SnO -BASED MEMRISTORS AND THE POTENTIAL SYNERGIES OF INTEGRATING MEMRISTORS WITH MEMS

SnO2-BASED MEMRISTORS AND THE POTENTIAL SYNERGIES
OF INTEGRATING MEMRISTORS WITH MEMS
a
David Zubia*a, Sergio Almeidaa, Arka Talukdara, Jose Mirelesb, and Eric MacDonalda
Electrical and Computer Engineering Department, University of Texas at El Paso, El Paso, TX
79968; bUniversidad Autonoma de Ciudad Juarez, Ciudad Juarez, Chihuahua, Mexico
ABSTRACT
Memristors, usually in the form metal/metal-oxide/metal, have attracted much attention due to their potential application
for non-volatile memory. Their simple structure and ease of fabrication make them good candidates for dense memory
with projections of 22 terabytes per wafer. Excellent switching times of ~10 ns, memory endurance of >10 9 cycles, and
extrapolated retention times of >10 yrs have been reported. Interestingly, memristors use the migration of ions to change
their resistance in response to charge flow, and can therefore measure and remember the amount of current that has
flowed. This is similar to many MEMS devices in which the motion of mass is an operating principle of the device.
Memristors are also similar to MEMS in the sense that they can both be resistant to radiation effects. Memristors are
radiation tolerant since information is stored as a structural change and not as electronic charge. Functionally, a MEMS
device’s sensitivity to radiation is concomitant to the role that the dielectric layers play in the function of the device. This
is due to radiation-induced trapped charge in the dielectrics which can alter device performance and in extreme cases
cause failure. Although different material systems have been investigated for memristors, SnO2 has received little
attention even though it demonstrates excellent electronic properties and a high resistance to displacement damage from
radiation due to a large Frenkel defect energy (7 eV) compared its bandgap (3.6 eV). This talk discusses recent research
on SnO2-based memristors and the potential synergies of integrating memristors with MEMS.
Keywords: Memristors, MEMS, radiation-hard, integration, SnO2
1. INTRODUCTION
Resistive switching in metal/metal-oxide/metal structures, also known as memristors, has recently attracted much
attention due to its potential application for non-volatile memory. 1 Their simple structure and ease of fabricating 2D
arrays make them good candidates for dense memory with projections of 0.5 F2 per cell where F is the feature size
corresponding to 22 terabytes per wafer. 2,3 The memristor has the potential to combine the best characteristics of the
hard drive, RAM and flash in terms of density, access speed and power, and resistance to radiation effects. Excellent
switching times of ~10 ns, memory endurance of >10 9 cycles, and extrapolated retention times of >10 yrs have been
reported.4
Memristors use the migration of ions to change their resistance in response to charge flow, and can therefore measure
and remember the amount of current that has flowed. 5 This is in contrast to CMOS-based flash RAM memory devices
which use electronic charge to store information. This makes memristors potentially radiation-hardened since
information is stored as a structural change and not as electronic charge. For example, in TiO 2-based memristors, defect
concentrations up to 1% are tolerated well compared to conventional CMOS that can only tolerate less than 0.1% at the
channel-gate oxide interface. 6 HfO2-based memristors also show potential radiation hardness. 7 Interestingly, memristors
are similar to MEMS devices in which the motion of mass is an operating principle of the device. Memristors are also
similar to MEMS in the sense that they can both be resistant to radiation effects. Functionally, a MEMS device’s
sensitivity to radiation is concomitant to the role that the dielectric layers play in the function of the device. This is due
to radiation-induced trapped charge in the dielectrics which can alter device performance and in extreme cases cause
failure.
Different material systems have been investigated for resistive switching, including Pt/TiO 2/Pt8, Ti/TiO2/Ti,
Pt/HfO2/Cr/Au9, Pt/MgZnO/Pt, Pt/TiO2/SrTi:Nb/Pt, and Pt/SnO2/Pt10. Although SnO2 has excellent electronic and
structural properties it has received little attention for resistive switching applications. 11 SnO2 demonstrates orders of
magnitude in resistivity change depending on its oxygen vacancy concentration.12 Moreover, it demonstrates a high
Updated 1 March 2012
resistance to displacement damage from radiation due to its Frenkel defect energy (7 eV) being much larger than its
bandgap (3.6 eV).13 Instead SnO2 and indium-tin-oxide have been extensively used as a transparent conducting oxide for
solar cell and flat panel displays which require high conductivity and high optical transmission. Resistivities as low as
10-2 and 10-4 Ohm-cm have been achieved on undoped and doped films, respectively. 14 This is compared to metallic tin
which has a resistivity of 10-5 Ohm-cm. At the high end, a resistivity of 3.3 Ohm-cm has been reported.15 Moreover, our
group has achieved resistivity as high as 106 Ohm-cm using RF magnetron sputtering of SnO2 at high oxygen content.16
This represents 11 orders of magnitude from metallic tin to highly stoichiometric SnO 2. This paper presents resistive
switching of SnO2-based memristors on silicon using a MOM structure with different materials (Ag, Ag paste, Au, and
Al) as a top electrode. Detailed structural and electrical characterization of the devices is presented. Potential synergies
of integrating memristors with MEMS are also discussed.
2. EXPERIMENTAL DETAILS
Three-inch Si wafers with (100) orientation were used as substrates. The wafers were cleaned by the standard RCA
process. The wafers were then immersed into a mixture of 50ml of H 2SO4 and 10ml of H2O2 for 10 minutes. A SiO2
layer with a thickness of 180 nm was grown on the wafers by exposing them to dry oxygen for 2.25 hours at 1323 K.
Metal/metal-oxide/metal (MOM) structures were fabricated on the oxidized wafers using Ti as the bottom electrode,
SnO2 as metal-oxide, and different materials (Au, Al, Ag and Ag paste) as the top electrode. The Ti bottom electrode
was deposited by RF magnetron sputtering using a Kurt J. Lesker depositor. A base pressure of 3.2 x 10 -6 Torr was
reached, a forward power of 100 W was used and the deposition pressure was maintained at 2.0 x 10 -3 with an Ar flow of
40 sccm. Subsequently SnO2 was deposited by RF reactive sputtering using a shadow mask in order to allow contact to
the bottom electrode. The forward power used was 40 W maintaining a deposition pressure of 5.3 X10 -3 Torr with an O2
flow of 70 sccm. The Au and Ag were deposited via thermal evaporation, the Al deposited by RF sputtering, and the Ag
paste was applied by hand application. In order to define the size of the Au, Ag and Al electrodes, a lift-off process was
used in which the wafers were coated with photoresist and patterned to create 100 x100 µm openings to expose the
SnO2. The photoresist was then removed after deposition using methanol in an ultrasonic bath. The thicknesses of the Ag
and Au were 100nm and 95 nm, respectively. The Al was deposited by RF sputtering using a forward power of 100W
maintaining a pressure of 2.0 x 10 -3 Torr and an Ar flow of 40 sccm. Finally, Ag paste drops were applied on SnO 2.
Microstructure analysis was performed and current-voltage (I-V) characteristics of the fabricated structures were
evaluated using a Keithley 2400 source meter and micro-manipulator probe station using software to control the
Keithley.
3. STRUCTURAL ANALYSIS
Figure 1 shows the grazing incidence x-ray diffraction (GIXRD) pattern of SnO2/Ti/SiO2/Si. Peaks for SnO2 and Ti are
observed indicating a crystalline growth for both layers.
Figure 1. Grazing incidence XRD pattern of the SnO 2/Ti/SiO2/Si structure. Peaks for SnO2 and Ti are observed indicating
nano-crystallinity in for both layers.
Cross-sectional SEM images of the Au/SnO2/Ti/SiO2/Si and Ag-paste/SnO2/Ti/SiO2/Si MOM structures are presented in Figure
2(a) and Figure 2(b), respectively. The SnO2 thickness is approximately 50 nm and the Ti film is ~100 nm thick. In order to
observe the grain size of the different top electrodes, SEM pictures of the surfaces are shown in Figure 3. Figure 3(a) shows the
thermally evaporated Ag with a 50 nm average grain size. Figure 3(b) shows the surface of RF sputtered Al with a 50 nm average
grain size and larger 100 nm grain inclusions. Figure 3(c) shows the surface of the thermally evaporated Au with a 25 nm
average grain size. Figure 3(d) shows the surface of the Ag-paste where particles from 1 µm to 5 µm can be observed.
In general, the thermally evaporated films (Au, Ag) showed the smallest and smoothest grain structure. This was followed by the
sputter deposited (Al) films which showed small grain size intermixed with much larder inclusions. Lastly, the Ag-paste showed
the most non-uniform structure with particles size ranging from 1 to 5 microns. Finally, energy dispersive x-ray spectroscopy of
the top contact metals confirmed the presence of elements in each of the structures.
(b)
(a)
Figure 2. Cross-sectional SEM images of (a) Au/SnO2/Ti/SiO2/Si and (b) Ag-paste/SnO2/Ti/SiO2/Si MOM structures.
(a)
(c)
(b)
(d)
Figure 3. SEM surface images of the (a) thermally evaporated Ag, (b) RF sputtered Al, (c) thermally evaporated Au, and (d) Agpaste top electrodes.
4. ELECTRICAL ANALYSIS
SnO2-based memristors have previously been reported to demonstrate unipolar resistive switching.10,11 For the structures
in this study, a positive voltage was applied to the top electrode in all the cases while the Ti bottom electrode was
grounded. Only structures with the RF sputtered Al and hand applied Ag-paste electrodes showed resistive switching. In
contrast, the MOMS with the thermally evaporated Ag and Au did not switch. Figure 4 below shows the IV
characteristics of the MOMS with Ag and Au top electrodes.
0.03
Ag
Au
0.025
Current (A)
0.02
0.015
0.01
0.005
0
-0.005
-0.5
0
0.5
1
1.5
2
2.5
3
3.5
Voltage (V)
Figure 4. IV curves for Ag/SnO2/Ti/SiO2/Si and Au/SnO2/Ti/SiO2/Si structures. It shows a space charge limited conduction for
voltages higher than 0.5 volts.
The structures with the RF sputtered Al and Ag-paste electrodes showed resistive switching. Figure 5(a) shows the
forming, reset, and set stages of the memristor with Ag-paste top electrode. For this structure, a resistor of 500 Ohms
was placed in series with the MOM during the forming stage and when switching from the high resistance state (HRS) to
the low resistance state (LRS) to implement current compliance and avoid drastic voltage changes during switching. The
forming process occurred at 2.26 V. After the forming stage, consistent unipolar switching was observed as shown in
Figure 5(b) which shows 40 switching endurance cycles. The resistance ratio was approximately one order of magnitude.
6
-2
10
10
(b)
(a)
Forming stage
-3
10
5
Resistance ()
Current (A)
10
-4
10
-5
10
4
10
3
10
-6
10
Forming stage
LRS
HRS
-7
10
0
0.5
1
1.5
2
2.5
Voltage (V)
3
3.5
2
4
10
0
5
10
15
20
25
30
Cycle
Figure 5. (a) I-V characteristic of the forming, set, and reset stages and (b) memory endurance of a typical Agpaste/SnO2/Ti/SiO2/Si memristor. The arrows indicate the direction of the applied voltage.
35
40
Figure 6(a) shows the switching behavior of the memristor that had Al as the top electrode. In these tests, a compliance
current of 10 mA was set to prevent damage to the memristor. Initial switching occurred at a bias of 5V during the
forming stage. Even though this structure exhibited resistive switching, it was not consistent as shown in the endurance
data in Figure 6(b).
6
-1
10
10
-2
10
HRS
LRS
Forming stage
(a)
(b)
5
10
-3
Resistance ()
Current (A)
10
-4
10
-5
10
-6
4
10
3
10
10
2
10
-7
10
1
-8
10
-1
0
1
2
3
4
5
10
6
0
2
4
6
8
10
12
14
16
18
Cycle
Voltage (V)
Figure 6. (a) I-V characteristic of the forming, set, and reset stages and (b) memory endurance of a typical Al/SnO2/Ti/SiO2/Si
memristor. The arrows indicate the direction of the applied voltage.
Researchers have reported different conduction processes for SnO 2 including, Ohmic, Poole-Frenkel (PF) emission, and
Schottky emission depending on the top contact material.10 The current density versus voltage in logarithmic scale for
the Ag-paste is shown in Figure 7(a) for the LRS and the Figure 8(a) for HRS. This correlates with the analysis presented
in [10] even though the compliance current was control by a 500 ohms resistor in series. The Al/SnO2/Ti/SiO2/Si
structure shows similar behavior however the HRS show Ohmic behavior at voltages below 0.4 V while the Agpaste/SnO2/Ti/SiO2/Si memristor showed Ohmic behavior at voltages below 0.2 V. A slope of unity indicates that the
current conduction is Ohmic in nature. All the cycles are described by this conduction process with a slight resistance
difference between cycles.
Al contact 11
(b)
(a)
-2
-3
10
Current (A)
Current (A)
10
-3
10
Slope = 1
J E
Slope = 1
JE
-4
10
-1
10
Voltage (V)
0
10
-1
10
0
10
Voltage (V)
Figure 7. (a) LRS conduction process analysis for the Ag-paste/SnO2/Ti/SiO2/Si structure. (b) LRS conduction process analysis
for the Al/SnO2/Ti/SiO2/Si structure. Constant Ohmic behavior is present in both structures.
-3
10
(b)
(a)
Ohmic
-3
Current (A)
Current (A)
10
Ohmic
-4
10
-4
10
-5
10
-5
10
-1
-1
0
10
0
10
10
10
Voltage (V)
Voltage (V)
Figure 8. (a) HRS conduction process analysis for the Ag-paste/SnO2/Ti/SiO2/Si structure. (b) HRS conduction process analysis
for the Al/SnO2/Ti/SiO2/Si structure. Constant Ohmic behavior is present in both structures at voltages indicated the dashed line.
5. INTEGRATION OF MEMRISTORS WITH MEMS
5.1 Introduction
This section discusses on the potential application of memristors to MEMS and microsystems. Many applications of
integrating memristors with MEMS are anticipated. These are divided into three categories: One is the use of memristors
in analog circuitry to create advanced nonlinear functions with the capability to recognize (sense) electrical patterns and
remember (memory) the patterns over a finite length of time. A second category is the application of memristors to
emulate synapses in artificial neural networks to create artificial intelligence (smart). The third category is the use of
memristors in digital circuitry for ultra-dense and radiation-hard nonvolatile memory, and as self-programmable
elements in memristor/MEMS circuits (smart reconfigurable circuitry).
5.2 Definition of Memristors
In 1971, Chua proposed, based on a purely mathematical symmetry argument, a new circuit element in the same
category as the resistor, capacitor and inductor.17 He called the new element the “memristor”. In the original formalism
by Dr. Chua, the memristor related the time rate of change in the magnetic flux to the time rate of change in the charge
through what is called the “memristance”. Unfortunately, using the original formalism to find the memristor proved to be
1
an elusive challenge. A contemporary formalism, given by the Equations
v = M(w)i
(1)
dw/dt = i
(2)
where v is the voltage across the device, i is the current, and M(w) is the memristance. w is a state variable (charge in this
case) is much more practical and led to the discovery of the memristor by Williams, et al., at HP laboratories. Equation
(1) is Ohm’s law, however, in a memristor the value of M changes depending on the charge, w, that flows through it. In
other words, memristors retain memory of the charge flow. This leads to one of the properties of memristors - they have
memory. This property implies that the i-v characteristic will demonstrate hysteresis. Another property, which comes out
of Equation (1) is that they cannot store energy. The manifestation of this property is that the i-v characteristics of
memristors must pass through the origin. Together, these two properties create the so called “bowtie” or “pinched
1
hysteresis” plots.
5.3 Memristor Technologies
One major type of technology is the metal-insulator-metal (MIM) structure. The MIM structure is composed of an
8
insulator material sandwiched between two electrodes, making the memristor a two-terminal device. The most mature
technology is that composed of Pt/TiO2/Pt. Overall, the transition-metal oxides are strong candidates due to their high
sensitivity to native vacancies and other defects.
5.4 Switching Mechanisms
Many switching mechanisms have been purported and they are highly dependent on the technology. Here, we will
concentrate on a few mechanisms that have been reported to operate in the MIM structure. One model proposes that
there is a moving boundary between doped and undoped regions of the insulator. For example, in this model the TiO 2 is
1
modeled as being composed of two resistors in series, corresponding to the doped and undoped regions. Stoichiometric
TiO2 is highly resistive while depleted TiO2-x is conductive. The boundary between these two species moves due to
current flow and causes the overall device resistance to change. Another mechanism espouses that the resistance change
comes from changes that occur at an interface. In this case, researchers believe that oxygen vacancies dope the interface
and cause a change in the nature of the Schottky barrier.
For many MIM technologies, including the Pt/TiO 2/Pt devices, studies have shown that the resistance change is localized
and manifested as the formation and rupture of conducting filaments. These devices show the filamentary switching
behavior.18 In this model, the initial metal-oxide is highly resistive and homogeneous. However, when a strong field is
applied across the device, conducting filaments composed of sub-oxides form a conducting path through the device. The
path can be ruptured and repaired many times by cycling the voltage bias in the device.
5.5 Memristor Application Examples
Memristors can be used to implement complex non-linear functions with few components due to their non-linear
behavior with memory. For example, the simple circuit shown in Figure 9 which contains a memristor, inductor, resistor
and capacitor was used to model the learning behavior of amoebas. The circuit is able to recognize and remember a
periodic electric pattern over a finite length of time. The voltage across the circuit, V(t), models temperature/humidity,
while the voltage across the capacitor, VC(t), models the amoeba’s response. When the input voltage fluctuation was nonperiodic, the response was damped and no memory was observed. In contrast, when the input was periodic over a finite
length of time, a response was given matching the periodicity of the input even after the input was removed. 19
V(t)
R
L
M(t)
C
Figure 9. Nonlinear analog circuit that models the learning of amoebas.
Memristors also have the potential to emulate (as opposed to simulate) neural synapses. Neural scientists have
discovered that a highly important parameter that links two neurons together is the timing of their respective electrochemical firing. This is the so called “cells that fire together, wire together” or plasticity. The strength of the link is
strongly related to how close in time the neurons fire together. Memristors’ ability to change resistance in relation to
how they are biased (fired) makes them exceptional candidates to emulate neural synapses.
Another example application of memristors is exemplified by a self-programmable logic circuit created by HP and the
State University of New York (SUNY).20 The circuit is composed of two crossbar arrays of memristors and several
MOSFECT transistors along the periphery. This architecture was used to implement the function, f = AB+CD, and also
to reprogram the resistance state of one of the memristors from high-resistance state (HRS) to low-resistance state
(LRS).
5.6 Memristor Applications for Microsystems
Memristors also have many potential applications for microsystems. The MOM structure of memristors is very simple
and the materials and manufacturing processes are highly compatible with MEMS. An underlying principle of
memristors is memory of charge flow. These features will have application in MEMS that also use flow of charge in
capacitors to convey information. Since the resistance in memristors can change from high to low, and vice versa, they
can be used as electrical switches to connect and disconnect MEMS components such as sensors and actuators. In this
case the memristor are fabricated in series with the MEMS components. Arrays of memristors can be used to reconfigure
MEMS architectures. Memristors can also be used a voltage and current limiters. As current limiters, the memristors will
act as resettable fuses which disconnect MEMS components when a certain current is exceeded. As voltage limiters,
memristors will act as resettable shunts which connect components when the voltage exceeds the switching voltage of
the memristor. Moreover, memristors can be used as sensors to detect charge flow, voltage, and current.
An interesting application is the intimate integration of the memristor with MEMS components to produce new circuits
with non-linear behavior and built-in memory. This will create complex functionality with few components. For
example, when intimately integrated with the MEMS components, memristors will enable the local storage of memory.
A possible example of such integration is shown in Figure 10, where the packaging of a MEMS device is surrounded
hermetically by a memristor device, or perhaps through the TSVs (electrical interconnection lines from MEMS devices
to exterior). Schematic representations of the integration of memristors with parallel plate capacitors are shown in Figure
11 for series and parallel connections. This integration has never been used before, and we believe has a strong potential
to create a new hybrid device for several applications, as discussed above.
Figure 10. A possible memristor integration into MEMS devices showing a cross section of a packaged MEMS
device with comb fingers sensing mechanisms shown in red.
k
(a)
(b)
Q
Q
k
CM
MR
V
CM
V
MR
Figure 11. A schematic representation of the integration between memristors and parallel plate capacitors for (a)
series and (b) parallel connections.
Memristors are potentially radiation-hard since information is stored as structural change. This will have important
applications in radiation-hard microsystems. For example, memristors arrays can be used as non-volatile radiation-hard
state memory for microsystems as shown in Figure 12 below. In conclusion, the combination of all the key features of
memristors (non-volatile radiation-hard memory, detection and memory of charge flow, simple structure, ease of
fabrication, materials and processing compatibility with MEMS) make them an excellent candidate for many
applications in microsystems.
CMOS-IC
MEMS
RH-NVM
Figure 12. Integration of memristors arrays (radiation-hard non-volatile memory) with MEMS and CMOS ICs to
create radiation-hard microsystems.
6. DISCUSSION AND CONCLUSIONS
MOM structures with Au, Ag, Ag-paste and Al as top electrodes were fabricated on SnO2/Ti/SiO2/Si substrates. The
memristors with the Ag-paste as top electrode demonstrated the most consistent switching behavior. Endurance data
showed good repeatability and indicated that more switching was possible. Conduction analysis showed a clear Ohmic
behavior for both LRS and HRS. However, at voltages higher than 0.4 V, the HRS behavior deviated from Ohmic
conduction.
In contrast the MOMs with evaporated Ag or Au as top electrodes did not switch. Instead their IV characteristics were
non-linear with current increasing exponentially at first and then linearly with applied voltage similar to Schottky
emission limited by a series resistance at higher voltages. Furthermore, the memristors with Al as top electrode showed
unipolar resistive switching however it was not consistent.
It is speculated that the difference in the microstructure played a role for the different behavior between the pure Ag and
Ag-paste, and the Al. In the Ag-paste, irregular grain size will give raise to a non-uniform local electrical field pattern
such that in some areas the electric field will exceed a critical value that will create the dielectric breakdown in the SnO2
and therefore form a conducting filament. In contrast the electrical field in the evaporated Ag MOM is much uniform
requiring a larger forming voltage. However due to current leakage in the device, IR drops effectively limit the electric
field from reaching a critical value needed for resistive switching.
Memristors are similar to MEMS devices in which the motion of mass is an operating principle of the device.
Memristors are also similar to MEMS in the sense that they can both be resistant to radiation effects. Moreover, the
combination of other key features such as potential as non-volatile radiation-hard memory, detection and memory of
charge flow, simple structure, ease of fabrication, materials and processing compatibility with MEMS will make
memristors an excellent candidate for many applications in microsystems.
7. ACKNOWLEDGEMENTS
This work is supported by Sandia National Laboratories under contract 1156850. Sandia National Laboratories is a
multi-program laboratory managed and operated by Sandia Corporation, a wholly owned subsidiary of Lockheed Martin
Corporation, for the U.S. Department of Energy's National Nuclear Security Administration under contract DE-AC0494AL85000.
REFERENCES
[1] D. B. Strukov, G. S. Snider, D. R. Stewart and R. S. Williams, “The missing memristor found”, Nature 453 (2008) 80-83.
[2] John Sanchez, “New CMOxTMCross-Point Memory Technology Based on a Novel Oxide Memory Element,” Thin-Film User
Group Meeting, Northern California Chapter of the American Vacuum Society, October 14, 2009, http://unitysemi.com/ and
UnitySemi Cross-point array, http://www.unitysemi.com/our-technology-cross-point-array.html, retrieved April 30, 2012.
[3] S. H. Jo, K –H. Kim, W. Lu, “High-density crossbar arrays based on a Si memristive system”, Nano Letters 9 No 2 (2009) 870874.
[4] Rainer Waser, Regina Dittmann, Georgi Staikov, and Kristof Szot , “Redox-Based Resistive Switching Memories –Nanoionic
Mechanisms, Prospects, and Challenges”, Adv. Mater. 21 (2009) 2632–2663.
[5] D. B. Strukov, R. S. Williams, “Exponential ionic drift: fast switching and low volatility of thin-film memristors”, Applied
Physics A 94 (2009) 515-519.
[6] W.M. Tong, J.J. Yang, P.J. Kuekes, D.R. Stewart, R.S. Williams, E. DeIonno, E.E. King, S.C. Witczak, M.D. Looper, and J.V.
Osborn, “Radiation Hardness TiO2 Memristive Junctions,” IEEE Transactions on Nuclear Science 57 (2010) 1640-1643.
[7] Yan Wang, Hangbing Lv, Wei Wang, Qi Liu, Shibing Long, Qin Wang, Zongliang Huo, Sen Zhang, Yingtao Li, Qingyun Zuo,
Wentai Lian, Jianhong Yang, and Ming Liu, “Highly Stable Radiation-Hardened Resistive-Switching Memory”, Electron Device
Letters 31 (2010) 1470-1472.
[8] J. J. Yang, M. D. Pickett, X. Li, D. A. A. Ohlberg, D. R. Stewart, and S. Williams, “Memristive switching mechanism for
metal/oxide/metal nanodevices”, Nature Nanotechnology 3 (2008) 429-433.
[9] M. Y. Chan, T. Zhang, V. Ho, P. S. Lee, “Resistive switching effects of HfO2 high-k dielectric”, Microelectronic Engineering 85
(2008) 2420-2424.
[10] K. Nagashima, T. Yanagida, K. Oka, and T. Kawai, “Unipolar resistive switching characteristics of room temperature grown
SnO2 thin films”, Applied Physics Letters 94 (2009) 242902.
[11] Sergio Almeida, Brandon Aguirre, Noel Marquez, John McClure and David Zubia, “Resistive Switching of SnO 2 Thin Films on
Glass Substrates”, Integrated Ferroelectrics, 126:1 (2011) 117.
[12] M. Buchanan, J. B. Webb and D. F. Williams, “The Influence of Target Oxidation and Growth Related Effects on the Electrical
Properties of Reactively Sputtered Films of Tin-Doped Indium Oxide”, Thin Solid Films 80 (1981) 373 – 382.
[13] Golovanov, V. , Khirunenko, L. , Kiv, A. , Fuks, D. , Soshin, M. and Korotchenkov, G., “Radiation effects in SnO2-Si sensor
structures”, Radiation Effects and Defects in Solids 161: 2, (2006) 85-89.
[14] H. Toyosaki, M. Kawasaki, and Y. Tokura, “Electrical properties of Ta-doped SnO2 thin films epitaxially grown on TiO2
substrate”, Appl. Phys. Lett. 93 (2008) 132109.
[15] C. G. Fonstad and R. H Rediker, “Electrical Properties of High Quality Stannic Oxide Crystals”, Journal of Applied Physics 42,
(1971) 2911.
[16] Franz Kuhlmann, “Deposition of SnO2 thin films using reactive RF sputtering”, Master’s thesis, December 2004, pp. 47.
[17] L. O. Chua, “Memristor – The missing circuit element”, IEEE Transactions on Circuit Theory CT-18 (5), (1971), 507 – 519.
[18] A. Sawa, “Resistive switching in transition metal oxides”, Materials Today 11 (6), 2008, 28-36.
[19] Yuriy V. Pershin, et al., “Memristive model of amoeba’s learning”, Nature Precedings : hdl:10101/npre.2008.2431.1 : Posted 22
Oct 2008.
[20] Julien Borghetti, et al, “A hybrid nanomemristor/transistor logic circuit capable of self-programming”, Proceedings of the
National Academy of Science 106:6, February 10, (2009) 1699–1703.