Nanoelectronic Devices

1
Nanoelectronic Devices
Gregory L. Snider Department of Electrical
Engineering University of Notre Dame
2
What are Nanoelectronic Devices?

• A rough definition is a device where
• The wave nature of electrons plays a significant
  (dominant) role.
• The quantized nature of charge plays a
  significant role.

3
Examples

• Quantum point contacts (QPC)
• Resonant tunneling diodes (RTD)
• Single-electron devices
• Quantum-dot Cellular Automata (QCA)
• Molecular electronics (sometimes not truly nano)

4
References

• Single Charge Tunneling, H. Grabet and M.
  Devoret, Plenum Press, New York, 1992
• Modern Semiconductor Devices, S.M. Sze, John
  Wiley and Sons, New York, 1998
• Theory of Modern Electronic Semiconductor
  Devices, K. Brennan and A. Brown, John Wiley and
  Sons, New York, 2002
• Quantum Semiconductor Structures, Fundamentals
  and Applications , C. Weisbuch and B. Vinter,
  Academic Press, Inc., San Diego, 1991

5
When does Quantum Mechanics Play a Role?
W V, pg. 12, Fig. 5
6
More Realistic Confinement
W V, pg. 13, Fig.6
7
Quantum Point Contacts
One of the earliest nanoelectronic devices QPCs
depend on ballistic, wave-like transport of
carriers through a constriction.
In the first demonstration surface split- gates
are used to deplete a 2D electron gas. The
confinement in the constriction produces subbands.
8
Quantized Conductance
When a bias is applied from source to drain
electrons travel ballisticly. Each
spin-degenerate subband can provide 2e2/h of
conductance.
Va Wees, PRL 60, p. 848, 1988
9
What About Temperature?
Thermal energy is the bane of all nanoelectronic
devices.
T2 gt T1
As the temperature increases more subbands become
occupied, washing out the quantized conductance.
All nanoelectronic devices have a characteristic
energy that must be larger than kT
10
Resonant Tunnel Devices
In a finite well the wavefunction penetrates into
the walls, which is tunneling
In the barrier
where
Transmission through a single barrier goes as
11
Two Barriers
Semiclassically a particle in the well oscillates
with
It can tunnel out giving a lifetime tn and
Now make a particle incident on the double
barrier
If Ei ? En then T T1T2 which is small
12
If Ei En then the wavefunction builds in the
well, as in a Fabry-Perot resonator
Which approaches unity for T1 T2
13
In Real Life!
Things are, of course, more complicated - No
mono-energetic injection - Other degrees of
freedom
In the well
In the leads
14
In k Space No One Can Hear You Scream!
For Transmission
To get through the barriers electrons must have E
gt Ec but must also have the correct kz. Only
states on the disk meet these criteria.
15
J is proportional to the number of states on the
disk, and therefore to the area of the disk
Note we have ignored the transmission probability
16
Scattering
Scattering plays an important but harmful role,
mixing in-plane and perpendicular states
BB p236
17
Single Electron Devices
The most basic single-electron device is a single
island connected to a lead through a tunnel
junction
If EC gt kT then the electron population on the
island will be stable. Usually we want Ec gt 3-10
times kT. For room temperature operation this
means C 1 aF.
18
If the temperature is too high, the electrons can
hop on and off the island with just the thermal
energy. This is uncontrollable.
19
What is an Island?

• Anywhere that an electron wants to sit can be
  used as an island
• Metals
• Semiconductors
• Quantum dots
• Electrostatic confinement

20
Single Electron Box
The energy of the configuration with n electrons
on the island is
Q Cs U
21
At a charge Q/e of 0.5 one more electron is
abruptly added to the island.
What does it mean to have a charge of 1/2 and
electron?
22
Single - Electron Transistor (SET)
When U0, no current flows.
Coulomb Blockade
When (CGU)/e 0.5 current flows.
Why?
One more electron is allowed on the island.
23
These are called Coulomb blockade peaks.
Is the peak the current of only one electron
flowing through the island?
No, but they flow through one at a time!
24
What about Temperature?
GD p181
As the temperature increases the peaks stay about
the same, while the valleys no longer go to zero.
This is the loss of Coulomb blockade. Finally
the peaks smear out entirely.
This shows the classical regime, such as for
metal dots. In semiconductor dots resonant can
cause an increase in the conductance at low
temperatures (the peak values increase).
25
SET Stability Diagram
You can also break the Coulomb blockade by
applying a large drain voltage.
26
Ultra-sensitive electrometers
Dot Signal
Add an electron
Lose an electron
Sensitivity can be as high as 10-6 e/sqrt(Hz)
27
Single Electron Trap
GD p123
This non-reversible device can be used to store
information.
28
Single Electron Turnstile
GD p124
This is an extension of the single electron trap
that can move electrons one at at time
29
Turnstile Operation
Why does it need to be non-reversible?
GD page 125
Can this be used as a current standard?
Issues Co-tunneling Missed transitions Thermal
ly activated events
30
Single Electron Pump
Here there are two coupled boxes, and an electron
is moved from one to the other in a reversible
process.
GD p128
Same Issues Missed transitions Thermally
activated events Co-tunneling
31
Background charge effect on single electron
devices
e-

• Nanometer scaled movements of charge in
  insulators, located either near or in the device
  lead to these effects.
• This offset charge noise (Q0) limits the
  sensitivity of the electrometer.

32
Background charge insensitive single electron
memory

• A bit is represented by a few electron charge on
  a floating gate.
• SET electrometer used as a readout device.
• Random background charge affects only the phase
  of the SET oscillations.
• The FET amplifier solves the problem of the high
  output impedance of the SET transistor.

K. K. Likharev and A. N. Korotkov, Proc. ISDRS95
33
Plasma oxide fabrication technique
34
Plasma oxide device

• Two step e-beam lithography on PMMA/MMA.
• Oxidation after first step in oxygen plasma
  formed by glow discharge.
• Oxide thickness characterized by VASE technique.

Ground
SET
BG
FG
CG

• 6 nm of oxide grown after 5 min oxidation in 50
  mTorr oxygen plasma at 10 W.

35
Hysteresis Loops

• SET conductance monitored on the application of a
  bias on the control gate.
• A back gate bias cancels the direct effect of the
  control gate on the SET.
• The change in the operating point of the SET is
  due to electrons charging and discharging the
  floating gate.

36
Zuses paradigm

• Konrad Zuse (1938) Z3 machine
• Use binary numbers to encode information
• Represent binary digits as on/off state of a
  current switch

The flow through one switch turns another on or
off.
37
Problems shrinking the current-switch
38
New paradigm Quantum-dot Cellular Automata
Represent information with charge configuration.

• Zuses paradigm
• Binary
• Current switch

• Revolutionary, not incremental, approach
• Beyond transistors requires rethinking circuits
  and architectures

39
Quantum-dot Cellular Automata

• Represent binary information by charge
  configuration

Tunneling between dots
Polarization P 1 Bit value 1
Bistable, nonlinear cell-cell response Restoration
of signal levels Robustness against disorder
Neighboring cells tend to align. Coulombic
coupling
40
Variations of QCA cell design
41
Clocking in QCA
Keyes and Landauer, IBM Journal of Res. Dev. 14,
152, 1970
1
0
Clock
0
Clock Applied
Input Removed
Small Input Applied
0
but Information is preserved!
42
Quasi-Adiabatic Switching

• Clocking Schemes for Nanoelectronics
• Keyes and Landauer, IBM Journal of Res. Dev. 14,
  152, 1970
• Lent et al., Physics and Computation Conference,
  Nov. 1994
• Likharev and Korotkov, Science 273, 763, 1996
• Requires additional control of cells.
• Introduce a null state with zero polarization
  which encodes no information, in contrast to
  active state which encodes binary 0 or 1.

Clocking signal should not have to be sent to
individual cells, but to sub-arrays of cells.
43
Power Will Be a Limiter

• Microprocessor power continues to increase
  exponentially

100000
10000
Transition from NMOS to CMOS
1000
Power (Watts)
Pentium
100
P6
286
486
10
8086
386
8080
8008
8085
1
4004
0.1
1971
1974
1978
1985
1992
2000
2004
2008

• Power delivery and dissipation will be
  prohibitive !

Slide author Mary Jane Irwin, Penn State
University
Source Borkar De, Intel?
44
Power Density will Increase
10000
1000
Power Density (W/cm2)
100
8086
10
P6
8008
Pentium
8085
4004
386
286
486
8080
1
1970
1980
1990
2000
2010

• Power densities too high to keep junctions at low
  temps

Slide author Mary Jane Irwin, Penn State
University
Source Borkar De, Intel?
45
QCA power dissipation
QCA architectures can operate at densities above
1011 devices/cm2 without melting the chip.
46
QCA devices
Binary wire
Majority gate
Inverter
Programmable 2-input AND or OR gate.
47
Metal-dot QCA implementation
70 mK
dot metal island
48
Tunnel junctions by shadow evaporation
Oxidation of aluminum
49
Metal-dot QCA cells and devices
Switch Point
Input Double Dot

• Demonstrated 4-dot cell

(1,0)
(0,1)
Top Electrometer
Bottom Electrometer
A.O. Orlov, I. Amlani, G.H. Bernstein, C.S. Lent,
and G.L. Snider, Science, 277, pp. 928-930,
(1997).
50
Switching of 4-Dot Cell
51
Majority Gate
Amlani, A. Orlov, G. Toth, G. H. Bernstein, C. S.
Lent, G. L. Snider, Science 284, pp. 289-291
(1999).
52
QCA Latch Fabrication
53
QCA Clocked Latch (Memory)
54
QCA Shift Register
55
Fan-Out
56
From metal-dot to molecular QCA
Metal tunnel junctions
dot metal island
70 mK
57
Charge configuration represents bit
HOMO
Gaussian 98 UHF/STO-3G
58
Double molecule
Considered as a single cell, bit is represented
by quadrupole moment. Alternatively consider it
a dipole driving another dipole.
59
Double molecule
HOMO
Isopotential ()
60
Core-cluster molecules
61
Core-cluster moleculesTheory of molecular QCA
bistability Allyl group
Variants with feet for surface binding and
orientation
62
Electron Switching in QCA
Molecular Dots
Metal Dots
Measure conductance
63
Electron Switching Demonstration
Capacitance peaks correspond to click-clack
switching within the molecule
JACS 125, 15250-15259, 2003
64
Clocked molecular QCA
65
Summary

• QCA may offer a promising paradigm for
  nanoelectronics
• binary digits represented by charge configuration
  
• beyond transistors
• general-purpose computing
• enormous functional densities
• solves power issues gain and dissipation
• Scalable to molecular dimensions
• Single electron memories represent the ultimate
  scaling