Over or under: Hydride attack at the metal versus the coordinated nitrosyl ligand in ferric nitrosyl porphyrins

Abstract

Hydride attack at a ferric heme–NO to give an Fe–HNO intermediate is a key step in the global N-cycle. We demonstrate differential reactivity when six- and five-coordinate ferric heme-NO models react with hydride. Although Fe–HNO formation is thermodynamically favored from this reaction, Fe–H formation is kinetically favored for the 5C case.

Introduction

Nitroxyl (HNO) is gaining attention as a significant player in the overall biology of nitric oxide (NO) due to its involvement as a reactive intermediate in the global N-cycle.^1-4 HNO can be generated by metal-mediated and organic processes.^{5, 6} Heme–HNO species are present in the reaction cycles of cyt c nitrite reductases (via proton attack on a heme-bound NO),⁷ and in fungal cyt P450 NO reductases (via hydride attack on a ferric heme–NO) en route to N₂O formation.⁸ Very little experimental information on heme model-HNO compounds is available.^{1-4, 9-12} Farmer and coworkers have reported the spectroscopic characterization of several heme protein-HNO adducts.¹³ Coordination non-heme compounds with HNO ligands have been reviewed.^{1, 2, 14} Related anionic (i.e., non-protonated) synthetic porphyrin-NO compounds of the form [(por)Fe(NO)]^{− 9, 10, 12, 15} have been characterized, and two such species, namely [(TFTTBr₈)Fe(NO)]⁻ (TFTTBr₈ = octabromo[tetrakis(pentafluorophenyl)porphyrinato dianion)¹⁶ and [(OEP)Fe(NO)]⁻ (OEP = octaethylpoprhyrinato dianion)¹⁷ have been structurally characterized by X-ray crystallography. Although density functional theory (DFT) calculations have aided significantly in our theoretical understanding of heme–HNO compounds, ^{3, 7, 18-20} the general lack of appropriate experimental heme–HNO models has hindered research in this important area.

We recently reported that hydride attack at the coordinated NO group in a ferric [(OEP)Fe(NO)(5-MeIm)]⁺ compound generates the Fe–HNO derivative (eq 1),²¹ modeling a key step in cyt P450nor catalysis. Of prime importance is an examination of factors that lead to successful hydride attack at

$${\left[\left(\text{OEP}\right)\text{Fe}\left(\text{NO}\right)\left(5-\text{Melm}\right)\right]}^{+}+{\mathrm{H}}^{-}\to \left(\text{OEP}\right)\text{Fe}\left(\text{HNO}\right)\left(5-\text{Melm}\right)$$ (1)

the coordinated NO of ferric–NO hemes. The availability of both six-coordinate and five-coordinate ferric–NO heme models has allowed us to investigate hydride attack at the coordinated NO and/or the Fe center experimentally and computationally (e.g., the 5C case shown in eq 2).

Results and discussion

Following up on our earlier report, we have determined that the attack of hydride at the coordinated NO groups of six-coordinate cationic ferric-NO heme models can be quite general. In this work, we prepared several [(por)Fe(NO)(L)]+ compounds (por = dianions of protoporphyrin IX dimethyl ester (PPDME), OEP, and TTP (tetratolylporphyrin)). The crystal structure of [(PPDME)Fe(NO)(5-MeIm)]SbF₆, an excellent model for histidine-ligated (PPIX)Fe, is shown in Fig. 1.^‡

Fig. 1.

Molecular structure of the cation of [(PPDME)Fe(NO)(5-MeIm)]SbF₆ in the $P\stackrel{‒}{1}$ space group (with thermal ellipsoids drawn at 50%). The H atoms (except for the imidazole N6 proton) and the anion have been omitted for clarity. Fe1–N7 = 1.654(5) Å, N7–O5 = 1.133(6) Å, Fe1–N5 = 1.990(5), ∠Fe1–N7–O5 = 174.8(5)°.

The reactions of these cations with borohydride (as the source of hydride) are best carried out at low temperature (−50 to −20 °C), as the products decompose readily at higher temperatures. For example, with the reaction [(PPDME)Fe(NO)(1-MeIm)]⁺ cation (υ_NO 1915 cm⁻¹) with borohydride generates the (PPDME)Fe(HNO)(1-MeIm) product characterized by a new band in its IR spectrum at 1384 cm⁻¹ assigned to the υ_NO of the newly-formed Fe–HNO group (Fig. 2). This Fe–HNO decomposes even at this temperature to the paramagnetic (PPDME)Fe(NO) species (υ_NO 1672 cm⁻¹) (see later).

Fig. 2.

IR spectroscopic characterization of the bound HNO ligand in (PPDME)Fe(HNO)(1-MeIm), showing formation of the υ_NO 1384 cm⁻¹ band (dashed line) upon hydride addition to the cationic precursor (υ_NO 1915 cm⁻¹). The new 1384 cm⁻¹ band slowly converts to the band at 1672 cm⁻¹.

The generated (PPDME)Fe(HNO)(1-MeIm) product is also characterized by a singlet peak at 13.64 ppm in the ¹H NMR spectrum that splits into a doublet (J_NH 77 Hz) with ¹⁵NO labeling (Fig. 3). The ¹H NMR spectra from the related reaction to generate the (PPDME)Fe(HNO)(5-MeIm) analogue are shown in Fig. S1a. The low-temperature IR and ¹H NMR spectral data for these and other Fe–HNO compounds prepared in this work are collected in Table 1.

Fig. 3.

¹H NMR spectra (in CDCl₃) of the reaction of [(PPDME)Fe(NO)(1-MeIm)]OTf with [NBu₄]BH₄ to generate (PPDME)Fe(HNO)(1-MeIm). The bottom spectrum showing the splitting of the Fe-HNO peak at 13.64 ppm into a doublet was obtained using [(PPDME)Fe(¹⁵NO)(1-MeIm)]OTf in the reaction. The peak labeled * is due to the H₂ decomposition product (see text and Fig. S2).

Table 1.

Spectral data for the precursor [(por)Fe(NO)(L)]⁺ and (por)Fe(HNO)(L) complexes.^a

		[(por)Fe(NO)L]+	(por)Fe(HNO)(L)
por	L	IR (cm−1)	IR (cm−1)	1H NMR, ppmb
OEP	ImH	1911	1381	13.93 (78)
	5-MeIm	1910	1383	13.99 (76)
	1-MeIm	1912	1388	13.72 (77)
PPDME	ImH	1915	1382	13.90 (76)
	5-MeIm	1912	1382	13.93 (77)
	1-MeIm	1915	1384	13.64 (77)
TTP	ImH	1917	1386	14.20 (76)
	5-MeIm	1912	1381	14.26 (76)
	1-MeIm	1914	1389	14.02 (76)

^aIR data in CHCl₃ (at −45 °C), and ¹H NMR data in CDCl₃ (at −20 °C).

^bThe J_N-H coupling constants (in Hz) for the Fe(H¹⁵NO) derivatives are in brackets.

The spectral data for the Fe–HNO complexes are reproduced well by DFT calculations using both the pyrrole-substituted (OEP) and meso-substituted (TTP) porphyrins, and an N-substituted imidazole (1-MeIm) and the histidine mimic (5-MeIm). Our experimental observations of a mild effect of OEP vs. TTP macrocycle and axial ligand (5-MeIm vs. 1-MeIm) type on ¹H NMR chemical shifts (range of 0.54 ppm; ^~4%), but essentially negligible (<1%) υ_NO shifts for these systems, are reproduced by the calculations (Table S1); with a range of proton shifts of 0.69 ppm (^~5%) and a range of υ_NO shifts of <1%, affirming that ¹H NMR spectroscopy is a more sensitive structural probe for these Fe–HNO systems.

In contrast, the reaction of borohydride with the five-coordinate [(OEP)Fe(NO)]OTf (Fig. 4)^‡ at −50 °C does not result in new ¹⁵N isotope sensitive peaks in the 11–15 ppm region of the ¹H NMR spectrum attributable to an Fe–HNO derivative. Rather, a new sharp peak at −4.11 ppm is observed that we attribute to the six-coordinate Fe-hydride product (OEP)Fe(NO)H (Fig. 5a). Importantly, the DFT-calculated ¹H NMR chemical shift of the hydride peak for (OEP)Fe(NO)H is at −3.64 ppm, which is in good agreement with the experimentally observed value of −4.11 ppm.

Fig. 4.

Molecular structure of the cation of [(OEP)Fe(NO)]OTf in the $P\stackrel{‒}{1}$ space group (with thermal ellipsoids drawn at 50%). The H atoms and the anion have been omitted for clarity. Fe1–N5 = 1.6371(15) Å, N5–O1 = 1.1473(19) Å, ∠Fe1–N5–O1 = 176.15(15)°. The related [(OEP)Fe(NO)]ClO₄ structure in the P2₁/n space group has been reported.^{22, 23}

Fig. 5.

¹H NMR spectra (in CDCl₃) of the reaction to generate (OEP)Fe(NO)H, highlighting the upfield and downfield regions of (a) (OEP)Fe(NO)H (top spectrum), (b) (OEP)Fe(H) byproduct (middle spectrum), and (c) (OEP)Fe(H) from the control reaction of [(OEP)Fe]OTf with hydride (bottom spectrum). The peak labeled * is due to the H₂ decomposition product (see text).

Interestingly, geometry optimization of the model (porphine)Fe(NO)H product reveals a core geometry not unlike that of the structurally characterized aryl derivative (OEP)Fe(NO)(C₆H₄F-p),²⁴ showing an off-axis tilt of the nitrosyl N atom, a bent FeNO moiety (∠FeNO = 155.2°), and an asymmetry of the equatorial Fe–N(por) core displaying longer Fe–N(por) distances in the direction of the bent FeNO moiety (Fig. 6).

Fig. 6.

Selected geometrical parameters (in Å and degrees) for DFT-calculated (OEP)Fe(NO)H. The tilting angles are with respect to the four-nitrogen porphyrin plane.

We observe, on occasion, an additional broad peak at −4.6 ppm in the ¹H NMR spectrum (Fig. 5b). We attribute this latter peak to the non-nitrosyl paramagnetic (OEP)Fe(H) compound, probably resulting from dissociation of NO from the ferric [(OEP)Fe(NO)]⁺ cation in solution prior to hydride attack. Indeed, a control experiment involving the reaction of [(OEP)Fe]⁺ with borohydride reproduces this peak (Fig. 5c).

The (OEP)Fe(NO)H product is very unstable even at −35 °C, with the Fe-H peak at −4.11 ppm in the ¹H NMR spectrum disappearing even after only ^~15 mins. In fact, this peak is not detectable in the ¹H NMR spectrum when the reaction is carried out at −20 °C. The (OEP)Fe(NO)H decomposition products are (OEP)Fe(NO) (87% total yield by IR) and H₂ (85% total yield by NMR). That diborane is the boron-containing by-product in both the reactions of the six-coordinate [(OEP)Fe(NO)(5-MeIm)]⁺ and five-coordinate [(OEP)Fe(NO)]⁺ with borohydride was verified by ¹¹B{¹H} NMR spectroscopy (Fig. 7), which showed identical ¹¹B NMR signals at −26 ppm when compared with that of authentic diborane (prepared from the reaction of borohydride with a reactive alkyl halide).²⁵

Fig. 7.

¹¹B{¹H} NMR spectra of the product mixtures from (a) the reaction of [(OEP)Fe(NO)(5-MeIm)]OTf with excess [NBu₄]BH₄ (signal at −40 ppm), and (b) the reaction of [(OEP)Fe(NO)]OTf with excess [NBu₄]BH₄, and (c) the control and known reaction of [NBu₄]BH₄ with 1,2-dichloroethane to generate diborane.²⁵

We employed DFT calculations to provide insight into the differential Fe–H versus Fe–HNO bond-forming reactions (Scheme 1) when the five-coordinate [(porphine)Fe(NO)]⁺ cation (R1) is reacted with borohydride (R2). We previously reported the calculated reaction path for the analogous six-coordinate compound [(porphine)Fe(NO)(5-MeIm)]⁺.²¹ The "N path" in Scheme 1 represents an attack of hydride at the coordinated NO, and the "H path" represents direct Fe–H bond formation. The calculated electronic energies (ΔE), zero-point energy corrected electronic energies (ΔE_ZPE), enthalpies (ΔH), and Gibbs free energy (ΔG) follow the same trends for the N- and the H-paths (Table S2).

Scheme 1.

DFT-calculated N- and H-pathways for hydride addition to the five-coordinate [(P)Fe(NO)]⁺ cation.

The first encounter intermediate in the N-path is represented by I-1_N in Scheme 1, with a distance of 2.984 Å (Table S3) between the nitrosyl N atom and the hydride to be transferred. This distance shortens to 2.441 Å in the transition state TS_N, with an accompanying very slight lengthening of the bond between boron and the hydride to be transferred. In fact, the similarity of the B–H bond lengths in TS_N and the reagent R2, and the large difference (of 1.405 Å) in the N–H bond lengths between that in TS_N and the final Fe–HNO product P-1_N suggests an early transition state along the N-path, as observed also for the six-coordinate systems.²¹

The H-path first generates the intermediate I-1_H with a distance of 1.594 Å (Table S3) between the Fe and the hydride to be transferred; this distance is much shorter than that seen in I-1_N, suggesting a stronger Fe…HBH₃ interaction in I-1_H (H-path) than the N…HBH₃ interaction in I-1_N (N-path). This trend is also maintained in both transition states TS_N and TS_H. While the data for the N-path suggests an early transition state, that for the H-path suggests a much later transition state. For example, for the H-path, the difference in the Fe–H distances between TS_H and P-1_H is only 0.115 Å (c.f., the analogous difference in N–H distances of 1.405 Å along the N-path). Further, the difference in the B–H (H to be transferred) bond distances in TS_H and in the initial reactant R2 is 0.098 Å, ^~33x the noted difference along the related N-path. Attempts to locate a distinct second intermediate I-2_H, using a shorter Fe…H length and longer FeH…BH₃ distance (en route to P-1_H), yielded the same I-1_H structure, probably due to the strong favorable electronic driving force between the ferric metal center and the hydride as discussed above, and the strong interaction between hydride and BH₃ (the B-H bond length difference between I-1_H/I-2_H and TS_H is only 0.003 Å).

Analysis of the data above shows that although both pathways of hydride attack on the five-coordinate [(por)Fe(NO)]⁺ are thermodynamically favorable, the H-path is kinetically more favorable than the N-path by 17.43 kcal/mol (i.e., ΔE∣TS_H–TS_N∣), supporting the experimental formation of the Fe–H hydride complex with no observation of the Fe–HNO product. The charge analysis data (Table S4) is also consistent with the experimental and energy results; in the precursor cation [(porphine)Fe(NO)]⁺ (R1), the Fe atom bears a more positive charge (0.999e) than the nitrosyl N atom (0.101e; i.e. by ^~9.9x), consistent with an easier hydride transfer to Fe rather than N. In accordance with such a strong electronic driving force difference, and based on the charge differences in transition states and reactants (Table S4), hydride transfer to Fe results in a donation of 0.652e from borohydride, while the hydride transfer to the nitrosyl N results in a much smaller donation of 0.136e from borohydride.

Both the (por)Fe(HNO)(L) (Table 1) and (OEP)Fe(NO)H systems are thermally unstable even at −35 °C. For example, although the Fe–HNO peaks for the (OEP)Fe(HNO)(L) compounds are persistent in the ¹H NMR spectra for ^~2 h (integrating to ^~5-21% variable yields), the compounds decompose to generate H₂ (by ¹H NMR) and the five-coordinate (OEP)Fe(NO) derivative (by IR), with overall yields of 85% (L = 5-MeIm), 52% (L = ImH), and 76% (L = 1-MeIm), respectively, based on the cationic [(OEP)Fe(NO)(L)]⁺ precursors. The analogous overall yields for the PPDME systems to give (PPDME)Fe(NO) are 41%, 65%, and 51%, respectively. We surmise that HNO dissociation from the six-coordinate (por)Fe(HNO)(imidazole) products does not occur, as N₂O (the HNO dimerization product) is not detected in the headspace gas by IR spectroscopy. Curiously, when borodeuteride is used for the reaction with the [(OEP)Fe(NO)(5-MeIm)]⁺ cationic precursor (eq 1), D₂ is similarly generated (identified by ²H NMR spectroscopy) together with (OEP)Fe(NO), but we also detected the presence of N₂O (by IR) and D₂O (by ²H NMR spectroscopy), indicative of both D–NO bond cleavage and partial DNO dissociation from the (OEP)Fe(DNO)(5-MeIm) complex. In contrast, the reaction of borodeuteride with the five-coordinate [(OEP)Fe(NO)]⁺ cation resulted in the net formation of only D₂ and (OEP)Fe(NO).

We then employed DFT calculations to probe the Fe-HNO decomposition pathways. Selected data are shown in Fig. 8. and in Tables 2 and S5. As seen from the enthalpy costs to break the H–NO and Fe–N(H)O bonds, the covalent H–NO bond is stronger than the Fe–N(H)O coordination bond, as expected from the bonding nature, and both bonds become stronger in the presence of the axial 5-MeIm ligand. Although the trans effect elongates the Fe–N(H)O bond length by 0.054 Å (Scheme 1) it also donates charge (0.248 e) to (P)Fe(HNO).

Fig. 8.

Bond strength energies (ΔH and ΔG) for the five-coordinate and six-coordinate Fe–HNO systems (L = 5-MeIm). Values are in kcal/mol.

Table 2.

Reaction energies (in kcal/mol).^a

C.N.	Decomposition pathway	δG
5-C	(P)Fe(HNO) → (P)Fe + 1/2(N2O + H2O)	−48.71
5-C	(P)Fe(HNO) → (P)Fe(NO)+ 1/2 H2	−4.78
6-C	(P)Fe(HNO)L → (P)Fe(L) + 1/2(N2O + H2O)	−47.76
6-C	(P)Fe(HNO)L → (P)Fe(NO) + L + 1/2 H2	−3.19
	HNO → 1/2(N2O + H2O)	−44.46

^aC.N. = coordination number. P = porphine. L = 5-MeIm.

The reaction pathway associated with Fe–N(H)O bond breaking and subsequent HNO dimerization is much more thermodynamically favorable, by ^~44 kcal/mol in Gibbs free energy, than that associated with the H–NO bond breaking and subsequent H₂ formation. This is largely due to the strong thermodynamic driving force of HNO dimerization (ΔG of −44.46 kcal/mol; Table 2). In fact, the ΔG of −4.25 kcal/mol for the decomposition of the Fe–N(H)O bond in the five-coordinate (porphine)Fe(HNO) compound is similar to that for the H₂ formation pathway overall energy of −4.78 kcal/mol. The corresponding data for Fe–N(H)O decomposition and for the H₂ formation pathway are similar for the six-coordinate (porphine)Fe(HNO)L, at −3.30 kcal/mol and −3.19 kcal/mol, respectively. This suggests that the experimentally observed relative preference of H₂ formation in the presence of the axial ligand, versus HNO loss and subsequent dimerization, is not due to thermodynamics alone, but could be due to the kinetic effect of the hydrogen radical formation and dimerization.

The formation and spectral characterization of the Fe–HNO complexes allowed us to probe their N–N bond-forming reactions with external NO, a key coupling reaction step that is at the center of NO detoxification by fungal NO reductases.¹⁹ As described above, the decomposition of the six-coordinate (OEP)Fe(HNO)(5-MeIm) complex generates H₂ and (OEP)Fe(NO) with no evidence of N₂O formation as judged by headspace IR spectroscopy. Interestingly, however, our preliminary results from the reactions of (OEP)Fe(HNO)(5-MeIm) with external NO show clear N₂O formation resulting from an N–N coupling reaction. To verify that N₂O was indeed forming from a coupling reaction involving the bound HNO and external NO, we employed various NO isotopomers (i.e., containing ¹⁵N and/or ¹⁸O) in these reactions. Our results reveal that the terminal N-atom of the mixed-isotope N₂O product originates from the Fe-HNO moiety, whereas the central N-atom and the O-atom of the N₂O product originate from the external NO reagent (sketched schematically at the top of Fig. 9). The mixed-isotopic N₂O products were identified by their characteristic gas-phase IR spectra.²⁶ For example, the reaction of unlabeled (OEP)Fe(HNO)(5-MeIm) with external and doubly-labeled ¹⁵N¹⁸O generates ¹⁴N¹⁵N¹⁸O (bands at 2185 and 2162 cm⁻¹) as shown in the solid trace in Fig. 9b. The related reaction of (OEP)Fe(H¹⁵N¹⁸O)(5-MeIm) with unlabeled NO generates ¹⁵N¹⁴N¹⁶O (Fig. 9c, broken line trace; bands at 2195 and 2169 cm⁻¹). Further, the reaction of (OEP)Fe(H¹⁵NO)(5-MeIm) with unlabeled NO generates ¹⁵N¹⁴N¹⁶O, namely the same reaction product from the (OEP)Fe(H¹⁵N¹⁸O)(5-MeIm)/NO reaction.

Fig. 9.

N₂O formation from the reactions of in situ generated (OEP)Fe(HNO)(5-MeIm) with external NO. (a) spectra of authentic samples of ¹⁴N₂O (solid line trace, 2237/2212 cm⁻¹), ¹⁵N₂O (broken line trace, 2167/2142 cm⁻¹), and ¹⁵N₂¹⁸O (dotted line trace, 2160/2138 cm⁻¹), (b) spectra of the headspace from the reactions of (solid line trace) (OEP)Fe(HNO)(5-MeIm) with ¹⁵N¹⁸O, and (broken line trace) (OEP)Fe(H¹⁵N¹⁸O)(5-MeIm) with NO; the newly formed mixed-isotope N₂O bands are highlighted. The solid line trace shows formation of ¹⁴N¹⁵N¹⁸O (2185/2162 cm⁻¹), while the broken line trace shows formation of ¹⁵N¹⁴N¹⁶O (2195/2169 cm⁻¹), (c) IR spectrum of the headspace from the reaction of (OEP)Fe(H¹⁵NO)(5-MeIm) with NO. The newly formed bands for the mixed-isotope ¹⁵N¹⁴N¹⁶O bands at 2195/2169 cm⁻¹ are highlighted. The singly labeled N₂O gases are also present in the headspace, and provide good internal reference spectra.

Our finding that the terminal N-atom of the mixed-isotope N₂O products originate from the Fe–HNO moiety, whereas the central N- and O-atoms originate from external NO is consistent with that predicted by DFT calculations for the related N–N coupling reaction catalyzed by fungal cyt P450nor.^{19, 27} The generation of the additional singly labeled (at the N-atom) N₂O isotopomers implies that other pathways may supplement the mixed-isotope Fe–HNO/NO coupling reactions observed in this non-protein system.

Conclusion

In summary, we have demonstrated, experimentally and computationally, varied outcomes for hydride attack at ferric–NO moieties (five- vs. six-coordinate) that result in either Fe–HNO or Fe–H formation. We have also demonstrated varied decomposition pathways for these species, one involving H₂ formation without Fe–N(H)O bond cleavage (to form Fe–NO and H₂) and the other involving Fe–N(H)O bond cleavage (to form N₂O). We have also shown that the Fe–HNO species can react with external NO to generate N₂O in which the central N and the O atoms of N₂O derive from the external NO reagent. These results have set up excellent framework for the study of the Fe–HNO complexes related to their observed biology. Further work to delineate the reaction mechanism of the N–N coupling reactions and related reactions are underway.

Experimental

General

The reactions were performed anaerobically in standard Schlenk glassware and/or in a glove box under a nitrogen atmosphere. Solvents used in the reactions were collected under nitrogen from a Pure Solv 400-5-MD Solvent Purification System (Innovative Technology) or distilled from appropriate drying agents under an atmosphere of nitrogen. Nitric oxide (NO; Air Gas Inc.) gas was passed through a potassium hydroxide column, then through a cold trap (dry ice/acetone) prior to its contact with the precursor solution to avoid the introduction of NO_x impurities. ¹⁵NO (Icon Isotopes Inc., 99% ¹⁵N) was used as received without further purification. FT-IR spectra were recorded on a Bruker Tensor 27 spectrometer. ¹H NMR experiments for Fe–HNO and Fe–H detection were performed on a 400 MHz Varian NMR spectrometer at −20°C and −50°C, respectively. ¹¹B{¹H} NMR experiments were performed on a 400 MHz Varian NMR spectrometer using quartz NMR tubes.

[(PPDME)Fe(NO)(5-MeIm)]SbF₆

A CH₂Cl₂ (5 mL) solution of [(PPDME)Fe]SbF₆ (13.8 mg, 0.016 mmol) with 1 equiv of 4/5-methylimidazole (1.3 mg, 0.016 mmol) was stirred for 2 h, followed by the introduction of NO gas, in a similar manner to that used to prepare other crystalline [(por)Fe(NO)(L)]⁺ complexes.²⁸ X-ray diffraction-quality crystals of the product [(PPDME)Fe(NO)(5-MeIm)]SbF₆ (9.0 mg, 58% isolated yield based on Fe) were obtained from a mixed CH₂Cl₂/methanol solvent system under an NO atmosphere. IR (KBr): υ_NO = 1905 cm⁻¹. Characteristic band of uncoordinated hexafluoroantimonate: υ_Sb-F = 659 cm⁻¹.²⁹

(PPDME)Fe(HNO)(5-MeIm)

To a CDCl₃ (0.5 mL) solution of [(PPDME)Fe(NO)(5-MeIm)]OTf (8.1 mg, 0.009 mmol; υ_NO 1912 cm⁻¹ (υ_15NO 1874 cm⁻¹)) in a J. Young NMR tube at −20 °C was added a CDCl₃ (0.2 mL) solution of [NBu₄]BH₄ (5.0 mg, 0.017 mmol) in a similar manner to that reported by us earlier.²¹ The ¹H NMR spectra were recorded immediately. ¹H NMR (−20° C, 400 MHz): 13.93 ppm (s, Fe–HNO; J_15N-H = 77 Hz). The characteristic NO stretching frequency of the (OEP)Fe(HNO)(5-MeIm) complex was obtained from a separate reaction in CHCl₃ at −45°C (IR υ_NO = 1384 cm⁻¹, υ_15NO = 1360 cm⁻¹).

The other (por)Fe(HNO)(L) derivatives were generated in a similar manner.

Detection of H₂ from the decomposition of (PPDME)Fe(HNO)(5-MeIm)

The headspace gas of the reaction mixture to generate (PPDME)Fe(HNO)(5-MeIm) described above was collected using a gas-tight syringe after warming the mixture to room temperature. The headspace gas was then injected into pre-cooled CDCl₃ in an NMR tube and the ¹H NMR spectrum was recorded (Fig. S2A). ¹H NMR: 7.26 (s, residual CHCl₃), 4.62 (s, dissolved H₂ gas).³⁰ The data was compared with an authentic commercial sample of 5% H₂/N₂ (Fig. S2B).

Detection of N₂O from the decomposition of (OEP)Fe(DNO)(5-MeIm) and (OEP)Fe(DNO)(1-MeIm)

The headspace gas of the reaction mixture to generate (OEP)Fe(DNO)(5-MeIm) (using [NBu₄]BD₄ as the deuteride source) was vacuum transferred to an IR gas cell (10 cm pathlength) after warming the mixture to room temperature. IR (gas phase): υ_as (N₂O) = 2237/2213 (¹⁵N₂O = 2167/2145) cm⁻¹. ²H NMR: Δ = 2.2 ppm assigned to D₂O which was confirmed by spiking the sample with authentic D₂O. In a separate reaction, formation of the known five coordinate (OEP)Fe(NO) (solution υ_NO 1665 cm⁻¹) was also observed during the decomposition of (OEP)Fe(DNO)(5-MeIm). Identical results above were obtained for the decomposition of (OEP)Fe(DNO)(1-MeIm).

[(OEP)Fe(NO)]OTf

In a manner similar to that used to prepare [(OEP)Fe(NO)]ClO₄,²³ NO gas was introduced to a CH₂Cl₂ (5 mL) solution of [(OEP)Fe]OTf (10.7 mg, 0.015 mmol). X-ray diffraction-quality crystals of the product [(OEP)Fe(NO)]OTf (6.7 mg, 61% isolated yield based on Fe) were obtained from CH₂Cl₂/n-hexane under an NO atmosphere. IR (KBr): υ_NO = 1856, 1841 cm⁻¹.

(OEP)Fe(NO)H

To a CDCl₃ (0.5 mL) solution of [(OEP)Fe(NO)]OTf ⁷ (7.9 mg, 0.013 mmol; IR υ_NO =1856 and 1841 cm⁻¹) in a J. Young NMR tube at −50 °C was added a CDCl₃ (0.2 mL) solution of [NBu₄]BH₄ (6.0 mg, 0.02 mmol). The ¹H NMR spectra were recorded immediately. ¹H NMR (−50° C, 400 MHz): −4.11 ppm (s, Fe–H).

Thermal decomposition of (OEP)Fe(NO)H

IR and ¹H NMR spectroscopy were utilized to monitor the decomposition of (OEP)Fe(NO)H in the product mixture from the reaction of [(OEP)Fe(NO)]OTf with [NBu₄]BH₄. The IR spectrum was recorded after warming the solution mixture to room temperature for 0.5 h. IR (CHCl₃): υ_NO = 1668 cm⁻¹. ¹H NMR (−50° C, 400 MHz): Δ 4.62 ppm assigned to the H₂ by-product of the decomposition (see Fig. S2).³⁰

Reaction of (OEP)Fe(HNO)(L) with NO and its isotopomers

(a) ¹⁵N¹⁸O gas was introduced to a CH₂Cl₂ (3 mL) solution of the in-situ prepared (OEP)Fe(HNO)(5-MeIm) (−95 °C) in a sealed Schlenk tube. The reaction mixture was stirred for 10 min at this temperature. The solution was then slowly warmed to room temperature and stirred for additional 30 min. The headspace gases formed during the reaction were vacuum transferred to a gas IR cell (10 cm path length). IR (gas phase): in addition to the bands due to singly labeled N-atom of N₂O (N₂O at 2237/2213 cm⁻¹ and ¹⁵N₂ ¹⁸O at 2160/2137 cm⁻¹), new bands were observed at 2185/2162 cm⁻¹ assigned to ¹⁴N¹⁵N¹⁸O (the band at 2162 cm⁻¹ overlaps with a ¹⁵N₂ ¹⁸O band).

(b) A separate reaction of (OEP)Fe(H¹⁵N¹⁸O)(5-MeIm) with NO was also conducted in a similar manner described above. IR (gas phase): in addition to the bands due to singly labeled N-atom of N₂O gases, new bands were formed at 2195/2169 cm⁻¹ assigned to ¹⁵N¹⁴N¹⁶O.

(c) In a similar manner, we conducted a reaction of singly labeled (OEP)Fe(H¹⁵NO)(5-MeIm) with unlabeled NO. in addition to the bands due to N₂O and ¹⁵N₂O, new bands were formed at 2195/2169 cm⁻¹ assigned to ¹⁵N¹⁴N¹⁶O.

Computational Details

All calculations were performed using the program Gaussian 09³¹ as reported previously.²¹ Full geometry optimizations were conducted for all chemical species studied with subsequent frequency calculations to verify the nature of the corresponding stationary states on their potential energy surfaces and provide zero-point energy corrected electronic energies, enthalpies, and Gibbs free energies. Based on a recent methodology study,²¹ geometries were optimized using the mPW1PW91³² method and the NMR properties were calculated using the B3LYP³³ method with solvent (CHCl₃) effect included using the PCM formalism,^34-37 similar to the approach used previously to study ¹H NMR shifts in various organometallic complexes.³⁸ The basis set used in the geometry optimization is Wachters’ basis³⁹ for iron, 6-311++G(2d,2p) for 1st shell atoms (atoms bonded to iron, HNO, and BH₃/B₂H₆), and 6-31G(d) for other atoms. The basis set used in the NMR calculation is similar with the only difference of using LanL2DZ⁴⁰ basis for Fe. The calculated NO frequencies of various iron porphyrin systems studied in this work were scaled using the experimental/computational NO frequency (1380/1568) for a related HNO Ru porphyrin system.¹¹ The atomic charges were calculated using the Merz-Singh-Kollman scheme⁴¹ as implemented in Gaussian 09.