Fanout in Quantum Dot cellular Automata

Download Report

Transcript Fanout in Quantum Dot cellular Automata

Fanout in Quantum-dot Cellular
Automata
Kameshwar K. Yadavalli, Alexei O. Orlov, Ravi K.
Kummamuru, John Timler, Craig Lent, Gary Bernstein,
and Gregory Snider
Department of Electrical Engineering
University of Notre Dame
Supported by DARPA, NSF, ONR, and W. Keck Foundation
Outline
 Quantum Cellular Automata paradigm
 Initial devices
 Clocked QCA
 Power Delay Product
 Metal tunnel junction implementation
 QCA latch
 QCA shift register
 Fanout in QCA
 Summary
Quantum-dot Cellular Automata
Information encoded in charge configuration
A cell with 4 dots
2 extra electrons
and inter-dot tunneling
Polarization PP == +1
-1
Polarization
Bit value
value “1”
“0”
Bit
Polarization P = 1
Bit value “1”
Neighboring cells tend to align by
Coulombic coupling
QCA simulations are
available at
www.nd.edu/~qcahome/
Initial QCA devices
Majority gate
A
B
C
M
A
B
Out
C
Programmable 2-input
AND or OR gate.
Binary wire
1
0
1
0
1
0
0
1
Inverter
Ground state computation
Possibility of metastable
states
Lack of power gain and signal
level degradation
Benefits of Clocked QCA
 Power Gain. Energy is supplied
directly to cells by the clock,
not by the signal inputs alone.
 Pipelined Architectures. Clocked
cells in locked state acts as a
memory controlled by the clock.
A large QCA array can be
divided into sub-arrays using
phase shifted clocks.
 Fanout in QCA. A single latch
can drive multiple latches to
create a complex circuit.
PCLK
Pin
Pbath
Pout
Adiabatic clocking with
reversible computation can
beat the limit of power
dissipation per bit operation,
kTln2.
Implementation of Clocking in Quantum-dot
Cellular Automata
Semiconductor dots:
Clocking achieved by modulating
inter-dot barriers
P= +1
P= –1
Null State
Metallic or molecular dots:
Clocking achieved by modulating
the energy of a third dot
Lent et al., Physics and Computation Conference, Nov. 1994
Likharev and Korotkov, Science 273, 763, 1996
Clock signals need not have to be sent to individual cells,
but to sub-arrays of cells.
Clocking in QCA
Keyes and Landauer, IBM Journal of Res. Dev. 14, 152, 1970
energy
Initial State
With clock applied
1
0
x
Clock
Null State
Clock barrier is
slowly raised
Input removed
0
but Information
is preserved!
Differential Input
applied
Clock Applied
0
Switching Energy in QCA
Quasi-adiabatic operation of QCA devices leads to very low power
dissipation.
QCA Latch: A Building Block
The third, middle dot acts as an adjustable barrier for tunneling
Electrometer
A
SET
D1
-VIN
Vg
-VIN
MTJ
+VIN
~
D1
MTJ
D2
+VIN
D2
MTJ
MTJ
+VIN
D3
MTJ=multiple tunnel junction
+VIN
D3
1mm
SEM Micrograph of a QCA latch
Experiment: Single-Electron Latch in Action
D3
+
-VIN
VIN (mV)
Latch
 Weak input signal sets the direction of switching
 Clock drives the switching
T=100 mK
0.5
0.0
Input
+VIN
D2
D1
E1
VD1 (mV)
VCLK
VCLK (mV)
-0.5
0
-3
Clock
0.2
Hold “1”
Switch to “neutral”
-0.2 Switch to “1”
0.0
0
SET electrometer
“High”
-6
2
Switch to “0”
Switch to “neutral”
4
6
Time (sec)
Hold “0”
8
 Memory Function demonstrated
 Inverter Function demonstrated
10
QCA Two-Stage Shift Register




Two latches with two electrometers for readout
Inter-latch coupling by means of inter-digited capacitors (CC)
A Two-phase Clock to control electron switching is used
One latch serves as input for the other
SEM micrograph
+VIN
D1
CC
VCLK1
VCLK2
D2
CC
1mm
D4
D5
CC
-VIN
CC
D3
D6
-VIN
 Small external input applied – SR
remains in neutral state
 CLK1 applied
1st latch switches. Input now can be
removed
 CLK2 applied
2nd latch switches
 Process is repeated for the inverted
input
VCLK1 (mV) V +(mV)
IN
VCLK2
VD1 (mV)
VCLK1
0.2
0.0
-0.2
VCLK2 (mV)
+VIN
0.5
0.0
-0.5
0
-6
VD4(mV)
Operation of QCA Shift Register
0.2
0.0
-0.2
0
-6
0.00 0.35 0.70 1.05 1.40 1.75 2.10 2.45
Time (sec)
Fanout in QCA
Writing Information into
first stage latch
Transfer information into
second stage latches
CLK2
NULL
CLK1
CLK1
+
Input
-
NULL
Fanout in QCA
• Fanout in QCA allows for
complex circuits to be
designed and operated.
• A latch in the first stage
(L1) is coupled to two
latches (L2, L3) in the
second stage.
• A two phase clock (VCLK1,
VCLK2) controls information
transfer between the two
stages.
• Information is first
written into L1, then
clocked into L2, L3 on the
application of VCLK2.
SET
E1
L1
SET
(E2)
L2
e
+VIN/2
e
VCLK1
VCLK2
-VIN/2
e
SET
E4
L3
E3
SET
Operation of Fanout in QCA
• All the latches are initially
in null state.
• A differential input is then
applied to L1, to define the
polarization state.
• On the application of VCLK1,
electron switches in L1 and
is locked after the input is
removed.
• VCLK2 is then applied to L2,
L3, with the locked state
of L1 providing the input to
L2, L3.
• Information from L1 is
written to L2 and L3.
QCA with Fanout
•
•
•
•
An implicit demonstration of
power gain and the benefit of
clocking.
In the absence of clocking, there
will be signal level degradation in
a fanout gate.
Also, the middle latch will see
two kinks resulting in a higher
energy state, stopping the
computation.
As the second stage latches are
driven by a weak input, the
power gain in the second stage
latches is greater than unity.
•Enables multi-tasking architecture.
•Affords the creation of complex circuitry.
Kinks
Summary
 Clocked QCA offers a working paradigm for digital
nanoelectronics in the quantum realm:
orders of magnitude lesser power dissipation than
FETs
power gain for signal level restoration
pipelining
 Latch and shift register elements for information
processing
 Fanout gate in QCA paradigm is demonstrated
 Future Work: Molecules? High resistive junctions for
QCA latches in place of MTJs, for higher charging
energy
 Future Work: High speed measurements on QCA
Fabrication of Metal-dot QCA cells
Simple 4-dot cell is shown
No clocking yet!
 Dots = small metal (Al) islands separated by tunnel junctions (Al203)
Junctions: area of about 100 x 100 nm2 ; thickness is 0.1-0.5 nm
Charging energy is small, so that operation temperature is low (<1K)
High yield and good reproducibility allows proof of concept demonstration