TiO2-based Memristors and ReRAM:

Materials, Mechanisms and Models (a Review)

arXiv:1611.04456v1 [cond-mat.mtrl-sci] 9 Nov 2016

Ella Gale1,2∗
1. Unconventional Computing Group,
University of the West of England, Frenchay Campus,
Bristol, BS16 1QY
2 Bristol Robotics Laboratory, Bristol, BS16 1QY
November 15, 2016

Abstract
The memristor is the fundamental non-linear circuit element, with
uses in computing and computer memory. ReRAM (Resistive Random
Access Memory) is a resistive switching memory proposed as a non-
volatile memory. In this review we shall summarise the state of the
art for these closely-related fields, concentrating on titanium dioxide,
the well-utilised and archetypal material for both. We shall cover
material properties, switching mechanisms and models to demonstrate
what ReRAM and memristor scientists can learn from each other and
examine the outlook for these technologies.

1 Introduction
The semiconductor industry has, for many years, managed to increase the
number of components per chip according to Moore’s law (geometrical scal-
ing). Further advances have been achieved through functional diversification
(‘More-than-Moore’), namely complementary technologies that increase the
usefulness of electronic systems by adding extra functionality. According
to the International Technology Roadmap for Semiconductors, to continue
∗
Current Address: School of Experimental Psychology, University of Bristol, 12a Priory
Road, Bristol, BS8 1TU, UK. E-mail: [email protected]

1
at this rate the semiconductor industry needs to combine further miniatur-
isation with this type of functional diversification, as well as investigate a
future beyond CMOS [1]. There are several technologies that might fit such
a desire, two of which are memristors and Resistive Random Access Memory
(ReRAM, also known as RRAM).
In this short review, we shall cover the material properties, elucidated
mechanisms and current models in order to explain the state of these fast-
moving fields. ReRAM is usually based on transition metal oxides such as
TiO2 , SrTiO3 [2, 3], NiO [4], CuO [5], ZnO [6], MnOx [7], HfOx [8], Ta2 O5 [9],
Ti2 O5−x /TiOy [10], TaOx /TiO2−x [11]; both binary and perovskite oxides are
capable of resistance switching. Memristors can be made out of TiO2 , chalco-
genides [12], polymers [13, 14], atomic switches [15], spintronic systems [16]
and quantum systems [17]. Biological material and mechanisms like sweat
ducts [18], leaves [19, 20], blood [21], slime mould [22, 23], synapses [24, 25]
and neurons [26] can be described as memristive. In this review we shall
restrict our focus to TiO2 -based memristors and ReRAM because TiO2 is
the material that has received most work in memristors and is a well-studied
and archetypal system within ReRAM [27]. We shall start by going through
a brief the history of the fields, then the materials and mechanisms before
looking at the models in more detail.
The resistor, capacitor, and inductor are the three well-known fundamen-
tal circuit elements, discovered in 1745, 1827 and 1831 respectively. Based on
an assumption of completeness (see figure 1), in 1971 Leon Chua postulated a
4th fundamental circuit element [28] which would relate charge q to magnetic
flux ϕ and would have the distinction of being the first non-linear circuit
element (where the non-linearity arises because q and ϕ are integrals of the
circuit measurables, current, I and voltage, V ). No physical instantiations
of the memristor were generally acknowledged until 2008 when Strukov et al.
realised that their molecular electronic switches’ behaviour was due to the
titanium electrodes and not the organic layer [29] and announced that they
had found the memristor [30]. At that time, it was believed that memristors
could not have been made contemporary with the other fundamental cir-
cuit elements as memristance was a nanoscale phenomenon [30], ‘essentially
unobservable at the millimetre scale’ and above [29] (since experimentally
disproved by the fabrication of macroscopic memristors [31, 32, 33, 34]). It
later became apparent that memristor-like devices had been fabricated be-
fore 2008: resistance switching was first observed in oxides in a gold-silicon
monooxide-gold sandwich in 1963 [35], the first metal oxide resistance switch
was reported in nickel oxide in 1964 [36], memristor-like switching curves were
observed in TiO2 thin films in 1968 [37] and the memistor [38], a 3-terminal
memristive system, was fabricated in 1960 (although is disputed as being an

2
 φ

L M ∫

I R V

∫ C
q
Figure 1: The four circuit measurables and the six relationships between
them.

example of a memristive system [39]). These examples inspired some searches

for the earliest known memristor [40, 41] with the singing arc [42] in 1880
and the coherer in 1899 [43] being the oldest devices identified so far.
When the memristor was announced the field of ReRAM was already
mature: two contemporary reviews [44, 45] summarise the state of the art.
ReRAM is generally made of a metal-insulator-metal structure, typically
using transition metal oxides as the insulator, while the electrodes tend to
be a noble metal, with Pt being a popular choice (but Al, Au, ITO (Indium
Tin Oxide) and Si are all also common). ReRAM has two resistance states:
a high resistance state (HRS) and low resistance state (LRS). It also has
two switching modes that go between them. The simplest to understand is
bipolar resistance switching (BRS) where the device switches depending on
the magnitude and direction of the applied voltage. The other is unipolar
resistance switching (UPS) which depends only on the magnitude of the
voltage. Both BPS and UPS V -I curves resemble memristor curves, a fact
which has led to much debate in scientific literature. In this review, we shall
compare and contrast the approaches of both fields, using titanium dioxide
devices as examples.

2 Materials
One reason for the confusion between the fields of memristors and ReRAM is
the different standard measurements. The term ReRAM refers to a suggested
use for a range of materials (namely as memory), whereas memristors are

3
named after a property and have been investigated not just for RAM, but
also for such uses as neuromorphic hardware (e.g. [46, 25, 47, 48, 49, 17, 50]),
chaotic circuits (e.g. [51, 52, 53, 54]), processors (e.g. [55, 56, 57, 58, 59, 60]),
vision processors (e.g. [25, 61, 62, 63, 64]) and robot control (e.g. [65, 66, 50])
to name a few. Memristor measurements tend to be a.c. measurements, often
done at different frequencies. This is partly because initially the memristor
was thought to be a purely a.c. device and only recently have the d.c. aspects
of the device received attention [48, 67, 68, 69]. ReRAM measurements are
reported in different ways depending on the type of device. BPS devices are
measured in a similar way to memristors; but UPS devices are taken to a high
voltage with a compliance current set, which is the method of forming the
device. After forming, the device is then taken around a positive voltage loop
from low to high voltage (where the high voltage is usually below the forming
voltage). As UPS ReRAM is not subjected to bipolar voltage waveforms, it
is difficult to compare them to memristors.
The first device called a memristor was made from TiO2 thin film [30],
but the first TiO2 thin film device that demonstrated memristor curves was
fabricated in 1963 [37], this work showed the film pass through a family of
hysteretic curves during repeated tests: from pinched open curves (resem-
bling a jelly-bean in shape), to BPS curves before collapsing to a single line
as the device degraded. Resistance switching in TiO2 materials had been
seen several times in ReRAM experiments before 2008, for example: sputter-
grown TiO2 with Pt electrodes [70] and Pt-TiO2 -Ru stacks [71]. But it was
not until Hewlett-Packard’s announcement that people became interested in
making memristors, and so the report in 2009 [72] of a flexible Al-TiO2 -
Al memristor fabricated using solution processing garnered a lot of interest
(a flexible ReRAM equivalent came out the same year [73]). The use of
aluminium electrodes however, did suggest that aluminium oxide might be
involved in the switching: XPS and EELS has shown the existence of a layer
of Al2 O3 in Al-TiO2 -Al ReRAM BPS devices [74], Al2 O3 had been added
to organic memory devices to improve switching [75] and been implicated in
resistance switching [76, 77, 78, 79]. There is evidence for Al being support-
ive in TiO2 resistive switches grown by atomic deposition [71] and there is
even intriguing evidence for a mixed phase [80]. In [72] the authors state
that the devices still switch if the electrode materials are changed to a no-
ble metal (Au) and [81] showed that changing the device electrode changes
the form and reversibility of the switching. Other work has concentrated on
making cheaper and easier to manufacture memristor devices [34, 82, 83],
but presently the memristor is a difficult device to manufacture for the pur-
poses of experimental testing, which has led to the field being centred around
simulation and theoretical work.

4
3 Mechanisms
Many suggested mechanisms have been put forward to explain the causes
of resistance switching and of memristance, and TiO2 -based devices have
examples of all of them. The high resistance material is generally believed to
be stoichiometric TiO2 [30], but there is debate about what the low resistance
material is, and, given that the memristors/ReRAM are made in several
different ways there is no expectation that there is a single explanation or
material cause.
The most popular suggested mechanisms can be split into two groups:
ionic and thermal.

3.1 Ionic
The ionic mechanism in TiO2 devices involve the migration of oxygen vacan-
cies (although suggestions of migrating OH− ions have been made [77, 84]).
This movement of vacancies creates auto-doped phases which are metallically
conducting for TiO(2−x) for x > 1.5 [85]. The oxygen ions may combine at
the anode and evolve O2 gas, this has been seen in ReRAM [86], the Strukov
memristor when fabricated with macroscale electrodes [87] and the sol-gel
memristor [81]. There is some evidence that the choice of electrode material
can hinder the production of O2 , for example, it is thought aluminium sup-
ports TiO2 memristance by acting as a source or sink of oxygen ions [74, 88].
Most memristor-based modelling has concentrated on modelling the flow of
these oxygen vacancies (see later). [89] suggests that switching is dominated
by ionic motion rather than charge trapping. A mixed mechanism involv-
ing oxygen vacancy transport between the tip of a conducting filament (itself
formed by thermal mechanism) and an electrode has been suggested [90]. Al-
ternative electrochemical mechanisms that explain memristance via a change
in titanium oxidation state, Ti(IV)→Ti(III), have been put forward [84] al-
though not widely adopted.

3.2 Thermal
Joule heating is the process where the application of an electric field and flow-
ing current heats the material and changes its structure. TiO2 atomic deposi-
tion thin films can form conducting filaments as extended defects along grain
boundaries [91], the ions drift, forming a path which breaks with excess heat
(i.e. too high a voltage) and can be reformed via the same mechanism. This
mechanism has been credited with causing both HRS and LRS states [71].
Single crystals of SrTiO3 show switching in the skin after being subjected to

5
an electric field under ultra-high vacuum. This causes ‘dislocations’ which
are broken by ambient oxygen and then reformed via electroforming. This
is present in a crystalline structure and suggests a poly-filamentary mecha-
nism [86]. The conduction channels are usually ohmic and it has been shown
that the resistance and reset current bears little relation to the composition
of the material or whether it is operated in a unipolar or bipolar mode, which
suggests, and has been verified by simulation, that the mechanism is ther-
mally activated dissolution of conducting filaments [92]. This could explain
why compliance current choice controls the resistance and reset current and
suggests that larger resistance values are more related with the heat sink
effects of the metal oxide, rather than its electronic properties.
The form of these conducting filaments / channels is debated. In 2009 A
Pt-TiO2 -Pt ReRAM showed a Magnéli phase [88], Tin O2n−1 , crystallographic
shear plane from Rutile [93], roughly 10-20nm in diameter and which shows
metallic (i.e. ohmic) conduction. Magnéli phases are considered likely to be
the material cause as they are more thermodynamically stable than vacan-
cies spread through-out the material [94]. Single crystal TiO2 nanorods have
shown bipolar switching [95] which have a rectifying effect. Conical Magnéli
phase filaments have been observed and shown to be the cause of fusing and
anti-fusing BPS [96]. Hourglass-shaped Magnéli phases were seen in [88],
which were formed by alternating polarity voltage and proved to be more
stable in operation. Fractal conducting filaments, have been suggested by
simulation [97], observed [98] and are thought to relate to dielectric break-
down [99]. From spectromicroscopy and TEM, electroforming the Strukov
memristor has been found to produce an ordered Ti4 O7 Magnéli phase [100].

3.3 Other mechanisms

Other mechanisms have been put forward. The UPS switching has cred-
ited to: metal-semiconductor transitions [101], crystalline TiO2 -amorphous
TiO2 phase transition via conduction heating and breaking [71], raising and
lowering Schottky barriers via bulk (UPS) or interface (BPS) transport of
oxygen [2], and conductance heating causing lateral transport of conduct-
ing filaments [92]. There is also a debate as to what the structure of the
less-conducting thin-film TiO2 is, with rutile (r-TiO2 ) [71] and amorphous
titanium dixoide (a-TiO2 ) being the most popular suggestions. Note, that
a-TiO2 has also been suggested as the conducting form of TiO2 and that
Magnéli phases are sheer planes from r-TiO2 .
Currently, the Magnéli conducting filaments are the most generally ac-
cepted mechanism, but as all the materials are treated differently, measured
differently, and titanium dioxide has a wealth of behaviours, it is unlikely

6
that there will emerge only one explanation for all published TiO2 devices.

4 Models
Generally, ReRAM modelling has concentrated on models based on the mate-
rials science and chemistry of the devices (and are described above), whereas
the memristor, by virtue of being predicted from electronic engineering the-
ory tends to include more physics- and mathematics-based approaches.

4.1 Models based on materials science

Chua’s 1971 paper [28] made the observation that there were only five re-
lations between the four circuit measurables: charge, q, magnetic flux, ϕ,
current, I and voltage, V . Two relations were the definitions of charge and
flux (as I = dq/dt and V = dϕ/dt), the other three were the constitu-
tive definitions of the resistor (V = IR), capacitor (q = CV ) and inductor
(ϕ = LI). From these facts, the missing 6th relationship would be ϕ = M q,
see figure 1. When Strukov et al. announced their discovery of a memris-
tor [30] they included a simple model based on treating the memristor as two
space-conserving variable resistors in which they considered the velocity of
the boundary w between TiO2 and TiO(2−x) phases. This model has been
both highly criticised and widely used, both as is and with extensions.
The model as reported [30] did not include the magnetic flux which is ex-
pected from Chua’s definition [28]. This lead to questions, including the ques-
tion of whether a memristor had actually been fabricated [102]. Attempted
solutions to this issue have included fixes to the model [103], statements that
the magnetic flux was a theoretical construct [30, 104] and not related to a
material property [40] or that it was the non-linear relation between I and V
that defined the device [30]. The Strukov model as presented assumes linear
dopant drift under a uniform field [30], which is widely believed to be unlikely
in devices with high electronic field [105]. Several people have undertaken to
fix this: [106] introduced the concept of a rectangular ‘window’ function that
would prevent w from going out of its bounds (0 and D – the thickness of the
semiconductor layer) and slowed the rate of motion down near the bound-
ary edges (making it non-linear), [107] improved on this to prevent w from
getting stuck at the boundaries and [108] adjusted it to allow the maximum
to be tuned to less than 1, thereby introducing more variability in device
modelling, [109] concentrated on applying the models to nanostructures and
[110] concentrated on neuromorphic systems. A further confusion arises from
the widely-interpreted implication that the uniform field is across the entire

7
device suggested by the wording in [30], which is not what was intended [111].
A field with a discontinuity is problematic from the electrochemical point of
view [112], especially as the discontinuity is located on w [113], and may
require further work with windowing functions.
Despite these issues, the Strukov model has been highly adopted, and
most work involves simulating with it or improving it with window func-
tions. An in-depth discussion of the uses of it is beyond the scope of this re-
view, but it has generally been used to model test circuits, often using SPICE
(Simulation Program for Integrated Circuits Emphasis, a standard electronic
engineering simulation package) and comparisons of different SPICE imple-
mentations of the models described in this section are given in [114, 115].
Three models have extended it, [116] demonstrated a SPICE model with
the initial state as a variable, Georgiou et al. extended the model using
Bernoulli formulation, introducing extra parameter to get a measure of the
device hysteresis [117] and [118] extended it by demonstrating a SPICE
model of a magnetic flux-controlled memristor.
Other models have been proposed for use in modelling TiO2 devices based
on more realistic models of the materials. A model based on the idea of set-
ting w to the tunnelling barrier length between TiO(2−x) and the electrode
was derived by experimentally fitting to the Strukov memristor after electro-
forming had created a localised conducting channel [119]. This model gives
good agreement with experimental data [119], but has the drawbacks of being
difficult to simulate (containing a hyperbolic sine term and two exponential
terms) and uses 8 fitting parameters for which the experimental analogues
are not completely clear (although they are related to nonlinearities due to
the high field and Joule heating [120]). The data in [119] has been used
to create other models: [121] presented a version implemented in SPICE;
[120] presented a more general model with a threshold that was easier to
simulate; [122] presented a tunnelling model that required only one fitting
parameter, which has also been implemented in SPICE [123]; [124] presented
a tunnelling model that included a threshold relation and was applied to
both memristors and ReRAM. A good model requires that ‘any parameters
in the model determined by fitting to experimental data should be intrinsic
to the device’ [125]. There is a trade-off between the quality of the fit, which
can require many fitting coefficients; and the relation to physical processes,
which requires that each fitting coefficient is strongly related to a real-world
property.
The trend in building memristor device models is currently towards the
more general (in that they can model more devices and types of devices),
experimentally-informed (in that the model can be understood in terms of
the physical processes happening in the device), easy-to-simulate models.

8
4.2 Models based on electronic engineering
As Strukov’s model is phenomenological, so the derivation of Chua’s model
was more rigorous and mathematical, nonetheless, the theoretical idea of a
memristor has not escaped the influence of real devices. Before a memristor
was identified, the concept of a memristor was expanded to that of a mem-
ristive system [126], which has two state variables, this concept was used to
describe a thermistor and neuron ion channels. Since the discovery of a mem-
ristor device, the terminology has changed to refer to memristive systems as
a type of memristor, rather than vice versa. In 2011, Chua published a pa-
per [40] in which he asserted that all ReRAM devices with pinched hysteresis
loops are memristors. In [40] Chua suggested that experimentalists should
record q-ϕ data because V -I curves are open to effects of input voltage (a
point well illustrated in [117], which offers an attempted theoretical solution)
and are not predictive. The focus shifted in [127] where the frequency depen-
dence (which is mentioned in the 1971 paper [28]) is elevated to being two of
the three identifying ‘fingerprints’ for experimentalists to search for (namely
that the hysteresis loop shrinks as the frequency tends to infinity). Further
expansions of the memristor idea have been undertaken. The memristor is
a passive device, but the idea of an active memristor has been shown to be
useful [54, 23, 51]. The pinched hysteresis loop, the most distinctive property
of the memristor, has been relaxed to a non-self-crossing pinched hysteresis
loop [128, 129, 18] (matching experimental data for memristors [37, 130] and
for ReRAM [131]). These changes have led to some criticism [132] that the-
oretical ideas cannot be redefined at whim to fit experimental data. This is
an interesting point to ponder. A theory that has no relevance to real world
systems is of little practical use, and adjusting the presentation and switch-
ing which properties are necessary to define device behaviour and which are
merely indicative in response to a larger group of discovered devices seems, to
me at least, to be entirely appropriate for theoretical science. Nonetheless,
none the changes in the definition of the memristor put forward by Chua
have changed the basic concept, that of a non-linear resistor which relates
the time integral of the current to the time integral of the voltage and pro-
duces pinched (although now not necessarily self-crossing) V -I curves which
shrink with increasing frequency.

5 Memristors and ReRAM - one field or two?

Having presented an overview of both fields it is worth asking if they are
one field or two. Initially the two communities ignored each other and

9
thus several memristor papers published early on presented phenomena al-
ready discovered within ReRAM (as described earlier) and, even now, some
ReRAM papers fail to reference memristor work [133]. Generally, opinion
has shifted to ReRAM scientists using and extending [131] memristor the-
ory, while memristor scientists look to ReRAM to explain device proper-
ties [134, 135], suggesting that even if they are not the same field there is at
least convergence between them. Some researchers are even using the terms
interchangeably [136] and some are taking memristors as just another type
of technology with which to build ReRAM [137].
Questions highlighting the differences remain however: The memristor
is defined as a nonlinear, analogue device; so devices that undergo ohmic
conduction (as seen in conductive filaments) are not strictly memristors, al-
though it is possible to model them by including conducting channels in the
device model [138]. However, the memristor equations at certain frequencies
do give more triangular shaped curves [30] which can match the observed
shape ( Figure 2d.). To match memristor theory (Figure 2c) to real devices,
different extensions to the theory, such as non-zero crossing [131], conduct-
ing filaments [138], active memristors [54] and analysis of use cases such as
reading and writing operations of memory [136, 139] are required. However,
memristor theory is so elegant and useful (for example, it neatly explains the
behaviour of I-V loops shrinking to a single valued function that has been
observed in ReRAM [37, 71]), it seems worth the effort to add real-world
considerations.
The question of whether the work belongs in one or two fields is far from a
settled one. It seems, however, that BPS ReRAM (Figure 2b) is a real-world
instantiation of the memristor and whether UPS ReRAM (Figure 2a) is con-
sidered a memristor depends which formulation of the memristor definition
you adopt: whilst in the ohmic conducting regime UPS ReRAM does not
satisfy the 1971 definition [28] of the memristor, because the memristance is
not changing with q. The system can, however, be described as a memristive
system as per the 1976 definition [126] with the state of conducting filaments
(i.e. fused or connecting) as a second state variable. Under the nomenclature
put forward in [40] these devices can thus be considered memristors.

6 The Future
The memristor has only been understood as an existent device for five and
half years and in that time it has drastically effected the outlook for ReRAM
devices by providing a deeper theoretical understanding of aspects of their
operation and suggesting uses for these devices other than merely as a novel

10
 on

on
Current
cc
Current

off
off off
off
cc

on
on

Voltage Voltage
(a) Typical unipolar ReRAM (b) Typical bipolar ReRAM
on

on
Current
Current

off off
off
off
on
on

Voltage Voltage
(d) An Example Memristor
(c) Theoretical Memristor V-I
curve [81]

Figure 2: Predicted and typical ReRAM and Memristor curves.Note that

the example from [81] switches from on to off, many memristors switch the
opposite way round, and would go around the I-V curve in the opposite
direction

11
type of storage. In turn, the rich history and vast amount of work on the
materials, their properties and how they can be controlled that comes from
ReRAM scientists have allowed the memristor concept to move from inter-
esting phenomenon to practical device far quicker than would otherwise have
happened. I suspect that the trend for ReRAM and memristors to be in-
creasingly be viewed as describing the same phenomena will continue.
Currently, work is going in several directions. The first computer con-
taining ReRAM memory (based on TaOx ) was announced by Panasonic in
2013, but many people and a few companies, are racing to produce their
competing commercial memristor/ReRAM memory, be it computer RAM,
flash drives or other form of storage, and so there is a lot of work on stabil-
ising and controlling the device’s properties and looking for other materials
for manufacturing.
The first commercial suggestion for memristors was memory because the
memristor’s small feature size could offer an increase in memory density,
wouldn’t require power to hold a value (so it is low energy electronics) and
could be used for multi-state memory. Alongside the usual manufacturing
faults, memristors suffer some unique issues on which there has been much
work and many attempted fixes, both based on memristor theory [139, 136]
and experiment [140]. For example, according to the theory, memristor state
will change when read, affecting data integrity, this can be solved by us-
ing a read pulse below a threashold time [139] or a reading algorithm that
reads and then rewrites the memory cells [141] (this is also useful for reduc-
ing sneak-path-related read errors). Other work has focused on improving
memory tests to identify undefined states [136] and fabrication errors [142]
(by making use of sneak paths to test several memristors at the same time).
A big problem for memristor-based cross-bar memory is sneak paths which
can result in erroneous memory values being read and are a big problem
for shrinking memory size. Attempted solutions include: Hewlett-Packard’s
reading algorithm [141] described above; unfolding the memory (only one
memristor per column/row, which greatly reduces the possible gains from
shrinking memory size; adding an active element like a diode [143], which
might add delay, or a transistor gate [137], which would remove the ad-
vantage of the memristor’s small size; using anti-series memristors as the
memory element so that the total resistance is always Ron + Roff [144], this
would require differentiating between Ron − Roff and Ron − Roff for logical
values; using multiple access points [145]; using a.c. sensing [146] at the cost
of increased complexity; making use of the memristors own non-linearity
(instead of a diode) [11, 140]; making use of a 3-terminal memristor [147]
(instead of a transistor) which forces sneak paths to have a higher than Roff
resistance (which is the best case failure for sneak paths) but requires an

12
additional column line. The authors of [147] concentrated on the memis-
tor [38] as an example device, but intriguingly the idea could be applied to
the 3-terminal plastic memristor [148, 149]. For memristor-based memory to
be adopted, there is much further work to be done in this area, especially on
experimentally testing these approaches.
Another big area is the design and fabrication of neuromorphic com-
puters using memristors, which will require novel approaches and hardware
instantiations. Because memristors can implement IMPLY logic, Bertrand
Russell’s logical system [150] is getting something of a renaissance. If mem-
ristors natively implement IMPLY logic [69], perhaps as spikes [151, 56],
then there will be a lot of work on engineering memristors into current cir-
cuit design approaches (which traditionally use AND, OR, NOT, NAND
and similar). Several workers have suggested doing digitised stateful logic in
CMOS-compatible cross-bar memory arrays (using the IMPLY and FALSE
operation set) and started to look at design methodologies: as in [118], mak-
ing encoders/decoders out of transistor-memristor arrays [152] and making
use of the parallelisability of the architecture for simultaneous bitwise vec-
tor operations [153]. Recently, several researchers have started to look at
hysteresis [154, 155, 117, 156] and other figures of merit [157, 158, 108] of
practical use to engineers.
An interesting and unexpected outcome of this work is the discovery that
evolution has made use of memristors and memristive mechanisms, where
living memristors have been identified in leaves, skin, blood, and eukaryotic
mould so far, and memristor theory has been used to understand learning
in the synapses and simple organisms and to update neuron models such
the Hodgkin-Huxley [26]. This could suggest that electrophysiology is a field
ripe for memristor research and perhaps the most ground-breaking memristor
work will come from linking the manufactured devices with living memristors,
either directly or via the creation of bio-inspired computers which may usher
in an entirely new paradigm of computational approaches.

Acknowledgements
This work was supported by EPSRC on grant EP/HO14381/1. The author
would like to thank Oliver Matthews and Ben de Lacy Costello for useful
discussions.

13
References
References
[1] International technology roadmap for semiconductors 2012 update
overview. Technical report, International Technology Roadmap for
Semiconductors 2012, 2012.

[2] M.H. Tang, Z.P. Wang, J.C. Li, Z.Q. Zeng, X.L. Xu, G.Y. Wang, L.B.
Zhang, Y.G. Xiao, S.B. Yang, B. Jiang, and J. He. Bipolar and unipolar
resistive switching behaviors of sol-gel-derived SrTiO3 thin films with
different compliance currents. Semiconductor Science and Technology,
26:075019 (4pp), 2011.

[3] Xianwen Sun, Guoqiang Li, Xin’an Zhang, Linghong Ding, and Weifeng
Zhang. Coexistence of the bipolar and unipolar resistive switching
behaviours in Au/StTiO3/Pt cells. Journal of Physics D: Applied
Physics, 44:125404 (5pp), 2011.

[4] L. Goux, J.G. Lisoni, M. Jruczak, D.J. Wouters, L. Courtade, and Ch.
Muller. Coexistence of bipolar and unipolar resistive-switching modes
in NiO cells made by thermal oxidation of Ni layers. Journal of Applied
Physics, 107:024512, 2010.

[5] R. Yasuhara, K. Fujiwara, K. Horiba, H. Kumigashira, and M. Kotsugi.

Inhomogeneous chemical states in resistance-switching devices with a
planar-type Pt/CuO/Pt structure. Applied Physics Letters, 95:012110,
2009.

[6] Wen-Yuan Chang, Yen-Chao Lai, Tai-Bor Wu, Sea-Fue Wang, Freder-
ick Chen, and Ming-Jinn Tsai. Unipolar resistive switching character-
istics of ZnO thin films for nonvolatile memory applications. Applied
Physics Letters, 92:022110, 2008.

[7] Sen Zhang, Shibing Long, Weihua Guan, Qi Liu, Qin Wang, and Ming
Liu. Resistive switching characteristics of MnOx-based ReRAM. Jour-
nal of Physics D: Applied Physics, 42:055112, 2009.

[8] C.H. Lien, Y.S. Chen, H.Y. Lee, P.S. Chen, F.T. Chen, and M.-J. Tsai.
The highly scalable and reliable hafnium oxide ReRAM and its future
challenges. In Proceedings of 2010 10th IEEE International Conference
on Solid-State and Integrated Circuit Technology (ICSICT), pages 1084
– 1087, 2010.

14
 [9] Feng Miao, John Paul Strachan, J. Joshua Yang, Min-Xian Zhang,
Ilan Goldfarb, Antonio C. Torrezan, Peter Eschbach, Ronald D. Kel-
ley, Gilberto Medeiros-Ribeiro, and R. Stanley Williams. Anatomy
of a nanoscale conduction channel reveals the mechanism of a high-
performance memristor. Advanced Materials, 23(47):5633–5640, 2011.

[10] Yuchao Yang, Patrick Sheridan, and Wei Lu. Complementary resistive
switching in tantalum oxide-based resistive memory devices. Applied
Physics Letters, 100:203112 (4pp), 2012.

[11] J. J. Yang, M.X. Zhang, M.D. Pickett, Feng Miao, J.P. Strachan,
W.D. Li, W. Yi an D. A.A. Ohlberg, B.J. Choi, W. Wu, J.H.Nickel,
G. Medeiros-Ribeiro, and R.S. Williams. Engineering nonlinearity into
memristors for passive crossbar applications. Applied Physics Letters,
100:113501 (4pp), 2012.

[12] A.S. Oblea, A. Timilsina, D. Moore, and K.A. Campbell. Silver chalco-
genide based memristor devices. In Neural Networks (IJCNN), The
2010 International Joint Conference on, pages 1–3, July 2010.

[13] Francesca Pincella, Paolo Camorani, and Victor Erokhin. Electri-

cal properties of an organic memristive system. Applied Physics A,
104:1039–1046, 2011.

[14] Hu Young Jeong, Jong Yun Kim, Jeong Won Kim, Jin Ok Hwang,
Ji-Eun Kim, Jeong Yong Lee, Tae Hyun Yoon, Byung Jin Cho,
Sang Ouk Kim, Rodney S. Ruoff, and Sung-Yool Choi. Graphene oxide
thin films for flexible nonvolatile memory applications. Nano Letters,
10(11):4381–4386, 2010.

[15] Tsuyoshi Hasegawa, Alpana Nayak, Takeo Ohno, Kazuya Terabe,

Tohru Tsuruoka, James K. Gimzewski, and Masakazu Aono. Mem-
ristive operations demonstrated by gap-type atomic switches. Appl.
Phys. A, 102:811–815, 2011.

[16] Y. V. Pershin and M. Di Ventra. Spin memristive systems: Spin mem-

ory effects in semiconductor spintronics. Phys. Rev. B, 78:113309–1 –
113309–4, 2008.

[17] Yuriy V. Pershin and Massimiliano Di Ventra. Neuromorphic, digital

and quantum computation with memory circuit elements. Proceedings
of the IEEE, 100:2071–2081, June 2012.

15
[18] C.A. Lütken S. Grimes and O.G. Martinsen. Memristive properties of
human sweat ducts. World Congress on Medical Physics and Biomedial
Engineering, 25/7:696–698, 2009.

[19] Shiv Prasad Kosta, Y.P. Kosta, Avinash Gaur, Yogesh M. Dube, J. P.
Chuadhari, Jignesh Patoliya, Shakti Kosta, Paresh Panchal, Piyush
Vaghela, Kalpesh Patel, Bhavin Patel, Ritesh Bhatt, and Vimal Patel.
New vistas of electronics towards biological (biomass) sensors. Inter-
national Journal of Academic Research, pages 511–526, 2011.

[20] Alexander G. Volkov, Clayton Tucket, Jada Reedus, Maya I. Volkova,

Vladislav S. Markin, and Leon Chua. Memristors in plants. Plant
Signalling and Behavior, 9:e28152 (8pp), 2014.

[21] S.P. Kosta, Y.P. Kosta, Mukta Bhatele, Y.M. Dubey, Avinash Gaur,
Shakti Kosta, Jyoti Gupta, Amit Patel, and Bhavin Patel. Human
blood liquid memristor. International Journal of Medical Engineering
and Informatics, pages 16–29, 2011.

[22] Yurij V. Pershin, Steven La Fontaine, and Massimiliano di Ventra.

Memristive model of amoeba’s learning. Phys. Rev. E, 80:021926 (6
pages), 2009.

[23] Ella Gale, Andrew Adamatzky, and Ben de Lacy Costello. Slime mould
memristors. arXiv, June:1306.3414v1, 2013.

[24] Bernabé Linares-Barranco and Teresa Serrano-Gotarredona. Memris-

tance can explain spike-time dependent plasticity in neural synapses.
Nature Precedings, 2009.

[25] Carlos Zamarreno-Ramos, Luis A. Carmu nas, Jose A. Pérez-Carrasco,

Timotheée Masquelier, Teresa Serrano-Gotarredona, and Bernabé
Linares-Barranco. On spike-timing dependent plasticity, memristive
devices and building a self-learning visual cortex. Frontiers in Neuor-
morphic engineering, 5:26(1)–26(20), 2011.

[26] Leon Chua, Valery Sbitnev, and Hyongsuk Kim. Hodgkin-Huxley axon
is made of memristors. International Journal of Bifurcation and Chaos,
22:1230011 (48pp), 2012.

[27] K. Szot, M. Rogala, W. Speier, Z. Klusek, A. Besmehn, and R. Waser.

TiO2 - a prototypical memristive material. Nanotechnology, 22:254001–
1 – 254001–21, June 2011.

16
[28] L. O. Chua. Memristor - the missing circuit element. IEEE Trans.
Circuit Theory, 18:507–519, 1971.

[29] R. Williams. How we found the missing memristor. Spectrum, IEEE,

45(12):28–35, Dec 2008.

[30] D. B. Strukov, G. S. Snider, D. R. Stewart, and R. S. Williams. The

missing memristor found. Nature, 453:80–83, 2008.

[31] W. Alan Doolittle, W.L. Calley, and W. Henderson. Complemen-

tary oxide memristor technology facilitating both inhibitory and ex-
citatory synapses for potential neuromorphic computing applications.
In Semiconductor Device Research Symposium, 2009. ISDRS ’09. In-
ternational, pages 1–2, Dec 2009.

[32] D. J. Kim and Z. Fisk. A kondo insulating memristor. Applied Physics

Letters, 101(1):–, 2012.

[33] H. Wu, K. Cai, J. Zhou, B. Li, and L. Li. Unipolar memristive switch-
ing in bulk negative temperature coefficient thermosensitive ceramics.
PLoS ONE, 8:e79832, 2013.

[34] Ella Gale, Richard Mayne, Andrew Adamatzky, and Ben

de Lacy Costello. Drop-coated titanium dioxide memristors. Materials
Chemistry and Physics, 143:524–529, January 2014.

[35] N. M. Bashara P. H. Nielsen. The reversible voltage-induced initial

resistance in the negative resistance sandwich structure. IEEE Trans-
actions on Electron Devices, 11:243–244, May 1964.

[36] W.E. Beadle. Switching properties of thin NiO films. Solid State Elec-
tronics, 7:785–797, 1964.

[37] F. Argall. Switching phenomena in titanium oxide thin films. Solid

State Electronics, 11:535–541, 1968.

[38] B. Widrow. An adaptive ‘adaline’ neuron using chemical ‘memistors’.

Technical report, Stanford University, 1960.

[39] Shyam Prasas Adhikari and Hyongsuk Kim. Memristor Networks,

chapter Why are Memristor and Memistor Different Devices? Springer
International Publishing Switzerland, 2014.

[40] Leon Chua. Resistance switching memories are memristors. Applied

Physics A: Materials Science & Processing, pages 765–782, 2011.

17
[41] Christofer Toumazou Themistoklis Prodromakis and Leon Chua. Two
centuries of memristors. Nature Materials, 11:478–481, 2012.

[42] Jean-Marc Ginoux and Bruno Rossetto. Chaos, CNN, Memristors and
Beyond, chapter The Singing Arc: The Oldest Memristor? World
Scientific, 2013.

[43] Gaurav Gandhi, Varun Aggarwal, and Leon O. Chua. Memristor Net-
works, chapter The Detectors Used in the First Radios Were Memris-
tors. Springer International Publishing Switzerland, 2014.

[44] Akihito Sawa. Resistive switching in transition metal oxides. Materials

Today, 11:28–36, 2008.

[45] Rainer Waser and Masakazu Aono. Nanoionics-based resistive switch-

ing memories. Nature Materials, 6:833–840, 2007.

[46] Sung Hyun Jo, Ting Chang, Idongesit Ebong, Bhavitavya B. Bhad-
viya, Pinaki Mazumder, and Wei Lu. Nanoscale memristor device as a
synapse in neuromorphic systems. Nanoletters, 10:1297–1301, 2010.

[47] Ella M Gale, Benjamin de Lacy Costello, and Andy Adamatzky. Ob-
servation and characterization of memristor current spikes and their
application to neuromorphic computation. In 2012 International Con-
ference on Numerical Analysis and Applied Mathematics (ICNAAM
2012), Kos, Greece, Sept 2012.

[48] Xin Jin, Alexander Rast, Franceson Galluppi, Sergio Davies, and Steve
Furber. Implimenting spike-timing dependent plasticity on SpiNNiker
neuromorphic hardware. 2010.

[49] Ella Gale, Ben de Lacy Costello, and Andrew Adamatzky. Memristor
Networks, chapter Spiking in Memristor Networks. Springer Interna-
tional Publishing Switzerland, 2014.

[50] Gerard David Howard, Ella Gale, Larry Bull, Benjamin

de Lacy Costello, and Andrew Adamatzky. Evolution of plastic
learning in spiking networks via memristive connections. IEEE
Transactions on Evolutionary Computation, 16:711–719, 2012.

[51] Bharathwaj Muthuswamy. Memristor based circuit chaos. IETE Tech-

nical Review, 26:1–15, 2009.

18
[52] Arturo Buscarino, Luigi Fortuna, Mattia Frasca, and Lucia Valentina
Gambuzza. A chaotic circuit based on Hewlett-Packard memristor.
Chaos, 22:023136, 2012.

[53] Bharathwaj Muthuswamy and Leon Chua. Simplest chaotic circuit.

International Journal of Bifurcation and Chaos, 20:1567, 2009.

[54] Makoto Itoh and Leon Chua. Memristor oscillators. International

Journal of Bifurcation and Chaos, 18:3183–3206, 2008.

[55] A. D. Rast, F. Galluppi, X. Jin, and S. B. Furber. The leaky integrate

and fire neuron: a platform for synaptic model exploration on the
SpiNNaker chip. 2010.

[56] Ella Gale, Ben de Lacy Costello, and Andrew Adamatzky. Boolean
logic gates from a single memristor via low-level sequential logic. In
Giancarlo Mauri, Alberto Dennunzio, Luca Manzoni, and Antonio E.
Porreca, editors, Unconventional Computation and Natural Computa-
tion, volume 7956 of Lecture Notes in Computer Science, pages 79–89.
Springer Berlin Heidelberg, 2013.

[57] S. Kvatinsky, Y. Nacson, Y. Etsion, E. Friedman, A. Kolodny, and

U. Weiser. Memristor-based multithreading. Computer Architecture
Letters, PP(99):1–1, 2013.

[58] T.M. Taha, R. Hasan, C. Yakopcic, and M.R. McLean. Exploring

the design space of specialized multicore neural processors. In Neural
Networks (IJCNN), The 2013 International Joint Conference on, pages
1–8, Aug 2013.

[59] Warren Robinett, Matthew Pickett, Julien Borghetti, Qiangfei

Xia, Gregory S Snider, Gilberto Medeiros-Ribeiro, and R Stanley
Williams. A memristor-based nonvolatile latch circuit. Nanotechnology,
21(23):235203, 2010.

[60] Yuriy V. Pershin and Massimiliano Di Ventra. Solving mazes with

memristors: A massively parallel approach. Phys. Rev. E, 84:046703,
Oct 2011.

[61] Chuan Kai Kenneth Lim and T. Prodromakis. Computing motion with
3D memristive grids. arxiv, page 1303.3067v1, 2013.

[62] Ioannis Georgilas, Ella Gale, Andrew Adamatzky, and Chris Melhuish.
UAV horizon tracking using memristors and cellular automata visual

19
 processing. In , editor, Advances in Autonomous Robotics, volume – of
Lecture Notes in Computer Science, pages –. Springer Berlin Heidel-
berg, 2013.

[63] Makoto Itoh and Leon O. Chua. Memristor cellular automata and
memristor discrete-time cellular neural networks. International Journal
of Bifurcation and Chaos, 19(11):3605–3656, 2009.

[64] A. Mittal and S. Swaminathan. Image stabilization using memristors.

In Mechanical and Electrical Technology (ICMET), 2010 2nd Interna-
tional Conference on, pages 789–792, Sept 2010.

[65] Ella Gale, Ben de Lacy Costello, and Andrew Adamatzky. Design of
a hybrid robot control system using memristor-model and ant-inspired
based information transfer protocols. In Proceedings of Workshop Fr-
Ws-09 on ‘Unconventional Approaches to Robotics and Automation In-
spired by Nature’ (UARACIN) at International Conference on Robotics
and Automation (ICRA), pages 34–36, May 2013.

[66] Lidan Wang, Xiaoyan Fang, Shukai Duan, and Xiaofeng Liao. Pid
controller based on memristive cmac network. Abstract and Applied
Analysis, page 510238 (6 pp), 2013.

[67] M. Rivas-Pérez, A. Linares-Barranco, J. Cerdá, N. Ferrando,

G. Jiménez, and A. Civit. Visual spike-based convolution processing
with a cellular automata architecture. 2010.

[68] Ella Gale, Benjamin de Lacy Costello, and Andrew Adamatzky. Ob-
servation, characterization and modeling of memristor current spikes.
Appl. Math. Inf. Sci., 7:1395–1403, 4,July 2013.

[69] Julien Borghetti, Gregory D. Snider, Philip J. Kuekes, J. Joshua Yang,

Duncan R. Stewart, and R. Stanley Williams. ‘memristive’ switches
enable ‘stateful’ logic operations via material implication. Nature,
464:873–876, 2010.

[70] Doo Seok Jeong, Herbert Schroeder, and Rainer Waser. Coexistence
of bipolar and unipolar resistive switching behaviors in a Pt/TiO2 /Pt
stack. Electrochemical and Solid State Letters, 10:G51–G53, 2007.

[71] B.J. Choi, D.S. Jeong, S.K. Kim, C.Rohde, S. Choi, J.H. Oh, H.J.
Kim, C.S. Hwang, K. Szot, R. Waser, B. Reichenberg, and S. Tiedke.
Resistive switching mechanism of TiO2 thin films grown by atomic-
layer deposition. Journal of Applied Physics, 98:033715, 2005.

20
[72] Nadine Gergel-Hackett, Behrang Hamadani, Barbara Dunlap, John
Suehle, Curt Richer, Christina Hacker, and David Gundlach. A flexible
solution-processed memrister. IEEE Electron Device Letters, 30:706–
708, 2009.

[73] Junggwon Yun, Kyoungah Cho, Byoungjun Park, Bae Ho Park, and
Sangsig Kim. Resistance switching memory devices constructed on
plastic with solution-processed titanium oxide. Journal of Materials
Chemistry, 19:2082–2085, 2009.

[74] Hu Young Jeong, Jeong Yong Lee, Sung-Yool Choi, and Jeong Won
Kim. Microscopic origin of bipolar resistive switching of nanocale tita-
nium oxide thin films. Applied Physics Letters, 95:162108, 2009.

[75] Sanghee Won, Seunghee Go, Kwanwoo Lee, and Jaegab Lee. Resis-
tive switching properties of Pt/TiO2 /n+-Si ReRAM for non-volatile
memory application. Electronic Materials Letters, 4:29–33, 2008.

[76] Micheal Colle, Michael Buchel, and Dago M. de Leeuw. Switching

and filamentary conduction in non-volatile organic memories. Organic
Electronics, 7:305–312, 2006.

[77] J. C. Bernede. Polarized memory switching in MIS thin films. Thin

Solid Films, 81:155–160, 1981.

[78] Kever T, U. Bottger, C. Schlindler, and R. Waser. On the origin if

bistable resistive switching in metal organic charge transfer complex
memory cells. Applied Physics Letters, 91:083506, 2007.

[79] Jeong Yong Lee Hu Young Jeong and Sung-Yool Choi. Interface-
engineered amorphous TiO2 based resistive memory devices. Advanced
Functional Materials, 20:3912–3917, 2010.

[80] Stefan T. Norberg, Stefan Hoffmann, Masahiro Yoshimura, and Nobuo

Ishizawa. Al6 Ti2 o13 O3 , a new phase in the Al2 —TiO2 system. Acta
Cryst. C, 61:i35–i38, 2005.

[81] E. Gale, D. Pearson, S. Kitson, A. Adamatzky, and B. de Lacy Costello.

Aluminium electrodes effect the operation of titanium oxide sol-gel
memristors. arXiv:1106.6293v2, 2011.

[82] K. Michelakis, T. Prodromakis, and C. Toumazou. Cost-effective fab-

rication of nanoscale electrode memristors with reproducible electrical
response. Micro Nano Letters, IET, 5(2):91–94, April 2010.

21
[83] Kyung-Hyun Choi, M.N. Awais, Hyung Chan Kim, and Yang Hui Doh.
Cost-effective printed memristor fabrication and analysis. In Cellular
Nanoscale Networks and Their Applications (CNNA), 2012 13th Inter-
national Workshop on, pages 1–4, Aug 2012.

[84] Jing Wu and Richard L. McCreery. Solid-state electrochemistry in

molecule/TiO2 molecular heterojunctions as the basis of the TiO2
“memristor”. Journal of the Electrochemical Society, 156:P29–P37,
2009.

[85] Rainer Waser, Tudor Baiatu, and Karl-Heinz Härdtl. dc electrical

degradation of perovskite-type titanates: I, ceramics. Journal of the
American Ceramic Society, 73(6):1645–1653, 1990.

[86] Krzysztof Szot, Wolfgang Speier, Gustav Bihlmayer, and Rainer Waser.
Switching the electrical resistance of individual dislocations in single-
crystalline SrTiO3 . Nature Materials, 5:312–320, 2006.

[87] J. Joshua Yang, Feng Miao, Matthew D. Pickett, Douglas A. A.

Ohlberg, Duncan R. Stewart, Chun Ning Lao, and R. Stanley Williams.
The mechanism of electroforming of metal oxide memristive switches.
Nanotechnology, 20:215201–1 – 215201–9, 2009.

[88] DH Kwon, J-M Jeon, JH Jang, KM Kim, CS Hwang, and M Kim.

Direct observation of conducting paths in TiO2 thin film by transmis-
sion electron microscopy. Microscopy and Microanalysis, null:996–997,
7 2009.

[89] Feng Miao, J. Joshua Yang, Julien Borghetti, Gilberto Medeiros, and
R. Stanley Williams. Observation of two resistance switching modes in
TiO2 memristive devices electroformed at low current. Nanotechnology,
22:254007 (7pp), 2011.

[90] Julien Borghetti, Dmitri B. Strukov, Matthew D. Pickett, J. Joshua

Yang, and Duncan R. Stewart. Electrical transport and thermometry of
electroformed titanium dioxide memristive switches. Journal of Applied
Physics, 106:125504 (5pp), 2009.

[91] K. Szot, W. Speier, R. Carius, U. Zastrow, and W. Beyer. Localized

metallic conductivity and self-healing during thermal reduction of Sr-
TiO. Phys. Rev. Lett., 88:075508, Feb 2002.

22
 [92] Daniele Ielmini, Federico Nardi, and Carlo Cagli. Universal reset char-
acteristics of unipolar and bipolar metal-oxide RRAM. IEEE Transac-
tions on Electron Devices, 58:3246–3253, 2011.

[93] L.A. Bursill and B.G. Hyde. Crystallographic shear in the higher ti-
tanium oxides: Structure, texture, mechanisms and thermodynamics.
Progress in Solid State Chemistry, 7(0):177 – 253, 1972.

[94] L. Liborio and N. Harrison. Thermodynamics of oxygen defective Mag-

neli phases in rutile: A first principles study. Phys. Rev. B, 77:104104,
2008.

[95] Feng Zhang, Xiaoyan Gan, Xiaomin Li, Liang Wu, Xiangdong Gao,
Renkui Zheng, Yong He, Xinjun Liu, and Rui Yang. Realization of
recitifying and resistive switching behaviors of TiO2 nanorod arrays for
nonvolatile memory. Electrochemical and Solid-State Letters, 14:H422–
H425, 2011.

[96] Deok-Hwang Kwon, Kyung Min Kim, Jae Hyuck Jang, Jong Mey-
ong Jeon, Min Hwan Lee, Gun Hwan Kim, Xiang-Shu Li, Gyeong-Su
Park, Bora Lee, Seungwu Han, Miyoung Kim, and Cheol Seong Hwang.
Atomic structure of conducting nanofilaments in TiO2 resistive switch-
ing memory. Nature Nanotechnology, 5:148–153, 2010.

[97] J.S. Lee, S.B. Lee, S.H. Chang, L.G. Gao, B.S. Kang, M.-L. Lee, C.J.
Kim, T.W. Noh, and B. Kahng. Scaling theory for unipilar resistance
switching. Physical Review Letters, 105:205701–1:205701–4, 2010.

[98] Seul Ji Song, Jun Yeong Seok, Jung Ho Yoon, Kyung Min Kim,
Gun Hwan Kim, Min Hwan Lee, and Cheol Seong Hwang. Real-time
identification of the evolution of conducting nano-filaments in TiO2
thin film ReRAM. Scientific Reports, 3:3443, 2013.

[99] L. Niemeyer, L. Pietronero, and H.J. Wiesmass. Fractal dimension of

dielectric breakdown. Physical Review Letters, 52:1033–1036, 1983.

[100] John Paul Strachan, Matthew D. Pickett, J. Joshua Yang, Shaul

Aloni, A. L. David Kilcoyne, Gilberto Medeiros-Ribeiro, and R. Stanley
Williams. Direct identification of the conducting channels in a func-
tioning memristive device. Advanced Materials, 22:3573–3577, 2012.

[101] M. J. Rozenberg, I. H. Inoue, and M. J. Snchez. Strong electron corre-

lation effects in nonvolatile electronic memory devices. Applied Physics
Letters, 88(3):–, 2006.

23
[102] ISCAS. The mythology of the memristor,, 2010.

[103] Ella Gale. The memory-conservation theory of memristance. In Pro-

ceedings of the 16th UKSim-AMSS International Conference of Mod-
elling and Simulation (UKSIM2014), pages 598–603, 2014.

[104] R. Stanley Williams. Aftermath of Finding the Memristor. In

A. Adamatzky and et al., editors, Chaos, CNN, Memristors and
Beyond: A Festschrift for Leon Chua (With DVD-ROM, Com-
posed by Eleonora Bilotta). Edited by Adamatzky Andrew et al. Pub-
lished by World Scientific Publishing Co. Pte. Ltd., 2013. ISBN
#9789814434805, pp. 490-493, pages 490–493, January 2013.

[105] Dmitri B. Strukov and R. Stanley Williams. Exponetial ionic drift: fast
switching and low volatility of thin-film memristors. Applied Physics
A, 94:515–519, 2009.

[106] Yogesh N. Joglekar and Stephen J. Wolf. The elusive memristor: prop-
erties of basic electrical circuits. Eur. J. Phys., 30:661–675, 2009.

[107] Z. Biolek, D. Biolek, and V. Biolkova. SPICE model of memristor with

nonlinear dopant drift. Radioengineering, 18:210–214, 2009.

[108] T. Prodromakis, B. P. Peh, C. Papavassiliou, and C. Toumazou. A ver-

satile memristor model with non-linear dopant kinetics. IEEE Trans.
Electron Devices, 30:3099–3105, 2011.

[109] Fernando Corinto and Alon Ascoli. A boundary condition-based ap-

proach to the modeling of memristor nanostructures. IEEE Transac-
tions on Circuits and Systems, 59:2713–2725, 2012.

[110] Chris Yakopcic, Tarek M. Taha, Guru Subramanyam, Robinson E.

Pino, and Stanley Rogers. A memristor device model. IEEE Electron
Device Letters, 32:1436–1438, 2011.

[111] Ella Gale. Uniform and peice-wise uniform fields in memristor models.
arXiv:1404.5581v2, 2014.

[112] Paul Meuffels and Rohit Soni. Fundamental issues and problems in the
realization of memristors. arXiv, page 207.7319v1, 2012.

[113] Stan Williams. Communication Relating to Brief Communication Aris-

ing Discussions.

24
[114] G.E. Pazienza and J. Albo-Canals. Teaching memristors to EE under-
graduate students [class notes]. IEEE Circuits and Systems Magazine,
11:36–44, 2011.
[115] J. Albo-Canals and G.E. Pazienza. A brief analysis of the main SPICE
models of the memristor. In Proceedings of 19th IEEE Int. Conf. Elec-
tronics, Circuits, and Systems (ICECS 2012), pages 861–864, 2012.
[116] Xuliang Zhang, Zhangeai Huang, and Juebang Yu. Memristor model
for SPICE. IEICE Trans. Electron., E93-C:355–360, 2010.
[117] P.S. Georgiou, S.N. Yaliraki, E.M. Drakakis, and M. Barahona. Quanti-
tative measure of hysteresis for memristors through explicit dynamics.
Proceedings of the Royal Society A, 468:2210–2229, August 2012.
[118] S. Kvatinsky, N. Wald, G. Satat, E. G. Friedman, A. Kolodny, and
U. C. Weiser. Memristor-based material implication (IMPLY) logic:
Design principles and methodologies. IEEE Transactions on Very Large
Scale Integration (VLSI), PP:1–13, 2013.
[119] Matthew D. Pickett, Dmitri B. Strukov, Julien L. Borghetti, J. Joshua
Yang, Gregory S. Snider. Duncun R. Stewart, and R. Stanley Williams.
Switching dynamics in titanium dioxide memristive devices. Journal
of Applied Physics, 109:074508, 2009.
[120] Shahar Kvatinsky, Eby G. Friedman, Avinoam Kolodny, and Uri. C
Weiser. TEAM: ThrEshold adaptive memristor model. IEEE Trans-
actions on Circuits and Systems, 60:222111–, 2013.
[121] E. Lehtonen and M. Laiho. CNN using memristors for neighborhood
connections. In Cellular Nanoscale Networks and Their Applications
(CNNA), 2010 12th International Workshop on, pages 1–4, Feb 2010.
[122] Ioannis Vourkas and Georgios Ch. Sirakoulis. A novel design and mod-
eling paradigm for memristor-based crossbar circuits. IEEE Trans.
Nanotechnology, 11:1151–1159, 2012.
[123] Ioannis Vourkas, Athanasios Batsos, and Georgios Ch. Sirakoulis.
SPICE modeling of nonlinear memristive behavior. International Jour-
nal of Circuit Theory and Applications, pages n/a–n/a, 2013.
[124] C. Yakopcic, T.M. Taha, G. Subramanyam, and R.E. Pino. Gener-
alized memristive device SPICE model and its application in circuit
design. Computer-Aided Design of Integrated Circuits and Systems,
IEEE Transactions on, 32(8):1201–1214, Aug 2013.

25
[125] R.Stanley Williams and MatthewD. Pickett. The art and science of
constructing a memristor model. In Ronald Tetzlaff, editor, Memristors
and Memristive Systems, pages 93–104. Springer New York, 2014.
[126] Leon O. Chua and Sung Mo Kang. Memristive devices and systems.
Proceedings of the IEEE, 64:209–223, 1976.
[127] Shyam Prasad Adhikari, Hyongsuk Kim Maheshwar Pd. Sah, and
Leon O. Chua. Three fingerprints of memristor. IEEE Transactions
on Circuits and Systems, 60:3008–3021, November 2013.
[128] Massimiliano Di Ventra and Yuriy V Pershin. On the physical prop-
erties of memristive, memcapacitive and meminductive systems. Nan-
otechnology, 24(25):255201, 2013.
[129] Yuriy V. Pershin and Massimiliano Di Ventra. Memory effects in com-
plex materials and nanoscale systems. Advances in Physics, 60:145–227,
2011.
[130] Ella M Gale, Benjamin de Lacy Costello, and Andy Adamatzky. The
effect of electrode size on memristor properties: An experimental and
theoretical study. In 2012 IEEE International Conference on Electron-
ics Design, Systems and Applications (ICEDSA 2012), Kuala Lumpur,
Malaysia, November 2012.
[131] I. Valov, E. Linn, S. Tappertzhofen, S. Schmelzer, J. van den Hurk,
F. Lentz, and R. Waser. Nanobatteries in redox-based resistive switches
require extension of memristor theory. Nature Communications, 4:1771
(9pp), 2013.
[132] Blaise Mouttet. The memristor and scientific method. viXra:1205.0004,
2012.
[133] H.-S.P. Wong, H.-Y. Lee, S. Yu, Y.-S. Chen, Y. Wu, P.-S. Chen, B. Lee,
F.T. Chen, and M.-J. Tsai. Metal-oxide RRAM. Proc. IEEE, 100:1951–
1970, 2012.
[134] J. L. Tedesco, L. Stephey, M. Hernandez-Mora amd C. A. Richter, and
N. Gergel-Hackett. Switching mechanisms in flexible solution-processed
TiO2 memristors. Nanotechnology, 23:305206 (7pp), 2012.
[135] Nathan R. McDonald, Robinson E. Pino, Peter J. Rozwood, and
Bryant T. Wysocki. Analysis of dynamic linear and non-linear mem-
ristor device models for emerging neuromorphic computing hardware
design. In IJCNN, pages 1–5, 2010.

26
[136] S. Hamdioui, M. Taouil, and N.Z.B. Haron. Testing open defects
in memristor-based memories. IEEE Transactions on Computers,
99:TC.2013.206, October 2013.

[137] Seungjun Kim, Hu Young Jeong, Sung Kyu Kim, Sung-Yool Choi, and
Keon Jae Lee. Flexible memristive memory array on plastic substrates.
Nano Letters, 11(12):5438–5442, 2011.

[138] Ella M Gale, Benjamin de Lacy Costello, and Andy Adamatzky. Fil-
amentary extension of the Mem-Con theory of memristance and its
application to titanium dioxide Sol-Gel memristors. In 2012 IEEE
International Conference on Electronics Design, Systems and Applica-
tions (ICEDSA 2012), Kuala Lumpur, Malaysia, November 2012.

[139] Y. Ho, G.M. Huang, and P. Li. Dynamical properties and design anal-
ysis for nonvolatile memristor memories. IEEE Trans. Circuits Syst. I,
58:724–736, 2011.

[140] K.-H. Kim, S. Gaba, D. Wheeler, J.M. Cruz-Albrecht, T. Hussain,

N. Srinivasa, and W. Lu. A functional hybrid memristor crossbar-
array/cmos system for data storage and neuromorphic applications.
Nano Lett., 12:389–395, 2012.

[141] Pascal O Vontobel, Warren Robinett, Philip J Kuekes, Duncan R Stew-

art, Joseph Straznicky, and R Stanley Williams. Writing to and reading
from a nano-scale crossbar memory based on memristors. Nanotech-
nology, 20(42):425204, 2009.

[142] S. Kannan, J. Rajendran, O. Sinanoglu, and R. Karri. Sneak path

testing of memristor-based memories. In Proceedings of the 2013 26th
International Conference on VLSI Design and the 12th International
Conference on Embedded Systems, pages 386–391, 2013.

[143] Harika Manem, Garrett S. Rose, Xiaoli He, and Wei Wang. Design
considerations for variation tolerant multilevel cmos/nano memristor
memory. In Proceedings of the 20th Symposium on Great Lakes Sym-
posium on VLSI, GLSVLSI ’10, pages 287–292, New York, NY, USA,
2010. ACM.

[144] Chul-Moon Jung, Jun-Myung Choi, and Kyeong-Sik Min. Two-step

write scheme for reducing sneak-path leakage in complementary mem-
ristor array. Nanotechnology, IEEE Transactions on, 11(3):611–618,
May 2012.

27
[145] M. A. Zidan, H. Fahmy, A. Eltawil, F. Kurdahi, and K. N. Salama.
Memristor multi-port readout: A closed-form solution for sneak-paths.
IEEE Transactions on Nanotechnology, 13:274–282, 2014.
[146] M.S. Qureshi, W. Yi, G. Medeiros-Ribeiro, and R.S. Williams. AC
sense technique for memristor crossbar. Electronics Letters, 48:757–
758(1), June 2012.
[147] M. A. Zidan, H. Fahmy, M.H Hossein, and K. N. Salama. Memristor
based memory: The sneak paths problem and solutions. Microelectron-
ics journal, 44:176–183, 2013.
[148] V. Erokhin, T. Berzina, and M. P. Fontana. Hybrid electronic de-
vice based on polyaniline-polyethylenoxide junction. J. Appl. Phys.,
97:064501, 2005.
[149] Victor Erokhin, Almut Schuz, and M.P. Fontana. Organic memris-
tor and bio-inspired information processing. International Journal of
Unconventional Computing, 6:15–32, 2009.
[150] Alfred North Whitehead and Bertrand Russell. Principia Mathematica:
volume one, pages 394–508. Marchant Books, 1910.
[151] Ella Gale, Ben de Lacy Costello, and Andrew Admatzky. Is spiking
logic the route to memristor-based computers? In 2013 International
Conference on Electronics, Circuits and Systems (ICECS), pages 297–
300, Abu Dhabi, UAE, 8-11 December 2013. IEEE.
[152] Ioannis Vourkas and Georgios Ch. Sirakoulis. Memristor-based com-
binational circuits: A design methodology for encoders/decoders. Mi-
croelectronics Journal, 45:59–70, 2014.
[153] Eero Lehtonen, JussiH. Poikonen, and Mika Laiho. Memristive state-
ful logic. In Andrew Adamatzky and Leon Chua, editors, Memristor
Networks, pages 603–623. Springer International Publishing, 2014.
[154] Joao Capela Duarte, Ernesto Ventura Martins, and Luis Nero Alves.
Frequency characterisation of memristive devices. In 2013 European
Conference on Circuit Theory and Design (ECCTD), pages 1–4, Dres-
den, Germany, 8-12 September 2013. IEEE.
[155] Joao Capela Duarte, Ernesto Ventura Martins, and Luis Nero Alves.
Amplitude characterization of memristive devices. In 2013 Interna-
tional Conference on Electronics, Circuits and Systems (ICECS), pages
45–49, Abu Dhabi, UAE, 8-11 December 2013. IEEE.

28
[156] Z.Biolek D. Biolek and V.Biolkova. Interpreting area of pinched hys-
teresis loop. Electronics Letters, 50:74–75, Jan 2014.

[157] T.A. Wey and W.D. Jemison. Variable gain amplifier circuit using
titanium dioxide memristors. IET Circuits and Systems, 5:59–65, 2011.

[158] Omid Kavehei, Said Al-Sarawi, Kyong-Rok Cho, Kamram Eshraghian,

and Derek Abbott. An analytical approach for memristive nanoarchi-
tectures. IEEE Transactions on Nanotechnology, 11:374–385, 2012.

29