In situ characterization of ALD processes and study of reaction mechanisms for high-k metal oxide formation

Yoann Tomczak

Laboratory of Inorganic Chemistry

Department of Chemistry

Faculty of Science

University of Helsinki

Finland

ACADEMIC DISSERTATION

To be presented with the permission of the Faculty of Science of the University of Helsinki for public criticism in Auditorium A110 of the Departement of Chemistry, A.I Virtasen aukio

1, on 06.06.2014 at 12 o,clock noon

Helsinki 2014

2

© Yoann Tomczak

ISBN: 978-952-10-9925-0 (paperback) ISBN: 978-952-10-9926-7 (PDF version)

http://ethesis.helsinki.fi

University of Helsinki

Helsinki 2014

3

Supervisors

Professor Mikko Ritala and

Professor Markku Leskelä

Laboratory of Inorganic Chemistry Department of Chemistry

University of Helsinki Helsinki, Finland

Reviewers

Professor Martyn Pemble Tyndall National Institute and University College Cork

Cork, Ireland

Professor Charles H. Winter Wayne State University

Detroit MI, U.S.A

Opponent

Professor Annelies Delabie Katholieke Universiteit Leuven

Leuven, Belgium

4

Abstract

Atomic Layer Deposition (ALD) is a thin-film deposition method allowing the growth of highly conformal films with atomic level thickness and composition precision. For most of the ALD processes developed, the reaction mechanisms occurring at each step of the deposition remain unclear. Learning more about these reactions would help to control and optimize the existing growth processes and develop new ones more quickly. For that purpose, in situ methods such as quartz crystal microbalance (QCM) and quadrupole mass spectrometer (QMS) present numerous advantages: They provide direct information on the growth mechanisms during the process. Additionally, they do not require separate experiments or large amounts of precursors to test the efficiency of new processes and could be very effective means to monitor industrial processes in real time. This thesis explores the most common in situ analytical methods used to study ALD processes. A review on the ALD metal precursors possessing ligands with nitrogen bonded to the metal center and their reactivity is provided. The results section reports the reaction mechanisms of the following ozone and water based ALD processes using precursors with nitrogen bonded ligands for the deposition of high-k and other oxide films:

The tBuN=M(NEt2)3 (M = Nb, Ta) water or ozone processes for the deposition of Nb2O5 and Ta2O5: The influence of the nature of the metal on the adsorption mechanism of precursor with similar heteroleptic ligands was highlighted.

The Si2(NHEt)6, Al(CH3)3 and water process for the deposition of ternary AlxSiyOz oxide: A

theoretical model based on the in situ results was used to relate the film composition with the proposed reaction mechanisms.

The LiN[Si(CH3)3]2 – O3 process for the deposition of Li2SiO3 : Dissociative adsorption of the bimetal precursor was observed. Additionally, as the temperature increases the adsorption of HN[Si(CH3)3]2 becomes more efficient and thus more silicon is incorporated in the film, decreasing the Li/Si ratio and increasing the growth rate.

The Ti(OiPr)3(NiPr-Me-amd) and Ti(NMe2)2(OiPr)2 with water and ozone processes for the

deposition of TiO2: In the water based processes, the amidinate precursor showed adsorption with decomposition of its ligand while the alkoxy/alkylamide precursor adsorbed through ligand exchange. The ozone processes showed combustion reactions during both metal precursor and ozone pulses. Additionally, the evolution of the reaction mechanism with temperature and its influence on the growth characteristics are reported.

The [Zr(NEtMe)2(guan-NEtMe)2] – water process for the deposition of ZrO2: Decomposition of the guanidinate ligand upon reaching the surface was observed during adsorption of the metal precursor.

5

Preface The studies reported in this thesis were carried out in the Laboratory of Inorganic Chemistry at the

Department of Chemistry of the University of Helsinki during the years 2010-2014. I am very grateful

to my supervisors Prof. Mikko Ritala, Prof. Markku Leskelä and Dr. Kjell Knapas for their guidance,

support and responsiveness during these 4 years. In addition, I would like to acknowledge the

reviewer of this thesis, Prof. Charles H. Winter and Prof. Martyn Pemble.

I would like to thank my co-workers at the inorganic chemistry laboratory who have helped me during

the thesis: Dr. Timothee Blanquart, Dr. Jani Hämäläinen and Dr Jaakko Niinistö for their contribution

to process development and their help with technical issues; Dr. Suvi Haukka for the constructive

discussions we had on reaction mechanisms; Dr. Marianna Kemell, Mikko Heikkilä and Markku

Sundberg for their time and assistance with EDS, XRD and computational results, respectively. I also

want to thank all the other members of the ALD and Catlab groups for their welcome and the good

ambiance in the lab.

The research leading to this thesis was funded by the European Community’s Seventh Framework

Program (FP7/2007-2013) under Grant Agreement ENHANCE-238409 and was also supported by

the Finnish Centre of Excellence in Atomic Layer Deposition. I wish to thank Dr. Harish Parala, Prof.

Anjana Devi, Prof. Erwin Kessels and all the other supervisors involved in ENHANCE for the

exceptional opportunities and trainings they provided during the program.

I also want to thank all my ENHANCE research fellows for the good times we had together: Matthieu

Weber, Valentino Longo, Marco Gavagnin, Dr. Daniela Beckermann, Dr. Quentin Simon, Brunot

Gouat, Dr. Nabeel Aslam, Dr Manish Banerjee and Babu Srinivasan, whishing them a bright and

successful future.

Special thanks also to my parents, my brother Pierre and my friends Tibi, Alexis, Franky and Fred for

their psychological support.

Finally, I would like to thank my wife Sorina for her love and understanding during all these years.

Helsinki, May 2014

Yoann Tomczak

6

Table of contents Abstract ................................................................................................................................................... 4

Preface .................................................................................................................................................... 5

Table of contents .................................................................................................................................... 6

List of publications .................................................................................................................................. 8

List of acronyms used ............................................................................................................................. 9

1. Introduction ...................................................................................................................................... 10

2. Background ....................................................................................................................................... 11

2.1 Atomic layer deposition .............................................................................................................. 11

2.2 In situ analytical methods used for reaction mechanism studies in ALD ................................... 12

2.3 Precursors with nitrogen donor ligands used in ALD .................................................................. 15

2.3.1 Homoleptic alkylamides ....................................................................................................... 16

2.3.2 Disilazides ............................................................................................................................. 17

2.3.3 Amidinates ........................................................................................................................... 17

2.3.4 Guanidinates ........................................................................................................................ 18

2.3.5 Heteroleptic alkylamides ..................................................................................................... 18

2.4 Metal precursor adsorption mechanisms in ALD ........................................................................ 19

2.4.1 Chemisorption and dissociative adsorption ........................................................................ 20

2.4.2 Ligand exchange mechanism ............................................................................................... 21

2.4.3. Precursor adsorption through ligand decomposition ......................................................... 22

2.4.4. Partial combustion .............................................................................................................. 22

2.5 Reactions of precursors with nitrogen donor ligands ................................................................. 23

2.5.1 Alkylamides .......................................................................................................................... 23

2.5.2 Hetroleptic alkylamides ....................................................................................................... 24

2.5.3 Amidinates ........................................................................................................................... 25

2.5.4 Guanidinate .......................................................................................................................... 27

2.5.5 Disilazide .............................................................................................................................. 28

7

3. Experimental ..................................................................................................................................... 30

4. Results and discussion ...................................................................................................................... 32

4.1 Nb2O5 and Ta2O5 .......................................................................................................................... 32

4.2 AlxSiyO .......................................................................................................................................... 33

4.3 Li2SiO3 .......................................................................................................................................... 35

4.4 TiO2 .............................................................................................................................................. 37

4.5 ZrO2 ............................................................................................................................................. 39

5. Conclusions ....................................................................................................................................... 41

6. References ........................................................................................................................................ 43

8

List of publications

This work is based on the following publications which are referred to with Roman numerals:

I. Y. Tomczak, K. Knapas, M. Sundberg, M. Ritala, M. Leskelä “In situ reaction mechanism

studies on the new tBuN=M(NEt2)3 -Water and tBuN=M(NEt2)3 - Ozone (M=Nb,Ta) Atomic

Layer Deposition processes.” Chem. Mater.(2012), 24(9), 1555-1561

II. Y. Tomczak, K. Knapas, S. Haukka, M. Kemell, M. Heikkilä, M. Ceccato, M. Leskelä, M.

Ritala “In situ reaction mechanism studies on atomic layer deposition of AlxSiyOz from

trimethylaluminium, hexakis ethylaminodisilane and water.” Chem. Mater.(2012), 24(20),

3859-3867

III. Y. Tomczak, K. Knapas, M. Sundberg, M. Leskelä, M. Ritala “In situ reaction mechanism

studies on lithium hexadimethyldisilazide and ozone atomic layer deposition process for

lithium silicate.” Journal of Physical Chemistry C (2013), 117(27), 14241-14246

IV. Y. Tomczak, K. Knapas, M. Ritala, M. Leskelä “In situ reaction mechanism studies on the

Ti(NMe2)2(OiPr)2-D2O and Ti(OiPr)3(NiPr-Me-amd)-D2O Atomic Layer Deposition

processes” Journal of Vacuum Science and Technology A: Vacuum, Surfaces, and Films

(2014), 32(1), 01A121-01A121-7

V. T. Blanquart, J. Niinistö, N. Aslam, M. Banerjee, Y. Tomczak, M. Gavagnin, V. Longo, E.

Puukilainen, H.D. Wanzenböck, W.M.M. Kessels, A. Devi, S. Hoffmann-Eifert, M. Ritala,

and M. Leskelä “[Zr(NEtMe)2(guan-NEtMe)2] as a novel ALD precursor: ZrO2 film growth

and mechanistic studies” Chem. Mater.(2013), 25(15), 3088-3095

VI. M. Kaipio, T. Blanquart, Y. Tomczak, J. Niinistö, M. Gavagnin, V. Longo, V. Pallem, C.

Dussarrat, M. Ritala, M. Leskelä “Atomic layer deposition, characterization and growth

mechanism of high quality TiO2 thin films” submitted

The publications are reprinted with the permission from the copyright holders. For publications I-IV,

the author has planned the research with the help of the coauthors, performed the experiments, wrote

the first versions of the papers and completed them for submission with the help of the coauthors. In

publications V-VI, the author has performed the reaction mechanism studies and written that part of

the article.

9

List of acronyms used

AFM Atomic force microscopy

ALD Atomic layer deposition

amd Amidinate, RN(CR’)NR

Cp Cyclopentadienyl, -C5H5

CN Coordination number

CVD Chemical vapor deposition

DFT Density functional theory

DRAM Dynamic random access memory

DL Detection limit

EDX Energy-dispersive X-ray spectroscopy

Et Ethyl, -C2H5

FT-IR Fourier transform infrared spectroscopy

GPC Growth per cycle

GR Growth rate

guan Guanidinate, NR(CNR’R’’)NR

HMDS Hexamethyldisilazane iPr Isopropyl, -CH(CH3)2

Me Methyl, -CH3

ML Monolayer

NOx Nirous oxides: NO, NO2, N2O…

PEALD Plasma enhanced ALD

QCM Quartz crystal microbalance

QMS Quadrupole mass spectrometer

SE Spectroscopic ellipsometry

SEM Scanning electron microscopy tBu tert-butyl, C(CH3)3

ToF-ERDA Time-of-flight elastic recoil detection analysis

TPD Temperature-programmed desorption

HV High vacuum

XPS X-ray photoelectron spectroscopy

XRF X-ray fluorescence

XRR X-ray reflectometry

10

1. Introduction In the microelectronics industry, the size of the devices continuously decreases to follow Moore´s law.

To further enhance their efficiency, the geometry of MOSFETs is shifting from planar to three

dimensional. Consequently, thin film deposition methods used for producing these devices have to

adapt to produce thinner and more conformal films. In this context, Atomic Layer Deposition (ALD) is

a method of choice because it allows precise control over the deposited film thicknesses down to the

atomic level and exhibits perfect conformality even on structures with high aspect ratios.1-3 This

method permits the deposition of a wide range of materials, from pure noble metals to transition metal

oxides.4

Many types of metal precursors have been investigated for the deposition of high-k metal oxides:

halides, alkyls, alkoxides, alkylamides, amidinates and guanidinates are a few examples of the types

of ligands used in chemical vapor deposition (CVD) and its pulsed version, ALD. Good volatility,

thermal stability and reactivity of the metal precursors are important prerequisites for their use in

industry. In addition, the purity of the films deposited with these precursors and the compatibility of the

ALD method with the other steps of the device fabrication are critical. Among the wide variety of

precursors, those containing nitrogen donor ligands are of particular interest because of their good

volatility and the high reactivity of their metal-nitrogen bonds.

A deep understanding of the reaction mechanisms occurring during the ALD processes is needed in

developing new precursors and in optimizing the already existing ALD processes. In situ

measurements are best suited for mechanistic studies of ALD processes as they provide chemical

information continuously during the process. However, a strong theoretical knowledge of the nature of

the potential reactions taking place during the deposition is required to solve the reaction mechanisms

completely and to explain the main characteristics of the process.

The goal of this work is to give some insight on in situ reaction mechanism studies on the ALD of

high-k metal oxide thin films. A particular focus is applied on the reactivity of precursors with metal-

nitrogen bonds such as alkylamides, disilazides, amidinates and guanidinates. Several decomposition

pathways are explicated and their possible influence on the process characteristics discussed.

The structure of this thesis is the following: Chapter 2 presents a literature review on the reactivity of

different types of precursors with nitrogen donor ligands used in ALD. Chapter 3 shows the

experimental information on the in situ mechanistic studies performed in this work. Finally, the results

obtained for this thesis are summarized in Chapter 4.

11

2. Background

2.1 Atomic layer deposition

The general principle of ALD for metal oxide deposition is as follows: first the metal containing

precursor is pulsed in the reaction chamber where it reacts with the substrate surface. The excess of

unreacted precursor molecules and the reaction by-products are removed from the chamber during

the following purge by inert gas (generally N2 or Ar). The oxygen containing precursor is then pulsed

and reacts with the adsorbed metal species on the surface, removing their remaining precursor

ligands and regenerating the reactive surface sites. Finally, another pulse of purge gas is supplied to

remove the reaction by-products and the excess of oxygen containing precursor from the chamber

and the substrate surface is ready for another ALD cycle to begin. For thermal ALD of metal oxides,

the oxygen precursors used are mainly water and ozone. Plasma Enhanced Atomic Layer Deposition

(PEALD) uses pulses of oxygen activated with plasma to deposit metal oxides.5 However, the

complex reactions involved in PEALD are beyond the scope of this work and will not be examined in

this thesis.

Figure 1. Typical growth rate evolution of ALD processes with temperature.

Figure 1 shows the different possible evolutions of an ALD growth rate with temperature:2 on one

hand, the metal precursors can condense on the surface at low temperature, eliminating the perfect

conformality and increasing the growth rate. On the other hand, low temperature can limit the

reactivity of the precursors and thus decrease the growth rate. There is a temperature range where

the deposition is surface-controlled and the surface reactions are self-limiting. This is usually

described as the “ALD window” of the process. Therefore, the film thickness can be precisely

controlled by changing the number of deposition cycles without the need to control the precursor

doses strictly. The self-limiting behavior allows deposition of perfectly uniform and conformal films on

12

large area substrates and on complex structures with high aspect ratio. In ideal ALD processes, the

growth per cycle (GPC) is constant in this window. For real ALD processes, the growth per cycle is

highly dependent on the nature of the metal precursor used and the deposition parameters

(temperature, pressure, reactor geometry): for instance, the reactivity of the precursor and the density

of reactive sites on the surface are strongly influenced by the deposition temperature. Other

properties of the deposited film such as the GPC, the density of the film and the amount of impurities

incorporated are directly linked to these factors and can vary even within the “ALD window”. However,

the main characteristic of ALD remains a self-limiting behavior where the adsorption reactions on the

surface can stop even if some potential reactive sites are still unreacted. The ligands of already

adsorbed metal precursors can hinder the access of unreacted precursor molecules to adjacent

hydroxyl groups and/or other potential reactive sites. Within real processes, the steric hindrance of the

adsorbed surface ligands can also impact some properties of the deposition (GPC, density).

At higher temperature, secondary decomposition reactions start to play an increasing role in the film

deposition characteristics: in Figure 1 the dashed area corresponds to a temperature range where the

decomposition reactions of the metal precursor become thermodynamically and kinetically favored

and modify the adsorption mechanism. Decomposition reactions open additional reactive sites that

can destroy the self-limiting behavior of the deposition, increasing the growth rate and possibly

incorporating more impurities in the films. High temperature can also enhance desorption of

precursors and surface dehydroxylation, thus decreasing the growth rate.

Although ALD has been used for several decades, many of the underlying mechanisms of the film

deposition remain mostly unknown. Firstly, the nature of the elemental reactions occurring during an

ALD cycle is often simplified: the common thought is that a single precursor molecule reacts with one

or several reactive sites upon adsorption, forming volatile by-products that leave the surface without

further interactions. Secondly, the surface itself is ideally depicted as two dimensional with a discrete

number of identical reactive sites, often considered as reacting separately. This ideal vision of the

reaction is contradicted by empirical evidence hinting to a more complex reality: the reactivity of one

adsorbed precursor can be strongly influenced by the neighboring molecules. Reaction mechanism

studies help to understand the influence of the substrate, co-reactants and temperature on the

process characteristics such as growth per cycle, crystallinity and film composition.

2.2 In situ analytical methods used for reaction mechanism studies in ALD

In situ analytical methods allow the investigation of ALD chemistry and also several characteristics of

the growing films directly during the process. They minimize the alteration of the film due to exposure

of the sample to air and constitute the only means to observe reaction mechanisms accurately as they

are taking place. Most of the in situ analytical methods require specific set-up conditions and proper

interpretation to provide valuable information on the reaction mechanisms. The following describes

the most common in situ techniques used to investigate atomic layer deposition processes.

13

Fourier transform infrared spectroscopy (FT-IR) is a non-destructive technique which provides direct

information about the nature of the chemical bonds formed on the reaction surface during an ALD

process. To a smaller extent, FT-IR can also inform on the volatile by-products released at each step

of the growth.6 It has been used to study reaction mechanisms in ALD metal oxide6-10 and nitride11-13

processes. Recording in situ FT-IR spectra requires a significant acquisition time that lengthens the

ALD cycles but also ensures that each reaction reaches completion before switching to the next

reactant. The data obtained are often differential spectra qualitatively assessing the evolution of the

chemical bonds on the growing film. However, FT-IR does not directly inform on the exact nature of

the surface species (especially regarding carbon containing bonds) as one chemical bond can be

present in several different surface species. Accordingly, a careful comparison with similar molecules

and surfaces from the literature is necessary to correctly interpret the observed IR bands.

Quadrupole mass spectrometry (QMS) allows the investigation of the gaseous species present in the

reactor during the process.14-18 Volatile by-products released at each step of the ALD process are

ionized and fragmented by electrons from a filament source and then sorted by the quadrupole

analyzer. The quadrupole requires high vacuum (HV) conditions obtained by differential pumping

through an orifice with a turbomolecular pump. The signal measured is expressed versus mass to

charge ratio, m/z. The main limitation of this technique is the difficulty to attribute an m/z signal to a

single fragment: several by-products originating from different sources (ozone combustion,

hydrogenated ligands…) can produce fragments at the same m/z and thus the signal observed with

the QMS can be composed of multiple contributions. It is possible to discriminate these signals when

comparing their evolution with a signal belonging to only one fragment. However, in most cases the

interpretation remains problematic, especially with combustion reactions occurring during the process.

The quadrupole analyzer allows monitoring only a few masses at the same time during a short ALD

cycle. This limitation could be eliminated by using another type of analyzer such as a Time-of-Flight

(ToF).

Quartz crystal microbalance (QCM) allows the detection of minuscule mass changes on the reaction

surface: deposition of less than a monolayer can be observed with a great precision. When a certain

mass is deposited on a quartz crystal, its resonance frequency changes accordingly. The frequency

change is also dependent on other factors such as temperature and deposition area.

Figure 2 exhibits an example QCM data obtained when measuring one typical ALD cycle: During the

metal precursor pulse, the mass increases abruptly in the first few seconds before it reaches a

plateau. The self-limiting behavior of the studied ALD process is demonstrated by this mass

stabilization during the metal precursor pulse. The mass can sometimes decrease slightly during the

following purge, illustrating desorption of unreacted molecules from the surface or decomposition of

the adsorbed ligands. The oxygen precursor pulse (D2O in Figure 2) induces another mass change

when the remaining adsorbed ligands are released from the surface and the reactive sites are

regenerated. Depending on the mass of the ligand compared to the mass of the reactive surface

species (hydroxyl groups, active oxygen) this mass change can be an increase, for instance in the

14

tetramethylaluminum and water process.19 However as portrayed in Figure 2, in most processes the

mass decreases during the oxygen precursor pulse due to the removal of heavy ligands.

Depending on the process, additional phenomena can influence the quartz resonance frequency.

Sudden pressure and temperature change in particular can directly influence the QCM data. For

instance, ozone pulse triggers a significant artifact on the QCM data. Accordingly, longer stabilization

times are needed for the purge after an ozone pulse.

Figure 2. Typical QCM data obtained for one ALD cycle. m1 corresponds to the mass increase after

the metal precursor pulse. m0 corresponds to the overall mass change after the full cycle.

X-ray photoelectron spectroscopy (XPS) is a surface sensitive quantitative spectroscopic technique

that measures the elemental composition, empirical formula, chemical state and electronic state of the

elements present on the growing film. The XPS measurements do not occur exactly in situ as the ALD

deposition would damage the XPS set-up but rather in vacuo.20-24 The substrate is moved from the

reaction chamber between each cycle or half cycle to the XPS chamber for analysis without breaking

the vacuum. Additionally, the acquisition requires UHV in the XPS chamber. This limits the

contamination of the sample due to air exposure. In situ XPS is especially useful to observe the

quantitative evolution of an element in the film as well as its direct environment: the binding energies

measured with XPS for an atom are directly related to its neighbors. A change in the shape of an XPS

peak illustrates a modification of the environment of the corresponding atom. This feature allows

precise investigation on the interfacial layer formation and on the effect of the substrate on the initial

growth rate.25-27 When measuring thicker films, quantitative compositional analysis can also be done

with ex situ XPS by ion beam etching: consecutive XPS spectra are measured as a function of the

etching time and give a compositional depth profile of the grown film. This is especially useful to

observe migration of a specific element (for instance a dopant) in a multilayer structure.28

In situ spectroscopic ellipsometry (SE) monitors the change in polarization of a light beam upon

reflection on the surface.29-33 With a model-based analysis, SE gives information on the evolution of

the optical dielectrical function including the film refractive index and the thickness of the growing film

after each deposition cycle. However, SE does not provide direct chemical information on the reaction

15

mechanism. The set-up required is generally non-invasive and very versatile which makes in situ SE

a valuable alternative to QCM to measure film thickness.

X-ray fluorescence spectroscopy (XRF) is a spectroscopic technique used for elemental analysis.

Using the high intensity of synchrotron generated X-rays increases the sensitivity of the XRF to

monitor the number of metal atoms deposited on the surface during each ALD cycle. This technique is

especially useful to study initial nucleation on planar substrates and nanoporous films.34, 35 However, it

is unsuitable for the detection of light atoms such as carbon, oxygen and nitrogen. High-resolution

synchrotron x-ray diffraction (XRD) is an in situ method that studies the crystallography of the growing

films.36 This technique is particularly important to investigate the nucleation of crystalline films and the

influence of reaction temperature on the nucleation of a specific crystalline phase. The limited access

to synchrotron facilities and the high cost of the experiments limit the use of these techniques to study

reaction mechanism of ALD processes routinely.

In situ measurements sometimes do not provide sufficient information to fully solve the reaction

mechanisms occurring during the ALD deposition: in complex adsorption and decomposition

reactions, several competitive pathways can possibly lead to the same by-products. In such cases,

computational modeling of the key process steps can help to assess the energetically most favorable

reaction pathways. Among all the atomic-scale simulation methods available, Density functional

theory (DFT) is the most commonly used to study ALD processes. This quantum mechanical method

allows the modeling of systems up to 1000 atoms, occupying roughly one cubic nanometre. It can

describe classical two-centre covalent bonds, non-classical multi-centre bonds, metallic bonding in

conductors and polar / ionic bonding. This is especially valuable when assessing structural

differences in metal precursors which could account for a difference in the reactivity of these

precursors. In addition, DFT can represent accurately atomic layer arrangement in the ALD deposited

semiconductors and insulators. Recently, Elliott reviewed the atomic-scale simulation studies on ALD

processes.37

2.3 Precursors with nitrogen donor ligands used in ALD

The traditional precursors used to deposit transition metal oxides possess halides, alkyls, β-

diketonate or alkoxy ligands. During the last decades, several new types of ligands have been

experimented to expand the deposition possibilities (ALD window, evaporation temperature, growth

per cycle…). Particularly, precursors possessing metal-nitrogen bonds present interesting advantages

in ALD. They are highly reactive with water, ozone and surface hydroxyls and usually lead to higher

growth rates than alkoxides and halides. In addition, the easy conversion of the M-N bonds into M-O

bonds leaves very small N contamination in the metal oxide films grown with such precursors.

16

2.3.1 Homoleptic alkylamides

Figure 3. Types of ligand with metal-nitrogen bonds employed in ALD processes (Ri correspond to

alkyl groups and L to another type of ligand such as alkoxide, cyclopentadienyl, amidinate…).

An alkylamide ligand consists of a nitrogen atom linked to the metal center of the precursor and to

small alkyl groups (see Figure 3). Homoleptic alkylamide precursors have been used for the ALD of a

wide range of materials: aluminium oxide,38-42 tungsten oxide43 and antimony oxide44-46 as well as

most of the group 4 and 5 transition metal oxides (namely titanium,47-59 hafnium,60-75 zirconium76-84 and

tantalum85-91) have been deposited thermally with water and ozone but also with oxygen plasma. The

same alkylamide precursors allow the deposition of transition metal nitrides (TiN,92-96 ZrNx,97-99 HfNx,98,

100-102 TaNx103-107 and WxN108) with ammonia, nitrogen plasma or hydrogen plasma.

The most common alkylamide precursors possess dimethylamide, diethylamide or ethylmethylamide

ligands. They are liquids with reasonably high vapor pressures and low evaporation temperatures due

to the small size of their alkyl moieties. Heavier and bulkier ligands would reduce their volatility

without necessarily increasing the stability of the precursor. The M-N bonds present in the homoleptic

alkylamides are generally weaker than the corresponding M-O bonds which render alkylamide ligands

highly reactive with surface hydroxyl groups, water and ozone, even at low temperatures. However,

they also induce poor thermal stability both in the gas phase and on the surface. Thermogravimetric

17

analysis of homoleptic alkylamides show mass losses in a single step and with low mass residue,

indicating evaporation of the precursors without decomposition.109

The nature of the metal center also plays an important role in the deposition. Late transition metals

are less oxiphilic and the M-N bonds of their metal precursor are stronger than early transition metals.

For instance among the group IV and V transition metals, titanium alkylamides seem to yield

significantly lower growth rates than their niobium and hafnium counterparts.109 Although the Ti-N

bond energy is less than those of the other group IV and V metal-nitrogen bonds, the small size of the

Ti atom can hinder nucleophilic attacks and account for the lower growth rates observed. For most of

the homoleptic alkylamides, the maximum temperature to obtain a proper ALD film is around 250

°C.67, 72, 76, 109 This upper limit is seemingly independent of the size of the alkylamide ligands but varies

with the metal: vanadium alkylamides decompose already at 200 °C110 when zirconium alkylamides

remain stable up to 250 °C.76 The amount of impurities in the films deposited with the homoleptic

alkylamides is small, usually below 2 at.% for carbon and 1 at.% for nitrogen.67, 72, 76, 109 However,

there are some discrepancies depending on the oxygen source, the metal and the ligands with no

clear trend.

2.3.2 Disilazides

Disilazide precursors, also known as silylamides, possess ligands consisting of a nitrogen atom

bonded to the metal center and to two trimethylsilyl groups (Figure 3). These precursors have been

used in ALD in conjunction with water to deposit zirconium oxide,111 hafnium oxide,7, 111 lanthanum

oxide112-115 and praseodymium oxide.116-118 In addition, ALD with lutetium and lithium disilazide

precursors and ozone as the oxygen source led to the formation of lutetium and lithium silicates.119-121

The reaction mechanism of the lithium disilazide and ozone process is discussed in publication III.

Zn(N(SiMe3)2)2 was also used to grow ZnSe but resulted in an important incorporation of nitrogen into

the deposited films.122 Disilazide precursors start to decompose partially at relatively low temperature

(200 °C) and thus the oxide films deposited from these precursors contain a significant amount of

silicon impurities.

2.3.3 Amidinates

Amidinate ligands are bonded to a metal center through two nitrogen atoms that themselves are

linked together by a carbon atom (Figure 3). The substituents of the ligand can be chosen to tailor the

volatility and the melting point of the precursor. The chelating effect of the amidinate ligand increases

the thermal stability of these precursors but the M-N bonds allow high reactivity even at low

temperatures. Amidinates are also non-pyrophoric but maintain high reactivity toward water. They

allow the ALD of metal films such as iron,123 cobalt,124-127 nickel,123 ruthenium128-130 and copper123, 124,

131-135 when used in combination with ammonia or hydrogen, with or without plasma. Many oxide films

(Al2O3,42 CoOx,123 Y2O3,136 La2O3,137-139 PrOx,116 Er2O3,140) have also been deposited with amidinate

precursors and water or ozone. The surface reactions involved in the adsorption of amidinates are

18

complex (see publications IV and VI) and can potentially lead to a significant incorporation of

impurities in the growing films.

2.3.4 Guanidinates

Guanidinate ligands resemble amidinates except their central bridge carbon is linked to a third

nitrogen atom (see Figure 3). Homoleptic guanidinate precursors present a high thermal stability but

also a high temperature of evaporation, generally above 120 °C. Similarly to amidinates, guanidinates

are versatile ligands with many tunable properties depending on the number and the nature of their

substituents. Guanidinates are novel promising precursors in MOCVD and PEALD of nitrides (ZrNx,

TiN, WxN,141TaNx) and metals (copper,142-144 ruthenium145). They have also been experimented in

thermal ALD for deposition of various oxides such as Gd2O3,146 ZrO2,147 HfO2,148 and lanthanide

oxides.149, 150 The thin films obtained with the guanidinate precursors show comparable nitrogen and

carbon contamination comparable to the films grown with the amidinate precursors (below 2%).

2.3.5 Heteroleptic alkylamides

Heteroleptic precursors possess metal center surrounded by several types of ligands. The

combination of these different ligands allows the tailoring of the precursor reactivity and its thermal

properties (vapor pressure, evaporation temperature, and thermal stability). Moreover, combination of

different metal-ligand bonds modify the electron distribution around the metal center and thus the

surface reaction mechanisms of heteroleptic precursors compared to homoleptic ones.

With heteroleptic alkylamides, several deposition characteristics depend on the nature of the other

ligands. Only a couple of ALD processes involving alkoxy/alkylamide precursors have been reported

in the literature: the deposition of titanium and hafnium oxides exhibits self-limiting growth rate and

excellent purity even above 300 °C.151, 152 In addition, many patents of different combinations of

alkoxy/alkylamides precursors have been filed by precursor manufacturers.153

Oxide ALD processes using zirconium and hafnium monocyclopentadienyl alkylamides result in

relatively high growth rates (0.8-0.9 Å/cycles) and low amounts of carbon and nitrogen remaining in

the films.154 Similarly to the homoleptic titanium alkylamides, titanium oxide ALD processes involving

monocyclopentadienyl alkylamides exhibit a lower GR109 (0.3-0.4 Å/cycle) than other metal oxide

processes. Precursors containing more than one Cp ligands have been evaluated for PEALD of

zirconium carbides155 and MOCVD of hafnium oxide.156 Generally, cyclopentadienyl alkylamides tend

to be thermally less stable than the other types of heteroleptic cyclopentadienyl precursors.

Decomposition reactions involving the alkylamide ligands are directly responsible for this decrease in

thermal stability. Such reactions will be investigated later in this thesis (Chapter 3).

Alkylamide/imido precursors have mostly been investigated in ALD in combination with ammonia to

deposit tantalum13, 157-160, tungsten,161-164 and molybdenum165-168 nitrides. Hausmann et al.91 deposited

19

Ta2O5 using EtN=Ta(NEt2)3 and water. More recently, Blanquart et al.169, 170 also used

alkylamide/imido precursors with water and ozone for the deposition of niobium and tantalum oxides.

In situ reaction mechanism studies of these processes are displayed in publication I.

Several alkylamide/amidinate and alkylamide/guanidinate precursors have been synthetized for use in

ALD and CVD: thermogravimetric studies of hafnium and zirconium precursors indicate significant

differences in volatility and thermal stability depending on the size of the ligands.171-173 So far, only a

couple of studies have reported precise growth characteristics for the deposition of transition metal

oxides. The ALD processes174 using (EtMeN)2Zr(guan-NEtMe)2 as the metal precursor exhibit high

growth rates (0.9 Å/cycle with water and 1.15 Å/cycle with ozone) and low carbon and nitrogen

contamination (< 3 at.%). The reaction mechanisms of these processes are investigated in publication

V. Similar processes, namely (Me2N)Hf(guan-NMe2)2 and water171 and dysprosium/gadolinium

alkylamides/guanidinates with water or ozone, have also been investigated and shown promising

results.175

2.4 Metal precursor adsorption mechanisms in ALD

The reaction of metal precursor molecules during a typical ALD cycle can be decomposed in several

simple steps according to the following equation:

gas-MLn surf-(MLn) surf-(MLx) surf-MOx bulk-MOx

where MLn is the metal-containing precursor with n ligands L, x is the number of ligands eliminated

from the surface so that 0 < x < n, and “bulk-MOx” refers to a metal atom with a local environment

(coordination number) corresponding to the structure of the film material. Firstly, the precursor

molecule adsorbs on the surface without chemically reacting with surface species. Secondly, the

adsorbed molecule can react with the surface and lose some of its ligands either by exchange with

the reactive surface sites, decomposition or dissociation. In parallel, the coordination number (CN) of

each metal center expands gradually during the adsorption to reach the configuration they would

possess in the bulk of the film. This gradual modification of the CN acts as a driving force of the

deposition and can be responsible for the crystalline state of the deposited film.37, 176 Unfortunately,

some of these steps are often difficult to clearly assess with in situ measurements and computational

studies are required to investigate their potential influence on the reaction mechanism.

During the ALD cycle, the metal precursors can adsorb by different mechanisms. These possible

mechanisms are summed up in Figure 4. Firstly, the precursor can chemisorb on the surface without

reaction with surface species (path A). The precursor molecule can also dissociate with its metal

center bonding to one reactive site and some of its intact ligands linking to the surface (path B).

However, most of the metal precursors possess highly reactive metal-ligand bonds which can interact

with surface reactive sites. For example, the metal precursors often chemisorb though the exchange

of their ligands with reactive surface sites such as hydroxyl groups (path C). The ligands are then

released intact in hydrogenated form. They can also decompose upon reacting with the surface and

20

release smaller volatile by-products (path D). Finally, for ozone based processes another type of

adsorption is possible: adsorbed oxygen atoms can react with metal precursors and partially combust

their ligands and release combustion by-products (path E).

Figure 4. Possible adsorption mechanisms of metal precursors that occur during an ALD cycle.

2.4.1 Chemisorption and dissociative adsorption

An extreme case of metal precursor adsorption is when the chemisorption occurs without reactions

between the metal precursors and the reactive surface sites. The removal of the precursor ligands is

then solely due to reaction with the oxygen containing precursors. The QCM results for such

mechanisms show a relatively small m0/m1 ratio because the precursor molecules stay intact on the

surface at the end of the metal precursor pulse (high m1). The QMS data exhibit the release of by-

products only during the oxygen precursor pulse.

Alternatively, the precursor molecule can dissociate and thus one or several of its ligand will bond to

neighboring reactive sites instead of remaining kinked to the adsorbing molecule. Dissociative

adsorption can occur due to the high reactivity of the metal-ligand bond, the lack of suitable reactive

sites or thermal energy necessary for ligand exchange reactions. Such dissociative adsorption was

observed for the deposition of copper amidinates on metallic surfaces at low temperature.134

21

2.4.2 Ligand exchange mechanism

Figure 5. Ligand exchange mechanism for the ALD of metal oxide with water.

Ligand exchange is the most common adsorption mechanism in metal oxide ALD. During the metal

precursor pulse, the precursor ligands react with surface hydroxyl groups and leave the surface

without decomposition: the hydrogen from the surface –OH groups is transferred to the ligand, a bond

is formed between the metal center and the oxygen of the surface –OH group and the hydrogenated

ligand is released to the gas phase (Figure 5). The number of molecules deposited before reaching

saturation is related to the number of available reactive sites. For ALD of oxides, the reactive sites

involved in ligand exchange reactions are almost exclusively hydroxyl groups. During the water pulse,

the remaining ligands are hydrogenated and the surface –OH groups regenerated to start a new

cycle.

The exchange reactions release gaseous by-products that can be detected by QMS. However, the

nature and repartition of by-products originating from ligand exchange can vary greatly from process

to process. An extreme case would be all the ligands reacting with the reactive surface sites and

being eliminated as by-products during the metal precursor pulse. For most processes, the release of

these specific by-products is divided between both metal and oxygen precursor pulses. In situ

analytical techniques are then necessary to quantify precisely this repartition and to draw accurate

half-reactions for the reaction mechanism. In this respect, the combination of QCM and QMS allows

the discrimination of by-products originating from different sources. This is especially interesting while

studying heteroleptic precursors with two or more types of ligands that behave differently upon

reaching the surface.

22

2.4.3. Precursor adsorption through ligand decomposition

More complex reactions can occur when a precursor reaches the surface. Amidinate, guanidinate and

disilazide precursors tend to decompose partially during their adsorption on reactive surfaces. These

decomposition mechanisms can take place without direct reaction with surface reactive groups

(hydroxyls) and lead to the release of numerous gas phase by-products that can further readsorb on

the surface and continue to react. This behavior leads to multiple pathways of adsorption in which

desorption, readsorption and rearrangement on the surface are as important in the surface saturation

as the initial reactions. For instance, gas thermolysis of amidinate precursors at normal pressure

shows carbodiimide deinsertion at a temperature as low as 120 °C (see pathway A in Figure 7).177-179

Moreover, dissociation of the acetamidinate ligand around -73°C on a metal surface leads to

adsorbed alkene, cyanamides and alkylamide groups on the surface. The removal of such adsorbed

species can be problematic with water as the oxygen co-reactant and ultimately lead to high carbon

and nitrogen contamination of the deposited films. The reaction mechanisms governing the

decomposition of amidinate ligands are discussed in detail in a later section.

The nature of the surface and temperature are the key factors in the thermodynamics and kinetics of

these surface reactions: for instance on an oxide surface, the density of hydroxyl groups decreases at

higher temperature, which can limit the initial adsorption of the precursor molecule through chemical

bonding or change the reaction mechanism by favoring decomposition of the metal precursor and

CVD type reactions.

Adsorption of the precursor with decomposition of the ligands is not always detrimental: in the case of

disilazides, the decomposition of the precursors through reaction with surface hydroxyl groups can be

used to passivate a reactive surface with trimethylsilyl groups.180, 181 Disilazides can also constitute

potential bimetal precursors for deposition of ternary silicon oxide.121, 182

2.4.4. Partial combustion

According to in situ FT-IR studies, surface impurities and hydroxyl groups can be removed during the

ozone pulse.10 Additionally, the number of hydroxyl groups on the surface diminishes with increasing

deposition temperature. Carbonate and formate surface groups originating from partial combustion of

alkyl substituent by ozone can replace the hydroxyls groups as nucleation sites.9, 10 Also, the

combustion of ligands by ozone creates H2O that can be readsorbed.183 Another alternative reactive

species on the surface is adsorbed oxygen radicals: similarly to ozone molecules, these active

oxygen species can partially combust ligands leading to the decomposition of the metal precursors

during their adsorption. Even though no in situ characterization techniques could formally prove their

presence, active oxygen radicals remain the best explanation available at this point for the

combustive adsorption reaction mechanism of several ozone based processes (see publication VI).

23

2.5 Reactions of precursors with nitrogen donor ligands

2.5.1 Alkylamides

The lone electron pair of the nitrogen atom in the alkylamide ligand is capable of initiating a

nucleophilic attack onto electron-deficient species. Additionally, the weaker M-N bonds strength

compared to the corresponding M-O bonds explain the superior reactivity of alkylamide compared to

alkoxy ligands: with heteroleptic precursors containing both types of ligands, adsorption of the

precursor occurs primarily through reaction with the alkylamide ligands (see publication V).

The length of the alkyl chain within the alkylamide ligand influences greatly the vapor pressure of the

precursor. However, there is no direct correlation with the growth rate of the process. For the same

type of alkylamide ligand, the growth rate depends mainly on the metal. The difference in the strength

of the metal-nitrogen bonds also results in a variation of the number of alkylamide ligands reacting

with surface groups. For instance in publication I, QMS and QCM results show that only one

deuterated ligand (DNEt2) is released during the adsorption of tBuN=Nb(NEt2)3 while 1,7 DNEt2 are

released during the adsorption of tBuN=Ta(NEt2)3 . In addition, the gradual expansion of the metal CN

during the adsorption step can also depend on the strength of the M-N bonds present in the ligands

remaining on the surface.

Alkylamide ligands can undergo partial decomposition reactions in the gas phase or on the surface

after adsorption of the precursor molecule. For the majority of metal alkylamides, gas phase

thermolysis studies reveal that decomposition occurs around 250-300 °C, depending on the length of

the alkyl chains.76 This decomposition often defines the upper limit of the ALD window as the

subsequent parasitic reactions occurring during adsorption above this threshold temperature avert a

saturative growth. Oppositely, surface decomposition reactions do not necessarily destroy the self-

limiting character of a process. They can be slow and remain marginal within the ALD window, or can

lead to unreactive surface intermediates that passivate the surface against further adsorption of the

precursor. However, these reactions can also lead to increased surface roughness and amount of

impurities deposited in the film.

Figure 6 exhibits some initial pathways for the surface decomposition of alkylamide precursors: an

alkylamide ligand can undergo β-hydride elimination that releases an alkylimine in the gas phase and

forms of a M-H bond as observed with in situ FT-IR184 (path A). For instance with Hf(NEtMe)4, the β-

hydride elimination above 270 °C leads to the release of ethylmethyleneimine or methylethyleneimine

and the formation of Hf-H surface species. These Hf-H species can later react with H2O to regenerate

surface –OH groups and release H2.184 QMS and FT-IR results and DFT calculations both indicate

that β-hydride elimination can lead to further intramolecular hydrogen transfer and release of

hydrogenated alkylamide185, 186 (path B). Although possible, such a H transfer between two alkylamide

moieties linked to the same metal center is unlikely compared to hydrogen incorporation from another

surface species (path C), especially on oxide surfaces covered with –OH groups.186, 187 Another

possible pathway requires the scission of a C-N bond and β-hydrogen migration within the same

24

alkylamide ligand to release volatile hydrogenated alkyls, mostly methane (path D). This forms an

imino surface group that was observed experimentally with FT-IR185 and predicted by DFT

calculations186 and can be involved in further decomposition reactions. Most of these reaction

pathways are more noticeable on H terminated metal or silicon surfaces than on hydroxylated oxide

surfaces.184, 188 At higher temperatures (> 300 °C), additional dehydrogenation reactions increase

surface species decomposition thus eliminating the self-limiting character of the deposition and to

higher growth rate and leaving more impurities in the films.75

Figure 6. Potential decomposition pathways for an adsorbed alkylamide molecule (R1 to R5 represent

hydrogen atoms or alkyl substituents).

2.5.2 Hetroleptic alkylamides

In typical ALD conditions, the reactivity of alkoxy/alkylamide precursors is governed by ligand

exchange. The higher reactivity of the M-N bonds of the alkylamide ligands compared to the M-O

bonds of the alkoxy ligands implies preferential reaction of the chemisorbed precursor through an

exchange of the alkylamide ligands (see publication V). Concerning amidinate/alkylamide and

guanidinate/alkylamide precursors, the surface reactions underline the decomposition of the

amidinate or guanidinate ligand.

Within the “ALD window”, heteroleptic cyclopentadienyl/alkylamide precursors can possibly react with

surface hydroxyl groups and release their hydrogenated alkylamide ligands, leaving the surface

covered with Cp groups. The presence of these bulky ligands on the surface limits further access of

precursor molecules to other reactive sites. Alternatively, the Cp ligands can also exchange with

surface hydroxyl groups and leave the surface as a whole hydrogenated Cp. The remaining

alkylamide ligands on the surface can lead to a more closely packed saturated surface and increase

the density and/or the growth rate of the film. At higher temperature, the Cp ligands start to

decompose and form reactive by-products that can readsorb on the surface. These reactive species

25

increase significantly the amount of carbon impurities in the film and remove the self-limiting character

of the surface reactions.

Similarly to the other heteroleptic alkylamides, the main mechanism for adsorption of imido/alkylamide

precursors is ligand exchange with the available reactive surface sites. Depending on the metal, the

charge repartition between the metal and the nitrogen of the imido ligand can allow its reaction with

surface hydroxyl groups during the metal precursor adsorption (see publication I). At high

temperatures, unimolecular decomposition reactions involving the imido ligand can occur: γ-H

migration would lead to the formation of an M=NH group and the release of alkene by-product.

Alternatively, β-H transfer from another alkylamide ligand would simply release a volatile alkylamide.

Both of these by-products have been observed with QMS during temperature-programmed desorption

(TPD) experiments.187

2.5.3 Amidinates

Metal amidinate precursors have been selected for ALD because of their volatility and reactivity with

hydrogen, water, ozone and ammonia. However, the adsorption of amidinates observed with in situ

analytical methods (QCM, QMS, and XPS) involves decomposition of the amidinate ligands when

they chemisorb and react with the surface sites. These decomposition reactions are critical to the

success of a deposition process. Although multiple and highly complex, several of these pathways

have been characterized with in situ analytical methods and/or calculation modeling and are

discussed in the following sections.

Hydrogen transfer decomposition pathways

Similarly to the alkylamides, decomposition of amidinates often involves hydrogen transfer within the

same ligand. Starting from 120 up to 250°C, some QMS studies during the gas phase thermolysis of

copper amidinates have led to a suggestion of carbodiimide deinsertion which is the reverse reaction

for the amidinate synthesis and leads to the formation of a metal-alkyl bond.179 This M-C bond

formation could be a major cause of carbon contamination in the deposited films. However, the

cleavage of a C-C bond at low temperature appears unprobable and carbodiimide by-products were

never observed during actual ALD cycles involving amidinates. Therefore, carbodiimide deinsertion of

amidinate ligands seems very unlikely. Additional XPS and TPD measurements189 of copper

amidinates point toward several other possible pathways displayed in Figure 7: β-hydrogen transfer

within the amidinate ligand can occur through a four member ring reaction (path A), forming a N-H

bond and releasing alkene by-products. Further β-hydrogen transfer leads to the removal of the

hydrogenated substituent R3 and formation of a N≡C triple bond (path D). Formation of this nitrile

species can also result from a six member ring with a δ-hydrogen transfer reaction (path C). Path B

illustrates another possible hydrogen transfer producing a M-H bond as calculated by DFT for cobalt

and nickel amidinates.190 The resulting moieties formed are not reactive with the other metal

precursors thus maintaining the self-limiting behavior of the process. They can later be removed from

the surface by the following oxygen precursor pulse. These decomposition reactions occur on the

surface but can also take place in the gas phase at high temperatures (>350 °C). In this case, the

26

deposition loses its self-limiting behavior and a drastic increase in the amount of impurities

incorporated in the film is observed.189

Figure 7. Potential amidinate hydrogen transfer decomposition pathways for Cu, Co and Ni amidinates.

Mechanisms leading to incorporation of impurities in the film

The decomposition of amidinate is quite complex and involves many intermediates and competitive

reaction steps. Depending on deposition temperature, the nature of the surface and the co-reactants,

some of these reactions can dominate and lead to more or less impurities being incorporated in the

film. The detrimental reaction pathways that have been characterized with in situ surface or gas

phase techniques are displayed in Figure 8.

For copper amidinates, XPS and TPD results189 indicate that upon the initial adsorption of the metal

precursor, the two nitrogen atoms which had a similar chemical environment in the gas phase, start to

differentiate into two distinct types: the amido type of nitrogen which is directly bonded to the metal

atom and the imido type of nitrogen which is double bonded to the central carbon and only weakly

bonded or not bonded to the surface. Surface decomposition starts with a scission of the N-C bond

from either the amido (path A) or the imido side (path B) with a H transfer from the alkyl substituent to

the nitrogen atom, allowing the alkyl substituent to readsorb on the surface as a bridged molecule.134,

189 These bridged alkyl groups can later be removed by a strong co-reactant (ozone, hydrogen

plasma) or dehydrogenate further and lead to carbon impurities in the film. The hydrogen formed this

way can later react with another adsorbed amidinate ligand and release alkylamidine in the gas phase

(path B in Figure 7). The remaining smaller amidinate-like species on the surface can continue to

decompose through β-hydrogen transfer as shown above (path D). Alternatively, another C-N bond

scission would produce bridged imino and alkylamide surface groups. These potential pathways do

27

not involve additional reaction steps with the surrounding reactive surface sites: a hydroxylated

surface would facilitate some reaction steps such as the formation of alkylamidine or surface alkoxy

groups while on the other hand a metal surface would enhance dehydrogenation and decomposition

reactions.

Figure 8. Potential copper amidinate decomposition pathways leading to incorporation of impurities in

the film.

Part of the amidinate ligand remaining on the surface after adsorption can also be hydrolyzed during

the water pulse and release alkylamine, as measured with QMS. This could be problematic while

studying heteroleptic precursors with alkylamide/amidinate ligands because the gaseous fragments of

hydrogenated alkylamides can originate from both the exchange of alkylamide ligands and the

decomposition of amidinate ligands. This renders quantitative interpretation of the reaction

mechanism extremely difficult (see publication V). The same issues arise with ozone processes as

the combustion by-products of amidinates and alkylamides are often the same (NOx). Generalization

of these reaction pathways to all amidinates should be avoided as the chemistry is likely strongly

dependent upon the nature of the metal. Additional in situ studies need therefore to be conducted on

a broader range of amidinates to clarify their surface reaction mechanisms.

2.5.4 Guanidinate

Guanidinate precursors possess a chemical structure very similar to the amidinates with the two

chelate N atoms bonded to the metal center. The presence of an additional alkylamide group linked to

the bridge carbon atom participates in the delocalization of the π bonding of the chelate nitrogen

atoms. For conjugation of this dialkylamide moiety with the NCN unit to occur, it should be sp2-

hybridized. A minimal torsion angle between the NR2 p-orbital and the NCN bridge is also required for

an optimal overlap with the chelate π-orbital of the nitrogen atoms.

28

Sublimation of homoleptic lanthanide guanidinates occurs in the same temperature range (120-165

°C at 0.05 Torr)149 as their amidinate counterparts although the mass of the guanidinate ligands is

higher. This indicates that steric crowding, electronic saturation and intermolecular interactions also

have a strong influence on the thermal behaviour of these precursors. Therefore, the increase in

thermal stability of guanidinate ligands allows deposition of conformal films up to 280°C.146, 191

The decomposition of the guanidinate ligand upon adsorption shares several reaction pathways with

the amidinates: β-hydrogen elimination can occur and form a M-H bond with a release of a

dehydrogenated guanidinate ligand in the gas phase. Carbodiimide deinsertion is also more favorable

for guanidinates compared to the alkyl deinsertion from the amidinates and would result in the transfer

of the exocyclic alkylamide moiety to the metal center. Two of these alkylamide moieties can

subsequently react together to release a hydrogenated alkylamide and an alkylimine.177 With copper

amidinates, this reaction proceeds with a disproportionation of Cu(I) in the guanidinate dimer into

Cu(0) and Cu(II) on the surface above 150 °C.177, 189 These adsorption reactions require many steps

and rearrangement on the surface, and thereby may explain why the saturation of mass observed by

in situ QCM is more gradual for guanidinates and amidinates than for precursors with simpler

adsorption chemistry (V).

Thin films deposited with the guanidinate precursors tend to contain less carbon and nitrogen

impurities than the films deposited at the same conditions from the amidinates. This can be explained

by the increased stability of the guanidinate ligands which diminishes detrimental decomposition

reactions such as dehydrogenation in the gas phase. In addition, carbodiimide deinsertion leads to

the formation of amido complexes. These complexes and the presence of the dialkylamide moiety can

both facilitate the removal of intermediate surface species through nucleophilic attack by reactive

surface groups or water.192

2.5.5 Disilazide

The metal centre of disilazide precursor initially exchanges with a surface –OH group to form

hexamethyldisilazane (HN[Si(CH3)3]2 or HMDS). Therefore, the subsequent reactivity of disilazide

precursors is linked to the behavior of its hydrogenated ligand. Hexamethyldisilazane further reacts

with a hydroxylated surface in a two-step mechanism: first it exchanges with a surface hydroxyl group

and dissociates into a surface-bound trimethylsilyl and a reactive intermediate trimethylsilylamine

(H2NSi(CH3)3). The second step consists of the reaction of the trimethylsilylamine with a neighbouring

–OH group to form another trimethylsilyl surface species (–Si(CH3)3) and release ammonia.193, 194

These timethylsilyl species on the surface are unreactive toward water. As shown in Figure 9, HMDS

reacts mainly with isolated –OH groups and leaves the strongly and weakly H-bonded surface –OH

groups unoccupied.193, 194 The reaction is initially fast as non-hydrogen bonded single and geminal

silanols are consumed. However, further adsorption is hindered because of slower reaction with

geminal silanols that already have one hydroxyl group silylated. At high deposition temperatures,

dehydrogenation of the trimethylsilyl groups can occur and trigger parasitic adsorption reactions,

leaving higher amounts of carbon and silicon impurities in the films.

29

Figure 9. Ligand exchange reactions that occur during the chemisorption of hexamethyldisilazane on

a hydrogenated silicon oxide surface.

30

3. Experimental All the in situ measurements and the depositions were performed with a specially modified

commercial flow-type F-120 SAT ALD reactor manufactured by ASM Microchemistry Ltd that is

integrated with QCM and QMS instruments. Argon or N2 (Oy AGA Ab, 99,999%) were used as the

carrier and purging gas. The gas species present in the reactor during the ALD cycle were

investigated using a Hiden HAL/3F 501 RC QMS with a Faraday cup detector and ionization energy

of 50 or 70 eV. The pressure in the QMS chamber was around 1*10-5 mbar, obtained by differential

pumping through a 100 μm orifice with a turbomolecular pump backed with a mechanical pump. The

mass changes on the substrate were recorded using a Maxtek TM 400 QCM with a sampling rate of

20 Hz. During a typical deposition cycle, the pressure inside the reactor was around 3-5 mbar. 3500

cm2 of soda lime glass substrates was required to obtain good signal intensities with the QMS. All the

in situ reaction mechanism studies were performed on the same oxide surface as the material being

deposited. For instance, tBuN=Nb(NEt2)3 and D2O or O3 processes were studied on Nb2O5 surface.

This allows the study of the ALD reaction mechanisms under stable growth conditions.

The quadrupole mass spectrometer allows the investigation of volatile by-products released during

the ALD process. The evolution in intensity of a few signals at different m/z ratio can be investigated

simultaneously. In order to confirm the actual release of by-products for these signals, background

pulses are necessary. The electron source in the QMS fragments all molecules, i.e also unreacted

metal precursors producing fragments that can correspond to signals with the same m/z ratio as

surface by-products. Accordingly, comparing the intensities of these signals during the background

and ALD pulses indicates the signal intensities measured during the ALD pulses belong partly or fully

to the fragments of by-products originating from the surface reactions. Five reference pulses of the

oxygen containing precursor and the metal precursor are respectively performed before and after the

studied cycles. They are separated by adequate purge time to match the length of the studied ALD

full cycles. This ensures a similar partial pressure of the metal and oxygen precursors during the

background pulses and the ALD cycle pulses.

Quartz crystal microbalance monitors the mass changes occurring on the quartz surface through

variations in the piezoelectric resonance frequency. The resonance frequency of the quartz crystal is

directly proportional to its mass and other factors such as temperature, surface area and shear

modulus. Ten ALD cycles are usually recorded and the frequency changes after the metal precursor

pulse and after a full ALD cycle are measured. Using a ratio of these two frequency changes removes

the influence of external factors and thus corresponds to mass ratios. For reaction mechanisms

involving ligand exchange, the experimental ratios are often compared to calculated mass ratios for

integer number of ligands exchanged. This provides indications on the number and type of ligands

reacting during the adsorption of the metal precursor.

The tBuN=Nb(NEt2)3 and tBuN=Ta(NEt2)3 precursors (I) were furnished by ATMI while (iPrO)3Ti(NiPr-

Me-amd), (iPrO)2Ti(NMe2)2 (IV, VI) and Si2(NHEt)6 (II) were obtained from Air Liquide. The

LiN[Si(CH3)3]2 (III) and Al(CH3)3 (II) were bought from Sigma. Finally, (EtMeN)2Zr(guan-NEtMe2)2 (V)

31

was synthetized by Manish Banerjee, a colleague from the ENHANCE program at the Ruhr-University

Bochum. Table 1 sums up the experimental details of each mechanistic study performed for this

thesis.

Table 1. Summary of processes studies in situ.

Precursor Subl. /Evap. Temp. (°C) at 2-5 mbar

Oxygen sources

Deposition temperatures tested (°C)

Material deposited Publication

tBuN=Nb(NEt2)3 65 (liquid) H2O, O3 250 Nb2O5 I tBuN=Ta(NEt2)3 65 (liquid) H2O, O3 250 Ta2O5 I

Si2(NHEt)6 / Al(CH3)3

65 (liquid) H2O 200 AlxSiyOz II

LiN[Si(CH3)3]2 65 (liquid) O3 200-300 Li2SiO3 III (iPrO)2Ti(NMe2)2 35 (liquid) H2O, O3 275-325 TiO2 IV,VI

(iPrO)3Ti(NiPr-Me-amd) 55 (liquid) H2O, O3 275-325 TiO2 IV,VI

(EtMeN)2Zr(guan-NEtMe2)2

120 (solid) H2O, O3 255 ZrO2 V

32

4. Results and discussion

4.1 Nb2O5 and Ta2O5 Nb2O5 and Ta2O5 thin films possess high-permittivity and have potential applications in DRAM

devices, capacitor dielectrics and as catalyst-supporting oxide materials. The reaction mechanisms of tBuN=Nb(NEt2)3 and

tBuN=Ta(NEt2)3 using D2O or ozone as the co-reactants in ALD were investigated

in situ with QCM and QMS at 250 °C (I).

The tBuN=Nb(NEt2)3 and D2O process exhibits a straightforward ligand exchange reaction as

demonstrated by the formation of D2NtBu and DNEt2 as the main by-products observed with QMS:

one out of three -NEt2 ligands is exchanged during the precursor pulse when the tert-butylimido ligand

remains on the surface. During the following D2O pulse, the remaining ligands on the surface are

released in deuterated form and the surface –OD groups are regenerated, leaving the surface ready

for another cycle to begin.

–OD(s) + tBuN=Nb(NEt2)3 (g) (-O-)Nb(tBuN)(NEt2)2 (s) + DNEt2 (g)

(-O-)Nb(tBuN)(NEt2)2 (s) + 2,5 D2O (g) (-O-)2,5Nb(OD) (s) + D2NtBu (g) + 2 DNEt2 (g)

Interestingly, for the tBuN=Ta(NEt2)3 and D2O process, the QCM and QMS results indicate that in

average 1,7 -NEt2 ligands and 0,3 =NtBu ligands are exchanged during the tBuN=Ta(NEt2)3 pulse:

2,3 –OD (s) + tBuN=Ta(NEt2)3 (g) (-O-)2,3Ta(tBuN)0,7(NEt2)1,3 (s) + 1,7 DNEt2 (g) +0,3 D2NtBu (g)

(-O-)2,3Ta(tBuN)0,7(NEt2)1,3 (s) + 2,5 D2O (g) (-O-)2,5Ta (OD)2,3 (s) + 1,3 DNEt2 (g) + 0,7 D2NtBu (g)

This is surprising as the two precursors contain exactly the same ligands, differing only by the nature

of their group V metal centers. As shown by our quantum chemical calculations, the discrepancy

between the reaction mechanisms can be explained by the differences in the polarity between the Ta-

N and Nb-N bonds: the double bond present in the tantalum precursor is more polar than the one in

the niobium precursor and hence is more likely to react with the surface –OD groups.

Concerning the ozone processes, QCM results point to a molecular adsorption of tBuN=Nb(NEt2)3 on

the surface during the metal precursor pulse while in the case of tBuN=Ta(NEt2)3 a release of at least

two ligands could be deduced. The QMS data indicate typical ozone combustion by-products, such as

CO2 (m/z = 44), CO (m/z = 28), H2O (m/z =18) and NO (m/z = 30) or N2O (m/z = 44, 30). A major

fragment of both types of ligands (m/z = 58) is also observed during the metal precursor and ozone

pulses. Further quantification of these by-products is rendered impossible due to an overlap of

different species with the same m/z value. However, by comparing the shape of the m/z = 58 signal

during the tBuN=Ta(NEt2)3 and tBuN=Nb(NEt2)3 precursor pulses (Figure 10) differences in the

adsorption mechanism of these precursors can be underlined: the sharp rise of the m/z = 58 signal at

the beginning of the tBuN=Ta(NEt2)3 pulse corresponds to a release of by-products originating from

ligand exchange reactions with reactive surface groups during the adsorption of the precursor. This

contribution is absent in the case of tBuN=Nb(NEt2)3 that adsorbs molecularly on the surface.

33

Figure 10. Comparison of the shapes of the signals at m/z = 58 during a metal precursor pulse in the

ozone processes (top curve is tBuN=Nb(NEt2)3 and bottom one is tBuN=Ta(NEt2)3).

4.2 AlxSiyO Binary SiO2 has been difficult to grow by ALD with only water as the oxygen source. By contrast, in

the ALD of ternary metal-silicon-oxides, the other metal oxide activates the ALD reactions of SiO2: as

shown in Figure 11, the growth stops after a few cycle of the Si2(NHEt)6-D2O process whereas it is

sustained in the Al(CH3)3-D2O-Si2(NHEt)6-D2O process. The origins of that peculiarity and the study of

reaction mechanisms beneath are important to optimize and develop processes for the deposition of

silicon based oxides.195, 195-204 Therefore, the reaction mechanisms of the Al(CH3)3-D2O-Si2(NHEt)6-

D2O ALD process were studied in situ with a QCM and QMS at 200 °C (II). Two additional pulsing

sequences were also investigated to assess the specific surface reactivity of Si2(NHEt)6 and Al(CH3)3.

Additional information on the reaction mechanisms was obtained from ex situ characterization of the

deposited films by XRR, XPS, EDX and FT-IR.

Figure 11. QCM data illustrating the supporting effect of aluminium oxide in silicon oxide growth.

34

The main by-products observed with QMS were CDH3 and NDHC2H5. They point to a ligand exchange

reaction with surface hydroxyl groups during the adsorption of Si2(NHEt)6 and Al(CH3)3. On average,

1,3 methyl ligands of Al(CH3)3 and 3,8 –NHEt ligands of Si2(NHEt)6, were released during the metal

precursor pulses, the rest being eliminated during the following D2O pulses. However, the low

reactivity of Si2(NHEt)6 with D2O leaves a small amount of –NHEt ligands present on the surface after

the last D2O pulse. The FT-IR spectra of the film showed the presence of Si–CH3 bonds which

indicates the incorporation of methyl groups from TMA into the film. This could be the result of either

opening of Si-O-Si bridge by TMA or exchange of –CH3 ligands with the few –NHEt ligands remaining

on the surface when the TMA pulse started.

Several pathways were considered for the adsorption of Si2(NHEt)6 on a hydroxylated surface (Figure

12). The high growth rate observed for the Al(CH3)3-D2O-Si2(NHEt)6-D2O process (2.2 Å/cycle) and

the detection of Si-H bonds in the films support the pathway B where both Si atoms in the dimeric

Si2(NHEt)6 precursor molecule bond to the surface with a cleavage of their Si-Si bond. This pathway

also explains why the Si2(NHEt)6-D2O process could not sustain itself: the adsorption of Si2(NHEt)6

forms Si-D surface groups that cannot be eliminated by the D2O pulse. This surface poisoning added

to the low reactivity of silicon precursors renders the deposition of binary SiO2 with water impossible.

However, in the Al(CH3)3-D2O-Si2(NHEt)6-D2O process the high reactivity of TMA induces a better

regeneration of the reactive surface sites and allows the deposition to take place.

Figure 12. Possible reaction pathways for the attachment of Si2(NHEt)6 on a hydroxylated surface.

By integrating the QCM, EDX and XPS results into a quantitative reaction model, the following

reaction mechanism could be drawn:

1.4 Al(CH3)3 + 1.8 –OD + 0.2 –Si(NHC2H5)

–O1.8-Al1.4(CH3)2.4 + 1.8 CH3D + 0.2 –Si(NHC2H5)

35

–O1.8-Al1.4(CH3)2.4+ 0.2 –Si(NHC2H5)

–O1.8-Al1,4(CH3)2.2(NHC2H5)0.2 + 0.2 –Si(CH3)

–O1.8-Al1,4(CH3)2.2(NHC2H5)0.2 + 0.2 –Si(CH3) + 2.4 D2O

–O1.8-Al1,4(OD)2.4 + 2.4 CH3D + 0.2 NDHC2H5 + 0.2 –Si(CH3)

–O1.8-Al1,4(OD)2.4 + 0.5 [Si(NHC2H5)3]2 + 0.2 –Si(CH3)

–O4.2-Al1,4SiD0,5(NHC2H5)1.1 + 0.2 –Si(CH3) + 1.9 NDHC2H5

–O4.2-Al1,4SiD0,5(NHC2H5)1.1 + 0.2 –Si(CH3) + 0.9 D2O

–O4.2-Al1,4SiD0,5(NHC2H5)0.2OD1.8 + 0.2 –Si(CH3) + 0.9 NDHC2H5

This mechanism illustrates the incorporation of Si-D and Si-CH3 bonds in the film and the incomplete

reaction of surface –NHEt ligands with D2O but cannot account for the high amount of carbon

impurities measured. This study demonstrates the importance of the growth surface in ternary oxide

deposition and that parasitic reactions can strongly influence the quality and repeatability of an ALD

process.

4.3 Li2SiO3

Lithium silicates can potentially be used in lithium ion batteries as solid state electrolytes, leading to

all solid state batteries with improved efficiency and safety. LiN[Si(CH3)3]2 was previously used with

ozone by our group to deposit amorphous Li2SiO3.205 Interestingly, the growth rate of the ALD process

and the composition of the deposited film (Li/Si ratio) seem to be highly dependent on the reaction

temperature. To investigate these characteristics, the reaction mechanisms were studied in situ with

QCM and QMS at several temperatures (III). In addition, the chemical bonds present in the deposited

films were identified with ex situ FT-IR.

QMS results reveal typical combustion by-products such as CO2 (m/z = 44), CO (m/z = 28), H2O (m/z

=18) and NO (m/z = 30) being released during the ozone pulse. Signals corresponding to the

fragments of the ligands were detected during the LiN[Si(CH3)3]2 pulse but their low intensities imply

that there are no direct ligand exchange reactions with the hydroxyl groups on the surface. The mass

ratios measured with QCM indicate a non-stoichiometric adsorption of the ligand thus confirming its

decomposition through complex reactions upon reaching the surface. Accordingly, several reaction

pathways were drawn (Figure 13) and DFT calculations were performed to assess the reactivity of

possible reaction intermediate.

36

Figure 13. Possible reaction pathways upon adsorption of LiN[Si(CH3)3]2 on an –OH covered surface.

The FT-IR results suggest that lithium is incorporated into the film as Li+ ions and show the presence

of Li2CO3 and Si-CH3 in the deposited films. The latter originates from an incomplete combustion of –

Si(CH3)3 surface groups during the ozone pulse.

The Li/Si ratio decreases and the growth rate increases at higher deposition temperatures as the

adsorption of HN[Si(CH3)3]2 (intermediate 1 in Figure 13) becomes more efficient and thus more

silicon is incorporated in the film. Faster desorption of other non-reactive intermediates can also play

a role by leaving more surface –OH groups available for further adsorption reactions.

The deposited lithium silicate film continues to react with LiN[Si(CH3)3]2 molecules beyond the

common monolayer adsorption reaction after the end of the ALD cycles. This is due to the high

reactivity of lithium oxide with H2O, especially in binary form but apparently also in lithium silicates: a

“bulk reactivity” occurs where the film bulk absorbs water according to the following reaction (the

underlined species denote the film bulk):

Li2O + H2O 2 LiOH

The LiOH thereby formed can further react with LiN[Si(CH3)3]2 molecules. In addition, CO2 created by

combustion of the LiN[Si(CH3)3]2 by ozone during the process can also react with the Li-OH bonds

present in the film, forming lithium carbonate and water:

2 Li-OH + CO2 Li2CO3 + H2O or Li2O + CO2 Li2CO3

37

These parasitic reactions induce reactions not only on the surface of the growing film but also within

the bulk. Since this publication, other groups have also observed this “bulk reactivity” for several other

lithium containing ternary oxides.182, 206, 207 Accordingly, the films produced with the LiN[Si(CH3)3]2 - O3

process tend to deteriorate rapidly when exposed to air. To prevent these parasitic reactions in the

bulk, the films should be capped to prevent their deterioration by exposure to water and CO2

molecules.

4.4 TiO2

Titanium dioxide (TiO2) thin films have promising applications as dielectric materials for the next

generation dynamic random access memories (DRAMs) and can also be used as photocatalysts. Two

processes for depositing TiO2 with heteroleptic precursors and D2O were studied in situ with QCM

and QMS at 275 °C: Ti(OiPr)3(NiPr-Me-amd) has three isopropoxy ligands and an amidinate ligand

whereas Ti(NMe2)2(OiPr)2 possesses two isopropoxy ligands and two alkylamide ligands. Comparison

of these precursors helps to understand the synergy between different types of ligands and their

effect on the adsorption reaction mechanisms (IV).

For the Ti(NMe2)2(OiPr)2-D2O process, the QCM mass ratios measured suggest either one –NMe2

ligand or one –OiPr ligand exchanged with surface –OD groups, or Ti(NMe2)2(OiPr)2 simply adsorbing

on the surface without ligand exchange. However, the main signals observed with QMS belonged to

fragments of deuterated ligands and allowed the discrimination between these different possibilities:

as only DNMe2 fragments were observed during the metal precursor pulse the adsorption of

Ti(NMe2)2(OiPr)2 occurs through an exchange of one alkylamide ligand:

Ti(NMe2)2(OiPr)2 + –OD -O-Ti(NMe2)(OiPr)2 + DNMe2

-O-Ti(NMe2)(OiPr)2 + 2 D2O TiO2 + –OD + DNMe2 + 2 DOiPr

These results show the importance of monitoring both the surface changes and the gas phase

species during the ALD process in order to draw a coherent reaction mechanism.

Figure 14. Comparison of QCM data during the Ti(OiPr)3(NiPr-Me-amd) and Ti(NMe2)2(OiPr)2 pulses

on hydroxylated surface.

38

Moreover, differences in the conclusions of the QCM and QMS results often reveal more complex

surface reactions: for the Ti(OiPr)3(NiPr-Me-amd)-D2O process the QMS results indicated one out of

three isopropoxy ligands being exchanged during the metal precursor pulse but this did not correlate

with the mass ratios measured with QCM. In addition, the lower intensity of the signals from (NiPr-Me-

amd) and the shape of the QCM mass increase suggested decomposition of the (NiPr-Me-amd)

ligand upon reaching the surface (Figure 14). This partial decomposition was also supported by our

calculation model: the relatively high growth rate observed for this process is incompatible with the

whole (NiPr-Me-amd) ligand remaining on the surface.

In this work, the influence of the other ligands on the release of isopropoxy ligand was demonstrated:

in Ti(NMe2)2(OiPr)2, an alkylamide ligand primarily reacts with the surface and the isopropoxy ligands

did not take part in the precursor adsorption whereas with Ti(OiPr)3(NiPr-Me-amd), the decomposition

of the amidinate ligand allowed one –OiPr ligand to exchange on the surface. For comparison, during

the adsorption of Ti(OiPr)4 two hydrogenated –OiPr ligands are exchanged with the surface hydroxyl

groups208

Figure 15. Comparison of the QCM data for one cycle of the (Me2N)2Ti(OiPr)2-O3 process at 275°C

(dashed line) and 325°C (full line).

The reaction mechanisms of the same precursors with ozone were also studied (VI). The in situ

results obtained at 275 and 325 °C were compared to explain the increase in GR observed in the

Ti(NMe2)2(OiPr)2-O3 process above 300 °C (see publication VI). The QMS results showed the release

of typical combustion by-products (CO2, CO, H2O, NO) during both the ozone pulse and the metal

precursor pulse in the two processes. Accordingly, partial combustion reactions seem to occur also

during the metal precursor adsorption. Various reactive surface sites are possibly involved in these

39

adsorption reactions: active oxygen radicals formed during the previous ozone pulse could remain on

the surface and partially combust the metal precursor ligands. Surface carbonates, formates and

hydroxyl groups originating from readsorption of combustion by-products (H2O and CO2) can also

react with metal precursor through ligand exchange.

Figure 15 shows a comparison of the QCM results obtained for the Ti(NMe2)2(OiPr)2-O3 process at

two temperatures. At 325 °C, the mass increases strongly and continuously during the whole

Ti(NMe2)2(OiPr)2 pulse and then steadily decreases during the following purge. This behavior supports

partial decomposition of the ligands remaining after the Ti(NMe2)2(OiPr)2 pulse at 325 °C while at 275

°C they remain intact. Additionally, the percentage of all by-products with QMS observed during the

metal precursor pulse increased at 325 °C, suggesting that more ligands react and leave the surface

during the adsorption of Ti(NMe2)2(OiPr)2. The higher GR at 325 °C can thus be explained by an

increased reactivity of Ti(NMe2)2(OiPr)2 molecules and a partial decomposition of the ligands.

4.5 ZrO2

ZrO2 is a promising high-k dielectric material for DRAM and MOSFETS applications. The ALD

processes available to deposit this material often use homoleptic metal precursors with poor thermal

stability. To circumvent this issue, heteroleptic precursors have been experimented and heteroleptic

Cp-alkyls, Cp-alkylamides and Cp-halides have shown good thermal stability. In addition, novel

guanidinate precursors, although relatively complex, produce high quality films but their surface

chemistry is mostly unknown. Accordingly, we studied the reaction mechanism of Zr(NEtMe)2(guan-

NEtMe)2 with D2O in situ with the combined QCM-QMS at 225 °C (V).

Figure 16 displays a comparison of the mass increases measured by QCM for the adsorption of

Zr(NEtMe)2(guan-NEtMe)2 and Zr(NEtMe)4 on a hydroxylated surface: in the case of

Zr(NEtMe)2(guan-NEtMe)2, the mass increase is less sharp and a longer time is needed to reach

saturation. This behavior shows the importance of the guanidinate ligands in the adsorption of

Zr(NEtMe)2(guan-NEtMe)2: much like the amidinate precursor of the previous study, the guanidinate

ligand most likely decomposes rapidly upon reaching the surface. This occurs through complex

surface reactions which induce gradual desorption of by-products and reorganization of the remaining

adsorbed molecules thus explaining the slower saturation. Oppositely, the adsorption of the

homoleptic Zr(NEtMe)4 occurs through simple exchange reactions of the alkylamide ligands with

surface –OH groups leading to a much faster saturation. The apparent two-step saturation is probably

due to differences in the mass transport of the metal precursors flowing through different parts of the

reaction chamber cross section: A small portion of the precursor molecules can bypass the glass

array and is responsible for the first step of the QCM mass increase but is not enough to reach

saturation. The main pulse arrives later and completes the saturation of the quartz crystal in the

second step.

40

Figure 16. Comparison of QCM data during the Zr(NEtMe)2(guan-NEtMe)2 and Zr(NEtMe)2 pulses on

hydroxylated surface.

The QMS results indicate a release of by-products from decomposition of the guan-NEtMe ligands

only during the metal precursor pulse. In addition, by-products originating from the deuterated –

NEtMe ligands are only observed during the water pulse. This supports the decomposition of the

guanidinate ligand during the adsorption of Zr(NEtMe)2(guan-NEtMe)2 and the removal of the

remaining alkylamide ligands during the following water pulse. The high growth rate observed for this

process (0.8 Å/cycle) also indicates close packing of the adsorbed molecules on the surface. A partial

decomposition of the guanidinate ligands upon adsorption of Zr(NEtMe)2(guan-NEtMe)2 leaves more

neighboring reactive surface sites available to react and thus lead to a higher number of precursor

molecules adsorbed on the surface.

Despite this partial decomposition of the Zr(NEtMe)2(guan-NEtMe)2 precursor during the process, the

films deposited exhibit remarkably low amount of impurities and the conformality is maintained even in

high aspect ratio trenches. These results emphasize the importance of using new heteroleptic

precursors to produce good quality thin films within a broader temperature range compared to the

commonly used homoleptic precursor. Although the reactions mechanisms occurring in these ALD

processes can be complex, in situ measurements allow their understanding and better control.

41

5. Conclusions This thesis is dedicated to the in situ analysis of some common reaction mechanisms occurring during

ALD processes. The reactivity of metal precursors possessing different types of nitrogen bonded

ligands is discussed. A particular emphasis was put on the typical adsorption mechanisms of metal

precursors such as ligand exchange and ligand decomposition. The in situ studies of homoleptic and

heteroleptic precursors for the deposition of binary and ternary high-k metal oxides conducted in this

thesis are regrouped in Table 2.

For the deposition of Nb2O5 and Ta2O5, the influence of the metal center on the adsorption

mechanism of tBuN=Nb(NEt2)3 and tBuN=Ta(NEt2)3 was highlighted: with the tantalum center the

imido ligand takes part in the adsorption of the precursor whereas it does not react during the

adsorption of the niobium precursor. The differences in polarity and charge repartition between the

two metal-nitrogen bonds explained this variation in the adsorption mechanism of the two otherwise

similar precursors.

The in situ results on the ALD of AlxSiyOz ternary oxides stressed the importance of TMA as a co-

reactant for the growth of silicate with Si2(NHEt)6 and water: without TMA, the low reactivity of the

silicon precursor and the formation of unreactive Si-D bonds on the surface stops the growth. TMA

allows the partial removal of these Si-D bonds and can potentially increase the number of reactive

sites through Si-O-Si bridge opening, allowing the deposition to continue.

The growth characteristics of the LiN[Si(CH3)3]2 – O3 process were linked to the reaction mechanisms

taking place during and after the ALD deposition: the evolution of the GPC, Li/Si ratio and carbon

content in the film with temperature were explained by the increasing efficiency of the adsorption of

HN[Si(CH3)3]2 and the more complete combustion by ozone. Parasitic reactions with water and CO2

occurring on the surface and within the bulk of the grown film were also explicated.

Adsorption through ligand exchange was characterized for alkylamide and isopropyl ligands in several

heteroleptic precursors (tBuN=Nb(NEt2)3, tBuN=Ta(NEt2)3, Ti(NMe2)2(OiPr)2…) within water processes.

Decomposition of metal precursors during their adsorption was also observed for the

Zr(NEtMe)2(guan-NEtMe)2 – D2O and Ti(OiPr)3(NiPr-Me-amd) – D2O processes. Even with hydroxyl

groups present on the surface, the amidinate and guanidinate ligands seem to decompose upon

reaching the surface. Regarding ozone processes, partial combustion was found to occur during both

metal and ozone pulses in the TiO2 processes studied. An increase in the GR with Ti(NMe2)2(OiPr)2

but not with Ti(OiPr)3(NiPr-Me-amd) above 300 °C was explained by the decomposition of the -NMe2

ligands during the metal precursor pulse.

42

Table 2. Summary of in situ reaction mechanisms studied in this thesis.

Metal Precursor Oxygen Precursor

Material Deposited Reaction Mechanism Publications

tBuN=Nb(NEt2)3 D2O Nb2O5 Ligand exchange I tBuN=Nb(NEt2)3 O3 Nb2O5

Molecular adsorption / combustion I

tBuN=Ta(NEt2)3 D2O Ta2O5 Ligand exchange I tBuN=Ta(NEt2)3 O3 Ta2O5

Ligand exchange / combustion I

Al(CH3)3 / Si2(NHEt)6 D2O AlxSiyOz Ligand exchange II LiN[Si(CH3)3]2 O3 Li2SiO3 Combustion III

Ti(NMe2)2(OiPr)2 D2O TiO2 Ligand exchange IV

Ti(OiPr)3(NiPr-Me-amd) D2O TiO2 Ligand decomposition /

ligand exchange IV

Zr(NEtMe)2(guan-NEtMe)2

D2O ZrO2 Ligand decomposition /

ligand exchange V

Ti(NMe2)2(OiPr)2 O3 TiO2 Combustion VI Ti(OiPr)3(NiPr-Me-amd) O3 TiO2 Combustion VI

43

6. References

(1) Ritala, M.; Leskelä, M. Handb. Thin Film Mater. 2002, 1, 103.

(2) Leskelä, M.; Ritala, M. Thin Solid Films 2002, 409, 138.

(3) Ritala, M.; Niinistö, J. Chem. Vap. Deposition 2009, 158.

(4) Miikkulainen, V., Leskelä, M., Ritala, M.; Puurunen, R. L. J. Appl. Phys. (Melville, NY, U. S. ) 2013, 113, 021301/1.

(5) Kessels, E., Profit, H., Ports, S.; van de Sanden, R. At. Layer Deposition Nanostruct. Mater. 2012, 131.

(6) Li, K., Li, S., Li, N., Klein, T.; Dixon, D. Abstracts, Joint 66th Southwest and 62nd Southeast Regional Meeting of the American Chemical Society, New Orleans, LA, United States, December 1-4 2010, SESW.

(7) Kang, S., Rhee, S.; George, S. M. J. Vac. Sci. Technol. , A 2004, 22, 2392.

(8) Du, X., Du, Y.; George, S. M. J. Vac. Sci. Technol. , A 2005, 23, 581.

(9) Goldstein, D. N., McCormick, J. A.; George, S. M. J. Phys. Chem. C 2008, 112, 19530.

(10) Wang, Y., Dai, M., Ho, M., Wielunski, L. S.; Chabal, Y. J. Appl. Phys. Lett. 2007, 90, 022906/1.

(11) Yokoyama, S., Goto, H., Miyamoto, T., Ikeda, N.; Shibahara, K. Appl. Surf. Sci. 1997, 112, 75.

(12) Snyder, M. Q., McCool, B. A., DiCarlo, J., Tripp, C. P.; DeSisto, W. J. Thin Solid Films 2006, 514, 97.

(13) Burton, B. B., Lavoie, A. R.; George, S. M. J. Electrochem. Soc. 2008, 155, D508.

(14) Rahtu, A., Alaranta, T.; Ritala, M. Langmuir 2001, 17, 6506.

(15) Rahtu, A., Kukli, K.; Ritala, M. Chem. Mater. 2001, 13, 817.

(16) Knapas, K.; Ritala, M. Chem. Mater. 2008, 20, 5698.

(17) Knapas, K., Hatanpää, T., Ritala, M.; Leskelä, M. Chem. Mater. 2010, 22, 1386.

(18) Knapas, K., Rahtu, A.; Ritala, M. Langmuir 2010, 26, 848.

(19) Rahtu, A., Alaranta, T.; Ritala, M. Langmuir 2001, 17, 6506.

(20) Swaminathan, S.; McIntyre, P. C. ECS Trans. 2010, 33, 455.

(21) Lee, S. Y., Jeon, C., Kim, S. H., Kim, Y., Jung, W., An, K.; Park, C. Jpn. J. Appl. Phys. 2012, 51, 031102/1.

(22) Li, K., Li, S., Li, N., Klein, T. M.; Dixon, D. A. J. Phys. Chem. C 2011, 115, 18560.

(23) McDonnell, S., Dong, H., Hawkins, J. M., Brennan, B., Milojevic, M., Aguirre-Tostado, F. S., Zhernokletov, D. M., Hinkle, C. L., Kim, J.; Wallace, R. M. Appl. Phys. Lett. 2012, 100, 141606/1.

(24) Brennan, B., Qin, X., Dong, H., Kim, J.; Wallace, R. M. Appl. Phys. Lett. 2012, 101, 211604/1.

44

(25) Tallarida, M., Karavaev, K.; Schmeisser, D. J. Appl. Phys. 2008, 104, 064116/1.

(26) Strehle, S., Schmidt, D., Albert, M.; Bartha, J. W. Chem. Vap. Deposition 2011, 17, 37.

(27) Shin, B., Clemens, J. B., Kelly, M. A., Kummel, A. C.; McIntyre, P. C. Appl. Phys. Lett. 2010, 96, 252907/1.

(28) Wu, Y., Hermkens, P. M., van de Loo, B. W. H., Knoops, H. C. M., Potts, S. E., Verheijen, M. A., Roozeboom, F.; Kessels, W. M. M. J. Appl. Phys. (Melville, NY, U. S. ) 2013, 114, 024308/1.

(29) Dakshinamurthy, S.; Bhat, I. J. Electron. Mater. 1998, 27, 521.

(30) Kumagai, H.; Toyoda, K. Reza Kagaku Kenkyu 1994, 16, 56.

(31) Langereis, E., Heil, S. B. S., Knoops, H. C. M., Keuning, W., van de Sanden, M. C. M.; Kessels, W. M. M. J. Phys. D: Appl. Phys. 2009, 42, 073001/1.

(32) Langereis, E., Heil, S. B. S., Knoops, H. C. M., Keuning, W., Van de Sanden, M. C. M.; Kessels, W. M. M. Bull. - Soc. Vac. Coaters 2010, 36.

(33) Lamagna, L., Wiemer, C., Perego, M., Spiga, S., Rodriguez, J., Santiago Coll, D., Grillo, M. E., Klejna, S.; Elliott, S. D. Chem. Mater. 2012, 24, 1080.

(34) Dendooven, J., Pulinthanathu Sree, S., De Keyser, K., Deduytsche, D., Martens, J. A., Ludwig, K. F.; Detavernier, C. J. Phys. Chem. C 2011, 115, 6605.

(35) Devloo-Casier, K., Dendooven, J., Ludwig, K. F., Lekens, G., D'Haen, J.; Detavernier, C. Appl. Phys. Lett. 2011, 98, 231905/1.

(36) Methaapanon, R., Geyer, S. M., Brennan, S.; Bent, S. F. Chem. Mater. 2013, 25, 3458.

(37) Elliott, S. D. Semicond. Sci. Technol. 2012, 27, 074008/1.

(38) Katamreddy, R., Inman, R., Jursich, G., Soulet, A.; Takoudis, C. J. Electrochem. Soc. 2006, 153, C701.

(39) Wade, C. R., Silvernail, C., Banerjee, C., Soulet, A., McAndrew, J.; Belot, J. A. Mater. Lett. 2007, 61, 5079.

(40) Katamreddy, R., Inman, R., Jursich, G., Soulet, A., Nicholls, A.; Takoudis, C. Thin Solid Films 2007, 515, 6931.

(41) Majumder, P., Katamreddy, R.; Takoudis, C. Electrochem. Solid-State Lett. 2007, 10, H291.

(42) Brazeau, A. L.; Barry, S. T. Chem. Mater. 2008, 20, 7287.

(43) Dezelah, C. L.,IV, El-Kadri, O. M., Szilagyi, I. M., Campbell, J. M., Arstila, K., Niinistö, L.; Winter, C. H. J. Am. Chem. Soc. 2006, 128, 9638.

(44) Elam, J. W., Baker, D. A., Hryn, A. J., Martinson, A. B. F., Pellin, M. J.; Hupp, J. T. J. Vac. Sci. Technol. , A 2008, 26, 244.

(45) Elam, J. W., Baker, D. A., Martinson, A. B. F., Pellin, M. J.; Hupp, J. T. J. Phys. Chem. C 2008, 112, 1938.

(46) Prasittichai, C.; Hupp, J. T. J. Phys. Chem. Lett. 2010, 1, 1611.

45

(47) Kääriäinen, T. O., Lehti, S., Kääriäinen, M. ; Cameron, D. C. Annu. Tech. Conf. Proc. - Soc. Vac. Coaters 2011, 54th, 357.

(48) Xie, Q., Musschoot, J., Deduytsche, D., Van Meirhaeghe, R. L., Detavernier, C., Van den Berghe, S., Jiang, Y., Ru, G., Li, B.; Qu, X. J. Electrochem. Soc. 2008, 155, H688.

(49) Xie, Q., Jiang, Y., Detavernier, C., Deduytsche, D., Van Meirhaeghe, R. L., Ru, G., Li, B.; Qu, X. J. Appl. Phys. 2007, 102, 083521/1.

(50) Alessandri, I., Zucca, M., Ferroni, M., Bontempi, E.; Depero, L. E. Small 2009, 5, 336.

(51) Gougousi, T.; Lacis, J. W. Thin Solid Films 2010, 518, 2006.

(52) Alessandri, I.; Depero, L. E. ACS Appl. Mater. Interfaces 2010, 2, 594.

(53) Bontempi, E., Zanola, P., Gelfi, M., Zucca, M., Depero, L. E., Girault, B., Goudeau, P., Geandier, G., Le Bourhis, E.; Renault, P. Nucl. Instrum. Methods Phys. Res. , Sect. B 2010, 268, 365.

(54) Gerasopoulos, K., Chen, X., Culver, J., Wang, C.; Ghodssi, R. Chem. Commun. (Cambridge, U. K. ) 2010, 46, 7349.

(55) Rose, M., Bartha, J. W.; Endler, I. Appl. Surf. Sci. 2010, 256, 3778.

(56) Lee, C., Kim, J., Son, J. Y., Choi, W.; Kim, H. Appl. Catal. , B 2009, 91, 628.

(57) Lee, C., Kim, J., Gu, G. H., Jo, D., Park, C. G., Choi, W.; Kim, H. Thin Solid Films 2010, 518, 4757.

(58) Rodriguez-Reyes, J. C. F.; Teplyakov, A. V. J. Phys. Chem. C 2008, 112, 9695.

(59) Lee, C., Kim, J., Son, J. Y., Maeng, W. J., Jo, D., Choi, W.; Kim, H. J. Electrochem. Soc. 2009, 156, D188.

(60) Afanas'ev, V. V., Badylevich, M., Stesmans, A., Brammertz, G., Delabie, A., Sionke, S., O'Mahony, A., Povey, I. M., Pemble, M. E., O'Connor, E., Hurley, P. K.; Newcomb, S. B. Appl. Phys. Lett. 2008, 93, 212104/1.

(61) Chang, C. , Chiou, Y. , Chang, Y. , Lee, K. , Lin, T. , Wu, T. , Hong, M.; Kwo, J. Appl. Phys. Lett. 2006, 89, 242911/1.

(62) Chang, Y. C., Chiu, H. C., Lee, Y. J., Huang, M. L., Lee, K. Y., Hong, M., Chiu, Y. N., Kwo, J.; Wang, Y. H. Appl. Phys. Lett. 2007, 90, 232904/1.

(63) Chen, Z., Sarkar, S., Biswas, N.; Misra, V. J. Electrochem. Soc. 2009, 156, H561.

(64) Chiou, Y., Chang, C.; Wu, T. J. Mater. Res. 2007, 22, 1899.

(65) Cho, M., Kim, J. H., Hwang, C. S., Ahn, H., Han, S.; Won, J. Y. Appl. Phys. Lett. 2007, 90, 182907/1.

(66) Delabie, A., Brunco, D. P., Conard, T., Favia, P., Bender, H., Franquet, A., Sioncke, S., Vandervorst, W., Van Elshocht, S., Heyns, M., Meuris, M., Kim, E., McIntyre, P. C., Saraswat, K. C., LeBeau, J. M., Cagnon, J., Stemmer, S.; Tsai, W. J. Electrochem. Soc. 2008, 155, H937.

(67) Deshpande, A., Inman, R., Jursich, G.; Takoudis, C. J. Vac. Sci. Technol. , A 2004, 22, 2035.

(68) Hackley, J. C., Demaree, J. D.; Gougousi, T. J. Vac. Sci. Technol. , A 2008, 26, 1235.

46

(69) Kamiyama, S., Miura, T.; Nara, Y. Electrochem. Solid-State Lett. 2006, 9, G285.

(70) Kim, J. C., Cho, Y. S.; Moon, S. H. Jpn. J. Appl. Phys. 2009, 48, 066515/1.

(71) Kirsch, P. D., Quevedo-Lopez, M. A., Li, H. -., Senzaki, Y., Peterson, J. J., Song, S. C., Krishnan, S. A., Moumen, N., Barnett, J., Bersuker, G., Hung, P. Y., Lee, B. H., Lafford, T., Wang, Q., Gay, D.; Ekerdt, J. G. J. Appl. Phys. 2006, 99, 023508/1.

(72) Kukli, K., Pilvi, T., Ritala, M., Sajavaara, T., Lu, J.; Leskelä, M. Thin Solid Films 2005, 491, 328.

(73) Lee, T., Ko, H., Ahn, J., Park, I., Sim, H., Park, H.; Hwang, H. Jpn. J. Appl. Phys. , Part 1 2006, 45, 6993.

(74) Liu, X., Ramanathan, S., Longdergan, A., Srivastava, A., Lee, E., Seidel, T. E., Barton, J. T., Pang, D.; Gordon, R. G. J. Electrochem. Soc. 2005, 152, G213.

(75) Nyns, L., Delabie, A., Swerts, J., Van Elshocht, S.; De Gendt, S. J. Electrochem. Soc. 2010, 157, G225.

(76) Hausmann, D. M., Kim, E., Becker, J.; Gordon, R. G. Chem. Mater. 2002, 14, 4350.

(77) Kim, S. K.; Hwang, C. S. Electrochem. Solid-State Lett. 2008, 11, G9.

(78) Kim, Y., Koo, J., Han, J., Choi, S., Jeon, H.; Park, C. J. Appl. Phys. 2002, 92, 5443.

(79) Henkel, C., Abermann, S., Bethge, O.; Bertagnolli, E. Semicond. Sci. Technol. 2009, 24, 125013/1.

(80) Meyer, J., Goerrn, P., Bertram, F., Hamwi, S., Winkler, T., Johannes, H., Weimann, T., Hinze, P., Riedl, T.; Kowalsky, W. Adv. Mater. (Weinheim, Ger. ) 2009, 21, 1845.

(81) Yun, S. J., Lim, J. W.; Lee, J. Electrochem. Solid-State Lett. 2004, 7, F81.

(82) Yun, S. J., Lim, J. W.; Lee, J. H. Electrochem. Solid-State Lett. 2005, 8, F47.

(83) Bethge, O., Abermann, S., Henkel, C., Straif, C. J., Hutter, H.; Bertagnolli, E. J. Electrochem. Soc. 2009, 156, G168.

(84) Hausmann, D. M.; Gordon, R. G. J. Cryst. Growth 2003, 249, 251.

(85) Kim, I., Tuller, H. L., Kim, H.; Park, J. Appl. Phys. Lett. 2004, 85, 4705.

(86) Maeng, W. J.; Kim, H. Electrochem. Solid-State Lett. 2006, 9, G191.

(87) Maeng, W. J., Park, S.; Kim, H. J. Vac. Sci. Technol. , B: Microelectron. Nanometer Struct. --Process. , Meas. , Phenom. 2006, 24, 2276.

(88) Maeng, W. J., Lim, S. J., Kwon, S.; Kim, H. Appl. Phys. Lett. 2007, 90, 062909/1.

(89) Kim, H., Sohn, S., Jung, D., Maeng, W. J., Kim, H., Kim, T. S., Hahn, J., Lee, S., Yi, Y.; Cho, M. Org. Electron. 2008, 9, 1140.

(90) Potts, S. E., Keuning, W., Langereis, E., Dingemans, G., van de Sanden, M. C. M.; Kessels, W. M. M. J. Electrochem. Soc. 2010, 157, P66.

(91) Hausmann, D. M., de Rouffignac, P., Smith, A., Gordon, R.; Monsma, D. Thin Solid Films 2003, 443, 1.

47

(92) Banerjee, P., Perez, I., Henn-Lecordier, L., Lee, S. B.; Rubloff, G. W. Nat. Nanotechnol. 2009, 4, 292.

(93) Jeon, S.; Park, S. J. Electrochem. Soc. 2010, 157, H1101.

(94) Kim, H. K., Kim, J. Y., Park, J. Y., Kim, Y., Kim, Y. D., Jeon, H.; Kim, W. M. J. Korean Phys. Soc. 2002, 41, 739.

(95) Kim, J. Y., Kim, Y.; Jeon, H. Jpn. J. Appl. Phys. , Part 2 2003, 42, L414.

(96) Kim, J. Y., Kim, H. K., Kim, Y., Kim, Y. D., Kim, W. M.; Jeon, H. J. Korean Phys. Soc. 2002, 40, 176.

(97) Kim, J.; Yong, K. Electrochem. Solid-State Lett. 2004, 7, F35.

(98) Becker, J. S., Kim, E.; Gordon, R. G. Chem. Mater. 2004, 16, 3497.

(99) Cho, S., Lee, K., Song, P., Jeon, H.; Kim, Y. Jpn. J. Appl. Phys. , Part 1 2007, 46, 4085.

(100) Kim, E.; Kim, D. Electrochem. Solid-State Lett. 2006, 9, C123.

(101) Consiglio, S., Zeng, W., Berliner, N.; Eisenbraun, E. T. J. Electrochem. Soc. 2008, 155, H196.

(102) Jeong, W., Ko, Y., Bang, S., Lee, S.; Jeon, H. J. Korean Phys. Soc. 2010, 56, 905.

(103) Wu, Y. Y., Kohn, A.; Eizenberg, M. J. Appl. Phys. 2004, 95, 6167.

(104) Bae, N., Na, K., Cho, H., Park, K., Boo, S., Bae, J.; Lee, J. Jpn. J. Appl. Phys. , Part 1 2006, 45, 9072.

(105) Kim, H., Detavenier, C., Van der Straten, O., Rossnagel, S. M., Kellock, A. J.; Park, D. -. J. Appl. Phys. 2005, 98, 014308/1.

(106) Xie, Q., Musschoot, J., Detavernier, C., Deduytsche, D., Van Meirhaeghe, R. L., Van den Berghe, S., Jiang, Y., Ru, G., Li, B.; Qu, X. Microelectron. Eng. 2008, 85, 2059.

(107) Langereis, E., Knoops, H. C. M., Mackus, A. J. M., Roozeboom, F., van de Sanden, M. C. M.; Kessels, W. M. M. J. Appl. Phys. 2007, 102, 083517/1.

(108) Dezelah, C. L., El-Kadri, O. M., Kukli, K., Arstila, K., Baird, R. J., Lu, J., Niinistö, L.; Winter, C. H. J. Mater. Chem. 2007, 17, 1109.

(109) Katamreddy, R., Wang, Z., Omarjee, V., Rao, P. V., Dussarrat, C.; Blasco, N. ECS Trans. 2009, 25, 217.

(110) Blanquart, T., Niinistö, J., Gavagnin, M., Longo, V., Heikkilä, M., Puukilainen, E., Pallem, V. R., Dussarrat, C., Ritala, M.; Leskelä, M. RSC Adv. 2013, 3, 1179.

(111) Nam, W.; Rhee, S. Chem. Vap. Deposition 2004, 10, 201.

(112) Triyoso, D. H., Hegde, R. I., Grant, J., Fejes, P., Liu, R., Roan, D., Ramon, M., Werho, D., Rai, R., La, L. B., Baker, J., Garza, C., Guenther, T., White, B. E., Jr.; Tobin, P. J. J. Vac. Sci. Technol. , B: Microelectron. Nanometer Struct. --Process. , Meas. , Phenom. 2004, 22, 2121.

(113) Lim, B. S., Rahtu, A., de Rouffignac, P.; Gordon, R. G. Appl. Phys. Lett. 2004, 84, 3957.

(114) Inman, R., Schuetz, S. A., Silvernail, C. M., Balaz, S., Dowben, P. A., Jursich, G., McAndrew, J.; Belot, J. A. Mater. Chem. Phys. 2007, 104, 220.

48

(115) Wang, T.; Ekerdt, J. G. Chem. Mater. 2010, 22, 3798.

(116) de Rouffignac, P.; Gordon, R. G. Chem. Vap. Deposition 2006, 12, 152.

(117) Jones, A. C., Aspinall, H. C., Chalker, P. R., Potter, R. J., Kukli, K., Rahtu, A., Ritala, M.; Leskelä, M. Mater. Sci. Eng. , B 2005, B118, 97.

(118) Kukli, K., Ritala, M., Pilvi, T., Sajavaara, T., Leskelä, M., Jones, A. C., Aspinall, H. C., Gilmer, D. C.; Tobin, P. J. Chem. Mater. 2004, 16, 5162.

(119) Scarel, G., Wiemer, C., Tallarida, G., Spiga, S., Seguini, G., Bonera, E., Fanciulli, M., Lebedinskii, Y., Zenkevich, A., Pavia, G., Fedushkin, I. L., Fukin, G. K.; Domrachev, G. A. J. Electrochem. Soc. 2006, 153, F271.

(120) Scarel, G., Wiemer, C., Fanciulli, M., Fedushkin, L., Fukin, G. K., Domrachev, G. A., Lebedinskii, Y., Zenkevich, A.; Pavia, G. Z. Anorg. Allg. Chem. 2007, 633, 2097.

(121) Hämäläinen, J., Munnik, F., Hatanpää, T., Holopainen, J., Ritala, M.; Leskelä, M. J. Vac. Sci. Technol. , A 2012, 30, 01A106/1.

(122) Rees, W. S., Jr., Green, D. M., Anderson, T. J., Bretschneider, E., Pathangey, B., Park, C.; Kim, J. J. Electron. Mater. 1992, 21, 361.

(123) Lim, B. S., Rahtu, A.; Gordon, R. G. Nat. Mater. 2003, 2, 749.

(124) Li, Z., Gordon, R. G., Farmer, D. B., Lin, Y.; Vlassak, J. Electrochem. Solid-State Lett. 2005, 8, G182.

(125) Lee, H., Kim, W., Lee, J. W., Kim, J., Heo, K., Hwang, I. C., Park, Y., Hong, S.; Kim, H. J. Electrochem. Soc. 2010, 157, D10.

(126) Lee, H.; Kim, H. Electrochem. Solid-State Lett. 2006, 9, G323.

(127) Kim, J., Lee, H., Lansalot, C., Dussarrat, C., Gatineau, J.; Kim, H. Jpn. J. Appl. Phys. 2010, 49, 05FA10/1.

(128) Wang, H., Gordon, R. G., Alvis, R.; Ulfig, R. M. Chem. Vap. Deposition 2009, 15, 312.

(129) Li, H., Farmer, D. B., Gordon, R. G., Lin, Y.; Vlassak, J. J. Electrochem. Soc. 2007, 154, D642.

(130) Farmer, D. B.; Gordon, R. G. J. Appl. Phys. 2007, 101, 124503/1.

(131) Li, Z., Rahtu, A.; Gordon, R. G. J. Electrochem. Soc. 2006, 153, C787.

(132) Kucheyev, S. O., Biener, J., Baumann, T. F., Wang, Y. M., Hamza, A. V., Li, Z., Lee, D. K.; Gordon, R. G. Langmuir 2008, 24, 943.

(133) Seitz, O., Dai, M., Aguirre-Tostado, F. S., Wallace, R. M.; Chabal, Y. J. J. Am. Chem. Soc. 2009, 131, 18159.

(134) Ma, Q., Guo, H., Gordon, R. G.; Zaera, F. Chem. Mater. 2010, 22, 352.

(135) Dai, M., Kwon, J., Halls, M. D., Gordon, R. G.; Chabal, Y. J. Langmuir 2010, 26, 3911.

(136) De Rouffignac, P., Park, J.; Gordon, R. G. Chem. Mater. 2005, 17, 4808.

(137) Lee, B., Park, T. J., Hande, A., Kim, M. J., Wallace, R. M., Kim, J., Liu, X., Yi, J. H., Li, H., Rousseau, M., Shenai, D.; Suydam, J. Microelectron. Eng. 2009, 86, 1658.

49

(138) Abermann, S., Henkel, C., Bethge, O., Pozzovivo, G., Klang, P.; Bertagnolli, E. Appl. Surf. Sci. 2010, 256, 5031.

(139) Kwon, J., Dai, M., Halls, M. D., Langereis, E., Chabal, Y. J.; Gordon, R. G. J. Phys. Chem. C 2009, 113, 654.

(140) Päiväsaari, J., Putkonen, M., Sajavaara, T.; Niinistö, L. J. Alloys Compd. 2004, 374, 124.

(141) Rische, D., Parala, H., Baunemann, A., Thiede, T.; Fischer, R. Surf. Coat. Technol. 2007, 201, 9125.

(142) Coyle, J. P., Monillas, W. H., Yap, G. P. A.; Barry, S. T. Inorg. Chem. 2008, 47, 683.

(143) Coyle, J. P., Johnson, P. A., DiLabio, G. A., Barry, S. T.; Mueller, J. Inorg. Chem. 2010, 49, 2844.

(144) Chen, T., Xu, C., Baum, T. H., Hendrix, B. C., Cameron, T., Roeder, J. F.; Stender, M. PCT Int. Appl. 2007, 2006-US62709; 2006-810578P, 47.

(145) Kukli, K., Ritala, M., Kemell, M.; Leskelä, M. J. Electrochem. Soc. 2010, 157, D35.

(146) Thiede, T. B., Krasnopolski, M., Milanov, A. P., de los Arcos, T., Ney, A., Becker, H., Rogalla, D., Winter, J., Devi, A.; Fischer, R. A. Chem. Mater. 2011, 23, 1430.

(147) Kukli, K., Niinistö, J., Tamm, A., Ritala, M.; Leskelä, M. J. Vac. Sci. Technol. , B: Microelectron. Nanometer Struct. --Process. , Meas. , Phenom. 2009, 27, 226.

(148) Milanov, A., Bhakta, R., Baunemann, A., Becker, H., Thomas, R., Ehrhart, P., Winter, M.; Devi, A. Inorg. Chem. 2006, 45, 11008.

(149) Edelmann, F. T. Chem Soc Rev 2009, 38, 2253.

(150) Edelmann, F. T. Chem. Soc. Rev. 2012, 41, 7657.

(151) Blanquart, T., Niinistö, J., Gavagnin, M., Longo, V., Pallem, V. R., Dussarrat, C., Ritala, M.; Leskelä, M. Chem. Mater. 2012, 24, 3420.

(152) Seo, M., Min, Y., Kim, S. K., Park, T. J., Kim, J. H., Na, K. D.; Hwang, C. S. J. Mater. Chem. 2008, 18, 4324.

(153) Pallem, V. R.; Dussarrat, C. PCT Int. Appl. 2012, 2011-US31360; 2010-981872, 38.

(154) Niinistö, J., Kukli, K., Tamm, A., Putkonen, M., Dezelah, C. L., Niinistö, L., Lu, J., Song, F., Williams, P., Heys, P. N., Ritala, M.; Leskelä, M. J. Mater. Chem. 2008, 18, 3385.

(155) Potts, S. E., Carmalt, C. J., Blackman, C. S., Abou-Chahine, F., Leick, N., Kessels, W. M. M., Davies, H. O.; Heys, P. N. Inorg. Chim. Acta 2010, 363, 1077.

(156) Carta, G., Rossetto, G., Sitran, S., Zanella, P., Crociani, L., Zherikova, K. V., Morozova, N. B., Gelfond, N. V., Semyannikov, P. P., Yakovkina, L. V., Smirnova, T. P.; Igumenov, I. K. Proc. - Electrochem. Soc. 2005, 2005-09, 260.

(157) Sreenivasan, R., Sugawara, T., Saraswat, K. C.; McIntyre, P. C. Appl. Phys. Lett. 2007, 90, 102101/1.

(158) Kim, J. Y., Lee, K. W., Park, H. O., Kim, Y. D., Jeon, H.; Kim, Y. J. Korean Phys. Soc. 2004, 45, 1069.

50

(159) Kumar, S., Greenslit, D., Chakraborty, T.; Eisenbraun, E. T. J. Vac. Sci. Technol. , A 2009, 27, 572.

(160) Choi, B. H., Lim, Y. H., Lee, J. H., Kim, Y. B., Lee, H.; Lee, H. K. Microelectron. Eng. 2010, 87, 1391.

(161) Li, Z., Gordon, R. G., Farmer, D. B., Lin, Y.; Vlassak, J. Electrochem. Solid-State Lett. 2005, 8, G182.

(162) Becker, J. S., Suh, S., Wang, S.; Gordon, R. G. Chem. Mater. 2003, 15, 2969.

(163) Becker, J. S.; Gordon, R. G. Appl. Phys. Lett. 2003, 82, 2239.

(164) Rugge, A., Becker, J. S., Gordon, R. G.; Tolbert, S. H. Nano Lett. 2003, 3, 1293.

(165) Miikkulainen, V., Suvanto, M.; Pakkanen, T. A. Chem. Mater. 2007, 19, 263.

(166) Miikkulainen, V., Suvanto, M.; Pakkanen, T. A. Thin Solid Films 2008, 516, 6041.

(167) Miikkulainen, V., Suvanto, M.; Pakkanen, T. A. Chem. Vap. Deposition 2008, 14, 71.

(168) Miikkulainen, V., Suvanto, M., Pakkanen, T. A., Siitonen, S., Karvinen, P., Kuittinen, M.; Kisonen, H. Surf. Coat. Technol. 2008, 202, 5103.

(169) Blanquart, T., Longo, V., Niinistö, J., Heikkilä, M., Kukli, K., Ritala, M.; Leskelä, M. Semicond. Sci. Technol. 2012, 27, 074003/1.

(170) Blanquart, T., Niinistö, J., Heikkilä, M., Sajavaara, T., Kukli, K., Puukilainen, E., Xu, C., Hunks, W., Ritala, M.; Leskelä, M. Chem. Mater. 2012, 24, 975.

(171) Xu, K., Milanov, A. P., Parala, H., Wenger, C., Baristiran-Kaynak, C., Lakribssi, K., Toader, T., Bock, C., Rogalla, D., Becker, H., Kunze, U.; Devi, A. Chem. Vap. Deposition 2012, 18, 27.

(172) Xu, K., Milanov, A. P.; Devi, A. ECS Trans. 2009, 25, 625.

(173) Eleter, M., Daniele, S., Brize, V., Dubourdieu, C., Lachaud, C., Blasco, N.; Pinchart, A. ECS Trans. 2009, 25, 151.

(174) Blanquart, T., Niinistö, J., Aslam, N., Banerjee, M., Tomczak, Y., Gavagnin, M., Longo, V., Puukilainen, E., Wanzenböck, H. D., Kessels, W. M. M., Devi, A., Hoffmann-Eifert, S., Ritala, M.; Leskelä, M. Chem. Mater. 2013, 25, 3088.

(175) Xu, K., Ranjith, R., Laha, A., Parala, H., Milanov, A. P., Fischer, R. A., Bugiel, E., Feydt, J., Irsen, S., Toader, T., Bock, C., Rogalla, D., Osten, H., Kunze, U.; Devi, A. Chem. Mater. 2012, 24, 651.

(176) Shirazi, M.; Elliott, S. D. Chem. Mater. 2013, 25, 878.

(177) Coyle, J. P., Johnson, P. A., DiLabio, G. A., Barry, S. T.; Mueller, J. Inorg. Chem. (Washington, DC, U. S. ) 2010, 49, 2844.

(178) Barry, S. T. Coord. Chem. Rev. 2013, 257, 3192.

(179) Brazeau, A. L., Wang, Z., Rowley, C. N.; Barry, S. T. Inorg. Chem. 2006, 45, 2276.

(180) Park, K.; Yao, Q. Abstracts of Papers, 234th ACS National Meeting, Boston, MA, United States, August 19-23, 2007 2007, ORGN.

51

(181) Yoshimura, Y., Kumamoto, H., Baba, A., Takeda, S.; Tanaka, H. Nucleic Acids Res. Suppl. 2003, 3, 17.

(182) Ostreng, E., Sonsteby, H. H., Sajavaara, T., Nilsen, O.; Fjellvåg, H. J. Mater. Chem. C 2013, 1, 4283.

(183) Comstock, D. J.; Elam, J. W. Chem. Mater. 2012, 24, 4011.

(184) Li, K., Li, S., Li, N., Klein, T.; Dixon, D. Abstracts, Joint 66th Southwest and 62nd Southeast Regional Meeting of the American Chemical Society, New Orleans, LA, United States, December 1-4 2010, SESW.

(185) Kan, B., Boo, J., Lee, I.; Zaera, F. J. Phys. Chem. A 2009, 113, 3946.

(186) Rodriguez-Reyes, J. C. F.; Teplyakov, A. V. J. Appl. Phys. 2008, 104, 084907/1.

(187) Kim, T.; Zaera, F. J. Phys. Chem. C 2011, 115, 8240.

(188) Kelly, M. J., Han, J. H., Musgrave, C. B.; Parsons, G. N. Chem. Mater. 2005, 17, 5305.

(189) Ma, Q., Lee, I.; Zaera, F. Abstracts of Papers, 239th ACS National Meeting, San Francisco, CA, United States, March 21-25, 2010 2010, COLL.

(190) Wu, J., Li, J., Zhou, C., Lei, X., Gaffney, T., Norman, J. A. T., Li, Z., Gordon, R.; Cheng, H. Organometallics 2007, 26, 2803.

(191) Barry, S. T., Gordon, P. G., Ward, M. J., Heikkilä, M. J., Monillas, W. H., Yap Glenn, P. A., Ritala, M.; Leskela, M. Dalton Trans 2011, 40, 9425.

(192) Kim, T., Yao, Y., Coyle, J. P., Barry, S. T.; Zaera, F. Chem. Mater. 2013, 25, 3630.

(193) Haukka, S.; Root, A. J. Phys. Chem. 1994, 98, 1695.

(194) Gun'ko, V. M., Vedamuthu, M. S., Henderson, G. L.; Blitz, J. P. J. Colloid Interface Sci. 2000, 228, 157.

(195) Senzaki, Y., Park, S., Tweet, D., Conley, J. F., Jr.; Ono, Y. Mater. Res. Soc. Symp. Proc. 2004, 811, 217.

(196) Rittersma, Z. M., Roozeboom, F., Verheijen, M. A., van Berkum, J. G. M., Dao, T., Snijders, J. H. M., Vainonen-Ahlgren, E., Tois, E., Tuominen, M.; Haukka, S. J. Electrochem. Soc. 2004, 151, C716.

(197) Kamiyama, S., Miura, T., Nara, Y.; Arikado, T. Electrochem. Solid-State Lett. 2005, 8, G215.

(198) Kamiyama, S., Miura, T.; Nara, Y. Electrochem. Solid-State Lett. 2005, 8, F37.

(199) Duenas, S., Castan, H., Garcia, H., Barbolla, J., Kukli, K., Ritala, M.; Leskelä, M. Span. Conf. Electron Devices, Proc. , 5th 2005, 45.

(200) Senzaki, Y., Park, S., Bartholomew, L.; Chatham, H. Proc. - Electrochem. Soc. 2004, 2004-01, 264.

(201) Nam, W.; Rhee, S. Stud. Surf. Sci. Catal. 2006, 159, 373.

(202) Zhang, H.; Solanki, R. J. Electrochem. Soc. 2001, 148, F63.

52

(203) Gladfelter, W. L.; Zhong, L. Abstracts of Papers, 227th ACS National Meeting, Anaheim, CA, United States, March 28-April 1, 2004 2004, INOR.

(204) Zhong, L., Daniel, W. L., Zhang, Z., Campbell, S. A.; Gladfelter, W. L. Chem. Vap. Deposition 2006, 12, 143.

(205) Hämälainen, J., Ihanus, J., Sajavaara, T., Ritala, M.; Leskelä, M. J. Electrochem. Soc. 2011, 158, P15.

(206) Ostreng, E., Vajeeston, P., Nilsen, O.; Fjellvåg, H. RSC Adv. 2012, 2, 6315.

(207) Miikkulainen, V., Nilsen, O., Laitinen, M., Sajavaara, T.; Fjellvåg, H. RSC Adv. 2013, 3, 7537.

(208) Rahtu, A.; Ritala, M. Chem. Vap. Deposition 2002, 8, 21.