MOCVD of Tungsten and Molybdenum Nitrides

Home , Tungsten hexachloride, Tungsten nitride

Daniel Rische; Doktorarbeit 2007

MOCVD of Tungsten and Molybdenum Nitrides

Dissertation zur Erlangung der Doktorwürde der Fakultät für Chemie der Ruhr-Universität Bochum

vorgelegt von Diplom-Chemiker Daniel Rische aus Bochum

Referenten: Prof. Dr. Roland A. Fischer Prof. Dr. Christof Wöll

Die vorliegende Arbeit entstand in der Zeit von Juli 2004 bis November 2007 am Lehrstuhl für Anorganische Chemie II der Ruhr-Universität Bochum

Mein besonderer Dank gilt an Prof. Dr. Roland A Fischer, für die interessante Aufgabenstellung, das in mich gesetzte Vertrauen bei der Bearbeitung eines interessanten und herausfordernden Forschungsthemas, und den Rat mit dem er mir zur Seite stand.

Danksagung

Juniorprof. Dr. Anjana Devi Anjana und Harish danke ich für die Unterstützung in Form & Dr. Harish Parala von konstruktiver Kritik in vielen Bereichen und der Unterstützung bei XRD und TG/DTA Andrian Milanov Danke für die Unterstützung bei TG/DTA Messungen , sowie häufigen Gedankenaustausch Daniela Bekermann Danke für die tatkräftige Unterstützung bei Syntheseproblemen Sabine Masukowitz Vielen Dank für die Unterstützung in allen organisatorischen Fragen und bei dem Umgang mit dem RUB’schen Büroapparat. Stephan Spöllmann Stephan danke ich für die tatkräftige Unterstützung bei den Experimenten am Aixtron-Reaktor Manuela Winter Dank an Manuela für die Messungen der Einkristallstrukturen und Hilfe bei der Lösung derselben Dr. Rolf Neuser Herrn Neuser danke ich für die SEM- und EDX Messungen.

H.C. Starck GmbH Danke an die H.C. Starck GmbH für die großzügige Chemikalienspende. Andreas Kempter Für den gemeinsamen Weg durch das Studium, von Vordiplom über Diplom bis jetzt.

Desweiteren möchte ich mich bei meinem gesamten Lehrstuhl für die schöne Zeit in der Gruppe bedanken. Dazu gehören Saeed Amirjalayer, Dr. Raghunandan Bhakta, Thomas Cadenbach, Mirza Cokoja, Rolf Deibert, Daniel Esken, Lina Freitag, Dr. Eliza Gemel, Malte Hellwig, Stephan Hermes, Ursula Herrmann, Todor Hikov, Heike Kampschulte, Andreas Kempter, Dr. Jayaprakash Khanderi, Dr. Emmanuel Lamouroux, Dr. Eva Maile, Mikhail Meilikhov, Andrian Milanov, Maike Müller, Dr. Rochus Schmid, Felicitas Schröder, Dr. Jelena Sekulic, Stephan Spöllmann, Tobias Thiede, Dr. Maxim Tafipolsky, Tim Wilmsen, Manuela Winter, Dr. Wenhua Zhang und Xiaoning Zhang Zudem danke ich allen die meinen Dank verdienen, die ich aber in meiner Liste vergessen habe.

“Science is like sex: Something useful comes out, but that’s not the reason we are doing it.” Richard P. Feynman

Für Jenny

Abbreviations

ALD: Atomic Layer Deposition tBu: tert- Butyl CVD: Chemical Vapor Deposition Cy: Cyclohexyl DME: 1,2-Dimethoxyethane DTA: Differential Thermal Analysis EI-MS: Electron Ionization Mass Spectroscopy Et: Ethyl FT-IR Fourier-Transform-Infrared Spectroscopy GC: Gas Chromatography MBE: Molecular Beam Epitaxy Me: Methyl Mes: Mesityl MOCVD: Metal Organic Chemical Vapor Deposition MS: Mass Spectroscopy NMR: Nuclear Magnetic Resonance Ph: Phenyl iPr: iso-Propyl PVD: Physical Vapor Deposition py: Pyridine SEM: Scanning Electron Microscopy SIMS: Secondary Ion Mass Spectroscopy S/N-ratio Signal to Noise ratio SNMS: Secondary Neutral Mass Spectroscopy THF: Tetrahydrofurane TMEDA: N,N,N’,N’-Tetramethylethylenediamine TG: Thermogravimetric Analysis XRD: X-Ray Diffraction

Table of Contents

1. Introduction ...... 1 1.1. Motivation and Goals...... 1 1.2. Tungsten Nitride Materials...... 2 1.3. Molybdenum Nitride Materials...... 3 1.4. Applications of Tungsten and Molybdenum Nitrides...... 4 1.4.1. Schottky Diodes ...... 4 1.4.2. Conductive Diffusion Barriers ...... 5 1.4.3. Hard Coatings...... 7 1.4.4. Catalysis ...... 7 1.5. Thin Film Deposition Techniques via gas Phase ...... 7 1.5.1. Physical Vapour Deposition (PVD) ...... 7 1.5.2. Chemical Vapour Deposition (CVD)...... 9 1.5.3. Atomic Layer Deposition (ALD) ...... 13 2. State of the Art...... 15 2.1. Precursors for MOCVD and ALD ...... 15 2.1.1. Tungsten nitride...... 15 2.1.2. Molybdenum nitride...... 20 2.2. Guanidinato ligands...... 22 3. Synthesis of the Tungsten Compounds...... 25

3.1. [WCl2(Nt-Bu)2py2] (1) ...... 25

3.2. [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2] (3) ...... 27

3.3. [W(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] (4) ...... 31

3.4. [W(Nt-Bu)2Cl{(Ni-Pr)2CNEt2}] (5)...... 33

3.5. [W(Nt-Bu)2Cl{(NCy)2CNEt2}] (6) ...... 35

3.6. [W(Nt-Bu)2NMe2{(Ni-Pr)2CNi-Pr2}] (7)...... 37

3.7. [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] (8)...... 39

3.8. [W(Nt-Bu)2(N3){(Ni-Pr)2CNi-Pr2}] (9) ...... 39

3.9. [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] (10)...... 40

3.10. [W(Nt-Bu)2Cl{NC(NMe2)2}]2 (11) ...... 42

3.11. [W(Nt-Bu)2(N3){NC(NMe2)2}]2 (12)...... 45 2 3.12. [(W(Nt-Bu)2(N3)(µ -N3)py)]2 (13) ...... 47 3.13. Conclusions for Chapter 3...... 50

Table of Contents

4. Synthesis of the Molybdenum Compounds...... 52 4.1. Starting Compound for the Molybdenum synthesis...... 52

4.2. [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] (15) ...... 55

4.3. [Mo(Nt-Bu)2I{(Ni-Pr)2CNMe2}] (16)...... 55

4.4. [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}] (17) ...... 57 4.5. Conclusions for Chapter 4...... 59 5. MOCVD-Experiments of Tungsten Nitride...... 60 5.1. Thermal Characterization of the precursors...... 60 5.2. Depositions on the home-built MOCVD reactor ...... 61 5.2.1. XRD analysis and surface morphology...... 62 5.2.2 Composition of the films (SNMS analysis) ...... 68 5.2.3 Growth rates & resistivities...... 73 5.3. Characterization of the Exhaust gases...... 74 5.4. Experiments on the Aixtron 200 RF reactor ...... 77 5.4.1 The Reactor ...... 77 5.4.2 The Bubbler Design ...... 79 5.4.3 The Experiments ...... 79 5.5. Comparison of the Guanidinato Precursors to Imido/Amido Precursors...... 84 5.6. Conclusions for Chapter 5...... 85 6. MOCVD-Experiments of Molybdenum Nitride...... 87 6.1. Thermal Characterization of the precursors...... 87 6.2. Depositions on the home-built MOCVD reactor ...... 88 6.2.1 XRD analysis and morphology ...... 89 6.2.2 Composition of the films (SNMS analysis) ...... 94 6.2.3 Growth rates & resistivities...... 97 6.3. Conclusions for Chapter 6...... 98 7. Summary and Outlook...... 100 7.1. Summary ...... 100 7.1.1. Synthesis and characterization of the new compounds...... 100 7.1.2. MOCVD-Experiments ...... 102 7.2. Outlook...... 104 8. Experimental Section ...... 105 8.1. Spectroscopic Characterization of the synthesized Compounds...... 105 8.2. Thin Film Analysis...... 106

Table of Contents

8.3. MOCVD-Experiments...... 109 8.3.1 Wafer Treatment...... 109 8.3.2. Handling of the self-built reactor...... 109 8.4. General Synthesis Procedures and Starting Compounds...... 111 8.5. Synthesis of the Tungsten-Compounds...... 112 8.6. Synthesis of the Molybdenum-Compounds...... 119 8.7. Crystallographic Data...... 122 9. References ...... 126

1. Introduction

Thin films of tungsten nitride and their application have gained increased interest, during the last 20 years. Although the first attempts to make tungsten nitrides were described over a hundred years ago[1, 2], the attention began to increase in the late 1980’s. The main driving force in tungsten nitride research was the great potential for applying tungsten nitride films in microelectronic devises like Schottky diodes[3-8]or diffusion barriers[9, 10]. Compared to tungsten nitride, molybdenum nitride has drawn much less attention, especially in microelectronics, although it possesses some interesting additional properties, not only in microelectronics[11-13]but also for catalytic applications[14].

1.1. Motivation and Goals

Beside their application as hard coatings as protection against wear and corrosion, an important application of early transition metal nitrides are thin films of these nitrides in microelectronic devices. Concerning to the International Technology Roadmap for Semiconductors 2005 which describes the technical developments that have to be achieved in order to keep the high growth rate, barrier materials used for Cu wiring must prevent its diffusion into the adjacent dielectric, but in addition must form a suitable, high quality interface with Cu to limit vacancy diffusion and achieve acceptable electromigration lifetimes[15]. Beside Tantalum nitride[16, 17] and Titanium nitride[18], Tungsten nitride is among the most promising materials for this purpose. The optimal technique for depositing thin films of these materials still are not known. All fabrication techniques available, have several advantages and disadvantages (see below). For the present study, MOCVD was selected as thin film fabrication technique for depositing films of W2N and Mo2N. To obtain better results in these processes, our approach was to investigate the behaviour of new classes of all-nitrogen coordinated precursors based on the guanidinato ligand in deposition techniques and to compare the results to common precursor systems that posses more simple structures. The introduction of a relatively complex guanidinato ligand, of course leads to more complex precursor molecules, with changed volatility, thermal stability and thermal decomposition behaviour. The main question was, how the increased complexity of the precursor molecule affects the deposition process and the properties of the desired material.

1 1. Introduction

The second motivation for this work was to investigate new complexes in the chemistry of guanidinato-complexes. Guanidinato ligands have recently attraction interest as nitrogen-rich chelating ligands with high steric and electronic tunability. Many guanidinato complexes have been identified as being excellent initiators or catalysts for electrophilic polymerization reactions, e.g., trimethylene carbonate and its copolymerization with ε-caprolactone[19] (which is initiated by an ytterbium guanidinate), the polymerization of lactide[20] (catalyzed by a homoleptic zinc guanidinate), the polymerization of α-olefins (which is catalyzed by a zirconium guanidinato complex[21]), or the polymerization of styrene[22] (which can be initiated by guanidinato complexes of neodymium or ytterbium). Thus, the chemistry of guanidinato complexes appears to be quite promising, particularly looking at novel types of tailored homogeneous catalysts for various reactions.

1.2. Tungsten Nitride Materials

After the discovery of tungsten in 1783[23], the first attempts in making tungsten nitride were th [24] made in the middle of the 19 century, but they all yielded in oxo-nitrides like WN2·WO3 . Langmuir reported a brown powder which was formed by gaseous tungsten from a filament in [2] the nitrogen atmosphere inside a light bulb , which he suggested to have the formula WN2 but no direct analyses were made. Smithells and Rooksby proofed this by titration on the product[25].

Table 1. Phases of tungsten nitride and their structure types Phase Structure type References

α-WxN solid solution of N in α-tungsten [38]

β-W2N cubic face centered structure [26] γ-WN tetragonal structure [30]

γ-W3N4 cubic structure [31] δ-WN hexagonal [32-35]

WN2 rhombohedral [36;37]

The first crystallographic characterizations were made in 1930 by Hägg[26], who identified β- tungsten nitride which has the composition W2N and was synthesized by the treatment of metal powder with ammonia. This β-phase later was confirmed by electron diffraction[27] Kiesling and Liu postulated a new γ-Phase in 1951[28] but this was disproved being an oxynitride[29]. The first well characterized γ-tungsten nitride was synthesized by Neugebauer et al. who nitrided β-tungsten powder with dry ammonia to obtain a tetragonal phase of [30] WN . A second, cubic γ-phase of the stoichiometry W3N4 was can be synthesized a heating

2 1. Introduction

α-W in an atmosphere of hydrogen and ammonia[31]. A hexagonal δ-phase of tungsten nitride [32-35] was described by Khitrova and Pinsker . The crystal structure of WN2 was determined to be rhombohedral[36, 37]. The α-phase is a solid solution of nitrogen in tungsten. Due to the low solubility of nitrogen in tungsten this phase was believed being non-existent for a long time and it is only described in a single reference[38]. The synthesis succeeded by nitridation of tungsten films at temperatures above 700 °C. Important for this reaction was complete preliminary dissociation of the ammonia and a sufficient flow rate. While WN2 is a ceramic material and an insulator, that

Figure 1: Crystal Structure of β-W2N decomposes to WO3 when exposed to moist air, the β-phase and also the γ-phase and the δ- phase are airstable metallic and conductive phases, where the resistivity of the nitride is increasing at a higher nitrogen content[3]. They are refractory materials that are acid resistant[39], and have the lowest resistivity compared to other transition metal nitrides.

1.3. Molybdenum Nitride Materials

Again it was Hägg[26] who made the first crystallographic studies on molybdenum nitrides. Beside the solid solution of nitrogen in molybdenum which was assigned to be the α-phase he found three different molybdenum nitrides. The tetragonal β-phase having the stoichiometry

Mo16N7, the cubic γ-molybdenum nitride with the composition Mo2N and a hexagonal δ- phase of the stoichiometry MoN. When nitriding pure molybdenum with ammonia, the metastable β-phase is only accessible at higher temperatures, but if the reaction is catalysed by calcium nitride, β-Mo16N7 is stable at room temperature, stabilized by the inclusion of calcium in the lattice[40]. By high pressure nitrididation at 1100 °C Ettmayer obtained the β-

3 1. Introduction phase in the absence of a catalyst[41] Several manufacturing methods lead to mixtures of γ- [11, 42, 43] Mo2N and δ-MoN . An accurate process control is required, where the formation of the

Table 2. Phases of Molybdenum nitride and their structure types Phase Structure type References

α-MoxN solid solution of N in α-molybdenum [26]

β-Mo16N7 tetragonal [40;41]

γ-Mo2N cubic face centered structure, analogue to β-W2N [26] δ-MoN hexagonal structure [11;42;43]

Mo2N hexagonal structure [44] MoN rock salt structure, high temperature superconductor [45;46]

Mo2N3 amorphous, gold coloured phases [12;13]

Mo3N2 Most likely γ-Mo2N with excess nitrogen in the lattice [47] more nitrogen rich δ-phase is favoured at higher temperatures and higher concentration of the nitrogen source. By reactive sputtering, a new hexagonal phase of Mo2N could be obtained by Fuller et al.[44]. The metastable cubic, rock salt structure MoN phase, which was predicted, being a high temperature superconductor[45] is accessible by ion assisted deposition [46] techniques . By chemical vapour deposition from Mo(NMe2)4 an additional phase of the [12, 13] stoichiometry Mo2N3 was obtained by Fix et al. . The obtained films were gold-coloured and conductive but due to their amorphous nature, no XRD examinations could be made.

Another phase, composed Mo3N2 made by ammonia nitridation of sublimed molybdenum films is most likely to be the γ-phase Mo2N containing excess nitrogen included in the [47] lattice . A ceramic insulating phase MoN2 of Mo(VI) comparable to tungsten in unknown.

1.4. Applications of Tungsten and Molybdenum Nitrides 1.4.1. Schottky Diodes

One of the first technical applications of tungsten nitride was its use as a metal contact in Schottky diodes[6], named after the German scientist Walter Schottky, also known as Schottky contacts or Schottky barriers. Diodes are electric rectifiers letting the current pass in only one direction. In a normal diode the rectifying is achieved by a contact between a p-semiconductor and an n-semiconductor (Figure 2a). Taking a Schottky diode, the p-semiconductor is replaced by metallic material (Figure 2b). As semiconductor in a Schottky diode normally silicon or a III/V semiconductor like GaAs or InP is chosen when operating at lower current. If the diode is operated at a higher current (>250 V) SiC or SiGe is chosen. Compared to normal diodes Schottky diodes can operate at higher frequencies (up to 100 GHz), but an inherent disadvantage is the higher leakage current compared to p-n-Diode.

4 1. Introduction

Figure 2: Examples of diodes. a) p-n diode; b) Schottky-Diode (reproduced after: Wikipedia – The free Encyclopedia)

1.4.2. Conductive Diffusion Barriers

According to the International Technology Roadmap for Semiconductors out of the year 1999 the device node in sub-quarter-micron devices has to shrink from 180 nm in 1999 to 35 nm in 2014[48]. “Moore’s law” states that the number of transistors, placed on an integrated circuit will double every 18 to 24 month. The International Roadmap for Semiconductors describes the technical developments that have to be achieved in order to keep this high growth rate. To achieve this goal one major focus is on the replacement of the wiring material aluminium, that has been used so far (Resistivity: 2.65 μΩ·cm[49]), by copper (Resistivity: 1.67 μΩ·cm[50]) or silver[51]. Due to its lower costs, the main focus of industry as a future wiring material is copper. The main problem of the introduction of copper into the structure of an integrated circuit is the highly favoured formation of insulating copper silicide (Cu3Si) caused by migration of copper into silicon at copper/silicon interfaces. Due to this fact, an effective, conductive diffusion barrier is necessary to be placed between the copper wiring and the Si-substrate (Figure 3).

5 1. Introduction

Figure 3: Changes in a MOSFET-Structure, left: Al-wired MOSFET, right: Cu-wired MOSFET including [52] a WxN diffusion barrier (after Ref. )

A good diffusion barrier has to fulfil high requirements, due to the several mechanisms of possible barrier failures. There are three different mechanisms described how a diffusion barrier might fail. The first is the loss of integrity due reaction with the wiring and / or the silicon substrate. Another possibility is the diffusion of copper through crystal defects in the barrier material. And finally, the diffusion of copper or silicon atoms along grain boundaries. So a diffusion barrier material has to be highly conductive and thermally stable, as well as inert to the Silicon Substrate and to the wiring material. To prevent copper diffusion along grain boundaries, which exist in polycrystalline material the diffusion barrier should be nanocrystalline or ideally amorphous[53]. All these requirements have to be fulfilled by a diffusion barrier of ideally no extension. A high melting point is one indicator of a good barrier material, due to experimental results that proved the proportional relation between lattice diffusion rates and the melting point of the material[54]. Due to this, the refractory early transition metals like Titanium, Niobium, Molybdenum, Hafnium, Tantalum and tungsten and their nitrides, as well as several [9] conducting carbides (e.g.: TiC, TaC, W2C) and carbonitrides (e.g.: TiCxNy, WCxNy) . Very detailed studies for the systems Ti and TiN are already available, and have been employed but their barrier properties are quiet poor, due to low thermal stability of Titanium nitride phases, low inertness of titanium against copper and a high barrier thickness that is required for adequate prevention of copper diffusion (>50 nm)[53]. So far the best barrier properties have been obtained for Tantalum nitride[55-57] and tungsten nitride[58, 59]. Tungsten nitride forms no known compounds with copper, it is acid resistant and it has the lowest bulk resistivity compared to other transition metal nitrides. Additionally it is possible to deposit tungsten nitride in an amorphous form, which means the lack of grain boundaries and the missing of fast diffusion pathways for copper. Another important parameter of the film is its work function, that must satisfy the threshold voltage requirements of specific devices. The work function can be tuned by the deposition method, by post deposition heat treatments, by the film thickness and its crystallinity or the 6 1. Introduction crystal orientations. In the case of metal nitrides, the work function can also be tuned by the nitride content of the film.[60-62]

1.4.3. Hard Coatings

Another application of early transition metal nitrides, that deals with their mechanical properties, are hard coatings on tools that should protect these tools (e.g. cutting or drilling tools) against wear and corrosion. The steel substrates (the tools) are coated with a metal nitride layers of a few microns thickness. Mostly these metal nitrides are used as composite [63] [64] materials e.g. W2N/Si3N4 , as ternary alloys like Cr1-xWxNy or even quaternary or higher alloys[65]. These coatings posses a hardness of about 50 GPa[63]. (For comparison: Hardness of diamond: 78 GPa; Hardness of Silicon Carbide: 24 GPa)

1.4.4. Catalysis

While tungsten nitride phases are mainly used in microelectronics, the main application for phases of molybdenum nitride is heterogeneous catalysis. For this purpose not smooth films like desired in microelectronics, but high surface area material is required. Molybdenum nitride phases for this purpose have been prepared on Zeolithes[66, 67], from thermal [68] [69-71] decomposition of molybdate salts or by ammonia nitridation of the oxides , but also films deposited by chemical[72] or physical vapour deposition[14] are suitable for catalytic applications. Possible reactions that are catalysed by the different molybdenum nitride phases are ammonia synthesis[73] and decomposition[14], hydrodesulfurization[66, 71] and alkine conversion[74]. The most employed phases for catalytic applications are the tetragonal β-

Mo16N7 and the cubic γ-Mo2N.

1.5. Thin Film Deposition Techniques via gas Phase 1.5.1. Physical Vapour Deposition (PVD)

Actually Physical Vapour Deposition is not a single technique, but is a variety of techniques of which the most prominent is sputtering. Several other techniques like Molecular Beam Epitaxy (MBE) and pulsed laser deposition also belong to the group of PVD techniques.

7 1. Introduction

Figure 4: Sputtering of tungsten nitride. Nitrogen atoms (blue) combine with tungsten atoms (yellow) to [52] form WxN-films (after Ref. )

All PVD processes have in common, that no molecular gas-phase or surface reactions are included into the deposition process. During a sputtering process, a target made of the desired element is bombed with ions (mostly argon), acting a the material source which releases the target material in the form of individual atoms into the gas phase. The material is now deposited everywhere in the reaction chamber including loaded substrates. The charged argon atoms used for bombardment of the target are created in a magnetron plasma, with the target acting as the cathode of the magnetron. The bias voltage applied at the magnetron cathode lies in the range of 300 V or higher. The high voltage causes an acceleration of the argon ions towards the target. When the ions hit the target, they strike out neutral atom from the surface of the target. For the deposition of metal nitrides dinitrogen is added to the argon gas.

Various methods are used to activate the N2 Molecule. For example in a plasma the dinitrogen forms nitrogen radicals that are incorporated in the deposited films. When an additional gas is added as a reactive gas, the process is called reactive sputtering. Sputtering techniques allow good control over the film’s stoichiometry by varying the nitrogen content and it allows high deposition rates. Due to the absence of any other elements incorporated in the process (e.g. carbon, oxygen), the deposited films are free of impurities. In the case of tungsten nitride and molybdenum nitride, phases with nearly any stoichiometry can be deposited by adjusting the flux of the sputtered atoms[75, 76]. The obtained stoichiometry has enormous influence on the physical properties like conductivity[75, 77]and hardness[75, 78]. The main disadvantage of sputtering processes is the incapability of depositing

8 1. Introduction uniform films on objects with high aspect ratios, e.g. DRAM-devices and more complex device structures. Additionally, the highly energetic particles involved in the process can damage underlying substrate structures. Effective WNx diffusion barriers have been deposited by this technique[75, 79]. Nevertheless PVD is a standard technique for device fabrication.

1.5.2. Chemical Vapour Deposition (CVD)

Chemical Vapour Deposition is a rather old method, dating much more than hundred years. Ludwig Mond was one of the first who applied this technique by using Nickeltetracarbonyl to deposit very pure Nickel[80].(Equation 1.1)

Δ ()CONi )g(4 ⎯⎯→ Ni(s) + 4 CO (g) (1.1 ) Till the day CVD processes for many different material are established like metals, III/V semiconductors, oxides, nitrides, silicates, borides or carbides[81]. CVD has established as the main fabrication method for materials in microelectronic industry.

Figure 5: Simplified Scheme of reactions involved in a CVD-Process

The Process distinguishes itself from physical methods like sputtering or sublimation by the occurrence of chemical reactions involving molecular precursors. Due to this, CVD is of course much more complex, than PVD methods. During a PVD process, the deposition is controlled more or less only by diffusion processes, while in a CVD process several thermally induced chemical reaction in the gas phase, as well as on the surface of the substrate have to be considered. In the first step, the precursor has to diffuse out of the main flow going through the reactor to the surface of the substrate. To reach the surface, the precursor molecule has to diffuse perpendicular to the surface through the so called diffusion zone. During the diffusion partial thermally induced decomposition of the precursor can occur in the gas phase, leading

9 1. Introduction to reactive intermediates. When the reactants are adsorbed to the surface, they can either desorp from the surface, diffuse on the surface, or react to form thin films. An important feature that has to be considered are the gas phase reactions. A gas phase reaction of the intermediate species might occur, where the species undergo a chemical reaction, forming solid powders that are collected on the surface of the substrate. This powder might act as a crystallization center[82]. The reaction kinetics of this highly complex system are heavily dependent on the reaction temperature. At a low temperature the film growth is controlled by surface kinetics. A

Figure 6: Arrhenius-Plot of the growth rate: 1: Surface reaction controlled zone 2: Diffusion controlled zone 3: Zone of depletion of the diffusion zone precursor molecule, adsorbed to the surface, does not necessarily decomposed immediately, but can diffuse on the surface and might also be desorbed without any decomposition. An increase of the temperature leads to a higher decomposition rate of the precursor thus to a higher growth rate. The first regime (1) refers to the kinetically controlled deposition (see figure 6). The Arrhenius-like behaviour of the temperature dependence of the growth rate relates to the activation energy of the rate-limiting steps of the decomposition mechanism of the molecule (typically surface reactions). The second regime (2) refers to the diffusion limited growth. All chemical reactions are much faster than the mass transfer through the gas phase to the surface. The third regime (3) refers to the depletion of the diffusion zone above the substrate.

10 1. Introduction

A classical CVD-Process of a binary compound consists of two reactants. One is normally a Metal containing, volatile precursor, the other one is a reactive gas, e.g. ammonia or hydrogen, depending on the desired material. Volatile homoleptic halides like WF6, MoCl5 or

MoF6 have been used in many studies. These compounds require high volatilisation temperatures and high deposition temperatures, thus high requirements to the reactors and to the thermal stability of the substrates have to be fulfilled. Halide containing by-products like HCl or HF are produced during the process, that might damage the substrate structures by etching. Additionally possible halide incorporation into the films has a negative influence on the electronic properties of the films, and so can be problematic. The use of metalorganic compounds was first described by Manasevit in 1968[83], who used Arsine and Trimethylgallium to deposit Gallium arsenide (Equation 1.2).

Δ ()CHGa 3 3(g) AsH 3 ⎯+ ⎯→ GaAs(s) + 3 CH 4(g) (1.2 ) Metal organic compounds are halide free and are often more volatile and less thermally stable compared to metal halides. So they allow rather mild deposition conditions. Some metal organic precursors are suitable as so called single source precursors. A single source precursor is able to form a binary material without a second reactant. Examples for single source [84] [85] precursors are Bisazido(dimethylaminopropyl)gallium and [(CH3)HGaN3]x which are able to form Gallium nitride by thermal decomposition without any additional nitrogen source. The second reactive compound can also be replaced by a plasma (e.g. hydrogen or nitrogen plasma). The use of a plasma yields in higher reactivity at lower temperatures. Nevertheless the reactive conditions in a plasma enhanced CVD process might damage complex structures of the substrate.

Figure 7: Examples for Single Source Precursors

By changing the parameters of the deposition experiment (temperature, pressure, flow rates) MOCVD offers high control over deposition rate, film composition and uniformity of the desired films. When deposition temperature is in the regime of kinetic control, in particular if it is surface reaction controlled the reactant diffusion on the surface of the substrate allows coverage of very complex structures with uniform films. The main disadvantage of MOCVD

11 1. Introduction is the air sensitivity of most precursors and the frequent incorporation of impurities (mostly carbon) into the films, resulted by the organic ligand sphere of the precursors. The standard method for MOCVD experiments is the deposition at low pressures. A suitable precursor for this purpose has to fulfil several requirements. A crucial point is the volatility of the precursor, which should evaporate significantly below its decomposition temperature, so that a transport from bubbler to the reaction zone, without decomposition can be ensured. It is also possible to apply less volatile compounds by dissolving them in a organic solvent (e.g. toluene, benzonitrile). This solution is abruptly evaporated, tearing the precursor molecules into the gas phase. This technique is called liquid injection CVD. Another important feature the precursor should offer, is the clean decomposition to the desired phase. This means low incorporation of unwanted ligand fragment into the films. Good leaving groups bonded to a thermally labile compound can fulfil this requirement. An example for this is given by the [(CH3)HGaN3]x, mentioned above. The methyl group and the hydrido ligand can form methane, while the azide can eliminate nitrogen. (Equation 1.3)

Δ []()3 HGaNCH 3 x(g) ⎯⎯→ Nx 2(g) x CH 4(g) ++ x GaN (s) (1.3 ) The use of a bidentate ligands normally increases the thermal stability, but often it leads to incorporation of ligand fragments into the film. The content of impurities normally decreases at higher deposition temperatures. The temperature that can be applied during a CVD- experiment is limited by the thermal stability of either the deposited film or the substrate. When different phases of a binary material are possible, the precursor should yield of course the desired crystal phase (e.g. β-W2N instead of γ-WN). It was claimed that pre-formed bonds like metal-nitrogen double bonds should favour the formation of certain phase, but till the day no proof for that could be observed. The crystallinity can be mainly controlled by the amount of reactive gas, and the substrate temperature. It is still a subject to speculation if a certain ligand system can influence the obtained crystallinity. For industrial application it is important that the synthesis of the precursor has got high yields and can be scaled up to large quantities easily (>50 g). A synthesis that requires only one or two steps is favourable compared to a multi step synthesis. Also a low toxicity of the applied ligands is preferred. A desirable feature of the precursors is also a low reactivity towards air and moisture, which simplifies the handling of the precursor. However this requirement depends on the type of CVD process. A high chemical stability at the evaporation temperature should also be given.

12 1. Introduction

1.5.3. Atomic Layer Deposition (ALD)

The first report about atomic layer deposition published by in 1977[86] by Suntola who is commonly known its inventor. This makes ALD the youngest technique among all chemical gas phase deposition techniques. The main difference to ordinary CVD techniques is the elimination of gas phase reactions during an ALD experiment. The closest parallel in PVD would be Molecular Beam Epitaxy (MBE) techniques. In a CVD-Experiment, the substrate is heated to temperatures that are significantly higher than the decomposition temperature of the precursor. The decomposition of the precursor occurs by thermal decomposition in the gas phase as well as by surface reaction on the substrate. In the case of an ALD-Experiment the temperature of the substrate is lower than the decomposition temperature of the precursor. The precursor is adsorbed onto the surface of the substrate preferable by chemisorbtion or strong physisorbtion forming a monolayer than ideally should inhibit further growth. Thus it is critical, that the adsorption thermodynamics allow the quantitative formation of a monolayer due to a strong interaction of the precursor with the surface. Otherwise desorption of the intact precursor is favoured by higher temperatures. Thermal decomposition of the precursor does not occur. After removing not absorbed excess precursor out of the reaction zone a second reactant (e.g. water, ozone for oxides; ammonia; hydrazine for nitrides) is pulsed over the substrate, leading to decomposition of the precursor and creating new absorption sites on the surface. By purging the reactor with inert gas, excess reactant and by-products is removed from the reactor. Now another cycle can be started by pulsing precursor over the substrate again. A well known example of an ALD-process comes close to the ideal behaviour mentioned above is the deposition of aluminium oxide by trimethylaluminium and water. On a silicon substrate the surface is terminated by OH-groups, that can react with the trimethylaluminium by eliminating methane. When the water is added it reacts with the residual methyl groups forming aluminium oxide. At the same time the water also creates new OH-groups on the surface that can act as new absorption sites in the next cycle. Though ALD has several indisputable advantages, it still suffers from serious problems that still have to be overcome. Thus it polarizes research groups as well as industry if it is an applicable technique. On the side of the advantages of ALD is the easy control of the film growth by the number of cycles per deposition. Due the separation of the precursor and the reactive gas no gas-phase

13 1. Introduction

Figure 8: Schematic view of an ALD Process (after www-rpl.stanford.edu) prereaction during the depositions can occur. Even on substrates with very high aspect ratios films of high uniformity can be obtained by ALD. Due to self-limiting of the process a precise control of pressure and gas flow rates is not necessary if the ALD-process exhibits ideal behaviour. Unfortunately most precursors do not exhibit ideal ALD behaviour. In many cases an increase of the growth rate is observed when the pulse time of the precursor is elongated. In the case of an ideal behaviour the growth rate would be independent of the pulse time. This indicates the occurrence of parasitic thermal decomposition like in CVD experiments. In this case the experimental parameters like flow rates, pulse times of the reactants and deposition temperatures play an important role on the film uniformity. The low deposition temperatures of ALD-experiments that are below the decomposition temperatures, favour the formation of amorphous films with low density. Post deposition annealing steps may be required… The deposition of nano crystalline films has turned out to be challenging. In the case of the deposition of very thin films of thicknesses below 10 nm, the first ALD cycles raise problems in terms of island growth. Also the theoretical knowledge of about the ALD-process is quiet low yet. The technique not very well established yet, and the technical application is still problematic. Nevertheless, though it has several disadvantages ALD is a promising technique.

14 2. State of the Art

2. State of the Art 2.1. Precursors for MOCVD and ALD 2.1.1. Tungsten nitride

Figure 9: Overview of all literature reported tungsten nitride precursors In 1987 Nakajima et al. reported the first synthesis of tungsten nitride by CVD[87]. They used tungsten hexafluoride as a tungsten source, ammonia as a nitrogen source plus hydrogen as an additional reducing agent. (Equation 2.1)

NH32WF 3 H ⎯++ ⎯→Δ N 12 HF ++ W N 2.1 6(g) 3 2 2 2(g) (g) (s)2 ( ) While hydrogen is the most common reducing agent[88-91], also silanes[92, 93] or boranes[94] can be applied. The reducing agent has to be dosed very carefully, because when it is overdosed the tungsten hexafluoride will be reduced to elemental tungsten. Due to this fact hydrogen is unsuitable as a carrier gas, mostly argon is used in these processes. Without additional reducing agent it was WF6 and ammonia gave only a yellow-brown adduct WF6·4NH3. A serious problem of this process is the occurrence of HF in the reaction, which might form ammonia adducts like NH4F and NH4HF2 as solid by-products, which might lead to fluorine incorporation into the films. Additionally it can damage a silicon or silicon oxide surface by etching (Equations 2.2 and 2.3)

2.2 2H SiF Si 4HF(g) Si (s) ⎯+ ⎯→ SiF 4(g) + 2H 2 ()2.2 (g) SiO 4HF(g) SiO 2(s) ⎯+ ⎯→ SiF 4(g) + O2H (g)2 ()2.3 Another problem is that tungsten hexafluoride also can react with silicon[88, 95] to silicon tetrafluoride leaving an etched surface and making it nearly impossible to obtain homogenous films. Tungsten hexafluoride melts at 1,9 °C and boils at 17,1 °C at atmospheric pressure[23] thus is a very volatile compound. Due to the very high volatility of the tungsten source high growth rates can be achieved, but it is also difficult to obtain very thin films, of thicknesses below

100 nm. The reduction of WF6 by hydrogen starts at 300 °C. By this method tungsten nitride

15 2. State of the Art layers with resistivities down to 100 μΩ·cm could be deposited[88, 89]. The stoichiometry of the films, which are mostly amorphous can be easily controlled by variation of the WF6/NH3 ratio. It was observed that a higher fluorine contamination in the films is obtained at higher deposition temperatures[96] Fluorine diffuses rapidly into copper, thus can be incorporated if used in conjunction with copper metallization. The contamination of the films can be reduced by post deposition annealing at high temperatures[91]. The problem of ammonium salt formation can be overcome by the use of a nitrogen plasma as a nitrogen source[95]. Tungsten hexafluoride was also successfully applied as a precursor for ALD[94] of tungsten nitride as well as of tungsten carbonitride[97-103]. Again a reducing agent is required to obtain tungsten nitride. For the deposition of tungsten carbonitride, triethylboron is applied acting as a reducing agent and as a carbon source at the same time. The films that could be obtained by this method were down to 10 nm thick and had conductivities of ~ 350 μΩ·cm[94].

Tungsten carbonitride films grown by ALD using WF6, NH3 and triethylboron had very low densities[100, 101] (~4,6 g cm-3) when the films of a thickness lower than 5 nm are deposited but increased to 13.1 g cm-3 at a film thickness of 40 nm. Resistivities are reported to be 350 [97] μΩ·cm at 24 nm layer thickness . Quiet recently the deposition of W2N by ALD using tungsten hexafluoride and ammonia without an additional pulse of a reducing agent was reported[104]. The reduction was induced by exposing the ammonia to a plasma pulse, producing reactive N-H species that are easily absorbed to the silicon substrate and react sequentially with the tungsten hexafluoride. This surface nitridation suppresses the silicon reduction by WF6. An interesting result was reported by Bystrova et al., who applied a [105] combination of ethylene and silane as reducing agents, pulsed WF6/NH3/C2H4/SiH4/NH3.

They obtained film had the stoichiometry of W3N2 and were phase mixtures of β−W2N and γ−WN. Both silicon and carbon content was found being below the detection limit of XPS. While all these experiments use a high oxidized tungsten (VI) source, that has to be reduced to form low valent tungsten nitride, it is also possible to form tungsten nitride from the tungsten (0) source tungsten hexacarbonyl. W(CO)6 is less volatile than WF6, (melting point: 150 °C[106]) but no additional reducing agent is required. Depositions can be carried out at temperatures as low as 200 °C [107, 108]. The dissociation energy of W-CO bonds is relatively low, with recombination being easily interrupted by the presence of nitrogen-bearing reactants, such as ammonia, leading to the formation of tungsten nitride. Due to the absence of halides, no fluorine contamination, and also no etching of the substrate can occur. Additionally, no acidic by-products are obtained, so no ammonia adducts will be formed. Depending on the process conditions the obtained films incorporated 2-16 atom percent of

16 2. State of the Art

[108] carbon and about 4 at. % of oxygen . Similar to the WF6 case the film stoichiometry can be easily controlled by adjusting the flow rates of W(CO)6 and ammonia. The films obtained from tungsten hexacarbonyl showed good properties. Resistivities as low as 123 μΩ·cm for 50 nm thick films[106] and 590 μΩ·cm for 15 nm thick films[107] were obtained. To reduce the carbon contamination of the films, it is desirable to have tungsten complexes with an all nitrogen coordination sphere. The first volatile complex of this king was reported by Nugent in 1980[109]. (Bis-tert-butylimido)(Bis-tert-butylamido) Tungsten is synthesized by adding 10 equivalents of tert-butyl amine to a hexane solution of tungsten hexachloride (equation 2.4)

WCl6 + 10 t-BuNH2→ W(Nt-Bu)2(NHt-Bu)2 + 6 t-BuNH3Cl (2.4) It is a white powder, that can be easily sublimed[110].The compound is able to form β-tungsten nitride without any additional ammonia[111], thus is acting as a single source precursor. Temperatures of 450 to 650 °C were used to grow highly conformal tungsten nitride. The films contained 66 at. % tungsten but only 25 at. % nitrogen. The residual content consisted of 6 at. % of carbon and 3 at. % oxygen. Though this is one of the lowest carbon content in a tungsten nitride film grown from a single-source precursor, that could be obtained till the day, mechanistic studies figured out that further reduction of the carbon content is impossible without an additional reactive gas like ammonia. The carbon containing fragments on the surface were produced by fragmentation of the tert-butyl-imido- respectively the amido ligands. Tert-butyl amine, isobutene and acetonitrile were the main decomposition products[112, 113]. γ−hydride activation and β-methyl activation are supposed to be the main reaction pathways supposed to be the dominating reaction pathways producing carbon contamination. Ammonia seemed to be essential for removing the carbon containing fragments from the surface[112]. Further metal organic precursors are all based on imido complexes featuring tungsten- nitrogen double bonds. (Bis-tert-butylimido)(Bis-tert-butylamido) Tungsten contains N-H bonds, making the compound relatively polar. Replacing the protons by alkyl group should lead to less polar compound that possesses an increased volatility and in fact (Bis-tert- butylimido)(Bis-diethylamido) Tungsten is a liquid. Wu et al. observed an interesting feature of this precursor. When they tried to deposit tungsten nitride by this precursor on a Si(100) surface, they obtained relatively pure β-W2N. But when the same precursor is deposited on a SiO2 surface, a mixture of tungsten carbides, tungsten

17 2. State of the Art nitrides and even tungsten oxides is obtained. They assumed this behaviour to a less selective bond cleavage on silicon dioxide compared to silicon.[114] Substitution of the diethylamido ligands by dimethylamido groups leads to a further increase of the volatility[59]. This precursor was applied by Becker et al. in ALD experiments to obtain tungsten nitride. When they applied ammonia as an additional reactive gas, surprisingly they did not obtain the β-phase, but an amorphous, carbon free phase of the stoichiometry WN[58, 59, 115, 116]. The films of 20 nm thickness and conductivities down to 480 µΩ·cm could be obtained. Even objects with an extremely high aspect ratio could be covered by this technique. It was reported that a 2 cm long fused silica capillary tube with an inner diameter of 20 µm could be homogeneously covered with tungsten nitride[59]. Ammonia was found to be essential for the growth of the nitride phase, but interestingly it was not incorporated into 15 the films, as was observed in deposition with NH3. When the ammonia was replaced by a hydrogen plasma, the obtained films possessed only very low contents of nitrogen, but consisted mainly out of tungsten carbide[117].

Winter et al. reported another variation precursors bearing the W(NtBu)2-fragment. They synthesized compounds containing pyrazolato ligands. Depending on the substituents on the pyrazolates, the compounds are either suitable to produce tungsten nitride nanoparticles[118] or are as volatile to be potential precursors for gas phase depositions[119]. Unfortunately deposition experiments with these compounds were not reported till the day.

Halide containing metal organic precursors of the form (RN)WCl4(NCR’) (R=i-Pr, Ph, Allyl; R’= Me, Ph) were reported by McElwee-White et al.[120-123]. These compounds were obtained by carbon dioxide elimination from WOCl4 in the presence of either acetonitrile or benzonitrile. (Equation 2.5)

'CNR WOCl4 RNCO⎯+ ⎯→⎯ () 4 (NCR'WClRN ) + 2 (2.5CO ) These class compounds is non volatile due to the coordinated nitrile but can be applied in MOCVD anyway by liquid injection. For this purpose the compound is dissolved in an organic solvent and pumped into a nebulizer. Interestingly, though the precursors posses high chlorine contents, the obtained films contained no detectable chlorine. This was supposed to be due to the use of hydrogen as a carrier gas, removing chlorine from the reaction as HCl[124]. When these compounds are use for depositions in the absence of ammonia, the films were contaminated with high amounts of carbon, up to 50 at. %[122, 123]. The carbon is mainly originated in the solvent, and the content of contamination was found being heavily dependent on the kind of solvent used. The use of a nitrile as a solvent even affects the nitrogen content[121].Despite the high carbon and also oxygen contamination level the conductivity of

18 2. State of the Art the films deposited at temperatures below 500 °C the resistivity of the films is as low as 750 µΩ·cm, but it increased up to 5500 µΩ·cm at higher deposition temperatures[123]. When the precursors are used in the presence of ammonia, the nitrogen content of the films could be significantly increased. But the level of carbon contamination could still not be reduced below 10 at. %[125]. Deriving from these kind of complexes, some guanidinato complexes have been synthesized by substitution of one chloride (equation 2.6)[126]. R'' Cl Cl Cl Cl N

RN W NCR'+ Li{(R''N)2CNMe2}x W N (2.6) - LiCl RN N - R'CN Cl Cl Cl R'' The guanidinato complexes have been suggested, being precursors for liquid injection MOCVD, but a their use in deposition experiments has not been reported till the day. Quiet recently the growth of tungsten nitride films by ALD using the low valent compound [127] W2(NMe2)6 and ammonia has been reported . This tungsten (III) compound shows ALD growth behaviour in the temperature window of 180-220 °C, which is the lowest deposition temperature of WxN-films reported. The films showed tungsten contents of only 39 at. % and high contamination levels of carbon, oxygen and even hydrogen of 5-13 at. %. The W:N ratios in these films were 1,2-1,35 which is clearly nitrogen rich towards W2N Resistivities of the as-deposited, amorphous films were about 800 µΩ·cm.

Table 3: Overview of all Literature-reported CVD experiments of WxN Additional Precursor Reagents Results Reference

WF6 NH3; H2 [87-91] β-W N. Often contaminated with fluorine; Condutivity > WF NH ; SiH 2 [92-93] 6 3 4 100 μΩ·cm; Problems with particle formation WF6 NH3; B2H6 (WF6·4NH3); Easy control of stoichiometry [94]

WF6 N2-Plasma β-W2N. No Particle Formation [95] very thin (<5 nm) β-WCxNy. Low density; Condutivity > WF6 NH3; BEt3 240 μΩ·cm; [97-103] NH3; SiH4; WF6 C2H4 Phase Mixture of β-W2N & γ-WN [105] β-W2N; 2-16 at. % Carbon incorporation; ~4% oxygen contamination, No Particle Formation; Condutivity > 123 W(CO)6 NH3 μΩ·cm; [106-108]

W(N-tBu)2(NHtBu)2 none β-W2N. ~ 6 % Carbon incorporation [111]

W(N-tBu)2(NEt)2 none relatively pure β-W2N [114]

W(N-tBu)2(NMe)2 NH3 Amorphous phase; stoichiometry WN [59]

W(N-tBu)2(pz)2 NH3 Nanoparticles of β-W2N [118-119]

(RN)WCl4(NCR') NH3; H2 β-WCxNy [120-124]

W2(NMe6) NH3 amorphous WxN [127] pz= pyrazolates; R = i-Pr; Ph, Allyl; R'= Me, Ph

19 2. State of the Art

2.1.2. Molybdenum nitride

Figure 10:Overview of all literature reported molybdenum nitride precursors

The number of CVD experiments aiming on molybdenum nitride is rather low, compared to tungsten nitride. Most of the compounds that were used as precursors are homologs of the analogous tungsten compounds.

By using molybdenum hexafluoride, cubic γ−Μo2N can be produced in an analogous way to tungsten nitride[128]. Both hydrogen and ammonia were necessary to deposit molybdenum nitride, and analogous to WF6, the amount of reducing agent has to be adjusted very careful or metallic molybdenum will be formed. The obtained films were porous and showed a very rough surface. At higher deposition temperatures (≥800 °C) hexagonal δ-MoN formation occurs beside the formation of metallic molybdenum occurs. Because these films were produced for hard coating purposes, nothing about the electronic properties was reported.

Molybdenum pentachloride could also be used to deposit MoxN. In contrast to MoF6, no reducing agent is needed to initiate the nitride formation. At 400 °C, the Mo:N ratio is 2:1, indicating the formation of cubic γ−Μo2N, while at higher temperatures more nitrogen rich

MoxN films are formed, until a maximum of 95 % δ−ΜoN and 5 %γ−Μo2N at 700 °C. Depositions at higher temperature lead to the formation of films containing metallic [43, 72] molybdenum beside the two MoxN phases Problematic in this process is the formation of ammonium chloride, that reduces the amount of ammonia available for MoxN formation.

In the temperature range of 350-500 °C MoCl5 is also suitable as a precursor for ALD purposes[129].Similar to the CVD experiments the nitrogen content is dependent on the deposition temperature. Powder-XRD patterns showed the formation of γ−Μo2N, while the film composition was observed being MoN1,23 at 350 °C and MoN1,1 an 500 °C indicating the presence of a nitrogen rich amorphous phase. At higher temperatures again the formation of δ−ΜoN was observed.

The application of Mo(CO)6 and ammonia in CVD experiments also leads to mixtures of [43] γ−Μo2N and δ−ΜoN , but in contrast to MoCl5 the formation of the δ-phase is stronger

20 2. State of the Art favoured at lower temperatures. It was even possible to obtain a pure δ−ΜoN phase at 700 °C.

Additionally the formation of pure molybdenum less favoured compared to MoCl5. Not only film deposition on flat substrates could be done by this precursor, but also MoxN deposition inside zeolite structures was reported[66].

Only three different MoxN-precursors bearing an all-nitrogen coordination sphere are reported [12, 13] in literature. Mo(NMe2)4 was reported by Fix et al. . When applied in CVD-Experiment in the presence of ammonia, this precursor yields a gold coloured, amorphous phase of the stoichiometry Mo2N3, at temperatures as low as 200 °C. The obtained films are conductive and free of carbon contamination. The structure of this phase is still unclear.

(Bisdiethylamido-)(Bis-tert-butylimido-)molybdenum is able to produce γ-Mo2N even in the absence of ammonia, but the films are highly contaminated with carbon[130]. The applied deposition temperatures were in the range of 450-600 °C. The carbon contamination of the films could be reduced, and the conductivity was improved by annealing these films an in ammonia atmosphere.

The dimethyl amido analogue of this compound, Mo(NtBu)2(NMe2)2 was recently reported as a potential precursor for ALD purposes[131]. Highly conformal, conductive molybdenum nitride films of down to 44 nm could be deposited by this method. The step coverage of these films was very high. The XRDs of these films showed the formation of the tetragonal β-

Mo16N7 phase, but the proportion of molybdenum and nitrogen content was observed being close to one. This indicated an amorphous, nitrogen rich phase being present in the films. Analogous molybdenum compounds to the tungsten imido- pyrazolato- compounds mentioned above were also reported by Winter et al.[118, 119]. These compounds were suggested being possible molybdenum nitride precursors, but their application in deposition experiments were not reported. An alternative approach to molybdenum nitride films was reported by Kim et. al.[132], who deposited films of pure molybdenum metal films. These films were converted into molybdenum nitride by heating them to 500-900 °C in an ammonia atmosphere. Similar to the

Table 4: Overview of all Literature-reported CVD experiments of MoxN Additional Precursor Reagents Results Reference

MoF6 NH3; H2 γ-Mo2N, At higher Temperatures δ-MoN is favored [128]

MoCl5 NH3 mostly phase mixtures of γ-Mo2N and δ-MoN [43,72,129] phase mixtures of γ-Mo2N and δ-MoN, but stronger Mo(CO)6 NH3 favoring of the δ-phase [43,66]

Mo(NMe2)4 NH3 gold-coloured, amourphous Mo2N3 [12,13] Mo(N- tBu)2(NEt2)2 none highly carbon contaminated γ-Mo2N [130] Mo(N- β-Mo16N7 as crystalline phase plus amorphous tBu)2(NMe2)2 NH3 nitrogen rich phase [131]

21 2. State of the Art

CVD experiments using MoCl5 and Mo(CO)6 described above, the dominant phase formed at lower temperatures (<800 °C) was γ-Mo2N, while the formation of δ−ΜoN was favoured at temperatures above 800 °C.

2.2. Guanidinato ligands

Since the first guanidinato complexes have been reported in 1970 by Lappert et al.[133],

− − guanidinato ligands, { )NC(NR 22 }and {(NR)2 (NRC 2 ) }, have been found to support all metal ions across the periodic table. The characteristic feature of all guanidinato complexes is a trigonal planar CN3 unit substituted by organic groups being isoelectronic to the monoanionic

− hydrogen carbonato HCO3 ligand. A detailed discussion about guanidinato complexes is far beyond the scope of this work, but can be found in some more recent reviews[134, 135], so only short overview will be given here.

The steric and electronic properties of the guanidinato group can be tuned by variation of the 2 hydrocarbon residues, R. In the case of a planar CN3 unit, all atoms are sp -hybridized, and thus, an overlap of all lone pairs at the nitrogen atoms is possible, which leads to very short C- N distances. This effect has been claimed as Y-aromaticity[134]. Guanidinato ligands are highly tuneable, and the bonding situation to metal centres can vary considerably. The type I

− 1 [136] ligands, { )NC(NR 22 } are known to bind via the imido N atom in an η mode . The type II

− 2 1 3 congeners, {()NR 2 C()NR 2 }, bind in an η fashion involving the N and N atoms, and the dialkylamido group of N2 is not coordinated[134, 137-141]. Type II represents the most common form of guanidinato ligands. There are also a few examples known where the guanidinate is bonded as a dianionic ligand; as well, neutral, not deprotonated guanidines can bind as donor ligands[134, 142].

22 2. State of the Art

R R

N R N R

LnM N LnM N (2.7)

N R N R

R R

1,3 - diazaallyl- Iminium-diamide- resonance-form resonance-form

The Y-aromaticity of the type II has two possible resonance forms (Equation 2.7). The 1,3 – diazaallyl-resonance-form is effectively an aminidate bearing an amido ligand on the C-atom.

The endocyclic CN2-group possesses delocalized π-electron system. The iminium-diamide resonance form exhibits the participation of the exocyclic NR2 group in the π-electron system of the CN3 group. A structural indication of the extent of π-delocalisation may be the extent of planarization of the amido group. An aminidate is an electronically flexible ligand system displaying compatibility with transition metals in a range of oxidation states[134]. The potential for a contribution from the iminium-diamide resonance form means an increase of flexibility. The π-donor ability of guanidinates and hence the compatibility with electron deficient metal ions is augmented by any contribution from this form. Of course, these conclusion can only be drawn from crystal structures or dynamic NMR studies. If the diazaallyl form is more dominant in the complex, the endocyclic C-N bonds have partial double bond character, thus being significantly shorter than the exocyclic one. On the other hand a shortening of the exocyclic C-N bond means an increased contribution of the 2 iminium diamide form. The complex [Ti(η -(NPh)2CNEt2)2Cl2] represents a special case. The guanidinato ligands in this complex are crystalographically equivalent and the bond length of the exocyclic C-N bond are indistinguishable from the endocyclic bonds, indicating considerable delocalisation of the uncoordinated nitrogen lone pair into the ligand π- system.[143]

23 2. State of the Art

R R

N N (A) LnMNR2 + C R2N LnM

N N

R R (2.8) R R R

N N N LnM-Cl LiNR + 2 C R2N Li R2N LnM (B) -LiCl N N N

R R R Guanidinato complexes are mainly synthesized by the insertion of a carbodiimide RN=C=NR into a metal amide bond (A) or alternatively the addition of a lithium amide to a carbodiimide, followed by salt elimination from a metal halide complex (B) (Equation 2.8). The reaction mechanism, which is similar to the formation of carboxylates from a grignard-reagent and

CO2 start with a nucleophilic attack of a nitrogen atom to the metal. After the formation of a four membered LnM-N-C-N transition state, the amido group migrates from the metal to the C-atom, and the guanidinate is formed. Both possible reaction pathway have been successfully applied in the synthesis several complexes. The insertion pathway (A) has been applied in the synthesis of guanidinato complexes from Aluminium and Gallium[141], Tantalum[144], Titanium[145] or Lanthanides[19, 138], while the salt elimination pathway was used for the synthesis of Titanium[18] or Tungsten[126]. Till the day there are only two reports on guanidinates of tungsten[126, 146]. The chemistry of molybdenum guanidinates is only slightly more developed. Cotton et al. reported the synthesis Molybdenum-Molybdenum multiple bonds, that were stabilized by bicyclic guanidinato ligands[147-151] It obvious that the chemistry of guanidinates of tungsten and molybdenum still bears many room for new explorations.

24 3. Synthesis of the Tungsten Compounds

3. Synthesis of the Tungsten Compounds

The metal organic precursors that have been applied for CVD of tungsten nitride, except

W(CO)6 are all based on imido ligands. Though the reactivity of these compounds towardsis reduced compared to Tantalum[144] or Hafnium[152], it still very high. The introduction of a chelating ligand into the coordination sphere of the metal centre, e.g. by replacing an amido ligand by the bidentate guanidinato ligand leads to an increase of the coordination number. Consequently the free coordination sited in the resulting compound are reduced, leading to a reduced reactivity towards oxygen and moisture, and possibly affect the reactivity towards ammonia as well as the decomposition pathways. Additionally the thermal stabilty of the complex should be increased. For the use of the compounds as CVD precursors, a main question is how the increased complexity of the compound influences the behaviour during deposition experiments. Addionally the more complex ligand bears a higher carbon content, that might be incorporated into the deposited films. The first approach to tungsten guanidinates that we tried, was the insertion of carbodiimides to (bis-tert-butylimido)(bisalkylamido)tungsten, but this reaction path did not lead to satisfactory results, presumably because of the steric bulk at the tungsten center and problems with the chemoselectivity of the insertion reaction (i.e., parallel presence of imido and amido groups). Alternatively, we investigated salt-elimination reactions, starting with octahedral

[WCl2(Nt-Bu)2py2] (1) The following compounds have been fully characterized by Elemental Analysis, 1H- and 13C-NMR and EI-MS. Some examples were also studied by single crystal X-ray diffraction. The details of the synthesis procedures and the full documentation of the analytical data are given in the experimental part (Chapter 8).

3.1. [WCl2(Nt-Bu)2py2] (1)

The tungsten compound 1 was synthesized by a reaction scheme first reported by Becker et al.[59], which we modified into a protocol for a one-pot synthesis (Equation 3.1 and 3.2).

2 WCl6 + 8 HNt-Bu(SiMe3) → [WCl2(Nt-Bu)2(H2Nt-Bu)]2 + 6 Me3SiCl + 2 H2Nt-Bu(SiMe3)Cl (3.1) [WCl2(Nt-Bu)2(H2Nt-Bu)]2 + 4 py → [WCl2(Nt-Bu)2py2] + 2 H2Nt-Bu (3.2)

1 The H NMR spectrum in C6D6 showed the expected singlet for the t-Bu moieties of the imido groups at 1.53 ppm and three signals with the expected intensity ratios at 6.51, 6.84,

25 3. Synthesis of the Tungsten Compounds and 9.10 ppm for the pyridine ligands in accordance with the monomeric octahedral trans- chloro structure revealed by single crystal X-ray diffraction, as discussed below.

Single Crystrals of [WCl2(Nt-Bu)2py2] 1 were obtained by crystallization from toluene. The

Complex crystalizes in space group Pna21 with four molecules per unit cell The molecular structure of 1 is shown in Figure 11. Bonding parameters are given in Table 5. Details of the X-ray diffraction data collection are compiled in Table 21 in the experimental part. The molecular structure reveals a tungsten center with a distorted octahedral coordination, two

Figure 11: Molecular Structure of [WCl2(Nt-Bu)2py2] (1) in the solid state as ORTEP plot, thermal ellipsoids at the 50 % probability level. For selected bondlenghts and angles see Table 3. trans-chloro ligands, and cis coodinated imido and pyridine ligands, which was the typical [153-156] coordination scheme for WCl2(NR)2Lx (x = 1, 2) compounds . The octahedral coordination can be specified by the sum of angles around the tungsten being 360.0(7) , 356.3(5), and 355.5(5)° for the three space directions. By looking at the two W-N-C angles, one can see that both imido ligands are bonded slightly different. The W(1)- N(4)-C(41) angle is nearly linear (174.9(4)°), whereas the other imido ligand is coordinated in a slightly bent fashion (161.3(3)°). None of the imido ligands is in the category of strongly bent ligands (bond angle < 150°). The difference of angles might also be

26 3. Synthesis of the Tungsten Compounds induced by interactions with the π systems of the pyridine ligand or by crystal-packing forces. A more detailed analysis of the structural features would request DFT-type quantum chemical calculation, which are beyond the scope of this work. Table 5: Selected Bond leghts (Å) and Angles (°) for Table 6: Comparison of W=N-C angles and W=N- [WCl2(Nt-Bu)2py2] (1). For the molecular structure see distances figure 11 angles Bondlenghts (°) (Å) Ref. W(1)-N(3) 1.761(4) W(1)-Cl(1) 2.4150(12)

W(1)-N(4) 1.746(4) W(1)-Cl(2) 2.4192(13) This W(1)-N(1) 2.414(4) N(3)-C(31) 1.461(6) WCl2(Nt-Bu)2py2 161.9(3) 1.761(4) Work W(1)-N(2) 2.396(4) N(4)-C(41) 1.454(7) 174.9(4) 1.746(4)

[154] N(4)-W(1)-N(3) 107.86(18) N(2)-W(1)-Cl(1) 83.23(10) WCl2(Nt-Bu)2bipy 166.4(5) 1.747(5) N(4)-W(1)-N(2) 162.60(17) N(1)-W(1)-Cl(1) 81.56(10) 162.7(4) 1.754(4)

N(3)-W(1)-N(2) 89.44(16) N(4)-W(1)-Cl(2) 94.67(14) [153] N(4)-W(1)-N(1) 87.79(17) N(3)-W(1)-Cl(2) 95.63(14) WCl2(NPh)2bipy 165.6(12) 1.781(11) N(3)-W(1)-N(1) 164.33(16) N(2)-W(1)-Cl(2) 81.50(10) 164.4(11) 1.787(14)

N(2)-W(1)-N(1) 74.89(14) N(1)-W(1)-Cl(2) 82.06(11) [156] WCl (Nt-Bu) (t-Bu-dab) 160.8(7) 1.756(8) N(4)-W(1)-Cl(1) 96.07(14) Cl(1)-W(1)-Cl(2) 160.01(5) 2 2 161.8(7) 1.758(8) N(3)-W(1)-Cl(1) 97.05(13) C(31)-N(3)-W(1) 161.9(3) C(41)-N(4)-W(1) 174.9(4) t-Bu-dab = N,N'-di-tert-butyl-1,4-diaza-1,3- butadiene The W-N distance to the pyridine ligand in a position trans to the linear imido group is slightly shorter than the same distance trans to the bent one. Also, the ligand trans to the bent imido group is twisted out of the N(1)-W(1)-N(3) plane a bit stronger than the other pyridine is out of the N(2)-W(1)-N(4) plane, which means that the less-twisted pyridine ligand trans to the linear imido group acts as a better π acceptor and thus exhibits a shorter W-N bond length. Comparing these angles and the W-N distances of 1 with the situation of other closely related

WCl2(NR)2Lx compounds (Table 6), all of the other reference compounds exhibit very similar angles and bond lengths, whereas both imido ligands in 1 are bonded distinctly differently. This difference is apparently due to the presence of the chelating ligand in all of the reference compounds of Table 6. The linkage between the two pyridine groups of the bipy ligand causes a twisting of the groups to be energetically unfavorable and consequently forces the imido ligands into a more symmetric environment.

3.2. [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2] (3)

Treatment of [WCl2(Nt-Bu)2py2] with i-Pr2NC(Ni-Pr)2Li(TMEDA) (2) in n-hexane and recrystallization from n-hexane produces bis(tert-butylimido)-chloro-(tetraisopropyl) guanidinato-tungsten (3) as yellow crystals in 66.6 % yield. Attempts to substitute the remaining chloro ligand by another guanidinato group, e.g., by using an excess of 2,

27 3. Synthesis of the Tungsten Compounds prolonged reaction times and harsher conditions (reflux) failed. The EI-MS of 3 shows the expected molecular ion peak [M+] at m/z = 587 (8% rel. int.) and another peak of similar + intensity at m/z = 572 that is assigned to the loss of one methyl group [M - CH3]. The peak at m/z = 446 (18% rel. int.) is attributed to a fragment ion containing the guanidinate ligand and the remaining chloro ligand bonded to the tungsten centre [M+ - 2Nt-Bu]. The peak at m/z = 291 (10% rel. int.) is likely to relate to a protonated W-Nt-Bu ion, resulting from a fragmented guanidinate ligand . All peaks with m/z ≤ 226 can be assigned to fragmentation of the guanidinato ligand, except the peak at m/z = 58 (35% rel. int.), which can have its origin in fragmentation of the guanidinato ligand as well as from the t-Bu residue at the imido ligand.

1 Figure 12: H NMR spectrum of [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2] (3). The 1H NMR spectrum (figure 12) of 3 displays one singlet for the imido groups and four different doublets for the CH3 groups of the four isopropyl rests of the guanidinate ligand. The spectrum also reveals three septets for the CH groups, two at 4.08 and 4.15 ppm, belonging to the isopropyl groups close to the tungsten, and one at 3.17 ppm for the isopropyl groups of the amide. The doublets at 0.97 and 0.98 ppm are pointing to very similar chemical environments and are assigned to the bis-isopropylamido group N(i-Pr)2. In addition, the rotation around the C-N(i-Pr)2 bond axis should be slightly hindered because of steric bulk. Upon heating to 30 °C, the splitting of the two signals disappears. The asymmetric pentacoordination of the tungsten centre causes the CH3 moieties of the Ni-Pr groups 28 3. Synthesis of the Tungsten Compounds

Figure 13: (top) Molecular Structure of [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2] (3) in the solid state as ORTEP plot, thermal ellipsoids at the 50 % probability level. For selected bond lengths and angles see table 7. (bottom) Projection of the molecular structure of [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2] (t-Bu and i-Pr groups omitted fo claritiy) perpendicular to the CN3 plane

29 3. Synthesis of the Tungsten Compounds coordinated to the tungsten to be different, exhibiting resonances at 1.11 and 1.54 ppm, which remains unchanged upon heating to 60 °C. This interpretation of the NMR data referring to the molecular structure in solution at more or less ambient conditions was substantiated by the molecular structure of [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2] in the solid state obtained from single- crystal X-ray diffraction studies. Obviously the complex does not exhibit structural fluctuation (e.g. Pseudorotation), as one would expect for a five coordinated complex, maybe induced by the bidentate guanidinato ligand anchoring the structure. Table 7: Selected Bond leghts (Å) and Angles (°) for [W(Nt- Bu)2Cl{(Ni-Pr)2CNi-Pr2] (3). See figure 13 for the molecular structure W(1)-Cl(1) 2.3760(16 ) N(13)-C(130) 1.461(7) W(1)-N(10) 2.089(5) N(14)-C(140) 1.429(9) W(1)-N(11) 2.215(4) N(10)-C(100) 1.338(7) W(1)-N(13) 1.750(5) N(11)-C(100) 1.307(7) W(1)-N(14) 1.755(5) N(12)-C(100) 1.414(7)

N(13)-W(1)-N(10) 100.4(2) W(1)-N(10)-C(100) 96.5(3) N(13)-W(1)-N(11) 135.0(2) W(1)-N(11)-C(100) 91.7(3) N(13)-W(1)-N(14) 117.7(2) W(1)-N(13)-C(130) 164.2(5) N(13)-W(1)-Cl(1) 94.62(17) W(1)-N(14)-C(140) 163.0(5) N(10)-W(1)-N(11) 60.58(17) N(10)-C(100)-N(11) 110.6(5) N(10)-W(1)-N(14) 103.7(2) N(10)-C(100)-N(12) 121.5(5) N(10)-W(1)-Cl(1) 14.93(14) N(11)-C(100)-N(12) 127.9(5) N(11)-W(1)-N(14) 112.3(2)

Single Crystrals of [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2] 3 were obtained by crystallization from hexane. The Complex crystalizes in space group P-1 with two molecules per unit cell The molecular structure of 3 is shown in Figure 13. Bonding parameters are given in Table 7. Details of the X-ray diffraction data collection are compiled in Table 21 in the experimental part.

The coordination at the tungsten centre of the mononuclear compound [W(Nt-Bu)2Cl{(Ni-

Pr)2CNi-Pr2] is described as strongly distorted trigonal bipyramidal, with the chloro ligand in the axial position and the imido groups and one of the coordinated N atoms of the guanidinato ligand in the equatorial position, with a sum of angles of 359.1(6) ° in the equatorial plane. The second N atom of the guanidinato ligand is shifted out of the ideal axial position due to its linkage to the rest of the ligand; nevertheless, the sum of angles around the tungsten in the Cl(1)-N(10)-N(11) plane is 362.3°, as expected for a trigonal bipyramidal coordination geometry. In contrast to [WCl2(Nt-Bu)2py2], for 3 now both imido ligands are bonded in a bent way with similar bond lengths (1.750(5) and 1.755(5) Å) and angles (163.0(5) and 164.2(5) °). Weaker π interactions of the metal with the bidentate guanidinato ligand might be an explanation; also, crystal-packing forces cannot be excluded. The angle between the two imido groups (117.7(2)°) is slightly enlarged compared to that of [WCl2(Nt-Bu)2py2] 30 3. Synthesis of the Tungsten Compounds

(107.9(2)°) due to the lower coordination number of the metal centre. The W-Cl distance

(2.376(2) Å) is slightly shorter compared to those of [WCl2(Nt-Bu)2py2] (2.415(1) and 2.419(1) Å, respectively) again due to the lower coordination number and the missing chorine atom in the trans position competing for electron density. As a consequence of the steric demand of the isopropyl group at N(12), the whole amide group is twisted 59.8(9) ° out of the N(10)-C(100)-N(11) plane, possibly resulting in a reduced conjugation of the lone pair at 2 N(12) which is sp -hybridised (Σangles= 357.6(16)° with the delocalized π system of the guanidinate ligand. The isopropyl groups at N(10) and N(11), close to the tungsten, are positioned in different chemical environments, which was already confirmed by 1H NMR studies. The first one, bonded to the equatorial N(11) atom, is close to the apical chloro ligand, whereas the other one, bonded to the axial N(10) atom, is positioned toward the imido ligands. The bite angle N(10)-C(100)-N(11) of the “free” CN3 moiety is slightly smaller

(110.6(5) °) than one would expect for a trigonal planar system 120° but the CN3 system it is still strictly planar (Σangles = 359.1(15)°). The bondlenght of the exocyclic C-N bond (1.414(7) Å) of the guanidintate is significantly longer than the two exocyclic bonds (1.307(7) and 1.338(7) Å, respectively), indicating the guanidinate binding dominantly in the 1,3 diazaallyl resonance form. (see Table 7).

3.3. [W(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] (4)

Reaction of W(NtBu)2Cl2py2 with one equivalent of the in situ formed Li(Ni-Pr)2CNMe2 and recrystallization from toluene gave [W(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] (4) as pale yellow crystals in 80 % yield. The proton NMR displays one singlet for the tert-butylimido group at 1.51 ppm and one at 2.19 ppm for two methyl groups at the guanidinato ligand. It also reveals two doublets (1.09 and 1.57 ppm) and two septets (3.79 and 3.91 ppm) belonging to two chemically different isopropyl groups. The EI-MS of [W(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] shows the expected molecular ion peak [M+] at m/z = 531 (2% relative intensity) and another peak of + similar intensity at m/z = 516 that was assigned to the loss of one methyl group [M - CH3]. The peak at m/z = 390 (9% relative intensity) was attributed to a fragment ion containing the guanidinato ligand and the remaining chloro ligand bonded to the tungsten centre [M+- 2Nt- Bu].

31 3. Synthesis of the Tungsten Compounds

Figure 14: Molecular Structure of [W(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] (4) in the solid state as ORTEP plot, thermal ellipsoids at the 50 % probability level. For selected bond lengths and angles see table 8. Table 8: Selected Bond leghts (Å) and Angles (°) for [W(Nt- Bu)2Cl{(Ni-Pr)2CNMe2}] (4). For the molecular structure see figure 14 W(1)-Cl(1) 2.385(2) N(1)-C(10) 1.482(11) W(1)-N(1) 1.753(7) N(2)-C(20) 1.500(10) W(1)-N(2) 1.693(7) N(3)-C(1) 1.339(8) W(1)-N(3) 2.125(6) N(4)-C(1) 1.354(9) W(1)-N(4) 2.168(7) N(5)-C(1) 1.366(8)

N(1)-W(1)-N(2) 112.1(3) N(4)-W(1)-Cl(1) 86.54(17) N(1)-W(1)-N(3) 99.9(3) W(1)-N(3)-C(1) 95.7(5) N(1)-W(1)-N(4) 124.3(3) W(1)-N(4)-C(1) 93.3(4) N(1)-W(1)-Cl(1) 96.7(2) W(1)-N(1)-C(10) 162.6(6) N(2)-W(1)-N(3) 101.3(3) W(1)-N(2)-C(20) 168.0(6) N(2)-W(1)-N(4) 121.6(3) N(3)-C(1)-N(4) 109.3(6) N(2)-W(1)-Cl(1) 97.5(2) N(3)-C(1)-N(5) 126.2(7) N(3)-W(1)-N(4) 61.6(2) N(4)-C(1)-N(5) 124.4(7) N(3)-W(1)-Cl(1) 148.01(17)

Single crystrals of [W(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] 4 were obtained by crystallization from hexane. The Complex crystalizes in space group C2/c with eight molecules per unit cell The molecular structure of 4 is shown in Figure 14. Bonding parameters are given in Table 8.

32 3. Synthesis of the Tungsten Compounds

Details of the X-ray diffraction data collection are compiled in Table 21 in the experimental part. The strongly distorted trigonal bipyramidal coordination mode at the tungsten centre is expectedly quite similar to that of the closely related congeners [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-

Pr2}], that was reported above and [W(Nt-Bu)2Cl{(Ni-Pr)2CNEt2}] and [W(Nt-

Bu)2Cl{(NCy)2CNEt2}] that are reported below. the N atoms of the imido groups and one N atom of the guanidinato ligand form the equatorial plane with a sum of angles of 359(9) °, while the chloro ligand occupies an axial position (see table 10). The same is true for [W(Nt-

Bu)2Cl{(Ni-Pr)2CNMe2}] with a sum of angles in the CN3 plane of 359.9(20)°. The second N atom of the guanidinato ligand is shifted out of the ideal axial position due to its linkage to the rest of the ligand. Nevertheless, the sum of angles in the Cl(1)-N(3)-N(4) plane around the tungsten is 360.1(5)°, as expected for a trigonal bipyramidal coordination geometry. The bondlength of the exocyclic C-N is not significantly different from the analogue bond in

[W(Nt-Bu)2Cl{(Ni-Pr)2CNEt2}] (see below), so the slightly decreased steric bulk in [W(Nt-

Bu)2Cl{(Ni-Pr)2CNMe2}] does not result in a shorter C-N bond (see table 11).The imido ligands are bonded more asymmetric compared to the other complexes [W(Nt-Bu)2Cl{(Ni-

Pr)2CNi-Pr2], [W(Nt-Bu)2Cl{(Ni-Pr)2CNEt2}] and [W(Nt-Bu)2Cl{(NCy)2CNEt2}], with an extremely short W(1)-N(2) bondlength. The N-C bondlenghts are significantly longer than in the imido ligands of the analogue complexes. The bond angle between the two imido ligands is the longest observed of all complexes W(Nt-Bu)2Cl{(NR’)2CNR2}. This observation can be explained be the low sterical demand of the guanidinato ligand in [W(Nt-Bu)2Cl{(Ni-

Pr)2CNMe2}], that allows the widening of the angle between the two imido ligands. The

Bondlength of the W(1)-Cl(1) is equivalent to the analogue bond in [W(Nt-Bu)2Cl{(Ni-

Pr)2CNEt2}], but it is shorter than in [W(Nt-Bu)2Cl{(NCy)2CNEt2}] and longer than in

[W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2].

3.4. [W(Nt-Bu)2Cl{(Ni-Pr)2CNEt2}] (5)

The reaction of [Li{(Ni-Pr)2CNEt2}]x, generated in situ by the reaction of LiNEt2 and i-

PrN=C=Ni-Pr, with one equivalent of [WCl2(Nt-Bu)2py2] gave [W(Nt-Bu)2Cl{(Ni-

Pr)2CNEt2}] (5) as a yellow powder. [W(Nt-Bu)2Cl{(Ni-Pr)2CNEt2}] is the Ethyl analogue of 1 [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2]. The H NMR exhibits one singlet for the imido groups and two different doublets for the CH3 groups of the isopropyl rests of the guanidinate ligand. It also display a quartet and a triplet that were assigned to the ethyl group at the guanidinato 33 3. Synthesis of the Tungsten Compounds

ligand, In contrast to [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2] no hindered rotation of the NR2 group at the guanidinate was observed, which is most likely due to the minor steric bulk of the ethyl groups compared to the iso-propyl groups. The EI-MS was more or less identical to the one of + [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2], exhibiting the [M ]-peak at m/z = 559.

Single crystrals of [W(Nt-Bu)2Cl{(Ni-Pr)2CNEt2}] 5 were obtained by crystallization from hexane. The Complex crystalizes in space group P-1 with six molecules per unit cell The molecular structure of 5 is shown in Figure 15 Bonding parameters are given in Table 9. Details of the X-ray diffraction data collection are compiled in Table 22 in the experimental part.

The coordination sphere of the tungsten centre is analogue to the one of [W(Nt-Bu)2Cl{(Ni-

Pr)2CNi-Pr2], but in contrast to [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2] the C(1)-N(5) bond is significantly shorter than the equivalent bond in [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2] (see table 8).

A difference to [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2] is also the crystal packing. While the [W(Nt-

Bu)2Cl{(Ni-Pr)2CNi-Pr2] possesses a crystal structure with two molecules per unit cell,

[W(Nt-Bu)2Cl{(Ni-Pr)2CNEt2}] has six molecules per unit cell, where three molecules are

Figure 15: Molecular Structure of [W(Nt-Bu)2Cl{(Ni-Pr)2CNEt2}] (5) in the solid state as ORTEP plot, thermal ellipsoids at the 50 % probability level. For selected bond lengths and angles see table 9.

34 3. Synthesis of the Tungsten Compounds

Table 9: Selected Bond leghts (Å) and Angles (°) for [W(Nt- Bu)2Cl{(Ni-Pr)2CNEt2}] (5) For the molecular structure see figure 15 W(1)-Cl(1) 2.3838(19) N(1)-C(10) 1.461(10) W(1)-N(1) 1.738(6) N(2)-C(20) 1.465(9) W(1)-N(2) 1.734(6) N(3)-C(1) 1.345(9) W(1)-N(3) 2.130(5) N(4)-C(1) 1.328(8) W(1)-N(4) 2.181(6) N(5)-C(1) 1.362(9)

N(1)-W(1)-N(2) 111.5(3) N(4)-W(1)-Cl(1) 87.32(15) N(1)-W(1)-N(3) 100.5(3) W(1)-N(3)-C(1) 94.9(4) N(1)-W(1)-N(4) 129.7(3) W(1)-N(4)-C(1) 93.1(4) N(1)-W(1)-Cl(1) 95.26(19) W(1)-N(1)-C(10) 163.7(5) N(2)-W(1)-N(3) 102.1(2) W(1)-N(2)-C(20) 166.0(6) N(2)-W(1)-N(4) 117.9(3) N(3)-C(1)-N(4) 110.5(6) N(2)-W(1)-Cl(1) 97.5(2) N(3)-C(1)-N(5) 123.1(6) N(3)-W(1)-N(4) 61.3(2) N(4)-C(1)-N(5) 126.4(7) N(3)-W(1)-Cl(1) 148.06(18) crystallographically inequivalent. The bond angle between the two imido ligands is significantly smaller and the W-N distances are smaller compared than in [W(Nt-Bu)2Cl{(Ni-

Pr)2CNi-Pr2]. The tungsten-chlorine distance is slightly longer than in [W(Nt-Bu)2Cl{(Ni-

Pr)2CNi-Pr2], but it is still significantly shorter than in [WCl2(Nt-Bu)2py2]. The guanidinato ligand is exactly planar with a sum of angles in the CN3 plane being exactly 360.0(19) °.

3.5. [W(Nt-Bu)2Cl{(NCy)2CNEt2}] (6)

Analogue to di-iso-propylcarbodiimide also dicyclohexylcarbodiimide could be used to form a guanidinato ligand. When added to LiNEt2 in n-Hexane follow the reaction of a solution of 1 [WCl2(Nt-Bu)2py2] the complex W(Nt-Bu)2Cl{(NCy)2CNEt2} (6) is formed. The H NMR displays a singlet at 1.49 ppm for two imido ligand. A quartet and a triplet, assigned to the ethyl groups and a broad combination of multiplets belonging to the cyclohexyl groups. The carbon NMR is very similar to that of [W(Nt-Bu)2Cl{(Ni-Pr)2CNEt2}] corresponding to the NMR shift, with the exception of four additional signals at ~ 26 ppm, and at ~ 34 ppm for the cyclohexyl group that has four C atom that are inequivalent in the NMR spectrum while the iso-propylgroup has only two. The 13C NMR spectrum also proofs that the cyclohexyl are chemically inequivalent, due to the number of signal shown in the spectrum. The EI-MS + shows the expected [M ]-peak at m/z = 640, and the loss of NEt2 at m/z = 566. The peak at was assigned to the peak at m/z = 278, while the peak at m/z = 206 belongs to dicyclohexylcarbodiimide. All peaks with m/z ≤ 206 could be assigned to fragments of the

35 3. Synthesis of the Tungsten Compounds

Figure 16: Molecular Structure of [W(Nt-Bu)2Cl{(NCy)2CNEt2}] (6) in the solid state as ORTEP plot, thermal ellipsoids at the 50 % probability level. For selected bond lengths and angles see table 10. Table 10: Selected Bond leghts (Å) and Angles (°) for [W(Nt- Bu)2Cl{(NCy)2CNEt2}] (6) . For the molecular structure see figure 16 W(1)-Cl(1) 2.3924(11) N(1)-C(10) 1.449(5) W(1)-N(1) 1.764(3) N(2)-C(20) 1.453(5) W(1)-N(2) 1.742(3) N(3)-C(1) 1.334(5) W(1)-N(3) 2.190(3) N(4)-C(1) 1.344(5) W(1)-N(4) 2.097(3) N(5)-C(1) 1.380(5)

N(1)-W(1)-N(2) 110.96(16) N(4)-W(1)-Cl(1) 145.01(10) N(1)-W(1)-N(3) 136.82(14) W(1)-N(3)-C(1) 92.4(2) N(1)-W(1)-N(4) 99.77(14) W(1)-N(4)-C(1) 96.3(2) N(1)-W(1)-Cl(1) 94.44(11) W(1)-N(1)-C(10) 168.7(3) N(2)-W(1)-N(3) 111.12(14) W(1)-N(2)-C(20) 160.9(3) N(2)-W(1)-N(4) 103.09(15) N(3)-C(1)-N(4) 109.8(3) N(2)-W(1)-Cl(1) 101.32(12) N(3)-C(1)-N(5) 123.7(3) N(3)-W(1)-N(4) 61.41(12) N(4)-C(1)-N(5) 126.5(3) N(3)-W(1)-Cl(1) 86.52(8) guanidinato ligand, except the main peak at m/z = 58 (100% relative intensity) which is originated from isobutene that results from the fragmentation of the imido ligands.

Single crystrals of [W(Nt-Bu)2Cl{(NCy)2CNEt2] 6 were obtained by crystallization from hexane. The Complex crystalizes in space group P21/n with four molecules per unit cell The

36 3. Synthesis of the Tungsten Compounds molecular structure of 6 is shown in Figure 16. Bonding parameters are given in Table 10. Details of the X-ray diffraction data collection are compiled in Table 22 in the experimental part.

The structure is again very similar to the ones of [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2] and [W(Nt-

Bu)2Cl{(Ni-Pr)2CNEt2}]. The Bondlength of the W(1)-Cl(1) is elongated compared to [W(Nt-

Bu)2Cl{(Ni-Pr)2CNEt2}], but it is still shorter than in [WCl2(Nt-Bu)2py2]. (see table 9) The exocyclic C-N bond of the guanidinato ligand is shorter than the analogue bond in [W(Nt-

Bu)2Cl{(Ni-Pr)2CNi-Pr2], but on the other hand it is significantly longer than the analogue bond in [W(Nt-Bu)2Cl{(Ni-Pr)2CNEt2}]. This indicates that this bondlenght is mainly directed by steric effects. A decrease in the steric demand of the alkyl groups might lead to a further decrease of this bondlength.

3.6. [W(Nt-Bu)2NMe2{(Ni-Pr)2CNi-Pr2}] (7)

Aiming at an all-nitrogen coordination sphere at the tungsten centre and to characterize the substitution chemistry of the remaining chloro ligand, [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2] was treated with one equivalent of LiNMe2 in diethyl ether. After one day of stirring at room temperature and workup, the substitution product bis(tertbutylimido)- dimethylamido(tetraisopropyl)guanidinato-tungsten (7) was obtained as a yellow viscous oil. The EI-MS shows the molecular peak [M+] at m/z = 596 (1% rel int). The following daughter peaks mostly show a clean cleavage of the ligands similar to those described above for compound [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2]. The peaks below m/z = 226 (30% rel int.) arise only from ligand fragmentation and do not contain tungsten. The 1H NMR spectrum (figure

17) of [W(Nt-Bu)2NMe2{(Ni-Pr)2CNi-Pr2}] displays an interesting feature in comparison with that of [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2]. Although the signals for the i-Pr groups of the guanidinate ligand of [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2] were splitted, the NMR spectrum of

[W(Nt-Bu)2NMe2{(Ni-Pr)2CNi-Pr2}] does not show a splitting but only two doublets and two septets of equal intensity corresponding to only two different kinds of i-Pr groups. This phenomenon was also observed in the carbon NMR. Assuming a monomeric pentacoordinated structure similar to that of [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2], this situation can only be explained by a more symmetric coordination of the guanidinato ligand. In particular, a structure derived from a square pyramidal WN5 arrangement with the NMe2 group at the apical position and the two t-BuN imido groups in a cis position at the basal N4 plane, likewise the two i-PrN groups of the guanidinato ligand, would explain the different

37 3. Synthesis of the Tungsten Compounds

NMR feature of [W(Nt-Bu)2NMe2{(Ni-Pr)2CNi-Pr2}]. This might be caused by the higher steric demand (cone angle) of the dimethylamido ligand compared to that of a chlorine atom, which forces the complex into a more symmetrical structure with equivalent environments for the i-PrN groups coordinated to the tungsten centre (compared figure 18). Unfortunately, we were not able to grow single crystals of [W(Nt-Bu)2NMe2{(Ni-Pr)2CNi-Pr2}] suitable for X- ray diffraction studies to unambiguously clarify the molecular structure of [W(Nt-

Bu)2NMe2{(Ni-Pr)2CNi-Pr2}] in the solid state.

1 Figure 17: H NMR spectrum of [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNi-Pr2] (7).

Figure 18: Trigonal bipyramidal form of [W(Nt-Bu)2X{(Ni-Pr)2CNR2]. (left) vs. square pyramidal form (right)

38 3. Synthesis of the Tungsten Compounds

3.7. [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] (8)

The treatment of [W(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] 4 with one equivalent of lithium dimethylamide yields [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] as a yellow powder in 68 % yield. The proton NMR shows the singlets for the dimethylamido group, the methyl groups at the guanidinate, and the imido groups. Similar to [W(NtBu)2(NMe2){(Ni-Pr)2CNi-Pr2}] (7) the two isopropyl groups have equivalent NMR shifts, which points to a more symmetrical environment of the guanidinato ligand, which can be explained by the higher steric demand of the dimethylamido group compared to that of a chloro ligand, forcing the complex into a more symmetrical structure with equivalent environments for the iPrN groups coordinated to the tungsten centre. In addition, the higher π-donor ability of the amido group with respect to the chloro ligand has to be taken into account too. However, more detailed computational analysis of the particular bonding situation is beyond the scope of this work here. The EI-MS of

[W(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] reveals the molecular peak at m/z = 540 (12% relative intensity). The peak at m/z = 495 (100% relative intensity) is assigned to the cleavage of a dimethylamido group, the peak at m/z = 482 (7% relative intensity) is attributed to the loss of a tert-butyl group, and the peak at m/z = 454 (10% relative intensity) most likely relates to the loss of two isopropyl groups. Peaks with m/z = 171 can be assigned to fragmentation of the guanidinato ligand, except the peak at m/z = 58 (26% relative intensity), which may have its origin in fragmentation of the guanidinato ligand as well as from the t-Bu residue at the imido ligand. Unfortunately, we were not able to grow single crystals of [W(Nt-Bu)2Cl{(Ni-

Pr)2CNMe2}] suitable for X-ray diffraction studies to unambiguously clarify the molecular structure of [W(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] in the solid state.

3.8. [W(Nt-Bu)2(N3){(Ni-Pr)2CNi-Pr2}] (9)

On refluxing compound [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2] in a mixture of toluene/THF together with an excess of sodium azide the complex bis(tertbutylimido)-azido-(tetra-isopropyl)guanidinato-tungsten (9) was obtained as a viscous oil. [W(Nt-Bu)2(N3){(Ni-Pr)2CNi- -1 Pr2}] exhibits the characteristic, intense asymmetric νas(NN) vibration at 2086 cm in the IR spectrum. In contrast to that of [W(Nt-Bu)2NMe2{(Ni-Pr)2CNi-Pr2}], the 1H NMR of [W(Nt-

Bu)2(N3){(Ni-Pr)2CNi-Pr2}] again exhibits the splitted signals for the i-PrN groups as observed for complex [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2], but with a slightly smaller difference

39 3. Synthesis of the Tungsten Compounds in the chemical shift. The sterically nondemanding azido group expectedly behaves more like the chloro than like the alkylamido ligand. We suppose that [W(Nt-Bu)2(N3){(Ni-Pr)2CNi-

Pr2}] again was coordinated in a distorted trigonal bipyramidal fashion rather than being derived from a more symmetric square pyramidal structure, as suggested for [W(Nt-

Bu)2NMe2{(Ni-Pr)2CNi-Pr2}], creating different chemical environments for the i-PrN groups attached to the W center. As in the case of [W(Nt-Bu)2NMe2{(Ni-Pr)2CNi-Pr2}] above, we failed to grow single crystals of [W(Nt-Bu)2(N3){(Ni-Pr)2CNi-Pr2}] suitable for X-ray diffraction studies to unambiguously clarify the molecular structure of [W(Nt-Bu)2(N3){(Ni-

Pr)2CNi-Pr2}] in the solid state.

3.9. [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] (10)

Adding one equivalenz of LiBEt3H solution to compound [W(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] and heating it overnight yields the hydrido compound [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}]. The proton NMR shows the signal of the hydrido ligand at 12.59 ppm, which could be identified 1 by its tungsten satellites, revealing a coupling constant of JW-H = 96.8 Hz. This is an unusual downfield shift of a hydrido ligand, with only a few examples in the literature.[157- 159]According to the smaller steric bulk of the hydrido ligand, the NMR exhibits two different signals for the Ni-Pr groups, like described above for [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] but with a smaller splitting of the signals. Gas evolution occurs when [W(Nt-Bu)2H{(Ni-

Pr)2CNMe2}] is dissolved in methanol, which indicates the hydridic character of the W-H bond. The EI-MS nicely displays the molecular peak at m/z = 497 (43% relative intensity).

The fragmentation pattern of [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] is very similar to that of

[W(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}], referring to the nearly identical set of ligands of the two complexes. The IR spectrum exhibits an absorption band at 1854 cm-1, which is a typical value for W-H stretching frequencies[159, 160].

Single crystrals of [W(Nt-Bu)2H{(Ni-Pr)2CNMe2] (10) were obtained by crystallization from methylene chloride. The Complex crystalizes in space group P21/n with eight molecules per unit cell The molecular structure of 10 is shown in Figure 19. Bonding parameters are given in Table 11. Details of the X-ray diffraction data collection are compiled in Table 22 in the experimental part.

40 3. Synthesis of the Tungsten Compounds

Figure 19: Molecular Structure of [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}]·CH2Cl2 (10) in the solid state as ORTEP plot, thermal ellipsoids at the 50 % probability level. For selected bond lengths and angles see table 11.

Table 11: Selected Bond leghts (Å) and Angles (°) for [W(Nt- Bu)2H{(Ni-Pr)2CNMe2}].}]·CH2Cl2 (10) For the molecular structure see figure 19. W(1)-N(1) 1.758(2) N(1)-C(10) 1.456(3) W(1)-N(2) 1.752(2) N(2)-C(20) 1.457(3) W(1)-N(3) 2.149(2) N(3)-C(2) 1.357(3) W(1)-N(4) 2.1603(19) N(4)-C(2) 1.327(3) N(5)-C(2) 1.364(3)

N(1)-W(1)-N(2) 112.83(9) W(1)-N(3)-C(2) 94.79(15) N(1)-W(1)-N(3) 107.28(9) W(1)-N(4)-C(2) 95.19(15) N(1)-W(1)-N(4) 121.88(9) W(1)-N(1)-C(10) 158.56(18) N(2)-W(1)-N(3) 106.97(9) W(1)-N(2)-C(20) 165.81(19) N(2)-W(1)-N(4) 125.19(9) N(3)-C(2)-N(4) 108.6(2) N(3)-W(1)-N(4) 60.77(8) N(3)-C(2)-N(5) 125.6(2) N(4)-C(2)-N(5) 125.8(2)

One solvent molecule (methylene chloride) per complex has been included in the elementary cell. The hydrido ligand could not be located in the difference Fourier maps but was included in the structure refinement in a typical position and then refined. Compound [W(Nt-

Bu)2H{(Ni-Pr)2CNMe2}] adopts a trigonal bipyramidal coordination geometry quite similar

41 3. Synthesis of the Tungsten Compounds

to that of [W(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}], with the estimated position of the hydrido group at the axial place and the imido ligands and the N(4) atom forming the equatorial plane, where the sum of angles around the tungsten is 359.9°. (table 11) Compared to the chloro derivative the bond angle N(1)-W(1)-N(2) between the two imido ligand is slightly enlarged, and so is the largest bond angle between the two imido groups observed in all described complexes in this work. The size of this angle is most likely dependent on the sterical demand of the residual ligands, as the ligands in [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] have the smallest sterical demand of all described compounds, and the largest bond angle between the imido groups.

As the bonding situation of the imido ligands in [W(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}], was described being quite asymetric, the bond lenghts in [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] are again very similar, with crystallographically equivalent C-N bondlenghts and similar W-N distance, like e.g. in [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2]. Table 12. Comparison of structural parameters of [W(Nt-Bu)2X{(NR')2CNR}] taken from 1H NMR and X-ray diffraction Density/ No. Compound NMR-Shifts / ppm X-Ray Data; Bondlenghts / Å Angles / ° g·cm-3 i-Pr @ Cguanidinate- Nimide-W- X R R' t-Bu guanidinate W-Nimide Nimide-C Nexocyclic Nimide 3 Cl i-Pr i-Pr 1.47 1.11 1.753(5) 1.429(9) 1.414(7) 117.7(2) 1.396 1.54 1.750(5) 1.461(9) 5 Cl Et i-Pr 1.49 1.12 1.738(6) 1.461(10) 1.362(9) 111.5(3) 1.461 1.57 1.734(6) 1.465(9) 4 Cl Me i-Pr 1.51 1.09 1.693(7) 1.482(11) 1.366(8) 112.1(3) 1.507 1.57 1.753(7) 1.500(10)

7 NMe2 i-Pr i-Pr 1.48 1.26 n.a. n.a. n.a. n.a. n.a.

8 NMe2 Me i-Pr 1.51 1.27 n.a. n.a. n.a. n.a. n.a.

9 N3 i-Pr i-Pr 1.46 1.06 n.a. n.a. n.a. n.a. n.a. 10 H Me i-Pr 1.54 1.22 1.758(2) 1.456(3) 1.364(3) 112.83(9) 1.536 1.26 1.752(2) 1.456(3) 6 Cl Et Cy 1.49 n.a. 1.742(3) 1.449(5) 1.380(5) 110.96(16) 1.448 1.764(3) 1.453(5)

3.10. [W(Nt-Bu)2Cl{NC(NMe2)2}]2 (11)

The addition of lithiated 1,1,3,3-tetramethyl-guanidine, [LiNC(NMe2)2]x, to [WCl2(Nt-

Bu)2py2] gives the new complex bis(bis(tert-butylimido)-chloro-(1,1,3,3- tetramethyl)guanidinato-tungsten) (11). The dimeric structure was established by single crystal X-ray diffraction studies (see below, Figure 20). In the EI-MS spectra, only the monomer was detected showing one-half of the molecular ion [M/2+] at m/z = 475 (1.6% rel int) and a clean fragmentation pattern of the ligands with similar features to those discussed above for the related compounds. The 1H NMR reveals one signal at 1,52 ppm for the imido

42 3. Synthesis of the Tungsten Compounds

groups and a very broad signal for the NMe2 groups, which might indicate a dynamic behaviour of the guanidinato ligand at room temperature. The CN3 unit of the particular guanidinato-ligand requires a C-N bond with partial double-bond character. Thus, in the case of [W(Nt-Bu)2Cl{NC(NMe2)2}]2, it should be possible to probe this by freezing the rotation of the ligand. For this purpose, a low-temperature 1H NMR was recorded from -50 to +50 °C in steps of 10 °C. The spectrum at -50 °C shows two sharp signals at 2.39 and 3.05 ppm for each NMe2 group of the guanidinato ligand. Upon rising the temperature, coalescence of the two signals was observed at room temperature, whereas at 45 °C one sharp singlet at 2.7 ppm was observed. (see figure 20)

1 Figure 20: Temperature dependent H NMR of [W(Nt-Bu)2Cl{NC(NMe2)2}]2 (11)

Single crystrals of [W(Nt-Bu)2Cl{NC(NMe2)2}]2, 11 were obtained by crystallization from hexane. The Complex crystalizes in space group P21/c with two molecules per unit cell The molecular structure of 11 is shown in Figure 21. Bonding parameters are given in Table 13. Details of the X-ray diffraction data collection are compiled in Table 23 in the experimental part. The molecule structure shows a dimeric species with the µ2-bonded guanidinato ligands

{NC(NMe2)2}. The coordination sphere of the tungsten atoms is best described as two distorted trigonal bipyramids linked over the edges, where the angles around the tungsten in

43 3. Synthesis of the Tungsten Compounds the N(1)-N(2)-N(3) plane add to 359.8(5)°. Perpendicular to that, within the plane defined by Cl(1) and the bridging nitrogen atoms, the sum of angles is 362.4(3)°. The angle of 110.5° between the two equatorial t-BuN imido groups is quite similar to the situation observed for the mononuclear compound [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2}], but the W-N-C-

Figure 21: Molecular Structure of [W(Nt-Bu)2Cl{NC(NMe2)2}]2 (11) in the solid state as ORTEP plot, thermal ellipsoids at the 50 % probability level. For selected bond lengths and angles see table 13.

Table 13: Selected Bond lengths (Å) and Angles (°) for [W(Nt-Bu)2Cl{NC(NMe2)2}]2 (11). See figure 21 for the molecular structure. W(1)-Cl(1) 2.4509(10) N(2)-C(20) 1.463(6) W(1)-N(1) 1.753(4) N(3)-C(30) 1.328(5) W(1)-N(2) 1.757(4) C(30)-N(4) 1.334(7) W(1)-N(3) 2.059(3) C(30)-N(5) 1.367(6) N(1)-C(10) 1.450(5)

N(1)-W(1)-N(2) 110.54(18) W(1)-N(3)-C(30) 122.8(3) N(1)-W(1)-N(3) 98.42(15) N(3)-C(30)-N(4) 123.6(4) N(1)-W(1)-Cl(1) 90.52(11) N(3)-C(30)-N(5) 119.8(5) N(2)-W(1)-N(3) 106.79(15) N(4)-C(30)-N(5) 116.6(4) N(2)-W(1)-Cl(1) 94.44(12) C(30)-N(4)-C(31) 119.5(4) N(3)-W(1)-Cl(1) 152.13(10) C(30)-N(4)-C(32) 121.6(4) W(1)-N(1)-C(10) 166.0(3) C(30)-N(5)-C(33) 119.8(4) W(1)-N(2)-C(20) 159.0(3) C(30)-N(5)-C(34) 122.1(4)

44 3. Synthesis of the Tungsten Compounds

angles are smaller than in [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2}]. Interestingly, though a second chlorine atom in the trans position is missing, the W(1)-Cl(1) bondlenghts is even longer than in [WCl2(Nt-Bu)2py2] and in [W(Nt-Bu)2Cl{(Ni-Pr)2CNi-Pr2}], which is maybe due to the presence the guanidinato ligand, pulling electron density out of the W-N bond. The CN3 unit of the guanidinato ligands is nearly perfectly trigonal planar, with all N-C-N angles near 120° and a sum of angles around C(30) of exactly 360.0(13)°. Also, all other bond angles in the guanidinato ligand, including the W(1)-N(3)-C(30) angle, are close to 120° (see Table 13). The sp2 hybridization of the atoms enables the lone pairs of the N atoms to effectively donate electron density toward the tungsten. The comparably short C-N bond distances also indicate the delocalized CN multiple bond character in the CN3 unit. The observed twisting of the guanidinato ligands toward the W(1)-N(3)-W(1) plane is likely to reduce the electron donation of the dimethylamido groups toward the tungsten centre. However, the temperature- dependent 1H NMR showed a hindered rotation of the ligand, pointing to the issue of electron-density donation to the tungsten centre. This dynamic situation finally allows a formal configuration of 18 valence electrons for each tungsten.

3.11. [W(Nt-Bu)2(N3){NC(NMe2)2}]2 (12)

Refluxing [W(Nt-Bu)2Cl{NC(NMe2)2}]2 in toluene with sodium azide leads to a chloride vs. azide exchange and results in the analogous azido compound [W(Nt-Bu)2(N3){NC(NMe2)2}]2.

[W(Nt-Bu)2(N3){NC(NMe2)2}]2 also exhibits a characteristic intense asymmetric νas(NN) vibration at 2072 cm-1 in the IR spectrum. The NMR spectra are very similar to those of

[W(Nt-Bu)2Cl{NC(NMe2)2}]2, but the guanidinato ligand exhibits one sharp signal for both

NMe2 groups already at room temperature. All structural details of [W(Nt-

Bu)2(N3){NC(NMe2)2}]2 referring to the dimeric structure and the W2N2 ring are very similar to those of [W(Nt-Bu)2Cl{NC(NMe2)2}]2, of course with the exception of the presence of the terminal azido ligand rather than a chloro group.

Single Crystals of [W(Nt-Bu)2(N3){NC(NMe2)2}]2 12 were obtained by crystallization from toluene. The Complex crystallizes in space group P21/n with two molecules per unit cell The molecular structure of 12 is shown in Figure 21. Bonding parameters are given in Table 14. Details of the X-ray diffraction data collection are compiled in Table 23 in the experimental part.

45 3. Synthesis of the Tungsten Compounds

Figure 22: Molecular Structure of [W(Nt-Bu)2(N3){NC(NMe2)2}]2 (12) in the solid state as ORTEP plot, thermal ellipsoids at the 50 % probability level. For selected bond lengths and angles see table 14. Table 14: : Selected Bond lengths (Å) and Angles (°) for [W(Nt-Bu)2(N3)l{NC(NMe2)2}]2 (12). See figure 22 for the molecular structure. W(1)-N(1) 1.750(6) N(3)-C(30) 1.337(9) W(1)-N(2) 1.758(6) C(30)-N(4) 1.354(9) W(1)-N(3) 2.069(6) C(30)-N(5) 1.354(9) W(1)-N(6) 2.137(6) N(6)-N(7) 1.195(8) N(1)-C(10) 1.462(9) N(7)-N(8) 1.152(9) N(2)-C(20) 1.460(9)

N(1)-W(1)-N(2) 109.3(3) N(3)-C(30)-N(4) 121.7(6) N(1)-W(1)-N(3) 100.1(2) N(3)-C(30)-N(5) 120.2(6) N(1)-W(1)-N(6) 95.4(3) N(4)-C(30)-N(5) 118.1(6) N(2)-W(1)-N(3) 104.9(2) C(30)-N(4)-C(33) 118.0(6) N(2)-W(1)-N(6) 94.1(3) C(30)-N(4)-C(34) 122.2(7) N(3)-W(1)-N(6) 149.8(2) C(30)-N(5)-C(31) 119.0(6) W(1)-N(1)-C(10) 154.6(5) C(30)-N(5)-C(32) 123.7(7) W(1)-N(2)-C(20) 167.2(5) W(1)-N(6)-N(7) 129.8(5) W(1)-N(3)-C(30) 124.1(5) N(6)-N(7)-N(8) 174.9(8)

The crystal structure data of [W(Nt-Bu)2(N3){NC(NMe2)2}]2 is very similar to the one of

[W(Nt-Bu)2Cl{NC(NMe2)2}]2, except the C-amido bonds in the guanidinato ligands that are equivalent in this structure, while the C(30)-N(4) bond in [W(Nt-Bu)2Cl{NC(NMe2)2}]2 was

46 3. Synthesis of the Tungsten Compounds significantly shorter than the C(30)-N(5) bond. (see Table 13 & Table 14). The bonding situation of the imido ligands is almost identical in both dimeric complexes [W(Nt-

Bu)2Cl{NC(NMe2)2}]2 and [W(Nt-Bu)2(N3){NC(NMe2)2}]2. The N(1)-W(1)-N(2) angle in

[W(Nt-Bu)2(N3){NC(NMe2)2}]2 is only slightly smaller compared to [W(Nt-

Bu)2Cl{NC(NMe2)2}]2, and the W-N bonds are crystallographically equivalent. A slight difference is observed in the N-C bonds, which are equivalent in [W(Nt-

Bu)2(N3){NC(NMe2)2}]2, while there is a significant difference between the two bondlenghts in [W(Nt-Bu)2Cl{NC(NMe2)2}]2. The W(1)-N(6) bondlength of the azide ligand is significantly shorter than the analogue W(1)-Cl(1) bond in [W(Nt-Bu)2Cl{NC(NMe2)2}]2. The bondlength of the W(1)-N(3) of the guanidinato ligand is slightly longer than in [W(Nt-

Bu)2Cl{NC(NMe2)2}]2. Two different resonance form are possible for the azide group. (Figure 23) Form A includes a N-N single bond close to the M-N bond and a N-N triple bond connecting the outer nitrogen atoms. This form would require two significantly different N-N bondlengths with the outer one being shorter. Form B possesses two N-N double bonds. This form require two equal N-N bondlenghts in the azido group. The bondlength of the N(6)-N(7) is longer (1.195(8) Å) than the N(7)-N(8) bond (1.152(9) Å) so obviously resonance form A is dominant in the complex.

Figure 23: Different resonance forms of an azide ligand

2 3.12. [(W(Nt-Bu)2(N3)(µ -N3)py)]2 (13)

Refluxing [WCl2(Nt-Bu)2py2] with an excess of NaN3 in toluene leads to the dimeric bis-(bis- tert-butylimido)-bisazido-pyridine tungsten (13) in 43 % yield. It shows the characteristic IR absorption band at 2082 cm-1. The 1H NMR shows a singlet for the tungsten and pyridine signals, corresponding to one pyridine ligand per tungsten. These observations match the 13C NMR data. 2 Single Crystals of [(W(Nt-Bu)2(N3)(µ -N3)py)]2 13 were obtained by placing a layer of hexane over a concentrated toluene solution of the crude product. The Complex crystallizes in

47 3. Synthesis of the Tungsten Compounds space group P-1 with three molecules per unit cell. The molecular structure of 13 is shown in Figure 24. A plot of the unit cell is given in figure 25. Bonding parameters are given in Table 15. Details of the X-ray diffraction data collection are compiled in Table 23 in the experimental part. The molecular structure reveals a dinuclear complex with two tungsten atoms now resting in a distorted octahedral environment, as indicated by the sum of angles around W(1) being 353.7(7) , 355.5(7) and 355.0(7) ° for all three space directions (see Table 15). These two

WN6 octahedrons are linked together in an edge-sharing mode. Two azido ligands are in bridging positions (typical head-bridging mode of azido ligands), whereas two other azido moieties are located head-bridging mode of azido ligands), whereas two other azido moieties 1 2 are located terminally and η -bonded. Comparing the W-N distances in [(W(Nt-Bu)2(N3)(µ -

N3)py)]2 with the analogous distances in the closely related known complex [W(Nt-

2 Figure 24 Molecular Structure of [(W(Nt-Bu)2(N3)(µ -N3)py)]2 (13) in the solid state as ORTEP plot, thermal ellipsoids at the 50 % probability level. For selected bond lengths and angles see table 15.

48 3. Synthesis of the Tungsten Compounds

2 Figure 25: Different molecules of [(W(Nt-Bu)2(N3)(µ -N3)py)]2 (13) in the unit cell. Imido and pyridine ligands are omitted for clarity. Table 15: Selected Bond lengths (Å) and Angles (°) for [(W(Nt- 2 Bu)2(N3)(µ -N3)py)]2 (13). See figure 24 for the molecular structure W(1)-N(1) 1.759(5) N(1)-C(10) 1.454(7) W(1)-N(2) 1.749(5) N(2)-C(20) 1.467(7) W(1)-N(3) 2.200(5) N(7)-N(8) 1.235(6) W(1)-N(7) 2.312(5) N(8)-N(9) 1.142(6) W(1)-N(10) 2.285(5) N(16)-N(17) 1.207(6) W(1)-N(16) 2.090(5) N(17)-N(18) 1.154(6)

N(1)-W(1)-N(2) 108.0(2) N(7)-W(1)-N(10) 65.04(15) N(1)-W(1)-N(3) 94.59(19) N(7)-W(1)-N(16) 81.54(18) N(1)-W(1)-N(7) 158.3(2) N(10)-W(1)-N(16) 86.95(18) N(1)-W(1)-N(10) 93.35(19) W(1)-N(1)-C(10) 166.9(4) N(1)-W(1)-N(16) 97.2(2) W(1)-N(2)-C(20) 169.1(4) N(2)-W(1)-N(3) 92.89(19) W(1)-N(7)-N(8) 118.0(4) N(2)-W(1)-N(7) 93.7(2) W(1)-N(10)-N(11) 122.7(4) N(2)-W(1)-N(10) 157.5(2) W(1)-N(16)-N(17) 125.8(4) N(2)-W(1)-N(16) 97.1(2) N(7)-N(8)-N(9) 179.0(6) N(3)-W(1)-N(7) 82.13(17) N(10)-N(11)-N(12) 178.2(6) N(3)-W(1)-N(10) 78.00(17) N(16)-N(17)-N(18) 175.9(6) N(3)-W(1)-N(16) 161.4(2) 2 [161] 2 Bu)2(N3)(η -N3)(t-BuNH2)]2 (Table 16), the new compound [(W(Nt-Bu)2(N3)(µ -N3)py)]2 reveals a somewhat more strongly bonded azido ligands, possibly due to the pyridine ligand which is a stronger π donor than the primary amine, and thus, back donation effects may be increased, which causes shorter W-Nazide bonds. Also, the pyridine ligands of [(W(Nt-

49 3. Synthesis of the Tungsten Compounds

2 Bu)2(N3)(µ -N3)py)]2 appear to be a bit more strongly bonded compared to those of

[WCl2(Nt-Bu)2py2].

2 Table 16: Comparison of W-N distances in [W(NtBu)2N3µ -N3L]2

L=py L=tBuNH2 W-Azide 2.090(5) 2.156(14) W-µ2Azide 2.285(5) 2.357(16) W-L 2.200(5) 2.185(14) W-Imide 1.749(5) 1.734(16) 1.759(5) 1.794(15) The W-N distance (2.20 Å) is significantly shorter than those found for [WCl2(Nt-Bu)2py2] (2.396 and 2.414 Å, respectively), which is correspondingly attributed to the azido group in the trans position being clearly a much stronger π acceptor than a chloro ligand. The bond 2 angles between the imido groups of [(W(Nt-Bu)2(N3)(µ -N3)py)]2 (angle N(1)-W(1)-N(2) =

107.98°) are identical with those found for the parent compound [WCl2(Nt-Bu)2py2]

(107.86°). A significant widening of that angle, similar to those of [W(Nt-Bu)2Cl{(Ni-

Pr)2CNi-Pr2}], [W(Nt-Bu)2Cl{NC(NMe2)2}]2, and [W(Nt-Bu)2(N3){NC(NMe2)2}]2, is not 2 observed, which is most likely due to the higher coordination number of [(W(Nt-Bu)2(N3)(µ - 2 N3)py)]2. [(W(Nt-Bu)2(N3)(µ -N3)py)]2 is one of the very few examples that has three crystallographic independent molecules per unit cell. In each of these, the η1-bonded azides are orientated differently so that the unit cell contains different chiral forms of the complex (Figure 25). 2 [(W(Nt-Bu)2(N3)(µ -N3)py)]2 is one of the very few compounds having three molecules per [162] unit cell. Other examples for this case are Cl(Mes)3W=N-N=W(Mes)3Cl or other compounds[163, 164]. A query in the Cambridge Crystal Structure Data base (CCDC) resulted only 54 hits for tungsten containing compounds having three molecules per unit cell, at a general content of 8101 tungsten containing compounds, so that Z=3 occurs in less than one percent of all crystal structures.

3.13. Conclusions for Chapter 3.

A series of new tungsten(VI)guanidinates focusing on all nitrogen coordination sphere of the type WN5 and WN6 was described. The compound [W(Nt-Bu)2Cl2py2] (1) was found to be a very good starting material for the introduction of guanidinato ligands at tungsten(VI). The mononuclear ,monomeric pentacoordinated complexes W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}

(8) and W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2} (10) are particularly interesting an candidates for subsequent studies aiming at depositions of WxN and WNxCy phases and thin film materials

50 3. Synthesis of the Tungsten Compounds by MOCVD, as has been described for related transition metal guanidinato compounds of titanium[18]or tantalum[144, 165]. The thermal characterization of these compounds as well as the MOCVD experiments are described in chapter 5.

51 4. Synthesis of the Molybdenum Compounds

4. Synthesis of the Molybdenum Compounds 4.1. Starting Compound for the Molybdenum synthesis

In order to transfer the results obtained for tungsten, to molybdenum chemistry, a suitable starting compound had to be found. The analogue synthetical route, described for tungsten

(see Chapter 3) could not be applied, because the existence of MoCl6 is still a subject of [23] speculation and the compound is not avialable . The Compound [MoCl2(Nt-Bu)2py2] that is analogue to [WCl2(Nt-Bu)2py2] as a starting compound, however is available by CO2- [131] elimination from MoCl2O2 and two equivalents of tert-Butylisocyanate , followed by the addition of pyridine (Equation 4.1). py 2BuNCO 2OMoCl t +−+ 2BuNCO py ⎯⎯⎯→ t − pyBu)(NMoCl (4.1 ) 22 2CO- 2 2 22

A very similar compound, being a suitable starting compound is [MoCl2(Nt-Bu)2(DME)] (DME = 1,2-Dimethoxyethane). Two possible reaction pathways were reported. The reaction of ammonium dimolybdate refluxed in DME in the presence of tert-Butylamine and trimethylchlorosilane was reported by Fox et al. (Equation 4.2)[166].

7224 14OMo)(NH t 2 +−+ 41BuNH 3SiClMe + 2 DME ⎯⎯⎯⎯⎯→2 2 t 2 (DME)Bu)-(NMoCl (4.2 ) ()3 2 OSiMe7- t 3ClBuNH-10- NH 2- NH 3 The second pathway is very similar. Here sodium molybdate is applied as a molybdenum source and a tertiary amine is applied as a base[167-169](Equation 4.3).

42 2MoONa t 2 +−+ 8BuNH 3SiClMe + 4 3 + DMENEt ⎯⎯⎯⎯⎯→ 2 t 2 (DME)Bu)-(NMoCl (4.3 ) ()3 2 OSiMe4- 2NaCl- 3NHCl4Et-

Another possible reaction pathway to obtain MoCl2(Nt-Bu)2py2 would be the reaction of [170] MoCl2(Nt-Bu)2(DME) with an excess of pyridine , but this chosing this pathway would mean a non-desirable additional reaction step. Due to high costs of MoCl2O2 and tert-Butyl isocyanate, that are commercially available only in relatively small batches, sodium molybdate was chosen as a cheap molybdenum source available in large quantities.

The starting compound [MoCl2(Nt-Bu)2(DME)] (14) was synthesized by refluxing a suspension of NaMoO4 in DME in the presence of tert-Butylamine, trimethylchlorosilane and triethylamine overnight. The 1H NMR displayed three singlets. The peak for the imido ligands was observed at 1.46 ppm, while the methylene groups of the DME ligand was observed at 3.22 ppm and the methoxy groups were shifted to 3.44 ppm. In the carbon NMR, the imido group was observed at 30.1 and 62.1 ppm while the DME ligand was observed at 70.8 and 71.8 ppm.

52 4. Synthesis of the Molybdenum Compounds

Figure 26: Molecular Structure of [MoCl2(Nt-Bu)2(DME)] (14) in the solid state as ORTEP-plot with thermal ellipsoids at the 50 % probability level. For selected bondlenghts and angles see Table 17.

Table 17: Selected Bond lengths (Å) and Angles (°) for [MoCl2(Nt-Bu)2(DME)] (14). See figure 26 for the molecular structure. Mo(1)-Cl(11) 2.4034(7) Mo(1)-O(1) 2.3943(18) Mo(1)-Cl(12) 2.4252(7) Mo(1)-O(2) 2.3773(18) Mo(1)-N(3) 1.722(2) N(3)-C(3) 1.440(3) Mo(1)-N(4) 1.735(2) N(4)-C(1) 1.451(3)

Cl(11)-Mo(1)-Cl(12) 158.95(3) N(3)-Mo(1)-N(4) 107.34(11) Cl(11)-Mo(1)-N(3) 93.56(7) N(3)-Mo(1)-O(1) 160.21(9) Cl(11)-Mo(1)-N(4) 99.27(7) N(3)-Mo(1)-O(2) 91.92(9) Cl(11)-Mo(1)-O(1) 79.50(5) N(4)-Mo(1)-O(1) 92.13(9) Cl(11)-Mo(1)-O(2) 83.45(5) N(4)-Mo(1)-O(2) 160.25(9) Cl(12)-Mo(1)-N(3) 99.25(7) O(1)-Mo(1)-O(2) 69.00(6) Cl(12)-Mo(1)-N(4) 92.80(7) Mo(1)-N(3)-C(3) 162.51(19) Cl(12)-Mo(1)-O(1) 82.90(5) Mo(1)-N(4)-C(1) 162.94(19) Cl(12)-Mo(1)-O(2) 79.52(5)

Single Crystrals of [MoCl2(Nt-Bu)2(DME)] 14 were obtained by crystallization from hexane. The Complex crystalizes in space group Pbca with eight molecules per unit cell The molecular structure of [MoCl2(Nt-Bu)2(DME)] is shown in Figure 26. Bonding parameters

53 4. Synthesis of the Molybdenum Compounds are given in Table 17. Details of the X-ray diffraction data collection are compiled in Table 24 in the experimental part. The complex has a distorted octahedral coordination sphere, with the chlorine ligands in trans position, and the imido groups are bonded in a cis position. The octahedral coordination is specified by the sum of angles around the molybdenum being 360.4(3), 355.2(2) and 355.0(2)° for the three space directions. (see table 17) Comparing this complex with

[WCl2(Nt-Bu)2py2] the bonding situation is very similar. The bond angle between the imido ligands is nearly identical (107.34(11)° in [MoCl2(Nt-Bu)2(DME)] vs. 107.84(18)° in

[WCl2(Nt-Bu)2py2]). The bond angle between the chlorine ligands is slightly smaller than in

[WCl2(Nt-Bu)2py2] (158.95(3)° in [MoCl2(Nt-Bu)2(DME)] vs. 160.01(5)° in [WCl2(Nt-

Bu)2py2]) due to the minimally smaller sterical demand of the DME ligand compared to the pyridine ligands.

Several similar molecular structures of the kind [MoCl2(NR)2(DME)] have already been described in literature[168, 171-175]. The imido ligands of these compounds are typicallysubstituted phenylimides like mesityl-[173], 2,6-iso-propylphenyl-[171] or 2,4,6- [175] tribromophenylimide . Comparing [MoCl2(Nt-Bu)2(DME)] to [MoCl2(Nt-Bu)(N(2,6-iso- propylphenyl)(DME)][171], the angle between the two imido ligands is a bit wider in

[MoCl2(Nt-Bu)2(DME)] than in the literature reported compound (107.34° vs 105.9°). Looking at the other reported compounds bearing aryl imido ligands the N=Mo=N bond angle is mostly even smaller, down to 103°. Most likely the aromatic systems of the imido has an influence on the angle. The bonding situation of the two imido ligands is obviously much more symmetric in

[MoCl2(Nt-Bu)2(DME)]. While the C-N bond of the tert-Butylimido ligand in [MoCl2(Nt-

Bu)(N(2,6-iso-propylphenyl)(DME)] is much shorter than in [MoCl2(Nt-Bu)2(DME)] (1.387 Å vs. 1.440 Å), the bondlength of the arylimido ligand is equivalent to the second tert-

Butylimido ligand in [MoCl2(Nt-Bu)2(DME)]. While the Mo=N-C angles are very different in the literature reported example, with a nearly linear Mo=N-Aryl bond angle, the analogue bond angles in [MoCl2(Nt-Bu)2(DME)] are crystallographically equivalent. Looking at the other reported complexes, the respective bond angles are very similar with [MoCl2(Nt-

Bu)2(DME)] if both imido ligands are equivalent. This indicates that these bond angles are mainly dependent on the substitution pattern on the imido ligands. Looking at the Cl-Mo-Cl bond angles, no significant differences could be observed, independent from the ligand.

54 4. Synthesis of the Molybdenum Compounds

4.2. [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] (15)

Analogue to the formation of [W(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}], the reaction of one equivalent of Li(Ni-Pr)2CNMe2 and [MoCl2(Nt-Bu)2(DME)] followed by recrystallization from hexane gave [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] (15) as a yellow powder in 65% yield. The proton NMR displays one singlet for the tert-butylimidogroups at 1.48 ppm and one at 2.26 ppm for two methyl groups at the guanidinato ligand. It also reveals two doublets at 1.04 ppm and at 1.62 ppm and a septet at 3.68 ppm belonging to the isopropyl groups. The EI-MS shows the expected molecular peak [M+] at m/z = 445 (2% relative intensity) and another peaks at + similar intensity at m/z = 401 assigned to the loss of the dimethylamido group [M -NMe2] and at m/z = 388 assigned to the loss of a tert-butylgroup. The peak at m/z = 302 (3% relative intensity) is assigned to the loss of both imido ligands, while the peak at m/z = 248 (3% relative intensity) derives from the loss of one imido ligand and the deinsertion of the di- isopropyl-carbodiimide. Peaks with m/z ≤ 170 could be assigned to fragmentation of the guanidinato ligand, except the peak at m/z = 71 (100% relative intensity) which derives from a tert-butylimido group and the peak at m/z = 58 which may have its origin in fragmentation of the guanidinato ligand as well as from the tBu residue at the imido ligand.

4.3. [Mo(Nt-Bu)2I{(Ni-Pr)2CNMe2}] (16)

Unfortunately we failed to get crystals suitable for single-crystal X-ray diffraction from

[Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}]. So we chose the iodo-derivative

[Mo(NtBu)2I{(NiPr)2CMe2] (16), for crystallographic studies. 16 is easily obtained by stirring

[Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] in THF with an excess of NaI for a day at room temperature. The spectroscopic properties of [Mo(NtBu)2I{(NiPr)2CMe2] are, as expected, nearly identical to those of [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}], so they are not discussed in detail.

Single Crystals of [Mo(NtBu)2I{(NiPr)2CMe2] 16 were obtained by crystallization from hexane. The Complex crystallizes in space group P21/c with four molecules per unit cell The molecular structure of [Mo(NtBu)2I{(NiPr)2CMe2] is shown in Figure 27. Bonding parameters are given in Table 18. Details of the X-ray diffraction data collection are compiled in Table 24 in the experimental part.

55 4. Synthesis of the Molybdenum Compounds

The strongly distorted trigonal bipyramidal coordination mode at the molybdenum centre is quite similar to those observed at the tungsten complexes 3, 9, 10 and 11, which were discussed above. The N atoms of the imido groups and one N atom of the guanidinato ligand form the equatorial plane with a sum of angles of 359.5(4) °, while the iodo ligand occupies

Figure 27 : Molecular structure of [Mo(Nt-Bu)2I{(Ni-Pr)2CNMe2}] (16) in the solid state as ORTEP-plot with thermal ellipsoids at the 50 % probability level. For selected bondlengths and angles see Table 18.

Table 18: Selected Bond lengths (Å) and Angles (°) for [Mo(Nt-Bu)2I{(Ni-Pr)2CNMe2}] (16). See figure 27 for the molecular structure Mo(1)-N(1) 1.742(3) N(3)-C(1) 1.341(4) Mo(1)-N(2) 1.735(3) N(4)-C(1) 1.344(4) Mo(1)-N(3) 2.112(3) C(1)-N(5) 1.357(4) Mo(1)-N(4) 2.187(3) N(1)-C(10) 1.465(4) Mo(1)-I(1) 2.7918(3) N(2)-C(20) 1.457(4)

N(1)-Mo(1)-N(2) 111.37(14) N(3)-Mo(1)-I(1) 153.06(7) N(1)-Mo(1)-N(3) 103.21(12) N(4)-Mo(1)-I(1) 92.25(10) N(1)-Mo(1)-N(4) 117.31(12) Mo(1)-N(1)-C(10) 155.9(2) N(2)-Mo(1)-N(3) 101.72(12) Mo(1)-N(2)-C(20) 173.7(2) N(2)-Mo(1)-N(4) 130.85(13) Mo(1)-N(3)-C(1) 95.6(2) N(3)-Mo(1)-N(4) 61.54(10) Mo(1)-N(4)-C(1) 92.21(19) N(1)-Mo(1)-I(1) 94.14(9) N(3)-C(1)-N(4) 110.1(3) N(2)-Mo(1)-I(1) 90.66(9) N(3)-C(1)-N(5) 125.6(3) N(4)-C(1)-N(5) 124.3(3)

56 4. Synthesis of the Molybdenum Compounds

An axial position. The second N atom of the guanidinato ligand is shifted out of the ideal axial position due to its linkage to the rest of the ligand. Nevertheless. The sum of angles in the I(1)-N(4)-N(3) plane around the molybdenum is 360.7(3) °. As expected for a trigonal bipyramidal coordination geometry. The CN3 plane of the guanidinato ligand possesses a sum of angles of exactly 360 ° in the CN3 plane. (Table 18).

4.4. [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}] (17)

Because [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] and [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] were the most promising tungsten compounds for MOCVD purposes, we tried to synthesized the molybdenum analogues of these compounds. But unfortunately neither the reaction with

LiBEt3H, nor with LiNMe2 lead to isolable products. A successful attempt obtain a derivative of [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] was its reaction with an excess of sodium azide, that

[Mo(Nt-Bu)2I{(Ni-Pr)2CNMe2}], distinguishing only in the chemical shifts. The EI-MS

Figure 98: Molecular structure of [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}] (17) in the solid state as ORTEP- plot with thermal ellipsoids at the 50 % probability level. For selected bondlengths and angles see Table 19.

57 4. Synthesis of the Molybdenum Compounds

Table 19: Selected Bond lengths (Å) and Angles (°) for [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}] (17). For the molecular structure see figure 28 Mo(1)-N(1) 1.7368(16) N(4)-C(1) 1.349(2) Mo(1)-N(2) 1.7510(15) C(1)-N(5) 1.366(2) Mo(1)-N(3) 2.1674(15) N(1)-C(10) 1.456(2) Mo(1)-N(4) 2.1384(15) N(2)-C(20) 1.468(2) Mo(1)-N(6) 2.1222(15) N(6)-N(7) 1.209(2) N(3)-C(1) 1.342(2) N(7)-N(8) 1.153(2)

N(1)-Mo(1)-N(2) 111.77(8) N(4)-Mo(1)-N(6) 145.40(6) N(1)-Mo(1)-N(3) 125.94(7) Mo(1)-N(1)-C(10) 174.00(15) N(1)-Mo(1)-N(4) 102.53(7) Mo(1)-N(2)-C(20) 158.54(14) N(2)-Mo(1)-N(3) 121.88(7) Mo(1)-N(3)-C(1) 93.52(11) N(2)-Mo(1)-N(4) 102.29(6) Mo(1)-N(4)-C(1) 94.59(11) N(3)-Mo(1)-N(4) 61.51(6) N(3)-C(1)-N(4) 109.84(15) N(1)-Mo(1)-N(6) 97.01(7) N(3)-C(1)-N(5) 124.15(16) N(2)-Mo(1)-N(6) 96.46(7) N(4)-C(1)-N(5) 126.01(16) N(3)-Mo(1)-N(6) 83.90(6) N(6)-N(7)-N(8) 176.1(2) spectrum displayed different fragmentation pattern to the halide complexes. It shows the [M+] peak at m/z = 450 and the loss of the azide ligand at m/z = 408. The β-elimination of two equivalent of iso-butene, coming from the imido ligands is assigned to the peak at m/z = 340, while complete loss of both imido ligand is observed at m/z = 311. The fragment [Mo(Nt- + Bu)2NH2] yielded by the loss of the guanidinato ligand and the elimination of dinitrogen from the azide ligand is displayed by the peak at m/z = 254. Peaks with m/z ≤ 171 again could be assigned to the fragmentation of the guanidinato ligand or the imido ligand as already discussed for [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}].

Single Crystals of [Mo(NtBu)2(N3){(NiPr)2CMe2] 17 were obtained by crystallization from hexane. The Complex crystallizes in space group P21/c with four molecules per unit cell The molecular structure of [Mo(NtBu)2(N3){(NiPr)2CMe2] is shown in Figure 28. Bonding parameters are given in Table 19. Details of the X-ray diffraction data collection are compiled in Table 24 in the experimental part. The structure is very similar to the structure of the analogue iodo complex

[Mo(NtBu)2I{(NiPr)2CMe2], of course with the exception of the azido ligand. The ligand is bonded in the direction towards the imido ligands, and the torsion angles towards both imido ligands are nearly identical. (see Table 19.) The azido complex was found to be sublimeable and so was chosen for testing in MOCVD experiment. Beside this complex also the chloro complex [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] exhibited volatility and was also applied in deposition experiment. Thermal characterization of [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] and [Mo(NtBu)2(N3){(NiPr)2CMe2] and deposition experiment of Molybdenum nitride is described in chapter 6.

58 4. Synthesis of the Molybdenum Compounds

4.5. Conclusions for Chapter 4

Though the synthesis of the molybdenum analogues of [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}]

(8) and [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] (10) were unsuccessful, we succeeded to obtain the two volatile guanidinato compounds [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] (15) [Mo(Nt-

Bu)2(N3){(Ni-Pr)2CNMe2}] (17). Though none of these compounds sublimes without residue, even in repeated sublimation, the sublimate does not show traces of decomposition, as proofed by NMR. These compounds were interesting candidates for depositing molybdenum nitride, molybdenum carbide, or molybdenum carbonitride phases by MOCVD methods. The thermal characterization of these compounds as well as the MOCVD experiments are described in chapter 6.

59 5. MOCVD-Experiments of Tungsten Nitride

5. MOCVD-Experiments of Tungsten Nitride 5.1. Thermal Characterization of the precursors

The thermal characterization of the compounds [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] (8) and

[W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] (10), that were chosen as tungsten nitride precursor, was carried out by simultaneous thermogravimetry and differential thermal analysis TG/DTA. Both compounds are solids having a low melting point (melting points: 62 °C (8) and 73 °C respectively (10)). The compounds begin to volatilize at about 105 °C (under ambient pressure). For both compounds the onsets of volatilization (~105 °C) fall in the same temperature range, but the temperature ranges of decomposition are slightly different. In the case of [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] (figure 29), the TG curve shows almost a single-step decomposition behaviour, while the TG curve of compound [W(Nt-Bu)2H{(Ni-

Pr)2CNMe2}] (figure 30) clearly reveals a multistep decomposition pattern. It can be seen that the volatilization of compound [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] starts at 105 °C with a rather constant evaporation up to 205 °C. An inflection point corresponding to the onset of a decomposition process can be seen around 205 °C. A second inflection point is

Figure 29: TG/DTA-Diagramm of W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2} (8) 60 5. MOCVD-Experiments of Tungsten Nitride

Figure 30: TG/DTA-Diagram of W(Nt-Bu)2H{(Ni-Pr)2CNMe2} (10) observed around 370 °C, and the weight loss remains almost constant at higher temperatures.

In the case of [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] the TG curve shows a constant weight loss up to 175 °C. Evaluation of the TG plot shows two shoulders curves corresponding to multistep decomposition behaviour of [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] (from 175 to 275 °C and from 275 to 370 °C). In both the cases, there is a sufficient temperature window between sublimation and decomposition, which renders these compounds suitable for MOCVD application. Residual masses of 25% and 22% were observed for compounds [W(Nt-

Bu)2(NMe2){(Ni-Pr)2CNMe2}] and [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}], respectively, in TG analyses, which are lower than the residual masses expected for the most likely W2N phase (38% and 35%).

5.2. Depositions on the home-built MOCVD reactor

To evaluate the suitability of compounds [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] and [W(Nt-

Bu)2H{(Ni-Pr)2CNMe2}] in thin film deposition application, several preliminary control

MOCVD experiments were carried out with and without NH3. All experiments were carried out in the temperature range of 400-800 °C at 1 mbar (if not stated otherwise), using nitrogen

61 5. MOCVD-Experiments of Tungsten Nitride as a carrier gas. A detailed description of the reactor and the conditions are given in the experimental part. All obtained films were shiny with a mirror-like appearance. Films deposited below 500 °C were amorphous without X-ray diffraction features, while those deposited at and above 500 °C were polycrystalline. The obtained films were characterized by XRD, SEM, cross-sectional SEM, SNMS depth-profiling analysis, and resistivity measurements.

5.2.1. XRD analysis and surface morphology

The first set of deposition experiments was carried out at a constant pressure of 1 mbar. Figure 31 shows the effect of the deposition temperature on the thin film crystallinity using

[W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] without ammonia. When the deposition temperature was increased from below 500 to 600-800 °C, the deposited material started to exhibit crystalline domains as indicated in the XRD; two broad peaks around 2θ = 37.73° (111) and

43.84° (200) were observed, corresponding to cubic β-W2N and the (111) and (200) lattice planes. The parallel formation of ternary β- WNxCy or binary β-WC (or related phases) cannot be ruled out from the XRD analysis, as the observed peaks are too broad and the (111) peak positions of the binary β-W2N and β-W2C phases are very close to each other with 2θ values of 37.74° (111=) and 36.98° (111), respectively. The additional peak at 33.03° represents

Si(200) Kα radiation. No other detectable peaks corresponding to metallic tungsten or tungsten oxide phases were observed. When additional NH3 was used as a reactive gas during the depositions, even at 500 °C a sharpening of the X-ray diffraction peaks was observed (Figure 32). Clearly the presence of ammonia is beneficial for the growth of larger crystalline domains (grains). This effect increases with increasing temperature possibly due to the insufficient surface diffusion of NH3 or NHx dissociated species at lower deposition temperatures.

SEM analysis (Figure 33) of films grown from [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] at 800

°C without NH3 indicates a smooth surface with uniform distribution of small grains having similar shapes and dimensions. The cross-sectional SEM image indicates the existence of columnar- type growth of the grains. The SEM micrograph of the films deposited at 800 °C using [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] in the presence of NH3 (Figure 34) shows a different surface morphology with almost uniform grain size and shape, as well as an increase of the average grain size (~350 nm) as anticipated from the XRD data. These larger crystals, i.e., elongated cone-shaped grains, are well conglomerated and uniformly distributed through-

62 5. MOCVD-Experiments of Tungsten Nitride

Figure 31: XRD of the films deposited using [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] at different temperatures and 1 mbar without NH3. The film thickness is in the range of 170 nm at 500 °C to 1200 nm at 800 °C For the SEM image of the film deposited at 800 °C see figure 33. The elemental composition, analysed by SNMS is shown in figure 42. The high S/N-ratio is due to higher magnification compareded to Figure 32

Figure 32: XRD of the films deposited using [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] at different temperatures and 1 mbar in the presence of NH3. The film thickness is in the range of 170 nm at 500 °C to 1200 nm at 800 °C For the SEM image of the film deposited at 800 °C see figure 34. The elemental composition, analysed by SNMS is shown in figure 44.

63 5. MOCVD-Experiments of Tungsten Nitride

Figure 33: SEM-image of the film deposited at 800 °C and 1 mbar using [W(Nt-Bu)2(NMe2){(Ni- Pr)2CNMe2}] in the absence of ammonia. The corresponding XRD is shown in figure 31. Elemental composition, analysed by SNMS is shown in figure 42.

Figure 34: SEM-image of the film deposited at 800 °C and 1 mbar using [W(Nt-Bu)2(NMe2){(Ni- Pr)2CNMe2}] in the presence of ammonia. The corresponding XRD is shown in figure 32. Elemental composition, analysed by SNMS is shown in figure 44. out the substrate surface. Again, cross-sectional SEM substantiates the columnar growth mode (Figure 34).

When [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] was used in MOCVD experiments, addition of NH3 again proved favourable in terms of selective growth of the desired tungsten nitride phase. The crystallinity of the films (Figures 35 & 36) increases drastically when ammonia is added. SEM images (Figures 37 & 38) document growth morphology similar to that discussed in the case of [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}].

Further deposition experiments using [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] were carried out in the temperature range between 400 and 600 °C at different pressures of 1, 4 and 8 mbar. All these experiments were carried out in the presence of ammonia (20 sccm). Figure 39 shows the effect of the deposition pressure on crystallinity of the films. Though all deposition experiments yielded crystalline β-W2N, there are some remarkable differences. When the pressure is increased from 1 to 4 and 8 mbar, there is a steady decrease in the crystallinity

64 5. MOCVD-Experiments of Tungsten Nitride

Figure 35: XRD of the films deposited using [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] at different temperatures and 1 mbar without NH3. The film thickness is in the range of 380 nm at 500 °C to 1200 nm at 800 °C For the SEM image of the film deposited at 700 °C see figure 37. The elemental composition ot the film depostied at 800 °C, analysed by SNMS is shown in figure 44. The high S/N-ratio is due to higher magnification compareded to Figure 36

Figure 36: XRD of the films deposited using [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] at different temperatures and 1 mbar without NH3. The film thickness is in the range of 200 nm at 500 °C to 1900 nm at 800 °C For the SEM image of the film deposited at 700 °C see figure 38. The elemental composition ot the film depostied at 800 °C, analysed by SNMS is shown in figure 45. observed. Comparing the depositions at 1 and 4 mbar the intensity of the (111) peak at 2θ = 65 5. MOCVD-Experiments of Tungsten Nitride

Figure 37: SEM-image of the film deposited at 800 °C and 1 mbar using [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] in the absence of ammonia. The corresponding XRD is shown in figure 35. Elemental composition of the film deposited at 700 °C, analysed by SNMS is shown in figure 43.

Figure 38: SEM-image of the film deposited at 800 °C and 1 mbar using [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] in the absence of ammonia. The corresponding XRD is shown in figure 36. Elemental composition of the film deposited at 700 °C, analysed by SNMS is shown in figure 45. observed. Comparing the depositions at 1 and 4 mbar the intensity of the (111) peak at 2θ = 37.73°, the intensity of the peaks at 4 mbar is decreased to less than the half compared to 1 mbar. The intensity of the peaks at 8 mbar are further decreased about the half. (This effect is best visible at the black curves, representing the depositions at 600 °C.) The SEM images also proofs this effect. Figure 40 shows the depositions at 600 °C at 1 and 8 mbar. The film deposited at 1 mbar exhibited crystallites of significantly bigger size compared to the films deposited at 8 mbar. Columnar growth of the layers, as already described above, was observed independent from the pressure as shown in figure 41. Figure 39 d) shows the effect of the use of hydrogen as a carrier gas, instead of nitrogen on the crystallinity. It shows a drastic increase compared to the experiments of figure 39 b) which was obtain at an identical set of parameters, except the kind of carrier gas.

66 5. MOCVD-Experiments of Tungsten Nitride

Figure 39: XRD of the different CVD experiments using [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] from 400 – 600 °C. a) Depositions at 1 mbar; b) Depositions at 4 mbar; c) Depositions at 8 mbar; d) Depositions at 4 mbar with hydrogen instead of nitrogen as a carrier gas.

Figure 40: SEM images of the films deposited at 600 °Cusing [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] in the presence of ammonia. (left): pressure: 1 mbar; (right) pressure: 8 mbar. For the corresponding XRD see figure 39 above

67 5. MOCVD-Experiments of Tungsten Nitride

Figure 41: Cross sectional SEM-image of a film deposited at 550 °C and 1 mbat using [W(Nt- Bu)2H{(Ni-Pr)2CNMe2}] in the presence of ammonia.

5.2.2 Composition of the films (SNMS analysis)

SNMS depth-profiling was used to determine the film composition throughout the layer down to the interface to the substrate. Without ammonia, both precursors yielded films with high carbon contents (50-60%) and surprisingly low nitrogen levels (<5%), which is consistent with previously reported results of Kim et al.[117] and of Bchir et al.,[120-123, 125] who used [117] [120-123, 125] W(Nt- Bu)2(NMe2)2 and Cl4(RCN)W(NC3H5) (R = Me, Ph) as precursors for tungsten nitride film growth, respectively (see figures 42& 43). However, this finding is in sharp contrast to related work on TaN MOCVD using similar guanidinato precursors, e.g., 2 Ta(NMeEt){η -(Ni-Pr)2C(NMeEt)}(Nt-Bu), which gave very pure, stoichiometric and carbon-free cubic TaN in the absence of ammonia![144, 165] From the SNMS data one can infer that the films are likely to contain the mixed the amorphous tungsten oxynitride phase,

W(N,O)x. However additional phases of β-WNxCy as well as β-W2N and β-WC cannot be ruled out. In addition, the presence of more or less of NH3 was used as a co reactant gas during the depositions with [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}]. In that case a nitrogen level of 23 atom % and a tungsten concentration of 73 atom % were obtained throughout the film thickness, pointing to the β-W2N composition (Figure 44). As expected, addition of NH3 dramatically increases the N concentration and decreases the C content in the film. This is probably due to the enhanced NH3 chemisorption (cracking/activation) at higher temperatures, which may favour the desorption of hydrocarbon

68 5. MOCVD-Experiments of Tungsten Nitride

Figure 42: SNMS-Analysis of a film deposited using [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] at 800 °C and 1 mbar in the abesence of ammonia. The corresponding XRD is shown in figure 31. The SEM images of this film are shown in figure 33.

Figure 43 SNMS-Analysis of a film deposited using [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] at 800 °C and 1 mbar in the abesence of ammonia. The corresponding XRD is shown in figure 35. The SEM images of this film are shown in figure 37.

69 5. MOCVD-Experiments of Tungsten Nitride

Figure 44: SNMS-Analysis of a film deposited using [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] at 800 °C and 1 mbar in the presence of ammonia. The corresponding XRD is shown in figure 32. The SEM images of this film are shown in figure 34.

Figure 45: SNMS-Analysis of a film deposited using [W(Nt-Bu)2(H){(Ni-Pr)2CNMe2}] at 800 °C and 1 mbar in the presence of ammonia. The corresponding XRD is shown in figure 36. The SEM images of this film are shown in figure 38.

be that the ammonia is acting as a catalyst for the cleavage of the ligands, as reported by that 70 5. MOCVD-Experiments of Tungsten Nitride containing fragments from the surface and likely scavenging of the carbon species from the film. Another possibility could be that the ammonia is acting as a catalyst for the cleavage of [59] the ligands, as reported by Becker et al. for [W(Nt-Bu)2(NMe2)2] .

SNMS depth-profiling of the deposition using [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] (Figure 45) reveals that in contrast to [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] the addition of NH3 causes an unexpected drop in the N level throughout the film with a surprisingly high carbon incorporation of 11 atom %. Nevertheless, the SNMS results indicate a substantial increase in

N concentration for these films relative to those grown without NH3, which is in consistent with the results obtained [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}]. The increase in the C concentration may be due to the prereaction of [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] with NH3

(where NH3 could act as a reducing agent at higher temperatures), resulting in precursor decomposition, prior to the thin film growth, which may increase the carbon content in the film. As observed in the TG analysis, the hydrido precursor showed multistep decomposition behaviour unlike in the case of compound [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}]. The composition of the films deposited at different pressures than 1 mar is similar to the ones reported above for 1 mbar and 800 °C (figure 37). Figure 46 shows the composition of the

Figure 46: SNMS-Analysis of a film deposited at 600 °C and 4 mbar using [W(Nt-Bu)2H{(Ni- Pr)2CNMe2}] with NH3. The corresponding XRD is shown in figure 39.

71 5. MOCVD-Experiments of Tungsten Nitride

Figure 47 SNMS-Analysis of a film deposited using [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] with NH3 at 600 °C and 4 mbar, with hydrogen as a carrier gas. The corresponding XRD is shown in figure 39. film deposited at 600 °C and 4 mbar. The content of both nitrogen and tungsten is ~45 at. %. The carbon content could be decreased to 3 at. % compared to ~ 11 at. % at1 mbar is observed, while the oxygen content is increased to 7 at. % (4 at. % at 1 mbar). While the XRD indicates the formation of β-W2N, the tungsten to nitrogen ratio of 1:1 indicates the presence of an amorphous, nitrogen rich tungsten nitride phase.

The composition of the films deposited using [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] at 4 mbar with hydrogen as a carrier gas is displayed in figure 47. The tungsten to nitrogen content is again near to 1:1. The carbon content is 3 at. % and the oxygen content 7 at. %.

When the same variations of the parameters are applied [W(Nt-Bu)2(NMe2){(Ni-

Pr)2CNMe2}], very similar trends in the obtained parameters as observed for [W(Nt-

Bu)2H{(Ni-Pr)2CNMe2}], like composition, resistivity and growth rate (discussed below) were observed. Due to this similarity the whole set of experiments should not be discussed in detail. Because films obtained with precursor [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] contained at least 10 at. % of impurities( 7 at. %, and 3 at.% of carbon), and showed tungsten to nitrogen ratios near 1:1, while the film deposited applying the [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] possessed oxygen contents at the measuring limit of SNMS measurements and contained

72 5. MOCVD-Experiments of Tungsten Nitride

below 2 at.% of carbon, the amido precursor[W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] was chosen for deposition on the semi-industrial reactor Aixtron 200 RF reactor (see below, chapter 5.4).

5.2.3 Growth rates & resistivities

Growth rates of MOCVD experiments were calculated by dividing the film thickness (obtained from cross-sectional SEM analysis) by the deposition time. Both compounds showed similar trends with respect to the deposition rate without NH3, and the growth rate was increased with an increase in the deposition temperature (both cases). In the case of

[W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] the growth rate was increased from 3 nm/min (at 500

°C) to 13 nm/min (at 800 °C), and in the case of [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] the range of the growth rate increased from 6 to 19 nm/min. As expected, with both the compounds, the addition of NH3 increases the growth rate with increasing deposition temperature, i.e., for the amido precursor from 4 to 58 nm/min and for the hydrido precursor from 3 to 32 nm/min.

However, in the case of depositions without NH3, [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] showed higher growth rates compared to [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}], while, in the case of depositions with NH3, [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] had higher growth rates than

[W(Nt-Bu)2H{(Ni-Pr)2CNMe2}]. Such an increase in the growth rate with NH3 depositions with [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] may arise from the precursor decomposition pathways such as β-hydrogen elimination from the dimethylamido ligand of [W(Nt-

Bu)2(NMe2){(Ni-Pr)2CNMe2}]. Resistivity measurements of the films deposited from both compounds showed fairly good values for films deposited with and without NH3. The resistivities were calculated by multiplying the sheet resistivity (from four-point probe measurement) with the film thickness

(from cross-sectional SEM). Films deposited from [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] at

500 °C (amorphous film) with NH3 revealed the lowest resistivity value of 207 μΩ·cm, and

660 μΩ·cm was measured for the film deposited with [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] at 500

°C with NH3. In the case of the depositions with NH3 the film resistivity increases with the deposition temperature, and at 800 °C films grown from the amido precursor and the hydrido precursor showed resistivity values of 1238 and 1337 μΩ·cm, respectively. This is likely due to an increase in the nitrogen concentration in the film, which is known to increase the film resistivity. Similar observations are reported in the literature, where a resistivity decrease is [88, 96] expected with decreasing nitrogen content in WNx films .

73 5. MOCVD-Experiments of Tungsten Nitride

When the deposition pressure was varied a dependency of the growth rate on the pressure was observed. While a growth rate of 8 nm/min was observed at 600 °C and 1 mbar, the growth rate decreases to 5 nm/min at 600 °C and 4 mbar and 1 nm/min at 600 °C and 8 mbar. This effect of decreased growth rate at increased pressure is well known from many precursor systems, having its origin in the increased velocity of the gas flow at lower pressures. A clear pressure dependency in the resistivity could not be observed in this series of experiments. While the set of depositions at 1 mbar exhibited a resistivity of about 3 mΩ·cm, it increases to ~4 mΩ·cm and decreases again to ~2 mΩ·cm when the pressure is increased further to 8 mbar. The growth rates that were obtained when hydrogen was used as a carrier gas were more than doubled compared to the use of nitrogen. While the use of nitrogen results in a growth rate of 5 nm/min at 600 °C, a rate of 12 nm/min is obtained when hydrogen is applied, resulting in a film thickness of up to 1,5 µm. Due to this, the specific resistivity is increased compared to the layers obtain with nitrogen as a carrier gas to values up to 10 mΩ·cm at 600 °C. Adhesion of the films to the substrates was tested by the Scotch tape test. It was observed that films deposited at lower temperatures (T ≤ 600 °C) had low adhesion to the substrate, while films deposited at high temperatures were impossible to remove from the substrate by this test.

5.3. Characterization of the Exhaust gases

To get a hint about the decomposition mechanisms of the two precursors, the condensable fraction of the exhaust gases were analysed. For this purpose an amount of precursor was filled into an evaporator of the same kind as was used for the MOCVD experiments. This evaporator was connected to a quartz glass tube, which was put into a tube furnace, as seen in figure 48. At the other end of the glass tube, a cooling trap, cooled with liquid nitrogen was connected to collect the decomposition products. The whole apparatus was evacuated by an oil pump. To provide a continuous flow of precursor an argon stream was attached to the evaporator. Heating tapes were wiped around the evaporator heating it to 100 °C. The furnace was heated to 500 °C. The decomposition products were dissolved in benzene and were analysed with GC/MS. A metallic film inside the glass tube was observed after the experiments.

74 5. MOCVD-Experiments of Tungsten Nitride

Figure 48: Schematic view of the setup for decompostion experiments

Figure 49: Gas chromatogram of the decomposition experiment of [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}]

Surprisingly the GC-MS of the condensed decomposition products of [W(Nt-

Bu)2(NMe2){(Ni-Pr)2CNMe2}] showed only one fraction (figure 49). The highest peaks of this fraction is found at m/z = 171 which corresponds to a complete, protonated guanidinato ligand H[(Ni-Pr)2CNMe2]. The next peak is found at m/z = 156 and represents the loss of a methyl ligand. The whole residual fragmentation pattern nicely fits to the patterns of guanidinato ligand, as already discussed in Chapter 3 and 4 (For a possible formation pathway of this fragment see figure 51). More volatile species like isobutene or tert-butylamine get lost during the GC analysis due to transfer of the sample.

The GC/MS of the condensed decomposition productsof [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] is more complicated (figure 50). In total, 7 fractions appeared in the GC. The first fraction (Retention time: 6.6 min) exhibited an MS, with and an [M+]-peak at m/z = 144 that could be assigned to a partially fragmented guanidinato ligand [(HNi-Pr)2CNH2], that lost two methyl groups. The second fraction had a much higher retention time of 21.9 min. The highest mass in the

MS was found at m/z = 558, which is interestingly higher than the mass of [W(Nt-Bu)2H{(Ni-

Pr)2CNMe2}] itself, which is 498. This indicates the formation of agglomerates, most likely

75 5. MOCVD-Experiments of Tungsten Nitride

Figure 50 Gas chromatogramm of the decomposition experiment of [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] containing more than one tungsten atom. Peaks that could be assigned to the guanidinato ligand could not be observed in the fragmentation pattern. A peak at m/z = 73 could be assigned to tert-butyl amine. The third (Retention time: 22.1 min) and the fifth fraction (Retention time: 22.6 min) showed nearly identical fragmentation patterns with an [M+]-peak at m/z = 547, making it likely that these two fractions were stereo isomers. In contrast to the second fraction this one exhibits a peak at m/z = 156 corresponding to a guanidinato ligand that lost one methyl group. The fourth fraction appearing at a retention time of 22.3 min showed a [M+]-peak at m/z = 419. Again no peaks assigned to the guanidinato ligand was observed, but a complete fragmentation pattern of tert-butyl amine was observed. The sixth and the seventh fraction appearing at retention times of 23.4 and 23.5 min respectively showed the highest mass peaks at m/z = 555 and m/z = 541. Fragments of the guanidinato ligand could not be found in these fraction, but again tert-butyl amine could be observed at m/z = 73.

Though the exact identity of the fractions 2-7 of the decomposition of [W(Nt-Bu)2H{(Ni-

Pr)2CNMe2}] could not be clarified exactly, these experiments indicate that [W(Nt-

Bu)2(NMe2){(Ni-Pr)2CNMe2}] may have a cleaner decomposition behaviour compared to

[W(Nt-Bu)2H{(Ni-Pr)2CNMe2}].

76 5. MOCVD-Experiments of Tungsten Nitride

Figure 51: Possible decomposition pathways of the tungsten guanidinato precursors

Possible decomposition pathways given in figure 51. Isobutene can be eliminated by β- hydride elimination from the tert-butylamine groups. From dimethylamido group, the imide

CH2=N-CH3 can be formed. Bond cleavage of the C-N bond of the amido groups might lead to the formation of methane. Beside these elimination pathways, several radical mechanisms are possible as decomposition pathways. These fragments having low molecular masses could not be observed in the GC/MS, most likely due to their loss during sample transfers, but are concluded from observations reported from the strucutrally similar imido/amido precursor [114] [W(Nt-Bu)2(NEt2)2] A direct observation of these fragments might be possible by in situ Spectroscopy (MS or IR) during a decomposition experiment. For a detailed study of decomposition of the imido/amido precursor [W(Nt-Bu)2(NEt2)2] that should posses similar decomposition pathways see the report by Wu et al.[114].

5.4. Experiments on the Aixtron 200 RF reactor 5.4.1 The Reactor

The semi-industrial Aixtron 200 RF reactor was originally designed for the deposition of III/V-semiconductors like gallium nitride, using high volatile precursors like

77 5. MOCVD-Experiments of Tungsten Nitride

Figure 52: Photograph of the reaction chamber of the Aixtron reactor. The substrate is clearly visible in the middle of the glowing susceptor.

Figure 53: Schematic View of the Aixtron 200 RF reactor (after Ref [52]) trimethylgallium (see figures 52 & 53) The type of the reactor is a horizontal cold-wall reactor. The susceptor is made of graphite and can be inductively heated to temperature up to 1300 °C. The substrate are placed on a rotating disc inside the susceptor, actuated by a hydrogen gas flow, to enhance the uniformity of the deposited thin films. Several discs are possible to use different sizes shapes of substrates, to a maximum size of round two inch wafers.

78 5. MOCVD-Experiments of Tungsten Nitride

Samples are loaded from an attached glovebox into the open reactor without air contact. The whole reactor is constantly flushed with nitrogen when it is not in use. All gas flows are regulated by computer controlled mass flow controllers and valves. All lines and valves were heated with heating tapes to a constant temperature of 100 °C. Hydrogen (99,999 % purity) was used as a carrier gas, and was precleaned in a palladium cell. To allow the application of precursors being less volatile than the originally planed group III-alkyls, some modification at the bubbler design have been made Further details regarding the experimental conditions and operation of the instrument are given in the experimental section.

5.4.2 The Bubbler Design

The standard bubblers, which are normally chosen for precursor storage in industrial and semi industrial MOCVD reactors are stainless steel containers. These containers have the advantage of being highly mechanically stable, but it is impossible to check the condition of the bubbler or its filling height in the container, unless the bubbler is disconnected from the reactor and cycled into a glove box. For the application of [W(Nt-Bu)2(NMe2){(Ni-

Pr)2CNMe2}] on the Aixtron-reactor a modified, glass made bubbler was used. The bubbler was placed in an oven, and was heated to 80 °C during the experiments, to liquefy the precursor. The glass made bubbler allows an easy control of the consumption of the precursor, as well as possible changes of the precursor, as many precursors decompose show partial decomposition after long exposure to evaporation temperatures, that are indicated by changes in colour and/or the formation of precipitates[52].

5.4.3 The Experiments

The screening studies discussed in section 5.2. using [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] showed very positive results, thus the compound was chosen for deposition experiments employing the AIXTRON reactor. The experiments were carried out in the temperature range of 500-700 °C in steps of 50 °C. All experiments were carried out at 1.5 mbar, which is the lowest pressure the reactor can be set to. A total hydrogen (carrier gas) flow of 700 sccm was applied. Whenever ammonia was used, a gas flow of 300 sccm was used. The deposition time was set to 15 minutes for all MOCVD runs. Resistivity of the obtained films, measured by four-point probe was measured directly after the depositions. The samples were cycled out of

79 5. MOCVD-Experiments of Tungsten Nitride

Figure 54: XRD of the films deposited on the Aixtron using [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] at different temperatures and 1.5 mbar with NH3. The film thickness is in the range of 20 nm at 500 °C to 108 nm at 700 °C For the SEM image of the film deposited at 700 °C see figure 56. The elemental composition ot the film depostied at 700 °C, analysed by SNMS is shown in figure 57.

Figure 55: XRD of the films deposited on the Aixtron using [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] at different temperatures and 1.5 mbar without NH3. The film thickness is in the range of 38 nm at 500 °C to 147 nm at 700 °C For the SEM image of the film deposited at 700 °C see figure 54 The elemental composition ot the film depostied at 700 °C, analysed by SNMS is shown in figure 58.

rather small crystal grains, thus corresponds to a layer thickness of 50-100 nm rather than >1 80 5. MOCVD-Experiments of Tungsten Nitride

Figure 56 Cross sectional SEM images of the films deposited on the Aixtron reactor using [W(Nt- Bu)2(NMe2){(Ni-Pr)2CNMe2}] at 700 °C and 1.5 mbar. (left): with ammonia, film thickness: 108 nm; (right) without ammonia, filmthickness: 147 nm. The corresponding XRDs are shown in figure 54 or 55 respectively. SNMS analysis is shown in figure 57 (with NH3) and 58 (without NH3). the glove box attached to the reactor and then were transferred to the measuring instrument. Figure 54 shows the effect of deposition temperature on the crystallinity The depositions in the absence of ammonia showed, similar to the experiments on the self-build reactor, very broad peaks (figure 55), where obtained phase cannot be clearly determined, due to the vicinity of the peaks of β-W2N and cubic tungsten carbide.. Similar as observed for the depositions on the self-built reactor, the XRD shows two peaks at 2θ = 37.73° and 43.84° were observed, corresponding to cubic β-W2N and its (111) and (200) lattice planes. The intensity of the peaks, expectedly increased, when the deposition temperature is increased, and / or ammonia was present. But in contrast to the deposition on the self-built reactor, the peaks are less intense and much broader. The broadness of the peaks indicates a formation of rather small crystal grains, thus corresponds to a layer thickness of 50-100 nm rather than >1 µm in case of the screening experiments. Peaks pointing to tungsten carbide or tungsten oxide phases could not be observed. The SEM-images (figure 56) show the morphology of the films. A very smooth and homogeneous surface can be seen for the film deposited at 700 °C in the presence of ammonia and unlike the deposition on the self-built reactor a columnar growth could not be observed. The absence of defined crystal boundaries is advantageous in tungsten nitride films made for diffusion barrier properties, because effective diffusion of metal atoms occurs along these boundaries, thus it is hindered in films with a few grain boundaries. Films deposited in the absence of ammonia showed a similar morphology to the films deposited without ammonia. SNMS depth profiling of the film deposited in the presence of ammonia at 700 °C is shown in figure 57. The composition of the films displays a tungsten to nitrogen ratio of 2:1 nicely cor-

81 5. MOCVD-Experiments of Tungsten Nitride

Figure 57: SNMS of the film deposited on the Aixtron reactor using [W(Nt-Bu)2(NMe2){(Ni- Pr)2CNMe2}] at 700 °C and 1.5 mbar in the presence of ammonia. The corresponding XRD is shown in figure 54. An SEM image of the film is shown in figure 56.

Figure 58: SNMS of the film deposited on the Aixtron reactor using [W(Nt-Bu)2(NMe2){(Ni- Pr)2CNMe2}] at 700 °C and 1.5 mbar in the absence of ammonia. The corresponding XRD is shown in figure 55. An SEM image of the film is shown in figure 56 at. %. The nitrogen content increases steadily throughout the film thickness. The oxygen 82 5. MOCVD-Experiments of Tungsten Nitride

responding to β-W2N, as already suggested by XRD analysis. The carbon contamination of the film is very low, the content is constantly below 2 at. %, similar to the depositions on the home-built reactor. The Oxygen content of the film is below 1 at. % throughout the films thickness. The SNMS of the film deposited at 700 °C, but in the absence of ammonia shows a surprising result (figure 58). While the depositions carried out on the home-built reactor exhibited a high carbon content of up to 60 at. % and low nitrogen content of about 10 at. %, the depositions on the Aixtron-reactor carried out in the absence of ammonia, showed a different picture. Similar to the depositions in the presence of ammonia the carbon content was constantly below 2 at. %. The tungsten content of the film is about 45 at. %while the nitrogen content is approx. 42 at. %. The nitrogen content increases steadily throughout the film thickness. The oxygen content on the surface is about 15 at. %, and it decreases constantly to a value of 5 at. % at the bottom of the film. The constant decrease clearly indicates that the oxygen content has its origin in post deposition oxidation, by contact of the films with air. As already indicated by XRD the films deposited in absence of ammonia exhibited a lower crystallinity. Obviously these low crystalline films are much more sensitive to oxidation, than the films deposited in the presence of ammonia. Growth rates of the depositions in the presence of ammonia were observed in a range of 2 nm/min at 500 °C to 7 nm/min at 700 °C. The linear increase of the growth rate, when the temperature is increased indicates that the growth is either surface reaction controlled or diffusion controlled in the chosen temperature window. (see chapter 1). When the films were deposited in the absence of ammonia, the growth rate was higher than the depositions from 3 nm/min at 500 °C to 9 nm/min at 700 °C. Additionally the behaviour seems to lose linearity. The resistivity measurement showed good values, for all films independent if ammonia is applied or not. The film deposited in the presence of ammonia at 500 °C revealed the lowest resistivity value of 94 µΩ·cm for 24 nm thick film, which is one of the lowest resistivities ever reported for CVD or ALD deposited β-W2N. The film deposited the same temperature in the absence of ammonia, showed a resistivity of 150 µΩ·cm for a film being 37 nm thick. When the deposition temperature was increased the resistivities increase to 606 µΩ·cm for a

60 nm thick film deposited in the presence of NH3 at 600 °C and 992 µΩ·cm for a 108 nm thick film at 700 °C. Remeasuring the resistivities after the films were stored in normal air for several weeks did not show a significant change of the values. The adhesion of the films to the substrate was again tested by the scotch tape test. Compared to the depositions on the self-built reactor, the adhesion was strongly improved. All films

83 5. MOCVD-Experiments of Tungsten Nitride passed the test without any problems, independent from the deposition conditions. (To recall: On the self-built reactor, films deposited at temperatures below 600 °C could be removed from the substrate by this test.) We also tried to deposit tungsten nitride onto a three dimensional structure. For this purpose the reactor was equipped with special patterned substrates, The films were grown at 700 °C in the presence of ammonia. While all flat surfaces were nicely covered with tungsten nitride, the room between the structures is filled unregularly. A smooth covering of these planes could not be observed.

5.5. Comparison of the Guanidinato Precursors to Imido/Amido Precursors

Only a few reports are focussed on the deposition properties of the imido/amido precursors of the composition [W(Nt-Bu)2R2] (R=NMe2; NEt2; NHt-Bu). Tsai et al. reported the deposition [111] of tungsten nitride using [W(Nt-Bu)2(NHt-Bu)2] under single source precursor condition. The obtained crystallographic phase was β-tungsten nitride, but they obtained tungsten to nitrogen ratios of ~ 0.9, indicating the film being nitrogen rich. In contrast to the guanidinato precursors, that yielded films with very high carbon contents of upto 60 at. % these films had carbon contents only of ~ 6 at % and an oxygen content of ~ 3 at %. The lowest resistivity was 620 µΩ·cm at a deposition temperature of 650, but it increased up to ~ 8000 µΩ·cm when the deposition temperature decreased, which are higher values than obtained form the guanidinato precursors. Deposition using [W(Nt-Bu)2(NHt-Bu)2] in the presence of ammonia are not reported.

The Precursors [W(Nt-Bu)2(NEt2)2] and [W(Nt-Bu)2(NMe2)2] show a behaviour, being very similar to the guanidinato precursors. Depositions using [W(Nt-Bu)2(NMe2)2] in the absence of ammonia yielded mainly tungsten carbide having only low nitrogen contents of 2-7 at. %[117], while the same precursor yields highly conformal tungsten nitride when ammonia is applied as an additional reactive gas[58, 59]. Resistivities were similar to the films obtained from the depositions using the guanidinato precursors.

Crane et al. postulated a three coordinated transition state of the formula [W(Nt-Bu)2(NR2)], being formed during the decomposition[112]. As shown above in the case of [W(Nt-

Bu)2(NMe2){(Ni-Pr)2CNMe2}], the guanidinato ligand seems to be dissociated cleanly during the deposition. The resulting transition state [W(Nt-Bu)2(NMe2)] would be identical to the transition state being formed from [W(Nt-Bu)2(NMe2)2] by the loss a dimethylamido group. An interesting feature on the role of ammonia during tungsten nitride film growth by ALD [59] was observed by Becker et al. , who deposited tungsten nitride using [W(Nt-Bu)2(NMe2)2] 84 5. MOCVD-Experiments of Tungsten Nitride and 15N marked ammonia. None of the marked ammonia was observed in the deposited films, so they suggested a mechanism with the ammonia acting as a lewis base, catalysing the decomposition of the precursor, In fact, they also obtained tungsten nitride films, when the ammonia was exchanged by pyridine. A detailed report on the role of ammonia in tungsten nitride growth under CVD conditions is not available, but it is very much likely that NHx-species adsorbed to the surface play a vital role, similar as reported for gallium nitride CVD[82].

5.6. Conclusions for Chapter 5

Both of the two new tungsten guanidinato complexes yielded β-tungsten nitride in the preliminary MOCVD experiments on the self-build reactor when applied in the presence of ammonia. High growth rates and fairly good conductivities were obtained. While the amide precursor [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] yielded β-W2N with a tungsten content of approx. 73 at. %, the tungsten to nitrogen ratio of the films deposited using the hydride precursor [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] was much closer to 1:1. The films deposited using

[W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] also exhibited a much higher carbon contamination of up to

11 at, %, than the films that were obtained from [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}], that possessed carbon contents of below 2 at %. When the precursors were applied under single source precursor conditions, both compound yielded highly carbon contaminated films with only low nitrogen contents and a carbon content of up to 60 at. %.

Consistent with the observation of [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] yielding films bearing less carbon and also consistent with the thermal analysis of the precursors, decomposition experiments of both precursors showed that the amide compound decomposes much cleaner than its hydride analogue [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}].

When [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}] was applied under semi-industrial conditions using the Aixtron RF-200 reactor, very homogeneous, high quality films were obtained. Even under single source precursor conditions, pure β-W2N with very low carbon contents was deposited. Resistivities as low as 93 μΩ·cm were measured. While the films deposited in the presence of ammonia were nearly oxygen free, the films deposited under single source precursor conditions exhibited an oxygen content of 15 at. % at the surface but decreasing steadily throughout the film thickness. This decrease clearly indicates post-deposition oxidation as the origin of the oxygen contamination. When we tried to deposit tungsten nitride onto three dimensional structures, we failed to obtain good step coverage under the chosen conditions.

85 5. MOCVD-Experiments of Tungsten Nitride

A detailed comparison of the guanidinato precursors to the standard imido/amido precursors under reproducible experimental conditions, especially to [W(Nt-Bu)2(NMe2)2], which is the most similar compound to the guanidinato compounds, could clarify the amount of similarity in the deposition behaviour

86 6. MOCVD-Experiments of Molybdenum Nitride

6. MOCVD-Experiments of Molybdenum Nitride 6.1. Thermal Characterization of the precursors

As already mentioned in Chapter 4, two volatile compounds [Mo(Nt-Bu)2Cl{(Ni-

Pr)2CNMe2}] (15) and its azide derivative [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}] (17) were obtained. Analogously to the tungsten precursors these two compounds were characterized by simultaneous TG/DTA-analysis. Both compounds are solids, but their melting points are higher than those of the tungsten precursors. The chloride compound [Mo(Nt-Bu)2Cl{(Ni-

Pr)2CNMe2}] melt at 120 °C nicely displayed in the DTA curve as endothermic peak (figure 59). This values was also verified by a standart measurement of the melting point using an oil bath. A loss of the mass is already visible below 100 °C. The peak at 192 °C in the DTA is most likely due to the onset of the decomposition of the precursor. A small shoulder in the TG curve is visible at around 260 °C, indicating a multistep decomposition behaviour. A sufficient temperature window between sublimation and decomposition was observed, which renders this compound suitable for MOCVD application. A residual mass of 43% is observed, which is higher than the residual mass expected for the most likely Mo2N and Mo2C phases, but is consistent with the observation of some residue during the preparative sublimation

Figure 59: TG/DTA-Diagramm of [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] (15)

87 6. MOCVD-Experiments of Molybdenum Nitride

Figure 60: TG/DTA-Diagramm of [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}] (17) experiments for purification. Obviously volatilization and decomposition are not fully decoupled in this case. The azide precursor [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}] shows a lower melting point. The DTA and the standart melting point measurement revealed a melting point of 79 °C (figure 60). The melting of the compound is accompanied by a slight mass loss of approx. 2 %. At ~ 135 °C volatilisation sets in. At about 200 °C, the TG curve exhibits an inflection possibly indicating the onset of the decomposition which is also reflected by the exothermic peak in the DTA curve. Finally at 350 °C a residue of ~ 30 % is shown. This compound also shows a sufficient temperature window between sublimation and decomposition, making the compound suitable for MOCVD purposes.

6.2. Depositions on the home-built MOCVD reactor

To evaluate the suitability of [Mo(NtBu)2Cl{(NiPr)2CMe2}] and

[Mo(NtBu)2(N3){(NiPr)2CMe2}] in thin film deposition applications, several preliminary MOCVD experiments on the self built reactor were carried out in the presence and in the absence of ammonia. All experiments were carried out in the temperature range of 400-800

88 6. MOCVD-Experiments of Molybdenum Nitride

°C. All deposited films were shiny with a mirror like appearance. All films deposited at temperatures below 500 °C amorphous exhibiting no X-ray diffraction features, while those deposited at 500 °C or higher temperatures were polycrystalline. The obtained films were characterized by XRD, SEM, cross-sectional SEM, SNMS depth-profiling analysis, and resistivity measurements. For detailed description of the reactor and the experimental conditions see the experimental section.

6.2.1 XRD analysis and morphology

In the deposition experiments using compound [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] hydrogen was applied as a carrier gas. All depositions were carried out at a hydrogen gas stream of 8 sccm. Whenever ammonia was applied, the reactive gas flow was set to 10 sccm. The evaporator was heated to 110 °C. In Figure 61, the effect of the deposition temperature on the film crystallinity in the absence of ammonia. By rising the deposition temperature from below 500 to 600 °C and higher, the deposited material started to exhibit crystalline domains, as

Figure 61: XRD of the films deposited using [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] at different temperatures and a pressure of 1 mbar without NH3. Film Thickness is in the range of 60 nm at 400 °C to 2000 nm at 800 °C. SEM image of the film deposited at 700 °C is shown in figure 63. Figure 69 shows the SNMS analysis of the film deposited at 800 °C.

89 6. MOCVD-Experiments of Molybdenum Nitride

Figure 62: XRD of the films deposited using [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] at different temperatures and a pressure of 1 mbar in the presence NH3. Film Thickness is in the range of 200 nm at 400 °C to 1300 nm at 800 °C. SEM image of the film deposited at 700 °C is shown in figure 64. Figure 70 shows the SNMS analysis of the film deposited at 800 °C. indicated in the XRD. Two Peaks at around 2θ = 37.24 ° and 43.37 ° were observed, which are either corresponding to the (111) and the (200) lattice plains of cubic γ-Mo2N. A formation of cubic Mo2C cannot be ruled out from the XRD analysis, as the peak are to broad, and peak positions of the (111) as well as the (200) lattice plains of Mo2N and Mo2C are very close to each other. Additionally, due the structural similarity of the β-phase, Mo16N7, has a very similar diffraction pattern to γ-Mo2N. Thus the formation of this phase cannot be ruled out completely by XRD. Peaks corresponding to metallic molybdenum or molybdenum oxide phases were not observed. When additional ammonia was used as a reactive gas during the depositions (figure 62), the X-ray diffraction patterns were looking more or less the same. It seems that the presence of ammonia is not beneficial for the growth of large crystalline domains. SEM analysis (figure 63) of films grown without NH3 indicates a surface with uniform distribution of grains having similar shapes and dimensions. The image indicates the existence of columnar-type growth of the grains. The SEM micrograph of the films deposited using NH3 shows a different surface morphology (Figure 64), which is even more smooth

90 6. MOCVD-Experiments of Molybdenum Nitride

Figure 63: SEM-images of the film deposited using [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] at 700 °C and a pressure of 1 mbar without NH3. (left) top view; (right) cross-sectional image. The corresponding XRD analysis is shown in figure 61. For film composition, analysed by SNMS, see figure 69.

Figure 64: SEM-images of the film deposited using [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] at 700 °C and a pressure of 1 mbar in the presence of NH3. (left) top view; (right) cross-sectional image. The corresponding XRD analysis is shown in figure 62. For film composition, analysed by SNMS, see figure 70. compared to the films deposited without ammonia. Again, cross-sectional SEM substantiates the columnar growth mode.

When [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}] was applied in MOCVD experiments, nitrogen was used as a carrier gas instead of hydrogen. For all depositions using [Mo(Nt-Bu)2(N3){(Ni-

Pr)2CNMe2}] a carrier gas flow of 25 sccm was applied. The evaporator was heated to 130 °C for all depositions using [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}]. Figure 65 shows the effect of deposition temperature on the crystallinity of the films deposited with ammonia. The two peaks at 2θ = 37.24 ° and 43.37 ° were observed similar to the depositions using [Mo(Nt-

Bu)2Cl{(Ni-Pr)2CNMe2}], but for the depositions at and below 700 °C the peak corresponding to the (200) plane has an unusually increased intensity compared to the (111) peak. The (200) peak, which is about 8 times more intense, is most likely due to an orientated growth of the MoxN films towards that plane. It is not clear why this effect disappears at

91 6. MOCVD-Experiments of Molybdenum Nitride higher temperatures. Further studies are necessary to clarify this. The depositions carried out in the absence of ammonia. Show a totally different picture (figure 66). The depositions at temperatures ≤ 700 °C showed the known peaks corresponding to γ-Mo2N, but they are much more broad and less intense than in diffraction patterns of the analogue depositions using

Figure 65: XRD of the films deposited using [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}] at different temperatures and a pressure of 1 mbar in the presence of NH3. Film Thickness is in the range of 300 nm at 400 °C to 1200 nm at 800 °C. SEM image of the film deposited at 700 °C is shown in figure 67. Figure 71 shows the SNMS analysis of the film deposited at 800 °C.

[Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}], indicating a smaller size of the crystal grains. In the diffraction pattern of the deposition at 800 °C, several new peaks appeared. Beside the two peaks of molybdenum nitride, three new peaks at 2θ = 34,2 °, 39,1 ° and at 51,8 ° could be assigned to the hexagonal α-phase of molybdenum carbide. Another crystalline phase could be found by peaks at 2θ = 27,1 ° and a broad peak at 2θ ~ 53 °. These peaks could be assigned to Tungarinovite, the monoclinic oxide phase of Mo(IV), MoO2. An intense peak corresponding to MoO2 could be found below the (111) peak of γ-Mo2N according to literature. In none of the other experiments other crystalline phase than Molybdenum nitride could be observed. While the carbide phase obviously has its origin in incomplete decomposition of the ligands, the origin of the oxide phase is more difficult to find. The oxygen content might be incorporated into the films, due to leakage in the reactor seals, but

92 6. MOCVD-Experiments of Molybdenum Nitride there is another possible origin. As Wu et al. observed, the deposition of tungsten nitride using [W(NtBu)2(NEt2)2] on Si(100) wafers, that still bear the native oxide layer, yields a phase mixture of tungsten nitride, tungsten carbide and tungsten oxide, due to relatively [114] uncontrolled surface reaction, with an oxygen content coming from the native SiO2-layer . It is very much likely that a similar reaction occurs in this case.

Figure 66: XRD of the films deposited using [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}] at different temperatures and a pressure of 1 mbar without NH3. Film Thickness is in the range of 300 nm at 400 °C to 1200 nm at 800 °C. SEM image of the film deposited at 700 °C is shown in figure 68. Figure 72 shows the SNMS analysis of the film deposited at 800 °C.

The SEM analysis of the depositions at 700 °C in the absence of ammonia is shown in figure 67. The picture shows a very rough surface bearing a large size of crystal grains. Compared to depositions using [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] the surface is noticeably rougher. The cross sectional SEM again substantiates the columnar growth, similar as observed for depositions using [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}]. When ammonia was used during the depositions, the surface exhibits expectedly a higher roughness. (figure 68). Large crystal grains are visible.

93 6. MOCVD-Experiments of Molybdenum Nitride

Figure 67: SEM-images of the film deposited using [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}] at 700 °C and a pressure of 1 mbar in the presence of NH3. (left) top view; (right) cross-sectional image. The corresponding XRD analysis is shown in figure 65. For film composition, analysed by SNMS, see figure 71.

Figure 68: SEM-images of the film deposited using [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}] at 700 °C and a pressure of 1 mbar without NH3. (left) top view; (right) cross-sectional image. The corresponding XRD analysis is shown in figure 66. For film composition, analysed by SNMS, see figure 72.

6.2.2 Composition of the films (SNMS analysis)

SNMS depth-profiling was used to determine the film composition throughout the layer down to the interface to the substrate. The composition of the film, deposited at 800 °C and 1 mbar using [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}] without ammonia is shown in figure 67. The precursor yields films with a high carbon content (~33%) and a surprisingly low nitrogen level of below 5%. The molybdenum content of 65% is fitting well to the stoichiometric formula Mo2C. Oxygen contamination was observed being below 3% and it decreases through the layer thickness, so that the oxygen content derives most likely from post deposition oxidation. Interestingly, though the precursor contains a chlorine atom, no chlorine contamination was observed throughout the layer thickness.A possible explanation for this is

94 6. MOCVD-Experiments of Molybdenum Nitride

Figure 69: SNMS of a film deposited using [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] at 800 °C and 1 mbar in the absence of ammonia. For the corresponding XRD analysis see figure 61.

Figure 70: SNMS of a film deposited using [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] at 800 °C and 1 mbar in a thereaction presence of ofthe ammonia. precursor For with the corresthe carrierponding gas, XRD inducing analysis thesee figureloss of 62. chlorine as HCl, as was a 95 6. MOCVD-Experiments of Molybdenum Nitride a reaction of the precursor with the carrier gas, inducing the loss of chlorine as HCl, as was observed by McElwee-White et al. for Chlorine containing tungsten precursors[120-123, 125, 176]. When ammonia was applied the composition of the films changed (figure 70). As expected, the carbon content decreased, while the nitrogen content increased. A nitrogen content of about 28% and a molybdenum level of ~ 64% are corresponding to the stoichiometric formula of Mo2N. Still a carbon contamination of ~9% is present. Also an oxygen level of about 4% is observed throughout the film. Again no chlorine could be detected throughout the film thickness.

SNMS depth-profiling of a film using [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}] deposited at 700 °C in the presence of ammonia is shown in figure 71. The molybdenum to nitrogen ratio of approx. 2:1 is pointing nicely to the elemental composition of γ-molybdenum nitride. A formation of the β-phase of molybdenum nitride can be excluded by this results, due to the higher molybdenum to nitrogen ratio that would be required for this phase. An oxygen content of approx. 7 at. % is observed throughout the film thickness, while the carbon content is about 5 at. %. The deposition in the absence of ammonia shows a different picture. A molybdenum to carbon ratio of about 2:1 with a molybdenum content of 67 at. % and a

Figure 71: SNMS of a film deposited using [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}] at 800 °C and 1 mbar in the presence of ammonia. For the corresponding XRD analysis see figure 65.

96 6. MOCVD-Experiments of Molybdenum Nitride

Figure 72: SNMS of a film deposited using [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}] at 800 °C and 1 mbar in the presence of ammonia. For the corresponding XRD analysis see figure 66. carbon content of 30 at. %is observed in the film. The surface of the films shows a different composition. While the carbon content is only ~ 12 at. % the nitrogen content is 28 at. %, but is constantly decreasing to ~ 3 at. % along the upper 200 nm of the film. The reason for this is not clear. The oxygen content is ~ 4 at. % but decreasing to a level below in the upper region of the film 2 at. % indicating that the oxygen content is coming from post deposition oxidation.

6.2.3 Growth rates & resistivities

Growth rates of MOCVD experiments were calculated by dividing the film thickness (obtained from cross-sectional SEM analysis) by the deposition time. Deposition using

[Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}], in the absence of ammonia showed a constant increase of the growth rate with an increase of the deposition temperature, up to a maximum at 700 °C. The growth rate increased from 1 nm/min at 400 °C to 35 nm/min at 700 °C. Rising the temperature further to 800 °C, the growth rated decreased to 8 nm/min. When ammonia was applied during the depositions, the growth rates decreased slightly. By rising the temperature 97 6. MOCVD-Experiments of Molybdenum Nitride from 400 °C to 800 °C the growth rate increased constantly from 3 nm/min to 21nm/min. A maximum of the growth rate at lower temperatures like observed for the depositions without ammonia was not observed. Obviously the presence of ammonia during the depositions slows down the decomposition reaction at higher temperatures. Resistivity measurements of the films deposited using [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] with and without ammonia showed fairly good values. The film deposited at 700 °C in the presence of ammonia showed revealed the lowest resistivity value of 2483 μΩ·cm. The lowest resistivity value of the films deposited without NH3 was measured for the film deposited at 800 °C being 1216 μΩ·cm.

For the depositions using [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}] in the presence ammonia as an additional reactive gas, a clear dependence in the growth rates could not be observed. At 500 °C a growth rate of 20 nm/min was observed. The rate decreased to 13 nm/min at 600 °C to increase again 20 nm/min. When the depositions were carried out in the absence of ammonia, a linear increase of the growth rate could be observed. At 400 °C a rate of 2 was observed, to increase up to 21 nm/min at 800 °C.

The resistivity measurements of the films deposited using [Mo(Nt-Bu)2(N3){(Ni-

Pr)2CNMe2}] showed inferior values compared to the deposition using [Mo(Nt-Bu)2Cl{(Ni-

Pr)2CNMe2}] as a precursor. The lowest resistivity of 2796 μΩ·cm was obtained without ammonia at 500 °C. All other resistivities were observed being at least 1000 μΩ·cm higher. Adhesion of the films to the substrates was tested by the Scotch tape test. All films deposited using [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] have high adhesion to the substrates. Films deposited without ammonia were impossible to remove from the surface by this method, while films deposited with NH3 at lower temperatures (≤ 600 °C) could be partially removed by the tape.

For the depositions using [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}], the test revealed similar results. All films deposited below 600 °C could be removed by this test, while the films deposited at higher temperatures were impossible to remove from the substrate.

6.3. Conclusions for Chapter 6

Two new molybdenum guanidinato complexes, the chloro compound

[Mo(NtBu)2Cl{(NiPr)2CMe2}] and its azido analogue [Mo(NtBu)2(N3){(NiPr)2CMe2}] have been tested in preliminary MOCVD experiments. Both compounds yield molybdenum carbide with only low nitrogen contents, when used under single source precursor conditions. In both cases a change of the composition to a more nitrogen rich phase. The reason for this behaviour is unclear, but because it is not observed in a single case, it seems to be a

98 6. MOCVD-Experiments of Molybdenum Nitride systematic behaviour. It could not be clarified, if this behaviour has its origin in the precursor structure or if it is a reactor specific behaviour. The difficulty of obtaining reproducible results on the self-built reactor, also makes it nearly impossible to give a clear statement about this problem. When the precursors are applied in the presence of ammonia, the composition of the deposited films changed. The nitrogen content increases to a content of approximately 30 at. %, with γ-

Mo2N being the dominant phase. In both cases the films are contaminated with a carbon content of about 9 at.% and an oxygen content of 4 at. %. Both precursors exhibited a multistep decomposition, leaving a high residue during thermal analysis, and in fact both compounds leave a brown-black residue in the evaporator after the deposition experiments. Thus an eventual use of residual precursor for another experiment was impossible and the evaporator has to be refilled with fresh precursor after each experiments. These observations make a scale up of these experiments undesirable, and so MOCVD experiments on the Aixtron reactor were not taken into consideration. Additionally the use of [Mo(NtBu)2(N3){(NiPr)2CMe2}] is not desirable on a large scale reactor, because the necessary presence of high amounts of an azide compound that is connected with the possible formation of hydrazoic acid, bears possible risks. So both compound are interesting mainly for small scale applications, e.g. catalyst preparations.

99 7. Summary and Outlook

7. Summary and Outlook 7.1. Summary

During the present study, new compounds of tungsten and molybdenum were developed and some of these complexes were applied for metal organic chemical vapour deposition experiments of tungsten and molybdenum nitride. These films find their application in the fabrication of microelectronic devices, namely as metal gate and diffusion barrier materials.

Both material β-W2N and γ-Mo2N feature high thermal stability, good conductivity and barrier properties. The challenge is to deposit carbon free, homogeneous and dense films that withstand post oxidation processes. The experimental work and its results of this dissertation can be divided in two parts. The first part deals with the realisation of a new concept for MOCVD precursors containing - the guanidinato ligand {(R’N)2CNR2} . The concept includes an increases of complexity of the metal organic complexes compared to literature reported, more simple compounds (mixed amido/imido complexes of the type M(Nt-Bu)2(NR)2 (M=W, Mo; R=Me, Et)), having low monodentate ligands and low coordination numbers. The influence on the chemical as well as thermal behaviour has been investigated. The second part of this work deals with the MOCVD experiments applying the newly synthesized compounds as precursors.

7.1.1. Synthesis and characterization of the new compounds

The synthetic routes to the new guanidinato complexes are based on ligand stabilized, dichloro-, di-tert-Butylimido complexes of molybdenum and tungsten. WCl2(Nt-Bu)2py2 was chosen as the tungsten starting compound, while MoCl2(Nt-Bu)2(dme) was chosen for the synthesis of the molybdenum complexes. By reaction with lithiated 1,1,3,3- Tetramethylguanidinate, or with [Li{(R’N)2CNR2}]x generated in situ by addition of a carbodiimide to a suspension of Lithiumdialkylamide, new guanidinato ligand containing tungsten and molybdenum complexes could be obtained. Further derivatives could be obtained by exchanging the residual chloro ligand by dimethylamido, hydrido, azido (see figure xx for tungsten), or iodo ligands (see figure xx for molybdenum) The compounds were analyzed by means of 1H-NMR, 13C-NMR, EI-Mass spectroscopy, FT-IR spectroscopy and CHN analysis. Whenever suitable single crystal could be obtained , single crystal X-ray

100 7. Summary and Outlook

Figure 73: Reaction Scheme for all synthesized tungsten compounds

diffractometry was also applied. The compounds that were chosen for MOCVD experiments were also characterized by TG/DTA analysis.

The reaction of WCl2(Nt-Bu)2py2 with {Li[NC(NMe2)]}x, as well as with sodium azide resulted in a dimeric complexes bearing a bridging µ2-guanidinato ligand, that were non- volatile. Thus they were not suitable for MOCVD purposes, and were not studied further. 2 - Among the numerous tungsten complexes, containing the η bonded{(R’N)2CNR2} ligand that could be obtained, the amido complex W(Nt-Bu)2(NMe2){(i-PrN)2CNMe2} and the hydrido complex W(Nt-Bu)2(H){(i-PrN)2CNMe2} exhibited the most promising thermal properties for CVD purposes and were chosen for application in deposition experiment. The crystal structure of W(Nt-Bu)2(H){(i-PrN)2CNMe2} reveals a distorted trigonal bipyramidal

101 7. Summary and Outlook coordination sphere with the imido ligands in equatorial positions and the hydride in an axial position. The structure of W(Nt-Bu)2(NMe2){(i-PrN)2CNMe2} could not be clarified by single crystal X-ray diffraction, but due different NMR features compared to the hydride complex, it is likely that the compound possessed a different structure with a square-pyramidal structure.

Figure 74: Reaction Scheme for all synthesized molybdenum compounds

The transfer of these results to molybdenum was not successful, neither the analogue hydride, nor amide could be synthesized out of Mo(Nt-Bu)2(Cl){(i-PrN)2CNMe2}, but it was found that Mo(Nt-Bu)2(Cl){(i-PrN)2CNMe2} itself was volatile and so was chosen for deposition experiments. By reaction with an excess of sodium azide Mo(Nt-Bu)2(N3){(i-PrN)2CNMe2} could be obtained that was also chosen to be a suitable precursor for deposition experiments.

7.1.2. MOCVD-Experiments

MOCVD experiments were performed using different reactors. Preliminary experiments were conducted with a self-built reactor for small substrates. W(Nt-Bu)2(NMe2){(i-PrN)2CNMe2} that showed the most promising preliminary results was tested with the industrial Aixtron 200 RF MOCVD reactor. The results of the tungsten nitride depositions using the self-built reactor showed films that were heavily contaminated with carbon and had only low nitrogen content, when deposited without additional ammonia. When ammonia was applied as an additional reactive gas the hydride precursor yielded tungsten nitride films that were still contaminated with a carbon level of ~ 10 at. %. These films also possessed an oxygen of ~4 at. %. In contrast to this, 102 7. Summary and Outlook

experiment in the presence of ammonia using W(Nt-Bu)2(NMe2){(i-PrN)2CNMe2} gave oxygen free, pure β-tungsten nitride films with carbon contents below 2 at. %.

When Mo(Nt-Bu)2(Cl){(i-PrN)2CNMe2}, was applied in CVD experiments, the films obtained in the absence of ammonia were contaminated with carbon, similar as observed for the tungsten nitride depositions, but the films were much more crystalline compared to the films deposited using the tungsten precursors. The presence of ammonia results in the formation of γ-Mo2N, but the films still contained ~9 at. % of carbon and 4 at. % of oxygen. Interestingly, though the precursor contains a chlorine atom, bonded to the molybdenum centre, the films did not contain detectable chlorine. This is most likely due to the removal of the chlorine as HCl by reaction with the carrier gas (hydrogen). The obtained surfaces exhibited a very high roughness compared to all other films, as indicated by SEM-images, which is an indicator for the formation of a corrosive byproduct. Table 20.: Compounds that were used for MOCVD including a short summary of the results. Resistivity Formula NH Reactor type Chapter Qualitative results 3 / µΩ·cm 207-1238 W(Nt-Bu)2(NMe2){(i-PrN)2CNMe2} yes Selfbuilt 5.2. β-W2N with low le- MOCVD- vels of impurities reactor

W(Nt-Bu)2(NMe2){(i-PrN)2CNMe2} no Selfbuilt 5.2. Low content of N; 380-550 MOCVD- but heavily carbon reactor contaminated films 660-1337 W(Nt-Bu)2(H){(i-PrN)2CNMe2} yes Selfbuilt 5.2. Carbon contamina- MOCVD- ted (11 at. %) β- reactor W2N

W(Nt-Bu)2(H){(i-PrN)2CNMe2} no Selfbuilt 5.2. Low content of N; 250-720 MOCVD- but heavily carbon reactor contaminated films 94-990 W(Nt-Bu)2(NMe2){(i-PrN)2CNMe2} yes Aixtron 5.4.3. Very homogeneous 200 RF β-W2N with low levels of impurities 150-570 W(Nt-Bu)2(NMe2){(i-PrN)2CNMe2} no Aixtron 5.4.3. β-W2N with low car- 200 RF bon content but sensitive to post deposition oxidation 2000- Mo(Nt-Bu)2(Cl){(i-PrN)2CNMe2} yes Selfbuilt 6.2. Carbon contamina- 4500 MOCVD- ted (9 at. %) γ-Mo2N reactor

Mo(Nt-Bu)2(Cl){(i-PrN)2CNMe2} no Selfbuilt 6.2. Molybdenum 1200- MOCVD- carbide. 5000 reactor 4500- Mo(Nt-Bu)2(N3){(i-PrN)2CNMe2} yes Selfbuilt 6.2. Carbon contamina- 10000 MOCVD- ted (5 at. %) γ-Mo2N reactor

Mo(Nt-Bu)2(N3){(i-PrN)2CNMe2} no Selfbuilt 6.2. Cystalline mixture of 2800- 8800 MOCVD- γ-Mo2N, α-MoC2, reactor monoclinic MoO2

103 7. Summary and Outlook

The use of Mo(Nt-Bu)2(N3){(i-PrN)2CNMe2} in the absence of ammonia resulted in a mixture of cubic γ-Mo2N, monoclinic MoO2 and α-MoC2. At higher deposition temperatures all these different phases could be observed by XRD, which is in sharp contrast to the depositions using the other precursors, where appearing oxide and carbide phases were amorphous. In the presence of ammonia this precursor also yields γ-Mo2N films contaminated oxygen (~4 at. %) and carbon (~ 5 at. %). Table 20. gives an overview over all featured precursors.

7.2. Outlook

The synthesis of the tungsten and molybdenum guanidinato complexes was elucidated to a satisfying extent, and the synthesis can be easily scaled up to a multigramm scale. But the mechanistic pathways of the decomposition of the precursor could only be clarified to a small extent. The different mechanisms of the two tungsten precursors could not be resolved fully. Further experiments are required, for a proper identification of the decomposition pathways. A suitable experiment for this purpose would be a direct coupling of a MOCVD reactor to an analytical method such as EI-mass spectroscopy or FT-IR for direct identification of the decomposition products.

Especially the tungsten precursor W(Nt-Bu)2(NMe2){(i-PrN)2CNMe2}, which was identified to be the most promising of all studied compounds is worth further studies. By variation of the parameters of the MOCVD experiment it might be possible to avoid the particle formation that was observed in depostion experiments using three dimensional substrates. Also the application of this compound in ALD experiments should be taken into consideration. An indication for the quality of the could be given by the deposition of copper films onto the obtained tungsten nitride films. While very promising results were obtained in tungsten nitride depositions, a scale up of the molybdenum nitride depositions was not promising, due the properties of the precursor. A possible application for these compound might be the production of nanoparticles or several sol-gel applications.

104 8. Experimental Section

8. Experimental Section 8.1. Spectroscopic Characterization of the synthesized Compounds

1H- and 13C-NMR. NMR spectra were measured on a Bruker Avance DPX 250 spectrometer.

All spectra were recorded at 25 °C, except measurements of {W(Nt-Bu)2Cl{NC(NMe2)2}}2 (6), which was measured at different temperatures from -50 to 25 °C. All spectra were integrated and analyzed using the “Mestrec®”software suite (Version 4.7 or higher). Deuterated benzene (99.5% purity) and deuterated toluene (99.8% purity) were purchased from Deutero GMBH and were degassed and dried over activated molecular sieves (4 Å)

EI-MS. Electronic Ionization (EI) mass spectra at ionization energies of 70 eV and 24 eV were recorded by Jutta Schäfer using a “CHS-Mass spectrometer” by “Varian MAT” (Bremen). Data were given as specific masses (m/z) based on the most abundant isotopes

1 12 14 98 184 (1 6 7 42 ;Mo;N;C;H 74W).

IR-Spectroscopy. All Infrared spectra were measured on a “Perkin-Elmer 1720 X” Fourier Transform Spectrometer.

Elemental Analysis. CHN-analyses of all compounds were performed by Karin Bartholomäus of the analytical service center of the faculty using a “CHNSO Vario EL” instrument (Elemental, Hanau).

TG/DTA. Thermal Properties of the precursors were analysed by using “Seiko - TG/DTA6200/SII” instrument, performing simultaneous thermogravimetric and differential thermal analysis. The Samples were filled in an aluminium crucible with a punched lid inside the glovebox. An empty aluminium crucible was used as a reference during the measurement. In all measurements a heating rate of 5 K/min and a nitrogen (99.9999 % purity) flow of 300 sccm was applied. The measurements were started at 25 °C and were heated to a target temperature between 500 and 600 °C.

Single Crystal X-Ray Diffraction. The single-crystal X-ray diffraction measurements of the compounds 1, 4, 5, 6, 10, 11, 12, 13, 15, 16 were performed on an Oxford Excalibur-2 diffractometer. For the measurement of 3, a Bruker SMART CCD 1000 diffractometer was

105 8. Experimental Section

used. Both diffractometers were run with monochromated Mo-Kα radiation (0.71073 Å). All measurements were performed by Manuela Winter. An ω-scan mode was applied for data collection, which was performed using the program package ChrysAlisPro, including ChrysAlis CCD for data collection and ChrysAlis RED for data reduction. Absorption correction was carried out using SADABS. XPREP was used for determination of the space groups, followed by solving the Structure with SHELXS. Structural refinement, introduction of anisotropic displacement parameters and determination of the positions of the protons were carried out applying the program SHELXL. The hydrogen atoms are fixed at geometric 2 2 2 position .The refinement was performed by minimizing the function Σw(FO -FC ) with the full-matrix least-squares methods against F2. The R-values describe the agreement between the crystallographic model and the diffraction data. They are defined as:

0.5 R1 = Σ(||FO|-|Fc||)/( Σ |FO|) and

2 2 2 2 2 0.5 wR2 = (Σw(FO -Fc ) / Σw(FO ) )

The Samples for single-crystal X-ray diffraction measurements were prepared as followed: After crystal of a suitable size were grown from a saturated solution of the product, some of the crystals were transferred into a perfluorated oil that avoids contact of the crystals to moisture and air. Under an optical microscope, suitable crystals were chosen for the measurements. Crystals that were to large, were cut to size with a razor blade. The chosen sample was mounted to the goniometer, covered with the perfluorated oil, where it was frozen in a constant stream of cold nitrogen (T ~ -170 °C).

8.2. Thin Film Analysis

X-ray diffraction (XRD) of thin (polycrystalline) films. XRD was applied to gather information about the crystalline structure of the films. Cu-Kα radiation emitted by an X-ray source interacts with the thin film deposited on the substrate. The incident radiation is reflected at the crystalline lattice of the thin film and the underlying Silicon (100). Constructive (intensifying) and destructive (attenuating) interference occurs between different reflected beams, depending on the crystal lattice (distance d) and the substrate / incident

106 8. Experimental Section radiation angle. Reflexes can be observed as a function of the angle 2θ. The positive interference leading to the observed reflexes is described by Bragg’s Law. sin 2d sin θ = n λ λ = wavelength An ideal poly-crystalline film with grain sizes between 500·nm and 5,000 nm would give reflexes with very small half-widths and high intensities. Deviations from this case occur because of several influences on the crystallinity of the thin film: 1. Impurities incorporated in the film (e.g.:carbon, oxygen) 2. Non-stoichiometric composition of the film. 3. Crystalline defects resulted by mixtures of crystalline phases 4. Lattice defects in the films 5. Grain boundaries in between poly-crystalline phases[177] Beside these reasons the line profile can be influenced by microstrain or a special mosaic structure, if several crystallites are tilted out of the ideal plane. These effects can be measured by an ω-scan.[178] X-ray diffraction analysis was performed by Dr. Harish Parala on a Bruker D8 Advance instrument with θ-2θ (Bragg-Brentano) geometry (figure 8.1). The X-ray source (Cu-Kα – radiation; λ = 1.5418 Å, acceleration voltage: 40 kV, heating current 30-40 mA) and the position sensitive detector are constrained to lie on a circle. A parabolic Göbel mirror is mounted in the primary beam path (slit width: 0.2 mm) to separate parallel, pure Cu-Kα radiation. Calibration in the flat sample mode was performed using standard quartz sample. The 2θ-angle was varied for the measurements in a range of 20 to 60° with a step size of 0.0141 °/step (0.3 s per step for standard measurements and 4 s for long time measurements).

Figure 75: Powder XRD-Detector in Bragg-Brentano-Geometry

107 8. Experimental Section

Secondary neutral mass spectroscopy (SNMS). SNMS depth profiling was used to determine the elemental composition of the deposited films. For SNMS measurements, the sample is bombarded with rare gas ions (in this case argon) at an energy range of 0.5 to 10 keV. Atoms are sputtered from the surface of the sample by this beam of argon ions. The flux of sputtered particles consists of ions and neutral atoms. The neutral atoms are ionized by electron bombardment when they enter the analyzer. SNMS gives more reliable results than SIMS (Secondary ion mass spectroscopy). Due to matrix effects, the probability of sputtering ionized particles from the sample heavily depends on the type of surface (0 - 10 % of all sputtered particles). The percentage (yield) of neutral particles is therefore in the range of 90 - 99.999 %, which drastically reduces the influence of the sample composition and constitution on the results, compared to SIMS. SNMS measurements show an accuracy of about ±1 atomic % with a detection limit in the range of 0.1 - 1 atomic %, depending on the element. The calibration is straight forward when the SNMS ion yields are available, as they are for most common species.[179] The samples were profiled using Sputtered neutral mass spectrometry (SNMS) on a VG SIMSLABB IIIA instrument by Dr. Simon Romani (CSMA-MATS Company, Stokes-on- Trent, England). The primary ion beam was argon at 10 keV, usually operated at high currents (0.8 - 1.0 microampere) over large areas (from 0.5 to 4 mm raster size generally, depending on total depth requirements).

Scanning electron microscopy (SEM). SEM is a powerful analytical instrument for analyzing thin films deposited by CVD. It is possible to acquire surface scans with resolutions of only a few nanometers. Scanning electron microscopy uses electrons that are emitted by a heated tungsten filament. They are focused as a beam and accelerated towards the sample. The incident beam of electrons interacts with the sample and results in an emitting of different types of signals. There is an emission of X-rays beams, Auger electrons, catholuminescense, backscattered electrons and so called secondary electrons. The backscattered electrons can be collected and give topographical information about the sample. The morphology of the surface (nanocrystalline, amorphous etc.) and crystal sizes can be determined. Cross sections of samples are useful to provide information about the behaviour of the vertical growth and the growth rate during the deposition. Samples with low conductivities (e.g. insulating and semiconducting materials) have to be sputtered with gold or carbon in order to guarantee certain conductivity. Otherwise, absorbed electrons cannot be removed and the sample will get charged (lowering of the resolution).[180]

108 8. Experimental Section

SEM - samples were measured by Dr. R. Neuser and M. Born on a 1530 Gemini Field Emission SEM instrument by Leo (Zeiss), at the “Zentrales Rasterelektronenmikroskop der Ruhr-Universität Bochum. All samples were sputtered with carbon. The thickness measurement of the films were carried out by cross section SEM. The accuracy of the measurements was 5 %.

Four-point probe measurement. All resistivity measurements were carried out using a standard four-point probe unit (Jandel). The accuracy of the measurements was 2 Ω/⁯. The specific resistivity was obtained by multiplying the resistivity obtained from four-point probe measurement with the film thickness obtained form cross section SEM.

8.3. MOCVD-Experiments. 8.3.1 Wafer Treatment.

All Depositions were carried out on Si(100) Wafers. For Depositions on the Self-Built Reactor 150 mm Wafers were cut to pieces of approximately 1.5 x 2.0 cm size. The Aixtron 200 FE reactor was equipped with round two inch wafers. Prior to the depositions the substrates were cleaned by washing with acetone and iso-propanol.

8.3.2. Handling of the self-built reactor.

Figure 76: Scheme of the home build MOCVD reactor

109 8. Experimental Section

All thin film depositions were carried out in a home-built horizontal cold wall, glass reactor[152] (Figure 74). The thin film depositions were carried out on Si(100) substrates of approximately 2 x 1,5 cm of size, using N2 (99.9999%) as a carrier gas or H2 (99.999%) were used as carrier gas.. NH3 (99.995%) was used as a reactive gas wherever indicated. Prior to the MOCVD experiments, the substrates were cleaned by treating them with ultrasound and subsequently washed and cleaned with 2-propanol and acetone and the native oxide layer was not removed The substrates were placed on a Si3N4 susceptor which was resistively heated to the desired temperature. The area of the reactor around the substrate holder is constantly cooled by a water filled thermostat, the temperature of the water was regulated to 80 °C. However, the inner wall of the reactor has probably a higher temperature due to radiation of the heated susceptor. A U-shaped glass tube with Teflon stoppers on both sides, a connection for the carrier gas on one side, and a connection to the reactor on the other side was used as an evaporator. Each evaporator was filled with an amount of precursor needed for one deposition (~100 mg) and was refilled after each experiment. To exclude cold spots, where the precursor could resublime, the bubbler and the delivery lines were heated by heating tapes set at 100 °C for the delivery lines. The evaporator was thoroughly dried and filled in the glove box under inert gas atmosphere in order to prevent an oxidation of the product. The amount of filling was aprox. 100 mg of precursor for each deposition. After the experiments the evaporator was cleaned and replaced by a freshly filled one. An evaporator temperature of 100 °C was used for [W(Nt-Bu)2(NMe2){(Ni-Pr)2CNMe2}]. [W(Nt-Bu)2H{(Ni-Pr)2CNMe2}] was heated to 90

°C. [Mo(Nt-Bu)2Cl{(Ni-Pr)2CNMe2}] was heated to 110 °C, while a temperature of 130 °C was applied for [Mo(Nt-Bu)2(N3){(Ni-Pr)2CNMe2}]. After loading the substrate into the reactor, the setup was evacuated by a turbo-molecular pump to a base-pressure of ~ 10-5 mbar. Afterwards the carrier gas flow was set to the desired flow rate while the reactor pressure is regulated by a throttle valve. After all the substrates and the heating tapes had reached their target temperatures and are constant for at least 15 minutes, the deposition was started by opening the teflon valves of the precursor reservoir. To maintain a continuous flow of the carrier gas and reactive gas (whenever used), mass flow controllers are applied. To achieve a more uniform mass flow distribution over the substrate, the substrate holder is twisted about 15 °. Unreacted precursor and decomposition products are collected in a cooling trap, that is cooled with liquid nitrogen. All depositions were carried out at 1, 4 or 8 mbar of pressure and a carrier gas (N2) flow rate of 20 sccm or a hydrogen flow rate of 8 ppm. When NH3 was used as a co-reactant gas, a flow rate of 25 sccm was used for tungsten nitride depositions.

110 8. Experimental Section

Molybdenum nitride depositions were carried out at an ammonia flow rate of 10 sccm. The deposition time was set to 60 min for all MOCVD experiments. The advantage of this reactor is the simplicity of the setup which enables up to three experiments per day, even at a long deposition time without causing the high costs associated with the use of semi-industrial equipment.,. Nevertheless, as a result of this rather simple reactor setup, a drawback is that the results are difficult to reproduce. Thereby, the most crucial point is the quantitative control over the precursor transport into the reactor. The teflon stoppers of the bubbler, as well as the some valves on the reactor were opened by hand, Small variations in the openings of the evaporator might have a high impact on the fluid dynamics of the carrier gas and thus of the mass transport of the precursor into the reactor, making the drawing of exact conclusions on the growth rate under a certain condition difficult. Additionally the temperature of the susceptor could only be regulated by adjusting the power of the resistive heater, and it found that the final temperature was heavily dependent on the gas flows. Thus the power had to be adjusted to maintain the temperature, if ammonia is applied or not.

8.4. General Synthesis Procedures and Starting Compounds.

All Manipulations were carried out using standard Schlenk and glovebox techniques. Hexane and toluene were dried over Al2O3 columns under Ar (99.99) using an automatic solvent purification system (M. Braun) directly connected to a glovebox (H2O below 1 ppm, Karl

Fischer). Pyridine was dried over Na/benzophenone, and TMEDA was dried over CaH2. Both solvents were distilled prior to use. CH2Cl2 was purchased from J.T.Baker N,N′- Diisopropylcarbodiimide (98% purity), N,N′-Dicyclohexylcarbodiimide (98% purity) and lithium triethylboronhydride (1 mol/L in THF) were purchased from Acros, and n- butyllithium (15 wt % solution in n-hexane), diethylamine (99% purity) and diisopropylamine

(99% purity) were purchased from Merck. WCl6 (99.99% purity) was supplied by H. C. Starck. N-tert-Butyltrimethylsilylamine (98% purity), and lithium dimethylamide (95% purity) were purchased from Aldrich. 1,1,3,3-Tetramethylguanidine and sodium azide were obtained from Fluka. NaI was purchased from Riedel-de-Haën and was dried by heating in [167-169] vacuo for 24h. Mo(NtBu)2Cl2(dme) was synthesized by a literature procedure . All starting compounds and reagents were used as received if not stated otherwise.

111 8. Experimental Section

8.5. Synthesis of the Tungsten-Compounds

Synthesis of [WCl2(Nt-Bu)2py2] (1). A sample of WCl6 (10 g, 25.6 mmol) was dissolved in

100 mL of toluene. A sample of HN(t-Bu)(SiMe3) (19.2 g, 0.1 mol) was added slowly, and the reaction mixture was allowed to stir overnight at room temperature. The resulting brown suspension was filtered through Celite, and the solvent and all volatile byproducts were removed by vacuum distillation at 25 °C. The residue was washed with 40 mL of n-hexane. The obtained yellow solid was allowed to settle, and the dark-brown washing solution was removed by filtration. The yellow intermediate was suspended in 50 mL of diethyl ether. After the addition of pyridine (20 mL), the reaction mixture immediately turned dark-green. After 1 h of stirring, the solvent was removed in vacuum to yield 7.55 g (54.2%) of a green microcrystalline powder. Pure [W(NtBu)2Cl2py2] was obtained by recrystallization from toluene at room temperature after one week. 1 H NMR (250 MHz, C6D6, 298 K): δ 1.53 (s 18 H, Nt-Bu); 6.51 (t, 4H, py); 6.84 (t, 2H, py), 9.09 (d, 4H, py). 13 C NMR (250 MHz, C6D6, 298 K): δ 31.5 (NC(CH3)3); 67.9 (NC(CH3)3); 123.3 (py); 136.9 (py); 152.1 (py).

Anal. calcd for WCl2N4C18H28: C, 38.94; H, 5.08; N, 10.09. Found: C, 39.14; H, 5.06; N, 9.56.

Synthesis of [i-Pr2NC(Ni-Pr)2Li(TMEDA)] (2). A sample of 31.3 mL of butyllithium (15 wt %, n-hexane solution) was transferred into a Schlenk tube and cooled to -78 °C. Diisopropylamine (7 mL, 50 mmol) was added to the solution, which was then allowed to warm to room temperature. One equivalent of N,N′- diisopropylcarbodiimide (7.7 mL 50 mmol) was added, and the solution was stirred for 1 h at 25 °C. Subsequently, N,N,N′,N′- tetramethylethylendiamine (7.4 mL) was added to the yellow solution. After 1 h of stirring, the solvent was removed, and the product 2 was obtained as a microcrystalline pale-yellow powder (17.02 g, 97.5%). 1 H NMR (250 MHz, C6D6, 298 K): δ 1.23 (d, 12H, CN(CHMe2)2); 1.38 (d, 12 H,

C(NCHMe2)2), 1.88 (s, 4H, CH2NMe2), 2.02 (s, 12 H CH2NMe2), 3.59 (sept, 2H,

CN(CHMe2)2); 3.90 (sept, 2 H, C(NCHMe2)2). 13 C NMR (250 MHz, C6D6, 298 K): δ 24.1(Me2CH); 28.3 (Me2CH); 45.9 (CH2NMe2); 46.1

(Me2CH); 48.1 (Me2CH); 56.7 (CH2NMe2); 164.6 (CN3).

112 8. Experimental Section

EI-MS [m/z (%)]: 227 (65) [(Ni-Pr)2CNi-Pr2) + H]; 212 (20) [(Ni-Pr)2CNi-Pr2) + H - Me];

184 (52) [(Ni-Pr)2CNi-Pr) + H -i-Pr]; 127 (12) [(Ni-Pr)2C + H]; 116 (6) [TMEDA]; 113 (4)

[i-Pr2NC + H]; 100 (19) [i-Pr2N]; 85 (30) [i-Pr2N - Me]; 69 (16) [i-PrNC]; 58 (100) [i-PrN + H]; 43 (57) [i-Pr]; 27 (10) [HCN].

Anal. calcd for LiN5C19H40: C, 65.29; H, 12.69; N, 20.04; Li,1.98. Found: C, 64.69; H, 11.12; N, 19.96; Li, 2.3.

Synthesis of [W(Nt-Bu)2Cl((Ni-Pr)2CNi-Pr2)] (3). Equimolar samples of [W(NtBu)2Cl2py2] (2.53 g, 4.5 mmol) and 2 (1.57 g, 4.5 mmol) were dissolved in 40 mL of n-hexane. During stirring the green solution for 1 day at 25 °C, the color gradually turns to black. After filtration, all volatile parts were removed by standard vacuum distillation at 25 °C. The black crude product was recrystallized from saturated n-hexane solution to yield pure 3 (1.76 g, 66.6%) in the form of pale-yellow crystals. 1 H NMR (250 MHz, C6D6, 298 K): δ 0.97 (d, 6H, CNCHMe2); 0.98 (d, 6H, CNCHMe2); 1.11

(d, 6H, WNCHMe2); 1.47 (s, 18H, Nt-Bu); 1.54 (d, 6H, WNCHMe2); 3.17 (sept, 2H,

CNCHMe2); 4.08 (sept, 1H, WNCHMe2); 4.15 (sept, 1H, WNCHMe2). 13 C NMR (250 MHz, C6D6, 298 K): δ 22.6 (Me2CH); 22.7 (Me2CH); 24.3 (Me2CH); 24.5

(Me2CH); 32.6 (NC(CH3)3); 45.3 (Me2CH); 47.8 (Me2CH); 48.7 (Me2CH); 67.4 (NC(CH3)3);

168.7 (CN3). EI-MS [m/z (%)]: 587 (8) [M+]; 572 (8) [M+ - Me]; 530 (4) [M+ - t-Bu]; 473 (4) [M+ - 2t-Bu]; + 446 (18) [M - 2Nt-Bu]; 331 (4) [((i-PrN)2 - W - Cl) - 2H]; 291 (10) [t-BuNH - W -Cl]; 226

(56) [(Ni-Pr)2CNi-Pr2)]; 184 (68) [(Ni-Pr)2CNi-Pr) +H-i-Pr]; 169 (38) [(Ni-Pr)2CNi-Pr) - Ni-

Pr]; 127 (25) [(Ni-Pr)2C+ H]; 113 (6) [i-Pr2NC + H]; 100 (24) [i-Pr2N]; 85 (40) [i-Pr2N- Me]; 69 (24) [i-PrNC]; 58 (35) [i-PrN + H]; 43 (100) [i-Pr].

Anal. calcd for WClN5C21H46: C, 43.42; H, 6.76; N, 12.05. Found: C, 43.15; H, 7.55; N, 11.97.

Synthesis of W(Nt-Bu)2(Cl){(i-PrN)2CNMe2} (4). A 0.55 g portion of LiNMe2 (10.8 mmol) was suspended in 60 mL of hexane. A 1.67 mL portion of N,N’ -diisopropylcarbodiimide (10.8 mmol) was added, and the mixture was stirred until a clear solution formed. This solution was transferred to another solution of 6 g of [W(Nt-Bu)2Cl2py2] (10.8 mmol) dissolved in 60 mL of toluene. The reaction mixture was allowed to stir overnight. The solution was filtered, and the solvent was stripped to leave a yellow crude product, which was recrystallized from toluene to yield 4.6 g of pale yellow crystals (80%).

113 8. Experimental Section

1 HNMR (250 MHz, C6D6, 298 K): δ 1.09 (d, 6H, Ni-Pr); 1.51 (s, 18H, Nt-Bu);1.57 (d, 6H,

Ni-Pr); 2.19 (s, 6H, NMe2); 3.79 (sept, 1H, Ni-Pr); 3.91(sept, 1H, Ni-Pr). 13 C NMR (250 MHz, C6D6, 298 K): δ 23.7 (Me2CHN); 24.5 (Me2CHN); 32.6 (NCMe3); 39.4

(Me2N); 45.7 (Me2CHN); 48.8 (Me2CHN); 67.5 (NCMe3); 168.4 (CN3). EI-MS [m/z (relative intensity, %)]: 531 (2) [M+]; 516 (1) [M+ - Me]; 390 (9) [M+ - 2Nt-Bu];

71 (18) [Nt-Bu]; 58 (100) [tBuH]; 44 (22) [NMe2]; 29 (10) [HCN].

Anal. calcd for WClN5C17H38: C, 38.39; H, 7.20; N, 13.17. Found: C, 38.69; H, 7.29; N, 12.59.

Synthesis of W(Nt-Bu)2(Cl){(i-PrN)2CNEt2} (5). 0.15 ml of HNEt2 (5.4 mmol) were dissolved in 60 mL of hexane. 3.38 mL of of butyllithium (15 wt %, solution in n-hexane, 5.4 mmol) was added, and a white precipitate formed. A 0.84 mL portion of N,N’ - diisopropylcarbodiimide (5.4 mmol) was added, and the mixture was stirred until a clear solution formed. This solution was transferred to another solution of 3 g of [W(Nt-

Bu)2Cl2py2] (5.4 mmol) dissolved in 50 mL of toluene. The reaction mixture was allowed to stir overnight. The solution was filtered, and the solvent was stripped to leave a yellow crude product, which was recrystallized from toluene to yield 1.52 g of pale yellow crystals (50.2%). 1 HNMR (250 MHz, C6D6, 298 K): δ 0.72 (t, 6H, CH2CH3); 1.12 (d, 6H, Ni-Pr); 1.49 (s, 18H,

Nt-Bu);1.57 (d, 6H, Ni-Pr); 2.68 (quart, 4H, CH2CH3); 3.78 (sept, 1H, Ni-Pr); 3.91(sept, 1H, Ni-Pr). 13 C NMR (250 MHz, C6D6, 298 K): δ 13.4 (CH2CH3) 23.9 (Me2CHN); 24.4 (Me2CHN); 32.6

(NCMe3); 43.3 (CH2CH3); 45.9 (Me2CHN); 48.6 (Me2CHN); 67.5 (NCMe3); 168.5 (CN3). + + + EI-MS [m/z (relative intensity, %)]: 559 (2) [M ]; 546 (2) [M - Me]; 418 (9) [M - (i-PrN)2C

- Et]; 362 (2) [(t-BuN)2WCl + H]; 291 (3) [(t-BuN)WCl + H]; 198 (17) [H(i-PrN)2CNEt2];

182 (21) [H(i-PrN)2CNEt2 – Me - H]; 167 (24) [H(i-PrN)2CNEt2 – 2Me - H]; 141 (11) [(i-

PrNCNEt2]; 126 (8) [(i-PrN)2C]; 113 (24) [i-PrNCNEt + H]; 99 (27) [i-PrNCNH2]; 85 (13) [i-

PrNCNH2]; 72 (18) [HNt-Bu]; 58 (100) [t-BuH]; 43 (39) [i-Pr]; 29 (12) [HCN].

Anal. calcd for WClN5C19H42: C, 40.76; H, 7.50; N, 12.51. Found: C, 40.84; H, 7.69; N, 11.94.

Synthesis of W(Nt-Bu)2(Cl){(CyN)2CNEt2} (6). 0.15 ml of HNEt2 (5.4 mmol) were dissolved in 60 mL of hexane. 3.38 mL of butyllithium (15 wt %, solution in n-hexane, 5.4 mmol) was added, and a white precipitate formed. A 1.11 g portion of N,N’-

114 8. Experimental Section dicyclohexylcarbodiimide (5.4 mmol) was added, and the mixture was stirred until a clear solution formed. This solution was transferred to another solution of 3 g of [W(Nt-

Bu)2Cl2py2] (5.4 mmol) dissolved in 50 mL of toluene. The reaction mixture was allowed to stir overnight. The solution was filtered, and the solvent was stripped to leave a yellow crude product, which was recrystallized from hexane to yield 2.2 g of pale yellow crystals (62.1%). 1 HNMR (250 MHz, C6D6, 298 K): δ 0.79 (t, 6H, CH2CH3); 1.20 – 1.78 (mult., 22H, NCy); 13 1.49 (s, 18H, C NMR (250 MHz, C6D6, 298 K): δ 13.5 (CH2CH3) 25.7 (CyN); 25.8 (CyN);

26.0 (CyN); 27.1 (CyN); 32.7 (NCMe3); 34.2 (CyN); 34.4 (CyN); 43.3 (CH2CH3); 54.7

(CyHN); 57.6 (CyN); 67.5 (NCMe3); 168.8 (CN3). + + + EI-MS [m/z (relative intensity, %)]: 660 (1) [M ]; 566 (2) [M - NEt2]; 544 (4) [M - Cy - H];

362 (2) [(t-BuN)2WCl + H]; 420 (10) [(t-BuN)2WNCy + 2H]; 362 (6) [(t-BuN)2WCl + H];

278 (29) [(CyN)2CNEt2]; 206 (14) [(CyN)2C]; 163 (22) [(CyN)2C - i-Pr]; 124 (24) [CyNCH];

99 (21) [CyNH2]; 83 (37) [Cy]; 72 (18) [HNt-Bu]; 58 (100) [t-BuH]; 41 (53) [CH3CHCH2]; 29 (23) [HCN].

Anal. calcd for WClN5C19H42: C, 46.92; H, 7.87; N, 10.94. Found: C, 46.89; H, 7.60; N, 11.04.

Synthesis of W(Nt-Bu)2NMe2((Ni-Pr)2CNi-Pr2) (7). Equimolar samples of 3 (1.66 g, 2.7 mmol) and LiNMe2 (0.14 g, 2.7 mmol) were placed in a Schlenk tube, and 40 mL of diethyl ether was added. After stirring the mixture at room temperature overnight, filtration, and removal of the solvent in vacuo, the analytically almost pure product was obtained as a yellow, viscous oil. Further attempts of purification by crystallization at low temperature and by distillation failed. Yield: 0.80 g (49.7%). 1 H NMR (250 MHz, C6D6, 298 K): δ 1.05 (d, 12H CNCHMe2); 1.26 (d, 12H, WNCHMe2);

1.48 (s, 18H, Nt-Bu); 3.25 (sept, 2H, CNCHMe2); 3.51 (s, 6H, NMe2); 4.02 (sept, 2H,

WNCHMe2). 13 C NMR (250 MHz, C6D6, 298 K): δ 22.9 (Me2CH); 24.9 (Me2CH); 34.3 (NC(CH3)3); 48.4

(Me2CH); 55.4 (Me2CH); 66.6 (NC(CH3)3); 168.6 (CN3). EI-MS [m/z (%)]: 596 (1) [M+]; 553 (4) [M+ - i-Pr]; 508 (5) [M+ - Nt-Bu - Me]; 454 (5) [M+ -

2Nt-Bu]; 427 (1.5) [W(Nt-Bu)2NMe2((Ni-Pr)CNi-Pr)]; 371 (1.5) [W(Nt-Bu)2NMe2 + H]; 300

(2,5) [W(Nt-Bu)NMe2]; 226 (30), [(Ni-Pr)2CNi-Pr2]; 184 (16) [((Ni-Pr)2CNi-Pr2) - i-Pr]; 169

(8) [((Ni-Pr)2CNi-Pr2) - Ni-Pr]; 127 (10) [C(Ni-Pr)2]; 100 (11) [Ni-Pr2]; 85 (30) [(Ni-Pr2) -

Me]; 69 (19) [i-PrNC]; 58 (100) [i-PrN + H]; 43 (55) [i-Pr]; 28 (16) [H2CN].

Anal. calcd for WN6C23H52: C, 46.31; H, 8.72; N, 14.09.

115 8. Experimental Section

Found: C, 45.34; H, 9.04; N, 13.37.

Synthesis of W(Nt-Bu)2(NMe2){(i-PrN)2CNMe2} (8). A 0.2 g portion of LiNMe2 (3.8 mmol) was suspended in 30 mL of hexane. A solution of 2 g of [W(Nt-Bu)2(Cl){(i-PrN)2CNMe2}] (3.8 mmol) in toluene was added, and the whole mixture was allowed to stir overnight. The reaction mixture was filtered, and the solvent was stripped off, leaving a yellow solid crude product which was sublimed at 100 °C and 10-3 mbar to yield 1.4 g (68 %) of a yellow solid. 1 H NMR (250 MHz, C6D6, 298 K): δ 1.27 (d,12H, Ni-Pr); 1.51 (s, 18H, Nt-Bu); 2.36 (s, 6H,

NMe2); 3.55 (s, 6H, NMe2); 3.83 (sept, 2H, Ni-Pr). 13 C NMR (250 MHz, C6D6, 298 K): δ 24.4 (Me2CHN); 32.6 (Me2CHN); 34.3 (NCMe3); 39.9

(Me2CHN); 46.5 (Me2CHN); 55.8 (Me2N); 66.7 (NCMe3); CN3 not detectable. + + + EI-MS [m/z (relative intensity, %)]: 540 (12) [M ]; 495 (100) [M - NMe2 - H]; 482 (7) [M - + + + t-Bu]; 454 (10) [M - 2i-Pr]; 438 (4) [M - 2i-Pr - Me - H]; 413 (6) [M - (i-PrN)2C]; 399 (34) + [M - 2t-Bu]; 273 (11) [W(NMe2)2)]; 171 (5) [H(i-PrN)2CNMe2]; 114 (11) [i-PrNHCNMe2];

99 (12) [i-PrNCN2H2]; 85 (9) [i-PrNCNH2]; 71 (19) [Nt-Bu]; 58 (26) [Ht-Bu]; 43 (18) [i-Pr]; 29 (6) [HCN].

Anal. calcd for WN6C19H44: C, 42.23; H, 8.21; N, 15.55. Found: C 42.14; H, 8.12; N, 15.13.

Synthesis of W(Nt-Bu)2(N3)((Ni-Pr)2CNi-Pr2) (9). A sample of 3 (1.0 g, 1.7 mmol) was treated with an excess of dry, finely powdered sodium azide (0.4 g, 6.1 mmol) in a mixture of toluene (40 mL) and THF (5 mL) in a Schlenk tube at reflux over 4 days. The formed NaCl and excess NaN3 were removed by filtration, and the solvent was stripped in vacuo. The residue was dissolved in n-hexane and filtrated from insoluble byproducts. Stripping the solvent again left a yellow oily substance. The crude product was short-path distilled at 10-2 Torr (1 Pa) and 80 °C to give a yellow solid, which sublimes unchanged at the same conditions. Crystallization from toluene at -30 °C failed to give single crystals suitable for X- ray structural characterization. Yield: 0.65 g (64.4%). 1 H NMR (250 MHz, C6D6, 298 K): δ 0.92 (d, 6H, CNCHMe2); 0.93 (d, 6H, CNCHMe2); 1.06

(d, 6H, WNCHMe2); 1.39 (d, 6H, WNCHMe2); 1.46 (s, 18H, Nt-Bu); 3.11 (sept, 2H,

CNCHMe2); 4.01 (sept, 2H, WNCHMe2). 13 C NMR (250 MHz, C6D6, 298 K): δ 22.5 (Me2CH); 22.6 (Me2CH); 24.3 (Me2CH); 24.4

(Me2CH); 33.3 (NC(CH3)3); 45.2 (Me2CH); 47.1 (Me2CH); 48.7 (Me2CH); 67.9 (NC(CH3)3);

168.5 (CN3).

116 8. Experimental Section

-1 FT IR (dry film, cm ): 2086, vs (νN-N).; 2970, vs (νC-H).

Anal. calcd for WN8C21H46: C, 42.42; H, 7.80; N, 18.84. Found: C, 41.82; H, 7.55; N, 18.50.

Synthesis of W(Nt-Bu)2(H){(i-PrN)2CNMe2} (10). A 4.6 g portion of [W(Nt-

Bu)2(Cl){(iPrN)2CNMe2}] (8.6 mmol) was dissolved in 40 mL of toluene, and 8.6 mL of

LiBEt3H in THF (1 mol/L) was added. The solution was heated to 80 °C overnight. Now about a quarter of the solvent volume was stripped in a vacuum, and the resulting solution was filtered. The residual solvent was stripped to yield to brown crude product which was sublimed at 100 °C and 10-3 mbar to yield 2 g (47%) of a beige solid. Crystals suitable for X- ray crystallography were grown by cooling a saturated solution of 11 in methylene chloride. 1 H NMR (250 MHz, C6D6, 298 K): δ 1.22 (d, 6H, Ni-Pr); 1.26 (d, 6H, Ni-Pr); 1.54 (s, 18H,

Nt-Bu); 2.15 (s, 6H, NMe2); 3.70 (sept, 2H, Ni-Pr); 12.59 (s, 1H, WH). 13 C NMR (250 MHz, C6D6, 298 K): δ 26.8 (Me2CHN); 27.8 (Me2CHN); 37.2 (NCMe3); 41.5

(Me2N); 47.7 (Me2CHN); 48.7 (Me2CHN); 68.8 (NCMe3); 172.8 (CN3). -1 FT-IR (KBr, cm ): 2967 s (νC-H); 1854 s (νW-H). EI-MS [m/z (relative intensity, %)]: 497 (43) [M+]; 482 (13) [M+ - Me]; 454 (12) [M+ - i-Pr]; + + 413 (5) [M - i-Pr - NMe2 + H]; 370 (4) [(t-BuN)2WNMe2]; 356 (12) [M - 2t-Bu]; 310 (4)

[W(Ni-Pr)2C]; 298 (4) [W(Ni-Pr)2]; 282 (4) [WHNi-PrCN2]; 255 (7) [WNt-Bu]; 171 (8)

[H(Ni-Pr)2CNMe2]; 156 (4) [H(Ni-Pr)2CNMe]; 113 (11) [i-PrNCNMe2]; 99 (13) [i-PrNC-

(NH)2]; 85 (9) [Me2NCN2H]; 71 (53) [Nt-Bu]; 58 (100) [t-BuH]; 41 (30) [CH3CHCH2]; 29 (9) [HCN].

Anal. calcd for WN5C17H39: C, 41.05; H, 7.90; N, 14.08. Found: C, 40.78; H, 8.53; N, 13.99.

Synthesis of [W(Nt-Bu)2Cl(NC(NMe2)2)]2 (11). A sample of 1.15 ml of 1,1,3,3- tetramethylguanidine (9 mmol) was dissolved in 40 mL of diethyl ether. A volume of 6 mL of butyllithium (15 wt %, solution in n-hexane, 9 mmol) was added at 0 °C, and the reaction mixture was allowed to stir for a period of 30 min. A white precipitate was formed, and the whole suspension was added to a solution of [W(Nt-Bu)2Cl2py2] (5.0 g) in toluene (30 mL). The mixture was refluxed for 24 h. After filtration, the solvent was stripped, and the crude product was recystallized from toluene to yield 2.95 g (68.9%) of a microcrystalline pale- yellow powder. Well-shaped single crystals of 11 were obtained by recrystallization from n- hexane at -30 °C over 1 day.

117 8. Experimental Section

1 H NMR (250 MHz, C7D8, 223 K): δ 1.54 (s, 18H, Nt-Bu); 2.39 (6H, NMe2); 3.05 (6H,

NMe2). 13 C NMR (250 MHz, C6D6, 298 K): δ 33.1 (NC(CH3)3); 40.6 (broad, NMe2); 66.8

(NC(CH3)3); 178.4 (CN3). EI-MS [m/z (%)]: 475 (1,6) [M+/2]; 460 (8) [M+/2 - Me]; 390 (1) [M+/2 - Me - NtBu + H]; 334 (2,5) [M+/2 - 2Nt-Bu + H]; 318 [M+/2 - 2Nt-Bu - Me]; 291 (1,2) [t-BuNH - W-Cl]; 184 + (20) [W ]; 115 (48) [NHC(NMe2)2]; 86 (16) [(Me2N)2C - Me]; 71 (95) [Nt-Bu]; 58 (100) [t-

Bu + H]; 44 (52) [NMe2]; 27 (17) [HCN].

Anal. calcd for W2Cl2N10C26H60: C, 32.83; H, 6.36; N, 14.72. Found: C, 32.74; H, 6.36; N, 14.61.

Synthesis of [(W(Nt-Bu)2(N3)(NC(NMe2)2)]2 (12). [W(Nt-Bu)2Cl{NC(NMe2)2}]2 (1.0 g, 1.05 mmol) was combined with dried and finely powdered NaN3 (0.5 g, 7.6 mmol) in a Schlenk tube. Toluene (50 mL) and THF (5 mL) were added, and the mixture was refluxed for 1 day. The precipitated NaCl and excess NaN3 were removed by filtration, and the solvent was stripped in vacuo at 25 °C. The crude product was recrystallized from toluene at -30 °C to give 0.86 g (84.4%) of yellow crystals. 1 H NMR (250 MHz, C6D6, 298 K): δ 1.46 (s, 18H, Nt-Bu); 2.71 (s, 12H, NMe2). 13 C NMR (250 MHz, C6D6, 298 K): δ 33.7 (NC(CH3)3); 40.2 (NMe2); 67.3 (NC(CH3)3);

176.8 (CN3). -1 FT IR (toluene solution, cm ): 2072, vs (νN-N).

Anal. calcd for W2N16C26H60: C, 32.37; H, 6.33; N, 23.23. Found: C, 32.76; H, 6.68; N, 22.74.

2 Synthesis of [(W(Nt-Bu)2(N3)(µ -N3)py)]2 (13). [W(Nt-Bu)2Cl2py2] (1.0 g, 1.8 mmol) was combined with dried and finely powdered NaN3 (0.5 g, 7.6 mmol) in a Schlenk tube. Toluene (50 mL) and THF (5 mL) were added, and the mixture was heated to reflux for 1 day. The precipitated NaCl and excess NaN3 were removed by filtration, and the solvent was stripped in vacuo at 25 °C. Crystals of 8 were obtained by diffusion crystallization at 25 °C, placing a layer of pentane over a concentrated toluene solution of the crude product. Yield: 0.38 g (43.1%). 1 H NMR (250 MHz, C6D6, 298 K): δ 1.41 (s 36 H, Nt-Bu); 6.62 (t, 4H, py); 8.64 (t, 2H, py); 8.96 (d, 4H, py).

118 8. Experimental Section

13 C NMR (250 MHz, C6D6, 298 K): δ 32.8 (NC(CH3)3); 68.3 (NC(CH3)3); 125.2 (py); 139.1 (py); 153.5 (py). -1 FT IR (toluene solution, cm ): 2082, vs (νN-N).

Anal. calcd for W2N18C26H46: C, 31.91; H, 4.74; N, 25.77. Found: C, 31.24; H, 4.96; N, 24.24.

8.6. Synthesis of the Molybdenum-Compounds.

Synthesis of Mo(Nt-Bu)2Cl2(DME) (14): 20 g of 20g Na2MoO4 (97,13mmol) were suspended in 200 ml of DME, and 54,1 ml Triethylamine (0,38 mol), 20,5 ml tert-Butylamine (0,19 mol) and 99,31 ml of Chlorotrimethylsilane were added. A white, voluminous precipitate was formed, and the whole mixture was heated to 70 °C overnight. The Suspension was filtered, and the filtrate was stripped to yield to yellow powder. The yellow crude product was recrystallized from pentane to yield 36,05 g (37,19 %) of Mo(Nt-

Bu)2Cl2(DME) as yellow crystals. 1 H NMR (250 MHz, CD3CN, 298 K): δ 1,41 (s, 18H, Nt-Bu); 3,22 (s, 4H,

(MeOCH2CH2OMe)); 3,42 (s, 4H, (MeOCH2CH2OMe)). 13 C NMR (250 MHz, CD3CN, 298 K): δ 30,1 (NCMe3); 62,1 (NCMe3); 70,8 (DME); 71,8 (DME).

Anal. calcd for MoC12H28N2Cl2O2: C, 36,10; H, 7,07; N, 7,02. Found: C, 35,49; H, 7,15; N, 7,69.

Synthesis of Mo(Nt-Bu)2Cl{(Ni-Pr)2CMe2} (15): A portion of 1.28 g LiNMe2 (25 mmol) was suspended in 40 mL of Hexane. A portion of 3.87 ml N,N-diisopropylcarbodiimide (25 mmol) was added, and the mixture was stirred until a clear solution formed. This solution was transferred to another solution of 10 g of [Mo(Nt-Bu)2Cl2(dme)] (25 mmol) dissolved in 60 mL of toluene. The reaction mixture was allowed to stir overnight. The solution was filtered, and the solvent was stripped to leave a brownish crude product, which was recrystallized from toluene to yield 7.3 g of yellow crystals (65 %). 1 H NMR (250 MHz, C6D6, 298 K): δ 1,04 (d, 6H, Ni-Pr); 1,48 (s, 18H, Nt-Bu); 1,62 (d, 6H,

Ni-Pr); 2,26 (s, 6H, NMe2); 3,68 (sept, 2H, Ni-Pr). 13 C NMR (250 MHz, C6D6, 298 K): δ 23,9 (Me2CHN); 24,6 (Me2CHN); 31,2 (NCMe3); 39,4

(Me2N); 45,9 (Me2CHN); 49,1 (Me2CHN); 70,1 (NCMe3); 169,5(CN3).

119 8. Experimental Section

+ + + EI-MS [m/z (relative intensity, %)]: 445 (2) [M ]; 401 (2.5) [M - NMe2]; 388 (1.5) [M - + + tBu]; 302 (3) [M - 2Nt-Bu- H]; 248 (3) [M - Nt-Bu- (i-PrN)2C]; 170 (58) [(Ni-Pr)2CMe2];

156 (9) [{(Ni-Pr)2CMe2}- Me + H]; 113 (17) [i-PrNCNMe2]; 99 (30) [i-PrNC(NH)2]; 85 (19)

[i-PrNCNH2]; 71 (100) [Nt-Bu]; 58 (42) [t-BuH]; 43 (51) [i-Pr]; 27 (12) [HCN].

Anal. calcd for MoClN5C17H38: C, 46.00; H, 8.63; N, 15.70. Found: C, 45.68; H, 8.89; N, 16.11.

Synthesis of Mo(Nt-Bu)2I{(Ni-Pr)2CMe2} (16). A portion of 2 g of 14 (4.5 mmol) was dissolved in 80 ml of THF. 6.7 g of NaI (44.7 mmol) was added and the reaction mixture was allowed to stir overnight. The THF was removed and 50 ml of hexane was added to the residual solid. The resulting solution was filtered, and the hexane was removed in vacuo leaving a yellow powder as a product. Crystals suitable for X-ray diffraction were grown by cooling a saturated solution of 15 in hexane. Yield: 2.0 g (83.1 %) 1 H NMR (250 MHz, C6D6, 298 K): δ 1,04 (d, 6H, Ni-Pr); 1,52 (s, 18H, Nt-Bu); 1,65 (d, 6H,

Ni-Pr); 2,23 (s, 6H, NMe2); 3,60 (sept, 2H, Ni-Pr). 13 C NMR (250 MHz, C6D6, 298 K): δ 23,8 (Me2CHN); 25,0 (Me2CHN); 31,4 (NCMe3); 39,5

(Me2N); 46,1 (Me2CHN); 49,7 (Me2CHN); 70,9 (NCMe3); 171,8(CN3). + + + EI-MS [m/z (relative intensity, %)]: 535 (8) [M ]; 491 (16) [M - NMe2]; 478 (15) [M - t- + + Bu]; 424 (27) [Mo(Nt-Bu)2(I)Ni-Pr+H]; 410 (72) [M - (i-PrN)2C] 395 (13) [M - 2Nt-Bu]; + 340 (23) [M - Nt-Bu- (i-PrN)2C]; 295 (6) [(t-BuN)MoI]; 171 (20) [H(Ni-Pr)2CMe2]; 156 (9)

[{H(Ni-Pr)2CMe2}- Me]; 126 (10) [(i-PrN)2C]; 111 (25) [(i-PrN)2C-Me]; 99 (9) [i-

PrNC(NH)2]; 83 (8) [i-PrNCN]; 71 (100) [Nt-Bu]; 57 (61) [t-Bu]; 41 (11) [CH2CHCH3].

Anal. calcd for MoIN5C17H38: C, 38.14; H, 7.15; N, 13.08. Found: C, 38.31; H, 7.25; N, 12.78.

Synthesis of Mo(Nt-Bu)2(N3){(Ni-Pr)2CMe2} (17). A portion of 2 g of 14 (4.5 mmol) was dissolved in 60 ml of Toluene. 2 g of NaN3 (44.7 mmol) and 20 ml THF was added and the reaction mixture was heated to 80 °C overnight. The volume of the solution was reduced to the half and the precipitated NaCl and excess NaN3 were removed by filtration, and the solvent was stripped in vacuo at 25 °C.. Crystals suitable for X-ray diffraction were grown by cooling a saturated solution of 15 in hexane. Yield: 1.5 g (73.3 %) 1 H NMR (250 MHz, C6D6, 298 K): δ 1,09 (d, 6H, Ni-Pr); 1,56 (s, 18H, Nt-Bu); 1,58(d, 6H,

Ni-Pr); 2,30 (s, 6H, NMe2); 3,70 (sept, 2H, Ni-Pr).

120 8. Experimental Section

13 C NMR (250 MHz, C6D6, 298 K): δ 23,8 (Me2CHN); 24,8 (Me2CHN); 32,0 (NCMe3); 39,2

(Me2N); 45,7 (Me2CHN); 48,4(Me2CHN); 70,6 (NCMe3); 169,1 (CN3). + + + EI-MS [m/z (relative intensity, %)]: 450 (1) [M ]; 408 (16) [M - N3]; 340 (28) [M - 2t-Bu +

4H]; 311 (18) [Mo(N3){(Ni-Pr)2CNMe2} + 3H]; 254 (18) [Mo(Nt-Bu)2NH2]; 171 (20) [H(Ni-

Pr)2CMe2]; 156 (18) [{H(Ni-Pr)2CMe2}- Me]; 114 (85) [i-PrNHCNMe2]; 99 (79) [i-

PrNC(NH)2]; 85 (8) [i-PrNCNH2]; 71 (100) [Nt-Bu]; 58 (61) [t-BuH]; 43 (80) [i-Pr].

Anal. calcd for MoN8C17H38: C, 38.14; H, 7.15; N, 13.08. Found: C, 38.31; H, 7.25; N, 12.78.

121 8. Experimental Section

8.7. Crystallographic Data Table 21: Crystallographic Data for Compounds (1),(3) & (4) (1) (3) (4)

empirical formula WCl2N4C18H28 WClN5C21H46 WClN5C17H38 fw 555,19 587,93 531,81

space group Pna21 P-1 C2/c a(Å) 16,4634(9) 9,059(2) 15,399(2) b(Å) 8,8987(4) 10,359(3) 9,5815(15) c(Å) 14,9136(8) 15,293(4) 32,617(5) α(°) 90 81,786(10) 90 β(°) 90 80,206(9) 103,140(15) γ(°) 90 86,217(7) 90 V(Å3) 2184,89(19) 1398,5(6) 4686,5(12) Z 4 2 8 -3 ρcalc (g cm ) 1,688 1,396 1,507 µ (mm-1) 5,540 4,240 5,052 F (000) 1088 596 2127 Reflections collected 4987 4792 5333 Reflections unique 4281 4301 2734 2 Goodness-of-fit on F 1,023 0,824 0,848 R1 0,0277 0,0336 0,0528 wR2 0,0566 0,0900 0,0685

122 8. Experimental Section

Table 22: Crystallographic Data for Compounds (5),(6)&(10)

(5) (6) (10)

empirical formula WClN5C19H42 WClN5C25H50 WCl2N5C18H41 fw 559,88 640,00 582,31

space group P-1 P21/n P21/n a(Å) 9,0672(10) 10,1152(4) 8,8327(6) b(Å) 19,4701(16) 16,8737(9) 19,0411(13) c(Å) 23,514(2) 17,5953(9) 15,2788(10) α(°) 70,892(8) 90 90 β(°) 79,572(9) 102,199(4) 101,430(6) γ(°) 78,998(9) 90 90 V(Å3) 3818,8(6) 2935,4(2) 2518,7(3) Z 6 4 4 -3 ρcalc (g cm ) 1,461 1,448 1,536 µ (mm-1) 4,654 4,046 4,810 F (000) 1692 1304 1168 Reflections collected 17385 6751 4414 Reflections unique 10086 4714 3938 2 Goodness-of-fit on F 0,877 0,709 1,954 R1 0,0499 0,0289 0,0263 wR2 0,0762 0,0603 0,0411

123 8. Experimental Section

Table 23 Crystallographic Data for Compounds (11),(12)&(13) (11) (12) (13)

empirical formula W2Cl2N10C26H60 W2N16C26H60 W2N18C26H46 fw 951,44 964,60 978,51

space group P21/c P21/n P-1 a(Å) 11,8101(10) 9,3581(8) 10,6021(7) b(Å) 17,1297(13) 21,4148(18) 15,9447(10) c(Å) 9,8014(6) 10,0342(9) 16,8824(10) α(°) 90 90 93,330(5) β(°) 109,816(7) 110,507(8) 93,330(5) γ(°) 90 90 96,270(5) V(Å3) 1865,4(2) 1883,4(3) 2780,7(3) Z 2 2 3 -3 ρcalc (g cm ) 1,694 1,436 1,753 µ (mm-1) 6,335 6,119 6,245 F (000) 936 952 1428 Reflections collected 4312 4295 15707 Reflections unique 3426 3842 7868 Goodness-of-fit on F2 1,001 1,105 0,666 R1 0,0291 0,0510 0,0385 wR2 0,0728 0,0997 0,0573

124 8. Experimental Section

Table 24: Crystallographic Data for Compounds (14), (16)&(17)

(14) (16) (17)

empirical formula MoCl2N2O2C12H28 MoIN5C17H38 MoN8C17H38 fw 399,20 535,36 450,49

space group Pbca P21/c P21/c a(Å) 12,4615(7) 10,3984(4) 10,4700(7) b(Å) 29,7947(18) 9,6827(3) 9,0556(6) c(Å) 9,8564(6) 23,0207(7) 24,4483(18) α(°) 90 90 90 β(°) 90 90,279(3) 93,525(6) γ(°) 90 90 90 V(Å3) 3659,6(4) 2317,80(13) 2313,6(3) Z 8 4 4

-3 ρcalc (g cm ) 1,449 1,534 1,293 µ (mm-1) 1,009 1,906 0,584 F (000) 1648 1080 952 Reflections collected 3209 5253 5095 Reflections unique 2513 4387 4301 Goodness-of-fit on F2 0,731 0,991 1,124 R1 0,0245 0,0372 0,0292 wR2 0,0537 0,0910 0,0740

125 9. References

9. References

[1] G. C. Henderson, J. C. Galletly, J. Soc. Chem. Ind. 1908, 27, 387. [2] I. Langmuir, J. Am. Chem. Soc. 1913, 35, 931. [3] F. C. T. So, E. Kolawa, X. Zhao, M. A. Nicolet, Thin Solid Films 1987, 153, 507. [4] S. K. Cheung, S. P. Kwok, A. Kaleta, K. M. Yu, J. M. Jaklevic, C. L. Liang, N. W. Cheung, E. E. Haller, J. Vav. Sci. Technol. B 1988, 6, 1779. [5] K. M. Yu, J. M. Jaklevic, E. E. Haller, S. K. Cheung, S. P. Kwok, J. Appl. Phys. 1988, 64, 1284. [6] J. S. Lee, C. S. Park, J. W. Yang, J. Y. Kang, D. S. Ma, J. Appl. Phys. 1990, 67, 1134. [7] A. Paccagnella, A. Callegari, A. Carnera, M. Gasser, E. Latta, M. Murakami, M. Norcott, J. Appl. Phys. 1991, 69, 2356. [8] K. Steiner, H. Mikami, K. Nishihori, M. Nagaoka, N. Uchitomi, J. Vav. Sci. Technol. B 1991, 9, 1601. [9] M. A. Nicolet, M. Bartur, J. Vac. Sci. Technol. 1981, 19, 786. [10] K. E. Howard, (Dow Chemical Co., USA). Application: US US 5114750, 1992, p. 5 pp. [11] M. J. Kim, D. M. Brown, W. Katz, J. Electrochem Soc. 1983, 130, 1196. [12] R. M. Fix, R. G. Gordon, D. M. Hoffman, J. Am. Chem. Soc. 1990, 112, 7833. [13] R. Fix, R. G. Gordon, D. M. Hoffman, Thin Solid Films 1996, 288, 116. [14] H. J. Lee, J.-G. Choi, C. W. Colling, M. S. Mudholkar, L. T. Thompson, Appl. Surf. Sci. 1995, 89, 121. [15] The International Technology Roadmap for Semiconductord 2005 Edition – Interconnect, , Semiconductor Ind. Assoc., Santa Clara, CA, 2005. [16] O. Van der Straten, Y. Zhu, K. Dunn, E. T. Eisenbraun, A. E. Kaloyeros, J. Mater. Res. 2004, 19, 447. [17] X. Zhao, N. P. Magtoto, J. A. Kelber, Thin Solid Films 2005, 478, 188. [18] C. J. Carmalt, A. C. Newport, S. A. O'Neill, I. P. Parkin, A. J. P. White, D. J. Williams, Inorg. Chem. 2005, 44, 615. [19] L. Zhou, H. Sun, J. Chen, Y. Yao, Q. Shen, J. Polym. Sci. A 2005, 43, 1778. [20] M. P. Coles, P. B. Hitchcock, Eur. J. Inorg. Chem. 2004, 2662. [21] A. P. Duncan, S. M. Mullins, J. Arnold, R. G. Bergman, Organometallics 2001, 20, 1808. [22] Y. Luo, Y. Yao, Q. Shen, Macromolecules 2002, 35, 8670. [23] A. F. Hollemann, E. Wiberg, Lehrbuch der anorganischen Chemie, 101 ed., Walter deGruyter-Verlag, Berlin New York, 1995. [24] S. Rideal, Journal of the Chemical Society, Transactions 1889, 55, 41. [25] C. J. Smithells, H. P. Rooksby, J. Chem Soc. 1927, 1882. [26] G. Hagg, Z. physik. Chem. 1930, 7, 339. [27] Z. G. Pinsker, S. V. Kaverin, Kristallografiya 1957, 2, 386. [28] R. Kiessling, Y. H. Liu, J. Metals 1951, 3, 639. [29] N. Schonberg, Acta Chem. Scand. 1954, 8, 204. [30] J. Neugebauer, A. J. Hegedus, T. Millner, Z. anorg. u. allgem. Chem. 1959, 302, 50. [31] F. A. Gunther, H. G. Schneider, Kristallografiya 1966, 11, 683. [32] V. I. Khitrova, Z. G. Pinsker, Kristallografiya 1958, 3, 545. [33] V. I. Khitrova, Z. G. Pinsker, Kristallografiya 1960, 5, 711. [34] V. I. Khitrova, Z. G. Pinsker, Kristallografiya 1961, 6, 882. [35] V. I. Khitrova, Kristallografiya 1963, 8, 873. [36] V. I. Khitrova, Kristallografiya 1961, 6, 549. [37] V. I. Khitrova, Kristallografiya 1962, 7, 374. 126 9. References

[38] V. I. Khitrova, Z. G. Pinsker, Kristallografiya 1959, 4, 545. [39] J. P. Lu, W. Y. Hsu, J. D. Luttmer, L. K. Magel, H. L. Tsai, J. Electrochem Soc. 1998, 145, L21. [40] R. Karam, R. Ward, Inorg. Chem. 1970, 9, 1385. [41] P. Ettmayer, Monatsh. Chem. 1970, 101, 127. [42] W. Lengauer, J. Cryst. Growth 1988, 87, 295. [43] S. L. Roberson, D. Finello, R. F. Davis, Thin Solid Films 1998, 324, 30. [44] W. W. Fuller, S. A. Wolf, D. U. Gubser, E. F. Skelton, T. L. Francavilla, J. Vav. Sci. Technol. A 1983, 1, 517. [45] D. A. Papaconstantopoulos, W. E. Pickett, B. M. Klein, L. L. Boyer, Phys. Rev. B 1985, 31, 752. [46] N. Savvides, J. Appl. Phys. 1987, 62, 600. [47] N. V. Troitskaya, Z. G. Pinsker, Kristallografiya 1961, 6, 43. [48] The International Technology Roadmap for Semiconductors 1999 Edition, Semiconductor Ind. Assoc., Santa Clara, CA, 1999. [49] Chemical Rubber Company handbook of chemistry and physics, 79 ed., CRC Press, Florida, 1998. [50] H.-C. Chung, C.-P. Liu, Surf. Coat. Technol. 2006, 200, 3122. [51] D. Adams, G. F. Malgas, N. D. Theodore, R. Gregory, H. C. Kim, E. Misra, T. L. Alford, J. W. Mayer, J. Vac. Sci.Technol., B 2004, 22, 2345. [52] A. Baunemann, PhD-Thesis thesis, Ruhr-Universität (Bochum), 2006. [53] A. E. Kaloyeros, E. Eisenbraun, Annu. Rev. Mater. Sci. 2000, 30, 363. [54] Diffusion in metals, Am. Soc. Met., Westerville, 1973. [55] S.-J. Im, S.-H. Kim, K.-S. Park, K.-B. Cho, K.-B. Kim, Mat. Res. Soc. Symp. Proc. 2000, 612, D6.7.1. [56] L. Peters, Semicond. Internat. 2003, 26, 50. [57] P. Ho, R. Rajagopalan, H. Chong, H. Chung, J. Yu, Semicond. Internat. 2004, 27, 61. [58] J. S. Becker, R. G. Gordon, Appl. Phys. Lett. 2003, 82, 2239. [59] J. S. Becker, S. Suh, S. Wang, R. G. Gordon, Chem. Mater. 2003, 15, 2969. [60] K. Choi, H. N. Alshareef, H. C. Wen, H. Harris, H. Luan, Y. Senzaki, P. Lysaght, P. Majhi, B. H. Lee, Appl. Phys. Lett. 2006, 89, 032113/1. [61] S. Haukka, H. Huotari, (Finland). Application: US US, 2007, p. 16pp. [62] M. Lemberger, A. Baunemann, A. J. Bauer, Microelectronics Reliability 2007, 47, 635. [63] S. Veprek, M. Haussmann, S. Reiprich, J. Vav. Sci. Technol. A 1996, 14, 46. [64] P. Hones, M. Diserens, R. Sanjines, F. Levy, J. Vav. Sci. Technol. B 2000, 18, 2851. [65] J. Endrino, V. Derflinger, C. Gey, (Unaxis Balzers AG, Liechtenstein). PCT WO 2006/ 005217, 2006, p. 24 pp. [66] T. Kadono, T. Kubota, Y. Okamoto, Catal. Today 2003, 87, 107. [67] D. McKay, J. S. J. Hargreaves, R. F. Howe, Catal. Lett. 2006, 112, 109. [68] P. Afanasiev, Inorg. Chem. 2002, 41, 5317. [69] L. Volpe, M. Boudart, J. Solid State Chem. 1985, 59, 332. [70] R. Kojima, K. i. Aika, Appl. Catal. A 2001, 219, 141. [71] S. Gong, H. Chen, W. Li, B. Li, Catal. Commun. 2004, 5, 621. [72] S. L. Roberson, D. Finello, R. F. Davis, Surf. Coat. Technol. 1998, 102, 256. [73] L. Volpe, M. Boudart, J. Phys. Chem. 1986, 90, 4874. [74] V. Keller, H. Lauron-Pernot, G. Djega-Mariadassou, J. Mol. Catal. A 2002, 188, 163. [75] B.-S. Suh, Y.-J. Lee, J.-S. Hwang, C.-O. Park, Thin Solid Films 1999, 348, 299. [76] B.-S. Suh, H.-K. Cho, Y.-J. Lee, W.-J. Lee, C.-O. Park, J. Appl. Phys. 2001, 89, 4128. [77] K. K. Shih, D. B. Dove, J. Vav. Sci. Technol. A 1990, 8, 1359.

127 9. References

[78] T. Yamamoto, M. Kawate, H. Hasegawa, T. Suzuki, Surf. Coat. Technol. 2005, 193, 372. [79] M. Uekubo, T. Oku, K. Nii, M. Murakami, K. Takahiro, S. Yamaguchi, T. Nakano, T. Ohta, Thin Solid Films 1996, 286, 170. [80] L. Mond, (UK). US 455230, 1891. [81] H. O. Pierson, Handbook of Chemical Vapor Deposition, 2 ed., Noyes Publication, Park Ridge,New Jersey 1999. [82] J. Khanderi, PhD-Thesis thesis, Ruhr-Universität (Bochum), 2006. [83] H. M. Manasevit, Appl. Phys. Lett. 1968, 12, 156. [84] A. Devi, W. Rogge, A. Wohlfart, F. Hipler, H. W. Becker, R. A. Fischer, Chem. Vap. Deposition 2000, 6, 245. [85] J. Kouvetakis, J. McMurran, C. Steffek, T. L. Groy, J. L. Hubbard, Inorg. Chem. 2000, 39, 3805. [86] T. Suntola, J. Antson, US 4058430, 1977. [87] T. Nakajima, K. Watanabe, N. Watanabe, J. Electrochem Soc. 1987, 134, 3175. [88] Y. T. Kim, S. K. Min, Appl. Phys. Lett. 1991, 59, 929. [89] Y. T. Kim, C. W. Lee, S. K. Min, Appl. Phys. Lett. 1992, 61, 537. [90] C. W. Lee, Y. T. Kim, J. Y. Lee, Appl. Phys. Lett. 1994, 64, 619. [91] C. W. Lee, Y. T. Kim, Solid-State Electronics 1995, 38, 679. [92] K.-M. Chang, T.-H. Yeh, I. C. Deng, C.-W. Shih, J. Appl. Phys. 1997, 82, 1469. [93] J. W. Klaus, S. J. Ferro, S. M. George, Appl. Surf. Sci. 2000, 162-163, 479. [94] S.-H. Kim, J.-K. Kim, N. Kwak, H. Sohn, J. Kim, S.-H. Jung, M.-R. Hong, S. H. Lee, J. Collins, Electrochem. Solid-State Lett. 2006, 9, C54. [95] S. Ganguli, L. Chen, T. Levine, B. Zheng, M. Chang, J. Vav. Sci. Technol. B 2000, 18, 237. [96] C. W. Lee, Y. T. Kim, S. K. Min, Appl. Phys. Lett. 1993, 62, 3312. [97] S.-H. Kim, S. S. Oh, K.-B. Kim, D.-H. Kang, W.-M. Li, S. Haukka, M. Tuominen, Appl. Phys. Lett. 2003, 82, 4486. [98] A. M. Hoyas, J. Schuhmacher, D. Shamiryan, J. Waeterloos, W. Besling, J. P. Celis, K. Maex, J. Appl. Phys. 2004, 95, 381. [99] S.-H. Kim, S. S. Oh, H.-M. Kim, D.-H. Kang, K.-B. Kim, W.-M. Li, S. Haukka, M. Tuominen, J. Electrochem Soc. 2004, 151, C272. [100] A. M. Hoyas, J. Schuhmacher, C. M. Whelan, T. Fernandez Landaluce, D. Vanhaeren, K. Maex, J. P. Celis, J. Appl. Phys. 2006, 100, 114903/1. [101] A. M. Hoyas, Y. Travaly, J. Schuhmacher, T. Sajavaara, C. M. Whelan, B. Eyckens, O. Richard, S. Giangrandi, B. Brijs, W. Vandervorst, K. Maex, J. P. Celis, A. M. Jonas, A. Vantomme, J. Appl. Phys. 2006, 99, 063515/1. [102] A. M. Hoyas, C. M. Whelan, J. Schuhmacher, J. P. Celis, K. Maex, Electrochem. Solid-State Lett. 2006, 9, F64. [103] K.-S. Kim, M.-S. Lee, S.-S. Yim, H.-M. Kim, K.-B. Kim, H.-S. Park, W. Koh, W.-M. Li, M. Stokhof, H. Sprey, Appl. Phys. Lett. 2006, 89, 081913/1. [104] C. W. Lee, Y. T. Kim, J. Vac. Sci.Technol., B 2006, 24, 1432. [105] S. Bystrova, A. A. I. Aarnink, J. Holleman, R. A. M. Wolters, J. Electrochem Soc. 2005, 152, G522. [106] J. E. Kelsey, C. Goldberg, G. Nuesca, G. Peterson, A. E. Kaloyeros, B. Arkles, J. Vav. Sci. Technol. B 1999, 17, 1101. [107] B. H. Lee, K. Yong, J. Electrochem Soc. 2004, 151, C594. [108] Y. M. Sun, E. R. Engbrecht, T. Bolom, C. Cilino, J. H. Sim, J. M. White, J. G. Ekerdt, K. Pfeifer, Thin Solid Films 2004, 458, 251. [109] W. A. Nugent, R. L. Harlow, Inorg. Chem. 1980, 19, 777. [110] W. A. Nugent, Inorg. Chem. 1983, 22, 965.

128 9. References

[111] M. H. Tsai, S. C. Sun, H. T. Chiu, S. H. Chuang, Appl. Phys. Lett. 1996, 68, 1412. [112] E. L. Crane, H.-T. Chiu, R. G. Nuzzo, J. Phys. Chem. B 2001, 105, 3549. [113] Y.-W. Yang, J.-B. Wu, J. Wang, Y.-F. Lin, H.-T. Chiu, Surf. Sci. 2006, 600, 743. [114] J.-B. Wu, Y.-W. Yang, Y.-F. Lin, H.-T. Chiu, J. Vav. Sci. Technol. A 2003, 21, 1620. [115] A. Rugge, J. S. Becker, R. G. Gordon, S. H. Tolbert, Nano Lett. 2003, 3, 1293. [116] R. G. Gordon, S. Suh, J. Becker, (President and Fellows of Harvard College, USA). Application: WO 2004007796, 2004, p. 37 pp. [117] D.-H. Kim, Y. J. Kim, Y. S. Song, B.-T. Lee, J. H. Kim, S. Suh, R. Gordon, J. Electrochem Soc. 2003, 150, C740. [118] C. L. Dezelah IV, O. M. El-Kadri, M. J. Heeg, C. H. Winter, J. Mater. Chem. 2004, 14, 3167. [119] O. M. El-Kadri, M. J. Heeg, C. H. Winter, Dalton Trans. 2006, 1943. [120] O. J. Bchir, K. M. Green, H. M. Ajmera, E. A. Zapp, T. J. Anderson, B. C. Brooks, L. L. Reitfort, D. H. Powell, K. A. Abboud, L. McElwee-White, J. Am. Chem. Soc. 2005, 127, 7825. [121] O. J. Bchir, K. M. Green, M. S. Hlad, T. J. Anderson, B. C. Brooks, L. McElwee- White, J. Cryst. Growth 2004, 261, 280. [122] O. J. Bchir, K. M. Green, M. S. Hlad, T. J. Anderson, B. C. Brooks, C. B. Wilder, D. H. Powell, L. McElwee-White, J. Organomet. Chem. 2003, 684, 338. [123] O. J. Bchir, S. W. Johnston, A. C. Cuadra, T. J. Anderson, C. G. Ortiz, B. C. Brooks, D. H. Powell, L. McElwee-White, J. Cryst. Growth 2003, 249, 262. [124] S. Won Yong, S. Kim Young, J. Anderson Timothy, L. Reitfort Laurel, I. Ghiviriga, L. McElwee-White, J. Am. Chem. Soc. 2006, 128, 13781. [125] O. J. Bchir, K. C. Kim, T. J. Anderson, V. Craciun, B. C. Brooks, L. McElwee-White, J. Electrochem Soc. 2004, 151, G697. [126] C. B. Wilder, L. L. Reitfort, K. A. Abboud, L. McElwee-White, Inorg. Chem. 2006, 45, 263. [127] C. L. Dezelah IV, O. M. El-Kadri, K. Kukli, K. Arstila, R. J. Baird, J. Lu, L. Niinistoe, C. H. Winter, J. Mater. Chem. 2007, 17, 1109. [128] T. Nakajima, T. Shirasaki, J. Electrochem Soc. 1997, 144, 2096. [129] P. Alen, M. Ritala, K. Arstila, J. Keinonen, M. Leskelae, J. Electrochem Soc. 2005, 152, G361. [130] S.-C. Sun, H.-T. Chiu, (Taiwan Semiconductor Manufacturing Company, Taiwan). Application: US 6114242, 2000, p. 10 pp. [131] V. Miikkulainen, M. Suvanto, T. A. Pakkanen, Chem. Mater. 2007, 19, 263. [132] M. J. Kim, D. M. Brown, W. Katz, J. Electrochem Soc. 1983, 130, 1196. [133] G. Chandra, A. D. Jenkins, M. F. Lappert, R. C. Srivastava, J. Chem. Soc. A 1970, 2550. [134] P. J. Bailey, S. Pace, Coord. Chem. Rev. 2001, 214, 91. [135] M. P. Coles, Dalton Trans. 2006, 985. [136] W. P. Kretschmer, C. Dijkhuis, A. Meetsma, B. Hessen, J. H. Teuben, Chem. Commun. 2002, 608. [137] C.-C. Chang, C.-S. Hsiung, H.-L. Su, B. Srinivas, M. Y. Chiang, G.-H. Lee, Y. Wang, Organometallics 1998, 17, 1595. [138] Y. Zhou, G. P. A. Yap, D. S. Richeson, Organometallics 1998, 17, 4387. [139] R. L. Zuckerman, R. G. Bergman, Organometallics 2000, 19, 4795. [140] T.-G. Ong, G. P. A. Yap, D. S. Richeson, Chem. Commun. 2003, 2612. [141] A. P. Kenney, G. P. A. Yap, D. S. Richeson, S. T. Barry, Inorg. Chem. 2005, 44, 2926. [142] M. K. T. Tin, G. P. A. Yap, D. S. Richeson, Inorg. Chem. 1998, 37, 6728.

129 9. References

[143] P. J. Bailey, K. J. Grant, L. A. Mitchell, S. Pace, A. Parkin, S. Parsons, J. Chem. Soc., Dalton Trans. 2000, 1887. [144] A. Baunemann, D. Rische, A. Milanov, Y. Kim, M. Winter, C. Gemel, R. A. Fischer, Dalton Trans. 2005, 3051. [145] S. M. Mullins, A. P. Duncan, R. G. Bergman, J. Arnold, Inorg. Chem. 2001, 40, 6952. [146] D. Rische, A. Baunemann, M. Winter, R. A. Fischer, Inorg. Chem. 2006, 45, 269. [147] F. A. Cotton, C. A. Murillo, X. Wang, C. C. Wilkinson, Dalton Trans. 2006, 4623. [148] F. A. Cotton, C. A. Murillo, X. Wang, C. C. Wilkinson, Inorg. Chem. 2006, 45, 5493. [149] F. A. Cotton, J. C. Durivage, N. E. Gruhn, D. L. Lichtenberger, C. A. Murillo, L. O. Van Dorn, C. C. Wilkinson, J. Phys. Chem. B 2006, 110, 19793. [150] F. A. Cotton, L. M. Daniels, C. A. Murillo, D. J. Timmons, C. C. Wilkinson, J. Am. Chem. Soc. 2002, 124, 9249. [151] F. A. Cotton, M. Daniels Lee, A. Murillo Carlos, J. Timmons Daren, C. Wilkinson Chad, J. Am. Chem. Soc. 2002, 124, 9249. [152] Y. Kim, A. Baunemann, H. Parala, A. Devi, R. A. Fischer, Chem. Vap. Deposition 2005, 11, 294. [153] D. C. Bradley, R. J. Errington, M. B. Hursthouse, R. L. Short, B. R. Aschroft, G. R. Clark, A. J. Nielson, C. E. F. Rickard, J. Chem. Soc., Dalton Trans. 1987, 2067. [154] W. Clegg, R. J. Errington, D. C. R. Hockless, C. Redshaw, J. Chem. Soc., Dalton Trans. 1993, 1965. [155] K. Dreisch, C. Andersson, C. Staalhandske, Polyhedron 1993, 12, 1335. [156] W.-H. Leung, M.-C. Wu, J. L. C. Chim, W.-T. Wong, Polyhedron 1998, 17, 457. [157] M. H. Chisholm, B. W. Eichhorn, J. C. Huffman, Organometallics 1989, 8, 67. [158] R. R. Schrock, K.-Y. Shih, D. A. Dobbs, W. M. Davis, J. Am. Chem. Soc. 1995, 117, 6609. [159] S. M. Holmes, D. F. Schafer, II, P. T. Wolczanski, E. B. Lobkovsky, J. Am. Chem. Soc. 2001, 123, 10571. [160] A. A. H. Van der Zeijden, C. Sontag, H. W. Bosch, V. Shklover, H. Berke, D. Nanz, W. Von Philipsborn, Helv. Chim. Acta 1991, 74, 1194. [161] H. W. Lam, G. Wilkinson, B. Hussain-Bates, M. B. Hursthouse, J. Chem. Soc., Dalton Trans. 1993, 781. [162] E. Solari, J. Hesschenbrouck, R. Scopelliti, C. Floriani, N. Re, Angew. Chem., Intern. Ed. 2001, 40, 932. [163] L. Hirsivaara, M. Haukka, J. Pursiainen, Inorg. Chem. Commun. 2000, 3, 508. [164] M. Scheer, J. Muller, M. Schiffer, G. Baum, R. Winter, Chem. Eur. J. 2000, 6, 1252. [165] A. Baunemann, M. Lemberger, A. J. Bauer, H. Parala, R. A. Fischer, Chem. Vap. Deposition 2007, 13, 77. [166] H. H. Fox, K. B. Yap, J. Robbins, S. Cai, R. R. Schrock, Inorg. Chem. 1992, 31, 2287. [167] A. Bell, W. Clegg, P. W. Dyer, M. R. J. Elsegood, V. C. Gibson, E. L. Marshall, J. Chem. Soc., Chem. Commun. 1994, 2547. [168] A. Bell, W. Clegg, P. W. Dyer, M. R. J. Elsegood, V. C. Gibson, E. L. Marshall, J. Chem. Soc., Chem. Commun. 1994, 2247. [169] P. W. Dyer, V. C. Gibson, W. Clegg, J. Chem.Soc., Dalton Trans. 1995, 3313. [170] H.-T. Chiu, G.-B. Chang, W.-Y. Ho, S.-H. Chuang, G.-H. Lee, S.-M. Peng, J. Chin. Chem. Soc. 1994, 41, 755. [171] R. C. B. Copley, P. W. Dyer, V. C. Gibson, J. A. K. Howard, E. L. Marshall, W. Wang, B. Whittle, Polyhedron 1996, 15, 3001. [172] V. C. Gibson, C. Redshaw, W. Clegg, M. R. J. Elsegood, J. Chem.Soc., Dalton Trans. 1997, 3207. [173] D. del Rio, F. Montilla, A. Pastor, A. Galindo, A. Monge, E. Gutierrez-Puebla, J. Chem. Soc., Dalton Trans. 2000, 2433.

130 9. References

[174] A. J. Nielson, M. W. Glenny, C. E. F. Rickard, J. M. Waters, J. Chem. Soc., Dalton Trans. 2000, 4569. [175] K. A. Rufanov, D. N. Zarubin, N. A. Ustynyuk, D. N. Gourevitch, J. Sundermeyer, A. V. Churakov, J. A. K. Howard, Polyhedron 2001, 20, 379. [176] Y. S. Won, Y. S. Kim, T. J. Anderson, L. L. Reitfort, I. Ghiviriga, L. McElwee-White, J. Am. Chem. Soc. 2006, 128, 13781. [177] R. Kellner, J.-M. Mermet, M. Otto, H. M. Widmer, Wiley-VCH, Weinheim, 1997, p. 691. [178] A. Wohlfart, PhD-Thesis thesis, Ruhr Universität (Bochum), 2002. [179] C. R. Brundle, C. A. Evans jr., S. Wilson, Butterworth-Heinemann, Boston London Oxford Singapore Sydney Toronto Wellington, 1992, p. 571. [180] R. Kellner, J.-M. Mermet, M. Otto, H. M. Widmer, Wiley-VCH, Weinheim, 1997, p. 653.

131 Table of Contents

Curriculum Vitae

Personal Data: Birthday: 09. 08.1978 Hobbies: Climbing, Cycling, Trecking, Place of birth: Bochum, Germany Music, Billard Family status: unmarried Nationality: german

Education 08/1985 to 06/1989 Elementary School in Bochum-Linden 08/1989 to 06/1998 Attendance of the Theodor-Körner Schule (High School) in Bochum-Dahlhausen. Degree:Abitur 07/1998 to 07/1999 Civilian service at the „St.Josefs-Hospital“ in Bochum-Linden 10/1999 to 05/2004 Studies in chemistry at the „Ruhr-Universität“ in Bochum 11/2003 to 05/2004 Diploma thesis in inorganic chemistry at the chair of inorganic chemistry II at the „Ruhr-Universität“ in Bochum Topic: „Precursorchemistry of Lanthanide Nitrides for CVD- Applications” 07/2004 to 12/2007 PhD-Studies at the chair of inorganic chemistry II at the „Ruhr- Universität“ in Bochum Topic: „MOCVD of Tungsten- and Molybdenum Nitrides“ Scholarships Scholarship for final degree of the „Verein zur Förderung der Chemie und Biochemie an der Ruhr-Universität Bochum“

Additional Qualifikations Minor subject studies in „Patent right in engineering sciences“

Languages German (native speaker) English (expert) French (basic knowledge) Latin (basic knowledge)

Publications [1] J. Khanderi, D. Rische, H. W. Becker, R. A. Fischer, J. Mater. Chem. 2004, 14, 3210. [2] A. Baunemann, D. Rische, A. Milanov, Y. Kim, M. Winter, C. Gemel, R. A. Fischer, Dalton Trans. 2005, 3051. [3] D. Rische, A. Baunemann, M. Winter, R. A. Fischer, Inorg. Chem. 2006, 45, 269. [4] D. Rische, H. Parala, E. Gemel, M. Winter, R. A. Fischer, Chem. Mater. 2006, 18, 6075. [5] D. Rische, H. Parala, A. Baunemann, T. Thiede, R. A. Fischer, Surf. Coat. Technol. 2007, 201, 9125.

Professional Experience 08/2001 to Student assistant at the „Ruhr-Universität“ in Bochum (Lab work; 06/2004 Mentoring of tutorial groups and practicals) 07/2004 to Scientific assistant at the „Ruhr-Universität“ in Bochum (Lab work; 12/2007 Mentoring of bachelor students and practicals)

132 Table of Contents

Ich erkläre hiermit, dass ich die vorliegende Dissertation selbst verfasst und mich dabei keiner anderen als der von mir ausdrücklich bezeichneten Quellen und Hilfen bedient habe. Ich erkläre hiermit des Weiteren, dass ich an keiner anderen Stelle ein Prüfungsverfahren beantragt bzw. die Dissertation in dieser oder anderer Form bereits anderweitig als Prüfungsarbeit verwendet oder einer anderen Fakultät als Dissertation vorgelegt habe.

Bochum, den 26.11.2007

Daniel Rische

133