Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
Phase Transformations and Switching
of Chalcogenide Phase-change
Material Films Prepared by Pulsed
Laser Deposition
Der Fakultät für Physik und Geowissenschaften
der Universität Leipzig
eingereichte
D I S S E R T A T I O N
zur Erlangung des akademischen Grades
Doctor Rerum Naturalium
Dr. rer. nat.
vorgelegt
von M. Sc. Xinxing Sun
geboren am 23.12.1986 in Zhejiang/China
Gutachter: Prof. Dr. Dr. h.c. Bernd Rauschenbach (Leipzig)
Prof. Dr. Petr Němec (Pardubice/Czech Republic)
Tag der Verleihung: 13. März 2017
Bibliographische Beschreibung
Sun, Xinxing
Phase Transformations and Switching of Chalcogenide Phase-change Material Films
Prepared by Pulsed Laser Deposition
Universität Leipzig, Dissertation
171 S., 207 Lit., 76 Abb., 6 Tab.
Review:
The thesis deals with the preparation, characterization and, in particular, with the switching
properties of phase-change material (PCM) thin films. The films were deposited using the
Pulsed Laser Deposition (PLD) technique. Phase transformations in these films were
triggered by means of thermal annealing, laser pulses, and electrical pulses. The five major
physical aspects structure transformation, crystallization kinetics, topography, optical
properties, and electrical properties have been investigated using XRD, TEM, SEM, AFM,
DSC, UV-Vis spectroscopy, a custom-made nanosecond UV laser pump-probe system, in
situ resistance measurements, and conductive-AFM.
The systematic investigation of the ex situ thermally induced crystallization process of pure
stoichiometric GeTe films and O-incorporating GeTe films provides detailed information
on structure transformation, topography, crystallization kinetics, optical reflectivity and
electrical resistivity. The results reveal a significant improvement of the thermal stability in
PCM application for data storage. With the aim of reducing the switching energy
consumption and to enhance the optical reflectivity contrast by improving the quality of the
produced films, the growth of the GeTe films with simultaneous in situ thermal treatment
was investigated with respect to optimizing the film growth conditions, e.g. growth
temperature, substrate type.
For the investigation of the fast phase transformation process, GeTe films were irradiated
by ns UV laser pulses, tailoring various parameters such as pulse number, laser fluence,
pulse repetition rate, and film thickness. Additionally, the investigation focused on the
comparison of crystallization of GST thin films induced by either nano- or femtosecond
single laser pulse irradiation, used to attain a high data transfer rate and to improve the
understanding of the mechanisms of fast phase transformation.
Non-volatile optical multilevel switching in GeTe phase-change films was identified to be
feasible and accurately controllable at a timescale of nanoseconds, which is promising for
high speed and high storage density of optical memory devices. Moreover, correlating the
dynamics of the optical switching process and the structural information demonstrated not
only exactly how fast phase change processes take place, but also, importantly, allowed the
determination of the rapid kinetics of phase transformation on the microscopic scale.
In the next step, a new general concept for the combination of PCRAM and ReRAM was
developed. Bipolar electrical switching of PCM memory cells at the nanoscale can be
achieved and improvements of the performance in terms of RESET/SET operation voltage,
On/Off resistance ratio and cycling endurance are demonstrated. The original underlying
mechanism was verified by the Poole-Frenkel conduction model. The polarity-dependent
resistance switching processes can be visualized simultaneously by topography and current
images. The local microstructure on the nanoscale of such memory cells and the
corresponding local chemical composition were correlated.
The gained results contribute to meeting the key challenges of the current understanding
and of the development of PCMs for data storage applications, covering thin film
preparation, thermal stability, signal-to-noise ratio, switching energy, data transfer rate,
storage density, and scalability.
I
Table of Contents
CHAPTER 1 INTRODUCTION ........................................................................ 1
CHAPTER 2 FUNDAMENTALS: PHASE-CHANGE MATERIALS AND
THEIR APPLICATIONS ........................................................................................ 5
2.1 Principle of phase-change materials ....................................................................................... 5
2.2 Ge-Sb-Te phases ....................................................................................................................... 7 Pseudo-binary GeTe-Sb2Te3 tie-line ................................................................................. 7 2.2.1
Crystal structure of GST and GeTe ................................................................................... 9 2.2.2
Concise compendium of the crystallization process ....................................................... 12 2.2.3
2.3 Applications of phase-change materials................................................................................ 17 Optical data storage ......................................................................................................... 17 2.3.1
Electronic memory .......................................................................................................... 19 2.3.2
Frontier applications of phase-change materials ............................................................. 22 2.3.3
CHAPTER 3 EXPERIMENTAL CONDITIONS AND METHODS ........... 23
3.1 Pulsed laser deposition of thin films ...................................................................................... 23
3.2 Thermal treatment ................................................................................................................. 25 In situ thermal treatment in deposition chamber ............................................................. 25 3.2.1
Ex situ thermal treatment in an infra-red irradiation oven............................................... 25 3.2.2
3.3 Laser pulse irradiation for optical switching ....................................................................... 26 Nano- and femtosecond laser irradiation ......................................................................... 26 3.3.1
Nanosecond laser pump-probe system ............................................................................ 27 3.3.2
3.4 Arrangement for electrical switching ................................................................................... 28
3.5 Characterization methods ...................................................................................................... 30 X-ray diffraction-based analysis methods ....................................................................... 30 3.5.1
Electron microscopy........................................................................................................ 34 3.5.2
Atomic force microscopy ................................................................................................ 36 3.5.3
Spectroscopy and calorimetry methods ........................................................................... 38 3.5.4
CHAPTER 4 RESULTS AND DISCUSSION ................................................ 41
4.1 Thermal treatment-induced crystallization of GeTe films .................................................. 41 Ex situ thermally induced crystallization of GeTe films ................................................. 41 4.1.1
Effect of O incorporation on structural transitions of GeTe films ................................... 50 4.1.2
In situ thermally induced crystallization of GeTe films .................................................. 57 4.1.3
Epitaxial growth of GeTe films on BaF2(111) ................................................................ 63 4.1.4
Summary of the results on thermal treatment-induced crystallization of GeTe films ..... 67 4.1.5
II
4.2 Laser-induced phase transitions of GeTe and GST films .................................................... 69 Nanosecond laser-induced phase transition in GeTe films .............................................. 69 4.2.1
Crystallization of Ge2Sb2Te5 films by nano- and femtosecond single laser pulse 4.2.2
irradiation ........................................................................................................................ 80 Simulation of temperature distribution after pulse laser irradiation ................................ 89 4.2.3
Summary of the results on laser-induced phase transitions of GeTe and GST films ....... 93 4.2.4
4.3 Dynamic optical switching of non-volatile multi-level memory .......................................... 95 Optical switching ............................................................................................................. 96 4.3.1
Structural transformation ............................................................................................... 100 4.3.2
Morphology ................................................................................................................... 102 4.3.3
Dynamic phase transformation ...................................................................................... 106 4.3.4
Retention ........................................................................................................................ 109 4.3.5
Optical switching model ................................................................................................ 111 4.3.6
Film thickness dependence of numbers of multi-levels ................................................. 113 4.3.7
Broad wavelength response of reflectivity ..................................................................... 114 4.3.8
Summary of the results on dynamic optical switching of non-volatile multi-level 4.3.9
memory ......................................................................................................................... 115
4.4 Nanoscale bipolar electrical switching of phase-change materials ................................... 117 Background .................................................................................................................... 117 4.4.1
Polarity-dependent resistance switching of phase-change materials ............................. 123 4.4.2
Threshold of resistance switching .................................................................................. 129 4.4.3
Mechanism of bipolar resistance switching in phase-change materials ......................... 131 4.4.4
Summary of the results on nanoscale bipolar electrical switching of GST films .......... 143 4.4.5
CHAPTER 5 SUMMARY ............................................................................... 145
BIBLIOGRAPHY ................................................................................................ 149
LIST OF ABBREVIATIONS .............................................................................. 163
ACKNOWLEDGEMENTS ................................................................................. 165
CURRICULUM VITAE ...................................................................................... 167
LIST OF PUBLICATIONS ................................................................................. 169
SELBSTSTÄNDIGKEITSERKLÄRUNG ........................................................ 171
1
Chapter 1
Introduction
With the rapid growth of global networking, large data centers, and the portable
devices application-driven data storage market, the memory industry has ushered in
a comprehensive opportunity for diverse developments. The ideal new type of
memory is desirably characterized by superior all-round capabilities, including non-
volatility, long cycling life, small component size, low power consumption, fast
read/write processes, multilevel storage, 3D integration, and affordability of the
final product. One of the most promising areas in this respect is data storage
technology based on phase-change materials (PCMs) [1-3], which rely on the
reversible transformation between the disordered amorphous and ordered crystalline
states of PCMs induced by local heating/cooling either with laser pulses or
electrical pulses. The phase transformation is accompanied by a significant change
in optical reflectivity and electrical resistivity between these two solid states. The
pioneering work on the research field of PCMs was published in 1986 by
Ovshinsky [4], and considerable explorations have been made on the topic of
chalcogenides for data storage. Currently, chalcogenide-based PCMs are already
being applied in commercial optical storage media such as CDs, DVDs, and Blu-ray
disks. At the same time, the reversible electrical resistivity change of PCMs can be
applied in non-volatile electronic memory, which is considered to succeed FLASH
memory technology and is becoming the most promising candidate for future data
storage technology. However, the development of PCM-based data storage
technology gives rise to key challenges related to exploring the potential limits of
this technology, such as lowest switching energy, fastest switching speeds, largest
storage density and smallest cell sizes [5, 6].
A widely investigated chalcogenide material system in PCM based memory
technology is the Ge-Sb-Te system, particularly Ge2Sb2Te5 (GST). Regarding Ge-
Sb-Te thin films as the active layer of a memory device, magnetron sputtering is
typically employed for the thin film deposition. As an alternative to this physical
Chapter 1 Introduction
2
deposition method, pulsed laser deposition (PLD) provides several advantages: high
quality films, options for multi-compositions, quick and accurate control of the
deposition process, simple deposition system, possibility of scaling up the process
[7, 8]. To date, there exist only few studies dealing with the preparation of phase-
change films by PLD [9-12], and systematic investigations of the phase
transformation and the accompanying physical properties of PLD-prepared PCM
films are lacking. The promising PCMs possess unique properties such as high
crystallization speed, prominent optical reflectivity and electrical resistivity
contrast, excellent reversibility, large cycling numbers of reversible transitions and
high archival lifetimes of more than ten years [2]. With respect to guidance about
the optimal materials for data storage application, however, several crucial
questions exist: i) How can the thermal stability of PCMs be improved (so that they
are suitable, e.g. for use in automotive applications)? ii) How can the switching
energy be reduced? iii) What is the mechanism of phase transformation, particularly
in ultrafast switching? iv) How can the storage density be increased? v) What are
the electrical switching behavior and mechanisms when scaling down the memory
cells to the nanoscale? Undoubtedly, investigations to address the above issues are
of great importance for the ongoing development of phase-change data storage
technology.
Objectives and organization of this thesis
The thesis aims to explore the performance and mechanisms of PCMs with data
storage applications in view, including such key features as thermal stability, signal-
to-noise ratio, switching energy, data transfer rate, storage density, and scalability.
Related to this, in particular four major physical aspects, namely the crystallization
kinetics, structure transformation, optical and electrical switching have been
investigated. The main objectives are formulated as follows:
To investigate the influences of in situ or ex situ thermal treatment on
the structural transformation and the properties on the crystallization
process of PLD-deposited GeTe films, in order to optimize the thermal
stability and signal-to-noise ratio.
Chapter 1 Introduction
3
To examine laser pulse-induced phase transitions of GeTe and GST
films for attaining high date transfer rates and to improve the
understanding of rapid phase transformation kinetics.
To investigate the multi-level optical switching of single layer GeTe
films by using a custom-made nanosecond UV pump-probe system, in
order to increase the storage density and to reveal the mechanism of
rapid phase transformation.
To explore the nanoscale electrical switching and underlying mechanism
of GST films based device by using an arrangement of conductive
atomic force microscopy tips as a probe that enables write/read/erase
operations at the nanoscale.
To clarify correlations of structural phase transformation with respect to
optical or electrical property characterization by appropriate analysis.
Scope
In the thesis, Chapter 2 provides a brief introduction on the fundamentals of phase-
change materials and their application. A description of the experimental setup and
the characterization methods are presented in Chapter 3. The investigations on
various aspects of physical properties in PCMs for data storage applications and the
obtained results are presented and discussed in Chapter 4. In Chapter 4.1, the results
of thermal treatment-induced crystallization of GeTe films are presented. Chapter
4.2 deals with laser pulse-induced phase transitions of GeTe and GST films. Results
on dynamic optical switching of non-volatile multi-level memory cells are provided
in Chapter 4.3. Findings about nanoscale bipolar electrical switching of PCMs are
presented in Chapter 4.4. Finally, the contributions of the present work are
summarized in Chapter 5.
5
Chapter 2
Fundamentals: phase-change materials
and their applications
Principle of phase-change materials 2.1
Phase-change materials (PCMs) exist in two different stable states, crystalline and
amorphous, which exhibit distinctly different physical properties. The most
significant feature of this material class is the ability to repeatedly and rapidly
switch between these states through heating or cooling. The concept of using the
amorphous state of PCMs as logical ‘1’ and the crystalline state as logical ‘0’, as
well as the capability to switch between both states, has been extensively explored
for data storage applications. By utilizing the non-volatility of the optical
reflectivity switching property, PCMs have already been applied in commercial re-
writable optical storage media such as CDs, DVDs, and Blu-ray discs [2]. Recently,
based on the non-volatility of the large contrast in the electrical resistance switching
property, PCMs have been demonstrated to be promising for use in non-volatile
electronic memory devices [3, 13]. The basic principle of data storage based on
PCMs is exemplified in Figure 2.1.
The state in which the PCM is in the crystalline phase is characterized by a high
optical reflectivity and a low electrical resistance. For amorphization of the PCM,
an intense and short laser or electrical pulse is applied (Record/RESET operation).
This intense pulse provides the power to rapidly increase the temperature in the
material above the melting point Tm. Because of the very short pulse duration, high
cooling rates exceeding 109 K/s are achieved and thereby the material is quenched
into the disordered amorphous state. A moderately intense and long laser or
electrical pulse is used to crystallize the PCM (Erase/SET operation). The material
is locally heated above the crystallization temperature Tc, which lies between the
glass transition temperature Tg and the melting point Tm. The pulse-induced
Chapter 2 Fundamentals: phase-change materials and their applications
6
temperature increase causes increased mobility of atoms in the material and results
in an energetically favorable atomic arrangement in the form of the crystalline
phase. Because the differences in the optical and electrical properties between the
amorphous and crystalline states are significant, the respective state can thus be
easily read out by applying a very low intensity laser or electrical pulse.
The atomic arrangement in the material is schematically illustrated in Figure
2.1. In the crystalline phase, the atoms show a long-range order, whereas the liquid
state is characterized by complete disorder (exception: GeTe, see Ref. [14]). For the
amorphous phase, the arrangement of atoms is disordered, but has a locally limited,
short-range ordered structure extending over only a few atomic distances. The
understanding of the atomic configurations and bonding mechanisms provides an
explanation for the large contrast in optical and electrical properties between the
two phases [15]. Further details on the structural properties of Ge-Sb-Te phases in
the amorphous and crystalline states can be found in Chapter 2.2.2. It should be
noted that a high crystalline quality of the PCM in the crystalline state could be
beneficial for reduction of the required switching energy [16, 17].
In addition, it is important that the timescale of phase transitions in PCMs is
considered. For the amorphization process, a high quenching rate of more than
109
K/s is needed, because upon the quenching from a liquid into a “frozen” state,
the resulting structural configuration must be below the glass transition temperature
Tg, in order to avoid the occurrence of re-crystallization. The typical crystallization
Figure 2.1: Principle of reversible switching in phase-change materials for non-volatile memory
applications. Tm melting temperature, Tc crystallization temperature and t time.
Crystalline phase
Ordered structureHigh reflectivityLow resistance
Liquid phase
Disordered structure
Low reflectivityHigh resistance
Amorphous phase
Tm
t
Laser or current pulse
Tc Tm
t
Laser or current pulse
Tc
Erase/SET
Record/RESET
2.2 Ge-Sb-Te phases
7
process in PCMs takes place in the nanosecond regime. On the other hand,
crystallization of the amorphous region at room temperature should not happen
within the first 10 years, or even longer, in terms of data retention. This means that,
in terms of the timescale, the difference between crystallization time and retention
time spans roughly 17 orders of magnitude, whereas, regarding the temperature
scale, the corresponding temperature range spans over just a few hundred Kelvin.
Moreover, ultrafast switching behavior relying on non-thermal phase transitions in
GST or GeTe-Sb2Te3 superlattices, triggered by an ultrashort laser pulse, has been
proposed [17-19]. Nevertheless, the understanding of certain material properties and
physical mechanisms of PCMs, as well as developments based upon them, broadens
the perspective for promising future applications.
Ge-Sb-Te phases 2.2
Among PCMs, the Ge-Sb-Te system is the most interesting, in general. Here, a
description of the pseudo-binary GeTe-Sb2Te3 tie-line (Chapter 2.2.1), of the crystal
structure of GST and GeTe (Chapter 2.2.2), and concise compendium of the
crystallization process (Chapter 2.2.3) is given.
Pseudo-binary GeTe-Sb2Te3 tie-line 2.2.1
Figure 2.2 illustrates the ternary phase diagram for this material system with
intermetallic alloys that lie on the tie-line between GeTe and Sb2Te3. Along this
line, the compounds Ge2Sb2Te5, Ge1Sb2Te4 and Ge1Sb4Te7 can be found, in which
the properties change from high crystallization temperature (high stability) to low
crystallization temperature (low stability) [20]. An additional opportunity to tailor
the properties of the compounds on the GeTe-Sb2Te3 tie-line is to dope them with
elements like O, N, Si, etc. [21-23]. As a preliminary example, it was found that the
crystallization temperature Tc of 60 nm thick PLD- deposited GeTe films on Si
substrates could be increased from 240 to 280°C by incorporation of ~16 at.% O
(for details, see Chapter 4.1) [24]. PCMs with such an enhanced thermal stability
are excellent candidates for applications in various fields (i.e., automotive,
aerospace), where the use of conventional GST is limited by its relatively low Tc
(about 140°C) [25].
Chapter 2 Fundamentals: phase-change materials and their applications
8
On the other hand, the materials with fast crystallization are in general of
interest, because a higher data transfer rate for data storage application can be
attained. Figure 2.2 also demonstrates the alteration of the crystallization time (in
terms of switching speed). It is known that the crystallization time can be reduced
from 50 ns for Ge2Sb2Te5 and 40 ns for Ge1Sb4Te7 to 30 ns for Ge1Sb2Te4 [20, 27].
However, stoichiometric GeTe was demonstrated to exhibit a short crystallization
time, but it is highly sensitive to deviations in the composition [28]. Even small
deviations lead to a dramatic increase of the crystallization time. The first reported
short crystallization time for GeTe is 30 ns with a high optical contrast [29];
recently, a value as low as 16 ns was found with a high electrical resistance contrast
[30].
In addition, Coombs et al. [31, 32] proposed that the compounds along the
GeTe-Sb2Te3 tie-line can be defined as materials with so-called nucleation-
dominated or so-called growth-dominated crystallization, depending on the material
properties (kinetics) during the crystallization process. According to these authors,
in a nucleation-dominated crystallization process, the majority of the total
crystallization time is taken by the nucleation process of crystallization while the
growth process of the crystallites is rather short. In contrast, in the growth-
dominated crystallization, nucleation takes place only in a very short time but the
Figure 2.2: Ge-Sb-Te ternary phase diagram illustrates the various phase change alloys.
Stoichiometric compositions lying on the pseudo-binary tie-line of GeTe and Sb2Te3 are presented
(adapted from Refs. [20, 26]).
2.2 Ge-Sb-Te phases
9
growth process takes the majority of the total crystallization time. In this respect,
the crystallization behavior in amorphous GST is, in general, characterized by a
nucleation-dominated growth process [33-35], whereas, in amorphous GeTe,
crystallization is, in general, determined by a growth-dominated growth process
[28, 30, 36]. It is also noteworthy that the interface of a complex PCM-based
memory cell structure can also play an important role with respect to the dominant
crystallization mechanism [30].
Regarding the appropriate phase-change material for the relevant application,
development of optical data storage devices is required in order to improve the
optical contrast for smaller laser wavelengths. Along the pseudo-binary line from
GeTe to Sb2Te3, the optical contrast has been shown to increase with increasing Ge
content at a wavelength of 405 nm [2]. This implies that, along the tie line, a higher
GeTe content is promising for high density optical recording applications. For
electronic memory applications, however, a good balance of the materials
properties, such as fast switching speed and high stability, is required. The most
promising and widely studied material proves to be Ge2Sb2Te5, which is located in
the middle of the tie-line in Figure 2.2. The structural properties of Ge2Sb2Te5 and
GeTe are described in the following.
Crystal structure of GST and GeTe 2.2.2
1.) Structure of GeTe
As shown in Figure 2.3a, the crystal structure of GeTe is a rhombohedrally distorted
rocksalt-type configuration (space group R3m) with Ge and Te atoms, which form
face-centered cubic (fcc) sublattices at room temperature. In this structure, the
lattice constant is a = 0.5985 nm, and a distortion angle of 88.37°, that deviates
from the usual 90° angle along the [111] direction, is present [37-39]. At
temperatures above 420°C, GeTe in the distorted rocksalt structure is transferred
into a higher symmetry rocksalt configuration. The mechanism of the distortion in
rhombohedral GeTe is connected with the Peierls effect [40]. This lattice distortion,
in conjunction with resonance bonding, determines most of the material properties
[41].
In the crystalline state, the arrangement of atoms exhibits an octahedral (six-
fold) coordination with three shorter (0.2843 nm) and three longer (0.3158 nm)
Chapter 2 Fundamentals: phase-change materials and their applications
10
bond distances. Each atom possesses three valence electrons that strive to form six
covalent bonds with the nearest neighbors. However, the number of valence
electrons is insufficient to fulfill the covalent bonding state. The electrons are
delocalized in this resonantly bonded network, leading to a significant increase of
the electronic polarizability. In contrast, upon the structural transition from
crystalline to amorphous state, Ge atom positions exhibit switching from octahedral
to tetrahedral coordination through rupturing of the longer Ge-Te resonance bonds
and strengthening of shorter covalent bonds (as shown in Figure 2.4). The
amorphous GeTe structure can be represented by a random covalent network model
[2, 43]. The average Ge-Te bond length of ~0.260 nm is shorter than that in the
crystalline state [44]. Furthermore, the coordination number of the Ge and Te atoms
shows a four-fold and two-fold coordination in the amorphous state, which is
typically less than in the crystalline state.
2.) Structure of GST
GST possesses a metastable crystalline structure and a stable hexagonal structure.
The crystal structure of metastable crystalline GST can be described approximately
by a rocksalt structure (space group Fm-3m) [2, 45, 46]. Te atoms occupy one fcc
sub-lattice, whereas the other sub-lattice is randomly occupied by Ge and Sb atoms,
as well as by randomly arranged vacancies (~20%), as exemplarily shown in Figure
2.3b. In this structure, the lattice parameter is 0.602 nm. The transformation from
the amorphous to the metastable crystalline phase occurs above a threshold
temperature of 140°C and the phase transition temperature from the metastable
Figure 2.3: Crystal structure of rhombohedral GeTe (a) and cubic GST (b). The red, blue, green and
white spheres symbolize Ge, Sb, Te atoms and vacancies, respectively (adapted from Ref. [42]).
(b)(a)
2.2 Ge-Sb-Te phases
11
crystalline phase to the hexagonal phase is about 310°C [25]. It is noted that the
metastable crystalline phase shows the same type of Peierls-like distortion as in
rhombohedral GeTe. However, it is not surprising that the distorted rocksalt
structure has been found in the GeTe-Sb2Te3 pseudo-binary phase diagram with a
Ge/Sb site vacancy concentration that is proportional to (1-x)/(3-2x) for
(GeTe)x(Sb2Te3)1-x [47].
In the crystalline phase, Ge atoms exhibit an octahedral coordination, as
depicted in Figure 2.4. The crystalline to amorphous transformation process, with
breaking of the longer Ge-Te bonds and subsequent transition of the atoms from
octahedral to tetrahedral symmetry positions, is described in the framework of the
umbrella-flip model [1]. According to Liu et al. [48], 35% of the Ge atoms in the
cubic phase show a tetrahedral coordination, indicating the coexistence of
octahedral and tetrahedral sites for Ge in the GST lattice. The atomic structure of
the amorphous state is still under discussion. Nevertheless, the umbrella-flip model
is an intuitive explanation of ultrafast transition and provides a possible scenario for
the change of the bonding mechanism. Upon amorphization, the average Ge-Te and
Sb-Te bond lengths are 0.261 nm and 0.285 nm, respectively [49]. Moreover, in the
crystalline phase, GST exhibits resonance bonding, whereas in the amorphous phase
(covalent bonding), resonance bonding disappears, due to the loss of the long-range
order. In this amorphous structure (Figure 2.4, right), Ge atoms possess four nearest
Te neighbors in a tetrahedral arrangement and the positions of Sb and Te atoms
resemble those in the rocksalt structure [50].
Figure 2.4: Typical structural switching mechanism in the “umbrella flip-model” of Ge atoms in
crystalline and amorphous GST/GeTe (adapted from Ref. [1]).
Te
Ge
Chapter 2 Fundamentals: phase-change materials and their applications
12
Concise compendium of the crystallization process 2.2.3
In this subchapter, the phase transition from the amorphous to the crystalline state is
shortly considered. The crystallization process in PCMs can be described with the
classical theory of crystallization, which involves the nucleation of crystal seeds
within an amorphous matrix and subsequent growth of the nuclei at the amorphous-
crystalline interface. Nucleation is the process whereby nuclei (seeds) act as
templates for crystal growth. Homogeneous nucleation occurs in the interior of the
parent phase without the involvement of a foreign substance. At temperatures below
a material’s melting point Tm, the driving force for solidification is the difference in
the Gibbs free energy ΔG between the liquid and the solid. If it is assumed that the
heat capacities of the liquid and solid are equal, then the molar enthalpy and molar
entropy of solidification will each remain constant as a function of temperature, and
ΔGhom can be calculated for a spherical seed of radius r as follows [51]
∆𝐺ℎ𝑜𝑚 = −4
3𝜋𝑟3∆𝐺𝑣 + 4𝜋𝑟2𝜎 , (2.1)
where ΔGv is the free energy of the seed crystal and σ is the surface energy. The
second term involves the increase in energy required to form a new surface. The
first term is negative and represents the decrease in Gibbs free energy upon
solidification. Because the second term is a function of the second power of the
radius, and the first term a function of the third power of the radius, the sum of the
two first increases, passes through a maximum, and then decreases as a function of
the seed radius.
The radius at which the Gibbs free energy curve is at maximum is called the
critical radius, r*, for a solid nucleus in a liquid environment
𝑟∗ = −2𝜎
∆𝐺𝑣 . (2.2)
The driving force of the Gibbs free energy will tends to cause a solid particle with a
smaller radius than r* to decrease in size. Thus, this is a particle of subcritical size
for nucleation. A viable nucleus is one with a radius larger than or equal to r*. The
critical Gibbs free energy corresponding to the radius r* is ΔG*. This term can be
shown to be
∆𝐺∗ =16𝜋
3
𝜎3
∆𝐺𝑣2 . (2.3)
2.2 Ge-Sb-Te phases
13
In practice, homogeneous nucleation is rarely encountered in solidification.
Instead, heterogeneous nucleation occurs at crevices, at interfaces, or at impurity
particles suspended in the liquid. The analysis of the energies involved in
heterogeneous nucleation follows the same method as that used for homogeneous
nucleation. In the case of a heterogeneous nucleus, the critical radius and the critical
Gibbs free energy are
𝑟ℎ𝑒𝑡∗ = −
2𝜎𝑠𝑙
∆𝐺𝑣 (2.4)
and
∆𝐺ℎ𝑒𝑡∗ =
16𝜋
3
𝜎𝑠𝑙3
∆𝐺𝑣2 ∙ 𝑓(𝜃) (2.5)
respectively, where σsl is the solid-liquid interface energy. It is important to note
that the critical radius of curvature, r*, does not change when the nucleation
becomes heterogeneous. The critical Gibbs free energy, ∆𝐺ℎ𝑒𝑡∗ , however, is strongly
influenced by the wetting that occurs at the surface of the material that catalyzes the
nucleation. A lower value of ∆𝐺ℎ𝑒𝑡∗ means a lower activation energy to be overcome
in nucleation; in other words, nucleation takes place more easily. The magnitude of
the effect can be appreciated by considering the values of f(θ). f(θ), the so-called
geometrical factor, is the ratio of the volume of the heterogeneous nucleus to the
volume of the sphere with the same radius of curvature.
In the solid-solid transformation (amorphous-crystalline transformation), a small
change in specific volume that occurs must be accommodated elastically, leading to
strain energy effects. Consequently, this strain energy must be involved, together
with the volume-free energy and the interfacial energy changes
∆𝐺(𝑟) = −4
3𝜋𝑟3∆𝐺𝑣 + 4𝜋𝑟2𝜎 + 𝑠𝑡𝑟𝑎𝑖𝑛 𝑒𝑛𝑒𝑟𝑔𝑦 . (2.6)
This strain energy can be calculated (see Ref. [52]) for a spherical seed as
4
3𝜋𝑟3∆𝐺𝐸. Then, in the case of homogenous nucleation, this contribution leads to
∆𝐺(𝑟) = −4
3𝜋𝑟3(∆𝐺𝑣 + ∆𝐺𝐸) + 4𝜋𝑟2𝜎 . (2.7)
Implications:
(a) The interfacial contribution to the nucleation barrier dominates for small seeds
and the volumetric contribution 4
3𝜋𝑟3(∆𝐺𝑣 + ∆𝐺𝐸) for large nucleus sizes. The
Chapter 2 Fundamentals: phase-change materials and their applications
14
competition between these two contributions can produce a complicated
sequence of states for developing phases.
(b) Typically, for solid-state nucleation, the elastic energy contribution dominates
for small particles sizes. As the particle grows, elastic energy can be reduced by
the introduction of misfit dislocations into the interface. Such interfacial
dislocations transform a coherent interface into a semi-coherent interface at the
expense of increased interfacial energy.
(c) With exception of ideal spherical particles, this elastic strain energy is not easy
to express with simple algebraic expressions.
The rate of nucleation per unit volume and unit time, J, is proportional to
(1) the probability, 𝑤 = exp(−∆𝐺∗/𝑘𝐵𝑇), that a thermodynamic fluctuation of
the critical free energy ΔG* is given,
(2) the number of the growth species per unit volume, n, that can be used as
nucleation seed points, given by Co (in the simplest case of the homogenous
nucleation, Co is given by the initial concentration),
(3) the jump frequency of the growth species Γ from one site to another is given
by 𝛤 = 𝑘𝐵𝑇 3𝜋𝜆3𝜂⁄ , where λ is the diameter of the growth species and η is
the viscosity of the solution (for simplest case that solid precipitates are
formed in a solution).
Thus, the rate of nucleation can be described by
𝐽 = 𝑛 ∙ 𝑤 ∙ 𝛤 = 𝐶𝑜𝑘𝐵𝑇
3𝜋𝜆3𝜂∙ 𝑒𝑥𝑝 (−
∆𝐺∗
𝑘𝐵𝑇) . (2.8)
This equation indicates that a high initial concentration (supersaturation), i.e., a
large number of nucleation sites, a low viscosity, and a low critical energy barrier
favor the formation of a large number of nuclei. For a given concentration of solute,
a larger number of nuclei means smaller-sized nuclei.
The size distribution of the formed crystallites is dependent on the subsequent
growth process of the nuclei. The growth process of the nuclei involves several
processes. This multi-step process can be separated into (i) the generation of seed
points, (ii) the diffusion of the growth species, (iii) the adsorption of the growth
species onto the growth surface, and (iv) growth of the seeds through irreversible
incorporation of growth species. These steps can be further grouped into two
processes as follows. Supplying the growth species to the growth surface is termed
2.2 Ge-Sb-Te phases
15
diffusion, which includes the generation, diffusion, and adsorption of growth
species, whereas incorporation of growth species adsorbed on the growth surface
into solid structure is denoted as growth.
Growth controlled by diffusion
When the concentration of growth species declines below the minimum
concentration for nucleation, the nucleation process stops, whereas growth
continues. If the growth process is controlled by the diffusion of growth species
from the bulk solution to the particle surface, the growth rate is given by [53]
𝑑𝑟
𝑑𝑡=
𝐷(𝐶−𝐶𝑠)
𝑟𝑉𝑚 , (2.9)
where a spherical nucleus with the radius r is assumed, D is the diffusion constant
of the growth species, Cs is the concentration on the surface of the formed
crystalline precipitate, and Vm is the molar volume of nuclei. The solution of this
differential equation under the assumption that the initial size of the nucleus, ro, and
the change of the bulk concentration are negligible leads to
𝑟 = √2𝐷(𝐶 − 𝐶𝑠)𝑉𝑚𝑡 + 𝑟𝑜2 . (2.10)
For two crystalline precipitates with an initial radius difference of Δro the radius
difference Δr decreases as time increases (or as precipitates grow larger), according
to
∆𝑟 =𝑟𝑜
𝑟∆𝑟𝑜 . (2.11)
Combining Equations 2.10 and 2.11, the evolution of the radius as result of a
diffusion-controlled process is
∆𝑟 =𝑟𝑜
√2𝐷(𝐶−𝐶𝑠)𝑉𝑚𝑡+𝑟𝑜2
∆𝑟𝑜 . (2.12)
This indicates that the radius difference decreases with the increase of radius of the
precipitation and prolonged growth time. Consequently, diffusion-controlled growth
promotes the formation of uniform crystallites in an amorphous ambience.
Growth controlled by interface reaction
When the diffusion of growth species from (as here) the amorphous bulk material to
the growth surface of the crystalline precipitate is sufficiently fast, i.e., the
concentration on the surface is the same as that in the bulk, the growth rate is
controlled by the interface reaction process. The growth rate is thus proportional to
the surface area
Chapter 2 Fundamentals: phase-change materials and their applications
16
𝑑𝑟
𝑑𝑡= 𝐾(𝐶) ∙ 𝑟2 , (2.13)
where K(C) is a proportionality constant that is dependent on the concentration of
the growth species. Solving this equation, the evolution of the crystalline precipitate
radius is given by
𝑟 =1
1𝑟𝑜
⁄ −𝐾(𝐶)𝑡 . (2.14)
The radius difference increases with an increasing radius of the nuclei according to
∆𝑟 =𝑟2
𝑟𝑜2 ∆𝑟𝑜 . (2.15)
Combining Equations 2.14 and 2.15 yields
∆𝑟 =𝛥𝑟𝑜
1−𝐾(𝐶)𝑡 . (2.16)
Equation 2.16 demonstrates that the radius difference increases with a prolonged
growth time, if an interface reaction-controlled process for the growth is assumed.
Consequently, this growth mechanism does not favor the synthesis of mono-sized
crystallites.
For the kinetics of phase transformation, an expression of the temperature
dependence of the growth velocity can be derived as [51, 54]
𝑢 ∝𝑇
𝜂∙ (1 − 𝑒𝑥𝑝 (−
∆𝐺
𝑘𝐵𝑇)) . (2.17)
This illustrates that a high growth velocity of phase transformation occurs for low
viscosity η and/or a high free energy ΔG to form a crystallite from the liquid state.
Self-evidently, the viscosity and the free energy are temperature dependent, as well.
Recently, J. Orava et al. investigated the kinetics of crystallization in GST by ultra-
fast DSC measurements [55, 56]. They demonstrated that the growth velocity gets
highest at an intermediate temperature between the glass transition temperature Tg
and the melting temperature Tm. At this intermediate temperature both viscosity and
free energy contribute significantly to the growth velocity. For temperatures slightly
above Tg the growth velocity is several orders of magnitude smaller than at the
above mentioned intermediate temperature, because the velocity is limited by the
still high viscosity [55]. For temperatures very close to Tm the growth velocity gets
again small. Consequently, in order to achieve the fastest crystallization, the laser
pulse induced or current pulse induced temperature rise must be optimized. In the
2.3 Applications of phase-change materials
17
case of data storage applications, this corresponds to attaining a high data transfer
rate.
Applications of phase-change materials 2.3
As mentioned above, PCMs can be rapidly and reversibly switched between an
amorphous and a crystalline state. Because both states are characterized by strong
differences in the optical and electrical properties, these materials can be employed
for data storage, including rewritable optical data storage (Chapter 2.3.1) and
electronic memory (Chapter 2.3.2). Based on this mechanism, a series of frontier
applications of phase-change materials (Chapter 2.3.3) is generated.
Optical data storage 2.3.1
Yamada et al. [20, 27] introduced the re-writable (RW) optical data storage based
on Ge-Sb-Te alloys. Within the last decades, commercial products based on PCMs
have been used in optical storage. They have progressed from compact discs (CDs)
with the operational wavelength λ = 780 nm and a storage capacity of 650 MB in
1997 over digital versatile discs (DVDs) with λ = 650 nm and 4.7 GB in 2000 to
recordable/erasable Blu-ray discs (BD-REs) with λ = 405 nm and 23.3 GB in 2003.
Based on the fixed operation wavelength of 405 nm, together with the so-called
multilayer technology, a dual-layer BD-RE with a capacity of 50 GB followed in
2004 and, later on, a triple-layer BD-RE with 100 GB storage capacity in the year
2011 [57]. These stages of development are depicted in Figure 2.5, which shows the
progress of data storage on optical discs, based on the phase-change optical
recording technology.
Basically, a higher storage density can be expected by either reducing the laser
wavelength or by increasing the numerical aperture of the objective lens (NA),
because of the storage density D ∝ (NA/λ)2, where λ is the laser wavelength. In this
thesis, the UV-wavelength of 248 nm is used for triggering the phase transition of
phase-change materials described in Chapters 4.2 and 4.3. Another approach uses
the near-field optical technique to realize an increase of the storage density,
whereby a nano-aperture or a solid immersion lens focuses the laser beam into a
very small spot on the disc surface. However, this technique is limited to a working
distance of < 40 nm, which is a challenge for practical fabrication and operation of
Chapter 2 Fundamentals: phase-change materials and their applications
18
an optical disc. Nonetheless, this led to the development of optical discs based on a
super-resolution near-field structure (so-called Super-RENS), which are able to
record and read out the marked regions beyond the diffraction limit [58]. In spite of
this, the detailed working mechanism of such Super-RENS is still not completely
understood [15]. Alternatively, multi-layer recording technology has been used for
improving the commercially available recording density from 23.3 to 100 GB, as
shown in Figure 2.5. More recently, multi-level data storage has been attracting
increasing attention because of the possibility to increase the storage density even
further. This method is based on a uniquely precise control of the amorphous to
crystalline transition process, in order to realize different (non-binary) reflectivity
levels in each single layer memory cell, instead of using multi-layers. Thus, multi-
level storage techniques are not only beneficial for the increase of storage density,
but they also can potentially increase the data transfer rate and decrease the cost per
bit (for more details, see Chapter 4.3).
Another important trend is the desire for a high maximum data transfer rate.
Over the last 20 years [57], an increase of the data transfer rate from about 10 to
Figure 2.5: Progress of phase-change based re-writable optical storage media (top part adapted from
Ref. [2]): A redesign of the optics, i.e., usage of shorter wavelengths and higher numerical apertures
(NA) allowed for a reduction of the cross-section of the focused laser beam and thereby for a
decrease of the minimal mark size and a corresponding increase in recording capacity.
1995 2000 2005 2010 2015
0
50
100
Cap
acity (
GB
)
Release date (Year)
λ=780 nm λ=650 nm λ=405 nm λ=405 nm λ=405 nm
NA=0.85 NA=0.85 NA=0.85NA=0.6NA=0.5
1.2 mm
CD-RW
0.6 mm 0.1 mm 0.075 mm 0.057 mm
DVD-RAMBD-RE (Single)
BD-RE (Dual)
BD-RE (Triple)
0.018 mm0.025 mm
112 nm149 nm149 nm410 nm833 nm
CD-RWDVD-RAM BD-RE
(Single)
BD-RE (Dual)
BD-RE (Triple)
2.3 Applications of phase-change materials
19
133 Mbit/s has been developed. The data transfer rate of a PCM is generally
associated with the amorphization and crystallization rates of its recording material
(see also Chapter 2.1). The amorphization process consists of an inherently quick
melt-quenching process, whereas crystallization is characterized by a minimum
time required for the generation of crystalline phase nuclei and their growth.
Therefore, the maximum attainable data transfer rate is mainly determined by the
crystallization kinetics of the PCM (see Chapter 2.2). According to classical
crystallization theory, nucleation and growth rates are dependent on the temperature
of the respective material. By using laser pulses as the energy source, an increase of
the temperature in the material and the induction of phase transformation in PCM
must be considered. In the case of optical switching, this energy transfer is usually
realized by supplying additional heat directly, via photon energy transfer to excited
electronic states, and subsequent relaxation [1, 59]. Coherent light is used in optical
switching to induce the phase transformation. Thereby, by reducing the irradiated
laser pulse duration, which leads to a short energy deposition time for phase
transitions in PCMs, performance at a high data transfer rate can be expected.
Various experimental studies on the property changes of PCMs upon nano-, pico-
and femtosecond laser pulse irradiation have been published in recent years [31, 60-
67]. However, comparative examinations are still lacking. Thus, the investigation of
an experimental design for a comparison of the process of GST crystallization
induced by single laser pulses in the time scale frame of ns and fs, as well as a
corresponding model of the average contribution of ns and fs laser photons to the
phase transition process, are presented in Chapter 4.2.2, with the aim of
understanding the switching process and increasing the data transfer rate.
Electronic memory 2.3.2
Similar to the optical data storage (as described in Chapter 2.3.1), PCMs are also
used as the active layer, in the form of phase-change random access memory
(PCRAM). In the year 1960, Ovshinsky observed I-V behavior in amorphous
chalcogenide glasses with a characteristic reversible switching mechanism between
high resistance state and low resistance state [4]; this principle is used for current
electronic data storage applications. Presently, the known remarkable properties of
PCRAM are non-volatility, good scalability (in terms of minimized device feature
Chapter 2 Fundamentals: phase-change materials and their applications
20
size F), long endurance, short write/read (W/E) time, and multi-level cell (MLC)
capability, leading to PCRAM becoming competitive with solid-state electronic
memory. A comparison of the device characteristics of various memory
technologies is summarized in Table 2.1.
PCRAM devices use the different values of electrical resistivity upon rapid and
reversible phase transitions between the amorphous and the crystalline states of the
PCMs. The physical principle of switching PCMs relies on the atomic configuration
changes illustrated in Chapter 2.1. For practical memory applications, with logical
“0” in the cell, in which the PCM is in the crystalline state that exhibits the low
resistivity state (LRS). To write a logical “1” in the cell, a short high power pulse is
applied to the PCM cell, leading to amorphization of a partial volume of the PCM
and generating the high resistivity state (HRS), named RESET operation. In reverse,
a pulse of moderate power but with sufficiently long duration is used for
crystallization of this amorphous volume, resulting in erasing the “1” and returning
to “0” memory state, named SET operation. Due to the significant difference
between the HRS and LRS (up to five orders of magnitude), the process logical
states are easily read out through a lower power pulse. In addition, the threshold
switching effect is essential for the operation of the PCM cell, because it is
Table 2.1: Comparison of various memory technologies [3]: Dynamic random-access memory
(DRAM), NAND fresh memory, phase-change random access memory (PCRAM), ferroelectric
random access memory (FeRAM), magnetoresistive random-access memory (MRAM), resistive
random-access memory (ReRAM).
Attributes DRAM NAND PCRAM FeRAM MRAM ReRAM
Non-Volatile No Yes Yes Yes Yes Yes
Cell size 6F2 4F
2 4F
2 22F
2 20F
2 4F
2
Read time <10 ns 0.1 ms 12 ns 40 ns 35 ns <50 ns
W/E time <10 ns 1/0.1 ms 100 ns 65 ns 35 ns <1 ns
Endurance >1015
>105 >10
12 >10
14 >10
12 >10
9
Retention 64 ms >10 yrs >10 yrs >10 yrs >10 yrs >10 yrs
Write Energy ~10
-15
J/bit
~10-16
J/bit
~10-12
J/bit
~10-14
J/bit
~10-12
J/bit
~10-13
J/bit
MLC No No Yes No No Yes
2.3 Applications of phase-change materials
21
accompanied by an abrupt transition from amorphous (HRS) to crystalline (LRS)
states. Otherwise, extremely high voltages would be required to introduce enough
power into the cell to heat it above the crystallization temperature, unless the
occurrence of threshold voltage switching [68, 69]. A corresponding typical current-
voltage (I-V) curve for a PCRAM and a schematic picture of a PCRAM in cross-
sectional view are shown in Figure 2.6.
With respect to the attributes of PCRAM (see Table 2.1), scalability is one of
the dominating issues that must be addressed in order to enhance PCRAM to next
generation non-volatile memory applications. Scaling down the feature size is not
only advantageous for increasing the storage density, but, even more important, for
the reduction of the operation energy consumption; in particular, cutting off the
RESET current. The structure design of the PCRAM cell plays an essential role,
here. When attempting to confine the current flow in the PCRAM device, Joule
heating of the PCM can be restricted and optimized. As an example, a typical
mushroom-cell structure of a PCRAM is shown schematically in Figure 2.6b.
Following an applied electric pulse (irrespective of its polarity) with a voltage
above the threshold voltage acts as Joule heating source, narrowing the contact area
between PCM and the heat contact can be used to gain an increased heating
efficiency in the PCM layer and to induce a local switching process from the high-
resistive amorphous phase to the low-resistive crystalline phase. However, scaling
down memory size below a certain dimension of < 50 nm is challenging because of
Figure 2.6: (a) Typical current-voltage (I-V) curve of SET and RESET operation of a PCM-based
device. The threshold switching behavior occurs at a threshold switching voltage (Vth). (b) A cross-
sectional schematic of a PCRAM mushroom-cell.
S
Bottom electrode
SiO2
PCM
Heater
Switchable region
Top electrode
Crysta
lline st
ate (L
RS)
Curr
ent (A
)
Voltage (V)
Amorphous state (HRS)
RESET
SET
Am
orp
hiz
ation
Cry
sta
llizatio
n
Threshold
voltage (Vth)
(a) (b)
Chapter 2 Fundamentals: phase-change materials and their applications
22
the spatial limitations of lithographic device fabrication. As an alternative
technology, probe-based storage (using an AFM tip as probe for writing, reading,
and erasing operations in local PCM layers) has been attracting increasing attention.
It exhibits remarkably advantageous features [70-74]: i) higher data storage density
> Tbit/in.2, ii) lower power consumption of < 100 pJ per bit written, iii) higher data
transfer rate with large arrays of tips operating in parallel, iv) less expensive, in
comparison with the process for conventional memory device fabrication. In the
present thesis, conductive-AFM is used as an active tool for the investigation of
GST phase-change films at the nanoscale. This is not only a breakthrough regarding
the bottleneck of device scalability but, more importantly, it reveals deep insights
into the switching mechanism by simultaneously measuring electrical and
topographical properties of phase-change films at nanoscale (for more details see
Chapter 4.4).
Frontier applications of phase-change materials 2.3.3
Over the last decades, the functionality of chalcogenide-based PCMs has been
further developed with respect to the targets of optical data storage and electronic
memory applications. Simultaneously, a remarkable improvement in the
understanding of structural, optical and electrical properties of the materials with
the capabilities of nanoscale, non-volatile, prominent contrast of optical and
electrical properties between amorphous and crystalline states, fast switching speed,
multilevel storage, and 3D integration has occurred. Such a significant functionality
extension of PCMs opens the door for several promising frontier applications. For
instance, Michel et al. demonstrated that, by introducing PCM thin layers, a non-
volatile tunability of the resonance wavelength in an antenna can be achieved [75].
Similarly, the combination of surface plasmon polaritons with PCMs, with the
purpose of controlling surface plasmon waveguides, has been explored [76].
Another insight was used to uniquely combine electro-optical properties of PCMs,
in order to reveal possible applications in displays and data visualization [77].
Furthermore, a memristor capable of Boolean logic function conjunction with
memory in a single PCM cell has been proposed [78], as well as the use of PCM
devices for emulating biological synapses and neurons [79-83], which paves the
way for neuromorphic computing.
23
Chapter 3
Experimental conditions and methods
Pulsed laser deposition of thin films 3.1
Pulsed Laser Deposition (PLD) is used to deposit thin films of the phase-change
material system Ge-Sb-Te. In contrast to other physical deposition technologies, the
main advantages of PLD are (1) high quality films can be grown reliably in a very
short time and (2) the deposition of films with unusual composition and with
stoichiometric transfer of a target material to the film is possible. In addition, (3) the
vacuum system and the laser beam source are decoupled, (4) a simple vacuum
system is required, and (5) the electrical properties of the target are not crucial (e.g.,
compared to dc sputtering), and (6) PLD enables precise control of the growth rate
(sub-monolayer deposition per pulse).
The basic experimental setup is shown in Figure 3.1. A pulsed KrF excimer
laser with a wavelength of 248 nm and 20 ns pulse duration is applied. The laser
beam is focused, via a plano-convex lens (focal length = 800 mm) through a quartz
window, onto the surface of the rotating target at an incident angle of 60° with
respect to the target normal. The pulse energy is varied between 160 and 200 mJ,
Figure 3.1: Schematic of a pulsed laser deposition system.
Rotating target
Rotating
substrate
Plume
Focus lens
Pulsed laser beam
Target
Substrate
Heater
Chapter 3 Experimental conditions and methods
24
which leads to a fluence within a range between 1.0 and 1.5 J/cm2. The pulse
repetition rate is typically set to 10 Hz. The target materials are arc melted GeTe,
Ge2Sb2Te5, and LaAlOX, targets, which are placed on the multi-target holder.
Ge2Sb2Te5 and GeTe have a high absorption coefficient in the UV range (~106
cm-1
)
[84, 85] and are therefore suitable for the PLD process. The working pressure in the
PLD chamber is typically 5×10-8
mbar during deposition. The substrates (1×1 cm2)
are ultrasonically cleaned in ethanol and deionized water prior to deposition and are
positioned inside the vacuum chamber parallel to the target surface at a target-to-
substrate distance of 7.4 cm. Both target and substrates are rotated in order to avoid
deep ablation of the target and to improve the thickness homogeneity of the films,
respectively.
A resistive heater source is mounted in a specific position on the substrate
holder, near the substrates, in order to ensure an effective thermal dissipation when
increasing the substrate temperature to values required for the experiments (max.
temperature up to 1100°C).
The basic processes of PLD are the absorption of the laser radiation by the
target material, generation of a plasma plume, expansion of the plasma, and
thermalization and condensation of the target material on the substrate. In detail, the
laser beam is focused on the target material surface, causing the near-surface region
of the target to melt and then vaporize during the early stage of the ablation
processes. The continuing laser energy transfer within one laser pulse and
absorption of the laser light cause an increase in the target material vapor to high
temperatures and high pressure and, eventually, the ionized gas expands from the
target in the direction of the substrate; this is the so-called plasma plume. The
plasma plume contains a number of different species, including electrons, ions,
atoms, molecules, etc. If background gas is present in the deposition chamber, the
thermalization of the plume takes place through reduction of the kinetic energy of
the plume species. Theoretical models of interaction between plasma expansion and
background gas have been described, e.g., in Ref. [86]. Finally, the plume particles
are deposited on the substrate surface. Especially at higher substrate temperatures,
these particles are able to diffuse onto the surface and find the energetically most
3.2 Thermal treatment
25
favorable sites. For more detailed information about PLD, the reader is referred, for
example, to Refs. [7, 8, 87, 88].
Thermal treatment 3.2
The thermal treatment of the thin films is carried out by in situ annealing in the
deposition chamber during thin film deposition (Chapter 3.2.1), and by ex situ
annealing in an infra-red-irradiation oven following the thin film deposition process
(Chapter 3.2.2).
In situ thermal treatment in deposition chamber 3.2.1
To investigate the influence of substrate temperature on the structural properties of
the deposited GeTe and GST films, the substrate is heated by a resistive heater
source, which is located behind the substrate holder in the deposition chamber, as
depicted in Figure 3.1. The heater temperature is normally set to temperatures in the
range between 100°C and 500°C. The temperature on the surface of the substrate
materials is calibrated by connecting a thin (0.25 mm) type C thermocouple wire
with the substrate surface. The measurement point is pinned to the substrate surface
using ceramic cement (Omega bond CC High Temp). The results of the calibrations
on Si (100) and BaF2 (111) substrates are shown in Figure 3.2. The uncertainty of
the measurements is estimated to be ±30°C [26].
Ex situ thermal treatment in an infra-red irradiation oven 3.2.2
The post-thermal annealing of deposited GeTe or GST films is realized by an infra-
Figure 3.2: Substrate temperature calibrations.
0 100 200 300 400 500
100
200
300
400
Si (100)
BaF2(111)
Tsub
= -19.7°C+0.98Theater
Tsub
= 1.65°C+0.65Theater
Substr
ate
Tem
pera
ture
(°C
)
Temperature Heater (°C)
Chapter 3 Experimental conditions and methods
26
red irradiation oven system. The heating furnace features a high-accuracy
temperature control and is capable of rapid heating (> 50 K/s). As-deposited films
were annealed under high vacuum condition (5×10-6
mbar) in the temperature range
between 100 and 500°C for 20 min at a heating rate of 10 K/min. Furthermore, in
order to study the temperature dependence of the electrical resistance in vacuo, the
furnace was connected with a Keithley 2400 source measure unit in a two-point
probe setup, using gold wires for contacts, as shown in Figure 3.3.
Laser pulse irradiation for optical switching 3.3
Nano- and femtosecond laser irradiation 3.3.1
A schematic of the static ns- and fs pulse laser irradiation experiment setup is
Figure 3.3: Schematic of oven system with the possibility of temperature-dependent resistance
measurement.
Figure 3.4: Schematic of the static laser irradiation experiment setup.
Gas
Heating source: halogen lamp
Thermocouples
Keithley 2400 Source
Measurement unit
Excimer laser 248 nm
Pulse duration 20 ns or 500 fs
Convex lens
Filter
Gaussian lens
GST or GeTe
SiO2/Si
3.3 Laser pulse irradiation for optical switching
27
shown in Figure 3.4. Two laser source systems at the same wavelength of 248 nm,
but with different pulse lengths, are applied to investigate the influence of the pulse
duration on the crystallization process in the phase-change films. For the ns laser, a
KrF excimer laser with pulse length of 20 ns is utilized. The fs laser is a hybrid
excimer-dye laser with a pulse length of 500 fs. The two laser beams are focused by
a convex lens onto the surface of the deposited thin films. The average fluence is
estimated as the laser energy divided by the spot size. The laser fluence can be
varied between 0 and 200 mJ/cm2 and between 0 and 19 mJ/cm
2 for the ns pulse
laser and the fs pulse laser irradiation, respectively.
Nanosecond laser pump-probe system 3.3.2
In order to investigate the dynamics of phase transition in phase-change films, an
optical pump-probe system, based on the ns pulse laser irradiation setup, was
constructed. The schematic arrangement of the ns pump-probe system for real-time
optical reflectivity experiments is shown in Figure 3.5. The pumping source is a
KrF excimer laser with a wavelength of 248 nm and a pulse duration of ~20 ns. The
pumping laser beam is focused by a convex lens on the surface of the sample, in
order to induce local phase changes. The spot size of the rectangular laser beam is
about 24 × 6 mm2 and the laser fluence is varied between 0 and 182 mJ/cm
2 by
adjusting the laser pulse energy from 0 to 300 mJ. A continuous-wave (CW)
Figure 3.5: Schematic of the ns laser irradiation optical pump-probe system setup.
Excimer laser (248 nm, 20 ns)
Convex lens
Filter
Gaussian lens
Oscilloscope
Diode laser
405 nmGeTe
SiO2/Si
LaAlOX
Photodiode
Chapter 3 Experimental conditions and methods
28
probing laser beam from a laser diode with a wavelength of 405 nm is focused on
the center of the irradiated spot (d ~2 mm) at an incident angle of about 30° with
respect to the sample surface. The reflected probing beam is collected by a
high-speed Si photo detector (rise/fall time: 1 ns) equipped with a front UV filter, to
reduce the influence of the UV laser, and is connected via a coax cable to a fast
digital 500 MHz oscilloscope. The total time resolution of the pump-probe system
is about 2 ns. A second photodetector (without filter) is employed to detect the
shape of the pumping excimer laser beam as well as to calibrate the time delay
between the measured signal of the excimer laser and the diode laser. Moreover, the
influence of CW probing laser irradiation with a maximum power density
~0.7 W/cm2 on the sample is examined by a long duration (t > 60 s) analysis of the
reflection signals; this indicated that the CW probing laser irradiation does not
influence the optical signal from the sample. The layer structure of samples, as
shown in Figure 3.5, is an example with LaAlOX serving as a capping layer to
protect degradation of the active GeTe layer on SiO2/Si substrate. It should be noted
that the absorption of the LaAlOx layer with thickness ~5 nm as used in this study
upon laser irradiation wavelength of 248 nm can be neglected.
Arrangement for electrical switching 3.4
In this work, the electrical resistance switching in the memory stack (50 nm
GST/100 nm Cr/Si) is systematically investigated at the nanoscale by means of a
conductive atomic force microscope (C-AFM) based setup for write, read and erase
Figure 3.6: Schematic of a conductive atomic force microscope configuration-based setup for
electrical pulse- induced experiments.
LaserPhotodetectorpA-Amplifier
to A/D
Cr 100 nm
GST 50 nm
Si
HeaterCeramic
N2 gas
Thermocouple
type C
DC bias voltage
Conductive tip
Au wire
AFM stage
3.4 Arrangement for electrical switching
29
operations, and is shown in Figure 3.6. The measurements are performed using an
AFM equipped with a tunneling atomic force microscopy extension that is used for
C-AFM-type investigations. A commercial Pt-Ir-coated Si AFM tip, with a tip
radius of 20-25 nm as the probe, serves as a nanoscale contact (top electrode) to the
investigated phase-change film. The spring constant of the utilized cantilever is 2.8
N/m. During the experiments, a variable bias voltage is applied across the phase-
change thin films. The backside electrode (Cr layer layer) and the top electrode (PtIr
coated tip) are in series with a current amplifier in the range of 1 pA to 1 µA. The
tip can be used to perform a static measurement. Alternatively, it can be scanned in
contact mode. When the tip is used in static mode, the local nanoscale I-V curve of
the phase-change film is measured. By programming a controlled motion scheme of
the tip, an array of marks can be written into the amorphous phase-change films by
applying a voltage that exceeds a certain threshold switching voltage. During the
scan process of the tip, and maintaining a constant force between the tip and
sample, it is possible to record topographic and current images of the films
simultaneously. This enables the direct correlation of local nanoscale topography
with local electrical properties. In this study, the ramp parameters of the voltage
sweep are between -3 V to +3 V with a ramp frequency between 0.15 Hz and 10
Hz. Scanning is usually performed on areas of 500 × 500 nm² or 1 × 1 μm² with a
scanning frequency between 0.15 Hz and 0.2 Hz at a DC sample bias of 0.5 V; this
is applied to record the topography-current-images. A calibration of measured I-V
curves, using a standard resistor (100 MΩ) with a measurement sensitivity of 1
nA/V (limited current range between -12 nA and +10 nA), is shown in Figure 3.7a.
Figure 3.7: (a) Calibration of a standard resistance with value of 100 MΩ in I-V characterization by
C-AFM, (b) Calibration of substrate temperature as a function of supply power.
0 2 420
40
60
80
Measurement points
Fit by y=27+11.3xSubstr
ate
tem
pera
ture
(°C
)
Power (W)
-3 -2 -1 0 1 2 3-15
-10
-5
0
5
10
Cu
rre
nt
(nA
)
Voltage (V)
Forward
Reverse
Calib. Res.100 M
Sen. 1 nA/V
(a) (b)
Chapter 3 Experimental conditions and methods
30
The sample temperature is controlled by heating with a Peltier element (calibration
substrate temperature vs. heating power is shown in Figure 3.7b). In order to
minimize the oxidation of the films, the film surface is surrounded by protective
nitrogen gas. With this experimental arrangement, the measurement of I-V curves
as a function of the sample temperature can be realized.
Characterization methods 3.5
To investigate the structure, chemical composition, morphology, crystallization
kinetics and optical properties of the as-deposited films and of the films that
underwent a phase transformation induced by thermal annealing via laser irradiation
or via an electric current flow, a variety of characterization methods is used. In this
subchapter, a brief description of these techniques, including X-ray diffraction-
based analysis (Chapter 3.5.1), electron microscopy (Chapter 3.5.2), atomic force
microscopy (Chapter 3.5.3), spectroscopy and calorimetry methods (Chapter 3.5.4)
is given.
X-ray diffraction-based analysis methods 3.5.1
X-ray diffraction (XRD) is applicable to the analysis of the crystalline structure of
the GeTe and GST thin films. The principle of XRD measurements is based on the
interference between incident X-rays and periodically arranged atoms in a lattice;
i.e., the beam is diffracted only in the case of specific specular incidence angles θ
with respect to the family of lattice planes. The condition of maximum
diffracted/reflected intensity is determined by the Bragg law
2𝑑ℎ𝑘𝑙 sin 𝜃 = 𝑛𝜆 , (3.1)
where dhkl is the interplanar spacing, n is the diffraction order, λ is the X-ray
wavelength.
In this thesis, the samples are measured with a Rikagu Ultima IV Type 3
diffractometer in parallel beam geometry, using a θ-θ goniometer and parallelized
Cu K radiation (λ = 0.15418 nm). A schematic of the goniometer axes is shown in
Figure 3.8. A special feature of this arrangement is that the detector arm can be
rotated around two axes: the usual 2θ axis, as well as the additional 2θχ axis. The
latter is called in-plane axis.
3.5 Characterization methods
31
In order to obtain information about various properties of the deposited thin
films, several corresponding measurement techniques are performed, on the basis of
the utilized XRD setup, as briefly described in the following:
2θ-ω scan measurements
2θ-ω scans allow the determination of the interplanar spacing of families of lattice
planes that are oriented parallel to the sample surface, the underlying crystal
structure(s) that are present and the preferred orientation of crystallites. This can be
achieved by altering the incident X-ray beam angle ω and simultaneously altering
the detector at the angle of (2θ), while the sample remains stationary. Importantly,
ω is equal here to the angle θ between the incident X-rays and the sample surface.
The 2θ angular step width of the scans is 0.05° and an angular resolution of 0.11° is
determined by the use of a parallel slit analyzer. According to Bragg’s law
(Equation 3.1), from a distinct 2θ position of a diffraction peak, the corresponding
interplanar spacing dhkl can be deduced. In addition, the full width at half maximum
(FWHM) of the detected diffraction peak, also called Δ(2θ) can be used to estimate
the approximate average size of crystallites. This relationship is expressed by the
Scherrer formula [89]:
∆(2θ) =0.9𝜆
τ∙cos 𝜃𝐵 , (3.2)
where τ is the average size of the crystallites (average grain size), and θB is half of
the Bragg angle.
Figure 3.8: Schematic of the goniometer geometry: ω angle between incident x-ray beam and sample
surface, 2θ angle between incident x-ray beam and detector, 2θχ in-plane angle, φ sample rotation
angle around the sample normal, χ sample tilt angle around the axis perpendicular to 2θ and χ axes.
φ
χ
X-ray detectorX-ray source
Sample
ω 2θ
2θχ
Chapter 3 Experimental conditions and methods
32
In order to obtain a higher diffracted intensity from the polycrystalline GeTe
films, despite the small film thickness, grazing incidence diffraction (GID) at a
fixed incidence angle ω of 1° with respect to the sample surface is employed as
well. In this measurement mode, only the X-ray detector scans the angle range of
interest during the measurement.
Rocking curve measurement
Rocking curve measurements of distinct Bragg reflections are used to obtain the
average tilt distribution of the related crystallites in the films. The precondition for
this type of measurement is that the 2θ and ω axes are oriented to the sample in such
a way that Bragg’s law is fulfilled for the chosen reflection. Then, in the case of the
utilized θ-θ goniometer, the rocking curve measurement is realized by a coupled
movement of the ω axis and the 2θ axis, whereby the angle between the two axes
remains constant. This scan is performed around the position given by the Bragg
law. For typical thin films, the maximum intensity of the detected rocking curve
peak can usually be found near (2θ)/2; the corresponding FWHM of the rocking
curve is an estimate of the average tilt of crystallites around the normal of the
chosen family of planes.
In-plane pole figure measurement
Pole figure measurements are typically used to determine the orientation
distribution of crystallites in textured or epitaxial thin films. First, a distinct Bragg
reflection is chosen. Then, the pole figure measurement is realized by rotating the
sample around its normal from 0 to 360° (complete φ-Scan) for polar angles χ
ranging from 0 to 90°. A combined movement of the ω, 2θ, and 2θχ axes is
performed to realize the various polar angles. Thus, the pole figure measurement is
called in-plane pole figure measurement. The collected diffracted intensity of the
{hkl} family of planes at each position (χ, φ), is dependent on the actual crystal
structure of the sample materials and on the orientation of the crystallites in the
sample. The recorded intensities of all (χ, φ) positions are then projected on the
sample plane in stereographic projection. The angles χ and φ are also called α and β,
respectively. In the resulting pole figure, the intensity at each angle is dependent on
the number of crystallites in the films that are oriented so that they fulfil the Bragg
condition for the pre-selected reflection. For a fiber textured film, continuous rings
3.5 Characterization methods
33
of higher intensity centered at or near the origin of the pole figure are typically
encountered. For an epitaxial film, a number of separate small spots of higher
intensity, the so-called pole density maxima, can be found in a well-defined
arrangement, and are characteristic of a highly preferred orientation of the film. In
order to evaluate actually measured pole figures, predicted pole figure models were
created by CaRIne crystallography software [90] and compared to the measured
ones.
φ-scan measurements
Instead of recording full pole figures, which include intensity data from the whole
northern pole sphere, it is, in many cases, sufficient and much more efficient to
measure partial pole figures or even single φ-scans. Such φ-scans are used here for
determining the orientation relationships between epitaxial films and the underlying
single crystal substrate. For this purpose, a distinct reflection from the film, as well
as from the substrate, is selected.
X-ray reflectivity measurement
Apart from the information about crystal structure and texture, the thickness,
density and surface/interface roughness of deposited thin films on a substrate can be
measured with an X-ray diffractometer by using X-ray reflectivity (XRR) [91]. In
terms of measurement geometry, an XRR measurement is, in principle, similar to a
2θ-ω scan, but in XRR mode, the X-ray beam is incident at a comparably small
angle relative to the sample surface. As long as the incidence angle lies below the
critical angle of total reflection (i.e., θc< 0.5°), total reflection of the X-ray beam
takes place, i.e., the X-ray reflectivity is maximal. When the incident angle
increases to values above θc, the reflectivity decreases drastically. In the case of a
homogeneous thin film on a flat substrate, the X-ray reflectivity curve exhibits an
intensity oscillation, due to constructive/destructive interference of X-rays reflected
at the surface/interfaces. The period of the oscillation is related to the film thickness
and the amplitude of the oscillation to the film density, substrate density and
surface/interface roughness. The film density can be determined from the critical
angle of total reflection. By evaluating the measured reflectivity profile, information
about film thickness, density and surface/interface roughness can be extracted. This
Chapter 3 Experimental conditions and methods
34
evaluation of the obtained measurements is performed by using the commercial
RayfleX software [92].
Electron microscopy 3.5.2
Scanning electron microscopy
Scanning electron microscopy (SEM) is used for the characterization of the surface
topography of the deposited thin films. In SEM, a focused beam of electrons with
an energy varying between 0.1 and 30 keV is scanned over the sample. Due to the
interaction of primary electrons with the atoms in the near surface region of the
sample, secondary electrons, bremsstrahlung and characteristic X-rays are
generated. Some of the primary electrons are rejected, as so-called backscattered
electrons. The most common use of SEM is for the detection of secondary electrons
that correspond, in their emission, with local height variations of the surface,
producing a type of surface topography images.
In the present work, a Carl Zeiss Ultra 55 field emission SEM is used for
obtaining cross-section and top-view images of the films. The instrument is
equipped with a GEMINI electron column as well as with two available secondary
electron detectors, an Everhart-Thornley detector and an in-lens detector. The SEM
is operated with an electron acceleration voltage of 15 kV and a working distance
between sample surface and electron column end of approximately 5 mm.
Energy-dispersive X-ray spectroscopy
Energy-dispersive X-ray spectroscopy (EDX) was used to determine the chemical
composition of the deposited films. The working principle of EDX is based on an
incident electron beam that knocks out an inner shell electron of an atom and,
hence, creates an electron hole where the electron was. Subsequently, the hole is
filled by an outer shell electron. The difference in energy between the inner and
outer shell produces a characteristic X-ray quantum. The emitted X-ray can be
detected by an energy dispersive spectrometer. Since each element has a unique
electron configuration of the atom, the energies are characteristic for the elements
contained in the sample and, therefore, they can be identified. The atomic
concentration can be calculated by using the signal intensities.
3.5 Characterization methods
35
In the present work, EDX (Quantax 400-EDX, Bruker Nano GmbH, attached to
the SEM system) is used for analyzing the chemical composition of the films. An
electron beam energy of 25 keV is used. In order to increase the detected intensity
from the films, the samples are tilted to an angle of 45° and a large aperture with
60 μm diameter is used for acquisition. The acquired data are analyzed in the
Bruker ESPRIT software.
Transmission electron microscopy
Transmission electron microscopy (TEM) is used to determine the morphology,
microstructure and crystal structure of the deposited films. In TEM, an electron
beam with a high acceleration voltage of a few 100 keV is transmitted through a 20
to 200 nm thin specimen. The interaction between the incident electrons and the
atoms within the specimen results in the generation of differing signals. The basic
TEM operation modes are imaging and diffraction, which can be chosen by
adjusting the apertures and electron lenses. In imaging mode, a bright field image
can be formed using the detected direct electrons, whilst a dark field image is
formed using the electrons scattered by a particular set of crystal planes. The images
can provide information about the film morphology, appearance of grains, grain
boundaries and the defect structure of the measured films. The resolution of the
images can be improved (down to the sub-Ångstrom scale) by using spherical
aberration corrector (Cs corrector) TEM, which allows the direct characterization of
the atomic structure. The unique properties of such a TEM facility can be combined
with the capability of scanning the electron beam over the specimen, resulting in the
scanning transmission electron microscopy (STEM) technique. In diffraction mode,
the obtained electron diffraction patterns result from coherent elastic scattering and
provide information about periodicity in the materials. For instance, an amorphous
film shows broad rings in the diffraction pattern that correspond to the inter-atomic
spacing, whereas crystalline materials show sharp diffraction spots that correspond
to Bragg diffraction at the periodic lattice. Using an electron beam with a small
diameter (~5 nm) makes it possible to take diffraction patterns from precisely
selectable locations of the specimen and, therefore, facilitates determining the local
structure on the nano scale, the so-called nano beam diffraction (NBD) technique.
Chapter 3 Experimental conditions and methods
36
In addition, STEM can also be combined with EDX, which makes a direct
correlation of imaged region and chemical analysis of a specimen feasible.
In the present work, a FEI Titan3 G
2 60-300 analytical TEM equipped with a Cs
probe corrector and annular dark-field (ADF) detectors is used for local structure
and elemental analysis of the GeTe and GST thin films. The cross-sectional TEM
specimens of the films are prepared by using focused, high-energy Ga and
subsequent low-energy Ar ion beam milling. Typically, a small specimen thickness
of approximately 25 nm is used for STEM studies. More details about the specimen
preparation using a focused ion beam (FIB) can be found in Ref. [93]. An ultra-thin
coating of electrically-conducting Pt is used to protect the surface of a FIB lamella
during the preparation. The TEM was operated at 300 kV accelerating voltage. A
probe-forming aperture of 25 mrad is used in the STEM studies. Middle-angle ADF
(MAADF) images are taken, using annular ranges of 40-200 mrad of the HAADF
detector, whereas low-angle ADF (LAADF) images are recorded using annular
ranges of 9.3-53.6 mrad of the ADF detector. Diffraction patterns of the specimens
are obtained in the NBD mode. The analysis of local chemical compositions is
identified by a Super-X EDX system.
Atomic force microscopy 3.5.3
Atomic Force Microscopy (AFM) is used to obtain the topography of the GeTe and
GST films at the nanoscale, as well as their local phase transformation behavior.
The basic principle of AFM refers to the interactions between the probe tip and
sample surface on the atomic level. Using a very simplified picture of the tip-
surface interaction, it can be described by a superposition of the attractive van-der-
Waals forces and the Pauli repulsion, the underlying potential is the Lennard-Jones
potential [94]
𝑈(𝑑) = 4ε [(𝜎
𝑑)
12
− (𝜎
𝑑)
6
] , (3.3)
where σ and ε are constants. This potential U is schematically depicted as a function
of the tip-sample distance in Figure 3.9a. AFM measurement modes can be
distinguished as a function of the sign of resulting force (repulsive or attractive):
Contact and Non-contact Mode. In this thesis, AFM measurements in contact as
well as in intermittent contact mode (the so-called tapping mode) are used. The
3.5 Characterization methods
37
contact mode is utilized in the conductive AFM setup, resulting in the opportunity
to register topographic and electric current images of the films simultaneously (for
details see Chapter 3.4). The tapping mode is used to gain information about the
microscopic topography of the film. In contact mode, the tip is dragged across the
surface of the sample, whereas, in tapping mode, a cantilever performs a certain
driven resonance oscillation with constant amplitude near the surface of a sample. A
schematic of an AFM setup is given in Figure 3.9b. A laser beam is reflected from
the back of the cantilever onto a position-sensitive photodetector. In general, the
position of the reflected laser beam on the detector shifts when the tip, which is
attached to the cantilever, is scanned over a sample surface with non-homogenous
topography. These shifts, caused by cantilever deflections, are recorded with help of
the photodetector and yield information about the movements of the cantilever.
These are fed into a feedback controller, which controls the cantilever’s up or down
motion, in order to compensate for the cantilever deflection (constant force mode).
A surface topography image is assembled, based on this signal. For the tapping
mode, the tip-sample interaction forces provoke changes of the cantilever
oscillation amplitude at a certain fixed resonance frequency. By maintaining the
preset constant amplitude of the oscillation during sample scanning, a surface
topography image is obtained. In this study, a cantilever with a spring constant
between 0.5 and 4.4 N/m and a resonance frequency between 315 and 366 kHz is
used for all tapping mode measurements.
Figure 3.9: (a) Tip-sample interaction and basic operation modes of an atomic force microscope, (b)
Schematic setup of an AFM.
AFM stage
Photodetector
Cantilever
(a)
Attractive
Po
ten
tia
l (V
)
Tip-sample distance (d)
Contact Mode
Non-contact Mode
Repulsive
(b)
Tip
Chapter 3 Experimental conditions and methods
38
Spectroscopy and calorimetry methods 3.5.4
Time-of-flight secondary ion mass spectrometry
Time-of-flight secondary ion mass spectrometry (TOF-SIMS) is used for the
analysis of the in-depth distribution of the chemical elements in the deposited thin
films. In TOF-SIMS, a focused beam of primary ions with low beam current
(commonly used energy: 10-30 keV; typical ions used in the source: Ga+, Au
+, Bi
+)
generates collision cascades near the sample surface. Due to ion sputtering,
energetic atoms and clusters are emitted from the surface. Most of them are neutral;
only the charged particles are detectable in SIMS. These emitted secondary ions are
analyzed with a time-of-flight mass spectrometer and provide qualitative
information about the presence of chemical elements. In order to obtain a depth
profile, a second ion beam source is used and operated at low ion energy (≤ 2 keV)
and a higher ion beam current, to remove surface atoms by sputtering. Typically
used ions are O2+, Cs
+, Ar
+ or Arn
+. Both ion beams are switched on alternately
during profiling. Thus, TOF-SIMS is able to obtain the in-depth composition of the
film surface as a function of the sputter time. In this study, short Ga ion pulses are
employed with an energy of 15 keV and an average beam current of about 1.5 pA
for scans in regions of 50 × 50 μm². The sputtering species are Cs ions with an
energy of 2 keV and the current range of nA is actually used. The polarity of the
system was set to detect negatively charged ions.
UV-Vis spectrophotometry
UV-Vis spectrophotometry is utilized to measure the optical reflectivity of the
films. The principle of the measurement is based on the well-known Beer-Lambert
law and on the measured intensities of the light reflected from a sample and a
standard reference material (100% absorption), respectively. In this work, the films
are measured with a Varian Cary 5000 UV-Vis spectrophotometer equipped with an
integrating sphere. The typically measured wavelength range is from 400 to 700 nm
and recorded at a step width of 1 nm.
Differential scanning calorimetry
Differential scanning calorimetry (DSC) is used for the thermal analysis of the
phase transformation occurring in the deposited films. In DSC measurements, both
sample and reference are always at the same temperature, with constant thermal
3.5 Characterization methods
39
heating. When the heat flow changes, due to a phase transformation occurring in the
sample, the amount of heat gain or loss can be observed with differential scanning
calorimeters. Thus, DSC allows a characterization of the crystallization temperature
of the films as well as the crystallization kinetics. In the present work, a Perkin
Elmer DSC 8500 setup is employed. For the preparation of the samples, roughly
800 nm thick GeTe films are directly deposited on an Al cup as a measure sample,
while an empty Al cup is used as a reference. A temperature range between room
temperature (RT) and 300°C and at various heating rates of 5 to 100 K/min with a
temperature accuracy of 0.2°C is used for the measurement.
41
Chapter 4
Results and discussion
Thermal treatment-induced crystallization of GeTe 4.1
films
Among the phase-change materials, the binary compound GeTe is characterized by
a significantly high crystallization velocity, data retention at comparably high
temperatures, stability, and an excellent contrast in terms of optical and electrical
properties between amorphous and crystalline states [28, 30, 95-98]. However,
regarding detailed knowledge on the thermally induced crystallization mechanism
of GeTe films grown by PLD, only few reports exist [9, 24]. To improve the
understanding of the phase-change mechanism in this specific material system, the
crystallization of PLD-deposited GeTe films is investigated in the present thesis that
classifies the thermal treatment as ex situ or in situ thermal treatments. The
investigations on ex situ thermal treatment will with structural transitions of GeTe
films and the effect of oxygen incorporation into GeTe on the phase transition. The
in situ thermal treatment investigation is separated into temperature-dependent
polycrystalline and epitaxial GeTe film growth, based on different types of
substrates.
Ex situ thermally induced crystallization of GeTe films 4.1.1
Nanosecond-PLD was used to deposit GeTe films on Si substrates (for experimental
details, see Chapter 3.1). The structural properties, the topography, the
crystallization kinetics, the optical and electrical properties of these GeTe films in
the as-deposited state as well as in the crystalline state induced by ex situ thermal
treatment are presented [24].
1.) Chemical composition
The GeTe films were deposited on Si substrates with a laser pulse number of 1200
at room temperature, resulting in a typical film thickness of 60 nm at a deposition
rate of ~30 nm/min. In Figure 4.1a, the chemical composition of the films, as
Chapter 4 Results and discussion
42
determined quantitatively by EDX, is shown. The characteristic x-ray lines from
Ge, Te and Si are annotated and the composition of the films is calculated to be
Ge51.6Te48.4. Considering the experimental error of the EDX technique (± 1.6 at.%),
the composition of the films is that of Ge50Te50. In addition, the atomic ratio of
Ge/Te remained nearly constant for an increase of the sample temperature up to
350°C, as shown in Figure 4.1b.
2.) Structure analysis
Figure 4.2a presents XRD spectra of an as-deposited GeTe film and of GeTe films
annealed for 20 min in high vacuum at various temperatures. The XRD spectra of
the as-deposited GeTe film and also of the films annealed at 200 and 220°C show a
very broad peak in the 2θ range between 25° and 30°, indicating an amorphous state
Figure 4.1: (a) EDX spectrum of an as-deposited GeTe film on a Si substrate. (b) Chemical
composition ratio (Ge/Te) of GeTe films, determined by EDX as a function of annealing
temperature.
0 200 250 300 350
0.4
0.8
1.2
1.6
2.0
Ato
mic
ra
tio
Temperature (°C)
Ge
Te
(b)
(a)
0 2 4 6 8 10 12
GeK
TeM Te
L
TeL
GeK
GeL
TeL
Inte
nsity (
arb
. unit)
Energy (keV)
SiK
4.1 Thermal treatment-induced crystallization of GeTe films
43
of the film. However, when the annealing temperature reaches 240°C, diffraction
peaks at angles of 26.1°, 29.9°, 42.4° and 43.3° emerge, which are correlated to the
lattice planes (021), (202), (024) and (220) of rhombohedral GeTe, respectively.
These results illustrate the onset of the structural transformation from the
amorphous to the crystalline state, which corresponds with a remarkable optical
reflectivity increase and resistance drop, as determined by temperature-dependent
reflectivity and resistance measurements (see sections 5 and 6 of this chapter).
After a slight increase of the annealing temperature to 280°C, the maximum
reflection intensity was found, as shown in Figure 4.2a, implying optimum
crystallization conditions. With a further increase of the annealing temperature up to
350°C, the reflection intensity gradually decreases. This intensity decrease can be
Figure 4.2: (a) XRD spectra of an as-deposited GeTe film and of GeTe films with a film thickness of
60 nm annealed at different temperatures. The respective annealing temperatures are denoted. The
Miller indices correspond to the rhombohedral crystal structure of GeTe with the space group R3m
(JCPDS no. 47-1079). (b) FWHM value and calculated grain size from the crystalline phase (202)
reflection as a function of annealing temperature. The dashed lines serve as optical guides for the
eye.
20 30 40 50
Inte
nsity (
arb
.un
its)
2 (°)
380°C
350°C
300°C
280°C
260°C
240°C
220°C
200°C
(021)
Si
(220)
(202)
as-dep.
(003)
(024)
(a)
(b)
240 260 280 300 320 340
0.28
0.30
0.32
0.34
0.36
0.38
0.40
Temperature (°C)
FW
HM
(°)
(202)
30
35
40
45
50
Gra
in s
ize
(n
m)
Chapter 4 Results and discussion
44
attributed to partial evaporation of the films in the vacuum oven environment.
Finally, for films annealed at 380°C, no diffraction peaks could be detected, due to
complete evaporation of the film. Moreover, no separation of Ge or Te crystalline
phases could be observed within the annealing temperature range of 240-380°C.
Because the full width at half maximum (FWHM) value of the diffraction peaks
correlates with the average grain size of the crystallized films, the average grain size
of the GeTe films, expressed as a function of the annealing temperature, was
estimated from the dominant (202) reflection by using the Scherrer formula [89]. As
shown in Figure 4.2b, the average grain size is inversely proportional to the FWHM
value, and it increases from about 30 to 50 nm as the annealing temperature exceeds
the critical temperature of crystallization at 240 to 350°C. Based on the XRD results
from films annealed at different temperatures, it becomes clear that the phase
transformation from the amorphous to the rhombohedral crystalline state starts
between 220 and 240°C. This temperature for GeTe films is significantly higher
than for GST (Ge2Sb2Te5) films, indicating high data retention at higher
temperatures [99]. Furthermore, it is also higher than typical crystallization
temperatures for sputter-deposited GeTe films [95, 99, 100].
3.) Topography
Figure 4.3 shows the topography and roughness evolution of the annealed GeTe
films at various annealing temperatures under high vacuum conditions. Between RT
and 220°C, large differences in topography and roughness of the GeTe films cannot
be observed (not shown here) and the films are characterized by a smooth,
featureless topography with low roughness. However, crystallites clearly emerge
when the annealing temperature is increased to 240°C. A further increase of the
annealing temperature leads to a uniformly enhanced coarsening of crystallites, as
shown in Figure 4.3. This agrees well with the results of XRD measurements, which
only reveal diffraction peaks for films annealed at 240°C and above. The root-
mean-square roughness (rms) of the films is constant, between 1.7 to 1.9 nm in the
crystallization temperature interval between 240 and 280°C; it is related to the
homogeneous crystal growth process. A significantly higher rms value of 7.0 nm is
obtained for the film annealed at 350°C. This could be associated with increased
crystal coarsening as well as a more intense re-evaporation of the film constituents.
4.1 Thermal treatment-induced crystallization of GeTe films
45
4.) Crystallization kinetics
Crystallization kinetics of the as-deposited GeTe films was investigated using DSC
at different heating rates. The results of the DSC measurements are summarized in
Figure 4.4a. With increasing heating rate, the exothermic peak corresponding to the
crystallization temperature Tc shifts gradually to higher values, i.e., from 181.4°C at
5 K/min to 198.4°C at 100 K/min. According to the homogenous nucleation theory
[101], the crystallization process of GeTe crystallites into the amorphous GeTe
matrix is diffusion-controlled. Crystalline seeds are formed as a result of this
diffusion process. For low heating rates, the crystallite dimensions are large and the
number of crystallites per unit volume small. These large crystallites are stable (the
disaggregation of crystallites by out-diffusion can be neglected). With increasing
heating rate, the probability of the formation of a larger number of smaller
crystallites per unit volume is increased. But the process of disaggregation also
becomes increasingly obvious and is contrary to the diffusion process that forms the
crystallites. This means that the crystallization is delayed, which leads to higher
measured crystallization temperatures.
Furthermore, such a time-dependent crystallization process [102] can be
expected, if the crystallization temperature Tc is influenced by the heating rate.
Figure 4.3: AFM images (500×500 nm2) of GeTe films with a thickness of 60 nm annealed at
different temperatures: (a) 240°C, (b) 260°C, (c) 280°C, (d) 350°C.
rms = 1.82 nm
Z = 18 nm
0 nm
a) rms = 1.90 nm
Z = 17 nm
b)
rms = 1.70 nmc) rms = 6.98 nmd)
Z = 16 nm Z = 48 nm
Chapter 4 Results and discussion
46
According to Kissinger’s formula [103], the activation energy Ea of crystallization
can be determined by
𝑙𝑛 (𝑎
𝑇𝑐2) = −
𝐸𝑎
𝑘𝐵𝑇𝑐+ 𝐴 , (4.1)
where a is the heating rate, Tc is the crystallization temperature, kB is the Boltzmann
constant, and A is a constant. The activation energy is calculated to be 3.14 eV, as
obtained from fitting the slope in the Kissinger plot, and shown in Figure 4.4b. In
the literature, measured activation energies between 1.7 and 3.9 eV for the
transition from the amorphous to the rhombohedral phase of GeTe have been
reported [104-108]. The measured activation energy obtained in this study fits
within this range. A comprehensive explanation for this relatively high activation
energy cannot be given yet. A high activation energy of crystallization could be
related to a long incubation time for crystallization.
Figure 4.4: (a) DSC diagrams of GeTe films recorded with different heating rates. (b) Kissinger plot
for determination of the activation energy Ea of crystallization.
140 160 180 200 220 240
He
at
Flo
w (
arb
.un
its)
Temperature (°C)
Heating rate
Tc=198.4°C
Tc=196.3°C
Tc=194.1°C
Tc=188.3°C
Tc=185.2°C
70 K/min
50 K/min
20 K/min
10 K/min
5 K/min
100 K/min
Tc=181.4°C
2.12 2.14 2.16 2.18 2.20 2.22-11.0
-10.5
-10.0
-9.5
-9.0
-8.5
-8.0
-7.5
ln(
2 c)
1000/Tc(K)
Ea = 3.14 eV
(a)
(b)
4.1 Thermal treatment-induced crystallization of GeTe films
47
5.) Optical reflectivity
The optical reflectivity contrast between the crystalline and amorphous state is one
of the crucial optical parameters considered in phase-change optical storage. The
reflectivity measurements were performed on both as-deposited and annealed films.
Based on previous studies [10], an optimal annealing time of 20 min was applied.
For all these measurements, a GeTe film thickness of 60 nm was chosen. The
temperature-dependent reflectivity measurements are presented in Figure 4.5a. The
reflectivity of an as-deposited film varies from 72 to 66 % in the wavelength range
between the near ultraviolet and the near infrared (360-800 nm) and it increases
slightly for films annealed at 220°C. In contrast, the reflectivity of the film after
annealing at 240°C is between 83 and 88% and much higher than the reflectivity of
Figure 4.5: (a) Optical reflectivity spectra of an as-deposited GeTe film and films annealed at
different temperatures, (b) reflectivity contrast between annealed and as-deposited films at a
wavelength of 650 nm for different annealing temperatures. The dashed line in (b) serves only as a
guide for the eye.
(a)
(b)
400 500 600 700 800
60
65
70
75
80
85
90
Re
fle
ctivity (
%)
Wavelength (nm)
Si substrate
300°C240°C
350°C220°C
380°C
as-dep.
0 80 160 240 320
0
5
10
15
20
Reflectivity c
ontr
ast (%
)
Temperature (°C)
Chapter 4 Results and discussion
48
the as-deposited film. An additional increase of the annealing temperature from 240
to 300°C does not significantly affect the reflectivity of the GeTe films, and the
values still remain between 83 and 88%. After annealing a film at 350°C, the
reflectivity starts to decrease. The probable reason for this decrease of the optical
reflectivity of films annealed at temperatures higher than 300°C could be the much
higher surface roughness caused by crystallite coarsening and evaporation (see
Figure 4.3d). A higher surface roughness scatters light more efficiently in all
directions and the chances of absorption on nearby hillocks are therefore increased.
With a further increase of the annealing temperature up to 380°C, the reflectivity of
the film shows a dramatic decrease that results from the evaporation of film
material. In fact, almost all material has evaporated, because the reflectivity spectra
almost overlap with that of a fresh, non-coated Si substrate.
In order to obtain a high signal-to-noise ratio in an optical application, a high
reflectivity contrast is required. To evaluate the reflectivity changes in more detail,
the reflectivity contrast between the as-deposited and annealed films at a
wavelength of 650 nm were calculated and it amounted to approx. 21%, as shown in
Figure 4.5b. In this context, it should be taken into account that a phase-change film
with an optical contrast above 15% is a promising candidate for commercial
applications [109].
6.) Electrical resistivity
The resistivity measurements were carried out on 60 nm thick as-deposited GeTe
films on SiO2/Si substrates at temperatures between 0 and 300°C. In Figure 4.6, the
resistivity of the GST thin films as function of the heating rate between 5-20 K/min
is shown. Independent of the heating rate, the resistance of ~106-10
7 Ω of the as-
deposited films decreases gradually as the annealing temperature increases up to
200-220°C, due to a thermally activated conduction mechanism. The coarsening of
grains induced by the annealing process for temperatures higher than 220°C (see
Figure 4.2b) leads to a reduced grain boundary density. Based on the assumption
that the grain boundaries are the dominant scattering center for the electrical current
transport, a decrease of the grain boundary density should be connected with a
reduction of the resistivity.
4.1 Thermal treatment-induced crystallization of GeTe films
49
Upon further temperature increase, a significant resistance drop, encompassing
3-4 orders of magnitude, can be observed at a transition temperature ranging from
220 to 255°C, where the decrease of the resistance is weakly dependent on the
heating rate (Figure 4.6). Here, the resistivity drop is strongly connected with the
amorphous to crystalline (rhombohedral) phase transition, which is mainly due to
the increase of mobility, rather than of carrier concentration [110, 111]. It is also
shown in Figure 4.6 that, with an increase of the heating rate, the drop of resistivity
and, therefore, the phase transition itself is shifted to slightly higher temperatures,
as follows from the discussion of the crystallization kinetics (see section 4 of this
chapter). Interestingly, the resistances in the crystalline state do not show any
dependence on the annealing temperature in the investigated temperature range.
The results are reasonably consistent with the optical reflectivity results of the
films (see Figure 4.5). This is because crystalline GeTe is a p-type semiconductor
and the Fermi level lies inside the valence band, resulting in metal-like behavior
[112, 113]. Furthermore, the transition temperature around 240°C for the film
heated at a heating rate of 10 K/min is in good agreement with the results of XRD
measurements, as discussed above, although a SiO2 buffer layer had to be added for
electrical insulation (see inset of Figure 4.6).
Figure 4.6: Resistance of 60 nm thick as-deposited GeTe films measured with heating rates of
5 K/min, 10 K/min and 20 K/min. The solid lines are only intended as guides for the eye. The inset
shows a schematic view of the samples used for resistance measurements.
0 50 100 150 200 250 30010
2
103
104
105
106
107
Re
sis
tan
ce
(
)
Temperature (°C)
5 K/min
10 K/min
20 K/min
Si (100)
SiO2 500 nm
GeTe 60 nmAu 20 nm
Chapter 4 Results and discussion
50
Effect of O incorporation on structural transitions of GeTe films 4.1.2
The phase-transformation of PLD-deposited GeTe films by an ex situ, thermally-
induced crystallization mechanism was described in detail in the previous section.
Based on experiences with GST alloys, O-doping may contribute to improve the
overwrite cyclability, stability and crystallization rate [114-117]. Strategies for GeTe
films doping by N, C, Cu, O were also explored [21-23, 118]. However, there are
limited reports about environmental oxidation of GeTe films [23, 119, 120]. Thus, it
is relevant to understand the influence of O incorporation on the structural
transitions of binary alloy GeTe films upon ex situ thermal treatment. This influence
is investigated in the following taking into account the properties of the pure GeTe
films described in the previous Chapter 4.1.1.
1.) Chemical composition
The Ge-Te-O (GeTeO) films were deposited by nanosecond-PLD on Si substrates
at room temperature. The number of laser pulses used was 1200, leading to a
deposition rate of 0.5 nm/s and a film thickness of 60 nm. The chemical
composition of the films was determined quantitatively by EDX. Exemplarily, an
EDX spectrum of a GeTe film with O incorporation is shown in Figure 4.7. The
characteristic X-ray lines from Ge, Te and O were detected, and the composition of
the films was calculated to be Ge34Te50O16. Considering the experimental error of
the EDX technique, the composition of the films is therefore 16 at.% O
Figure 4.7: EDX spectrum of an as-deposited 60 nm thick GeTeO film on a Si substrate. Inset: SIMS
depth profile of the film (negative secondary ions).
0 50 100 150 200
102
103
104
105
Inte
nsity (
counts
)
Sputtering time (s)
O
Si
Ge
Te
0 2 4 6 8 10 12
GeL
SiK
GeK
GeK
OK
Inte
nsity [arb
. unit]
Energy [keV]
TeL
TeL
TeL
4.1 Thermal treatment-induced crystallization of GeTe films
51
incorporation in GeTe, as compared to the nominal composition of the used target
(GeTe). The O content can be partly explained by surface oxidation, but a
substantial amount is transferred from the target (as the composition of the target is
Ge46.32Te43.72O9.96). SIMS measurements, as shown in the inset of Figure 4.7, were
carried out to confirm that an O incorporation in GeTe exists throughout the
complete film. Apart from the oxidation of the film surface and the oxide at the
interface between film and substrate, the O content is almost constant over the
entire film thickness. Because the atomic radius of O (r = 66 pm) is relatively small,
compared to those of Ge (r = 122 pm) and Te (r = 137 pm) [121], O atoms in the
GeTeO films are expected to be located only at the tetrahedral interstitial sites rather
than as substitutes for Ge or Te atoms.
The series of SEM images in Figure 4.8 shows the surfaces of an as-deposited
GeTeO film and of GeTeO films after thermal treatment at different temperatures.
Starting with the as-deposited film and then annealing at up to 280°C the surfaces
of films appear to be comparably smooth. However, after annealing at 350°C, gap
formation set in and the gaps significantly enlarge with further increases in
temperature up to 380°C. Following annealing at 450°C, the films obviously
disintegrated completely by evaporation, except for a small number of pre-existing
particulates (originating from the laser ablation process itself), with a diameter
between 0.3 and 10 µm; these can still be discerned on the surface of the sample.
Figure 4.8: SEM images of GeTeO films as a function of annealing temperature.
260oCAs-dep.
350oC 380oC 450oC
280oC
Chapter 4 Results and discussion
52
2.) Structure analysis
XRD measurements were performed to obtain structural information about the
films. Figure 4.9a presents XRD spectra of an as-deposited GeTeO film and of
GeTeO films annealed at different temperatures. In each case, the films were
annealed for 20 min in high vacuum. The corresponding annealing temperatures are
shown in Figure 4.9a. The XRD spectrum of the as-deposited GeTeO film shows a
broad peak in the 2θ range between 25 to 30°, suggesting an amorphous state of the
film. XRD spectra of the films annealed at 200 and 260°C resemble that of the as-
deposited film. However, when the annealing temperature reaches 280°C,
reflections at angles of 26.1°, 29.9° and 43.3° emerge that can be attributed to the
lattice planes (021), (202) and (220) of the rhombohedral GeTe phase (JCPDS no.
47-1079), respectively. The appearance of these reflections marks the structural
transformation from the amorphous to the crystalline state. After a slight increase of
Figure 4.9: (a) XRD spectra of as-deposited GeTeO film and GeTeO films with a film thickness of
60 nm annealed at different temperatures. The corresponding annealing temperatures are denoted.
The Miller indices correspond to the rhombohedral crystal structure of GeTe with the space group
R3m (JCPDS no. 47-1079). (b) Corresponding (202) lattice plane spacing d.
280 300 320 340 360 38029.85
29.90
29.95
30.00
Temperature (°C)
2(
°)
(202)
0.2975
0.2980
0.2985
d
(nm
)
(a)
(b)
20 30 40 50
as-dep.
200°C
260°C
280°C
300°C
350°C
380°C
Inte
nsity (
arb
. u
nits)
2 (°)
450°C(22
0)
(20
2)
(02
1)
4.1 Thermal treatment-induced crystallization of GeTe films
53
the annealing temperature to 300°C, peak intensities reached maximum values. The
intensities gradually decreased with a further increase of annealing temperature to
380°C. This intensity decrease can be attributed to partial evaporation of the films
in the vacuum oven environment. Finally, for films annealed at 450°C, diffraction
peaks no longer occur due to complete evaporation of the films (demonstrated by
SEM measurements in the previous chapter). It should be noted that, despite the
significant O content of the films, no additional reflections related to crystalline
oxide phase formation appeared. Thus, as confirmed by the XRD measurements
presented in Figure 4.9a, phase segregation could be detected. However, for the
crystalline state of the films, it is interesting to note that all the peak positions
shifted to lower diffraction angles with increasing annealing temperature, as shown
in Figure 4.9b. Such peak position shifts were not observed with the GeTe films
described in Chapter 4.1.1 (Figure 4.2). Regarding, for instance, the (202)
reflection, its position in 2θ is 30.02° at an annealing temperature of 280°C and
decreases to 29.87° at an annealing temperature of 380°C. This shift of peak
position with increasing annealing temperature is caused by a change of the lattice
parameter of the thin film. Using Bragg’s law, the lattice spacing d of the crystal
plane (202) was calculated and plotted as a function of the annealing temperature in
Figure 4.9b. The d value increases from 0.2973 to 0.2988 nm as the annealing
temperature increases from 280 to 380°C. Regarding the crystallization of the
investigated GeTeO films, here, phase transformation from the amorphous to the
rhombohedral crystalline state only starts at 280°C. This is a much higher transition
temperature compared to those of films with slightly Te-rich stoichiometry prepared
by magnetron sputtering and based on different stoichiometry, such as GeTe and
GeTe4. It is also higher in comparison to the transition temperature of films
discussed in Chapter 4.1.1 (a comparison of data is given in Table 4.1 in section 4
of this Chapter 4.1.2).
3.) Topography
The topography and roughness evolution of the annealed GeTeO films at different
annealing temperatures (Figure 4.10) is provided by AFM measurements. With
increasing temperature, from RT to 260°C, no large differences in the topography
and roughness of the GeTeO films appear (not shown here). However, some grains
Chapter 4 Results and discussion
54
clearly emerge when the annealing temperature is increased to 280°C, as in Figure
4.10a. A further increase of the annealing temperature leads to enhanced grain
growth with small protrusions appearing at 350°C and 380°C, as shown in Figure
4.10. This agrees well with the result of XRD studies for annealing temperatures
below 280°C; the GeTeO films are in the amorphous state and the crystalline state
appears after annealing at a minimum temperature of 280°C. Correspondingly, the
rms surface roughness of the films shows a gradual increase from 1.24 to 4.18 nm
with the increase of the annealing temperature from 280 to 380°C, which is related
to the heterogeneous crystal growth process. In comparison with the pure GeTe
films (in Figure 4.3), the topographies differ, although the tendencies for crystalline
grain growth with increasing temperature are similar. Rationally, the influence of
incorporated O in a GeTe film may serve to suppress the crystal growth process,
because the incorporated oxygen atoms probably inhibit the nucleation of crystals
GeTe elements clusters and it continues growing. That can be due to a smaller grain
size (in Figure 4.10) as well as to a higher critical crystallization temperature of a
GeTeO film, in contrast to a pure GeTe film. Furthermore, the surface roughness
values of GeTeO films are higher than those of pure GeTe films. This assumes that
Figure 4.10: AFM images (500×500 nm2) of GeTeO films with a thickness of 60 nm annealed at
different temperatures: (a) 280°C, (b) 300°C, (c) 350°C and (d) 380°C.
rms = 2.08 nm
rms = 2.33 nm
Z = 13 nm
Z = 17 nm Z = 25 nm
rms = 4.18 nm
a) b)
c) d)
Z = 9 nm
0 nm
rms = 1.24 nm
4.1 Thermal treatment-induced crystallization of GeTe films
55
O can partially enter into the grain boundaries and exert bi-axial strain on the
crystal; this influences the growth direction of the materials, according to Ref. [23].
4.) Crystallization kinetics
In Figure 4.11a, the as-deposited GeTeO films were investigated using DSC at
various heating rates. With increasing heating rate, the crystallization temperature
Tc shifts gradually to higher values, i.e., from 258.8°C at 5°C/min to 281.1°C at
100°C/min. This indicates that the crystallization of GeTeO films proceeds in
diffusive nucleation and growth within the range of the heating rate used [23, 122].
The influence of the heating rate on the crystallization temperature Tc also shows
that the crystallization process is not independent of time [123]. According to
Kissinger’s formula (as described in Equation 4.1), an activation energy Ea of
crystallization of 3.18 eV was extracted by fitting the slope in the Kissinger plot in
Figure 4.11: (a) DSC diagrams of GeTeO films for different heating rates. (b) Kissinger plot of the
crystallization from amorphous to rhombohedral phase reveals the activation energy Ea for
crystallization of GeTeO films.
240 250 260 270 280 290 300
50 K/min
Tc= 272.5
oC
70 K/min
Tc= 275.5
oC
He
at
Flo
w (
arb
.un
its)
Temperature (oC)
5 K/min
10 K/min
20 K/min
30 K/min
Tc= 264
oC
Tc= 262.6
oC
Tc=258.8
oC
Tc= 266
oC
Tc= 281.1
oC
Heating rate
100 K/min
1.80 1.82 1.84 1.86 1.88
-11.0
-10.5
-10.0
-9.5
-9.0
-8.5
-8.0
ln(a
/T2 c)
1000/Tc(K)
Ea = 3.18 eV
(a)
(b)
Chapter 4 Results and discussion
56
Figure 4.11b. Comparison of the Ea and Tc values of GeTeO and GeTe films in this
study, as well as Ea and Tc values for GeTe and GST films reported in literature, are
summarized in Table 4.1. It has to be considered that a high Tc, together with a high
Ea, means high data retention. Based on the data in Table 4.1, GeTeO films exhibit
the highest Ea and Tc values, indicating that they possess the highest thermal
stability, i.e., higher data retention.
5). Optical reflectivity
Figure 4.12a shows the results of reflectivity measurements from the near
ultraviolet to the near infrared (wavelength range from 360 to 800 nm) of both as-
deposited and annealed GeTeO films. The reflectivity of as-deposited GeTeO films
varies from 55 to 58%, whereas the reflectivity of the films annealed at a
temperature of 280°C shows a significant increase and ranges from 68 to 78%.
However, a gradually decreasing reflectivity, ranging from 69 to 72%, appeared
when the annealing temperature increased from 280 to 380°C. One probable reason
for the decrease of the optical reflectivity of films annealed at increasing
temperature could be the greater surface roughness, according to AFM
measurements. On the other hand, the average film thickness slightly decreased
with increased annealing temperature due to evaporation. This may also contribute
to the observed decrease in reflectivity in Figure 4.12a. With a further increase of
the annealing temperature up to 450°C, the reflectivity of the films shows a
significant decrease, which is similar to the tendencies of a pure GeTe film at 380°C
Table 4.1: Comparison of Tc and Ea of GeTeO, GeTe and GST films.
Material Tc (°C) Ea (eV)
GeTeO 280 3.18
GeTe 240 3.14
GeTe4 a 232 2.7
GeTeb 184.8 -
GSTc 154.8 2.2
[a] H. Jiang et al., J. Appl. Phys. 109, 066104 (2011).
[b] N. Yamada et al., J. Appl. Phys. 69, 2849 (1991).
[c] L. Perniola et al., IEEE Electron Device Lett. 31, 488 (2010).
4.1 Thermal treatment-induced crystallization of GeTe films
57
or a bare Si substrate (presented in Figure 4.5). The result is explained by the
evaporation of a large film fraction, as demonstrated by SEM images (section 1 of
this Chapter 4.1.2). As shown in Figure 4.12b, the corresponding reflectivity
contrasts of the GeTeO films, determined at a wavelength of 650 nm, were 18.8%
above the critical crystallization temperature within the range between 280°C and
350°C.
In situ thermally induced crystallization of GeTe films 4.1.3
The quality of phase-change films potentially benefits from the reduction of energy
consumption when switching between amorphous and crystalline states with respect
to RESET and SET operations, and from increasing the optical and electrical
contrast in terms of the high signal-to-noise ratio, leading to higher distinction
Figure 4.12: (a) Optical reflectivity spectra of as-deposited GeTeO film and of films annealed at
different temperatures, (b) reflectivity contrast at a wavelength of 650 nm of GeTeO films annealed
at different temperatures, compared to that of GeTe films. The dashed line is only intended as a
guide for the eye.
0 100 200 300 400
0
4
8
12
16
20
24
GeTe
Reflectivity c
ontr
ast (%
)
Temperature (°C)
GeTeO
400 500 600 700 800
56
60
64
68
72
76
80
84
As deposited
450°C260°C
380°C
350°C
Reflectivity (
%)
Wavelength (nm)
280°C
(a)
(b)
Chapter 4 Results and discussion
58
levels capability in multilevel memory applications [120]. Unlike the previously
described ex situ thermally induced crystallization, in situ thermally induced
crystallization of GeTe films was also explored in order to improve the film quality.
During PLD film growth, the growth temperature and the choice of the substrate
have a crucial influence on the growth mechanism [26].
In this subchapter, GeTe thin films PLD-deposited on Si(100) substrates (with a
very thin native surface oxide layer) at different substrate temperatures are
investigated. The effect of growth temperature on GeTe thin films’ structure,
morphology and reflectivity is characterized by using XRD, AFM and UV-Vis
spectrophotometry.
1.) Structure analysis
Figure 4.13 shows the XRD spectra of GeTe films deposited on Si(100) at different
substrate temperatures that ranged between 100 and 290°C. The diffraction
spectrum of the GeTe film deposited at 100°C is typical for the amorphous state.
However, the onset of polycrystalline phase formation appears at the growth
temperature of 130°C, as can be seen from the significant presence of (021) and
(220) reflections of GeTe in Figure 4.13. With a further increase of the substrate
temperature to 160°C, the dominant XRD reflections depict a transition to
preferably (003) oriented films. A continuous increase of the intensity of the (003)
reflection is clearly visible at a substrate temperature of 180°C. GeTe film growth at
substrate temperatures in the range from 200 to 240°C exhibits a significant
Figure 4.13: XRD spectra of GeTe films deposited on Si(100) substrates at different temperatures.
20 30 40 50 60
Si
(006)
200°C
Inte
nsity (
cps)
2(°)
130°C
150°C
160°C
180°C
290°C
220°C
240°C
100°C
(003)
(220)
(202)
(021)
(024)
(04
2)
4.1 Thermal treatment-induced crystallization of GeTe films
59
increase of the preferred (003) reflection intensity, which is so intense that also the
corresponding second order reflection is present and of similarly high intensity.
Thus, a pronounced (003) preferred orientation has formed. For films deposited
above this temperature range (e.g., at 290°C) reflections are no longer detectable,
due to the small amount of remaining material, a consequence of the enhanced
desorption of atoms at such high substrate temperatures.
A more detailed analysis of the films at growth temperatures in the range from
200 to 240°C is shown with a zoom of the (003) reflection in Figure 4.14. It appears
that the reflection possesses an extremely high intensity of more than 104 cps,
indicating a higher crystallinity of these films. Based on the Scherrer formula, the
calculated average grain size of these highly oriented polycrystalline films is 4.7 nm
Figure 4.14: GeTe(003) reflection of the GeTe films on Si(100) deposited in the substrate
temperature range from 200 to 240°C.
Figure 4.15: GeTe(003) rocking curves of films deposited at different substrate temperatures.
23 24 25 26 27 28
0
1x104
2x104
3x104
4x104
(111
)
Inte
nsity (
cp
s)
2 (°)
200°C
220°C
240°C
(003
)
-2 -1 0 1 2
0
1x103
2x103
3x103
Inte
nsity (
cp
s)
(°)
200°C, FWHM: 1.9°
220°C, FWHM: 1.3°
240°C, FWHM: 1.6°
Chapter 4 Results and discussion
60
at 200oC, 7.1 nm at 220°C and 5.9 nm at 240
oC. From the corresponding rocking
curves of the GeTe (003) reflection, as shown in Figure 4.15, the FWHM was
extracted as 1.9° at 200oC, 1.3° at 220
oC and 1.6° at 240
oC. For polycrystalline
films exhibiting a preferred orientation, these values are relatively low. In order to
clarify the present crystalline texture of such highly oriented films, an in-plane pole
figure measurement was performed of a representative film deposited at a growth
temperature of 220°C. The pole density distribution in this pole figure (Figure
4.16a) exhibits a concentric ring at an α angle of 76°. Thus, the film is characterized
by a pronounced fiber texture with the [003] direction being the fiber axis. The pole
figure also contains eight distinct pole density maxima at α = 46° as well as
additional four pole density maxima at α = 76°. These are identified by the modeled
pole figure in Figure 4.16b to originate from the Si substrate and not from the film.
In conclusion, GeTe film growth in the narrow temperature window of 200-240°C
leads to a pronounced fiber texture of the films. Furthermore, at a growth
temperature of 220°C, the films possessed the highest reflection intensity and the
smallest rocking curve FWHM.
2.) Topography
In order to increase the understanding of polycrystalline GeTe film growth by PLD
in terms of the topography, AFM measurements on a series of selected films were
carried out (summarized in Figure 4.17). It is clear that different topographies result
from different growth temperatures in the range between 130 and 240°C. At a
Figure 4.16: (a) In-plane pole figure of the GeTe film on Si substrate using the {422} reflection at
2θ= 71.27°. Note that the intensity (cps) is depicted on a logarithmic scale. (b) Modeled bare Si
substrate pole figure from the Si(331) reflection.
1.5E+02
5.1E+02
8.6E+02
1.2E+03
GeTe {422}
(a) (b)
β
α
α = 76°
α = 46°
GeTe film
Si sub. Si sub.
4.1 Thermal treatment-induced crystallization of GeTe films
61
growth temperature of 130°C, the film surface is relatively smooth in spite of the
several crack-like features evident in Figure 4.17a (rms roughness is 1.4 nm). With
a higher substrate temperature of 180°C, as shown exemplarily in Figure 4.17b, the
film surface is no longer smooth and it exhibits an increased rms roughness of 2.3
nm, similar to that of films deposited at temperatures of 220°C (rms = 2.6 nm) and
240°C (rms= 2.0 nm) in Figure 4.17c and Figure 4.17d, respectively. This trend of
the topography change with rising substrate temperature becomes clearer by
comparing spatial height profiles obtained from representative surface regions of
the films, as depicted in Figure 4.17e. The film surfaces obviously changed, due to
Figure 4.17: AFM images of the topography of GeTe films grown on Si(100) at different substrate
temperatures: (a) 130°C, (b) 180°C, (c) 220°C, (d) 240°C and (e) height profiles taken along the red
lines in the images of (a) to (d).
0 40 80
0
2
4
6
He
igh
t (n
m)
Length (nm)
130°C 180°C
220°C 240°C
a)
(e)
b)
c) d)rms =2.6 nm
Z=19Z=21
Z=18
Z=11
rms =2.0 nm
rms =2.3 nmrms =1.4 nm
Chapter 4 Results and discussion
62
the fiber-textured growth at deposition temperatures of 220°C and 240°C, in
contrast to the films obtained at 130°C. This illustrates that the higher the level of
the crystallite orientation in the films is, the higher the peak-to-valley values in the
height profiles and, consequently, the greater the surface roughness.
3.) Optical reflectivity
Figure 4.18a shows the optical reflectivity of GeTe films deposited on SiO2/Si(100)
substrates and grown at different temperatures. The reflectivity of as-deposited
films varies from 48 to 52% in the wavelength range from the near ultraviolet to the
near infrared (400-700 nm). In contrast, the reflectivity of films grown at a substrate
temperature of 130°C lies between 64 and 80%, and is thus distinctly higher than
the reflectivity of as-deposited films. A further increase of the growth temperature
Figure 4.18: (a) Optical reflectivity of GeTe films deposited on Si(100) substrates covered with
500 nm SiO2 at different growth temperatures. (b) Reflectivity contrast at a wavelength of 650 nm
between GeTe films annealed in situ during growth at different growth temperatures, compared to
amorphously deposited GeTe films post-growth annealed at different annealing temperatures. The
dashed lines are only intended as guides for the eye.
0 50 100 150 200 250 300 350 400
0
8
16
24
32
Reflectivity c
ontr
ast (%
)
Temperature (°C)
post-anneal GeTe
during growth GeTe
400 500 600 700
50
60
70
80
Optical re
flectivity (
%)
Wavelength (nm)
130°C 180°C
220°C 240°C
as-dep.
(a)
(b)
4.1 Thermal treatment-induced crystallization of GeTe films
63
to 180°C obviously does not affect the reflectivity of the GeTe films. Only in the
temperature range between 220 and 240°C does the film reflectivity exhibit a
further increase, with values between 66 and 84% within the measured wavelength
range. For even higher growth temperatures, e.g. 290°C, a dramatic decline of the
optical reflectivity of films appears, caused by decomposition and partial desorption
of film components. The results agree well with those obtained from the structure
analysis presented in section 1 of this chapter. Thus, the higher optical reflectivity
of films deposited at substrate temperatures in the range from 220 to 240°C is
caused by the higher crystalline quality of the GeTe films.
In order to obtain a high signal-to-noise ratio for optical application of GeTe
films, a high reflectivity contrast is required. The reflectivity contrast, determined at
a wavelength of 650 nm, for as-deposited amorphous films and films grown at
elevated substrate temperatures, compared to post-growth annealed GeTe films, is
presented in Figure 4.18b. An optical reflectivity contrast of 28% between
amorphous and polycrystalline GeTe films was obtained at lower growth
temperatures, and of approx. 31.5% between amorphous and fiber-textured GeTe
films at higher growth temperatures. These results demonstrate that the reflectivity
contrasts of the films obtained from fiber-textured growth (in situ thermal
treatment) are approx. 10% higher than for the polycrystalline films after post-
growth annealing (ex situ thermal treatment). Moreover, the critical crystallization
temperature of films with in situ thermal treatment lies at 130°C, which is close to
100°C lower than that of amorphous films crystallized by ex situ thermal treatment
at 220-240°C (see Chapter 4.1.1). Consequently, a considerably higher optical
reflectivity contrast (31.5%) of the fiber-textured GeTe films is achieved with
relatively lower crystallization temperature. Such a high optical reflectivity is not
only beneficial for the increase of the signal-to-noise ratio, but also promising for
multilevel PCM based data storage applications.
Epitaxial growth of GeTe films on BaF2(111) 4.1.4
A further improvement of the crystalline quality can be realized by PLD deposition
of epitaxial GeTe films. A small lattice mismatch between GeTe lattice and the
lattice of the substrate is the main condition. Consequently, freshly cleaved
BaF2(111) is used as a substrate, to provoke epitaxial GeTe film growth, because
Chapter 4 Results and discussion
64
the lattice mismatch is only about 3.5 %. Based on the results in the previous
chapter 4.1.3 regarding the structural and optical properties of GeTe films deposited
at different substrate temperatures, the substrate temperature was selected to be
205°C, constantly, for the epitaxial growth experiments. The deposition rate in this
case was 0.2 nm/s. The typical film thickness was again 60 nm.
1.) Composition and morphology
The chemical composition of a representative PLD-deposited GeTe film on a
BaF2(111) substrate was determined from EDX, as shown in Figure 4.19. The
characteristic x-ray peaks from the elements Ge, Te, Ba and F are present and the
resulting stoichiometry of the film is determined to be Ge50.1Te49.9. As the measured
composition is almost identical to the specified target composition, this indicates
that a stoichiometric transfer from the ablation target to the film was easily
achievable. The topography of GeTe films on BaF2(111) was measured by AFM at
a scanning range of 1×1 um², as presented in the inset image of Figure 4.19. The
film surface is relatively smooth, with a rms roughness of ~1.0 nm.
2.) Structure analysis
Figure 4.20 presents a representative XRD spectrum of a crystalline GeTe film
deposited on a BaF2(111) substrate at a substrate temperature of 205°C. The main
reflections of the GeTe film, well distinguishable from those of the BaF2(111)
substrate, are situated at 2θ angles of 25.1° and 51.4°, which are attributed to the
Figure 4.19: EDX spectrum of a GeTe film deposited on a BaF2(111) substrate. Inset: AFM
topography image.
0 2 4 6 8 10 12
BaL
BaL
BaL
FK
GeL
TeL
GeK
Inte
nsity (
arb
.unit)
Energy (keV)
4.1 Thermal treatment-induced crystallization of GeTe films
65
(111) and (222) lattice planes of rhombohedral β-GeTe, respectively. This result
suggests that the GeTe film grows in the β-phase. This is in agreement with
previous studies on GeTe films growing preferably in the cubic [111] direction
when providing (111) oriented, single crystalline substrates [124, 125].
A more detailed analysis is possible on the basis of XRD pole figure
measurements. The in-plane pole figure using the {422} reflection of a
representative GeTe film from the (111) orientation-dominated growth regime is
presented in Figure 4.21a. The intensity is plotted on a logarithmic scale. Six
intense {422} pole density maxima at a polar angle α of 76° are clearly visible, each
exhibiting a three-fold symmetry. Each couple of pole density maxima is
azimuthally separated by 40°, and the centers in between the couples are separated
by an azimuthal angle β = 120°. This is in good agreement with the modelled
GeTe{422} pole figure shown in Figure 4.21b. It is noted that the remaining, much
weaker intensity pole density maxima in the pole figure in Figure 4.21a are intensity
contributions originating from the BaF2 substrate. This is confirmed by the pole
figure of a bare BaF2(111) substrate (Figure 4.21c), measured at exactly the same 2θ
angle of 71.27° as the pole figure in Figure 4.21a. Unequivocally, the six high-
intensity pole density maxima present in Figure 4.21a are not observed in Figure
4.21c. This result is also clearly explained by selected φ scans of the GeTe film and
of the BaF2(111) substrate shown in Figure 4.21d, which were measured at the
identical 2θ angle of 71.27° (corresponding to the GeTe{422} reflection) and the
Figure 4.20: XRD spectrum of a GeTe thin film deposited on BaF2(111) at a substrate temperature of
205°C.
20 30 40 50 60
102
103
104
105
(222)
BaF2
GeTe (222)GeTe (111)
Inte
nsity (
cps)
2(°)
BaF2
(111)
Chapter 4 Results and discussion
66
identical polar angle α of 76°. The red-colored φ scan presents the previously
described 40° interval between couples of most intensive peaks with a three-fold in-
plane orientation symmetry of the GeTe film, as well as six much weaker peaks
consistent with the peaks from the BaF2(111) substrate appearing in the black-
colored φ scan. Therefore, a distinguishable epitaxial relationship between the α-
GeTe film and the BaF2(111) substrate is confirmed. This epitaxial relationship is
given by α-GeTe(111) || BaF2(111) and α-GeTe[1-10] || BaF2[1-10].
To estimate the average tilt distribution of crystallites in the films, rocking
curves of the β-GeTe(111) reflection were measured. One representative
measurement is shown in Figure 4.22. The very narrow peak in the middle of the
rocking curve measurement is a contribution from the β-GeTe(111) reflection. The
FWHM is 0.044°, indicating a rather small degree of mosaicity of the epitaxial
Figure 4.21: (a) In-plane pole figure of the GeTe film at the {422} reflection at 2θ = 71.27°. Intensity
(cps) is depicted on a logarithmic scale. (b) Modeled GeTe{422} pole figure provided by the CaRIne
Crystallography software. (c) Corresponding bare BaF2 substrate pole figure measured at 2θ =
71.27o. (d) φ scans of the GeTe film and of the BaF2 substrate, both measured at 2θ = 71.27° and α =
76°.
Inte
nsity (
cps)
(a) (b)
(c) (d)
BaF2 sub.
GeTe film
100
1.0E+03
1.1E+04
1.0E+05
GeTe {422}
3.5E+02
2.4E+03
1.6E+04
1.0E+05
GeTe {422}
BaF2 sub.
β
α
α = 76°
0 80 160 240 320
102
103
(°)
BaF2
102
103
104
105
80° 40°
GeTe
4.1 Thermal treatment-induced crystallization of GeTe films
67
GeTe film. Compared to the FWHM in the case of fiber-textured growth of GeTe
film on Si(100) (~1.3°) and in the case of an epitaxial GST film on a BaF2(111)
substrate (~0.2°) [12], a reduction was registered.
Summary of the results on thermal treatment-induced crystallization of 4.1.5
GeTe films
The process of PLD-deposited PCM films on Si substrates at room temperature has
been established. The ex situ thermally induced crystallization of GeTe and GeTeO
phase-change films was systematically investigated, including structure analysis,
topography, crystallization kinetics, as well as optical and electrical properties. The
obtained values for the crystallization temperature Tc, the activation energy for
crystallization Ea, the optical reflectivity contrast ORa-c and the electrical resistivity
contrast ERa-c are summarized in Table 4.2. The experimental results reveal that
PLD-deposited GeTeO films exhibit a higher crystallization temperature (280°C)
and a slightly higher activation energy for crystallization (3.18 eV), compared to
pure GeTe films. The achievable higher thermal stability makes them excellent
Figure 4.22: β-GeTe(111) rocking curve of an epitaxial GeTe film on BaF2(111).
Table 4.2: Comparison of crystallization temperature Tc, activation energy Ea, optical reflectivity
contrast ORa-c and electrical resistivity contrast ERa-c between PLD-deposited GeTe and GeTeO
films.
PLD-deposited
films
Tc
[°C]
Ea
[eV]
ORa-c
[%]
ERa-c
GeTe 240 3.14 21 103 – 10
4
GeTeO 280 3.18 18.8 _
12.4 12.6 12.8
5.0x103
1.0x104
1.5x104
2.0x104
2.5x104
Inte
nsity (
cp
s)
(°)
FWHM: 0.044°
β-GeTe (111)
Chapter 4 Results and discussion
68
candidates for phase change applications.
The quality of PLD-deposited GeTe films on Si(100) upon in situ thermal
treatment was investigated. A structure evolution from as-deposited amorphous over
polycrystalline (130°C) to fiber textured (220-240°C) in GeTe films during growth
at elevated temperatures was demonstrated. It should be noted that, following the
in situ thermal treatment, the GeTe films showed a critical temperature of
crystallization Tc that was approx. 100°C lower than in ex situ thermally treated
films. Moreover, the higher the achieved crystalline quality of the GeTe films was,
the higher the obtained optical reflectivity. The optical reflectivity contrast of
polycrystalline and fiber textured GeTe films, compared to amorphous films, was
approx. 28% and 31.5%, respectively. The latter is 10% higher than that of ex situ
thermally induced crystalline films (21%). Finally, an improvement of the film
quality, in order to influence the crystallization behavior, was achieved by the
epitaxial growth of GeTe films on BaF2(111) at a substrate temperature of as low as
205°C. The epitaxial growth of GeTe films was confirmed by X-ray texture analysis
measuring GeTe{422} pole figures. These epitaxial films exhibited a relatively
small polar mosaicity component of as low as 0.044o, obtained from the FWHM of
rocking curve measurements.
4.2 Laser-induced phase transitions of GeTe and GST films
69
Laser-induced phase transitions of GeTe and GST 4.2
films
The application of phase-change materials for data storage relies on the fast
reversible transformation between the disordered amorphous and ordered crystalline
states, which can be achieved by local heating/cooling at high rates, either with a
laser beam or electrical pulses. As described in Chapter 2.2, distinct chalcogenides
such as GeTe and GST, which are highly attractive for such applications, exhibit
markedly different crystallization behaviors. Compared to the thermally annealing
treatment-induced crystallization discussed in Chapter 4.1, laser pulse induced
crystallization is characterized by much higher heating/cooling rates (> 109 K/s [66,
126, 127]) and resembles more closely practical crystallization kinetics in PCMs.
Furthermore, regarding the physical process of laser-induced phase transition, ns
laser pulse irradiation can be considered as a thermal process, whilst upon
excitation with ultrashort pulses, i.e., fs laser pulses, non-thermal effects are
probably involved in the phase transition process [128-130]. Therefore, it is of
extreme relevance to investigate in detail the phase transformation of phase-change
films for the particular case of laser-induced crystallization, as well as to compare
the crystallization process in the time scale frame of ns and fs single laser pulses by
variation of the laser fluence at the same wavelength.
In this chapter, results and conclusions about the laser-induced phase transition
of phase-change films are presented [127, 131]. The chapter is divided into three
sections: ns laser-induced phase transition in GeTe films (Chapter 4.2.1),
crystallization of GST films by ns and fs single laser pulse irradiation (Chapter
4.2.2), and simulation of temperature distribution after pulsed laser irradiation
(Chapter 4.2.3).
Nanosecond laser-induced phase transition in GeTe films 4.2.1
1.) Crystallization process
Crystallization as a function of pulse number
Figure 4.23a shows the optical reflectivity of an as-deposited amorphous GeTe
film and of films after 20 ns laser pulse irradiation with pulse numbers between 0
Chapter 4 Results and discussion
70
and 30 at a constant fluence of 130 mJ/cm2 and a repetition rate of 10 Hz. The
optical reflectivity of the as-deposited film lies in the range between 72 and 74% for
wavelengths between 400 and 700 nm. The reflectivity of the film is slightly
increased, by 3%, when the number of applied laser pulses increases to 3. A more
pronounced effect is discernable after irradiation with 5 pulses. For this film, the
reflectivity is in the range between 85 and 88%, which is significantly higher than
that of the as-deposited film. Upon further increase of the number of applied laser
pulses the reflectivity increases further, but the effect is no longer as obvious. The
corresponding results from XRD analysis are depicted in Figure 4.23b. The XRD
patterns of an as-deposited film and of a film irradiated with 3 laser pulses only
show a several degrees broad and weak peak centered around 28.3°, typical for the
Figure 4.23: (a) Optical reflectivity of as-deposited GeTe film and films after 20 ns laser pulse
irradiation with a fluence of 130 mJ/cm2 as a function of the number of pulses. The inset shows the
corresponding optical microscope image of the irradiated area after 5 pulses. (b) Corresponding
XRD spectra of GeTe films as a function of the pulse number.
20 30 40 50
20 pulses
10 pulses
5 pulses
3 pulses
Inte
nsity (
arb
. u
nits)
2(°)
(220
)
(200
)
as-dep.
(111
)
Si
400 500 600 70070
75
80
85
90 30 pulses 20 pulses
10 pulses
5 pulses
as-depositedRe
fle
ctivity (
%)
Wavelength (nm)
3 pulses
(a)
(b)
0 N
5 N
4.2 Laser-induced phase transitions of GeTe and GST films
71
amorphous state of GeTe films. In contrast, when the film is irradiated by more than
5 pulses, clearly visible Bragg peaks located at 26.4°, 30.2° and 43.7° appear, which
correspond to the (111), (200) and (220) reflections of cubic phase GeTe,
respectively. Therefore, at least 5 laser pulses (not excluding the case of 4 pulses)
are required to induce the crystallization process. The XRD results are in good
agreement with the optical reflectivity measurements. Consequently, it can be
concluded that an increase in the number of laser pulses continuously increases the
crystalline fraction of the films. It can be assumed that the crystallization process is
incomplete after 20 ns single pulse laser irradiation [31], because the diffusional
processes involved in crystallization require sufficient time above the threshold
temperature typical of the GeTe material. According to this, the superposition of
laser pulses leads to an increasing crystallized volume and, consequently, to the
increase of the optical contrast.
Crystallization as a function of laser fluence
In order to examine in detail the influence of the laser fluence on the phase
transition, a series of fluence-dependent optical reflectivity and XRD measurements
of GeTe films was performed. A laser pulse number of 20 at a pulse duration of
20 ns was chosen for all samples, in order to obtain comparable data. The optical
reflectivity was evaluated at a constant wavelength of 650 nm, as presented in
Figure 4.24a. With fluences < 11 mJ/cm2, only a slight increase in reflectivity is
observed. However, the reflectivity rises abruptly to 86 % after laser irradiation
with a fluence of 14 mJ/cm2. Subsequently, the reflectivity increases only slowly
upon a gradual fluence increase from 14 and 130 mJ/cm2, indicating a relatively
large stable region with high optical contrast (up to 130 mJ/cm²). From these
results, it can be deduced that the fluence threshold for crystallization of films
irradiated with 20 pulses lies between 11 and 14 mJ/cm2. Above this threshold, the
series of laser pulses heats the films above the crystallization temperature. The
atoms in the GeTe films thus become increasingly mobile and reach the
energetically favorable crystalline state, leading to a partial crystallization of the
film. With an increase of laser fluence, the degree of crystallization increases,
corresponding to the increase of film reflectivity. The highest optical contrast
between an as-deposited GeTe film and a laser-irradiated GeTe film is calculated to
Chapter 4 Results and discussion
72
be ~15%, which is much higher than the optical contrast values for laser-induced
phase transition of GST [66]. From the view of application, the higher reflectivity
contrast, the higher signal-to-noise ratio and the high absolute reflectivity of the
films are independent of the irradiation fluence in the range between 36 and
130 mJ/cm2, which indicates a high stability of optical contrast and, as a
consequence, a long data retention. Also, as demonstrated in Figure 4.24a, the
reflectivity of the laser-irradiated films with fluences ≥ 130 mJ/cm2 drops to 82%.
This is attributed to ablation of the film surface, as verified by the SEM image in
the inset of Figure 4.24a. Holes in the film with sizes ranging from 200 nm to 1 μm
are clearly visible. Additionally, an influence of laser-induced crystallization on the
film surface roughness can be excluded, based on evidence provided in Figure 4.25.
Figure 4.24: (a) Optical reflectivity of GeTe films at the wavelength of 650 nm after irradiation with
20 laser pulses (pulse duration: 20 ns) as a function of the laser fluence between 4 and 162 mJ/cm2.
The films were amorphous in the as-deposited state. The inset shows an SEM image of the film
topography after laser irradiation with 20 pulses and a fluence of 162 mJ/cm2. (b) XRD spectra of
GeTe films after laser irradiation with 20 pulses as function of the laser fluence between 4 to
130 mJ/cm2. The Miller indices correspond to the rhombohedral crystal structure of GeTe with the
space group R3m (JCPDS no. 47-1079). (c) Details of the XRD patterns around the 2θ position of the
rhombohedral (202) or/and cubic (200) reflections, respectively. (d) Corresponding interplanar
spacings d of the rhombohedral (202) planes or/and cubic (200) planes as a function of the laser
fluence. The dashed line serves only as a guide for the eye.
0 40 80 120
0.294
0.296
0.298
d (
nm
)
Laser fluence (mJ/cm2)
0 40 80 120 16070
75
80
85
90
Re
fle
ctivity (
%)
Laser fluence (mJ/cm2)
20 30 40 50
Inte
nsity (
arb
.un
its)
2 (°)
(00
3)
98 mJ/cm2
4 mJ/cm2
(22
0)
(20
2)
130 mJ/cm2
78 mJ/cm2
54 mJ/cm2
36 mJ/cm2
as-dep.
(02
1)
26 mJ/cm2
22 mJ/cm2
17 mJ/cm2
14 mJ/cm2
11 mJ/cm2
(02
4)
(a) (b)
29.5 30.0 30.5 31.0
Inte
nsity (
arb
.un
its)
2 (°)
98 mJ/cm2
4 mJ/cm2
130 mJ/cm2
78 mJ/cm2
54 mJ/cm2
36 mJ/cm2
as-dep.
26 mJ/cm2
22 mJ/cm2
17 mJ/cm2
14 mJ/cm2
11 mJ/cm2
(c) (d)
4.2 Laser-induced phase transitions of GeTe and GST films
73
It shows that the surface roughness of the ns laser-irradiated GeTe films is quite low
(~0.5 nm) and it remains constant, even when the laser fluence increases, in contrast
to the thermally annealed GeTe films.
Figure 4.24b presents the evolution of XRD patterns of films irradiated with
laser fluences between 0 and 130 mJ/cm2 after irradiation with 20 pulses. The
diffraction spectrum of the as-deposited film is typical for the amorphous state, as
shown before. No apparent differences in the diffraction spectra can be found up to
an applied fluence of 11 mJ/cm2. Above this value, crystalline diffraction peaks at
26.1°, 30.0°, 42.3°, 43.6°, which correspond to the (021), (202), (024) and (220)
lattice planes of the rhombohedral phase, respectively, emerge. The rhombohedral
structure can be interpreted as a slightly distorted cubic phase. This is in good
agreement with the results of the optical reflectivity measurements regarding the
threshold of the optical contrast increase. The intensity of diffraction peaks
increases slightly when the laser irradiation fluence is continuously increased from
14 to 36 mJ/cm². Simultaneously, the intensity of the (024) reflection gradually
disappears (as depicted in Figure 4.24b) because of the transformation from the
rhomhohedral GeTe phase into the rocksalt GeTe phase. Above a fluence of
36 mJ/cm2, only the latter phase is found. The measured fluence dependence
correlates with the assumption that a lower fluence is necessary for construction of
a low-symmetry rhombohedral unit cell than for a high-symmetry cubic unit cell,
because there are large atomic displacements required [132]. Nevertheless, it is
interesting that a peak shift to higher diffraction angles can be observed for fluences
Figure 4.25: Surface roughness of the GeTe films after laser irradiation at various fluences, compared
to the thermal annealing results at different annealing temperatures.
1 2 3 40
50
100
150
200
La
se
r flu
en
ce
(m
J/c
m2)
An
ne
alin
g t
em
pe
ratu
re (
°C)
RMS surface roughness (nm)
250
300
350
400
Chapter 4 Results and discussion
74
between 14 to 36 mJ/cm². Exemplarily, the shift of the 2θ position of the
rhombohedral phase (202) reflection at 30° towards the cubic phase (200) reflection
at 30.2° is plotted in Figure 4.24c and d. The corresponding interplanar spacing d
changes from about 0.297 nm to about 0.294 nm after this fluence increase (Figure
4.24d). The peak shift to higher diffraction angles or smaller interplanar spacings is
the result of the rhombodedral to cubic phase transformation induced by fluences
between 14 to 36 mJ/cm². It is assumed that this reduction of the interplanar spacing
is probably due to the release of film stress as a result of the transformation process
in this particular fluence range [133]. With further increase of the irradiating
fluence, from 36 to 130 mJ/cm², the interplanar spacing increases slightly, due to
the higher thermal expansion, resulting in the peak shifting back to a relatively low
angle, as shown in Figures 4.24c and d. To summarize these findings, the
crystallization phenomena require several laser pulses instead of only one, even at
the highest laser fluence used in this study.
Crystallization as a function of pulse repetition rate
The influence of the repetition rate in the range from 1 to 400 Hz on the reflectivity
of films triggered by 20 ns laser irradiation with an identical laser fluence of
78 mJ/cm2 (i.e., above the threshold for crystallization) and a fixed pulse number of
20 is shown in Figure 4.26a. Compared to the reflectivity of the as-deposited film, a
prominent increase in the reflectivity of the irradiated films of approx. 15% over the
measured wavelength range between 400 and 700 nm is obtained. As confirmed by
XRD analysis of these films (shown in Figure 4.26b), this increase in reflectivity is
due to the crystallization of the amorphous GeTe film. More interestingly, the
reflectivity increase of the films is consistently high for all chosen repetition rates
between 1 and 400 Hz, corresponding to delay times between pulses of 1 and
0.0025 s, respectively, and the resulting reflectivity curves match almost completely
over the entire measured wavelength range. Coincident with these findings, the
intensities of the GeTe phase diffraction peaks in the XRD measurements in Figure
4.26b are almost identical for the chosen laser pulse repetition rates. This indicates
that the transformation process is independent of the delay time between delivered
laser pulses in the investigated range. This is consistent with the findings of
4.2 Laser-induced phase transitions of GeTe and GST films
75
Coombs and co-workers [31], who determined the probability of nucleation to be
almost unity when the delay between pulses exceeds 35 ns. In principle, each laser
pulse induces an irreversible modification in the region of the amorphous film and
contributes to the overall crystallization process; this is the so-called energy
accumulation process [81].
The initial pulses might already cause some
modifications in the form of initial crystal nucleation as well as in the early growth
of nuclei that already existed within the overall amorphous matrix of the as-
deposited film [134]. During the investigations on the crystallization as a function of
the pulse number, it was found that the threshold number of pulses for
crystallization is ≥ 5 pulses, for which a higher optical reflectivity, caused by a
higher crystalline fraction, could only be observed for a higher number of pulses
Figure 4.26: (a) Optical reflectivity of an as-deposited GeTe film and of films after 20 ns laser pulse
irradiation with a fluence of 78 mJ/cm2 and 20 pulses as a function of the pulse repetition rate
varying between 1 Hz and 400 Hz. (b) Corresponding XRD patterns of GeTe films as a function of
the pulse repetition rate varying between 1 Hz and 400 Hz.
20 30 40 50
400 Hz
100 Hz
10 Hz
Inte
nsity (
arb
. units)
2(°)
1 Hz
(22
0)
(20
0)
(11
1)
400 500 600 70070
75
80
85
90
as-depositedRe
fle
ctivity (
%)
Wavelength (nm)
Repetition rate 1 Hz
10 Hz
100 Hz
400 Hz
(a)
(b)
Chapter 4 Results and discussion
76
(≥ 5). Such a pulse number dependence of the optical reflectivity in GeTe films
could be beneficial for the potential application of multilevel storage. In recently
published literature, a similar characteristic has also been found with AIST and
GST materials [83, 135]. A more detailed discussion follows in Chapter 4.3.
Crystallization as a function of film thickness
Apart from the influence of the pulse laser beam parameters on the crystallization
process presented above, another important influential variable for the application
of phase-change films is the use of very thin films. A reduction of the film thickness
can affect several material properties including crystallization temperature,
crystallization rate, optical constants, etc. [136]. In order to investigate the influence
of the film thickness on laser-induced crystallization, a series of GeTe films with
different thicknesses were prepared. Film thicknesses of 20, 51, 68, and 120 nm
were obtained by precisely controlling the amount of deposited material by using
pulse numbers of 400, 1000, 1200, and 2400 pulses, respectively. As a first example
regarding film thickness dependencies, an as-deposited GeTe film with a thickness
of 51 nm and the same film after 20 ns pulse laser irradiation with 20 pulses at a
fluence of 67 mJ/cm2 were characterized by XRR, as shown in Figure 4.27a. The
film thickness-related oscillations of the XRR curve from the as-deposited film are
present in the whole displayed angular range, whereas the oscillations in the XRR
curve from the laser-irradiated film are characterized by a smaller oscillation period
caused by a shrunken film thickness upon laser irradiation. This decrease in the
thickness of the films is due to the disordered amorphous-ordered crystalline phase
transformation of GeTe films, as evidenced by the XRD patterns of the irradiated
films shown in Figure 4.27c. Moreover, a shift of the critical angle of total
reflection towards higher angles is clearly visible in the XRR curve after laser
irradiation, compared to that of the as-deposited GeTe film. This indicates that the
laser irradiated film possesses a higher density than the as-deposited one, since the
critical angle θc for total reflection is proportional to the square root of the mass
density ρ as defined by the following equation [137]: 𝜃𝑐 ∝ λ√𝜌, where λ is the
wavelength of the used X-ray beam. By corresponding modeling fitting to the
measured data in Figure 4.27a, the films thickness and mass density can be
extracted as 51 nm and 5.5 g/cm3 for the as-deposited film and 45 nm and 6.0 g/cm
3
4.2 Laser-induced phase transitions of GeTe and GST films
77
for the post laser-irradiated film, respectively. This means that there is a decrease in
film thickness of about 10% upon laser-induced crystallization.
Figures 4.27b, c and d present the investigation on film thickness-dependent
optical reflectivity and XRD measurements for the 20 ns pulse laser-irradiated GeTe
films at 20 pulses and 67 mJ/cm2. The optical reflectivity contrast of the films at a
constant wavelength of 650 nm is depicted in Figure 4.27b. The reflectivity of the
as-deposited film remains as a reference (i.e., at 0% in terms of reflectivity
contrast). With an increase in film thickness from 20 to 51 nm, the reflectivity
contrast rises significantly. The highest achievable optical reflectivity contrast of
17% (with errors ± 2%) was observed for a film thickness of around 68 nm. With
further increase in thickness up to 120 nm, the reflectivity contrast shows no
obvious change and stays constant within the measurement error at this high
reflectivity contrast level. The corresponding XRD spectra of the laser-irradiated
Figure 4.27: (a) XRR measurements and the corresponding modeled curves for an as-deposited
GeTe film and the film after laser irradiation with 20 pulses and 67 mJ/cm2. The inset shows the
values of density and film thickness extracted from the measurement. (b) Optical reflectivity contrast
of GeTe films at the wavelength of 650 nm after irradiation with 20 pulses (pulse duration: 20 ns)
and 67 mJ/cm2 as a function of the film thickness that varies between 20 and 120 nm. (c) XRD
patterns of GeTe films after irradiation with 20 pulses as a function of the film thickness between 20
to 120 nm. (d) FWHM of the (200) reflection and corresponding calculated average grain sizes as a
function of the film thickness between 20 and 120 nm.
20 30 40 50
Inte
nsity (
arb
.units)
2 (°)
(220)
(200)
(111)
Si
120 nm
68 nm
51 nm
20 nm
as-deposited:experiment data
modeling data
laser-irradiated:experiment data
modeling data
0.2 0.4 0.6 0.8 1.0 1.2
101
102
103
104
105
Inte
nsity (
cps)
(°)
Si (100)
GeTe = 6.0 g/cm
3
t = 45 nm
Si (100)
GeTe = 5.5 g/cm
3
t = 51 nm
laser-irradiation
(a)
(c)
0 20 40 60 80 100 1200.20
0.25
0.30
0.35
0.40
0.45
0.50
0.55
Film thickness (nm)
FW
HM
(°)
20
30
40
50
60
70
80
Gra
in s
ize
(n
m)
(b)
(200)
(d)
0 20 40 60 80 100 120
0
5
10
15
20
Reflectivity c
ontr
ast (%
)
Film thickness (nm)
Chapter 4 Results and discussion
78
GeTe films with film thickness varied between 20 and 120 nm are shown in Figure
4.27c. First, it is obvious that there are reflections present for all film thicknesses
and the detected crystalline diffraction peaks of GeTe films are annotated. As
expected, the peak intensity of the crystalline phase decreases with decreasing film
thickness. It can be seen that the peak intensity from the 20 nm thick film is very
low due to the lower crystallized volume, when compared to the thicker one. It
should be noted that an increased effect of the interface on GeTe crystallization with
reduced film thickness below a critical thickness is also possibly involved. It was
reported in the literature that the crystallization temperature increases as the film
thickness is reduced below a critical thickness [138, 139]. In Figure 4.27d, the
average grain size of laser-irradiated GeTe films is plotted as a function of the film
thickness, estimated from the FWHM value of the (200) reflection by using the
Scherrer formula. The average grain size decreases with reduction of the film
thickness, with the grain size varying between 21 and 62 nm in the examined film
thickness range. Therefore, it can be concluded that the crystallization process of
GeTe films can be tailored with appropriate control by the irradiated ns pulse laser
parameters (i.e., pulse number, laser fluence) and by the film thickness.
2.) Amorphization process
It was reported in the literature that amorphization of a crystalline GeTe film can
be achieved with single laser pulse irradiation with a pulse length of 8 ns, a
wavelength of 800 nm and a fluence of 71 mJ/cm² [140]. For the present study, the
GeTe thin films, which were crystallized by ns laser irradiation as described in the
previous sections, were re-amorphized by a single 20 ns laser pulse irradiation at a
wavelength of 248 nm and at various laser fluences up to 195 mJ/cm². In Figure
4.28a, the optical reflectivity curves in the spectral region from 400 up to 700 nm of
single pulse irradiated GeTe films are shown. For comparison, the reflectivity
curves of an as-deposited GeTe film and of a GeTe film after 20 min furnace
annealing at a temperature of 300 °C are also shown. The reflectivity of this latter
crystalline film varies between 90 and 93%. Figure 4.28 also reveals that an obvious
reflectivity change after irradiation by a single laser pulse at a fluence of 98 mJ/cm²
cannot be observed. However, with an increase of the laser fluence up to about
160 mJ/cm², a continuous and gradual decrease in reflectivity is detected. Gawelda
4.2 Laser-induced phase transitions of GeTe and GST films
79
et al. have also observed such a decreasing trend of the reflectivity of GeTe films
after 800 nm single pulse irradiation for fluences > 50 mJ/cm², which corresponds
to the formation of amorphous regions but also to increased ablation [140].
With further increase of the fluence from 160 to 182 mJ/cm2, a dramatic
decrease in reflectivity can be detected in Figure 4.28a. Apparently, the atoms in the
GeTe films are increasingly displaced from their lattice sites above a laser fluence
of 98 mJ/cm². The displaced atoms are quenched during the very short laser pulse
interaction with the film, which leads to a partial amorphization of the film. With
increasing laser fluence, the degree of amorphization, i.e. the amorphized volume
fraction, increases, corresponding to the decrease of film reflectivity. However,
Figure 4.28: (a) Optical reflectivity of GeTe films in the wavelength range between 400 and 700 nm
after irradiation with a single 20 ns laser pulse as function of the laser fluence. For comparison, the
optical reflectivity curves of an as-deposited GeTe film and of a crystalline GeTe film after thermal
annealing at 300°C for 20 min are shown. The inset displays an SEM image of the surface after
single laser pulse irradiation with a fluence of 195 mJ/cm2. (b) XRD patterns of ns laser pulse
irradiated GeTe films as a function of the laser pulse fluence.
20 30 40 50
Inte
nsity (
arb
. u
nits)
2(°)
182 mJ/cm2
(220)
(202)
98 mJ/cm2
130 mJ/cm2
160 mJ/cm2
(024)
(003)
(021)
400 500 600 70070
75
80
85
90
Reflectivity (
%)
Wavelength (nm)
98 mJ/cm2
160 mJ/cm2
130 mJ/cm2
182 mJ/cm2
oven annealed
as-dep.
195 mJ/cm2
(a)
(b)
Chapter 4 Results and discussion
80
when the fluence increases to much higher values (≥195 mJ/cm2), the reflectivity of
the laser-irradiated films shows only a slight further decrease, due to the increased
fraction of amorphization and a significant ablation, as visible in the inserted SEM
images in Figure 4.28a. In order to confirm the structural transition from the
crystalline to the amorphous state, those films irradiated with laser fluences between
98 and 182 mJ/cm2 were studied by XRD, as shown in Figure 4.28b. It can be
observed that, with an increase of the fluence from 98 to 182 mJ/cm2, the intensity
of diffraction peaks decreases gradually. Thus, the amorphous fraction in the
crystalline films increases with increasing laser fluence. These consistent results
demonstrate that amorphization of crystalline GeTe films can also be achieved with
single 20 ns UV laser pulse irradiation at sufficiently high fluences.
Crystallization of Ge2Sb2Te5 films by nano- and femtosecond single 4.2.2
laser pulse irradiation
As known from the discussion in Chapter 2.2, GST possesses a potentially higher
switching speed, compared to GeTe. Moreover, the crystallization-dominated
mechanisms differ between GST (nucleation-dominated) and GeTe (growth-
dominated). In contrast to the previous sub-chapter (4.2.1), in this sub-chapter
investigations of the crystallization characteristics of PLD-deposited GST films
with either ns or fs single pulse laser irradiation at the same wavelength of 248 nm
are compared. The details of the crystallization process are characterized by
investigating the film structure, morphology and optical reflectivity as a function of
the laser fluence [131].
1.) Structure analysis
XRD measurements were employed to investigate the influence of the laser fluence
on the crystallinity of as-deposited 90 nm thick GST films upon single pulse ns and
fs laser irradiation. Figure 4.29a presents the evolution of XRD patterns of films
irradiated with increasing laser fluences. The fluence was varied between 0 and
130 mJ/cm2 and between 13 and 19 mJ/cm
2 for pulse durations of 20 ns and 500 fs,
respectively. The diffraction spectrum of the as-deposited film shows that the film is
in the amorphous state. The diffraction peak at ~33° is a contribution from the Si
substrate. For 20 ns laser irradiation with fluences ≥ 36 mJ/cm2, crystalline phase
reflections at 29.6° and 42.7° are visible, which correspond to the (200) and (220)
4.2 Laser-induced phase transitions of GeTe and GST films
81
lattice planes of the metastable fcc phase of GST, respectively. For laser fluences
above 98 mJ/cm2, the diffracted intensity is significantly increased and the (111)
reflection also becomes visible. It should be noted that, although partial
crystallization of the films is confirmed from optical reflection measurements (as
described in section 4 of this sub-chapter) for fluences between 15 and 36 mJ/cm2,
the XRD reflections of the crystalline phase cannot be detected, due to the low
volume fraction of the crystalline phase in this particular fluence range. With a
continuous increase of the laser irradiation fluence from 36 to 98 mJ/cm2, the
intensity of the reflections becomes gradually proportional to the volume fraction of
the crystalline phase. For the films irradiated with a single fs pulse, the diffraction
spectra are qualitatively the same with the exception that the first diffraction peaks
are already observed at a fluence of 13 mJ/cm2. A further increase of the laser
Figure 4.29: (a) XRD patterns of an as-deposited GST film and of films after 500 fs and 20 ns laser
single pulse irradiation at different fluences. (b) FWHM of the rocksalt phase GST(200) reflection
and the corresponding calculated average grain sizes as a function of the laser fluence (fs and ns
denoted by the symbols with solid and open interiors, respectively).
Laser fluence (mJ/cm2)
(a)
20 30 40 50
fcc(1
11)
fcc(2
00)
fcc(2
20)
Si(2
00)
as-dep.
15 mJ/cm2
36 mJ/cm2
67 mJ/cm2
98 mJ/cm2
13 mJ/cm2
16 mJ/cm2
19 mJ/cm2
ns single pulse
fs single pulse
130 mJ/cm2
Inte
nsity (
arb
. u
nits)
2 (°)
40 60 80 100 120
ns single pulse
30
35
40
45
50
55
60
Gra
in s
ize
(n
m)
12 15 180.24
0.28
0.32
0.36
0.40
fs single pulse
FW
HM
at
(20
0)
(°)
(b)
Chapter 4 Results and discussion
82
fluence promotes crystallite growth, as evidenced from the increase in diffracted
intensity. The average grain size of GST films after ns and fs irradiation as a
function of laser fluence was estimated from the FWHM of the (200) reflection by
using the Scherrer formula and the results are presented in Figure 4.29b. It follows
that the average grain size decreases from approx. 50 to 35 nm after single pulse ns
laser irradiation in the fluence range between 36 and 130 mJ/cm². In the left part of
Figure 4.29b, this tendency can also be detected for the fs irradiated films. For GST
films after fs pulse laser irradiation, the average grain size gradually increases from
about 50 to 60 nm with an increase of the fluence from 13 to 19 mJ/cm2. These
experiments demonstrate that a crystallization of the GST films is possible in both
cases: after ns single pulse and also after fs single pulse laser irradiation for laser
fluences larger than 36 mJ/cm² and larger than 13 mJ/cm², respectively.
2.) Morphology
To understand the crystallite growth process in the GST thin films after ns and fs
single pulse laser irradiation, the microstructure evolution was examined by cross-
sectional TEM. Figure 4.30 shows a cross-sectional bright-field (BF) TEM image
and a NBD pattern of an as-deposited GST thin film (Figures 4.30a and c), a film
after ns single pulse laser irradiation with fluences of 36 mJ/cm2 (Figures 4.30b, d
and e) and a film with 67 mJ/cm2 (Figure 4.30f) as well as after fs single pulse laser
irradiation with a fluence of 19 mJ/cm2
(Figure 4.30g). In Figure 4.30a, a uniform
and smooth GST film above the SiO2 / Si layer is shown; the corresponding NBD
pattern in Figure 4.30c is typical for an amorphous structure (the top layer of Pt was
only used to stabilize the samples during the TEM preparation). In contrast,
crystalline grains with an oval shape developed within the amorphous GST matrix
after laser irradiation by a 20 ns single laser pulse with a fluence of 36 mJ/cm2
(Figure 4.30b). With a further increase of the laser fluence to 67 mJ/cm2, the
crystalline precipitates in the amorphous regions grow to grains with an oval shape,
covering the film over its entire thickness. Between these grains, small angle grain
boundaries are visible (see Figure 4.30f). Similarly, the GST film occurs to be
completely crystalline and the grain diameters in the film vary between 15 and 60
nm when the laser fluence increases up to 130 mJ/cm2. In contrast, the morphology
after 500 fs single pulse irradiation of amorphous GST thin films differs from the
4.2 Laser-induced phase transitions of GeTe and GST films
83
morphology of GST thin films irradiated by a ns single pulse. Figure 4.30g shows a
cross-sectional BF image of the GST thin film after single pulse fs laser irradiation
with a fluence of 19 mJ/cm2. A homogeneous polycrystalline structure with
columnar grains can be observed over the complete film thickness of ~90 nm. The
diameter of these columnar grains lies in the range between 20 and 60 nm. It should
be noted that the near surface region of the films was altered by natural oxidation in
air. However, this oxidation can be avoided by depositing a 5 nm thin capping layer
Figure 4.30: Cross-sectional BF-TEM images and NBD patterns of GST thin films: (a) and (c) as
deposited; (b), (d) and (e) after ns laser pulse irradiation with a fluence of 36 mJ/cm2; (f) after ns
laser pulse irradiation with a fluence of 67 mJ/cm2; the inset shows a magnified region; (g) after fs
laser pulse irradiation with a fluence of 19 mJ/cm2.
Chapter 4 Results and discussion
84
of LaAlOx onto the top of the GST films. As an example, GST films with and
without a LaAlOx layer are shown in Figure 4.31.
3.) Local atomic arrangement
High-resolution STEM (HRSTEM) investigations allow a direct insight into the
local atomic arrangement. In Figure 4.32a, a middle-angle annular dark-field
(MAADF) HRSTEM image along the [001] viewing direction of an as-deposited
GST film after single fs laser pulse irradiation is depicted. The real-space lattice
fringes in the image indicate that the GST film is characterized by a high crystalline
quality. The crystal plane spacing of the film is approx. 0.302 nm, which
corresponds to the {002} crystal plane spacing (0.302 nm) of the cubic GST
crystalline phase. It is in good agreement with the results from XRD measurements;
the interplanar distance of the {002} crystal plane is calculated to be 0.301 nm,
according to the Bragg equation. It should be noted that no additional sub-lattices
[141] are formed by Ge atoms located at tetrahedral positions or Ge atoms located at
off-octahedral positions [142] as evident from atomic-resolution low-angle annular
dark field (LAADF) HRSTEM studies of fs laser pulse irradiated GST films (see
Figure 4.32b). This is in contrast to the results of Liu et al., which claim a
coexistence of octahedral and tetrahedral sites for Ge in the GST lattice [48]. In
Figure 4.31: Contrast between the layer structures of laser-irradiated LAO/GST/SiOx and GST/SiOx
samples in cross-section TEM images.
Pt
GST ~90 nm
SiOX
GST:O
98 mJ/cm2
ns single pulse
Pt
GST ~90 nm
SiOX
LaAlO3
schematic
90 mJ/cm2
4.2 Laser-induced phase transitions of GeTe and GST films
85
addition, detailed analyses of local distortions in the GST lattice showed that a
distorted octahedral atomic arrangement of GeSb is the main structural motif for the
metastable and amorphous GST phases [141]. Thus, due to laser irradiation, these
distortions of the GST lattice can even be enhanced. This will induce a breakage of
long-range order and lead to the transition from the metastable to the stable GST
phase, as was proposed by Kolobov et al. [18]. Another possibility for phase
transition is melting of the GST lattice, due to lattice heating caused by the transfer
of electron energy to the covalent backbone by applying an ultrashort laser pulse
Figure 4.32: Aberration-corrected (a) MAADF- and (b) LAADF-HRSTEM images of a
representative fs single laser pulse irradiated GST film viewed along the [001] zone axis. The insets
show a magnification of the marked regions.
Chapter 4 Results and discussion
86
[59]. Moreover, a non-thermal amorphization process caused by displacement of Ge
atoms from octahedral to tetrahedral sites, induced by ultrashort laser pulses, was
also recently reported [19].
4.) Optical reflectivity
Figure 4.33 shows the reflectivity of Ge2Sb2Te5 films deposited at room temperature
after 20 ns single pulse laser irradiation with fluences of 15 and 67 mJ/cm2, and
after 500 fs single pulse laser irradiation with a fluence of 15 mJ/cm2. The optical
reflectivity of the as-deposited, amorphous GST film is approximately 43% for
wavelengths between 400 and 700 nm. In contrast, the reflectivity of the film after
ns pulse laser irradiation with a fluence of 15 mJ/cm2 varies between 52 and 58%.
With increasing laser fluence, up to 67 mJ/cm2, an even higher reflectivity in the
range between 65 and 67% is observed. For the film that was fs-irradiated with a
fluence of 15 J/cm², the optical reflectivity is in the range from 64 to 67%, which is
comparable to that of the ns-irradiated film with a fluence of 67 mJ/cm2. These
results demonstrate that a phase transition from the amorphous to the crystalline
state can be achieved by single ns laser pulse irradiation with a fluence ≥ 15
mJ/cm². With an increase of the laser fluence (67 mJ/cm², Figure 4.33) the optical
reflectivity is further increased. This is an indication that the initially small
crystalline volume embedded in the amorphous matrix increased. A comparison of
the reflectivity after ns and fs single laser pulse irradiation in Figure 4.33 leads to
Figure 4.33: Optical reflectivity of GST films after deposition (as-dep.), after 20 ns single laser pulse
irradiation at fluences of 15 and 67 mJ/cm2, and after 500 fs single laser pulse irradiation at a fluence
of 15 mJ/cm2.
400 500 600 700
40
50
60
70 ns-induced, 67 mJ/cm2
ns-induced, 15 mJ/cm2
fs-induced, 15 mJ/cm2
Reflectivity (
%)
Wavelength (nm)
as-dep.
4.2 Laser-induced phase transitions of GeTe and GST films
87
the conclusion that obviously single pulse fs laser irradiation is more effective
regarding the crystallization process. Similar results were also found by XRD
characterization (see section 1 of this sub-chapter).
In Figure 4.34, a series of fluence-dependent optical reflectivity contrast values
at a wavelength of 650 nm of GST films is shown. As expected, the reflectivity
contrast of the films increases with increasing laser fluence, where the reflectivity
rise after fs irradiation is very steep, up to 15 mJ/cm², in contrast to the ns laser
irradiation. A simple rescaling procedure of the laser fluence yields the number of
photons per unit area Np
𝑁𝑝 =∅
𝐸𝑝= ∅
𝜆
ℎ∙𝑐 , (4.2)
where Φ is the laser fluence, Ep the energy of the photons, h the Planck constant, c
the speed of light in vacuum and λ the laser light wavelength (here: 248 nm).
Further, the change of the optical reflectivity contrast is proportional to the amount
of the crystalline fraction implied by the laser photons [17, 83]. Consequently, it can
be assumed that the crystallization of the thin GST films is completed when no
further change of the optical contrast can be determined as a function of the laser
fluence. The results of the rescaling procedure are summarized in Figure 4.34.
In the idealized case, it might be assumed that every photon that hits the surface
produces a cylinder-like crystallized region into the surrounding amorphous matrix
with a sufficiently large crystalline area ac at the surface. Then, the formation rate of
Figure 4.34: Optical reflectivity contrast of GST films at a wavelength of 650 nm after ns and fs
laser pulse irradiation with a single pulse as a function of the laser fluence and the number of laser
photons per area.
0 20 40 60 80 100 120
0
5
10
15
20
25
ns single pulse
fs single pulse
Re
fle
ctivity c
on
tra
st
(%)
Laser fluence (mJ/cm2)
0 5 10 15
0.00
0.25
0.50
0.75
1.00
Cry
sta
llin
e f
ractio
n
No. of photon/area [
cm-2]
Chapter 4 Results and discussion
88
the crystalline fraction with the number of photons per area (laser fluence) can be
given by
𝑑𝐴𝑐
𝑑𝑁𝑝= 𝑎𝑐 (1 −
𝐴𝑐
𝐴𝑜) , (4.3)
where Ac is the area covered by crystalline clusters and Ao is the total irradiated
area. The integration of this simple differential equation (4.3) leads to
𝐴𝑐
𝐴𝑜= 1 − 𝑒𝑥𝑝(−𝑎𝑐𝑁𝑝) , (4.4)
where Ac/Ao represents the crystalline fraction on the total area. Equation (4.4) was
used to fit the experimental data in Figure 4.34, where the black (dash) and red
(dash dot) lines indicate the crystalline fraction according to Equation (4.4) after
single ns and fs pulse laser irradiation, respectively. Good fits are obtained, if the
circular areas of ac = 0.0025 nm² and ac = 0.0115 nm² are used for ns and fs laser
irradiation, respectively. These results correspond to the average radii of the
cylinder-like crystalline clusters of rc = 0.02 nm and rc = 0.043 nm, respectively.
The maximum length of the cylinder-like region is given by the total absorption of
the incident laser light (wavelength is 248 nm and photon energy is about 5 eV).
According to the Beer-Lambert absorption law, the absorption length can be
determined as 50 nm, when an absorption coefficient of 8·105 cm
-1 at the given
wavelength is assumed for crystalline and amorphous GST phases [143].
Consequently, the average volume that is transferred by a single laser photon with a
wavelength of 248 nm is 0.125 nm³ and 0.575 nm³ after ns and fs laser irradiation,
respectively. Assuming that the atomic density of the amorphous phase is 3.34×1022
atoms/cm³ [144], about 5 and 20 GST atoms participate in the amorphous-
crystalline phase transition per photon by ns and fs laser irradiation, respectively.
This result illustrates that the single fs pulse laser irradiation appears to be more
effective in inducing crystallization of GST thin films than the single ns pulse laser
irradiation. Explanations could be an extreme suppression of the thermal diffusion
[145, 146], because the shorter laser pulse duration is beneficial for reducing the
dwell time of the laser spot. Another possibility is the effect of multi-photon
absorption occurring during fs laser irradiation [147].
4.2 Laser-induced phase transitions of GeTe and GST films
89
Simulation of temperature distribution after pulse laser irradiation 4.2.3
To correlate the thermal analysis and laser irradiation experiments in Chapters 4.2.1
and 4.2.2, the temperature distribution of the laser beam interaction with the films is
simulated and discussed in this sub-chapter. To verify the crystalline growth
mechanism of GeTe and GST films in terms of the heat flow condition after laser
irradiation, the finite element method (program package COMSOL) was used to
estimate the temperature distribution in the films.
1.) Temperature distribution after ns laser irradiation in GeTe film
The phase transitions in GeTe films are strongly determined by the temperature rise
caused by laser irradiation. The spatial distribution of the temperature T was
calculated by a finite element method [148] as function of the pulse laser fluence Φ
by solving the heat conductivity equation [149-151]
GeTeat
Siat
z)α(αΔt
)R()rΦ(t,
)rQ(t,))rT(t,(κ)r(t,Tcρ
p
opts
ssGeTe,Sisp;GeTe,SiGeTe,Si
0
exp1
, (4.5)
where Q is the laser beam induced heat source, rs is the spatial coordinate (rs =
rs(x,y,z), z is the depth, x is the distance from the laser spot center. The assumed
material parameters of GeTe are summarized in Table 4.3.
Figure 4.35 shows the calculated temperature distribution in a 60 nm thick GeTe
film on a Si substrate after 20 ns single pulse laser irradiation with a fluence of
14 mJ/cm², which represents the threshold fluence for crystallization of amorphous
GeTe. It is obvious that the temperature field in the GeTe film is comparably
uniform to a depth of about 60 nm (corresponding to the film thickness) and, in the
lateral dimension, 8 mm in the beam profile at the top heat. Consequently, it is
assumed that also for higher laser fluences, the complete films can be described by
a defined temperature. At these low laser fluences, the temperature of the GeTe film
Table 4.3: Summary of the assumed GeTe material parameters [127, 152].
density ρ 6.18 g·cm-3
heat capacity at constant pressure cp 250 J·kg-1
·K-1
thermal conductivity κ 80 W·m-1
K-1
reflectivity R 74 %
absorption coefficient α 0.053 nm-1
Chapter 4 Results and discussion
90
increases by only approx. 30 K. Considerably higher temperature rises can be
expected for higher laser fluences.
2.) Temperature distribution after ns/fs laser irradiation in GST film
For GST films irradiated by ns and fs single pulse laser, one- and two-temperature
models [150] are used, respectively. The calculated temperatures are for layered
samples (GST/SiO2/Si) as well as for the results of the time-dependent temperature
distribution in the surface and interface of the films. For ultrashort laser pulses
smaller than 10 ps, a disparity of the electron temperature and the lattice
temperature is supposed, and the laser-solid interaction can be physically described
by a two-temperature model [150].
Two-temperature model:
Electron temperature (Te) in GST:
)()exp()1(),(
),()),((),( ,,;,
ep
p
opts
sseeGSTseeGSTpeGST
TTzt
Rrt
rtQrtTrtTc
(4.6)
Phonon / lattice temperature (Tp) in GST:
)()),((),( ,;, pesppGSTsppppGST TTrtTrtTc (4.7)
At short laser pulses, e.g., ns laser pulses, an equality of the Te and Tp can be
assumed and a one-temperature model allows a good description of the physical
process.
Figure 4.35: Modeled two-dimensional temperature distribution of a 60 nm thick GeTe film on a Si
substrate after single laser pulse irradiation with a fluence of 14 mJ/cm2 for 20 ns. z is the depth, x is
the distance from the laser spot center.
0 1 2 3 4 5 6-200
-150
-100
-50
0
273
Z (
nm
)
X (mm)
T (K)296
4.2 Laser-induced phase transitions of GeTe and GST films
91
One-temperature model (Tp = Te=T) in GST:
)exp()1(),(
),()),((),(;
zt
Rrt
rtQrtTrtTc
p
opts
ssGSTsGSTpGST
, (4.8)
where Q is the laser beam induced heat source, rs is the spatial coordinate (rs =
rs(x,z), z is the depth, x is the distance from the laser spot center), t is the time and
Δtp the laser pulse duration.
The assumed material parameters of GST are summarized in Table 4.4. At this
first estimation, the material parameters are assumed to be constant. The material
parameters of SiO2 and Si were taken from the COMSOL material database.
Further, the thermal diffusivity into the Si and SiO2 layer was calculated by a one-
temperature model in both cases. For the laser beam profile, a homogeneous beam
profile was assumed. The aim of the model is an estimation of the maximum
achievable temperature induced by laser radiation. Further energy loss mechanisms,
such as laser-plasma absorption and melting and evaporation enthalpy, were not
considered. The simulated result of ns and fs single pulse laser irradiation is
depicted in Figure 4.36.
Figure 4.36a shows the two-dimensional lattice temperature distribution in a
90 nm GST/500 nm SiO2/Si substrate system 50 ps after the single pulse
femtosecond laser irradiation with a fluence of 19 mJ/cm2, as compared to ns one
(see in Figure 4.36b). The temperature distribution shows a surface-localized heated
region, where there is a relatively rapid temperature rise and a higher temperature in
the top surface of the GST film, with a larger temperature gradient through the layer
thickness, as illustrated Figure 4.36a and in left of Figure 4.36c. Although the lattice
Table 4.4: Summary of the assumed GST material parameters [66, 143, 153, 154].
density ρ 5.5 g·cm-3
heat capacity at constant pressure cp 1.55 J·kg-1
·K-1
heat capacity at constant pressure cp;p/e 1.395/0.155 J·kg-1
·K-1
thermal conductivity κ 0.27 W·m-1
K-1
reflectivity R 65 %
absorption coefficient α 0.08× nm-1
electron-photon coupling strength γ 1017
W/(m3∙K)
Chapter 4 Results and discussion
92
temperature increases by about 864 K (591°C) in the top surface, which is much
higher than the conventional crystallization temperature (~140°C) for GST films,
the crystallization temperature with increasing heating rate shifts to higher
temperatures (the measured Tc of GST is 356.5° at a heating rate of 4 × 104 K/s)
[25, 55].
Figure 4.36c shows the simulation results of electron and lattice temperature
distribution at the surface of the film as a function of time. The electron temperature
immediately rises up to extremely high values (above ~6500 K) during fs pulse
laser irradiation on the GST film, then it quickly cools down (~860 K) after
40-50 ps, whereas the lattice temperature gradually increases with increasing
irradiation time, but it shows the same temperature range ~860 K at the time scale
of 50 ps. It is noted that the lattice temperature in the interface keeps almost
unchanged (only up to approximately 295 K). The heating and cooling rate of
carrier thermalization is calculated to be approx. 1.3×1016
K/s and 1.7×1013
K/s,
Figure 4.36: Modeled two-dimensional temperature distribution of a 90 nm thick GST film on a
SiO2/Si substrate: (a) after single fs pulse laser irradiation with a fluence of 19 mJ/cm2 for 50 ps, (b)
after single ns pulse laser irradiation with a fluence of 36 mJ/cm2 for 20 ns. z is the depth, x is the
distance from the laser spot center. (c) A comparison of temperature time dependence at the top
surface center of the GST film: carrier and lattice temperature as function of time after one fs single
pulse at a laser fluence of 19 mJ/cm2, (right) after one ns single pulse at a laser fluence of 36 mJ/cm
2.
(d) Corresponding comparison of the cooling rates directly after the laser pulse as well as after a
delay time of 100 ns for ns and fs single laser pulse irradiation.
0 200 400 600 800 1000
-600
-400
-200
0
293
864
T (K)
z (
nm
)
x (nm)
Si
SiO2
Ge2Sb
2Te
5
0 200 400 600 800 1000
-600
-400
-200
0
293
1288
T (K)
z (
nm
)
x (nm)
Si
SiO2
Ge2Sb
2Te
5
(a) (b)
(c) (d)
0 20 40 60 80 100
400
600
800
1000
1200
1400
8 x108 K/s
4 x108 K/s
1.5 x1011
K/s
Te
mp
era
ture
(K
)
Time (ns)
5 x1010
K/s
Cooling rates
ns single pulse
fs single pulse
10-14
10-13
10-12
10-11
10-9
10-8
10-7
1000
2000
3000
4000
5000
6000
7000
Tp
t= 500 fs
t= 20 ns
fs pulse
Tem
pera
ture
(K
)
Time (s)
ns pulseT
e
4.2 Laser-induced phase transitions of GeTe and GST films
93
respectively. For comparison, the simulated temperature distribution as a function of
time in a GST film after single pulse ns laser irradiation is shown in the right part of
Figure 4.36c. Moreover, it is found that after single pulse fs laser irradiation within
approximately 100 ns room temperature is reached as shown in Figure 4.36d. In the
case of the fs single pulse irradiation, the cooling rate decreases from approx.
1.5 ×1011
K/s within the initial 50 ps to 8 ×108 K/s after a delay time of 100 ns,
whereas in the case of the ns single pulse irradiation, a decrease of the cooling rate
from initially 5 ×1010
K/s to 4 ×108 K/s after a delay time of 100 ns results (see
Figure 4.36d). The above represented results of the two-temperature model analysis
for fs laser irradiation describe the non-equilibrium process, which is more accurate
than used normal thermal diffusion equations. During laser-material interaction, the
electrons in the materials are first excited by the deposited laser energy, and these
“hot” electrons are then cooled by transferring their energy to the lattice through
electron-phonon coupling [155, 156]. Nevertheless, the temperature in the GST film
upon ns/fs laser irradiation, as calculations show, exceeds the crystallization
temperature of GST. Consequently, a crystallization process in the GST films can be
expected.
Summary of the results on laser-induced phase transitions of GeTe and 4.2.4
GST films
In this chapter, the investigation of the phase transformation of GeTe and GST films
upon pulsed laser irradiation was presented. Additionally, simulated temperature
distributions of laser irradiated samples were used to confirm the physics behind the
phase transformation after laser irradiation.
Nanosecond laser-induced phase transition in GeTe films
Phase transformation processes of PLD-grown GeTe films on silicon substrates
were initiated by irradiation with 20 ns laser pulses at a wavelength of 248 nm. The
investigations of these films revealed that a reversible phase transition was realized
by using pulse numbers ≥ 5 at fluences above the threshold fluence between 11 and
14 mJ/cm2 for the crystallization process and by using single pulses at fluences
between 162 and 182 mJ/cm² for the amorphization process. Moreover, a fluence-
dependent structural evolution from the rhombohedral to cubic phase transition was
observed. The reflectivity contrast between the laser irradiated and as-deposited
Chapter 4 Results and discussion
94
films was studied; it reached a maximum of 14.7% for laser fluences between 36
and 130 mJ/cm2. It should be noted that the crystallization of GeTe films could be
shown to be independent of the irradiated laser pulse repetition rate between 1 and
400 Hz. The influence of film thickness variation between 20 and 120 nm on
crystallization was investigated. A reduction of approximately 10% in the film
thickness of laser-irradiated crystallized GeTe films, compared to as-deposited
amorphous films, was found. Moreover, with decreasing film thickness, smaller
crystallized volumes were detected, because the probability of an effect of the
interface increased with smaller film thickness, it envisions an enhancement of their
crystallization temperature.
Crystallization of GST films by nano- and femtosecond single laser pulse
irradiation
Crystallization of GST thin films deposited by PLD was triggered by ns and fs
single UV laser pulses. The phase transition from the amorphous to the crystalline
state was studied as a function of the laser fluence. The appearance of a crystalline
state was found after ns-laser and fs-laser irradiation with fluences ≥ 36 mJ/cm² and
≥ 13 mJ/cm², respectively. Various crystallite shapes and growth mechanisms
following irradiation with fs and ns pulse durations were observed. GST films
irradiated with fs-laser pulses show a columnar growth, whereas in ns-laser
irradiated films, oval shaped crystallites were produced. In addition, neither
tetrahedral nor off-octahedral positions of Ge-atoms were detected in the defect-free
crystallites after fs-laser irradiation by HRSTEM examination. The optical
reflectivity contrast increased with the laser fluence for both ns and fs pulse laser
irradiation. A simple model of the irradiation-induced crystalline phase formation
was used to assess the average contribution of ns and fs laser photons to the phase
transition process. A change in the optical reflectivity contrast of 25% could be
obtained by single pulse fs-laser irradiation with a fluence ≥ 15 mJ/cm² or ns-laser
irradiation with a fluence ≥ 67 mJ/cm². This reflectivity contrast change of 25%
corresponds to a complete crystallization of the GST film.
4.3 Dynamic optical switching of non-volatile multi-level memory
95
Dynamic optical switching of non-volatile multi-level 4.3
memory
Information storage relying on chalcogenide-based alloys and performed by
switching between fully amorphous and crystalline phases was originally reported
by S. R. Ovshinsky [4]. Increasing the memory storage density achieved by
accumulated switching of phase-change materials, so called multi-level storage, has
recently been attracting extensive attention for applications in non-volatile memory
devices [67, 81, 83, 157-162]. In multi-level storage, the crystallization rate, driven
by appropriate amplitude and pulse duration of laser or electric current, reflects
distinct multi-levels of either electrical resistivity or optical reflectivity, which are
utilized for the identification of storage levels. The possibility to identify different
storage levels is determined by the maximum number of levels in a memory cell.
Therefore, an accurate control of the crystallization process and a high degree of
reproducibility are crucial for the obtained distinct intervals of intermediate levels
upon rapid phase transformation of phase-change materials. In this chapter, it is
mainly demonstrated that the optical reflectivity of GeTe films changes
accumulatively with multi ns laser pulse-induced crystallization between logic state
‘1’ (fully amorphous) and logic state ‘0’ (fully crystalline). An analogous behavior
has been observed for electrical multi-level storage [77, 81, 161].
Besides the benefit of increasing storage density, multi-level storage within a
single layer phase-change film also brings the inherent advantage of low power
consumption. Routinely, the recording (amorphization) process is obtained by
applying an intense and short pulse, either a laser or a current pulse, which induces
melting and fast quenching, whilst a moderately intense and long laser/current pulse
is used to thermally anneal the material above the crystallization temperature within
a sufficient amount of time in the erasing (crystallization) process. With the
application of multi-level storage, each appropriate excitation with a pulse
contributes to achieving a well-controlled partial crystallization/amorphization and
this allows multiple bits for recording/erasing of information in each memory cell,
based on the respective mode of energy accumulation. Consequently, the energy
consumption per bit can be reduced.
Chapter 4 Results and discussion
96
The work in this chapter is based on results obtained with the previously
described UV ns laser pump-probe setup system presented in Chapter 3.3.2. The
aim is to describe the key aspects of the proposed simple memory cell: 1) Non-
volatile memory. 2) Feasible and reliable control of the crystallization of GeTe thin
films for multi-level storage in a single memory cell. 3) Dynamic switching within a
few nanoseconds. 4) Detailed insights into the switching mechanism of structural
transformation and morphology.
Optical switching 4.3.1
Figure 4.37 displays the On and Off switching states operation of the optical
reflectivity as a function of laser pulse number and fluence, respectively. The PLD-
prepared GeTe films on Si substrates with a LaAlOx capping layer for this type of
Figure 4.37: Optical switching (On and Off states) operation of GeTe films as a function of the laser
pulse number and the fluence: (a) Switching On process: low to high reflectivity change of the film
as a function of the pulse number at a constant fluence of 26 mJ/cm2. (b) Switching Off process: high
to low reflectivity change of the films as a function of the laser fluence with only one single pulse,
(c) and (d) the reflected pump laser intensity with a constant fluence of 26 mJ/cm2 and various
fluences of 0-135 mJ/cm2 as a function of time, corresponding to (a) and (b), respectively.
0 2 4 6 8 10 12 80100
0
25
50
75
100
Rela
tive r
eflectivity (
%)
Number of pulses
0 1 2 3
UV
la
se
r in
tensity
Relative time (s)
26 mJ/cm2
0 20 40 60 80 100120140
Fluence (mJ/cm2)
0.0 0.2 0.4 0.6 0.8 1.0 1.2Relative time (s)
135 mJ/cm2
112 mJ/cm2
90 mJ/cm2
26 mJ/cm2
67 mJ/cm2
Off
On
(a) (b)
(c) (d)
4.3 Dynamic optical switching of non-volatile multi-level memory
97
experiments. The composition close to that of Ge50Te50 films was confirmed. The
samples were loaded into the custom-made ns laser pump-probe setup, as a detail
described in Chapters 3.3.2. It turns out that, as shown in Figure 4.37a, a steep
increase of the GeTe film reflectivity as a function of the number of pulses occurs,
when ns laser irradiation at a fluence of 26 mJ/cm2 is used. To evaluate the
reflectivity changes more directly, the relative reflectivity of the films (Rf) is
defined by the following equation:
𝑅𝑓 =𝑅𝑥−𝑅𝑎
𝑅𝑐× 100% , (4.9)
where Rx, Ra and Rc are the probed reflectivity of laser irradiated, amorphous and
completely crystallized GeTe films, respectively. The structurally amorphous and
completely crystallized GeTe films are referred to the as-deposited film and the
films after 20 laser pulses, confirmed by analysis with XRD and TEM (for details
see Chapters 4.3.2 and 4.3.3). During the initial four pulses, the relative reflectivity
increases gradually from 0% to almost 80% and remains constant at this high
reflectivity level for higher pulse numbers, up to 100 pulses. In the corresponding
optical microscope images of the films as show in Figure 4.38, optical reflectivity in
the large area is obviously different after irradiation with different pulse numbers.
The image of the as-deposited film shows a relatively dark region, whereas in the
images of films after laser irradiation with an increasing number of pulses, the
regions appear increasingly bright until after four pulses. The image brightness does
not change any more with larger laser pulse numbers. This result is in good
Figure 4.38: Optical microscope images of GeTe films upon ns laser pulse irradiation illustrate the
corresponding regional changes of film reflectivity upon irradiation with different pulse numbers.
as-dep. 2nd pulse
3rd pulse 20th pulse
200 μm
Chapter 4 Results and discussion
98
agreement with the probing reflectivity change. The evolution of reflectivity
increase is further demonstrated for phase transformation from the amorphous to the
crystalline state, as confirmed by results of both XRD and TEM (details will be
given in Chapters 4.3.2 and 4.3.3, respectively). In other words, such a change in
reflectivity, from low to high, is considered to be the logical switching On process
(crystallization). The reverse trend of optical reflectivity is the switching Off
process (amorphization), as described in detail in the following.
The amorphization of GeTe films was achieved by irradiation with a sufficiently
high laser fluence and with only one single pulse, as shown in Figure 4.37b. As an
initial stage, a high reflectivity of GeTe films by full crystallization was provided,
as in the On process described above, which means as-deposited GeTe films were
laser irradiated with 20 pulses at a constant fluence of 26 mJ/cm2. Subsequently, the
reflectivity change was investigated by applying single laser pulses at different
fluences. It seems that the reflectivity remains at a high level, despite a small
deviation after single pulse irradiation at a fluence of 26 mJ/cm2. With an increase
of the laser fluence up to 67 mJ/cm2, a decrease of reflectivity by about 50% is
detected, as shown in Figure 4.37b. For a further increase of the fluence, a gradual
decrease of the reflectivity is observed. In particular, at a fluence of 112 mJ/cm2, the
relative reflectivity of the films decreased to almost zero, meaning that the
reflectivity of the amorphized film is now equal to that of an as-deposited film. The
decrease in reflectivity upon single pulse irradiation at increasing fluences is
correlated with a reduction of the degree of crystallinity, i.e., the fraction of the
amorphized material, in GeTe films. A detailed discussion of this dynamic
crystalline-to-amorphous transition can be found in Chapter 4.3.4. With a further
increase of the fluence to 135 mJ/cm2, the reflectivity exhibits a slight increase
caused by the appearance of melting and ablation of the film (not shown here).
The reflected laser intensity of the pump laser source, measured as a function of
time, is given in Figures 4.37c and d. The pumping source consisted of an UV ns
pulse laser with a pulse duration of 20 ns and a wavelength of 248 nm. A constant
intensity of delivered laser pulses for an exemplary identical fluence of 26 mJ/cm2
is demonstrated in Figure 4.37c; this is correlated to the use of the laser as a
pumping source for the irradiation of the film in Figure 4.37a (switching On
4.3 Dynamic optical switching of non-volatile multi-level memory
99
process). In contrast, a gradual increase of the measured intensity with an increase
of the laser fluence from 0 to 135 mJ/cm2 was registered, as depicted in Figure
4.37d, which corresponds to the pumping source for irradiating the films in Figure
4.37b (switching Off process).
In order to clarify the accuracy of achieving distinctive levels of reflectivity
change upon the crystallization process as a function of the laser fluence and the
number of pulses, a series of pulse number-fluence-dependent optical reflectivity
measurements of GeTe films was investigated. Figure 4.39a shows the relative
reflectivity of GeTe films after laser irradiation with fluences of 0-36 mJ/cm2 as a
function of the laser pulse number. No apparent difference in the relative optical
Figure 4.39: (a) Optical reflectivity and crystalline fraction of GeTe films after ns laser pulse
irradiation at various fluences of 0-36 mJ/cm2 as a function of the pulse number, (b) Relation
between laser pulse fluence, number of pulses and the resulting relative reflectivity change for the
initial five pulses in (a).
0 4 8 12 16 20
0
20
40
60
80
100
Re
lative
re
fle
ctivity (
%)
Number of pulses
6 mJ/cm2 11 mJ/cm
2
14 mJ/cm2
17 mJ/cm2
26 mJ/cm2
36 mJ/cm2
0.0
0.2
0.4
0.6
0.8
1.0
Cry
sta
llin
e f
ractio
n
(a)
(b)
Chapter 4 Results and discussion
100
reflectivity after laser irradiation with 0 to 20 pulses at a fluence of ≤ 6 mJ/cm2 can
be recognized. With a slightly increased fluence of 11 mJ/cm2 and with an increase
of the number of pulses, the reflectivity gradually increases to a saturation value of
about 40%, reached after eight applied laser pulses. In contrast, the reflectivity
curve exhibits a significant rise to a level close to 85% after only four pulses, when
the laser fluence was 14 mJ/cm2. In a fluence window between 14 and 36 mJ/cm
2,
the observed trend of reflectivity change did not display very marked differences,
with the saturation level varying only by around 10%. Thus, a relatively high
reflectivity can be realized by using a higher fluence for the first 4 pulses; for higher
pulse numbers, the achieved reflectivity level remains constant. A more detailed
evolution of fluence-pulse number-dependent reflectivity of GeTe films after ns
laser irradiation within the first five pulses is shown in Figure 4.39b. It can be
observed that a continuous reflectivity increase of GeTe films can be attributed to
either the increase of the pulse number or to the increase of the laser fluence,
excluding the fluences smaller than 14 mJ/cm2, which is the threshold fluence for
crystallization (see Chapter 4.3.2). For fluences above this threshold, five distinct
levels (including the as-deposited amorphous state) of relative reflectivity of the
film can be achieved by varying the pulse number between zero and four pulses. In
addition, the possible number of different levels of reflectivity is not severely
dependent on the laser fluence in the region between 14 and 36 mJ/cm2. This
illustrates that a multilevel operation with 5 distinct levels is highly stable and also
reproducible (see Chapter 4.3.5). Consequently, these investigations demonstrate
that not only binary states of either high reflectivity or low reflectivity of GeTe
films are obtainable by using an irradiation fluence at the threshold (14 mJ/cm2), but
also non-binary states (multi-level states) that are based on accumulative
crystallization processes controlled by the laser pulse number.
Structural transformation 4.3.2
After having gained insights into the optical switching of GeTe films, the structural
transformation, as a function of pulse number and fluence, was investigated by
XRD measurements. The obtained results are given in Figure 4.40. Figure 4.40a
presents the evolution of XRD patterns of films irradiated with an increasing laser
fluence between 0 and 54 mJ/cm2 after irradiation with 20 pulses (pulse duration of
4.3 Dynamic optical switching of non-volatile multi-level memory
101
20 ns and wavelength of 248 nm). The chosen fluence values equal those in the
previous chapter (4.3.1). The diffraction pattern of the as-deposited film shows the
usual features that are characteristic for the amorphous state. No distinct differences
in the diffraction spectra appear for fluences up to 11 mJ/cm2. For GeTe films
irradiated with a laser fluence larger than 14 mJ/cm², crystalline phase reflections
emerge at the angular positions 26.3°, 30.2° and 43.8°, which correspond to the
(111), (200) and (220) lattice planes of the cubic GeTe phase, respectively. This is
in good agreement with the results of the optical reflectivity measurements related
to the threshold of the significant optical contrast change. It should be noted that
although the increased optical reflectivity of films irradiated with a laser fluence of
11 mJ/cm² is presumably due to the partial crystallization of the films, XRD
Figure 4.40: XRD patterns of ns laser pulse irradiated GeTe films: (a) with 20 pulses as a function of
the laser pulse fluence between 0 and 54 mJ/cm2, (b) with a fluence of 26 mJ/cm
2 as a function of the
pulse number.
20 30 40 50
20th pulse
4th pulse
3th pulse
Inte
nsity (
arb
. units)
2 (deg)
as-dep.
2nd pulse
fcc (
220
)
Si (2
00
)
fcc (
200)
fcc (
111
)
20 30 40 50
54 mJ/cm2
36 mJ/cm2
26 mJ/cm2
17 mJ/cm2
14 mJ/cm2
11 mJ/cm2
Inte
nsity (
arb
. u
nits)
2 (deg)
as-dep.
6 mJ/cm2
fcc (
22
0)
Si (2
00
)
fcc (
20
0)
fcc (
11
1)
(a)
(b)
Chapter 4 Results and discussion
102
reflections of the crystalline phase cannot be detected yet, due to the low fraction of
the crystallized volume induced within this fluence range. Figure 4.40b presents the
evolution of XRD patterns of films irradiated with an increasing number of pulses
between 0 and 20 at a constant fluence of 26 mJ/cm2. In contrast to the pattern of
the as-deposited amorphous film, when the film was laser-irradiated with two
pulses, weak Bragg peaks located at 26.4°, 30.3° and 43.7° appeared, which
correspond to the (111), (200) and (220) lattice planes of cubic phase GeTe,
respectively. With a further increase of the number of pulses to four and higher (up
to 20), the corresponding Bragg peaks became clearly visible and reached an
intensity saturation level, depicted in Figure 4.40b. This indicates that a continuous
increase of the crystalline fraction in the films results from the increase of the
number of laser pulses, due to a longer duration of energy absorption of the film.
The XRD results, evidencing a phase transformation in the laser-irradiated GeTe
films, are consistent with the optical reflectivity measurements presented in the
previous Chapter (4.3.2). Unequivocally, the observed higher optical reflectivity of
the films originates from a higher degree of crystallization.
Morphology 4.3.3
In order to reveal the On and Off processes of optical switching related to
crystalline phase growth and melt-quench to the amorphous state of GeTe phase-
change films on the microscopic scale, the morphology of the previously described
simple memory element was analyzed by cross-sectional TEM. Figure 4.41
summarizes cross-sectional bright-field (BF) TEM images and the corresponding
calculated 2D fast Fourier transformation (FFT) images of such GeTe thin films. A
uniform and smooth as-deposited GeTe film lying between the Si substrate and the
LaAlOx capping layer is shown in Figure 4.41a (the Pt layer is only used for TEM
specimen preparation). The entire accessible as-deposited film was confirmed to be
in the amorphous state by the FFT diagram (Figure 4.41g); this is characterized by
diffuse circular intensity distribution rings. However, it is obvious that crystalline
grains formed in the amorphous GeTe matrix after ns laser irradiation with 2 pulses
at a constant fluence of 26 mJ/cm2
(Figure 4.41b). The average grain dimensions of
the oval-shaped grains are approx. 62 nm in height and 90 nm in width, based on
the TEM results. This is because, when the crystalline grain size within the GeTe
4.3 Dynamic optical switching of non-volatile multi-level memory
103
film reaches the film thickness, grain growth along the direction perpendicular to
the substrate surface will be prevented by the top and bottom interfaces of the GeTe
film. Consequently, due to these geometric restrictions, any further grain growth
can only occur in the lateral direction. Furthermore, since such restricted growth of
the crystalline regions might induce mechanical stress and involve shrinkage in film
thickness upon amorphous-crystalline transition, both forces, as a result of the
altered elastic energy, possibly lead to the inhomogeneous complex height
distribution on the film’s surface. With an increase of the pulse number from two to
Figure 4.41: Cross-sectional BF-TEM images and FFT patterns of GeTe thin films on Si: (a) as-
deposited; (b), (c), (d), (e) after ns laser pulse irradiation at an identical fluence of 26 mJ/cm2
with 2,
3, 4, 20 pulses, respectively; (f) re-amorphization of 20 pulse-crystallized film by ns laser pulse
irradiation with 112 mJ/cm2
and a single pulse; (g), (h), (i) and (j) are corresponding FFT patterns to
(a), the marked regions P1 and P2 of (b) after magnification, and to (f), respectively. All GeTe films
were coated with a thin protective LaAlOx layer.
Chapter 4 Results and discussion
104
four pulses, the crystalline precipitates in the amorphous regions gradually grow
further in the lateral direction and the width of grains extends from 90 to 200 nm,
while, consequently, the amorphous regions between the crystalline grains are
reduced in volume, as observed in Figures 4.41b, c and d. The corresponding
structure of two selected representative regions, P1 and P2, in Figure 4.41b was
confirmed by FFT patterns to be amorphous and polycrystalline, as shown in
Figures 4.41h and i, respectively. This can explain the contribution to the
multi-level reflectivity of the film that is dependent on the pulse number chosen for
the irradiation. Since a higher volume of crystallized material is obtained upon
applying higher pulse numbers, the prolonged accumulated photon absorption time
allows the crystalline precipitates to grow and merge until a continuously
polycrystalline film is formed. Almost complete crystallization of the GeTe film can
exemplarily be observed after 20 pulses of irradiation, as shown in Figure 4.41e.
Moreover, complete amorphization of such a fully crystallized film is
achievable by ns laser irradiation with one single pulse at a fluence of 112 mJ/cm2.
This is confirmed by the corresponding results of cross-sectional TEM in Figures
4.41f and j. A homogeneous and smooth GeTe film without crystalline regions
within the film is visible that closely resembles the original as-deposited film. The
corresponding FFT pattern (Figure 4.41j) is typical for amorphous phase material,
indicating that the crystallized GeTe film was melt-quenched to a disordered
amorphous state upon the laser irradiation. These findings agree well with the
results from the optical reflectivity measurements (presented in Chapter 4.3.5).
However, it should be noted that there are differences between the as-deposited
amorphous and the melt-quenched amorphous films regarding the optical
reflectivity change upon repetitive crystallization and amorphization (see Figure
4.45). The resulting FFT images in Figures 4.41g and j are similar in that they
illustrate the amorphous nature of the thin films. It is, however, to be expected that
as-deposited and melt-quenched amorphous structures possess slightly different
short-range order that cannot be detected by FFT due to differences in TEM
specimen thickness.
To gain more information about the local structure of as-deposited, re-
amorphized and crystallized GeTe thin films, atomic resolution cross-sectional
4.3 Dynamic optical switching of non-volatile multi-level memory
105
TEM images of the GeTe films were recorded. The representative crystallized GeTe
film is the as-deposited film after ns laser irradiation with 20 pulses at 26 mJ/cm2
and the re-amorphized film is the crystallized film after subsequent ns laser
irradiation with single pulses at 112 mJ/cm2. The result is shown in Figure 4.42.
Compared to the as-deposited GeTe film exhibiting the presence of an
amorphous structure with a lack of long-range order, as shown in Figure 4.42a, the
HRTEM image in Figure 4.42b of the disordered structure reveals no
nanocrystalline components after melt-quenching. Note that the structural features
of melt-quenched and as-deposited amorphous are very similar and distinguishing
between them is difficult [163, 164]. Some reports [134, 165-168] have claimed the
existence of a nanocrystalline phase, and demonstrated a higher medium-range
order in the melt-quenched amorphous state, compared to the as-deposited
amorphous state in other chalcogenide materials, such as AgInSbTe or GST. Other
authors also hypothesized [169] that the ordering of Ge and Te in melt-quenched
GeTe is much higher than in as-deposited. Nevertheless, compared to a
representative crystallized GeTe film, a long-range order with a lattice fringe
Figure 4.42: Cross-sectional HRTEM images of: (a) as-deposited GeTe film, (b) re-amorphization of
the crystallized GeTe film after ns laser irradiation with one pulse at a fluence of 112 mJ/cm2, (c) a
representative GeTe film after ns laser irradiation with 20 pulses at a fluence of 26 mJ/cm2. The
corresponding FFT pattern is shown in the inset, (d) The electron diffraction patterns of GeTe films
after ns laser irradiation with 20 pulses at a fluence of 26 mJ/cm2.
Chapter 4 Results and discussion
106
spacing of ~0.34 nm along the [110] viewing direction can be observed in Figure
4.42c. This indicates that the crystalline phase of the GeTe film induced by ns pulse
laser irradiation possesses a high crystalline quality. A corresponding Fourier
transformed pattern obtained from this HRTEM intensity region of the GeTe thin
film is shown in the inset of Figure 4.42c and a measured electron diffraction
pattern is additionally shown in Figure 4.42d, which confirms this finding. In the
electron diffraction pattern, a local lattice spacing is identified, which is
characteristic for a typical crystalline phase of GeTe.
Dynamic phase transformation 4.3.4
In order to determine exactly how fast the phase transformations of the PLD-
deposited GeTe films take place, the dynamics of crystallization and amorphization
of as-deposited GeTe thin films driven by ns laser pulses was investigated. Figure
4.43 shows the reflectivity evaluation of a GeTe film upon laser irradiation with
each ns pulse (first to fifth pulses are shown) being at a constant fluence of 26
mJ/cm2 on a fixed irradiation region of the size of about 24 × 6 mm
2. Without the
pumping laser irradiation (zero pulses), probing the reflectivity change from an
as-deposited, amorphous GeTe film is referred to as Baseline in Figure 4.43a and it
appears as a relatively smooth line within the observed timescale range
between -200 and 200 ns. In contrast, a distinguishable level of reflectivity increase
from pre-existing to post-existing states upon each successive single pulse
irradiation for five pulses can be measured. A final total relative reflectivity of
~85% can be achieved after four pulses and an unstable increase of the relative
reflectivity with a variation of 5% after the 5th pulse to a level of more than 90% is
observed. This indicates that the crystallization process was almost completed
within the first four (or five) pulses. With a further increase of the number of pulses
to more than five, the relative reflectivity change becomes negligible, due to the
crystallization being accomplished throughout the whole film (see Figure 4.39a).
For all distinct pulse-dependent reflectivity changes in Figure 4.43, the transition
time from the lower to the higher reflectivity level upon laser irradiation lies within
an interval of ~60 ns, which is around three times longer than the pulse duration
(~20 ns) of the applied laser pulse. Possibly, this transition time is the time required
for atoms to reorganize into an adequate ordered arrangement, assuming that the
4.3 Dynamic optical switching of non-volatile multi-level memory
107
increased reflectivity is representative of the crystallization process in the form of a
crystalline fraction increase of the films. The time for complete crystallization of a
GeTe film after ns laser irradiation with five pulses is ~300 ns after the complete
transition process. Moreover, it is interesting to note that the gradients of the
transitions from lower to higher reflectivity levels decreased with an increase in the
number of pulses (first to fifth pulse), which implies that the velocity of
crystallization decreased with each increase of the crystalline fraction. Assuming
that the crystalline fraction is proportional to the measured reflectivity of the film, a
simple representation of the correlation between growth velocity and the number of
irradiation pulses can be extracted from the results of Figure 4.43a, as shown in
Figure 4.44. These results might be helpful for the further understanding of the fast
Figure 4.43: (a) Optical reflectivity of GeTe films after laser pulse irradiation with different pulse
numbers from 0 to 5 pulses and at a constant fluence of 26 mJ/cm2 as a function of time, (b) Optical
reflectivity of GeTe films after single laser pulse irradiation with a fluence of 112 mJ/cm2
as a
function of time. The inset displays a magnification of the curve in the time-scale range from -7 to
12 ns.
-200 -100 0 100 200
0
20
40
60
80
100 Amorphization
Re
lative
re
fle
ctivity (
%)
Time (ns)
-5 0 5 10
Time (ns)
3 ns
(a)
(b)-200 -100 0 100 200
0
20
40
60
80
100
Re
lative
re
fle
ctivity (
%)
Time (ns)
5th pulse
4th pulse
3d pulse
2nd pulse
1st pulse Baseline
Crystallization
Chapter 4 Results and discussion
108
phase transition mechanism and for the design of multi-level phase-change memory
elements.
The reverse process, i.e., the amorphization process (switching Off state) of a
GeTe film, characterized by a change of the optical reflectivity from the highest to
the lowest level, can be achieved by irradiation with only one single laser pulse.
Figure 4.43b shows the reflectivity change of a GeTe film with such a single laser
pulse irradiation at a fluence of 112 mJ/cm2
as a function of the time. The pre-
existing high reflectivity level of ~100% was provided by the crystallization of an
as-deposited GeTe film using ns laser irradiation with 20 pulses at a fluence of
26 mJ/cm2 (switching On process described above). When the film is irradiated
with a single laser pulse at a sufficient fluence, the film reflectivity drops sharply to
a much lower level of ~40% after about 3 ns (as seen in the inset with magnified
timescale) and then gradually decreases further to ~20% within a delay time of
~200 ns. It should be noted that the experimental setup of the pump-probe system
(as described in Chapter 3.3.2) makes it possible to provide a time resolution as fast
as about 2 ns. This would mean that the processes are maybe even faster than 3 ns
but cannot be resolved. The result of a dramatically decreased reflectivity is
assumed to substantially reduce atomic mobility caused by melt-quenching of the
crystallized GeTe film back to the stable amorphous phase, after application of an
intensive single laser pulse. Furthermore, a structural transformation was
demonstrated by the analysis of cross-sectional TEM images of such amorphized
GeTe films and this is also confirmed by the results related to the retention of
Figure 4.44: Crystal growth velocity of GeTe film after ns pulse laser irradiation with a fluence of
26 mJ/cm2 as a function of the pulse number.
0 1 2 3 4 50
50
100
150
2000 60 120 180 240 300
Crystallization time (ns)
Number of pulses
Gro
wth
velo
city (
m/s
)
0.0
0.2
0.4
0.6
0.8
1.0
Rela
tive r
eflectivity (
%)
4.3 Dynamic optical switching of non-volatile multi-level memory
109
reversible switching (Chapters 4.3.3 and 4.3.5, respectively).
Retention 4.3.5
The retention of the actual present state upon reversible phase transformation
between amorphous and crystalline states in phase-change films is an eminently
important parameter for data storage applications based on phase-change materials.
Repetitive optical On and Off switching processes of GeTe films with a various
number of cycles driven by ns laser pulses was investigated and is shown in Figure
4.45a. During the first cycling, crystallization (On state, high reflectivity) of an as-
deposited amorphous GeTe film was obtained at a moderate fluence of 26 mJ/cm2
with multi-pulses, i.e., a short series of five successive ns pulses; subsequently,
amorphization (Off state, low reflectivity) of the film was triggered by a single
pulse at a higher fluence (112 mJ/cm2), as discussed above. It is interesting that the
Figure 4.45: (a) Optical reflectivity of a GeTe film as a function of the number of laser irradiation
cycles (forward: with a certain number of pulses at a constant laser fluence of 26 mJ/cm2 and
reverse: a single laser pulse at a laser fluence of 112 mJ/cm2) as function of the pulse number. (b)
Representation of the optical reflectivity switching On and Off processes with various laser pulse
numbers as a function of the number of cycles.
0 5 10 15 20
0
20
40
60
80
100
Off
1,2,3
,4,5
Re
lative
re
fle
ctivity (
%)
Number of pulses
1st cycle
2nd cycle
3d cycle
4th cycle
5th cycle
On
1
On 2
,3,4
,5
0.0
0.2
0.4
0.6
0.8
1.0
Cry
sta
llin
e f
ractio
n
1 2 3 4 5
0
20
40
60
80
100
3d pulse
2nd pulse
1st pulse
0th pulse
Rela
tive r
eflectivity (
%)
Number of cycles
No. of pulses
20th pulse
5th pulse
4th pulse
0.0
0.2
0.4
0.6
0.8
1.0
Cry
sta
lline fra
ction
(a)
(b)
Chapter 4 Results and discussion
110
recrystallization (switching On) of a melt-quenched amorphous GeTe film requires
a smaller number of pulses (three pulses), compared to as-deposited ones (five
pulses) at the same fluence of 26 mJ/cm2. This indicates that the total amount of
energy (or crystallization time) required for the re-crystallization of a melt-
quenched film is smaller than that required to crystallize an as-deposited,
amorphous film for the first time. Thus, a difference between both amorphous
structures and the formation mechanism of crystalline nuclei can be expected.
According to previous reports [60, 170, 171], nanocrystalline precipitates were
detected in melt-quenched amorphous Ge-Sb-Te films, which should accelerate the
crystallization process. For instance, the crystallization from a melt-quenched
amorphous film was found to be an order of magnitude faster than that from an
as-deposited amorphous film [60]. Similarly, it was found that the crystallization
times for Ge-Te melt-quenched amorphous material is shorter than of as-deposited
amorphous material [28, 140]. The dependence of the crystallization kinetics of
amorphous GeTe films on their thermal history was also confirmed [169]. It has
been hypothesized that the existence of crystalline seeds is responsible for the
increase in crystallization time. An extended discussion on the local structure of as-
deposited amorphous and melt-quenched amorphous is provided in Chapter 4.3.3.
The switching behavior of On and Off processes was found to be reproducible
and stable during the investigated cycles. The intermediate levels of reflectivity
were repeatedly achieved by laser irradiation with a varied number of pulses, as
shown in Figure 4.45b. The 1st pulse of the 1st cycle did not provoke an obvious
reflectivity change, after the 2nd pulse an almost middle level of reflectivity was
reached, an then, after the 3rd trigger pulse, the reflectivity of the film increased
close to the highest (saturated) level. Compared to the 1st cycle, with the 2nd to 5th
cycle the respective 1st pulse already altered the reflectivity of the melt-quenched
amorphous GeTe film to the medium level and the following 2nd pulses were
sufficient to achieve the final high reflectivity level. Apparently, repeatedly
specified intermediate reflectivity levels are achievable in phase-change films
driven by the numbers of laser pulses, and the identification of these levels is
meaningful for multilevel storage applications.
4.3 Dynamic optical switching of non-volatile multi-level memory
111
Optical switching model 4.3.6
The principle of a non-volatile memory cell with multilevel optical switching of
GeTe films using ns laser pulse irradiation is illustrated in Figure 4.46. The unique
property of reversible switching between the amorphous and crystalline states upon
specific heating/cooling by applying appropriate laser pulses is the basis of the
functionality of such a memory cell. Precisely, crystallization can be considered as
accumulative switching (On states). Each laser pulse excitation can induce a partial
crystallization, and complete crystallization can be achieved by a succession of such
Figure 4.46: (a) A schematic of the multi-level optical switching of a GeTe film driven by ns laser
pulses for non-volatile memory application. (b) Corresponding multi-states evaluation of reversible
amorphous-to-crystalline phase transitions of a phase-change film using ns laser pulses. The a-GeTe
and c-GeTe represents amorphous and crystalline GeTe, respectively.
Phase-change cell
Input pulses
Cell switch On
States
Cell switch Off
Tm
Tc
t
Tm
Tc
t
I
II
III
IV
t
States
t
Input pulse
V
V
I
(a)
a-GeTeLaAlOx c-GeTe Si
States I II III IV V
Top-view
Cross-section
PCM
(b)
Chapter 4 Results and discussion
112
excitations. As depicted in Figure 4.46a, a moderately intense laser beam (fluence
of 26 mJ/cm2) and distinct sequences of pulse numbers (pulse duration of 20 ns for
each pulse) is used for crystallization of the phase-change memory cell. The
supplied laser pulse energy heats the material above the crystallization temperature
(but below the melting temperature) for a sufficiently long time, resulting in
transformation from the amorphous to the crystalline state. The resulting
crystallization rate of the growth-dominated GeTe film is dependent on the
irradiation time, which corresponds to the number of applied laser pulses. Thus, a
higher degree of crystallization, i.e., a higher volume fraction of crystallized
material, can be achieved by increasing the pulse number at a constant fluence,
providing an important prerequisite for realizing multi-level phase-change memory
cells.
The discrimination of these levels can be realized by reading out the optical
reflectivity of the memory cell, as shown in the example with five different
levels/states. The corresponding schematic diagram for the crystallization process of
such a phase-change GeTe film is shown in Figure 4.46b. Initially, stage I shows a
completely amorphous GeTe film. With each succeeding laser pulse excitation,
small nuclei are formed and then the crystalline regions gradually grow until the
film is completely polycrystalline (from stages II to V), induced by using ns laser
pulse irradiation with an increase of the number of pulses. The high optical
reflectivity contrast between amorphous and crystalline states is used for tuning the
distinctly intermediate levels in reflectivity as a result of the changed ratio of the
crystalline volume fraction to the amorphous fraction of the film. It should be
pointed out that the time response (~60 ns) of the phase-transition in GeTe upon
laser pulse excitation was found to be longer than the duration of the applied pulse
(~20 ns) (see Chapter 4.3.4). Further, the optical switching can be regarded as
wavelength independent from the near ultraviolet to the near infrared (400 and
700 nm) range, as confirmed in the present study (see Chapter 4.3.8).
In the reverse process, i.e., the reverse transformation from the crystalline to the
amorphous state, an intense (fluence of 112 mJ/cm2) single laser pulse is used for
amorphization of the phase-change cell, as shown in Figure 4.46a. The pulse energy
induces melting of the material, followed by rapid quenching to freeze the atoms in
4.3 Dynamic optical switching of non-volatile multi-level memory
113
the disordered structure. The significantly different reflectivity between the
crystalline state (high reflectivity) and amorphous state (low reflectivity) can be
easily detected. Such a sharp reflectivity drop from high to low upon amorphization
of GeTe film was determined to take 3 ns (see Chapter 4.3.4). The corresponding
evaluation of schematic state changes in the case of amorphization is also shown in
Figure 4.46b. The reversal of the state V (crystallization) back to state I (amorphous
state) is depicted. In summary, the possibility to reversibly switch a GeTe film for
practical non-volatile memory application with multi-level functionality is
provided.
Film thickness dependence of numbers of multi-levels 4.3.7
The aforementioned five distinct optical reflectivity levels of the memory cell were
obtained with a GeTe film thickness of 68 nm. By sacrificing the high signal-to-
noise ratio, an increase of the number of intermediate levels is possible through
slightly reducing the laser fluence to a subthreshold value (~11 mJ/cm2), as shown
in Figure 4.39. An alternative approach to increase the number of possible levels in
a memory cell could be to decrease the film thickness. The example in Figure 4.47
shows the optical reflectivity of an only 50 nm thick GeTe film upon ns laser pulse
irradiation at the identical fluence of 26 mJ/cm2 as a function of the pulse number.
As a fully crystallized film can be obtained only after 12 applied pulses, each pulse-
induced crystallization of GeTe film leads to up to ten different reflectivity levels
Figure 4.47: Optical reflectivity and possible indicated multi states of a 50 nm thick GeTe film as a
function of the laser pulse number.
0 5 10 15 20 80 100
0
20
40
60
80
100
Re
lative
re
fle
ctivity (
%)
Number of pulses
X
IX
VIII
VII
VI
V
IV
III
I
II
Chapter 4 Results and discussion
114
(as indicated in Figure 4.47) within a single memory cell, which is much higher in
comparison to those of the thicker film. For a smaller thickness of phase-change
films, more possible intermediate levels of reflectivity can be expected. On the
other hand, it should be noted that the levels must be accurately identifiable so that
they can be used for reading out stored data. A possible explanation for the effect of
film thickness on the varying number of intermediate levels is the fact that
nucleation and growth of crystalline regions in GeTe films is strongly influenced by
the film thickness, as already discussed in Chapter 4.2.1.
Broad wavelength response of reflectivity 4.3.8
As described in the sub-chapters above, the reflectivity readout of the laser
irradiated GeTe film was realized with a probing diode laser at a wavelength of
405 nm in the setup of the pump-probe system. To examine the multilevel
reflectivity of phase-change films over a wide wavelength range, a series of pulse
number-dependent optical reflectivity measurements of GeTe films in the
wavelength range from the near ultraviolet to the near infrared (400 and 700 nm)
was investigated. Each of the conditions of the ns laser-excited GeTe films was
prepared and the broad wavelength response of reflectivity was performed by UV-
Vis spectroscopy measurements. Figure 4.48 shows the optical reflectivity of an as-
deposited GeTe film and of films after 20 ns laser pulse irradiation with pulse
numbers up to 20 at a constant fluence of 26 mJ/cm2. The optical reflectivity of the
as-deposited film varies from 66 to 70% in the wavelength range between 400 and
Figure 4.48: Optical reflectivity of GeTe films as a function of the pulse number within the
wavelength range from 400 to 700 nm
400 500 600 700
65
70
75
80
85
90
20 pulses
4 pulses
3 pulses
2 pulses
Re
fle
ctivity (
%)
Wavelength (nm)
as-dep.
4.3 Dynamic optical switching of non-volatile multi-level memory
115
700 nm. After two laser pulses are applied, the reflectivity of the film within the
measured wavelength range is slightly increased, by around 3%. A further
enhancement by 6% appears after three pulses. A more distinct reflectivity of the
film after irradiation with four pulses is in the range from 80 to 82% (reflectivity
contrast ~14% at 650 nm). Upon further increase of the number of applied laser
pulses to, for example, 20 pulses, the reflectivity increases only slightly further by
~3%, but the effect is no longer so pronounced. Moreover, the changed optical
reflectivity of GeTe films after ns laser excited pulses is almost independent of the
wavelength, except for the maximum variation of around 4% increase from the
wavelength of 400 nm to 650 nm in the case of the films after 20 pulses irradiation.
The highest optical reflectivity contrast of 20% at a wavelength of 600 nm is
observed between as-deposited and films after 20 pulses. Therefore, in general, the
broad wavelength response (400-700 nm) of multi-states reflectivity increase in the
GeTe film is possible for the controllable crystallization process in the GeTe film
with induction by the appropriate number of applied laser pulses.
Summary of the results on dynamic optical switching of non-volatile 4.3.9
multi-level memory
Rapid and reversible amorphous-crystalline phase transformations in thin single-
layered PLD-deposited phase-change GeTe films were studied. Based on the
construction of a UV ns laser pump-probe system, non-volatile switching in GeTe
film based cells was determined by distinct changes of the optical reflectivity on the
time scale of nanoseconds. Reversible optical switching was demonstrated; the
results included a switching threshold fluence of ~14 mJ/cm2 and an energy
accumulation as function of the applied number of pulses, leading to switching On
parameters of 26 mJ/cm2
with 20 pulses and switching Off parameters of
112 mJ/cm2 with a single pulse. Thus, by appropriately applying laser fluence and
pulse number, the crystallization and amorphization processes in GST films could
be feasibly controlled, which is important for realizing multi-level phase-change
memory cells. Next, the structural transformations between amorphous and
crystalline states, as well as the local cross-sectional microstructure changes
correlated with the optical switching process, were investigated, which provided an
equally important understanding of the fast kinetics of the phase transition
Chapter 4 Results and discussion
116
mechanisms that were present. Upon the detailed examination of the dynamics of
phase transformation, the speed of switching On process (crystallization) was
determined to be ~300 ns and that of the switching Off process (amorphization) to
be ~3 ns. The dependency of the velocity of crystallization was found to be
inversely proportional to the number of irradiation pulses (≤ 5 pulses). Moreover,
the crystallization of a melt-quenched GeTe film required less energy than that of
an as-deposited amorphous film. On the basis of the confirmed experimental results,
a general optical switching model was established, which was helpful for
understanding the mechanism of multilevel operation. Increasing numbers of multi-
levels (up to 10 distinct levels) could be produced by choosing a GeTe film
thickness of 50 nm. The multi-level optical reflectivity was realizable at various
wavelengths between 400 and 700 nm and the highest contrast of ~20% was
obtained at a wavelength of 600 nm. Unequivocally, these experimental results
illustrate that dynamic optical switching of non-volatile multi-level memory, based
on PCMs, is not only promising for the next generation of optical data storage
memory applications, but also for improving the understanding of the physical
aspects of these unique material properties.
4.4 Nanoscale bipolar electrical switching of phase-change materials
117
Nanoscale bipolar electrical switching of phase-change 4.4
materials
The laser pulse-induced structure transition from the amorphous to the crystalline
state in phase-change materials for the optical switching process was illustrated in
the previous chapters (4.2 and 4.3). In this chapter, the phase transformation
between the amorphous and the crystalline state of the chalcogenide materials
induced by electrical pulses is demonstrated. This transformation process is the
basis of phase-change random access memory (PCRAM) devices. The
investigations on electrical switching of phase-change materials at the nanoscale
comprise, in particular, polarity-dependent resistance switching, the threshold of the
voltage for resistance switching, as well as the underlying mechanisms.
Background 4.4.1
Currently, floating-gate FLASH memory devices dominate the non-volatile
memory market, due to their low cost. However, significant challenges emerge for
the scaling of these devices, when the characteristic dimensions are to be reduced
towards the 32 nm size and below. Alternative candidates for the conventional, non-
volatile memory technology utilize the magneto-resistance (MRAM) [148],
ferroelectricity (FRAM) [172], phase-change random access memory (PCRAM) [4]
and resistive random-access memory (ReRAM) [173] technologies. Among these
memory technologies, PCRAM and ReRAM are becoming increasingly attractive
for the next generation of non-volatile memory.
PCRAM is based on dramatically different electrical resistivity states (up to five
orders of magnitude) between the amorphous and the crystalline phases of
chalcogenides materials. This large difference makes phase-change alloys ideal
candidates for solid-state memory devices. The renewed interest in PCRAM
technology was triggered by the discovery of very fast crystallizing materials such
as GST or nitrogen and/or oxygen doped GST. These materials can crystallize in
less than 100 ns. In PCRAM, the amorphous phase-change material is crystallized
by heating it above its crystallization temperature (the so-called SET operation) and
it is melt-quenched to re-obtain the amorphous phase (the so-called RESET
operation). These operations are controlled by application of an electrical current.
Chapter 4 Results and discussion
118
High power pulses for the RESET operation place the memory into the high
resistivity state, whilst pulses of moderate power but longer duration return the cell
to the low resistivity SET state. The remarkable, advantageous properties of
PCRAM processing include: non-volatility, long cycling life, relatively low power
consumption, fast read/write speed (ps range), and prominent resistivity contrast.
Without the threshold voltage for resistance switching, a PCRAM cannot be
used because, in the high resistivity state, extremely high voltages would be
required to deliver sufficient Joule heating power to the cell to heat it above the
crystallization temperature [68, 69]. This represents a major difference between
PCRAM and ReRAM. However, both non-volatile memory technologies – PCRAM
and ReRAM – share some similarities in function and performance (including non-
volatility, scalability and switching speed) that potentially overcome the limitations
of FLASH memory, as previously discussed in the comparison of emerging
memory technologies in Chapter 2.3.2.
ReRAM is based on an electrical resistance switching process caused by the
formation and rupture of a local conduction path in a solid electrolyte, the so-called
conductive filament, which can be formed into a conductor generated by the
application of a sufficiently high voltage. When the conductive filament is
generated, it is possible to switch towards a high resistivity state by breaking the
filament through application of an opposite voltage (RESET) and back to the lower
resistivity state through reconstruction of the filament by the original voltage (SET;
for more details see Ref. [174]). Because the respective high or low resistivity state
remains unchanged until a reverse polarity voltage is applied, this type of memory
has a non-volatile character. The write time, at approximately 10 ns, is
fundamentally shorter than those needed to change the charge state of a floating
gate in FLASH memory. Such behavior has been found in various material systems:
binary transition metal-oxides, perovskites, solid-state electrolytes and, crucially,
also in phase-change chalcogenides such as GST and AgInSbTe [72, 174-178]. The
promising properties of ReRAM are the simplicity of the structure, low switching
voltage, fast switching speeds, excellent scalability and high compatibility with the
CMOS technology.
4.4 Nanoscale bipolar electrical switching of phase-change materials
119
With the development of PCRAM and ReRAM technologies, several key
challenges arise. For PCRAM, high density integration is of utmost importance and
it requires a small memory cell size. However, in practice scaling down is mainly
limited by the process of lithography technology during the fabrication of PCRAM.
In addition, a high RESET current is required in order to melt/quench the
crystallized PCM and bring it back to the amorphous state. Another critical problem
is the possible mechanical failure of the interface between the PCM and the heating
contact after a certain numbers of cycles that can be easily induced by the repeated
melting/quenching and recrystallization processes. For ReRAM, once they are
formed, the nature of such filaments fluctuates; this was described in the literature
on the basis of the filamentary theory [174], and leads to non-uniformities of the
high/low resistance switching states. Furthermore, the on/off resistance contrast is
relatively low, which is unfavorable for the signal-to-noise ratio in an electrical
resistance switching memory application. A concise comparison of PCRAM and
ReRAM regarding their advantages and disadvantages is presented in Table 4.5.
In order to eliminate these issues and optimize the cell performance, it is
proposed here that the concepts of the PCRAM and ReRAM could be combined
(we call such a combination “PC/Re-RAM”). As a prospective view, a unified
approach for studying correlations between phase-change and electrolytic behavior
Table 4.5: Advantages and disadvantages of PCRAM and ReRAM.
Performance PCRAM ReRAM
Advantages multi-bit cell
high on/off ratio
long term retention
simple structure
low operation voltage
Disadvantages high RESET current
cycling endurance
stable retention
low on/off ratio
Figure 4.49: Schematic of the sample structure used in C-AFM experiments.
Conductive tip
GST
Cr bottom electrodeSiO2/Si
V
Chapter 4 Results and discussion
120
may prove crucial for improving both device performance and reliability [179]. In a
combined PC/Re-RAM device, the chalcogenide-based amorphous GST layer is
introduced as a solid electrolyte between metal contacts, which is similar to the
typical ReRAM device of the metal/insulator/metal type. An exemplary
experimental setup using a Pt-Ir coated AFM tip as top electrode and a Cr bottom
electrode is schematically shown in Figure 4.49. Based on the C-AFM experimental
setup, a sharp tip contact on the phase-change film is not only beneficial for the
switching operation with a demanded low power consumption, due to the small
contact area and corresponding to an increase in heating efficiency, but it also
models the device switching in a nanoscale investigation, which may help to
develop a cost-effective high-density memory. During the sample preparation
process, any additional thermal treatment of the samples was avoided. The detailed
configuration of the C-AFM setup, its calibration and operation can be found in the
experimental section (see Chapter 3.4).
For clarification of the following text, low resistance state (LRS) means
crystallization of GST with atomic migration of ions forming conductive bridges,
and high resistance state (HRS) is either the initially amorphous state or the state
with nanocrystals existing in the GST film but without any conductive bridge. The
switching mechanism is depicted schematically in Figure 4.50, and the typical I-V
behavior for the combination of phase-change material as the active layer in a
memory device is shown in Figure 4.51. Figures 4.50 and 4.51 (A) represent the
pristine state. The memory cell features a HRS, which is typical for an amorphous
GST layer (as confirmed by TEM shown in Figure 4.61a) between the top electrode
(PtIr) and the bottom electrode (Cr). During the initial electroforming process (in
Figure 4.50 (B) and A B C in the I-V path in Figure 4.51), an applied electric
pulse (irrespective of its polarity) with a voltage around the subthreshold voltage
between electrodes acts as Joule heating source. This electrical pulse induces local
crystallization in the amorphous GST film. In the ideal case, the result of this
crystallization process is a crystalline low-resistivity GST filament produced in the
initial high-resistivity amorphous phase matrix. The switching process is based on
the difference between the high- and low-resistivity states. Simultaneously, an
electric pulse induced by an electrical field can lead to electrochemical reactions
4.4 Nanoscale bipolar electrical switching of phase-change materials
121
inside the chalcogenide material, which result in a local ionic conduction, involving
Te or Ge ions. In GST, Te atoms act as anions, whereas Ge atoms act as cations
[180]. Thus, under the influence of the electrical field, migration of these ions is
along opposite directions. The mobility of the ions in the filaments of locally
crystallized material inside the GST film depends on the electrical field distribution
present between the electrodes. Only if the electric field strength is sufficiently
large, both the phase-change and the ionic migration mechanism can contribute to
form such electrically conductive filamentary pathways between the electrodes. It is
known that the crystallization in GST is nucleation-dominated in the case of formed
nanocrystallites. Such nanocrystalline GST exhibits a conductivity that is several
Figure 4.50: Schematic representation of the switching mechanism in PC/Re-RAM: (A) Pristine,
(B) Filament forming / SET state process. The red feature indicates local Joule heating. (C) ON
state (LRS), (D) RESET process, (E) OFF state (HRS).
Figure 4.51: Schematic representation of the current-voltage (I-V) curve for the combination of
phase-change material as an active layer of a bipolar resistive switching memory cell. The states
from (A) to (E) correspond to the schematic switching mechanism depicted in Figure 4.50.
A CB D E
Metal (e.g. Cr, Pt, Au, etc.)c-GSTTe- a-GST
OFF stateON state RESET processSET processPristine state
Curr
ent [a
.u.]
0 Voltage [a.u.]
Low state
high state
RESET
SET
A
C
B
D
E
Chapter 4 Results and discussion
122
orders of magnitude higher than the surrounding GST amorphous matrix [181, 182].
The movement of ions in the electrical field therefore follows the most favorable
diffusion path, such as the GST nanograin boundaries or/and the nanograins and
electrodes [183, 184], which might act as an electron conductive bridge, causing a
LRS of the cell, depicted as stage C (ON state) in Figure 4.50. As confirmed by
TEM and EDX analysis of PLD-deposited GST films (see section 3 in Chapter
4.4.4), the presence of generated GST nanocrystallites is evident, as well as a Te-
rich and Ge-depleted environment after electrical switching, when compared to
stoichiometric GST (225). However, if the voltage applied to the memory cell is not
high enough to form nanocrystalline GST grains, and also not sufficient to induce
migration of ions, the cell remains in the HRS.
When the polarity of the external electric field is reversed during the RESET
operation (Figures 4.50 and 4.51 D), the pre-existing electrical conductive filament
is disrupted, because of the ion repulsion from the GST nanocrystal grains
boundaries that is caused by electrostatic force. As a result, the electrical resistance
turns to the HRS (Figures 4.50 and 4.51 E). By application of a positive voltage, the
Te or Ge ions can be re-positioned between GST nanocrystallites and, thus, can
form a conductive filament again (see Figures 4.50 and 4.51 C). In conclusion, it
should be noted that a low bias voltage for the switching process is connected with
(1) the existence of small crystalline regions (grains) after the electroforming and
(2) the local movement of Te or Ge ions as a condition for the
construction/disruption of the conductive filaments between crystalline GST grains
and/or electrodes. This is the special quality of PC/Re-RAM memory cells, because
they require neither re-crystallization/re-amorphization (as is the principle of
PCRAM) nor purely an ion-built/dissolved filament (as is the principle of ReRAM)
for accomplishing the switching process. Apart from the physical switching
mechanism, possibly attributed to the combination of the phase-change of the GST
layer and the ionic migration of elements (e.g., Te, Ge) within the GST layer, as
described above, an effect of the redox reaction in the interface layer (e.g.,
formation of a thin GSTO layer associated with a filamentary conducting channel)
[177, 185], together with a phase-change in the GST layer between the electrodes,
4.4 Nanoscale bipolar electrical switching of phase-change materials
123
could also explain the bipolar memory device switching. Detailed experiments to
elucidate the latter case are in progress.
Nevertheless, a comparison between the separated PCRAM or ReRAM and the
combined PC/Re-RAM device shows that the PC/Re-RAM provides the following
advantages:
(i) The RESET and/or operation voltage is smaller, because the resistivity
switching is dependent on the polarity and the heat loss is minimized by
the locally formed nano-contact between PCM and tip electrode, rather
than on the heat required for re-amorphization by local melt-quenching.
(ii) The On/Off resistance ratio is higher because of the confined current
flow with a higher conductivity of the generated GST nanocrystals,
reducing the non-uniformity of filament formation (owing to the
dependence of LRS on the filament) in ReRAM.
(iii) Excellent cycling endurance, since a smaller RESET/SET voltage is
required for operation, which can reduce the commonly encountered
problems in PCRAM, i.e. the largely thermal effects causing mechanical
device failure. In addition, the principle of ions migrating between
locally crystallized regions can avoid the crucial problem of randomly
forming filaments in ReRAM.
Polarity-dependent resistance switching of phase-change materials 4.4.2
The development of a concept for the combination of PCRAM and ReRAM is the
most important consequence of the presentation in the previous Chapter, in which
thin films of phase-change material were introduced as the active layer for bipolar
electrical switching memories. If this combination is realized in a PCM-based Re-
RAM (PC/Re-RAM), it can lead to applications in the fields of next generation non-
volatile memory, as well as cognitive computing, reconfigurable logic circuits, and
more.
Features of a PC/Re-RAM memory, such as excellent scalability (< 15 nm), low
SET/RESET currents (several nA), long retention (> 180 days), excellent cycling
endurance (> 1000 cycles) and high On/Off resistance ratios (about four orders of
magnitude), as well as a high estimated data density (0.7 Tbit/inch2), should be
ultimately attainable. Based on conductive AFM as an active switching tool, it is
Chapter 4 Results and discussion
124
possible to examine and to quantify the switching process on the nanoscale in a
laboratory environment. A schematic of the prototype cell (GST/Cr/SiO2/Si) and the
configuration based on C-AFM was already shown in Figure 4.49 and discussed in
detail in the experimental section (see Chapter 3.4). As an example, a measured I-V
characteristic of GST films is shown in Figure 4.52. The bias voltage is applied
between the bottom Cr electrode and the top electrode consisting of a PtIr coated
AFM tip, where the top electrode is connected to the ground and an amperemeter is
attached between the latter two, in order to measure the electric current flowing
through the GST film. By sweeping the voltage from -3 V→ 0 → +3 V → 0 → -3
V at a sweeping rate of 1.8 V/s (as demonstrated in the top left inset of Figure 4.52),
the I-V curve of the PtIr/GST/Cr stack shows typical bipolar switching with a
pinched hysteresis loop. There are two segments that correlate with the two
distinguishable resistance levels of a low ohmic resistance state (LRS) and a high
ohmic resistance state (HRS), characterized by the different slopes of the curves.
Figure 4.52: Measured I-V characteristic of GST/Cr/SiO2/Si arrangement with PtIr coated C-AFM
tip (diameter = 20 nm) as top electrode. The inset in the top left of the figure shows the
corresponding applied sweeping voltage as a function of time (at a sweep rate of 1.8 V/s) and the
inset in the bottom right of the figure represents the corresponding I-V curve with respect to the
absolute norm of the current at a logarithmic scale. The arrows indicate the directions of the voltage
sweep.
-3 -2 -1 0 1 2 3
1E-12
1E-11
1E-10
1E-09
1E-08
|Cu
rre
nt| (
A)
Voltage (V)
0 1 2 3 4 5 6 7-3
-2
-1
0
1
2
3
Vo
lta
ge (
V)
Time (s)
swee
p ra
te 1
.8 V
/s
-3 -2 -1 0 1 2 3-15
-10
-5
0
5
10
RESETCurr
ent (n
A)
Voltage (V)
HRS
current limit
Vth
current limit
LRS
Vh
SET
4.4 Nanoscale bipolar electrical switching of phase-change materials
125
Initially, the cell stage of the as-deposited GST film is HRS (see black curve) and
sharp switching to LRS occurs after the bias voltage exceeds the threshold voltage
Vth, suggesting that the conduction is due to the formation of a filament in the GST
film. For the negative voltage polarity (black curve in Figure 4.52), the decreased
current indicates that the cell switches to the HRS and remains in HRS until the bias
voltage becomes positive again. With a further voltage sweep in this direction, the
current of the cell increases abruptly at the value of SET or of the threshold voltage
(about +1.55 V) caused by switching from the HRS to the LRS. If the voltage
exceeds +1.55 V, the sample current rises dramatically up to >10 nA, which is the
current limit for preventing electrical breakdown. Following reversal of the voltage
sweeping (red line in Figure 4.52), the cell initially maintains the LRS at the
positive voltage polarity, then the current gradually declines with the decrease of
the bias voltage with an almost linear I-V behavior, as observed for the voltage
sweep in the range between around +0.9 V and 0.8 V. For voltages below 0.8 V,
the current is limited to a constant level of 12 nA, similar to the positive current
limit. It should be noted that with the very low current at the voltage below the so-
called holding voltage (Vh), the HRS is reached. It is possible for the state of the
cell to return back to HRS by the RESET operation in the negative voltage polarity
regime. The representation of the I-V curve on a logarithmic scale, using the
absolute value of the current, is shown in the inset of Figure 4.52 (bottom right).
The I-V curve loops in a pinch to zero current at zero bias voltage and it changes in
the bias-polarity-dependence current of the curves, which is identical to a common
capacitor-like cell structure. Moreover, the current through the GST film rises
suddenly up to a level of three orders of magnitude higher than the prior state, when
the voltage is swept above the threshold voltage for crystallization. Similar pinched
hysteresis I-V curves were also found during investigations of the I-V curve as a
function of the repeated switching, voltage sweeping rate, and film thickness
(shown in the following), indicating that the experimental results for polarity-
dependent switching coincide well with the proposed switching mechanism as
introduced in Chapter 4.4.1.
The intrinsic memory effect of the cell was found to be reproducible. The I-V
curves of the device for the representative first ten sequences of repeatable
Chapter 4 Results and discussion
126
switching are plotted together in Figure 4.53a. It shows that the bipolar switching
characteristics of the PtIr/GST/Cr cell are well maintained during repeated cycling.
The cell can be SET (HRS) by applying a positive bias and RESET by applying a
negative bias. Extended endurance characteristics were determined by repeated
sweeping of the voltage between -3 V and +3 V at a sweeping rate of 1.8 V/s and at
a temperature of 85°C up to 1000 cycles, as shown in Figure 4.53b. The resistance
of the HRS shows an enhanced scattering after 600 cycles, whereas the resistance of
the LRS remains mostly constant. However, a considerable difference between HRS
and LRS is clearly maintained during the 1000 cycles. The two levels of resistance
switching range between about 8 MΩ and 100 GΩ, which corresponds to
approximately 3-4 orders of magnitude between the LRS and the HRS.
The switching probability of the prototype cell is also obtained from data in
Figure 4.53b and is shown in Figure 4.53c. The probability is the calculated
cumulative resistance distribution probability of the different cycle numbers (up to
Figure 4.53: (a) Repeatable switching I-V curves of a Pt-Ir/GST/Cr cell. Only the first ten repeated
switching cycles are shown, for the sake of clarity. Insert: corresponding log-scale switching I-V
curves show LRS/HRS ratio of ~factor 5×103. (b) Cycling endurance characteristics of a Pt-
Ir/GST/Cr cell at a cell temperature of 85°C for 1000 cycles. (c) Corresponding cumulative
probability of resistance distribution of the cell. (d) Data retention characteristics of the cell at room
temperature for more than 180 days.
(a) (b)
(c) (d)
-2 -1 0 1 2-15
-10
-5
0
5
10
current limit
current limit
RE
SE
T
Cu
rre
nt
(nA
)
Voltage (V)
SE
T
Repeated switching
-2 -1 0 1 210
-12
10-11
10-10
10-9
10-8
0 200 400 600 800 100010
7
108
109
1010
1011
1012
Resis
tance (
)
Number of cycles
LRS
HRS
~ 5
1E7 1E8 1E9 1E10 1E11 1E12
0
20
40
60
80
100
Cum
ula
tive p
robabili
ty (
%)
Resistance ()
1 10 10010
8
109
1010
1011
1012
Ron
Roff
Re
sis
tan
ce
(
)
Retention time (days)
4.4 Nanoscale bipolar electrical switching of phase-change materials
127
1000 cycles). The resistance of the HRS shows a slight fluctuation between 1011
Ω
and 1012
Ω. The resistance of the LRS is quite uniform at 8×108 Ω. The ratio of the
HRS to the LRS remains without remarkable degradation over the range of 3-4
orders of magnitude of the resistance.
Furthermore, to investigate the data retention characteristics of the PtIr/GST/Cr
cell, the electrical properties have been measured for a time interval exceeding 6
months. The variation of the resistance as a function of the retention time is
presented in Figure 4.53d. It shows that in such a cell both HRS (Roff) and LRS
(Ron) can be maintained at a high ratio of 3-4 orders of magnitude for several tens of
days. Subsequently, the values of HRS are slightly reduced, but are still clearly
distinguishable from the LRS (one order of magnitude difference) after a retention
time of more than 180 days. In general, a higher ratio of HRS to LRS can be of
advantage not only for the signal-to-noise ratio in data transformation but it is also
promising for multilevel memory technology applications (analogous to the
principle of optical multi-level storage; see Chapter 4.3). The achieved ratio of HRS
to LRS of 3-4 orders of magnitude in this work is higher than in the conventional
ReRAM [186]. Such a high HRS/LRS ratio can possibly be explained by the
proposed electrical resistance switching mechanism in GST films that is described
in detail in Chapter 4.4.4.
Figure 4.54 shows the measured I-V curves of typical features of pinched
hysteresis loops and sweeping rate-dependent hysteresis in the Pt-Ir/GST/Cr cell.
The bipolar switching behavior of the cell, in principle, seems to be almost
independent of the voltage sweeping rate in the investigated range between 1.8 and
120 V/s (see Figures 4.54a-e). Moreover, it is worth noting that, with respect to the
resistance values of the cell obtained from Figures 4.54a-e, the values of the LRS
(Ron) lie very close on a common resistance level, whereas the level of the HRS
(Roff) can be adjusted by changing the sweeping rate, as demonstrated in Figure
4.54f. At a sweeping rate ≤ 3.6 V/s, the resistances (Roff) are approximately
5.2×1011
Ω, which is ~4×103 times higher than the LRS (Ron). In contrast, at a
higher sweeping rate (e.g., ≥ 12 V/s), the Roff values are approximately 1.8×109 Ω,
which is ~15 times higher than the HRS (Roff) values. Such distinct levels (I and II)
of Roff obtained at different sweeping rates can allow for possible multi-bits per cell
Chapter 4 Results and discussion
128
(multi-level storage), as indicated in Figure 4.54f. The reduction of the HRS (Roff)
at higher sweeping rates, as compared to lower sweeping rates, may be explained by
the shorter duration of the applied pulse, progressively narrowing the filaments
within the GST layer that connects both electrodes and, consequently, resulting in a
lower resistance.
Figure 4.54: (a) to (e) representative I-V curves of a Pt-Ir/GST/Cr cell at different voltage sweeping
rates from 1.8 V/s to 120 V/s. (f) corresponding On/Off resistance as a function of the sweeping rate.
The arrows in (a) to (e) indicate the directions of voltage sweep. The dashed lines in (f) are only
intended as a guide for the eye.
-3 -2 -1 0 1 2 3-15
-10
-5
0
5
10
Curr
ent (n
A)
Voltage (V)
1.8 V/s
-3 -2 -1 0 1 2 3-15
-10
-5
0
5
10 3.6 V/s
Curr
ent (n
A)
Voltage (V)
-3 -2 -1 0 1 2 3-15
-10
-5
0
5
10 12 V/s
Curr
ent (n
A)
Voltage (V)
-3 -2 -1 0 1 2 3-15
-10
-5
0
5
10 60 V/s
Curr
ent (n
A)
Voltage (V)
-3 -2 -1 0 1 2 3-15
-10
-5
0
5
10 120 V/s
Curr
ent (n
A)
Voltage (V)
(a) (b)
(d)
(f)
(c)
(e)
1 10 100
108
109
1010
1011
1012
~factor 4 103
Level I
Roff
Ron
Resis
tance (
)
Sweeping rate (V/s)
Level II
~factor 15
4.4 Nanoscale bipolar electrical switching of phase-change materials
129
Threshold of resistance switching 4.4.3
The threshold of resistance switching plays an essential role in the operation and
performance of resistance switching memory cells. As shown in Figure 4.52 (in
Chapter 4.4.2), the resistance of the HRS is suddenly switched to the LRS at a
specific electrical voltage, known as the threshold voltage (Vth). The resistance in
the LRS up to this critical voltage is called the holding voltage (Vh). As long as the
sweeping voltage is lower than Vh, the current across the GST layer tends towards a
value close to zero, due to the change of resistance from the LRS to the HRS.
Figure 4.55 summarizes the statistic values of Vth and Vh measured at 12 different
points on the sample surface, as marked by crosses in the topography image (500 ×
500 nm²) of the ~50 nm thick GST film. On average, a threshold voltage Vth of 1.55
± 0.4 V and a holding voltage Vh of 0.5 ± 0.1 V are obtained by sweeping the
voltage from 0 up to +3 V and from +3 V down to 0, with a sweep rate of 1.8 V/s.
Thus, the results for Vth and Vh are stable and reproducible for a homogeneous GST
film of a given film thickness.
However, it is worth noting that the Vth values are tunable by varying the GST
film thickness. These experimental results are shown in Figure 4.56. Values of Vth
in the range between 1.2 and 4.2 V are obtained for GST film thicknesses varied
between 10 and 300 nm. Figure 4.56a shows the dependence of Vth on the film
thickness. For film thicknesses > 50 nm, the threshold voltages are approximately
constant. However, once the film thickness is less than 50 nm, the values of the
threshold voltage increase exponentially with decreasing film thickness. It is noted
Figure 4.55: Distribution of threshold voltage and holding voltage measured at various probe points,
marked in the inset, showing the surface topography of the investigated GST film. The numbers in
the inset mark the first, fifth and ninth measuring points.
4 nm
0 2 4 6 8 10 120.0
0.4
0.8
1.2
1.6
2.0
Number of probe point
Th
resh
old
vo
lta
ge
Vth (
V)
0.0
0.4
0.8
1.2
1.6
2.0
Ho
ldin
g v
olta
ge
Vh (
V)
1
5
9
Chapter 4 Results and discussion
130
that the positive polarity Vth is slightly different from the negative polarity Vth. This
is possibly related with the formation of an asymmetric trap-band structure within
GST under both positive and negative electrical fields, because structural defects in
the interface of GST films and electrodes are involved [187]. The values for the
threshold voltage in Figure 4.56a were extracted by fitting the data points of the
current at the critical sweeping voltage, as described in Figures 4.56b and c. It is
obvious that, above Vth, I-V characteristics of GST films at various film thicknesses
show a linear, i.e., ohmic relationship, indicating the respective resistance states of
the films switched to LRS. Consequently, the relationship of resistance and film
thickness, as shown in Figure 4.56d, illustrates that the LRS is uniform with a
change of the film thickness, whereas the HRS is slightly increased when the
thickness of GST films is reduced to ≤ 50 nm. Nevertheless, the divergence of the
ratio of HRS and LRS is within less than one order of magnitude.
In a previous report, it was shown that the crystallization temperature of GST
films is dependent on the film thickness in an inverse exponential relationship
Figure 4.56: (a) The threshold voltage as a function of film thickness. (b), (c) Positive and negative
I-V curves as a function of film thickness. (d) The electrical resistance of HRS and LRS as a
function of film thickness.
(a)
(d)0 1 2 3 4 5
0
2
4
6
8
10 10 nm
20 nm
80 nm
150 nm
300 nm
Cu
rre
nt (n
A)
Voltage (V)
Vth
-5 -4 -3 -2 -1 0-10
-8
-6
-4
-2
0
10 nm
20 nm
80 nm
150 nm
300 nm
Curr
ent (n
A)
Voltage (V)
Vth
0 50 100 150 200 250 300
1
2
3
4
Th
resh
old
Vo
lta
ge
(V
)
Film thickness (nm)
Positive Vth
Negative Vth
Fit by y=1.56+7.31e-0.082x
Fit by y=1.65+6.48e-0.081x
0 50 100 150 200 250 300
108
109
1010
1011
1012
Re
sis
tan
ce
(
)
Film thickness (nm)
HRS
LRS
(b)
(c)
4.4 Nanoscale bipolar electrical switching of phase-change materials
131
[136]. The crystallization temperature was measured by using either in situ
resistivity or in situ XRD during annealing at identical heating rates. On the basis of
these experiments, it was concluded that the role of interfaces will certainly increase
with reduced film thickness, which leads to a larger influence on the crystallization
process of PCMs. This is in agreement with the trend of increasing Vth with reduced
GST film thicknesses in this study.
Mechanism of bipolar resistance switching in phase-change materials 4.4.4
The conventional physical mechanism for resistive switching based on phase-
change materials is Joule heating of the materials, resulting in a phase
transformation between amorphous and crystalline states. Correspondingly, the
current driven through the device switches the thin GST film to LRS when T > Tc
(SET), and switches back to HRS when T > Tm (RESET). This is unipolar resistive
switching, which is not dependent on the polarity of the bias voltage, because
negative voltages can also be applied for SET or RSET. However, the experimental
results for the I-V characteristics of a PC/Re-RAM device (see Chapter 4.4.2)
demonstrate that the resistive switching of the GST film presented here is, indeed,
dependent on the polarity of the bias voltage. Thus, bipolar resistive switching is
found here; this has previously been attributed to electrochemical metallization
(ECM) or to solid-state electrolytic behavior. Therefore, it is suggested that the
switching mechanism is a combination of both crystalline phase change in the GST
film and solid-state electrolytic behavior. The investigation is structured in the
following sections according to: 1) the Poole-Frenkel conduction mechanism, 2)
visualization of the polarity-dependent resistance switching processes provided by
C-AFM as an active switching tool, and 3) correlation of microstructure and
chemical composition of the C-AFM tip-induced marker by analysis of cross-
sectional HRTEM, NBD and focused EDX.
1.) Poole-Frenkel conduction mechanism
The Poole-Frenkel emission is one common conduction mechanism used to explain
the HRS current [188-190]. Defect-related localized states in amorphous phase-
change materials (here a-GST) act as electronic traps that are able to capture
electrons (so-called Anderson localization). However, when thermally excited, the
Chapter 4 Results and discussion
132
trapped electrons can be emitted into the conduction band, as depicted in Figure
4.57. The rate of this excitation of electrons into the conduction band is dependent
on the applied electrical field. The electron emission in the framework of the Poole-
Frenkel mechanism [191] is given by
𝐽 ∝ 𝐸 exp [−𝑞(∅𝐵−√
𝑞𝐸
𝜋𝜀)
𝑘𝑇] , (4.10)
where J is the current density, E is the electric field across the a-GST film, k is the
Boltzmann constant, T is the temperature, øB is the voltage barrier, q is the electron
charge, and ε is the dielectric constant of the material. Considering E = V/d and
J = I/d, where d is the film thickness, V is the voltage, and I is the current, equation
(4.10) can be rearranged to
𝑙𝑛(𝐼/𝑉) ∝ −𝑞∅𝐵−√ 𝑞3
𝜋𝜀𝑑 √𝑉
𝑘𝑇 . (4.11)
Figure 4.58 shows that this behavior fits well with the experimental results under
the assumption of the Poole-Frenkel effect being present in this case. Figure 4.58a
shows the relation of ln(I/V) versus √V in the HRS under negative bias for different
temperatures. The sweeping voltage is below -1.0 V, which is smaller than the
threshold switching voltage (~ -1.5 V). For an a-GST film thickness of ~18 nm (as
determined from the TEM image in Figure 4.61), the electric field is derived as
~105 V/cm. This is a sufficiently high applied electric field to drive the carriers
from Poole-Frenkel emission across the entirely amorphous GST matrix [188, 189,
192]. Here, the fit of the experimental results indicate that the Poole-Frenkel
Figure 4.57: Schematic representation of the Poole-Frenkel emission. Electrons at the valence band
edge inside amorphous GST are localized in trap states. Due to an external voltage (EF (left) –
EF(right)), the emission barrier for a jump towards the next trap state (or the electrode) is lowered
and can be overcome by thermal excitation. Electrons can thus flow across the insulating layer
without the external voltage exceeding the breakdown voltage.
qøB
a-GST
EF EC
EF
EV
4.4 Nanoscale bipolar electrical switching of phase-change materials
133
emission is the dominant current conduction mechanism below Vth.
To gain more insight into the conduction mechanism, the conduction in the
amorphous GST film, as function of the bias voltage and temperature, is shown in
Figure 4.58: Analysis of the experimental results on the conductivity in the HRS, based on the
Poole-Frenkel effect: (a) ln(I/V) vs. sqrt(V) plots for a negative bias voltage, (b) ln(I/V) vs. 1/kT to
extract the activation energy Ea for carrier emission, (c) the activation energy plotted as a function of
sqrt(V) to determine the dielectric constant of the material.
0.2 0.4 0.6 0.8 1.030
31
32
33
34
negative bias
80°C
70°C
60°C
50°C
PF fit
ln (
I/V
) (S
)
sqrt (V) (V)
33 34 35 36-30
-31
-32
-33
-34
ln (
I/V
) (S
)
1/kT (eV-1)
bias - 0.2 V
- 0.3 V
- 0.4 V
- 0.5 V
0.4 0.5 0.6 0.70.40
0.45
0.50
0.55
0.60
Ea (
eV
)
sqrt (V) (V)
slope= -0.53
Intercept = 0.81 eV
(a)
(b)
(c)
Chapter 4 Results and discussion
134
Figure 4.58b. Low bias voltages between -0.2 and -0.5 V and temperatures between
50 and 80°C are investigated, as shown in the ln(I/V) versus 1/kT plots. Due to the
Arrhenius type of equation (4.10), the activation energy Ea for Poole-Frenkel carrier
emission of the amorphous GST film can be given as (considering again E = V/d)
𝐸𝑎 = 𝑞∅𝐵 − √𝑞3
𝜋𝜀𝑑 √𝑉 . (4.12)
The value of Ea can be obtained as extracted from the slopes of the Arrhenius plots
in Figure 4.58b, and is in the range from 0.43 to 0.56 eV for external voltages
between -0.5 and -0.2 V. Such values of Ea are consistent with the case in which the
Fermi level is pinned around the middle of the bandgap, as predicted by previously
reported bandgap energies of amorphous GST [68, 192-194]. Moreover, the
activation energy decreases as the bias voltage increases, as represented in the plot
of Ea vs. √V in Figure 4.58c. This is in agreement with previous reports [195],
indicating a hopping transport of carriers by thermal emission over a potential
barrier, which is generally lowered by the applied electric field [194, 196]. The
slope of the curve in Figure 4.58c provides the dielectric constant 𝜺 of the material.
With a very small measuring frequency (scan rate of 10 Hz), the dielectric constant
was accordingly estimated to be ~1.15. This value is quantitatively smaller than the
one predicted in the literature [197, 198]. By extrapolating to the zero bias point, the
trap energy (ET= qΦB) in the a-GST film is extracted to be around 0.81 eV. The
Fermi level is pinned close to the valence band edge in a degenerate p-type
semiconductor (i.e., close to EV, but electrons cannot occupy EF). Since the GST
conductivity is p-type, acceptor-like trap states determine the electronic conduction.
ET is located between EF and EV [68]. In summary, the conduction mechanism of
the HRS in a-GST films, based on the I-V characteristics as dependent on
temperature and applied electrical field, is identified as being based on the Poole-
Frenkel effect. For more details about this electrical conduction mechanism, see
Refs. [192, 194].
2.) Polarity-dependent resistance switching
Conductive-AFM is used to determine the surface topography, as well as the local
electrical conductivity of the phase-change thin films, as described above. Arrays of
marker points at the surface of amorphous phase-change films can be written by a
4.4 Nanoscale bipolar electrical switching of phase-change materials
135
computer-controlled movement of the tip in the scanning contact mode and by
application of a voltage (exceeding the threshold switching voltage). Moreover, it is
also possible to record topography and current images simultaneously (see
description in Chapter 3.4) in order to visualize the resistance switching process.
Figure 4.59a shows an 8×10 array of marker points in the amorphous GST film
with a distance of 80 nm between the marker points. Figure 4.59b shows the current
profile of a chosen marker point row (dashed line in Figure 4.59a). The tip-induced
current markers in the amorphous GST films were realized by application of a
sweeping voltage of +3 V at a sweeping rate of 1.8 V/s. The local marker points are
evidenced by the electric current image that was obtained from the scanning contact
mode over the region (1×1 μm²) with an applied bias voltage of +0.5 V. The current
Figure 4.59: (a) A current image of a C-AFM tip-induced 8×10 array of markers in an amorphous
GST film and (b) a current profile plotted along the dashed line in (a).
0.4 nA
(a)
(b)
0 200 400 600 800 10000.0
5.0E2
1.0E3
1.5E3
2.0E3
2.5E3
Cu
rre
nt
(pA
)
Lateral distance (nm)
Chapter 4 Results and discussion
136
image exhibits a considerably high contrast between the tip-induced local marker
points (bright spots indicate a high conductivity) and the surrounding amorphous
material. It is possible to change the local electrical resistance/conductivity of the
amorphous GST film significantly by applying a writing current to the AFM tip.
The bright spots represent the contact area of the current-induced crystalline
filaments at the GST surface (see Chapter 4.4.1). The spots on the conductive
filaments at the surface have a diameter between 25 and 50 nm and they are
frequently non-homogeneous. This inhomogeneity of the markers at the surface
may be caused by improper tip-sample electrical contact, including a rough surface
topography, wearing-off of the conductive tip coating, or lateral spread of the Joule
heating in the presence of small inhomogeneities in the film. The current profile
extracted along a row of a current image, as marked in Figure 4.59a, demonstrates
that the tip-induced points are characterized by an approximately three times higher
current response, in comparison with the amorphous surroundings. The size of the
observed bright spots (in-plane filament dimension) is approximately proportional
to the probing current across the GST film. From the current profile at the
respectively formed in-plane filament dimension in Figure 4.59b (as plotted from
Figure 4.59a), the localization of the switching spots of, on average ~15 nm, is
reproducibly observed in the full width at half maximum. This implies that the
minimum dimension of the locally formed atomic conductive pathways within the
intra-connection of the GST film may be less than 15 nm. Details of the local
geometry of a conductive filament are described in the subsequent section.
Considering an in-plane filament diameter dimension of 15 nm, the resistance
switching behavior potentially allows for very high data storage densities up to
0.7 Tbit/inch².
The next step in this study was to analyze the switching behavior of these tip-
induced markers by changing the electrical polarity. The electrical resistance of the
tip-induced points can be switched between a low and a high resistance level by
reversing the polarity of the applied electric field. The experimental results are
summarized in Figure 4.60. For an operation bias voltage of ±0.5 V, topography (a-
c) and current images (d-f) were simultaneously recorded over the region of
4.4 Nanoscale bipolar electrical switching of phase-change materials
137
500×500 nm². The corresponding current profiles (g-i) were plotted from the
current images along the dashed lines in (d-f). One exemplary tip-induced marker
point was selected to study the polarity-dependent resistance switching. The
topography (see Figures 4.60, a-c) remains unchanged during a switching process
between +0.5 V and 0.5 V and the reverse. By applying a positive bias voltage of
+0.5 V, the LRS of the marked region is clearly distinguished from the surrounding
amorphous phase, as shown in the current image (Figure 4.60d). Upon the polarity
change of the applied bias voltage to 0.5 V, the current flow through the GST film
at the marked point is in the range of several pA only (Figure 4.60e); i.e., it is much
lower in comparison to the previously produced LRS (several nA). This indicates
that the electrical resistance of the marked point switches from the LRS to the HRS.
The electrical contrast between LRS and HRS is about three orders of magnitude
Figure 4.60: Polarity-dependent resistance switching of a GST film at the nanometer scale: (a-c)
topography images for bias voltages of +0.5 V, 0.5 V and +0.5 V, respectively; (d-f) current
images, corresponding to the images (a-c); (g-i) current profiles extracted from the marked lines in
the current images (d-f), respectively.
+0.5 V -0.5 V +0.5 V
0 100 200 300 400 500
0
1
2
3
4
Cu
rre
nt
(nA
)
Lateral distance (nm)
0 100 200 300 400 500
0
1
2
3
4
Lateral distance (nm)
Cu
rre
nt
(nA
)
0 100 200 300 400 500
0
1
2
3
4
Cu
rre
nt
(nA
)
Lateral distance (nm)
Z = 5 nm
Z = 4 nm Z = 3 nm
Z = 2 nA
Z = 2 pA Z = 2 nA
(a) (b) (c)
(d) (e) (f)
(g) (h) (i)
Chapter 4 Results and discussion
138
(see corresponding current profiles in Figures 4.60g and h, respectively).
Furthermore, it is also clear from Figure 4.60f that the marker point has been
brought back to the LRS after another positive bias scanning with +0.5 V. The
current profile in Figure 4.60f indicates that the current across the GST film is in the
order of several nA; i.e., in the same range as before the process of switching
(compared with Figure 4.60g). The magnitude of the bias voltage in contact-mode
scanning for polarity switching is 0.5 V, which is lower than the threshold voltage
of about 1.55 V. This indicates a resistance switching of the GST film, as discussed
in the case of the I-V characteristic in Chapter 4.4.1. This implies that the observed
bipolar resistance switching is due to the change of the voltage polarity, causing
ionic migration, rather than to thermal heating of the material, resulting in a phase-
change process. In detail, only sufficient power is required to drive the local Te or
Ge ion movement as the functionality for bridging or rupturing conduction channels
between GST nanocrystal grains or grains and electrodes in the amorphous GST
film. Consequently, the bipolar electrical switching of the memory cell can be
realized with a lower bias voltage than expected. This required voltage is less than
the voltage needed to provoke a reverse phase-change in GST.
3.) Structural changes connected to the switching process
In order to understand the correlation between the local switching process and the
microstructure of phase-change GST films, thin lamellae of the phase-change
memory films were prepared with a FIB and analyzed by cross-sectional HRTEM,
NBD and EDX analysis. To locate the conductive channel regions in the GST
memory, the vicinity of tip-induced regions was tracked using pre-prepared FIB
markers. Results of the cross-sectional HRTEM studies for both as-prepared and
tip-induced regions of the GST film are shown in Figures 4.61a and b, respectively.
For the as-prepared GST films, including metal electrodes, a uniformly amorphous
and smooth GST thin film was found that had a thickness of ~20 nm between the Pt
top layer (that was deposited during the FIB process in order to protect the GST
surface from subsequent ion irradiation during FIB preparation) and the Cr bottom
electrode. A thin natural surface oxide of the GST layer, as well as at the surface of
the Cr layer, in the form of CrOx was detected (the effect of oxygen in this system is
being investigated further), as shown in Figure 4.61a. The long-range disordered
4.4 Nanoscale bipolar electrical switching of phase-change materials
139
amorphous structure in the GST layer was confirmed from the 2D fast Fourier
transformed (FFT) image region shown in the inset of Figure 4.61a, corresponding
to the region marked by the square. In sharp contrast, after introducing an electrical
current locally into the tip-induced region, the GST presents nanocrystals with
lattice planes clearly visible in HRTEM imaging mode (see Figure 4.61b).
Figure 4.61: Cross-sectional TEM images, NBD and EDX analysis of GST based memory cell: (a)
HRTEM image of as-prepared GST film. (b) HRTEM image of tip-induced region. The insets are
the fast Fourier transformation patterns corresponding to the marked square regions in the HRTEM
images. (c) NBD images 1-6 corresponding to the numbered positions in the HAADF-STEM image
above. The diffraction pattern (4) can be identified as the [434] zone axis of the cubic Ge2Sb2Te5
phase with a tilt angle of ~1°. There are some residual diffraction spots in (5), resulting from the
electron beam running across the edge of the crystalline area. (d) Local EDX analysis corresponding
to the marked rectangle regions of (a) and (b); the dashed lines mark the composition of Ge2Sb2Te5.
(d)
(c)
(a) (b)
As-prepared Tip-induced0
10
20
30
40
50
60
70
Co
nce
ntr
atio
n(a
t.%
)
Ge
Sb
Te
1 2 3 4 5 6
1
4 5 6
2 3
Pt-Ir tip
GST
Cr
Pt-Ir tip
CrCr
GSTGST
As-prepared Tip-induced
CrOx CrOx
5 nm 5 nm
Chapter 4 Results and discussion
140
The spacing of these planes (~0.21 nm) corresponds well with the (022) planes
of cubic GST225. The corresponding FFT image (inset in the top left panel) is
extracted from the square region of the HRTEM image marked by the dashed box.
The presence of the spots and their specific arrangement in the FFT image indicates
a metastable cubic structure (Fm-3m). The corresponding local tip-induced
resistivity switching of the GST film with a bias voltage (> Vth) is illustrated by the
resulting I-V characteristic, as detailed in Chapter 4.4.2. It is clear that a sufficiently
high voltage applied between the conductive tip and the Cr bottom electrode leads
to the formation of a conductive nano-filament within the amorphous GST thin
film. Consequently, a high current flows through this conductive channel, which
generates Joule heating of its surroundings, and once the temperature and the
holding time that is required for GST crystal nucleation and growth is reached, the
formation of the crystalline phase occurs in the amorphous GST film. The geometry
of the conductive channel that accompanies the crystalline region is strongly
dependent on the distribution of the electric field [199-202]. The shallow valley
visible in the film surface of the tip-modified marker in Figure 4.61b may be caused
by the effect of the tip loading force and Coulomb attraction. From the cross-
sectional view, it appears that the contact radius of the tip and sample is around 10
to 15 nm and the penetration depth into the film is ~4 nm. Because of the shape of
the apex at the tip-film contact, the application of a voltage results in a radial
distribution of the electric field towards the bottom electrode, leading to crystallite
nucleation and successive growth starting from the tip-film contact (as sketched in
Figure 4.50). A quantitative investigation of the temperature profile in the presence
of an electrical nanoprobe, using a theoretical model, can be found in Ref. [200].
The estimated electric field of ~150 MV/m in this study exceeded the critical
electric field (~50 MV/m) for inducing crystallization of the GST film [202, 203].
In this scenario of a local, electric field-induced conductive channel and subsequent
thermal crystallization, the primary conductive channel may have measured 10 nm
or less in width. The NBD measurements further corroborate the information given
above.
A broadening in the lateral direction of the tip-induced crystallization is also
confirmed by the NBD measurement shown in Figure 4.61c. The displayed NBD
4.4 Nanoscale bipolar electrical switching of phase-change materials
141
images 1 to 6 correspond to different positions in the STEM image marked by red
circles. For the NBD measurements, the STEM beam was adjusted to a very small
convergence semiangle of 1.6 mrad by means of the smallest available condenser
aperture (20 um) and appropriate lens settings. The resulting diffraction patterns
were recorded with a camera length of 145 mm, while a HAADF-STEM image
under the same instrument conditions was also acquired (upper part of Figure
4.61c). These NBD patterns thus individually carry electron diffraction information
from a region in the sample roughly 5 nm in diameter along the entire TEM sample
thickness of approximately 40 nm. The diffuse halo patterns obtained from spot
positions 1 to 3, which are located to the left side of the tip-film contact region,
demonstrate the presence of an amorphous phase. More specifically, the first diffuse
ring maximum is at 3.232 nm-1
, the second at 5.114 nm-1
, both with significant
broadening. As a reference, the octahedral sites in cubic GST225 have nearest-
neighbor distances of 3.316 nm-1
, whilst hypothetical tetrahedral sites would have
nearest neighbor distances of 3.829 nm-1
. It thus seems likely that the short range
order is of octahedral coordination (also in the light of the second diffraction ring),
as expected. Nevertheless, the structures from spot positions 1 to 3, as the NBD
patterns show, are all uniformly amorphous. Position 4 corresponds to the tip-
induced area at which the electron diffraction reveals the presence of the cubic
phase of GST. The positions and intensities of the diffraction spots suggest a [434]
zone axis orientation of the largest grain contributing to the diffraction pattern, with
a tilt misorientation of ~1°. Along the [434] zone axis, the dominant diffraction
spots from the (‒202), (‒3‒35), (‒1‒33), (‒35‒1) (‒551), (‒4‒68), (‒2‒66), (4-82)
and (6-80) planes are clearly identified, as indicated in Figure 4.61c. When the
beam position is moved to the right side of the tip-film contact region (position 5),
the amorphous ring contribution to the diffraction pattern reappears, with only a
small crystalline component running across by the NBD beam. By moving the beam
position further towards the right, no more crystalline diffraction is observed, as
seen for position 6. Thus, it can be concluded that the GST nanocrystals formed by
tip-induced switching probably form a cylindrical region across the film. This is
consistent with the observations shown in Figure 4.61b.
Chapter 4 Results and discussion
142
The local compositional changes between as-prepared and after tip-induced
modification of the film were analyzed by STEM-EDX. The quantification results
shown in Figure 4.61d are extracted from the regions in Figures 4.61a and b that are
enclosed by dashed lines. The measured relative atomic concentrations of elements
Ge, Sb and Te in the as-prepared and tip-induced region are Ge22.1Sb21.5Te56.3 and
Ge17.7Sb20.3Te62.0, respectively. Considering the experimental error of the EDX
technique (up to ± 4 at.%, depending on S/N ratio and peak overlap), the
composition of the as-prepared film can be assumed to be that of Ge22.2Sb22.2Te55.5,
whereas the composition of the tip-induced region is slightly over-stoichiometric,
regarding the Te content, and under-stoichiometric, regarding the Ge content. The
results suggest that, during the voltage application in the tip-induced local switching
experiment, either Te or Ge atoms might migrate, due to ionic migration that
depends on the electric field (i.e., electropositive Ge with lowest ionization potential
among GST and electronegative Te ions) or to segregation of material components
during local Joule heating. This could be attributed to the movement of Te or Ge
ions forming bridges or ruptures of conduction channels between GST crystal
grains or between grains and electrodes in the amorphous GST film upon polarity
switching by voltage application, resulting in resistance switching (for more details,
see Chapter 4.4.1). Very recently, it has been reported [72, 178, 204-207] that, in
some compositions of the ternary system Ge-Sb-Te, bipolar resistive switching was
also observed, which was attributed to either Sb or Te ion migration, based on solid-
state electrolytic transport. Moreover, if the entire current in the I-V characteristic
flows along the ~10 nm wide channel formed by the GST nanocrystallites between
the Pt-Ir tip and the Cr electrode, as found, the current density and resistance would
be 104 A/cm² and 10
8 Ω, respectively. These values are two orders of magnitude
lower than those at which electromigration typically occurs in metal interconnects
(e.g., of Cr, Pt, etc.). Therefore, it should be possible to exclude an influence of in-
diffusing electrode material in the present study. In other words, bipolar resistance
switching behavior of a PC/Re-RAM device should, in this regard, be independent
of the metal electrode materials.
4.4 Nanoscale bipolar electrical switching of phase-change materials
143
Summary of the results on nanoscale bipolar electrical switching of 4.4.5
GST films
A concept for a PC/Re-RAM memory device was developed that was based on the
combination of both phase change mechanisms and the electrolytic properties of
PCMs. For this purpose, the active layer of a GST phase-change film based memory
was prepared by PLD. The properties and mechanism of the simple device were
verified for actual bipolar electrical switching. Based on using advanced C-AFM
techniques, bipolar electrical switching on the nanoscale (~15 nm) could be
demonstrated, and high data storage densities (0.7 Tbit/inch2) could be estimated on
the base of the present device setup. A high On/Off resistance ratio of ~5×103 was
obtained and the switching process could be influenced by voltage sweeping rates in
the regions ≤ 3.6 V/s and ≥ 12 V/s, showing distinct resistance values that reveal the
possibility of realizing several multi-levels per cell. A retention time of > 180 days
and a cycling endurance of > 103 times of such memory cells was measured.
Regarding the switching energy, above the obtained threshold voltage of ~1.5 V,
the current increased to a level in the range of only a few nA; this limited the
required power to a relatively small amount. Moreover, the threshold voltage was
influenced by film thickness. The conduction mechanism in amorphous GST films
was explored by an analysis based on the Poole-Frenkel conduction model. In
addition, the operation of a polarity-dependent switching process was revealed by a
local visualization using simultaneously measured current and topography images
of the GST film. The investigation of local conductive AFM tip-induced switching,
in correlation with a local microstructure and chemical composition analysis by
STEM, suggested the formation of cubic GST(225) nanocrystals within a
cylindrical region with a diameter of ~10 nm across the film, as well as a slight
deviation from stoichiometric GST (225) to a Te-rich and Ge-depleted composition.
From these results, the present switching mechanism was proposed to consist of a
combination of the usual phase changes in GST films and the migration of mobile
Ge/Te ions. The demonstrated creation of this novel type of PC/Re-RAM memory
does not only open up new opportunities for exploring physical mechanisms but
also proves PCMs to be promising candidates for next generation non-volatile
memory applications.
145
Chapter 5
Summary
This thesis covers the key challenges of PCMs in science and technology
development, regarding usage of PCMs in data storage applications. For this
purpose, PCM thin films of GST and GeTe were prepared with the PLD technique.
These films were the basis of a thorough investigation of phase transformations of
PCMs with respect to structural change, crystallization kinetics as well as local
microscopic mechanisms. The structural, optical, and electrical properties of these
films were examined with a particular focus on the switching processes. Finally, the
thesis explores topics such as controllable multi-level storage, scalability, and new
switching mechanism of PCMs for next generation non-volatile data storage
applications. Briefly, the contributions of this thesis include the following:
1. The ex situ thermally induced crystallization process of PLD-prepared GeTe
films was characterized regarding relevant film properties such as structure,
chemical composition, topography, optical reflectivity, and electrical resistivity.
The crystallization temperature for pure GeTe film was found to be 240°C.
Investigation of the crystallization kinetics revealed the activation energy to be
3.14 eV. Moreover, a higher crystallization temperature Tc (280°C) and a higher
activation energy for crystallization Ea (3.18 eV) could be obtained with oxygen
incorporation into GeTe films. The correspondingly higher achievable thermal
stability makes them excellent candidates for phase change application.
2. The in situ thermally induced crystallization process of GeTe films was explored
for dominant growth temperature regimes related to the amorphous, polycrystalline,
and fiber textured states. It was found that the optical reflectivity increased with the
corresponding degree of crystalline quality. An even higher degree of crystalline
quality was achieved by realization of epitaxial growth of GeTe films on BaF2(111)
substrates by PLD.
Chapter 5 Summary
146
3. The ns UV laser-induced phase transformations of PLD-prepared GeTe films
were identified by optical reflectivity and structural evaluation. It could be
demonstrated that the processes of crystallization and amorphization are
controllable as functions of the laser irradiation parameters pulse number, fluence,
and pulse repetition rate. With a threshold fluence of as low as 11-14 mJ/cm2 with
pulse numbers ≥ 5, the amorphous films crystallize in the rhombohedral GeTe
phase. In contrast, with higher threshold fluences of above 36 mJ/cm2 with pulse
numbers ≥ 5, the amorphous films crystallize in the cubic GeTe phase. The re-
amorphization of the crystallized films could be achieved by using single pulses at
fluences between 162 and 182 mJ/cm². By using these determined parameters,
reversible phase transformations of GeTe films were realized.
4. The crystallization of GST thin films induced by either ns or fs single laser pulse
irradiation was compared at the same wavelength of 248 nm. The appearance of a
crystalline state was found after ns-laser and fs-laser irradiation with fluences
≥ 36 mJ/cm² and ≥ 13 mJ/cm², respectively, by evaluation of structural and optical
reflectivity. Local microstructure analysis revealed that GST films irradiated with
fs-laser pulses exhibit a columnar growth, whereas, in ns-laser irradiated films, oval
shaped crystallites were produced. The corresponding growth mechanisms are
clearly different. Based on the fluence dependence of the optical reflectivity of GST
films, a model of irradiation-induced crystalline phase formation was used to assess
the average contribution of ns and fs laser photons to the phase transition process.
The obtained results demonstrate that single fs pulse laser irradiation appears to be
more effective in inducing crystallization of GST thin films than single ns pulse
laser irradiation, enabling the attainment of a higher data transfer rate. In particular,
the results gained from the fs single laser pulse irradiation provide a better
understanding of the rapid phase transformation mechanism in practical PCM
applications.
5. Inspired by the multi-level storage technique to increase storage density and
reduce energy consumption, non-volatile multi-level phase-change memory based
on single-layer GeTe films was developed. Distinct multi-levels, instead of the
commonly used binary states, could be achieved by generating intermittent states,
using ns UV laser irradiation with a controlled combination of laser parameters.
Chapter 5 Summary
147
Each laser pulse excitation induced a partial crystallization, and complete
crystallization was achieved by a succession of such excitations. Various
reflectivity states could be realized by producing different volume fractions of
crystallized material. A fluence of 26 mJ/cm2 and a specific sequence of pulse
numbers were used for the crystallization of the films. In the reverse process, a
single laser pulse at a higher fluence of 112 mJ/cm2 was used for amorphization of
the films. The formation of these distinct levels was shown to be reproducible.
Regarding the dynamics of the described phase transformations, the switching times
for the ON operation (complete crystallization) and OFF operation (amorphization)
were found to be ~300 ns and ~3 ns, respectively. The local microstructure phase
transformation processes were correlated to the observed reversible switching of the
optical reflectivity, which are important for understanding the underlying
mechanisms of the PCM-based multilevel phase-change memory.
6. A concept for a PC/Re-RAM memory device was developed that was based on
the combination of both phase-change mechanisms and the electrolytic properties of
PCMs. The memory cell exhibited excellent scalability (< 15 nm), low
RESET/SET current (several nA), small RESET/SET operation voltages (0.5 V),
long retention (> 180 days), excellent cycling endurance (> 1000 cycles) and high
On/Off resistance ratios (~5×103), as well as a high estimated data density (0.7
Tbit/inch2). Bipolar electrical switching of PCMs at the nanoscale was performed
by using C-AFM techniques for write, read, and erase operations. The physical
switching mechanism of PC/Re-RAM was validated by the Poole-Frenkel
conduction model. The polarity-dependent resistance switching processes could be
visualized by simultaneous topography and electric current images. The local
microstructure on the nanoscale of such memory cells and the corresponding local
chemical composition were correlated. The formation of cubic GST nanocrystals
within a cylindrical region with a diameter of ~10 nm across the film was observed,
as well as a slight deviation from stoichiometric GST (225) to a Te-rich and Ge-
depleted composition. From these results, the switching mechanism was proposed
to consist of a combination of the usual phase changes in GST films and the
migration of mobile Ge/Te ions. The demonstrated creation of this type of PC/Re-
RAM memory does not only open up new opportunities for exploring physical
Chapter 5 Summary
148
mechanisms but also proves PCMs to be promising candidates for next generation
non-volatile memory applications.
149
Bibliography
[1] A. V. Kolobov, P. Fons, A. I. Frenkel, A. L. Ankudinov, J. Tominaga, and
T. Uruga, Understanding the phase-change mechanism of rewritable optical
media, Nat. Mater. 3, 703 (2004).
[2] M. Wuttig and N. Yamada, Phase-change materials for rewriteable data
storage, Nat. Mater. 6, 824 (2007).
[3] H. S. P. Wong, S. Raoux, S. Kim, J. Liang, J. P. Reifenberg, B. Rajendran,
M. Asheghi, and K. E. Goodson, Phase change memory, Proc. IEEE 98,
2201 (2010).
[4] S. R. Ovshinsky, Reversible electrical switching phenomena in disordered
structures, Phys. Rev. Lett. 21, 1450 (1968).
[5] M. Wuttig, Exploring the limits of phase change memories, Proc. IEEE
Design, Automation and Test in Europe Conference and Exhibition
(DATE), Dresden, Germany, (2014), DOI: 10.7873/DATE2014.280.
[6] S. R. Elliott, Chalcogenide phase-change materials: past and future, Int. J.
Appl. Glass Sci. 6, 15 (2015).
[7] R. W. Eason, Pulsed laser deposition of thin films: Applications-led growth
of functional materials, Wiley-Blackwell, Hoboken (NJ), 2006.
[8] J. A. Greer, History and current status of commercial pulsed laser deposition
equipment, J. Phys. D: Appl. Phys. 47, 034005 (2014).
[9] H. Jiang, K. Guo, H. Xu, Y. Xia, K. Jiang, F. Tang, J. Yin, and Z. Liu,
Preparation and characterization of GeTe4 thin films as a candidate for phase
change memory applications, J. Appl. Phys. 109, 066104 (2011).
[10] H. Lu, E. Thelander, J. W. Gerlach, D. Hirsch, U. Decker, and B.
Rauschenbach, Ge2Sb2Te5 phase-change films on polyimide substrates by
pulsed laser deposition, Appl. Phys. Lett. 101, 031905 (2012).
[11] E. Thelander, J. W. Gerlach, U. Ross, F. Frost, and B. Rauschenbach,
Epitaxial growth of Ge-Sb-Te films on KCl by high deposition rate pulsed
laser deposition, J. Appl. Phys. 115, 213504 (2014).
[12] E. Thelander, J. W. Gerlach, U. Ross, A. Lotnyk, and B.
Rauschenbach, Low temperature epitaxy of Ge-Sb-Te films on BaF2 (111)
by pulsed laser deposition, Appl. Phys. Lett. 105, 221908 (2014).
[13] M. H. R. Lankhorst, B. W. S. M. M. Ketelaars, and R. A. M. Wolters, Low-
cost and nanoscale non-volatile memory concept for future silicon chips,
Nat. Mater. 4, 347 (2005).
[14] J. Y. Raty, V. Godlevsky, P. Ghosez, C. Bichara, J. P. Gaspard, and J. R.
Chelikowsky, Evidence of a reentrant Peierls distortion in liquid GeTe,
Phys. Rev. Lett. 85, 1950 (2000).
[15] D. Lencer, M. Salinga, and M. Wuttig, Design rules for phase-change
materials in data storage applications, Adv. Mater. 23, 2030 (2011).
[16] J. Tominaga, P. Fons, A. Kolobov, T. Shima, T. C. Chong, R. Zhao, H. K.
Lee, and L. Shi, Role of Ge switch in phase transition: Approach using
atomically controlled GeTe/Sb2Te3 superlattice, Jpn. J. Appl. Phys. 47, 5763
(2008).
Bibliography
150
[17] R. E. Simpson, P. Fons, A. V. Kolobov, T. Fukaya, M. Krbal, T. Yagi, and
J. Tominaga, Interfacial phase-change memory, Nat. Nanotechnol. 6, 501
(2011).
[18] A. V. Kolobov, M. Krbal, P. Fons, J. Tominaga, and T. Uruga, Distortion-
triggered loss of long-range order in solids with bonding energy hierarchy,
Nat. Chem. 3, 311 (2011).
[19] M. Hada, W. Oba, M. Kuwahara, I. Katayama, T. Saiki, J. Takeda, and
K. G. Nakamura, Ultrafast time-resolved electron diffraction revealing the
nonthermal dynamics of near-UV photoexcitation-induced amorphization in
Ge2Sb2Te5, Sci. Rep. 5, 13530 (2015).
[20] N. Yamada, E. Ohno, K. Nishiuchi, N. Akahira, and M. Takao, Rapid‐phase
transitions of GeTe‐Sb2Te3 pseudobinary amorphous thin films for an
optical disk memory, J. Appl. Phys. 69, 2849 (1991).
[21] G. B. Beneventi, L. Perniola, V. Sousa, E. Gourvest, S. Maitrejean, J. C.
Bastien, A. Bastard, B. Hyot, A. Fargeix, C. Jahan, J. F. Nodin, A. Persico,
A. Fantini, D. Blachier, A. Toffoli, S. Loubriat, A. Roule, S. Lhostis, H.
Feldis, G. Reimbold, T. Billon, B. De Salvo, L. Larcher, P. Pavan, D.
Bensahel, P. Mazoyer, R. Annunziata, P. Zuliani, and F. Boulanger, Carbon-
doped GeTe: A promising material for phase-change memories, Solid-State
Electron. 65-66, 197 (2011).
[22] C. Peng, P. Yang, L. Wu, Z. Song, F. Rao, S. Song, D. Zhou, and J. Chu,
Nitrogen-doped Ge3Te2 materials with self-restricted active region for low
power phase-change memory, J. Appl. Phys. 113, 034310 (2013).
[23] X. Zhou, W. Dong, H. Zhang, and R. E. Simpson, A zero density change
phase change memory material: GeTe-O structural characteristics upon
crystallisation, Sci. Rep. 5, 11150 (2015).
[24] X. Sun, E. Thelander, J. W. Gerlach, U. Decker, and B. Rauschenbach,
Crystallization kinetics of GeTe phase-change thin films grown by pulsed
laser deposition, J. Phys. D: Appl. Phys. 48, 295304 (2015).
[25] I. Friedrich, V. Weidenhof, W. Njoroge, P. Franz, and M. Wuttig, Structural
transformations of Ge2Sb2Te5 films studied by electrical resistance
measurements, J. Appl. Phys. 87, 4130 (2000).
[26] E. Thelander, Epitaxial Ge-Te-Sb Thin films by pulsed laser deposition,
Ph.D thesis, Leipzig University, 2015.
[27] N. Yamada, E. Ohno, N. Akahira, K. Nishiuchi, K. Nagata, and M. Takao,
High speed overwritable phase change optical disk material, Jpn. J. Appl.
Phys. 26, 61 (1987).
[28] S. Raoux, H.-Y. Cheng, M. A. Caldwell, and H.-S. P. Wong, Crystallization
times of Ge–Te phase change materials as a function of composition, Appl.
Phys. Lett. 95, 071910 (2009).
[29] M. Chen, K. A. Rubin, and R. W. Barton, Compound materials for
reversible, phase-change optical data storage, Appl. Phys. Lett. 49, 502
(1986).
[30] G. Bruns, P. Merkelbach, C. Schlockermann, M. Salinga, M. Wuttig, T. D.
Happ, J. B. Philipp, and M. Kund, Nanosecond switching in GeTe phase
change memory cells, Appl. Phys. Lett. 95, 043108 (2009).
Bibliography
151
[31] J. H. Coombs, A. P. J. M. Jongenelis, W. van Es-Spiekman, and B. A. J.
Jacobs, Laser‐induced crystallization phenomena in GeTe‐based alloys. I.
Characterization of nucleation and growth, J. Appl. Phys. 78, 4906 (1995).
[32] J. H. Coombs, A. P. J. M. Jongenelis, W. van Es‐Spiekman, and B. A. J.
Jacobs, Laser‐induced crystallization phenomena in GeTe‐based alloys. II.
Composition dependence of nucleation and growth, J. Appl. Phys. 78, 4918
(1995).
[33] J. Kalb, F. Spaepen, and M. Wuttig, Atomic force microscopy
measurements of crystal nucleation and growth rates in thin films of
amorphous Te alloys, Appl. Phys. Lett. 84, 5240 (2004).
[34] B. J. Kooi, W. M. G. Groot, and J. T. M. De Hosson, In situ transmission
electron microscopy study of the crystallization of Ge2Sb2Te5, J. Appl. Phys.
95, 924 (2004).
[35] J. A. Kalb, F. Spaepen, and M. Wuttig, Kinetics of crystal nucleation in
undercooled droplets of Sb- and Te-based alloys used for phase change
recording, J. Appl. Phys. 98, 054910 (2005).
[36] R. E. Simpson, P. Fons, A. V. Kolobov, M. Krbal, and J. Tominaga,
Enhanced crystallization of GeTe from an Sb2Te3 template, Appl. Phys. Lett.
100, 021911 (2012).
[37] S. S. Kabalkin, L. F. Vereshch and N. R. Serebrya, Phase transition in
germanium telluride at high pressures, J. Exp. Theor. Phys. 24, 917 (1967).
[38] T. Chattopadhyay, J. X. Boucherle, and H. G. Vonschnering, Neutron
diffraction study on the structural phase transition in GeTe, J. Phys. C: Solid
State Phys. 20, 1431 (1987).
[39] F. Tong, J. D. Liu, X. M. Cheng, J. H. Hao, G. Y. Gao, H. Tong, and X. S.
Miao, Lattice strain induced phase selection and epitaxial relaxation in
crystalline GeTe thin film, Thin Solid Films 568, 70 (2014).
[40] R. Peierls, Quantum theory of solids, Oxford University Press, Oxford,
1956.
[41] M. Wuttig, Phase change materials: Chalcogenides with remarkable
properties due to an unconventional bonding mechanism, Phys. Status Solidi
B 249, 1843 (2012).
[42] K. Perumal, Epitaxial growth of Ge-Sb-Te based phase change materials,
Ph.D. thesis, Humboldt-Universität zu Berlin, 2013.
[43] W. H. Zachariasen, The atomic arrangement in glass, J. Am. Chem. Soc. 54,
3841 (1932).
[44] A. V. Kolobov, P. Fons, J. Tominaga, A. L. Ankudinov, S. N. Yannopoulos,
and K. S. Andrikopoulos, Crystallization-induced short-range order changes
in amorphous GeTe, J. Phys.: Condens. Matter 16, S5103 (2004).
[45] N. Yamada and T. Matsunaga, Structure of laser-crystallized Ge2Sb2+xTe5
sputtered thin films for use in optical memory, J. Appl. Phys. 88, 7020
(2000).
[46] W. Welnic, A. Pamungkas, R. Detemple, C. Steimer, S. Blugel, and M.
Wuttig, Unravelling the interplay of local structure and physical properties
in phase-change materials, Nat. Mater. 5, 56 (2006).
[47] T. Matsunaga, H. Morita, R. Kojima, N. Yamada, K. Kifune, Y. Kubota, Y.
Tabata, J. J. Kim, M. Kobata, E. Ikenaga, and K. Kobayashi, Structural
Bibliography
152
characteristics of GeTe-rich GeTe-Sb2Te3 pseudobinary metastable crystals,
J. Appl. Phys. 103, 093511 (2008).
[48] X. Q. Liu, X. B. Li, L. Zhang, Y. Q. Cheng, Z. G. Yan, M. Xu, X. D. Han,
S. B. Zhang, Z. Zhang, and E. Ma, New structural picture of the Ge2Sb2Te5
phase-change alloy, Phys. Rev. Lett. 106, 025501 (2011).
[49] A. V. Kolobov, P. Fons, J. Tominaga, A. I. Frenkel, A. L. Ankudinov, and
T. Uruga, Local structure of Ge-Sb-Te and its modification upon the phase
transition, J. Ovonic Res. 1, 21 (2005).
[50] K. Ohara, L. Temleitner, K. Sugimoto, S. Kohara, T. Matsunaga, L. Pusztai,
M. Itou, H. Ohsumi, R. Kojima, N. Yamada, T. Usuki, A. Fujiwara, and M.
Takata, The roles of the Ge-Te core network and the Sb-Te pseudo network
during rapid nucleation-dominated crystallization of amorphous Ge2Sb2Te5,
Adv. Funct. Mater. 22, 2251 (2012).
[51] J. W. Christian, The theory of transformations in metals and alloys,
Pergamon Press, Oxford, 1965.
[52] R. W. Balluffi, S. Allen, and W. C. Carter, Kinetics of Materials, Wiley,
Hoboken (NJ) 2005.
[53] A. E. Nielsen, Kinetics of Precipitation, Pergamon Press, New York, 1964.
[54] M. Salinga, E. Carria, A. Kaldenbach, M. Bornhoefft, J. Benke, J. Mayer,
and M. Wuttig, Measurement of crystal growth velocity in a melt-quenched
phase-change material, Nat. Commun. 4, 2371 (2013).
[55] J. Orava, A. L. Greer, B. Gholipour, D. W. Hewak, and C. E. Smith,
Characterization of supercooled liquid Ge2Sb2Te5 and its crystallization by
ultrafast-heating calorimetry, Nat. Mater. 11, 279 (2012).
[56] J. Orava and A. L. Greer, Fast and slow crystal growth kinetics in glass-
forming melts, J. Chem. Phys. 140, 214504 (2014).
[57] N. Yamada, Origin, secret, and application of the ideal phase-change
material GeSbTe, Phys. Status Solidi B 249, 1837 (2012).
[58] J. Tominaga, T. Nakano, and N. Atoda, An approach for recording and
readout beyond the diffraction limit with an Sb thin film, Appl. Phys. Lett.
73, 2078 (1998).
[59] L. Waldecker, T. A. Miller, M. Rude, R. Bertoni, J. Osmond, V. Pruneri,
R. E. Simpson, R. Ernstorfer, and S. Wall, Time-domain separation of
optical properties from structural transitions in resonantly bonded materials,
Nat. Mater. 14, 991 (2015).
[60] V. Weidenhof, I. Friedrich, S. Ziegler, and M. Wuttig, Laser induced
crystallization of amorphous Ge2Sb2Te5 films, J. Appl. Phys. 89, 3168
(2001).
[61] J. Siegel, A. Schropp, J. Solis, C. N. Afonso, and M. Wuttig, Rewritable
phase-change optical recording in Ge2Sb2Te5 films induced by picosecond
laser pulses, Appl. Phys. Lett. 84, 2250 (2004).
[62] G. J. Zhang, D. H. Gu, X. W. Jiang, Q. X. Chen, and F. X. Gan,
Crystallization of amorphous Ge2Sb2Te5 films induced by a single
femtosecond laser pulse, Solid State Commun. 133, 209 (2005).
[63] H. Huang, F. Zuo, F. Zhai, Y. Wang, T. Lai, Y. Wu, and F. Gan, Fast phase
transition process of Ge2Sb2Te5 film induced by picosecond laser pulses
with identical fluences, J. Appl. Phys. 106, 063501 (2009).
Bibliography
153
[64] R. L. Cotton and J. Siegel, Stimulated crystallization of melt-quenched
Ge2Sb2Te5 films employing femtosecond laser double pulses, J. Appl. Phys.
112, 123520 (2012).
[65] Y. Liu, M. M. Aziz, A. Shalini, C. D. Wright, and R. J. Hicken,
Crystallization of Ge2Sb2Te5 films by amplified femtosecond optical pulses,
J. Appl. Phys. 112, 123526 (2012).
[66] H. Lu, E. Thelander, J. W. Gerlach, U. Decker, B. Zhu, and B.
Rauschenbach, Single pulse laser-induced phase transitions of PLD-
deposited Ge2Sb2Te5 films, Adv. Funct. Mater. 23, 3621 (2013).
[67] Q. Wang, J. Maddock, E. T. F. Rogers, T. Roy, C. Craig, K. F. Macdonald,
D. W. Hewak, and N. I. Zheludev, 1.7 Gbit/in.2
gray-scale continuous-
phase-change femtosecond image storage, Appl. Phys. Lett. 104, 121105
(2014).
[68] A. Pirovano, A. L. Lacaita, A. Benvenuti, F. Pellizzer, and R. Bez,
Electronic switching in phase-change memories, IEEE Trans. Electron
Devices 51, 452 (2004).
[69] A. Redaelli, A. Pirovano, A. Benvenuti, and A. L. Lacaita, Threshold
switching and phase transition numerical models for phase change memory
simulations, J. Appl. Phys. 103, 111101 (2008).
[70] C. D. Wright, M. Armand, M. M. Aziz, H. Bhaskaran, B. C. Choo, C.
Davies, M. Despont, S. Gidon, M. Klein, O. Lemonnier, A. Maisse, A.
Pauza, M. Salinga, A. Sebastian, L. Wang, and M. Wuttig, Scanning probe-
based phase change terabyte memories, E/PCOS Conf. Proc. (2008).
[71] W. W. Koelmans, J. B. C. Engelen, and L. Abelmann, Probe-based data
storage, IBM Research Report RZ 3891 (2015), arXiv:1511.08755.
[72] R. Pandian, B. J. Kooi, G. Palasantzas, J. T. M. De Hosson, and A. Pauza,
Nanoscale electrolytic switching in phase-change chalcogenide films, Adv.
Mater. 19, 4431 (2007).
[73] P. Vettiger, G. Cross, M. Despont, U. Drechsler, U. Durig, B. Gotsmann, W.
Haberle, M. A. Lantz, H. E. Rothuizen, R. Stutz, and G. K. Binnig, The
"millipede"-nanotechnology entering data storage, IEEE Trans.
Nanotechnol. 1, 39 (2002).
[74] L. Wang, J. Wen, C. Yang, S. Gai, and Y. Peng, The route for ultra-high
recording density using probe-based data storage device, Nano 10, 1550118
(2015).
[75] A.-K. U. Michel, P. Zalden, D. N. Chigrin, M. Wuttig, A. M. Lindenberg,
and T. Taubner, Reversible optical switching of infrared antenna resonances
with ultrathin phase-change layers using femtosecond laser pulses, ACS
Photonics 1, 833 (2014).
[76] M. Rude, R. E. Simpson, R. Quidant, V. Pruneri, and J. Renger, Active
control of surface plasmon waveguides with a phase change material, ACS
Photonics 2, 669 (2015).
[77] P. Hosseini, C. D. Wright, and H. Bhaskaran, An optoelectronic framework
enabled by low-dimensional phase-change films, Nature 511, 206 (2014).
[78] M. Cassinerio, N. Ciocchini, and D. Ielmini, Logic computation in phase
change materials by threshold and memory switching, Adv. Mater. 25, 5975
(2013).
Bibliography
154
[79] B. L. Jackson, B. Rajendran, G. S. Corrado, M. Breitwisch, G. W. Burr, R.
Cheek, K. Gopalakrishnan, S. Raoux, C. T. Rettner, A. Padilla, A. G.
Schrott, R. S. Shenoy, B. N. Kurdi, C. H. Lam, and D. S. Modha, Nanoscale
electronic synapses using phase change devices, ACM J. Emerging Technol.
Comput. 9, 12 (2013).
[80] D. Kuzum, R. G. D. Jeyasingh, B. Lee, and H. S. P. Wong, Nanoelectronic
programmable synapses based on phase change materials for brain-inspired
computing, Nano Lett. 12, 2179 (2012).
[81] S. R. Ovshinsky, Optical cognitive information processing - A new field,
Jpn. J. Appl. Phys. 43, 4695 (2004).
[82] C. D. Wright, P. Hosseini, and J. A. V. Diosdado, Beyond von-Neumann
computing with nanoscale phase-change memory devices, Adv. Funct.
Mater. 23, 2248 (2013).
[83] C. D. Wright, Y. Liu, K. I. Kohary, M. M. Aziz, and R. J. Hicken,
Arithmetic and biologically-inspired computing using phase-change
materials, Adv. Mater. 23, 3408 (2011).
[84] J. W. Park, S. H. Baek, T. D. Kang, H. Lee, Y. S. Kang, T. Y. Lee, D. S.
Suh, K. J. Kim, C. K. Kim, Y. H. Khang, J. L. F. Da Silva, and S. H. Wei,
Optical properties of (GeTe, Sb2Te3) pseudobinary thin films studied with
spectroscopic ellipsometry, Appl. Phys. Lett. 93, (2008).
[85] T. Tsafack, E. Piccinini, B. S. Lee, E. Pop, and M. Rudan, Electronic,
optical and thermal properties of the hexagonal and rocksalt-like Ge2Sb2Te5
chalcogenide from first-principle calculations, J. Appl. Phys. 110, 063716
(2011).
[86] R. K. Singh and J. Narayan, Pulsed-laser evaporation technique for
deposition of thin films: Physics and theoretical model, Phys. Rev. B 41,
8843 (1990).
[87] D. B. Chrisey and G. K. Hubler, Pulsed laser deposition of thin films, Wiley,
New York, 1994.
[88] J. M. Warrender and M. J. Aziz, Effect of deposition rate on morphology
evolution of metal-on-insulator films grown by pulsed laser deposition,
Phys. Rev. B 76, (2007).
[89] A. L. Patterson, The Scherrer formula for X-ray particle size determination,
Phys. Rev. 56, 978 (1939).
[90] CaRIne Crystallography: http://carine.crystallography.pagespro-orange.fr/.
[91] M. Yasakan, X-ray thin film measurement techniques: V. X-ray reflectivity
measurement, The Rigaku Journal 26, 1 (2010).
[92] RayfleX: http://www.ge-mcs.com/de/radiography-x-ray/analytical-x-ray-
xrd/1050-siweb-pl425.html.
[93] A. Lotnyk, D. Poppitz, U. Ross, J. W. Gerlach, F. Frost, S. Bernütz, E.
Thelander, and B. Rauschenbach, Focused high- and low-energy ion milling
for TEM specimen preparation, Microelectron. Reliab. 55, 2119 (2015).
[94] J. E. Jones, On the determination of molecular fields. II. From the equation
of state of a gas, Proc. R. Soc. London, Ser. A 106, 463 (1924).
[95] S. Raoux, B. Muñoz, H.-Y. Cheng, and J. L. Jordan-Sweet, Phase transitions
in Ge–Te phase change materials studied by time-resolved X-ray diffraction,
Appl. Phys. Lett. 95, 143118 (2009).
Bibliography
155
[96] G. C. Sosso, G. Miceli, S. Caravati, F. Giberti, J. Behler, and M. Bernasconi,
Fast crystallization of the phase change compound GeTe by large-scale
molecular dynamics simulations, J. Phys. Chem. Lett. 4, 4241 (2013).
[97] J. Hu, G. M. Vanacore, Z. Yang, X. Miao, and A. H. Zewail, Transient
structures and possible limits of data recording in phase-change materials,
ACS Nano 9, 6728 (2015).
[98] G. C. Sosso, M. Salvalaglio, J. Behler, M. Bernasconi, and M. Parrinello,
Heterogeneous crystallization of the phase change material GeTe via
atomistic simulations, J. Phys. Chem. C 119, 6428 (2015).
[99] L. Perniola, V. Sousa, A. Fantini, E. Arbaoui, A. Bastard, M. Armand, A.
Fargeix, C. Jahan, J.-F. Nodin, A. Persico, D. Blachier, A. Toffoli, S.
Loubriat, E. Gourvest, G. B. Beneventi, H. Feldis, S. Maitrejean, S. Lhostis,
A. Roule, O. Cueto, G. Reimbold, L. Poupinet, T. Billon, B. De Salvo, D.
Bensahel, P. Mazoyer, R. Annunziata, P. Zuliani, and F. Boulanger,
Electrical behavior of phase-change memory cells based on GeTe, IEEE
Electron Device Lett. 31, 488 (2010).
[100] G. Navarro, V. Sousa, A. Persico, N. Pashkov, A. Toffoli, J. C. Bastien, L.
Perniola, S. Maitrejean, A. Roule, P. Zuliani, R. Annunziata, and B. De
Salvo, Material engineering of GexTe100−x compounds to improve phase-
change memory performances, Solid-State Electron. 89, 93 (2013).
[101] D. Kashchiev, Nucleation: Basic theory with applications, Butterworth-
Heinemann, Oxford, 2000.
[102] S. Senkader and C. D. Wright, Models for phase-change of Ge2Sb2Te5 in
optical and electrical memory devices, J. Appl. Phys. 95, 504 (2004).
[103] H. E. Kissinger, Reaction kinetics in differential thermal analysis, Anal.
Chem. 29, 1702 (1957).
[104] M. Libera and M. Chen, Time‐resolved reflection and transmission studies
of amorphous Ge‐Te thin‐film crystallization, J. Appl. Phys. 73, 2272
(1993).
[105] Q. M. Lu and M. Libera, Microstructural measurements of amorphous GeTe
crystallization by hot-stage optical microscopy, J. Appl. Phys. 77, 517
(1995).
[106] Z. Fan, L. Wang, and D. V. Laughlin, Modeling of crystallization activation
energy for GeTe-Sb2Te3-based phase change materials, Proc. SPIE 5380,
493 (2004).
[107] T. Matsunaga, R. Kojima, N. Yamada, K. Kifune, Y. Kubota, Y. Tabata, and
M. Takata, Single structure widely distributed in a GeTe-Sb2Te3
pseudobinary system: A rock salt structure is retained by intrinsically
containing an enormous number of vacancies within its crystal, Inorg.
Chem. 45, 2235 (2006).
[108] J. J. Gervacio Arciniega, E. Prokhorov, F. J. Espinoza Beltran, and G.
Trapaga, Crystallization of Ge:Sb:Te thin films for phase change memory
application, published in M. R. B. Andreeta (Ed.), Crystallization - Science
and Technology, InTech, Vienna, 2012.
[109] M. Xu, S. J. Wei, S. Wu, F. Pei, J. Li, S. Y. Wang, L. Y. Chen and Y. Jia,
Theoretical and experimental investigations of the optical properties of GST
for multi-state optical data storage, J. Korean Phys. Soc. 53, 2265 (2008).
Bibliography
156
[110] H. K. Lyeo, D. G. Cahill, B. S. Lee, J. R. Abelson, M. H. Kwon, K. B. Kim,
S. G. Bishop, and B. K. Cheong, Thermal conductivity of phase-change
material Ge2Sb2Te5, Appl. Phys. Lett. 89, 151904 (2006).
[111] B. S. Lee and S. G. Bishop, Optical and electrical properties of phase change
materials, published in S. Raoux and M. Wuttig (Eds.), Phase change
materials: Science and Applications, Springer, New York, 2009.
[112] S. K. Bahl and K. L. Chopra, Amorphous versus crystalline GeTe films. II.
Optical properties, J. Appl. Phys. 40, 4940 (1969).
[113] T. Siegrist, P. Jost, H. Volker, M. Woda, P. Merkelbach, C. Schlockermann,
and M. Wuttig, Disorder-induced localization in crystalline phase-change
materials, Nat. Mater. 10, 202 (2011).
[114] T. H. Jeong, H. Seo, K. L. Lee, S. M. Choi, S. J. Kim, and S. Y. Kim, Study
of oxygen-doped GeSbTe film and its effect as an interface layer on the
recording properties in the blue wavelength, Jpn. J. Appl. Phys. 40, 1609
(2001).
[115] D. Z. Dimitrov, Y. H. Lu, M. R. Tseng, W. C. Hsu, and H. P. D. Shieh,
Oxygen and nitrogen co-doped GeSbTe thin films for phase-change optical
recording, Jpn. J. Appl. Phys. 41, 1656 (2002).
[116] B. Liu, Z. T. Song, T. Zhang, S. L. Feng, and B. M. Chen, Effect of O-
implantation on the structure and resistance of Ge2Sb2Te5 film, Appl. Surf.
Sci. 242, 62 (2005).
[117] S. J. Park, M. H. Jang, S. J. Park, M. Ahn, D. B. Park, D.-H. Ko, and M.-H.
Cho, Effect of amorphization on the structural stability and reversibility of
Ge2Sb2Te5 and oxygen incorporated Ge2Sb2Te5 films, J. Mater. Chem. 22,
16527 (2012).
[118] Z. Zhang, C. Peng, S. Song, Z. Song, Y. Cheng, K. Ren, X. Li, F. Rao, B.
Liu, and S. Feng, Characterization of Cu doping on GeTe for phase change
memory application, J. Appl. Phys. 114, 244311 (2013).
[119] L. V. Yashina, R. Puettner, V. S. Neudachina, T. S. Zyubina, V. I. Shtanov,
and M. V. Poygin, X-ray photoelectron studies of clean and oxidized alpha-
GeTe(111) surfaces, J. Appl. Phys. 103, 094909 (2008).
[120] E. Gourvest, B. Pelissier, C. Vallee, A. Roule, S. Lhostis, and S. Maitrejean,
Impact of oxidation on Ge2Sb2Te5 and GeTe phase-change properties, J.
Electrochem. Soc. 159, H373 (2012).
[121] https://www.capital.net/com/vcl/periodic/periodic.htm.
[122] S. Privitera, S. Lombardo, C. Bongiorno, E. Rimini, and A. Pirovano, Phase
change mechanisms in Ge2Sb2Te5, J. Appl. Phys. 102, 063506 (2007).
[123] S. W. Ryu, J. H. Oh, J. H. Lee, B. J. Choi, W. Kim, S. K. Hong, C. S.
Hwang, and H. J. Kim, Phase transformation behaviors of SiO2 doped
Ge2Sb2Te5 films for application in phase change random access memory,
Appl. Phys. Lett. 92, 142110 (2008).
[124] J. Stoemeno and R. Vincent, Twinning faults in epitaxial films of
germanium telluride and GeTe-SnTe alloys, Phys. Status Solidi A 11, 545
(1972).
[125] A. Giussani, K. Perumal, M. Hanke, P. Rodenbach, H. Riechert, and R.
Calarco, On the epitaxy of germanium telluride thin films on silicon
substrates, Phys. Status Solidi B 249, 1939 (2012).
Bibliography
157
[126] H. Sun, L. Hou, Y. Wu, and J. Wei, Structural change of laser-irradiated
Ge2Sb2Te5 films studied by electrical property measurement, J. Non-Cryst.
Solids 354, 5563 (2008).
[127] X. Sun, E. Thelander, P. Lorenz, J. W. Gerlach, U. Decker, and B.
Rauschenbach, Nanosecond laser-induced phase transitions in pulsed laser
deposition-deposited GeTe films, J. Appl. Phys. 116, 133501 (2014).
[128] P. Fons, P. Rodenbach, K. V. Mitrofanov, A. V. Kolobov, J. Tominaga, R.
Shayduk, A. Giussani, R. Calarco, M. Hanke, H. Riechert, R. E. Simpson,
and M. Hase, Picosecond strain dynamics in Ge2Sb2Te5 monitored by time-
resolved x-ray diffraction, Phys. Rev. B 90, 094305 (2014).
[129] S. K. Sundaram and E. Mazur, Inducing and probing non-thermal transitions
in semiconductors using femtosecond laser pulses, Nat. Mater. 1, 217
(2002).
[130] J. P. Callan, A. M. T. Kim, C. A. D. Roeser, E. Mazur, J. Solis, J. Siegel,
C. N. Afonso, and J. C. G. de Sande, Ultrafast laser-induced phase
transitions in amorphous GeSb films, Phys. Rev. Lett. 86, 3650 (2001).
[131] X. Sun, M. Ehrhardt, A. Lotnyk, P. Lorenz, E. Thelander, J. W. Gerlach, T.
Smausz, U. Decker, and B. Rauschenbach, Crystallization of Ge2Sb2Te5 thin
films by nano- and femtosecond single laser pulse irradiation, Sci. Rep. 6,
28246 (2016).
[132] D. Lencer, M. Salinga, B. Grabowski, T. Hickel, J. Neugebauer, and M.
Wuttig, A map for phase-change materials, Nat. Mater. 7, 972 (2008).
[133] I.-M. Park, J.-K. Jung, S.-O. Ryu, K.-J. Choi, B.-G. Yu, Y.-B. Park, S. M.
Han, and Y.-C. Joo, Thermomechanical properties and mechanical stresses
of Ge2Sb2Te5 films in phase-change random access memory, Thin Solid
Films 517, 848 (2008).
[134] B.-S. Lee, G. W. Burr, R. M. Shelby, S. Raoux, C. T. Rettner, S. N. Bogle,
K. Darmawikarta, S. G. Bishop, and J. R. Abelson, Observation of the role
of subcritical nuclei in crystallization of a glassy solid, Science 326, 980
(2009).
[135] P. Zalden, A. von Hoegen, P. Landreman, M. Wuttig, and A. M.
Lindenberg, How supercooled liquid phase-change materials crystallize:
snapshots after femtosecond optical excitation, Chem. Mater. 27, 5641
(2015).
[136] S. Raoux, Scaling properties of phase change materials, published in S.
Raoux and M. Wuttig (Eds.), Phase Change Materials: Science and
Applications, Springer, New York, 2009.
[137] L. G. Parratt, Surface studies of solids by total reflection of X-rays, Phys.
Rev. 95, 359 (1954).
[138] S. Raoux, J. L. Jordan-Sweet, and A. J. Kellock, Crystallization properties
of ultrathin phase change films, J. Appl. Phys. 103, (2008).
[139] X. Wei, L. Shi, C. Chong Tow, R. Zhao, and H. K. Lee, Thickness
dependent nano-crystallization in Ge2Sb2Te5 films and its effect on devices,
Jpn. J. Appl. Phys. 46, 2211 (2007).
[140] W. Gawelda, J. Siegel, C. N. Afonso, V. Plausinaitiene, A. Abrutis, and C.
Wiemer, Dynamics of laser-induced phase switching in GeTe films, J. Appl.
Phys. 109, 123102 (2011).
Bibliography
158
[141] A. Lotnyk, S. Bernütz, X. Sun, U. Ross, M. Ehrhardt, and B. Rauschenbach,
Real-space imaging of atomic arrangement and vacancy layers ordering in
laser crystallised Ge2Sb2Te5 phase change thin films, Acta Mater. 105, 1
(2016).
[142] E. T. Kim, J. Y. Lee, and Y. T. Kim, Investigation of the structural
transformation behavior of Ge2Sb2Te5 thin films using high resolution
electron microscopy, Appl. Phys. Lett. 91, 101909 (2007).
[143] B. S. Lee, J. R. Abelson, S. G. Bishop, D. H. Kang, B. K. Cheong, and K. B.
Kim, Investigation of the optical and electronic properties of Ge2Sb2Te5
phase change material in its amorphous, cubic, and hexagonal phases, J.
Appl. Phys. 97, 093509 (2005).
[144] W. K. Njoroge, H. W. Woltgens, and M. Wuttig, Density changes upon
crystallization of Ge2Sb2.04Te4.74 films, J. Vac. Sci. Technol., A 20, 230
(2002).
[145] J. Siegel, W. Gawelda, D. Puerto, C. Dorronsoro, J. Solis, C. N. Afonso,
J. C. G. de Sande, R. Bez, A. Pirovano, and C. Wiemer, Amorphization
dynamics of Ge2Sb2Te5 films upon nano- and femtosecond laser pulse
irradiation, J. Appl. Phys. 103, 023516 (2008).
[146] T. Ohta, M. Birukawa, N. Yamada, and K. Hirao, Optical recording; phase-
change and magneto-optical recording, J. Magn. Magn. Mater. 242, 108
(2002).
[147] K. Miura, J. R. Qiu, H. Inouye, T. Mitsuyu, and K. Hirao, Photowritten
optical waveguides in various glasses with ultrashort pulse laser, Appl. Phys.
Lett. 71, 3329 (1997).
[148] G. A. Prinz, Magnetoelectronics, Science 282, 1660 (1998).
[149] P. Lorenz, M. Ehrhardt, A. Wehrmann, and K. Zimmer, Laser-induced front
side etching of fused silica with XeF excimer laser using thin metal layers,
Appl. Surf. Sci. 258, 9138 (2012).
[150] D. Bäuerle, Laser Processing and Chemistry, Springer, Heidelberg, 2011.
[151] H. S. Carslaw and J. C. Jaeger, Conduction of Heat in Solids, Oxford
Science Publications, Oxford, 2008.
[152] E. M. Levin, M. F. Besser, and R. Hanus, Electronic and thermal transport
in GeTe: A versatile base for thermoelectric materials, J. Appl. Phys. 114,
083713 (2013).
[153] J. Lee, E. Bozorg-Grayeli, S. Kim, M. Asheghi, H. S. P. Wong, and K. E.
Goodson, Phonon and electron transport through Ge2Sb2Te5 films and
interfaces bounded by metals, Appl. Phys. Lett. 102, 191911 (2013).
[154] E. Morales-Sanchez, E. F. Prokhorov, A. Mendoza-Galvan, and J.
Gonzalez-Hernandez, Determination of the glass transition and nucleation
temperatures in Ge2Sb2Te5 sputtered films, J. Appl. Phys. 91, 697 (2002).
[155] Q. Yang, Z. Cai, Y. Wang, H. Huang, and Y. Wu, Controllable
crystallization of Ge2Sb2Te5 phase-change memory thin films driven by
multiple femtosecond laser pulses, Mater. Sci. Eng., B 193, 189 (2015).
[156] S. I. Anisimov, B. L. Kapeliovich, and T. L. Perelman, Electron emission
from metal surfaces exposed to ultrashort laser pulses, Sov. Phys. JETP 39,
375 (1974).
Bibliography
159
[157] B. J. Choi, S. Choi, T. Eom, S. H. Rha, K. M. Kim, and C. S. Hwang, Phase
change memory cell using Ge2Sb2Te5 and softly broken-down TiO2 films for
multilevel operation, Appl. Phys. Lett. 97, 132107 (2010).
[158] K. Nakayama, M. Takata, T. Kasai, A. Kitagawa, and J. Akita, Pulse
number control of electrical resistance for multi-level storage based on
phase change, J. Phys. D: Appl. Phys. 40, 5061 (2007).
[159] N. Papandreou, H. Pozidis, A. Pantazi, A. Sebastian, M. Breitwisch, C.
Lam, and E. Eleftheriou, Programming algorithms for multilevel phase-
change memory, Proc. IEEE Int. Symp. Circuits Syst. 329-332 (2011)
[160] C. Rios, M. Stegmaier, P. Hosseini, D. Wang, T. Scherer, C. D. Wright, H.
Bhaskaran, and W. H. P. Pernice, Integrated all-photonic non-volatile multi-
level memory, Nat. Photonics 9, 725 (2015).
[161] C. D. Wright, L. Wang, M. M. Aziz, J. A. V. Diosdado, and P. Ashwin,
Phase-change processors, memristors and memflectors, Phys. Status Solidi B
249, 1978 (2012).
[162] Y. Yin and S. Hosaka, Controlled promotion of crystallization for
application to multilevel phase-change memory, Appl. Phys. Lett. 100,
253503 (2012).
[163] J. Akola and R. O. Jones, Amorphous structures of Ge/Sb/Te alloys: Density
functional simulations, Phys. Status Solidi B 249, 1851 (2012).
[164] J. Akola, J. Larrucea, and R. O. Jones, Polymorphism in phase-change
materials: melt-quenched and as-deposited amorphous structures in
Ge2Sb2Te5 from density functional calculations, Phys. Rev. B 83, 094113
(2011).
[165] M. H. Kwon, B. S. Lee, S. N. Bogle, L. N. Nittala, S. G. Bishop, J. R.
Abelson, S. Raoux, B. K. Cheong, and K. B. Kim, Nanometer-scale order in
amorphous Ge2Sb2Te5 analyzed by fluctuation electron microscopy, Appl.
Phys. Lett. 90, 021923 (2007).
[166] B.-S. Lee, R. M. Shelby, S. Raoux, C. T. Retter, G. W. Burr, S. N. Bogle, K.
Darmawikarta, S. G. Bishop, and J. R. Abelson, Nanoscale nuclei in phase
change materials: Origin of different crystallization mechanisms of
Ge2Sb2Te5 and AgInSbTe, J. Appl. Phys. 115, 063506 (2014).
[167] B.-S. Lee, K. Darmawikarta, S. Raoux, Y.-H. Shih, Y. Zhu, S. G. Bishop,
and J. R. Abelson, Distribution of nanoscale nuclei in the amorphous dome
of a phase change random access memory, Appl. Phys. Lett. 104, 071907
(2014).
[168] M. Naito, M. Ishimaru, Y. Hirotsu, and M. Takashima, Local structure
analysis of Ge-Sb-Te phase change materials using high-resolution electron
microscopy and nanobeam diffraction, J. Appl. Phys. 95, 8130 (2004).
[169] S. Privitera, A. M. Mio, G. D'Arrigo, E. Carria, and E. Rimini, Impact of
annealing induced structural relaxation on the electrical properties and the
crystallization kinetics of amorphous GeTe films, Appl. Phys. Lett. 103,
071901 (2013).
[170] P. K. Khulbe, E. M. Wright, and M. Mansuripur, Crystallization behavior of
as-deposited, melt quenched, and primed amorphous states of Ge2Sb2.3Te5
films, J. Appl. Phys. 88, 3926 (2000).
[171] S. Raoux, R. Robert Shelby, B. Munoz, M. Hitzbleck, D. Daniel Krebs, M.
Salinga, M. Woda, M. Austgen, K. Chung, and M. Wuttig, Crystallization
Bibliography
160
times of as-deposited and melt-quenched amorphous phase change
materials, E/PCOS Conf. Proc. (2008).
[172] G. R. Fox, F. Chu, and T. Davenport, Current and future ferroelectric
nonvolatile memory technology, J. Vac. Sci. Technol. B 19, 1967 (2001).
[173] K. L. Chopra, Avalanche-induced negative resistance in thin oxide films, J.
Appl. Phys. 36, 184 (1965).
[174] D. B. Strukov, G. S. Snider, D. R. Stewart, and R. S. Williams, The missing
memristor found, Nature 453, 80 (2008).
[175] K. Szot, W. Speier, G. Bihlmayer, and R. Waser, Switching the electrical
resistance of individual dislocations in single-crystalline SrTiO3, Nat. Mater.
5, 312 (2006).
[176] K. Terabe, T. Hasegawa, T. Nakayama, and M. Aono, Quantized
conductance atomic switch, Nature 433, 47 (2005).
[177] R. Waser and M. Aono, Nanoionics-based resistive switching memories,
Nat. Mater. 6, 833 (2007).
[178] S. Yoo, T. Eom, T. Gwon, and C. S. Hwang, Bipolar resistive switching
behavior of an amorphous Ge2Sb2Te5 thin films with a Te layer, Nanoscale
7, 6340 (2015).
[179] A. L. Greer and N. Mathur, Materials science: Changing face of the
chameleon, Nature 437, 1246 (2005).
[180] T.-Y. Yang, I.-M. Park, B.-J. Kim, and Y.-C. Joo, Atomic migration in
molten and crystalline Ge2Sb2Te5 under high electric field, Appl. Phys. Lett.
95, 032104 (2009).
[181] F. Xiong, A. D. Liao, D. Estrada, and E. Pop, Low-power switching of
phase-change materials with carbon nanotube electrodes, Science 332, 568
(2011).
[182] S.-W. Nam, H.-S. Chung, Y. C. Lo, L. Qi, J. Li, Y. Lu, A. T. C. Johnson, Y.
Jung, P. Nukala, and R. Agarwal, Electrical wind force-driven and
dislocation-templated amorphization in phase-change nanowires, Science
336, 1561 (2012).
[183] J. J. Yang, M. D. Pickett, X. Li, D. A. A. Ohlberg, D. R. Stewart, and R. S.
Williams, Memristive switching mechanism for metal/oxide/metal
nanodevices, Nat. Nanotechnol. 3, 429 (2008).
[184] G. Bersuker, D. C. Gilmer, D. Veksler, J. Yum, H. Park, S. Lian, L.
Vandelli, A. Padovani, L. Larcher, K. McKenna, A. Shluger, V. Iglesias, M.
Porti, M. Nafria, W. Taylor, P. D. Kirsch, and R. Jammy, Metal oxide
RRAM switching mechanism based on conductive filament microscopic
properties, IEDM Tech. Dig. 456-459 (2010).
[185] J. Woo, S. Jung, M. Siddik, E. Cha, S. M. Sadaf, and H. Hwang, Effect of
interfacial oxide layer on the switching uniformity of Ge2Sb2Te5-based
resistive change memory devices, Appl. Phys. Lett. 99, 162109 (2011).
[186] D. Ielmini and R. Waser (Eds.), Resistive switching: From fundamentals of
nanoionic redox processes to memristive device applications, Wiley-VCH
Verlag, Weinheim, 2016.
[187] S. Lee, J.-H. Jeong, T. S. Lee, W. M. Kim, and B.-K. Cheong, Bias polarity
dependence of a phase change memory with a Ge-doped SbTe: A method
for multilevel programing, Appl. Phys. Lett. 92, 243507 (2008).
Bibliography
161
[188] G. B. Beneventi, L. Guarino, M. Ferro, and P. Fantini, Three-dimensional
Poole-Frenkel analytical model for carrier transport in amorphous
chalcogenides, J. Appl. Phys. 113, 044506 (2013).
[189] D. Ielmini and Y. Zhang, Analytical model for subthreshold conduction and
threshold switching in chalcogenide-based memory devices, J. Appl. Phys.
102, 054517 (2007).
[190] S. Menzel, U. Boettger, M. Wimmer, and M. Salinga, Physics of the
switching kinetics in resistive memories, Adv. Funct. Mater. 25, 6306
(2015).
[191] H. H. Poole, On the dielectric constant and electrical conductivity of mica in
intense fields, Philos. Mag. Ser. 6, 32, 112 (1916).
[192] M. Nardone, M. Simon, I. V. Karpov, and V. G. Karpov, Electrical
conduction in chalcogenide glasses of phase change memory, J. Appl. Phys.
112, 071101 (2012).
[193] J.-W. Park, S. H. Eom, H. Lee, J. L. F. Da Silva, Y.-S. Kang, T.-Y. Lee, and
Y. H. Khang, Optical properties of pseudobinary GeTe, Ge2Sb2Te5,
GeSb2Te4, GeSb4Te7, and Sb2Te3 from ellipsometry and density functional
theory, Phys. Rev. B 80, 115209 (2009).
[194] D. Ielmini, Phase change memory device modeling, published in S. Raoux
and M. Wuttig (Eds.), Phase change materials: Science and Applications,
Springer, New York, 2009.
[195] D. Ielmini and Y. Zhang, Evidence for trap-limited transport in the
subthreshold conduction regime of chalcogenide glasses, Appl. Phys. Lett.
90, 192102 (2007).
[196] C. B. Thomas, The temperature dependence of the non-Ohmic current and
switching characteristics of a chalcogenide glass, J. Phys. D: Appl. Phys. 9,
2587 (1976).
[197] K. Shportko, S. Kremers, M. Woda, D. Lencer, J. Robertson, and M. Wuttig,
Resonant bonding in crystalline phase-change materials, Nat. Mater. 7, 653
(2008).
[198] M. Le Gallo, M. Kaes, A. Sebastian, and D. Krebs, Subthreshold electrical
transport in amorphous phase-change materials, New J. Phys. 17, 093035
(2015).
[199] H. Satoh, K. Sugawara, and K. Tanaka, Nanoscale phase changes in
crystalline Ge2Sb2Te5 films using scanning probe microscopes, J. Appl.
Phys. 99, 024306 (2006).
[200] C. D. Wright, M. Armand, and M. M. Aziz, Terabit-per-square-inch data
storage using phase-change media and scanning electrical nanoprobes, IEEE
Trans. Nanotechnol. 5, 50 (2006).
[201] K. Tanaka, Smallest (~ 10 nm) phase-change marks in amorphous and
crystalline Ge2Sb2Te5 films, J. Non-Cryst. Solids 353, 1899 (2007).
[202] K. Kohary and C. D. Wright, Electric field induced crystallization in phase-
change materials for memory applications, Appl. Phys. Lett. 98, 223102
(2011).
[203] D. Krebs, S. Raoux, C. T. Rettner, G. W. Burr, M. Salinga, and M. Wuttig,
Threshold field of phase change memory materials measured using phase
change bridge devices, Appl. Phys. Lett. 95, 082101 (2009).
Bibliography
162
[204] R. Pandian, B. J. Kooi, G. Palasantzas, J. T. M. De Hosson, and A. Pauza,
Polarity-dependent reversible resistance switching in Ge-Sb-Te phase-
change thin films, Appl. Phys. Lett. 91, 152103 (2007).
[205] J. W. Park, Y. S. Kim, S. W. Ryu, J. H. Lee, J. Heo, and H. J. Kim,
Threshold switching in Te-SbO films for the selection device of crossbar
resistive memory applications, ECS Solid State Lett. 4, N5 (2015).
[206] A. Pradel, N. Frolet, M. Ramonda, A. Piarristeguy, and M. Ribes, Bipolar
resistance switching in chalcogenide materials, Phys. Status Solidi A 208,
2303 (2011).
[207] N. Ciocchini, M. Laudato, M. Boniardi, E. Varesi, P. Fantini, A. L. Lacaita,
and D. Ielmini, Bipolar switching in chalcogenide phase change memory,
Sci. Rep. 6, 29162 (2016).
163
List of Abbreviations
ADF
AFM
BF
C-AFM
CW
DRAM
DSC
EDX
FeRAM
FFT
FIB
FWHM
GeTeO
GID
GST
HAADF
HRS
HRTEM
HRSTEM
I-V
LAADF
LRS
MAADF
MRAM
NBD
PCM
PCRAM
PLD
ReRAM
SEM
SIMS
STEM
TEM
XRD
XRR
annular dark-field
atomic force microscopy
bright-field
conductive atomic force microscopy
continuous-wave
dynamic random access memory
differential scanning calorimetry
energy-dispersive X-ray spectroscopy
ferroelectric random access memory
fast Fourier transformation
focused ion beam
full width at half maximum
Ge-Te-O
grazing incidence diffraction
Ge2Sb2Te5
high-angle annular dark-field
high resistivity state
high-resolution transmission electron microscopy
high-resolution scanning transmission electron microscopy
current-voltage
low-angle annular dark-field
low resistivity state
middle-angle annular dark-field
magnetoresistive random access memory
nano beam diffraction
phase-change material
phase-change random access memory
pulsed laser deposition
resistive random access memory
scanning electron microscopy
secondary ion mass spectrometry
scanning transmission electron microscopy
transmission electron microscopy
X-ray diffraction
X-ray reflectivity
165
Acknowledgements
This dissertation would not have been possible without our team which contributed
to and gave valuable assistance during the period of this study. I would like to
acknowledge all of them. In the case I forgot someone to mention, I do apologize
for sincerely.
First and foremost, my utmost gratitude goes to my supervisor, Prof. Dr. Dr. h.c.
Bernd Rauschenbach, for having offered me the opportunity to work on this
interesting and challenging topic at the IOM and for his consistent support and
guidance throughout my Ph.D. study. Prof. Rauschenbach has been leading me to
the right direction to go when I faced various obstacles during the completion of
this research work. I really appreciate him propelling me in my research efforts,
being engaged in the deep physics regarding the experimental findings and
conclusions, as well as encouraging me to think creative and critical, which I am
benefitting from throughout my academic career, significantly. In addition, I would
also like to acknowledge my second supervisor, Prof. Dr. Stefan G. Mayr, for his
general support and valuable advices on my research.
I especially thank Dr. Jürgen W. Gerlach, for his friendly and patient offer of all
kinds of help with his broad knowledge, from teaching me how to use XRD to
assistance in vacuum technology, from results discussion to all manuscripts
corrections as well as the proof-reading of the present thesis. Moreover, his rigorous
style of scientific work has certainly influenced me.
I would like to thank Dr. Erik Thelander, as a senior member of the PLD-group, for
training me how to utilize the PLD technique, for discussing experiments and
results regularly, for having introduced me to the surroundings and for valuable
suggestions, which I benefitted from to quickly adapt to the working environment.
The other PLD-group members, Isom Hilmi and Philipp Schumacher, are
acknowledged as well.
I would like to thank Dr. Andriy Lotnyk and Ulrich Ross for performing the STEM
characterization, for the subsequent valuable discussions during the data analysis
and for proof-reading the manuscript. I also thank Agnes Mill for TEM lamellae
preparation.
Dr. Martin Ehrhardt I would like to thank for sharing his extensive knowledge on
laser applications with me and for always being willing to help me. I would also
like to thank Dr. Pierre Lorenz for performing COMSOL simulations and for
166
valuable discussions. I thank Dr. Joachim Zajadacz for performing E-beam
lithography, Eva Salamatin for performing optical interference microscopy, Dr.
Klaus Zimmer for comments on laser experiments. Dr. Tomi Smausz (University of
Szeged, Hungary) is acknowledged for help on the fs laser system.
I would also like to acknowledge the other collaborators, though some results have
not been presented in the final thesis: Prof. Dr. Bernd Abel and Dr. Tilmann Häupl
for performing the Raman spectroscopy measurements, Prof. Dr. Marius
Grundmann and Dr. Holger von Wenckstern (University of Leipzig) for performing
the impedance spectroscopy measurements, Dr. Carsten Bundesmann for
conducting the ellipsometry measurements, Dr. Heidemarie Schmidt and Tiangui
You (Technische Universität Chemnitz) for initially trying C-AFM measurements,
Dr. Wolfgang Knolle for trying electron beam irradiation of several samples.
I am also grateful to many more colleagues at the IOM for either contributing to this
work or for providing me with many useful suggestions: Dr. Ulrich Decker and
Nadja Schönherr for their help regarding UV-Vis spectroscopy; Ingrid Reinhardt for
performing DSC measurements; Dr. Frank Frost, Dr. Jan Lorbeer, Marina
Sarmanova and Dr. Ariyan Arabi-Hashemi for tips on AFM measurements. Dietmar
Hirsch for conducting SIMS analyses and teaching me how to operate EDX/EDS;
Petra Hertel for performing the magnetron sputtering, providing the sufficient
sample boxes and helping me in the clean-room; Toni Liebeskind for supplying me
with wafer material; Marco Müller for tips on electronic circuit design.
I feel very grateful for my former and current officemates: Dr. Annemarie Finzel,
Susann Liedtke, Michael Mensing, who warmly offered all kinds of help, and in
particular Christoph Grüner for his patient and consistent help, as well as his
assistance on experimenting with the annealing system and the resistance
measurements.
I also want to acknowledge all members of the mechanical workshop and of the
administration, as well as all other PhD candidates and staff at the IOM for
providing me with a convenient work atmosphere from their lots of times being
helpful. I gratefully acknowledge the “BuildMona” graduate school for providing
various training activities, which benefitted me with the concept of an
interdisciplinary approach in conducting research.
Last, but not least, I want to deeply express my gratitude to my family members and
all my friends. They gave me continuous support and encouragement during the
time I was pursuing my Ph. D. degree.
167
Curriculum Vitae
Person
Name: Xinxing Sun
Date of Birth: 23. December 1986
Place of Birth: Zhejiang, China
Nationality: China
Education
2013-2017 Ph.D. in Physics, Leibniz Institute of Surface Modification and
Leipzig University, Germany
Doctoral candidate and member of the graduate school
“BuildMoNa”
Title of the thesis: Phase Transformations and Switching
of Chalcogenide Phase-change Material Films Prepared
by Pulsed Laser Deposition
Supervisor: Prof. Dr. Dr. h.c. Bernd Rauschenbach
2009-2012 Master’s degree (M.Sc.) in nanophysics, East China Normal
University, China
Title of the thesis: Scattering and Antireflection Coatings
for Opto-electronic Device Application
Supervisor: Prof. Dr. Zhuo Sun
2004-2008 Undergraduate diploma in Electronic Engineering, Hangzhou
Dianzi University and Self-taught Higher Education Examinations
of Zhejiang Province, China
Work Experience
2007-2008 Engineer, State Key Lab of Silicon Materials and Department of
Materials, Zhejiang University, China
Working project: Transmission Electron Microscopy and
its Applications
Supervisor: Prof. Dr. Xiaobin Zhang and Youwen Wang
169
List of publications
1. X. Sun, U. Ross, J. W. Gerlach, A. Lotnyk and B. Rauschenbach, Nanoscale
bipolar electrical switching of Ge2Sb2Te5 phase-change material thin films,
submitted to Nature nanotechnology.
2. X. Sun, A. Lotnyk, M. Ehrhardt, J.W. Gerlach and B. Rauschenbach,
Realization of multi-level states in phase-change thin films by fast laser
pulse irradiation, Adv. Optical Mater. 1700169 (2017). DOI:
10.1002/adom.201700169.
3. X. Sun, M. Ehrhardt, A. Lotnyk, E. Thelander, J.W. Gerlach, T. Smausz, U.
Decker and B. Rauschenbach, Crystallization of Ge2Sb2Te5 thin films by
nano- and femtosecond single laser pulse irradiation, Sci. Rep. 6, 28246
(2016).
4. A. Lotnyk, S. Bernütz, X. Sun, U. Ross, M. Ehrhardt and B. Rauschenbach,
Real-space imaging of atomic arrangement and vacancy layers ordering in
laser crystallized Ge2Sb2Te5 phase change thin films, Acta Mater. 105, 1
(2016).
5. X. Sun, E. Thelander, J.W. Gerlach, U. Decker, B. Rauschenbach,
Crystallization kinetics of GeTe phase-change thin films grown by pulsed
laser deposition, J. Phys. D: Appl. Phys. 48, 295304 (2015).
6. X. Sun, E. Thelander, P. Lorenz, J.W. Gerlach, U. Decker and B.
Rauschenbach, Nanosecond laser-induced phase transitions in pulsed laser
deposition-deposited GeTe films, J. Appl. Phys. 116, 133501 (2014).
7. X. Sun, X. Chen, Z. Zhang and Z. Sun, Plasmon based antireflection
coatings containing nanostructured Ag and silica medium, Appl. Surf. Sci.
258, 3785 (2012).
8. L. Zhou, X. Chen, F. Zhu, X. Sun and Z. Sun, Improving temperature-stable
AZO-Ag-AZO multilayer transparent electrodes using thin Al layer
modification, J. Phys. D: Appl. Phys. 45, 505103 (2012).
9. B. Xue, R. Liu, Z. Xu and X. Sun, Synthesis, characterization and
photocatalytic activity of hierarchical spherical TiO2 nanostructures, Chinese
J. Inorg. Chem. 25, 1 (2009).
10. X. Sun, Z. Sun, L. Pan, M. Zhang, X. Chen, P. Guo. A measuring device for
thermal conductivity of solid materials, China Invention Patent, Publication
number: CN101949873A.
170
Conference contributions (talks and posters):
1. X. Sun, A. Lotnyk, M. Ehrhardt, P. Lorenz, J.W. Gerlach, B. Rauschenbach,
Accurately controllable phase transitions in PLD-deposited GeTe films for
multi-level memory application, talk, DPG spring meeting, Dresden (2017).
2. X. Sun, M. Ehrhardt, A. Lotnyk, E. Thelander, J.W. Gerlach, T. Smausz, U.
Decker, B. Rauschenbach, Phase transition on crystallization of PLD-
deposited Ge2Sb2Te5 films induced by nano- and femtosecond single laser
pulse irradiation, poster, DPG spring meeting, Regensburg (2016).
3. X. Sun, E. Thelander, J.W. Gerlach, B. Rauschenbach, Nanosecond laser-
induced phase transitions in PLD-deposited GeTe films, talk, DPG spring
meeting, Berlin (2015).
4. X. Sun, E. Thelander, J.W. Gerlach, U. Decker, B. Rauschenbach, Pulsed
laser deposited GeTe phase-change films on Silicon, poster, European
Phase Change and Ovonics Symposium, Amsterdam (2015).
5. X. Sun, E. Thelander, J. W. Gerlach, U. Decker, D. Hirsch, B.
Rauschenbach, Effect of O-incorporation on structural transitions of PLD-
deposited GeTe films, poster, Annual BuildMoNa Conference, Leipzig
(2015).
6. X. Sun, E. Thelander, H. Lu, J.W. Gerlach, B. Rauschenbach, Laser-induced
phase transitions of PLD-deposited GeTe films, poster, DPG spring
meeting, Dresden (2014).
7. X. Sun, E. Thelander, H. Lu, J.W. Gerlach, B. Rauschenbach, Ultraviolet
laser-induced phase transitions of PLD-deposited GeTe films, talk, Annual
Conference of the Graduate School BuildMoNa, Leipzig (2014).
8. X. Sun, E. Thelander, H. Lu, J.W. Gerlach, U. Decker, D. Hirsch, B.
Rauschenbach, Structural investigation of pulsed laser deposited phase-
change films, poster, 18th International Summer School on Vacuum,
Electron and Ion Technologies, Sozopol (2013).
171
Selbstständigkeitserklärung
Hiermit versichere ich, dass die vorliegende Arbeit ohne unzulässige Hilfe und ohne
Benutzung anderer als der angegebenen Hilfsmittel angefertigt, und dass die aus
fremden Quellen direkt oder indirekt übernommenen Gedanken in der Arbeit als
solche kenntlich gemacht wurden.
Ich versichere, dass alle Personen, von denen ich bei der Auswahl und Auswertung
des Materials sowie bei der Herstellung des Manuskripts Unterstützungsleistungen
erhalten habe, in der Danksagung der vorliegenden Arbeit aufgeführt sind.
Ich versichere, dass außer den in der Danksagung genannten, weitere Personen bei
der geistigen Herstellung der vorliegenden Arbeit nicht beteiligt waren, und
insbesondere von mir oder in meinem Auftrag weder unmittelbar noch mittelbar
geldwerte Leistungen für Arbeiten erhalten haben, die im Zusammenhang mit dem
Inhalt der vorliegenden Dissertation stehen. Außerdem versichere ich, keinen
Promotionsberater in Anspruch genommen zu haben.
Ich versichere weiterhin, dass die vorliegende Arbeit weder im Inland noch im
Ausland in gleicher oder in ähnlicher Form einer anderen Prüfungsbehörde zum
Zwecke einer Promotion oder eines anderen Prüfungsverfahrens vorgelegt und in
ihrer Gesamtheit noch nicht veröffentlicht wurde.
Ich versichere außerdem, dass keine früheren erfolglosen Promotionsversuche
stattgefunden haben.
Leipzig, 12. May 2017
Xinxing Sun