University of Texas at El PasoDigitalCommons@UTEP

Open Access Theses & Dissertations

2017-01-01

Structural And Electrical Characterization Of TinOxide Resistive SwitchingArka TalukdarUniversity of Texas at El Paso, arka.talukdar@gmail.com

Follow this and additional works at: https://digitalcommons.utep.edu/open_etdPart of the Computer Engineering Commons, Electrical and Electronics Commons, Materials

Science and Engineering Commons, and the Mechanics of Materials Commons

This is brought to you for free and open access by DigitalCommons@UTEP. It has been accepted for inclusion in Open Access Theses & Dissertationsby an authorized administrator of DigitalCommons@UTEP. For more information, please contact lweber@utep.edu.

Recommended CitationTalukdar, Arka, "Structural And Electrical Characterization Of Tin Oxide Resistive Switching" (2017). Open Access Theses &Dissertations. 563.https://digitalcommons.utep.edu/open_etd/563

STRUCTURAL AND ELECTRICAL CHARACTERIZATION OF TIN OXIDE

RESISTIVE SWITCHING

ARKA TALUKDAR

Doctoral Program in Electrical and Computer Engineering

APPROVED:

David Zubia, Ph.D., Chair

Jose Mireles, Ph.D.

John C. McClure, Ph.D.

Charles H. Ambler, Ph.D.

Dean of the Graduate School

Copyright ©

by

Arka Talukdar

2017

Dedication

To my family and friends

STRUCTURAL AND ELECTRICAL CHARACTERIZATION OF TIN OXIDE

RESISTIVE SWITCHING

by

ARKA TALUKDAR, MS, BSE

DISSERTATION

Presented to the Faculty of the Graduate School of

The University of Texas at El Paso

in Partial Fulfillment

of the Requirements

for the Degree of

DOCTOR OF PHILOSOPHY

Department of Electrical and Computer Engineering

THE UNIVERSITY OF TEXAS AT EL PASO

December 2017

v

Acknowledgements

I would like to thank my advisors Dr David Zubia, Dr Jose Mireles and Dr John C.

McClure. I would like to extend a special that to Dr. David Zubia for his research expertise,

guidance and support.

I would like to thank all the present and past members of NanoMIL lab at UTEP for

helping me with training, support and discussion.

Finally, I would like to thank UTEP’s ORSP, Sandia National Lab and UTEP’s Electrical

and Computer Engineering department for their financial support during my PhD tenure,

vi

Abstract

Resistive switching in metal oxide is a phenomenon in which the metal oxide changes its

resistance upon application of electric field and thus giving two states; high resistance state

(HRS) and low resistance state (LRS). Many metal oxides have been investigated however very

little is known about unipolar resistive switching in SnO2 though it has shown excellent resistive

switching characteristics. Defects in the material play a vital role in resistive switching of the

metal oxides. In this work, the role of defects in resistive switching of SnO2 are investigated in

Ti/SnO2/Au structures. Two methods were used to control the concentration of defects in the

SnO2 layer. One method was by changing the concentration of O2 during reactive sputtering

deposition of SnO2, and the other was through post-deposition heat treatment in an oxygen

atmosphere.

The structural study of each of the devices was conducted using X-ray diffraction

spectroscopy (XRD) and photoluminescence spectroscopy (PL). XRD analysis revealed that

devices fabricated with low oxygen concentration have Sn particles in the deposited films while

high oxygen concentration did not have Sn crystals. The PL analysis further confirmed that the

devices fabricated with low oxygen concentration have energy states within the band gap

attributed to Sn interstitial defects.

Electrical characterization showed that resistive switching occurred only in the devices

with the Sn defects which were fabricated with low oxygen concentration. Furthermore, the

switching type is unipolar. The switching mechanism is of the filament formation type. The

current conduction mechanism in the HRS state is Ohmic at low applied electric field but

vii

transitions to Schottky emission at higher electric field strength. In contrast, the current

conduction in the LRS state is Ohmic throughout the applied voltage range.

The contribution of this work is the development of a method to study the switching

mechanism in SnOX using reactive sputtering of Sn. Through application of this technique, it was

revealed that Sn defects play a dominant role in resistive switching in SnOX via the formation of

conductive filaments.

viii

Table of Contents

Acknowledgements ..................................................................................................v

Abstract .................................................................................................................. vi

Table of Contents ................................................................................................. viii

List of Tables ...........................................................................................................x

List of Figures ........................................................................................................ xi

Chapter 1: Introduction and Motivation ..................................................................1

Chapter 2: Resistive Switching Technical Background...........................................6

2.1 Resistive Switching ................................................................................6

2.2 Metal oxide Metal Resistive Switching Mechanisms ..........................10

2.3 Comparison of RRAM Technologies ..................................................12

2.4 SnO2 Resistive Switching State-of-the-Art ..........................................15

2.5 Contribution of this Dissertation ..........................................................17

Chapter 3: Device Fabrication ...............................................................................19

3.1 Summary of Sample Preparation Conditions .......................................19

3.2 Oxidation of Si ........................................................................................20

3.3 Deposition of Cr/Au Bottom Electrode ..................................................21

3.4 Deposition of SnO2 Layer .......................................................................22

3.5 Heat Treatment........................................................................................23

Chapter 4: Structural Analysis ...............................................................................24

4.1 X-ray Diffraction Analysis .....................................................................25

4.2 Photoluminescence .................................................................................28

4.3 Summary of Structural analysis ..............................................................32

Chapter 5: Electrical Analysis ...............................................................................33

5.1 Switching Characteristics........................................................................33

5.2 Current Conduction Mechanisms ...........................................................37

5.3 Compliance Current Test ........................................................................42

5.3 Summary of Electrical Characterization .................................................45

ix

Chapter 6: SnO2-Resistive switching device Model ..............................................47

Chapter 7: Conclusion............................................................................................52

References ..............................................................................................................53

Vita 57

x

List of Tables

Table 1.Comparison between current and emerging memory technologies ................................... 5 Table 2. Comparison of resistive switching properties of metal oxides ....................................... 14 Table 3.Summarized reports on SnO2 resistive switching ............................................................ 17 Table 4. Sample deposition and heat-treatment conditions .......................................................... 19 Table 5. Summary of XRD data ................................................................................................... 27

Table 6. Summarized structural analysis ...................................................................................... 32 Table 7.Summary of electrical characterization of the devices .................................................... 46 Table 8.Summarized structural and electrical analysis ................................................................. 48

xi

List of Figures

Figure 1. Block diagram of memory architecture ........................................................................... 1 Figure 2. The different types of memory. The speed of memory decreases from top to base of the

triangle while density and cost increase from base to top of the triangle. ...................................... 3 Figure 3. MOM structure. ............................................................................................................... 7 Figure 4. Unipolar resistive switching ............................................................................................ 8

Figure 5.Bipolar resistive switching ............................................................................................... 9 Figure 6.Interface model schematic diagram ................................................................................ 10 Figure 7. Filament model schematic diagram ............................................................................... 12

Figure 8. Periodic table. The oxides of the elements used for resistive switching is marked in the

periodic table [18] ......................................................................................................................... 15 Figure 9. (Left panel) Cross-sectional diagram of the MOM structure where SnO2 is the oxide

layer and Au and Ti are top and bottom electrodes, respectively. Cr was the adhesion layer

between SiO2 and Au. (Right panel) Top image of a wafer containing the MOM devices. ......... 20 Figure 10. Schematic diagram of the sputtering system. .............................................................. 21

Figure 11.Schematic diagram of Evaporation system. ................................................................. 22 Figure 12. Structural analysis experimental design ...................................................................... 24 Figure 13. XRD patter of SnO2/Au/Cr/SiO2/Si structure.............................................................. 26

Figure 14. The XRD data to the heat treated analysis. 12%-O2 was as-grown, HT-200 was

annealed at 200ᴼC, and HT-300 was annealed at 300ᴼC .............................................................. 27 Figure 15.Different defect levels within band gap of SnO2. ......................................................... 29

Figure 16. PL spectra of SnOx deposited at different O2 flow. .................................................... 30

Figure 17. Temperature-dependent PL study of individual 12% sample heated to several

temperatures .................................................................................................................................. 31 Figure 18. PL data of the heat treated samples compared to 12%-O2 as-grown. ......................... 32

Figure 19. I-V characteristics of the (a) 12%-RT, (b) 18%-RT and (c) 12%-200HT devices. The

red traces are the forming stages, the blue are the HRS and the green are the LRS states. .......... 35 Figure 20. The endurance of the (a) 12%-RT, (b) 18%-RT and (c) 12%-200HT devices.

Endurance measured at 0.04 V. .................................................................................................... 36 Figure 21.I-V characteristics of the (a) 30%-RT and (b) 12%-300HT devices. These devices did

not switch. ..................................................................................................................................... 37

Figure 25. The LRS analysis of 12%-RT device showing Ohmic conduction with a slope of 0.97.

....................................................................................................................................................... 41 Figure 26.The LRS analysis of 18%-RT device showing Ohmic conduction with a slope of 1.02

....................................................................................................................................................... 41 Figure 27.The LRS analysis 12%-200HT device showing Ohmic conduction with a slope of 0.97

....................................................................................................................................................... 42 Figure 28. Compliance current (CC) study of the 12%-RT device. High compliance require

higher re-set voltage ...................................................................................................................... 43 Figure 29. Compliance current study of the 18%-RT device. High compliance require higher re-

set voltage ..................................................................................................................................... 44 Figure 30. Compliance current study of the 12%-200HT device. High compliance require did not

change re-set voltage..................................................................................................................... 45

xii

Figure 31. Ohmic conduction at low HRS voltage. Electrons are generated due to ionization of

Sn atom. Sn+ flows towards the negative electrode (Au) while electrons flow towards the

positive (Ti) ................................................................................................................................... 49 Figure 32. Schottky conduction. ϕ is the barrier height that electrons have to overcome. The Sn+

continues to flow towards the negative electrode, i.e. Au ............................................................ 50 Figure 33.At reset voltage filament is formed and the electron conduction takes place through the

metallic filament. The devices switched to LRS .......................................................................... 51 Figure 34. At re-set voltage filament ruptures indicated by red-cross and device goes back to

HRS ............................................................................................................................................... 51

1

Chapter 1: Introduction and Motivation

The semiconductor memories play a vital role in storing and processing data in a vast

amount of electronic circuits used in society. Furthermore, the role that they play can be further

refined depending on the application of the electronic circuit and the needs of the data processor

within an electronic circuit. Essential characteristics of memories are; the physical size and

density, write and erase speeds, power consumption, and fabrication cost. The amount of

physical space depends on the size of each memory cell and how many bytes of storage are

required. Different types of memory technologies have a range of erasing and write times, and

the speed that is required depends heavily on the application of the electronic circuit. Power

consumption is also a factor that has significantly increased in importance as the densities of

memories and numbers of portable applications have increased over time. Finally, the cost of

fabrication is also an important economic consideration. For example, the demand for affordable

and more functional mobile cellphones has dramatically increased the importance of low-power

and low-cost memories.

FIGURE 1. BLOCK DIAGRAM OF MEMORY ARCHITECTURE

In digital electronic circuits that have a data processing unit, memory can be divided into

three categories depending primarily on how the processor accesses, uses and communicates

with the different type of memories as shown in Figure 1 which illustrates a typical memory

architecture. The three memory categories are the cache, the main, and the secondary memory.

processor

cach

e

Main memorySecondary memory

data data

2

The cache memory is embedded inside the central processing unit (CPU). The cache memory is a

very high-speed memory and is used to store data and programs that the processor uses very

frequently. These are volatile memory, so the data are lost when power is turned off. The leading

technology used for cache memory is static random access memory (SRAM). The main memory

is slower than the cache memory and is used to store information which the processor is

currently working on but doesn’t use as frequently as the data in the cache memory. The main

memory is also volatile memory similar to the cache memory. The main memory interacts with

the CPU, the cache memory, and the secondary memory. The dominant technology used for

main memory is dynamic random-access memory (DRAM). The secondary memories are non-

volatile which means that is data is not lost when power is turned off. This type of memory is

much slower than the main memory. The processor cannot access data directly from secondary

memory. Instead, the data needs to be loaded into the main memory from where the processor

can access data. The technologies used for secondary memories are magnetic, optical, and solid-

state. Examples of secondary memory technologies are magnetic hard disk drives, optical

compact disks, and solid-state flash drives.

3

FIGURE 2. THE DIFFERENT TYPES OF MEMORY. THE SPEED OF MEMORY DECREASES FROM TOP TO

BASE OF THE TRIANGLE WHILE DENSITY AND COST INCREASE FROM BASE TO TOP OF THE TRIANGLE.

Figure 2 describes how cost, density, and speed vary among the 3 types of memory. The

cache memories have the highest speed among all the memories, but it is very costly and has

very low density due to its large cell size. A cache memory unit cell is made up of six transistors

(6T) and a minimum cell size of 140 F2 where F is the feature size usually expressed in

nanometers. The SRAM technology has the highest speed, but the drawbacks are density and the

cost. The main memory is slower compared to the cache, but it has lower cost and higher density.

The DRAM technology which is used to make main memory is made up of only one transistor

and one capacitor (1T1C) with a compact unit cell size of 6F2. The most commonly used

secondary memory is the magnetic hard disk. The magnetic disk is relatively very slow and

requires high power compared to the other memories. However, it provides high density and is

very affordable. To reduce power consumption and increase speed, magnetic memory has been

increasingly replaced by flash technology within the last 20 years. Flash technology uses one

transistor and one resistor (1T1R) architecture. The power consumption by the flash cell is lower

than DRAM cell. To have energy efficient electronic devices flash technology was also used in

cache

Main memory

Secondary memory

speeddensity

cost

4

main memories, but they have low density and were slower than DRAM technology. summarizes

and compares the current and emerging memory technologies based on density, speed,

endurance, and application. [1]

A current focus in memory technology development is to integrate main with secondary

memory. To address this challenge several new technologies are emerging as listed in Table 1.

The emerging technologies listed in Table 1 offer the potential for replacing and/or integrating

main and secondary memories but not the cache due to their relatively slower speed compared to

the SRAM. A key to integrating main with secondary memory is to replace DRAM with a

technology that has a higher density, faster erase and write speeds (less than 10 ns), lower power

consumption and lower cost compared to DRAM. The technologies that are emerging are

magnetic random access memory (MRAM), phase change random access memory (PRAM) and

the resistive-switching random access memory (RRAM) as listed in Table 1. The MRAM has

good potential to be the next generation secondary memory since it has good erase and write

speeds. However, since it has a lower density than DRAM, it cannot be used as main memory.

On the other hand, PRAM has a higher density than DRAM but its erase and write speeds are

slower than DRAM technology; therefore it is an excellent candidate to be used in secondary

memory but not main memory. In comparison, the RRAM offers both higher speed and higher

density compared to DRAM. Other positive attributes of RRAM are; low fabrication cost, low

power consumption and non-volatility. However, one drawback is the endurance is lower

compared to the DRAM. Since RRAM is a potential candidate to be the next generation main

and second memory we have focused our research on RRAM technology in this work.

However, in order to implement any new technology, we need to know how the devices

work and what type of material systems and structures will be easy to implement in a cost-

5

effective manner. Three main challenges faced by RRAM before they can be inserted into

CMOS are; (1) the development of an accurate and predictive model of the fundamental

switching mechanism, (2) the most suitable device structure to yield high density with low

manufacturing cost, (3) and the development of electrical characteristics suitable for CMOS

circuit applications.

Table 1.Comparison between current and emerging memory technologies

Type Cell

elements

Minimum

size (F2)

Erase

speed (ns)

Write

speed (ns)

Endurance

(cycles)

Memory

Application

Current technologies

SRAM 6T 140 F2 0.3 0.3 >1016 Cache

DRAM 1T1C 6F2 10 10 >1016 Main

NOR-Flash 1T 10F2 105 107 >105 Secondary

NAND-Flash 1T 5F2 106 105 >105 Secondary

Emerging technologies

MRAM 1T1R 20F2 10 10 >1016 Secondary

PRAM 1T1R 5F2 5 5 108 Secondary

RRAM 1T1R 4F2 5 5 >1012 Main and

Secondary

6

Chapter 2: Resistive Switching Technical Background

2.1 Resistive Switching

Resistive switching is a phenomenon in which the resistance in a device or material

structure abruptly changes by several orders of magnitude upon application of an electric field.

The change in resistance can be highly reversible, and this has motivated much research to use

resistive switching for memory applications. These memory devices are called resistive RAM

(RRAM). Many different materials and material structures exhibit resistive switching; however,

the metal-oxide-metal (MOM) structure shown in Figure 3 has received the most attention due to

its simplicity, ease of fabrication and compatibility with CMOS. Devices based on the MOM

structure are two terminal devices where a dielectric material is sandwiched between two metal

electrodes. The dielectric layer consists of metal oxide. The resistance across the MOM structure

changes in response to the application of voltage and gives rise to two distinct resistance states; a

high resistance state (HRS) and a low resistance state (LRS).

The voltage at which the devices transition from HRS to LRS is known as the set voltage.

In contrast, the voltage at which the device goes from LRS to HRS is known as the reset voltage.

Resistive switching can be classified into two broad categories depending on the polarity of set

and reset voltages required to cause the device to change in resistance state. Resistive switching

behavior is called unipolar when only one polarity is needed to switch the device. In this case,

the set and reset voltages have the same polarity. In contrast, bipolar switching requires the set

and reset voltage to have opposite polarities.

7

FIGURE 3. MOM STRUCTURE.

2.1.1. Unipolar

Figure 4 shows a typical current-voltage characteristic of unipolar switching. High-

quality metal-oxides are typically good insulators with very high resistivity. They usually do not

show any change in resistance when subjected to voltages below a critical value. However, if the

voltage is high enough, a physical change occurs in the MOM structure which causes it to

suddenly start to conduct significant electric current. This is called electroforming. If left

unchecked, the high electric current dissipates a high degree of joule heating that can cause

permanent damage to the materials. In the past, the critical value was described as the dielectric

breakdown voltage, and care was taken not to exceed it. However, it is now well known that

damage can be prevented if an external circuit is used to limit the electric current after the

dielectric material switches to LRS. The value to which the external circuit limits the current is

called the “compliance current”. Current compliance is normally achieved by quickly reducing

the applied voltage when the device switches to LRS as shown by the blue trace in Figure 4.

To reset the device, a voltage is applied again with the same polarity and this causes the

MOM structure to now suddenly switch to HRS as illustrated by the red trace in Figure 4.

Because the compliance current prevents damage to the dielectric, the process is reversible.

Much research has been performed to study resistive switching. Although models have been

proposed, many details of the switching behavior have yet to be fully understood.

Metal

Oxide

Metal

8

FIGURE 4. UNIPOLAR RESISTIVE SWITCHING

2.1.2. Bipolar

As stated above, bipolar resistive switching requires the set and reset voltages to have

opposite polarity as shown by the I-V characteristic in Figure 5. One interesting attribute of

bipolar resistive switching is that the I-V curves pass through the origin. This implies that there

is no charge or energy stored in the device and it is a purely resistive device. However, due to the

transitioning between HRS and LRS, the I-V characteristic curve has a hysteresis form as shown

in Figure 5 which has been described as a “pinched hysteresis” or “bowtie” characteristic. One

disadvantage of bipolar switching compared to unipolar is that two power supplies are needed to

reverse the applied electric field. Nevertheless, much of the research has focused on bipolar

switching due to the superior characteristics of the bipolar devices compared to unipolar

Set

Re-set

Voltage

Cu

rre

nt

9

FIGURE 5.BIPOLAR RESISTIVE SWITCHING

Set

Re-set

-Voltage

Cu

rre

nt

+Voltage

10

2.2 Metal oxide Metal Resistive Switching Mechanisms

Two primary mechanisms have been proposed that give rise to resistive switching in

MOM structures. In one mechanism, the interface between a metal electrode and the metal-oxide

plays the dominant role. In the other mechanism, the creation of conductive filaments in the bulk

of the metal-oxide governs switching.

Figure 6 describes the interfacial switching mechanism. In this mechanism, the electrode

plays a vital role. The electrode reduces the metal oxide to metal ions or oxygen vacancies

indicated by the white circles in Figure 6. The switching mechanism occurs due to movement of

this oxygen vacancies or metal ions. No filament is formed in this type of switching. The

mechanism is localized to the interface, the HRS and LRS resistances (and respective currents)

are proportional to the cross-sectional area of the interface. So in that sense, the mechanism is

planar in nature. Interestingly, the interface mechanism gives rise to only bipolar type I-V

characteristics but not unipolar. MOM structures composed of perovskite oxides [2] [3] tend to

manifest the interface switching mechanism.

FIGURE 6.INTERFACE MODEL SCHEMATIC DIAGRAM

Figure 7 shows the fundamentals of the filament formation mechanism. In this

mechanism, a conductive filament is formed within the metal-oxide which bridges the two metal

electrodes when an electric field is applied to the MOM structure. In the virgin state, the metal-

Metal

Oxide

Metal

OxygenVacancies

OrMetal ions

11

oxide is homogeneous as shown in Figure 7(a) and is initially in the HRS. However, when an

electric field is applied, conductive filaments start to grow in the metal oxide. When one of the

filaments bridges completely across the two metal electrodes as shown in Figure 7(b), the MOM

structure transitions to the LRS. In order to transition to the HRS state again, current is flowed

through the filament producing joule heating to rupture the filament. Rupture of the filament

causes the MOM structure to transition to the HRS state as shown in Figure 7(c). The process of

transitioning between HRS and LRS is highly repeatable.

The filament formation mechanism depends strongly on the materials used for the metal-

oxide and metal electrodes layers. Moreover, the thickness of the metal-oxide layer is also a

crucial parameter. If the metal-oxide is too thick, resistive switching is not observed because the

electric field is too low and distance between metal electrodes is too long. On the other hand, if

the metal oxide layer is too thin, pinholes become a concern. The minimum cell size in any of the

devices is still unknown but using 3D architecture very high-density memory has been achieved.

A company named Crossbar showed that RRAM could be organised in 3D cross-point array

achieving terabytes of memory cell in a single die. [4] The two leading theories behind filament

formation are electrochemical metallization and oxygen ion defects and are described below.

In the electrochemical metallization model, the metal-oxide layer is sandwiched between

an electrochemically active metal and an inert metal. When a voltage is applied between the

electrodes, the electrochemically active metal is oxidized at the interface and form metal cations.

The metal cations then migrate towards the inert electrode forming a metallic path between the

two electrodes thus switching the device into LRS. Due to high current at the LRS enough joules

heat is produced to fuse the filament and thus bringing the device to HRS. [5] [6] [7]

In the oxygen ion defects model, when the device is under an applied voltage, oxygen

ions from the metal oxide layer migrate towards the anode, and oxygen vacancies migrate

towards the cathode. The oxygen ions are discharged as gas bubbles into the environment.

However, the oxygen vacancies which are metal-rich MOx form the conductive filaments

between the electrodes. [8] [9] [10]

12

FIGURE 7. FILAMENT MODEL SCHEMATIC DIAGRAM

The interface and filament formation models give a good working framework to study

resistive switching in MOM structures. However many details of the switching mechanism are

still not well understood. The filament and interface models indicate that defects play a vital role

in the resistive switching phenomenon through the ideal concentrations of defect density are yet

to be determined. In filament formation switching, filaments start growing randomly and cause

the set and reset voltages vary from device-to-device and even from cycle-to-cycle. Lastly, the

properties of the materials themselves have a strong influence. In the next section, the different

material systems that have been explored for RRAMs are surveyed.

2.3 Comparison of RRAM Technologies

The performance of current technologies listed in Table 1 can be used as benchmarks for

their replacement. Based on these criteria, RRAM technologies have the potential to emerge as

replacements for main and secondary memories. For example, RRAMs have demonstrated high

density with their MOM structure. They also have excellent erase and write speeds (0.3 ns).

However, one parameter in which RRAMs are lagging is in endurance with the highest reported

being 1012 cycles compared to 1016 cycles for DRAM. Moreover, other important parameters not

Metal

Oxide

Metal

Ions or vacancies

Formed filament

Metal

Oxide

Metal

Metal

Oxide

Metal

Ions or vacancies

Ruptured filament

Virgin state LRS HRS

13

list in Table 1 are the on/off ratio and the set/reset voltages. Importantly, a material system and

MOM structure has yet to be developed which surpasses all the CMOS memory benchmarks. In

this section, we explore the various material systems that have been investigated for RRAM and

the type of resistive switching demonstrated with the different technologies.

Figure 8 highlights 20 metals that have been studied as metal-oxides for resistive

switching. Among the 20 metals marked in the periodic table, eight have shown solely bipolar

switching while the other nine have shown both bipolar and unipolar switching. The oxides that

have shown exclusive bipolar switching are Mg, Mn, Fe, Co, Zn, Zr, Nb, and Mo. The remaining

oxides that have shown both bipolar and unipolar behavior are Ti, Hf, Ta, W, Al, Ni, Cu, Cr, and

Sn. Of all these materials, most of the research has focused on TiOx, TaOx, HfOx, AlOx, SiOx,

CuO2 and NiOx. [11] [12] [13] [14] This is presumably due to the fact that these materials are

commonly used for the fabrication of CMOS which is the most advanced technology in the

world.

Table 2 compares the performance of several material systems that have received a high

degree of attention. All the materials on this list showed both unipolar and bipolar switching

characteristics. In general, the bipolar devices are superior to the unipolar for all the material

systems in the list. This has caused most of the research to focus on bipolar switching although

two voltage polarities are required for switching.

TiOx was one the earliest materials studied and showed great promise in terms of

switching speeds. However, the highest endurance is only 106 which is far less than the 1016

benchmark for DRAM. Moreover, the On/Off ratio is 105 which is less than the 106 needed to

limit leakage off-current to acceptable levels. The bipolar device demonstrates superior

characteristics compared to the unipolar device.

14

More recently, TaOx was studied and showed excellent resistive switching behavior. In

fact, bipolar TaOx devices have demonstrated the highest On/Off ratios (106) and endurance

(1012 cycles) of all the MOMs studied [15]. However, a significant drawback of bipolar TaOx is

the high DC peak voltage.

HfOx has also shown excellent resistive switching behavior with a superior speed of 0.3

ns which is comparable to SRAM used for cache memory. [16] However the On/Off ratio and

endurance still need improvement to replace main or secondary memory.

Lastly, nickel-oxide also shows promise for RRAMs. However, the performance of the

bipolar NiOx device lags behind TaOx and HfOx. Among the unipolar devices, however, NiO has

the best endurance and On/Off ratio. It has been seen that NiO unipolar resistive switching

devices are not reliable and the resistive switching depends on the choice and size of electrodes

[17].

Table 2. Comparison of resistive switching properties of metal oxides

Metal

oxide

Mode of

switching

On/OFF

ratio

Write speed

(ns)

Erase

speed (ns)

DC peak

voltage (V)

Endurance

(cycles)

TiOx Unipolar >102 Unknown Unknown Unknown <105

Bipolar >105 5 10 1 106

TaOx Unipolar >103 100 100 3 Unknown

Bipolar >109 50 50 6 >1012

HfOx Unipolar >103 Unknown Unknown Unknown >103

Bipolar >105 0.3 0.3 2.5 >1010

NiO Unipolar >103 180 180 2 >103

NiOx Bipolar >106 10 20 Unknown >106

15

FIGURE 8. PERIODIC TABLE. THE OXIDES OF THE ELEMENTS USED FOR RESISTIVE SWITCHING IS

MARKED IN THE PERIODIC TABLE [18]

2.4 SnO2 Resistive Switching State-of-the-Art

Among the 20 metals mentioned in Figure 8, Sn has received very little research

attention. This is presumably to due to the fact that SnOx is a conductive metal-oxide and not

used in CMOS manufacturing. Instead, SnOx has been used for gas sensors and as a transparent

conductive layer in solar cells. Nevertheless, SnOx possesses many interesting semiconductor

properties that make it an attractive material for RRAM applications. In this section, the material

properties of SnOx are presented first, followed by its resistive switching performance. [19]

16

2.4.1. SnOx as Semiconductor Material

The SnOx has proved to be an excellent semiconductor material. It has shown both n and

p-type behavior, have a wide band gap (2.2-3.6 eV) and a resistivity rage between 10-5 to 106

Ω.cm. [19] [20].Importantly, this wide range in resistivity was achieved with magnetron reactive

sputtering from a Sn target in an O2-Ar atmosphere by varying the percent oxygen. SnO2 is

regarded as n-type while SnO (oxygen deficient oxide) is a p-type semiconductor [21] [22].

Presence of defects in SnOx depends on the deposition condition of SnOx [23] and the types of

defects in SnOx determine resistivity [24]. It has demonstrated a high resistance to displacement

damage from neutron radiation. [25]. All the above properties indicated that SnOx is a promising

candidate for resistive switching behavior.

2.4.2. Resistive Switching Behavior of SnOx

SnOx has shown unipolar as well as bipolar resistive switching behavior. The first

unipolar resistive switching behavour iour was observed by K. Nagashima, et,al. in which SnO2

was sandwiched between two Pt electrodes. The SnO2 layer was deposited on a Pt/Ti/Si substrate

by pulse lesser deposition method using SnO2 pallets. [26] The On/Off ratio was ~102 and the

switching behavior was independent of the electrode material. The forming set and reset voltages

were 5V, 1.5-2 V and 0.5-1 V respectively. The LRS current analysis showed Ohmic behavior

and HRS current analysis indicated that current conduction was dominated by the Pool-Frankel

mechanism.

Sergio Almeida, et.al., observed resistive switching behavior on Ag paste/SnO2/Ti/glass

MOM structures [27]. The SnO2 was deposited by reactive magnetron sputtering using a Sn

target in a 100% O2 atmosphere on a Ti/glass substrate. The Ag paste was painted on the SnO2

surface. The On/Off ratio was 10 and the forming, set and reset voltages were 1.96 V, 0.8 and

1.1V, respectively. LRS current analysis showed Ohmic behavior.

A. Talukdar et,al. observed unipolar resistive switching behavior in Ti/SnO2/Au/Cr/SiO2

structure [28]. The SnO2 was deposited by reactive magnetron sputtering using a Sn target in a

17

100% O2 atmosphere. The On/Off ratio was ~103 and the set, reset voltages were .8-1.2 V and

0.3-0.6 V, respectively. The LRS current was Ohmic while the Pool-Frankel conduction

dominated the HRS.

Bipolar resistive switching behavior was observed by Jindong Jin, et.al. in a

Al/SnO2/Pt/glass structure where SnO2 was deposited by DC sputtering using a SnO target in an

Ar atmosphere. [29] The On/Off ratio was 102 and the set, reset voltages were +5 and -5.5 V,

respectively. To study resistive switching behavior, annealing was performed in O2 and N2

atmospheres. X-ray photoelectron spectroscopy (XPS) data indicated that the sample annealed in

O2 had fewer oxygen defects than the one annealed in N2 atmosphere. A significant finding was

that the devices with fewer defects did not switch indicating the critical role of oxygen vacancies

in resistive switching. Table 3 summarises the reports on SnO2 resistive switching devices

Table 3.Summarized reports on SnO2 resistive switching

Lead Author

and Institution

Switching

Type

Set

Voltage

(V)

Reset

Voltage

(V)

On/Off

Ratio SnO2 Deposition Structure

Nagashima, Unipolar 1.5-2 0.5-1 <102 Pulse laser using

SnO2 pallets

Pt/SnO2/Pt/Ti/Si

Almedia,

UT El Paso

Unipolar 1.1 0.8 10 RF sputtering with

Sn target

Ag/SnO2/Ti/glass

Talukdar,

UT El Paso

Unipolar 0.8-1.2 0.3-0.6 103 RF sputtering with

Sn target

Ti/SnO2/Au/Cr/SiO2/Si

Jin, Bipolar +5 -5 102 RF sputtering with

SnO target

Al/SnO2/Pt/glass

2.5 Contribution of this Dissertation

To date, only four reports have been published regarding resistive switching of SnO2. In

three of the reports, the switching was unipolar while one reported bipolar switching. Pool-

Frankel current conduction was observed in two of the structures indicating the presence of deep

level energy states in SnO2. Heat treatments in O2 and N2 atmospheres revealed that devices with

fewer defects did not switch indicating that oxygen vacancies play an important role in resistive

18

switching. However, a detailed model of the mechanism for unipolar resistive switching is

lacking. In this dissertation, research is focused on unipolar resistive switching behavior of SnO2.

Reactive sputtering of Sn in a variable O2-Ar atmosphere is used to control the level of oxidation

and defect concentration in the films. Moreover, the layers were heat-treated in an oxygen

environment after deposition also to affect the defect concentration. X-ray diffraction,

Photoluminescence, and Raman spectroscopy are used to study the structural characteristics.

This is complemented with detail current-voltage characterization. Together, these data are

combined to propose a model for resistive switching in the devices.

19

Chapter 3: Device Fabrication

3.1 Summary of Sample Preparation Conditions

Two primary methods were used to control the concentration of defects in the SnO2 layer

of the MOM structure. One method was through the amount of O2 in the deposition chamber

during reactive sputtering of Sn. The second method was through post-deposition heat-treatment

of the SnO2 layer in a tube furnace in an O2 atmosphere. Table 4 shows the conditions used to

during deposition and heat-treatment of the samples. In general, four O2 concentrations were

used during deposition consisting of 12%, 18%, 30% and 100%. Two samples where were

deposited in a 12% O2 atmosphere, were heat-treated for 20 minutes at temperatures of 200C and

300C, respectively. During the deposition of each sample, a constant pressure of 2.4 mTorr was

maintained. The power used for each deposition was 40 Watts. Since the deposition rate

depended on the condition, the deposition times were adjusted to maintain a constant thickness

between 45 to 50 nm for all the samples.

Table 4. Sample deposition and heat-treatment conditions

Sample % O2 % Ar Ar-flow

(sccm)

O2 flow

(sccm)

Time

(mins)

Temperature

ᵒC

12%-RT 12 88 10.5 1.6 25 NA

12%-200HT 12 88 10.5 1.6 25 200

12%-300HT 12 88 10.5 1.6 25 300

18%-RT 18 82 9.5 2.5 23 NA

30%-RT 30 70 8.3 3.7 21 NA

100%-RT 100 0 0 50 28 NA

Figure 9 shows a cross-sectional schematic and a top view image of the MOM devices.

The starting substrate consists of a 3-inch diameter silicon wafer. The silicon wafer is first

oxidized to create an insulating SiO2 layer. Chromium and gold were then deposited to form the

20

bottom electrode of the MOM structure. The SnO2 layer is then deposited using reactive

sputtering in an oxygen atmosphere. A shadow mask is used to protect the outer areas of the gold

layer, so it can be contacted as a bottom electrode during electrical testing. Two wafers with

SnO2 were heat treated in an oxygen atmosphere to investigate the effect of oxidation on

resistive switching behavior. Finally, an array of circular titanium top electrodes are sputter

deposited through a shadow mask with holes as shown in Figure 9.

FIGURE 9. (LEFT PANEL) CROSS-SECTIONAL DIAGRAM OF THE MOM STRUCTURE WHERE SNO2 IS

THE OXIDE LAYER AND AU AND TI ARE TOP AND BOTTOM ELECTRODES, RESPECTIVELY. CR WAS

THE ADHESION LAYER BETWEEN SIO2 AND AU. (RIGHT PANEL) TOP IMAGE OF A WAFER

CONTAINING THE MOM DEVICES.

3.2 Oxidation of Si

The first step of the device fabrication was dry-oxidation of p-type Si wafer. The p-type

Si wafer was oxidized in a tube furnace at a temperature of 1100ᴼC for 3 hrs under an O2 flow of

60 sccm. The thickness of the SiO2 was measured using a Filmetrics F20 interferometer thin

film analyzer and was between 330 nm-350 nm.

Ti

SnO2

AuCr

21

3.3 Deposition of Cr/Au Bottom Electrode

All the layers of the MOM structure were deposited in a Kurt J. Lester vacuum depositor

with sputtering and thermal evaporation capabilities. The first layer to be deposited was

chromium which acts as a wetting layer for gold. Due to low binding energy between Au and

oxygen of SiO2, gold does not have good adhesion to SiO2. On the other hand the binding energy

between chromium and oxygen of SiO2 is high so it was used as an intermediate layer to promote

adhesion of the gold. [30] The chromium layer was deposited using magnetron sputtering at

room temperature. Figure 10 is the schematic diagram of a typical magnetron sputtering

deposition chamber. It is a type of plasma vapor deposition in which the materials that have to be

deposited is used as the negative electrode or the target and an inert gas is used to create the

plasma. To deposit the wetting layer a 99.99% pure Cr was used as the target. Ar gas was used to

create the plasma. The flow of Ar gas was 40 sccm and a constant pressure of 5 mTorr was

maintained during deposition. The power used for deposition was 40 Watts. The deposition time

was 12 minutes which gave Cr thickness of 120-140 nm.

FIGURE 10. SCHEMATIC DIAGRAM OF THE SPUTTERING SYSTEM.

Substrate or wafer

Sputtering Target

Ar+

Sputtering gas

Exhaust

Target material

Ar+

22

Thermal evaporation was used to deposit the gold layer which acts as the bottom

electrode of the device. Figure 11 shows an evaporation system where the material pellets are

being heated by a DC power supply. Thermal evaporation is a type of physical vapour deposition

in which the material that has to be deposited is heated in a vacuum chamber until the atoms

have enough thermal energy to leave the surface. To deposit gold, 99.5% pure gold pellets were

heated to a temperature above 1200ᵒC. The deposition pressure of 10-6 mTorr was maintained

using a cryopump. The distance between the substrate and the material was 15 feet and

deposition time was 5 minutes. The thickness of the deposited gold was 100-130 nm.

FIGURE 11.SCHEMATIC DIAGRAM OF EVAPORATION SYSTEM.

3.4 Deposition of SnO2 Layer

Reactive magnetron sputtering was used to deposit the SnOx layer at room temperature.

The target material was 99.999% pure tin (Sn) while the atmosphere was a mixture of Ar and O2.

The Ar is inert, but O2 is reactive with Sn. Tin oxide has two stable oxides; SnO and SnO2. It has

been shown that by changing the concentration of O2, different concentration of Sn, SnO and

SnO2 can be found in the deposited film [31]. As discussed in the previous chapter, resistive

Substrate or wafer

Material

Power Supply

23

switching appears to depend on the concentration of energy states within the bandgap of the

metal-oxide. It is also well known that interstitial Sn and oxygen vacancies have energy levels

within the bandgap [32]. Therefore, by varying the O2 concentration during reactive sputtering, it

is anticipated that the concentration of energy states within the bandgap of SnOx can be

controlled.

Four different O2 concentrations were used during reactive sputter of SnOx as mentioned

in Table 4. The deposition power was 40 Watts and the deposition pressure was 2.4 mTorr. To

keep the deposition pressure constant the flow of the Ar was varied as shown in Table 4. The

deposition rate differed under different O2-Ar concentration. The time of deposition was varied

to keep the thickness between 45-50 nm for all the samples as shown in Table 4.

3.5 Heat Treatment

The heat treatment was the second method used to control the defects in SnOx. Heat

treatment is a well-known process for reducing defects in a material [33]. Two samples with the

SnOx deposited with an oxygen pressure of 12% were heat traded in a tube furnace with an O2

atmosphere at 200ᴼC and 300ᴼC for 20 minutes respectively. The temperature was kept low so

that no diffusion of Au or Cr into SnO2 occurred.

24

Chapter 4: Structural Analysis

Structural analysis was performed to determine the effect of the deposition and heat-

treatment conditions on the type and concentration of defects within the SnOX layer. X-ray

diffraction analysis (XRD) was performed to determine the crystalline structure and composition

of the films. Photoluminescence (PL) was performed to probe energy states within the bandgap

of SnO2 molecules.

Figure 12 is a graphical representation of six different samples prepared for systematic

study as shown in Table 4. Three samples were deposited at three distinct oxygen concentrations;

12%, 18%, and 30%. These samples did not receive a post-deposition heat-treatment and are

called the “as-grown” samples. The other three samples were all deposited at 12% oxygen

concentration but were heat-treated at room-temperature, 200 °C and 300 °C. These are the

“heat-treated” samples

FIGURE 12. STRUCTURAL ANALYSIS EXPERIMENTAL DESIGN

Structural Analysis

Heat treated

12%-RT

12%-200HT

12%-300HT

As-grown devices

12%-RT

18%-RT

30%-RT

25

4.1 X-ray Diffraction Analysis

XRD is a non-destructive technique that gives information about crystal structure and

composition of thin films. The Bruker Advance D8 X-Ray Diffractometer was used for 2θ scan

of the samples. The scan speed was 3 angles per minute. In order to reduce noise multiple scans

were done on each sample.

Figure 13 shows the XRD pattern of the three as-grown samples. Peaks corresponding to

Au and SiO2 from the substrate are observed in all the samples as expected. Moreover, all the

samples have SnO and SnO2 peaks indicating that the presence of oxygen during deposition was

effective in reacting and oxidizing Sn. However, the presence of both SnO and SnO2 peaks

indicates that reactive sputtering using Sn target produced a non-stoichiometric film. SnO is an

oxide of Sn with an oxidation state of +2 and SnO2 has an oxidation state of +4.

In contrast, a peak corresponding to elemental Sn (211) is only present in the 12% and

18% samples but not the 30% sample. This indicates that the amount of oxygen present in the

12% and 18% samples were insufficient to oxidize the Sn completely. The presence of Sn

particles is attributed to incomplete oxidization of Sn during the deposition process. On the other

hand, no Sn(211) peak was observed in the 30% sample indicating that it was more thoroughly

oxidized.

26

FIGURE 13. XRD PATTER OF SNO2/AU/CR/SIO2/SI STRUCTURE

Figure 14 compares the XRD data of the heat-treated samples. Similar to the as-grown

samples, peaks corresponding to Au and SiO2 from the substrate are observed in all the samples

as expected. Moreover, all the samples have peaks corresponding to SnO and SnO2. A

differentiation occurred however in the presence of the Sn (211) peak. The Sn peak was observed

in room temperature (RT) and 200 °C samples but not in the 300 °C sample. Similar to the as-

grown samples, this is interpreted as the 12%-RT and HT-200 samples being incompletely

oxidized and HT-300 sample being completely oxidized. A final observation is that the SnO2

peaks in HT-300 sample are relatively more intense compared to the SnO peaks.

30%-RT

18%-RT

12%-RT

27

FIGURE 14. THE XRD DATA TO THE HEAT TREATED ANALYSIS. 12%-O2 WAS AS-GROWN, HT-200

WAS ANNEALED AT 200ᴼC, AND HT-300 WAS ANNEALED AT 300ᴼC

The Table 5 summarized the XRD data. The as-grown XRD data showed that as O2

percentage was increased during deposition, Sn particle concentration is reduced. The heat

treatment analysis also showed the same result. The high concentration of O2 during deposition

and heat treatment both oxidize the Sn particles present in the deposited thin film.

Table 5. Summary of XRD data

The as-grown analysis The heat-treated analysis

XRD peaks 12%-RT 18%-RT 30%-RT 12%-RT 12%-200HT 12%-300 HT

Sn (211) Present Present Absent Present Present Absent

SnO Present Present Present Present Present Present

SnO2 Present Present Present Present Present Present

Au Present Present Present Present Present Present

SiO2 Present Present Present Present Present Present

Au

(20

0)

Sn(2

20

)

Sn(2

11

)

SnO

(11

0)

Au

(11

1)

SiO

2(2

20

)

SnO

(21

1)

SiO

2(2

20

)

SnO

2(3

10

)Sn

02(3

01

)

Inte

nsi

ty (

a.u

)

2θ

12%-RT

12%-200HT

12%-300HT

SnO

2(1

01

)

28

4.2 Photoluminescence

The photoluminescence (PL) spectroscopy is an effective technique to study energy states

within the bandgap of semiconductors. The energy states are usually attributed to structural

defects in the semiconductor. In essence, PL can be used to probe for the presence of structural

defects in a semiconductor by observing the luminescence at different energy levels. Figure 15

contains a schematic diagram of the energy levels in SnO2 attributed to three types of structural

defects; interstitial Sn, singly-ionized oxygen vacancies and, doubly-ionized oxygen vacancies.

The energy state at 1.65-2 eV is associated with interstitial Sn in SnO2. The energy state at 2.5-

2.75 eV is attributed to singly-ionized oxygen vacancies. Finally, the energy state at 3.09 eV is

attributed to doubly ionized-oxygen vacancies. [32] [34]. [35]

In this study, a Horiba LABRAM-HR Evolution spectrometer was used. The

accumulation time for each data point was 15 seconds, and the intensity of the laser was 10%.

For each sample the experiment was repeated five times to reduce noise. However, the photon

energy of the laser light source was 2.33 eV (532 nm), which limited excitation of only the

lowest energy state (1.65-2 eV) due to interstitial Sn. The photon energy of the laser source was

too low to excite the states associated with the singly- and doubly-ionized oxygen vacancies.

Nevertheless, useful information regarding the concentration of the interstitial Sn defect was

obtained as described below.

29

FIGURE 15.DIFFERENT DEFECT LEVELS WITHIN BAND GAP OF SNO2.

Figure 16 shows the PL spectra of the O2 as-grown samples deposited with 12%, 18%,

and 30% oxygen concentration. The sample with 12% oxygen has an emission peak centred at

2.05 eV which corresponds to the interstitial Sn defect in Figure 15. In contrast, the 18% and

30% samples do not manifest a peak associated with the Sn defect. Instead the 18% and 30%

samples show peaks at energies between 2.2 eV and 2.3 eV.

~3.6 eV

~1.65 eV

Valance band

Conduction band

~2 eV

~2 .5eV

~2 .75eV

~3.09eV

~2.8 eV

30

FIGURE 16. PL SPECTRA OF SNOX DEPOSITED AT DIFFERENT O2 FLOW.

PL was performed on an individual 12% sample heated to several temperatures to track

the evolution of the as a function of temperature. In this measurement, the sample stage was

heated to room temperature, 100ᵒC, 200ᵒC and 300ᵒ C prior to measurements. The starting

temperature was room temperature. After the PL measurement at room temperature the sage was

heated to 100ᵒC. Once the stage stabilized the PL measure measurement was conducted at 100ᵒC.

The same procedure was followed for 200ᵒC and 300ᵒC. Figure 17 shows the intensity of the Sn

peak decreased and shifted to higher energy as the temperature was increased. The decreased in

intensity of the peak is attributed to oxidation of the Sn defects present in the sample. [32]

The peak shift can be due to change in band-gap of deposited SnO2. The band gap of

SnO2 varies from 3.0-3.8 depending on deposition condition and post-deposition heat-treatment.

Incomplete oxidation of Sn in the deposited film has low band-gap than well oxidized deposited

film [36] [37] The PL peak shift due to temperature was also observed by Suhua Luo et.al have

in their nanowire and nanobelts SnO2. [38]. However, no further investigation was conducted to

understand the band gap-peak shift relation.

Energy (eV)

Inten

sity (

a.u)

2.03 e

V

2.25 e

V

30%-O2

18%-O2

12%-O2

31

FIGURE 17. TEMPERATURE-DEPENDENT PL STUDY OF INDIVIDUAL 12% SAMPLE HEATED TO

SEVERAL TEMPERATURES

Figure 18 shows the PL of the 12%-O2 samples that were heat-treated under an oxygen

atmosphere. A peak associated with the interstitial Sn defect at ~1.9 eV is observed only in the

sample that was maintained a room-temperature. In contrast, the samples that were heat-treated

at 200 °C and 300 °C did not show the 1.9 eV peak. The fact that the 1.9 eV energy state is

present only in the 12% as-grown sample and not in the heat-treated samples suggests that 1.9

eV energy state is caused by interstitial Sn defects due to incomplete oxidation of Sn during

deposition. This is also consistent with the XRD data.

32

FIGURE 18. PL DATA OF THE HEAT TREATED SAMPLES COMPARED TO 12%-O2 AS-GROWN.

4.3 Summary of Structural analysis

The Table 6 summarizes the XRD and PL analysis. In both the analysis it was observed

that the high concentration of O2 during deposition and heat treatment oxidize the Sn particles

present in the deposited thin film.

Table 6. Summarized structural analysis

Type of analysis Sample XRD Sn(211) peak PL Sn defect

peak

Sn defect

intensity

As-grown analysis 12%-RT Present Present high

18%-RT Present Present low

30%-RT Absent Absent None

Heat treated analysis 12%-RT Present Present high

12%-200HT Present Present Very low

12%-300HT Absent Absent None

1.7 1.8 1.9 2 2.1 2.2 2.30

500

1000

1500

2000

2500

3000

3500

4000

4500

5000

12%-RT

12%-200RT

12%-300RT

33

Chapter 5: Electrical Analysis

Electrical testing was performed on completed MOM devices to gain insight into several

aspects of the switching behavior. Two basic types of electrical biasing tests were performed. In

one test, the voltage was linearly increased using a constant compliance current to cause repeated

switching between the HRS and LRS. The data from these I-V curves were then analyzed to

determine several switching characteristics such as; switching mode (unipolar vs bipolar),

set/reset voltages, endurance (number of switching cycles), HRS/LRS resistance ratio, and

current conduction in the HRS and LRS states. In this test, a positive voltage was applied to the

top electrode (Ti), and a negative voltage was applied to the bottom electrode (Au). The data

obtained from several current-voltage graphs were fitted with different conduction mechanism to

obtain the conduction mechanisms.

In a second biasing test, the compliance current was varied during the HRS to LRS

switching to distinguish between the filament and interface switching. In this biasing test, if the

switching is due to filament formation, the thickness of the filament increases with the

compliance current. In turn, the thicker filament should require more Joules heating and

therefore larger reset voltage to rupture the filament and switch back to HRS [39] [40]. This is a

very efficient way of distinguishing between the interface and filament switching. To conduct

this test, the compliance current should not be too high else the device will be permanently

damaged.

5.1 Switching Characteristics

For the resistive switching analysis, a Keithley 4200 SCS and a Micromanipulator model

6000 was used. The Keithley was used to create a potential difference between the electrodes and

also for the compliance current. The Micromanipulator 6000 was used to connect the Keithley

34

4200SCS with the device via probes. Au plated probes of tip diameter 0.5 μm were used. The

positive terminal of the Keithley 4220 SCS was connected to the top electrode (Ti), and the

negative terminal was connected to the bottom electrode.

The current-voltage characteristics of the devices are presented in Figure 19 and Figure

21. Three of the devices (12%-RT, 18%-RT, and 12%-HT200) switched and showed unipolar

behavior (Figure 19). The other two devices (30%-RT, 12%-HT300) did not switch Figure 21.

Figure 19 shows the I-V characteristics of the devices that switched. The red trace in each

case is the initial I-V or “forming stage” characteristic. Initially, after deposition or heat-

treatment but prior to electrical biasing, the SnOx films are homogeneous. However, after the

initial biasing, filaments are formed, and this first I-V characteristic is called the “forming stage”

because it creates the filaments. Usually, the set voltage of the forming stage is higher compared

to subsequent HRS stages. This is true for the 12%-RT but not necessarily for the 18%-RT

device. In the 18%-RT device, some of the HRS traces had larger set voltages than the forming

stage. Also, usually, the forming stage is in the HRS. This is true for the 12%-RT and 18%-RT

but not for the 12%-200HT devices. In the 12%-200HT device, the forming stage is in LRS.

The switching behavior is consistent with the presence of interstitial Sn defects observed

via the XRD and PL materials characterization. On the other hand, the devices did not switch did

not show any Sn.

35

FIGURE 19. I-V CHARACTERISTICS OF THE (A) 12%-RT, (B) 18%-RT AND (C) 12%-200HT

DEVICES. THE RED TRACES ARE THE FORMING STAGES, THE BLUE ARE THE HRS AND THE GREEN

ARE THE LRS STATES.

The blue and red traces in Figure 19 are the HRS and LRS curves, respectively. Many

differences can be observed between the devices. The 12%-RT and 18%-RT devices had highly

consistent switching. However the 12%-200HT only switched for a maximum of 6 cycles. After

the sixth cycle, the device went to HRS and did not switch back to LRS. Other differences are

observed between the set and reset voltages. Table 7 summarizes the set and reset voltages of all

the devices.

Figure 20 shows the endurance cycles of the devices that switched. The HRS and LRS

resistances were measured at 0.04 V. The HRS/LRS resistance ratio of 12%-RT, 18%-RT, and

12%-200HT were 102, 10 and 4, respectively. Although only 35 cycles were recorded for the

12%-RT and 18%-RT devices, they showed good potential for much more switching. In contrast,

the 12%-200HT device only switched 6 cycles.

0 1 2 3 4 5 6 710

-6

10-4

10-2

Voltage (V)

Cu

rre

nt

(A)

Re-set

Set

forming

0 0.5 1 1.5 2 2.5 30

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

0.09

Current (A)

Vo

lta

ge

(V

)

set

Re-set

36

FIGURE 20. THE ENDURANCE OF THE (A) 12%-RT, (B) 18%-RT AND (C) 12%-200HT DEVICES.

ENDURANCE MEASURED AT 0.04 V.

Figure 21 shows the I-V characteristics of the devices that did not switch. The 30%-RT

and 12%-RT both have Ohmic conduction in low voltage stage. At higher voltage, the

conduction of 30%-RT and 12%-300HT changed to 𝐼∞𝑉2 and 𝐼∞𝑉3 respectively.

0 5 10 15 20 25 30 3510

1

102

103

104

105

Cycle

Res

ista

nce

(O

hm

)

HRS

LRS

102

0 5 10 15 20 25 30 3510

1

102

Cycle

Re

sis

tan

ce

(O

hm

) HRS

LRS

10

37

FIGURE 21.I-V CHARACTERISTICS OF THE (A) 30%-RT AND (B) 12%-300HT DEVICES. THESE

DEVICES DID NOT SWITCH.

5.2 Current Conduction Mechanisms

The HRS and LRS I-V characteristics of the three devices that switched (12%-RT, 18%-

RT, and 12%-200HT) were analyzed to determine the types of current conduction through the

devices. For each case, equations which describe current conduction mechanisms were compared

and fitted to the measured data. Analysis of the HRS data is presented first followed by analysis

of LRS.

Figure 22 is a log-log plot of representative HRS curves from the three devices that

switched. Log-log graphs are highly useful for determining if two parameters have a power-law

dependence (𝑦 = 𝑘𝑥𝑛) or not. If the relationship between the two parameters has a power-law

dependence, the log-log graph will yield a straight line and the slope is the power n. Otherwise, if

the graph is not straight, the parameters have dependence besides the power-law.

-1 -0.5 0 0.5 1-7

-6

-5

-4

-3

log (V)

log

(I)

slope=1.5

slope=1.2

slope=2

30%-O2

HT-300

slope=3.1

slope=1.1

slope=1.6

38

The 12%-RT trace (green) has two regions corresponding to two levels of applied voltage

(electric field) as shown in Figure 22. At the lower voltage level (<0.32 V), where the electric

field is approximately 𝐸 ≤ 6 × 106 𝑉/𝑚, the current-voltage characteristic is straight with a

slope of approximately 1. This implies that in that lower region, the current conduction follows

an 𝐼 ∝ 𝑉1 dependence or Ohmic dependence. Above 0.32 V, the current-voltage characteristic is

not straight with increasing slope greater than 1 and therefore is not Ohmic Or at least not purely

Ohmic. The 18%-RT device follows a similar trend as shown by the (red) trace in Figure 22.

Below an applied voltage of ~0.25 V, the trace is straight with a slope of approximately 1

indicating Ohmic current conduction. Above 0.25 V, the log-log trace is not straight but curves

upward with increasing voltage. In contrast, the 12%-200HT device shows an Ohmic

dependence throughout the applied voltage range.

Figure 22. The log-log scale of the I-V characteristics of the devices that switched. The red is for

18%-RT HRS I-V, blue squares are for 12%-200HT HRS I-V and green is for 12%-RT.

-2 -1.5 -1 -0.5 0 0.5 1-6

-5

-4

-3

-2

-1

log (V)

log

(I)

18%-O2

HT-200

12%-O2

slope=1.4

slope=1.06

slope=1.01

slope=1.3

slope=1.05

Schottky

Schottky

Transition

Transition

Ohmic

Ohmic

Ohmic

39

If the log(I)-log(V) graph is non-linear with increasing slope greater than 1, then many

current conduction mechanisms are possible. In this work Schottky effect (field enhance

thermionic emission), Poole-Frankel, Space Charge Limited, and Fowler-Nordheim were

investigated as potential mechanisms. Out of these four mechanisms, the Schottky effect fit the

data the best. The Schottky effect equation can be written in the following form,

ln (𝐽) = ln (𝐴𝑇2) −𝑞𝜑

𝑘𝑇+

√ 𝑞3

𝜋𝜀𝑑

√4𝑘𝑇√𝐸 (1)

to facilitate linearized plotting where ln(I) is plotted versus √𝑉. In this equation, I is

current, V is the applied voltage, q is the charge of an electron, k is Boltzmann’s constant, T is

the temperature in Kelvin, and A is Richardson’s constant. Other parameters which depend on

the materials and structure of the device are; the barrier height between the gold electrode and

the SnOx layer was varied between 1-2 eV, the distance between the electrodes (thickness of

SnOx layer) 𝑑 = 50 × 10−9 𝑚, and constant. The only parameter left which was used as a fitting

parameter is the dielectric permitivity, ε. The value of the dielectric constant reported for SnO2 is

between 8.5-14 [41]. The dielectric constant used to fit the data was also in the range of the

dielectric constant of SnO2. The dielectric constant used to fit the data was varied from 8.5-11.8.

Figure 23 plot ln(J) versus √𝐸 of the measured data and Schottky Effect equation of the 12%-RT

and 18%-RT devices, Straight lines with excellent fit to the Schottky model are observed

indicating that the Schottky Effect is the dominant mechanism in this voltage range. The range of

the applied voltage is between 0.6 to 3.6 V corresponding to an applied electric field of 1 ×

107 𝑉/𝑚 to 7 × 107 𝑉/𝑚 which is within the range for Schottky Effect due to barrier lowering

but below for Fowler-Nordheim tunneling.

40

Figure 23.The I-V data of the device 12%-RT and 18%-RT fitted with linearized Schottky

equation.

Figure 24, Figure 25 and Figure 26 are log(I)-log(V) plots of representative LRS data

from the three devices that switched. All the graphs are straight with a slope of approximately 1,

indicating that Ohmic conduction occurred in the LRS state of all the devices. Taking into

consideration the filament model for resistive switching, the LRS data is interpreted as indicating

that Ohmic conduction is occurring through conductive filaments in the SnOx layer between the

electrodes.

0 2000 4000 6000 8000 10000-4

-2

0

2

4

6

8

E (V)

ln J

(A

)

Transition region

Transition region

Ohmicregion

Ohmicregion

18%-RT

12%-RT

LinearizedSchottky

LinearizedSchottky

41

FIGURE 24. THE LRS ANALYSIS OF 12%-RT DEVICE SHOWING OHMIC CONDUCTION WITH A SLOPE

OF 0.97.

FIGURE 25.THE LRS ANALYSIS OF 18%-RT DEVICE SHOWING OHMIC CONDUCTION WITH A SLOPE

OF 1.02

-1 -0.9 -0.8 -0.7 -0.6 -0.5-2.8

-2.7

-2.6

-2.5

-2.4

-2.3

-2.2

-2.1

log (V)

log

(A

)

LRS data

linear slope 0.97

42

FIGURE 26.THE LRS ANALYSIS 12%-200HT DEVICE SHOWING OHMIC CONDUCTION WITH A SLOPE

OF 0.97

5.3 Compliance Current Test

The analysis of the current-voltage characteristic in the LRS state indicated that

conduction is Ohmic possibly due to current flow through metallic filaments. To investigate this

further a compliance-current test was performed. The compliance-current test is used to

distinguish between the filament and interface type switching.

The compliance current test assumes a conical metallic filament is formed between the

electrodes and the compliance current determines the thickness of the filament. Since resistance

is inversely proportional to the cross-sectional area, a thicker filament will have higher

conductivity. The thicker filament will also require higher re-set voltage to rupture it when

compared to thinner filament.

Figure 27 shows the compliance current test on 12%-RT. The compliance current used

was 5 mA to 10 mA which changes the re-set voltage from 0.8 V to 1.62 V. The conductivity of

LRS also increased when 10mA of compliance current was used to switch the device from HRS

to LRS. This test confirms that switching is filament type in the 12%-RT device. [42] [43]

-1.5 -1 -0.5 0 0.5 1-3

-2.5

-2

-1.5

-1

-0.5

0

log (V)

log

(I)

LRS data

linear slope .97

43

The current compliance test done in device 18%-RT is shown in Figure 28. The

compliance current used was 60 mA and 70 mA. The re-set voltage for 60 mA compliance

current voltage was 1.1V and when 70mA was used it changed to 1.5 V. In this device also the

conductivity in LRS was directly proportionality to the compliance current. The change in re-set

voltage due to change in compliance current confirms that the switching is due to filament

formation between the electrodes.

FIGURE 27. COMPLIANCE CURRENT (CC) STUDY OF THE 12%-RT DEVICE. HIGH COMPLIANCE

REQUIRE HIGHER RE-SET VOLTAGE

.

10-1

100

10-6

10-5

10-4

10-3

10-2

10-1

Voltage (V)

Cu

rre

nt

(A)

HRS CC 10mA

LRS CC 10mA

HRS CC 5mA

Reset=1.62 V

Reset=0.8 V

LRS CC 5mA

44

FIGURE 28. COMPLIANCE CURRENT STUDY OF THE 18%-RT DEVICE. HIGH COMPLIANCE REQUIRE

HIGHER RE-SET VOLTAGE

Figure 29 shows the current compliance test of 12%-HT200. The re-set voltages and the

resistivity remained constant when compliance current was changed from 100mA to 70 mA. This

indicated that the switching mechanism in 12%-200HT is not filament type.

0 0.5 1 1.5 2 2.50

0.02

0.04

0.06

0.08

Voltage (V)

Cu

rre

mt

(A)

HRS(60mA)

LRS(60mA)

HRS(70mA)

HRS(70mA)

1.1.V

1.5 V

45

FIGURE 29. COMPLIANCE CURRENT STUDY OF THE 12%-200HT DEVICE. HIGH COMPLIANCE

REQUIRE DID NOT CHANGE RE-SET VOLTAGE

5.3 Summary of Electrical Characterization

Table 7 summarizes the electrical characteristics of the devices. Good resistive

switching behavior was observed in the 12%-RT and 18%-RT devices. Both devices showed

Ohmic and Schottky conduction in the HRS and Ohmic conduction in LRS. The current

compliance study on both the devices showed that the switching mechanism is filament type.

Although the 12%-200HT device switched for 6 cycles, its conduction mechanism and

compliance-current results were distinct from the other two devices indicating a different

switching mechanism. The 12%-RT had the best yield among all the devices.

0 2 4 6 8 10 120

0.02

0.04

0.06

0.08

0.1

0.12

Voltage (V)

Cu

rre

nt

(A)

HRS(100mA)

LRS(100mA)

HRS(70mA)

HRS(70mA)

2.9 V

46

Table 7.Summary of electrical characterization of the devices

Sample Switching

Set

voltage

(V)

Reset

voltage

(V)

HRS

conduction

mechanism

LRS

conduction

mechanism

Type of

switching

Yield

(%)

12%-RT Unipolar 1.6-6.1 0.6-1.2 Ohmic and

Schottky

Ohmic Filament 60

18%-RT Unipolar 1.8-2.5 1.5-1.8 Ohmic and

Schottky

Ohmic Filament 30

30%-RT None N/A N/A N/A N/A N/A 0

12%-200HT Unipolar 6-11 1.8-3.7 Ohmic Ohmic Unknown <10

12%-300HT No N/A N/A N/A N/A N/A 0

47

Chapter 6: SnO2-Resistive switching device Model

A general phenomenological model for current conduction and resistive-switching was

created based on the structural and electrical analysis. In this study, the following observations

were made which form the basis of the model:

• The amount of oxygen during deposition and temperature during heat-treatment in

oxygen affected the level of Sn particles and interstitial in the SnOx layer. SnOx layers

with lower levels of oxygen exposure appeared to be incompletely oxidized. This was

manifested by Sn particles as indicated by XRD data, and Sn interstitials as indicated by

PL emission at energies levels associated with Sn interstitials.

• Electrical current-voltage data revealed that devices with Sn particles and interstitials

switched. In contrast, the devices that were more completely oxidized did not switch or

did not switch as well.

• Current conduction in the HRS of the devices that switched well is Ohmic at low voltages

and transitions to Schottky as the electric field increases.

• Current conduction in the LRS of the devices that switched well is Ohmic throughout the

bias range from zero to the reset voltage.

• Current-compliance testing indicates a filament formation mecanism.

These observations are summarized in Table 8.

48

Table 8.Summarized structural and electrical analysis

Sample XRD

Sn(211)

peak

PL peak

intensity

of Sn

Switching Low voltage

HRS current

High voltage

HRS current

12%-RT Yes High Unipolar Ohmic Schottky

18%-RT Yes Low Unipolar Ohmic Schottky

30%-RT No None None NA NA

12%-200HT Yes None Unipolar Ohmic Ohmic

12%-300HT No None None NA NA

Based on the observations presented above, the model in Figure 30 is proposed for the

Ohmic current conduction in the device in the HRS state at low applied electric field (𝐸 ≤

6 × 106 𝑉/𝑚). In this state and bias condition, the primary current flow is due to free electrons

in the conduction band of the SnO2 towards the Ti electrode. [44] The source of the free

electrons are the Sn interstitial energy states as well as oxygen vacancies. The electrons are

replentished by thermal generation from the valence band to the defect energy levles. At the

same time, holes that are created in the valence drift towards the Au electrode where they are

annihilateed by electrons there. It is also possible that some ionized Sn atoms migrate towards

the Au electrode but this current would very low.

49

FIGURE 30. OHMIC CONDUCTION AT LOW HRS VOLTAGE. ELECTRONS ARE GENERATED DUE TO

IONIZATION OF SN ATOM. SN+ FLOWS TOWARDS THE NEGATIVE ELECTRODE (AU) WHILE

ELECTRONS FLOW TOWARDS THE POSITIVE (TI)

While still in HRS but with the applied electric field increasing above (𝐸 ≥ 6 × 106 𝑉/

𝑚), the current conduction transitions to Schottky emission and the model in Figure 31 is

proposed for this conduction. In this state and bias condition, the primary current flow is due to

field enhanced thermionic emission from the Au electrode towards the SnO2 as shown in Figure

31. [45] [46] The conduction is dominated by the electrons from the metal.

Also, as the electric field increases, the Sn cations continue to flow towards the negative

electrode where they can form neutral Sn atom by gaining electron from metal as shown in

Figure 32. It is speculated that at high enough electric field 𝐸 ≥ 1 × 108 𝑉/𝑚, the Sn ion

migration becomes dominant and this is the primary mechanism for creating a conductive

filament which leads to resistive switching from HRS to LRS.

SnO2

Ohmic conduction HRS

Sn+ holes electronsSn

Electron flow

hole flow

Sn+

flow

Ef

Ti SnO2 Au

5.1eV

Unknown4.2eV

ϕTiSnO2

ϕAuSnO2eV

50

FIGURE 31. SCHOTTKY CONDUCTION. Φ IS THE BARRIER HEIGHT THAT ELECTRONS HAVE TO

OVERCOME. THE SN+ CONTINUES TO FLOW TOWARDS THE NEGATIVE ELECTRODE, I.E. AU

Once a metal Sn filament is created bridging the electrodes, the device will be in the LRS

state as shown in Figure 33. The change in resistance occurs due to the formation of metallic

filament between the electrodes as shown by the current compliance analysis. The cations form

the filament between the electrodes, and the current conduction is Ohmic similar to a simple thin

wire. However, if the current in the thin filament is too high, significant Joules heating will occur

and will rupture a section of the filament causing the device to switch to HRS as shown inFigure

33. This process of creating the rupturing the filament is repeatable and gives rise to the observed

resistive switching behavior. [47]

51

FIGURE 32.AT RESET VOLTAGE FILAMENT IS FORMED AND THE ELECTRON CONDUCTION TAKES

PLACE THROUGH THE METALLIC FILAMENT. THE DEVICES SWITCHED TO LRS

FIGURE 33. AT RE-SET VOLTAGE FILAMENT RUPTURES INDICATED BY RED-CROSS AND DEVICE GOES

BACK TO HRS

SnO2

Ohmic conduction LRS

Sn+ holes electronsSn

Electron flow

hole flow

Sn+

flow

Ti SnO2 Au

5.1eV

Unknown4.2eV

ϕTiSnO2

ϕAuSnO2

SnO2

Re-set filament rupture

Sn+ holes electronsSn

Electron flow

hole flow

Sn+

flow

Ti SnO2 Au

5.1eV

Unknown4.2eV

ϕTiSnO2

ϕAuSnO2

52

Chapter 7: Conclusion

The reactive sputtering technique was used to deposit SnO2 using Sn target, in the O2-Ar

atmosphere on Au/Cr/SiO2/Si. The reactive sputtering deposition was very useful to do a

systematic study and to find the switching mechanism in SnO2 devices. The structural and

electrical characterization showed that the Sn concentration in SnOx had a critical role in

unipolar resistive switching behavior. The Sn concentration in the SnOx was varied by changing

the O2 –Ar concentration during deposition. Heat-treatment analysis further investigated the role

of Sn.

The LRS current conduction was Ohmic, and the re-set voltage was highly dependent on

the compliance current. The compliance current test established that the switching in SnO2

devices is filament type. In filament type switching a metallic filament is formed between the

electrodes. The HRS analysis gave us more detail about the filament.

The HRS conduction in low voltage was Ohmic and at high voltage was Schottky

conduction. The Ohmic conduction ionized the Sn to form Sn+ ions. The movement of the Sn+

was possible due to the potential difference between the electrodes. The migration of Sn+ created

a filament between the electrodes, and the devices switched to LRS. During the LRS Ohmic

conduction, the filament ruptured at the re-set voltage to bring the device back to HRS. The

rupture of the filament was due to Joules heating.

This study reported for the first time about the switching mechanism in SnO2 device

using the reactive sputtering deposition technique. The findings concluded that SnO2-RRAM

have a promising future for next-generation memory.

53

References

[1] D. S. Jeong, R. Thomas, R. S. Katiyar, J. F. Scott, H. Kohlstedt and C. S. Hwang,

"Emerging memories: resistive switching mechanisms and current status," Reports on

Progress in Physics, vol. 75, p. 7, 2012.

[2] J. J. Yang, M. D. Pickett, X. Li, D. A. Ohlberg, D. R. Stewart and R. S. Williams,

"Memristive switching mechanism for metal/oxide/metal nanodevices," Nature

nanotechnology, vol. 3, no. 7, pp. 429-433, 2008.

[3] S. Tsui, A. Baikalov, J. Cmaidalka, Y. Y. Sun, Y. Q. Wang, Y. Y. Xue, C. W. Chu, L. Chen

and A. J. Jacobson., "Field-induced resistive switching in metal-oxide interfaces.," Applied

physics letters, vol. 85, no. 2, pp. 317-319, 2004.

[4] [Online]. Available: https://www.crossbar-inc.com/en/technology/reram-overview/.

[5] C. Schindler, G. Staikov and R. Waser, " Electrode kinetics of Cu–Si O 2-based resistive

switching cells: Overcoming the voltage-time dilemma of electrochemical metallization

memories.," Applied physics letters, vol. 94, no. 4, p. 072109, 2009.

[6] P. Shanshan, F. Zhuge, X. Chen, X. Zhu, B. Hu, L. Pan, B. Chen and R.-W. Li, "Mechanism

for resistive switching in an oxide-based electrochemical metallization memory," Applied

Physics Letters , vol. 100, no. 7, p. 072101., 2012.

[7] Yang, Yuchao, P. Gao, S. Gaba, T. Chang, X. Pan and W. Lu., "Observation of conducting

filament growth in nanoscale resistive memories.," Nature communications, vol. 3, p. 1737,

2012.

[8] R. Waser and M. Aono, "Nanoionics-based resistive switching memories.," Nature

materials, vol. 6, no. 11, pp. 833-840, 2007.

[9] Park, Seong-Geon, B. Magyari-Kope and Y. Nishi., ""Impact of Oxygen Vacancy Ordering

on the Formation of a Conductive Filament in $\hbox TiO _ 2 $ for Resistive Switching

Memory.," IEEE Electron Device Letters, vol. 32, no. 2, pp. 197-199, 2011.

[10] W. Rainer and M. Aono, "Nanoionics-based resistive switching memories.," In Nanoscience

And Technology: A Collection of Reviews from Nature Journals, pp. 158-164, 2010.

[11] Z. Wei, Y. Kanzawa, K. Arita, Y. Katoh, K. Kawai, Muraoka, S. Mitani, S. F. S., K. K., I.

M. and T. Mikawa, "Highly reliable TaOx ReRAM and direct evidence of redox reaction

mechanism," in In Electron Devices Meeting, 2008.

[12] K. D.H., K. K.M., J. Jang, J. Jeon, Lee, M. Kim, G. Li, X. Park, G. Lee, B., S. Han and K.

M., "Atomic structure of conducting nanofilaments in TiO2 resistive switching memory.,"

Nature nanotechnology, vol. 5, no. 2, pp. 148-153, 2010.

[13] W. Yan, Q. Liu, S. Long, W. Wang, Q. Wang, M. Zhang, S. Zhang and M. Liu,

"Investigation of resistive switching in Cu-doped HfO2 thin film for multilevel non-volatile

memory applications," Nanotechnology, vol. 21, no. 4, p. 045202., 2009.

[14] A. Sawa, " Resistive switching in transition metal oxides," Materials today, vol. 11, no. 6,

pp. 28-36, 2008.

[15] Y.-B. Kim, S. R. Lee, D. Lee, C. B. Lee, M. Chang, J. H. Hur and M.-J. Lee, "Bi-layered

54

RRAM with unlimited endurance and extremely uniform switching," Symposium onVLSI

Technology (VLSIT), pp. 52-53, 2011.

[16] H. Lee, P. Chen, T. C. Wu, Y. Wang, C. Tzeng, P. Lin, C. Chen, F. Lien and T. M.J., "Low

power and high speed bipolar switching with a thin reactive Ti buffer layer in robust HfO2

based RRAM.," Electron Devices Meeting, pp. 1-4, 2008.

[17] D. Ielmini, F. Nardi, C. Cagli and A. L. Lacaita, "Size-dependent retention time in NiO-

based resistive-switching memories," EEE Electron Device Letters, vol. 31, no. 4, pp. 353-

355, 2010.

[18] [Online]. Available: www.ptable.com.

[19] Z. Jarzebski and J. Marton, "Physical properties of SnO2 materials II. Electrical properties,"

Journal of the electrochemical Society, vol. 123, no. 9, pp. 299C-310C, 1976.

[20] A. F. Khan, M. Mehmood, A. M. Rana and ". M. T. Bhatti, "Effect of annealing on

electrical resistivity of rf-magnetron sputtered nanostructured SnO 2 thin films," Applied

Surface Science, vol. 255, no. 20, pp. 8562-8565, 2009.

[21] K. Mary, J. Sajiz and A. P. Reena, "Tin Oxide Based P and N-Type Thin Film Transistors

Developed by RF Sputtering," ECS Journal of Solid State Science and Technology, vol. 4,

no. 9, pp. Q101-Q104, 2015.

[22] Z. Wang, S. Liu, T. Jiang, X. Xu, J. Zhang, C. An and C. Wang, "N-type SnO2 nanosheets

standing on p-type carbon nanofibers: a novel hierarchical nanostructures based hydrogen

sensor," RSC Advaces, vol. 79, no. 5, pp. 64582-64587, 2015.

[23] B. S. Tosun, R. K. Feist, A. Gunawan, K. A. Mkhoyan, S. A. Campbell and E. S. Aydil,

"Sputter deposition of semicrystalline tin dioxide films," Thin Solid Films, vol. 520, pp.

2554-2561, 2012.

[24] D. Leng, L. Wu, H. Jiang, Y. Zhao, J. Zhang, W. Li and L. Feng, "Preparation and

Properties of SnO2 Film Deposited by Magnetron Sputtering," International Journal of

Photoenergy, vol. 2012, no. 235971, p. 6, 2012.

[25] V. GOLOVANOV, L. KHIRUNENKO, A. KIV, D. FUKS, M. SOSHIN and K. G,

"Radiation effects in SnO2–Si sensor structures," Radiation Effects & Defects in Solids,

Vols. 00,No0, pp. 1-5, 2006.

[26] K. Nagashima, T. Yanagida, Oka and T. Kawai, "Unipolar resistive switching

characteristics of room temperature grown SnO 2 thin films," Applied Physics Letters, vol.

94, no. 24, p. 242902, 2009.

[27] S. Almeida, B. Aguirre, N. Marquez, J. McClure and D. Zubia, ""Resistive switching of

SnO2 thin films on glass substrates," Integrated Ferroelectrics, vol. 126, no. 1, pp. 117-124,

2011.

[28] A. Talukdar, S. Almeida, J. Mireles, E. MacDonald, J. Pierluissi, E. Garcia and D. Zubia,

"Unipolar resistive switching and current flow mechanism in thin film SnO2," in Nanotech,

Washington,DC, 2014.

[29] J. Z. J. K. R. Jin, L. Y., P. Bao, M. Althobaiti, D. Hesp, V. Dhanak, Z. Zheng, M. I.Z. and S.

Hall, "Effects of annealing conditions on resistive switching characteristics of SnO x thin

films," Journal of Alloys and Compounds, vol. 673, pp. 54-59, 2016.

[30] F. Colas, D. Barchiesi, S. Kessentini, T. Toury and M. L. d. l. Chapelle, "Comparison of

adhesion layers of gold on silicate glasses for SERS detection," Journal of Optics, vol. 17,

55

p. 122, 2015.

[31] P. Hsu, C. Hsu, Chang, C.H., Tsai, S.P, Chen, W.C., H. H.H and C. Wu, "Sputtering

deposition of p-type SnO films with SnO2 target in hydrogen-containing atmosphere.," ACS

applied materials & interfaces, vol. 6, no. 16, pp. 13724-13729., 2014.

[32] M. Bhatnagar, V. Kaushik, A. Kaushal, M. Singh and B. R. Mehta1, "Structural and

photoluminescence properties of tin oxide and tin oxide: C core–shell and alloy

nanoparticles synthesised using gas phase technique," AIP Advances , vol. 6, p. 095321 ,

2016.

[33] A. F. Khan, M. Mehmood, A. M. Rana and M. T. Bhatti, "Effect of annealing on electrical

resistivity of rf-magnetron sputtered nanostructured SnO 2 thin films.," Applied Surface

Science, vol. 255, no. 20, pp. 8562-8565, 2009.

[34] A. Kar, S. Kundu and A. Patra, "Surface Defect-Related Luminescence Properties of SnO2

Nanorods and Nanoparticles," Journal of physical chemistry, vol. 115, no. 1, p. 118–124,

2011.

[35] E. R. Viana, J. C. González, G. M. Ribeiro and A. G. d. Oliveira, "Photoluminescence and

High-Temperature Persistent Photoconductivity Experiments in SnO2 Nanobelts," The

Journal of Physiscal Chemistry, vol. 117, pp. 7844-7849, 2013.

[36] J. Yadav, R. Patil, R. Puri and V. Puri, "Studies on undoped SnO2 thin film deposited by

chemical reactive evaporation method," Materials Science and Engineering , vol. 139, pp.

69-73, 2007.

[37] E. Fortunato, R. Barros, P. Barquinha, V. Figueiredo, S.-H. K. Park, C.-S. Hwang and R.

Martins, "Transparent p-type SnOx thin film transistors produced by reactive rf magnetron

sputtering followed by low temperature annealing," Applied Physics letters, vol. 97, p.

052105, 2010.

[38] S. Luo, J. Fan, W. Liu, M. Zhang, Z. Song, C. Lin, X. Wu and P. K. Chu, "Synthesis and

low-temperature photoluminescence properties of SnO2 nanowires and nanobelts,"

Nanotechnology, vol. 17, pp. 1695-1699, 2006.

[39] D. Ielmini, "Resistive switching memories based on metal oxides: mechanisms, reliability

and scaling," Semiconductor Science and Technology, vol. 31, p. 25, 2016.

[40] S. Mei, M. Bosman, R. Nagarajan, X. Wu and K. L. P. a, "Compliance current dominates

evolution of NiSi2 defect size in Ni/dielectric/Si RRAM devices," Microelectronics

Reliability , vol. 61, pp. 71-77, 2016.

[41] H. Daal, "The Static Dielectric Constant of Sn02," Journal of Applied Physics, vol. 39, pp.

4467-4469, 1968.

[42] X. Cao, X. Li, X. Gao, Y. Zhang, X. Liu, Q.Wang and L. Chen, "Effects of the compliance

current on the resistive switching behavior of TiO2 thin films," Applied Physics A, vol. 97,

pp. 883-887, 2009.

[43] K. M. Kim and C. S. Hwang, "The conical shape filament growth model in unipolar

resistance switching of TiO2 thin film," Applied Physics Letters, vol. 94, p. 122109, 2009.

[44] F.-C. Chiu, "A Review on Conduction Mechanisms in Dielectric Films," Advances in

Materials Science and Engineering, vol. 578168, p. 2014, 2014.

[45] F.-C. Chiu, "A Review on Conduction Mechanisms in Dielectric Films," Advances in

Materials Science and Engineering, vol. 2014, p. 578168, 2014.

56

[46] M. Wittmer, "Conduction mechanism in PtSi/Si Schottky diodes," PHYSICAL REVIEW B,

vol. 40, no. 5, pp. 4385-4395, 1991.

[47] U. Russo, D. Ielmini, C. Cagli and A. L. Lacaita, "Self-accelerated thermal dissolution

model for reset programming in unipolar resistive-switching memory (RRAM) devices.,"

IEEE Transactions on Electron Devices, vol. 56, no. 2, pp. 193-200, 2009.

57

Vita

Arka Talukdar earned his Bachelors in Electrical and Electrons Engineering from West

Bengal University of Technology in 2009. In August 2010 he joined integrated Masters and

doctoral program in Electrical and Computer Engineering at UTEP. In May 2016 he received his

Masters of Science in Electrical Engineering from University of Texas at El Paso.

While pursuing his degree Arka worked as research associate and assistance instructor for

the department of electrical and computer engineering. He joined the US ARMY Reserve in

May,2016. Arka gave an Oral presentation of his research at 2014 Nanotechnology Conference

and Expo help at Washington DC. Additionally, he has authored in two conference proceeding

articles.

Arka’s dissertation entitled, “Electrical and Structural Characterization of Tin Oxide

Resistive Switching” was supervised by Dr. David Zubia.

Permanent address: 1532 Upson Drive, Apt #A

El Paso, TX-79902

This thesis/dissertation was typed by Arka Talukdar