ECE 6332 Design Review: QCA logic

chatventriloquistΤεχνίτη Νοημοσύνη και Ρομποτική

1 Δεκ 2013 (πριν από 3 χρόνια και 4 μήνες)

138 εμφανίσεις


ECE 6332 Design Review: QCA logic

by Wenlu Sun & Jiyati Verma

Topic

Logic device implementation using quantum cellular automata (QCA)

Description

Quantum Cellular Automata (QCA) or Quantum
-
dot Cellular Automata (QDCA) consist of an
array of quantum
-
dot ce
lls on a surface, connected locally by the interactions of the electrons contained in
each cell, the quantum state of each cell is used to encode binary information (see Fig. 1). The Coulomb
interaction connects the state of one cell to the state of its ne
ighbors, thus Logic gates and transmission
wires can be built (see Fig. 2), and the problems associated with small output currents and parasitic
capacitances of connecting wires do not occur. “Binary wires” composed of linear array of cells are
effective i
n transmitting information, coded in the cell states, from one place to another. As shown in Fig.
3, the wire is robust in the sense that it transmits a binary value, since the nonlinear response of the cell
to the polarization of its neighbors plays the r
ole of gain in a conventional digital circuit.



Figure 1. Schematic structure of unit QCA cell.
1

Figure 2. Implementation of a) transmission wire;








b) unit
-
inverter; c) simple fan
-
out.
1

Notice that








the bolded squares are fixed values.



Figure 3. Cell
-
Cell response function for the basic four
-
site cells. This shows


the polarization of P1 induced in cell 1 by the fixed polarization of its neighbor P2.
1


In nano
electronics, QCA technology comes under he branch of molecular electronics which has
come under much artillery in the past decade. At its inception, the promise and possibility of discovering
and fully implementing a new idea that could match silicon
-
based

technology fueled a staggering amount
of research between the 1980s and 2000s. But over time, as many of those promises turned out empty,
many outsiders began to treat molecular electronics like a pipe
-
dream. QCA technology is one example
of an idea that

initially showed promise in a multitude of ways. It was orders of magnitude smaller than
the transistors being manufactured at its time. Intel’s 75
-
MHz Pentium® Processor had a channel width
of 0.8 μm (800 nm). After the seminal paper of QCA technology
was published in 1994, 50 x 50 nm
2

tunnel junctions were experimentally demonstrated in 1997: a drastic reduction in size, with many other
benefits in sight.

In ECE6332, we are learning many new skills about analyzing, simulating, and designing ready
-
to
-
fa
bricate digital ICs, based on the supremely well
-
developed silicon technology infrastructure. Ideally,
once we are done with this class, we will be equipped with analytical skills and information that would
help us ameliorate underdeveloped technologies t
o a point where hopefully, they too can be implemented
on larger scales. Many of the papers we read so far impressed on us the successes of QCA technology:
clearly, despite these successes, other things stood in the way of the success of this elegantly si
mple,
transistorless approach to implement logic functions. Yet, QCA technology is not used anywhere, by
anyone apparently. So, to effectively size QCA technology up with silicon technology, a number of
questions must be addressed. What did QCA technolog
y need to make it big in the electronics industry?
What insurmountable problems stood in the way of commercializing this technology? Regardless of
limitations, other similarly odd
-
ball technologies found niche markets where their new features opened up
n
ew possibilities for electronics where silicon technologies could not. How come QCAs couldn’t do so as
well? Why did this technology get discarded in the end? Whatever it is about current silicon
-
technology,
something about it completely eclipsed QCA techn
ology, which was a pretty awesome idea to begin with.

Although we are milestones away from answering these questions, we scope out what issues of
this technology have and have not been addressed by literature in this first design review. To address
the br
oader context surrounding this topic, the purpose of our project would be more geared towards
using our new
-
found analytic skills to analyze the results published in literature for usability, scalability,
the ease of implementation, robustness, relative co
st. Alongside these, we will probably think of other
factors that determine the ultimate success of a technology when a super
-
successful technology such as
silicon technology already exists. In the next assignment, the proposal, we will extract the pertine
nt data
for the preliminary simulations. Simulations will begin in small
-
scale (10s of transistors) & medium
-
scale
(100s of transistors) circuit design.

Literature Review

Summaries of 12 publications considered, in chronological order (date published), to

reflect the
development of this field over time.


1. P. D. Tougaw and C. S. Lent, “Logical devices implemented using quantum cellular automata”,
J. Appl.
Phys.
75(3), 1994.

In this cornerstone paper, the authors theorize how to implement basic logic devic
es using
quantum cellular automata (QCA), like AND, OR and XOR gates, as well as designing wires. The
concept begins with the thought experiment of using a 2 by N square grids of quantum dots to function as
a quantum wire. Electrons would want to reside
on two diagonal islands due to electrostatic repulsion.
Using this Coulombic repulsion, a NOT gate could be made by rotating the unit 45 degrees. A majority
gate could be made by bringing in three leads to one unit, and the output on the fourth side woul
d give
the appropriate output. Thus, majority gates and NOT gates are the most easy to build using this
technology idea. Biasing the majority gate with a 0 or 1 results in an OR and AND gate, respectively.
Using these basic concepts, a one
-
bit full
-
adder
and XOR gates are illustrated also illustrated in the
paper.

This idea allows engineers to use quantum tunneling rather than fight it. To explore new logic
styles and arithmatic structures, we will probably be using this paper as the conceptual core of ou
r project
as it is the mother of the all the work following it.


2. P. D. Tougaw and C.S. Lent, “Dynamic behavior of quantum cellular automata”,
J. Appl. Phys.

Vol. 80,
No. 15, 1996.

This paper examines the dynamic behavior of quantum cellular automata, it

demonstrates how
dissipative coupling to thermal reservoirs makes switching times for real devices faster or slower. After
increasing switching time, the output state will be more deterministic. And when implementing with more
and more cells, the switchin
g time needs to be long enough so the input states can be completely
transmitted. The paper also spoke of how to reduce the size of the basis set required to accurately model
large QCA devices.



3. T. I. Kamins and R. S. Williams “Lithographic positioning

of self
-
assembled Ge islands on Si(001)”
Appl. Phys. Lett. Vol. 71, No. 9, 1997.


One of the first and greatest criticisms of this technology was that, at the time, there was no way
of mass assembling quantum dots (QDs) in an orderly way. So going behind
the scenes on the materials
science side of things, these authors attempt to make working QDCA using Ge quantum dots on a silicon
surface. The two
-
dimensional array of QDs is made from lithographically patterned features on thermally
grown oxide. The depos
ition of germanium dots forms the islands. AFM analysis proved that dome
-
like
quantum dots were indeed formed in a semi
-
regular manner, just not orderly enough to move on to
verification of AND/OR logic behavior from the given alignment of QDs.



4. C. S
. Lent and P. D. Tougaw, “A device architechture for computing with quantum dots”,
Proceedings
of The IEEE
, Vol. 85, No. 4, 1997.

This paper revisited how QCA cells work and interact in other computational architectures,
showing quantitatively how two adja
cent QCA cells respond to one another, and how to implement basic
logic gates and transmission wires using this. Basically, the ground states of QCA cells are labelled as “0”
and “1”, therefore the stable states are considered to be pretty robust even with

disordered wire and
randomly distributed tunneling energies and dot sizes. The timing issues in QCA computing are also
introduced in details, and the discussion of overly fast switching, hard
-
to
-
control and inefficient relaxation
of energy, resulting in m
eta stable states occurring in the system where the local energy minimum is not
the true ground state. The cells may then be stuck in a metastable state for a considerable period of time.
Adiabatic switching technique is introduced in this paper, which all
ows transitions to take place gradually
enough so the system state can track the instantaneous input states.


5. I. Amlani, A. O. Orlov, G. L. Snider, C. S. Lent, G. H. Bernstein, “Demonstration of a functional
quantum
-
dot cellular automata cell”
J. Vac.
Sci. Technol. B
, Vol. 16, No. 6, Nov/Dec. 1998

Another win for QCA technology was the experimental demonstration of a six
-
dot QCA using
metal. The six
-
dot version of the system consists of a pair of double dots and two single dots. This
implemented design

of the QCA cell is considered more symmetric than the previous four
-
dot system as it
is based on two capacitively
-
coupled identical double dots. Dolan
-
bridge technique was used to
fabricated the dots, and electron beam lithography was used to make the al
uminum islands and contacts,
followed by shadow evaporation processes. The resultant tunnel junctions are 50 x 50 nm
2
, and IV
characteristics revealed they had a resistance of 750 kΩ each. At low temperature (75 K), the QDCA
experimental operation is show
n to pretty closely match theoretical results for basic operations. They
have evidence that such technology can support high operating frequencies, the lower limit of the
maximum operating frequency being 13 MHz, and the higher limit, 1 GHz.


6. I. Amlani
, A. O. Orlov, G. Toth, G. H. Bernstein, C. S. Lent, G. L. Snider, “Digital Logic Gate Using
Quantum
-
Dot Cellular Automata”
Science
, Vol. 284, April 1999.

As a continuation to the previous paper, the authors continue to write of the results behind the
expe
riment verifying the functions 2x2 QDCA. Here, the
authors adhere to the original 4
-
dot structure, and connect
them in a ring by tunnel junctions, and verify operation
using 2 single dot electrometers. The device is operated by
applying inputs to the ga
tes of the cell. The operation of
logical AND and OR gates, using the majority gate setup, is
experimentally verified from electrometer outputs, showing
high correlation with theoretical simulations. Logic gate
output characteristics can be extracted fro
m the information
in this paper. Other strengths of this technology are
illustrated, such as a clear divide between the output high
(V
OH
) and output low (V
OL
) voltages, which is absolutely
necessary for digital logic. Simulated switching behavior is
also
compared to actual switching behavior in this paper.


Figure 4. An actual output characteristic showing majority
gate operation, where T = 20 s is the switching period. The
dashed line shows what theory predicts for 70 mK.
6




7. G. L. Snider, A. O. Or
lov, I. Amlani, X. Zuo, G. H. Bernstein, C. S. Lent, J. L. Merz, and W. Porod,
“Quantum
-
dot cellular automata: Review and recent experiments” in J. of App. Phys., Vol. 85, No. 8, April
1999.


This paper reveals a little bit more about the technical decisio
ns made to illustrate the operation
of QCA technology. For example, since the operation of a QCA cell depends on the position of each
electron, tracking the position of electrons within the cells becomes necessary. There are two ways to do
this: one is to

measure the conductance through each pair of dots within the cell. The conductance peaks
as the gate voltages are changed, indicating that the Coulomb blockade has been lifted for both dots
simultaneously, and a change in the dot population has occurred.
The other way to detect the change of
an electron position within the cell is using the electrometers. Using electrometers is what this group
decided to go with, although they have a sample of what a plot of conductance looks like in their 1998
paper.

The

authors surmise that QCA cells are scalable to molecular dimensions, and since the
performance improves as the size shrinks, a molecular QCA cell should operate at room temperature.
While the device they demonstrated operates using single electrons, they
thought that implementing QCA
technology using magnetic domains would also be possible.



8. John Timler and Craig S. Lent, “Power gain and dissipation in quantum
-
dot cellular automata”
J. Appl.
Phys.

Vol. 91, No. 2, 2001.

This paper developed a theoretic
al approach to examine the
energy flow in QCA devices and the role of power gain and power
dissipation in QCA cells. These authors correctly identify that any new
approach in nanoelectronics must address two questions: (1) do the
device exhibit power gain?

and (2) how much power is dissipated?
Their theoretical approach utilizes coherence vector formalisms to
come up with a detailed solution illustrating calculates polarization in
a QCA shift register. Using that solution, they examine how energy
flows betwe
en cells, from the clock, and to the environment, focusing
on steady
-
state behavior. Using that analysis, they extrapolate
energy dissipation and propogation delay times for QCA devices and
compare them to the future technology at the time. They clearly
show
that achieving ultra
-
low energy dissipation in QCA circuits is possible.


Figure 5. Power dissipation in QCA devices and how they compare to projected CMOS technology. The
vertical line corresponds to the theoretical limit of a cell’s switching speed
. Points A and B are SIA’s
predictions for high
-
performance CMOS devices for 2001 & 2014, respectively. Points C and D represent
the power delay properties of 30
-

and 20
-
nm transistor technology fabricated at Intel labs of the time.
8


9. J. M. Tour,
Molec
ular electronics: commercial insights, chemistry, devices, architecture and
programming
, World Scientific: 2003.

This book goes into some of the real world, higher
-
level issues that QCA logic must face to
succeed in industry, elaborating on both pros and c
ons of the technology. Wins include that Amlani et. al.
has successfully experimentally demonstrated switching behavior in six
-
dot QCAs, along with a
functioning majority gate, verifying the AND and OR operations. This technology also has hope because
QCA
s could work with as little as one millionth of an electron per bit of information. Even still, fan out is a
problem, as is cheap fabrication, since the experimentalists use e
-
beam lithography.




10. C. S. Lent and B. Isaksen, “Clocked Molecular Quantum
-
Dot Cellular Automata,” in
IEEE
Transactions on Electron Devices
, Vol. 50, No. 9, pp. 1890
-
1896, Sept. 2003.

Some weakness of the QCA idea are revealed in this paper, namely reliability and defect
tolerance. The way the negative slope of an inverter’s gai
n helps make conventional circuits robust, there
exists a need to recover signals in QCA logic. Clocking QCA logic is seen as one way of increasing the
power gain of signals, to ensure information is not lost.

The researchers have presented results on ot
her possible simplified molecular systems which
can also demonstrate the key requirements for QCA operation using a 6
-
quantum dot cells, and even
mixed
-
valence molecules. These QCA molecules can be designed to exhibit appropriate bistable
switching behavio
r, and clocked control of molecules is possible using an external electric field.


11. K. Walus, T. J. Dysart, et al. “QCADesigner: A rapid design and simulation tool for quantum
-
dot
cellular automata”,
IEEE Trans on Nanotech.

Vol. 3, No. 1, 2004.

This pap
er decribes the novel design and simulation tool for quantum
-
dot cellular automata,
namely, QCADesigner, QCADesigner gives the ability to quickly layout a QCA design by providing an
extensive set of CAD tools. This tool has already been used to design full
-
adders, barrel shifters, random
-
access memories, etc. The paper has showed in detail how to interpolate experimental data for QCA
systems with a large number of cells, and how to use the three most important simulation engines. Their
next step is to compi
le a DRC for QCA’s so this paper’s follow up is critical to our project.


12. S. K. Lim, R. Ravichandran, and M. Niemier,
Partitioning and Placement for Buildable QCA Circuits
,
ACM Journal on Emerging Technologies in Computing Systems
, Vol. 1, No. 1, 2005.

After enumerating the advantages of QCA technology, this article goes into their hurdles at
length. Their purpose is to explain how CAD can help research move from small circuits to small systems
of quantum
-
dot cellular automata (QCA) devices. Using thei
r ties to physical scientists who are working
to build real QCA devices, a set of near
-
term buildability constraints are compiled

much like the design
constraints used in DRC.

Progress

So far we have completed a literature review of published work done in our area of interest. The
strengths of QCA technology. Has a world of advantages. Beyond this, apparently there is a universe of
disadvantages, so from here on out we will be focusin
g on them. To go beyond thinking like researchers
and ultimately think like developers.

Remaining Tasks

[1] Extract pertinent numerical data from the graphs in literature


[2] Find PDK and DRC to simulate QCA logic in Cadence. The follow
-
up paper to publi
cation #6
(“QCADesigner: A rapid design and simulation tool for quantum
-
dot cellular automata”) will probably be
helpful in this endeavor as it was mentioned in their future research.


[3] How is power dissipated in such circuits? Is capacitance an issue?

Results from Amlani et. al. would
contribute to such device
-
level concerns.


[4] Compare the characteristics of same logic gates implemented with CMOS techniques and QCA
systems and weigh the trade
-
offs.


[5] Elaborate on the limitations of the technology
: design
-
wise, on the circuit
-
level, and on the fabrication
-
level. Work by the Amlani group would probably touch on this.