This document summarizes the results of simulations performed to analyze the performance of graphene nanoribbon tunnel field-effect transistors (TFETs). Key findings include: (1) Subthreshold swing and on/off current ratio vary inversely with channel width and positively with channel length; (2) Contact doping positively correlates with subthreshold swing and on/off ratio up to 0.28eV doping; and (3) Drain bias weakly positively correlates with subthreshold swing and negatively correlates with on/off ratio. The best performance was obtained for a device with length of 40nm, width of 5nm, drain bias of 0.1V, and contact doping of 0.24eV, achieving a subthreshold swing
10. Relevant Functions (analytical)
SS=
𝜕𝑉𝑔𝑠
𝜕log(𝐼𝑑)
= 𝐥𝐧 𝟏𝟎 ∗ (
𝝏𝑰𝒅
𝝏𝑽𝒈𝒔
∗
𝟏
𝑰𝒅,𝒐𝒏
)−𝟏
𝝏𝑰𝒅
𝝏𝑽𝒈𝒔
= 𝒆
𝝏𝑰𝒅
𝝏𝑬𝒗
𝒄𝒉 =
𝟐𝒆𝟐
𝒉
(
𝝏𝑻𝑾𝑲𝑩
𝑬𝒗
𝒄𝒉 𝑭 𝑬𝒗
𝒄𝒉 + 𝑻𝑾𝑲𝑩
𝝏𝑭(𝑬𝒗
𝒄𝒉)
𝝏𝑬𝒗
𝒄𝒉 )
𝑻𝑾𝑲𝑩 = 𝒆
−
𝟒𝜦 𝟐𝒎∗𝑬𝒈
𝟑
𝟐
𝟑𝒉 ∆𝝓+𝑬𝒈
F 𝑬𝒗
𝒄𝒉 = 𝐸𝑐
𝑠
𝐸𝑣
𝑐ℎ
𝑓𝑠 𝐸 − 𝑓𝑑 𝐸 𝑑𝐸
10
J. Knoch, S. Mantl and J. Appenzeller, "Impact of dimensionality on the performance of tunneling
FETs: Bulk versus one-dimensional devices," ScienceDirect, vol. 51, pp. 572-78, 2007.
11. Literature Review: MOSFET/TFET IV
of different material system
A. M. Ionescu and H. Riel, "Tunnel field-effect transistors as energy-efficient
electronics switches," Nature, vol. 479, pp. 329-337, 2011.
11
12. Literature Review: varying gate
overlap & differential voltage
Gate overlap improves SS
without degrading Ion and Ioff
Differential voltage between top and bottom gate
for a double gate TFET correlates positively with Ion/Ioff
Fiori, G.; Iannaccone, G., "Ultralow-Voltage Bilayer Graphene Tunnel FET," Electron Device Letters,
IEEE , vol.0, no.10, pp.1096,1098, Oct. 2009 doi: 10.1109/LED.2009.2028248
12
13. Literature Review: varying drain-
side gate underlap & drain doping
X. Yang, J. Chauhan, J. Guo, and K. Mohanram “Graphene tunneling FET and its applications in
low-power circuit design,” VLSI, pp. 263-268, 2010
13
Drain-side gate underlap and drain doping reduce the
ambipolar IV characteristics without sacrificing Ion/Ioff and SS
14. Result: varying channel width
14
Channel width varies inversely with
SS and correlates negatively
(exponential) with Ion/Ioff
𝐼𝑑 =
2𝑒
ℎ
𝑊
𝐸𝑐
𝑠
𝐸𝑣
𝑐ℎ
𝑇 𝐸 − 𝑈 𝑓𝑠 𝐸 − 𝑓𝑑 𝐸 𝑑𝐸
15. Result: varying channel width
y = 381.85e-0.554x
R² = 0.9697
y = 3E+08e-2.043x
R² = 0.979
1
10
100
1,000
10,000
100,000
1,000,000
0
50
100
150
200
250
0 1 2 3 4 5 6 7 8
ratio
SS(mV/dec)
width (nm)
SS (mV/dec)
Ion/Ioff
15
Channel width varies inversely with SS and
correlates negatively (exponential) with Ion/Ioff
17. Results varying channel length
17
y = 30782ln(x) - 70513
R² = 0.7531
1
10
100
1,000
10,000
100,000
0
20
40
60
80
100
120
140
160
180
0 20 40 60 80 100 120 140 160 180
Ratio
SS
(mV/dec)
length (nm)
SS (mV/dec)
Ion/Ioff
Channel length varies inversely with SS and
correlates positively (logarithmic) with Ion/Ioff
18. Results: varying doping in contacts
18
Channel doping correlates positively with SS (exponential) and
positively with Ion/Ioff (exponential) up until doping of around 0.28eV
Off
On
𝐸𝑐
𝐸𝑣
q∆𝑉𝐺
λ
Channel
Source Drain
𝑓𝑠 𝐸
19. Results: varying doping in contacts
y = 0.1836e15.587x
R² = 0.8899
y = 20.708e32.662x
R² = 0.9263
y = 3E+07e-33.01x
1
10
100
1,000
10,000
100,000
1,000,000
0
10
20
30
40
50
60
70
0.2 0.22 0.24 0.26 0.28 0.3 0.32 0.34 0.36 0.38 0.4
ratio
SS
(mV/dec)
doping (eV)
SS (mV/dec)
Ion/Ioff
19
Channel doping correlates positively with SS (exponential) and
positively with Ion/Ioff (exponential) up until doping of around 0.28eV
20. Results: varying drain bias
20
Drain bias correlates positively with SS (linear & weak)
and negatively with Ion/Ioff (exponential)
Off
On
𝐸𝑐
𝐸𝑣
q∆𝑉𝐺
λ
Channel
Source Drain
𝑓𝑠 𝐸
21. Results: varying drain bias
y = 366373e-26.58x
R² = 0.9464
1
10
100
1,000
10,000
100,000
1,000,000
0
2
4
6
8
10
12
0 0.05 0.1 0.15 0.2 0.25
ratio
SS
(mV/dec)
vd (V)
SS (mV/dec)
Ion/Ioff
21
Drain bias correlates positively with SS (linear & weak)
and negatively with Ion/Ioff (exponential)
22. Conclusion
SS of 6.4 mV/dec and Ion/Ioff of >25,000 were
obtained for length=40nm, width=5nm, vd=0.1 V, and
doping=0.24eV.
Further analysis is required to balance the trade-offs
among size, power, and performance.
In comparison to a MOSFET, high Ion/Ioff ratio and
steep SS over several decades indicate GNR TFET’s
superiority for ultra-low-voltage applications.
22
23. Future direction
Link experimental results with analytical equations
Adjust simulation to account for experimental
challenges
Include scattering (inelastic & elastic)
Alternative TFET designs
23
24. Appendix: Simulation Design
(continue)
Tight-binding Hamiltonian model
TFET setup:
Channel doping
Tri-gate
Non-equilibrium green function (NEGF)
Assumptions:
Room temperature
ballistic transport
electrodes are infinite electron reservoir
steady state
24
25. 𝐼𝑑 =
2𝑒
ℎ
𝑊 𝐸𝑐
𝑠
𝐸𝑣
𝑐ℎ
𝑇 𝐸 − 𝑈 𝑓𝑠 𝐸 − 𝑓𝑑 𝐸 𝑑𝐸
𝐺 = (𝐸𝐼 − 𝐻 − Σ 1 − Σ 2) −1
E : energy matrices from the electronic band structure
H : hamiltonian matrix
Σ 1,2 : self energy matrices from the contacts
Σ 1=Γ1𝑓1 , Σ 2=Γ2𝑓2
Γ: broadening matrices due to coupling with contacts
f: fermi functions describing number of electrons
𝐺𝑛 = 𝐺 Γ1𝑓1 + Γ2𝑓2 𝐺+
Electron density per unit energy
Appendix: NEGF
25
26. Appendix: NEGF (continue)
T(E)=Trace(Γ1𝐺Γ2𝐺+
)
Average transmission at different energy
U=𝑈𝐿 + 𝑈𝑁
Potential energy effecting the DOS , and hence the transmission T
𝑈𝐿 =
𝐶𝐺
𝐶𝐸
(−𝑞𝑉𝐺)+
𝐶𝐷
𝐶𝐸
(−q𝑉𝐷)
𝑈𝑁 =
𝑞2
𝐶𝑒
∆N
𝑓(𝐸) =
1
1+𝑒
𝐸−µ
𝑘𝑇
Probability that an electron will be at an energy state E given the
fermi level µ, and temperature T
𝐼𝑑 =
2𝑒
ℎ
𝑊 𝐸𝑐
𝑠
𝐸𝑣
𝑐ℎ
𝑇 𝐸 − 𝑈 𝑓𝑠 𝐸 − 𝑓𝑑 𝐸 𝑑𝐸
26