, 𝑛+ source and drain in MOSFETs
Net recombination rate: Similar to the direct recombination case,
except with the requirement of 3rd carrier

𝑅 = 𝑎𝑛 𝑛 𝑛𝑝 − 𝑛𝑖2 + 𝑎𝑝 𝑝 𝑛𝑝 − 𝑛𝑖2 𝑎𝑛 ≈ 2.8 × 10−31 𝑐𝑚6 𝑠 −1 for n+ Si

𝑎𝑝 ≈ 0.99 × 10−31 𝑐𝑚6 𝑠 −1 for p+ Si
For n+ material 𝑅 ≈ 𝑎𝑛 𝑛𝑜 (𝑛𝑜 ∆𝑝)

∆𝑝 1 Causes the life time to decrease at a much

𝑅≅ 𝜏=
𝜏 𝑎𝑛 𝑛𝑜2 faster rate with doping concentration
Surface Recombination (Topic not covered properly in class)

Surface

→ Termination of crystal → Large number of electrically active states

→ Exposed during fabrication → Likely to contain many impurities

⇒ Defects and trap states near the surface of a semiconductor device – say, NST cm-2

⇒ Carrier recombination rate will be high

Recombination rate: Expression similar to that of SRH recombination rate

1 1 −1 = 𝑣
Consider holes at the surface. We defined, for SRH: 𝜏𝑝𝑜 ≡ = ⇒ 𝜏𝑝𝑜 𝑡ℎ𝑝 𝜎𝑝 𝑁𝑡
𝑐𝑝 𝑁𝑡 𝑣𝑡ℎ𝑝 𝜎𝑝 𝑁𝑡

Similarly, we define ‘surface recombination velocity’ here as: 𝑆 = 𝑣𝑡ℎ𝑝𝑠 𝜎𝑝𝑠 𝑁𝑡𝑠

𝑁𝑡𝑠 : No. of surface states per unit area Dimensions of S: cm/s