Zinc oxide thin-films by spray pyrolysis with low deposition ...

Zinc oxide thin-films by spray pyrolysis

with low deposition temperature

by

Jonas Köhling

a Thesis submitted in partial fulfillment

of the requirements for the degree of

Doctor of Philosophy

in Physics

Approved Dissertation Committee

__________________________________

Prof. Dr. Veit Wagner

Prof. Dr. Gerd-Volker Röschenthaler

Prof. Dr. Ralf Anselmann

Date of Defense: 22 April 2021

Department of Physics and Earth Sciences

Statutory Declaration

Family Name, Given/First Name Köhling, Jonas

Matriculation number 20331904

What kind of thesis are you submitting: Bachelor-, Master- or PhD-Thesis

PhD-Thesis

English: Declaration of Authorship I hereby declare that the thesis submitted was created and written solely by myself without any external support. Any sources, direct or indirect, are marked as such. I am aware of the fact that the contents of the thesis in digital form may be revised with regard to usage of unauthorized aid as well as whether the whole or parts of it may be identified as plagiarism. I do agree my work to be entered into a database for it to be compared with existing sources, where it will remain in order to enable further comparisons with future theses. This does not grant any rights of reproduction and usage, however. This document was neither presented to any other examination board nor has it been published. German: Erklärung der Autorenschaft (Urheberschaft) Ich erkläre hiermit, dass die vorliegende Arbeit ohne fremde Hilfe ausschließlich von mir erstellt und geschrieben worden ist. Jedwede verwendeten Quellen, direkter oder indirekter Art, sind als solche kenntlich gemacht worden. Mir ist die Tatsache bewusst, dass der Inhalt der Thesis in digitaler Form geprüft werden kann im Hinblick darauf, ob es sich ganz oder in Teilen um ein Plagiat handelt. Ich bin damit einverstanden, dass meine Arbeit in einer Datenbank eingegeben werden kann, um mit bereits bestehenden Quellen verglichen zu werden und dort auch verbleibt, um mit zukünftigen Arbeiten verglichen werden zu können. Dies berechtigt jedoch nicht zur Verwendung oder Vervielfältigung. Diese Arbeit wurde noch keiner anderen Prüfungsbehörde vorgelegt noch wurde sie bisher veröffentlicht.

………………………………………………………………………………………………………. Date, Signature

Abstract Many different future applications like bendable and transparent displays or functional clothing require

relatively low process temperatures (e.g. ~ 249 °C as upper temperature limit for polyimide) for

deposition of the active material, e.g. solution-processed semiconductors. Metal oxides such as zinc

oxide have good electrical properties but processing those from solution requires relatively high

temperatures (e.g. spray pyrolysis of zinc acetate at ~ 360 °C). This thesis pursues three fundamentally

different approaches to lower the process temperature of zinc oxide produced by spray pyrolysis.

Tailored organic molecules are used as a post-deposition treatment to passivate surface traps, i.e.

hydroxy groups or chemisorbed water, as the first approach. A successful passivation of surface defects

improve the electrical properties of the zinc oxide that occur at low process temperatures. Therefore

tailored passivation molecules, i.e. 1,3-diketones, are presented. Their binding towards zinc is studied

and their passivating properties of solution processed zinc oxide thin-film transistors analyzed.

Fluorinated zinc carboxylate derivates are analyzed as novel potential zinc oxide precursors with focus

on lower deposition temperatures as a second approach. FTIR (Fourier transform infrared spectroscopy)

and TGA (Thermogravimetric analysis) reveal whether a precursor thermally decomposes to zinc oxide

and identifies the decomposition temperature.

The third approach: High-speed picoliter droplet analysis gives deeper understanding of droplet

interactions with the substrate depending on the temperature under real deposition conditions. A novel

model for top-view analysis of dynamic and static advancing contact angles and a comprehensive

determination of thermodynamic properties like Leidenfrost point, critical heat flux and thermodynamic

boiling regimes is presened. Additionally, a novel hovering state of very small droplets above the

substrate at room temperature is presented. This state is similar to the Leidenfrost point and enables

the deposition of smooth layers at low temperatures (T < 100 °C). Overall, this analysis allows a fast

screening for suitable solvent and substrate combinations for the deposition of precursors that are not

processable with standard solvents to find beneficial deposition conditions at low temperatures.

Table of Contents

Acronyms and chemical formulae ......................................................................... ix

1 Introduction ......................................................................................................... 1

2 Spray pyrolysis ................................................................................................... 5

2.1 Airblast atomization ................................................................................................................. 6

2.2 Ultrasonic atomization ............................................................................................................. 7

2.3 Atomization of liquids and behavior of falling droplets ............................................................ 8

2.4 Wetting, contact angles, and evaporation of droplets at hot surfaces ................................... 11

2.5 Thermodynamic boiling regimes ............................................................................................ 14

2.6 Deposition of precursor ......................................................................................................... 15

3 Metal oxide thin-film transistors ..................................................................... 19

3.1 History of and basics of field-effect transistors ...................................................................... 19

3.2 Charge transport .................................................................................................................... 22

4 Experimental and theoretical methods ........................................................... 24

4.1 Density functional theory ....................................................................................................... 24

4.2 Atomic force microscopy ....................................................................................................... 27

4.3 X-ray photoelectron spectroscopy ......................................................................................... 28

4.4 Nuclear magnetic resonance spectroscopy .......................................................................... 29

4.5 Thermogravimetric analysis ................................................................................................... 32

4.6 Fourier-transform infrared spectroscopy ............................................................................... 32

4.7 Scanning electron microscopy ............................................................................................... 35

5 Tailored organic surface passivation for metal oxide semiconductors ...... 37

5.1 Concept of surface traps passivation by tailored organic molecules .................................... 37

5.2 1,3-diketone molecules: theoretical characterization ............................................................ 39

5.3 Chemical binding of passivation molecules to zinc oxide surfaces ....................................... 42

5.4 Passivation process monitoring ............................................................................................. 49

5.5 Performance of passivated thin film transistors ..................................................................... 52

5.6 Fabrication and characterization of thin-film transistors ........................................................ 55

5.7 Conclusion and outlook ......................................................................................................... 57

6 Fluorinated Carboxylates as zinc oxide precursor ........................................ 59

7 High-speed real-time analysis of picoliter droplets under spray pyrolysis

conditions ............................................................................................................... 63

7.1 Cross-section analysis of droplet diameter after atomization ................................................ 63

7.2 Raytracing and droplet appearance ...................................................................................... 66

7.3 Sessile and moving droplets: Indicator for Leidenfrost temperature ..................................... 68

7.1 Formation of sessile droplets and their interaction with the environment ............................. 71

7.2 Evaporation model of sessile droplets: Ad hoc extraction of the contact angle .................... 73

7.3 Thermodynamic boiling regimes, critical heat flux and Leidenfrost temperature .................. 80

7.4 Novel meta-stable hovering state of very small droplets – Bypass to the Leidenfrost effect 81

7.5 Experimental setup and video acquisition ............................................................................. 85

7.6 Conclusion and outlook ......................................................................................................... 87

8 Concluding remarks ......................................................................................... 90

List of publications ................................................................................................. 93

8.1 Journal paper ......................................................................................................................... 93

8.2 Patent applications ................................................................................................................ 93

8.3 Conference contributions ....................................................................................................... 94

References .............................................................................................................. 96

Acknowledgments ................................................................................................ 105

ix

Acronyms and chemical formulae

m : Micrometer XPS : X-ray photoelectron spectroscopy

AFM : Atomic force microscopy ZnO : Zinc oxide

BE : Binding energy

CHF : Critical heat flux

DFT : Density functional theory

EDG : electron-donating group

Eh : Hartree

EPD : Electronic paper display

EWG : electron-withdrawing group

FT-IR : Fourier-Transform Infrared

spectroscopy

FWHM : Full width half maximum

GTO : Gaussian type orbital

H2O : water

HOMO : highest occupied orbital

IC : Integrated circuit

KE : Kinetic energy

LCD : Liquid crystal display

LED : Light-emitting diode

LFP : Leidenfrost point

LTPS : Low-temperature polycrystalline

silicon

LUMO : lowest unoccupied orbital

NMR : Nuclear magnetic resonance

OLED : Organic light-emitting diode

pL : picoliter

SiSiO2 : Silicon/Silicon dioxide

STO : Slater type orbital

TGA : Thermogravimetric analysis

TIR : Total internal reflection

UV/O : Ultraviolet ozone

UV-VIS : Ultraviolet-visible

VLSI : very largescale integration

VMD : Volume mean diameter

Introduction

1

1 Introduction

Robert Noyce applied for the first patent on an integrated circuit (IC) fabricated on a single-

crystal substrate in 1959.[1] He used for his first chip novel photolithography and diffusion

processes and enabled new applications. In the early 1960s, the first microprocessors went

into mass production, and a technical revolution had begun. Since then, the integration of

passive elements like resistors and advanced manufacturing technologies have made

transistors smaller, and hence the number of transistors on a chip has increased. Today, the

transistor density is one of the primary indicators for the performance of ICs. Nowadays, the

first manufacturers produce chips with more than 11.8 billion transistors per chip in a 5 nm

process.

However, many application areas, such as displays or sensors, which utilize ICs and

semiconductors per se, do not require remarkably shrunk dimensions and high transistor

densities. State of the art displays of high-end smartphones, for example, consists of pixels that

are about 50 m in size (Figure 1.1). Such small pixels cannot be recognized by the human eye

anymore and imitate a continuous realistic picture to the observer. Much larger displays with a

comparable pixel resolution like TVs or monitors have a lower pixel density, and the individual

pixel size is up to a magnitude larger.

Figure 1.1 Microscopy image of pixel arrays. Each pixel consists of three colored subpixels (red, green,

and blue). The pixels' size and alignment vary with technology (a: OLED: Organic light emitting diode,

b: LCD: Liquid crystal display).

Thin-film transistors that switch the pixels in displays thus have much more space compared to

transistors in processors. In most cases, they require different properties like high transparency

and mechanical durability. Therefore, thin-film transistor materials that can replace the standard

material silicon and their deposition technologies are of particular interest in these fields instead

of further shrinking their size.

Introduction

2

Extensive research has revealed that some metal-oxides meet the optoelectronic requirements

to serve as silicon substitute in thin-film transistors. Nowadays, research focuses on

significantly better understanding and optimizing existing deposition technologies for those

metal oxides as the active material or as electrodes, while they already substitute silicon. There

is great potential to save resources by switching from vacuum-based technologies like

CVD/PVD (chemical/physical vapor deposition) or sputtering to liquid deposition processes like

spray pyrolysis, spin coating, and inkjet printing. Another main focus is the reduction of the

usually high process temperatures of metal oxides.

One of the main application areas for metal oxide thin-film transistors is the display market. It

includes liquid crystal displays (LCD), organic light-emitting diodes (OLED), and electronic

paper displays (EPD), also known as e-ink displays. Each of these display technologies has its

strengths, and all are part of today's products. For example, LCDs can provide contrast-rich

images with a broad color spectrum even with intense ambient light combining dynamic

backlighting with anti-reflective surfaces. OLEDs can be switched off entirely and therefore

achieve an almost perfect black. Since OLEDs do not require backlighting, they make the

construction of transparent displays possible. Compared to classic LCD displays, they consume

significantly less energy. EPDs, on the other hand, can display content without energy

consumption for an unlimited time. Energy is only required to update the displayed content.

Compared to LCDs and OLEDs, EPDs have a prolonged refresh rate and are suitable for eBook

readers or digital price displays in retail stores, for example.

Since 2005 displays based on field-effect transistors dominate the market. They belong to the

class of thin-film transistors. Since then, they have undergone rapid development, and the

demands for improved modern displays technologies increase almost every year. Technical

engineers mastered the hurdles for properties such as a 90-inch screen diagonal or extreme

dynamic range (contrast 1.000.000: 1), with fast response time (< 1 ms), high viewing angle

(178°), a resolution of 8K (7680 x 4320 pixels), or 3D technology already by today.

In novel displays, millions of thin-film transistors switch the individual pixels and control their

brightness. Initially, low-cost a-Si:H (hydrogenated amorphous silicon) was the primary active

semiconductor material. It achieves with its amorphous structure a charge carrier mobility of

about = 1 cm2/Vs. With such low carrier mobility, no transparency, and mechanical rigidity, it

is not suitable for high-end display applications.[2–4]

One way to overcome parts of the issues is further crystallization and conversion into LTPS

(low-temperature polycrystalline silicon) after the deposition of amorphous silicon. This method

increases the charge carrier mobility of silicon to = 50 cm2/Vs and higher. Different sized

crystals and an uneven distribution of grain boundaries lead to fluctuations in performance.

Introduction

3

Therefore, LTPS works only for relatively small displays. Also, only a few substrate classes

resist the crystallization process temperatures above 650 °C.[3,5]

With good electrical conductivity and optical transparency in the electromagnetic spectrum's

visible region, metal oxides represent the groups of TSOs (transparent semiconducting oxides)

and TCOs (transparent conducting oxides). They combine the necessary properties for use in

optoelectronic devices like displays and are suitable materials for transparent electrodes and

transparent active semiconducting layers. With their properties, they find use in high-end

display applications. The charge carrier mobility of metal oxides at lower process temperatures

is comparable to LTPS.

Recent highlights enabled by new materials are curved and, to a certain extent, flexible displays

in 2015, rollable OLED displays in 2019, and in 2020, subways with transparent OLED displays

instead of regular windows. The latter transparent displays allow the familiar outside view and

provide live-information about the next stops and the railway network additionally at the same

time.

However, many different future applications are conceivable. Automotive concepts include

interactive transparent windscreens with live traffic and navigation information. Ideas of

adaptive billboards that adapt to the individual viewer, functional clothing, or displays printed

on the skin that, for example, provides vital data of a patient or surgery information and

instructions to the treating doctor, are considered realistic. Finally, the further development of

devices that are already in use today, such as monitors, cell phones, and e-book readers,

continues and leads to further advances and new technologies/categories. Such exciting

applications rely on novel materials like metal-oxide semiconductors.

This thesis aims to reduce process temperatures for processing metal oxide semiconductors

active layers used in thin-film transistors by improving transistor performances, finding novel

precursors, and a deeper understanding of the deposition process itself. Therefore the following

three approaches are pursued.

1. Improved stability and performance of solution-processed zinc oxide thin-film transistors

processed at low temperatures by organic surface defect passivation:

• Theoretical description of organic passivating molecules

• Electrical characterization of passivated zinc oxide thin-film transistors

• Monitoring of passivation process and identifying the binding situation at the surface

2. Experimental evaluation of possible novel zinc oxide precursors

• Experimental evaluation of decomposition temperatures for suitable zinc oxide

precursor candidates

Introduction

4

3. Gain a deeper understanding of picoliter droplets at hot surfaces under real application

conditions

• Investigate the Leidenfrost point and contact angles of water under deposition

conditions

• Understanding of picoliter size droplet behavior at hot surfaces

• Identifying beneficial deposition conditions

The following chapters introduce the core topics: spray pyrolysis, metal oxides, and thin-film

transistors. An explanation of the basics of the used theoretical and experimental methods

follows. The final chapters present and discuss the theoretical and experimental results on the

previously listed approaches. Concluding remarks, as a final chapter, summarize the work.

Spray pyrolysis

5

2 Spray pyrolysis

Spray pyrolysis is, in general, a processing method for the production of (thin) coatings and

powders from precursor solutions. The process itself usually divides into three phases. First:

Droplet generation, second, droplet transport, and third, material deposition and conversion.

The sprayed precursor solution undergoes within these phases one or more of the following

processes: atomization and recombination of droplets, evaporation of the solvent, distribution

of solutes, thermolysis, or decomposition of the precursor, annealing of the precursor, and

forming particles.

Four standard methods exist for the droplet generation from a solution, which is also known as

atomization: Airblast atomization, ultrasonic atomization, pressurized nozzles, and electrostatic

atomization. Depending on the method, a typical spray pyrolysis setup consists of an atomizing

nozzle generating droplets or a nebulizer with piezoelectric crystal generating aerosols. After

generation, droplets fall accelerated by an electric field, gravity, or carried by pressurized gas

towards a target. Depending on droplet size and density, they can create a laminar or turbulent

flow within the spray stream. Common precursor solutions contain pure or mixtures of dissolved

metallic salts, organic molecules, or dispersed colloidal particles.

After reaching the target, droplets suddenly heat up to elevated temperatures, and the solvents

evaporate when they enter an oven or reach a heated surface. The solutes precipitate or

thermally convert to produce either powders or single/multi-component layers. Droplets from

atomizing nozzles usually pyrolyze at heated substrate surfaces producing layers. Aerosols

from nebulizers usually flow through tube furnaces and pyrolyze producing powders before

reaching their target destination. In general, the spray pyrolysis process can produce

morphologies including spheres, 1D structures like nanorods or porous, ceramic, or dense

layers. Produced layers for electrical applications mainly act as electrodes or active material in

thin-film transistors, sensors, and solar-cells.

Chapters 2.1 and 2.2 describe the nozzles' design and the atomization methods for this work:

air blast and ultrasonic atomization. Chapter 2.3 introduces some of the essential characteristic

parameters of fluid mechanics and fields that use fine droplets. Chapter 2.4 describes the

wetting and dewetting of arrived droplets at surfaces, the contact angle, and evaporation.

Chapter 2.5 introduces the thermodynamic boiling regimes and their impact on spray pyrolysis.

Finally, Chapter 2.6 explains how precursors are deposited on surfaces by spray pyrolysis.

Spray pyrolysis

6

2.1 Airblast atomization

One possibility to atomize a fluid into small droplets is the airblast atomization. Compressed air

or other pure (mostly inert) gases like N2, Ar, or O2 atomize a dosed liquid by high pressure and

high flow rate. The atomizing gas flow also has the second use to carry the droplets towards a

target. It usually generates droplets with a wide droplet diameter distribution and accelerates

them to a high velocity. Typical values are mean droplet diameters of 80 m and velocities of

more than 10 meters per second.[6] The distribution of the droplets becomes broader with higher

velocity. Various industry applications and artists use this popular and cost-efficient method to

deposit functional layers, coatings, and paint. Deposition of functional coatings/layers or spray

cooling are examples of scientific applications. Even the simplest version of the airbrush guns,

a perfume dispenser, can deposit fully functional zinc oxide layers.[7] Figure 2.1 shows a typical

airbrush gun and schematically its construction as a cross-sectional sketch. The main parts are

the needle, a double-action trigger, an air valve, and the liquid source.

Figure 2.1 a) Sketch of an airbrush construction as a cross-sectional sketch b) Commercially available

IWATA HP-B Plus airbrush used in this work, connected through a particle filter to pressurized nitrogen.

The liquid source feeds the gun with a precursor solution either by gravity or through a siphon.

The gravity feed can atomize even the tiniest volumes (V < 25 L) at comparatively low

pressures, resulting in finer droplets. In comparison, a siphon liquid feed can provide larger

volumes but requires higher pressures because the solution is pulled through a stem and thus

atomized. The gravity feed described first is more suitable for this work due to the small volumes

and finer droplets. For both cases, atomization occurs because the airflow rips the solution

apart.

Spray pyrolysis

7

The droplets can mix after atomization with the carrier gas either internally or externally. Internal

mixing means that the fluid is atomized at the airflow center, providing a more spherical pattern.

External mixing means that the gas comes sidewise and provides a more elliptical pattern.

Finally, a double-action trigger controls the whole spraying process. It can turn the airflow and

activate the liquid feed. In this work, a commercially available IWATA HP-B Plus airbrush with

a nozzle diameter of 200 m deposits the thin-film zinc oxide for thin-film transistors.

2.2 Ultrasonic atomization

Ultrasonic atomization is a method to atomize fluids into fine droplets with an ultrasonic atomizer

nozzle (described here) or an ultrasonic nebulizer. Therefore, the fluid enters through a narrow

capillary the atomization platform. Piezoelectric transducers induce oscillations at the

atomization platform at very high frequencies (30 kHz < f < 130 kHz). Figure 2.2 shows a typical

ultrasonic atomizer with spray shaper and schematically its construction as a cross-sectional

sketch. The main parts are the liquid source with a flow restrictor, piezoelectric transducers,

and the atomization platform. The atomization process at the atomization platform is also

illustrated.

Figure 2.2 a) Sketch of an ultrasonic atomizer cross-section b) atomization process c) Commercially

available ultrasonic atomizer Sonaer NS130K (left) with self-made spray shaper (right).

The liquid solution is wetting the atomization platform with a constant flow rate forming a film.

Piezoelectric transducers induce the oscillation of the film until it gets atomized. Generated

droplets then fall, accelerated by gravity until they reach equilibrium velocity due to friction with

Spray pyrolysis

8

the surrounding atmosphere. A spray shaper located around the atomizing platform can inject

air or inert gases to accelerate the droplets. Droplets accelerated this way form into a spray

that is shaped circular or elongated, depending on the spray shaper geometry. Ultrasonic

atomizers yield finer droplets at higher frequencies. Mean droplet diameters of approximately

20 – 25 m are achievable with a comparable high frequency (f = 130 kHz). Increasing power

at the atomizing platform sharpens the droplet diameter distribution. Ultrasonic atomizers are

limited to Newtonian fluids with low viscosities. Exemplary scientific and industrial applications

are thin-film coating, micro etching, microbead production, and spray drying. The atomizing

nozzle used in this work is a Sonaer NS130K.

2.3 Atomization of liquids and behavior of falling droplets

The understanding and observation of small droplets and their impingement on heated solid-

state surfaces is an already well-established research field. However, there are still many open

questions. It is also a very crucial topic for many areas in the industry. Exemplary applications

and conducted research studies are:

• Spray pyrolysis[7–10]: E.g., Electrical stability enhancement of produced thin zinc oxide

layers and analysis of layer morphologies produced with different droplet sizes.

• Direct fuel injection[11–14]: E.g., Analysis of droplet size, velocity, and volume flux for

predominant port fuel injection technology used in combustion engines.

• Desalination and refrigeration[15–18]: E.g., Effects of spray density, overall heat transfer

coefficient, inlet stream velocity for use in horizontal-tube falling film evaporators

• Spray cooling[19–21]: Studies of droplet diameter and velocity impact heat flux in

thermodynamic regimes: film, transitional, and nucleate boiling.

In those applications, wall temperatures can range from 60 °C to 500 °C, e.g., for aqueous

droplets. The broad temperature range, different boiling regimes, and many process

parameters underline that these applications' processes are highly complex and depend on

numerous parameters. The main parameters are droplet diameter and impact velocity[22] and

material properties like density, viscosity, and surface tension of the selected liquid[23] and wall

properties like wettability,[24] surface roughness,[25] and surface temperature.[26] Those complex

processes in real application conditions were already recorded with a high-speed camera

approach[20], but detailed droplet behavior is not analyzed so far on such datasets.

Furthermore, environmental parameters like pressure and turbulences have to be considered.

Top-view observation of droplets can often be easily integrated in existing set-ups. Droplet

Spray pyrolysis

9

observation close to hot surfaces in real-time under real process conditions gives valuable

insights into the crucial droplet dynamics.

Previous research already allows a good understanding of the formation and the expected

behavior of droplets in motion with varying sizes for intensively used regimes, where others still

lack fundamental research.

A few basic descriptors help with an initial assessment for a better comparison between

different technologies and setups. Therefore, it is essential to know that in principle, six forces

affect falling droplets through a gas: Gravity, surface tension, viscosity and inertia of the liquid,

and inertia and viscosity of the gas. The often applied descriptors and dimensionless

parameters Weber number, Reynolds number, and Ohnesorge number set theses forces into

meaningful ratios that support the mathematical description of droplets.

The Weber number, named after Moritz Weber, describes the ratio of inertia and surface

tension (drag force and cohesion force) of a fluid. That ratio can describe, for example, the

deformation behavior of a flowing fluid when it interfaces with another fluid. A simple example

is a falling raindrop in air. Its inertia promotes deformation, while the surface tension stabilizes

its shape. The Weber number is

𝑊𝑒 = Inertia

Surface tension=

𝜌𝑣2𝐿

𝜎

with is the density [kg/m3] of the fluid, v [m/s] its velocity, L its characteristic length [m] and σ

its surface tension [N/m]. The Weber number qualitatively characterizes the droplet properties

after atomization, for example, by an ultrasonic atomizer. In general, as the Weber number

increases, a liquid breaks down more strongly into smaller droplets. Also, with larger Weber

number (We ≥ 1)[27], the droplets are subject to greater deformation and move further away from

the ideal spherical shape (We << 1).[27]

The Reynolds number predicts whether the flow pattern of fluids and gases tends to be

preferably laminar or turbulent. The concept itself was introduced by George Stokes in 1851

and named after Osborne Reynolds. The Reynolds number is

𝑅𝑒 = 𝜌𝐿𝑣

𝜇

with is the density [kg/m3], L its characteristic length [m], v the velocity [m/s] and μ the dynamic

viscosity [Pa ∙ s] of the liquid. The flow of a liquid or gas at Reynold numbers below the critical

Reynolds number (Re < Recrit, Recrit = 2340 for pipes) is usually laminar, and the droplets move

within a sheet-like flow. With increasing Reynold number (Re > Recrit), the flow becomes

turbulent, where the flow and thus the droplets strongly vary in speed and movement direction.

It is necessary to set the Reynolds number for the experiment below the turbulence onset

Spray pyrolysis

10

(Re = Recrit) in spray pyrolysis. In this regime, the droplets need no direction to the target

position and are mainly falling because of gravity. The turbulent flow does not change to a

laminar flow does not occur exactly at Recrit it is more like a transition point. Turbulences can

occur below Recrit but decay fast. Otherwise, at higher Reynold numbers (Re > Recrit) or in a

turbulent environment, a required carrier gas dominates the droplets' behavior and directs them

to the target.

Wolfgang von Ohnesorge defined in 1936 the Ohnesorge number (Oh) that combines Weber

number and Reynolds number to eliminate the velocity of the liquid. It relates viscous forces to

inertial and surface tension forces.

𝑂ℎ = viscous forces

√inertia x surface tension=

√𝑊𝑒

𝑅𝑒=

𝜇

√𝜌𝐿𝜎

Where μ is the dynamic viscosity [Pa ∙ s] of the liquid, ρ is the density [kg/m3] of the liquid, L its

characteristic length [m], and σ is the surface tension [N/m]. In fluid dynamics, the interpretation

of the Ohnesorge number, together with the Reynolds number, describes the formation of

droplets at nozzles before applying external power in three regimes: droplets (dripping), waves,

and droplets, and atomization, which Figure 2.3 visualizes.

Figure 2.3 Ohnesorge number vs. Reynolds number for an aqueous system. The diagram delimits the

three regimes under which a water stream without external energy breaks up into droplets, into waves

and droplets and atomizes.[28]

The droplets analysis in this thesis (Chapter 7) requires droplets with a mean diameter around

D0 = 25 m and a falling velocity of u0 < 50 mm/s, corresponding to We = 10-4 and

Oh = 2 x 10-2. The thin-film transistor production (Chapter 5) requires to atomize small volumes

(V ≤ 50 L).

Spray pyrolysis

11

Generating water droplets in air (ρwater = 997.0 kg/m^3, water = 0.890 × 103 Ns/m2, air Ns/m2,

1.225 kg/m³, air 18.5 Ns/m2, at 25 °C) with a conventional high-pressure nozzle with an internal

diameter of 0.8 mm (Oh = 4 x 10-3) requires to press the solution through the nozzle with

pressures of up to 100 bar to operate in the atomization regime, according to Figure 2.3. This

high pressure delivers droplets with a broad diameter distribution and larger diameters than

required. Also, the high pressures accelerate the droplets to velocities that are 100 m/s and

faster. Also, for small volumes, it is technically hard to build up such high pressures. Because

of that reason, additional energy from the ultrasonic atomizer or pressurized gas in the airblast

atomization is required to provide the right size of droplets for analysis and small atomized

volumes for deposition.

2.4 Wetting, contact angles, and evaporation of droplets at hot surfaces

The wetting of surfaces by droplets and their evaporation (at heated surfaces) has been subject

to previous theoretical and experimental research for decades.[29–31] Figure 2.4 shows the most

common scenarios droplets undergo from the first contact to the surface until evaporation.

Depending on the solvents' polarity and surface properties, the droplets either wet the surface

or are repelled from the surface after impact (Figure 2.4 a and b). For example, a water droplet

repels from a hydrophobic surface and wets a hydrophilic surface due to matching or

mismatching surface energies. After wetting the surface, the droplets settle with a defined

contact angle[32,33].

This is shown in (Figure 2.4 a) or it undergoes secondary atomization[34], where they break

down into secondary and much smaller droplets (Figure 2.4 c). After settling, droplets begin to

evaporate, which takes place for larger droplets (V > pL), where surface tension is not the

dominating force, in a bubble and pattern boiling[35] with visible gas evolution within the droplet

(Figure 2.4 d). The evaporation mechanism of droplets depends on their contact angle. In

general, a systems' contact angle Θ takes a value between the larger advancing Θa and the

smaller receding Θr contact angle.

Droplets with a contact angle larger than their receding contact angle have a pinned edge[36]

(Figure 2.4 e). A droplet with a pinned edge evaporates with a constant contact area towards

the substrate until the contact angle is equal to the receding contact angle.

Otherwise, the system evaporates with a constant contact angle and a shrinking contact area

towards the substrate. The constant contact angle is equal to the receding contact angle (Figure

2.4 f). The droplets' contact angle can decrease while evaporation and the evaporation

Spray pyrolysis

12

mechanism can switch from a decreasing contact angle with a constant area to a constant

contact angle with a shrinking area until the whole droplet volume is evaporated.

Figure 2.4 Possible scenarios of droplets after colliding with the surface: a) wetting of the surface, b)

repelling at the surface, c) secondary atomization, d) boiling with or without bubbles, e) pinned

evaporation (with decreasing contact angle), and f) evaporation with constant contact angle.

Standard methods for contact angle determination (Θa: advancing, Θr: receding) are a droplet

on a tilted substrate[37], a dynamic sessile droplet into which volume is pumped in/out, or the

Wilhelmy method, which studies the wettability by a dipping a plate[38] (Figure 2.5).

Figure 2.5 Standard methods for contact angle measurement: a) droplet on a tilted substrate, b)

dynamically pumped droplet on a surface, and c) Wilhelmy method with a dipped plate.

A sessile droplet on a leveled substrate has, in the ideal case, the same contact angle in every

direction (pointing from the center). If the substrate is now slightly tilted, gravity deforms the

droplet towards the substrate's lowest position. The deformation causes a larger contact angle

at the lowest position and a smaller contact angle at the highest position than the initial contact

angle. These changes increase with further tilting of the substrate until the droplet starts moving.

Spray pyrolysis

13

The larger contact angle corresponds to the advancing and the lower to the receding contact

angle by reaching this point.

For the second method, the droplets' volume on a substrate increases through a capillary. The

contact angle increases meanwhile until it reaches the advancing contact angle. The contact

angles are equal when the contact area starts to increase, too. Afterward, a pump reduces the

volume through the capillary until the contact area starts to shrink. The contact angle is now

equal to the receding contact angle.

For the third method, a substrate dips into the liquid. The contact angle increases until the liquid

wets more surface of the substrate. The contact angle corresponds to the advancing contact

angle. Afterward, the substrate retracts out of the solution until the dewetting starts. The contact

angle then corresponds to the receding contact angle.

The previous descriped advancing and receding contact angles are denoted as static contact

angles where the contact line is not in motion. For the case that the contact line is in motion the

advancing and receding contact angles are denoted as dynamic. Therefore the contact angle

depends on the velocity of the contact line as shown in Figure 2.6. The increase/decrease of

the advancing and receded contact angle is limited at high velocities of the contact line.

Figure 2.6 Dependence of static and dynamic advancing and receding contact angles Θ on the velocity u

of the contact line.[39]

Also, more complex methods exist as cryo-AFM (atomic force microscopy)[40–45], ESEM

(environmental scanning microscopy)[46–48], and cryo-SEM (scanning electron

microscopy).[49–51] These methods analyze solidified static momentary states of droplets in

frozen environments in vacuum and ambient atmosphere. Top-view analysis of large droplets'

light interference fringes allows the extraction of contact angles for huge droplets.[52]

Spray pyrolysis

14

However, very few reports investigate pico-liter droplets in general[53–55] even though an

increasing number of fields use pico-liter droplets due to their extraordinary properties and their

generation methods' availability.[56–58]

In Chapter 7, a new method for a simple and statistically meaningful determination of static and

dynamic advancing and receding contact angles Θ is introduced.

2.5 Thermodynamic boiling regimes

The previous description of droplets reaching the surface states that repelling occurs in the

simplest case because the phobicity of liquid and surface doesn't match, inhibiting the wetting

itself.[59] A more complex scenario is the Leidenfrost behavior at higher temperatures. To

explain Leidenfrost behavior, three different thermodynamic boiling regimes of liquids at hot

surfaces are introduced (see Figure 2.7). An often made but wrong assumption is that with

increasing temperature, the heat flux increases proportional, and thus, a drop evaporates

faster. After an expected increase in heat transfer, it reaches a maximum, also known as the

critical heat flux (CHF). The thermodynamic boiling regime until the CHF is the nucleate boiling.

Droplets' lifetime decreases with increasing temperature.

At higher temperatures than the CHF temperature, the heat flux decreases. The heat flux

reaches its minimum at the Leidenfrost point. The thermodynamic regime between those two

points is the transition boiling. Droplets' lifetimes increase with increasing temperature.

At this point, with increasing temperature, droplets evaporate in the film boiling regime. The

previously mentioned scenario of the Leidenfrost behavior starts. A droplet stops wetting the

surface but also stops to repel from it. It starts to levitated close above the surface. This is

possible due to the droplet's evaporation towards the surface, creating an own sacrificial vapor

phase below. The additional gas layer between the surface and droplet is isolating the heat

flux. The droplets experience a drastically increased lifetime, because of the general bad heat

conduction through gaseous phases.

Spray pyrolysis

15

Figure 2.7 Droplet lifetime / Evaporation time and heat flux/evaporation rate vs. superheat T of the heated

surface. Critical heat flux (CHF, maximum evaporation rate, minimum lifetime), and the Leidenfrost point

(LFP, minimum evaporation rate, maximum lifetime) delimit the thermodynamic boiling regimes (nucleate

boiling, transition boiling, and film boiling).

2.6 Deposition of precursor

From the heat transfer perspective, the wall temperature plays a significant role and determines

the lifetime of droplets on the surface and whether the droplets are sessile or levitating.

Therefore, the temperature has to be determined and controlled as precisely as possible.

The deposition of precursor layers (or their thermally converted form) from picoliter droplets for

different applications is possible in all mentioned temperature regimes. It is of high importance

to distinguish whether the deposition occurs at temperatures below the LFP in the nucleate

boiling and transition boiling regime where the droplets wet the surface. Or if the surface

temperature is higher than the LFP and deposition occurs in the film boiling regime with droplets

levitating above the surface (Figure 2.8).

Spray pyrolysis

16

Figure 2.8 Layer deposition at temperatures below LFP in a) nucleate boiling and b) transition boiling

regimes. Deposition at temperatures above LFP in the film boiling regime c).

Sessile droplets slowly evaporate at relatively low temperatures (T < LFP) in the nucleate

boiling regime, and the solutes start to precipitate at the border. In this case, a droplet leaves

behind the very well known coffee stain ring. Residuals of multiple droplets don't interfere unless

their evaporating position overlaps.

Sessile droplets experience rapid evaporation at elevated temperatures close to the CHF and

above in the nucleate boiling regime. Coffee stains are not observable anymore. Instead,

droplets burst, and their solutes spread around the landing position. The residuals of multiple

droplets can now interfere even if the initial droplets are separated.

Functional layers deposited in these temperature regimes from sessile droplets are usually very

rough because the solutes are distributed unevenly. Because of the low surface temperatures,

especially in the nucleate boiling regime, deposition on various substrates such as flexible and

transparent polymers or even fabrics is possible. A drawback might be that many precursor

systems might not thermally convert at these low process temperatures. These films (layers)

perform fundamentally different from smooth films because of the high surface roughness,

resulting in grain boundaries, and not or only partly converted precursors.[60]

Deposition at temperatures above the LFP undergoes different mechanisms since the droplets

are now levitating above the surface and move laterally. Three known mechanisms are the

One-droplet-one-particle (ODOP), One-droplet-many-particle (ODMP), and a chemical vapor

deposition like mechanism (CVD) (Figure 2.9). In the case of ODOP and ODMP, a droplet

Spray pyrolysis

17

shrinks due to evaporation. If the precursor concentration exceeds the solubility, one or more

particles precipitate, respectively. In the latter case of a CVD-like mechanism, the levitating

droplet distributes single molecules, which reach the surface by Brownian motion.

Figure 2.9 Deposition mechanisms at temperatures above the Leidenfrost point LFP in the film boiling

regime a) One-droplet-one-particle ODOP, b) One-droplet-multiple-particles, c) chemical vapor deposition

CVD.

All mechanisms usually distribute the precursor (or its thermally converted form) evenly over

the surface and deliver smooth films. Deposition at these high temperatures often already

provides the necessary energy for a precursor's thermal conversion. Otherwise, post-annealing

processes convert precursors that require higher energies for conversion.

After introduction of mechanisms by which material can be applied to surfaces by spray

pyrolysis, suitable solution-processable precursors for metal-oxides are briefly introduced.

Popular inorganic precursors are halides and nitrate salts or nanoparticles of the target

compound. Popular organic precursor compounds are alkoxides, carboxylates, and

acetylacetonate derivatives. These precursors can be dissolved and processed from aqueous

solutions and dissolved in most polar organic solvents like alcohols. Some combinations need

adding of the corresponding acid to stabilize the solution, increase the solubility, and avoid

precipitation of insoluble reaction products before deposition. There is not an optimal precursor

for every application. A decision for a precursor depends much more on which morphology is

to be obtained, which deposition process and which reaction mechanism are used or whether

the final layer needs doping. If one precursor candidate is to be highlighted in the present

context of zinc oxide production, it is zinc acetate, because it can be deposited with different

Spray pyrolysis

18

methods and is widely used throughout different applications. Its advantageous deposition

mechanism of zinc acetate is explained in Chapter 6, together with fluorinated derivatives.

Besides the heat transfer in different thermodynamic boiling regimes and the precursor, multiple

droplets' interaction is also relevant. If one droplet meets another droplet before it has

evaporated, its lifetime increases and influences the heat transfer from the surface into the

droplet. Resulting temperature inhomogeneities within a droplet leads to the Marangoni

Effect[61], further influencing the evaporation behavior for depositions at higher droplet densities.

At higher densities, knowledge of the processes between droplets is just as essential as the

heat transfer.

The evaporation regimes and droplet interactions and the properties of a produced layer via a

precursor in the droplets, e.g., morphology or electrical performance, depend highly on

deposition temperature. The deposition of ultra-thin metal oxide semiconducting layers from

solution-based spray pyrolysis require a wall temperature above LFP to achieve CVD-like

(chemical vapor deposition) growth of the oxide layer to achieve acceptable electrical

performance.[10,62] The deposition temperature for this process is also limited by the requirement

to provide sufficient activation energy to convert precursor molecules (e.g. ~ 350 °C as lower

temperature limit for zinc acetate) and prevent thermal decomposition of solutes and the

substrate (e.g. ~ 249 °C as upper temperature limit for polyimide).

To match the right surface temperature for the solvent and precursor can become challenging.

For solutes with strict thermal requirements and selective solubility, it can even become

impossible.

Previous systematic research describes the suitable deposition parameters for metal oxide

semiconductors such as zinc oxide[62] and indium oxide[10]. It indirectly describes the Leidenfrost

point and the optimal deposition temperature by investigating the produced films' morphology,

electrical properties, and precursor residues.

Metal oxide thin-film transistors

19

3 Metal oxide thin-film transistors

3.1 History of and basics of field-effect transistors

The history of field-effect transistors began with their invention by William B. Shockley, John

Bardeen, and Walter H. Brattain in the 1940s in the Bell Lab.[63–65] It is undisputed that modern

devices such as processors or displays based on integrated circuits are only possible through

field-effect transistors' invention. It is consequent that the three inventors are members of the

Hall of Fame for patents.

Field-effect transistors have undergone enormous development since then, but the general

operating principle is the same: A transistor, in general, is a three-electrode device. The gate

contact controls the channel resistance between source and drain contacts. One member of

the field-effect transistor is the metal oxide semiconductor field-effect transistor (MOSFET). The

channel resistance is controlled capacitively by applying an electric field between source and

gate. The gate is separated from the other parts by a gate dielectric forming a capacitor. A very

well-known MOSFET system is silicon with silicon dioxide as gate dielectric (SiSiO2) introduced

by Atalla in 1960.[66]

A field-effect transistor can be built in various geometries. Typically, the position of the source,

drain, and the gate is described with respect to the semiconductor. A bottom-gate, bottom-

contact transistor, for example, implies that the dielectric has been deposited on top of a gate.

The source and drain contacts are located directly on top of the dielectric. The semiconductor

is deposited as the last layer on top. This configuration has advantages if the semiconductor

properties are studied because, after its deposition, no further process steps are necessary that

could influence the properties itself. This configuration is shown in Figure 3.1, together with the

measurement circuit and a microscopy image of source and drain contacts.


20

Figure 3.1 a) Schematic side view of bottom-gate bottom-contact MOSFET b) a simplified version of

measurement circuit for thin-film transistor measurements c) Microscopic top view of the finger-like source

and drain contacts with W/L = 100 on SiO2 (scalebar = 500 m) for a L = 2000 m transistor.

In field-effect transistors, the conducting channel is formed either by e- (n-type channel) or holes

(p-type channel). For n-type channels a positive gate-source voltage pulls the electrons into the

conduction channel. Conversely, for p-type, the channel becomes conductive at negative gate-

source voltages. Conductivity increases in both cases with increasing absolute voltage.

The transistor is normally-on if it is conductive at 0 V gate bias and operates in depletion mode.

In case the channel conductivity is very low at 0 V gate bias, the transistor is described as

normally-off and operates in enhancement mode. The transistor switches its state when the

gate-source voltage exceeds the threshold voltage Vth.

Zinc oxide, which serves in this thesis as a model semiconductor for the group of metal-oxide

semiconductors, is an n-type semiconductor that is normally-off and operates in enhancement

mode.

If there is a voltage applied between Drain and Source, the channel material acts as a resistor.

For small Drain-Source voltages, the transistor operates in the linear regime, and the current

ideally scales proportionally to the drain applied voltage.[67]

𝐼𝐷,𝑙𝑖𝑛 = 𝑊𝐶𝑖

𝐿𝜇 (𝑉𝐺𝑆 − 𝑉𝑡ℎ −

𝑉𝐷𝑆

2) 𝑉𝐷𝑆

Where W and L are the channel width and length [m], Ci is the gate insulator capacitance

[F/cm2], and the charge carrier mobility [cm2/Vs]. From this, the linear mobility of the charge

carriers calculates as

µ𝑙𝑖𝑛 = 𝐿

𝑊 𝑉𝐷𝑆𝐶𝑖

𝑑𝐼𝐷

𝑑𝑉𝐺𝑆

and characterizes how quickly an electron moves for a given electric field. The linear mobility

is proportional to the conductivity. A few electrons with high mobility can result in the same

conductivity as many electrons with low mobility. In a semiconductor, the electron mobility


21

depends on impurity concentrations, defect concentrations, temperature, and charge carrier

concentration. High electron mobility is desirable for semiconductors in transistors.

Increasing Drain-Source voltage leads to a transistor operating in the saturation regime, where

the Drain current ID saturates. This point is called pinch-off and is reached when VDS = VGS – Vth.

Substituting VDS into the equation describing ID,lin describes the current in the saturation ID,sat

regime:

𝐼𝐷,𝑠𝑎𝑡 = 𝑊𝐶𝑖

2𝐿µ(𝑉𝐺𝑆 − 𝑉𝑡ℎ)2

With further increase of the Drain-Source voltage, the Drain current remains essentially

constant and independent of the Drain-Source voltage.

Figure 3.2 Transfer (current between Drain and Source for variable gate bias) (a) and output (current

between Drain and Source for fixed gate bias) (b) characteristic of a zinc oxide thin-film transistor. Onset-

voltage VOnset (ID = 10-7 A) and threshold voltage Vth (at onset of current) are shown in the half-logarithmic

and linear plots of the drain current.

The Vissenberg-Matters model[68] for charge transport provides additional physical information

for highly disordered systems. It assumes that the charge transport occurs in the tail of an

exponential density of states, specifically between localized states. Charge carrier mobility is

then correlated to VGS by a power law:

µ = µ0 (𝑉𝐺𝑆 − 𝑉𝑡ℎ

𝑉𝑎𝑎

)𝛾

where 0 is the charge carrier mobility at VGS – Vth = Vaa and the disorder parameter. is

directly related to the width of DOS where charge transport occurs and, therefore, a measure

for the system's disorder.[69] For an ideal semiconductor is 0.


22

3.2 Charge transport

Charge transport in solids is a broad research topic with already many existing models of charge

transport mechanisms. For this topic, the differentiation between highly ordered and disordered

semiconductors is essential. Extreme examples are defect-free crystalline solids for highly

ordered and disordered molecular or amorphous solids for disordered systems.

Figure 3.3 a) Schematic diagram of the density of energy states of a crystalline compound. Ec denotes as

the conduction band's energy, Ev as the valence band's energy, EF as Fermi energy. b) Band diagram of

the previous DOS. EG is the band gap energy. For n-type band transport occurs in the conduction band

upon the applied electric field.

Band transport occurs in the first case. Charges move through delocalized molecular

wavefunctions (valence and conduction bands) that expand over the entire crystal volume. This

molecular wavefunctions form the valence and conduction band. The difference between the

valence band's highest energetic states and the conduction band's lowest energetic state is the

band gap with the energy Egap. Materials with band gaps smaller than 0 eV (overlapping valence

and conduction band) or larger than 4 eV are metals and isolators. Conduction in materials with

band gaps between those energies is possible when free charges are available. Charges are

available after thermal excitation or through absorption of photons. The introduction of filled

states close to the conduction band (n-type) or empty states close to the valence band (p-type)

by dopants also provides free charges. Prominent examples are silicon doped with either

Phosphorus (n-type) or Boron (p-type).

With an increasing number of defects, more and more localized states exist within the band

gap. Defects can have multiple natures. A distortion of the lattice by extra, missing, or misplaced

atoms is one cause. Grain boundaries, where the lattices of multiple grains don't match or

dangling bonds that are satisfied with problematic groups like hydroxyl (OH), is another. A few


23

traps already influence the transport of charge carriers, and the transport is called multi-trap

and release transport. Many electron-accepting states can cause the exponential tail in the

density of states that the Vissenberg Matters model accounts for, as described in the previous

chapter, and the transport is then hopping transport. In this case, conduction is energy

activated, and charges have to overcome barriers or tunnel between two states. Passivation of

such trap states can be done, for example, by chemical reaction with suitable molecules or

adding layers of passivating materials.

Figure 3.4 a) Schematic diagram of the density of states of a disordered compound. Exponential tail states

(shallow states) are marked in green; deep states within the gap are marked in red. b) Band diagram of

the previous DOS. Upon applying an electric field, Multi-trap and release transport or hopping transport

can occur.

Experimental and theoretical methods

24

4 Experimental and theoretical methods

4.1 Density functional theory

Density functional theory (DFT) is a quantum mechanical ab initio method to investigate the

electronic structure (i.e., the ground state) of many-body systems based on the spatially

dependent electron density. It is widely used in physics, chemistry, and material science to

predict molecules and crystals' properties. The two Hohenberg-Kohn-Theorems[70] are the

theoretical basis of DFT. The first states that a specific electron density allows a connection to

the underlying external potential. The second states that the basic properties such as the

ground state energy of a many-electron system directly depends on the electron density (𝑟).

The energy of a system E[] is calculated as:

E[ρ] = F[ρ] + Ene[ρ]

Where F is the universal functional and E𝑛𝑒 the nucleus-electron interaction. Exchange-

correlation and kinetic energy are calculated by Kohn-Sham functions[71], which generate the

same density for a system of non-interacting electrons as any given system with interacting

electrons.

(−ℏ2

2𝑚∇2 + υeff(𝑟)) 𝜑𝑖(𝑟) = 휀𝑖𝜑𝑖(𝑟)

𝜌(𝑟) = ∑|𝜑𝑖(𝑟)|2

𝑁

𝑖

i is the orbital energy of a Kohn-Sham orbital i, eff is the Kohn-Sham potential and N is the

number of particles in the system. With the given Kohn-Sham theory the energy of a system

E[] is calculated as:

E[ρ] = T[ρ] + Ene[ρ] + Eee[ρ]

with kinetic energy T[], nucleus-electron interaction Ene[] and electron-electron interaction

Eee[]. The latter is the sum

E𝑒𝑒[ρ] = J[ρ] + EXC[ρ]

where J[] is the classical electrostatic repulsion of electrons and EXC is the exchange-

correlation term that accounts for the non-classical effects. Kohn-Sham equation are solved in


25

self consistent manner. This is done by starting with an initial guess, i.e. of a molecule geometry,

and is followed by a Self Consistent Field (SCF) cycle. Within the cycle the Kohn-Sham orbitals

i are calculated for the given guess of and eff .for the geometry. For the calculated orbitals

the and eff are obtained. The self consistent field calculation is carried out until convergence

criteria are met and the energy and electron density for the given geometry is found. The

properties of the system are described by a functional of the electron density. The first

functionals used a local-density approximation (LDA). The functional uses the properties of a

free electron gas with constant electron density given by the local density at the point where it

is evaluated. It was further developed towards a local spin-density approximation (LSDA)

functional with accounting also for the spin. However, more accurate results are obtained by

generalized gradient approximation (GGA) functionals, which are taking the non-homogeneity

of the electron density into account. Another option are hybrid functionals that include a portion

of the exact exchange energy calculated by the Hartree-Fock theory. A widely used hybrid

functional is B3LYP (Becke three-parameter Lee-Yang-Parr),[72–76] which is also used for this

work's calculations. The exchange-correlation part calculates as

𝐸𝑥𝑐𝐵3𝐿𝑌𝑃 = (1 − 𝑎)𝐸𝑥

𝐿𝐷𝐴 + 𝑎𝐸𝑥𝐻𝐹 + 𝑏∆𝐸𝑥

𝐵 + (1 − 𝑑)𝐸𝑐𝐿𝐷𝐴 + 𝑐𝐸𝑐

𝐿𝑌𝑃

where a = 0.1161, b = 0.9262, d = 0.8133, x is the exchange part, c the correlation, B stands

for Becke functional and LYP for Lee-Yang-Parr.

In practical DFT calculations, the wave-function is represented with a set of known functions,

i.e. basis set:

𝜑𝑖 = ∑ 𝑐𝑖𝑗𝜒𝑗

𝑗

A distinction is made between basis sets composed of localized functions (linear combination

approach) for molecules or plane waves, which are used for solid-state calculations. In the

former case, typically Slater-type orbitals (STOs) or Gaussian-type orbitals (GTOs) are used.

Integrals involving Gaussian-type orbitals can be evaluated with less computational effort and

are thus usually preferred. However, GTOs make compared to STOs two qualitative mistakes:

1. GTOs have a no cusp close to the core (r → 0)

2. GTOs decay faster then STOs (r → ∞)

To compensate for this, one uses a linear combination of GTOs (contract atomic orbitals) to

instead of pure GTOs.

A commonly used Gaussian basis set family is the Pople basis set, [77–80] which is designated

as x-yzG. x stands for the number of Gaussians combined per core atomic orbital. Valence

orbitals are described by two functions (y and z) which are a linear combination of y or z

Gaussians, respectively. In this case, it is a split valence double zeta basis set. Higher zeta


26

basis sets can also be used. Depending on the atoms being calculated, additional d and f

orbitals can be added to account for diffusion and polarization.

DFT provides not only the energy and the ground state density, but also forces on atoms which

can be used for geometry optimization. The geometry optimization is finished when for sum of

forces acting on the atoms is below a given threshold. Once this is done, for of a single molecule

and its relaxed electronic ground state in vacuum has been obtained, basic properties like bond

lengths, bond angles, electron density, or molecular orbitals (HOMO, LUMO) can be extracted

and visualized (Figure 4.1).

Figure 4.1 Left: DFT input of a molecular structure of a candidate for zinc oxide passivation. Right:

Visualized electronic ground state properties of a single molecule in a vacuum after geometry optimization

with B3LYP/6-31G.

The obtained electron density from the DFT calculation allows with further calculations to

determine of the following parameters among others:

• Accurate predictions of NMR shielding tensors and magnetic susceptibilities are usually

made with the Gauge-Independent Atomic Orbital (GIAO) method[81–85]. The Self-

Consistent Reaction Field (SCRF) is used with a polarizable continuum model (PCM)[86]

to account for solvation effects.

• Reliable estimation of XPS Core energies is done using the Koopmans theorem[87],

which states that the ionization energy is equal to the orbital's negative energy. After

geometry optimization, the requested electron is 'lifted' into the LUMO. The steepest

descent SCF calculates the difference in energy of this unstable state compared to the

ground state.[88]

• Calculation of total charge per atom in the molecule by e.g. Atomic polar tensors[89] (this

work), Mulliken[90], Hirshfeld[91] population analysis and others.


27

4.2 Atomic force microscopy

Atomic force microscopy (AFM) is a high-resolution method[92] for raster-imaging the

topography of surfaces by monitoring the interaction of an (oscillating) probe with the sample.

AFMs can operate in three different modes: contact mode, tapping mode, and non-contact

mode. Today, tapping mode is the most common AFM mode and the applied mode in this

thesis. A cantilever oscillates at its resonant frequency (frequency modulation) or near its

resonant frequency (amplitude modulation) in tapping mode. Van der Waals forces, dipole-

dipole interactions, and electrostatic forces modulate the amplitude during the probe's

intermittent contact with the surface. A photodetector is measuring a laser reflection by the tip

of the cantilever during the oscillation. A controller compensates for the measured change in

the amplitude of the oscillation by adjusting the cantilever's height above the surface. Monitoring

the height of the controller gives the topography image. Additionally, the phase changes reveal

varying stiffness or different adhesion of the probe at the surface (Figure 4.2). Due to the precise

control with piezo elements, a resolution of sub-nanometers is achievable, which is up to 1000

times better than the resolution of an optical microscope at the diffraction limit.

Figure 4.2 a) Schematic setup of an AFM with height controlled oscillating cantilever, laser, photodiode,

and moving sample surface b) exemplary topography of ZnO.

By statistical analysis of the topography, different moment-based parameters like the mean

height Ra and the root mean square roughness Rq can be determined.

Ra =1

N∑ zn

N

n=1

and Rq = √1

N∑(zn − z̅)2

N

n=1


28

4.3 X-ray photoelectron spectroscopy

X-ray photoelectron spectroscopy (XPS)[93] is a qualitative and quantitative element analysis

method at the surface of a sample. The intensity and energetic position of core photoelectron

peaks reveal the atomic composition of a sample and its chemical bonding. For this purpose, a

sample is irradiated in ultra-high vacuum (UHV) with X-rays generated by e.g. a Al or Mg source

with characteristic photon energies. Other common photon sources are UV lamps (UPS), lasers

or synchroton radiation. Due to the photoelectric effect, core electrons in the first 1 - 3 nm of

the surface absorb the energy and escape the material at the surface into the vacuum. The

kinetic energy (KE) of those emitted electrons is

𝐾𝐸 = ℎ𝑣 − 𝐵𝐸 − 𝛷𝑠𝑎𝑚𝑝𝑙𝑒

where BE is binding energy with respect to the samples Fermi energy, and Φsample is the work

function of the measured sample. Photoelectrons that pass an aperture and an entrance slit

enter a hemispherical electron analyzer. This analyzer filters photoelectrons by their kinetic

energy and is coupled with a channel electron multiplier to count the photoelectrons that pass

the exit slit.

A single peak is observable for s orbital electrons. Due to spin-orbit splitting, two peaks are

observable for electrons with non-zero angular momentum, i.e. p, d, and f orbitals. These peaks

show a defined area ratio depending on the orbital. Additionally, X-ray induced Auger peaks

and satellites are visible while measuring XPS spectra. Electrons from higher shells fill in the

vacancies of lower shells. The Auger electron from a third energetic shell is emitted by the

excess energy. Their peak positions are other than the X-ray photoelectrons discussed before

independent from the excitation energy. Other peaks that can be observed in the XPS spectra

are satellite peaks. They come from a second transition in the X-ray source. E.g., the less

intense radiation of Al Kα1,3 has 9.6 eV less energy and only 7.8% of the intensity.

For analysis of the chemical bonding, the chemical shift is calculated from the expected and

measured BE. This shift is usually in the range of a few meV to eV. Removing valence electrons

(oxidation), adding valence electrons (reduction), and the electronic effects of neighboring

atoms influence it the most. A fitted pseudo-Voigt function with Shirley background function

allows the precise extraction of peak position and full width half maximum (FWHM).


29

Figure 4.3 a) Typical XPS measurement setup with photon source, sample, aperture, entrance and exit

slit, hemispherical analyzer, and detector and b) examplary XPS measurement plot.

4.4 Nuclear magnetic resonance spectroscopy

Nuclear magnetic resonance (NMR) spectroscopy is a standard method for the structure

elucidation mainly of organic molecules, but also inorganic compounds.[94] NMR also allows

verifying if a reaction occurs, quantifying the progress of a reaction and whether products are

pure or mixtures. In the health sector, magnetic resonance imaging is nowadays one of the

most important medical imaging methods.

In general, atomic nuclei give rise to a magnetic dipole moment that stems from their associated

spin-angular momentum (spin). When a nucleus is exposed to an external magnetic field 𝐵0⃗⃗⃗⃗ ⃗ it

splits into multiple energetic states. E.g., for a nucleus with a degenerated spin of 1/2 it can split

into the states -1/2 and 1/2. The energetically favored state is a parallel alignment with the

external magnetic field (spin α), and the less energetically favored state is the anti-parallel

alignment (spin ß). The energy difference between two spins is the Zeeman-energy and is

defined as

∆𝐸 = 𝛾 ∙ ℏ ∙ 𝐵0


30

Where 𝛾and ℏ denote the gyromagnetic ratio and the reduced Planck's constant, respectively.

Because of quantization, the nuclear spin does not fully align with the external magnetic field

𝐵0⃗⃗⃗⃗ ⃗ and precesses around the z-axis of the magnetic field with the Larmor frequency. The

transition from the state spin α to the state spin ß can be excited by providing the resonance

energy ∆𝐸. Since the resonance energy is typically small compared to thermal energy (kB ∙ T)

and relatively easy to overcome, the population of both spin states is almost equal at room

temperature. The ratio depending on the absolute temperature T [K] can be described as a

Boltzmann distribution (kB = Boltzmann constant).

𝑁ß

𝑁𝛼

= 𝑒−∆𝐸𝑘𝐵𝑇 ≈ 1 −

∆𝐸

𝑘𝐵𝑇= 1 −

𝛾 ∙ ℏ ∙ 𝐵0

𝑘𝐵𝑇

Of course, only nuclei with a spin-quantum number 𝐼 ≠ 0 can be investigated by NMR. The

spin-quantum number depends on the exact composition of the respective nucleus, as

summarized in Table 4.4.1. Nuclei like 12C or 16O with an even number of protons and an even

number of neutrons have a nuclear spin of 0 and are thus NMR inactive. Popular NMR active

nuclei for liquid NMR measurements are 1H, 13C, 19F, 15N, and 31P (see Table 4.4.1).

Table 4.4.1 Overview of NMR active and not active nuclei. Red indicates no nuclear spin, thus no magnetic

moment and not NMR active. Green indicates a nuclear spin, thus a magnetic moment and NMR active.

Z (atomic number), N (neutrons) Nuclear spin number I examples

Z + N = even number Further distinction needed,

Z and N = even number I = 0 12C, 16O

Z and N = uneven number I = 1, 2, 3, … 2H, 10B, 14N

Z + N = uneven number I = 1/2, 3/2, 5/,2 …. 1H, 13C

An important factor for performing NMR experiments is the natural abundance of the respective

isotope. For example, the natural abundance of 1H is 99.9%, and thus high when compared to

13C, which only has a natural abundance of 1.1%. A lower abundance of course increases the

experimental time, but also, due to the spin dilution, does not allow to easily extract distance

information based on dipolar couplings between adjacent nuclei.

For the measurement, the sample is exposed to an electromagnetic wave to perturbate the

equilibrium nuclear magnetic dipole moment multiple times. The frequency of that pulse needs

to be sufficiently close to the Larmor frequency for the nuclear spins (depends on the measured

nucleus), and, for typically magnetic field strengths, lies in the radio-frequency (rf) range (MHz).


31

The sample emits a time-dependent rf signal while return to Boltzmann equilibrium. The typical

NMR spectrum is then obtained by a Fourier transformation of the time-domain signal.

Figure 4.4.1 NMR setup with magnetic field B0 induced by magnets, sample, rf transmitter and rf receiver.

The resonance of for an isotope, namely the chemical shift δ, depends on its surrounding

electron density and is independent of the external magnetic field strength.

𝛿 =frequency shift from the reference [Hz]

spectrometer resonance frequency [MHz]ppm

A higher electron density shields the nucleus and leads to resonances at higher fields. Vice

versa, unshielded nuclei with lower surrounding electron density have a resonance towards

higher frequencies. Chemical binding, hybridization, and different neighboring atoms vary the

electron density around a nucleus. For example, an electron-donating neighboring group results

in a more shielded nucleus. Thus, the chemical shift is characteristic of nuclei in similar binding

situations.

Since NMR is quantitative, the integral ratios in e.g. 1H NMR spectra allow to determine the

number of protons in different chemical surroundings. Integral ratios of hydrogen resonance

signals reveal the atomic ratio in the measured substance. This information can help to identify,

for example, whether a keto- or an enol tautomeric form of a molecule is present by analyzing

the number of hydrogen atoms next to the ketone functional group.

The so-called direct spin-spin coupling, or J-coupling causes a splitting of the signal for

chemically bonded atomic nuclei. The number of observed resonances, the multiplicity, can be


32

calculate for spin -1/2 isotopes like 1H as 1 + N, where N represents the number of coupled

nuclei. The coupling constants J are usually smaller than 5 Hz for 1H-1H-coupling. More

complex 2D experiments furthermore allow correlating the signals of different isotopes in

different chemical environments with each other.

4.5 Thermogravimetric analysis

Thermogravimetric analysis (TGA) is a method to monitor the change in mass of a sample over

time and temperature. This method allows measuring physical processes like evaporation or

sublimation. An increase or decrease in a sample's mass can also occur due to thermal

decomposition or chemical reduction/oxidation. Other processes that occur are phase

transitions, absorption, adsorption, chemisorption, and desorption.

During the measurement, a microscale continuously monitors the weight of the sample. An

oven heats the sample at a constant rate or a controlled rate with constant mass loss. Thermal

reactions can be monitored under defined atmospheres like ambient air, reducing atmosphere

(hydrogen), oxidating atmosphere (oxygen), or inert atmospheres (nitrogen, argon) and

different pressure regimes like a high vacuum, high pressure, and constant or controlled

pressure.

A TGA curve is understood as the measurement plotted as absolute or relative mass loss

versus time or temperature. The first derivative of that curve, the differential thermal analysis

(DTG) curve, allows further analysis and more in-depth insight into a reaction's kinetics.

Other analysis methods can be a useful addition to the TGA. Popular methods for TGA are

mass and IR spectroscopy and, lately, NMR. A coupled method can detect and identify reaction

and degradation products and assign them to a certain point of mass loss.

4.6 Fourier-transform infrared spectroscopy

Fourier-transform infrared spectroscopy is a method to obtain infrared absorption, emission, or

reflection spectra of solids, liquids, and gases.[94] It is based on the fact that molecules absorb

electromagnetic radiation at frequencies that match their vibrational frequency. These

frequencies depend, for example, on the molecular shape and the mass of the atoms. This


33

method is useful for identifying substances or functional groups by analyzing their characteristic

molecular or lattice vibrations (see Figure 4.2). The measured spectra are usually plotted

against wave numbers (1/ in reciprocal centimeters [cm-1])

Figure 4.2 Overview of typical infrared absorptions of different functional groups and bond types.[95]

Molecules have 3N degrees of freedom depending on the number of atoms (N). These degrees

are divided into translational, rotational and vibrational degrees of freedom. A monoatomic

“molecule” has only the three translational degrees of movement which arrives from the ability

to move in space (three directions in Cartesian coordinate system). Linear and non-linear

molecules have additionally 2 and 3 rotational degrees of freedom, respectively. Subtracting

the translational and rotational degrees of freemdom from the total amount of degrees of

freedom (3N) results in 3 x N - 5 vibrational degrees of freedom for linear molecules and

3 x N – 6 vibrational degrees of freedom for non-linear molecules. To be IR active, a vibration

must have a dipolemoment.

For example, N2 has 1 degree of freedom, and H2O has 3 degrees of freedom. The nitrogen

molecule would not be visible in an IR spectrum because the vibration causes no change in the

dipole. It is Raman active instead. A two atomic molecule that is IR active would be carbon

monoxide (CO). IR spectroscopy is not limited to molecules with 2 or 3 atoms and can also

measure large molecules' molecular vibrations.

The molecules' vibrations are classified into stretching (symmetric and asymmetric) and

bending or deformation (scissoring, twisting, wagging, rocking, and torsion), as Figure 4.3

visualizes.


34

Figure 4.3 Vibrational modes: a) symmetrical stretching and asymmetrical stretching, and b) in-plane

bending (rocking and scissoring) and out-of-plane bending (wagging and twisting).

A Michelson interferometer is a popular setup to measure an FTIR (Fourier transform infrared

spectroscopy) spectrum (Figure 4.4). Other than a traditional spectrometer, it measures an

interferogram, which is afterward converted to a useful spectrum by (fast) Fourier

transformation (FFT). Instead of measuring the absorption by changing the wavelength like a

traditional dispersive spectrometer, the sample is exposed to a broad IR radiation range. The

recorded interferogram is a function of the relative position of the scanning mirror in the setup.

This way, the FTIR spectrometer can collect data with a high resolution over a wide spectral

range.

The right sample preparation is essential for IR measurements to avoid the scattering effects

of large crystals. Standard sample preparation for solids is to 'dilute' a ground powder of the

sample with salts like KBr (potassium bromide), which is IR active at wave numbers below

< 200 cm-1. The mixture is then pressed at high pressures to a pellet, which IR radiation can

penetrate.


35

Figure 4.4 a) Standard Michelson interferometer setup for measurement of transmission FTIR spectra.

The basic setup consists of a (broad range) IR source, collimator, beam splitter, two mirrors, and a

detector. It yields b) interferograms converted by Fourier transformation (FFT) to c) IR absorption spectra.

4.7 Scanning electron microscopy

Scanning electron microscopy (SEM) is a method of raster-imaging surfaces in a UHV chamber

using a focused electron beam.[96]

Electrons are emitted by either a thermionic, a Schottky or a field-emission cathode and

accelerated by a voltage difference (0.1 eV – 50 keV) between anode and cathode. This beam

is then demagnifed by an electron-lens system from its initial size (50 m – 10 nm, depending

on the source) to a probe size of 1 – 10 nm. The current at the speciem usually varies between

10-9 to 10-12 A


36

Figure 4.5 SEM setup with Cathode, Anode, Condenser,Probe-forming lens, speciem and secondary

electron analyzer.

Elastic and inelastic scattered electrons at the speciem surface are analyzed. They are divided

by their gradual energy loss into secondary electrons (main 2 – 5 eV and up to 50 eV) ,

backscattered electrons (above 50 eV), Auger electrons (50 eV – 2 keV) and low-loss electrons

(energy close to the energy of primary electrons). The signal from secondary electrons were

used in this work and therefore explained in more detail. They origin from a few top layers (a

few nm) of a surface and are very susceptible for scattering. They can also be generated by

back scattered electrons besides their generation by primary electrons. The secondary

electrons are then collected by a positively biased collector grid and accelerated onto a

scintillator for quantization. The material contrast is achieved by the varying yield of secondary

electrons for different materials (depending on atomic number, chemical bonding and charges).

Tailored organic surface passivation for metal oxide semiconductors

37

5 Tailored organic surface passivation for metal oxide

semiconductors

This chapter aims for the first approach to reduce process temperatures by utilizing a

specifically for this purpose tailored organic molecule class for surface trap passivation of zinc

oxide thin-film transistors. The theoretical characterization of the molecules, the passivation

process, and performance of the resulting thin-film transistors are discussed in chronological

order.

At first, the molecule class 1,3-diketone is introduced in general and theoretically described by

density functional theory calculations. This chapter's focus lies in the description of effective

charges at the oxygen atoms and the possibility to tailor these by different substituents.

Afterward, their chemical binding towards zinc oxide nanoparticles is investigated by 1H NMR

experiments to evaluate how and if they bind to a zinc atom.

For the passivation of zinc oxide thin-film transistors, the chemical bonding at the surface is

monitored by X-ray photoelectron spectroscopy and the stacking of the molecules at the surface

by atomic force microscopy.

At the end of the chapter, the increase in the transistors' performance and their stability against

positive and negative bias stress is evaluated. This is followed by a brief overview of the

experimental parameters used, a conclusion, and an outlook.

5.1 Concept of surface traps passivation by tailored organic molecules

Amorphous and polycrystalline transparent semiconducting oxides such as ZnO,[97–99] SnO2,[100]

In2O3, or In-Ga-Zn-O (IGZO)[101] are promising materials for optoelectronic applications.[102,103]

The combination of their semiconducting properties and transparency in the visible part of the

electromagnetic spectrum makes them suitable candidates for novel applications such as

transparent displays and electronics. Compared to expensive vacuum deposition methods,

cost-efficient alternatives for thin film deposition are solution-processing methods such as spray

pyrolysis. This deposition technique can be used with water as a solvent and therefore offers

the advantage of complying with green chemistry standards. However, as the main drawback,

these materials reveal instability against moisture at ambient atmosphere under electrical

operation.[104,105] Since the fabrication of electric circuits requires stable transistor


38

performances, several investigations about the origin of trap states and their passivation have

been carried out, showing that with decreasing layer thickness (d < 30 nm), the instability

increases. The instability for thin layers is caused by surfaces states that act as trap states

when they are sufficient close to the conducting channel.[8]

Besides annealing at elevated temperatures,[106] one approach to fabricate electrically stable

oxide layers under ambient atmosphere is to deposit a second oxide layer like Y2O3,[107]

Al2O3,[108] SiO2, [108]

HfO2[108,109] or MgO,[110] a nitride layer SiN[111] or a fluoride layer by SF6

plasma treatment ZnF[112] on top of the semiconducting oxide layer to minimize the influence of

the surrounding atmosphere. However, by using such an approach, the newly introduced

surface between the two oxide layers or the oxide/nitride layer or oxide/fluoride layer becomes

the main drawback besides the cost-intensive vacuum processes and higher material

requirements.

Earlier, Ortel et al. found that a self-assembled monolayer of hexafluoropropylene oxide

(HFPO) is increasing the long-term stability and electrical performance of solution-processed

zinc oxide thin-film transistors by passivating surface trap states.[113] HFPO is applied as gas

and has no chemical positions to vary the molecule properties via side groups systematically.

Similar approaches to successfully functionalize zinc oxide surfaces through the chemical

binding of organic molecules through an oxygen anchor atom are reported in the literature.[114]

However, the electrical performance and stability of such functionalized zinc oxide

nanoparticles was not investigated.

In the following parts of this chapter, the passivation process of 1,3-diketones on solution-

processed zinc oxide transistors is monitored. The electrical performance and stability are

investigated after passivation. There general structure and possible binding are shown in

(Figure 5.1).

Figure 5.1 a) Hydroxide group adsorbed at a dangling bond of zinc oxide b) the molecular structure of

functionalized 1,3-diketones coordinating to a zinc atom at the zinc oxide surface to passivate surface

state induced trap states. Depending on the surface structure, sigma bonding to one of the oxygens might

be possible as well (not shown). [reused with permission[9]]

Compared to the previous binding of organic molecules with a single oxygen anchor atom,

diketones offer two oxygen anchor atoms. These diketones can also be systematically tailored


39

compared to HFPO or alcohols for the specific goal of surface passivation. Changing between

four different functional groups at the phenyl ring influences the partial charge distribution of the

molecule. Functional groups vary from strong electron-withdrawing properties (EWG) (NO2, F,

and Cl) towards weak electron-donating properties (EDG) (Me) to tailor the partial charge

distribution of the molecule. Diketones of this class appear under ambient conditions as an

amorphous solid and can be handled therefore as more safe and controllable passivation than

a gas.

5.2 1,3-diketone molecules: theoretical characterization

The following part describes the electronic ground states of 1,3-diketones with focus on the

effective charge at the oxygen anchor atoms and the manipulation of those by tailoring with

different substituents.

Molecular structures often represent only one of the multiple possible organic molecules'

arrangements, even though other molecular structures can describe the same molecule. In

most cases, the chosen molecular structure represents the most probable configuration (the

predominant form under given conditions). Other molecular structures are energetically less

favorable or inhibited forms and called tautomers. Tautomeric structures represent these.

Ketones like the 1,3-diketones are subject to keto-enol tautomerism and form four different enol

structures (each oxygen atom can form an E and a Z enol form) in addition to the diketone

structure shown in Figure 5.2. The predominant and experimentally observable forms of

1,3-diketones are Enol 2 and 4. A coordinative hydrogen-oxygen bond is forming a six-ring and

thus energetically stabilizing the tautomer. This stabilization does not occur for the diketone as

well as Enol 1 and 3. However, all tautomeric forms are possible candidates to bind to the zinc

oxide surface chemically.


40

Figure 5.2 Molecular structures of diketone structure, where Enol 1 & 3 indicate the E configuration and

Enol 2 & 4 indicate the Z configuration (experimentally observable free molecule form). Red arrows

indicate the tailoring of the electron density at both oxygen atoms depending on the substituent and a

trifluoromethyl group as a strong electron-withdrawing group at the other side. [reused with permission[9]]

From a theoretical point of view and evaluating the total energy calculated by DFT is that the

Enol 1 configuration is energetically favored. This trend is valid for all substituents. The total

energy is reduced by 7.6–7.7 mEh (millihartree), compared to the Diketone form as a reference.

However, further evaluation is done only for the Diketone form because it exists in liquid and

gaseous states[115] and is the intermediate state the keto-enol tautomerism undergoes. It is

representing a trend throughout the different substituents that all five tautomeric forms follow.

The calculated total energies and the net charges at the oxygen atoms are listed in Table 5.2.1.

Table 5.2.1 Calculated energies using the DFT functional B3LYP and the basis set 6-31-G* for the

functionalized 1,3-diketones and their tautomeric forms and the calculated effective charge for both

oxygen atoms.

Form Substituent E (B3LYP) [Eh] ΔE compared

to Diketone [Eh] Effective charge

oxygen #1 Effective charge

oxygen #2

Diketone

CH3 −874.5581 0 −0.725 −0.542

Cl −1294.8343 0 −0.720 −0.540

F −934.4726 0 −0.709 −0.539

NO2 −1039.7365 0 −0.691 −0.537

Enol 1

CH3 −874.5657 −0.0077 −0.808 −0.733

Cl −1294.8419 −0.0076 −0.804 −0.729

F −934.4802 −0.0076 −0.793 −0.720

NO2 −1039.7442 −0.0077 −0.777 −0.715

Enol 2

CH3 −874.5369 0.0211 −0.768 −0.741

Cl −1294.8131 0.0211 −0.776 −0.742

F −934.4512 0.0215 −0.766 −0.731

NO2 −1039.7052 0.0314 −0.773 −0.726

Enol 3

CH3 −874.5384 0.0196 −0.706 −0.604

Cl −1294.8136 0.0207 −0.700 −0.601

F −934.4517 0.0210 −0.690 −0.597

NO2 −1039.7045 0.0320 −0.683 −0.593

Enol 4

CH3 −874.5425 0.0156 −0.775 −0.612

Cl −1294.8126 0.0217 −0.770 −0.659

F −934.4509 0.0217 −0.763 −0.662

NO2 −1039.7037 0.0328 −0.763 −0.662


41

As introduced before, one advantage of 1,3-diketones is the possibility of tailoring the net

charge at the oxygen anchor atoms. Substitution at the para-position of the phenyl ring is

directly influencing the electronic structure of the whole molecule. Different substituents

increase or decrease the effective charge, described by the partial atomic tensor, at both

oxygen anchor atoms. For example, a methyl group is a weak electron-donating group (EDG)

with a positive inductive effect. A substitution with a methyl group results in higher absolute net

charges. Chlorine and fluorine are weak electron-withdrawing groups (EWG) that can be

substituted. Their negative inductive effect results in lower absolute net charges at the oxygen

anchors. Its negative mesomeric effect makes the nitro group a strong EWG. Substitution with

such strong EWGs results in the smallest absolute net charges. Figure 5.3 visualizes the results

of the calculation. The calculated electron density of the substituted molecules is on the right

side next to the relaxed geometries. Red, white and blue colors indicate the electrostatic

potential.

The trend of systematic changes in the electronic structure is visible in the electrostatic potential

at the substituent itself and the para-carbon position in the phenyl ring. The substituents replace

the hydrogen at this carbon atom. The other carbon atoms within the ring are exposed less to

the electron-withdrawing or donating effect but pass on the effect towards the oxygen anchor

atoms. All modifications from negative (methyl group) to neutral (chlorine) and to

positive/neutral (fluorine and nitro) can reach the oxygen atoms. However, the relaxed

geometries of all molecules are not affected by the substitution of the para position of the phenyl

and ring and show no observable change.

A comparison between the oxygen atoms reveals that the oxygen atom (#2) near the strongly

electron-withdrawing trifluoromethyl group generally shows fewer negative net charges and

smaller relative changes than oxygen atom (#1). The phenyl ring next to the oxygen atom can

explain the smaller effect with its general rather electron-donating nature. By modifying the

phenyl ring with the groups mentioned above, it acts as a “buffer” against the oxygen atom (#1).

This buffer explains two effects. First, the absolute net charge at the oxygen atom (#1) does

not decrease as much as it does near the trifluoromethyl group. Second, the systematic

variation at the phenyl ring causes the net charge at both oxygen atoms to decrease less than

in a comparable structure where there is no phenyl ring between the substituent and carbon

atom bonded to the oxygen atom (#1). The final effective net charge of the oxygen atom (#1)

decreases from -0.725, -0.720, -0.709 to -0.691 by increasing the substituent’s electron-

withdrawing effect.


42

Figure 5.3 Relaxed structures (left) and electron density surfaces (0.1 e−/a03) (right) (variated at the para-

carbon position) calculated with DFT (B3LYP/6-31G*) with the substituent a) CH3 b) Cl c) F and d) NO2

(color coding: atoms: H: white, C: grey/black, N: blue, O: red, F: cyan and Cl: green, surface potential:

red = negative, blue = positive). [reused with permission[9]]

5.3 Chemical binding of passivation molecules to zinc oxide surfaces

The previous chapter discussed the electronic structure of the passivation molecules' ground

states and quantified their electron withdrawing/donating character. This chapter analyzes the

binding of passivation molecules towards zinc and should answer the following questions:

1. Does the passivation molecule chemically bind towards zinc oxide?


43

2. Can NMR give information about the exact binding situation or at least narrow down

the possible binding situations?

The analysis is done on the passivation molecule with R = F because of the additional fluorine

resonance in 19F NMR (one signal comes from CF3 and the other from the fluorine bond to the

para-carbon position of the phenyl ring).

For the monitoring of the reaction, pure passivation molecules were dissolved in deuterated

acetonitrile (acetonitrile-D3). Several of these solutions are mixed with ZnO nanoparticles in

different molar ratios and reacted for 1 day. 1H, 13C, and 19F NMR spectra were measured for

the resulting solutions.

The previous chapter introduced the tautomerism of 1,3-diketones. Therefore, all spectra that

are discussed in this chapter show resonances of energetically less favorable tautomeric forms.

Additionally to those signals, other signals from different residuals, additional resonance signals

from ZnO nanoparticles, and the deuterated solvent are visible throughout the different spectra.

However, all these resonances (besides the deuterated solvent resonance signal) show a minor

intensity (<< 1 %) and are not commented on. Only resonances with relevant integral (protons)

or significant resonance intensity (fluorine, carbon) are considered for the discussion. The

residual solvent resonance is used as an internal reference.

At first, the 1H NMR spectra of the pure passivation molecule (Figure 5.4) are discussed. The

resonance signals of the protons in the molecule are the status quo compared to the resonance

signals of molecules bound to zinc.

Figure 5.4 1H NMR of pure passivation molecule with R = F in acetonitrile-D3.


44

The three main resonance signals are visible at chemical shifts around 7 – 8 ppm in the low

field and are marked in blue, red, and green. These signals integrate for 5 protons. The signal

shifted to the lowest field at a chemical shift of d 𝛿 = 8.12 ppm are assigned to the two protons

in meta-carbon position at the phenyl ring and the signal at 𝛿 = 7.31 ppm to the remaining two

protons of the phenyl ring at the ortho-carbon position. The signal at 𝛿 = 6.80 ppm integrates

for only one proton and is assigned to the carbon atom between the ketones. The resonance

signal of the hydrogen atom bond to oxygen as an alcohol functional group is not visible and is

typically expected as a very broad resonance signal for this substance class at a chemical shift

of ~ 16 ppm[115], which is beyond the measured frequency range. Thus, the integral of the single

proton resonance signal proves that the most energetically favorable form of the molecule is an

enol form, where only one hydrogen atom is bound to the bridging carbon (see Figure 5.5:

Hydrogen atoms highlighted by a green dotted circle). The previous DFT calculations showed

that Enol 1 is the energetically favorable form. This NMR experiment confirms the presence of

mainly an enol form.

Figure 5.5 Keto-enol tautomerism of 1,3-diketones with highlighted hydrogen atoms at the bridging carbon

atom. Both enol forms have a cis and trans form.

At this point, some hypothetical binding situations of diketones towards a zinc atom are

discussed before analysis of the 1H NMR spectra of reacted molecules with zinc oxide. This

discussion describes which NMR spectra to expect for a particular binding situation. Reactions

leading to some of those binding situations require hydrogenation of the compound or other

additional reactions (which are unlikely under the given reaction conditions). Some might be

energetically very unfavorable and thus would not be measurable under normal conditions.

However, they are still mentioned to present a comprehensive list of possible binding

configurations. Sixteen (two possibilities per molecular structure) of these hypothetical binding

situations, where the molecule binds through one or both oxygen atoms (coordinative and/or

covalent) towards the zinc atom, are visualized in Figure 5.6.


45

Figure 5.6 Sixteen (two per enumeration) hypothetical binding situations of diketone molecules towards a

zinc atom (other bonds of zinc atom not displayed). CF3 or p-C6H4F are either bound as R´ or R´´.

If the passivation molecules bind in any of those molecular structures, additionally to the

resonance signals for the 4 protons of the phenyl ring, additional characteristic resonance

signals for the remaining protons are expected. Each particular binding situation (Figure 5.6

a) – e)) results in a combination of a number of resonance signals and a number of protons in

the 1H NMR spectra. Cases a) and d) can involve additionally a coordinative binding of the other

oxygen atom towards zinc (not shown). The protons bond to the bridging carbon for the cases

d), e), and f) are diastereotopic hydrogens and thus count as non-equivalent hydrogens

because their neighboring carbon atom(s) is/are chiral. They are expected to yield two signals.

These combinations for the shown molecular structures are listed in Table 5.3.1.

Table 5.3.1 Number of expected resonance signals and protons for hypothetical molecular structures

(Proton signals and counts of R-groups excluded). The expected chemical shift 𝛿 [ppm] is additionally

given as an machine learning approach estimation.[116]

Molecular structure a) b) c) d) e) f) g) h)

Number of resonance signals 1 2 1 3 4 5 3 3

Number of protons 1 2 2 3 4 5 3 3

Expected chemical shift [ppm] ~ 6.2 ~ 5.0

~ 6.0 ~ 4.0

~ 3.3

~ 3.7

~ 5.5

~ 2.2

~ 2.4

~ 4.0

~ 5.1

~ 2.1

~ 2.3

~ 2.8

~ 4.4

~ 4.9

~ 3.2

~ 5.6

~ 5.8

~ 5.2

~ 5.9

> 10


46

At this point, the result of the 1H NMR of the reaction between passivation molecules with zinc

oxide nanoparticles (d < 100 nm) in a 1:1 ratio (molecules per zinc atom) is discussed, and the

identified relevant signals are compared to the resonance signals of the pure molecule and a

reference measurement of pure zinc oxide nanoparticles (see Figure 5.7).

Figure 5.7 Comparison of 1H NMR spectra of zinc oxide nanoparticles (top), pure passivation molecules

(bottom) and the reaction product of a 1:1 mixture (middle). Highlighted protons belong to the phenyl ring

(blue), the bridging carbon (red), hydroxyl groups at zinc oxide surface (green), and different solvent

residuals (grey & yellow).

The spectra reveal 5 groups of signals (blue, red, green, grey, and yellow). The resonance

signals highlighted in blue and red (shifted to the lower field) correspond to the protons of the

passivation molecule as discussed before (blue: phenyl ring, red: bridging carbon between the

ketones) and integrate in both measurements for 4 and 1 protons, respectively. All three signals

shift after the reaction with zinc oxide towards a higher field and are now located at 𝛿 = 8.05,

7.24, and 6.43 ppm. These shifted signals are the first indicator for chemical binding of the

molecules towards zinc oxide nanoparticles. Since there is only one resonance signal of the

passivation molecule that integrates for one proton (additional to the signals for the protons of

the phenyl ring), the measurement indicates the binding situation a) (see Figure 5.6 and Table

5.3.1). This binding situation can be reached by a simple condensation reaction between

hydroxyl at the zinc oxide surface and the passivating molecule.

The resonance signals highlighted in green are assigned to hydroxyl groups and chemisorbed

H2O present at the surface of zinc oxide.[114,117] The broadening of those peaks arises due to

different chemical environments for hydroxyl groups on the surface of the ZnO nanoparticles.

The signals shifts from 𝛿 = 2.13 ppm before passivation to 𝛿 = 2.33 ppm. The existence of the

peak after passivation in a 1:1 ratio indicates that not all chemisorbed water molecules and


47

hydroxyl groups are replaced by passivations molecules yet. However, the resonance signal is

shifted due to the new chemical environment of the zinc oxide surface, which is now partly bond

to passivation molecules. The remaining resonance signals are assigned to the deuterated

solvent acetonitrile-d3 (grey) and n-hexane or n-pentane (yellow) residuals.

The binding situation a) of the molecule does not specify whether the molecule is bound towards

the oxygen atom close to the phenyl ring or on the other side of the molecule close to CF3.

Comparing the expected chemical shifts of either molecule will not finally answer this question

but gives a tendency (see Figure 5.8).

Figure 5.8 Comparison of simulated 1H NMR resonance signals with the experimentally measured

resonance signals.

Variation of the molar ratio between zinc and the molecules by increasing the molecule

concentration reveals how many molecules can bind per zinc atom. For example, if two

molecules can bind to one zinc atom and the molar ratio is 1:4, then the ratio of free molecules

per bound molecules is expected to be 1:1. The expected ratio is plotted against the molar ratio

in Figure 5.9 for the cases that 1 molecule binds per zinc atom (red), 1.5 molecules bind per

zinc atom (yellow), 2 molecules bind per zinc atom (green), and three molecules bind per zinc

atom (blue). The measured ratios extracted from 1H NMR by integrating the hydrogen between

both carbonyl groups are shown as black stars and are in good agreement with the case that

one zinc atom can bind up to two molecules for the case that the zinc atome is isolated.


48

Figure 5.9 Molar ratio of zinc and passivation molecules vs. the ratio of free molecules to bound molecules

plotted for the cases where 1 (red), 1.5 (yellow), 2 (green), or 3 (blue) molecules chemically bind per zinc

atom. Measured ratios are extracted from the 1H NMR integrals are plotted as black stars.

Finally, all measurements are shown as an overview in Figure 5.10.

8 7 6 5 4 3 2 1

diketone-F

5:1

4:1

3:1

2:1

1:1

chemical shift [ppm]

ZnO

1H NMR in ACN

mol ratio molecule : zinc atom

exce

ss o

f p

assiv

atio

n

JK509

JK510

JK511

JK512

JK513

JK514

JK515

Figure 5.10 1H NMR measurements of zinc oxide nanoparticles (pure), 1,3-diketone with R = F, and

mixtures of both in different molar ratios. The experiments were measured in acetonitrile.


49

5.4 Passivation process monitoring

The previous chapter describes the chemical binding of diketone passivation molecules to zinc

oxide in solution. This chapter describes the morphology and presence of the diketone

passivation molecules on a solid-state zinc oxide surface. It also confirms whether the zinc

oxide weakly adsorbs the passivation molecules or chemically binds to them when they

passivate through a vacuum instead of reacting in solution.

AFM measurements help identify the surface's coverage by the diketones and characterize

their growth behavior. For that purpose, the diketone molecule with the methyl substituent

passivated the zinc oxide surface for a series of deposition times (0 s, 30 s, and 3600 s). The

other variations of the diketones are assumed to behave similarly, based on the similar

geometric and electronic ground state properties calculated by the DFT. The passivation

process divides into three phases, which Figure 5.11 visualizes in parts a, b, and c. Each part

shows a sketch of the observed surface, the respective topography with a dimension of

2.5 x 2.5 m2, and extracted height profiles. The non-passivated sample (Figure 5.11 a,

passivation time = 0 s) shows a surface of accumulated small spheres typical for zinc oxide

after deposition with a roughness of Ra = 0.57 nm. After a passivation time of 30 s (Figure 5.11

b), an almost closed self-assembled passivation layer is visible on top of the bare zinc oxide

surface. This surface is with a roughness of Ra = 1.74 nm, slightly rougher than zinc oxide. The

two extracted height profiles show the 2D topography of stacked molecules (red) and still

exposed underlying zinc oxide (blue). Both profiles differ in terms of the hill/valley distance and

height. The step height between a stack of molecules and zinc oxide is around 4 – 5 nm and,

therefore, as large as the height difference between two stacks of molecules. That corresponds

to approximately 4 – 5 molecules for their longest dimension from the optimized geometry by

DFT. After a passivation time of 3600 s (Figure 5.11 c), a fully closed bulk layer of passivation

molecules is visible on top of zinc oxide. The surface roughness significantly increased to

Ra = 9.5 nm. Zinc oxide spheres and molecule stacks are not visible anymore. The bulk

passivated surface is essential for electrical evaluation of the passivation performance since

incomplete organic layers close to the surface reduce the performance and even cause an

initial worsening of the zinc oxide.[118]


50

Figure 5.11 The schematic side view (left), AFM topography (center), and extracted height profile (right)

of a) 12 nm thick bare ZnO layer deposited by spray pyrolysis, b) ZnO surface passivated for 30 s with

the molecule shown above (R = CH3) and c) ZnO surface passivated for 60 min with the same molecule.

[reused with permission[9]]

The NMR spectra indicated that diketones chemically bind to zinc oxide nanoparticles from

solution. It is essential for passivated transistors to know whether the zinc oxide weakly adsorbs

in vacuum-deposited organic passivating molecules or if, in this case, a chemical binding

occurs. Comparing three XPS spectra of bare ZnO, bulk passivated ZnO, and the passivation

itself gives information about the different binding situations. Bare zinc oxide is compared to the

bulk passivated sample to achieve a good contrast between the binding energies. A sample

with shorter passivation time includes signals of both passivated and non-passivated areas.

Silicon peaks and part of the oxygen signal come from silicon with natural oxide, which is the

substrate in all three cases. The measurement of pristine diketones with methyl substituent is

the reference spectra. Figure 5.12 shows the measured overview spectra, and Table 5.4.1 lists

the detailed information about the fitted binding energies and FWHM.


51

Figure 5.12 XPS measurement of top/black: bare zinc oxide, middle/blue: passivated zinc oxide surface

60 min (R = CH3) and bottom/red: passivation molecule only (R = CH3). [reused with permission[9]]

The highlighted signals (Zn 2p 1/2, Zn 2p 3/2, and F 1s) allow following the molecules'

presence and thus to monitor the passivation process. A deposited passivation layer at the

surface reduces the Zn 2p signal compared to bare zinc oxide, and additionally, an F 1s signal

is observable. The binding energy of Zn 2p 3/2 decreases from 1029.5 eV (bare zinc oxide) to

1025.3 eV (passivated zinc oxide). The first is the characteristic binding energy of zinc bonds

in zinc oxide, and the second is zinc bonds to oxygen from an organic molecule.

The 1H NMR results of zinc oxide showed resonance signals that belong to the surface species

of zinc oxide, i.e. chemisorbed water or hydroxyl groups. Before passivation, two O 1s signals

with binding energies of 532.2 eV and 533.9 eV are attributed to the oxygen (O2−) within the

zinc oxide crystal and the aforementioned oxygen species (OH or H2O) at the surface,

respectively. Both signals disappear after passivation. The now observable O 1s has a new

binding energy of 534.6 eV, which corresponds to the passivation molecule's oxygen. The

passivation molecules (binding energy of 534.6 eV) replace hydroxyl groups or chemisorbed

water (533.9 eV) and other species at the surface.

Two C 1s signals of the molecule shift by 0.2 – 0.3 eV towards higher binding energy while the

binding energy of the C 1s signal from the CF3 group decreases by 0.4 eV. These findings

validate the NMR results that the molecules bind chemically to the ZnO surface.


52

Figure 5.13 XPS measurement of Zn 2p 3/2, F 1s, O 1s and C 1s for top: bare zinc oxide, middle:

passivated zinc oxide surface 60 min (R = CH3) and bottom: passivation molecule only (R = CH3).

Considering the shift of all binding energies of Zn 2p, F 1s, and C 1s reveals an unsystematic

shift that seems not to be induced by a change in the work function. A shift induced by

workfunction would shift all binding energies by the same amount.

The resulting composition of Zn:C:O:F is 2.0 : 12.3 : 3.6 : 3.1, indicates one passivation

molecule per two zinc atoms present in the XPS measurement volume.

Table 5.4.1 Measured Zn 2p 3/2, F 1s, O 1s, and C 1s binding energies (FWHM) for bare ZnO, passivated

ZnO, and the molecule (both R = CH3).

Bare ZnO (FWHM) [eV]

passivated ZnO (FWHM) [eV]

Molecule (R = CH3)

(FWHM) [eV]

Zn 2p 3/2 1029.5 (2.78) 1025.3 (2.90) –

F 1s – 690.8 (2.78) 690.5 (2.80)

O 1s 532.2 533.9

(2.58) 534.6 (2.85) 534.6 (2.82)

C 1s – 287.3 290.1 294.8

(2.61) 287.0 289.9 294.4

(2.61)

5.5 Performance of passivated thin film transistors

Electrical measurements of the ZnO thin-film transistors in the form of transfer curves are

carried out to evaluate the electrical performance and stability against BIAS stress before and

after passivation. From these data, the on-set voltage and hysteresis are extracted.

A theoretical curve has been fitted to the IV transfer characteristic according to the equation in


53

chapter 3.1 [68,69] to extract the electron mobility μ0 and disorder exponent for fixed Vaa = 50 V.

The passivated sample by the diketone with CH3 as a substituent is investigated regarding the

exposure time as representative for the other molecules. Stability after positive and negative

bias stress (PBS/NBS) was investigated for all molecules after two days of storage under an

ambient atmosphere. Some improvements in electrical properties are attributed to storing thin

zinc oxide films, which explains the lower off-current compared to the initial sample after

production and passivation. Figure 5.14 a shows that the hysteresis is clearly reduced

compared to a not passivated thin film transistor with about H = 6.0 V to H = 2.2 V after 5 s

exposure time and to H < 1 V for longer passivation times. The threshold voltage shifts from

Vth = 4.4 V to Vth = −7.3 V with increasing exposure time. Also, with longer passivation times,

the off-current increases and does not depend on the gate-voltage anymore. The disorder

parameter is raising from = 1.53 for 5 s passivation time to = 1.74 after 30 s passivation time

and then decreases to = 1.32 for 1 min and even further to = 1.08 after 60 min passivation

time. The smaller value of the disorder parameter indicates that there are fewer trap states

within the channel, and thus a more ideal TFT behavior is observable. Also, the electron mobility

follows the same trend and is dropping from μ0 = 3.14 cm2/V for a 5 s exposure time to

μ0 = 2.28 cm2/V for 30 s passivation time. For 1 min passivation time, the mobility is rising to

μ0 = 3.60 cm2/V and for 60 min further to μ0 = 5.22 cm2/V. TFTs utilizing the other three

passivation molecules achieve μ0 = 5.71 cm2/V and = 1.35 for Cl, μ0 = 4.60 cm2/V

and = 1.13 for F and μ0 = 4.54 cm2/V and = 1.23 for NO2 as substituents after 60 min of

passivation time. The initial worsening of mobility and disorder parameter below 1 min of

passivation time is caused by incomplete coverage of the surface by passivation molecules.

Improvement of the performance is observed after full coverage[118].


54

Figure 5.14 a) Transfer characteristic for different passivation times at VDS = 2 V for the molecule

(R = CH3) b) transfer characteristics for a 60 min passivated TFT before and after PBS/NBS stress with

VGS = ± 20 V for 4000 s at VDS = 5 V. [reused with permission[9]]

Analyzing the effect of negative and positive bias stress (4000 s) on the transistors electrical

performance after 60 min of passivation time shows an onset-voltage (ΔVOnset) shift (NBS/PBS)

of −1/3 V for CH3 (Figure 5.14 b), 0/4 V for Cl, 0/10 V for F and −0.5/12 V for NO2 (Figure 5.15).

Compared to an unpassivated zinc oxide TFT, which experiences an onset-voltage shift of

ΔVOnset = 14 V/19 V, all passivation molecules result in improved stress stability. All molecules

strongly reduce the effect of the negative BIAS stress, most of them even to ΔVOnset = 0 V. The

effect of positive BIAS stress correlates to the effective charge at the oxygen anchor atoms

calculated by DFT. Utilizing molecules with more electron density in the phenyl Ring (CH3 and

Cl) and, therefore, a more negative effective charge at the oxygen atoms results in higher

stability against positive BIAS stress. The best performance is observed for the molecule

(R = CH3).


55

Figure 5.15 Onset shift after PBS/NBS stress for bare ZnO and passivated samples with the substituent

R = CH3, Cl, F, and NO2 respectively after 60 min passivation and the effective charge for oxygen atom

#1 of the respective molecule. [reused with permission[9]]

5.6 Fabrication and characterization of thin-film transistors

Parts of the experimental parameters for this section are adapted with permission.[9]

Synthesis of 1,3-diketones is described in literature R = Me,[119–121] F,[119] Cl,[119–121], and NO2.[122]

Fabrication of ZnO thin-film transistors: Highly n-doped silicon wafers with predefined

source/drain contacts were used as substrates for ZnO thin-film transistors in a bottom-gate-

bottom-contact configuration. The thermally grown SiO2 layer (230 nm) and the silicon served

as gate dielectric and gate contact. 30 nm Au and 10 nm ITO served as source/drain contacts.

The predefined channel length was 20 m with a channel width of 2000 m. Before ZnO

deposition, the substrates were cleaned by 2 min ultrasonication in acetone (tech. grade),

followed by 10 min ultrasonication in a mixture of acetone and isopropanol (1:1, VLSI grade)

and finally rinsed with deionized water (18.3 MΩ) for 1 min. The substrates were dried first with

nitrogen and then on a hotplate at 120 °C before 10 min UV/ozone treatment. After the cleaning

steps, ZnO was deposited by spray pyrolysis using a custom setup utilizing an IWATA HP-B

Plus airbrush. The distance between the airbrush and the substrates was set to 250 mm, and

nitrogen was used as the carrier gas. 350 L of a 0.3 M zinc acetate dihydrate in deionized

water was sprayed under ambient conditions with 14 spray pulses (25 L each) at a substrate


56

temperature of 360 °C to deposit 12 nm thin zinc oxide layers. The samples were then annealed

at 500 °C for 45 min before analyzing and passivation. During the passivation process, samples

were placed in an evacuated desiccator at a pressure below 40 mbar upside down above the

passivation molecules with a distance of 20 mm. Passivation times of 5 s, 30 s, 1 min, and

60 min were used.

Chemicals: The Si/SiO2 substrates with predefined source and drain electrodes were obtained

from Fraunhofer IPMS. Zinc acetate dihydrate (99.999% metal trace basis) and used solvents

were ordered from Sigma-Aldrich. All materials were used as received, if not other mentioned,

without further purification.

Characterization methods: Electrical characterization of the transistors were conducted under

ambient atmosphere with a Keithley 2612A sourcemeter. If not other mentioned, samples were

characterized directly after production. The drain-source voltage VDS was kept constant at 2 V,

while the gate-source voltage VGS was swept from -20 V to 50 V and reverse. Onset-voltage

(Vonset) was extracted at 10-7 A and the hysteresis (H) at 10-6 A.

Positive and negative bias stress were conducted by applying a drain-source voltage of

VDS = 5 V and a gate-source voltage VGS = 20 V and -20 V, respectively, for 4000 s. Surface

topography imaging was carried out using a Veeco atomic force microscope (AFM). For X-ray

photoelectron spectroscopy, a Specs PHOIBOS 100 hemispherical energy analyzer and non-

monochromatized Al Kα1,2 radiation (E = 1486.6 eV) was used. The fitting of the XPS data was

done by the program CASA:XPSTM. A pseudo-Voigt-function was employed to fit the curves,

and a homogenous layer model was assumed.

1H NMR (400 MHz), 13C NMR (100 MHz), and 19F NMR (376 MHz) were recorded on a JEOL

ECX 400 MHz spectrometer.1H, 13C, and 19F NMR chemical shifts (𝛿) are reported in ppm from

tetramethylsilane (TMS), and CFCl3, using the residual solvent resonance as an internal

reference. Coupling constants (J) are given in Hz.

Computational methods: For all molecules and their tautomeric structures, geometries were

optimized. Calculations were done by density functional theory (DFT) with B3LYP (Becke

three-parameter Lee-Yang-Parr)[72–76] exchange-correlation functional and the Gaussian 09

code[123]. 6-31G*[79] was used as a basis set for all atoms (H, C, N, O, F and Cl). Afterward, the

frequency calculations were carried out, and it was ensured that no imaginary vibrations were

calculated. Avogadro[124] software was used to visualize the computational structures, where

the electron density is visualized with an iso value of 0.1 e-/a03. The electrostatic potential is

visualized by a color-coding with red, white, and blue referring to negative, neutral and positive

potentials, respectively. Atom colors are assigned as the following: H: white, C: grey/black, N:

blue, O: red, F: cyan, and Cl: green.


57

NMR resonance shifts were simulated by a machine learning algorithm of commercially

available software.[116]

5.7 Conclusion and outlook

This chapter described the passivation of solution-processed zinc oxide thin film transistors.

DFT was applied to calculate diketone properties before experimental passivation. Those

calculations revealed the diketones' systematic variation by utilizing substituents with an

electron-donating group (CH3) and electron-withdrawing groups (Cl, F, and NO2). The

substitution took place at the para carbon position of the phenyl ring. The methyl group as an

electron-donating group caused the highest electron density (most negative effective charge)

at the oxygen anchor atoms. The substitution by chloride, fluoride, and nitro continuously

decreased the electron density at the oxygen anchor atoms due to their increasing electron-

withdrawing effect. The electrostatic potential revealed that the electron density in the phenyl

ring significantly varies with different substituents.

The chemical binding of 1,3-diketones was shown for zinc oxide nanoparticles by 1H NMR

experiments and compared to theoretically expected and simulated resonance signal shifts.

The passivation of the zinc oxide surface was realized by evaporation of the diketone

molecules. Chemical bonding was further confirmed by XPS and the presence of the molecules

on the surface could be clearly followed by monitoring the F 1s peak. AFM measurements

additionally showed that the molecules deposited at the surface in stacks of 4 - 5 molecules.

Passivated thin-film transistors show clearly improved electrical properties regarding electron

mobility, hysteresis, and disorder parameter gamma. Also, there is a systematic shift of the

threshold-voltage to more negative values indicating less trapped electrons. The thin-film

transistors are stable under electrical stress, i.e., hysteresis is reduced to H < 1 V, and onset-

voltages do not shift under negative BIAS stress. Overall, the unreliable electrical performance

of solution processed zinc oxide thin film transistor induced by the trap states induced by, e.g.,

hydroxyl groups at the surface, can be significantly improved by diketone passivation. The

smallest improvement corresponds to diketones having side groups with electron-withdrawing

groups, while the best results are shown for thin-film transistors passivated by diketones with

the methyl group as a substituent due to the most negative effective charge at the anchoring

oxygen atoms.


58

Follow up research on this topic might be utilizing additional carbonyl groups in the molecules

with the same functional distance, namely triones or quadrones. The latter might polymerize at

the surface as they are known to produce polymers by reaction with zinc. Another possibility is

to reduce or increase the carbonyl groups' functional distance, which would inhibit or enhance

the enol tautomerism, respectively. The introduction of heteroatoms could increase the affinity

of the molecules towards zinc and is known from medicinal approaches.

Fluorinated Carboxylates as zinc oxide precursor

59

6 Fluorinated Carboxylates as zinc oxide precursor

This chapter aims for the second approach to reduce spray pyrolysis process temperatures by

identifying precursors with decomposition temperatures. In this thesis, perfluorinated zinc

carboxylate derivatives are investigated as new precursors for zinc oxide synthesis. After a

short motivation, the presence of FTIR absorption bands of zinc oxide for annealed precursor

is analyzed to prove that these compounds are suitable for zinc oxide production. Afterward,

the necessary reaction temperature for the conversion of the precursors to zinc oxide is

determined by TGA measurements. At the end of the chapter, a conclusion and an outlook are

given.

Perfluorinated zinc carboxylates are an interesting molecule class with the potential to be

utilized as a precursor for zinc oxide synthesis. The molecule class is a derivative of zinc

carboxylates, such as zinc acetate, that was used as the zinc oxide precursor in chapter 5. It is

known for zinc acetate that the transition to zinc oxide occurs at temperatures higher than

T = 325 °C in two main steps. First, acetic anhydride is split off after heating, and a tetranuclear

‘basic’ zinc acetate cluster is formed. After reaching temperatures higher than T = 360 °C, this

cluster decomposes into zinc acetate and acetic anhydride (see Figure 6.1).

Figure 6.1 Thermal conversion of zinc acetate to zinc oxide. (Ac = Acetic).

Such tetranuclear cluster analogue are also known as a stable intermediate for fluorinated zinc

carboxylate derivates in literature and were synthesized for different purposes, such as

transesterification of ß-keto-esters.[125] Also, methyl zinc-containing or TMEDA

(Tetramethylethylendiamin) containing fluorocarboxylates were used as atomic layer deposition

(ALD) zinc oxide precursors.[126] These precursors have the main drawback of decomposing to

zinc fluoride at higher temperatures. This thesis is utilizing the pure perfluorinated carboxylate

complexes without additional methyl or TMEDA groups. An exemplary optimized structure by

DFT (B3LYP/6-31G*) of a basic tetranuclear trifluoro zinc acetate cluster is shown in

Figure 6.2.


60

Figure 6.2 Optimized geometry of a tetranuclear basic trifluoro zinc acetate cluster by DFT

(B3LYP/6-31G*). Color coding: C: grey/black, O: red, F: cyan and Zn: violet.

For the investigation, whether the fluorinated zinc carboxylates Zn(RCOO2)2 (R = CF3, C2F5,

C3F7) are suitable precursors for thermal conversion to zinc oxide, the precursor were annealed

to 500 °C for 1 hour in a quartz crucible and the residue was analyzed by FTIR. The spectra

were compared to the spectra of zinc oxide nanoparticles (200 nm) and the spectra of zinc

acetate and trifluoro zinc acetate as examples of unreacted precursor molecules in Figure 6.3.

Figure 6.3 FTIR spectra of pure precursor zinc acetate (pink) and trifluoro zinc acetate (yellow), zinc oxide

nanoparticles (200 nm, black) and annealed precursor (500 °C, 1 h) zinc acetate (red), trifluoro zinc

acetate (green), pentafluoro zinc butyrate (blue) and heptafluoro zinc propionate (cyan).

All spectra show at around 3500 cm-1 a weak absorption band from adsorbed water or the water

of the atmosphere . Some spectra also show a weak absorption band for atmospheric CO2 at

around 2300 – 2400 cm-1. Both precursors show an absorption band for C=O at around


61

1500 cm-1 for zinc acetate and 1670 cm-1 for zinc trifluoro acetate. Zinc acetate shows four more

absorption bands at 1010 cm-1, 953 cm-1, 699 cm-1 and 615 cm-1 which are attributed to C-C

and C-H bonds. For trifluoro zinc acetate analogue absorption bands can be seen at

1440 cm-1, 1160 cm-1, 850 cm-1, 800 cm-1, and 725 cm-1. They are attributed to C-C and C-F

bonds. Furthermore, the zinc oxide reference shows the three typical merged strong absorption

bands around 500 cm-1.

The annealed spectra show that a major part of the pure precursors' absorption bands almost

completely disappear and the characteristic absorption band for ZnO is visible in all cases. This

can be interpreted as an almost full conversion of the precursors into zinc oxide, which makes

the precursors suitable for zinc oxide production. The presence of some impurities can be

caused due to the closed quartz crucible, where byproducts and fumes cannot fully evaporate

and escape. Pulsed spray pyrolysis allows processing of such precursors, where byproducts

can evaporate between two pulses and the impurities of the resulting layers are reduced.

The thermal degradation of the fluorinated zinc carboxylates and zinc acetate was then

analyzed by a TGA to determine the thermal degradation temperature.

50 100 150 200 250 300 350 400

0

20

40

60

80

100

mass lo

ss p

erc

en

t (%

)

temperature (°C)

Zn(OC(O)CH3)

2

Zn(OC(O)CF3)

2

Zn(OC(O)C2F

5)

2

Zn(OC(O)C3F

7)

2

Normalized to (0,100)

Figure 6.4 TGA measurements of zinc acetate (black), trifluoro zinc acetate (red), penta fluoro zinc

butyrate (green) and heptafluoro zinc propionate (blue).

Zinc acetate (black) as standard precursor shows that the first conversion into basic zinc

acetate occurs between 75 – 100 °C and is followed by a plateau. The second reaction step,

the conversion to zinc oxide, then begins at 200 °C and is fully converted at 360 °C. Zinc

trifluoroacetate (red) shows the initial conversion to basic zinc trifluoroacetate over a broad


62

temperature range between 50 – 275 °C. The final thermal conversion to zinc oxide occurs at

360 °C, which is higher then the process temperature of zinc acetate.

Pentafluoro zinc butyrate (green) shows analog to the trifluoro zinc acetate a thermal

conversion that occurs over a wide temperature range of 50 - ~250 °C. The higher temperature

limit can not be clearly defined, as there is no clear plateau. This means that the basic zinc

pentafluoro butyrate already reacts to zinc oxide while the first conversion step is still ongoing.

The final temperature where zinc oxide is obtained is 325 °C.

The two conversion steps of heptafluoro zinc propionate (blue) merge even more together and

temperature limits can be clearly defined. The final conversion to zinc oxide occurred for this

precursor at 250 °C, which is advantageous compared to zinc acetate for zinc oxide production.

From the point of view to reduce process temperatures for thin-film zinc oxide production by

spray pyrolysis, the latter precursors heptafluoro zinc propionate is a suitable candidate.

However, the precursor does not dissolve in water or other non-hazardous solvents like

alcohols and thus, the processing as an aqueous solution or other environment-friendly solution

is at this point not possible and was not further pursued.

High-speed real-time analysis of picoliter droplets under spray pyrolysis conditions

63

7 High-speed real-time analysis of picoliter droplets under

spray pyrolysis conditions

This chapter aims for the third approach to reduce process temperatures by (mainly) top-view

video of recording picoliter droplets with a high-speed camera under real application conditions,

i.e., in a spray pyrolysis setup. The primary deposition relevant processes from droplet

generation to evaporation mechanisms at hot surfaces are discussed in chronological order.

At first, the extraction of meaningful physical parameters from droplet images is explained and

then directly applied to determine the droplet diameter distribution after atomization. The mean

diameter is a fundamental parameter for this chapter.

Afterward, raytracing is introduced. Throughout the rest of the chapter, the droplet appearance

is compared to theoretically rendered droplets with extracted physical parameters to verify the

experimental findings.

After a short discussion of the two types of observed droplets, static and moving, an explanation

of the formation and interaction of sessile droplets follows. An evaporation model for sessile

droplets is presented and allows straightforward extraction of static and dynamic contact angles

depending on the contact line's velocity.

At the end of the chapter, a novel Leidenfrost like meta-stable hovering state of picoliter droplets

at low temperatures is presented, followed by experimental parameters, a conclusion, and an

outlook.

7.1 Cross-section analysis of droplet diameter after atomization

After detecting a droplet in an image, different methods allow the extraction of its edge position

and subsequently properties like diameter, circumference, area, roundness, or the mean

intensity of the pixels within a droplet. A straight forward approach is a binarization of the image

by a threshold value. In the simplest case, the threshold is a fixed value. In an 8-bit grayscale

image (the recorded images in this work) where the minimum value is 0 and the maximum is

255, a fixed threshold lies in between these two values. For example, if the threshold should be

at 50% of the intensity value, it would take the value 127. After binarization, pixel values below

the threshold become 0, and values above become 1.


64

Ideally, for the droplet analysis, droplets now appear as black circles separated from a white

background. However, this approach is prone to unstable or uneven illumination. The

background intensity can vary, for example, because of droplet clusters flying through the

optical path, reducing the illumination intensity temporarily. Also, droplets' pixels can become

blurred and vary in their intensity depending on how well focused a droplet is by the optics.

Droplets that are entirely in focus appear darker. Also, multiple droplets can overlay and appear

darker together. A filter could remove droplets that are not suitable for analysis, which would

drastically reduce the number of drops included in the statistics.

Advanced binarization algorithms can tackle the illumination problem with adaptive threshold

values. The adaptive threshold value is calculated, for example, as a Gaussian kernel for each

pixel individually depending on the pixel value and the values of its pixel neighborhood. It is not

fixed for a whole image like a static threshold value and depends highly on the number of

considered neighboring pixels and a subtracted offset constant. The optimal parameters can

fluctuate within a dataset and between different datasets.

However, varying the threshold value for each pixel might help detect droplets’ edges in images

with varying illumination and throughout a dataset with thousands of images where the

background is unstable. On the other hand, with varying threshold values, the precise extraction

of droplets’ edges is not accurate and does not yield physically meaningful contact lines.

At this point, the cross-section analysis is introduced. For example (see Figure 7.1), the cross-

sections of a droplet in X and Y direction (a), and the same rotated be 45° (b), are analyzed.

The droplet edge is located where the slope of the intensity is the highest. This can be easily

mathematically determined by the first derivative of the cross-sections that reveal the highest

intensity increase or decrease in the profile with the position of its maximum and minimum

(marked in Figure 7.1 as big colored circles). Thus, both extreme points in a cross-section

represent the exact position of the contact line (c), and the distance between both is the

diameter of the droplet. The latter is statistically determined by this technique from multiple

cross-sections and can also handle droplets that are not entirely focused. Thus, this technique

delivers meaningful and consistent physical data.


65

Figure 7.1 Cross-section analysis of sessile droplet. Cross-sections in X and Y direction a), and cross-

sections in XY and YX plane b) yield a mean diameter of 19.7 m. The extracted position of the contact

line is superimposed onto the droplet c) (1 pixel ~ 3 m).

As the starting point for the droplet analysis, the atomized droplets are characterized by

applying the cross-section analysis. Therefore, Figure 7.2 a) shows a recorded side-view image

of droplets that leave the nozzle after atomization. The nozzle's atomization platform is visible

as a blurred black silhouette at the top of the image. Droplets in focus are visible as black

circles. Other not entirely focused droplets appear blurred and lighter in intensity. Analysis of

droplets in 5000 frames with 2 kHz framerate for different atomization power (varying amplitude

of the atomization platform) gives a histogram of the droplet diameter distribution. The diameter

distribution is plotted in Figure 7.2 b). For visualization purposes, the curve was smoothed by

a kernel density estimation (by Gaussian mixture model). The used ultrasonic atomization

nozzle generates asymmetric distributions of droplet sizes at an atomization frequency

f = 130 kHz. Each distribution has a shoulder to larger diameters (D > 30 m). Higher power

generates droplets with a sharper distribution and a larger mean diameter compared to lower

power. The majority of droplets have diameters from 10 to 30 m. However, the calculation of

droplet properties in the following chapters is done only for a single droplet diameter of

d = 25 m.


66

Figure 7.2 a) Generated droplets leaving the atomizing platform (black silhouette at the top of the image)

with scale bar = 50 m b) Kernel density estimation (by Gaussian mixture model) of diameter distribution

for different atomization powers.

7.2 Raytracing and droplet appearance

Rendering of photorealistic images is based on the principle of raytracing and was first

introduced by Appel in 1968[127] and Goldstein and Nagel in 1971[128]. Raytracing is mostly

applied in the entertainment industry for digital film production and video games, in which the

image quality has been significantly improved in recent years through real-time raytracing. In

this thesis, ray tracing is used to render certain top-view scenes of drops near a silicon surface

to simulate droplets' appearance at a different height. It also allows the comparison of measured

sessile droplets with known contact angle Θ and flying droplets to their theoretically expected

appearance to verify the extracted droplet parameters. The basic principle of raytracing is that

rays emitted by a source are tracked through a scene, in this case, the droplet on the substrate

and the surrounding atmosphere, until they reach an observer or camera. Raytracing was done

in this thesis using in the working group already existing self-coded raytracing scripts.

Physical processes such as reflection at object surfaces or refraction while propagating from

one medium into another can be precisely calculated using the material properties. With this

approach, images can be rendered whose realism is very close to real pictures. Besides the

scene's geometry, the wavelength of the light ( = 447.5 nm), the optical constants n and k

depending on the wavelength (nair = 1.00, nwater = 1.33, nSi = 4.70, kSi = 1.04 x 10-9 @ 447.5 nm),

and the F-number of the optics (F = 8) are needed to render the image. More realism can be


67

achieved by taking spherical aberrations or interference into account. Also, the number of

reflections until a ray is being dismissed can be increased. 4 Reflections were taken into

account in this work. Figure 7.3 a) shows an incoming tracked ray 1 (black) with a path through

the air towards a sessile water droplet on a silicon surface. At the surface of the droplet,

reflection 1 and refraction occur 2. Lighter red indicates attenuated light intensities due to beam

splitting. After reflecting at the silicon surface, part of the ray undergoes TIR (total internal

reflection) within the droplet 3, and a significant part of the ray is leaving the droplet 4. Figure

7.3 b) illustrates the fully rendered image showing the sessile droplet in top view (XY-plane) on

the silicon surface. Light is observable in the middle of the droplet, where a major part of the

light propagates through the droplet and is reflected at the silicon surface towards the camera.

The intensity of the reflected light reduces drastically when proceeding towards the edge of the

droplet, where major parts of the light are reflected at the surface of the droplet at an angle that

is not received by the optics.

Figure 7.3 Raytracing example of a 65.45 pL H2O sessile droplet on Si surface with a contact angle

Θ = 19.8°, illuminated with monochromatic blue light ( = 447.5 nm), and a F-number of 8: a) Side view

of traced ray. Black indicates initial light intensity where decreasing intensities are shown by lighter getting

red. The scene contains reflection, refraction, and propagation through media (1,2,4) and total internal

reflection (TIR, 3). b) Rendered top view image (XY-plane) of the sessile droplet where more white

indicates higher light intensities and blacker the opposite.

Recorded top-view images in a perpendicular arrangement with respect to the monitored silicon

surface do not provide height information about the droplet above the surface until it becomes

sessile. Raytracing allows limiting the height of a droplet into 4 categories, without further need

of side-view images (see Figure 7.4). A droplet that is approaching the surface (1) but is still far

away (100 – 1000 m) appear blurred because of the used optics. Droplets closer to the

surface (10 – 100 m) appear as sharp black circles (2). A droplet that crosses the height of

10 m and approaches the surface very close (3) can be identified by a vigorous white intensity

that is almost filling the full visible circle. Here the drop acts like a microlens in focal distance

above the surface. Once a droplet makes contact with the surface (4), the larger visible area is

clearly visible due to spreading and a small spot of high light intensity in the center.


68

Figure 7.4 Illustrated droplets at different heights a) and the expected appearance rendered with raytracing

b). 1) 100 – 1000 m: blurred appearance, “) 10 – 100 m: sharp black visible circle, 3) 10 m: strong

light intensity filling almost the full visible circle, and 4) the sessile droplet: enlarged visible area with a

small spot in the center of high light intensity.

7.3 Sessile and moving droplets: Indicator for Leidenfrost temperature

A significant change in the behavior of droplets at hot surfaces occurs at the Leidenfrost

temperature TL. Droplets start hovering on their own (sacrificial) vapor layer without direct

contact with the surface. The initial and most trivial assessment of the Leidenfrost temperature

is to monitor whether droplets come into direct contact with the surface forming a sessile

droplet, or hover above the surface. For that, three states can be distinguished in the droplet

analysis: 1. spherical droplet moving at velocity v above the surface, 2. from a side perspective

(slightly) elliptically shaped droplet experiencing the Leidenfrost phenomenon and moving near

the surface, and 3. sessile droplet with a direct interface to the heated substrate surface.

Representative image sequences for droplets with status 1 and 3 are shown in Figure 7.5 a).

The schematic side view in Figure 7.5 b) illustrates that the droplets at the bottom of the field

of view do not experience the Leidenfrost phenomenon and therefore wet the substrate surface

after making contact. This can be seen in the frame sequence by following the red dotted

droplet. After wetting the surface, this droplet is heated immediately to boiling temperature on

the surface and begins to evaporate. This evaporation process is visualized for 4.5 ms.


69

Figure 7.5 a) Continuously evaporating sessile evaporating water drop on a silicon wafer (red dotted circle)

visualized over a 4.5 ms time interval recorded below the Leidenfrost temperature for a substrate

temperature of T = 140 °C. b) Schematic side view of drops below the Leidenfrost temperature. Shown

are sessile drops with a direct interface to the heated surface, moving drops close to the surface (single

appearance and optically sharp), and further above moving drops, which are quadrupled due to pulsed

illumination and/or optically blurred along their trajectory due to their velocity. [reused with permission[129]]

In addition to the highlighted sessile droplet, other free moving droplets are visible. Optically

sharp droplets can be recognized as being close to the surface (10 – 100 m from raytracing).

Some droplets are very close to the surface ( < 10 m from raytracing) and can be identified

due to the intense white light reflex in the droplet center. The main force acting on the droplets

is the surface tension, which maintains for droplets with such small volumes (V < 100 pL) their

spherical shape. Their area appears smaller than the initial area of a droplet after wetting the

surface, where it spreads. These droplets are illustrated in the schematic as droplets close to

the surface but not in contact. They differ from the higher moving droplets by a noticeably slower

speed of movement. The remaining droplets move faster and with a larger distance to the

surface. They appear more blurred when they are beyond the field of view of the used optics.


70

It can also be observed that due to their relatively high speed, these droplets blur along their

trajectory and thus appear stretched. The fastest droplets appear as four single sharp droplets

with higher brightness along their trajectory. The quadruple pulsed exposure achieves this

optical superimposing during the acquisition of a frame. These droplets are illustrated in the

schematic in the upper part of the field of view.

Figure 7.6 a) Moving drop experiencing the Leidenfrost phenomenon (red dotted circle) with an

observation time of 7.5 ms recorded above the Leidenfrost temperature for a substrate temperature of

T = 210 °C. b) Schematic side view of drops at and above the Leidenfrost temperature. Shown are moving

drops experiencing the Leidenfrost phenomenon, moving drops close to the surface (individually and

sharply visible), and further above moving drops, which are quadrupled due to pulsed illumination and

optically blurred along their trajectory due to their velocity. [reused with permission[129]]

Figure 7.6 a) highlights with the red dotted circle a droplet that is moving through the camera's

lateral field of view. This droplet experiences the Leidenfrost phenomenon and glides with

constant speed on its own gas phase. The base area of the droplet does not decrease during

the 7 ms observation time. The evaporation rate is very low because of the vapor layer that


71

thermally isolates the droplet from the surface. A small light reflex is visible in the droplet center

that gives the information that the droplet hovers at a height around 10 m above the surface.

Droplets experiencing the Leidenfrost phenomenon are illustrated in Figure 7.6 b) at the bottom

close to the substrate surface. Droplets that move at a greater distance from the substrate

surface are described and illustrated in the same way as in Figure 7.5. By observing the droplets

of interest as described in Figure 7.5 and Figure 7.6, the temperature range of the LFP can be

narrowed down with little effort. From this simple approach, a Leidenfrost temperature of

TL = 210 °C is measured for picoliter droplets on a silicon wafer. An exact determination based

on the observed evaporation rate is made by statistical evaluation in chapter 7.3.

7.1 Formation of sessile droplets and their interaction with the

environment

The formation of sessile droplets (as described Figure 7.7) is a key factor for the quality of a

process in many applications below the Leidenfrost temperature. The following formation

scenarios could be observed.

Figure 7.7 Temporal sequence of sessile droplets formation at a heated surface. Incoming droplet(s) at 0

ms, spreading at the surface at 0.5 ms, and the full spread sessile droplet at 1 ms. Showing for the trivial

case (a) of one droplet hitting the surface and complex cases of up to 5 droplets (b - e) incoming and

spreading simultaneously resulting in a single sessile droplet. Droplet events were observed in a

temperature range of T = 115 – 170 °C. [reused with permission[129]]

A single droplet approaches the surface in the trivial case until it makes contact. Wetting creates

an interface with the substrate surface (Figure 7.7 a). The droplet is visible before wetting and

overlaps as a sessile droplet with a formed interface at the substrate surface within a frame due

to the quadruple pulsed exposure. In this scenario, it can be assumed that the volume of the


72

sessile droplets Vsub is equal to the volume of the droplet in air Vair before making contact. For

droplets of this size, the contact and spreading on the surface takes ~ tr (introduced in the next

subchapter), which is less than 125 s (experimental resolution, time period between two light

pulses), and their appearance is proportional to the rate of incoming droplets.

A sufficiently high droplet flow and a high local and temporal droplet density near the substrate

allow further sessile droplet formation scenarios. Single droplets that do not follow the trivial

case described above now reach the substrate surface by merging with two or more droplets.

Figure 7.7 (b-e) shows how several droplets (up to five in this example) collide and combine

their volumes on the substrate surface. The wetting and spreading of these droplets on the

substrate surface take place in the same time frame as the trivial case. The volume of the

resulting droplet Vsub is the sum of the initial droplets in air Vair,n. Multi-droplet formations result

in an increased lifetime of the sessile droplets compared to sessile droplets formed from a

single droplet.

In a third scenario (shown in the evaporation model discussion), droplets that have not yet

followed the trivial case meet already sessile evaporating. The combined volume and the

adjusted temperature of both droplets affect the already running evaporation process and

contact angle. The second and third scenarios extend the trivial view of the first scenario that

the lifetime and evaporation dynamic of the droplets depend not only on the average volume of

the generated droplets but also on the interaction of several droplets with each other. At low

temperatures, the third scenario leads to complete coverage of the surface by the liquid if the

total incoming flow rate is higher than the evaporation rate and the volume increases faster

than the evaporation process occurs.

In Figure 7.8 a), the interaction of an evaporating sessile droplet with neighboring droplets

below the Leidenfrost temperature is recognizable. As known by literature, a sessile droplet

evaporates predominantly at the edge.[55] Due to the additional hot gas volume created locally

around the droplet, the local pressure increases, resulting in the acceleration of the neighboring

droplets (red) near the substrate away from the droplet. This effect is visible within the first

500 s after evaporation starts. The horizontal flow of air along the substrate surface creates

further turbulence, creating a suction towards the evaporating droplet. Nearby droplets (green)

at the suction level are accelerated in the evaporating droplet's direction, which takes place with

a time delay of about 500 – 1000 s after the evaporation starts.

At surface temperatures above the LFP, the interaction of evaporating droplets with nearby

droplets is neither repulsive nor attractive. This finding suggests that the droplets that come

critically close to the substrate at temperatures greater than or equal to the LFP do not

evaporate locally at the edges, and not such unequal pressure conditions are created. In

Figure 7.8. b) a droplet is shown, which comes critically close to the substrate with a surface


73

temperature just above the LFP and undergoes secondary atomization. The secondary smaller

droplets are accelerated so fast that they are only weakly recognizable in the next light pulse

and have already left the field of view in the following frame.

Figure 7.8 Interaction of evaporating drops a) below the Leidenfrost temperature and b) above Leidenfrost

temperature (T = 210 °C). While below the Leidenfrost temperature, drops near the substrate surface are

repelled from the evaporation center, higher moving drops are attracted to the evaporation center with a

time delay. Above the Leidenfrost temperature, no interaction with neighboring drops can be observed.

Here the evaporating droplet undergoes secondary atomization (yellow arrows). [reused with

permission[129]]

7.2 Evaporation model of sessile droplets: Ad hoc extraction of the

contact angle

Temperature evolution of the substrate

For pico-liter droplets (rair ≤ 12.5 m) gas bubbles or bubble ring formations are not observed,

in contrast to other studies on evaporation behavior of larger droplets. This holds for boiling

below the Leidenfrost temperature. Therefore, the observed area and diameter can be directly


74

used for analysis without the need to account for any gas bubbles inside the droplet, which

would increase the volume.

Droplet processes as the impact at the surface, spreading, oscillations of the droplet shape at

the surface, and flow of droplets is commonly characterized by the Ohnesorge number (Oh),

Weber number (We), and Reynolds number (Re).

The characteristic heat diffusion length L for time t after contact with a hot surface is L = (αt)0.5

with α the thermal diffusivity.

For water (α = 0.143x10-6 m2/s at 25 °C) and after a frame-to-frame period t = 500 s, the heat

diffusion length is L = 8.5 m. This is about 2/3 of the droplet height at the center. The

evaporation is expected to start immediately within the first frame at the edge of the droplet

where the local height goes to zero. Correspondingly, for the further evaluation evaporation at

the edge of the droplet is assumed as also reported in literature.[55]

To verify, that the silicon substrate temperature remains essentially unaffected by the drop

evaporation process, the corresponding heat equation in the substrate was numerically solved

(details in experimental part). The calculation shows that the substrate temperature variation

remains small and less than 1 °C (see Figure 7.3).

Figure 7.9 a) Temperature change over time at different positions relative to the droplet, b) spatial

temperature change at specific times at r = 0. [reused with permission[129]]

An initial cooling of the substrate can be seen close to t = 0 ms, when the droplet is initially

heated up to boiling temperature and starts to evaporate. This depends on the place on the

substrate, i.e., the local drop height. Afterward, the temperature change stays almost constant

for the full evaporation time. The substrate heats up to the initial temperature as soon as there

is no more contact with the drop. Figure 7.3 a) shows that this occurs at different places on the

substrate at different times and with different temperature transfer. The substrate provides the

remaining energy for the evaporation of the droplet in the center of the droplet and cools down


75

further until it heats up to the initial temperature. Figure 7.3 b) shows that the change of

temperature in the center of the drop is the highest at any time. However, as mentioned above,

the temperature change of the substrate is less than 1 °C and thus for the experimental

temperatures observerd in this thesis less then 1% and is neglected in the further analysis.

Evaporation model

The experimental setup allows to measure the visible droplet radius in air rair from phase 1 and

the visible droplet radius on the substrate rsub after landing while evaporating (phase 2 and

phase 3).

Thus, the temporal change of rsub(t) during the droplet's lifetime T on the surface is known. For

the following evaporation model it is assumed that the volume evaporation rate is proportional

to the circumference of the droplet

�̇� = −𝒄 ∗ 𝒓𝒔𝒖𝒃 (4)

with proportional constant c. Knowing the initial volume 𝑽𝟎 =𝟒𝝅

𝟑𝒓𝒂𝒊𝒓

𝟑 we can determine this

constant c to

𝒄 = 𝑽𝟎

∫ 𝒓𝒔𝒖𝒃(𝒕)𝒅𝒕𝑻

𝟎

(5)

The volume of the droplet on the surface during evaporation (phase 2 and 3) is given by

integrating over time as

𝑽(𝒕) = 𝑽𝟎 − 𝒄 ∫ 𝒓𝒔𝒖𝒃(𝒕′)𝒅𝒕′𝒕

𝟎 (6)

Knowing the volume V and the contact radius rsub and assuming a spherical cap shape and

contact angles smaller than 90° allows straight forward calculation of the contact angle Θ for

any time while the droplet resides on the surface.

𝐭𝐚𝐧 (𝜽

𝟐) = 𝒌 −

𝟏

𝒌 with 𝒌 = (√𝟏 + (

𝟑𝑽

𝝅𝒓𝒔𝒖𝒃𝟑 )

𝟐

+𝟑𝑽

𝝅𝒓𝒔𝒖𝒃𝟑 )

𝟏

𝟑

(7)

Finally, the ß spread factor calculates as ß = 𝑟𝑠𝑢𝑏

𝑟𝑎𝑖𝑟 .

The visible area of a droplet as a function of time is divided into three phases as mentioned

before (Figure 7.10). In the initial phase the incoming droplet is spherical and freely moving

above the surface. In this phase, the droplet has no direct contact with the substrate and

experiences no significant temperature increase. The second phase begins after the droplet

hits the surface, and the enlarged visible area has formed. For picoliter droplets in this work no

relevant over-spreading is expected.[130] For a droplet with a diameter d = 25 m the receding


76

motion of the contact line starts after impact and initial spreading at tr ~ 15 s.[130] Thus, the

overall formation time is very short (< 125 s) and below the experimental resolution. The

droplet starts to heat up, and significant evaporation begins. This is attributed to phase 2. The

beginning of this phase is marked on the time scale as t = 0.

Figure 7.10 Temporal evolution of the visible area of an evaporating sessile droplet divided into three

phases. 1: Droplet in air, 2: evaporation with pinned area beginning with static contact angle Θa until it

decreases to static contact angle Θr, 3: evaporation with moving contact line and a dynamic receding

contact angle Θdr. Calculated contact angle is shown in blue and calculated volume is shown in red. ß

spread factor is shown as crosses. Transition between static advancing and receding contact angle and

dynamic receding contact angle is indicated by a grey zone. Above the graph selected images from the

droplet over time together with simulated images by raytracing. Enlarged images of the mapped droplet

are available in the SI. [reused with permission[129]]

This phase is characterized by a constant visible area indicating a pinned outer contact line.

The evaporation where the visible area is decreasing linearly is attributed as phase 3. The

evaporation occurs with dynamic receding contact angle Θdr. Finally, the droplet is fully

evaporated. The total lifetime of a sessile droplet is therefore the sum of phase 2 and 3. The

observed linear decrease of the contact area with time correlates mathematically with an

evaporation rate proportional to the contact radius when assuming a constant contact angle.

Obtained contact angles agree with raytracing simulation of the observed droplet images

(Figure 7.10 Upper part). Droplets evaporate in phase 2 with a pinned constant contact area


77

and a dynamically decreasing contact angle starting with the advancing contact angle Θa. until

Θr is reached. After transition to phase 3 the contact angle now depends on the contact line

movement and is referred to as dynamic receding contact angle Θdr

For DI-water on silicon with natural oxide at T = 110 °C an advancing contact angle

Θa = 37.2 ± 3.3° and a receding contact angle Θdr = 9.1 ± 1.6° was measured by analysis of

36 droplets and is consistent with the apparent contact angle reported in literature for very thin

SiO2 layers.[131]

The transition from a static contact angle regime to the dynamic contact angle regime is

visualized in Figure 7.11.

Figure 7.11 Droplet appearance at different evaporation points: a) static contact angle evaporation regime

(Θa ≥ Θ ≥ Θr), b) transition between static and dynamic evaporation regime, and c) dynamic contact angle

evaporation regime (Θr ≥ Θ ≥ Θdr). Each stage is visualized by simulated top-view, experimental top, a

side-view sketch and side-view with reflection image. [reused with permission[129]]

As mentioned in chapter 7.1 as the third scenario, under real conditions it can occur that

droplets can collaps into already evaporating sessile droplets. They combine their volume and

start to evaporate as a sessile droplet with increased volume. In general, this evaporation event

of multi-droplets can be analyzed analogous to single droplets . V0 is the combined volume of

all droplets that evaporated together as a single sessile droplet. For ß spread factor calculation,

rair is calculated from the combined volumes from the time points on, where the second (or

further droplets) droplet merged into the sessile droplet. For this kind of event four different

cases can be distinguished (see Figure 7.12).


78

Figure 7.12 Different cases of a droplet that collides into a sessile evaporating droplet. Either the new

droplet hits the evaporating droplet at the contact line with a) Vsecond drop < Vevap (small volumes), and b)

Vsecond drop >> Vevap. Or it hits the droplet from the top with c) Vsecond drop ≤ Vevap, and d) Vsecond drop > Vevap.

The incoming droplet can hit the evaporating sessile droplet at the contact line (Figure 7.12

a and b) or from the top (Figure 7.12 c and d). For cases a and b, the visible area of the droplet

is increased in any case. If the volume of the incoming droplet is too small to compensate for

the evaporated volume and the increased visible area, the contact angle Θ is smaller than the

static advancing contact angle Θa. This case is visualized in Figure 7.13.

Figure 7.13 Temporal evolution of the visible area of an evaporating sessile droplet that is hit by a second

droplet. The evaporation is divided into three phases. 1: Droplet in air, 2: evaporation of a single droplet

analogue to figure 4, and 3: droplet is hit by another droplet and volume, area and contact angle increase,

before evaporation continues. Calculated contact angle is shown in blue and calculated volume is shown

in red. ß spread factor is shown as crosses (d0 is the combined volume of both droplets from phase 3 on).

Above the graph selected images from the droplet over time together with simulated images by raytracing.


79

For case b, in which the volume of the incoming droplet is sufficient large enough, the contact

angle Θ can take values up to dynamic advancing contact angle Θda. The latter occurs if the

incoming volume is so large that the contact angle for the new volume and contact area is larger

than the static advancing contact angle Θa. The contact line velocity increases and the visible

area is spreading. While spreading and with moving contact line, the system can take contact

angles up to the maximum dynamic advancing contact angle Θda. As soon as the contact line

velocity is 0 the contact angle is equal to the static advancing contact angle Θa.

For case c, the volume of the evaporating droplet is refilled if the volume of the incoming droplet

is smaller or equal to the already evaporated volume. The contact angle can become at

maximum the static advancing contact angle Θa. In case d, the volume of the incoming droplet

is larger than the already evaporated volume, the contact line will have to expand to avoid larger

contact angles than the dynamic advancing contact angle Θda. During expansion the contact

angle will take a value between the static and dynamic advancing contact angles. Case d is

analog to the landing event and evaporation of a single droplet and is very suitable for

monitoring the contact angles.

Finally, the detailed analysis of such dynamic events allows the extraction of the static and

dynamic receding and advancing contact angle Θ depending on the velocity of the contact line,

which is visualized exemplary for one evaporating picoliter droplet at T = 90 °C in Figure 7.14.

Figure 7.14 Contact angle measured as a function of the contact line velocity for an evaporating picoliter

droplet at T = 90 °C.

This method uses for the dynamic contact angle Θd extraction the very short spreading phase

(contact line velocity >> 0) and a significantly longer evaporation phase with reducing visible

area (contact line velocity < 0). Thus the number of measured contact angles for the latter is

significantly larger. The static contact angles Θa and Θr are extracted when the contact line is


80

not in motion as maximum and minimum, respectively. In summary this contact angle Θ

extraction allows to fully record all dynamic and static contact angles an observed system can

take within a few milliseconds and could deliver a considerable quantity of statistical contact

angle data for various systems.

7.3 Thermodynamic boiling regimes, critical heat flux and Leidenfrost

temperature

Essential for the use of sessile droplets or droplets experiencing the Leidenfrost effect, the

knowledge of the thermodynamic properties of the used solvent, i.e., the boiling temperature

Tb, the temperature where critical heat flux (CHF) occurs, and the Leidenfrost temperature. The

latter two depend not only on the liquid properties itself but also on the used substrate. All of

them depend on the surrounding atmospheric conditions. They mark the temperature span of

the three thermodynamic boiling regimes, i.e., nucleate boiling, transition boiling, and film

boiling. By monitoring the evaporation of sessile droplets, the rate at which the visible area

decreases linearly as a measure of the evaporation rate allows determining the CHF and

Leidenfrost temperature and the three aforementioned thermodynamic boiling regimes.

Therefore in Figure 7.15, the linear area decrease rate in phase 3 is plotted against the

temperature range of 110 °C ≤ T ≤ 205 °C as a measure of the evaporation rate.


81

Figure 7.15 Visible area evaporation rate as a function of substrate temperature. Divided into three

regimes: Nucleate boiling (red), transition boiling (green) and film boiling (blue). Leidenfrost temperature

at T = 200°C. [reused with permission[129]]

Based on the average evaporation rate, the CHF can be recognized as maxima at T = 155°C

with a rate of ~ 1550 m2/ms. Below this temperature, the nucleate boiling regime is located

and above this temperature starts the transition boiling regime. The latter ends at the local

minimum at a temperature of T = 200 °C with a rate of ~ 1000 m2/ms, corresponding to the

LFP. Above this temperature, the evaporation rate increases, and sessile droplets are still

occasionally observed until the temperature exceeds 210 °C. The temperature range above the

LFP is assigned to the film boiling regime. The droplets show a strongly increased lifetime and

start to glide on their own gas phase. The observed temperatures are consistent with the

reported temperatures for water and silicon wafers in the literature.[132]

7.4 Novel meta-stable hovering state of very small droplets – Bypass to

the Leidenfrost effect

This chapter presents a novel hovering meta-stable droplet state that is not described yet by

the standard literature as shown in the previous chapters. Droplet evaporation was categorized

before at warm or hot surfaces by the boiling temperature, critical heat flux, and the Leidenfrost


82

point. Critical heat flux was described as the temperature where maximum heat flux from the

surface into the droplet occurs, and Leidenfrost point as the temperature where the droplets

are lifted on its evaporating vapor and begin to hover. Direct contact to the surface does not

exist from this temperature on anymore.

The Leidenfrost effect is only observable for temperatures higher than the boiling temperature

and the critical heat flux and simultaneously represents the minimum in the observed heat flux

into the droplet above the critical heat flux temperature. As extracted before (see Chapter 7.3),

the Leidenfrost temperature for water is found to be around 205 °C on a silicon wafer covered

with natural oxide.

It was also shown that the droplets become sessile below this temperature after making direct

contact with the surface. At such low temperatures, the material deposition by droplets

frequently suffers from the coffee stain effect or ring-like inhomogeneities due to material

deposition being restricted to the droplet landing site. So far this unwanted effects can only be

avoided at higher temperatures beyond the Leidenfrost effect where sessile droplets don't exist.

However, these higher temperatures are often not compatible with the surface material, process

requirements, the solvent used or the material being deposited.

The new meta-stable state allows small droplets to hover close to a warm surface if their

approach was slow enough. They evaporate slowly at the surface without forming a sessile

droplet and experience an increased lifetime. Within this state, the droplets avoid direct contact

with the surface. Thus material deposition from droplets in this meta-stable state does not suffer

from the coffee stain effect or layer inhomogeneities. Therefore, by utilizing the new meta-stable

hovering effect the Leidenfrost effect and the with it related temperature limitations are

bypassed.

This new discovered meta-stable state can not be described as a Leidenfrost effect anomaly

for small droplets at low temperatures. It clearly differs from the Leidenfrost situation in that

point, that if a direct contact of the droplet with the surface was made, a droplet below the

Leidenfrost temperature, i.e. in the meta-stable state, remains irreversibly in direct contact with

the surface and the meta-stable state is lost. In contrast, a Leidenfrost droplet is lifted again by

its own evaporated vapor and breaks direct contact with the surface again.

The meta-stable state requirements are that the droplet must be small enough. Gravitational,

inertial, or other driving forces towards the surface remain small enough compared to repelling

forces occurring during the warm or hot surface approach and the onset of accelerated droplet

evaporation. E.g. it was found that water droplets driven by gravity towards a surface at 60 °C

require a droplet diameter smaller than d = 40 m for the formation of the meta-stable state

(see Figure 7.16).


83

Figure 7.16 Camera side-view images of freely falling water droplets approaching a Si-wafer heated to

60 °C well below the Leidenfrost temperature. Due to light reflection at the substrate, the droplet is seen

twice. 1) a larger droplet normally lands with direct contact to the surface and forms a static sessile droplet.

2) a smaller droplet approaches the surface but enters the hovering meta-stable state above the surface

and remains mobile for lateral movements. [133]

The meta-stable state formation is demonstrated by observing first a normal landing of a larger

droplet and second a sufficiently small droplet with a slower approach entering the hovering

meta-stable state. All images show the water droplet twice, i.e. directly and via the substrate

reflection, to monitor the distance between the droplet and the surface accurately.

The first droplet makes contact with the surface. Afterward, it evaporates as a sessile droplet.

The droplet and its reflection image appear as one object. In contrast, the second droplet does

not directly contact the surface and remains in the novel hovering meta-stable droplet state.

The droplet and its reflection image remain as two distinct objects.

From the image series, the height of the approaching droplet can be extracted quantitatively

over time. For the second case of the novel hovering meta-stable state, this approach is shown

in Figure 7.17.


84

Figure 7.17 Monitored height (m) of the center of the meta-stable hovering droplet showed before and

the observed height of its reflection relative to the surface at 0 m. The droplet does not come into contact

with the surface at any time.[133]

To utilize this meta-stable state it is important to investigate whether material can be deposited

on the surface in the same quality as it can be done exploiting the Leidenfrost effect. Therefore,

magnesium acetate tetrahydrate was dissolved in DI-water (0.05 M) and atomized by an

ultrasonic atomization nozzle with the same parameters used for previous experiments.

Generated droplets fell driven by gravity towards a silicon wafer that was held at a surface

temperature Ts = 150 °C. This temperature is significantly lower than the in the previous chapter

determined Leidenfrost temperature for this system (water/silicon wafer), which is at

T = 205 °C. The deposition lasted for 2 min with a constant precursor solution feed rate at

50 l/min. The result of the deposition was analyzed by SEM (scanning electron microscopy).

The residues of two different deposition modes were found by analysis of the material contrast

and shown in Figure 7.18.


85

Figure 7.18 SEM images of a) deposited coffee stain structure by conventional sessile evaporating droplet

and b) deposited defect-free layer utilizing droplets with the novel hovering meta-stable state. [133]

For droplets that exceed the minimum required size a coffee stain is visible (see Figure 7.18

a)). It results from a droplet that evaporated sessile at the surface and leaves the material as

an inhomogenoues structure that is not applicable for producing homogenous smooth layers.

If the deposition occurs with sufficient small droplets the intended smooth featureless layer is

produced and is comparable to layers deposited at high temperatures where the Leidenfrost

effect occurs. The existence of a deposited homogeneous layer was verified by a simple

scratching experiment.

Thus, the novel meta-stable state allows the deposition of homogenous layers at low

temperatures. This can possibly open up a new field of applications with materials that cannot

withstand high temperatures. The previous mentioned maximum droplet diameter was so far

observed as d = 40 m for a surface temperature of 60 °C.

7.5 Experimental setup and video acquisition

Video acquisition was done using a Teledyne Dalsa Genie Nano camera with an ONsemi

Python 300 CMOS monochromatic sensor (640 x 480 pixel, pixel size = 4.8 x 4.8 m). If not

mentioned otherwise, a video clip consists of 5000 frames. A recorded frame has a resolution

of 256 x 256 pixels. The side length of a pixel is between 1.47 to 3.00 m depending on the

optical magnification. The pixels are recorded in monochromatic 8-bit color dept and thus take

values between 0 and 255. The exposure time of the pixels was set to 400 s, while the

illumination pulses and frame rate were synchronized and triggered externally by a Keithley


86

K3390 function generator at 1 or 2 kHz. The real illumination time of a pixel is due to pulsed

operation way shorter and accumulated to 10 s while the illuminating LED is turned on. A

Luxeon Rebel LED (910 mW, royal blue, 447.5 nm, LXML-PR01-0500) was used as a light

source and pulsed by a homemade LED driver. The pulse intervals were set to 125 s for

quadruple illuminated videos and 500 s for single illuminated videos. The recording starts with

a delay after atomization is turned on to allow the stream to stabilize. The optical system has a

theoretical optical resolution of 4 m. A schematic representation of the beginning of a

recording for a video with quadruple pulse per frame is shown in Figure 7.19.

Figure 7.19 Schematic representation of the recording sequence. The spray nozzle is stabilized in a run-

up time of 5 s. During the recording of a frame, the LED is pulsed four times with a duration of 2.5 s with

a period of 125 s. The time per frame is 500 s. [reused with permission[129]]

As ultrasonic atomizer a Sonaer 130 kHz ultrasonic atomization nozzle (model: NS130K) was

used and was operated with a power of 1.0 - 3.0 Watt to atomize the solvents. The atomization

platform has a diameter of 3.8 mm. The fine droplets are accelerated by gravity towards the

substrate. DI-water (18.3 MΩ) is used as the solvent and pumped with a rate of 0.6 ml/min into

the nozzle. 5 x 5 cm2 in lateral dimension and 375 m thick silicon with natural oxide at the

surface was used as a substrate and held at the target temperature by a thermocouple-

controlled hotplate. The surface roughness is Sa = 130.6 pm and Sq = 164.4 pm for a 5 x 5 m

area as measured by a Veeco Picoforce AFM (see Figure 7.20).


87

Figure 7.20 AFM topography image of silicon with natural oxide.

The atomization starts after the heating plate's temperature is stable and deviates maximum

± 0.5 °C from the target temperature. The volume mean diameter (VMD) was according to the

manufacturer (Sonaer Inc) simulated by the controller to be VMD = 20 – 25 m.

For image data evaluation the 8-bit single channel images were resized with a bicubic

interpolation over a 4x4 pixel neighborhood for droplet detection. Subsequently, an adaptive

threshold was used for the binarization of the images. This threshold is composed of the sum

of the mean pixel intensity in a suitable pixel neighborhood size and a suitable offset, depending

on the illumination and droplet density on the respective frame. Image processing was done

utilizing the OpenCV library (v. 4.4.0.46)[134].

7.6 Conclusion and outlook

This chapter described the cross-section analysis of picoliter droplets under real application

conditions. Beginning with the droplet diameter distribution and the extraction of droplet

properties like diameter and area by cross-section analysis.

It was shown that the droplet appearance close to surfaces (distance ≤ 100 m) could be

described well by raytracing, which has been used to confirm the calculated properties of the

droplets. Monitoring of these droplets close to the surface revealed that sessile droplets below

Leidenfrost point and moving droplets above Leidenfrost could be clearly identified and used

as a indicator for the boiling regime.

The knowledge of the volume of the droplet before becoming sessile allows the straightforward

calculation of the contact angle directly after landing. With an evaporation model the contact


88

angle could be obtained at any time in the sessile stage. The advancing and receding contact

angles could then be extracted as the minimum or maximum of the calculated contact angles

for the contact line velocity of 0 (static) and the contact line in motion (dynamic) for the

evaporation of water on a silicon wafer at T = 110 °C.

Analysis of sessile droplet evaporation allows determining evaporation properties for almost

any solvent/substrate system precisely and quickly. The thermodynamic regimes nucleate

boiling, transition boiling, and film boiling regimes are determined as well as the critical heat

flux and the Leidenfrost point, by analyzing the recorded visible droplet area over time.

The observation of many very small droplets shows which non-trivial processes take place on

the substrate surface. It was shown that the formation of sessile droplets depends not only on

the volume flow arriving at the substrate surface but also on the local droplet density. At higher

droplet densities, several droplets can combine to form sessile droplets with larger volumes and

longer lifetime.

For the observed system, it was found that very small droplets can experience an Leidenfrost-

like phenomenon far below their actual Leidenfrost temperature and levitate at surface

temperatures that are even 150 °C lower than the Leidenfrost point.

In depth understanding of sessile and hovering droplet evaporation can be especially helpful in

understanding the deposition of functional layers, where an understanding of the evaporation

kinetics is crucial. Microscopically, droplets can have significantly longer lifetimes than is

macroscopically assumed from the incoming volume flow and the energy available for

evaporation over time.

At higher droplet densities, the interaction of evaporating droplets with their environment can

be observed. Picoliter droplet evaporation can create turbulence at the edge of the droplet,

accelerating droplets near the surface and repels them from the evaporation center, attracting

higher droplets by a suction. The latter favors that an evaporating droplet increases its volume

by suctioning other droplets and extending its lifetime. The evaporation of such multi-droplet

events can be evaluated analog to single sessile droplet evaporation.

The new method of top-view monitoring of very small droplets opens up new possibilities and

fields for further and deeper research. The recording of contact angles versus the contact line

movement should be done with temperature as a further dimension and compared to results

from macroscopically methods. Also the measurement of various combinations of liquids and

substrates should give more insight into the physics behind the evaporation of very small

droplets. The method itself would benefit from video recordings at higher framerates and better

resolutions.


89

The tracking of droplets and the data evaluation can highly benefit from automated evaluation

by advanced deterministic algorithms, machine learning or a combination of both. Either can

be assisted by raytracing.

Sophisticated research can be done on the new meta-stable droplet state in terms of physics

behind this effect and maximum droplet size. Gaining control over the droplet movement can

open up possibilities like lab on the chip, mini reactors or 3D deposition techniques. Other

exciting applications might be possible as well.

Concluding remarks

90

8 Concluding remarks

Chapter one began with a general introduction and named the core topics of the thesis. The

focus was on investigating the following topics:

1. Improved stability and performance of solution-processed zinc oxide thin-film transistors

processed at low temperatures by organic surface defect passivation:

2. Experimental evaluation of possible novel zinc oxide precursors

3. Gain deeper understanding of picoliter droplets at hot surfaces under real application

conditions

Chapters two, three, and four provided general information about spray pyrolysis, metal oxide

thin-film transistors and basics of applied theoretical and experimental methods in this thesis.

The chapter Tailored organic surface passivation for metal oxide semiconductors

(chapter 5) discussed investigations concerning the first core topic. Tailored diketones were

used to passivate defect states at the surface of solution-processed zinc oxide thin film

transistors. Trifluoroacetylacetophenone was tailored for this purpose by the introduction of

electron-withdrawing or electron-donating substituents at the phenyl ring. After quantifying the

relative change of electron density by DFT, the general bonding of the molecules was examined

by NMR experiments. NMR also narrowed down the possible binding situations to one solution,

where zinc binds to an oxygen of the enol replacing water or hydroxide at the surface. The

passivation on a zinc oxide thin-film was achieved by diketone deposition in a vacuum. The

presence and morphology of the final passivating layer were confirmed by XPS and analyzed

by NMR. Finally, the electrical characterization of the passivated zinc oxide active layer

revealed that all diketones improve the transistor stability for positive and negative bias stress,

in general. Also, the passivated transistors showed improved electrical properties with

increased electron mobility and decreased disorder parameter. The threshold voltages shift

systematically to negative values and the hysteresis clearly reduces. The improved electrical

properties correlate with the effective electron density at the anchoring oxygen atoms of the

passivation molecules. The highest negative charge, which was found for methyl substitution

at the phenyl ring, resulted in the best performance.

These findings show that specifically tailored passivation molecules stabilize solution-

processed thin-film transistors. Further investigations on this topic could lead to different

organic molecules that can further manipulate the electrical properties of thin-film transistors by

surface passivation.

Concluding remarks

91

The chapter Fluorinated Carboxylates as zinc oxide precursor (chapter 6) showed that the

fluorination of zinc carboxylates leads to suitable precursors for zinc oxide productions. All

precursors show the characteristic absorption band of zinc oxide after thermal decomposition.

FTIR identified impurities for some precursors, which most likely originate from the annealing

procedure. TGA measurements revealed that especially heptafluoro zinc propionate might be

a beneficial precursor, which decomposes at a lower temperature at 250 °C. It can be

potentially processed at significantly lower temperatures compared to zinc acetate, which

decomposes at 360 °C. However, these precursors can not be dissolved in water or other

environment-friendly solvents and were therefore not used for active layer deposition.

Further investigations and the application as precursors are possible after the identification of

a suitable solvent for deposition. The search for such solvent can be assisted by the findings of

chapter 7, which allows the quick determination of thermodynamic properties of a solvent under

deposition conditions.

The chapter High-speed real-time analysis of picoliter droplets under spray pyrolysis

conditions (chapter 7) discussed the investigations on the third core topic. For that purpose,

high-speed images of free-falling picoliter and subpicoliter water droplets generated by an

ultrasonic atomizer were recorded. The physically meaningful extraction of properties of these

droplets was done by cross-section analysis of the droplet borderline and compared to the

expected appearance of the droplets obtained from raytracing. For the analysis of such a small

droplet falling onto a heated silicon substrate, an almost perfect spherical shape was assumed

due to the dominating surface tension. Because of this, a spherical cap was also assumed for

sessile drops. With these constraints, a new model was presented that allowed the extraction

of the contact angle Θ directly after landing and the calculation of static and dynamic advancing

and receding contact angles. The recording yield the velocity of the contact line at any time of

observation. In addition, thermodynamic properties like critical heat flux (CHF) and Leidenfrost

point (LFP) were determined by the extracted evaporation rate. The evaporation regimes:

nucleate boiling, transition boiling and film-boiling can be clearly defined by CHF and LFP.

Furthermore, different complex scenarios of interactions between droplets were observed for

sessile droplets formation and during the evaporation process.

Finally, a novel meta-stable hovering state of very small droplets at ambient temperatures was

discovered, which clearly differs from the Leidenfrost effect. Droplets with a sufficient small

droplet diameter hover close to the surface and evaporate without direct contact.

These hovering droplets allow drastically reduced process temperatures for the processing of

solution-processed thin-film transistors. Also, manipulating droplets at the surface or complex

scenarios like a lab on the chip are possible future scenarios.

Concluding remarks

92

It is also very interesting to monitor the contact angle by the top-view analysis versus

temperature and especially above the individual boiling point for further investigations. The

short lifetime of the droplets and the possibility of simultaneously monitoring a large number of

droplets allow a statistically meaningful determination of dynamic and static advancing and

receding contact angles.

The top-view analysis also facilitates the adaption of new spray pyrolysis systems

(solvent/substrate combinations) because the determination of necessary thermodynamic

properties and beneficial deposition parameters can be easily measured under deposition

conditions.

List of publications

93


8.1 Journal paper

Köhling, J., Kalinovich, N.,Pajkert, R., Lork, E., Wagner, V., Röschenthaler, G.-V., Oxamates

as 1,2-Diketone Equivalents: The Effect of Fluorine, ChemistrySelect, 2021, 6(8), 1882-1886

Köhling, J., Wagner, V., High speed picoliter droplet top-view analysis for advancing and

receding contact angles, boiling regimes and droplet-droplet interaction. Int. J. Heat Mass

Transf. 2021, 169, 120939.

Köhling, J., Jovanov, V., Kalinovich, N., Röschenthaler, G.-V., Wagner, V., Tailored β-

diketones as effective surface passivation for solution processed zinc oxide thin film transistors.

Org. Electron. 2020, 105906.

Köhling, J., Kozel, V., Jovanov, V., Pajkert, R., Tverdomed, S. N., Gridenco, O., Fugel, M.,

Grabowsky, S., Röschenthaler, G.-V., Wagner, V. Synthesis and Characterization of

Oxazaborinin Phosphonate for Blue OLED Emitter Applications ChemPhysChem, 20(5), 665–

671, 2019.

Huth, M., Chen, C.-W., Köhling, J., Taffa, D. H., Wark, M., Wagner, V. Prediction of

delamination state of 2D filler materials in cyclic olefin copolymer for enhanced barrier

applications. Compos. Struct., 202, 853–859, 2018.

Huth, M., Chen, C.-W., Köhling, J., Wagner, V. Influence of Hansen solubility parameters on

exfoliation of organophilic fluoromica. Appl. Clay Sci., 161, 412–418, 2018.

8.2 Patent applications

Wagner, V., Köhling, J., Novel hovering droplet state at warm and hot surfaces for small

droplets, 2020, DE 10 2020 133 536.9

Köhling, J., Merkulov, A., Brendt, J., Direkt-strukturierbare Formulierungen auf der Basis von

Metalloxid-Prekursoren zur Herstellung oxidischer Schichten, WO/2018/145907, 2018.


94

8.3 Conference contributions

Jonas Köhling and Veit Wagner, Poster: Influence of long chain carboxylates as precursor on

the performance of ZnO TFTs, HL 64.8, Frühjahrestagung der Deutsche Physikalische

Gesellschaft (DPG) Dresden, Germany, 2017

Cristian Telescu, Jonas Köhling and Veit Wagner, Poster: Influence of the pH value of the

precursor solution on ZnO TFTs, HL 64.7, Frühjahrestagung der Deutsche Physikalische

Gesellschaft (DPG), Dresden, Germany, 2017

Torsten Balster, Jonas Köhling, Marlis Ortel, and Veit Wagner, Oral presentation: Polymer

passivated zinc oxide thin film transistors, HL 38.3, Frühjahrestagung der Deutsche

Physikalische Gesellschaft (DPG), Dresden, Germany, 2017

Talha Nisar, Torsten Balster, Jonas Köhling, and Veit Wagner, Oral presentation:

Characterization of electrochemically deposited MoSx layers for thin film transistors, HL 86.4,

Frühjahrestagung der Deutsche Physikalische Gesellschaft (DPG), Dresden, Germany, 2017

Jonas Köhling, Marlis Ortel, Nataliya Kalinovich, Gerd-Volker Röschenthaler and Veit Wagner,

Poster: ß-diketone passivated solution processed zinc oxide nano-layers for application in thin

film transistors, International Meeting on Information Display (IMID), Seoul, Korea, 2017.

Jonas Köhling, Marlis Ortel, Nataliya Kalinovich, Gerd-Volker Röschenthaler and Veit Wagner,

Oral presentation: Effect of β-diketone passivation on solution processed zinc oxide nano-

layers used in thin film transistors, ASS I T5, 16th International Conference on the Formation

of Semiconductor Interfaces (ICFSI), Hannover, Germany, 2017.

Jonas Köhling, Nataliya Kalinovich, Gerd-Volker Röschenthaler and Veit Wagner, Oral

presentation: DFT assisted tailoring of fluorine-containing molecules for passivation of zinc

oxide layers in thin film transistors, HL 40.2, Frühjahrestagung der Deutsche Physikalische

Gesellschaft (DPG), Berlin, Germany, 2018

Michael Huth, Jonas Köhling, and Veit Wagner, Oral presentation: Influence of Hansen

solubility parameters on a shear exfoliation process of organophilic layered silica in chloroform,

DS 12.12, Frühjahrestagung der Deutsche Physikalische Gesellschaft (DPG), Berlin, Germany,

2018

Jonas Köhling, Nataliya Kalinovich, Gerd-Volker Röschenthaler and Veit Wagner, Oral

presentation: Binding mechanism of fluorine-containing ketones on zinc oxide surfaces for thin

film transistor passivation, DS 3.8, Frühjahrestagung der Deutsche Physikalische Gesellschaft

(DPG), Regensburg, Germany, 2019


95

Juliana Nair, Jonas Köhling, Gerd-Volker Röschenthaler and Veit Wagner, Poster: Molecular

design of 𝜆6-phosphorous compounds for OLED applications, CPP 37.22, Frühjahrestagung

der Deutsche Physikalische Gesellschaft (DPG), Regensburg, Germany, 2019

Jonas Köhling, Gerd-Volker Röschenthaler and Veit Wagner, Oral presentation: Molecular

design strategy of λ5σ6-phosphorous compounds for OLED applications, MT03.13.10, Materials

Research Society Fall Meeting (MRS), Boston, USA, 2019

Jonas Köhling, Kavish Tyagi and Veit Wagner, Oral presentation: High speed real time single

droplet analysis to improve spray pyrolysis deposition process, FF04.03.04, Materials Research

Society Fall Meeting (MRS), Boston, USA, 2019

Jonas Köhling, Gerd-Volker Röschenthaler, and Veit Wagner, Oral presentation: Substituent

approach in molecular design of phosphorous compounds for OLED emitters, CPP 3.11,

Frühjahrestagung der Deutsche Physikalische Gesellschaft (DPG), Dresden, Germany,

2020 – Canceled due to Corona crisis

Jonas Köhling and Veit Wagner, Oral presentation: Real time video analysis of droplets in

spray pyrolysis deposition process, DS 13.2, Frühjahrestagung der Deutsche Physikalische

Gesellschaft (DPG), Dresden, Germany, 2020 – Canceled due to Corona crisis

References

96

References

[1] Noyce, R. N., Semiconductor Device-and-Lead Structure, publ. date 1961.

[2] Fortunato, E., Barquinha, P., Martins, R., Oxide Semiconductor Thin-Film Transistors: A Review of Recent Advances. Adv. Mater. 2012, 24, 2945–2986.

[3] Kuo, Y. (Ed.), Thin Film Transistors. Springer US, Boston, MA 2004.

[4] Wager, J. F., Yeh, B., Hoffman, R. L., Keszler, D. A., An amorphous oxide semiconductor thin-film transistor route to oxide electronics. Curr. Opin. Solid State Mater. Sci. 2014, 18, 53–61.

[5] Kamiya, T., Nomura, K., Hosono, H., Present status of amorphous In–Ga–Zn–O thin-film transistors. Sci. Technol. Adv. Mater. 2010, 11, 44305.

[6] Rizk, N. K., Lefebvre, A. H., Airblast Atomization: Studies on Drop-Size Distribution. J. Energy 1982, 6, 323–327.

[7] Ortel, M., Trostyanskaya, Y. S., Wagner, V., Spray pyrolysis of ZnO–TFTs utilizing a perfume atomizer. Solid. State. Electron. 2013, 86, 22–26.

[8] Ortel, M., Pittner, S., Wagner, V., Stability and spacial trap state distribution of solution processed ZnO-thin film transistors. J. Appl. Phys. 2013, 113, 154502.

[9] Köhling, J., Jovanov, V., Kalinovich, N., Röschenthaler, G.-V., Wagner, V., Tailored β-diketones as effective surface passivation for solution processed zinc oxide thin film transistors. Org. Electron. 2020, 86, 105906.

[10] Isakov, I., Faber, H., Grell, M., Wyatt-Moon, G., Pliatsikas, N., Kehagias, T., Dimitrakopulos, G. P., Patsalas, P. P., Li, R., Anthopoulos, T. D., Exploring the Leidenfrost Effect for the Deposition of High-Quality In2O3 Layers via Spray Pyrolysis at Low Temperatures and Their Application in High Electron Mobility Transistors. Adv. Funct. Mater. 2017, 27, 1606407.

[11] Panão, M. R. O., Moreira, A. L. N., Experiments in Fluids. Springer 2005, pp. 364–374.

[12] Pan, H., Xiao, D., Hung, D., Xu, M., Li, X., Experimental investigations of wall jet droplet impact on spray impingement fuel film formation. Fuel 2019, 241, 33–41.

[13] Oliveira Panão, M. R., Moreira, A. L. N., Durão, D. F. G., Effect of a cross-flow on spray impingement with port fuel injection systems for HCCI engines. Fuel 2013, 106, 249–257.

[14] Tang, C., Qin, M., Weng, X., Zhang, X., Zhang, P., Li, J., Huang, Z., Dynamics of droplet impact on solid surface with different roughness. Int. J. Multiph. Flow 2017, 96, 56–69.

[15] Shen, S., Liang, G., Guo, Y., Liu, R., Mu, X., Heat transfer performance and bundle-depth effect in horizontal-tube falling film evaporators. Desalin. Water Treat. 2013, 51, 830–836.

[16] Xu, Q., Li, Z., Wang, J., Wang, R., Characteristics of Single Droplet Impact on Cold Plate Surfaces. Dry. Technol. 2012, 30, 1756–1762.

References

97

[17] Chunqiang, S., Shuangquan, S., Changqing, T., Hongbo, X., Development and experimental investigation of a novel spray cooling system integrated in refrigeration circuit. Appl. Therm. Eng. 2012, 33–34, 246–252.

[18] Ding, B., Wang, H., Zhu, X., Chen, R., Liao, Q., Water droplet impact on superhydrophobic surfaces with various inclinations and supercooling degrees. Int. J. Heat Mass Transf. 2019, 138, 844–851.

[19] Visaria, M., Mudawar, I., 2008 11th IEEE Intersociety Conference on Thermal and Thermomechanical Phenomena in Electronic Systems, I-THERM. 2008, pp. 275–283.

[20] Tenzer, F. M., Roisman, I. V., Tropea, C., Fast transient spray cooling of a hot thick target. J. Fluid Mech. 2019, 881, 84–103.

[21] Breitenbach, J., Roisman, I. V., Tropea, C., From drop impact physics to spray cooling models: a critical review. Exp. Fluids 2018, 59, 1–21.

[22] Mehdizadeh, N. Z., Chandra, S., Boiling during high-velocity impact of water droplets on a hot stainless steel surface. Proc. R. Soc. A Math. Phys. Eng. Sci. 2006, 462, 3115–3131.

[23] Baumeister, K. J., Simon, F. F., Leidenfrost Temperature—Its Correlation for Liquid Metals, Cryogens, Hydrocarbons, and Water. J. Heat Transfer 1973, 95, 166–173.

[24] Kim, H., Truong, B., Buongiorno, J., Hu, L.-W., On the effect of surface roughness height, wettability, and nanoporosity on Leidenfrost phenomena. Appl. Phys. Lett. 2011, 98, 83121.

[25] Bernardin, J. D., Stebbins, C. J., Mudawar, I., Effects of surface roughness on water droplet impact history and heat transfer regimes. Int. J. Heat Mass Transf. 1996, 40, 73–88.

[26] Liang, G., Mudawar, I., Review of drop impact on heated walls. Int. J. Heat Mass Transf. 2017, 106, 103–126.

[27] Norouzi, M., Abdolnezhad, H., Mandani, S., An experimental investigation on inertia motion and deformation of Boger drops falling through Newtonian media. Meccanica 2019, 54, 473–490.

[28] Ohnesorge, W. V., Die Bildung von Tropfen an Düsen und die Auflösung flüssiger Strahlen. ZAMM - J. Appl. Math. Mech. / Zeitschrift für Angew. Math. und Mech. 1936, 16, 355–358.

[29] Brutin, D., Starov, V., Recent advances in droplet wetting and evaporation. Chem. Soc. Rev. 2018, 47, 558–585.

[30] Liang, G., Mudawar, I., Review of drop impact on heated walls. Int. J. Heat Mass Transf. 2017, 106, 103–126.

[31] Bonn, D., Eggers, J., Indekeu, J., Meunier, J., Rolley, E., Wetting and spreading. Rev. Mod. Phys. 2009, 81, 739–805.

[32] Bernardin, J. D., Mudawar, I., Walsh, C. B., Franses, E. I., Contact angle temperature dependence for water droplets on practical aluminum surfaces. Int. J. Heat Mass Transf. 1997, 40, 1017–1033.

References

98

[33] Park, J. K., Ryu, J., Koo, B. C., Lee, S., Kang, K. H., How the change of contact angle occurs for an evaporating droplet: effect of impurity and attached water films. Soft Matter 2012, 8, 11889–11896.

[34] Cossali, G. E., Marengo, M., Santini, M., Thermally induced secondary drop atomisation by single drop impact onto heated surfaces. Int. J. Heat Fluid Flow 2008, 29, 167–177.

[35] Khavari, M., Sun, C., Lohse, D., Tran, T., Fingering patterns during droplet impact on heated surfaces. Soft Matter 2015, 11, 3298–3303.

[36] Armstrong, S., McHale, G., Ledesma-Aguilar, R., Wells, G. G., Pinning-Free Evaporation of Sessile Droplets of Water from Solid Surfaces. Langmuir 2019, 35, 2989–2996.

[37] Furmidge, C. G. L., Studies at phase interfaces. I. The sliding of liquid drops on solid surfaces and a theory for spray retention. J. Colloid Sci. 1962, 17, 309–324.

[38] Wilhelmy, L., Ueber die Abhängigkeit der Capillaritäts-Constanten des Alkohols von Substanz und Gestalt des benetzten festen Körpers. Ann. Phys. 1863, 195, 177–217.

[39] Rein, M. (Ed.), Drop-Surface Interactions. Springer Vienna, Vienna 2002.

[40] Heim, L.-O., Bonaccurso, E., Measurement of Line Tension on Droplets in the Submicrometer Range. Langmuir 2013, 29, 14147–14153.

[41] Checco, A., Guenoun, P., Daillant, J., Nonlinear Dependence of the Contact Angle of Nanodroplets on Contact Line Curvature. Phys. Rev. Lett. 2003, 91, 186101.

[42] Checco, A., Cai, Y., Gang, O., Ocko, B. M., High resolution non-contact AFM imaging of liquids condensed onto chemically nanopatterned surfaces. Ultramicroscopy 2006, 106, 703–708.

[43] Wang, R., Cong, L., Kido, M., Evaluation of the wettability of metal surfaces by micro-pure water by means of atomic force microscopy. Appl. Surf. Sci. 2002, 191, 74–84.

[44] Wang, R., Takeda, M., Kido, M., Micro pure water wettability evaluation with an AC no-contact mode of atomic force microscope. Mater. Lett. 2002, 54, 140–144.

[45] Connell, S. D. A., Allen, S., Roberts, C. J., Davies, J., Davies, M. C., Tendler, S. J. B., Williams, P. M., Investigating the Interfacial Properties of Single-Liquid Nanodroplets by Atomic Force Microscopy. Langmuir 2002, 18, 1719–1728.

[46] Stelmashenko, N. A., Craven, J. P., Donald, A. M., Terentjev, E. M., Thiel, B. L., Topographic contrast of partially wetting water droplets in environmental scanning electron microscopy. J. Microsc. 2001, 204, 172–183.

[47] Lau, K. K. S., Bico, J., Teo, K. B. K., Chhowalla, M., Amaratunga, G. A. J., Milne, W. I., McKinley, G. H., Gleason, K. K., Superhydrophobic Carbon Nanotube Forests. Nano Lett. 2003, 3, 1701–1705.

[48] Jung, Y. C., Bhushan, B., Wetting behaviour during evaporation and condensation of water microdroplets on superhydrophobic patterned surfaces. J. Microsc. 2008, 229, 127–140.

[49] Isa, L., Lucas, F., Wepf, R., Reimhult, E., Measuring single-nanoparticle wetting properties by freeze-fracture shadow-casting cryo-scanning electron microscopy. Nat. Commun. 2011, 2, 438.

References

99

[50] Rykaczewski, K., Landin, T., Walker, M. L., Scott, J. H. J., Varanasi, K. K., Direct Imaging of Complex Nano- to Microscale Interfaces Involving Solid, Liquid, and Gas Phases. ACS Nano 2012, 6, 9326–9334.

[51] Ensikat, H. J., Schulte, A. J., Koch, K., Barthlott, W., Droplets on Superhydrophobic Surfaces: Visualization of the Contact Area by Cryo-Scanning Electron Microscopy. Langmuir 2009, 25, 13077–13083.

[52] Sundberg, M., Månsson, A., Tågerud, S., Contact angle measurements by confocal microscopy for non-destructive microscale surface characterization. J. Colloid Interface Sci. 2007, 313, 454–460.

[53] Lim, T., Jeong, J., Chung, J., Chung, J. T., Evaporation of inkjet printed pico-liter droplet on heated substrates with different thermal conductivity. J. Mech. Sci. Technol. 2009, 23, 1788–1794.

[54] Talbot, E., Berson, A., Brown, P., Bain, C., Evaporation of Picolitre Droplets on Surfaces with a Range of Wettabilities and Thermal Conductivities. Phys. Rev. E 2012, 85, DOI: 10.1103/PhysRevE.85.061604.

[55] Saverchenko, V. I., Fisenko, S. P., Khodyko, Y. A., Evaporation of a picoliter droplet on a wetted substrate at reduced pressure. J. Eng. Phys. Thermophys. 2011, 84, 723.

[56] Atencia, J., Beebe, D. J., Controlled microfluidic interfaces. Nature 2005, 437, 648–655.

[57] Dittrich, P. S., Manz, A., Lab-on-a-chip: microfluidics in drug discovery. Nat. Rev. Drug Discov. 2006, 5, 210–218.

[58] Galliker, P., Schneider, J., Eghlidi, H., Kress, S., Sandoghdar, V., Poulikakos, D., Direct printing of nanostructures by electrostatic autofocussing of ink nanodroplets. Nat. Commun. 2012, 3, 890.

[59] Antonini, C., Bernagozzi, I., Jung, S., Poulikakos, D., Marengo, M., Water Drops Dancing on Ice: How Sublimation Leads to Drop Rebound. Phys. Rev. Lett. 2013, 111, 14501.

[60] Mathew, S., Vijayakumar, K. P., Ellipsometric studies of microscopic surface roughness of CdS thin films. Bull. Mater. Sci. 1994, 17, 421–427.

[61] Pan, Z., Wang, H., Symmetry-to-asymmetry transition of Marangoni flow at a convex volatizing meniscus. Microfluid. Nanofluidics 2010, 9, 657–669.

[62] Ortel, M., Wagner, V., Leidenfrost temperature related CVD-like growth mechanism in ZnO-TFTs deposited by pulsed spray pyrolysis. J. Cryst. Growth 2013, 363, 185–189.

[63] Bardeen, J., Brattain, W. H., The Transistor, A Semi-Conductor Triode. Phys. Rev. 1948, 74, 230–231.

[64] Shockley, W., Semiconductor Amplifier, US 2502488 A, publ. date 1950.

[65] Bardeen, J., Brattain, W. H., Three-Electrode Circuit Element Utilizing Semiconductive Materials, US 2524035 A, publ. date 1950.

[66] Atalla, M. M., Semiconductor Devices Having Dielectric Coatings, US 3206670 A, publ. date 1965.

[67] Sze, S. M., Ng, K. K., MOSFETs. Phys. Semicond. Devices 2006, 293–373.

References

100

[68] Vissenberg, M. C. J. M., Matters, M., Theory of the field-effect mobility in amorphous organic transistors. Phys. Rev. B 1998, 57, 12964–12967.

[69] Gburek, B., Wagner, V., Influence of the semiconductor thickness on the charge carrier mobility in P3HT organic field-effect transistors in top-gate architecture on flexible substrates. Org. Electron. 2010, 11, 814–819.

[70] Hohenberg, P., Kohn, W., Inhomogeneous Electron Gas. Phys. Rev. 1964, 136, B864–B871.

[71] Kohn, W., Sham, L. J., Self-Consistent Equations Including Exchange and Correlation Effects. Phys. Rev. 1965, 140, A1133–A1138.

[72] Becke, A. D., Density-functional exchange-energy approximation with correct asymptotic behavior. Phys. Rev. A 1988, 38, 3098–3100.

[73] Lee, C., Yang, W., Parr, R. G., Development of the Colle-Salvetti correlation-energy formula into a functional of the electron density. Phys. Rev. B 1988, 37, 785–789.

[74] Vosko, S. H., Wilk, L., Nusair, M., Accurate spin-dependent electron liquid correlation energies for local spin density calculations: a critical analysis. Can. J. Phys. 1980, 58, 1200–1211.

[75] Stephens, P. J., Devlin, F. J., Chabalowski, C. F., Frisch, M. J., Ab Initio Calculation of Vibrational Absorption and Circular Dichroism Spectra Using Density Functional Force Fields. J. Phys. Chem. 1994, 98, 11623–11627.

[76] Becke, A. D., Density‐functional thermochemistry. III. The role of exact exchange. J. Chem. Phys. 1993, 98, 5648–5652.

[77] Hariharan, P. C., Pople, J. A., The influence of polarization functions on molecular orbital hydrogenation energies. Theor. Chim. Acta 1973, 28, 213–222.

[78] Krishnan, R., Binkley, J. S., Seeger, R., Pople, J. A., Self‐consistent molecular orbital methods. XX. A basis set for correlated wave functions. J. Chem. Phys. 1980, 72, 650–654.

[79] Rassolov, V. A., Ratner, M. A., Pople, J. A., Redfern, P. C., Curtiss, L. A., 6-31G* basis set for third-row atoms. J. Comput. Chem. 2001, 22, 976–984.

[80] Clark, T., Chandrasekhar, J., Spitznagel, G. W., Schleyer, P. V. R., Efficient diffuse function-augmented basis sets for anion calculations. III. The 3-21+G basis set for first-row elements, Li-F. J. Comput. Chem. 1983, 4, 294–301.

[81] London, F., Théorie quantique des courants interatomiques dans les combinaisons aromatiques. J. Phys. le Radium 1937, 8, 397–409.

[82] McWeeny, R., Perturbation Theory for the Fock-Dirac Density Matrix. Phys. Rev. 1962, 126, 1028–1034.

[83] Ditchfield, R., Self-consistent perturbation theory of diamagnetism. Mol. Phys. 1974, 27, 789–807.

[84] Wolinski, K., Hinton, J. F., Pulay, P., Efficient implementation of the gauge-independent atomic orbital method for NMR chemical shift calculations. J. Am. Chem. Soc. 1990, 112, 8251–8260.

References

101

[85] Cheeseman, J. R., Trucks, G. W., Keith, T. A., Frisch, M. J., A comparison of models for calculating nuclear magnetic resonance shielding tensors. J. Chem. Phys. 1996, 104, 5497–5509.

[86] Cancès, E., Mennucci, B., Tomasi, J., A new integral equation formalism for the polarizable continuum model: Theoretical background and applications to isotropic and anisotropic dielectrics. J. Chem. Phys. 1997, 107, 3032–3041.

[87] Koopmans, T., Über die Zuordnung von Wellenfunktionen und Eigenwerten zu den Einzelnen Elektronen Eines Atoms. Physica 1934, 1, 104–113.

[88] Tardio, S., Cumpson, P. J., Practical estimation of XPS binding energies using widely available quantum chemistry software. Surf. Interface Anal. 2018, 50, 5–12.

[89] Cioslowski, J., A new population analysis based on atomic polar tensors. J. Am. Chem. Soc. 1989, 111, 8333–8336.

[90] Mulliken, R. S., Electronic Population Analysis on LCAO–MO Molecular Wave Functions. I. J. Chem. Phys. 1955, 23, 1833–1840.

[91] Hirshfeld, F. L., Bonded-atom fragments for describing molecular charge densities. Theor. Chim. Acta 1977, 44, 129–138.

[92] Jalili, N., Laxminarayana, K., A review of atomic force microscopy imaging systems: application to molecular metrology and biological sciences. Mechatronics 2004, 14, 907–945.

[93] Hüfner, S., Photoelectron Spectroscopy. Springer Berlin Heidelberg, Berlin, Heidelberg 1995.

[94] Hesse, M., Meier, H., Zeeh, B. (Eds.), Spektroskopische Methoden in Der Organischen Chemie. Georg Thieme Verlag, Stuttgart 2005.

[95] Radeglia, R., Hesse, M., Meier, H. und Zeeh, B.: Spektroskopische Methoden in der organischen Chemie. VI, 478 S. 169 Abb., 86 Tab., 12 × 19 cm. Stuttgart: Georg Thieme Verlag 1979. (Thieme Taschenlehrbuch der organischen Chemie, A Theoretische und allgemeine Gebiete, B. J. für Prakt. Chemie 1981, 323, 175–175.

[96] Reimer, L., Scanning Electron Microscopy. Springer Berlin Heidelberg, Berlin, Heidelberg 1998.

[97] Hoffman, R. L., Norris, B. J., Wager, J. F., ZnO-based transparent thin-film transistors. Appl. Phys. Lett. 2003, 82, 733–735.

[98] Carcia, P. F., McLean, R. S., Reilly, M. H., Nunes, G., Transparent ZnO thin-film transistor fabricated by rf magnetron sputtering. Appl. Phys. Lett. 2003, 82, 1117–1119.

[99] Norris, B. J., Anderson, J., Wager, J. F., Keszler, D. A., Spin-coated zinc oxide transparent transistors. J. Phys. D. Appl. Phys. 2003, 36, L105–L107.

[100] Presley, R. E., Munsee, C. L., Park, C.-H., Hong, D., Wager, J. F., Keszler, D. A., Tin oxide transparent thin-film transistors. J. Phys. D. Appl. Phys. 2004, 37, 2810–2813.

[101] Nomura, K., Ohta, H., Ueda, K., Kamiya, T., Hirano, M., Hosono, H., Thin-Film Transistor Fabricated in Single-Crystalline Transparent Oxide Semiconductor. Science (80-. ). 2003, 300, 1269 LP – 1272.

References

102

[102] Hosono, H., Ionic amorphous oxide semiconductors: Material design, carrier transport, and device application. J. Non. Cryst. Solids 2006, 352, 851–858.

[103] Fortunato, E., Barquinha, P., Martins, R., Oxide Semiconductor Thin-Film Transistors: A Review of Recent Advances. Adv. Mater. 2012, 24, 2945–2986.

[104] Jeong, J. K., Won Yang, H., Jeong, J. H., Mo, Y.-G., Kim, H. D., Origin of threshold voltage instability in indium-gallium-zinc oxide thin film transistors. Appl. Phys. Lett. 2008, 93, 123508.

[105] Liu, P.-T., Chou, Y.-T., Teng, L.-F., Environment-dependent metastability of passivation-free indium zinc oxide thin film transistor after gate bias stress. Appl. Phys. Lett. 2009, 95, 233504.

[106] Woo, H. K., Kang, M. S., Park, T., Bang, J., Jeon, S., Lee, W. S., Ahn, J., Cho, G., Ko, D.-K., Kim, Y., Ha, D.-H., Oh, S. J., Colloidal-annealing of ZnO nanoparticles to passivate traps and improve charge extraction in colloidal quantum dot solar cells. Nanoscale 2019, 11, 17498–17505.

[107] Nomura, K., Kamiya, T., Hosono, H., Highly stable amorphous In-Ga-Zn-O thin-film transistors produced by eliminating deep subgap defects. Appl. Phys. Lett. 2011, 99, 053505.

[108] Nomura, K., Kamiya, T., Hosono, H., Stability and high-frequency operation of amorphous In–Ga–Zn–O thin-film transistors with various passivation layers. Thin Solid Films 2012, 520, 3778–3782.

[109] Polydorou, E., Botzakaki, M., Drivas, C., Seintis, K., Sakellis, I., Soultati, A., Kaltzoglou, A., Speliotis, T., Fakis, M., Palilis, L. C., Kennou, S., Fakharuddin, A., Schmidt-Mende, L., Davazoglou, D., Falaras, P., Argitis, P., Krontiras, C. A., Georga, S. N., Vasilopoulou, M., Insights into the passivation effect of atomic layer deposited hafnium oxide for efficiency and stability enhancement in organic solar cells. J. Mater. Chem. C 2018, 6, 8051–8059.

[110] Yoo, D. Y., Chong, E., Kim, D. H., Ju, B. K., Lee, S. Y., Effect of magnesium oxide passivation on the performance of amorphous indium–gallium–zinc-oxide thin film transistors. Thin Solid Films 2012, 520, 3783–3786.

[111] Kim, K.-D., Shin, H.-K., Chang, S.-M., Thin Film Passivation Characteristics in OLED Using In-situ Passivation. Trans. Electr. Electron. Mater. 2012, 13, 93–97.

[112] Xu, Y., Bo, B., Gao, X., Qiao, Z., Passivation Effect on ZnO Films by SF6 Plasma Treatment. Crystals 2019, 9, 236.

[113] Ortel, M., Kalinovich, N., Röschenthaler, G.-V., Wagner, V., Long-term stabilization of sprayed zinc oxide thin film transistors by hexafluoropropylene oxide self assembled monolayers. J. Appl. Phys. 2013, 114, 094512.

[114] Ali, M., Donakowski, M. D., Mayer, C., Winterer, M., Chemical vapor functionalization: a continuous production process for functionalized ZnO nanoparticles. J. Nanoparticle Res. 2012, 14, 689.

[115] Matsuzawa, H., Nakagaki, T., Iwahashi, M., Intramolecular Hydrogen Bonding (Proton Transfer) of 1-Phenyl-1,3-butanedione. J. Oleo Sci. 2007, 56, 653–658.

[116] Cobas, C., NMR signal processing, prediction, and structure verification with machine learning techniques. Magn. Reson. Chem. 2020, 58, 512–519.

References

103

[117] Wang, L.-Q., Exarhos, G. J., Windisch, C. F., Yao, C., Pederson, L. R., Zhou, X.-D., Probing hydrogen in ZnO nanorods using solid-state H1 nuclear magnetic resonance. Appl. Phys. Lett. 2007, 90, 173115.

[118] Muck, T., Wagner, V., Bass, U., Leufgen, M., Geurts, J., Molenkamp, L. W., In situ electrical characterization of DH4T field-effect transistors. Synth. Met. 2004, 146, 317–320.

[119] Ünlü, S., Banoglu, E., Ito, S., Niiya, T., Eren, G., Ökçelik, B., Sahin, M. F., Synthesis, characterization and preliminary screening of regioisomeric 1-(3-pyridazinyl)-3-arylpyrazole and 1-(3-pyridazinyl)-5-arylpyrazole derivatives towards cyclooxygenase inhibition. J. Enzyme Inhib. Med. Chem. 2007, 22, 351–361.

[120] Lee, S.-G., Kim, J.-J., Kim, H.-K., Kweon, D.-H., Kang, Y.-J., Cho, S.-D., Kim, S.-K., Yoon, Y.-J., Recent Progress in Pyridazin-3(2H)-Ones Chemistry. Curr. Org. Chem. 2004, 8, 1463–1480.

[121] Uekawa, M., Miyamoto, Y., Ikeda, H., Kaifu, K., Nakaya, T., Synthesis and luminescent properties of europium complexes. Synth. Met. 1997, 91, 259–262.

[122] Büttner, S., Riahi, A., Hussain, I., Yawer, M. A., Lubbe, M., Villinger, A., Reinke, H., Fischer, C., Langer, P., First synthesis of functionalized 5-aryl-3-(trifluoromethyl)phenols by regioselective [3+3] cyclocondensations of 1,3-bis(silyloxy)-1,3-butadienes with 3-aryl-3-silyloxy-1-trifluoromethyl-2-en-1-ones. Tetrahedron 2009, 65, 2124–2135.

[123] Frisch, M. J., Trucks, G. W., Schlegel, H. B., Scuseria, G. E., Robb, M. A., Cheeseman, J. R., Scalmani, G., Barone, V., Mennucci, B., Petersson, G. A., Nakatsuji, H., Caricato, M., Li, X., Hratchian, H. P., Izmaylov, A. F., Bloino, J., Zheng, G., Sonnenberg, J. L., Hada, M., Ehara, M., Toyota, K., Fukuda, R., Hasegawa, J., Ishida, M., Nakajima, T., Honda, Y., Kitao, O., Nakai, H., Vreven, T., Montgomery, J. A., Peralta, J. E., Ogliaro, F., Bearpark, M., Heyd, J. J., Brothers, E., Kudin, K. N., Staroverov, V. N., Kobayashi, R., Normand, J., Raghavachari, K., Rendell, A., Burant, J. C., Iyengar, S. S., Tomasi, J., Cossi, M., Rega, N., Millam, J. M., Klene, M., Knox, J. E., Cross, J. B., Bakken, V., Adamo, C., Jaramillo, J., Gomperts, R., Stratmann, R. E., Yazyev, O., Austin, A. J., Cammi, R., Pomelli, C., Ochterski, J. W., Martin, R. L., Morokuma, K., Zakrzewski, V. G., Voth, G. A., Salvador, P., Dannenberg, J. J., Dapprich, S., Daniels, A. D., Farkas, Foresman, J. B., Ortiz, J. V, Cioslowski, J., Fox, D. J., Gaussian 09, Revision B.01. Gaussian 09, Revis. B.01, Gaussian, Inc., Wallingford CT 2009.

[124] Hanwell, M. D., Curtis, D. E., Lonie, D. C., Vandermeersch, T., Zurek, E., Hutchison, G. R., Avogadro: an advanced semantic chemical editor, visualization, and analysis platform. J. Cheminform. 2012, 4, 17.

[125] Agura, K., Hayashi, Y., Wada, M., Nakatake, D., Mashima, K., Ohshima, T., Studies of the Electronic Effects of Zinc Cluster Catalysts and Their Application to the Transesterification of β‐Keto Esters. Chem. – An Asian J. 2016, 11, 1548–1554.

[126] Johnson, A. L., Kingsley, A. J., Kociok-Köhn, G., Molloy, K. C., Sudlow, A. L., Inorganic and Organozinc Fluorocarboxylates: Synthesis, Structure and Materials Chemistry. Inorg. Chem. 2013, 52, 5515–5526.

[127] Appel, A., Proceedings of the April 30--May 2, 1968, Spring Joint Computer Conference on - AFIPS ’68 (Spring). ACM Press, New York, New York, USA 1968, p. 37.

[128] Goldstein, R. A., Nagel, R., 3-D Visual simulation. Simulation 1971, 16, 25–31.

[129] Köhling, J., Wagner, V., High speed picoliter droplet top-view analysis for advancing and

References

104

receding contact angles, boiling regimes and droplet-droplet interaction. Int. J. Heat Mass Transf. 2021, 169, 120939.

[130] Yarin, A. L., Drop Impact Dynamics: Splashing, Spreading, Receding, Bouncing…. Annu. Rev. Fluid Mech. 2006, 38, 159–192.

[131] Hermansson, K., Lindberg, U., Hok, B., Palmskog, G., TRANSDUCERS ’91: 1991 International Conference on Solid-State Sensors and Actuators. Digest of Technical Papers. IEEE 2014, pp. 193–196.

[132] Bourrianne, P., Lv, C., Quéré, D., The cold Leidenfrost regime. Sci. Adv. 2019, 5, eaaw0304.

[133] Köhling, J., Wagner, V., Verfahren Zum Aufbringen Einer Substanz Auf Einer Oberfläche, DE 10 2020 133 536.9, publ. date 2020.

[134] Bradski, G., The OpenCV Library. Dr. Dobb’s J. Softw. Tools 2000.

Acknowledgments

105

Acknowledgments

First of all, I would like to express my gratitude to my PhD supervisor Prof. Dr. Veit Wagner. It

was a particularly educational time with a great atmosphere in your research group. The

continuous support, patience and guidance could not be given for granted. I am especially

grateful for the interesting multidisciplinary research topic, that I could also follow my own

research interests and present my results at many conferences.

I would like to thank Prof. Dr. Gerd-Volker Röschenthaler for being part of my dissertation

committee and for reviewing my dissertation. It was a very interesting collaboration during my

time at Jacobs University, with many productive discussions, a lot of support and with a great

insight into Fluorine chemistry.

Furthermore, I would like to thank Prof. Dr. Ralf Anselmann for his participation in my

dissertation committee and for reviewing my dissertation. Furthermore, also for my previous

time at Evonik GmbH, during which I got to know Jacobs University and which made this

dissertation possible.

Special thanks also go to my colleagues in the research group for moral and scientific support,

as well as a motivating and relaxed atmosphere. Special thanks go to Dr. Torsten Balster, Dr.

Vladislav Jovanov, Dr. Michael Huth, Dr. Oliver Gomez, Dr. Nataliya Kalinovich, Talha Nisar,

Arne Müller, Vladimir Bacic and Jonas Koppe, as well as all the others including Master and

Bachelor students I had the chance to work with.

Big thanks also go to Britta and Ronja for their constant organizational support.

From Jacobs University I would also like to thank Prof. Dr. Dr. Arnulf Materny,

Prof. Dr. Ulrich Kleinekathöfer, Prof. Dr. Jürgen Fritz, Prof. Dr. Ulrich Kortz, and

Dr. Achim Gelessus whose resources I was allowed to use.

Of course, I would also like to thank my family for their endless support in all kinds of situations.

This includes especially my parents Annette and Ludger. Thanks to my dad for endless proof-

reading. Thanks to my sisters, brothers-in-law with my nieces and nephew. Special thanks to

my cousin Christoph, who accompanied my whole studies. I think it is clear that I would not be

here without you all.

I want to also thank my fiancée Sara, for all the emotional support und keeping me focused.

I am very excited for our time together yet to come.

Last but not least, I would like to thank Chava and Ivan, whom I got to know and appreciate as

really good friends during my time at Jacobs University.

Zinc oxide thin-films by spray pyrolysis with low deposition ...

Documents