UvA-DARE (Digital Academic 3-Methylindole-Based Tripodal Tetraphosphine Ruthenium Complexes in N2 Coordination and Reduction and Formic Acid Dehydrogenation

: The ruthenium(II) complexes RuCl 2 L1 H , RuCl 2 L1 CF3 , RuCl 2 L1 OMe and RuCl 2 L2 H were synthesized from [Ru( η 6 -benzene)Cl( µ -Cl)] 2 and the corresponding tripodal tris-3-methylindolephosphine-based ligands L1 H , L1 CF3 , L1 OMe , and L2 H . Stoichiometric reduction of these complexes with KC 8 yielded the corresponding ruthenium(0) dinitrogen complexes. The latter complexes were studied in the N 2 reduction with chlorosilanes and KC 8 , yielding stoichiometric amounts of the silylamines. The synthesized ruthenium(II) complexes are also active catalysts for the formic acid dehydrogenation reaction. gave


Introduction
Electrocatalytic reduction reactions that store energy by the conversion of H + , CO 2 , or N 2 , are important in the context of the transformation to a society based on sustainable energy. Therefore, the development of efficient catalysts that enable these reactions is needed. Homogeneous tripodal ligand transition metal complexes have shown beneficial reactivity over mono-or bidentate ligands in many organic transformations [1][2][3]. In recent years, tripodal ligands coordinated to molybdenum, iron, and cobalt were shown to lead to complexes that are capable of the reduction of N 2 to ammonia or silylamines [4][5][6][7][8][9][10]. Also, reversible hydrogen storage in CO 2 was shown to be possible with iron and cobalt based on tripodal ligands [11][12][13]. Previously, we reported the coordination of ligand L1 H (Figure 1) to rhodium [14,15] and ruthenium [16]. This ligand, which features three 1-(3-methylindolyl)diphenylphosphine groups tethered to a central phosphorus atom, stabilizes ruthenium in the oxidation states +II, +I, and 0. The isolated low oxidation state Ru I ClL1 H and Ru 0 (N 2 )L1 H complexes revealed interesting one-electron reactivity, as they were able to abstract chlorine atoms from either chloroform or dichloromethane, leading to formation of the corresponding Ru II complex. In this paper, we expand the coordination chemistry of ruthenium and 3-methylindole-based ligands [17][18][19][20][21][22][23] and investigate how modification of the ligand scaffold and its electronic properties influence the stability and activity of these complexes in dinitrogen coordination and subsequent reduction as well as in formic acid dehydrogenation. Ligand L1 H was modified using functionalized P(Ar) 2 units, wherein the arene carries either electron-withdrawing -CF 3 groups (L1 CF3 ) or electron-donating -OMe groups (L1 OMe ) in the para-position ( Figure 1). Additionally, the known structural isomer, L2 H ( Figure 2) [14], in which the central P-donor is bound to the three nitrogen atoms of the 2-(3-methylindolyl)diphenylphosphine moiety, was also used in these studies.

Results
Ligands L1 H and L2 H were prepared in two steps from 3-methylindole as previously reported by our group [14,15]. The new ligands L1 CF3 and L1 OMe were prepared in a similar way to L1 H , by reacting tris-2-(3-methylindolyl)phosphine [17] with the corresponding chlorodiarylphosphine ( Figure 1). known structural isomer, L2 H ( Figure 2) [14], in which the central P-donor is bound to the three nitrogen atoms of the 2-(3-methylindolyl)diphenylphosphine moiety, was also used in these studies.

Results
Ligands L1 H and L2 H were prepared in two steps from 3-methylindole as previously reported by our group [14,15]. The new ligands L1 CF3 and L1 OMe were prepared in a similar way to L1 H , by reacting tris-2-(3-methylindolyl)phosphine [17] with the corresponding chlorodiarylphosphine ( Figure 1).

Results
Ligands L1 H and L2 H were prepared in two steps from 3-methylindole as previously reported by our group [14,15]. The new ligands L1 CF3 and L1 OMe were prepared in a similar way to L1 H , by reacting tris-2-(3-methylindolyl)phosphine [17] with the corresponding chlorodiarylphosphine ( Figure 1).
The reaction of L1 H with [Ru(η 6 -benzene)Cl(µ-Cl)] 2 in refluxing THF for 16 h led to the formation of the octahedral complex RuCl 2 L1 H in quantitative yield ( Figure 2). Surprisingly, the other ligands did not react with this ruthenium precursor in refluxing THF. Even after 36 h, only trace amounts of the target complexes were observed using ex-situ 31 P NMR spectroscopy, and mainly signals of the unreacted ligand were present. Reactions in refluxing THF/toluene (1:3) mixture at 120 • C for a prolonged time of 64 h did lead to the formation of complexes RuCl 2 L1 CF3 , RuCl 2 L1 OMe , and RuCl 2 L2 H in >50% isolated yields.
All three L1-based RuCl 2 L1 complexes show similar 31 P NMR spectra: three signals in a ratio 1(td):2(t):1(td) (Figure 3). This splitting pattern indicates the formation of a symmetric complex. Ruthenium(II) complexes tend to form octahedral 18-valence electron complexes, and X-ray analysis and NMR spectroscopy confirmed that complex RuCl 2 L1 H indeed features an octahedral geometry with the two chlorides in mutual cis position and all four phosphines bound to the metal center [15]. The above factors and the similarity of the 31 P NMR spectra of the other L1-based complexes point to the same octahedral geometry around the metal center. The reaction of L1 H with [Ru(η 6 -benzene)Cl(μ-Cl)]2 in refluxing THF for 16 hours led to the formation of the octahedral complex RuCl2L1 H in quantitative yield ( Figure 2). Surprisingly, the other ligands did not react with this ruthenium precursor in refluxing THF. Even after 36 hours, only trace amounts of the target complexes were observed using ex-situ 31 P NMR spectroscopy, and mainly signals of the unreacted ligand were present. Reactions in refluxing THF/toluene (1:3) mixture at 120 °C for a prolonged time of 64 h did lead to the formation of complexes RuCl2L1 CF3 , RuCl2L1 OMe , and RuCl2L2 H in >50% isolated yields.
All three L1-based RuCl2L1 complexes show similar 31 P NMR spectra: three signals in a ratio 1(td):2(t):1(td) (Figure 3). This splitting pattern indicates the formation of a symmetric complex. Ruthenium(II) complexes tend to form octahedral 18-valence electron complexes, and X-ray analysis and NMR spectroscopy confirmed that complex RuCl2L1 H indeed features an octahedral geometry with the two chlorides in mutual cis position and all four phosphines bound to the metal center [15]. The above factors and the similarity of the 31 P NMR spectra of the other L1-based complexes point to the same octahedral geometry around the metal center. Layering of a dichloromethane solution of RuCl2L1 CF3 with pentane resulted in the formation of crystals suitable for single-crystal X-ray diffraction. The crystal structure (Figure 4) of RuCl2L1 CF3 indeed shows an overall octahedral geometry with the chlorido ligands in cis position, in accordance with the observed 31 (17) Å, respectively) compared with RuCl2L1 H (2.3189(9) Å, 2.3727(9) Å and 2.2671(9) Å). The -CF3 groups clearly point away from the metal center, so steric effects do not play a significant role in the coordination chemistry. The Cs symmetry observed in solution by 31 P NMR spectroscopy for both RuCl2L1 H and RuCl2L1 CF3 is not present in the solid state, as the unit cell contains two rotamers. This becomes evident when the symmetry-equivalent molecules in the crystal structure of RuCl2L1 CF3 are viewed from the bottom: the indolyl moieties either all point clockwise, or counter-clockwise ( Figure  5, showing the bottom view of RuCl2L1 CF3 and a schematic representation). In solution, these two rotamers seem to rapidly interchange on the NMR timescale, resulting in one signal for the two Layering of a dichloromethane solution of RuCl 2 L1 CF3 with pentane resulted in the formation of crystals suitable for single-crystal X-ray diffraction. The crystal structure (Figure 4) of RuCl 2 L1 CF3 indeed shows an overall octahedral geometry with the chlorido ligands in cis position, in accordance with the observed 31 (2.3189(9) Å, 2.3727(9) Å and 2.2671(9) Å). The -CF 3 groups clearly point away from the metal center, so steric effects do not play a significant role in the coordination chemistry. The C s symmetry observed in solution by 31 P NMR spectroscopy for both RuCl 2 L1 H and RuCl 2 L1 CF3 is not present in the solid state, as the unit cell contains two rotamers. This becomes evident when the symmetry-equivalent molecules in the crystal structure of RuCl 2 L1 CF3 are viewed from the bottom: the indolyl moieties either all point clockwise, or counter-clockwise ( Figure 5, showing the bottom view of RuCl 2 L1 CF3 and a schematic representation). In solution, these two rotamers seem to rapidly interchange on the NMR timescale, resulting in one signal for the two mutually trans phosphines. Likely, the methyl groups of the indolyl moiety do not create enough steric repulsion to suppress this process. Despite several attempts, no crystals suitable for X-ray diffraction of the RuCl 2 L1 OMe complex were obtained. mutually trans phosphines. Likely, the methyl groups of the indolyl moiety do not create enough steric repulsion to suppress this process. Despite several attempts, no crystals suitable for X-ray diffraction of the RuCl2L1 OMe complex were obtained.   mutually trans phosphines. Likely, the methyl groups of the indolyl moiety do not create enough steric repulsion to suppress this process. Despite several attempts, no crystals suitable for X-ray diffraction of the RuCl2L1 OMe complex were obtained.    [a] Taken from reference [16].
Interestingly, complex RuCl 2 L2 H , which is based on the linkage isomer of L1 H , ligand L2 H , displays a different 31 P NMR spectrum ( Figure 6). The four signals, two triplets of doublets and two doublets of doublets of doublets with an AX pattern, each integrating one phosphine, indicate the formation of a complex with a different symmetry. Crystals obtained by layering of a DCM solution of RuCl 2 L2 H with pentane at 5 • C were of sufficient quality for X-ray crystallography. The corresponding structure of complex RuCl 2 L2 H (Figure 7) helps to understand the origin of the AX pattern, as the bottom view displays much more steric hindrance around the central phosphorus atom P4 (vide infra).   (3) [a] Taken from reference [16].
Interestingly, complex RuCl2L2 H , which is based on the linkage isomer of L1 H , ligand L2 H , displays a different 31 P NMR spectrum ( Figure 6). The four signals, two triplets of doublets and two doublets of doublets of doublets with an AX pattern, each integrating one phosphine, indicate the formation of a complex with a different symmetry. Crystals obtained by layering of a DCM solution of RuCl2L2 H with pentane at 5 °C were of sufficient quality for X-ray crystallography. The corresponding structure of complex RuCl2L2 H (Figure 7) helps to understand the origin of the AX pattern, as the bottom view displays much more steric hindrance around the central phosphorus atom P4 (vide infra). However, different than the L1-based complexes, where the methyl groups are pointing towards the central phosphine ( Figure 5), in complex RuCl2L2 H , the bulkier phenyl rings of the indolyl moiety point towards the central phosphine ( Figure 8). From the crystal structure, it appears that these phenyl rings should encounter more hindrance toward inversion and it therefore seems likely that this process is associated with a high energy barrier in solution. As a result, the two phosphine atoms (P1 and P2, Figure 7) in trans position become inequivalent, leading to the observed AX pattern in the 31 P NMR spectrum as a result of the P1-P2 coupling. However, different than the L1-based complexes, where the methyl groups are pointing towards the central phosphine ( Figure 5), in complex RuCl 2 L2 H , the bulkier phenyl rings of the indolyl moiety point towards the central phosphine ( Figure 8). From the crystal structure, it appears that these phenyl rings should encounter more hindrance toward inversion and it therefore seems likely that this process is associated with a high energy barrier in solution. As a result, the two phosphine atoms (P1 and P2, Figure 7) in trans position become inequivalent, leading to the observed AX pattern in the 31 P NMR spectrum as a result of the P1-P2 coupling.  The crystal structure of RuCl2L2 H further reveals minor differences in the coordination around the ruthenium center compared with RuCl2L1 H ( Table 1). The P1-Ru, P2-Ru, and P3-Ru bond lengths  The crystal structure of RuCl2L2 H further reveals minor differences in the coordination around the ruthenium center compared with RuCl2L1 H ( Table 1). The P1-Ru, P2-Ru, and P3-Ru bond lengths The crystal structure of RuCl 2 L2 H further reveals minor differences in the coordination around the ruthenium center compared with RuCl 2 L1 H ( Table 1). The P1-Ru, P2-Ru, and P3-Ru bond lengths become slightly larger (2.3295(8) Å, 2.4079(8) and 2.2913(8) Å), but the P4-Ru bond (2.1363(8) Å) shortens, which is in line with a more π-acidic character of the tris(amido)phosphine P4 atom.
Two-electron reduction of RuCl 2 L1 H under an N 2 atmosphere led to the formation of the RuN 2 L1 H complex, which features a characteristic infrared band corresponding to the coordinated N 2 at 2125 cm −1 ( Table 2) [15]. To explore if the electronic properties of the ligands L1 CF3 , L1 OMe , and L2 H have an influence on the binding of N 2 to Ru 0 , the corresponding N 2 complexes were generated in situ by reacting the ruthenium(+II) complexes with two equivalents of KC 8 (Figure 9). The extent of activation of the N 2 bond was quantified by measuring the N≡N stretch frequency with infrared spectroscopy, as this is a direct measure for the ligand-dependent π-back-donation. Table 2 shows the N 2 stretch frequencies of the Ru 0 N 2 L complexes. shortens, which is in line with a more π-acidic character of the tris(amido)phosphine P4 atom. Two-electron reduction of RuCl2L1 H under an N2 atmosphere led to the formation of the RuN2L1 H complex, which features a characteristic infrared band corresponding to the coordinated N2 at 2125 cm −1 ( Table 2) [15]. To explore if the electronic properties of the ligands L1 CF3 , L1 OMe , and L2 H have an influence on the binding of N2 to Ru 0 , the corresponding N2 complexes were generated in situ by reacting the ruthenium(+II) complexes with two equivalents of KC8 ( Figure 9). The extent of activation of the N2 bond was quantified by measuring the N≡N stretch frequency with infrared spectroscopy, as this is a direct measure for the ligand-dependent π-back-donation. Table 2 shows the N2 stretch frequencies of the Ru 0 N2L complexes.  As expected, the complex coordinated with the less electron-rich, CF3 substituted ligand shows a band with a higher wavenumber in the infrared spectrum (2136 cm −1 ), as a result of weaker metalto-N2 π-back-donation. The complex coordinated to the more electron-rich OMe substituted ligand shows a band at 2113 cm −1 , thus revealing stronger π-back-donation to N2. Clearly, the peripheral phosphines induce measurable electronic effects on the d-orbitals of the ruthenium center; the phosphines cis to the N2 ligand become stronger σ-donors and weaker π-acceptors in L1 OMe and vice  As expected, the complex coordinated with the less electron-rich, CF 3 substituted ligand shows a band with a higher wavenumber in the infrared spectrum (2136 cm −1 ), as a result of weaker metal-to-N 2 π-back-donation. The complex coordinated to the more electron-rich OMe substituted ligand shows a band at 2113 cm −1 , thus revealing stronger π-back-donation to N 2 . Clearly, the peripheral phosphines induce measurable electronic effects on the d-orbitals of the ruthenium center; the phosphines cis to the N 2 ligand become stronger σ-donors and weaker π-acceptors in L1 OMe and vice versa for L1 CF3 , which changes the binding properties of the N 2 ligand. Going from RuN 2 L1 H (2125 cm −1 ) to the structural isomer RuN 2 L2 H (2136 cm −1 ), the N 2 stretch frequency increases by 11 cm −1 . This is likely a result of the more π-acidic pivotal P-atom (atom P4, Figure 10, top) of the tris(amido)phosphine donor ligand, which competes with the N 2 ligand for π-back-donation from the same metal d-orbitals. Thus, substitution of the phosphine trans to the N 2 ligand for a more π-acidic group leads to weaker π-back-donation of Ru to N 2 . These results show that the electronic properties of the dinitrogen complexes can be tuned by ligand design, either by manipulating the substitutions on the donor atoms coordinated cis to N 2 or by changing the π-acidity of the atom bound trans to N 2 .
Layering of a THF solution of RuN 2 L1 CF3 with pentane resulted in the formation of single crystals suitable for X-ray diffraction ( Figure 10). Similar to the RuN 2 L1 H complex reported previously [15], RuN 2 L1 CF3 also shows a trigonal bipyramidal geometry. The N≡N bond is slightly shorter in Table 1), which is in accordance with the observed trend in the activation of the N 2 stretch frequency ( Table 2). In solution, these complexes remain trigonal bipyramidal, as judged from the occurrence of a doublet and a quartet in all 31 P NMR spectra. versa for L1 CF3 , which changes the binding properties of the N2 ligand. Going from RuN2L1 H (2125 cm −1 ) to the structural isomer RuN2L2 H (2136 cm −1 ), the N2 stretch frequency increases by 11 cm −1 . This is likely a result of the more π-acidic pivotal P-atom (atom P4, Figure 10, top) of the tris(amido)phosphine donor ligand, which competes with the N2 ligand for π-back-donation from the same metal d-orbitals. Thus, substitution of the phosphine trans to the N2 ligand for a more πacidic group leads to weaker π-back-donation of Ru to N2. These results show that the electronic properties of the dinitrogen complexes can be tuned by ligand design, either by manipulating the substitutions on the donor atoms coordinated cis to N2 or by changing the π-acidity of the atom bound trans to N2.
Layering of a THF solution of RuN2L1 CF3 with pentane resulted in the formation of single crystals suitable for X-ray diffraction ( Figure 10). Similar to the RuN2L1 H complex reported previously [15], RuN2L1 CF3 also shows a trigonal bipyramidal geometry. The N≡N bond is slightly shorter in RuN2L1 CF3 (1.064(5)°) than in RuN2L1 H (1.085(5)°) ( Table 1), which is in accordance with the observed trend in the activation of the N2 stretch frequency ( Table 2). In solution, these complexes remain trigonal bipyramidal, as judged from the occurrence of a doublet and a quartet in all 31 P NMR spectra. The dinitrogen ligand in all of these ruthenium complexes is weakly coordinating as judged by the relatively small IR-shifts compared to that of free N2 (ν = 2359 cm −1 ) ( Table 2). Preliminary experiments show no dinitrogen reduction to ammonia when complex RuCl2L1 H was subjected to excess reductant (KC8) in the presence of acid ({H(OEt2)2}{BArF24}). Therefore, we set out to explore the reactivity of the tripodal tetraphosphine ruthenium(II) complexes for dinitrogen reduction with chlorosilanes ( Figure 11), which follows a different mechanism [24].  The dinitrogen ligand in all of these ruthenium complexes is weakly coordinating as judged by the relatively small IR-shifts compared to that of free N 2 (ν = 2359 cm −1 ) ( Table 2). Preliminary experiments show no dinitrogen reduction to ammonia when complex RuCl 2 L1 H was subjected to excess reductant (KC 8 ) in the presence of acid ({H(OEt 2 ) 2 }{BArF 24 }). Therefore, we set out to explore the reactivity of the tripodal tetraphosphine ruthenium(II) complexes for dinitrogen reduction with chlorosilanes ( Figure 11), which follows a different mechanism [24]. versa for L1 CF3 , which changes the binding properties of the N2 ligand. Going from RuN2L1 H (2125 cm −1 ) to the structural isomer RuN2L2 H (2136 cm −1 ), the N2 stretch frequency increases by 11 cm −1 . This is likely a result of the more π-acidic pivotal P-atom (atom P4, Figure 10, top) of the tris(amido)phosphine donor ligand, which competes with the N2 ligand for π-back-donation from the same metal d-orbitals. Thus, substitution of the phosphine trans to the N2 ligand for a more πacidic group leads to weaker π-back-donation of Ru to N2. These results show that the electronic properties of the dinitrogen complexes can be tuned by ligand design, either by manipulating the substitutions on the donor atoms coordinated cis to N2 or by changing the π-acidity of the atom bound trans to N2.
Layering of a THF solution of RuN2L1 CF3 with pentane resulted in the formation of single crystals suitable for X-ray diffraction ( Figure 10). Similar to the RuN2L1 H complex reported previously [15], RuN2L1 CF3 also shows a trigonal bipyramidal geometry. The N≡N bond is slightly shorter in RuN2L1 CF3 (1.064(5)°) than in RuN2L1 H (1.085(5)°) ( Table 1), which is in accordance with the observed trend in the activation of the N2 stretch frequency ( Table 2). In solution, these complexes remain trigonal bipyramidal, as judged from the occurrence of a doublet and a quartet in all 31 P NMR spectra. The dinitrogen ligand in all of these ruthenium complexes is weakly coordinating as judged by the relatively small IR-shifts compared to that of free N2 (ν = 2359 cm −1 ) ( Table 2). Preliminary experiments show no dinitrogen reduction to ammonia when complex RuCl2L1 H was subjected to excess reductant (KC8) in the presence of acid ({H(OEt2)2}{BArF24}). Therefore, we set out to explore the reactivity of the tripodal tetraphosphine ruthenium(II) complexes for dinitrogen reduction with chlorosilanes ( Figure 11), which follows a different mechanism [24].  A mixture of the RuCl 2 L complexes with 100 equiv of KC 8 and 100 equiv of chlorotrimethylsilane in 10 mL THF at room temperature under an atmosphere of dinitrogen was stirred for 1 day, whereafter an aliquot of the reaction mixture was analyzed by GC. Modest yields of~1.5 equivalents of (Me 3 Si) 3 N relative to Ru were detected, regardless of the complex used ( Table 3). The use of metallic sodium as reductant or longer reaction times did not increase the overall yield. In the absence of ruthenium only traces of (Me 3 Si) 3 N could be detected. To gain more insight on the N 2 activation using these novel ruthenium complexes, the reduction was followed in time for complexes RuCl 2 L1 H and RuCl 2 L2 H , analyzing aliquots of the reaction mixture every hour. The reaction profile shows an incubation period of 1 h before product formation is observed ( Figure 12). Such an incubation period was also reported for various other iron and cobalt systems that were investigated for N 2 reduction under comparable conditions [24][25][26], and which all showed similar activities for the formation of tris(trimethylsilyl)amine, regardless of the initial catalyst structure. It was speculated that the catalytically active species are generated after ligand dissociation from these iron and cobalt pre-catalysts. The exact nature of the reported catalysts could not be elucidated, as no dinitrogen-containing species were detected in these previous studies; however, test experiments pointed to a molecular nature of the catalyst [24,25].
Given that our complexes also show an incubation period and that electronic effects induced by ligand modification do not affect the catalytic activity, we speculate that our Ru-complexes are all pre-catalysts. To further test this hypothesis, we performed a catalytic reaction with [Ru(η 6 -benzene)Cl(µ-Cl)] 2 in the absence of any tripodal indolyl-phosphine ligand under the same reaction conditions. Comparable amounts (1.29 equiv relative to Ru) of silylamine were formed in this control experiment, which indeed suggests ligand dissociation to activate the catalyst. Thus, these ligands cannot prevent metal leaching under the strongly reducing conditions used. Ex-situ 31 P NMR spectroscopy of the RuCl 2 L1 H catalytic system showed the absence of the signals of the ruthenium complex pointing to decomposition of the pre-catalyst under the applied conditions. Next, we investigated the potential of these Ru-species in the catalytic dehydrogenation of formic acid (Figure 13), given our interest in the reversible storage of H 2 into liquid fuels [27][28][29][30][31][32][33]. For Ru-catalysts, formic acid dehydrogenation is accelerated when more electron-rich ligands are employed, as was shown by Himeda et al., who used a series of bipyridine ligands with various substituents at the para position (-OH, -OMe, -Me, -CO 2 H, and -H) [34]. With the various substituted tripodal tetraphosphine ligands L1 and L2 in hand, we anticipated to observe similar effects for the corresponding complexes in the formic acid dehydrogenation. The RuCl 2 L complexes are coordinatively saturated, which necessitates chloride abstraction using AgBF 4 to allow for substrate activation. Addition of 1.2 mmol of formic acid under nitrogen to a refluxing solution of a mixture of Ru-precatalyst (8.5 µmol) and AgBF 4 (17 µmol) in 3 mL THF pre-activated for 2 h (initial concentration of formic acid: 0.4 M, substrate to catalyst ratio: 140) led to gas production within 1-5 min. The reactions were monitored volumetrically with a gas burette for one hour (see experimental section for a schematic representation). A typical time profile of the dehydrogenation reaction with catalyst RuCl 2 L1 OMe is displayed in Figure 14. The overall results of the different catalysts used in the dehydrogenation of formic acid are presented in Table 4. The gas outlet was analyzed using gas chromatography, which confirmed presence of both CO 2 and H 2 , while no traces of CO were detected (detection limit of the GC for carbon monoxide was δ = 10 ppm). Under the described conditions, all the substrate is used within less than 2 h. Subsequent addition of aliquots of formic acid allowed for continuation of the reaction without an observed decrease in activity for at least 6 h. chromatography, which confirmed presence of both CO2 and H2, while no traces of CO were detected (detection limit of the GC for carbon monoxide was δ = 10 ppm). Under the described conditions, all the substrate is used within less than 2 h. Subsequent addition of aliquots of formic acid allowed for continuation of the reaction without an observed decrease in activity for at least 6 h.   chromatography, which confirmed presence of both CO2 and H2, while no traces of CO were detected (detection limit of the GC for carbon monoxide was δ = 10 ppm). Under the described conditions, all the substrate is used within less than 2 h. Subsequent addition of aliquots of formic acid allowed for continuation of the reaction without an observed decrease in activity for at least 6 h.   chromatography, which confirmed presence of both CO2 and H2, while no traces of CO were detected (detection limit of the GC for carbon monoxide was δ = 10 ppm). Under the described conditions, all the substrate is used within less than 2 h. Subsequent addition of aliquots of formic acid allowed for continuation of the reaction without an observed decrease in activity for at least 6 h.    Complexes RuCl 2 L1 OMe , RuCl 2 L1 H , and RuCl 2 L2 H follow the same trend as noticed by Himeda [34], with the most electron-rich complex RuCl 2 L1 OMe giving the highest activity (Turnover frequency (TOF) = 121 h −1 ) followed by RuCl 2 L1 H (TOF = 76 h −1 ) and the least electron-rich RuCl 2 L2 H (TOF = 33 h −1 ).

Tris-2-(3-methyl-N-di[4-(trifluoromethyl)phenyl]phosphinoindolyl)phosphine (L1 CF3 )
was prepared in the analogous way to L1 H using chlorobis [4-(trifluoromethyl)phenyl]phosphine: To a solution of tris-2-(3-methylindolyl)phosphine (1.9 g, 4.53 mmol) in THF (50 mL) was added n-BuLi (2.5 M in hexanes, 5.7 mL, 14.3 mmol) at −78 • C. The resulting solution was stirred for 1 h and chlorobis [4-(trifluoromethyl)phenyl]phosphine (3.5 mL, 14.0 mmol) was added. The reaction mixture was stirred for 16 h and allowed to warm slowly to room temperature. The resulting suspension was concentrated in vacuo and dissolved in CH 2 Cl 2 (total amount of 50 mL including washing of the pads). The suspension was filtered through a pad of basic alumina and subsequently through a pad of SiO 2 . The solvent was removed under reduced pressure and resulted in a yellow foam. This yellow foam was purified by SiO 2 chromatography using a gradient from pure hexane to 2% Et 2 O in hexane. Yield: 2.66 g (42%). 1 13 13 (3 mL) and toluene (6 mL) and stirred at 120 • C for 64 h. After cooling, 20 mL of hexane was added, which resulted in the precipitation of a yellow solid. The solid was filtered, washed with hexane (3 × 3 mL), and dried overnight in the vacuum oven at 60 • C. Yield: 477.5 mg (69.1%) of a yellow solid. Recrystallization of the complex by layering a DCM solution with pentane at 5 • C gave crystals suitable for X-ray diffraction analysis. 1   3.1. Reduction of RuCl 2 L to RuN 2 L RuCl 2 L (±17 µmol) and KC 8 (2-3 equiv) were transferred to a flame-dried Schlenk flask in the glovebox. THF (2 mL) was added and the suspension was stirred for two to three hours; after filtration, a red solution was obtained. Part of this solution was used for in-situ infrared spectrometry, part of the solution was used for in-situ 31 P NMR analysis, and the rest of the solution was set for crystallization via slow diffusion evaporation of pentane. Applying this procedure to RuCl 2 L1 OMe led to formation of a purple solution after filtration, which did not show an infrared band after two hours. However, after one day, the solution turned more red and the corresponding infrared band could be detected. The infrared analysis of the thus obtained solutions yielded the corresponding infrared bands as given in Table 2. The 31 P-NMR analysis showed the presence of one doublet and one quartet, indicative of a symmetric species, which corresponds with the proposed trigonal bipyramidal geometry.

Catalysis: Dinitrogen Reduction
The catalyst (33 µmol) and KC 8 (100 equiv) were transferred to a 100 mL flame-dried Schlenk flask equipped with a glass stirring bar in the glovebox. 10 mL of dry THF was added, followed by the addition of the chlorotrimethylsilane (100 equiv). The suspension was stirred overnight. A quantity of 10 µL of n-decane was added as internal standard, whereafter a filtered aliquot of the reaction mixture was measured on the Gas-GC.

Catalysis in Time: Dinitrogen Reduction
The same procedure as for the normal nitrogen reduction was used, adding the 10 µL of n-decane right away and taking a filtered aliquot of the reaction mixture every hour, which was subsequently measured on the Gas-GC.

Catalysis: Formic Acid Dehydrogenation
Hydrogen evolution was initiated by the addition of formic acid (45 µL, 1.2 mmol) to a solution of catalyst (±8.5 µmol) in refluxing THF (3 mL) in a 10 mL reaction flask equipped with a condenser.
The gas was cooled via a condenser, with the outlet connected to a burette filled with water ( Figure 15). The displacement of the water level in the burette was measured in time. The second burette was also filled with water and used to compensate for pressure buildup, keeping the water levels at equal height. The results are shown in Table 5. also filled with water and used to compensate for pressure buildup, keeping the water levels at equal height. The results are shown in Table 5.

X-ray Diffraction Analysis
Crystallographic data was obtained using a Bruker D8 Quest Eco diffractometer equipped with a Triumph monochromator. The intensities were integrated with the SAINT software package [41].

X-ray Diffraction Analysis
Crystallographic data was obtained using a Bruker D8 Quest Eco diffractometer equipped with a Triumph monochromator. The intensities were integrated with the SAINT software package [41]. Multiscan absorption correction and scaling was performed with SADABS [42]. The structure was solved with Intrinsic Phasing Methods using SHELXT [43]. Least-squares refinement was performed with SHELXL 2013 [44] against F 2 of all reflections. Non-hydrogen atoms were refined freely with anisotropic displacement parameters. All hydrogen atoms were located in difference Fourier maps and refined with a riding model. All structures have solvent-accessible voids filled with disordered solvent. Their contribution to the structure factors in the refinement was taken into account with the PLATON/SQUEEZE approach [45] (Table 6). RuCl 2 L1 CF3 : The structure contains voids (2544 Å 3 per unit cell) filled with disordered solvent molecules. Their contribution to the structure factors was secured by back-Fourier transformation using the SQUEEZE routine of the PLATON package [45], resulting in 671 electrons per unit cell which corresponds to 16 disordered molecules of CH 2 Cl 2 per unit cell. Alerts A and most Alerts B generated by the IUCr checkCIF program resulted from large displacement parameters of F atoms implying intense rotation of the CF 3 groups of the L1 CF3 ligand. Thirteen FCF reflections below θ-min were obscured by the beam-stop which generated Alert B.
RuCl 2 L2 H : The structure contains voids (1518 Å 3 per unit cell) filled with disordered solvent molecules. Their contribution to the structure factors was secured by back-Fourier transformation using the SQUEEZE routine of the PLATON package [45], resulting in 503 electrons per unit cell which corresponds to 12 disordered molecules of CH 2 Cl 2 per unit cell. Eleven FCF reflections below θ-min were obscured by the beam-stop which generated Alert B by the IUCr checkCIF program.
RuN 2 L1 CF3 : The structure contains voids (2224 Å 3 per unit cell) filled with disordered solvent molecules. Their contribution to the structure factors was secured by back-Fourier transformation using the SQUEEZE routine of the PLATON package [45], resulting in 514 electrons per unit cell which corresponds to 12 disordered molecules of pentane per unit cell. Alerts A generated by the IUCr checkCIF program resulted from large displacement parameters of F atoms implying intense rotation of the CF 3 groups of the L1 CF3 ligand. Thirteen FCF reflections below θ-min were obscured by the beam-stop which generated Alert B.

Conclusions
In conclusion, we have shown the formation of new ruthenium complexes based on a tripodal 3-methylindolephosphine scaffold. The ligand has two structural ("linkage") isomers, and, for one of those isomers, several analogues with electron-withdrawing and -donating groups were prepared. These ligands were successfully coordinated to [Ru(η 6 -benzene)Cl(µ-Cl)] 2 , resulting in the formation of octahedral complexes with the two chlorides in cis position. The stoichiometric reduction of these complexes with KC 8 yielded the corresponding dinitrogen complexes. Electronic effects in the ligands translated to a change in the N≡N stretch frequency of the coordinated dinitrogen. The complexes were studied in the N 2 reduction with chlorosilanes and KC 8 , yielding stoichiometric amounts of the silylamines. When the reaction was followed in time, an incubation period was observed. This incubation period, and the near-identical activities found for all complexes tested, suggests that these species are mere pre-catalysts. The complexes were also employed for formic acid dehydrogenation. Turnover frequencies between 33 and 124 h −1 were reached, depending on the ligand used, clearly showing that the electronic properties of the ligand do have an effect in this reaction. Further studies with these complexes have to be performed to elucidate the reaction mechanism and stability of the catalysts.