Australian Journal of Basic and Applied Sciences, 10(1) January 2016, Pages: 207-213

AUSTRALIAN JOURNAL OF BASIC AND

APPLIED SCIENCES
ISSN:1991-8178 EISSN: 2309-8414
Journal home page: www.ajbasweb.com

Design of Multiplexer Based 64-Bit SRAM using QCA

1
K. Pandiammal and 2D. Meganathan

1
Dept of Electronics and Communication, Jerusalem College of Engineering, Chennai, India.
2
Dept of Electronics Engineering, M.I.T, Anna University, Chennai, India.

Address For Correspondence:

K. Pandiammal, Dept of Electronics and Communication, Jerusalem College of Engineering, Chennai, India.
E-mail: [email protected]

ARTICLE INFO ABSTRACT

Article history: Quantum-dot Cellular Automata is a new trend in nanotechnology and investigated as
Received 10 December 2015 an alternative to the current CMOS technology. QCA provides an attractive
Accepted 28 January 2016 computational paradigm to design digital system architecture. This work proposes a
Available online 10 February 2016 novel multiplexer-based 64-bit SRAM architecture in which the area and clock delay
are optimized. The loop-based memory in motion paradigm is accomplished in each
Keywords: QCA memory cell and clocking zones are shared between memory cells. The read/write
QCA, SRAM, CMOS. operations are implemented by simplified control circuitry with less area for the higher
order memory array. The Proposed QCA memory cell with control circuit uses 104-
QCA cells and reduced area. The advantages of the proposed work is sharing the clock
zones in a column reduces the number of clock delay. For 8x8 memory utilizes only 5-
clock cycles to complete read/write operations.

INTRODUCTION

Advancement in CMOS fabrication technology in the last few decades reduces the feature size of the
transistor and scales down the supply voltage. Due to the decreasing supply voltage, the power consumption
from leakage current is a big challenge for transistor circuits (Rairigh, D., 2005). As the device size is reduced,
power dissipation, leakage current, interconnect wiring and capacitances become a potential bottlenecks to the
circuit performance. The International Technology Roadmap for Semiconductors (ITRS) predicts that the size
limit for CMOS technology to be 5 to 10 nm by 2016 (Lent, C., 1994) and summarizes several possible
technology solutions ((IRTS), 2007) Nanotechnology is an alternative to these problems and the Quantum-Dot
cellular automata is one of the attractive alternatives. QCA technology has been introduced in 1993 (Amlani, I.,
1999) and the experimental devices for semiconductor, molecular, and magnetic approaches have been
developed. The molecular QCA devices can achieve high density of 103 per cm2 and clock speed could be from
1 to 10THz at room temperature (Cowburn, R. and M. Welland, 2000) and (Frost, S., 2002). In terms of feature
size, it is claimed that the size of the basic QCA cell can be implemented by few nanometres molecular
fabrication at room temperature (Lent, C., 2003). The power dissipation of a QCA device with 1011 cells is
approximately 100mW at 10THz (Timler, J. and C.S. Lent, 2002; Kummamuru, K., 2002).
QCA cell is arranged as a square pattern with four aluminum metal islands connected via tunnel junctions
made of Al/AlOx as shown in Fig. 1. Thermal fluctuation is avoided by keeping quantum – dot charging energy
is much higher than thermal energy. Though the four dots are located at the corners of the cell and only two
electrons are injected into a cell and makes two possible configurations of polarization which is encoded as
binary ‘0‘ and ‘1’.
The QCA clock is used to control the signal propagation among QCA cells. QCA clocking has been
operated using quasi-adiabatic switching, which allows recycling of energy by returning stored energy back to

Open Access Journal

Published BY AENSI Publication
© 2016 AENSI Publisher All rights reserved
This work is licensed under the Creative Commons Attribution International License (CC BY).
http://creativecommons.org/licenses/by/4.0/

To Cite This Article: K. Pandiammal and D. Meganathan., Design of Multiplexer Based 64-Bit SRAM using QCA. Aust. J. Basic & Appl.
Sci., 10(1): 207-213, 2016

Electroniccopy
Electronic copy available
available at:
at: https://ssrn.com/abstract=2790462
https://ssrn.com/abstract=2790462
208 K. Pandiammal and D. Meganathan, 2016
Australian Journal of Basic and Applied Sciences, 10(1) January 2016, Pages: 207-213

the supply and reduces the total energy drawn from the power supply. The computation process has been
enabled by quantum-mechanical tunneling and columbic interaction between QCA cells.

Fig. 1: QCA cell.

Fig. 2: Polarization of QCA Cells for different Clocking zones.

There are four-different clocking zones such as Switch, Hold, Release and Relax. The Polarization of QCA
clocking is shown in Fig. 2. During Switch phase, the inter-dot barriers are slowly raised and the computation is
performed. During the phase of Hold, the inter-dot barriers are kept high and the QCA cells retain their states
and in the next phase of Release, the barriers are lowered and the cells are relaxed to an unpolarized state.
During relax, the barriers are kept low and the cells remain in unpolarized state.
A set of equally sized tiles together forms complex circuits in the square formalism. The QCA memory can
be designed using square formalism (Berzon, D. and T. Fountain, 1999). The main advantage of this technique
is simple geometric layout and disadvantage of this methodology is lower density and spatial redundancy and
additional control circuit.
The H-memory architecture has been introduced by researchers from Notre Dame University (Frost, S.,
2002), it uses a squared shaped spiral structure that loops back itself for storing data and the sections of each
layer of the spiral can be in the same clocking zone. It is a complete binary tree structure and the additional
control circuit is provided at each node to perform read/write operation. Though the word size of each memory
cell is increased by adding extra layers, the number of clocking zones is not increased. The architecture
achieves high density and uniform access time but requires complex control logic circuit and the memory size is
not linear with cell count.
The parallel memory architecture was proposed in (Walus, K., 2003). In this architecture, one-bit
information can be stored at each memory cell which is designed using 158 QCA cells with simple read/write
circuit. The limitation of this approach is the requirement of the large number of clocking zones.
The design of parallel QCA memory was introduced in (Vankamamidi, M., 2005), the number of clocking
zones was independent of the memory size and shared among all memory cells in a column. Also, the read/write
control circuit was very simple and requires two additional clock signals. In the line-based architecture, the
storage was achieved by moving data back and forth in QCA line. This line-based memory architecture design
needs additional three zones for storage and the four-step process whose timing was different from the adiabatic
switching.
Tile-based memory architecture has been discussed in (Vankamamidi, M., 2008) contains three tiles and
placed in the order of input tile, output tile and memory tile to store one-bit information. The number of memory
tiles can be extended between input and output tiles according to the storage of N-bit information in which the
input tile is used to multiplex new input values into the memory loop. The memory tile is designed using 74
QCA cells, whereas 24 and 54 QCA cells are utilized for the input and output tiles.
The synchronization of line-based memory architecture has been improved in (Taskin, B. and B. Hong,
2006) using dual phase clocking signal and the clock zone required for data storage was reduced to two. The
dual phase clocking signal was excited using single clock generator.

Multiplexer-Based Memory Architecture:

This work proposes a novel multiplexer-based 64-bit SRAM architecture in which the area and clock delay
are optimized. The one-bit memory cell is designed using a 2 to 1 multiplexer with control circuit. The circuit

Electroniccopy
Electronic copy available
available at:
at: https://ssrn.com/abstract=2790462
https://ssrn.com/abstract=2790462
209 K. Pandiammal and D. Meganathan, 2016
Australian Journal of Basic and Applied Sciences, 10(1) January 2016, Pages: 207-213

diagram of the memory cell and its QCA layout are shown in Fig. 3 & Fig. 4 respectively. The address decoder
circuit selects the entire row of the memory cells and the additional control is provided for read/write operations.
The input control circuit is designed using two majority gates and the output control circuit consists of one
majority gate and both functions efficiently to perform read/write operations. In QCA-based memory logic,
memory must be kept in motion, i.e., the memory state has to be continually moved through a set of QCA cells.
These cells are connected in a loop and are partitioned into four clocking zones. At any point of time, one of the
zones must be in hold phase to retain the information. The Proposed QCA memory cell with control circuit uses
104-QCA cells and has reduced area.

Fig. 3: Circuit diagram of one-bit memory cell.

Fig. 4: QCA layout of memory cell.

Table I: One-Bit Memory Cell.

SEL R/W Operation Data out
High High Write Low
High Low Read Data In
Low High or Low Holding Low

Table I shows the operation of one-bit memory cell. When the R/W and row select signals are logically
high, the new input state is written into memory cell and hold it until it receives the next control signal. If R/W
goes low and the row select remains high, the memory state is read out by the data out pin. When row select
signal goes low, data will be held in the memory loop. The Read/Write control circuit is implemented using
majority gates which is defined mathematically in Equation - (1) and (2). The equation – (3) describes the
funtion of Multiplexer-based memory loop which is aquiring control signal CS from control circuit.

CMOS based SRAM is designed using parallel architectures, in which the all the bits of information are
read-out simultaneously. But in QCA-based memory design, data read and write in the memory cell is
performed in different clock zones which makes it impossible to read the data simultaneously. Introducing
delay in memory cells will synchronize the read operation. In CMOS SRAM, wire delay is not considered, but
in QCA SRAM, wire is designed using QCA cells which will increase the latency and size. Area of QCA
circuits can be determined using QCA designer tool. The size of the one-bit memory cell, including the control

Electroniccopy
Electronic copy available
available at:
at: https://ssrn.com/abstract=2790462
https://ssrn.com/abstract=2790462
210 K. Pandiammal and D. Meganathan, 2016
Australian Journal of Basic and Applied Sciences, 10(1) January 2016, Pages: 207-213

circuit is 0.17 µm2 and the size of 64-bit memory array including the control circuit is 13.6 µm2.The QCA layout
of 64-bit SRAM is shown in Fig. 5. Table III lists the performance of the proposed one-bit memory cell, 4X4
and 8X8 memory arrays. It has been observed that the higher order memory arrays occupies less area and
complete the process in less number of clock cycles because of sharing the control circuit. The performance of
the one-bit cell is compared with the best published works and is listed in Table IV. The proposed one-bit
memory cell occupies less area, realized by less number of cell count and reduced clock cycles to complete the
operation than the existing works. Table IV shows the comparison of 1-bit QCA SRAM with 6T SRAM using
CMOS technology.

Fig. 5: QCA Layout of 64-bit SRAM.

The inverted and non inverted signals are obtained using coplanar wire crossing option which is shown in
Table II and its QCA layout is displayed in Fig. 6.

Table II: Coplanar Wire Crossing.

Data in Data out
X1 X1
Y1 ~Y1
X2 X2
Y2 Y2

Fig. 6: QCA Coplanar wire crossing.

Table III: Performance Of The Proposed Work.

Performance metrics
Size Area
Architecture Complexity Cell count Clock cycle
(µm2 )
Proposed work one-bit Multiplexer-based parallel Simple read/write 1.5
104 0.17
(dia=18nm) memory operations (6 clock zones)
Proposed work Multiplexer-based parallel Simple read/write 3 (11 clock
2862 6.31
(4x4 Par) (dia=18) memory operations zones)
Proposed work Multiplexer-based parallel Simple read/write 5 (21 clock
9264 13.6
(8x8 Par) (dia=18) memory operations zones)

Electroniccopy
Electronic copy available
available at:
at: https://ssrn.com/abstract=2790462
https://ssrn.com/abstract=2790462
211 K. Pandiammal and D. Meganathan, 2016
Australian Journal of Basic and Applied Sciences, 10(1) January 2016, Pages: 207-213

Simulation Results:
Fig. 7(a) shows the number of memory bits Vs QCA cells. It has been observed that the QCA cell count
including control circuits is linearly varies with respect to number of memory bits. Fig. 7(b) shows plot of
memory Vs Clock delay. The clock delay is specified in terms of clock zones i.e one clock delay is equal to
four clock zones. It reveals that the clock delay is almost varies linearly for higher order bits and it is a sharp
curve slightly for the lower order memory bits which is obtained due to area of the control circuit. Fig. 7(c)
shows the memory bits Vs area. It has been observed that the area of the arrays including control circuits varies
linearly with respect to number of memory bits. For bistable approximation and coherence vector engine, the
clock low and clock high is defined 9.800000e-22s and 3.800000e-23s respectively.

Table IV: comparison of proposed work with best published works.

Performance metrics
Comparisons Area Clock
Architecture Complexity Cell count
(µm2 ) cycle
K. Walus (one– bit)
Parallel memory Complex circuitry 181 0.29 8
[11]
Vankamamidi (one-
Line-Based parallel memory Complex read/write circuitry 158 unknown unknown
bit) [12]
Vankamamidi (one -
Tile-based serial memory Complex input-output tiles 152 unknown unknown
bit) [16]
Moein Kianpour
(one-bit)(dia=10nm) Loop-based parallel memory Simplified control circuit 140 0.25 2
[17]
1.5
Proposed work one-bit Multiplexer-based parallel
Simple read/write operations 104 0.17 (6 clock
(dia=18nm) memory
zones)

Memory bits Vs QCA cell count of Proposed Architecture

10000

9000

8000

7000
QCA cell count----->

6000

5000

4000

3000

2000

1000

0
0 10 20 30 40 50 60 70

. No of memory bits---->

Fig. 7: (a) Number of memory bits Vs cell count.

Memory bits Vs clock delay of Proposed Architecture
5

4.5

4
clock delay----->

3.5

2.5

1.5
0 10 20 30 40 50 60 70
No of memory bits---->

Fig. 7: (b). Number of memory bits Vs delay.

Memory bits Vs Area of Proposed Architecture
14

12

10
Area (um 2)----->

0
0 10 20 30 40 50 60 70
No of memory bits---->

Fig. 7: (c) Number of memory bits Vs area.

Electroniccopy
Electronic copy available
available at:
at: https://ssrn.com/abstract=2790462
https://ssrn.com/abstract=2790462
212 K. Pandiammal and D. Meganathan, 2016
Australian Journal of Basic and Applied Sciences, 10(1) January 2016, Pages: 207-213

The simulation result of 1-bit memory is displayed in Fig. 8. It can be inferred from Table-V that the
proposed QCA memory cell occupies much lesser area than other Conventional SRAM designs. The speed of
QCA is high compared to the conventional SRAMs. The Simulation result of coplanar wire crossing is shown in
Fig. 9. In the Fig. 6, there is no change in electron flow direction hence the out coming signal is not inverted. If
the QCA cell is placed in such a way that the electron flow direction gets changed then the signal is inverted.

Conclusion:
The 64-bit Parallel memory architecture is designed efficiently using multiplexer-based memory loop. It
reduces the cell count, and area considerably. The speed has been increased due to reduced clock cyle in
mulplexer-based memory loop. A simple control algorithm is developed to synchnonize read write memory
operations. The higher oder 64x64 bit SRAM will be developed in future using different methods of cross over
options by increasing temperature stabilty.

Table V: Comparison Of 1-Bit Qca Sram With 6t Sram Using Cmos Technology.
Proposed 1-bit QCA memory cell 6T SRAM using CMOS Technology
Parameters
(including R/W control circuit) 0.25 um [18] 0.18 um [19] 65 nm [20] 40 nm [20] 32nm [20]
Area 0.17um^2 2.4x4.1 um^2 5.59 um^2 0.4 um^2 0.33 um^2 0.124 um^2
Clock delay 1.5 clock cycle 1.8 ns 1.8ns - - -

Fig. 8: Simulation results of 1-bit memory cell.

Fig. 9: Simulation results of coplanar wire crossing.

REFERENCES

Rairigh, D., 2005. “Limits of CMOS Technology Scaling and Technologies Beyond-CMOS,”
http://www.drlock.com/paper/cmos_survey.pdf.
Lent, C., P. Tougaw, W. Porod, 1994. "Quantum cellular automata: The physics of computing with arrays
of quantum dot molecules," In Proc. workshop physics and compution (pp: 5-13), Dallas, TX, 17-20.
International Technology Roadmap for Semiconductor (IRTS), 2007. http://www.irts.net.
Amlani, I., A. Orlov, G. Toth, G. Bernstein, C. Lent, G. Snider, 1999. “Digital Logic Gate Using Quantum-
Dot Cellular Automata,” Science, 284: 289-291.
Cowburn, R. and M. Welland, 2000. “Room Temperature Magnetic Quantum Cellular Automata,” Science,
287: 1466-1468.
Lent, C., B. Isaksen, M. Lieberman, 2003. “Molecular Quantum-Dot Cellular Automata,” J. Am. Chemical
Soc., 125: 1056-1063.
Timler, J. and C.S. Lent, 2002. “Power gain and dissipation in quantum-dot cellular automata,” Journal of
Applied Physics, 91(2): 823–831.

Electroniccopy
Electronic copy available
available at:
at: https://ssrn.com/abstract=2790462
https://ssrn.com/abstract=2790462
213 K. Pandiammal and D. Meganathan, 2016
Australian Journal of Basic and Applied Sciences, 10(1) January 2016, Pages: 207-213

Kummamuru, K., J. Timler, G. Toth, C.S. Lent, R. Ramasubramaniam, A.O. Orlov, G.H. Bernstein, G.L.
Snider, 2002. “Power gain in a quantum-dot cellular automata latch,” Applied Physics Letters, 81-7.
Berzon, D. and T. Fountain, 1999. “A memory design in qcas using the squares formalism,” in Proc. 9th
Great Lakes Symp. VLSI, 168–172.
Frost, S., A. Rodrigues, A. Janiszewski, R. Rausch, P. Kogge, 2002. Memory in motion: A study of storage
structures in qca. In First Workshop on Non-Silicon Computing.
Walus, K., A. Vetteth, G. Julien, V. Dimitrov, 2003. “Ram design using quantum-dot cellular automata,” in
Proceedings of the nanotechnology Conference and Trade Show, 2: 160–163.
Vankamamidi, M., Ottavi, F. Lombardi, 2005. “A line-based parallel memory for qca implementation,”
IEEE Transactions on Nanotechnology, 4(6): 690–698.
Vankamamidi, M., Ottavi, F. Lombardi, 2008. “A serial memory by Quantum cellular Automata,” in IEEE
Transaction on computers, 57(5): 606–618.
Taskin, B. and B. Hong, 2006. "Dual-phase line-based QCA memory design," In Proceedings of the IEEE
Conference on Nanotechnology, 302305.
Ottavi, M., V. Vankamamidi, F. Lombardi, S. Pontarelli, A. Salsano, 2005. “Design of a qca memory with
parallel read/serial write,” in Proceedings of the IEEE Symposium on VLSI, 292–294.
Vankamamidi, V., M. Ottavi, F. Lombardi, 2005. "Tile-based design of a serial memory in QCA," In
Proceedings of the ACM Great Lakes Symposium on VLSI, 201206.
Moein Kianpour and Reza Sabbaghi-Nadooshan, 2012. “A Novel Design and Simulation of 16 Bits RAM
Implementation In Quantum-dot Cellular Automata (QCA)”, IEEE conference in.
Hiroaki Nambu, Kazuo Kanetani, Kaname Yamasaki, Keiichi Higeta, Masami Usami, Yasuhiro Fujimura,
1998. “A 1.8-ns Access, 550-MHz, 4.5-Mb CMOS SRAMIEEE”, Journal Of Solid-State Circuits, 33-11: 1650-
1659.
Cangsang Zhao, Uddalak Bhattacharya Martin Denham, Jim Kolousek, Yi Lu, Yong-Gee Ng, Novat
Nintunze, Kamal Sarkez, and Hemmige D. Varadarajan, 1999. “An 18-Mb, 12.3-GB/s CMOS Pipeline-Burst
Cache SRAM with 1.54 Gb/s/pin” Ieee Journal Of Solid-State Circuits, 34-11: 1564-1571.
Chang, L., 2005. Stable SRAM Cell Design for the 32 nm Node and Beyond. Symposium on VLSI Circuits
Digest of Technical Papers, 128–129.

Electroniccopy
Electronic copy available
available at:
at: https://ssrn.com/abstract=2790462
https://ssrn.com/abstract=2790462