Manganese(I) Complex with Monodentate Arylisocyanide Ligands Shows Photodissociation Instead of Luminescence.

Recently reported manganese(I) complexes with chelating arylisocyanide ligands exhibit luminescent metal-to-ligand charge-transfer (MLCT) excited states, similar to ruthenium(II) polypyridine complexes with the same d6 valence electron configuration used for many different applications in photophysics and photochemistry. However, chelating arylisocyanide ligands require substantial synthetic effort, and therefore it seemed attractive to explore the possibility of using more readily accessible monodentate arylisocyanides instead. Here, we synthesized the new Mn(I) complex [Mn(CNdippPhOMe2)6]PF6 with the known ligand CNdippPhOMe2 = 4-(3,5-dimethoxyphenyl)-2,6-diisopropylphenylisocyanide. This complex was investigated by NMR spectroscopy, single-crystal structure analysis, high-resolution electrospray ionization mass spectrometry (HR-ESI-MS) measurements, IR spectroscopy supported by density functional theory (DFT) calculations, cyclic voltammetry, and time-resolved as well as steady-state UV-vis absorption spectroscopy. The key finding is that the new Mn(I) complex is nonluminescent and instead undergoes arylisocyanide ligand loss during continuous visible laser irradiation into ligand-centered and charge-transfer absorption bands, presumably owed to the population of dissociative d-d excited states. Thus, it seems that chelating bi- or tridentate binding motifs are essential for obtaining emissive MLCT excited states in manganese(I) arylisocyanides. Our work contributes to understanding the basic properties of photoactive first-row transition metal complexes and could help advance the search for alternatives to precious metal-based luminophores, photocatalysts, and sensors.

Introduction

Metal-to-ligand charge-transfer (MLCT) excited states play a key role in many coordination complexes and organometallic compounds because they enable a range of different applications in photophysics and photochemistry. Precious and rare elements such as ruthenium(II),[1−500] osmium(II),[5−9] rhenium(I),[10−18] or iridium(III)[15,19−25] in polypyridine or cyclometalating coordination environments often feature a long-lived MLCT excited state,[26,27] whereas among first-row d6 transition metal elements, this is yet a very rare occurrence. Iron(II) is by far most investigated in this regard,[28−44] yet only a handful of iron(II) complexes with MLCT lifetimes in the nanosecond time regime are known.[28,29,41] Building on early reports of hexakis(arylisocyanide) manganese(I) complexes with a focus on UV–vis absorption and electrochemical properties,[45−52] we recently discovered that manganese(I) complexes with chelating bi- and tridentate ligands have luminescent and photoredox active MLCT states.[53] Until now, the synthesis of these chelates has remained laborious,[54] and therefore it seemed attractive to explore the possibility of using more directly accessible monodentate arylisocyanides for manganese(I) complexes with luminescent MLCT states. Recently, there has been increased interest in first-row transition metal elements, partly because they are cheaper and more abundant than the traditionally used metals from the platinum group,[55] and because they seem to offer ample opportunities for groundbreaking discoveries.[56−59] Advances in ligand design, photophysical techniques, and theoretical understanding[60] have made an unexpectedly broad range of transition metals in different oxidation states useable for photophysical or photochemical applications. In particular, luminescent first-row transition metal complexes with V,[61−64] Cr,[65−73] Mn,[53,74−76] Fe,[28,34,42,77−79] Co,[71,80−800] Ni,[83−86] and Cu[87−96] with different types of electronically excited states featuring promising photoreactivity and photoluminescence behavior have been discovered recently.[56,57,59,97,98]

As early as 1977, Mann, Gray, and Hammond reported on tungsten(0) arylisocyanide complexes that showed MLCT luminescence in solution at room temperature.[99] More recently, the GRAY group revisited this topic and discovered a whole range of [W(CNAryl)6],[100−105] complexes with outstanding photoluminescence properties and very strong photoreducing behavior with applications in catalysis. In addition to diisopropyl substituents at the ortho-position to the ligating isocyanide functional group, the attachment of further phenyl rings at the para-position proved valuable to enhance the chemical stability and the photophysical properties of this compound class, for example in [W(CNdippPhOMe2)6] (CNdippPhOMe2 = 4-(3,5-dimethoxyphenyl)-2,6-diisopropylphenylisocyanide).[101] This specific tungsten(0) complex featured a high photoluminescence quantum yield (0.42) paired with a long MLCT lifetime (1.65 μs in toluene at room temperature), hence our interest in this particular ligand for manganese(I).[101]

In the recent past, other oxidation states of manganese received attention in the context of photoluminescence; for instance, many manganese(II) compounds are known to emit in the solid state.[74] Furthermore, a bis-(tris(carbene)borate) manganese(IV) complex featured dual luminescence from LMCT (ligand-to-metal charge-transfer) and metal-centered (MC) excited states.[76] At present, the photophysics of manganese(I) isocyanide complexes seems underexplored yet, particularly in comparison to isoelectronic iron(II) polypyridines.[28,29,31−33,37−44,79,106−116] By contrast, tricarbonyl complexes of manganese(I) represent a fairly well-investigated class of compounds, which can undergo phototriggered loss of CO ligands. Consequently, they belong to the so-called photoCORM family of compounds, i. e., photoactive carbon monoxide-releasing molecules, that are interesting for the light-controlled delivery of CO for therapeutic purposes.[75,117−130]

Here, we report on the new complex [Mn(CNdippPhOMe2)6]PF6 (CNdippPhOMe2 = 4-(3,5-dimethoxyphenyl)-2,6-diisopropylphenylisocyanide), including detailed structural, vibrational, electrochemical, and photophysical studies.

Experimental Section

Physical Measurements

NMR spectra were recorded in deuterated solvents on a Bruker Avance III NMR spectrometer operating at a 1H frequency of 600 or 400 MHz, a 13C frequency of 151 or 100 MHz, 31P and 19F frequencies of 565 and 243 MHz, respectively, at 298 K. All chemical shifts are reported in δ values in ppm relative to tetramethylsilane (TMS), referred to protons of the residual nondeuterated solvent used or its carbon atoms, respectively (1H: δ(CD2Cl2) = 5.32 ppm, δ(CDCl3) = 7.26 ppm), the solvent signal (13C: δ(CD2Cl2) = 53.84 ppm, δ(CDCl3) = 77.16 ppm), or an external standard (31P: 85% aqueous H3PO4 in sealed capillary; 19F: CFCl3 in CDCl3).[131−133] Signals were assigned with the help of DEPT-135 and two-dimensional correlation spectra (1H,1H-COSY, 1H,13C-HSQC, and 1H,13C-HMBC). Signal multiplicities are abbreviated as s (singlet), d (doublet), t (triplet), sept (septett), m (multiplet), and br. (broad signal).

The 55Mn (I = 5/2) NMR experiments were performed on a 14.1 T instrument at a frequency of 148.56 MHz using a broad-band direct observe probe (BBFO). The transmitter offset was chosen as −1400 ppm and a spectral width of 100 ppm (14.8 kHz) was applied. The 90° pulse was 13 μs at a power level of 100 W. We used a 90° excitation mode, an acquisition time of 2.0 s, and a recycling delay of 100 ms. Sixty four scans were acquired and multiplied with an exponential line broadening function (lb = 5 Hz) before Fourier transformation. The 55Mn chemical shifts were referenced to a saturated solution of KMnO4 in D2O as δ = 0.00 ppm.[134] The recommended reference concentration of 0.82 mol kg–1 (13% w/w) is more than twice higher than the solubility in water at room temperature and seems incorrect.

Elemental analyses were performed using a vario MICRO cube CHN element analyzer from Elementar. Samples were burned in sealed tin containers using a stream of oxygen.

High-resolution ESI mass spectra were recorded on a Bruker maXis 4G ESI-Q-TOF spectrometer under direct injection conditions with CH3CN as a solvent.

IR spectra were recorded on a Shimadzu IRTracer-100 with QATR 10 spectrometer. Signal intensities are marked as vs (very strong), s (strong), m (medium), w (weak), br (broad), and vw (very weak).

X-ray powder diffraction (XRPD) measurements were performed with a STOE STADI P diffractometer with a microfocused Cu Kα radiation source (λ = 1.542 Å) equipped with a DECTRIS MYTHEN 1K detector.

Cyclic voltammetry was performed using a Versastat4-200 potentiostat from Princeton Applied Research in a glovebox. A glassy carbon disk electrode served as a working electrode, and two silver wires served as counter and pseudo-reference electrodes. Ferrocene was added as an internal reference (E1/2 vs saturated calomel electrode (SCE) in dichloromethane (DCM) = 0.475 V).[135] The solvent was dry and deaerated CH2Cl2 with 0.1 M TBAPF6 (tetra-n-butylammonium hexafluorophosphate) as an electrolyte. Potential sweep rates were, unless otherwise stated, 100 mV s–1.

All photophysical measurements were performed, unless otherwise stated, under an Ar atmosphere in either Schlenk or screw cap cuvette with dry and argon-saturated solvents (CH3CN and CH2Cl2).

UV/vis absorption spectroscopy was performed using a Cary 5000 instrument from Varian.

For photostability experiments, a Flame UV/vis spectrometer system from Ocean optics was used with a continuous-wave (cw) laser from Roithner Lasertechnik emitting at 405 nm (optical output up to 375 mA and 526 mW, circular beam dimension with a diameter of 2.5 mm) and a cw-laser emitting at 447 nm (optical output up to 1050 mA and 1.1 W, rectangular beam dimension with a diameter of 2 × 5 mm2). Spectroscopic experiments with that cw-laser as a light source were carried out at 293 K using self-built cuvette holders allowing temperature control and stirring of the solution with a small magnetic stir bar. Photochemical processes involving free radicals and triplet-excited states are usually sensitive to dissolved oxygen. Therefore, oxygen was removed by saturating the solution with argon (for 5 min) before the measurements.

Single-Crystal X-ray Structure Analysis

[Mn(CNdippPhOMe2)6]PF6 was recrystallized from DCM and cyclohexane at room temperature. Single yellow block-shaped crystals were obtained after a few days. A suitable crystal with dimensions 0.20 × 0.16 × 0.12 mm3 was selected, and the crystal was mounted on a mylar loop in perfluoroether oil on a Cu-Stoe diffractometer. The crystal was kept at a steady T = 150 K during data collection. The structure was solved with the olex2.solve 1.5[136] solution program using iterative methods and using Olex2 1.5[137] as the graphical interface. The model was refined with ShelXL 2018/3[138] using full-matrix least-squares minimization on F2.

Crystallographic drawings were made with the software packages Mercury[139,140] and CYLview.[141] Selected crystal data and details of the structure determinations are presented in Tables S1 and S2.

CCDC 2165895 ([Mn(CNdippPhOMe2)6]PF6 at 150 K) contains the supplementary crystallographic data for this article. These data can be obtained free of charge from the Cambridge Crystallographic Data Centre via http://www.ccdc.cam.ac.uk/data_request/cif.(142−144)

Density Functional Theory (DFT) Calculations

All calculations were carried out for the gas phase using the ORCA 4.1.2 package.[145−147] Geometry optimization and the calculation of infrared modes were performed with the composite approach PBEh-3c,[148] the basis def2-mSVP(C,H,N,O)[148] and def2-mTZVP(Mn)[149] with the atom pairwise dispersion correction with the Becke–Johnson damping scheme (D3BJ),[150,151] the geometrical counterpoise correction gCP,[152] the RIJCOSX approximation, the auxiliary basis def2/J,[153] fine numerical integration grids (grid5, gridX7, and NoFinalGrid in ORCA 4 nomenclature), and KDIIS+SOSCF. The crystallographic structure of the complex was used as a starting point. The libint2 library was used for the computation of 2-el integrals.[154] The calculated frequencies were scaled by a factor of 0.95.[148] No imaginary frequencies were obtained. Calculations were performed without symmetry. The Cartesian coordinates from the geometry optimization are in the Supporting Information (SI). The output results were handled with the ChemCraft software package.[155]

General Synthesis Procedures and Product Characterizations

The synthesis and characterization of the ligand CNdippPhOMe2 is described in the SI (see Schemes S1 and S2). The complexation was carried out in dry solvents and under a nitrogen atmosphere using standard Schlenk techniques. [Mn(CO)5Br] and KPF6 were purchased and used without further purification. THF was purified by a solvent purification system (SPS) by Innovative Technology and further degassing by at least three cycles of the freeze–pump–thaw method using nitrogen.

Synthesis of [Mn(CNdippPhOMe2)6]PF6

[Mn(CO)5Br] (20.4 mg, 72.7 μmol) and CNdippPhOMe2 (141 mg, 436 μmol) were dissolved in dry and degassed THF (5 mL) and refluxed for 24 h. After cooling to room temperature, a saturated aqueous solution of KPF6 (5–10 mL) was added. Overnight, a bright yellow to orange solid formed. The precipitate was collected by suction filtration and washed with deionized water. The solid was dissolved in DCM, dried over Na2SO4, filtered off, and washed with DCM. The solvent was removed in vacuum, resulting in an orange solid. This solid was recrystallized in DCM and cyclohexane at room temperature. Yellow crystals were obtained after a few days. The crystals were filtered off and dried under air, leading to a yellow powder. Yield: 82 mg (38.3 μmol, 53% relative to [Mn(CO)5Br]).

Elemental analysis calculated for C126H150F6MnN6O12P: C 70.70, H 3.93, N 7.06%, found: C 70.41, H 3.62, N 6.68%.

HR-ESI-MS(+)(CH3CN): m/z (%) = [M – PF6]+ calcd 1995.0720, found 1995.0679 (100), [M – PF6 – L + CH3CN]+ calcd 1712.9106, found 1712.9063 (36), [M – PF6 – L + CO]+ calcd 1699.8790, found 1699.8749 (20.9).

1H NMR (600 MHz, CD2Cl2): δ/ppm = 7.42 (s, 2H, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn), 6.74 (d, 3JH–H = 2.3 Hz, 2H, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn), 6.51 (t, 3JH–H = 2.2 Hz, 1H, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn), 3.86 (s, 6H, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn), 3.43 (sept, 3JH–H = 6.9 Hz, 2H, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn), 1.23 (d, 3JH–H = 7.0 Hz, 12H, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn).

13C{1H} NMR (151 MHz, CD2Cl2): δ/ppm = 176.0–182.0 (m, Cq, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn), 161.7 (Cq, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn), 145.0 (Cq, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn), 142.6 (Cq, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn), 142.1 (Cq, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn), 125.3 (Cq, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn), 122.9 (CH, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn), 106.1 (CH, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn), 99.5 (CH, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn), 55.9 (CH3, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn), 31.0 (CH, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn), 22.6 (CH3, {(Me2CH)2C6H2(C6H3OMe2)(NC)}6Mn).

55Mn NMR (148.5 MHz, CD2Cl2): δ/ppm = −1389.52 (s, Mn).

19F NMR (565 MHz, CD2Cl2): δ/ppm = −73.68 (d, J = 710.2 Hz, PF6–).

31P NMR (243 MHz, CD2Cl2): δ/ppm = −144.54 (sept, J = 710.2 Hz, PF6–).

Ir (ATR, 298 K): υ/cm–1 = 3075 (w, υ[=C–H]), 2964 (m, υ[−C–H]), 2932 (m, υ[−C–H]), 2875 (w, υ[−C–H]), 2846 (w, υ[−C–H]), 2070 (s, υ[−N≡C]), 1592 (s), 1573 (w), 1457 (m), 1434 (m), 1403 (m), 1386 (w), 1355 (m), 1324 (w), 1310 (w), 1266 (m), 1203 (s), 1153 (s), 1133 (w), 1111 (w), 1064 (m), 1038 (m), 992 (w), 962 (w), 943 (w), 928 (w), 904 (w), 891 (w), 877 (w), 836 (vs, ν[PF6]), 788 (s), 757 (m), 735 (m), 697 (m), 660 (w), 586 (vs), 556 (m), 525 (w), 480 (w), 464 (w).

The experimental XRPD pattern as well as the calculated pattern obtained from the single-crystal structure are shown in Figure S7. Not surprisingly, the XRPD pattern of the single-crystal structure containing 0.8 water and 2 cyclohexane molecules per unit cell does not match the XRPD pattern of the yellow powder of [Mn(CNdippPhOMe2)6]PF6 because of the loss of solvent by drying the crystals under air.

Results and Discussion

Synthesis, Characterization, Crystal Structure, and DFT Calculations

The ligand 4-(3,5-dimethoxyphenyl)-2,6-diisopropylphenylisocyanide (CNdippPhOMe2) was synthesized following a previously published procedure,[101] which was adapted with modifications inspired by other studies (Schemes S1 and S2).[156,157] The complex was prepared by reacting 6 equiv of the monodentate isocyanide ligand CNdippPhOMe2 with [Mn(CO)5Br] in THF under reflux and inert gas for 24 h (Scheme ). Afterward, the complex was precipitated with an aqueous hexafluorophosphate solution and yellow crystals were obtained from DCM/cyclohexane. As a solid, the compound [Mn(CNdippPhOMe2)6]PF6 can be handled under an ambient atmosphere without noticeable degradation, and 1H, 13C, 19F, and 31P NMR spectra in deuterated DCM confirm that the compound is pure (Figures S1–S5). The ESI-HR measurements in CH3CN exhibit the expected mass peak for [M – PF6],+ as well as the mass peaks for [M – PF6 – L + CH3CN]+ and [M – PF6 – L + CO]+ (Figure S8 and Table S3). Over time, the solution stored under ambient conditions and ambient light becomes intensely yellow colored, and measurements performed after a few days exhibit the mass peaks for [M – PF6 – L + CH3CN]+, [M – PF6 − 2L + 2CH3CN]+, [M – PF6 − 3L + 2O]+, and [M – PF6 − 4L + 2O]+, along with some additional minor mass peaks (Figure S9 and Table S3). Evidently, ligand exchange occurs in coordinating CH3CN solvent over time, and furthermore, the presence of oxygen seems to be detrimental.

Scheme 1

Synthesis of [Mn(CNdippPhOMe2)6]PF6

In the 13C NMR spectrum of [Mn(CNdippPhOMe2)6]PF6 in CD2Cl2 (Figures S2 and S3), the coordinating C-atoms of the aromatic isocyanide ligands appear between 170 and 185 ppm in the form of a multiplet, likely reflecting an overlay of 55Mn (I = 5/2) and 14N (I = 1) coupling patterns. This finding fits perfectly with the previously reported three compounds [Mn(CNPh)6]BF4, [Mn(CNPh)6]BF4, and [Mn(CNPh)6]BF4 in CD3CN, which exhibit identical signals in the chemical shift range from 170 to 185 ppm.[158]

The 55Mn NMR spectrum in CD2Cl2 (Figure S6) exhibits a remarkably sharp resonance at -1390 ppm with a linewidth (fwhh) of only 69 Hz, in line with an almost perfect O symmetry of the primary coordination sphere around Mn. The observable chemical shift is in the expected range for Mn(I) between -1000 ppm and -1500 ppm and thus confirms the +I oxidation state. This signal is not changing over a few hours in CD2Cl2 at 298 K in the dark, hence ligand exchange is evidently slower in this solvent than in CH3CN. The chemical shift observed for [Mn(CNdippPhOMe2)6]+ (−1390 ppm) lies between the values found for our two previously investigated manganese(I) complexes with chelating arylisocyanides,[53] which featured 55Mn resonances at -1225 and -1419 ppm, suggesting that the new complex has an electron density at Mn(I) somewhere in between the two previously studied related compounds. More striking is the very large difference in linewidth associated with the 55Mn resonances, which is only 69 Hz for [Mn(CNdippPhOMe2)6]+ but amounts to 2100 Hz and 5900 Hz in the previously studied chelate complexes, and this reflects the higher symmetry of the new Mn(I) complex with monodentate arylisocyanide ligands. Furthermore, this finding of a relatively narrow 55Mn resonance bandwidth is similar to the abovementioned other Mn(I) complexes with monodentate arylisocyanide ligands of the type [Mn(CNR)6]BF4. For example, [Mn(CNPh)6]BF4 featured its 55Mn resonance at -1382 ppm with a linewidth of 42 Hz in CD3CN at 18 °C.[158]

An additional resonance at -1387 ppm is observable in the 55Mn NMR spectrum of [Mn(CNdippPhOMe2)6]PF6 (Figure S6), but its integral is more than 10 times lower than that of the main resonance at -1389 ppm. We tentatively attribute the minor resonance at −1387 ppm to [Mn(CO)(CNdippPhOMe2)5]PF6, the mass peak of which is furthermore observable in the high-resolution electrospray ionization mass spectrometry (HR-ESI-MS) measurements.

The compound [Mn(CNdippPhOMe2)6]PF6 crystallizes in the P1̅ space group with Z = 1 molecule in the unit cell. The obtained crystal structure contains 0.8 water and 2 cyclohexane molecules per formula unit. The asymmetric unit consists of the manganese cation, three CNdippPhOMe2 ligands, and the PF6– anion. The Mn(I) cation is coordinated by six neutral CNdippPhOMe2 ligands (Figure ). The Mn–C bond distances are between 1.878(8) and 1.881(7) Å at 150 K (Table S2). The slightly distorted octahedral geometry is evident from the small differences in Mn–C bond lengths and the small deviations in the C–Mn–C bond angles from 90° (Table S2). For a more quantitative analysis of the deviation from perfect octahedral symmetry, we employed the Σ and Θ parameters, where Σ is a general measure for the deviation from ideal octahedral geometry and Θ quantifies the distortion from an octahedral toward a trigonal prismatic structure (eq S1 and Table S2).[159−161] A perfectly octahedral complex would give Σ = Θ = 0.[159−161] From the crystal structure of [Mn(CNdippPhOMe2)6]PF6, we obtain Σ = 22.55° and Θ = 58.37°, relatively close to O symmetry. For reference, the [Mn(CNPh)6]I3 compound has Σ = 25.84° and Θ = 65.51° (Table S2).[162] Furthermore, the structure of [Mn(CNdippPhOMe2)6]PF6 is also somewhat less distorted than the analogous W(0) complex [W(CNdippPhOMe2)6] (Σ = 26.52° and Θ = 66.33°, Table S2).[101]

Figure 1

Crystal structure of [Mn(CNdippPhOMe2)6]PF6. Hydrogen atoms and solvents (H2O and cyclohexane) are omitted for clarity. Color Code: C (gray), O (red), N (blue), Mn (purple), F (green), and P (orange).

The DFT geometry optimization of [Mn(CNdippPhOMe2)6]PF6 (PBEh-3c/def2-mSVP(C,H,N,O)/def2-mTZVP(Mn)) yields similar Mn–C bond lengths as the X-ray crystal structure (Table S2). Furthermore, the DFT-optimized geometry features a similar deviation from ideal octahedral geometry as the X-ray crystal structure (Σ = 20.83° and Θ = 56.41°, Table S2).

[Mn(CNdippPhOMe2)6]PF6 can be isolated as a yellow crystalline powder by filtration and subsequent drying under air. Not surprisingly, the experimental XRPD pattern of that yellow crystalline powder does not match the calculated X-ray pattern based on the single-crystal structure (Figure S7), due to the loss of crystal solvents during filtration and air drying. Combustion analysis of the crystalline yellow powder confirms that it is solvent-free.

Vibrational Spectroscopy and DFT Analysis

Infrared spectra of the complex and the free ligand were measured in the range of 400–3500 cm–1 (Figure ). The C≡N stretching frequency (υ(C ≡ N)) of isocyanides is usually strongly influenced by the metal coordination,[46,52,53,156,163−178] and this is also the case here. In the IR spectrum of the free ligand, υ(C≡N) is observed at 2116 cm–1, which is nearly the same for the analogous W-complex (2115 cm–1), whereas in the complex [Mn(CNdippPhOMe2)6]PF6, it is shifted to 2070 cm–1, due to π-backbonding from the metal center to the π* antibonding orbital of the isocyanide moiety. Furthermore, the respective IR band is significantly broader in the complex than in the free ligand, presumably due to the somewhat distorted octahedral coordination geometry resulting in slightly different metal–ligand bond lengths and angles (Table S2), and consequently to a continuum of slightly differing vibrational frequencies (see below). Another explanation for the broadening could be the fact that the symmetry of the vibrational band in the complex is T1, whereas in the free ligand the symmetry is simply A. Two prominent IR bands at 586 and 789 cm–1 appearing in the spectrum of the complex but not in the free ligand are tentatively attributed to Mn(I)-C vibrational modes, in analogy to a recently explored isoelectronic Cr(0) arylisocyanide complex, where the Cr(0)–C vibration appeared at 590 cm–1.[66]

Figure 2

Solid-state ATR-IR spectra of the compound [Mn(CNdippPhOMe2)6]PF6 (top) and the free ligand (4) (bottom). The red marked frequencies of the complex are assigned in Table .

Table 1

Selected Infrared Vibrational Frequencies for [Mn(CNdippPhOMe2)6]PF6 (Red Marked Frequencies in Figure ) in Comparison with DFT Calculated Frequencies (PBEh-3c/def2-mSVP(C,H,N,O)/def2-mTZVP(Mn))a Scaled by a Factor of 0.95[148],b

complex	mode	IRexpc (cm–1)	IRDFT (cm–1)
[Mn(CNdippPhOMe2)6]PF6	υ(C≡N) T1u	2070	2176 (2973), 2181 (2863), 2184 (1231)d and 2184 (2134)d
	υ(C≡N) Eg		2184 (1231),d 2184 (2134)d and 2188 (218)
	υ(C≡N) A1g		2261 (1.07)
	PF6–	834
	υ(Mn–C) T1u	789	789.72 (118), 789.77 (121) and 790.38 (118)
	υ(Mn–C) Eg		792.46 (0.29) and 792.70 (1.66)
	υ(Mn–C) A1g		793.12 (0.71)
	δ(C–Mn–C) and δ(Mn–C≡N) T1u	586	599.57 (99), 601.19 (169), 608.21 (41), 610.13 (168), 613.87 (91) and 618.43 (131)
[Mn(Ltri)2]PF6[53] e	υ(C≡N) T1u	2081
	Mn–C	584
[Mn(Lbi)3]PF6[53] f	υ(C≡N) T1u	2064
	Mn–C	568
[Mn(CNPh)6]PF6[46]	υ(C≡N) T1u	2088 (vs)
	PF6–	840 (m)
[Mn(CNPh)6]I/Br/Cl[46]	υ(C≡N) T1u	2088 (vs)
[Mn(CNPh)6]Cl[52]	υ(C≡N) T1u	2080g
[Mn(CNPh)6]BF4[171]	υ(C≡N) T1u	2084
[Mn(CNPh)6]I[177]	υ(C≡N) T1uh	2085 (vs) and 1993 (sh)
	υ(N(isocyanide)-C(pheny)) T1uh	1210
	δ(Mn–C≡N) T1uh	600 (vs)
	υ(Mn–C) T1uh	297 and 319 (multiplet)i
	δ(C–Mn–C) T1uh	113 (w)

The intensities of the vibrational frequencies obtained from DFT calculations (PBEh-3c/def2-mSVP(C,H,N,O)/def2-mTZVP(Mn)) are indicated in parentheses. (Abbreviations: bend., bending; str., stretching; ip, in-plane).

IR data from previously reported comparable compounds are included for comparison.

of the solid state.

One T1 and Eg modes are mixed together at 2184 cm–1 leading to two IR observable modes.

Ltri = 5,5′-(2-isocyano-5-methyl-1,3-phenylene)bis(2-(3,5-di-tert-butyl-2-isocyanophenyl)thiophene).

Lbi = 2,5-bis(3,5-di-tert-butyl-2-isocyanophenyl)thiophene.

The peak shows a shoulder at lower wavenumbers.[52]

Approximated local symmetry O.[177]

Due to a breakdown of the T1 degeneracy.[177]

For an ideal octahedral complex of the type [Mn–(C≡–N–C(Ar))6], six IR-active T1 modes are expected (υ(C≡N), υ(Mn–C), υ(N-C(Ar)), δ(Mn–C≡N), δ(C–Mn–C), and δ(C≡N-C(Ar))).[177] The infrared spectrum was analyzed only in terms of [Mn(−C≡N)] core vibrations (Table ). The core vibrations of [Mn(CNdippPhOMe2)6]+ were obtained from DFT calculations (PBEh-3c/def2-mSVP(C,H,N,O)/def2-mTZVP(Mn)) (Figures S11–S14) and compared with the experimentally obtained frequencies (Table and Figures S12–S14). In addition, we compared the assigned vibrational modes with previous reports.[46,52,53,171,177]

For the complex [Mn(CNdippPhOMe2)6]+ (approximated local symmetry O), two symmetrical Raman-active (A1 and doubly degenerate E) and one asymmetrical IR-active (triply degenerate T1) metal–ligand stretching modes (υ(Mn–C)) are expected.[177,179,180] DFT vibrational analysis reveals a splitting of the IR-active υ(Mn–C) T1 mode into three stretching modes with observable Mn motion in the x, y, and z directions for [Mn(CNdippPhOMe2)6]+ (Table and Figures S12–S14) at 789.72, 789.77, and 790.38 cm–1, probably because of the small deviation from an ideal octahedral symmetry (Table S2). These modes are forming one band at 790 cm–1 in a simulated IR spectrum (Figure S11).

A complex with O symmetry should furthermore exhibit one Raman-active (triply degenerate T2), one forbidden (triply degenerate T2), and one IR-active (triply degenerate T1) metal–ligand bending mode (δ(Mn–C≡N) as well as for δ(C–Mn–C)).[177,179,180] The calculations reveal between 599 and 618 cm–1 six-core vibration bending modes in x, y, and z directions with dominant Mn motions (Table ). Specifically, these modes are at 599.57, 601.19, 608.21, 610.13, 613.87, and 618.43 cm–1. It seems that these modes originate from the combination of the δ(Mn–C≡N) and δ(C–Mn–C) bending modes, related to the fact that the complex shows deviation from ideal octahedral geometry (Table S2). All computed modes form together one band at 610 cm–1 in a simulated IR spectrum (Figure S11) and correspond to the expected T1 bending mode.

Finally, the υ(C≡N) T1 mode is also split into a few IR-active motions in x, y, and z directions. The IR-active modes can be identified in DFT calculations at 2176 and 2181 as well as at 2184 and 2184 cm–1 due to a mixing of one T1 and one Eg motion (Table and see SI, Figures S12–S14). These modes are forming one intense broad band at 2181 cm–1 in a simulated IR spectrum (Figure S11) and correspond also to an expected T1 stretching mode.

On the basis of these DFT results supported by literature reports, the most relevant vibrational modes observed experimentally for [Mn(CNdippPhOMe2)6]PF6 can be assigned (Table ). The calculated T1 modes are split into three components due to small deviations from the O symmetry but these splittings cannot be observed experimentally. The observed band at 2070 cm–1 is attributed to the υ(C≡N) T1 mode, calculated to appear at ca. 2181 cm–1. This assignment is in agreement with several similar manganese(I) isocyanide complexes, which exhibit this mode between 2064 and 2088 cm–1.[46,52,53,171,177] The respective IR band exhibits a weak shoulder at lower wavenumbers, which is also reported in the literature, and which results from some deviation from linearity of the CNR axis at the N-atom.[52,177] The X-ray crystal structure of [Mn(CNdippPhOMe2)6]PF6 shows that such a distortion is indeed present in the solid state. Alternatively, the respective IR shoulder could stem from a CO vibration of [Mn(CO)(CNdippPhOMe2)5]PF6, for which the 55Mn NMR spectrum in Figure S6 provides some evidence (see above). The prominent broad band at 834 cm–1 is attributable to the PF6– anion, supported also by the fact that for [Mn(CNPh)6]PF6 this frequency is reported at 840 cm–1.[46] The υ(Mn–C), δ(Mn–C≡N), and δ(C–Mn–C) T1 modes resulting from the octahedral coordination skeleton of [Mn(CNdippPhOMe2)6]PF6 are observed at 789 and 586 cm–1 and can be assigned to the calculated IR-active T1 modes at ca. 790 (υ(Mn–C) T1) and ca. 610 (δ(Mn–C≡N) and δ(C–Mn–C) T1) cm–1. In the two previously investigated manganese(I) complexes with bidentate and tridentate isocyanides ligands, Mn–C vibrations at 568 and 584 cm–1 were found.[53] For [Mn(CNPh)6]I, splitting into the two T1 bending modes δ(Mn–C≡N) (600 cm–1) and δ(C–Mn–C) (113 cm–1) was reported.[177] As mentioned above, in our calculations, these two modes appear together around 610 cm–1 (see gifs in the SI).

Electrochemistry

The cyclic voltammogram of [Mn(CNdippPhOMe2)6]PF6 in dry and deaerated CH2Cl2 containing 0.1 M TBAPF6 shows an oxidation wave at 1.21 V vs SCE (E1/2), which can be attributed to the one-electron oxidation of Mn(I) to Mn(II) (Figure A). A value of E1/2 = 1.22 V was determined by differential pulse voltammetry (Figure B). In the potential range between 0 and −2.3 V vs SCE (Figure S15), no redox waves are detectable, and thus the reduction potentials of the coordinated ligands cannot be determined under these conditions. [Mn(CNdippPhOMe2)6]PF6 undergoes a second one-electron oxidation at E1/2 = 1.74 V vs SCE, likely attributable to the redox couple Mn(II/III) (Figure S15). The shape of the wave in Figure A and the peak–current ratio indicate a quasi-reversible chemical and electrochemical process. The difference between the anodic (Epa) and cathodic (Epc) peak potentials (peak-to-peak separation, ΔEp) is 92 mV (at a scan rate of 100 mV s–1) instead of the expectable 57 mV for a one-electron process at 25 °C, and the ratio of the anodic and cathodic peak currents is not equal to one (|Ipa/Ipc| ≠ 1).[181]

Figure 3

(A) Cyclic voltammogram for the Mn(I)/Mn(II) redox couple of [Mn(CNdippPhOMe2)6]PF6 in dry and deaerated CH2Cl2 containing 0.1 M TBAPF6 at 20 °C, recorded with a scan rate of 100 mV s–1. (B) Differential pulse voltammetry for the same redox couple under identical conditions.

To investigate the electrochemical reversibility of the single electron transfer process at 1.21 V in more detail, the cyclic voltammogram was measured at different scan rates (Figure S16A), and the respective data were used to make a Randles–Ševcík plot (Figure S16B).[181] The observed slight deviation from linearity in the plots of Ipa and Ipc vs υ1/2 (Figure S16B) suggests electrochemical quasi-reversibility, further confirmed by the observation that the peak-to-peak separation gets larger with increasing potential scan rate (see Trumpet plot in Figure S17).

In [Mn(CNPh)6]PF6, the quasi-reversible oxidation of Mn(I) to Mn(II) occurs at 1.01 V vs SCE in CH2Cl2 (Table ).[45,49] The E1/2 value of [Mn(CNPh)6]PF6 is close to the Mn(I/II) potentials reported for our previously investigated [Mn(Ltri)2]PF6 and [Mn(Lbi)3]PF6 compounds (1.00 and 1.05 V vs SCE in CH2Cl2; Ltri = 5,5′-(2-isocyano-5-methyl-1,3-phenylene)bis(2-(3,5-di-tert-butyl-2-isocyanophenyl)thiophene) and Lbi = 2,5-bis(3,5-di-tert-butyl-2-isocyanophenyl)thiophene).[53] In [Mn(CNMe)6]PF6, the oxidation of Mn(I) to Mn(II) occurs already at 0.47 V vs SCE in CH2Cl2 due to the stronger electron-donating character of the aliphatic isocyanide ligands compared to arylisocyanides.[45,49] For [Mn(CNMe)6]PF6 and [Mn(CNPh)6]PF6, a second one-electron wave was detected at 1.59 and 1.91 V vs SCE, respectively, and assigned to the oxidation of Mn(II) to Mn(III).[45] Against this background, it seems clear that the first and the second one-electron oxidation events observed for [Mn(CNdippPhOMe2)6]PF6 are metal-based and furthermore that the arylisocyanide ligand CNdippPhOMe2 makes oxidation easier compared to the other discussed arylisocyanide manganese(I) complexes.

Table 2

Electrochemical Potentials of [Mn(CNdippPhOMe2)6]PF6 in Comparison with Similar Compounds from the Literature

compound	E1/2 (MnI/II) vs SCE (V)	E1/2 (MnII/III) vs SCE (V)
[Mn(CNdippPhOMe2)6]PF6a	1.21	1.74b
[Mn(Ltri)2]PF6[53]  a,c	1.00
[Mn(Lbi)3]PF6[53] a,d	1.05
[Mn(CNPh)6]PF6[45] e	1.01	1.91
[Mn(CNMe)6]PF6[45] e	0.47	1.59

In (dry and deaerated) CH2Cl2 at 20 °C with TBAPF6 (0.1 M) as the supporting electrolyte.

Only observable during one cycle (irreversible).

Ltri = 5,5′-(2-isocyano-5-methyl-1,3-phenylene)bis(2-(3,5-di-tert-butyl-2-isocyanophenyl)thiophene).

Lbi = 2,5-bis(3,5-di-tert-butyl-2-isocyanophenyl)thiophene.

In CH2Cl2 (5 × 10–3 M) with TBAClO4 (0.1 M) as the supporting electrolyte.

Optical Spectroscopy

The UV–vis spectrum of [Mn(CNdippPhOMe2)6]PF6 exhibits two absorption band maxima at 278 nm and 370 nm (Figure , blue line), similar to [Mn(CNPh)6]Cl.[52] Thus, the yellow color stems from the broad absorption band maximizing at 370 nm, which tails into the violet region and which is not present in the UV–vis spectrum of the free ligand (Figure B, black trace). In CH2Cl2, the band at 278 nm has an extinction coefficient (ε) of 1.5 × 105 M –1 cm–1, whereas the band at 370 nm shows a slightly lower ε-value of 1.4 × 105M–1 cm–1. For [Mn(CNPh)6]Cl in ethanol/methanol/diethyl ether (8:2:1), an absorption band at 225 nm (ε = 7.5 × 104 M–1 cm–1) with shoulders at 234 (ε = 7.1 × 104 M–1 cm–1) and 249 nm (ε = 5.1 × 104 M–1 cm–1) was assigned to an intraligand (IL) transition (π → π*), whereas a band centered at 322 nm (ε = 6.6 × 104 M–1 cm–1) with a shoulder at 340 nm (ε = 6.1 × 104 M–1 cm–1) was attributed to MLCT (dπ → π*) transitions.[99] Our previously reported Mn(I) compounds [Mn(Ltri)2]PF6 and [Mn(Lbi)3]PF6 in CH2Cl2 showed absorption bands near 400 nm with similar extinction coefficients ((2–4) × 104 M–1 cm–1) and were likewise assigned to MLCT transitions.[53] Against this background, it seems plausible that the absorption band observable for [Mn(CNdippPhOMe2)6]PF6 at 278 nm is predominantly due to intraligand (IL) transitions (π → π*), whereas the band at 370 nm could be due to a delocalized MLCT or a mixed MLCT/IL transition.

Figure 4

UV/vis absorption spectra of [Mn(CNdippPhOMe2)6]PF6 in dry and argon-saturated (A) CH2Cl2 and (B) CH3CN before (blue) and after (green) irradiation with the Xenon lamp of (A) a spectrofluorometer or (B) a picosecond laser. The UV/vis spectrum of the free ligand in CH2Cl2 is included in (A). The absorbance axes (colored in green) apply to all measured spectra.

Continuous photoexcitation into the band at 370 nm or into its tail at 410 nm causes a decrease of the absorption band at 370 nm and leads to the appearance of a shoulder near 425 nm, both in CH3CN and in CH2Cl2 (green traces in Figure ). This behavior is observable under laser irradiation (Figure A) as well as under illumination with the Xenon lamp of a commercial fluorimeter (Figure B), suggesting that photoinduced ligand dissociation occurs.

In contrast to our previously reported manganese(I)[53] and isoelectronic chromium(0)[65,66] complexes with chelating arylisocyanide ligands,[97] [Mn(CNdippPhOMe2)6]PF6 does not show any detectable photoluminescence, neither in solution at room temperature (both in CH2Cl2 and CH3CN) nor at 77 K in a 2-methyl-THF glass. Consequently, we explored whether there are any dark (i. e., nonluminescent) states that could be detectable prior to the abovementioned photodecomposition. Light-induced ligand dissociation typically occurs from dissociative metal-centered excited states, and those states are usually nonemissive.[124,182−184] However, all our attempts to detect any such dark state with ns, ps, and fs laser transient absorption spectroscopy were unsuccessful. These findings are reminiscent of photoCORM behavior, such as observed, for example, in tricarbonyl manganese(I) complexes of the type [Mn(CO)3(Rbpy)Br] (Rbpy = 4,4′-disubstituted-2,2′-bipyridyl ligand, where R = Bu, H or CF3).[182] Upon photoexcitation, these complexes undergo loss of one CO ligand on the femtosecond time scale, followed by solvent coordination on the picosecond time scale. The proposed mechanism of CO extrusion involves ultrafast internal conversion from an initially populated MLCT excited state to a dissociative ligand-field (d–d) excited state.[182]

Photoinduced Ligand Dissociation

To investigate the photodissociation in more detail, we irradiated [Mn(CNdippPhOMe2)6]PF6 in CH2Cl2 (Figures A and S18A) and in CH3CN (Figure B) with a blue continuous-wave (cw) laser (405 nm, 526 mW). The absorbance at the excitation wavelength was adjusted close to 0.1 (Table S4), resulting in concentrations of 7.29 μM in CH2Cl2 (7.15 μM for a second measurement) and 7.13 μM in CH3CN. The measurements were done at 25 °C and the absorption spectra were measured each 10 s during continuous irradiation at 405 nm.

Figure 5

Photodegradation of (A) 7.29 μM [Mn(CNdippPhOMe2)6]PF6 in dry and argon-saturated CH2Cl2 and (B) 7.13 μM of the same compound in dry and argon-saturated CH3CN upon irradiation with a cw-laser (405 nm, 526 mW) for 30 min. Time steps between each spectrum: 10 seconds. The dips at 405 nm in (A) and (B) are artefacts caused by the cw-laser (marked by λexc and purple arrows).

Based on the series of UV–vis spectra in Figure , [Mn(CNdippPhOMe2)6]+ undergoes nearly complete photodegradation within 30 min. Aside from the very prominent absorbance decrease at 370 nm, the absorption band shift from 278 nm to 273 nm seems important, because the free ligand has an absorption band maximum at 273 nm (black trace in Figure A). Thus, the spectral evolution over time is compatible with the loss of one or more CNdippPhOMe2 ligands. Furthermore, in CH3CN, a new absorption band at 425 nm becomes increasingly prominent with increasing irradiation time. When CH2Cl2 and CH3CN solutions of [Mn(CNdippPhOMe2)6]PF6 are kept under ambient conditions without direct 405 nm irradiation, the UV–vis absorption spectra show comparatively minor changes over much longer time periods (Figure S19), compatible with the view that ligand loss is phototriggered. MS-ESI-HR measurements of a CH3CN sample prior and after photoirradiation (Figure S10 and Table S3) show that the initially dominant [M – PF6]+ mass peak (relative intensity of 85% with respect to 100% for [M – PF6 – L + CH3CN]+; L = CNdippPhOMe2) vanishes with continuing irradiation, finally resulting in relative intensities of 4.3% for [M – PF6]+ and 64% for [M – PF6 – L + CH3CN]+. Furthermore, a new mass peak attributable to [M – PF6 − 2L + 2CH3CN]+ with a relative intensity of 15.7% becomes observable. Collectively, the irradiation experiments and HR-ESI-MS measurements demonstrate that light-induced ligand dissociation from an electronically excited state is much faster than spontaneous ligand dissociation in the electronic ground state.

The prototypical d6 metal compound [Ru(bpy)3]2+ is known to undergo photodegradation as well,[4,185−187] and consequently it seemed meaningful to compare the photostabilities of [Mn(CNdippPhOMe2)6]PF6 and [Ru(bpy)3](PF6)2 in CH3CN under similar conditions (SI pages S27 to S31 and Figures S16B and S20 to S22). The key finding is that in CH3CN our manganese(I) complex undergoes photodegradation far more efficiently than [Ru(bpy)3](PF6)2 under these conditions of very intense cw-laser irradiation.

Conclusions

The primary coordination sphere of the [Mn(CNdippPhOMe2)6]+ complex comes close to O symmetry according to single-crystal X-ray diffraction, 55Mn NMR spectroscopy, IR spectroscopy, and vibrational analysis with DFT calculations. A detailed vibrational analysis permitted the identification of the IR-active and asymmetrical υ(C≡N), υ(Mn–C), δ(C–Mn–C), and δ(Mn–C≡N) vibrational modes (T1), and furthermore reveals substantial differences in the π-backbonding interaction between the two previously investigated Mn(I) complexes with chelating isocyanide ligands and the new compound studied here (Table ). Based on cyclic voltammetry, the Mn(I) center of [Mn(CNdippPhOMe2)6]PF6 is substantially more electron-rich than in previously reported hexakis(arylisocyanide)manganese(I) complexes.[45−51] Unlike our recently reported manganese(I) complexes with tris(bidendate) and bis(tridentate) coordination environments (Table ),[53,97] and unlike its third-row congener [W(CNdippPhOMe2)6],[100−105] the [Mn(CNdippPhOMe2)6]+ complex is nonemissive. Instead, photoirradiation into its lowest energy UV–vis absorption band leads to ligand dissociation, similar to what is well-known from manganese(I) carbonyl compounds.[75,117−119,121,123−130] The much weaker ligand field in our 3d6 complex compared to that experienced by W(0) in the same coordination environment likely leads to fast and efficient population of a dissociative metal-centered excited state, which represents a common challenge in first-row transition metal compounds.[56,59] The photophysical properties of the new Mn(I) complex differ substantially from those of the recently reported analogous complexes with bi- and tridentate isocyanide ligands (Table ).[53] Aside from the chelating nature of the previously used ligands, their thiophene units in the ligand skeleton likely affect the electronic structures of the resulting Mn(I) complexes.

Table 3

Comparison of [Mn(CNdippPhOMe2)6]PF6 to Related Mn(I) Compounds with Tris(bidendate) and Bis(tridentate) Coordination Environmentsa

In CH2Cl2 at 20 °C.

Thus, our study suggests that monodentate isocyanide ligands are not well suited for obtaining manganese(I) complexes with emissive MLCT excited states. Chelating arylisocyanides, as used previously for chromium(0),[65,66] manganese(I) (Table ),[53,97] and molybdenum(0),[156,188,189] seem better suited for this purpose.