Unraveling the Origins of Catalyst Degradation in Non-heme Iron-based Alkane Oxidation

A series of potentially tetradentate and pentadentate ligands modelled on BPMEN has been prepared and their iron(II) bis(triflate) complexes have been isolated and characterised by spectroscopic and crystallographic techniques (BPMEN = N,N’bis(pyridylmethyl)ethylenediamine). Changes to the BPMEN ligand have invariably led to complexes with different coordination modes or geometries and with inferior catalytic efficiencies for the oxidation of cyclohexane with H2O2. The reaction of an iron(II) complex containing a pentadentate BPMEN-type ligand with O2 has resulted in ligand degradation via oxidative N-dealkylation and the isolation of a bis(hydroxo)-bridged dinuclear iron(III) complex with a picolinate-type ligand. Page 1 of 34 Dalton Transactions D al to n Tr an sa ct io ns A cc ep te d M an us cr ip t


Introduction
Catalyst deactivation is a common but often ignored problem in catalyst development, in particular in oxidation catalysis.Oxidative ligand degradation of a homogeneous molecular oxidation catalyst during its lifetime can be a limiting factor for high turnover and the activity of the catalyst is often directly related to its stability under the oxidising reaction conditions. 1 In living systems, ligand degradation of oxidation catalysts also occurs in enzymatic systems where both heme and nonheme oxidases have a limited lifetime, but are regenerated in vivo by complex mechanisms. 2,3 thorough understanding of the various oxidative ligand degradation processes will be essential for the design and development of more robust oxidation catalysts.5][6][7][8] While the exact deactivation pathway in the large variety of non-heme metal catalysts is not known at this stage, oxidative degradation of the ligand is probably one of the main causes for catalyst deactivation.Another possible deactivation pathway that has been invoked in a number of non-heme catalyst systems is the formation of inactive dinuclear μ-oxo iron(III) complexes. 9,10arious oxidative ligand degradation pathways have been observed in non-heme iron systems, for example peripheral alkyl and aryl C-H bond oxidations, [11][12][13] and more importantly for amine-based ligands, oxidative N-dealkylation.5][16] After initial oxidation of Fe(II) to Fe(III), the C-H bonds adjacent to an amine moiety (CHR-NR′ 2 ) (A in Scheme 1) are prone to oxidation resulting in the formation of hemi-aminal (C(OH)R-NR′ 2 ) complexes (B). 17Particularly vulnerable to oxidative degradation are methylene or methine protons adjacent to an amine that are also in alpha position to a carbonyl group, for example in amino acids, [18][19][20][21] or next to an aromatic unit such as phenyl (benzylic), [22][23][24][25] phenolate, 26 or pyridine moieties, as shown in Scheme 1. [27][28][29][30][31][32][33][34] Occasionally, the hemi-aminal products are stable enough and can be isolated as O-bound hemi-aminal complexes of type C. [35][36][37][38][39] Alternatively, hemi-aminals are prone to dissociation into an aldehyde (or ketone) and a secondary amine (D). 40In many cases where ligand degradation has occurred via oxidative N-dealkylation, degradation products such as aldehydes, [23][24][25][41][42][43][44] or ketones 30,45 have been isolated. A thid possibility is further oxidation of the hemi-aminal intermediate (B) to an amide complex (F). Hydroysis of this amide intermediate would result in the formation of picolinic acid from pyridylmethylamine moieties, which appears to be a common occurrence, sometimes resulting in the formation of stable metal picolinate complexes (G).42,44,46,47 The prevention of oxidative ligand degradation is an on-going challenge in the design of more efficient oxidation catalysts.In an attempt to prepare more robust biomimetic catalysts for the oxidation of alkanes, we have previously reported various derivatives of the benchmark catalyst [Fe(BPMEN)(OTf ) 2 ], [48][49][50] developed by Que and co-workers (BPMEN = N,N′-bis( pyridylmethyl)ethylenediamine).51 The selectivity and stability of [Fe(BPMEN)(CH 3 CN) 2 ](SbF 6 ) 2 is improved by adding electron donating para-methoxy groups to the pyridine moieties of the BPMEN ligand.50,52 Electron donating ligands are preferred because they can support highvalent iron oxo intermediates, which are generally assumed to be the active oxidant in these systems.
Our continuing efforts to understand and improve the stability of non-heme iron-based catalyst have led us to explore three aspects of the BPMEN ligand.Firstly, the effect of N-Me versus N-H substitution in alkane oxidation has been evaluated (ligand 1, Fig. 1).4][55] Secondly, methylene (CH 2 ) moieties adjacent to amine donors have been identified as vulnerable to oxidation. 17To avoid such methylene linkages, we have prepared tetradentate ligands with a C(Me)H or a CvO linkage between the pyridine and the amine donor (ligands 2-4).Replacing vulnerable C-H bonds with C-Me bonds has previously been used to improve the catalyst stability in other non-heme catalysts, 56 and a recent report on a related iron(II) catalyst with C(Ph)H linkages has shown promising results in asymmetric epoxidations. 57In a third series, we have explored the application of linear pentadentate ligands in oxidation catalysis (5-8).An additional donor could result in greater catalyst stability and within this series we have examined the effect of N-H (5) versus N-Me substitution (6), the removal of CH 2 linkages ( 7) and an alternative donor set (8) on the catalytic behaviour in cyclohexane oxidation.

Synthesis and characterisation of ligands and metal complexes
9][60][61][62] The N-methyl carboxamide ligands 3, 4, 7 and 8 were prepared by methylation of the carboxamides, which were obtained by reacting pyridine dicarboxylic acid or acid chloride with the corresponding amine (see Experimental section).N-methyl carboxamides show restricted rotation around the CO-NMe bond, which can lead to 3 rotamers and complicated NMR spectra. 63The ratio between the three different rotamers was determined as 4 : 3 : 3 in the case of ligand 3, as reported previously. 64For the new ligands 4, 7 and 8, the ratio was determined as 10 : 6 : 1 for ligand 4 (Fig. S1 †) and as 3 : 2 : 1 for ligand 8.Only one isomer was observed for ligand 7 (Fig. S2 †).

Dalton Transactions Paper
thiocyanate ligands undergoes 5 T 2 → 1 A 1 relaxation at a very low transition temperature of 70 K, 69 and later studies have revealed a very slow spin transition process in the solid state. 70ompared to [Fe(BPMEN)(CH 3 CN) 2 ] 2+ , the change in magnetic susceptibility of [Fe(1)(CH 3 CN) 2 ] 2+ is much more gradual with μ eff = 2.1μ B at 233 K and μ eff = 3.9μ B at 343 K (Fig. 2).This suggests that [Fe(1)(CH 3 CN) 2 ] 2+ undergoes geometrical changes with temperature, most likely to form other HS complexes with different geometries.Geometrical rearrangements of iron(II) complexes with tetradentate ligands similar to ligand 1 are quite common and we have seen isomerisation between cis-α and other geometries cis-β and trans, for example when the chelate rings size or donor strength has been changed. 49,50Here, the weaker basicity of the N-H versus N-Me donor and the resulting weaker Fe-N bond strength, results probably in a similar rearrangement process.For example the analogous complex with O instead of NH donors forms the trans complex exclusively. 502][73][74] The activation barriers (ΔH ‡ ) for spin transitions between HS and LS iron(II) complexes can vary in solution between 2 and 34 kJ mol −1 and may become competitive with geometric rearrangements. 75,76Coupling between the spin relaxation process and the geometrical rearrangement of the ligand, as seen in a related system, 77 cannot be excluded.
VT-19 F NMR analysis of complex [Fe(1)(OTf ) 2 ] in CD 2 Cl 2 shows a broad signal at −40 ppm at room temperature, which coalesces at 258 K and reveals multiple species in equilibrium, most likely cis-β and trans isomers in addition to the main cis-α complex (see Fig. S5 †).We conclude that the non-methylated complex [Fe(1)(OTf ) 2 ] must be conformationally less rigid compared to [Fe(BPMEN)(OTf ) 2 ], probably due to the weaker basicity of NH versus NMe donors.The ligand flexibility results in a combination of spin crossover and geometric isomerisation upon changing the temperature, which explains the anomalous magnetic behaviour seen in Fig. 2.
Ligand 2, with two chiral centres, forms a pair of enantiomers (R,R and S,S) and a meso form (R,S or S,R).Upon coordination to the iron(II) centre, two additional N-chirogenic centres are created at the amino nitrogen atoms.This could result in 16    The complexes [Fe(3)(OTf ) 2 ] and [Fe(4)(OTf ) 2 ] are high spin over the temperature range from 243-343 K, according to VT-1 H-NMR studies.The NMR spectra are rather complicated (see for example Fig. S10 †) and indicate the presence of more than one species in solution, possibly different isomers due to the different ligand rotamers.IR spectroscopy shows a change in ν(CvO) from 1636 and 1648 cm −1 for 3 and 4, respectively, to 1608 and 1610 cm −1 for the corresponding iron(II) complexes, indicating coordination of the carbonyl oxygen, which is not uncommon in picoline amide ligands. 27,78In the case of complex [Fe(4)(OTf ) 2 ], X-ray quality crystals were obtained from a THF solution and analysis revealed the formation of a dinuclear complex [Fe(4)(OTf ) 2 ] 2 with carbonyl oxygen coordination and the ligand binding in a bridging bis(bidentate) mode (see Fig. 4 and 5).The dinuclear complex has adopted a C 2 symmetric geometry about an axis that passes through the middle of the Fe 2 O 4 N 4 C 8 macrocycle.Each iron centre has a distorted octahedral geometry with cis angles in the range 75.19(5)-102.84(5)°,and is bound to two bidentate N,O donor ligands as well as two triflate groups.The only π-π stacking interaction of note is an intramolecular contact between the N(18) pyridyl ring and its C 2 related counterpart with centroid⋯centroid and mean interplanar separations of ca.3.67 and 3.61 Å, the two rings inclined by ca.12°(interaction a in Fig. 4).
We postulate a similar coordination for ligand 3 in complex [Fe(3)(OTf ) 2 ] 2 (see Fig. 5).Coordination via the carbonyl oxygen donor is clearly preferred over coordination by the weakly basic amide nitrogen donors.In acetonitrile, formation of acetonitrile-coordinated complexes occurs and the triflate anions are not coordinated to the metal centre in the temperature range from 233-343 K, as shown by a single peak between −70 and −80 ppm in the VT- 19 F NMR spectra (see Fig. S9 and S11 †).0][81] Coordination via the carbonyl oxygen donors was also seen in a palladium(II) complex containing ligand 3, 82 which is the only previously reported structure of a metal complex containing 3.
The six-coordinate octahedral complex [Fe(5)(OTf)](OTf) is the major species at low temperature, as indicated by two signals in the 19 F NMR spectrum, one for coordinated triflate at 35 ppm and one for non-coordinated triflate anions at −68 ppm at 203 K (Fig. 6).As the temperature is raised, the rate of exchange between coordinated and non-coordinated triflate anions will become faster, but there is also a shift in the equilibrium from a six-coordinate complex [Fe(5)(OTf)] + to a five-coordinate complex [Fe(5)] 2+ , as illustrated in eqn (2).Both complexes are high spin, as can be seen from the VT 1 H NMR spectra in CD 2 Cl 2 (Fig. S12 †).The two iron(II) complexes are similar to the previously reported six-and five-coordinate zinc(II) complexes with the same pentadentate ligand, [Zn(5)Cl] + and [Zn(5)] 2+ . 60n acetonitrile, the octahedral iron(II) complex [Fe( 5)-(CH 3 CN)] 2+ is formed with non-coordinating triflate anions (see VT 19 F NMR in Fig. S13 †).At low temperature this is a low spin complex, which shows SC behaviour as the temperature is increased (see Fig. 7).The HS [Fe(5)(CH 3 CN)] 2+ complex is more labile and the equilibrium will shift between the sixcoordinate complex [Fe(5)(CH 3 CN)] 2+ and a five-coordinate complex [Fe(5)] 2+ , resulting in an anomalous magnetic behav-iour similar to complex [Fe(1)(CH 3 CN) 2 ] 2+ , as shown by the magnetic moment measurements in Fig. 2(×).
We have previously reported the remarkable coordination behaviour of complex [Fe(6)(OTf ) 2 ], which is distinctly different from [Fe(5)(OTf )](OTf ).In non-coordinating solvents such as DCM, an equilibrium is observed between a sevencoordinate complex [Fe(6)(OTf ) 2 ] with pentagonal bipyramidal coordination and a five-coordinate trigonal bipyramidal complex [Fe( 6)](OTf ) 2 with two non-coordinating triflate anions (see eqn (3)). 62In acetonitrile, a seven-coordinate [Fe( 6)-(CH 3 CN) 2 ] 2+ complex is formed, which is in equilibrium with the five-coordinate complex [Fe(6)] 2+ .These seven-and fivecoordinate complexes are all high spin, as shown by their magnetic moment measurements in Fig. 2(◆), which appears to be the preferred spin state for this complex rather than forming an octahedral LS complex [Fe(6)(CH 3 CN)] 2+ .ð2Þ Iron(II) complexes of the potentially pentadentate ligands 7 and 8 are paramagnetic (see Fig. S14 and S16 †) and contain coordinated carbonyl donors according to IR analysis.The ν(CvO) signal of ligand 7 decreases from 1647 to 1603 cm −1 upon coordination.Similar changes were observed in a related nickel(II) complex of a pyridine dicarboxamide ligand with coordinated carbonyl oxygens. 83Attempts to react the nonmethylated precursor of ligand 7 with iron(II) bis(triflate) were unsuccessful, probably due to the poor solubility of this precursor.The 19 F NMR spectrum of [Fe(7)(OTf ) 2 ] in CD 3 CN indicates non-coordinating triflate anions (Fig. S15 †) and the UVvis spectrum in CH 3 CN shows a relatively weak MLCT absorption at 425 nm (ε = 260 M −1 cm −1 ).X-ray analysis of the iron(II) bis(triflate) complex of ligand 8 revealed a tridentate coordination of the ligand via the carbonyl oxygen donors with an additional THF ligand to complete the octahedral coordination in [Fe(8)(OTf ) 2 (thf )] (see Fig. 8).The iron centre has a severely distorted octahedral geometry with cis angles in the range 74.00( 5 S21; see the ESI †).Tridentate coordination of pyridine dicarboxamide ligands via the carbonyl oxygens appears to be the dominant coordination mode for nickel, copper and cobalt complexes, 84,85 and based on the available data on the complex with ligand 7, we postulate a similar coordination mode in this case.
Several conclusions can be drawn from these structural studies.Firstly, the BPMEN ligand appears to have a unique ability to provide a very rigid ligand framework with a cis-α coordination mode and a strong binding of the iron centre.Any flexibility in the ligand framework or weakening of the N-donor strength and a consequent weakening of the metalligand interaction (e.g.ligand 1) can lead to changes in the coordination behaviour (cis-β and trans isomers).Secondly, the addition of a pyridine donor in the pentadentate ligands 5 and 6 does not improve the strength of the metal-ligand interaction and disfavours the formation of octahedral complexes.Thirdly, if a pyridylmethylene amine unit is changed to a pyridyl carboxamide unit (ligands 3, 4, 7 and 8), the amide N-donor will be too weak to coordinate effectively and the complexes rearrange to a preferred coordination of another donor.

Catalytic oxidation of cyclohexane
The catalytic properties of the iron(II) bis(triflate) complexes containing ligands 1-8 for the oxidation of cyclohexane with H 2 O 2 have been evaluated (eqn (4)).

ð4Þ
The oxidation reactions were carried out under our standard conditions in order to compare the results with previously reported data (see ESI †). 48,49Hydrogen peroxide solution (10 equiv.or 100 equiv.)was added to an acetonitrile solution containing the catalyst (1 equiv.)and cyclohexane (1000 equiv.).A large excess of substrate was used to minimize over-oxidation of cyclohexanol (A) to cyclohexanone (K).The addition of dilute H 2 O 2 was carried out slowly using a syringe pump, in order to minimise H 2 O 2 decomposition.The yields are based on the amount of oxidant (H 2 O 2 ) converted into oxygenated products.All the individual catalytic runs were performed at least twice.
Catalytic experiments were carried out initially using 10 equiv. of H 2 O 2 .The amount of cyclohexanol (A) and cyclohexanone (K) are measured by GC and the percentage conversion of H 2 O 2 into oxidised products (A + K) for the different catalysts is shown in Table 1.The iron bis(triflate) complex [Fe(BPMEN)(OTf ) 2 ] is used as a benchmark against which the other catalysts are compared.We have previously reported that this catalyst, when using 10 equiv. of H 2 O 2 , converts 65% of the added H 2 O 2 into oxygenated products, with a large ratio of cyclohexanol to cyclohexanone (A/K ratio) of 9. 86 These results are consistent with those reported previously by Que and coworkers for the complex [Fe(BPMEN)(CH 3 CN) 2 ](ClO 4 ) 2 . 87The addition of more H 2 O 2 (100 equiv.)results in a lower percentage conversion and a lower A/K ratio.The results obtained with [Fe(OTf ) 2 (CH 3 CN) 2 ] have been added for comparison (run 8), which shows only low conversions and A/K ratios, indicative of Fenton-type behaviour.
The non-methylated complex [Fe(1)(OTf ) 2 ] shows a lower catalytic conversion and a smaller A/K ratio.This complex was also evaluated using 100 equiv.H 2 O 2 , showing comparable results to those reported by Shteinman an co-workers, who reported a ratio of A/K = 2 using 140 equiv.H 2 O 2 . 67The lower activity of this complex compared to the benchmark catalyst is tentatively ascribed to the ease of ligand degradation in the case of secondary amines.Complex [Fe(2)(OTf ) 2 ] gave the best conversion and A/K ratio within the series, although still lower compared to [Fe(BPMEN)(OTf ) 2 ].Complex [Fe(2)(OTf ) 2 ] exists as a mixture of isomers (see Fig. 3) that are likely to have different individual catalytic oxidation activity.Evaluation of the catalytic activity of the individual isomers would require the synthesis of enantiomerically pure ligands and complexes.The complexes [Fe(3)(OTf ) 2 ] and [Fe(4)(OTf ) 2 ] with H 2 O 2 showed no conversion under these conditions.The pentadentate complexes [Fe(5)(OTf ) 2 ] and [Fe(6)(OTf ) 2 ] showed a comparable but moderate conversion and A/K ratios, indicating that the two cis-labile sites are not essential for catalytic oxidation activity in these complexes.Relatively low conversions and A/K ratios were observed with complexes [Fe(7)(OTf ) 2 ] and [Fe(8)(OTf ) 2 ], where the pyridyl diamides are coordinated as tridentate ligands.Previous experiments using iron(II) complexes with tridentate ligands have shown similar results for the oxidation of cyclohexane. 88

Catalyst decomposition
Previous studies and the structural analysis carried out here for complexes [Fe(1)(OTf ) 2 ] and [Fe(2)(OTf ) 2 ] have shown that changes to the BPMEN ligand framework generally lead to a change in ligand flexibility, such that different coordination modes (cis-β and trans) become accessible.An increase in ligand flexibility results in complexes that show inferior catalytic activity in alkane oxidation. 48,49,86As a result of these studies, it has become increasingly clear that catalyst stability, under the harsh oxidising conditions required to oxidise alkanes, is a major factor that determines the catalytic efficiency of a given catalyst.One possible deactivation pathway that has been invoked in a number of related nonheme catalyst systems is the formation of inactive dinuclear μ-oxo iron(III) complexes. 9,104][95][96] The formation of dinuclear complexes may be minimized by steric congestion around the metal centre. 10,91,97,98Ligand rigidity, a strong ligand field and low chemical reactivity of the ligand appear to be critically important for the stability and lifetime of non-heme catalysts.

ð5Þ
When complex [Fe(6)(OTf ) 2 ] was exposed to air and moisture in acetonitrile, oxidative ligand degradation occurred and an isolable complex could be obtained, which was characterised by X-ray diffraction.A dinuclear μ 2 -(OH) 2 -iron(III) complex [Fe(6′)(OH)] 2 (OTf ) 2 was obtained (see eqn (5)  The pentadentate ligand 6 has undergone oxidative C-N cleavage to give a picolinate-type ligand 6′, presumable together with (N-methyl)-2-aminomethyl pyridine as the byproduct.The iron(II) centre has been oxidised, most likely by oxygen which in turn is reduced and in the presence of moisture results in the formation of hydroxide anions, as observed in complex [Fe(6′)(OH)] 2 (OTf ) 2 .The oxidation of pyridylmethylamine moieties appears to be a common occurrence, sometimes resulting in the formation of stable metal picolinate complexes. 42,44,46,47Ligand degradation reactions via oxidative N-dealkylation was recently observed during catalytic toluene oxidation with H 2 O 2 with a related iron complex featuring the BPMCN ligand, an analogue of BPMEN with a cyclohexyl backbone. 34he mechanism by which oxidative ligand degradation occurs in pyridylamine complexes such as [Fe(6)(OTf ) 2 ] with oxidants is not yet understood, but the results obtained here can be explained according to the general mechanism shown in Scheme 1. Oxidation of a methylene unit in ligand 6 results in a hemi-aminal complex (B), which can react further or rearrange to an O-bound hemi-aminal complex C. Further oxidation of the tertiary C-H bond in N-bound hemi-aminal complex B results in the formation of complex (E), which will dehydrate to an amide complex (F).As we have seen here for complexes [Fe(4)(OTf ) 2 ] and [Fe(8)(OTf ) 2 ], the weak basicity of amide nitrogen donors will likely result in a rearrangement due de-coordination of the nitrogen donor.Hydrolysis of the amide complex would give a picolinate complex G.In the case of complex [Fe(6)(OTf ) 2 ], this results in the picolinate ligand 6′ (see eqn (5)).6][37][38][39] Further oxidation and C-N cleavage would also result in the picolinate complex G.
In conclusion, ligand degradation in non-heme oxidation catalysts is an important factor that affects catalyst stability and lifetime under the oxidising reaction conditions.In order to investigate potential ligand degradation pathways and to improve catalyst stability, we have prepared a series of iron(II) complexes with tetradentate and pentadentate pyridylaminetype ligands 1-8.Compared to [Fe(BPMEN)(OTf ) 2 ], all complexes have shown lower activities as catalysts for the oxidation of cyclohexane with H 2 O 2 as the oxidant.The BPMEN ligand ensures a strong coordination environment with a cis-α geometry at the iron centre, stabilises intermediates in various oxidation states along the catalytic oxidation cycle and undergoes negligible oxidative ligand degradation.The NH donors in [Fe(1)(OTf ) 2 ] provide a weaker ligand field resulting in a conformationally less rigid complex with different geometrical high spin isomers (cis-β and trans in addition to cis-α).Secondary amines are also believed to be more vulnerable to oxidative degradation.A mixture of isomers was obtained in the case of [Fe(2)(OTf ) 2 ], probably with different catalytic activities.Amide donors in ligands 3 and 4 result in dinuclear oxygen-bound complexes with no catalytic activity.Pentadentate ligands in [Fe(5)(OTf ) 2 ] and [Fe(6)(OTf ) 2 ] provide moderate catalytic activity, despite the absence of two cis-labile sites, where ligands 7 and 8 were found to coordinate as tridentate ligands and showed only low catalytic activities.The reaction of complex [Fe(6)(OTf ) 2 ] with O 2 has shown that ligand degradation can occur via oxidative N-dealkylation, based on the isolation of an iron(III) complex [Fe(6′)(OH)] 2 (OTf ) 2 with a picolinate-type ligand 6′.We are continuing our efforts to develop robust non-heme iron-based catalysts for the selective oxidation of alkanes.
Stirring was continued at room temperature for 3 h.Water (200 mL) was added (initially dropwise to quench the NaH) and the aqueous layer was extracted with diethyl ether (200 mL) and then chloroform (200 mL).The organic extracts were combined and concentrated in vacuo, leaving a yellow DMF residue.Diethyl ether (200 mL) was added followed by pentane (200 mL) and the mixture was left standing at room temperature.After 3 days, yellow crystals had formed, which were collected by filtration and identified as the product by 1 H-NMR spectroscopy.Yield: 1.57 g (74%); 1 H-NMR (400 MHz, 14.7 mmol) were dissolved in abs.toluene.The second solution was added to the first and the reaction mixture stirred at 80 °C overnight.After cooling the mixture to room temperature, and removal of all volatiles, the residue was taken up in chloroform and saturated aqueous sodium hydrogencarbonate.The phases were separated and the aqueous phase was extracted three times with chloroform.The organic phase was dried over MgSO 4 , filtered and the solvent removed under reduced pressure.The crude product was recrystallised from chloroform yielding the product as a white solid (2.7 g, 85%).(b) 2,6-Bis[(methylanthranilate)carboxamide]pyridine (700.0 mg, 1.5 mmol) was dissolved in abs.THF and NaH (110.0 mg, 4.6 mmol) were added.After stirring the suspension for 4 h, MeI (0.3 mL, 4.8 mmol) was added and the suspension was stirred at room temperature overnight.After removal of all volatiles at reduced pressure, the residue was taken up in chloroform and saturated aqueous sodium hydrogencarbonate.After the phases were separated the aqueous phase was extracted three times with chloroform.The organic phase was dried over MgSO 4 , filtered and the solvent removed under reduced pressure.The product 8 was obtained as viscous yellow oil (430 mg, 62%). 1 H NMR (CDCl 3 , 400 MHz) (due to the presence of three rotamers in a 3 : 2 : 1 ratio, a large number of signals appear at room temperature): δ ( ppm) = 7.97 (d, J = 8.2 Hz, 0.4H), 7.85-7.64(m, 2.7H), 7.60 (t, J = 8.0 Hz, 0.6H), 7.51 (d, J = 7.8 Hz, 0.4H), 7.44-7.17(m, 5.8H), 7.12 (d, J = 7.8 Hz, 0.4H), 6.84 (d, J = 8.0 Hz, 0.4H), 6.61 (d, J = 8.0 Hz, 0.2H), 6.52 (t, J = 8.0 Hz, 0.2H), 3.79 (m, 6H), 3.44 (m, 1.5H), 3.29 (m, 3H), 2.89 (m, 1.5H). 13 General synthesis of metal triflate complexes.To prepare [Fe(L)(OTf ) 2 ], the relevant ligand L and 1.0 molar equivalent of [Fe(NCMe) 2 (OTf ) 2 ] were placed in different Schlenk flasks and dissolved in dry tetrahydrofuran under nitrogen.After adding the solution of the ligand to the suspension of the metal precursor, the reaction mixture was stirred overnight at room temperature.The resulting solution was concentrated to one third of the initial volume.Diethyl ether was added to precipitate the product as solid, which was dried under vacuum.
isomers, but the cis-α coordination mode only supports the [(N)S*,(N)S*] configuration at the central amine donors.This reduces the number of possible isomers to 8 (4 pairs of enantiomers).The four possible isomers with [(N)S,(N)S] configuration at the central amines are shown in Fig. 3.The two meso isomers C are identical, which reduces the number of complexes to three diastereomeric complexes A, B and C in a 1 : 1 : 2 ratio.The presence of diastereomers complicates the 1 H NMR spectrum of [Fe(2)(OTf ) 2 ] in CD 3 CN.At 228 K, the spectrum appears to consist of a mixture of two complexes (or groups of
complexes), which we presume are A and B on the one hand, and C on the other (see Fig.S6 †).One complex is diamagnetic at this temperature, with a chemical shift range between 0 and 10 ppm and the other is partially high spin (HS) with a shift range from −5 to 40 ppm.The latter complex is fully HS at 278 K and follows Curie behaviour upon further temperature increase.The spin crossover temperature for this complex is at T c ≈ 200 K.The diamagnetic complex shows 9 signals at 228 K and as the temperature is raised, undergoes spin crossover at T c ≈ 235 K, whereby the chemical shift range increases and the signals become gradually broader.At 288 K, the signals for this complex are extremely broad, due to exchange between coordinated and non-coordinated CD 3 CN ligands.At the highest temperature of 338 K, the exchange is sufficiently fast for the signals to sharpen again.From 298 K onwards, all complexes are paramagnetic and follow Curie behaviour with further temperature increase.VT19  F NMR spectroscopy (Fig.S8 †) and magnetic susceptibility measurements (see ( • ) in Fig.2) also indicate a different behaviour compared to [Fe(BPMEN)(OTf ) 2 ], indicative of multiple complexes, each with a different spin crossover behaviour and T c values.
)-117.74(16)°, and the metal atom lies ca.0.74 Å out of the N(1)-pyridyl ring plane.Noteworthy intermolecular interactions are a pair of F⋯π contacts across independent centres of symmetry.The F(51) triflate fluorine atom in one molecule approaches the N(1) pyridyl ring in a C i -related counterpart (F⋯π 3.59 Å), whilst F(41) approaches the N(8)bound aryl ring across a different centre of symmetry (F⋯π 3.51 Å), resulting in an extended chain of molecules (interactions a and b respectively in Fig.
Fig. 9) with a centre of symmetry in the middle of the Fe 2 O 2 ring.The unique iron centre has a distorted octahedral geometry with cis angles in the range 73.70(8)-109.25(8)°, and is bound to one tetradentate N,N′,N″,O donor ligand as well as to two bridging hydroxo ligands.The Fe-O-Fe bridges are symmetric [Fe(1)-O(20) 1.9664(17), Fe(1A)-O(20) 1.9700(17)] and subtend an angle of 103.40(8)°at the oxygen.The triflate anions sit in the clefts formed by the pyridyl rings on each iron centre with O⋯centroid separations of ca.3.07 and 3.68 Å (interactions a and b respectively in Fig. S23 †).

Table 1
Oxidation of cyclohexane with H 2 O 2 catalysed by [Fe(L)(OTf ) 2 ] a Total percentage yield of cyclohexanol (A) + cyclohexanone (K), expressed as moles of product per mole of H 2 O 2 .c Ratio of moles of cyclohexanol (A) to moles of cyclohexanone (K).d Data taken from ref. 49. e Data taken from ref. 89.