The International Journal Of Engineering And Science (IJES)

|| Volume || 5 || Issue || 9 || Pages || PP 61-68|| 2016 ||

ISSN (e): 2319 1813 ISSN (p): 2319 1805

Emerging Memory Technologies

O.D. Alao1, J.V. Joshua2, D.O. Kehinde3, E.O. Ehinlafa4, M.O. Agbaje5,
J.E.T Akinsola6
1, 2, 5 ,6
3

Department of Computer Science, Babcock University, Ilishan-Remo, Nigeria

Department of Basic Science, Babcock University, Ilishan-Remo, Nigeria
4
Department of Physics, University of Ilorin, Ilorin, Nigeria.

--------------------------------------------------------ABSTRACT----------------------------------------------------------The processor caches, main memory and storage system is an integral part of any computer system. As
information begins to accumulate, higher density and long term storage solutions are necessary. Due to this,
computer architects face some level of challenges in developing reliable, energy-efficient and high performance
memories. Also, existing storage devises are degrading in performance, cost, and sizes. Power consumption
from the factory has increased, as newer codes are written, and server hardware capabilities are not adequate
to handle big data of the future. New emerging memories (NEMs) are presently with its properties likely to open
doors to innovative memory designs to solve the problems. This paper looks at the features of the emerging
memory technologies, and compares incumbent memories types with the expected future memories.
Keywords: Memory Storage, New Emerging Memory Technologies, Spin-Transfer Torque Magnetic RandomAccess Memory (STT-MRAM), Resistive Random Access Memory (ReRAM) and Phase Change Memories PCM.
------------------------------------------------------------------------------------------------------------------------------------Date of Submission: 17 May 2016
Date of Accepted: 20 September 2016
--------------------------------------------------------------------------------------------------------------------------------------

I. INTRODUCTION
The pecking order of memory and storage device is a critical component of various computer systems. Processor
caches act as a subset of data and instructions stored in the memory. Data stored in the main memory are stored
in large, slow storage devices, such as disks and flash. Data from modern applications such books, maps, photos,
audios, videos, references, facts, and conversations rely on both real and offline processing and their dataset can
be in gigabytes, terabytes, zettabytes or even larger in size.
Regrettably, the scaling of conventional memory technologies is at risk. Memory technologies, such as SRAM
(Static Random Access Memory) and DRAM (Dynamic Random Access Memory), are experiencing scalability
challenges as a result to the limitations of their device cell size and power dissipation.
NEMs offer several benefits such as low power (especially low leakage), high density, and the ability to retain
the stored data over long time periods (non-volatility) that have made them attractive for use as secondary
storage. Flash memory is already widely used in consumer electronics and in solid-state disks due to its low cost
and extremely high density [2]
The dynamic and increasing power of DRAM over the power leakage of SRAM is a threat to circuit and
architecture designers of future memory hierarchy designs. Energy consumption has become key design limiters
as the memory hierarchy continues to contribute a significant fraction of overall system energy and power. The
lack of memory technology scaling can make it difficult for the memory hierarchy to achieve high capacity and
efficiency at low cost. As a result, it remains a very attractive technology for data archiving, with a sustainable
roadmap for the next ten to twenty years, well beyond the anticipated scaling limits of current conventional
technology [1].
There is a fundamental trend towards designing entire systems such that they are optimized for particular workloads, departing from the traditional general-purpose architecture. The typical system, with standard CPUs
consisting of a small number of identical cores with a common set of accelerators and relying on a memory and
storage hierarchy has reached its limits in terms of delivering competitive performance improvements for an
increasingly diverse set of workloads: future systems will be built out of increasingly heterogeneous components.
This article examines todays memory storage requirements, reviews recent research efforts on computer
architecture design with New Emerging Memories design.

www.theijes.com

The IJES

Page 61

Emerging Memory Technologies

II. NEW EMERGING MEMORY(NEM) TECHNOLOGIES
NEM technology is brand of computer hardware memory systems being developed with a view to either
extending the existing memory technologys capabilities or eventually replacing the whole technology all
together. Scalability beyond the incumbent is a critical requirement.
Most NEM technologies have resistive storage elements. Resistive RAM (RRAM or ReRAM) is a type of nonvolatile RAM that is highly promising in the next generation of memories for computers of the future. It works
by changing the resistance across a dielectric solid-state material often referred to as a Memristor.
2.1. STT-RAM
Spin-Transfer Torque Magnetic Random-Access Memory (STT-MRAM) is the latest design of the magnetic
RAM (MRAM). The information carrier of STT-MRAM is a Magnetic Tunnel Junction (MTJ) instead of
electric charges which makes the difference between it and the conventional RAM. See Fig 1

Fig. 1 A conceptual view of MTJ structure. (a) Anti-parallel (high resistance), which indicates 1 state; (b)
Parallel (low resistance), which indicates 0 state.
Each MTJ contains two ferromagnetic layers and one tunnel barrier layer. Figure 1 shows a conceptual
illustration of an MTJ. One of the ferromagnetic layer (reference layer) has fixed magnetic direction while the
other one (free layer) can change its magnetic direction by an external electromagnetic field or a spin-transfer
torque [5].
In case, the two ferromagnetic layers have different directions, the MTJ resistance is high, indicating a 1 state
(the anti-parallel case in Fig. 1 (a)); if the two layers have the same direction, the MTJ resistance is low,
indicating a 0 state (the parallel case in Fig. 1 (a)).
STT-MRAM changes the magnetic direction of the free layer by directly passing a spin-polarized current
through the MTJ structure. Comparing to the previous generation of MRAMs that uses external magnetic fields
to reverse the MTJ status, STT-MRAMs has the advantage of scalability, which means the threshold current to
make the status reversal will decrease as the size of the MTJ becomes smaller.
In the STT-MRAM memory cell design, the most popular structure is composed of one non metal oxide semiconductor (NMOS) transistor as the access controller and one MTJ as the storage element.
In Fig. 2, the storage element, MTJ, is connected in series with the NMOS transistor. The NMOS transistor is
controlled by the word-line (WL) signal. The detailed read and write operations for each MRAM cell is
described as follows:

Fig. 2 An illustration of an STT-MRAM cell with read/write circuitry.

www.theijes.com

The IJES

Page 62

Emerging Memory Technologies

There are two operations that happens which are:
i. Read Operations: This operation happens when the NMOS is turned on and a voltage is applied between
the bit-line (BL) and the source-line (SL). It will be observed that the voltage is negative and usually small
which will cause a current passing through the MTJ, but it is not small enough to invoke a disturbed write
operation. A sense amplifier compares this current with a reference current and then decides whether a 0
or a 1 is stored in the selected MRAM cell.
ii. Write Operation: This operation happens when a positive voltage difference is established between SL and
BL for writing for a 0 or a negative voltage difference is established for writing a 1. The current
amplitude required to ensure a successful status reversal is called threshold current. The current is related to
the material of the tunnel barrier layer, the writing pulse duration, and the MTJ geometry [5].
2.2. Phase Change Memory(PCM)
PCM is a type of non-volatile memory that exploits the property of chalcogenide glass to switch between two
states, amorphous and crystalline, with the application of heat using electrical pulses. The phase change material
can be switched from one phase to another reliably, quickly, and a large number of times. The amorphous phase
has low optical reflexivity and high electrical resistivity. Whereas, the crystalline phase (or phases) has high
reflexivity and low resistance [3].
For storage of information, PCM uses chalcogenide-based material. It has a wide range of resistivity, about three
orders of magnitude, and this forms the basis of data storage. The amorphous, high resistance state is used to
represent a bit 0, and the crystalline, low resistance state represents a bit 1. When germanium-antimonytellurium (GeSbTe or GST) is heated to a high temperature (normally over 600 C), it gets melted and its
chalcogenide crystallinity is lost.
Once cooled, it is frozen into an amorphous and its electrical resistance becomes high. This process is called
RESET. A way to achieve the crystalline state is by applying a lower constant-amplitude current pulse for a time
longer than the so-called SET pulse. This is called SET process. The time of phase transition is temperaturedependent [5].

Fig 3: illustration of a PCM cell

There are two operations that happens which are;
i. Read Operation: To effectively read data stored in the PCM cells across the GST, a small voltage is
applied. The read voltage to invoke detectable current must be strong but low enough to avoid write
disturbance. In Fig. 3, every basic cell contains one GST and one NMOS access transistor. This structure
has a name of 1T1R where T stands for the NMOS transistor and R stands for GST. The GST in each
PCM cell is linked to the drain-region of the NMOS in series so that the data stored in the cells can be
accessed [4].
ii.
a.
b.

Write Operation: There are two kinds of write operations,

The SET operation that switches the GST into crystalline phase when heated
The RESET operation that switches the GST into amorphous phase.

Note: Both operations are controlled by electrical current: high-power pulses for the RESET operation heat the
memory cell above the GST melting temperature; moderate power but longer duration pulses for the SET
operation heat the cell above the GST crystallization temperature but below the melting temperature. The
temperature is controlled by passing through a certain amount of electrical current and generating the required
heat.
www.theijes.com

The IJES

Page 63

Emerging Memory Technologies

2.3 Resistive Random Access Memory(ReRAM)
Any memory technology that represent digital information using its variable resistance is what ReRAM is meant
for. An insulating dielectric is conducted via conduction path by applying an adequate high voltage. The
conduction path can be generated by different mechanisms, including defects, metal migration, etc. The filament
may be reset (broken, resulting in high resistance) or set (reformed, resulting in lower resistance) by applying an
appropriate voltage [5].
ReRAM structure is a one cell i.e. a one metal oxide layer sandwiched by two metal electrodes - the top
electrode (TE) and the bottom electrode (BE), shown in figure 4. A low resistance state (LRS) represents digital
1 while a high resistance state (HRS) represents digital 0.

Fig.4: ReRAM structure

III. OTHER EMERGING MEMORY TECHNOLOGIES

3.1. Programmable Metallization Cell (PMC)
According to [7], PMC also known as the Electrochemical Metallization Memory Cells (ECM) rely on
electrochemical growth and dissolution of a conducting filament within the insulating layer. They consist of an
active Cu or Ag electrode, a cation conducting insulating layer, and an inert counter electrode. PMC cells exhibit
multibit data storage capability, scalability almost down to the atomic level, and very low programming power.
Moreover, potential backend of line compatible integration has been demonstrated making ECM cells of high
interest for future nonvolatile memory.
3.2. Polymer memory
Throughout the last few years, polymers have found growing interest as a result of the rise of a new class of
nonvolatile memories. In a polymer memory, a layer consists of molecules and/or nanoparticles in an organic
polymer matrix is sandwiched between an array of top and bottom electrodes as illustrated in Figure 5.
Moreover, polymer memory has the advantage of a simple fabrication process and good controllability of
materials [9]. Polymer memory could be called digital memory with the latest technology. It is not possible for a
silicon-based memory to be established in less space, but it is possible for polymer memory. The non
volatileness and other features are inbuilt at the molecular level and offers very high advantages in terms of cost.
But turning polymer memory into a commercial product would not be easy.

Fig 5: Structure of a polymer memory device.

www.theijes.com

The IJES

Page 64

Emerging Memory Technologies

3.3. Molecular Memory
A molecular memory is a nonvolatile data storage memory technology that uses molecular species as the data
storage element, rather than, e.g., circuits, magnetics, inorganic materials, or physical shapes. In a molecular
memory, a monolayer of molecules is sandwiched between a cross-point array of top and bottom electrodes as
shown in Fig 6. The molecules are packed in a highly ordered way, with one end of the molecule electrically
connected to the bottom electrode and the other end of the molecule connected to the top electrode, and this
molecular component is described as a molecular switch. Then, regarding the molecular memory operation, by
applying a voltage between the electrodes, the conductivity of the molecules is altered, enabling data to be stored
in a nonvolatile way. This process can then be reversed, and the data can be erased by applying a voltage to the
opposite polarity of the memory cell.

Fig 6: Cell structure of a molecular memory device

3.4. MNW
The molecular nanowire array (MNW) memory is fundamentally different from other semiconductor memories;
information storage is achieved through the channel of a nanowire transistor that is functionalized with redoxactive molecules rather than through manipulation of small amounts of charge. It is relatively slow and lacks the
random access capability, wherein data that can be randomly read and written at every byte are being actively
pursued. Figure 7 shows the schematic design of a MNW memory cell.

Fig 7: A MNW memory cell structure

3.5. QD (QUANTUM DOTS) Memory
Memory made from tiny islands of semiconductors - known as quantum dots - could fill a gap left by today's
computer memory, allowing storage that is fast as well as long lasting. Researchers have shown that they can
write information into quantum dot memory in just nanoseconds. New research shows that memory based on
quantum dots as shown in figure 8 can provide the best of both: long-term storage with write speeds nearly as
fast as DRAM. A tightly packed array of tiny islands, each around 15 nm across, could store 1 terabyte
(1,000 GB) of data per square inch, the researchers say. Dieter Bimberg and colleagues at the Technical
University of Berlin, Germany, with collaborators at Istanbul University, Turkey, demonstrated that it is possible
to write information to the quantum dots in just 6 ns. The key advantages of quantum dot (QD) are the high
read/write speed, small size, low operating voltage, and, most importantly, multibit storage per device.

Fig 8 Structure of quantum dot memory [8].

www.theijes.com

The IJES

Page 65

Emerging Memory Technologies

3.6. 1T1R-RRAM
One-transistor one-resistor (1T1R)-RRAM is also one class of emerging memory technology with impressive
characteristics. It meets the demands for next-generation memory systems. It is expected that 1T1R-RRAM
could be able to meet the demand of high-speed (e.g., performance) memory technology. The 1T1R structure is
chosen because the transistor isolates current to cells, which are selected from cells which do not. The basic cell
structure of 1T1R is depicted in Figure 9. 1T1R-RRAM consists of an access transistor and a resistor as a
storage element. [6] posited that the 1T1R cell structure is similar to that of a DRAM cell in that the data is
stored as the resistance of the resistor and the transistor serves as an access switch for reading and writing data.
Moreover, the 1T1R structure is more compact and may enable vertically stacking memory layers, ideally suited
for mass storage devices. But, in the absence of any transistor, the isolation must be provided by a selector
device, such as a diode, in series with the memory element, or by the memory element itself.

Fig 9: The basic cell structure of 1T1R-RRAM.

IV. COMPARISON BETWEEN THE DIFFERENT NEMS

A comparison between some of the different NEMs as against the main current memory hierarchy components
(SRAM, DRAM, disk and flash) is drawn
The following set of criteria is employed;
1. Maturity: Whether the technology is currently used in the market or it is in early or later search stages
before being commercially mature
2. Cell size: the cell size, using standard feature size
3. Read Latency: the speed for reading values from memory cell
4. Write Latency: the speed for writing values to a memory cell
5. Endurance: the number of write cycles that a memory cell endures before eventually wearing out
6. Energy: energy spent per bit access. Related to dynamic power
7. Static power: whether power needs to be spent while accessing the memory device. This includes
refreshing solid memory contents due to energy leakage or keeping disks spinning to achieve lower access
latencies.
8. Non-volatility: whether the memory technology is volatile or not.
Table 1 compares NEMs with traditional memory and storage technologies, in terms of performance, energy,
density, and endurance [5].
FEATURE SRAM DRAM DISK
FLASH STT-MRAM PCM
ReRAM
Product Product Product
Product Advanced
Advanced
Advanced
Maturity
Development Development Development
<10ns
10-60ns 8.5ms
25 s
<10ns
48ns
<10ns
Read
latency
<10ns
10-60ns 9.5ms
200s
12.5ns
40-150ns
10ns
Write
latency
2pJ
10010nJ
2pJ
100pJ
0.02pJ
Energy Per >1pJ
Bit Access
1.000mJ
High
Medium High
Low
Low
Low
Low
Leakage
Power
>1015
>1015
104
>1015
105 109
105 1011
Endurance >1015
No
No
Yes
Yes
Yes
Yes
Yes
Non
volatility
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Scalability
www.theijes.com

The IJES

Page 66

Emerging Memory Technologies

V. REPLACING TECHNOLOGIES WITH NEMS
Since NEMs are persistent and have similar mechanics (no moving parts) and lower energy consumption, its
natural to refer to them as persistent storage. NEMs would also perform as a disk replacement.
5.1. PCM as a Disk Replacement
According to [3], when considering PCM as a disk replacement, the following would be considered;
1. Advantage of solid-state implementation (non-moving parts), very low latencies, low cost per bit and
physical durability.
2. Poor write endurance (can be attacked through wear- leveling) and scaling down to a point where that can
compete with disks (using multi-layers, multi-bit cells, etc.)
3. The prices of NEMs such as PCM will be much higher than disks, but over time it tends to decrease
dramatically s market adoption progresses.
4. Its I/O are accessed as a block device.
5. Runtime error detection or correction mechanisms
6. Read or write should be asymmetrical, the read being as a single word.
Table 2 below is a characteristic of NEM replacement PCM
Characteristics of PCM
1 TB
Capacity
100 ns
Read or write access time
> 1 GB/s
Data rate
238 000 sio/s (SIO: start I/O)
Sustained I/O rate
975 mb/s
Sustained bandwidth
1012 write
Write endurance
5.2. PCM or STT-MRAM as Disk Replacement
To evaluate as a persistent storage, five evaluation criteria that should be taken into consideration;
1. Density: It is related with cost/GB which is the most important parameter. The cost/GB scales in proportion
to the density, which is cell size divided by the number of bits per cell.
2. Power efficiency important for mobile/embedded devices as well as for datacenters.
3. Access time: time interval between write/read request and the writing or availability of the data: Specially
important for datacenters
4. Endurance: the number of times a bit can be rewritten
5. Retention: the length of time a bit remains stable.
Table 3: Comparison of NEM Technologies for Disk Replacement
Device Type
DRAM
Flash
PCM
STT-MRAM
Maturity
Product
Product
Product
Product
Density
83gb/chip
64gb/chip
512mb/chip
2mb/chip
Cell size
6F2
4F2
5F2
4F2
MLC Capacity
No
4 bits/cell
4bits/cell
4biits/cell
Energy
2pJ
10pJ
100pJ
0.02pJ
Access time
10/10ms
200/25ns
100/20ns
10/10ns
Retention
1016/64ms
105/10yr
105/10yr
1016/10yr
From Table 3 above, the technologies with the best opportunity have a small cell size and the capability of
storing multiple bits per cell.
Phase Change Memory (PCM) and Spin Transfer Torque Magnetic Random Access Memory (STT-MRAM)
appear to meet these criteria. It is concluded that the PCM and STT-MRAM are the technologies with higher
probabilities of being a feasible replacement for disk in the future since they meet the criteria (small cell size and
capability of storing multiple bits per cells).

VI. CONCLUSION
Since the limitations of the traditional technologies threatens the sustainable growth of performance and energy
efficiency of computer systems, memory technology should be non-volatile, low-cost, high dense energy
efficient, fast and within high endurance.
www.theijes.com

The IJES

Page 67

Emerging Memory Technologies

Therefore, superior density, power, and non-volatility characteristics of emerging NEM technologies provide
opportunities to break this traditional system organization.
The problems in performance and strength as compared with traditional memory technologies is as a result of rearchitecting processor caches, main memory, and other storage system. By taking the full advantages of the
NEM technologies, computer performance can be highly enhanced.

REFERENCES
[1].
[2].

[3].
[4].

[5].

[6].
[7].
[8].
[9].

Chris H. Kim (2011): Modeling, Architecture, and Applications for Emerging Memory Technologies. Pp 44-45
Clinton W. Smullen, V , Vidyabhushan M, Anurag N, Sudhanva G, Mircea R. Stan.C (2011): Relaxing Non-Volatility for Fast and
Energy-Efficient STT-RAM Caches. HPCA11 Proceedings of the 2011 IEEE 17th International Symposium on High Performance
Computer Architecture Pp 50-61
Qureshi, M, Srinivasan,V and Rivers,J (2009): Scalable High Performance Main Memory System Using Phase-Change Memory
Technology, Proc. 36th Intl Symp. Computer Architecture (ISCA 09), ACM Press, pp. 24-33.
Xiaoxia Wu, Jian Li, Lixin Zhang, Evan Speight, Ram Rajamony, Yuan Xie (2009): Hybrid Cache Architecture with Disparate
Memory Technologies:ISCA proceedings of the 36th annual international symposium on computer architecture pages 34-45 ACM
New York, NY, USA
Zhao, J; S. Li, D. H. Yoon, Y. Xie, and N. P. Jouppi Kiln (2015): Closing the performance gap between systems with and without
persistence support. In Proceedings of the 46th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO13).
ACM, New York, NY, 421 432. DOI
Zangeneh M, Joshi A. Design and optimization of nonvolatile multibit 1T1R resistive RAM. IEEE Trans VLSI System.
2014;22(8):18151822.
Chen, A., Hutchby, J., Zhirnov, V., and Bourianoff, G. (2015). Emerging nanoelectronic devises, First Edition. John Wiley &
Sons, Ltd.
Manna S, Aluguri R, Katiyar A, Das S, Laha A, Osten H, Ray S. MBE-grown Si and Si1-xGex quantum dots embedded within
epitaxial Gd2O3 on substrate for floating gate memory device. Nanotechnology. 2013;24:505709
Prakash A, Ouyang J, Lin JL, Yang Y.(2006) Polymer memory device based on conjugated polymer and gold nanoparticles.
Journal of Applied Physics. 2006;100(5):054309054314.

www.theijes.com

The IJES

Page 68