Octahedral Iron‐Acyl‐Nitrenoid Intermediates in Sulphur −Nitrogen Coupling and Hydrogen Atom Transfer Reactions

Iron‐nitrenoid complexes can activate C−H bonds and, thus, potentially directly transform C−H into C−N bonds. Low‐coordination complexes are the main reported systems; however, octahedral iron complexes can be air‐stable alternatives forming the iron‐nitrenoid species. Here, a novel octahedral non‐heme iron‐acyl‐nitrenoid complex is presented. Its reactivity was studied using a flow chemistry setup with online mass spectrometry detection, which allows qualitative reaction kinetics evaluation and the detection of the intermediates. The iron‐nitrenoid complex reacts with thiophenol by H‐Atom transfer, followed by a rebound to form a new N−S bond. The rebound intermediate could be detected and distinguished from the product complex. The reaction of the iron‐acyl‐nitrenoid complex with hydrocarbons results in the activation of the C−H bond via hydrogen‐atom transfer with no subsequent rebound process.


Introduction
High-valent iron-oxo complexes are powerful oxidants of organic substrates. Over the last decades, metal-complex optimization by ligand tuning led to catalysts with striking oxidizing reactivities and selectivities. [1,2] In analogy, isoelectronic iron-nitrenoid complexes can also abstract an H-atom, opening a path to imido transfer reactions. However, the reactivity and characterization of the iron-nitrenoid complexes have been underexplored. Up to now, most current synthetic and mechanistic studies have relied on low-coordination complexes. [3][4][5][6][7][8][9] Scarce studies comprising octahedral complexes, widely used for iron-oxo chemistry, show that the reactivity of the iron-nitrenoid complexes slightly differs from that of their oxo analogs. [10][11][12][13][14][15] Ligand design is a crucial tool to improve the catalytic activity of metal complexes. However, for iron-nitrenoid complexes, the nature of the imido moiety can also affect the formation of the catalytic species and their speciation as monomers or dimers; and, ultimately, modify their reactivity. [16] An efficient and extensively used imido precursor is Ntosyliminobenzyliodinane (PhINTs), but its use limits the ability to modify the imido moiety. [12,13,17] More versatile precursors are organic azides, [18][19][20][21][22][23][24] hydroxylamine derivatives [25,26] and dioxazolones, which have recently become popular thanks to their chemical stability and ability to generate metal-acyl-nitrenoid complexes. [27] For years, these species were elusive because they dominantly undergo the Curtius rearrangement, giving isocyanates as decomposed products. [27][28][29][30] However, an appropriate ligand design of the metal complexes allowed the stabilization of the acyl-nitrenoid intermediates, favouring the formation of amide organic products. Currently reported ironacyl-nitrenoid complexes are based on heme systems or were generated by photoirradiation with visible light. [31][32][33][34] Here, we study the reactivity of iron-acyl-nitrenoid complexes with the TPA ligand (tris-(2-pyridylmethyl)amine). TPA ligands have been frequently used to stabilize and study ironoxo species. [35][36][37] We apply mass spectrometry techniques to track the formation of reactive complexes and their reactions with substrates. Mass spectrometry is a powerful method to detect elusive reaction intermediates that escape detection by other techniques due to their high reactivities and short lifetimes. [38]

Results and Discussion
As demonstrated previously, an effective way to detect and study reactive metal complexes is to generate them in a flow reactor with a direct connection to a mass spectrometer with an electrospray ionization interphase (ESI-MS). [39][40][41] Reaction of (TPA)Fe(OTf) 2 with phenyl dioxazolone (A) in a flow reactor (Figure 1a Adding D 2 O into the flow reactor promotes a partial H/D exchange of one hydrogen atom for all detected nitrenoid complexes ( Figure S5). The H/D exchange points to the presence of a labile hydrogen atom in these species. Therefore, the detected ions partially also correspond to the complexes with a self-functionalized ligand with the À NHCOPh function resulting from the internal CÀ H activation followed by the rebound of the C-radical and the amido radical. Analogous selffunctionalization has also been observed for other nitrenoid systems. [16,42] We have further characterized the isolated ions corresponding to [(TPA)Fe(NCOPh)(MeCN)] 2 + (m/z 253) using helium tagging photodissociation spectroscopy ( Figure S17). Comparing the experimental spectrum with theoretical spectra of different isomers of the nitrenoid complex and possible products of the internal CÀ N and CÀ O couplings suggests that the nitrenoid complexes do not withstand the transfer to the gas phase. The detected ions report on the nitrenoid complexes in solution, as evidenced by the incomplete H/D exchange reaction (and the reaction kinetics below), but the desolvation promotes the intramolecular reactions. Interestingly, the experimental spectrum suggests that the major gas-phase product corresponds to the H-atom abstraction from one of the pyridine rings, followed by the CÀ O coupling ( Figure S18). This reaction would not be observed in solution but is likely kinetically favored in the gas phase. Collision-induced dissociation of the detected complexes leads to the elimination of acetonitrile, followed by the loss of PhCONH 2 . We did not observe any elimination of PhNCO, which attests that the Curtius rearrangement proceeds via a larger energy barrier than internal hydrogen atom transfer reactions ( Figure S1A, B).
The formation of an iron-acyl-nitrenoid species could be additionally supported by UV-Vis spectroscopy. The addition of phenyl dioxazolones (A) to an acetonitrile solution of (TPA)Fe-(OTf) 2 at 0°C resulted in the formation of a new band at around 730 nm ( Figure S16). The newly formed band agrees with the previously reported band of two other reported octahedral iron(IV) complexes (660 nm and 750 nm). [12,13] Further spectroscopic characterization would be necessary to assign the + IV oxidation state due to the redox non-innocent character of the nitrenoid moiety. [43] Besides the monomeric iron-acyl-nitrenoid complexes, we also detected a dimeric complex formally corresponding to [((TPA)Fe) 2 (NCOPh)(O) 2 (H)] 3 + (m/z = 281.40, denoted with a star in Figure 1b, top). Based on its fragmentation pattern, we can speculate that the iron centers are bridged with an oxygen atom and either phenylhydroxamate (PhC(O)NHŌ) formed by oxidation of the nitrenoid ligand or phenyl carbamate (PhNHCOŌ) formed by the Curtius rearrangement ( Figure S1C).
The ability of self-functionalization of the iron-acyl-nitrenoid species suggests a high reactivity of the intermediate. In comparison, [(TPA)Fe(O)(MeCN)] 2 + complexes are stable and do not undergo self-oxidation under the same reaction conditions. [39] Therefore, we tested its reactivity towards organic substrates. We first studied the reaction of the transient nitrenoid complexes with thiophenol ( Figure 1  With the flow setup, we can track the kinetics of these transformations by changing the molar excess of thiophenol (Figure 1c). With an increasing excess of the substrate, the ironacyl-nitrenoid complex progressively depletes while the initial iron(II) complex is regenerated. At the same time, the trends in the formation of the two newly formed species, ([(TPA)Fe-(PhSNHCOPh)] 2 + (4) and [(TPA)Fe(NH 2 COPh)] 2 + ) (3), match with the conversion of the iron-nitrenoid complex.
The maximum conversion with the given reaction time (4 s, see the scheme in Figure 1a) is reached with about 50 equiv. of PhSH. We also detect protonated functionalized ligand ([TPA-NHCOPh]H + (B)), suggesting that the iron-acyl-nitrenoid complex promotes self-functionalization. The abundance of the selffunctionalized ligand depletes with the increasing molar excess of thiophenol ( Figure S6). Hence, the bimolecular reaction of iron-acyl-nitrenoid intermediate with the substrate is faster than the self-functionalization, which cannot compete at larger substrate concentrations.
Detecting the [(TPA)Fe(PhSNHCOPh)] 2 + (4) and [(TPA)Fe-(NH 2 COPh)] 2 + (3) ions suggests that the iron-acyl-nitrenoid complex abstracts a hydrogen atom from thiophenol. We have been unable to detect the expected single HAT intermediate [(TPA)Fe III (NHCOPh)(X)] 2 + / + (X = À , MeCN, TfO À ); nevertheless, we could detect the [(TPA)Fe(NH 2 COPh)] 2 + amide, which results either from a protodemetallation step followed by coordination of the benzamide product; or after an additional HAT process, proposed to be feasible in a previously described system. [16] The    Figure S10-S12). The setup is identical, as depicted in Figure 1. The rise of the relative abundances is proportional to the rate constants (k multiplied by an unknown ionization efficiency factor, identical for the isotopologs).
formation of the ions. The ratio of the trends provides a kinetic isotope effect of~15 (Figure 3 and Figure S12).
This result is consistent with the H-atom abstraction process being the rate-determining step. We can compare this estimated value with previously reported KIE values of other high-valent iron complexes. The KIE value for iron-oxo species supported by the TPA ligand in the oxidation of benzyl alcohol is 58 at À 40°C. [44] The KIEs values of previously reported ironnitrenoid species supported by different ligands range from 4 to 13, thus consistent with our estimated value. [12,13,45,46] The lower KIE value has been attributed to a large amount of charge transfer during the H-atom abstraction process. [12] Next, we tested the reactivity of the iron-acyl-nitrenoid complex with cyclohexene. With the increasing molar excess of cyclohexene, the iron-acyl-nitrenoid abundance decreases. However, adduct complexes suggesting a similar rebound mechanism as described for the thiol were not observed. Instead, the decreasing abundance of the iron-acyl-nitrenoid complex correlates with the rise of iron(III) species ( Figure S15). This behavior is also typical for iron(IV)oxo complexes. [47] The results are consistent with the HAT reaction between the complex and hydrocarbon, followed by the iron(III) complex formation. The so-formed organic radical is oxidized by dissolved oxygen in the solution (Figure 4). In agreement, we can also detect adducts of iron complexes with oxidized cyclohexene. A control batch reaction showed the formation of 26 % of 2-cyclohexen-1-one and 15 % of 2-cyclohexen-1-ol, with no detection of amination or aziridination product ( Figure S14). Working under argon did not help in promoting the rebound mechanism. Instead, a lower conversion of phenyl dioxazolone (A) and traces of oxidized compounds were observed. Most likely, the generated radicals caused a faster degradation of the iron complex. These results prove that the present iron-acylnitrenoid species can promote HAT but do not rebound to the radical carbon center. Hence, the reactivity with hydrocarbons is analogous to that of Fe(IV)-oxo species. [37]

Conclusions
We present a new approach to form non-heme iron(IV)-acylnitrenoid species using dioxazolones as stable precursors. Using a flow reactor-mass spectrometry coupling, we could track the formation of the TPA-supported iron-acyl-nitrenoid intermediates and their reactivities towards organic substrates. We demonstrate that the iron-acyl-nitrenoid complexes react via the H-atom abstraction followed by the rebound process with thiophenol yielding the SÀ N product in a catalytic manner. We have detected the key rebound intermediates and could monitor the reductive elimination of the SÀ N coupling products. The reactivity of the iron-acyl-nitrenoid complexes with hydrocarbons is analogous to that of iron(IV)-oxo, showing no rebound products.

Experimental Section
All solvents and reagents were purchased from Acros, Alfa Aesar, Thermo Fischer, or Sigma Aldrich and were used without further purification. ([TPA)Fe(OTf) 2 ], [48] phenyl dioxazolone, [32] and thiophenol-d1 [49] were synthesized according to the previous reports. 1 H-NMR spectra were recorded at room temperature on a Bruker Avance III 400 MHz and referenced to the residual solvent peaks.
UV-Vis spectra were recorded on a JASCO V-630 UV-Vis spectrophotometer equipped with a thermostated cell holder at 0°C.
Electrospray ionization mass spectrometry (ESI-MS) experiments were performed in a trapped ion mobility-quadrupole time-of-flight mass spectrometer (timsTOF pro, Bruker Daltonics, Bremen, Germany). Ions were generated by electrospray ionization with the following settings: capillary voltage 5.5 kV, dry gas 3.0 L/min, dry temperature 200°C, and nebulizer gas 0.3 bar. TIMS experiments were performed in N 2 using imeX Detect mode by scanning ion mobility from 0.2 to 1.25 V.s.cm À 2 .
The flow chemistry setup was from the Labm8 company tailored for the coupling with ESI-MS as described previously. 40 For more details, see the Supporting Information.