Aims of this Thesis

Rare earth metal complexes bearing NHC ligands were first described in the rush that followed the discovery of the free stable NHC. Today, although they are still less common than late transition metals bearing NHC ligands^[2b], the investigations on REE NHCs paved the way to a significant number of complexes, which equipped the scientists with a repertory of approaches to future compounds.^{[4, 19c]} However, the knowledge concerning synthetic methods and especially reactivity is still somewhat limited in this field. The reports of catalytic activity are so far mostly limited to polymerization reactions; the systematic investigations on catalytic utility could provide further insight into this still under-explored complex class. Furthermore, REE NHC complexes show very promising reactivity for a range of very interesting reactions^{[19c, 19e]}, such as small molecule activation[93b, 114] or C–H bond activation^[166] and their applicability should be further explored by gaining access to compounds containing polydentate ligands, which might show superior stability due to stabilization by anionic anchors or/and additional carbene moieties. Such sites promote a robust attachment of NHCs to hard Lewis acidic centre allowing the exposition of NHC coordination sites without complete dissociation of the whole ligand.^[4] This generates an exciting possibility of a liberated NHC moiety participating in catalytic reactions as a non-innocent ligand.

The generation of NHC complexes with Lewis acidic metal centres, however, often requires a detour via alkali metal NHC intermediates, which are now widely recognised as efficient transmetallation agents.^[19e] Therefore, the work described in second chapter of this thesis focuses on the synthesis and characterization of various donor-functionalised chelating NHC pro-ligands and the investigations towards their reactivity upon subjection to alkali metal bases.

The deprotonation products should be fully characterised and, if possible, isolated, especially since fully characterised alkali metal adducts with poly(dentate) NHC ligands, particularly neutral ones, are rare. In order to maximize the chances for a successful transmetallation procedure investigations on the stability of the generated compounds should be conducted and optimal reaction conditions determined.

Subsequently, the respective REE compounds should be synthesized using various approaches, ether by transmetallation of alkali metal NHC adducts obtained in the second chapter or by a direct reaction of the ligand precursors with REE compounds acting as internal base. Optimal conditions for the synthesis of the respective complexes should be determined.

Since the structural information on REE NHC compounds is limited, all complexes should be fully characterised und used for further studies towards their reactivity with small molecules or in redox chemistry. This work will be described in the third chapter of this thesis.

Linear coordinated Au(I) complexes show promising antiproliferative activity in cancer cells.

Because of high complexity of biological medium, the fine-tuning of the steric and electronic properties of these compounds is crucial to cytotoxic properties and selectivity. Hereby, NHCs offer a possibility for efficient and fast optimization of respective complexes. Furthermore, the high stability of Au(I) NHC complexes makes them suitable candidates for rapid screening of potential metal-based drugs.^[167] Although a variety of Au(I) NHC compounds have been reported to show high potential as anti-cancer drugs, they still suffer from several disadvantages (low stability again thiols, low selectivity and etc., see Section 1.4.3.2).

Therefore, further investigations towards other ligand systems combining liphophilic and hydrophilic functional groups, which would allow simultaneously high membrane permeability as well as stability and solubility in biological medium, should be conducted. Additionally, ligand systems offering a possibility of post-synthetic modifications towards coupling with biological markers or anchoring of additional metals should be explored. Therefore, the fourth chapter of this thesis focuses on homo-dinuclear Ag(I) and Au(I) complexes supported by bridge-functionalised bis(NHC) ligands. Detailed investigations on conformational behaviour of these compounds are performed and correlated to possible antiproliferative activity in human cancer cell lines. Furthermore, first preliminary experiments towards post-synthetic modifications are conducted.

Synthesis and Characterization of Donor-Functionalised N-Heterocyclic Carbenes

and their Group I Adducts

Chapter 2

2.1 Results and Discussion

This chapter describes the work towards modification of NHC ligand precursors with additional neutral or anionic donor sites. Reactivity studies towards generation of free NHCs or alkali metals NHC adducts are conducted in order to subsequently use these intermediates salt elimination reactions with rare earth metals. In favour of enhanced clarity and content structuring the reactions all bis(imidazolium) salts with internal rare earth metal bases as well as in multi-component systems will be discussed in Chapter 3. This section summarizes the reactions of these pro-ligands with alkali metal bases and, if applicable, the conversion of the obtained products with various organic substrates.

2.1.1 N-(3,5-Di-tert-butyl-2-hydroxyphenyl) and N-Bis(3,5-di-tert-butyl-2-hydroxybenzyl) Functionalised Mono(imidazolium) Salts H2(L1^R)Br and H3(L2)Br

2.1.1.1 Synthesis and Characterization

The bidentate N-(3,5-di-tert-butyl-2-hydroxyphenyl)-functionalised pro-ligands H2(L1^Me)Br and H2(L1^Mes)Br are obtained according to published procedures (Scheme 2.1.1).^[168]

Scheme 2.1.1. Synthesis of bidentate N-(3,5-di-tert-butyl-2-hydroxyphenyl)-functionalised imidazolium bromides.

The exact mechanism of this reaction is unknown. Tashiro examined the conversion of 4-bromo-2,4,6-tri-tert-butyl-2,5-cyclohexadien-1-one 1a with various substituted pyridines and made an observation that the presence of ethylene glycol (EG) is not necessary for the reaction to proceed.^[169] Nevertheless, its presence increases the yields dramatically, but only by using a fixed molar ratio of 1a:EG:pyridine (1:1:2). Based on the ESR experiments the authors proposed a radical mechanism for this conversion. Hereby the formation of isobutylene generated by debutylation at the C2 position was confirmed by trapping this gas in toluene and converting it with AlCl3 to tert-butyltoluenes. The regioselectivity of the reaction was explained

with kinetic arguments. Furthermore, various side products such as 2,4,6-tri-tert-butylphenol, 2-bromo-4,6-di-tert-butylphenol, 2,4,6-tri-tert-butyl-4-hydroxycyclohexa-2,5-dien-1-one and 2,6-di-tert-butylcyclohexa-2,5-diene-1,4-dione are observed. It seems likely that the reaction of 1a with substituted imidazoles could proceed via the same mechanism (Scheme 2.1.2), and the formation of the side-products mentioned earlier is probably the reason for low yields in this synthetic procedure (36 % for R = Me and 16 % for R = Mes).

Scheme 2.1.2. Possible mechanism for the reaction of 4-bromo-2,4,6-tri-tert-butyl-2,5-cyclohexadien-1-one 1a with substituted imidazoles based on the mechanism proposed by Tashiro for the similar reaction of 1a with pyridines.^[169]

The synthesis of the related tridentate N-bis(3,5-di-tert-butyl-2-hydroxybenzyl) functionalised mono(imidazolium) salt H3(L2)Br has been already described and proceeds via double nucleophilic substitution (Scheme 2.1.3).^[170]

Scheme 2.1.3. Synthesis of tridentate N-functionalised imidazolium bromide H3(L2)Br.

All compounds have been characterised by means of multinuclear NMR spectroscopy and ESI-MS. The purity of the material was confirmed by elemental analysis.

2.1.1.2 Deprotonation Studies Bidentate H2(L1^R)Br

H2(L1^Me)Br can be successfully deprotonated on NMR scale with KNʺ or NaNʺ at RT in THF (Scheme 2.1.4). Hereby Na(L1^Me) and K(L1^Me) are obtained in quantitative yields. The ¹H NMR spectra of both compounds show the generation of symmetrical compounds with the formation of only one species for the potassium congener (Figure 2.1.1). In comparison to K(L1^Me), the Na NHC adduct is formed as a product mixture in ratio of 0.2:1.0 (A:B), which however could also indicate a formation of a single oligomeric product with different structural motifs. Such reactivity would be not surprising as a formation of complex alkali metal clusters was repeatedly observed (see Introduction, Section 1.2.1).^[65]

Scheme 2.1.4.Formation of alkali metal NHC adducts derived from H2(L1^R)Br. The structures of M(L1^R) are simplified suggestions based on experimental data.

The value for K–CC resonance (212.67 ppm) in K(L1^Me) lies within the range reported in the literature.^[65b] Notably, a splitting of the carbon resonances attributed to methyl groups suggest that K NHC adduct could also exist as an oligomer in solution. In comparison to K(L1^Me) the

13C NMR spectrum of Na(L1^Me) displays a second minor set of the singlets (see Experimental, SI, Figure 5.3.1-2). No resonance attributed to a carbene is detected in this case, probably due to longer relaxation time and lower sensibility of Na NHCs in comparison to K NHCs. Besides the missing carbene singlet and the resonance of one of the quaternary aryl-group-carbon (probably due to overlapping) for the main species (B), the resonance pattern of the product B is similar to K(L1^Me), indicating that also with Na an alkali metal NHC adduct could have been formed.

According to ²⁹Si INEPT NMR the conversion of H2(L1^Me)Br with KNʺ is quantitative as only a signal corresponding to HNʺ is visible. On the other hand, the spectrum of Na(L1^Me) shows a number of different resonances ranging from –28.17 till 8.34 ppm. The most prominent are at –11.97 and 1.96 ppm, corresponding to NaNʺ and HNʺ, which suggests that either NaNʺ is not completely consumed or it is incorporated into the compound. Such behaviour have been previously observed for the Na NHC adduct 21 (see Introduction, Scheme 1.2.1).^[57] In any

case, these findings rather support the assumption that Na(L1^Me) forms a complex polynuclear solvate in solution.

Both compounds, Na(L1^Me) and K(L1^Me), are stable in solution at RT under inert atmosphere conditions for at least 4 days. Hereby, neither the ¹H NMR spectra of these alkali metal NHC adducts nor the appearance of the samples show any signs of decomposition. Unfortunately, to date no crystals suitable for SC-XRD could be obtained for both compounds.

Figure 2.1.1. Comparison of ¹H NMR spectra of Na(L1^Me) (top) and K(L1^Me) (bottom) synthesized on NMR scale in THF-d8 at RT.

Also by treating the mesityl-substituted analogue H2(L1^Mes)Br with 2.0 eq. of KNʺ a clean reaction is observed in THF at RT yielding the deprotonated K NHC adduct K(L1^Mes) (Scheme 2.1.4).The ¹H NMR spectrum of the product suggests a symmetrical arrangement of the ligand in the obtained product (Figure 2.1.2). Also ¹³C NMR proves high symmetry and besides the carbene resonance all expected singlets can be clearly assigned (Figure 5.3.3). Finally, ²⁹Si NMR confirms a complete consumption of KNʺ by displaying solely the singlet attributed to HNʺ.

K(L1^Mes) is stable under inert conditions for at least 2 days. Unfortunately, this compound could not be characterised by SC-XRD due to decomposition during crystallization by slow diffusion of pentane into a solution of K(L1^Mes) in THF at RT. Therefore, the exact structure of K(L1^Mes) remains unknown.

In comparison to KNʺ the use of 2.0 eq. of NaNʺ for the deprotonation of 1.0 eq. of H2(L1^Mes)Br in THF at RT is not satisfactory. The NMR scale reaction contains still a lot of not-deprotonated

pro-ligand, and the multiplets attributed to deprotonated species are difficult to assign due to probable high degree of oligomerization (see SI, Figure 5.3.4). ¹³C NMR spectrum further confirms the observations made in the proton NMR, also the carbonic carbon resonance is not detectable (SI, Figure 5.3.5). According to ²⁹Si NMR most of the NaNʺ was consumed and converted to HNʺ, but a singlet at –11.93 ppm indicates a possible incorporation of NaNʺ into the cluster, since a free NaNʺ would react with the remaining pro-ligand. It is possible that due to complicated structure of the obtained Na NHC product the stoichiometry of the reaction was not exact. A formation of anionic dicarbenes cannot be excluded as well, as it has been previously observed for heavy alkali metals (see Introduction, Section 1.2.1).^[65b] Therefore, further investigations on this reaction are necessary.

Figure 2.1.2. ¹H NMR spectrum of K(L1^Mes) formed in a NMR scale reaction at RT in THF-d8.

In summary, although structural information could not be obtained, the NMR experiments prove a clean formation of K NHC adducts with N-(3,5-di-tert-butyl-2-hydroxyphenyl)-functionalised imidazol-2-ylidenes on NMR scale. The sodium compounds have been also successfully synthesized but in contrast to K NHCs they exhibit higher degree of oligomerization or are possibly more prone to rearrangement.

Tridentate H2(L2)Br

The deprotonation of tridentate pro-ligand H3(L2)Br can be conveniently performed in THF by addition of 3.0 eq. of KNʺ (Scheme 1.2.5). On NMR scale at RT the reaction proceeds cleanly and yields a highly symmetric product (Figure 2.1.3, bottom spectrum). ¹³C NMR spectrum of K(L2) confirms the formation of desired K NHC as well, as the chemical shift value for carbenic

carbon (δ (K–C) = 212.76 ppm) lies within the range of other reported K NHC compounds.^[57,

65b] Further spectroscopic characterization by ²⁹Si INEPT NMR confirms a clean conversion of H3(L2)Br with KNʺ since besides the singlet attributable to HNʺ only a very minor impurity at 4.83 ppm is present.

Scheme 2.1.5. Reaction of KN" with the pro-ligand H3(L2)Br. The structure of K(L2) is a simplified suggestion.

On preparatory scale, the ¹H NMR spectrum of crude product obtained after the treatment of H2(L2)Br with 3.0 eq. of KNʺ at RT for 18 h using standard Schlenk and glove box techniques reveals a formation of different products than in the NMR scale reaction described above (Figure 2.1.3). It is unlikely that K(L2) or other K NHC adducts are still present at this stage since no carbene resonance is visible in the ¹³C NMR spectrum of the residue. Also the appearance of the brown filtrate stands in contrast to the bright orange suspension of the NMR scale reaction.

To obtain more information about the formed compound the ESI-MS analysis of the crude product formed after 2.5 h of the reaction time at RT was performed (see Figure 5.3.43, SI).

The spectrum exhibits a main signal at 505.41 m/z, which belongs to ether the cation of the imidazolium salt H3(L2)Br or its product by 1,2-shift of N-(3,5-di-tert-butyl-2-hydroxybenzyl) substituents plus H⁺. The first case would indicate a presence of a K NHC adduct in the sample at the time of the injection and its re-protonation by formic acid. The second possibility,

however, would explain the asymmetrical set of resonances in ¹H NMR spectrum and the absence of the singlet attributed to carbenic carbon in ¹³C NMR. The 1,2-rearragement is further strongly supported by the literature as Kawaguchi also observed the same reactivity of Na(L2) upon warming it up to RT.^[171] Further signal at 723.50 m/z in the MS spectrum could not be assigned, but its isotope pattern suggests an organic molecule without bromide or potassium. Interestingly, a signal at 287.14 m/z indicates a cleavage between the imidazol-2-ylidene and one of N-substituents resulting in formation of [2b+H⁺] (Scheme 2.1.3). Such decay has been already previously observed for similar asymmetrically substituted pro-ligand 79 by treating it with Li[YN(i-Pr2)4] and n-BuLi at RT (see Introduction, section 1.3.1.2).^[97]

Figure 2.1.3. Comparison of the ¹H NMR spectra of the products of the reactions of H3(L2)Br with KNʺ at various conditions. The spectra were recorded in THF-d8 at RT.

Further modification of the synthetic procedure by lowering the reaction temperature to –78 °C and shortening the reaction time to 2 h yields K(L2) as major but impure product. The impurities were the same as seen in the conversions at RT and their formation is probably indebted by warming up of the filtrate during the removal of the solvent.

Interestingly, the comparison of ²⁹Si NMR spectra of the reaction on preparatory scale at –78 °C with NMR scale reaction conducted at RT shows the increase in the intensity of the singlet at 4.6 ppm in comparison to HNʺ resonating at 1.9 ppm (Figure 2.1.4). Finally, by conducting the same experiment on preparatory scale at RT the singlet at 4.6 ppm is replaced by a number of new resonances in the positive region, suggesting a possible direct involvement of amide in the decomposition process.

Figure 2.1.4. Comparison of ²⁹Si INEPT NMR spectra of the crude products obtained by the reaction of H3(L2)Br with KNʺ at various conditions.

Unfortunately, the washing of the crude product with toluene results in further decomposition of K(L2), similar to the Na(L2) adduct reported by Kawaguchi.^[171] Due to thermal sensitivity of K(L2) no crystals suitable for SC-XRD could be obtained and therefore, the exact structure of this alkali metal NHC adduct remains unknown.

On exploratory NMR scale the deprotonation of H3(L2)Br is also achieved in THF at RT by addition of NaNʺ. Hereby, in contrast to a clean symmetric product K(L2) a formation of structurally diverse motifs is evident (Figure 2.1.5). In contrast to potassium analogue, the appearance of carbenic carbon is not detected in ¹³C NMR spectrum of Na(L2), but the resonance pattern of the remaining carbon nuclei suggests that at least 3 species of ligand framework are formed (SI, Figure 5.3.6). The comparison of the spectra to the spectra of the decomposition product of K(L2) excludes the formation of the same 1,2-rearragement products. Therefore, in this case either structurally different Na(L2) adducts have been formed or other decomposition pathways are present. Additionally, in comparison to K(L2), the ²⁹Si NMR spectrum of Na(L2) displays singlets at –11.18 and 7.57 ppm. Therefore, similar to the reaction with H2(L1^R)Br described above, NaNʺ could be incorporated into the oligomeric structure, as seen e.g. for compound 21 (see Introduction, Scheme 1.2.1).^[57] Moreover, the resonances in the positive region of ²⁹Si NMR spectrum indicate possible presence of (poly)organosiloxane in the sample as well.^[172]

Figure 2.1.5. Comparison of the ¹H NMR spectra of the products obtained by the reactions of H3(L2)Br with KNʺ and NaNʺ by at RT in THF-d8 (NMR scale reactions).

In summary, N-(3,5-di-tert-butyl-2-hydroxyphenyl) and N-bis(3,5-di-tert-butyl-2-hydroxybenzyl) functionalised mono(NHC) ligands readily form alkali metal NHC adducts with potassium or sodium. However, although on smaller NMR scale these obtained compounds seem to be stable for at least 4 days at RT, on preparatory scale the isolation of respective complexes is impeded by their thermal instability. Therefore, the crystallographic characterization of M(L1^R) and M(L2) (M = K, Na; R = Me, Mes) could not be performed so far due to decomposition processes during crystallization. It is possible that due to higher Lewis acidity of Li, an isolation of respective Li NHC compound could be more straightforward. Although the exact molecular structures remain unknown, especially the potassium compounds K(L1^R) and K(L2) show great potential for application as transfer reagents in situ at –78 °C.

2.1.2 1,1’-(2-Hydroxyethane-1,1-diyl)-Bridge Functionalised Bis(imidazolium) Salts H3(L3^R)X2

2.1.2.1 Synthesis and Characterization

The synthesis of 1,1’-(2-hydroxyethane-1,1-diyl)-bridged bis(NHC) pro-ligand H3(L3^Me)Cl2 has been previously reported by Zhong et al.^[173] To investigate the implication of different wingtips on the solubility of potential complexes as well as the conformational changes, this class of ligands has been expanded for other N-substituents (i-Pr, Mes). Using 1-isopropylimidazole and 1-mesitylimidazole as precursors analogue synthetic method is applied (Scheme 2.1.6).

Scheme 2.1.6. Synthesis of 1,1’-(2-hydroxyethane-1,1-diyl)-functionalised bis(imidazolium) chlorides.

Unfortunately, the yields for H3(L3^i-Pr)Cl2 and H3(L3^Mes)Cl2 are quite low (48 % and 26 % respectively), especially for sterically demanding mesityl-substituents. Chloride-anion is a poor leaving group, but since the 2,2-dibromo or 2,2-diiodoethanol cannot be used due to their instability, high temperatures and long reactions times are required in this case. With the goal of increasing the yields and reducing the reaction times, the application of microwave irradiation was also investigated. For mesityl-substituted compounds the operation procedure at 110 °C and 100 W is found to yield small amounts of the desired compound after already 4 hours of the reaction time. Therefore, given the possibility of the usage of a microwave with sufficiently big enough reaction vessels and longer operation hours, this synthetic protocol could significantly improve the efficiency.

At first glance toluene seems like an unlikely solvent since the mono-substituted intermediates should have low solubility in such apolar solvent. Unfortunately, stirring 1-mesitylimidazole and 2,2-dichloroethanol in benzonitrile at 130 °C for 3 months does not yield bis(imidazolium) chlorides, also increasing the temperature to 140 °C and stirring the reaction mixture for another week under these conditions yields only an oily residue containing mono-substituted imidazolium salts. As a result, toluene proved itself as the best choice of the solvent. Apart from long reaction times, the procedure in the ACE pressure tubes is simple and all imidazolium salts can be obtained in high purity, as they precipitate out of toluene in contrast to the impurities.

If necessary, a salt metathesis reaction for the anion exchange to PF6— and BPh4— ions for all bis(imidazolium) chlorides can be performed in H2O. For the small amounts of H3(L3^Me)Cl2 an anion exchange with stoichiometric amounts of AgPF6 in acetonitrile shows better results due to good solubility of H3(L3^Me)(PF6)2 in H2O. All imidazolium salts were fully characterised by means of multinuclear NMR spectroscopy, ESI or FAB-MS and elemental analysis.

1H NMR spectrum of H3(L3^i-Pr)Cl2 in DMSO-d6 exhibits the expected shift of the imidazolium protons to high frequency (10.13 ppm), indicating the high tendency for easy deprotonation of NCHN-group with mild bases. In comparison to OH-proton corresponding to the triplet at 6.46 ppm the triplet attributed to the CH-group of the bridge is significantly shifted to higher frequency (7.25 ppm) as well. Therefore, the selective deprotonation of the OH group and the

C2 positions of the heterocyclic rings might pose some difficulties. The same tendencies in ¹H

C2 positions of the heterocyclic rings might pose some difficulties. The same tendencies in ¹H