Upload
truongkhanh
View
224
Download
3
Embed Size (px)
Citation preview
MOSFET and HFET devices are both very similar to a plain capacitor
Let the area of the capacitor plates be A. The induced charge Q can be expressed asQ = q × A × ∆nS,where q = 1.6 ×10-19 C is the electron charge, ∆nS is the SURFACE CONCENTRATION of induced electrons, ∆nS = Q / (q × A);
1x1 cm2
a
What is the surface concentration?The bulk charge density, nthe layer thickness, a;then the surface concentration,nS = n × a
09 Heterojunction FET (HFET) principles
V
Metal
Semiconductor
dA
For the PLAIN CAPACITOR, C = ε ε0 ×A/d Q = C × V = ε ε0 ×A×V/d,
The induced concentration of electrons (which are negatively charged) in the top (metal) plate:
∆nSM = - ε ε0 ×V/(q×d) <0;in the bottom (semiconductor) plate:
∆nS = ε ε0 ×V/(q×d) >0;
For a given voltage, V, the induced charge increases as we decrease d
V
Metal
Semiconductor
dA
Estimation of induced charge
Suppose the semiconductor plate is doped with donor concentration ND;
The equilibrium electron concentration in the semiconductor, n0 = ND;For the layer thickness, a, the surface concentration nS0 = ND ×a;
The voltage needed to deplete the entire active layer ( the semiconductor plate) is referred to as the THRESHOLD VOLTAGE of the FET
For the n-doped layer the threshold voltage is negative in order to repulse the electrons.
The induced concentration at the threshold has to compensate the equilibrium one:∆nST = ε ε0 ×VT/(q×d) = - nS0
Therefore,
VT = - q×d ×nS0/ (ε ε0)
The threshold voltage of FETs
At the threshold the net concentration in the channel is zero:∆nST – nS0 = 0, where ∆nST = ε ε0 ×VT/(q×d)
When the applied voltage is above the threshold, V > VT,∆nS = ε ε0 ×V/(q×d)
nS = ∆nS – ∆nST = ε ε0 /(q×d) × (V – VT)
Note, ε ε0 /d = C1 the gap capacitance per unit areaTherefore,
nS = (C1/q) × (V – VT)
The above model is referred to as “charge control model” of FETs
The charge control model of FETs
The channel current is then: I = V0 (q nS µ Z) /L = V0 q µ Z (C1/q) × (V – VT) /L
I = V0 µ Z C1 × (V – VT) /L
FETs: general design considerations
The current through the channel is
RV
I 0= where V0 is the voltage applied between the DRAIN and the SOURCE
We are assuming that V0 << VT (we will see why, later on)
The channel resistance, R (Z is the device width):
ZnqL
ZanqLR
s µµ==
-+ G
Semiconductor
The gate length L
DS
+-
V0
V
Low drain bias
The main factors affecting FET performance (for any FET type):
µ I and gm
I = V0 µ Z C1 × (V – VT) /L
The transconductance, gm = dI/dV;The gm is the “responsivity” of FET.In the linear mode under consideration (V0 << V),
gm = V0 µ Z C1 /L
FETs: general design considerations
-+ G
Semiconductor
The gate length L
DS
+-
V0
V
L I and gm
Carrier mobility in the channel and the gate length are crucial parameters of any FET
Low drain bias
Different types of FETs
Metal - Oxide - Semiconductor FET (MOSFET)
d
W
The gate-channel insulator is made out of dielectric (SiO2), ε = 3.9
Junction FET (JFET)
The gate-channel insulator consists of the DEPLETION REGION, i.e. the same material as the channel.For GaAs, ε ~ 12; for GaN ε ~ 9.
a0a W
Different types of FETs
Metal-Semiconductor FET (MESFET)
The gate is formed by Schottky barrier to the semiconductor layer.The gate-channel insulator consists of the DEPLETION REGION, i.e. the same material as the channel. Very similar to the JFET
a0 a
Different types of FETs
electrons
The Heterojunction Field-Effect Transistor (HFET)
The channel of HFETs is formed by 2D electron gas (2DEG)
induced channel (2DEG)
Channel
HFET JFET, MOSFET, MESFET
Effects of high drain bias on FET characteristics
VD
VG+
+
DrainSource Gate
VD
VG+
+
DrainSource Gate
The gate- to drain voltage difference depends on the position along the gateSo does the induced charge
JFETMOSFET
Effects of high drain bias on FET characteristics
The channel narrowing at the drain edge of the gate causes current saturation in the FETs
The particular range of the gate voltage depends on the device type
Effects of high drain bias on FET characteristics
Velocity saturation due to high electric field in the channel also results in the I-V saturation
Electron velocity saturation due to high electric field in the channel
The average electric field in the channel, Eav ~ V0 /L
Can be extremely high for small L
I = V0 µ Z C1 × (V – VT) /L I = vS Z C1 × (V – VT)
v = µ × E ~ µ × V0 /L
after T.A. Fjeldly, T. Ytterdal and M. Shur, 1998
Undoped active layer
Very high NS; very high µ; very high vS (in sub-µ HFETs)
1960 - Accumulation layer prediction (Anderson)1969 - Enhanced mobility of 2DEG prediction (Esaki & Tsu)
1978 Enhanced mobility observed (Dingle et. al.)
1980 The first Heterojunction FET (HFET)1991 The first GaN based HFET (A. Khan)
electrons
The Heterojunction Field-Effect Transistor (HFET)
The channel of HFETs is formed by 2D electron gas (2DEG)
induced channel (2DEG)
The Heterojunction Field-Effect Transistor (HFET)
1) Mobility depends on the interactions between electrons and phonons and impurities. For the phonon scattering, the dependence of mobility on temperature:
A.k.a. High electron mobility FETs: why?
For the impurity scattering, the dependence of mobility on impurity concentration, N:
When the dependence on both temperature and impurities is taken into account,
The Heterojunction Field-Effect Transistor (HFET)Concentration dependence of electron mobility
T = 300 K
The Heterojunction Field-Effect Transistor (HFET)
Electron Drift velocity
mnvn max2
2= En − Eo ≈ hω l
The electron accelerates in the electric field until it gains enough energy to excite lattice vibrations:
where vnmax is the maximum electron drift velocity. Then the scattering process occurs, and the electron loses all the excess energy and all the drift velocity. Hence, the electron drift velocity varies between zero and vnmax, and average electron drift velocity (vn = vnmax/2) becomes nearly independent of the electric field:
vn ≈
hω l2mn
= vsn
Typically, vsn ≈ 105 m/s. Indeed, the measured drift velocity becomes nearly constant in high electric fields
The Heterojunction Field-Effect Transistor (HFET)
Electron Drift velocity
electric field (kV/cm)
elec
tron
vel
ocity
(100
,000
m/s
)
T = 300 K
Si
GaAs
InP
InGaAs3
2
1
00 5 10 15 20
In the heavily doped materials the peak electron velocity is lower
Heavily doped
The HFET basics
qφb
di
qVFB
qϕb
∆Ec
Ec
GaAs
qVN
AlGaAsmetal
EFi
EFp
Ec
Ev
Considering first the band diagram of an AlGaAs/GaAs HFET with flat bands in the GaAs buffer. As can be seen from this figure, the flat-band voltage is given by
VFB = φb − VN − ∆Ec + ∆EF( )/ q
The HFET basics
The HFET basics
Band diagram Charge and field profiles
From the Poisson equation, s
si
qnFεε0
=
Fi
F
At the threshold, ns~0 --> Fi ~0
qVN
The HFET basics
HFET threshold voltage
VN = qNd (x)εi (x)
x dx0
di
∫
When ns is close to zero, the Fermi level in the GaAs is close to the bottom of the conductance band. Therefore,
VT ≈ φb −qNd di
2
2εi− ∆Ec / q
For non-uniform doping profile,
VT ≈ φb − qnδdδ / εi − ∆Ec / q
For the “delta-doped” barrier layer,
The HFET basics
HFET I-V characteristics
qns = ci VGT − V x( )[ ]Above the threshold the HFET is similar to MOSFET, i.e.
where VGT = VG - VT
Id = Wµnqns F = Wµnci VGT − V( ) dV
dx
The drain current:
Id =Wµnci
L×
VGT VDS − VDS2 / 2[ ], for VDS ≤ VSAT
VGT2 / 2 , for VDS > VSAT
⎧ ⎨ ⎪
⎩ ⎪
where VSAT = VGT
The HFET basics
HFET I-V characteristics
gm =dId
dVGS VDS
The transconductance,
gm =βVDS , for VDS ≤ VSATβVGT , for VDS > VSAT
⎧⎪⎨⎪⎩
where β = Wµnci /L is called the transconductance parameter.
The HFET basics
HFET I-V characteristics
Velocity saturation in HFETs
v F( ) =
µF , F < Fsvs , F ≥ Fs
⎧ ⎨ ⎩
A two-piece model is a simple, first approximation to a realistic velocity-field relationship:
More realistic velocity-field relationships :
v F( ) =µF
1+ µF / vs( )m[ ]1/m
where m = 1….2 0.0
0.4
0.8
1.2
3210Normalized Field
m = 1
m = 2
m = �
The HFET basics
HFET I-V characteristics
Velocity saturation in HFETs
Id =Wµnci
L×
VGT VDS − VDS2 / 2[ ], for VDS ≤ VSAT
VL2 1 + VGT VL( )2 − 1⎡
⎣ ⎢ ⎤ ⎦ ⎥ , for VDS > VSAT
⎧
⎨ ⎪
⎩ ⎪
VSAT = VGT − VL 1+ VGT / VL( )2 − 1⎡
⎣ ⎢ ⎤ ⎦ ⎥
where VL = Fs L.
For VL >> VGT, we arrive to the same expression as with the constant mobility case.
The HFET basics
HFET I-V characteristics
Velocity saturation in HFETs
Id =Wµnci
L×
VGT VDS − VDS2 / 2[ ], for VDS ≤ VSAT
VL2 1 + VGT VL( )2 − 1⎡
⎣ ⎢ ⎤ ⎦ ⎥ , for VDS > VSAT
⎧
⎨ ⎪
⎩ ⎪
VSAT = VGT − VL 1+ VGT / VL( )2 − 1⎡
⎣ ⎢ ⎤ ⎦ ⎥
where VL = Fs L.
In the opposite limit, when VL << VGT, we obtain VSAT ≈ VL
Isat ≈ βVLVGT
where β = Wµnci /L is the transconductance parameter.