lha

rK

nai

baF

sei

tre

por

p tr

ops

nart

lac

irtc

ele

dna

ta

eh l

ort

noc

ot

sml

fi ni

ht

des

ab-

On

Z d

erey

alitl

uM

yti

srev

in

U otl

aA

1202

ecneicS slairetaM dna yrtsimehC fo tnemtrapeD

desab-OnZdereyalitluMtaehlortnocotsmlfiniht

tropsnartlacirtcelednaseitreporp

lharKnaibaF

LAROTCODSNOITATRESSID

seires noitacilbup ytisrevinU otlaASNOITATRESSID LAROTCOD 62 / 1202

ot smlfi niht desab-OnZ dereyalitluM tropsnart lacirtcele dna taeh lortnoc

seitreporp

lharK naibaF

fo rotcoD fo eerged eht rof detelpmoc noitatressid larotcod A eht fo noissimrep eht htiw ,dednefed eb ot )ygolonhceT( ecneicS

cilbup etomer a ta ,gnireenignE lacimehC fo loohcS ytisrevinU otlaA fo ht62 no loohcs eht fo 3eK llah erutcel eht ni dleh noitanimaxe

.enilno dedivorp eb lliw knil mooZ .21 ta 1202 hcraM

ytisrevinU otlaA gnireenignE lacimehC fo loohcS

ecneicS slairetaM dna yrtsimehC fo tnemtrapeD yrtsimehc slairetam cinagronI

Printed matter4041-0619

NO

RDIC

SWAN ECOLABE

L

Printed matter1234 5678

rosseforp gnisivrepuS dnalniF ,ytisrevinU otlaA ,nenippraK tiraaM

srenimaxe yranimilerP

ainotsE ,utraT fo ytisrevinU ,kiraA naaJ .me .forP ASU ,notgnihsaW fo ytisrevinU ,namsgreB .S divaD .forP tnatsissA

tnenoppO

ynamreG ,nedserD hcraeser slairetam dna etats dilos rof etutisni zinbieL ,hcsleiN suilenroK .forP

seires noitacilbup ytisrevinU otlaASNOITATRESSID LAROTCOD 62 / 1202

© 1202 lharK naibaF

NBSI 0-4820-46-259-879 )detnirp( NBSI 7-5820-46-259-879 )fdp( NSSI 4394-9971 )detnirp( NSSI 2494-9971 )fdp(

:NBSI:NRU/if.nru//:ptth 7-5820-46-259-879

segamI : mn 001 a fo noitces ssorc MET a si erucip revoc ehT enezneb-OnZ dna edixo muinimula evitucesnoc htiw mlif OnZ

morf sevruc tnemerusaem RRX dna DRX htiw rehtegot sreyal .siseht siht ni I noitacilbuP

yO aifarginU

iknisleH 1202

dnalniF

tcartsbA otlaA 67000-IF ,00011 xoB .O.P ,ytisrevinU otlaA if.otlaa.www

rohtuA lharK naibaF

noitatressid larotcod eht fo emaN seitreporp tropsnart lacirtcele dna taeh lortnoc ot smlfi niht desab-OnZ dereyalitluM

rehsilbuP gnireenignE lacimehC fo loohcS

tinU ecneicS slairetaM dna yrtsimehC fo tnemtrapeD

seireS seires noitacilbup ytisrevinU otlaA SNOITATRESSID LAROTCOD 62 / 1202

hcraeser fo dleiF yrtsimehC

dettimbus tpircsunaM 0202 rebmeceD 4 ecnefed eht fo etaD 1202 hcraM 62

)etad( detnarg ecnefed cilbup rof noissimreP 1202 yraurbeF 71 egaugnaL hsilgnE

hpargonoM noitatressid elcitrA noitatressid yassE

tcartsbAlacissalc nI .etats klub eht morf yltaerg reffid taht seitreporp evah nac slairetam neewteb secafretnI fo klub eht ni si ytirojam tsav eht elihw ecafretni eht ta era smota fo noitcarf ynit a ylno slairetam

secafretni fo tnuoma hgih yllaicfiitra na htiw slairetam reenigne ot ytilibapac ehT .lairetam eht gnillortnoc yb seitreporp lairetam eht roliat dna stceffe ecafretni eht yfilpma ot yawhtap a pu snepo

osla reyal yb reyal ro pets yb pets slairetam reenigne ot hcaorppa sihT .secafretni fo tnuoma eht dluow taht lairetam dirbyh a otni slairetam tnereffid yrev fo noitanibmoc dellortnoc a rof swolla

.seitreporp wen dna tnereffid yllatnemadnuf wohs nac hcihw dna yllarutan mrof tonresal deslup dna )DLM( noitisoped reyal ralucelom ,)DLA( noitisoped reyal cimota siseht siht nI ehT .seitisned ecafretni hgih htiw smlfi niht desab-OnZ reenigne ot dezilitu era )DLP( noitisoped nortcele noissimsnart dna )DRX( noitcarffid yar-x ,)RRX( ytivitcefler yar-x htiw desylana era smlfi )RTDT( ecnatcefleromreht niamod emiT .erutcurts lanretni rieht ot sdrager ni )MET( ypocsorcim llah a htiw derusaem era seitreporp lacirtcele eht ,ytivitcudnoc lamreht eht erusaem ot dezilitu si

dna OnZ enillatsyrcylop fo smlfi niht dereyal ni sucof eht si rettal ehT .putes tnemerusaem OnZaGnI suohproma 4 noitartnecnoc reirrac egrahc eht ni esaercni elbaredisnoc a hcihw ni

.detartsnomed eb dluoc ytisned ecafretni eht gniwollofOlA dna enezneb-OnZ ,xirtam OnZ a neewteb secafretni ehT x dirbyh a ni liated ni deiduts era sreyal

OlA/enezneb-OnZ/OnZ x eb nac OnZ ni sreyal eseht taht ,setartsnomed krow siht hcihw ni metsys sreyal niht yrev eseht ,revewoH .sreyal evitcnitsid mrof llits tey ,elucelom/mota elgnis a sa niht sa

OlA dna enezneb-OnZ fo x tca nac tub ,OnZ fo htworg latsyrc eht no tcapmi elttil evah ot dnuof era era selcyc DLM/DLA evitucesnoc erom ro 01 nehw htworg latsyrc OnZ rof sreirrab evitceffe sa

OlA hcae rof dezilitu x enezneb/OnZ ni ytivitcudnoc lamreht eht yllaniF .ylevitcepser reyal enezneb/ yltnacfiingis si metsys taht ni ytivitcudnoc lamreht eht rof esabatad eht ,desiretcarahc si smlfi niht emit tsrfi eht rof detroper era serutcurts dereyal ylralugerri rof seitivitcudnoc lamreht dna dednetxe

esabatad eht fo sisylana atad etairavitlum htiw sisylanA .smlfi niht dirbyh enezneb-OnZ/OnZ ni .ytivitcudnoc lamreht eht no tceffe decnuonorp tsom eht sah ytisned ecafretni eht taht smrfinoc

sdrowyeK DLP ,DLM ,DLA ,mlfi nihT ,slairetam dirbyH ,secafretnI

)detnirp( NBSI 0-4820-46-259-879 )fdp( NBSI 7-5820-46-259-879

)detnirp( NSSI 4394-9971 )fdp( NSSI 2494-9971

rehsilbup fo noitacoL iknisleH gnitnirp fo noitacoL iknisleH raeY 1202

segaP 711 nru :NBSI:NRU/fi.nru//:ptth 7-5820-46-259-879

Preface

This work was carried out in the laboratory of inorganic chemistry atAalto university and at the functional thin film laboratory at Hokkaidouniversity. I am very grateful for my supervisor Maarit for taking me asa doctoral student after I visited the lab in 2015 and the whole group ofcolleagues here that made (and still make) my time at Aalto very enjoyable.Special thanks goes of course to my long time office mate Jarno and to my„reactor mate“ Giovanni for their good company and help during the years.And also Antti who has been acting as the co-advisor.

Thank you Otto, Anne, Esko, Janne, Mikko and Tommi for organisingcommon events like bad movie nights, ski trips and pea soup cooking (Anne,Otto and Esko) as well as introducing me to the R reactor (Janne, Mikkoand Tommi). I tried to keep this spirit up and hope the reactor and thecooking will stay part of the group after my departure! Also thank you(in no particular order) Juho, Aida, Chris, Linda, Elli, Taneli, Nina, Ira,Jenna, Amr, Ramin, Safdar and Niko for organising/participating in thepeasoup cooking, the chemistry sports days, the pikkujoulus etc. you weregreat colleagues!

I also want to thank Prof. Hiromichi Ohta and his lab namely YuzhangWu, Gowoon Kim, Hai Jun Cho, Ishino Matsumi, Yugo Takashima, TakakiOnozato, Qian Yang, Yu-Qiao Zhang, Gong Lizhikun, Takashi Fujimoto,Wei Mian and all the others that took so much time and such great care ofme during my time in Japan.

Also a big thank you for my parents, sister and grandma for their supportduring all my (long) education and in Finland thank you Ninja and theUnreality kuoro for providing distraction from work.

Last but not least I want to thank Annika, Sirje, Anita, Kimmo, Kaisa andIsabella for all their technical help with contracts, ordering and keepingthe university running!

For everyone that should be but isn’t in this list — sorry!I’m very grateful for the personal financial support from the Aalto school

of chemical engineering, the Kaute and the Sasakawa foundations.

Helsinki, February 19, 2021,

Fabian Krahl

i

Contents

Preface i

Contents iii

List of Publications v

Author’s Contribution vii

Abbreviations ix

Symbols xi

1. Introduction 1

2. Background 52.1 Electrical conductivity at interfaces . . . . . . . . . . . . . 52.2 Thermoelectric materials and thermal conductivity . . . 62.3 ZnO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102.4 Atomic and molecular layer deposition . . . . . . . . . . . 112.5 Pulsed laser deposition . . . . . . . . . . . . . . . . . . . . 14

3. AlOx & benzene layers in ZnO 173.1 Analysing the structure of layered films . . . . . . . . . . 20

3.1.1 XRD & GIXRD . . . . . . . . . . . . . . . . . . . 203.1.2 XRR . . . . . . . . . . . . . . . . . . . . . . . . . . 223.1.3 TEM . . . . . . . . . . . . . . . . . . . . . . . . . . 24

3.2 Structure of ZnO/ZnO-benzene and ZnO/AlOx superlatticefilms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25

4. Electrical properties in layered a-IGZO/ZnO thin films 294.1 InGaZnO4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304.2 Characterising physical properties . . . . . . . . . . . . . . 31

4.2.1 Transport measurements . . . . . . . . . . . . . . 314.2.2 UV-visible spectrophotometry . . . . . . . . . . . 32

iii

Contents

4.3 Charge carriers at interfaces . . . . . . . . . . . . . . . . . 32

5. Thermal conductivity in layered thin films 375.1 Time domain thermoreflectance . . . . . . . . . . . . . . . 385.2 Multivariate data analysis . . . . . . . . . . . . . . . . . . 395.3 Thermal conductivity of ZnO/ZnO-benzene thin films . . . 40

6. Conclusions and Outlook 45

7. Appendix 47

References 51

Publications 63

iv

List of Publications

This thesis consists of an overview and of the following publications whichare referred to in the text by their Roman numerals.

I F. Krahl, Yanling Ge and Maarit Karppinen. Characterization of Multi-layer ZnO/AlOx/benzene Thin-Film Structures Grown through ALD/MLD.Semiconductor Science and Technology, 36, 025012, 2020. DOI: https://doi.org/10.1088/1361-6641/abcee2.

II Fabian Krahl, Yuzhang Wu, Hai Jun Cho, Maarit Karppinen and Hi-romichi Ohta. Spontaneous Generation of Carrier Electrons at the Inter-face between Polycrystalline ZnO and Amorphous InGaZnO4. AdvancedElectronic Materials, 6, 10, 2000404, 2020. DOI: https://doi.org/10.1002/aelm.202000404.

III Fabian Krahl, Ashutosh Giri, John A. Tomko, Tommi Tynell, PatrickE. Hopkins and Maarit Karppinen. Thermal Conductivity Reductionat Inorganic–Organic Interfaces: From Regular Superlattices to Irreg-ular Gradient Layer Sequences. Advanced Materials Interfaces, 5, 11,1701692, 2018. DOI: https://doi.org/10.1021/acs.jpcc.0c06461.

IV Fabian Krahl, Ashutosh Giri, Md Shafkat Bin Hoque, Linda Sederholm,Patrick E. Hopkins and Maarit Karppinen. Experimental Control andStatistical Analysis of Thermal Conductivity in ZnO-Benzene MultilayerThin Films. Journal of Physical Chemistry C, 124, 45, 24731–24739,2020. DOI: https://doi.org/10.1021/acs.jpcc.0c06461.

v

Author’s Contribution

Publication I: “Characterization of Multilayer ZnO/AlOx /benzeneThin-Film Structures Grown through ALD/MLD”

F. Krahl and M. Karppinen planned the sample structures. All sampleswere deposited with ALD/MLD and characterised with XRR and XRD by F.Krahl. Y. Ge prepared the samples for and operated the TEM to obtain theTEM images. F. Krahl wrote the manuscript with support of M. Karppinenand Y. Ge.

Publication II: “Spontaneous Generation of Carrier Electrons at theInterface between Polycrystalline ZnO and Amorphous InGaZnO4”

H. Ohta and F. Krahl planned the sample structures. F. Krahl depositedand analysed 9 of the 11 samples with with help of Y. Wu and H. J. Cho.The pure ZnO and pure a-InGaZnO4 samples (samples 10 and 11) weredeposited and characterised by Y. Wu. The first first draft of the manuscriptwas written by F. Krahl. All authors contributed critically to the finalversion which was written by H. Ohta.

Publication III: “Thermal Conductivity Reduction atInorganic–Organic Interfaces: From Regular Superlattices toIrregular Gradient Layer Sequences”

F. Krahl and M. Karppinen planned the samples. F. Krahl deposited andcharacterised all samples with the exception of the TDTR measurementswhich were conducted by A. Giri and J. A. Tomko. The article was writtenby F. Krahl with critical input and comments from all other authors

vii

Author’s Contribution

Publication IV: “Experimental Control and Statistical Analysis ofThermal Conductivity in ZnO-Benzene Multilayer Thin Films”

F. Krahl and M. Karppinen planned the sample structures. F. Krahldeposited and characterised all samples with the exception of the TDTRmeasurements. The TDTR measurements were conducted by M. S. B.Hoque with support of A. Giri. L. Sederholm did the MVDA analysis of thesamples with support from M. Karppinen and F. Krahl. F. Krahl wrote themanuscript with support of all other authors and original input from L.Sederholm (MVDA) and M. S. B. Hoque (TDTR).

viii

Abbreviations

2DEG Two dimensional electron gas (typically at an interface)

a-IGZO Amorphous indium gallium zinc oxide (InGaZnO4)

ALD Atomic layer deposition

DEZ Diethylzinc, Zn(C2H5)2

GIXRD Grazing incidence X-ray diffraction

GM Gradient material

HQ Hydroquinone, C6H4(OH)2

IR Infrared light

MLD Molecular layer deposition

PLD Pulsed laser deposition

SL Superlattice

TCO Transparent conductive oxide

TDTR Time domain thermoreflectance

TE Thermoelectric

TMA Trimethyl-aluminium, Al(CH3)3

UV-Vis Ultraviolet and visible light

XRD X-ray diffraction

XRR X-ray reflectivity

ix

Symbols

C Phonon heat capacity

c Speed of light

d Distance

E Energy

h Planck constant

H Hamiltonian operator

K Shape factor (Scherrer equation)

kB Boltzmann constant

L Lorentz number

M Molar Mass

m∗ Effective mass

NA Avogadro number

n Charge carrier concentration

nDiff. Diffraction order

q Charge of a charge carrier (in this thesis the electron charge)

rel. Electron radius

S Seebeck coefficient

T Temperature

V Voltage

Z Atomic number

zT Figure of merit for a thermoelectric material

β Peak broadening

Θ Diffraction angle

Θc Critical Angle

κ Thermal conductivity

κl Thermal conductivity due to lattice vibrations

xi

Symbols

κe Thermal conductivity due to electrons

λ Wavelength

μ Charge carrier mobility

ν Phonon velocity

ρ Density

ρel. Electron density

σ Electrical conductivity

τ Crystallite diameter

ψ Wavefunction

xii

1. Introduction

Materials and the ability to understand and control them are a definingfeature of humanity. From the stone, bronze and iron age, over to theproduction of paper and gunpowder to oil refining up to nuclear fission andthe current information age, that is sometimes also dubbed silicon age, theunderstanding of the material world has steadily increased.

A great part of the recent technological advances are due to miniaturiza-tion, especially in computer hardware, resulting in an increase in compu-tational power on smaller and smaller circuits. This would not have beenpossible without the rapid development of methods in physics, chemistryand related sciences to control and characterise materials down to thenanoscale. At the current frontier of science about the miniaturization ofstructures are insights and capabilities to control structures down to thesize of a nanometre. A true control of materials on that level touches thefoundations of chemistry, because atomic bonds, unit cells of crystals andmolecules are at the lower end of this range.

A subset of nanotechnology is that of thin films which are the main focusof this thesis. Thin films of well known materials are crucial in electronicssuch as computer chips and have a great impact also on catalysts and opti-cal coatings. Due to the progressing capabilities to control the film growthdown to the scale of unit cells different materials can be deposited togetheron such a small scale, that it becomes possible to engineer essentially newmaterials. An example are materials with artificial interfaces, createdby stacking different materials layer by layer on a molecular level. Withthat kind of control it is possible to combine and tailor physical propertiesas well as engineer materials with completely new properties. At theinterfaces of nanometre thin layers, new questions arise that need to beanswered to understand the nature of these materials. The interfaces couldhave abrupt borders that are strictly separated, or smooth and gradualtransitions between two materials. Thin interfaces could be strong, pinholefree layers and effective separators between two materials on either side,or they could be more like a net with cavities, through which the materialsat both sides are connected. The answers are specific to the materials of

1

Introduction

the layers and the conditions under which they were formed.The amount of possible materials that could be combined is staggering;

this thesis focusses on ZnO-based thin films with AlOx and ZnO-benzenesublayers. To create these multilayered thin films this thesis utilizedatomic layer deposition (ALD) and molecular layer deposition (MLD). ALDwas invented by Tuomo Suntola [1] and has since become an importanttechnique for semiconductor microelectronics and research [2]. MLD is avariation of ALD that deposits whole molecules instead of single atomseach cycle and is used in this thesis to include ZnO-benzene layers intoZnO thin films and thus enabling deposition of hybrid inorganic-organicthin films.

Zinc oxide is a transparent conductive oxide (TCO) and a common dopantfor it is aluminium which can substitute Zn atoms in the ZnO lattice oroccupy interstitial spots, in this work AlOx layers in ZnO were deposited.Al in ZnO stands in strong contrast to benzene, which is also investigated inthis work. Benzene is an organic aromatic molecule that can not be easilyintegrated into the ZnO lattice. Consequently the ZnO/AlOx system couldbe in stark contrast to ZnO/ZnO-benzene system. An investigation of thesechemically very different interface layers has been conducted in PublicationI. As transparent conductive oxide, the electrical conductivity of ZnO isoften investigated, at the same time interfaces of metal oxides have shownexciting electric properties. At the interface between artificially layeredoxides the electrical conductivity can vary considerably, even forming a socalled two dimensional electron gas (as for example in LaAlO3/SrTiO3 [3]).The possibility to utilize such interface effects to increase the electricalconductivity at interfaces is especially interesting for materials that havethe need for a high electrical conductivity, but have other restrictions suchas transparency or benefit from anisotropy. This is the topic of PublicationII in which superlattices of ZnO and amorphous indium gallium zincoxide (a-IGZO) are deposited with pulsed laser deposition (PLD). a-IGZOitself is an interesting TCO that can have extraordinarily high electronmobilities [4]. ZnO and InGaZnO4 also naturally form crystalline layeredstructures but within this thesis the focus is prospecting for possibleinterface effects between polycrystalline-amorphous interface superlatticestructures built from polycrystalline ZnO layers and amorphous InGaZnO4

layers.Lastly interfaces can significantly reduce the thermal conductivity of

a material and together with the reasonably good electrical conductivitythis makes layered thin films based on ZnO a great model system for ther-moelectric (TE) materials, where balancing and possibly even decouplingof the thermal and electrical conductivity could lead to great gains [5,6].To better understand how the thermal conductivity is suppressed andhow this effect can be controlled to tailor the thermal conductivity inthe ZnO/ZnO-benzene hybrid superlattice system an extensive sample

2

Introduction

series is designed and systematically investigated. Notably this work alsoinvestigates samples that deviate from the strictly repetitively orderedsuperlattice structures and also considers materials that have varyinginterface spacings which are called gradient materials in this thesis.

The ZnO/ZnO-benzene system combines ZnO, a well known inorganicTCO, with benzene interfaces, the most basic organic aromatic ring struc-ture. Together they form a hybrid material that can not be synthesizedwithout this layered approach. It has been suggested that nearly the entirespectrum of heat carrying phonons in ZnO scatters at the inorganic-organicinterface between ZnO and benzene [6], making research on this systemvery interesting. Publication III investigates the effect that deviation fromthe regular superlattice structure has on the thermal conductivity whereasPublication IV adds thicker and thinner samples to the set and utilizesmultivariate data analysis to find the main factors for the suppression inthermal conductivity.

The effects of interfaces and their layered structure is the common ele-ment linking everything together and the present thesis tries to illuminatethese effects in the ZnO systems with ZnO-benzene, a-IGZO and AlOx

layers. An overview on the challenges of those questions is given in thenext Chapter that introduces some of the scientific background this thesisis based upon, together with a short introduction to ZnO as well as thedeposition methods, ALD/MLD and PLD, utilized in this work to createthese intricate layered thin films.

3

2. Background

To understand the work and the challenges that are tackled in this thesis,some background in this field is necessary. Basic concepts about theelectrical and thermal conductivity at interfaces are at the foundation ofthis thesis and the following sections aim to give a brief overview on them.

2.1 Electrical conductivity at interfaces

Thin films differ in several ways from bulk materials. This is important tounderstand because where surfaces and interfaces in bulk materials makeup just a tiny fraction of the material and are sometimes even omitted intheir discussion, in thin films they make up a large fraction of the materialand can’t be neglected.

The electrical conductivity (σ) is classically expressed as the product ofthe charge carrier concentration (n), the charge (q) and the mobility (μ) ofthe charge carriers.

σ = nqμ (2.1)

In this thesis the charge carriers are electrons and the discussion willbe limited to the solid state when discussing electronic properties. Thecharge of an electron is fixed (1e ≈ 1.6021 · 10−19C), but the charge carrierconcentration (n) and mobility (μ) are determined by the electronic bandstructure of the material. In principle this electronic „band“ is the combi-nation of the allowed quantum states of the involved electrons describedby the Schrödinger Equation:

H|Ψ〉 = E|Ψ〉 (2.2)

With H being the Hamiltonian operator, Ψ the wavefunction and E theenergy of the electron state. For all but the simplest situation (a singlehydrogen atom with 1 electron) the equation can’t be solved analyticallybut with the quickly developing computational power and methods withinthe last decades increasingly good approximations can be made. Howeverone fundamental assumption in many of these calculation is periodicity

5

Background

(Blochs theorem) meaning that there is a repeating unit (a unit cell). Thisis tremendously important because it means that not every single electronin a solid has to be taken into account and the Schrödinger equation canbe approximated for the electrons within this repeating unit. Obviouslythis theorem works especially well for crystalline materials. Deviationsoccur at interfaces between two materials (including the surface, the inter-face between the solid and the surrounding atmosphere) because at theinterface periodicity breaks down and the quantum states of the electronsare different from those within the material. Pioneering theoretical workon these surface states was done nearly 90 years ago by Igor Tamm [7] andWilliam Shockley [8], shortly after Felix Bloch introduced his descriptionon the Bloch waves in a continuous medium [9]. Ultimately these interfacestates are the reason why interfaces can have fundamentally differentproperties than the bulk of the material and why engineering materialswith many interfaces is so interesting.

An aspect, that is closely entwined with the effect of surface states, isthat the band structure can be limited and it’s amount of quantum statescan be significantly reduced if the structure in the physical space is thinenough. Then the amount of possible quantum states isn’t described bestby a continuous band but by discrete quantum states. This confinement ofelectrons into small areasis called quantum confinement [10]1.

For the electric conductivity this means that the charge carrier concen-tration as well as mobility at interfaces and very thin layers can differsignificantly from their values in the bulk. A prominent example of thisare the so called 2D electron gases (2DEG), in which electrons can movefreely within two dimensions, but are confined in the third.

More specific examples of such systems will be given in Chapter 4 to-gether with studies on ZnO/a-IGZO superlattice structures.

2.2 Thermoelectric materials and thermal conductivity

A related field with great hopes for major breakthroughs with the helpof nanostructures is the field of thermoelectricity. More than 60% ofthe worlds energy production is estimated to be lost, most of it as wasteheat [11].

Thermoelectric (TE) devices can generate electricity from temperaturegradients and thereby from waste heat, at their heart are thermoelectricmaterials. That a temperature gradient can lead to a voltage differencewithin a material was discovered by Thomas Johann Seebeck already

1Not limited to 2-dimensional layers, nanorods and nanodots utilize quantumconfinement in one and zero dimensions and also not necessarily limited toelectrons.

6

Background

in 18212 [13] and is named thermoelectric effect or Seebeck effect. Thecoefficient relating the voltage generated with the temperature differenceis named Seebeck coefficient, S after its discoverer and defined as follows:

S =ΔV

ΔT(2.3)

A good thermoelectric material should not only have a high Seebeckcoefficient, but also a decent electrical conductivity. The product σS2 istherefore sometimes called the power factor, a value that has a greateffect on the amount of energy that could be extracted. Lastly the thermalconductivity (κ) is important because it determines how well a temperaturegradient (ΔT in equation 2.3) can be kept across the material. While ahigh Seebeck coefficient and the electrical conductivity are beneficial for agood thermoelectric material, the thermal conductivity is ideally very lowto enable a steep temperature gradient. This relation is expressed in thedimensionless figure of merit for thermoelectric materials.

zT =σS2

κT (2.4)

The higher the zT value is the better the material is suited for thermoelec-tric applications. The same figure of merit can be done for thermoelectricdevices in which case it is often written with a capital Z to distinguishthe device from the material. For actual thermoelectric devices n-type andp-type legs connected in series are the standard configuration. [14]

Basically all materials show thermoelectric behaviour, alas the effect isvery weak in most cases and to increase the zT value or find a new materialwith a high zT value is a challenge. Currently the most widespreadmaterials are doped Bi2Te3 alloys, but their efficiency is rather low besidesome other obvious problems such as bismuth and tellurium being toxicheavy metals.

The core problem in finding a better thermoelectric material is finding away to maximize z = σS2/κ. In metals and degenerate semiconductors S

can be derived from the band structure and expressed as follows [15]:

S =8π2k2B3qh2

m∗T( π

3n

) 23 (2.5)

kB is the Boltzmann constant, h is the planck constant, q is the chargeof the charge carrier, m∗ is the effective mass of the charge carrier andn is the charge carrier concentration. The only variables in equation 2.5are m∗, T and n. As in the equation for the zT value (equation 2.4) hightemperatures are beneficial.

The effective mass of the charge carrier would be a way to increase theefficiency but unfortunately a higher m∗ often leads to lower mobilities

2Alessandro Volta is also sometimes mentioned as the first to discover the ther-moelectric effect already in 1794 [12]

7

Background

colliding with the need for good electrical conductivities as expressed inequation 2.1. [16,17]. A similar, more direct, issue arises with the chargecarrier concentration n. According to equation 2.5 n should be as low aspossible for a high Seebeck coefficient, but the electrical conductivity isdirectly linked to n. Looking at equation 2.4 shows that a high electricalconductivity as well as a high Seebeck coefficient would be ideal. But for ahigh σ the charge carrier concentration n should be high whereas for a highS it should be low. So a compromise must be made because too low carrierconcentrations will limit the efficiency due to a low conductivity whereastoo high carrier concentrations will lead to a Seebeck coefficient that is solow that barely any voltage is generated from a heat gradient. This showsthat S and σ have a fundamental connection through the charge carriersthat prevents them to be maximized independently.

A similar problem occurs between σ and κ. The ratio of thermal andelectrical conductivity in metals is proportional to the temperature, asexpressed in the Wiedemann-Franz law

κ

σ=

π2

3

(kBq

)2

T = LT (2.6)

kB is the boltzmann constant and L = π2k2B/3q2 is the Lorentz number.

The Wiedemann-Franz law shows that σ and κ are proportional to eachother, yet ideally high electrical but low thermal conductivity is required(compare with equation 2.4).

The challenge with thermoeletric materials is that in order to maximizethe efficiency the electrical conductivity and the Seebeck coefficient shouldbe as high as possible while the thermal conductivity should be minimized.For a high electrical conductivity a high concentration of charge carriers isdesired, but that contradicts high values for the Seebeck coefficient, whichbenefits from low carrier concentrations. Maximizing one while minimizingthe other can’t easily be done, this is the conundrum of thermoelectrics,the variables are connected and maximizing one will often come at a costfor the others.

Fortunately the Wiedemann-Franz law is valid for metals, but in in-sulators the contribution of charge carriers to κ is minuscule, anotherheat transfer mechanism is dominating: Lattice vibrations, in line withquantum chemistry these vibrations can also be described as quasipar-ticles — phonons. Phonons can be very effective in the transmission ofheat, in fact the material with the highest known thermal conductivity isdiamond, a classical insulator that conducts heat by phonons not by chargecarriers. [18]

Obviously classic insulators are not great thermoelectric materials be-cause their electrical conductivity is, by definition, rather low. On theother end of the spectrum are classic metals, which are typically also poorthermoelectric materials because their concentration of charge carriers is

8

Background

high and accordingly the Seebeck coefficient tends to be low (eq. 2.5) andthe thermal conductivity is, in accordance to the Wiedemann-Franz law(eq. 2.6) very high, consequently their zT value tends to be quite small.

The best thermoelectric materials are typically found among semiconduc-tors that have acceptable levels of electrical conductivity and yet can havea low thermal conductivity dominated by phonons not the charge carrierswhich means that the thermal conductivity is partly decoupled from theelectrical conductivity. The thermal conductivity can be expressed as thesum of the thermal conductivity that is due to phonons, κl, and the partthat is due to charge carriers, κe.

κ = κl + κe (2.7)

Especially κl is interesting because minimizing κl will not directly affectthe charge carriers. It is also linked to the materials structure as strongbonds in a regular crystal lattice such as diamond are ideal pathways forphonons while unordered amorphous substances such as glass typicallyshow low phonon conductivities. The thermal conductivity due to latticevibrations, κl, can be expressed by kinetic theory as:

κl =1

3Cνlmfp (2.8)

Here C is the heat capacity, ν the average phonon velocity (often approx-imated by the speed of sound in the material) and lmfp the mean freepath of a phonon, indicating what distance a phonon travels before it getsscattered [19].

To express the need for materials that behave like a glass for phonons,but like a crystal for electrons the nicely descriptive term „phonon glass,electron crystal“ has been coined [20]. This is currently one of the mostpromising ways to improve thermoelectric materials and there are manyapproaches to this including alloying, isoelectric substitution and clathratestructures. [16,21,22]

Also layered materials can have good thermoelectric properties, as in-terface scattering can be a very effective way to suppress thermal con-ductivity [21, 23, 24]. The layering can be due to the crystal structure(e.g. in complex oxides [25]), but for this thesis layered nanostructuresare achieved by artificial growth of different layers. Introducing layersperpendicular to the temperature gradient would ideally act as strongphonon scatterers but have as little impact as possible on charge carriers(cross plane). More examples will be given in Chapter 5 that investigatesthe thermal conductivity of ZnO/ZnO-benzene samples fabricated withinthis thesis. Thermoelectric properties are not the focus of this thesis butare a direct connection between two of the main topics: Electrical interfaceeffects and the control of thermal conductivity by interfaces.

9

Background

2.3 ZnO

Zinc oxide crystallizes readily under almost any deposition or growth con-dition. The most prevalent structure is the hexagonal wurtzite structureshown in Figure 2.1. Also a cubic zinc blende structure of ZnO is known,

Figure 2.1. Crystal structure of ZnO along the crystallographic a and c-axis. The pictureswere generated using Vesta [26] and the CIF file on the crystallography opendatabase [27] originally from Weber [28].

but was not observed within this work. In the hexagonal form ZnO ispiezoelectric along the distinguished c-axis. ZnO is used in many fields be-cause it shows a wide array of interesting physical properties, is relativelyabundant, stable and also non-poisonous [29,30]. For this work the focusis on the electrical and thermal properties of ZnO (always in the hexagonalwurtzite structure unless otherwise noted).

ZnO is a semiconductor with a wide (and direct) band gap and is alsostable at high temperatures, making it an interesting candidate for ther-moelectric research. As described earlier, semiconductors tend to be themost promising candidates for thermoelectric purposes and the efficiencyof the thermoelectric conversion is better at higher temperatures, ZnOfulfills these requirements. Another interesting aspect is that the ther-mal conductivity in ZnO is largely dominated by phonons, not electrons,which means that one of the big challenges of thermoelectric materials,the entanglement of electric and thermal conductivity, is mitigated a bitcompared to materials that have the thermal conductivity dominated bytheir charge carriers [31–37].

ZnO is also interesting as transparent conductive oxides and can bedoped (e.g. with aluminium and/or magnesium) for higher electrical con-ductivities [29,30]. It should be noted that Zn is much also more abundantthan indium, which is currently used for the predominant TCO indium tinoxide (ITO) and non-toxic.

One of ZnO’s main issues as semiconductor is that it is an intrinsicn-type conductor and so far no easily accessible and stable p-doping of

10

Background

ZnO is known. The nature of the intrinsic n-doping is still discussed inliterature — oxygen vacancies, interstitial Zn atoms and/or interstitialhydrogen are the main suspects for the natural n-doping in ZnO [29].Finally of great parctical importance for this work was also that ZnO has awell established deposition processes for atomic layer deposition [38] andpulsed laser deposition [39].

2.4 Atomic and molecular layer deposition

Atomic layer deposition (ALD) is uniquely suited for layer engineeringbecause it allows, in principle, an extraordinary level of layer sequencecontrol down to monoatomic layers under comparatively mild depositionconditions and with high conformality also on structured substrates [2,40].The mild deposition conditions and precise layer control were critical forthis thesis research and most of the work herein depends strongly on ALDand it’s counterpart, molecular layer deposition (MLD) for the organiclayers.

Atomic and molecular layer deposition techniques have quickly gainedtraction in science and industry, not only but especially, because they canyield conformal and pinhole free electric barrier layers of high-k oxides onirregular surfaces essential in the semiconductor industry [41].

ALD was invented in Finland by Tuomo Suntola [1] and independentlyalso in the Soviet Union [42]. It is mostly based on chemical reactions atthe surface and often used for binary metal oxides, sulfides and nitrides.A first precursor is pulsed as a gas into the reaction chamber with thesubstrate, where it reacts and sticks to the substrate surface. Here it isof utmost importance that the reaction stops once the surface sites arefully covered, meaning the reaction is self limiting. The excess precursoris then pumped out of the chamber before the second precursor is pulsedinto the chamber and reacts with the previous precursor on the surface.This reaction must be self limited so that no further reaction occurs oncethe surface is covered in the reaction product of the latest precursor. Thesimplest ALD processes need only two precursors, but depending on thedesired film more precursors can be pumped repeatedly and in sequenceinto the chamber.

In practice there is always a purging step between the precursor pulses toensure the previous precursor really is only on the surfaces and not in thegas phase any more, which could lead to particle formation by spontaneousreaction between the precursors in the gas phase. These separated steps ofpulsing an purging are cycled, while this ensures the excellent thicknesscontrol it also makes the whole process rather slow. For ZnO the growth isusually ≤ 2 Å/cycle, the cycle time heavily depends on the setup and wasa bit below 10 seconds (most of it purging) for the work conducted within

11

Background

this thesis. A sketch of the process and photos of the reactor employed inthis research are shown in Figure 2.2.

n cycles

Pulse

Pulse

Purge

Purge

Diethylzinc

Water

Precursor

Nitrogen

Purging &transport gas

Ethane

Reaction products

ZnO[ ]n

Substrate Substrate

SubstrateSubstrate

(a) A sketch of the basic ZnO ALD process.

(b) The ALD reactor from theoutside

(c) The ALD deposition chamber

Figure 2.2. The basic ZnO process sketched (a) and pictures of the reactor used (PicosunR-100).

Molecular layer deposition works with the same principle, but insteadof a single atom a complete molecule, based on an organic precursor, isdeposited on the surface. In both cases ALD and MLD the temperaturerange at which the precursors can be used is critical to prevent decom-position and ensure the self limiting characteristics of the reaction. Thecombination of ALD and MLD allows a great structural freedom to combineinorganic and organic materials at the nanometer scale into new hybridmaterials [43,44]. It should be mentioned that the exact reaction mecha-nisms during the ALD/MLD and the surface saturations of the differentprecursors are subject of active research [45–49] and also differ with eachprecursor, but the basic assumption — a layer by layer deposition of self

12

Background

limiting surface reactions, is undisputed.Because the precursor pulse times can be controlled and the precursor

can diffuse through the chamber until it hits an unreacted surface site,complex shapes can be coated. The self limiting surface reaction ensuresmore exposed parts don’t experience faster growth resulting in the excellentthickness control with high conformality. Another advantage is that thedeposition depends on the chemical reaction between the precursors on thesurface and as long as the precursors are reactive enough and in the gasphase, no high temperatures are necessary which enables the depositionof organic molecules.

The basic reactor setup is relatively simple: The sources with the precur-sor must be connected to the deposition chamber and in some way (e.g. avalve) their flow into the chamber must be controlled. The chamber itselfis typically at pressures of a few hPa due to the constant changing flow ofpurging gas (that often also acts as transport gas for the precursor). Lastlyat the exhaust are particle traps and/or scrubbers to get rid of unreactedprecursors. A challenge in ALD can be the purity of the films, because thematerials are deposited with a chemical reaction it is nearly impossibleto completely prevent reaction products from being trapped in the film.Small levels of impurities stemming from the precursor can nearly alwaysbe found if one looks close enough.

Another big challenge in ALD can be finding the correct precursors, asthey must fulfil rather strict requirements, they must have partial pressurein the gas phase, they must of course react at the surface but they also mustbe stable enough, in the deposition condition to guarantee the self limitingreaction once the surface is saturated. Often these materials are highlyreactive metal organic molecules (for example diethylzinc — DEZ andtrimethylaluminium — TMA used in this thesis) that need careful handling.For the work in this thesis, namely the ZnO/ZnO-benzene hybrid filmshydroquinone (HQ, C6H4(OH)2) was utilized as precursor. Hydroquinone isa benzene molecule with two hydroxy groups in para-position which reactswith ethylzinc surface groups under the formation of ethane linking theZn atom and the benzene molecule via an oxygen bridge Zn–C2H5(Surface) +C6H4(OH)2(gas) −−→ C2H6(gas) + Zn – O – C6H4 –OH(Surface). The challengehere is that HQ is a powder at room temperature and has a very low partialpressure, which is an issue because ALD and MLD are gas-phase reactions.To overcome this the HQ source had to be kept at a constant temperatureof 150°C during deposition and pulse times for the HQ half cycle were 15seconds, much longer than those for DEZ (0.5 seconds pulse time).

Finding the right deposition conditions for new precursors can be chal-lenging, however since this thesis predominantly studies the effects of newnanostructures on known materials, well known precursors such as TMAand DEZ could be used and also HQ had previously been reported as aprecursor [50], hence the deposition conditions were not an issue for this

13

Background

thesis. Because ALD is not a "line of sight" deposition method but depositspractically everywhere large amounts of precursors react on the chamberwalls or are pumped through the chamber without reacting, this meansALD/MLD deposition has a relatively large material consumption and thechamber needs frequent cleaning.

The self-limiting surface reaction is the key for the excellent layer controlbecause through the order in which the reactants are pulsed into the depo-sition the layer sequence is determined, together with the comparativelylow deposition temperatures ALD and MLD are extremely well suited tofabricate nanolayered thin films of a wide variety of materials.

2.5 Pulsed laser deposition

One material investigated in this thesis is InGaZnO4 (IGZO), a ternarymetal oxide and for ternary or even quarternary compounds ALD processesquickly become complicated because they need four compatible precursors.Recently the possibilities for ternary or even quarternary compounds withALD have been demonstrated even for IGZO [51–53]3

More common for ternary or quarternary compounds are physical depo-sition methods such as pulsed laser deposition (PLD). This has severaladvantages, the desired material can be synthesized before deposition, itis quicker and has a great freedom to vary the oxygen partial pressureduring deposition. The control of the partial pressure of oxygen is of specialimportance for IGZO because the conductivity of IGZO (amorphous or not)is closely linked to oxygen vacancies and the partial pressure of oxygenduring deposition has a strong influence on those. To get high conductivi-ties in IGZO systems typically the oxygen partial pressure is very low tofacilitate oxygen vacancies [54], to have this control pulsed laser depositionwas utilized in Publication II to deposit ZnO/IGZO structures.

In contrast to ALD pulsed laser deposition (PLD) is a physical depositionmethod and no chemical reaction takes place during the process. One ormore target materials (ZnO and IGZO in this case) are placed in a vacuumchamber below the substrate (e.g. a glass slide or a silicon wafer). Then alaser shoots at the target material with enough power to evaporate someof it. The evaporated material forms an evaporation plume and hits thesubstrate where it is deposited as thin film.

The target is typically rotated to minimise the effect the laser has on thetargets surface morphology and in consequence the amount of materialevaporated. The substrate must be tested and calibrated to find the idealposition. If it is too far from the target barely any material ejected fromthe target will reach the substrate, if it is too close it might block the laser

3The paper cited here [53] doesn’t name the precursors but those are probablyIn(CH3)3, Ga(CH3)3, Zn(C2H5)2 and H2O.

14

Background

from hitting the target, or the film homogeneity will suffer because theevaporation plume is not covering the whole substrate. Also the laserpower and pulse frequency has to be calibrated for the material to bedeposited and the temperature at which the deposition happens. When thelaser ionizes and evaporates the material on the target not all of it mightcondense, some of it can get pumped out in the process. This happensespecially for elements that form gases or evaporate relatively easy. In thework done for this thesis a pressure of 1Pa O2 was kept in the chamber toensure a low amount of oxygen vacancies in the deposited material (ZnOand InGaZnO4).

All these make this, in principle straightforward, deposition a technicalchallenge. Once set up, the laser power can be finely tuned to control theevaporation plume and controlling the amount of pulses lead to a goodcontrol of the film thickness. The target holder can hold several targets atthe same time and switch their position so it is possible to have the laserhit different materials in succession. A superlattice can e.g. be done bypulsing ten times at the ZnO target, then switching to the IGZO targetpulsing 100 times and repeating the whole process x times.

Compared to ALD/MLD a big strength of PLD is that it can be extremelyclean and that complex oxides can be deposited since the target materialsare prepared beforehand with all necessary requirements in mind andthe actual deposition process is „only“ an elaborate method to transportmaterial from the target to the substrate without any reaction productsthat could be trapped in the film. The film deposition can be relatively quickand there are mostly no limits on the used deposition temperature otherthan those imposed by the material itself. The laser can burn and damagematerials and therefore this method is not well suited for materials thatare not thermodynamically stable and could disintegrate, such as organicmolecules.

Figure 2.3 shows the chamber during deposition at the moment shortlyafter laser impact when the ionised plume from the target is visible. Thesubstrate is above the plume. Under the chosen deposition parameters theIGZO grows amorphous (a-IGZO) while ZnO grows polycrystalline with astrong preference for growth along the c-axis (see also [55]).

15

Background

Figure 2.3. A PLD chamber with the ionised plume visible on the substrate due to thelaser impact (Photo from the webpage of the Ohta lab in Sapporo, wherethis work was done — http://functfilm.es.hokudai.ac.jp/english/equipment/.Printed with permission. Creator: Prof. Ohta).

16

3. AlOx & benzene layers in ZnO

For all the studies in this thesis one thing is of utmost importance: thestructure of the sample must be known. Thin films can be stacked on top ofeach other creating a thin film consisting of several stacked layers. Atomiclayer deposition theoretically ensures that the structures are built up ontop of the substrate layer by layer, but there are possible ways to undo thedeposited structure and characterisation methods are needed to confirmthe wanted structure is indeed there. The terms nanolaminate and super-lattice are often (but not always) used in literature to distinguish betweena material with approximately equally thick stacked layers (nanolaminate)and a material where one layer-type is substantially thinner than theother, essentially acting as a thin separator between thicker layers. Thevast majority of research focusses on regular repeating stacks, but withinthis thesis irregular stacking is investigated and those are called gradientmaterials to emphasize their deviation from regular superlattice structures,sometimes the term random multilayer structure can also be found in lit-erature for such irregular structures. The different structure types aresketched in Figure 3.1 and this nomenclature will be used throughout thisthesis.

Figure 3.1. Sketch of a superlattice, a gradient material and a nanolaminate.

17

Superlattice(SL)

Gradient Material(GM)

Nanolaminate(NL)

Layer 1(matrix)Layer 2

AlOx & benzene layers in ZnO

With structures such as the ones sketched in Figure 3.1 there are someimportant details that are not always the centre of attention - the natureof the interfaces between the layers. Will they be sharp interfaces (assketched) or will they be more gradual through diffusion of layer 1 intolayer 2 and vice versa? And how strictly will the layers be separatedfrom each other by the other layers? These types of questions are ofgreat importance especially in hindsight to possible future applicationsutilizing quantum effects such as tunneling through thin layers (e.g. inmetal-insulator-metal diodes [56]). This is of course strongly dependant onthe (i) individual layer materials, (ii) deposition technique, (iii) depositionconditions and also related to the overall structure. Single crystals thathave different layers grown epitaxially can be expected to have well definedlayers, partly polycrystalline or amorphous samples not so much. Inthe present work those questions were investigated for a ZnO/AlOx/ZnO-benzene layer system.

Aluminium is one of the most common dopands to increase the electricalconductivity of ZnO, it can substitute Zn in the ZnO lattice, but with Alhaving three valence electrons instead of two Al acts as classical electrondonor (n-doping), with ideal doping levels typically being between 1-3%Al [29, 30, 57–59]. Sometimes this is labelled as Al2O3, but this thesiswill use AlOx to indicate the uncertainty on the structure and immediatesurrounding of Aluminium in these very thin layers.

Apart from irregularities and (always present) deviations from an idealgrowth during deposition, solid state diffusion could be an issue. Solidstate diffusion is notoriously slow, but the distances in thin film structurescan be very short (even below 1 nm). The, by design, large shared surfaceof the layers in thin films provide ideal conditions for solid state diffusion.The thinner the layers are the quicker the reaction can be and severalof the samples in this thesis were deposited using only one single cyclemaking the layer as thin as possible. Finally one of the main reasons formaking these layered thin film structures is that it enables structures thatare not accessible by other synthesis methods, these structures are notnecessarily the energetic most favourable configuration and mixing mightbe the thermodynamically preferred option. All of the papers in this thesistherefore include a section demonstrating that the structure is really there(mostly by showing superlattice peaks in the XRR pattern).

The ZnO/AlOx/ZnO-benzene structures were grown with atomic layerdeposition. ZnO grows (poly)crystalline when thicker than approximately1 nm, while AlOx tends to form amorphous layers with ALD [60, 61]. Inregards to the structure of AlOx layers within a ZnO-matrix, electronmicroscopy images are reported that show those layers are distinct andsharp even if only single a layer of AlOx is deposited between thicker ZnOlayers at temperatures below 250°C [62,63]. At deposition temperaturesof 250°C distinct AlOx layers are reported to show up only when thicker

18

AlOx & benzene layers in ZnO

than ~4Å [64]. With Aluminium contents > 1% a shift in the ZnO peakindicating Aluminium substituting Zn in the ZnO lattice has been observedalready around 200°C [58] which opens up the question how much of thelayer is participating in the ZnO-crystal structure.

Another interesting aspect, specific to the ALD process of ZnO and AlOx,is that during the deposition the precursor of AlOx, trimethylaluminium(TMA), can etch the ZnO surface [65,66] and that the specific cycle sequencewhen doping ZnO with Al has an impact on the conductivity [67].

The question investigated within this thesis about this interesting in-terface chemistry was: How well are thin AlOx layers separating the ZnOlayers? At 250°C only layers > 4 Å could be deposited and blocked ZnOgrowth [64] but what about the thinner monolayers deposited below 250°C.This has been studied in this thesis together with a comparison betweenaluminium — a small atom with the possibility to substitute Zn in the ZnOstructure, that etches the ZnO surface upon deposition and benzene — abulky organic molecule that does not show any of these properties.

Benzene is a somewhat exotic molecule in ZnO. Using hydroquinoneand DEZ as precursors the inorganic-organic hybrid material of benzenerings linked via oxygen bridges to Zn atoms was first deposited with MLDin 2011 [50] and has shown electrical conductivity that is credited tothe aromatic π − π system in the benzene [68]. In the MLD literaturestructures that can be described as [Zn – O – R]n with R being any organicmolecule are often termed zincones, but to be specific this thesis will alwaysrefer to ZnO/ZnO-benzene and not use the term zincone to indicate thatR is benzene.Notably the molecule inserted can have an influence on thepreferred growth of the ZnO grains [69].

ZnO/ZnO-benzene interfaces have first been utilized in 2013 by Tynell etal. [59,70] for thermoelectric purposes and Choudhury et al. [71]. Togetherwith TiO2/TiO-benzene superlattices, ZnO/ZnO-benzene has since servedas model system for the reduction of thermal conductivity in hybrid multi-layer structures, a collection of ZnO/ZnO-benzene structures investigatedfor thermal conductivity can be found in table 7.1 in the appendix on page47. Most interestingly for the structure, calculations suggest that only ev-ery 2nd possible reaction site on a ZnO surface will be occupied by benzenedue to steric hindrance blocking every other possible reaction site [72,73]and experimental studies could show that thin ZnO-benzene layers do notcompletely separate ZnO layers but that 9 (or more) consecutive hybridZnO-benzene cycles are necessary to effectively separate ZnO layers [74].

The structuring in the nanoscale is one of the most tantalizing aspectsof these films and during this thesis the interface of ZnO with AlOx andZnO-benzene layers in direct comparison within one sample were studiedin Publication I.

19

AlOx & benzene layers in ZnO

3.1 Analysing the structure of layered films

Many challenges regarding thin films are linked to characterisation. Char-acterisation is complicated by the scale of the films, in more „classicalchemistry“ the materials can be weighed with an fine but ordinary scale,measured with a ruler, or at least an optical microscope and so forth. A lotof this is complicated when working with materials that are only 100 nmthick1. The thickness, let alone the layered structure inside, is far belowthe wavelength of visible light so optical microscopes won’t do and electronmicroscopy techniques (SEM or TEM) or atomic force microscopy (AFM) isneeded for pictures. Silver paste contacts to measure electrical propertieswill heavily outweigh and dwarf the thin film in sheer volume and mass.Another difficulty in analysis is that this also makes it harder to pinpointreasons for measurement failures that are due to cracks in the sample, oreven flakes that might have broken loose from the film. Flaws like thatare not easily spotted but can have a pronounced impact on the measureddata.

Good methods to analyse the layered structure of thin films are trans-mission electron microscopy (TEM), x-ray reflectivity (XRR) and x-raydiffraction (XRD). Their general principles are presented in the followingsubsections.

3.1.1 XRD & GIXRD

X-ray diffraction has been one of the major breakthroughs of the lastcentury enabling the understanding of materials from inorganic crystals todecoding the structure of DNA. It is based on the diffraction patterns thatoccur when x-rays are scattered by regular crystal lattices as described byMax von Laue and then developed further by William and Henry Braggwho came up with the famous Bragg equation [75]

nDiff.λ = 2d sinΘ (3.1)

that describes when diffraction maxima occur and how they relate to thedistance (d) between lattice planes (nDiff. is the order of diffraction, λ thewavelength and Θ the scattering angle in the equation 3.1).

This method is necessary to determine the crystallinity and what crys-talline phase is present in the film. In this thesis the only crystalline phaseobserved was that of ZnO in the wurtzite structure (Figure 2.1). From thewidth of the diffraction peaks the average crystallite can also be estimatedwith the Scherrer formula:

τ =Kλ

β cosΘ(3.2)

1If you are reading the paper version of this thesis then the page you are readingis ca. 120 μm thick, about 1200 times thicker than the films deposited for thisthesis.

20

AlOx & benzene layers in ZnO

With τ being the crystallite diameter, K is a shape factor (0.9 for sphericalshapes), λ the wavelength, β the peak broadening and Θ the scatteringangle of the peak in question. This only works for small crystallites asfor big crystallites or even single crystals the peak broadening due tocrystallite size is far below the peak broadening due to the instrument.

XRD can not only be used for a confirmation on the crystallinity of theZnO in the film (and that there are no other phases present in the film),but with the Scherrer equation also give an estimate of the crystallite size.The crystallinity, the crystal phase and insights into the crystallite sizeare valuable insights into the internal structure of the film and can alsogive hints for the question on whether or not interface layers separate theZnO layers.

A variety of XRD is grazing incidence x-ray diffraction (GIXRD). Thisis useful because at high angles the x-ray passes mostly through the filmand the signal can become very weak, especially if light elements make upthe film. Keeping the angle of incidence low and only moving the detectorleads to a much higher fraction of the film being constantly exposed to thex-ray and a better signal from the film, this is especially useful for verythin films, films with a low crystallinity and films with a high amount oflight elements that do not scatter x-rays strongly. The basic principle issketched in Figure 3.2.

incoming beam

detec

tor

scattering vector

scattering plane

scatte

red be

am

scatte

red beam

incoming beam

scattering vector

scattering plane

detec

tor

GIXRDXRD

Figure 3.2. Beam path, sample and scattering vector in XRD (Bragg Brentano) andGIXRD configuration.

One important difference between the two methods is that the scatteringvector in the GIXRD setup is moving which means that each peak is fromcrystal planes having different orientations within the sample [76, 77]).This makes assumptions on the crystal orientation much less reliable thanin the classic Bragg-Brentano setup and should therefore be used withgreat caution. However this doesn’t impede the usefulness of GIXRD todetect crystalline structures even in thin films below 20 nm which is thegreat strength of this method.

21

AlOx & benzene layers in ZnO

3.1.2 XRR

X-ray reflectivity is a non-destructive method to obtain the thicknessof a thin film and information about the internal structure, especiallysuperlattice structures can be determined very well. By simulating theXRR curve a good estimate of the interface roughness and density is alsopossible. This made it probably the most powerful non-destructive methodcurrently available for the analysis of superlattice thin film structures.

As the name suggests XRR is based on reflections from the film inter-faces, which is also the reason why it can be used for multilayer struc-tures, but the main interfaces are typically the surface of the film andthe film/substrate interface. The reflected beams from those interfacesinterfere destructively or constructively very much alike to the crystalplanes used in the Bragg equation.

Because XRR depends on the reflection of the interfaces it must be doneat very low angles (typically Θ stays below 3°) as the reflections get weakerat higher angles and scattering starts to dominate. Conveniently in firstapproximation the refractive index for x-rays is directly coupled to theelectron density and therefore the electron density (ρel) can be derivedfrom the angle of total reflection (critical angle, Θc), the wavelength (λ)and the radius of the electron (rel) according to the following equation.

ρel =Θ2

cπ

λ2rel(3.3)

If the composition of the film is known, for example from the depositionparameters, then this electron density can be linked to the material density(ρ) with the Avogadro number (NA), the atomic number (Z) and the molarmass (M ) [78].

ρ =ρelM

NAZ(3.4)

At angles > Θc a declining wave like curve appears with the wave maxima(called Kiessig fringes) being due to constructive interference and the fringepositions (essentially the „wavelength“) can be connected to the distancebetween the reflecting interfaces (compare d in the Bragg equation, eq.3.1) and thereby with the overall film thickness. In case of a superlatticebigger fringes will appear from constructive interference of x-rays reflectedat the film-surface, the film-substrate and the film-internal superlatticeinterfaces simultaneously. Finally the overall decline of the signal intensitycan be used to give an estimate of the films roughness because a higherroughness leads to more scattering and less specular reflection whichmeans the overall x-ray intensity at the detector, positioned in the reflectedbeam, will decrease quickly with increasing angles.

To calculate the curve requires solving the Fresnel equations and the pio-neering work has been done by in the beginning and mid of the 20th century

22

AlOx & benzene layers in ZnO

by Heinz Kiessig and Lyman G. Parrat [79,80]; more recent overviews onthe mathematical background can be found in the following sources [81–84].Thanks to their work and modern day computational power these curvescan nowadays conveniently be simulated and matched to the measureddata.

The practical limits of XRR are films getting too thick as then the inter-ference maxima become too small and close together until the resolution ofthe machine doesn’t reliable distinguish any more between the peaks andthe resulting curve resembles a flat declining curve, for the machine usedin this thesis this limit was about 200 nm. It could be extended by a changeof the x-ray wavelength or a machine with finer measurement steps butultimately any XRR setup will reach it’s limit when the films become toothick. XRR also crucially depends on a reasonable smooth surface wherespecular reflection of the x-rays can be assumed.

Another issue is the reliability of the curve simulation of complex films,meaning films that have many non-repeating layers. This is becauseeach layer has (at least) three variables to fit (density, roughness andthickness) a multistacked film with many different layers will end up withan egregious amount of variables to fit, not only does this considerablyincrease the calculation time, but the more practical problem is that alot of variables often also lead to many different good fits and the dangerof ending up in a local minimum, that provides a reasonable fit but isultimately wrong, is rising the more variables there are. The user interfaceof the program used for fitting samples in this thesis (Panalytical x-rayreflectivity) with a measured and fitted curve is shown in Figure 3.3.

Figure 3.3. A screenshot from fitting a superlattice sample with the Panalytical Reflectiv-ity software. The blue line is the measured data, the red line the simulatedcurve. This one already has 18 independent variables.

Within those limits the big strength of XRR lies in its capability to beused for quick thickness checks confirming everything went well with thedeposition as well as for film analysis with a comprehensive simulationof the film structure and the measurement is non-destructive. Therefore

23

AlOx & benzene layers in ZnO

XRR was extensively used in all studies during the thesis.

3.1.3 TEM

With (GI)XRD the crystal structure (if present) of the film can be deter-mined and with XRR the film layering (at least for superlattices) can beobserved and then matched with a simulation. However, what none ofthese techniques can do is provide images.

Transmission electron microscopy is another analytic technique that hasbecome available in labs around the world and enabled great progressin the understanding of nanomaterials. As the name suggests it worksin principle like a microscope but instead of physical light electrons areutilized because their wavelength is far below that of optical microscopesand the resolution can therefore be much higher. Of course utilizing elec-trons instead of light comes with a whole set of technical and theoreticalchallenges and operating and preparing samples for, especially a transmis-sion electron microscope, takes a lot of expertise. Electron microscopy wastherefore one of the two measurement methods in this thesis for whichcollaboration was crucial (the other being time domain thermoreflectancewhich will be described in Section 5.1).

Transmission electron microscopy means that the electrons transmit thesample, detector and beam source are therefore on different sides of thesample. To get any meaningful information the sample must be extremelythin (ca. 70 nm in this work) to allow electrons to pass, cutting such asmall and thin sample out of the thin film was done with a focussed ionbeam. During this process and also later during the TEM measurementthe heavy bombardment of ions (and electrons respectively) can lead tostructural damage in the sample. Thermodynamically unstable materials,like organic molecules are especially susceptible, finally the sample mustbe conducting to prevent being charged from the constant electron beam.

The images of the structures that can be obtained offer a much moredirect insight into the internal structure of the films than indirect methods,such as XRR or XRD, ever could. Therefore TEM is incredible valuable butalso very time and resource expensive and not suited for everyday checkups on samples. TEM was the final cornerstone in the structure analysisin the ZnO/ZnO-benzene/AlOx thin film system conducted in Publication I.

24

AlOx & benzene layers in ZnO

Figure 3.4. A high resolution TEM picture of a sample (only ZnO visible).

3.2 Structure of ZnO/ZnO-benzene and ZnO/AlOx superlattice films

With ALD/MLD the desired samples with mixed layers could be prepared,a sketch of the structure is shown, next to the TEM image in Figure3.5a. Because earlier reports indicated that ZnO-benzene layers will onlyeffectively separate ZnO layers when 9 (or more) consecutive MLD cyclesare used for each ZnO-benzene layer [74] a cycle sequences with 1 MLD(ALD) cycle for the ZnO-benzene (and AlOx) layers and with 10 MLD(ALD) cycles for the ZnO-benzene (AlOx) layers respectively were chosen.That should ensure a sample with permeable ZnO-benzene layers and onesample with ZnO-benzene layers that are just thick enough to separatethe ZnO layers.

Benzene

AlOx

(a) A sketch of the internallayering of a ZnO samplewith ZnO-benzene andAlOx layers.

(b) TEM picture of the superlattice structuresketched next to it, 10 ALD/MLD cycles foreach interface layer respectively.

Figure 3.5. The planned sample sketched (a) next to a TEM of the deposited sample (b).The matrix is ZnO.

25

AlOx & benzene layers in ZnO

Figure 3.6. A TEM image and the ideal structure (structure according to the calculationsfrom Karttunen et al. [73] )

A previous study looking into AlOx layers in respect to their ability toblock the ZnO grain growth at 250°C could not observe AlOx layers belowa thickness of ca. 4 Å [64]. This might be because at 250°C diffusion couldplay a role as another study about annealing such structures showed [63].Within this thesis the deposition temperature was kept at 230°C which isvery close, but still below 250°C and layers formed by single ALD cyclesdepositing AlOx were observed. These findings are reported in PublicationI. The results show that the ZnO grains in the films matrix are able topenetrate both layers (AlOx and ZnO-benzene) deposited with only onesingle ALD/MLD cycle but that the layers deposited with 10 ALD/MLDcycles effectively halt ZnO grain growth.

This is an interesting find as initially the bulkier layers with benzenewere expected to be more effective barriers than the AlOx layers, espe-cially because Al can easily diffuse into the ZnO matrix taking interstitialpositions or even substituting Zn in the ZnO lattice, something that canhardly be expected from a benzene ring. The findings show no fundamentaldifference between those two layers in that respect, structurally they seemto be equally effective in blocking ZnO grain growth2. A reason for the ZnO-benzene layers showing similar permeabilities for the ZnO layers mightbe that their bulkiness leads to steric hindrance, calculations suggest thateven in an ideal case only every second possible surface reaction site canbe occupied by a linked benzene molecule (Fig 3.6) [72,73].

This penetration of ZnO grains is possibly the reason that more cy-cles/layer lead to higher interface resistances [6] because thicker layers aremore efficient at hindering grain growth (and thereby phonon pathways)through the interface (this will be discussed later in Chapter 5).

So ultimately the ZnO-benzene and the AlOx layers are structurally verycomparable on the nanoscale (at least when deposited at 230°C) as bothlayers could be penetrated by ZnO grains when only one cycle was used todeposit the respective layer and both layers effectively hindered ZnO grain

2At least within the precision achieved in this work.

26

AlOx & benzene layers in ZnO

growth when 10 successive ALD/MLD cycles were used for each layer.Unfortunately at higher resolution the interface layers become very

smeared in the TEM images, the reason for this possibly lies in the factthat they have a slight waviness in the range of 1 nm (see Figures 3.6 and3.5b) which of course would also be perpendicular to the image plane andbecause the beam goes through the sample (ca. 70 nm) the waviness canexplain that the interfaces appear smeared, this means that no good imageof the interface at the atomic scale could be obtained. Another problemwith increasingly higher resolutions is the stability of the film as the strongelectron beam burns through the samples. The smearing could also bepartly a result of the cutting process with the focussed ion beam duringsample preparation. Ultimately the images with higher resolution couldn’tprovide any further insight.

An encouraging find was that the structure obtained from XRR and TEMwas basically identical, confirming the capability of XRR to accuratelydeduce the structure by simulating the measured reflectivity curve. Alsothe Scherrer equation could be used to confirm smaller crystallites inthe sample with thicker interfaces, which could already be interpretedas a good hint towards a better separation of the ZnO layers which wasultimately confirmed by the TEM images.

These results show that even single layers of AlOx are still formingdistinct, separate, layers from the ZnO surrounding, in fact that on ananometre scale structurally they are very similar to chemically funda-mentally different, organic, ZnO-benzene layers. This study also confirmsthat XRD and XRR are very powerful tools for structural determinationof these films as they were the „working horse“ for the structural controlmeasurements in all studies within this thesis.

27

4. Electrical properties in layereda-IGZO/ZnO thin films

Interfaces can have a strong impact on the electrical conductivity, as hasbeen mentioned in Chapter 2. Thin localized interface structures areknown for example from naturally occurring layered oxides, most famouslyin the high temperature superconductors where copper oxide layers playsuch a crucial role [85,86] but can also be found in artificial superlattices[3, 87]. The basic principle, especially in the oxide heterostructures, isthat the different metals vary in electronegativity, charge and chemicalpotential leading to exotic band structures at the interface between twodifferent metal-oxide layers. [88,89]

Early studies on quantum confinement in thin AlxGa1–xAs–GaAs struc-tures were already conducted in the 1970s by Dingle et al. [90] and the dis-covery of a 2D electron gas at the interfaces of SrTiO3/LaAlO3 by Ohtomoin 2004 [3] spurred research in this direction and several more similaroxide systems have been reported since [91–94]. The vast majority of theseinterface effects are studied in epitaxial systems because single crystalsare by definition very well defined and the interface is smooth. Howeverthere is no fundamental reason why single epitaxial layers are the onlypossibility as long as the interface is sharp enough and consistent, in factgrowing epitaxial films is often a great challenge and limited in regardsto the films size. To study possible interface effects on the conductivity,ZnO/a-IGZO superlattice/nanolaminate films were deposited and anal-ysed. These films are not epitaxial, but polycrystalline (ZnO)/amorphous(a-IGZO), making them structurally very different from the classic singlecrystalline epitaxial layers such as AlGaN/GaN [95], or SrTiO3/LaAlO3 [3]structures among the oxides.

InGaZnO4 is, like ZnO, a transparent conductive oxide and has garneredincreased interest because of its good charge carrier mobilities even in theamorphous state. Publication II is based on this material and some of itsbasic properties will be presented here.

29

Electrical properties in layered a-IGZO/ZnO thin films

4.1 InGaZnO4

InGaZnO4 is a layered oxide with natural ZnO, In2O3 and Ga2O3 layersin the crystal structure. Growing single crystals of this ternary metalsystem is possible but not easy [96]. The system can be grown withdifferent In:Ga:Zn ratios leading to variations in the layering (e.g moreZnO layers). In this thesis all variants will be abbreviated IGZO, and if thematerial is amorphous a-IGZO. In the samples deposited for this thesisthe stoichiometry of In:Ga:Zn was 1:1:1 — InGaZnO4.

The main reason that the IGZO system has gained interest within thelast decade is that these phases can have excellent electrical mobilitiesin the amorphous state, which makes it applicable to fabricate transpar-ent flexible thin film transistors (TFTs) [97, 98] that are now replacingamorphous silicon transistors in flat screen devices [99].

The reason for the good mobility of charge carriers, is suspected to stemfrom the orbital structure of the involved metals [100]. The orbitals inthe lowest conduction band of e.g. the classic semiconductor silicon aresp3 orbitals that are highly directional and the overlap between them isstrongly reduced in the unordered structure of amorphous materials. Inindium, gallium and zinc oxides this is different, the lowest conductionband orbitals are s-orbitals, which are spherical, hence their overlap is notas strongly influenced by directional irregularities that are characteristicfor amorphous materials. While the electron mobility in amorphous silicondrops by several orders of magnitude compared to the crystalline stateit stays rather high for a-IGZO [99,100]. In, Ga and Zn are not the onlymetals with that property, the effect is proposed for all metals with a(n− 1)d10ns0 shell, which includes the metals from groups 11-14 below the3rd period of the periodic table or specifically: Cu, Ag, Au, Zn, Cd, Hg, Ga,In, Tl, Ge, Sn, Pb, As, Sb and Bi [100]. Notably this includes ZnO andalso the current state of the art TCO — indium tin oxide for which theIGZO system could become a strong contender. Furthermore an increasein Seebeck effect has been demonstrated in amorphous interfaces betweenIn-Zn-O and In-Ga-Zn-O systems [101] encouraging further investigationsregarding the interfaces.

Because this thesis investigates effects of internal structuring and foreffects similar to the 2DEG found in systems like SrTiO3/LaAlO3, the goodelectric mobility in the bulk of a-IGZO wasn’t as interesting and could evenbe seen as detrimental and therefore for this thesis pulsed laser depositionparameters with a relatively high partial pressure of oxygen (1Pa) werechosen. This allowed the deposition of a-IGZO/ZnO superlattices andnanolaminates at room temperature ensuring that the resulting a-IGZOhas few oxygen vacancies and thereby a low conductivity that ideallydoesn’t interfere with possible electric interface effects.

30

Electrical properties in layered a-IGZO/ZnO thin films

4.2 Characterising physical properties

The structure of the films deposited by PLD was checked with XRD andXRR as already described in Section 3.1 with the result that the ZnO waspreferentially growing in the expected c-direction and that the layeredstructure was present. The focus in this study was on the electrical prop-erties and to that regard Van der Pauw/Hall, UV-Vis and thermopowermeasurements were conducted that are shortly described in the followingsubsections.

4.2.1 Transport measurements

The Van der Pauw method is one of the standard 4 point probe configura-tions, combined with a magnetic field (in this case the field was 0.76 T) thatis perpendicular to the electrical current it allows for Hall measurements.Small droplets of an In/Ga alloy, that is liquid at room temperature, wereused to ensure good electrical contacts between the electrodes and thesample. This combined setup enables the experimenter to measure thesheet resistance, the charge carrier concentration (and type) as well as themobility of the charge carriers. It is standard laboratory equipment andtherefore won’t be discussed in greater detail here.

The setup for the thermopower consists of two Peltier elements that canbe used for heating and cooling. The sample is placed so that one end ison the cold Peltier element and the other end on the heated side, thena thermocouple and an electrode are placed on the sample on each endto measure the temperature at both sides simultaneously to the voltagedifference occuring upon inducing a temperature gradient. The measuredvoltage can then be plotted against the temperature difference with theslope being the Seebeck coefficient (Figure 4.1).

31

Electrical properties in layered a-IGZO/ZnO thin films

V

T

S = -131 μv/K

Figure 4.1. Measured values to calculate the Seebeck coefficient (from sample ZnO/IGZO1/6) potted.

4.2.2 UV-visible spectrophotometry

Another useful and common analysis tool is the UV-Vis setup, especiallyfor wide bandgap semiconductors and TCOs. The absorption spectrum of asample can give insights into the optical bandgap of the material since thewavelength at which absorption begins can be directly connected to theenergy of a photon of that wavelength with

hc

λ= Ephoton (4.1)

With E being the energy, h the planck constant, λ the wavelength and c

the speed of light. To be absorbed that energy must be larger or equalthan the bandgap (Ephoton ≥ Eg), thus enabling a measurement of theapproximate bandgap of the material together with the transparency ofthe sample in the visible and ultra-violet range. Another effect that canbe observed (if present) is the change from a crystalline material that hasa clear absorption edge due to its regular ordered structure towards amore amorphous material which often has a more gradual increase in theabsorption (also called Urbach tail). This can be seen in Figure 4.2 in thenext section.

4.3 Charge carriers at interfaces

With PLD a series of differently layered ZnO/IGZO structures was fab-ricated. The naming follows the structure such that ZnO/IGZO 1/6 is asample where 6 nm a-IGZO are followed by 1 nm ZnO and the sampleZnO/IGZO 10/1 is a sample where 10 nm ZnO are followed by 1 nm a-IGZO.These patterns are repeated until the overall thickness is ca. 100 nm,

32

Electrical properties in layered a-IGZO/ZnO thin films

always beginning and ending with the thicker material to keep the samplesymmetric.

With XRD the crystalline structure of the ZnO layers (growing strictlyin c-direction) could be confirmed. As expected the samples with highIGZO content were amorphous. This can also nicely be seen in the UV-Visspectra (Figure 4.2) where the sharp absorption edge from crystalline ZnOmellows to a more gradual absorption curve with increasing IGZO content.

The layered structure was confirmed by XRR and additionally a highthermal interface resistance of 1.35 m2K/GW was measured, which isconsiderably higher than measured for ZnO/In2O3 superlattices [102]. Allthis indicates a good separation of the layers and confirms that the sampleswere structured as intended with well defined layers of ZnO and a-IGZO.

The most interesting part was to look at the electronic properties andeffects the layering might have. The electrical conductivity, the Hallmobility, the charge carrier concentration, the Seebeck coefficient and theabsorption coefficient in the UV-Vis range of the samples are plotted inFigure 4.2.

The ZnO rich samples are the more interesting specimen. The purea-IGZO has a very low electrical conductivity, charge carrier concentrationand mobility and simply by „rule of mixture“ an increasing amount of ZnOin the sample would lead to higher values. If interface effects are at playthey are too small to notice in those samples. In the ZnO-rich samplesthings are different, by „rule of mixture“ a decline with increasing a-IGZO content in mobility, carrier concentration and conductivity would beexpected, instead a plateau in the Hall mobility is observed and an explicitrise in the conductivity driven mostly by the increase in charge carrierconcentration with increasing interface density. Instead of an interfaceeffect however the increase in charge carriers could also point to a dopingeffect. Doping was tried to prevent as much as possible by depositingthe films at room temperature. The XRR measurements confirmed theseparation of layers, yet doping could still occur. What speaks againstdoping being the only mechanism is that a rise in conductivity up tosample ZnO/IGZO 2/1 is measured, a sample in which 1/3 of the materialis a-IGZO, while typically the conductivity for doping of In and Ga in ZnOpeaks around 5-6% [103,104].

However the charge carrier mobility stays very constant in the ZnO-richsamples and then declines as the a-IGZO content in the samples getshigher. Oxide systems with such interface effects have shown very highmobilities [3, 93, 94]. However in some cases the measured mobilitieswere rapidly declining with increasing temperature [91] and might notbe visible at room temperature (at which these samples were measured).Furthermore in the polycrystalline/amorphous interfaces investigated inthis work more structural irregularities than in epitaxial layers can beexpected which could counteract the electron mobility.

33

Electrical properties in layered a-IGZO/ZnO thin films

x 1e20

ZnO

19/1

10/16/1

2/11/1

1/11/2

1/61/10

1/19

a-IGZO

1 2 3 4 5 6Energy in eV

IGZO

Substrate

ZnO-IGZO-1-1-igzo

ZnO-IGZO-1-1

ZnO-IGZO-1-10

ZnO-IGZO-1-19

ZnO-IGZO-1-2

ZnO-IGZO-1-6

ZnO-IGZO-10-1

ZnO-IGZO-19-1

ZnO-IGZO-2-1

ZnO-IGZO-6-1

ZnO

Optic

al A

bsor

ptio

n co

ecie

nt

Figure 4.2. Electrical properties of the ZnO/IGZO samples and the UV-Vis spectra (Datafrom [II].)

Ultimately this system shows increased charge carrier concentration thatis attributed, at least partly, to the difference in chemical potential at theinterfaces, but less clear defined than those observed in SrTiO3/LaAlO3

due to the complicated surface structure of polycrystalline ZnO next toa-IGZO.

As noted in [II] doping effects could unfortunately not unambiguously beexcluded from being at least partly responsible for the increase in chargecarrier concentration, but even if only part of the increase is due to theinterface effect this work has given strong support to electric interfaceeffects occurring between polycrystalline and amorphous interfaces. Due

34

Electrical properties in layered a-IGZO/ZnO thin films

to the many factors that can have an effect on the electronic properties ofthese unconventional thin films more research will be needed to accuratelyunravel all possible effects. The possibility to utilize interface effects alsoin amorphous and polycrystalline materials is tantalizing.

35

5. Thermal conductivity in layered thinfilms

One of the most promising routes to enhance the thermoelectric propertiesof materials is the reduction of thermal conductivity by controlling thephonon pathways. Studies to control the phonon properties in layeredthin films go back several decades [105,106]. Also the nature of phononscattering (coherent and incoherent) has been actively discussed [107].Coherent phonon transport in superlattice structures was experimentallydemonstrated only recently [108,109].

The thermal conductivity of a great number of superlattice structuressystems have been studied over the years, e.g. CaTiO3/SrTiO3 [108],W/Al2O3 [110], PbTe/PbSe [111], IrSb3/CoSb3 [112], AlN/GaN and AlGaN/-GaN [113, 114], TiN/MgO [115], TiNiSn/HfNiSn [116], Si/Ge (includingalloyed systems) [117–123], GaAs/AlAs (and various alloys) [105,109,124,125], Bi2Te3/Sb2Te3 [126, 127], InAs/AlSb [128], InGaAs/InGaAsP [129],Al2O3/TiO2 [130] and Ge2Te3/Sb2Te3 [131]. The systems vary from singlecrystalline to purely amorphous, and even crystalline/amorphous inter-faces of the same material have been studied [132]. Typically the thermalconductivity of these systems is measured cross plane, meaning perpendic-ular to the layers as this is the direction in which the phonons have to passthe interfaces, but as studies by Hicks and Dresselhaus suggest the ther-moelectric properties parallel to the layers are also worth noting [133,134].A good overview on the different principles and mechanisms investigatedwithin these inorganic material systems is given in a review by Norris etal. [135].

All of these thin film systems are purely inorganic, yet materials con-sisting of classical inorganic and organic structures combined (hybridmaterials) have emerged to expand the inorganic-inorganic layers withinorganic-organic materials that could have excellent thermoelectric prop-erties [5,6,59,136–138] and even add flexibility to the films [139–145]. Thehope is that the inorganic-organic interface acts as excellent scatterer forphonons due to the strong difference in the chemical bond and thermalimpedance between organic and inorganic materials. One of the mostversatile methods to deposit these hybrid thin films is ALD/MLD [43,146].

37

Thermal conductivity in layered thin films

All of these materials are superlattices, especially with the findings oncoherent scattering [108,109] the idea that deviations from the superlatticestructure could be beneficial to suppress phonons has been studied [147–150] sometimes under the name of random multilayer systems. No studiesin this direction had been done for hybrid materials and this is the workpresented in this thesis.

To deposit and analyse the films this work relied on the ALD/MLDmethod as well as XRR and GIXRD as described already in sections 2.4 and3.1. Because lowering the thermal conductivity was the aim of introducingdifferent layer structures into the ZnO films the cross plane thermalconductivity had to be measured. To measure this cross plane thermalconductivity time domain thermo-reflectance (TDTR) was utilized andfinally 39 samples were collected from [III], [IV] and previous studies [5,6]and analysed with a statistical method named multivariate data analysis.Both, TDTR and multivaraite data analysis will be introduced briefly inthe following sections.

5.1 Time domain thermoreflectance

Measuring the thermal conductivity of thin films is very challenging be-cause the thin film makes up such a small fraction of the total material ofthe sample, as most of it typically is substrate.

The time domain thermo-reflectance (TDTR) method overcomes thischallenge by heating and subsequently measuring the cooling of the filmwith extremely high temporal resolution. This is achieved with pico tofemtosecond laser pulses. The whole process has to be finely calibratedto the reflecting material and therefore the samples are typically coatedwith a thin metal film of known reflective properties for which the setuphas been calibrated; this metal film is called transducer. Within this thesistransducers of Al and Mo were used.

When the pump laser beam hits the prepared sample the localized heat-ing leads to a change in the reflectance of the transducer, this effect iscalled thermoreflectance and can be monitored by a separate beam (theso called probe beam). The reflectance change over time after the heatingcan be used to deduce the heat propagation in the sample and thereby thethermal conductivity [151]. This is achieved by modelling the reflectancecurve and therefore the thermal properties of substrate and transducershould be well known beforehand so that the only parameters to fit arethose of the thin film to be tested. The method is well suited for thin filmswith very low thermal conductivity [152, 153], this ability to accuratelymeasure thin films with very low conductivity made this method the bestpossible choice for our samples. Figure 5.1 shows a measurement curveform such a measurement.

38

Thermal conductivity in layered thin films

Figure 5.1. A measurement curve for a ZnO/IGZO sample

A physical model is simulated and matched to the measured curve toextract the thermal conductivity. The different possible configurationsof TDTR setups and the underlying mathematics of thermoreflectanceand it’s connection to phonons, electrons and the piezo-optic effect are anextensive subject on their own and an introduction to the subject can befound e.g. in the TDTR tutorial paper by Jiang et al. [154] or in Section7.4.3 of the textbook „Nano/Microscale Heat Transfer“ by Zhang [155].

5.2 Multivariate data analysis

When collecting all the data from thermal conductivity measurements inthe ZnO/ZnO-benzene system investigated over the years and comparingtheir measured thermal conductivities to their internal structure for thisthesis it soon became obvious that there are a lot of variables to consider.In literature most often the interface density is cited and plotted with thethermal conductivity. However there is more available data, e.g. on (i)the total film thickness, (ii) the interface thickness, (iii) the thickness ofthe ZnO blocks separated by the ZnO-benzene interfaces and (iv) the totalamount of interface layers. Just as the interface density is the product ofthe total amount of interface layers divided by the film thickness manyother parameters could be constructed from that data, for example thestandard deviation of the ALD cycle numbers for each ZnO layer was usedas a measure of disorder. The question was if those have any influenceon the thermal conductivity. For this sort of problem that includes manycross-correlated variables multivariate data analysis (MVDA) has beenused. The basic mathematics can be traced back to Karl Pearson in 1901

39

Thermal conductivity in layered thin films

and Harold Hotelling in 1933 under the name of principal componentanalysis (PCA). More recently the method has gained considerable tractionand applicability with the wide availability of computing power and theincrease of ever increasing datasets [156].

MVDA can consider a great number of observable variables (weight,volume, number of interfaces etc.) and check which ones have a stronginfluence on a measured outcome (in the present case the thermal con-ductivity) according to the data and finds a mathematical (not necessarilyphysically sensible!) way to combine the measured variables in an approx-imative model to describe the system and predict the outcome (thermalconductivity in this case).

MVDA includes several mehods, not only PCA but also the closely relatedpartial least squares or projection to latent structures (PLS)1 method. [157].PLS can correlate multiple sets of data; for this thesis the influence ofcertain properties (interface density etc.) on the thermal conductivity wasinvestigated. To do this PLS relies heavily on linear algebra expressingall input variables as a matrix (interface density etc.) and the variablescorrelating as another matrix (thermal conductivity). Then new param-eters (combinations of interface density etc.) and weights are calculatedand optimized to model the relation between the input and the outputmatrices. A good model can give insights into which input variables (filmthickness, interface density etc.) have a great effect on the outcome (thethermal conductivity in our case). Readers interested in the mathematicalbackground are referred to the overview articles by Jackson, Wold andRosipal et al. [156–160] or the textbook by Eriksson et al. [161].

The method is general and has been used previously on structure-propertyrelations e.g. in connection with superconductors [162], magnetic proper-ties [163,164] and even the chemical composition of perovskites [165]. Inthis thesis it is utilized to check for variables that might have an effect onthe thermal conductivity but might have been overlooked. All variablesused in the PLS are depictured in Figure 5.4 in the next section.

5.3 Thermal conductivity of ZnO/ZnO-benzene thin films

The samples were grown with ALD/MLD, their structure was determinedwith XRR and the crystallinity of the ZnO in the samples confirmed withGIXRD. TDTR and MVDA were used to measure the thermal conductivityand analyse connections between the thermal conductivity and structuralproperties of the samples respectively.

The potential of organic layers in inorganic materials to suppress the

1A bit confusing, the PLS method was first (and sometimes still today) referredto partial least squares, but then slightly modified and now projection to latentstructures is deemed the better and more descriptive name. [157]

40

Thermal conductivity in layered thin films

thermal conductivity of the material had been reported in several systemsby our group and others [5,6,136–139,142,166,167]. Missing was a system-atic study on many comparable samples with different interface densitiesand layering sequences as the by far predominant reported structure isthe superlattice.

This gap was filled by the work conducted in this thesis. In PublicationIII the first hybrid gradient materials were reported and their thermalconductivity values do not deviate significantly from superlattice struc-tures, supporting the notion of incoherent scattering and showing thatthere is no intrinsic reason to prefer superlattices over other layeringstructures. In Publication IV those samples, are analysed together withnewly deposited samples that extended the sample set to 200 nm and 50nm thick films (earlier reported samples were all around 100 nm thick)lifting the total number of ZnO/ZnO-benzene samples with data on theirthermal conductivity to 42 (of which 32 were deposited during this thesis,all are sketched in the Appendix in Figure 5.2).

From this 42 samples three were excluded because they were eitheramorphous due to too many ZnO-benzene layers (samples SL48(1) 100and SL96(1) 100 see appendix) or the ZnO-benzene layers were too closetogether in part of the sample rendering a reasonable distinction on thepoint at which they should count as separate layers too difficult (SL12(1)-Fib 100nm). The 39 remaining samples made up an appreciably largesample-set of thermal conductivities measured in a layered hybrid system;the complete list is in table 7.1 in the Appendix. The thermal conductivityof these samples is plotted against the interface density in Figure 5.2.

Reported thermal conductivities of ALD-grown ZnO thin films are above40 W m−1K−1 [168], therefore all samples with ZnO-benzene layers showa strongly reduced thermal conductivity. The clear outliers here are thesamples with thicker interface layers: (SL(3) 6, SL(5) 6, SL(10) 6 and GM(5)5 +20 100 nm) that are more effective in reducing the thermal conductivitythan their counterparts with monolayer thin interfaces. This observationhas already been noted in [6] and it fits well with the observation of thinZnO-benzene layers being penetrated by ZnO crystals (see Chapter 3,Publication I and Niemelä et al. [74]). Figure 5.3 plots the samples withthicker ZnO-benzene layers together with some comparison samples thatused similar amounts of MLD cycles but instead of having few thickerlayers have them spread out.

The results suggest that to reduce the thermal conductivity scatteringinterfaces is the slightly more advantageous route, however it should alsobe noted that a SL6(10) 100nm uses 10 cycles each for the 6 interfacelayers, using those 60 MLD cycles to fabricate a sample with 60 monolayerinterfaces would almost certainly result in a completely amorphous film atwhich point the comparison becomes less meaningful.

The variables shown in Figure 5.4 of the 39 samples all have weights in

41

Thermal conductivity in layered thin films

0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14Interface density (interfaces/nm)

0

2

4

6

8

10

12

14

Ther

mal

con

duct

ivity

[]

Wm

K11

GradientsSuperlatticesSamples with thick benzene layers

Figure 5.2. Thermal conductivity plotted against the interface density for all 39 analysedZnO/ZnO-benzene samples. The three samples at an interface density slightlyabove 0.12 with thicker ZnO-benzene layers are superlattices with 3, 5 and 7consecutive MLD cycles per ZnO-benzene layer.

Thick benzene layer samples

Ordinary benzene Layer samples

Figure 5.3. Thermal conductivity of samples with thicker interface layers compared tosamples with thinner, but more interface layers

42

Thermal conductivity in layered thin films

SL10(1) 200125 nm

GM5(5)-20 100

196

nm

# of Benzene layers:10 5

Interface density(interfaces/100nm):

10/196 = 0.051;

Cycles/Bz layer1 5

Std.DeviationMin ZnO layer thickness:

17.5 nm 9.6 nm

Max. ZnO Layer thickness:17.5 nm 28.8 nm

ALD-Cycles (ZnO)

150130110

907050

110110110110110110110110110110110

37.4

ALD-Cycles (ZnO)

Std.Deviation

0

Interface density(interfaces/100nm):

5/125 = 0.04

Avg. ZnO Layer thickness:17.5 nm 19.2 nm

Figure 5.4. Measured and calculated variables of the samples used for MVDA.

avg Z

nO (n

m)min

ZnO

(nm)

max Z

nO (n

m)std

ev Z

nOlm

thick

ness

cycle

s/bz l

ayer

# of

bz la

yers

intf. d

ensit

y0.3

0.2

0.1

0.0

0.1

0.2

Wei

ght c

oecie

nt

Increase in

Decrease in

Figure 5.5. Variable weights of the MVDA analysis (bz = benzene). variables connected tothe interface density have by far the biggest impact on the model.

the MVDA model which are presented in Figure 5.5.The model shows good predictive power and judges the interface density,

and the average ZnO-layer thickness as the most influential variablesfollowed by the minimun and maximum ZnO-layer thickness and the totalnumber of ZnO-benzene layers. The standard deviation, the total layerthickness and notably the ZnO-benzene layer thickness are of much lowerimportance according to the MVDA. Obviously the average ZnO thicknessin this work is basically the inverse of the interface density2 and both beingthe most influential variables is therefore reasonable. The interface densityis also directly connected to the number of interfaces. Finally minimum,

2With a very small correction factor due to the thickness of the ZnO-benzenelayers themself.

43

Thermal conductivity in layered thin films

maximum and average ZnO thickness are also seen as important factorsby the model, yet for the superlattice samples, which make up 24 of the 39considered samples in the set, all three variables are identical and againdirectly connected to the interface density, meaning that those are alsomostly covered by the interface density.

These MVDA results suggest that maximizing the interface densityis more effective than maximizing the thickness (and thereby phonon-blocking) of the layers and that the interface density (or the interfacespacing) is indeed the most appropriate way to predict the thermal conduc-tivity in the ZnO/ZnO-benzene system.

From these measurements the interface resistance (Kapitza resistance)could also be extracted by plotting the interface density against the thermalresistivity (1/κ) only samples with a ZnO-benzene monolayer were includedhere as the thickness of the ZnO-benzene layer has an obvious impact onthe interface resistance [6]. The interface resistance was determined as1.6 m2K/GW. This is a comparatively high value e.g. higher than the 1.35found for the ZnO/a-IGZO samples in [II].

Overall the experimental data collected in this work suggests it is ben-eficial to maximize the number of interfaces over the thickness of theinterfaces in the ZnO/ZnO-benzene system. This is despite the findingthat thicker ZnO-benzene interfaces, deposited with several consecutiveMLD cycles, are more effective in reducing the thermal conductivity, onan per interface level, than ZnO-benzene layers deposited with a singleMLD cycle. Yet the latter is an interesting connection between PublicationI and Publication IV as the observation of ZnO grains growing throughsingle cycle ZnO-benzene layers but being blocked by thicker ZnO-benzenelayers is a probable cause for this difference in regards to the thermalconductivity across these layers.

44

6. Conclusions and Outlook

The nanoscale structure of thin films depends on the deposition conditionsand the used materials, which can have a strong impact on the propertiesof the whole film. This was investigated in artificial superlattice strcturesconsisting of ZnO and a-IGZO, two important TCOs resulting in a dedicatedstructured thin film with polycrystalline/amorphous interfaces. The workin this thesis shows an increase in the charge carrier concentration withincreasing interface density, together with clear XRR superlattice fringesand a high Kapitza resistance that indicate well defined layers. Theincrease in charge carrier concentration is therefore suggested to stemfrom interface effects between the polycrystalline ZnO layers and theamorphous InGaZnO4 layers. This nano-structuring of materials could bea promising pathway to tailor the electrical conductivity of materials in ananisotropic fashion and complementary to classical doping.

The structure is investigated in more detail for ALD-grown ZnO withlayers of ZnO-benzene and AlOx with the first reported direct structuralcomparison between those two interfaces. This also provides valuablecross confirmation between XRR, XRD and TEM data, highlighting thegreat analytical utility of XRR when examining films based on knownmaterial components. In the films a great structural similarity betweenthe AlOx and ZnO-benzene layers on the nanoscale could be demonstrated.This similarity is attributed to the strictly layered deposition method ofALD and MLD. This study also reveals the permeability of ALD/MLDmonolayers as ZnO crystallites pass seemingly unhindered through thinlayers both of ZnO-benzene and AlOx while thicker ZnO-benzene and AlOx

layers grown with 10 consecutive ALD/MLD cycles seem to separate ZnOon both sides of the layer effectively.

These structural insights support the data on the thermal conductivityin ZnO/ZnO-benzene superlattices which was extended to now 42 reportedsamples, constituting the largest dataset on inorganic-organic thin filmswith respect to their thermal conductivity. The dataset shows that thickerZnO-benzene layers lead to a significant lower thermal conductivity, yetmaximizing the interface density has an even greater effect. The different

45

Conclusions and Outlook

ways the interfaces can be spread throughout the film, be it a gradientor superlattice like was shown to be of little importance; gradient ma-terials are just as effective as superlattices in suppressing the thermalconductivity. Lastly multivariate data analysis was introduced to anal-yse various variables for the deposited samples in the end confirming thespecial importance of the interface density for the thermal conductivity.

As so often yet many open questions remain. How much of the increasedconductivity in high interface density ZnO/a-IGZO films is really due tothe interface and how much is due to doping or other effects? How canthat be optimized? The thermal conductivity follows the interface densitymore than other factors in ZnO/ZnO-benzene thin films, but does thattranslate to a bulk material? Do the interfaces in layered materials varywhen comparing ALD and other deposition methods and if so how and inwhat respect?

Often the work conducted during this thesis is very specific for the ma-terial system (ZnO/ZnO-benzene), as is unavoidable when making actualsamples. The investigations focussed on fundamental properties such asthermal and electrical conductivity and connected to the structure of thesamples in the hope that techniques and insights from the material systemin this thesis can be transferred with relative ease to other layered thinfilm materials. Filling the blanks in experimental data (e.g. inorganic-organic gradient structures) are small steps on the quest to understandand engineer materials from the atom upwards and the author hopes thiswork will prove useful not only as exercise for himself but also as a steppingstone for others that attempt to extend the understanding and abilities tocreate novel materials for the benefit of all.

46

7. Appendix

Table 7.1. Reported thermal conductivities for ZnO/ZnO-benzene thin films withALD/MLD. (Bz = benzene)

Name (thisthesis)

Name (1streport)

Thick-ness(nm)

Bzlay-ers

cycles/bzlayer

κ

(W/mK)Error Source

SL 6(1)100

99:1 100 6 1 7.2 1.4 [5]

SL 12(1)100

49:1 100 12 1 4.2 0.4 [5]

SL 6(1)b100

- 91.1 6 1 6.9 1.4 [6]

SL 12(1)b100

- 93.3 12 1 4.2 0.49 [6]

SL 24(1)100

- 97.2 24 1 2.3 0.23 [6]

SL 48(1)100*

- 93.8 48 1 0.8 0.15 [6]

SL 96(1)100*

- 82.7 96 1 0.4 0.06 [6]

SL 12(3)100

- 97 12 3 2.4 na [6]

SL 12(5)100

- 97 12 5 1.6 na [6]

SL12(7)100

- 97 12 7 1.2 na [6]

SL 5(1)100

SL5(1) 105 5 1 11.8 1.8 [III]

SL 12(1)c100

SL12(1) 95 12 1 3.9 0.4 [III]

GM 5(1)-20 100

GM5(1)-20 115 5 1 9.3 0.9 [III]

47

Appendix

Table 7.1. Continued

Name (thisthesis)

Name (1streport)

Thick-ness(nm)

Bzlay-ers

cyc-les/bzlayer

κ

(W/mK)Error Source

GM 5(1)+20 100

GM5(1)+20 100 5 1 9.1 0.9 [III]

GM 5(1)+20b 100

GM5(1)+20-2

93 5 1 8.1 1.2 [III]

GM 5(1)20S 100

GM5(1)20S

87 5 1 8.2 1.3 [III]

GM 5(5)-20 100

GM5(5)-20

125 5 5 4.1 0.3 [III]

GM 12(1)+7 100

GM12(1)+7 95 12 1 4.6 0.5 [III]

GM 12(1)Fib 100**

GM12(1)Fib

95 12 1 7.9 1.1 [III]

GM 12(1)M±8 100

GM 12(1)M±8

92 12 1 3.2 0.3 [III]

GM 12(1)Mrev±8100

GM 12(1)Mrev±8

92 12 1 3.3 0.3 [III]

SW 1(12)100

Sandwich1(12)

92 1 12 8.9 0.9 [III]

SL 5(1)b100

SL 5(1)100

90 5 1 9.2 0.9 [IV]

SL 5(1)200

SL 5(1)200

195 5 1 10 1 [IV]

SL 10(1)200

SL 10(1)200

196 10 1 6.7 0.7 [IV]

SW SL 5(1)200

SW SL 5(1)200

201 5 1 9.5 1.0 [IV]

GM5(1)+50200

GM5(1)+50200

205 5 1 9.7 1.0 [IV]

Doublet10(1) 200

Doublet10(1) 200

210 10 1 7.6 0.8 [IV]

GM 10(1)+18 200

GM 10(1)+18 200

203 10 1 7.3 0.7 [IV]

48

Appendix

Table 7.1. Continued

Name (thisthesis)

Name (1streport)

Thick-ness(nm)

Bzlay-ers

cyc-les/bzlayer

κ

(W/mK)Error Source

GM 10(1)M±30 200

GM 10(1)M±30 200

202 10 1 6.9 0.7 [IV]

SL 12 (1)d100

SL 12 (1)d100

98 12 1 4.1 0.5 [IV]

SL 12 (1)200

SL 12 (1)200

196 12 1 6.7 0.8 [IV]

SL 24 (1)200

SL 24 (1)200

186 24 1 4 0.4 [IV]

SW SL 12(1) 200

SW SL 12(1) 200

188 12 1 6 0.5 [IV]

SL 18 (1)200

SL 18 (1)200

204 18 1 5.5 0.6 [IV]

SL 10 (1)100

SL 10 (1)100

94 10 1 4.4 0.6 [IV]

SL 8 (1)100

SL 8 (1)100

96 8 1 5.6 0.8 [IV]

SL 2 (1) 50 SL 2 (1) 50 45 2 1 8.9 5 [IV]

SL 3 (1) 50 SL 3 (1) 50 47 3 1 6.9 3 [IV]

SL 5 (1) 50 SL 5 (1) 50 49 5 1 4.7 1 [IV]

SL 4 (1)100

SL 4 (1)100

97 4 1 9.5 2 [IV]

SL 3 (1)100

SL 3 (1)100

94 3 1 11.3 3 [IV]

* These samples are amorphous and therefore omitted in the MVDA.

** Following the Fibonacci-Series the first ZnO-benzene barrier layers areso close together that they are practically indistinguishable. In practicethat means it has less than 12 barrier layers, but one rather thick layerand a few thin layers. Because he effective number of interfaces couldn’tbe determined this sample has also been omitted

49

Appendix

SL12(1) 100L12(1)10

SL6(1) 100SL6(1)100

100 nm Tynell et al. 2014

SL12(1)c 100

GM5(1) 20 100

GM5(1)b 20 100

GM12(1) 7 100

GM12(1) Fib 100

GM5(1) -20 1005(1)

201

5(1)b20

M12(1)7

1

GM12(1) ±8M

100GM

12(1) ±8M

rev 100

SL5(1) 100SL5(1)100

L12(1)c10

M5(1)20

1

SW1(12) 100

GM5(5) -20 1005(5)

201

12(1)Fib

GM12(1)

GM5(1) 20S 100

Publication III

SL12(1)b 100L12(1)b

10

SL6(1)b 100L6(1)b

10

SL24(1) 100

SL12(3) 100

SL12(5) 100

SL12(7) 100

Giri et al. 2016SL48(1) 100 & SL98(1)not pictured.

SL2(1) 50

SL3(1) 50

SL5(1) 50

50 nm

SL12(1)200SL18(1) 200

SL24(1) 200

SL10(1) 200SL5(1) 200

Doublet 10(1) 200

SW SL12(1)200

200 nm

SW SL5(1)200

GM10(1)

M±30 200

GM5(1) 50 200

GM10(1)

+18 200

SL12(1)d 100L12(1)d

10SL10(1) 100L10(1)10

SL4(1) 100SL4(1)100

SL3(1) 100SL3(1)100

SL5(1)b 100SL8(1) 100

Publication IV

Repetitions for cross validation

Figure

7.1.AllZnO

/ZnO-benzene

samples

with

reportedtherm

alconductivity.

50

References

[1] T. Suntola, “Atomic layer epitaxy,” Materials Science Reports, vol. 4, no. 5,pp. 261–312, 1989.

[2] M. Leskelä and M. Ritala, “Atomic layer deposition (ALD): from precursorsto thin film structures,” Thin Solid Films, vol. 409, no. 1, pp. 138–146, 2002.

[3] A. Ohtomo and H. Y. Hwang, “A high-mobility electron gas at theLaAlO3/SrTiO3 heterointerface,” Nature, vol. 427, no. 6973, pp. 423–426,2004.

[4] K. Nomura, H. Ohta, K. Ueda, T. Kamiya, M. Hirano, and H. Hosono,“Thin-film transistor fabricated in single-crystalline transparent oxide semi-conductor,” Science, vol. 300, no. 5623, pp. 1269–1272, 2003.

[5] T. Tynell, A. Giri, J. Gaskins, P. E. Hopkins, P. Mele, K. Miyazaki, andM. Karppinen, “Efficiently suppressed thermal conductivity in ZnO thinfilms via periodic introduction of organic layers,” Journal of MaterialsChemistry A, vol. 2, no. 31, pp. 12150–12152, 2014.

[6] A. Giri, J.-P. Niemelä, T. Tynell, J. T. Gaskins, B. F. Donovan, M. Karppinen,and P. E. Hopkins, “Heat-transport mechanisms in molecular buildingblocks of inorganic/organic hybrid superlattices,” Physical Review B, vol. 93,no. 11, p. 115310, 2016.

[7] I. Tamm, “Über eine mögliche Art der Elektronenbindung an Kristallober-flächen,” Zeitschrift für Physik, vol. 76, no. 11, pp. 849–850, 1932.

[8] W. Shockley, “On the surface states associated with a periodic potential,”Physical Review, vol. 56, no. 4, pp. 317–323, 1939.

[9] F. Bloch, “Über die Quantenmechanik der Elektronen in Kristallgittern,”Zeitschrift für Physik, vol. 52, no. 7, pp. 555–600, 1929.

[10] M. S. Dresselhaus, G. Chen, M. Y. Tang, R. G. Yang, H. Lee, D. Z. Wang,Z. F. Ren, J.-P. Fleurial, and P. Gogna, “New directions for low-dimensionalthermoelectric materials,” Advanced Materials, vol. 19, no. 8, pp. 1043–1053,2007.

[11] C. Forman, I. K. Muritala, R. Pardemann, and B. Meyer, “Estimating theglobal waste heat potential,” Renewable and Sustainable Energy Reviews,vol. 57, pp. 1568–1579, 2016.

[12] C. Goupil, H. Ouerdane, K. Zabrocki, W. Seifert, N. F. Hinsche, andE. Müller, Continuum Theory and Modeling of Thermoelectric Elements.John Wiley & Sons, first edition ed., 2016.

51

References

[13] T. J. Seebeck, “Magnetische polarisation der metall und erze durchtemperatur-differenz,” Abhandlungen der Königlichen Akademie der Wis-senschaften zu Berlin, 1822.

[14] D. Rowe, Thermoelectrics Handbook: Macro to Nano. CRC PressTaylor &Francis Group 6000 Broken Sound Parkway NW, Suite 300 Boca Raton, FL33487-2742: CRC Press Taylor & Francis, 1st ed., 2006.

[15] M. Cutler, J. F. Leavy, and R. L. Fitzpatrick, “Electronic transport insemimetallic cerium sulfide,” Physical Review, vol. 133, no. 4A, pp. A1143–A1152, 1964.

[16] G. J. Snyder and E. S. Toberer, “Complex thermoelectric materials,” NatureMaterials, vol. 7, no. 2, pp. 105–114, 2008.

[17] A. V. Sanchela, T. Onozato, B. Feng, Y. Ikuhara, and H. Ohta, “Thermopowermodulation clarification of the intrinsic effective mass in transparent oxidesemiconductor BaSnO3,” Physical Review Materials, vol. 1, no. 3, p. 034603,2017.

[18] L. Wei, P. K. Kuo, R. L. Thomas, T. R. Anthony, and W. F. Banholzer,“Thermal conductivity of isotopically modified single crystal diamond,” Phys.Rev. Lett., vol. 70, pp. 3764–3767, Jun 1993.

[19] C. Kittel, Introduction to Solid State Physics. New York: John Wiley &Sons, Inc., 8th edition ed., 2005.

[20] G. A. Slack and D. M. Rowe, CRC Handbook of Thermoelectrics. CRC Press,1. ed., 1995.

[21] G. S. Nolas, J. Poon, and M. Kanatzidis, “Recent developments in bulkthermoelectric materials,” MRS Bulletin, vol. 31, no. 3, pp. 199–205, 2006.

[22] X. Chen and H. Baumgart, “Advances in atomic layer deposition (ald)nanolaminate synthesis of thermoelectric films in porous templates forimproved seebeck coefficient,” Materials, vol. 13, no. 6, p. 1283, 2020.

[23] A. Giri and P. E. Hopkins, “A review of experimental and computationaladvances in thermal boundary conductance and nanoscale thermal trans-port across solid interfaces,” Advanced Functional Materials, vol. 30, no. 8,p. 1903857, 2020.

[24] K. Kothari and M. Maldovan, “Phonon surface scattering and thermalenergy distribution in superlattices,” Scientific Reports, vol. 7, no. 1, pp. 1–15, 2017.

[25] K. Koumoto, I. Terasaki, and R. Funahashi, “Complex oxide materials forpotential thermoelectric applications,” Materials Research Society Bulletin,vol. 31, no. 3, pp. 206–210, 2006.

[26] K. Momma and F. Izumi, “Vesta 3 for three-dimensional visualization ofcrystal, volumetric and morphology data,” Journal of Applied Crystallogra-phy, vol. 44, no. 6, pp. 1272–1276, 2011.

[27] S. Gražulis, D. Chateigner, R. T. Downs, A. F. T. Yokochi, M. Quirós, L. Lut-terotti, E. Manakova, J. Butkus, P. Moeck, and A. Le Bail, “Crystallographyopen database - an open-access collection of crystal structures,” Journal ofApplied Crystallography, vol. 42, no. 4, pp. 726–729, 2009.

[28] L. Weber, “Die Struktur von ZnO.,” Zeitschrift fuer Kristallographie,Kristallgeometrie, Kristallphysik, Kristallchemie (-144,1977), vol. 58,pp. 398–403, 1923.

52

References

[29] . Özgür, Y. I. Alivov, C. Liu, A. Teke, M. A. Reshchikov, S. Dogan, V. Avrutin,S.-J. Cho, and H. Morkoç, “A comprehensive review of ZnO materials anddevices,” Journal of Applied Physics, vol. 98, no. 4, pp. 041301–1 – 041301–103, 2005.

[30] H. Morkoç and Ümit Özgür, Zinc Oxide - Fundamentals, Materials andDevice Technology. Weinheim: Wiley-VCH Verlag GmbH & Co KgaA, 2007.

[31] J. Serrano, F. J. Manjón, A. H. Romero, A. Ivanov, M. Cardona, R. Lauck,A. Bosak, and M. Krisch, “Phonon dispersion relations of zinc oxide: Inelas-tic neutron scattering and ab initio calculations,” Physical Review B, vol. 81,no. 17, p. 174304, 2010.

[32] X. Wu, J. Lee, V. Varshney, J. L. Wohlwend, A. K. Roy, and T. Luo, “Thermalconductivity of wurtzite zinc-oxide from first-principles lattice dynamics - acomparative study with gallium nitride,” Scientific Reports, vol. 6, p. 22504,2016.

[33] Y. Geng, L. Guo, S.-S. Xu, Q.-Q. Sun, S.-J. Ding, H.-L. Lu, and D. W. Zhang,“Influence of al doping on the properties of ZnO thin films grown by atomiclayer deposition,” The Journal of Physical Chemistry C, vol. 115, no. 25,pp. 12317–12321, 2011.

[34] M. Ohtaki, T. Tsubota, K. Eguchi, and H. Arai, “High-temperature ther-moelectric properties of (Zn1–xAlx)O,” Journal of Applied Physics, vol. 79,no. 3, pp. 1816–1818, 1996.

[35] M. Ohtaki, K. Araki, and K. Yamamoto, “High thermoelectric performanceof dually doped ZnO ceramics,” Journal of Electronic Materials, vol. 38,no. 7, pp. 1234–1238, 2009.

[36] K. F. Cai, E. Müller, C. Drašar, and A. Mrotzek, “Preparation and ther-moelectric properties of al-doped ZnO ceramics,” Materials Science andEngineering: B, vol. 104, no. 1, pp. 45–48, 2003.

[37] T. Tsubota, M. Ohtaki, K. Eguchi, and H. Arai, “Transport properties andthermoelectric performance of (Zn1–yMgy)1–xAlxO,” Journal of MaterialsChemistry, vol. 8, no. 2, pp. 409–412, 1998.

[38] T. Tynell and M. Karppinen, “Atomic layer deposition of ZnO: a review,”Semiconductor Science and Technology, vol. 29, no. 4, p. 043001, 2014.

[39] A. Ohtomo and A. Tsukazaki, “Pulsed laser deposition of thin films andsuperlattices based on zno,” Semiconductor Science and Technology, vol. 20,no. 4, pp. S1–S12, 2005.

[40] S. M. George, “Atomic layer deposition: An overview,” Chemical Reviews,vol. 110, no. 1, pp. 111–131, 2010.

[41] M. Jenkins, D. Z. Austin, J. F. Conley, J. Fan, C. H. de Groot, L. Jiang,Y. Fan, R. Ali, G. Ghosh, M. Orlowski, and S. W. King, “Review—beyondthe highs and lows: A perspective on the future of dielectrics research fornanoelectronic devices,” ECS Journal of Solid State Science and Technology,vol. 8, no. 11, pp. N159–N185, 2019.

[42] R. L. Puurunen, “A short history of atomic layer deposition: Tuomo Suntola’satomic layer epitaxy,” Chemical Vapor Deposition, vol. 20, no. 10-11-12,pp. 332–344, 2014.

[43] P. Sundberg and M. Karppinen, “Organic and inorganic-organic thin filmstructures by molecular layer deposition: A review,” Beilstein Journal ofNanotechnology, vol. 5, pp. 1104–1136, 2014.

53

References

[44] D. S. Bergsman, J. G. Baker, R. G. Closser, C. MacIsaac, M. Lillethorup,A. L. Strickler, L. Azarnouche, L. Godet, and S. F. Bent, “Structurally stablemanganese alkoxide films grown by hybrid molecular layer depositionfor electrochemical applications,” Advanced Functional Materials, vol. 29,no. 43, p. 1904129, 2019.

[45] S. M. George, A. W. Ott, and J. W. Klaus, “Surface chemistry for atomic layergrowth,” The Journal of Physical Chemistry, vol. 100, no. 31, pp. 13121–13131, 1996.

[46] S. D. Elliott, “Atomic-scale simulation of ald chemistry,” SemiconductorScience and Technology, vol. 27, no. 7, p. 074008, 2012.

[47] V. Cremers, R. L. Puurunen, and J. Dendooven, “Conformality in atomiclayer deposition: Current status overview of analysis and modelling,” Ap-plied Physics Reviews, vol. 6, no. 2, p. 021302, 2019.

[48] C. W. Wiegand, R. Faust, A. Meinhardt, R. H. Blick, R. Zierold, andK. Nielsch, “Understanding the growth mechanisms of multilayered sys-tems in atomic layer deposition process,” Chemistry of Materials, vol. 30,no. 6, pp. 1971–1979, 2018.

[49] S. M. George, B. Yoon, and A. A. Dameron, “Surface chemistry for molec-ular layer deposition of organic and hybrid organic-inorganic polymers,”Accounts of Chemical Research, vol. 42, no. 4, pp. 498–508, 2009.

[50] B. Yoon, Y. Lee, A. Derk, C. Musgrave, and S. George, “Molecular layer de-position of conductive hybrid organic-inorganic thin films using diethylzincand hydroquinone,” ECS Transactions, vol. 33, no. 27, pp. 191–195, 2011.

[51] A. J. M. Mackus, J. R. Schneider, C. MacIsaac, J. G. Baker, and S. F. Bent,“Synthesis of doped, ternary, and quaternary materials by atomic layerdeposition: A review,” Chemistry of Materials, vol. 31, no. 4, pp. 1142–1183,2019.

[52] F. Larsson, L. Stolt, A. Hultqvist, M. Edoff, J. Keller, and T. Törndahl,“Atomic layer deposition of ternary compounds on Cu(In, Ga)Se2: An in situquartz crystal microbalance study,” ACS Applied Energy Materials, vol. 3,no. 7, pp. 7208–7215, 2020.

[53] M. H. Cho, H. Seol, A. Song, S. Choi, Y. Song, P. S. Yun, K.-B. Chung, J. U.Bae, K.-S. Park, and J. K. Jeong, “Comparative study on performance ofigzo transistors with sputtered and atomic layer deposited channel layer,”IEEE Transactions on Electron Devices, vol. 66, no. 4, pp. 1783–1788, 2019.

[54] A. Suresh, P. Gollakota, P. Wellenius, A. Dhawan, and J. F. Muth, “Trans-parent, high mobility InGaZnO thin films deposited by pld,” Thin SolidFilms, vol. 516, no. 7, pp. 1326–1329, 2008.

[55] K. L. Narasimhan, S. P. Pai, V. R. Palkar, and R. Pinto, “High quality zincoxide films by pulsed laser ablation,” Thin Solid Films, vol. 295, no. 1,pp. 104–106, 1997.

[56] N. Alimardani, E. William Cowell, J. F. Wager, J. F. Conley, D. R. Evans,M. Chin, S. J. Kilpatrick, and M. Dubey, “Impact of electrode roughnesson metal-insulator-metal tunnel diodes with atomic layer deposited Al2O3tunnel barriers,” Journal of Vacuum Science & Technology A, vol. 30, no. 1,p. 01A113, 2011.

[57] C. F. Klingshirn, A. Waag, J. Geurts, A. Hoffmann, and B. K. Meyer, ZincOxide - From Fundamental Properties Towards Novel Applications. SpringerHeidelberg, Dordrecht, London, New York: Springer, 2010.

54

References

[58] N. P. Dasgupta, S. Neubert, W. Lee, O. Trejo, J.-R. Lee, and F. B. Prinz,“Atomic layer deposition of Al-doped ZnO films: Effect of grain orientationon conductivity,” Chemistry of Materials, vol. 22, no. 16, pp. 4769–4775,2010.

[59] T. Tynell, I. Terasaki, H. Yamauchi, and M. Karppinen, “Thermoelectriccharacteristics of (Zn, Al) O/hydroquinone superlattices,” Journal of Materi-als Chemistry A, vol. 1, no. 43, pp. 13619–13624, 2013.

[60] J. M. Jensen, A. B. Oelkers, R. Toivola, D. C. Johnson, J. W. Elam, andS. M. George, “X-ray reflectivity characterization of ZnO/Al2O3 multilayersprepared by atomic layer deposition,” Chemistry of Materials, vol. 14, no. 5,pp. 2276–2282, 2002.

[61] R. L. Puurunen, “Surface chemistry of atomic layer deposition: A case studyfor the trimethylaluminum/water process,” Journal of Applied Physics,vol. 97, no. 12, pp. –, 2005.

[62] D.-J. Lee, H.-M. Kim, J.-Y. Kwon, H. Choi, S.-H. Kim, and K.-B. Kim,“Structural and electrical properties of atomic layer deposited al-doped znofilms,” Advanced Functional Materials, vol. 21, no. 3, pp. 448–455, 2011.

[63] C. H. Ahn, S. H. Kim, Y. K. Kim, H. S. Lee, and H. K. Cho, “Effect ofpost-annealing temperatures on thin-film transistors with ZnO/Al2O3 su-perlattice channels,” Thin Solid Films, vol. 584, pp. 336–340, 2015.

[64] J.-H. Lee, T.-H. Park, S.-H. Lee, N.-W. Park, W.-Y. Lee, J.-H. Lim, S.-K. Lee,and J.-S. Park, “Enhancing the thermoelectric properties of super-latticeAl2O3/ZnO atomic film via interface confinement,” Ceramics International,vol. 42, no. 13, pp. 14411–14415, 2016.

[65] J. W. Elam and S. M. George, “Growth of ZnO/Al2O3 alloy films usingatomic layer deposition techniques,” Chemistry of Materials, vol. 15, no. 4,pp. 1020–1028, 2003.

[66] D. R. Zywotko and S. M. George, “Thermal atomic layer etching of zno bya “conversion-etch” mechanism using sequential exposures of hydrogenfluoride and trimethylaluminum,” Chemistry of Materials, vol. 29, no. 3,pp. 1183–1191, 2017.

[67] J.-S. Na, Q. Peng, G. Scarel, and G. N. Parsons, “Role of gas doping sequencein surface reactions and dopant incorporation during atomic layer deposi-tion of Al-doped ZnO,” Chemistry of Materials, vol. 21, no. 23, pp. 5585–5593,2009.

[68] B. Yoon, B. H. Lee, and S. M. George, “Highly conductive and transpar-ent hybrid organic-inorganic zincone thin films using atomic and molec-ular layer deposition,” Journal of Physical Chemistry C, vol. 116, no. 46,pp. 24784–24791, 2012.

[69] R. Ghiyasi, G. C. Tewari, and M. Karppinen, “Organic-component dependentcrystal orientation and electrical transport properties in ALD/MLD grownZnO-organic superlattices,” The Journal of Physical Chemistry C, 2020.

[70] T. Tynell and M. Karppinen, “ZnO: Hydroquinone superlattice structuresfabricated by atomic/molecular layer deposition,” Thin Solid Films, vol. 551,pp. 23–26, 2014.

[71] D. Choudhury and S. K. Sarkar, “The ald-mld growth of a ZnO-zinconeheterostructure,” Chemical Vapor Deposition, vol. 20, no. 4-5-6, pp. 130–137,2014.

55

References

[72] A. J. Karttunen, T. Tynell, and M. Karppinen, “Atomic-level structural andelectronic properties of hybrid inorganic-organic ZnO:hydroquinone super-lattices fabricated by ald/mld,” Journal of Physical Chemistry C, vol. 119,no. 23, pp. 13105–13114, 2015.

[73] A. J. Karttunen, T. Tynell, and M. Karppinen, “Layer-by-layer design ofnanostructured thermoelectrics: First-principles study of ZnO:organic su-perlattices fabricated by ALD/MLD,” Nano Energy, vol. 22, pp. 338–348,2016.

[74] J.-P. Niemelä, M. Aghaee, W. M. M. E. Kessels, M. Creatore, and M. A.Verheijen, “Transition in layer structure of atomic/molecular layer depositedZnO-zincone multilayers,” Journal of Vacuum Science & Technology A,vol. 37, no. 4, p. 040602, 2019.

[75] P. P. Ewald, ed., 50 Years of X-Ray Diffraction. N.V. A. Oosthoek’s Uitgev-ersmaatschappij, Utrecht, The Netherlands, 1 ed., 1962.

[76] A. J. Blake, W. Clegg, J. M. Cole, J. S. O. Evans, P. Main, S. Parsons, andD. J. Watkin, Crystal Structure Analysis - Principles and Practice, vol. 13 ofIUCr Texts on Crystallography. Oxford University Press, 2 ed., 2009.

[77] M. Birkholz, Thin Film Analysis by X-Ray Scattering. Wiley-VCH VerlagGmbH & Co. KGaA, Weinheim, 2006.

[78] V. Holý, U. Pietsch, and T. Baumbach, High-Resolution X-Ray Scatteringfrom Thin Films and Multilayers, vol. 149 of Springe tracts in modernphysics. Spinger, 1999. page 127.

[79] H. Kiessig, “Interferenz von Röntgenstrahlen an dünnen Schichten,” An-nalen der Physik, vol. 402, no. 7, pp. 769–788, 1931.

[80] L. G. Parratt, “Surface studies of solids by total reflection of x-rays,” PhysicalReview, vol. 95, no. 2, pp. 359–369, 1954.

[81] P. F. Fewster, “X-ray analysis of thin films and multilayers,” Reports onProgress in Physics, vol. 59, no. 11, pp. 1339–1407, 1996.

[82] P. F. Fewster, X-Ray Scattering from Semiconductors. River Edge, NJ:Imperial College Pr, 2nd edition ed., 2003.

[83] E. Chason and T. M. Mayer, “Thin film and surface characterization byspecular x-ray reflectivity,” Critical Reviews in Solid State and MaterialsSciences, vol. 22, no. 1, pp. 1–67, 1997.

[84] M. Yasaka, “X-ray thin-film measurement techniques v. x-ray reflectivitymeasurement,” Rigaku journal, vol. 26(2), pp. 1–9, 2010.

[85] J. G. Bednorz and K. A. Müller, “Possible high tc superconductivity in theBa-La-Cu-O system,” Zeitschrift für Physik B Condensed Matter, vol. 64,no. 2, pp. 189–193, 1986.

[86] M. K. Wu, J. R. Ashburn, C. J. Torng, P. H. Hor, R. L. Meng, L. Gao, Z. J.Huang, Y. Q. Wang, and C. W. Chu, “Superconductivity at 93 k in a newmixed-phase Y-Ba-Cu-O compound system at ambient pressure,” PhysicalReview Letters, vol. 58, no. 9, pp. 908–910, 1987.

[87] H. Y. Hwang, Y. Iwasa, M. Kawasaki, B. Keimer, N. Nagaosa, and Y. Tokura,“Emergent phenomena at oxide interfaces,” Nature Materials, vol. 11, p. 103,2012.

[88] A. Giampietri, G. Drera, and L. Sangaletti, “Band alignment at heteroepi-taxial perovskite oxide interfaces. experiments, methods, and perspectives,”Advanced Materials Interfaces, vol. 4, no. 11, p. 1700144, 2017.

56

References

[89] N. Nakagawa, H. Y. Hwang, and D. A. Muller, “Why some interfaces cannotbe sharp,” Nature Materials, vol. 5, no. 3, pp. 204–209, 2006.

[90] R. Dingle, W. Wiegmann, and C. H. Henry, “Quantum states of confined car-riers in very thin AlxGa1–xAs–GaAs –AlxGa1–xAs heterostructures,” Phys-ical Review Letters, vol. 33, no. 14, pp. 827–830, 1974.

[91] Y. Z. Chen, N. Bovet, F. Trier, D. V. Christensen, F. M. Qu, N. H. Andersen,T. Kasama, W. Zhang, R. Giraud, J. Dufouleur, T. S. Jespersen, J. R. Sun,A. Smith, J. Nygård, L. Lu, B. Büchner, B. G. Shen, S. Linderoth, andN. Pryds, “A high-mobility two-dimensional electron gas at the spinel/per-ovskite interface of γ-Al2O3/SrTiO3,” Nature Communications, vol. 4, no. 1,p. 1371, 2013.

[92] H. Ohta, S. Kim, Y. Mune, T. Mizoguchi, K. Nomura, S. Ohta, T. Nomura,Y. Nakanishi, Y. Ikuhara, M. Hirano, H. Hosono, and K. Koumoto, “Giantthermoelectric seebeck coefficient of a two-dimensional electron gas inSrTiO3,” Nature Materials, vol. 6, p. 129, 2007.

[93] P. D. C. King, R. H. He, T. Eknapakul, P. Buaphet, S.-K. Mo, Y. Kaneko,S. Harashima, Y. Hikita, M. S. Bahramy, C. Bell, Z. Hussain, Y. Tokura,Z.-X. Shen, H. Y. Hwang, F. Baumberger, and W. Meevasana, “Subbandstructure of a two-dimensional electron gas formed at the polar surface ofthe strong spin-orbit perovskite KTaO3,” Physical Review Letters, vol. 108,no. 11, p. 117602, 2012.

[94] A. Tsukazaki, A. Ohtomo, T. Kita, Y. Ohno, H. Ohno, and M. Kawasaki,“Quantum hall effect in polar oxide heterostructures,” Science, vol. 315,no. 5817, pp. 1388–1391, 2007.

[95] I. P. Smorchkova, C. R. Elsass, J. P. Ibbetson, R. Vetury, B. Heying, P. Fini,E. Haus, S. P. DenBaars, J. S. Speck, and U. K. Mishra, “Polarization-induced charge and electron mobility in AlGaN/GaN heterostructuresgrown by plasma-assisted molecular-beam epitaxy,” Journal of AppliedPhysics, vol. 86, no. 8, pp. 4520–4526, 1999.

[96] K. Nomura, H. Ohta, K. Ueda, T. Kamiya, M. Orita, M. Hirano, T. Suzuki,C. Honjyo, Y. Ikuhara, and H. Hosono, “Growth mechanism for single-crystalline thin film of InGaO3(ZnO)5 by reactive solid-phase epitaxy,” Jour-nal of Applied Physics, vol. 95, no. 10, pp. 5532–5539, 2004.

[97] K. Nomura, H. Ohta, A. Takagi, T. Kamiya, M. Hirano, and H. Hosono,“Room-temperature fabrication of transparent flexible thin-film transistorsusing amorphous oxide semiconductors,” Nature, vol. 432, no. 7016, pp. 488–492, 2004.

[98] H. Ohta and H. Hosono, “Transparent oxide optoelectronics,” MaterialsToday, vol. 7, no. 6, pp. 42–51, 2004.

[99] H. Hosono, “How we made the igzo transistor,” Nature Electronics, vol. 1,no. 7, pp. 428–428, 2018.

[100] H. Hosono, M. Yasukawa, and H. Kawazoe, “Novel oxide amorphous semi-conductors: transparent conducting amorphous oxides,” Journal of Non-Crystalline Solids, vol. 203, pp. 334–344, 1996.

[101] H. Ohta, R. Huang, and Y. Ikuhara, “Large enhancement of the thermoelec-tric seebeck coefficient for amorphous oxide semiconductor superlatticeswith extremely thin conductive layers,” physica status solidi (RRL) - RapidResearch Letters, vol. 2, no. 3, pp. 105–107, 2008.

57

References

[102] X. Liang, M. Baram, and D. R. Clarke, “Thermal (kapitza) resistance ofinterfaces in compositional dependent ZnO-In2O3 superlattices,” AppliedPhysics Letters, vol. 102, no. 22, p. 223903, 2013.

[103] V. Bhosle, A. Tiwari, and J. Narayan, “Electrical properties of transparentand conducting Ga doped ZnO,” Journal of Applied Physics, vol. 100, no. 3,p. 033713, 2006.

[104] A. Hafdallah, F. Yanineb, M. S. Aida, and N. Attaf, “In doped ZnO thinfilms,” Journal of Alloys and Compounds, vol. 509, no. 26, pp. 7267–7270,2011.

[105] V. Narayanamurti, H. L. Störmer, M. A. Chin, A. C. Gossard, and W. Wieg-mann, “Selective transmission of high-frequency phonons by a superlattice:The "dielectric" phonon filter,” Physical Review Letters, vol. 43, no. 27,pp. 2012–2016, 1979.

[106] M. J. Kelly, “Acoustic phonon transmission in superlattices,” Journal ofPhysics C: Solid State Physics, vol. 18, no. 32, p. 5963, 1985.

[107] R. Cheaito, C. A. Polanco, S. Addamane, J. Zhang, A. W. Ghosh, G. Balakr-ishnan, and P. E. Hopkins, “Interplay between total thickness and periodthickness in the phonon thermal conductivity of superlattices from thenanoscale to the microscale: Coherent versus incoherent phonon transport,”Physical Review B, vol. 97, no. 8, p. 085306, 2018.

[108] J. Ravichandran, A. K. Yadav, R. Cheaito, P. B. Rossen, A. Soukiassian, S. J.Suresha, J. C. Duda, B. M. Foley, C.-H. Lee, Y. Zhu, A. W. Lichtenberger, J. E.Moore, D. A. Muller, D. G. Schlom, P. E. Hopkins, A. Majumdar, R. Ramesh,and M. A. Zurbuchen, “Crossover from incoherent to coherent phononscattering in epitaxial oxide superlattices,” Nature Materials, vol. 13, no. 2,pp. 168–172, 2014.

[109] M. N. Luckyanova, J. Garg, K. Esfarjani, A. Jandl, M. T. Bulsara, A. J.Schmidt, A. J. Minnich, S. Chen, M. S. Dresselhaus, Z. Ren, E. A. Fitzgerald,and G. Chen, “Coherent phonon heat conduction in superlattices,” Science,vol. 338, no. 6109, pp. 936–939, 2012.

[110] R. M. Costescu, D. G. Cahill, F. H. Fabreguette, Z. A. Sechrist, and S. M.George, “Ultra-low thermal conductivity in W/Al2O3 nanolaminates,” Sci-ence, vol. 303, no. 5660, pp. 989–990, 2004.

[111] D. G. Cahill, A. Bullen, and S.-M. Lee, “Interface thermal conductance andthe thermal conductivity of multilayer thin films,” High Temperatures -High Pressures, vol. 32, pp. 135–142, 2000.

[112] D. W. Song, W. L. Liu, T. Zeng, T. Borca-Tasciuc, G. Chen, J. C. Caylor,and T. D. Sands, “Thermal conductivity of skutterudite thin films andsuperlattices,” Applied Physics Letters, vol. 77, no. 23, pp. 3854–3856, 2000.

[113] Y. K. Koh, Y. Cao, D. G. Cahill, and D. Jena, “Heat-transport mechanisms insuperlattices,” Advanced Functional Materials, vol. 19, no. 4, pp. 610–615,2009.

[114] H. Ohta, S. W. Kim, S. Kaneki, A. Yamamoto, and T. Hashizume, “Highthermoelectric power factor of high-mobility 2d electron gas,” AdvancedScience, vol. 5, no. 1, p. 1700696, 2018.

[115] R. M. Costescu, M. A. Wall, and D. G. Cahill, “Thermal conductance ofepitaxial interfaces,” Physical Review B, vol. 67, no. 5, p. 054302, 2003.

58

References

[116] P. Hołuj, C. Euler, B. Balke, U. Kolb, G. Fiedler, M. M. Müller, T. Jaeger,E. Chávez Angel, P. Kratzer, and G. Jakob, “Reduced thermal conductivity ofTiNiSn/HfNiSn superlattices,” Physical Review B, vol. 92, no. 12, p. 125436,2015.

[117] T. Koga, S. B. Cronin, M. S. Dresselhaus, J. L. Liu, and K. L. Wang, “Experi-mental proof-of-principle investigation of enhanced Z3DT in (001) orientedSi/Ge superlattices,” Applied Physics Letters, vol. 77, no. 10, pp. 1490–1492,2000.

[118] S. T. Huxtable, A. R. Abramson, C.-L. Tien, A. Majumdar, C. LaBounty,X. Fan, G. Zeng, J. E. Bowers, A. Shakouri, and E. T. Croke, “Thermal con-ductivity of Si/SiGe and SiGe/SiGe superlattices,” Applied Physics Letters,vol. 80, no. 10, pp. 1737–1739, 2002.

[119] T. Borca-Tasciuc, W. Liu, J. Liu, T. Zeng, D. W. Song, C. D. Moore, G. Chen,K. L. Wang, M. S. Goorsky, T. Radetic, R. Gronsky, T. Koga, and M. S.Dresselhaus, “Thermal conductivity of symmetrically strained Si/Ge su-perlattices,” Superlattices and Microstructures, vol. 28, no. 3, pp. 199–206,2000.

[120] S.-M. Lee, D. G. Cahill, and R. Venkatasubramanian, “Thermal conductivityof Si-Ge superlattices,” Applied Physics Letters, vol. 70, no. 22, pp. 2957–2959, 1997.

[121] S. Chakraborty, C. A. Kleint, A. Heinrich, C. M. Schneider, J. Schumann,M. Falke, and S. Teichert, “Thermal conductivity in strain symmetrizedSi/Ge superlattices on Si(111),” Applied Physics Letters, vol. 83, no. 20,pp. 4184–4186, 2003.

[122] G. Pernot, M. Stoffel, I. Savic, F. Pezzoli, P. Chen, G. Savelli, A. Jacquot,J. Schumann, U. Denker, I. Mönch, C. Deneke, O. G. Schmidt, J. M. Ramp-noux, S. Wang, M. Plissonnier, A. Rastelli, S. Dilhaire, and N. Mingo,“Precise control of thermal conductivity at the nanoscale through individualphonon-scattering barriers,” Nature Materials, vol. 9, no. 6, pp. 491–5, 2010.

[123] A. Giri, B. F. Donovan, and P. E. Hopkins, “Localization of vibrationalmodes leads to reduced thermal conductivity of amorphous heterostruc-tures,” Physical Review Materials, vol. 2, no. 5, p. 056002, 2018.

[124] W. S. Capinski, H. J. Maris, T. Ruf, M. Cardona, K. Ploog, and D. S. Katzer,“Thermal-conductivity measurements of GaAs/AlAs superlattices using apicosecond optical pump-and-probe technique,” Phys. Rev. B, vol. 59, no. 12,pp. 8105–8113, 1999.

[125] M. N. Luckyanova, J. A. Johnson, A. A. Maznev, J. Garg, A. Jandl, M. T.Bulsara, E. A. Fitzgerald, K. A. Nelson, and G. Chen, “Anisotropy of thethermal conductivity in GaAs/AlAs superlattices,” Nano Letters, vol. 13,no. 9, pp. 3973–3977, 2013.

[126] R. Venkatasubramanian, “Lattice thermal conductivity reduction andphonon localizationlike behavior in superlattice structures,” Physical Re-view B, vol. 61, no. 4, pp. 3091–3097, 2000.

[127] R. Venkatasubramanian, E. Siivola, T. Colpitts, and B. O’Quinn, “Thin-film thermoelectric devices with high room-temperature figures of merit,”Nature, vol. 413, no. 6856, pp. 597–602, 2001.

[128] T. Borca-Tasciuc, S. S. Pei, D. Achimov, W. L. Liu, G. Chen, H.-W. Ren, andC.-H. Lin, “Thermal conductivity of InAs/AlSb superlattices,” MicroscaleThermophysical Engineering, vol. 5, no. 3, pp. 225–231, 2001.

59

References

[129] Y. Zhang, Y. Chen, C. Gong, J. Yang, R. Qian, and Y. Wang, “Optimizationof superlattice thermoelectric materials and microcoolers,” J. Microelec-tromech. Syst., vol. 16, no. 5, pp. 1113–1119, 2007.

[130] S. Ali, T. Juntunen, S. Sintonen, O. M. E. Ylivaara, R. L. Puurunen, H. Lip-sanen, I. Tittonen, and S.-P. Hannula, “Thermal conductivity of amorphousAl2O3/TiO2 nanolaminates deposited by atomic layer deposition,” Nanotech-nology, vol. 27, no. 44, p. 445704, 2016.

[131] H. Tong, X. S. Miao, X. M. Cheng, H. Wang, L. Zhang, J. J. Sun, F. Tong,and J. H. Wang, “Thermal conductivity of chalcogenide material with su-perlatticelike structure,” Applied Physics Letters, vol. 98, no. 10, p. 101904,2011.

[132] A. France-Lanord, S. Merabia, T. Albaret, D. Lacroix, and K. Termentzidis,“Thermal properties of amorphous/crystalline silicon superlattices,” Journalof Physics: Condensed Matter, vol. 26, no. 35, p. 355801, 2014.

[133] L. D. Hicks and M. S. Dresselhaus, “Effect of quantum-well structureson the thermoelectric figure of merit,” Physical Review B, vol. 47, no. 19,pp. 12727–12731, 1993.

[134] L. D. Hicks, T. C. Harman, X. Sun, and M. S. Dresselhaus, “Experimentalstudy of the effect of quantum-well structures on the thermoelectric figureof merit,” Physical Review B, vol. 53, no. 16, pp. R10493–R10496, 1996.

[135] P. M. Norris, N. Q. Le, and C. H. Baker, “Tuning phonon transport: Frominterfaces to nanostructures,” Journal of Heat Transfer, vol. 135, no. 6,pp. 061604–061604–13, 2013.

[136] A. Giri, J.-P. Niemelä, C. J. Szwejkowski, M. Karppinen, and P. E. Hopkins,“Reduction in thermal conductivity and tunable heat capacity of inorgan-ic/organic hybrid superlattices,” Physical Review B, vol. 93, no. 2, p. 024201,2016.

[137] J.-P. Niemelä, A. J. Karttunen, and M. Karppinen, “Inorganic-organic su-perlattice thin films for thermoelectrics,” Journal of Materials Chemistry C,vol. 3, no. 40, pp. 10349–10361, 2015.

[138] J.-P. Niemelä, A. Giri, P. E. Hopkins, and M. Karppinen, “Ultra-low thermalconductivity in TiO2:C superlattices,” Journal of Materials Chemistry A,vol. 3, no. 21, pp. 11527–11532, 2015.

[139] R. Tian, C. Wan, Y. Wang, Q. Wei, T. Ishida, A. Yamamoto, A. Tsuruta,W. Shin, S. Li, and K. Koumoto, “A solution-processed TiS2/organic hy-brid superlattice film towards flexible thermoelectric devices,” Journal ofMaterials Chemistry A, vol. 5, no. 2, pp. 564–570, 2017.

[140] C. Wan, X. Gu, F. Dang, T. Itoh, Y. Wang, H. Sasaki, M. Kondo, K. Koga,K. Yabuki, G. J. Snyder, R. Yang, and K. Koumoto, “Flexible n-type ther-moelectric materials by organic intercalation of layered transition metaldichalcogenide TiS2,” Nature Materials, vol. 14, no. 6, pp. 622–627, 2015.

[141] C. Wan, R. Tian, M. Kondou, R. Yang, P. Zong, and K. Koumoto, “Ultrahighthermoelectric power factor in flexible hybrid inorganic-organic superlat-tice,” Nature Communications, vol. 8, no. 1, pp. 1–9, 2017.

[142] C. Wan, R. Tian, A. B. Azizi, Y. Huang, Q. Wei, R. Sasai, S. Wasusate,T. Ishida, and K. Koumoto, “Flexible thermoelectric foil for wearable energyharvesting,” Nano Energy, vol. 30, pp. 840–845, 2016.

60

References

[143] G. Marin, T. Tynell, and M. Karppinen, “Flexible thermoelectric modulesbased on ald-grown ZnO on different substrates,” Journal of Vacuum Science& Technology A, vol. 37, no. 2, p. 020906, 2019.

[144] G. Marin, R. Funahashi, and M. Karppinen, “Textile-integrated ZnO-basedthermoelectric device using atomic layer deposition (in press),” AdvancedEngineering Materials, 2020.

[145] A. J. Karttunen, L. Sarnes, R. Townsend, J. Mikkonen, and M. Karp-pinen, “Flexible thermoelectric ZnO-organic superlattices on cotton tex-tile substrates by ALD/MLD,” Advanced electronic Materials, vol. 3, no. 6,pp. 1600459–n/a, 2017.

[146] O. Nilsen, K. Klepper, H. Nielsen, and H. Fjellvåg, “Deposition of organic-inorganic hybrid materials by atomic layer deposition,” ECS Transactions,vol. 16, no. 4, pp. 3–14, 2008.

[147] B. Qiu, G. Chen, and Z. Tian, “Effects of aperiodicity and roughness oncoherent heat conduction in superlattices,” Nanoscale and Microscale Ther-mophysical Engineering, vol. 19, no. 4, pp. 272–278, 2015.

[148] P. Chakraborty, L. Cao, and Y. Wang, “Ultralow lattice thermal conductivityof the random multilayer structure with lattice imperfections,” ScientificReports, vol. 7, no. 1, p. 8134, 2017.

[149] Y. Wang, H. Huang, and X. Ruan, “Decomposition of coherent and incoher-ent phonon conduction in superlattices and random multilayers,” PhysicalReview B, vol. 90, no. 16, p. 165406, 2014.

[150] Y. Wang, C. Gu, and X. Ruan, “Optimization of the random multilayerstructure to break the random-alloy limit of thermal conductivity,” AppliedPhysics Letters, vol. 106, no. 7, p. 073104, 2015.

[151] C. A. Paddock and G. L. Eesley, “Transient thermoreflectance from thinmetal films,” Journal of Applied Physics, vol. 60, no. 1, pp. 285–290, 1986.

[152] P. E. Hopkins, J. R. Serrano, L. M. Phinney, S. P. Kearney, T. W. Grasser,and C. T. Harris, “Criteria for cross-plane dominated thermal transport inmultilayer thin film systems during modulated laser heating,” Journal ofHeat Transfer, vol. 132, no. 8, pp. 081302–081302–10, 2010.

[153] D. G. Cahill, “Analysis of heat flow in layered structures for time-domainthermoreflectance,” Review of Scientific Instruments, vol. 75, no. 12,pp. 5119–5122, 2004.

[154] P. Jiang, X. Qian, and R. Yang, “Tutorial: Time-domain thermoreflectance(tdtr) for thermal property characterization of bulk and thin film materials,”Journal of Applied Physics, vol. 124, no. 16, p. 161103, 2018.

[155] Z. M. Zhang, Nano/Microscale Heat Transfer. Gewerbestrasse 11, 6330Cham, Switzerland: Springer Nature Switzerland AG, 2nd ed., 2020.

[156] J. E. Jackson, A Users Guide to Principal Components. John Wiley & Sons,Inc., 1991.

[157] S. Wold, M. Sjöström, and L. Eriksson, “PLS-regression: a basic tool ofchemometrics,” Chemometrics and Intelligent Laboratory Systems, vol. 58,no. 2, pp. 109–130, 2001.

[158] S. Wold, A. Ruhe, H. Wold, and W. J. Dunn, Iii, “The collinearity problem inlinear regression. the partial least squares (PLS) approach to generalizedinverses,” SIAM Journal on Scientific and Statistical Computing, vol. 5,no. 3, pp. 735–743, 1984.

61

References

[159] N. Kettaneh-Wold, “Analysis of mixture data with partial least squares,”Chemometrics and Intelligent Laboratory Systems, vol. 14, no. 1, pp. 57–69,1992.

[160] R. Rosipal and N. Krämer, “Overview and recent advances in partial leastsquares,” in Subspace, Latent Structure and Feature Selection. (C. Saun-ders, M. Grobelnik, S. Gunn, and J. Shawe-Taylor, eds.), Lecture Notes inComputer Science, pp. 34–51, Springer, 2006.

[161] L. Eriksson, T. Byrne, E. Johansson, J. Trygg, and C. Vikström, Multi-and Megavariate Data Analysis Basic Principles and Applications. Malmö:Umetrics Academy, 2013.

[162] K. Lehmus and M. Karppinen, “Application of multivariate data analysistechniques in modeling structure-property relationships of some supercon-ductive cuprates,” Journal of Solid State Chemistry, vol. 162, no. 1, pp. 1–9,2001.

[163] N. Imamura, T. Mizoguchi, H. Yamauchi, and M. Karppinen, “Multivariatedata analysis approach to understand magnetic properties of perovskitemanganese oxides,” Journal of Solid State Chemistry, vol. 181, no. 5,pp. 1195–1203, 2008.

[164] T. Tiittanen, S. Vasala, and M. Karppinen, “Assessment of magnetic prop-erties of A2B′B′′O6 double perovskites by multivariate data analysis tech-niques,” Chemical Communications, vol. 55, no. 12, pp. 1722–1725, 2019.

[165] S. Cacovich, F. Matteocci, M. Abdi-Jalebi, S. D. Stranks, A. Di Carlo,C. Ducati, and G. Divitini, “Unveiling the chemical composition of halideperovskite films using multivariate statistical analyses,” ACS Applied En-ergy Materials, vol. 1, no. 12, pp. 7174–7181, 2018.

[166] J. Liu, B. Yoon, E. Kuhlmann, M. Tian, J. Zhu, S. M. George, Y.-C. Lee, andR. Yang, “Ultralow thermal conductivity of atomic/molecular layer-depositedhybrid organic-inorganic zincone thin films,” Nano Letters, vol. 13, no. 11,pp. 5594–5599, 2013.

[167] H. Jin, G. Marin, A. Giri, T. Tynell, M. Gestranius, B. P. Wilson, E. Kontturi,T. Tammelin, P. E. Hopkins, and M. Karppinen, “Strongly reduced thermalconductivity in hybrid ZnO/nanocellulose thin films,” Journal of MaterialsScience, vol. 52, no. 10, pp. 6093–6099, 2017.

[168] J. Alvarez-Quintana, E. Martínez, E. Pérez-Tijerina, S. A. Pérez-García, andJ. Rodríguez-Viejo, “Temperature dependent thermal conductivity of poly-crystalline zno films,” Journal of Applied Physics, vol. 107, no. 6, p. 063713,2010.

62

-otl

aA

DD

62

/ 1

20

2

+aeica

e*GMFTSH

9

NBSI 0-4820-46-259-879 )detnirp(

NBSI 7-5820-46-259-879 )fdp(

NSSI 4394-9971 )detnirp(

NSSI 2494-9971 )fdp(

ytisrevinU otlaA

gnireenignE lacimehC fo loohcS

ecneicS slairetaM dna yrtsimehC fo tnemtrapeD

fi.otlaa.www

+SSENISUBYMONOCE

+TRA+NGISED

ERUTCETIHCRA

+ECNEICSYGOLONHCET

REVOSSORC

LAROTCODSNOITATRESSID