Synthesis of the bromide and iodide starting platforms

In Chapter II.1, we activated N2 by means of electrochemistry starting from [ReCl2(PNP)] (1^Cl) and performed an extensive mechanistic study accounting for all minimal occurring reaction steps to form [ReNCl(PNP)] (3^Cl). In this Chapter, the platform is extended by changing the halide ligands to bromide and iodide to examine if and how their different properties influence the (electro)chemical N2-splitting performance and to substantiate or expand our understanding of the corresponding mechanism.

Access to the iodide-substituted analogue of 1^Cl was established in our group via salt metathesis, which is a well-known strategy to exchange halogen substituents. Fergusson and Hevelt for instance showed the exchange of chloride to bromide or iodide in [ReOCl3(Et3P)2], by reacting it with LiBr or NaI, respectively.¹⁶⁶ Analogous, Dr. J. Abbenseth reacted 1^Cl with an excess of NaI, upon which halogen-substitution readily occurs, see Scheme 14. Repetitive extractions with benzene afford [ReI2(PNP)] (1^I) in 90 % yield.¹⁶⁷ In a slightly modified procedure used herein, the second benzene extraction is performed over Celite, yielding 1^I in circa 85 %.

Scheme 14. Synthesis of 1^I via salt-metathesis starting from 1^Cl.¹⁶⁸

The bromide congener [ReBr2(PNP)] (1^Br) was not reported before, and its synthesis was attempted by the same halogen exchange strategy. No reactivity was however observed upon reacting 1^Cl with NaBr. Changing to more soluble LiBr resulted in full conversion, with concomitant formation of two species according to two new singlets in ³¹P{¹H} NMR spectroscopy at −81 ppm and −110 ppm (1^BrCl and 1^Br respectively, vide infra).

Unfortunately, independent of the equivalents of added LiBr, no product was formed selectively.

Therefore, halogen exchange to access 1^Br was discarded, and we attempted a synthesis pathway similar to synthesis of 1^Cl.

Figure 18. Top: Synthesis of 1^Br. Bottom left: NMR spectroscopy of 1^Br: ³¹P{¹H} NMR spectrum (top), and ¹H NMR spectrum (bottom). Bottom right: Molecular structure of 1^Br: ORTEP plot with anisotropic displacement parameters drawn at the 50% probability level. Hydrogen atoms omitted for clarity. Selected bond lengths (Å) and angles (deg.):

Re-Br1: 2.4635(5), Re-Br2: 2.5123(5), Re-N1: 1.917(4), Re-P1: 2.4109(11), Re-P2: 2.4111(12), P2-Re-P1: 162.30(4), Br1-Re-Br2: 102.301(17), N1-Re-Br2: 144.04(12).

The rhenium(III)-precursor [Re(PPh3)2Br3(MeCN)]^169,170 reacts with the PNP-ligand in presence of base, resulting in a colour change from orange to dark brown (Figure 18). The corresponding

31P{¹H}NMR spectrum shows a main feature at −110 ppm and three signals by ¹H NMR spectroscopy that integrate 36:4:4, indicating a C2v-symmetric species. After work up, a brown solid is obtained and LIFDI mass spectroscopic analysis confirms successful synthesis of 1^Br. By layering a benzene solution with pentane, single crystals suitable for X-ray spectroscopy of 1^Br where obtained (Figure 18). 1^Br has a short PNP Namide-Re distance of 1.917(4) Å and a pyramidalised Namide (Σbond angles = 359.8 °), indicating strong N  Re π-donation, analogous to its chloride and iodide congeners (N-Re [Å] = 1.923(7) (1^Cl), 1.926(2) (1^I) ).^69,167 A τ5-value of 0.30 indicates distortion from both trigonal bipyramidal and square pyramidal geometry.¹⁷¹ Comparing the molecular structures of 1^X (X = Cl, Br, or I), we observe a trend towards a square pyramidal geometry being a better description going to the heavier halide homologues based on the τ5-values (τ5 = 0.37 (1^Cl), 0.18 (1^I)). This is tentatively attributed to the increased atomic

radius down the halogen group: in a square pyramidal geometry, steric hindrance is minimised by increasing the distance between the tBu-groups and the halogen atoms.

Both ³¹P{¹H} and ¹H NMR shifts are unusual for a diamagnetic complex, and this observation is attributed to TIP, as proposed for several Re(III)-species (Section II.1.1). SQUID magnetometry was performed on 1^Cl and 1^I, revealing TIP with molar susceptibilities of 282.6 10^-6 cm³ mol^-1 (1^Cl) and 248.4 10^-6 cm³ mol^-1 (1^I).¹⁶⁸ Although no SQUID magnetometry was measured for 1^Br, it is expected to possess similar values.

Although 1^Br is synthesised analytically pure according to combustion analysis, we observe a Cs -symmetric impurity at δ31P{1H} = −81 ppm in about 2-5 %. Strikingly, its NMR features are precisely in between 1^Br and 1^Cl (δ31P{1H} = −110 ppm (1^Br), −51 ppm (1^Cl)), pointing in the direction of partial halide exchange. Indeed, when in a separate experiment 1^Br and 1^Cl are mixed in equimolar amounts, we see formation of the same species with a LIFDI mass of 661.2 m/z, matching to [ReBrCl(PNP)] (1^BrCl) (see Scheme 45). The exchange reaction does not go to completion, and due to the same solubility properties, 1^BrCl cannot be isolated. Similar behaviour is observed when 1^Cl and 1^I were mixed, yet upon heating (60 ˚C): formation of a Cs-symmetric species with a LIFDI mass of 709.1 m/z, corresponding to [ReClI(PNP)] (1^ICl). The chloride that needs to be present to form 1^BrCl in the synthesis of 1^Br most likely originates from traces of chlorinated solvents or agents in the glovebox atmosphere. Performing the work up or storage of 1^Br in a chlorinating-agents free glovebox decreases the presence of 1^BrCl, yet it never fully disappears. A comparable problem is described by Schrock, where a persistent chloride impurity prohibits full purification of [ReBr(N(CH2CH2NC6F5)3)].¹⁷² Upon characterisation via XRD, the electron density of the axial halide can only be fitted when a ratio between bromide and chloride is considered. Although their synthetic approach is a template synthesis from a related chloride oxo complex, they attribute the chloride presence to residual DCM. For the parallel developed six-coordinate [ReX3(PNP^iPr)] platforms within our group (X = Cl, Br, I),these problems were not encountered,^h hinting towards an associative mechanism for formation of 1^XCl.

h Unpublished results of M. Fritz, M.Sc.

Scheme 45. Reacting 1^Br or 1^I with 1^Cl results in formation of 1^BrCl or 1^ICl as evidenced spectroscopically.

2.2. (Electro)chemical N2-splitting of 1^Br and 1^I

Upon establishing synthetic access to 1^Br and 1^I, we explored their N2-splitting ability.

Chemically, 1^Br and 1^I are readily reduced under 1 atm N2 using equimolar amounts of Na/Hg (−2.36 V)⁵⁵ or Co(Cp*)2 (E1/2 = −1.84 V in THF).⁵⁹ In addition, Cr(Cp*)2 (E1/2 = −1.47 V in THF)⁵⁹ also reduces 1^I. In all cases, reduction is accompanied by a colour change via red to light brown and the formation of [ReNBr(PNP)] (3^Br) or [ReNI(PNP)] (3^I) as confirmed via NMR spectroscopy and comparison to independently synthesised nitrides, as extensively described in Section II.2.3. Atmospheric N2-uptake is confirmed by reduction of 1^Br and 1^I under a ¹⁵N2 -atmosphere, resulting in a singlet resonance in the ¹⁵N NMR spectrum that compares well with the chemical shift as found for the chloride congener (δ15N{1H} = 375 ppm (3^Br), 381 ppm (3^I)), 371 ppm (3^Cl)⁶⁹). In the bromide case, where minor 1^BrCl is present, we see formation of some 3^Cl (≈ 2% yield), (Figure A8), indicating bromide as leaving group during N2-splitting.

By addition of a ³¹P- and/or a ¹H-NMR standard the N2-splitting yields for 3^Br and 3^I were determined. Yield determination via ¹H NMR spectroscopy was not possible for every reaction, e.g. reduction of 1^I with Na/Hg, because of overlapping side-product peaks. In other cases, yield determination via both ³¹P- and ¹H-NMR spectroscopy was accessible, e.g. for reduction of 1^I with Cr(Cp*)2, giving matching results. In general, 3^Br and 3^I can be obtained in moderate to high yields via chemical N2-splitting (80% for 3^Br (Na/Hg), and 60% for 3^I (Cr(Cp*)2 or Co(Cp*)2). Decreased spectroscopic yield is found when using Na/Hg for synthesis of 3^I (30 %), or Co(Cp*)2 for 3^Br (45 %). These yields are at least reproduced once and spread between the individual runs around 10 % (i.e. 76 % and 85 % for synthesis of 3^Br using Na/Hg as reductant, further exact numbers are given in the Experiment Section IV.5).

Scheme 46. Left: (Electro)chemical N2-splitting from 1^Br into 3^Br using Na/Hg, Co(Cp*)2, or via CPE with an applied potential of Eappl. = –1.72 V. Right: (Electro)chemical N2-splitting from 1^I into 3^I using Na/Hg, Co(Cp*)2, Cr(Cp*)2 or via CPE with an applied potential of Eappl. = –1.72 V or –1.58 V.

Electrochemical synthesis of 3^Br and 3^I was explored by CPE in a N2-atmosphere containing MBraun Glovebox at the potential of the first reduction as obtained via CV analysis (see Section II.2.5). CPE for 1^Br at Eappl. = −1.72 V results in a colour change from purple to yellow and the transfer of circa 1.2 electrons per Re. Concomitantly, in situ CV shows a new, prominent, and irreversible peak in the oxidative region with Ep = −0.02 V (Figure 19), close to the Re(VI/V) oxidation of 3^Br (E1/2 = −0.05 V). Although these potentials are slightly apart, no iR-compensation is applied when measuring CV in a CPE set up. Upon addition of separately synthesised 3^Br following CPE, the current of this new wave increases confirming nitride formation (Figure A11). The irreversibility of this feature is not clear at this point. In a control experiment, 3^Br was examined in presence of (nHe4N)Br to mimic bromide release as under CPE conditions (Figure 19). However, the Re(VI/V) oxidation couple maintains reversible (at higher bromide concentrations (i.e. 5 eq.) the oxidations of 3^Br and Br⁻ are overlapping). This is in contrast to titration of 3^Cl with Cl⁻, which indicates chloride coordination upon oxidation. An explanation might be the increased halide radius in case of bromide. To test this hypothesis, it would be interesting to add a smaller ligand and explore if the Re^VI/V reversibility is lost in that case. We tentatively attribute the irreversible oxidation of 3^Br under CPE conditions to coupled chemistry with unknown side products. ³¹P{¹H}-NMR confirmed the formation of 3^Br in 55 % spectroscopic yield, which represents a FE of circa 45 %.

Figure 19. Left: CPE of 1^Br at Eappl. = −1.72 V as monitored by CV. Inset: CV of the Re(VI/V)-oxidation of isolated 3^Br in THF with 0.2 M (nBu4N)PF6 in presence of 1 and 5 eq. of (nHe4N)Br, ν = 0.1 Vs^-1. Right: CPE of 1^I at Eappl. = −1.58 V as monitored by CV. Inset: CV of the Re(VI/V)-oxidation of isolated 3^I in THF with 0.2 M (nBu4N)PF6.

CPE of 1^I was performed at Eappl. = −1.58 V, which is accompanied by a colour change from green to brown and a transfer of 1.07 electrons per Re. In the CV traces in Figure 19, we see various oxidative waves appearing of which the wave of +0.06 V is close to the Re(VI/V) oxidation of 3^I (E1/2 = 0.01 V) (see Section II.2.3). Upon addition of some separately synthesised 3^I after a CPE experiment, this feature also clearly increases in intensity (Figure A11), confirming nitride synthesis. The nitride was spectroscopically quantified in about 50 % yield by ³¹P{¹H} NMR spectroscopy, which corresponds to 47 % FE. Notably, CPE at more negative potential (Eappl. = −1.72 V) yields more 3^I (65 %), which corresponds to a FE of 60 %. For both the bromide- and the iodide-case, CPE under argon only resulted in trace amounts of the corresponding nitrides.

Having established an electrochemical yield of 3^Br and 3^I of about 60 %, the question rises which other products are formed. The CV traces show the presence of multiple peaks in the oxidative region. Unfortunately, only crude spectroscopic information is obtained as it was found very hard to separate the nitrides from the electrolyte: the nitrides are barely soluble in pentane and column chromatography resulted in partial decomposition. Mass spectroscopic analysis is hampered by the presence of the (nBu4N)-cation, since this obstructs the tubing. The ³¹P{¹H}

NMR spectra for quantification are low in intensity (see Figures A8 and A9), and only reveal a few side products present in low amounts. Likely, additional compounds remain in the noisy baseline or are paramagnetic (¹H NMR spectroscopy was not measured after electrolysis).

Although no side-product could be identified, we showed electrochemical N2-splitting also for the 1^Br and 1^I platform in quite good yields.

In Chapter II.1, we presented a pathway as minimal mechanistic model for (electro)chemical N2 -splitting for 1^Cl. To answer if the bromide and iodide congeners split N2 via a similar model, we need to collect spectroscopic evidence if an analogous N2-bound dimer as 2^Cl is found as intermediate, which is discussed in Section II.2.4.