Section 12: Intro to Devices - University of California ...ee143/fa08/lectures/Section 12... ·...
Transcript of Section 12: Intro to Devices - University of California ...ee143/fa08/lectures/Section 12... ·...
EE143 – Ali Javey
Section 12: Intro to Devices
Extensive reading materials on reserve, includingRobert F. Pierret, Semiconductor Device Fundamentals
EE143 – Ali Javey
Bond Model of Electrons and Holes• Silicon crystal in
a two-dimensionalrepresentation.
Si Si Si
Si Si Si
Si Si Si
Si Si Si
Si Si Si
Si Si Si
Si Si Si
Si Si Si
Si Si Si
• When an electron breaks loose and becomes a conduction electron, a hole is also created.
EE143 – Ali Javey
Semiconductors, Insulators, and Conductors
• Totally filled bands and totally empty bands do not allow current flow. (Just as there is no motion of liquid in a totally filled or totally empty bottle.)
• Metal conduction band is half-filled.• Semiconductors have lower EG’s than
insulators and can be doped
E c
Ev
Eg=1.1 eV
E c
Eg= 9 eV empty
Si, Semiconductor SiO2
, insulator Conductor
Ecfilled
Top ofconduction band
E v
EE143 – Ali Javey
Dopants in Silicon
Si Si Si
Si Si
Si Si Si
Si Si Si
Si Si
Si Si Si
As B
• As, a Group V element, introduces conduction electrons and creates N-type silicon,
• B, a Group III element, introduces holes and creates P-type silicon, and is called an acceptor.
and is called a donor.
EE143 – Ali Javey
Ionized Donor
IonizedAcceptor
Immobile Chargesthey DO NOTcontribute to current flow with electric field is applied. However, they affect the local electric field
Hole
Electron
Mobile Charge Carriersthey contribute to current flow with electric field is applied.
Types of charges in semiconductors
EE143 – Ali Javey
Fermi Function–The Probability of an Energy State Being Occupied by an Electron
f(E)0.5 1
Ef
Ef – kTEf – 2kT
Ef – 3kT
Ef + kT
EfEf + 2kTEf + 3kT
E
kTEE feEf /)(1
1)( −+= Ef is called the Fermi energy or
the Fermi level.
( ) kTEE feEf −−≈)( kTEE f >>−
( ) kTEE feEf −−−≈ 1)( kTEE f −<<−
Boltzmann approximation:
( ) kTEE feEf −−≈)(
( ) kTEE feEf −−−≈ 1)(
EE143 – Ali Javey
Electron and Hole Concentrations
Remember: the closer Ef moves up to Ec , the larger n is; the closer Ef moves down to Ev , the larger p is.For Si, Nc = 2.8×1019cm-3 and Nv = 1.04×1019 cm-3.
kTEEc
CFeNn /)( −=
kTEEv
FVeNp /)( −=
Nc is called the effectivedensity of states.
Nv is called the effectivedensity of states of the valence band.
EE143 – Ali Javey
Shifting the Fermi Level
EE143 – Ali Javey
I. (i.e., N-type)
If ,
II. (i.e., P-type)
If ,
ad NNn −=
nnp i2=
iad nNN >>−
ad NN >> dNn = di Nnp 2=and
ida nNN >>− da NNp −=
pnn i2=
da NN >> aNp = ai Nnn 2=and
General Effects of Doping on n and p
EE143 – Ali Javey
• When an electric field is applied to a semiconductor, mobile carriers will be accelerated by the electrostatic force. This force superimposes on the random thermal motion of carriers:
E.g. Electrons drift in the direction opposite to the E-fieldCurrent flows
Average drift velocity = | v | = μ ECarrier mobility
12
3
45
electron
E
123
45
electron
E =0
Carrier Drift
EE143 – Ali Javey
• Mobile carriers are always in random thermal motion. If no electric field is applied, the average current in any direction is zero.
• Mobility is reduced by1) collisions with the vibrating atoms
“phonon scattering”
2) deflection by ionized impurity atoms “Coulombic scattering”
-
Si
-
As+
-B--
Carrier Mobility
EE143 – Ali Javey
1E14 1E15 1E16 1E17 1E18 1E19 1E20
0
200
400
600
800
1000
1200
1400
1600
Holes
Electrons
Mob
ility
(cm
2 V-1 s
-1)
Total Im purity Concenration (atom s cm -3)Na + Nd (cm-3)
impurityphonon
impurityphonon
μμμ
τττ
111
111
+=
+=
Total Mobility
EE143 – Ali Javey
Jp,drift = qpv = qpμp
Jn,drift = –qnv = qnμn
Jdrift = Jn,drift + Jp,drift = σ =(qnμn+qpμp)
conductivity of a semiconductor is σ = qnμn + qpμp
Resistivity, ρ = 1/ σ
∴
Conductivity and Resistivity
EE143 – Ali Javey
N-type
P-type
ρ = 1/σ
DOPA
NT D
ENSI
TY cm
-3
RESISTIVITY (Ω⋅cm)
Relationship between Resistivity and Dopant Density
EE143 – Ali Javey
• Rs value for a given conductive layer (e.g. doped Si, metals) in IC or MEMS technology is used– for design and layout of resistors– for estimating values of parasitic resistance in a device or
circuit
WLR
WtLR s== ρ
tRs
ρ≡
Rs is the resistance when W = L (in ohms/square)
if ρ is independent of depth x
V+ _
L
tW
I
Material with resistivity ρ
V+ _
L
tW
I
Material with resistivity ρ
V+ _
L
tW
I
Material with resistivity ρ
Sheet Resistance
EE143 – Ali Javey
Particles diffuse from higher concentration to lower concentration locations.
Diffusion Current
EE143 – Ali Javey
dxdnqDJ ndiffusionn =, dx
dpqDJ pdiffusionp −=,
D is called the diffusion constant. Signs explained:n p
x x
Diffusion Current
EE143 – Ali Javey
Generation/Recombination Processes
Recombination continues until excess carriers = 0.Time constant of decay is called recombination lifetime
EE143 – Ali Javey
A PN junction is present in almost every semiconductor device.
N- type
P- type
Donors
V
I
Reverse bias Forward bias
N P
VI
diodesymbol
– +
PN Junctions
EE143 – Ali Javey
N-region P-region(a)
(b)
(c)
(d)
Depletionlayer
NeutralP-region
NeutralN-region
Ef
n 0 and p 0in the depletion layer
Ec
EfEv
Ec
Ev
Ef
Ec
Ev
Ef
Ev
Ec
≈ ≈
Energy Band Diagram and Depletion Layer
2lni
adbi n
NNq
kT=φ
EE143 – Ali Javey
Qualitative Electrostatics
Band diagram
Built in-potential
From ε=-dV/dx
EE143 – Ali Javey
Effect of Bias on Electrostatics
EE143 – Ali Javey
Current Flow - Qualitative
EE143 – Ali Javey
dep
depir τ
WqnAII += 0
)1(0 −= kTVqeII
⎟⎟⎠
⎞⎜⎜⎝
⎛+=
an
n
dp
pi NL
DNL
DAqnI 2
0
PN Diode IV Characteristics
Solar Cells
N P
-
+
short circuit
lightIsc
(a)
V0.7 V
–IscMaximumpower-output
Solar CellIV
IDark IV
0
(b)
Ec
Ev
Eq.(4.9.4)
Eq.(4.12.1)
p-i-n Photodiodes•Only electron-hole pairs generated in depletion region (or near depletion region) contribute to current
•Only light absorbed in depletion region contributes to generation–Stretch depletion region–Can also operate near avalanche to amplify signal
Light Emitting Diodes (LEDs)
•LEDs are typically made of compound semiconductors–Why not Si
EE143 – Ali Javey
MOS: Metal-Oxide-Semiconductor
SiO2
metalgate
Si body
Vg
SiO2
gate
P-body
Vg
N+N+
MOS capacitor MOS transistor
MOS Capacitors
Ideal MOS Capacitor
– Oxide has zero charge, and no current can pass through it.– No charge centers are present in the oxide or at the oxide-
semiconductor interface.– Semiconductor is uniformly doped– ΦM = ΦS = χ + (EC – EF)FB
Ideal MOS CapacitorAt Equilibrium:
Ideal MOS CapacitorUnder Bias
– Let us ground the semiconductor and start applying different voltages, VG, to the gate
– VG can be positive, negative or zero with respect to the semiconductor
– EF,metal – EF,semiconductor = – q VG
– Since oxide has no charge (it’s an insulator with no available carriers or dopants), d Eoxide / dx = ρ/ε = 0; meaning that the E-field inside the oxide is constant.
Inversion conditionIf we continue to increase the positive gate voltage, the bands at the semiconductor bends more strongly. At sufficiently high voltage, Ei can be below EF indicating large concentration of
electrons in the conduction band.
We say the material near the surface is “inverted”. The “inverted” layer is not gotten by chemical doping, but by applying E-field. Where did we get the electrons from?
When Ei(surface) – Ei(bulk) = 2 [EF – Ei(bulk)], the condition is start of “inversion”, and the voltage VG applied to gate is called VT (threshold voltage). For VG > VT, the Si surface is inverted.
Ideal MOS Capacitor –n-type Si
Electrostatic potential, φ(x)Define a new term, φ(x) taken to be the potential inside the semiconductor at
a given point x. [The symbol φ instead of V used in MOS work to avoid confusion with externally applied voltage, V]
)]((bulk)[1)( ii xEEq
x −=φ
(surface)](bulk)[1iiS EE
q−=φ
](bulk)[1FiF EE
q−=φ
Potential at any point x
Surface potential
φF > 0 means p-type φF < 0 means n-type
| φF | related to doping concentration
Electrostatic potential
φS = 2φF at the depletion-inversion
transition point (threshold voltage)
φS is positive if the bands bend\ …….?
Charge Density - Accumulation
p-type silicon accumulation condition
The accumulation charges in the semiconductor are ……. , and appear
close to the surface and fall-offrapidly as x increases.
One can assume that the free carrier concentration at the oxide-semiconductor
interface is a δ-function.
M O S
VG < 0p-Si
Accumulation of holes
Charge on metal = −QM
Charge on semiconductor = − (charge on metal) |QAccumulation| = |QM|
x
Charge Density - Depletionp-type Si,
depletion conditionThe depletion charges in Si are
immobile ions - results in depletionlayer similar to that in pn
junction or Schottky diode. VG > 0
M O Sp-Si
Depletion of holes
wQM|q NA A W| = |QM|(−) (+)
If surface potential is φs, then the depletion layer width W will be
SA
Si2 φεqN
W = Does this equation look familiar?
Charge Density - Inversion
VG>>0
M O Sp-Si
Depletion of holes
wQM
Inversion electrons:δ-function-like
p-type Si, strong inversion
Once inversion charges appear, they remain close to the
surface since they are …….. Any additional voltage to the gate results in extra QM in gate and get compensated by extra inversion
electrons in semiconductor.
So, the depletion width does not change during inversion. Electrons appear as δ-function near the surface. Maximum depletion layer width W = WT
MOS C-V characteristics
•The measured MOS capacitance (called gate capacitance) varies with the applied gate voltage
– A very powerful diagnostic tool for identifying threshold voltage, oxide thickness, substrate doping concentration, and flat band voltage.
– It also tells you how close to an ideal MOSC your structure is.
•Measurement of C-V characteristics– Apply any dc bias, and superimpose a small (15 mV) ac
signal (typically 1 kHz – 1 MHz)
C-V: under accumulation
VG < 0
M O Sp-Si
Accumulation of holes
x
Consider p-type Si under accumulation.
VG < 0.Looks similar to parallelplate capacitor.
CG = Cox
where Cox = (εox A) / xox
CG is constant as a function of VG
C-V: under depletionDepletion condition:VG > 0
CG is Cox in series with Cswhere Cs can be defined as “semiconductor capacitance”
Cox= εox A / xoxCs = εSi A / W
CG = Cox Cs / (Cox + CS)s
A
Si2φ
ε=
qNW
where φs is surface potential
VG > 0
M O S
p-type Si
Depletion of holes
WQM
Cox Cs
CG decreases with increasing VG
C-V: under inversion (high frequency)
VG >>0
M O Sp-Si
Depletion of holes
WQM
Inversion electronsδ- function
Cox Cs
VG = VT and VG > VTInversion condition φs = 2 φF
FA
SiT 22
φε
==Nq
WW
At high frequency, inversionelectrons are not able to respondto ac voltage. Cox= εox A / xox
Cs = εSi A / WT
CG (ω →∞) = Cox Cs / (Cox + CS)
CG will be constant for VG ≥ VT
At low frequency, the inversion electrons will be able to respond to the ac voltage. So, the gate capacitance will be equal to the “oxide capacitance”(similar to a parallel plate capacitance).
CG (ω → 0) = Cox= εox A / xox
Ideal MOSC (p-type Si)
C-V: under inversion (low frequency)
CG increases for VG ≥ VT until it reaches Cox
CCox
accumulation depletion inversionVgVfb Vt
high frequency
low frequency
The First Transistor
1956 Physics Nobel Prize
Invention of the Field-Effect Transistor
In 1935, a British patent was issued to Oskar Heil. A working MOSFET was not demonstrated until 1955.
Today’s MOSFET Technology
Gate oxides as thin as 1.2 nm can be manufactured reproducibly.Large tunneling current through the oxide limits oxide-thicknessreduction.
Introduction to the MOSFET
Basic MOSFET structure and IV characteristics
Introduction to the MOSFET
Two ways of representing a MOSFET:
Complementary MOSFETs (CMOS)
When Vg = Vdd , the NFET is on and the PFET is off. When Vg = 0, the PFET is on and the NFET is off.
NFET PFET
Qualitative discussion: n-MOSFETVG > VT ; VDS ≈ 0ID increases with VDS
VG > VT; VDS small, > 0ID increases with VDS , but rate of increase decreases.
VG > VT; VDS ≈ pinch-offID reaches a saturation value, ID,satThe VDS value is called VDS,sat
VG > VT; VDS > VDS,satID does not increase further, saturation region.
Threshold voltage for NMOS and PMOSWhen VG = VT, φs = 2 φF; we get expression for VT.
FSi
A
ox
SioxFT 222 φ
εεε
+φ=NqxV
Ideal n-channel(p-silicon) deviceboth terms positive
⎟⎟⎠
⎞⎜⎜⎝
⎛φ
εεε
−+φ= |2|22 FSi
D
ox
SioxFT
NqxVIdeal p-channel(n-silicon) deviceboth terms negative
εSi / εox = = 11.9 / 3.9 ≈ 3
](bulk)[1FiF EE
q−=φ
φF > 0 means p-type φF < 0 means n-type
How to Measure the VT of a MOSFET
Vt is measured by extrapolating the Ids versus Vgs (at low Vds) curve to Ids = 0.
tgsdsnstgsoxedsat VVVVVCL
WI −∝−= μ)(
Vds = 10mV
Ids
Vgs
Vt
Quantitative ID-VDS Relationships – 1st attempt“Square Law”
( )⎥⎥⎦
⎤
⎢⎢⎣
⎡−−
μ=
2
2DS
DSTGoxn
DVVVVC
LZI satDS,DS0 VV << TG VV >;
ID will increase as VDS is increased, but when VG – VDS = VT, pinch-off occurs, and current saturates when VDS is increased further. This value of VDS is called VDS,sat. i.e., VDS,sat = VG – VT and the current when VDS= VDS,sat is called IDS,sat.
( )2TGox
satD, 2VV
LCZI −
μ= satDS,D VV > TG VV >;
Here, Cox is the oxide capacitance per unit area, Cox = εox / xox
ID-VDS characteristics expected from a long channel (ΔL << L) MOSFET (n-channel), for various values of VG
EE143 – Ali Javey
N-channel MOSFET
Layout (Top View)4 lithography steps are required:1. active area2. gate electrode3. contacts4. metal interconnects
EE143 – Ali Javey
1) Thermal oxidation(~10 nm “pad oxide”)
2) Silicon-nitride (Si3N4)deposition by CVD(~40nm)
3) Active-area definition(lithography & etch)
4) Boron ion implantation(“channel stop” implant)
Simple NMOS Process Flow
EE143 – Ali Javey
5) Thermal oxidation to growoxide in “field regions”
6) Si3N4 & pad oxideremoval
7) Thermal oxidation(“gate oxide”)
8) Poly-Si deposition by CVD
9) Poly-Si gate-electrodepatterning (litho. & etch)
10) P or As ion implantationto form n+ source and drainregions
Top view of masks
Simple NMOS Process Flow
EE143 – Ali Javey
11) SiO2 CVD
12) Contact definition(litho. & etch)
13) Al deposition by sputtering
14) Al patterningby litho. & etchto form interconnects
Top view of masks
Simple NMOS Process Flow