A Synthetic Oxygen Atom Transfer Photocycle from a Diruthenium Oxyanion Complex

Abstract

Three new diruthenium oxyanion complexes have been prepared, crystallographically characterized, and screened for their potential to photochemically unmask a reactive Ru–Ru=O intermediate. The most promising candidate, Ru₂(chp)₄ONO₂ (4, chp = 6-chloro-2-hydroxypyridinate), displays a set of signals centered around m/z = 733 amu in its MALDI-TOF mass spectrum, consistent with the formation of the [Ru₂(chp)₄O]⁺ ([6]⁺) ion. These signals shift to 735 amu in 4*, which contains ¹⁸O-labeled nitrate. EPR spectroscopy and headspace GC-MS analysis indicate that NO₂^• is released upon photolysis of 4, also consistent with the formation of an Ru₂(chp)₄O species. Photolysis of 4 in CH₂Cl₂ at room temperature in the presence of excess PPh₃ yields OPPh₃ in 173% yield; control experiments implicate Ru₂(chp)₄O (6), NO₂^•, and free NO₃⁻ as the active oxidants. Notably, Ru₂(chp)₄Cl (3) is recovered after photolysis. Since 3 is the direct precursor to 4, the results described herein constitute the first example of a synthetic cycle for oxygen atom transfer that makes use of light to generate a putative metal oxo intermediate.

Graphical abstract

INTRODUCTION

The use of light in chemical synthesis is a topic of significant current interest.¹ We² and others³ have recently used light to access putative, highly reactive transition metal-containing intermediates. The most well-known examples are metal nitride compounds (I), which can be photochemically accessed from metal azide precursors (Scheme 1).⁴ Metal terminal oxo species (II) are related reactive intermediates that are of great importance for biological and synthetic oxidation reactions.⁵ Typically, chemical or electrochemical redox reactions^5a,6 are used to access II, but here we probe the possibility of photochemically unmasking a highly reactive species II, as in the bottom of Scheme 1.

Scheme 1.

Formation of mononuclear metal nitride and oxo species after exposure of metal azides and oxyanions to light, respectively.

Important precedents for this work include the elimination of nitrite ion from an O=Ru–ONO₂ complex to produce a dioxo Ru species⁷ and disproportionation of nitrite to form both Ru=O and Ru–NO complexes.⁸ Furthermore, Suslick⁹ and Vogler¹⁰ have reported formation and reactivity of metal oxo species by photolytic cleavage of complexes with O-coordinated oxyanions, while Newcomb¹¹ and Bakac¹² have used flash photolysis to observe similar, highly reactive species. A major difficulty hindering the synthetic utility of this approach is the question of how to regenerate the M–OEO_x precursor (Scheme 2). This transformation, shown with a question mark in Scheme 2, is problematic because substrate (S) oxidation yields a coordinatively unsaturated metal complex that is reduced by one-electron from the original M–OEO_x species. Installation of OEO_x⁻ with concomitant one-electron oxidation is problematic because the OEO_x⁻ ions themselves can act as two-electron oxidants. We report here a solution to this problem that makes use of the properties of metal-metal bonded compounds.

Scheme 2.

Possible stoichiometric cycle illustrating the synthetic limitations of current oxyanion systems.

We recently described the first examples of M–M=O species (in which M = Mo or W) and we found them to display unusual reactivity.¹³ Related Ru–Ru=O intermediates have recently been proposed to be important in sulfide oxygenation¹⁴ and water oxidation¹⁵ catalysis; furthermore, Ru₂(II/II) complexes have been shown to reversibly bind O₂.¹⁶ Putative Ru–Ru=O species are highly reactive and have not been observed or isolated. Due to their important catalytic applications and our previous expertise in the chemistry of Ru–Ru≡N intermediates, we sought to use the photochemical methods outlined above to access Ru–Ru=O species in order to explore their fundamental reactivity. We also find the Ru–Ru=O species to be unstable, but it can nonetheless be utilized in a synthetic photocycle for oxygen atom transfer.

RESULTS AND DISCUSSION

To date, only two discrete (non-polymeric) Ru₂ compounds bearing axial oxyanionic ligands are known: Ru₂(DMBA)₄(ONO₂)₂ (DMBA = N,N′-dimethylbenzamidinate),¹⁷ which contains two η¹-coordinated nitrate ligands, and Ru₂(OAc)₄(ONO₂)(H₂O),¹⁸ which forms polymeric chains upon loss of the aquo ligand. We are instead keen to explore whether a discrete, anhydrous mono-oxyanion complex with an Ru–Ru–O–EO_x structure could be prepared. As such we turned to the DPhF (DPhF = N,N′-diphenylformamidinate) and chp (chp = 6-chloro-2-hydroxypyridinate) ligands, for their steric bulk near the axial coordination site should eliminate the possibility of forming polymeric chains and only one oxyanionic ligand should bind to the Ru₂ core based on charge balance. The compounds discussed herein are given in Chart 1.

Chart 1.

Compounds discussed herein.

Entry	Complex	Ligands
1	Ru2(DPhF)4Cl
2	[Ru2(DPhF)4][NO3]
3	Ru2(chp)4Cl
4	Ru2(chp)4ONO2
5	Ru2(chp)4OClO3

Synthesis and IR Spectroscopic Characterization

Compounds 2, 4, and 5 were prepared from the chloride precursors 1 and 3 by metathesis with the appropriate Ag⁺ salts (AgNO₃ for 2 and 4; AgClO₄ for 5) as seen in Scheme 3. The IR spectrum for 2 displays a signal at 1773 cm⁻¹, which is well within the characteristic range of 1700-1800 cm⁻¹ for free nitrate anions.¹⁹ Furthermore, the MALDI-TOF mass spectrum of 2 shows an isotopic envelope indicative of the free [Ru₂(DPhF)₄]⁺ core (Figure S1), and the crystal structure of 2 consists of well-separated [Ru₂(DPhF)₄]⁺ cations and NO₃⁻ anions.²⁰ The steric bulk of the DPhF phenyl rings therefore prevents NO₃⁻ coordination. In contrast, the IR spectra of 4 and 5 display signals consistent with O-coordinated nitrate^19b and perchlorate ligands^19b,21 (1278 cm⁻¹ [ν_asym (ONO)] for 4; ν₄ = 1154, 1134, 1025 cm⁻¹, and ν₂ = 894 cm⁻¹ for 5).

Scheme 3.

Formation of nitrate complex 2 from chloride precursor 1 and oxyanion complexes 4 and 5 from chloride precursor 3.

Crystallography

Compounds 4 and 5 have been characterized by X-ray crystallography (Figures 1, 2 and Table S1). As with the chloride precursor 3²² and azide analog Ru₂(chp)₄N₃,^2d the equatorial chp ligands in 4 and 5 are bound in a (4,0) orientation, which allows for the oxyanionic ligand to bind only to the exposed Ru axial site.

Figure 1.

Thermal ellipsoid plot of 4•2CH₂Cl₂ with ellipsoids drawn at the 50% probability level. Hydrogen atoms and molecules of solvation are omitted for clarity.

Figure 2.

Thermal ellipsoid plot of 5•2CH₂Cl₂ with ellipsoids drawn at the 50% probability level. Hydrogen atoms and molecules of solvation are omitted for clarity.

The Ru–Ru distances in 4 and 5 (2.2633(3) Å and 2.2540(7) Å,²³ respectively, listed in Table 1) are in accord with all other previously characterized Ru₂⁵⁺ oxypyridinate complexes.²⁴ The Ru(2)–Ru(1)–O(5) bond angle in 4 deviates from linearity to 170.07(7)° and that in 5 deviates to 170.9(2)°, likely due to crystal packing effects. At 1.221(3) Å and 1.236(3) Å, the nitrate N–O distances (N(5)–O(6) and N(5)–O(7), respectively) in 4 are slightly shorter than, but in line with, those of an unbound nitrate anion (1.241(2) Å).²⁵ At 1.306(3) Å, the O(5)–N(5) distance is significantly longer, causing the nitrate anion to lose 3-fold symmetry due to the Ru(1)–O(5) interaction. The O(6)–N(5)–O(7) angle is 122.1(3)° and, when combined with the other O–N–O angles (118.4(3)° and 119.4(2)°), sum to 359.9°, indicating a planar nitrate group, which is oriented parallel to the Ru–Ru bond. The perchlorate anion in 5 exhibits O–Cl–O angles that range from 107.0(3)° – 112.8(4)°, which are close to the idealized geometry of 109.5° for a tetrahedral anion. The O(5)–Cl(5) distance is slightly elongated at 1.464(4) Å compared to the other O–Cl bond distances of 1.370(5) Å, 1.392(5) Å, and 1.445(6) Å.

Table 1.

Selected crystallographic bond lengths and angles for 4 and 5. E = N for 4, Cl for 5.

Compound	4	5
Ru(1)–Ru(2) Å	2.2633(3)	2.2540(7)
Ru(1)–O(5) Å	2.200(2)	2.249(4)
O(5)–E(5) Å	1.306(3)	1.464(4)
E(5)–O(6) Å	1.221(3)	1.445(6)
E(5)–O(7) Å	1.236(3)	1.370(5)
E(5)–O(8) Å	–	1.392(5)
Ru(2)–Ru(1)–O(5) (°)	170.07(6)	170.9(2)

UV/Vis Spectroscopy

The most intense absorption feature for compounds 3-5 in CH₂Cl₂ shifts from 529 to 532 to 542 nm and increases in intensity as the coordinating strength of the anion increases (5 < 4 < 3) (Figure 3). Though this electronic transition has not been definitively assigned, the variation in energy and intensity observed here suggests that it has significant LMCT character involving the axial ligand. The secondary feature in the spectrum also shifts from 666 to 678 to 689 nm in the same manner (5 < 4 < 3), but the intensity of this feature remains relatively consistent throughout the series.

Figure 3.

UV/Vis for compounds 3-5 in CH₂Cl₂. The spectrum for 3 was previously reported²² but is included here for direct comparison.

Cyclic Voltammetry

Cyclic voltammograms of 4 and 5 were recorded under two distinct sets of conditions. Solutions of 3-5 in CH₂Cl₂ with 0.1 M NBu₄PF₆ electrolyte were examined, as well as solutions of 4 in 0.1 M NBu₄NO₃ and 5 in 0.1 M NBu₄ClO₄. These varying conditions give insight into the binding of NO₃⁻ and ClO₄⁻ anions to the [Ru₂(chp)₄]⁺ core in solution. In analyzing these data, we can consider two limiting cases. If, first, the NO₃⁻ and ClO₄⁻ anions dissociate completely from 4 and 5 in CH₂Cl₂, then we would expect both compounds to have identical CV traces in 0.1 M NBu₄PF₆. If, on the other hand, we assume that NO₃⁻ and ClO₄⁻ do not dissociate at all from 4 or 5 in solution, then we would expect the CV traces of 4 with NBu₄PF₆ and NBu₄NO₃ electrolytes to be identical, and the CV traces of 5 with NBu₄PF₆ and NBu₄ClO₄ to be identical as well. The data shown in Figure 4 (see also Table 2) show clearly that neither of these limiting cases reflects reality. The fact that the CVs of 4 and 5 in NBu₄PF₆ appear distinct indicates that these solutions do not simply contain free [Ru₂(chp)₄]⁺ cations. However, the fact that the redox potentials change when the electrolyte is changed to NBu₄NO₃ or NBu₄ClO₄ indicates that both 4 and 5 undergo the following equilibrium in solution: Ru₂(chp)₄X ⇌ [Ru₂(chp)₄]⁺ + X⁻. Other diruthenium compounds show similar behavior.²⁶ Addition of excess X⁻ shifts this equilibrium to the left. This is clearly a fast equilibrium relative to the timescale of the electrochemical measurement since we do not observe distinct waves that may be assigned to Ru₂(chp)₄X and [Ru₂(chp)₄]⁺. These results indicate that both Ru₂(chp)₄X species and free X⁻ are chemically relevant on the timescale of photolysis experiments in fluid solution at room temperature (vide infra).

Figure 4.

Cyclic voltammograms of Ru₂^5/4+ couple for 3-5 versus Fc/Fc⁺. E_1/2 for each species is marked with a dashed vertical line.

Table 2.

E values for Ru₂^5/4+ couple for 1-5 vs. Fc/Fc⁺ in CH₂Cl₂ with scan rate = 100 mV/s. Electrolyte is 0.1 M.

Compound	Electrolyte	E1/2 (V)	Ec (V)	Ea (V)
1		–	−1.1#	–
2	NBu4PF6	−0.687	–	–
3	NBu4PF6	−0.525‡	−0.581	−0.469
4	NBu4NO3	−0.401*	−0.465	−0.338
4	NBu4PF6	−0.267*	−0.388	−0.146
5	NBu4ClO4	−0.179*	−0.230	−0.127
5	NBu4PF6	−0.145*	−0.197	−0.094

^*Indicates quasi-reversible.

^#Previously reported against SCE^26b and converted here to be against Fc/Fc⁺.²⁷

^‡Previously reported against Ag/AgCl²⁸ and re-measured here for direct comparison.

Photolysis Screening by Mass Spectrometry

MALDI-TOF mass spectrometry is a particularly useful method for screening photolytically-active compounds. In this case, the MALDI-TOF nitrogen laser (337 nm) is the light source, and photolysis products can be directly observed in the mass spectrum. The most prominent set of signals for 4 and 5 in MALDI-TOF mass spectrometry experiments corresponds to the Ru₂(chp)₄⁺ core with the characteristic Ru₂ isotopic distribution centered around m/z = 717 amu (Figure 5 for 4; Figure S2 for 5), indicating facile loss of the axially-bound oxyanions. In the case of 4, there is a weak Ru₂ signal centered at m/z = 733 amu, which is consistent with an [Ru₂(chp)₄O]⁺ ion, suggesting that an Ru₂O species could be formed upon photolysis of 4. This is also similar to the case of Ru₂-azide compounds that show Ru₂N signals in their mass spectra due to photolysis.² To further support this hypothesis, we used labeled ¹⁸O-nitrate (N¹⁸O₃⁻) to form Ru₂(chp)₄¹⁸ON¹⁸O₂ (4*), and we found that the Ru₂O feature shifts in the mass spectrum to m/z = 735 amu (Figure 5). This shift clearly indicates that the axial O atom in the [Ru₂(chp)₄O]⁺ ion derives from nitrate and is fully consistent with photo-dissociation of NO₂^• or NO₂⁻ from 4. The m/z = 733 amu feature is present but barely discernable in the spectrum of 5 (Figure S2), suggesting to us that perchlorate may not be as suitable a precursor for the formation of an Ru₂O intermediate. Therefore only 4 was investigated further.

Figure 5.

MALDI-TOF mass spectrum for 4 (black, top). Simulation (red) indicates isotope pattern at m/z = 733 amu is due to [Ru₂(chp)₄O]⁺. Upon isotopic labeling, MALDI-TOF mass spectrum for 4* (black, below) shifts by 2 units, as confirmed by simulation (red, bottom).

In photolysis experiments of 4 there are two possible sites at which bonds could break homolytically: (a) the Ru–ONO₂ bond or (b) the RuO–NO₂ bond (Scheme 4). Previous studies have indicated that either site is possible.^11a,11b,29 Heterolytic RuO–NO₂ bond cleavage is another possible pathway (pathway c in Scheme 4), which has been demonstrated for perchlorate compounds.^11b Both pathways b and c would explain our mass spectrometry results. Upon homolysis (pathway b), the resulting radical species NO₂^• should be EPR active, unlike its Ru₂O counterpart Ru₂(chp)₄O (6), which is anticipated to have a ground state of S = 0.³⁰ Heterolysis (pathway c), on the other hand, would give rise to an EPR-active Ru₂O⁺ species. Thus, we decided to use EPR spectroscopy to assess the mechanism of photolysis for 4.

Scheme 4.

Possible products upon exposing nitrate complexes to photolytic conditions.

Photolysis of 4 in Frozen Solution at 77 K and in Fluid Solution at Room Temperature

The EPR spectra of 2, 4, and 5 (Figures S3, 6, and S4, respectively) were measured at 10 or 15 K in a frozen solution of CH₂Cl₂. Though electrochemical measurements (vide supra) suggest the presence of an equilibrium between Ru₂–bound and –unbound nitrate species, the 10 K EPR spectrum of 4 can be modeled with a single axial signal, with effective g values of 3.86 and 1.95, consistent with an S = 3/2 ground state where D ≫ hν (Table 3). Thus, the nitrate-bound 4 is most likely the dominant species in solution at this temperature, and its prominent EPR signal can be used to assess the efficacy of photolysis of 4 in frozen solution.

Table 3.

EPR spectral simulations for 2, 4, and 5.

Compound	2	4	5
g⊥	2.125	2.01	2.00
g║	1.94	1.95	1.93
E/D	0.006	0.035	0.033
HStraina	850, 350, 450	1200, 900, 1200	1550, 850, 1100

^aHStrain accounts for anisotropic line broadening due to unresolved hyperfine coupling

Exposure of a frozen CH₂Cl₂ solution of 4 to 350 nm wavelength light at 77 K for 8 hours yields no change in the intensity of the EPR signal of 4. Switching to 254 nm light and photolyzing for 16 hours at 77 K yields the EPR spectrum shown in Figure 7. There is a slight decrease in the signal intensity from the S = 3/2 signal of 4 and, more significantly, a new feature in the g = 2 region is present. This new signal is an isotropic 1:1:1 triplet, clearly indicating a radical species with hyperfine coupling involving an I = 1 ¹⁴N nucleus, and it compares favorably to known literature values for NO₂^• versus those of NO₃^• (Figure 7, inset).³¹ The presence of NO₂^• suggests that, though photolysis of 4 proceeds in low yield at 77 K, homolytic cleavage of the O(5)–N(5) nitrate bond is the primary photolysis pathway (pathway b in Scheme 4).

Figure 7.

EPR spectrum of 4 in CH₂Cl₂ taken at 10 K after 16 hours of frozen photolysis using 254 nm light. Inset: simulations³¹ of NO₃^• (blue, above) and NO₂^• (red, middle) compared to 4 after frozen photolysis (black, below).

In order to improve the yield of photolysis, subsequent experiments were performed on fluid solutions of 4 in CH₂Cl₂ at room temperature under N₂. Under these conditions, photolysis is complete in a shorter time period of 4 hours using 350 nm wavelength light. Using the isotopically-enhanced analog 4*, GC-MS analysis of the headspace of the reaction confirms formation of N¹⁸O₂^• (m/z = 50) in significant quantities (Figure 8), further supporting the homolytic cleavage mechanism (pathway b in Scheme 3) suggested by our EPR results.

Figure 8.

GC-MS headspace analysis for the formation of N¹⁸O₂^• (m/z = 50) after photolysis of 4* at room temperature under N₂ for 4 hours using 350 nm wavelength light. As compared to the counts for m/z = 50 amu for the N₂ control, N¹⁸O₂^• is clearly formed under reaction conditions.

Oxygen Atom Transfer via Room Temperature Photolysis of 4

Room temperature fluid solution photolytic experiments were also performed in the presence of an excess of the well-known oxygen atom acceptor PPh₃ and monitored for the formation of oxygen atom transfer products. A CH₂Cl₂ solution of 4 and PPh₃ (20 eq. vs. 4) was photolyzed for 4 hours using 350 nm light at room temperature under N₂. Analysis by ³¹P{¹H} NMR spectroscopy indicates the formation of the intermolecular oxygen atom transfer product OPPh₃ in > 100% yield (determined by ³¹P NMR integration against a standard of PPh₄Cl) (Figure 9). The identity of this product was confirmed by doping with an authentic sample of OPPh₃ (Figure S5) and is further corroborated by ESI mass spectral data, which shows that the product formed in this reaction with 4 has a mass of m/z = 279 amu (consistent with OPPh₃) (Figure S6) and shifts by 2 mass units to m/z = 281 amu (consistent with ¹⁸OPPh₃) upon using 4* (Figure S7).

Figure 9.

Yield of OPPh₃ after exposure to different oxygen atom sources, both with and without exposure to photolytic conditions.

The > 100% yield of OPPh₃ implies that more than one active oxidant is formed under these conditions. As mentioned above, some amount of free NO₃⁻ is expected to be present based on our electrochemistry results. Upon UV irradiation at ~ 300 nm, unbound nitrate is known to eliminate either ³P O or O^{− 32} or to isomerize to peroxynitrite.³³ To test the possibility that unbound nitrate is the sole oxygen atom source, photolysis of NBu₄NO₃ was performed under identical conditions as previously expressed. OPPh₃ is indeed generated but in less than one-third the yield as when 4 was used (Figure 9). We also considered the possibility that the Ru₂ complex could simply act as a photosensitizer for free NO₃⁻. A control reaction was therefore performed in which NBu₄NO₃ and PPh₃ were photolyzed in the presence of Ru₂(chp)₄Cl (3) as a photo-sensitizer (Figure 9). OPPh₃ is generated in better yields here than without 3, but this result still does not match our findings with 4. Therefore, OPPh₃ can be formed from free NO₃⁻ anion but is formed more efficiently from 4.

It is conceivable that O₂ could be formed from either nitrate photolysis or bimetallic reductive coupling of Ru₂O intermediates and could be the oxygen atom source for formation of OPPh₃. To test this possibility, 4* was photolyzed in a CH₂Cl₂ solution at room temperature and analysis of the reaction headspace was performed using GC-MS. These results indicate that ¹⁸O₂ is not formed (Figure S8). Another potential oxygen atom source is the photolysis byproduct NO₂^•.^9b,10 To probe this possibility, PPh₃ was exposed to NO₂^•. ESI mass spectrometry data indicate that the resulting product is indeed OPPh₃ (Figure S9). To see whether NO₂^• is fully consumed under our reaction conditions, 4 was photolyzed in a CH₂Cl₂ solution containing 20 equivalents of PPh₃ at room temperature, and analysis of the reaction headspace was performed using GC-MS (Figure S10). Free NO₂^• is still detected under these conditions, though in smaller yields than without PPh₃ present. Thus NO₂^• is likely to be an oxygen atom source in this system but clearly cannot be the sole source. We therefore propose that the oxygen atom transfer in this system derives from three active oxidants: free nitrate ion, NO₂^•, and an Ru–Ru=O intermediate, giving yields of OPPh₃ greater than 100%.

Finally, we were interested in determining the identity of the resulting Ru₂ species after completion of room temperature photolysis in the presence of PPh₃. Only one Ru₂ product was observed and isolated. MALDI-TOF mass spectrometry and UV/Vis confirm this product to be the precursor 3. This result was surprising to us since oxygen atom transfer from an Ru–Ru=O species should yield an Ru₂(II/II) product such as Ru₂(chp)₄PPh₃, a species that we have recently characterized.³⁴ However, low-valent metal-metal bonded dimers are known to react with light and halogenated solvents to yield one-electron oxidized compounds bearing an axial halide ligand.³⁵ This reaction sequence explains the formation of 3, as shown in Scheme 5. As emphasized by Scheme 5, our results present a complete synthetic photocycle for oxygen atom transfer from a photoactive oxyanion complex.

Scheme 5.

Proposed mechanistic scheme for the oxygen atom transfer reaction presented here. Proposed intermediates 6 and Ru₂(chp)₄ are in dashed boxes.

CONCLUSIONS

Whereas the [Ru₂(DPhF)₄]⁺ core is not sterically accessible to bind non-linear anions as ligands, the [Ru₂(chp)₄]⁺ core does support the formation of discrete, mono-oxyanion complexes with η¹-nitrate and -perchlorate anions. Mass spectrometry, EPR, and GC-MS analysis data suggest that photolysis of 4 proceeds to generate Ru–Ru=O species 6 and NO₂^•. Room temperature photolysis of 4 allows for oxygen atom transfer to PPh₃ to form OPPh₃ in a synthetic photocycle that regenerates 3. We therefore propose that PPh₃ is directly oxidized by a combination of three active oxidants: photosensitized NO₃⁻, reactive intermediate 6, and additionally by NO₂^•. This work demonstrates for the first time that the generation of a reactive metal oxo species from an oxyanion complex can be incorporated into a synthetic cycle. Further exploitation of these reaction conditions, along with efforts to stabilize and characterize the putative Ru–Ru=O intermediate are underway in our lab.

EXPERIMENTAL SECTION

General Methods

All syntheses were conducted under a dry N₂ atmosphere using Schlenk line techniques unless otherwise noted; product workup and isolation were achieved in air unless otherwise noted. Dichloromethane (CH₂Cl₂) was dried with CaH₂ and distilled before use. Hexanes were obtained from a Vacuum Atmospheres Solvent System and degassed prior to use. All materials were commercially available and used as received, unless otherwise noted. Compounds 1^26b,36 and 3^2d,22 were prepared according to literature procedures. Photolysis of frozen samples of 4 were performed in a Rayonet RPR-200 photochemical reactor with light from 254 and 350 nm mercury vapor lamps.

[Ru₂(DPhF)₄][NO₃] (2)

Ru₂(DPhF)₄Cl (200 mg, 0.196 mmol, 1 eq.) was dissolved in CH₂Cl₂ and added to a suspension of AgNO₃ (500 mg, 2.94 mmol, 15 eq.) in 3 mL of THF in air. The color of the mixture changed from green to purple-blue almost instantaneously and was allowed to continue stirring at RT overnight. The solids were filtered off and the remaining solution was left to evaporate slowly to yield dark purple crystals of the product. Yield: 180 mg, 87.7%. MW: 1045.10 g mol⁻¹. MALDI-TOF (m/z): ([M – NO₃]⁺) 984. IR (ATR): 3053, 1773 [ν_sym NO₃], 1591, 1520, 1486, 1449, 1364, 1349, 1314, 1214, 1175, 1156, 1078, 1027, 936, 827, 761, 693, 668, 658, 619. UV-Vis (CH₂Cl₂):. λ_max(ε) = 437 (4250), 527 (5950), 681 (7170 mol⁻¹ L cm⁻¹). [C₅₂H₄₄N₉O₃Ru₂•CH₂Cl₂•2H₂O]: calcd. C 54.59, H 4.32, N 10.81; found C 54.31, H 3.88, N 11.37. Crystals suitable for X-ray diffraction were grown via slow evaporation from a concentrated solution of CH₂Cl₂ at room temperature.

Ru₂(chp)₄ONO₂ (4)

Ru₂(chp)₄Cl (200.0 mg, 0.266 mmol, 1 eq.) and AgNO₃ (54.6 mg, 0.321 mmol, 1.2 eq.) were dissolved in 30 mL of freshly distilled CH₂Cl₂ and allowed to stir for 96 h at RT under static N₂. A white precipitate (AgCl) formed. The reaction mixture was filtered through a fine sintered glass frit. Excess CH₂Cl₂ was washed through the frit until it was no longer colored. The filtrate was removed under reduced pressure, and the resulting purple solid was washed with hexanes and collected. Yield: 195.5 mg, 94.4%. MW: 778.29 g mol⁻¹. MALDI-TOF (m/z): ([M – NO₂]⁺) 733, ([M – ONO₂]⁺) 717. IR (ATR): 3108, 2964, 1596, 1534, 1465, 1434, 1390, 1350, 1278 [ν_asym (ONO)], 1262, 1181, 1085, 1011, 964, 942, 931, 866, 795, 789, 724, 630 cm⁻¹. UV-Vis (CH₂Cl₂): λ_max(ε) = 532 (3890), 678 (1520 mol⁻¹ L cm⁻¹). [C₂₀H₁₂Cl₄N₅O₇Ru₂]: calcd. C 30.86, H 1.55, N 8.99; found C 30.64, H 1.44, N 8.81. Crystals suitable for X-ray diffraction were grown from a saturated CH₂Cl₂ solution at −80 °C.

Ru₂(chp)₄OClO₃ (5)

Ru₂(chp)₄Cl (200.0 mg, 0.266 mmol, 1 eq.) and AgClO₄ (66.7 mg, 0.322 mmol, 1.2 eq.) were dissolved in 30 mL of freshly distilled CH₂Cl₂ and allowed to stir for 96 h at RT under static N₂. A white precipitate (AgCl) formed. The reaction mixture was filtered through a fine sintered glass frit. Excess CH₂Cl₂ was washed through the frit until it was no longer colored. The filtrate was concentrated under reduced pressure, and the resulting purple solid was washed with hexanes and collected. Yield: 205.3 mg, 94.6%. MW: 815.74 g mol⁻¹. MALDI-TOF (m/z): ([M – ClO₃]⁺) 733, ([M – OClO₃]⁺) 717. IR (ATR): 3107, 1596, 1536, 1433, 1391, 1337, 1263, 1154 [ν₄, 1134 [ν₄, 1075, 1025 [ν₄, 1013, 932, 894 [ν₂, 790, 723, 705, 668, 632, 610 cm⁻¹. UV-Vis (CH₂Cl₂): λ_max(ε) = 529 (2150), 666 (1310 mol⁻¹ L cm⁻¹). [C₂₀H₁₂Cl₅N₄O₈Ru₂]: calcd. C 29.45, H 1.48, N 6.87; found C 29.20, H 1.68, N 6.26. Crystals suitable for X-ray diffraction were grown from a saturated CH₂Cl₂ solution layered with hexanes at room temperature. Safety warning: anhydrous metal perchlorate complexes are potentially explosive.

AgN¹⁸O₃. 0.1 g HN¹⁸O₃ (95 atom %, 65 weight % in H₂¹⁸O), 0.1 g H₂¹⁸O, Ag powder, and a stir bar were added to a vial that was subsequently capped. This heterogeneous mixture was stirred at 50 °C in air for 48 hours; periodically the mixture was manually rotated and the solvent mixture forced to the bottom of the vial as it had crept up the walls of the vial. A white solid became visible as the reaction reached completion, which was re-dissolved in the H₂¹⁸O. The supernatant liquid was then decanted away (using a pipette) from the heterogeneous solid mixture into a new vial and allowed to evaporate to dryness, leaving a crystalline product, which was dried under vacuum at 50 °C overnight. Yield: 126.8 mg (76.5%). IR (ATR): 1676, 1657 [ν_asym (¹⁸ON¹⁸O)], 1293 [ν_sym (¹⁸ON¹⁸O)], 792 [ν (N¹⁸O)], 694 cm⁻¹.

Ru₂(chp)₄¹⁸ON¹⁸O₂ (4*)

AgN¹⁸O₃ was ground into fine powder and then used in an identical procedure as 4. Yield: 141.2 mg, 67.7%. MW: 784.29 g mol⁻¹. MALDI-TOF (m/z): ([M – N¹⁸O₂]⁺) 735. IR (ATR): 3108, 2964, 1596, 1534, 1465, 1434, 1390, 1350, 1253 [ν_asym (¹⁸ON¹⁸O)], 1262, 1181, 1085, 1011, 942, 931, 866, 795, 789, 724, 630 cm⁻¹.

Solution-phase Reactions

A 0.5 mM CH₂Cl₂ solution of the nitrate source (4, 3 + NBu₄NO₃, or NBu₄NO₃) with a 20-fold excess of PPh₃ was prepared on a 30 mL scale in a quartz Schlenk tube (flask “A”). A 10 mL aliquot was transferred to a different flask (flask “B”), which was wrapped in foil, and allowed to stir under static N₂ for 4 hours without exposure to photolytic conditions. The remaining 20 mL solution was photolyzed in flask “A” under static vacuum at room temperature using 350 nm light for 4 hours. After this time a 10 mL aliquot was removed and transferred to a 25 mL Schlenk flask (flask “C”). Solutions in flasks “B” and “C” were reduced to dryness under vacuum and analyzed by NMR with a known quantity (1 eq. based on 4) of PPh₄Cl for use as a comparative standard. ³¹P{¹H} NMR (298 K, 400 MHz, CDCl₃): δ 29 (OPPh₃), 23 (PPh₄Cl), −5.5 (PPh₃).

Physical Measurements

Matrix-assisted laser desorption/ionization (MALDI) time-of-flight (TOF) mass spectrometry data were obtained using an anthracene matrix on a Bruker ULTRAFLEX® III mass spectrometer equipped with a SmartBeam® laser in positive ion detection mode. ESI mass spectrometry data was collected on a Thermo Q Exactive Plus™ mass spectrometer and GC-MS data were collected on a Shimadzu GCMS-QP2010 Ultra spectrometer. ¹H and ³¹P{¹H} data were obtained using a Bruker Avance III 400 MHz spectrometer. UV/Vis spectra were obtained using a StellarNet Miniature BLUE-wave UV-Vis dip probe with a Tungsten-Krypton light source and a 10 mm path length tip. IR spectra were taken on a Bruker Tensor 27 spectrometer using an ATR adapter (no matrix). Cyclic voltammograms were taken on a BASi Potentiostat using Epsilon software in CH₂Cl₂ solutions with 0.1 M electrolyte and 1.0 mM substrate. The electrodes were as follows: glassy carbon (working), Pt wire (auxiliary) and Ag/Ag⁺ in CH₃CN (reference). The potentials were referenced versus the ferrocene/ferrocenium redox couple by externally added ferrocene. Elemental analysis was performed by Midwest Microlab, LLC in Indianapolis, IN, USA.

EPR Spectroscopy

EPR data were acquired on a Bruker ELEXSYS E500 EPR spectrometer equipped with a Varian E102 microwave bridge interfaced with a Linux system. An Oxford Instruments ESR-900 continuous-flow helium flow cryostat and an Oxford Instruments 3120 temperature controller were used to control the sample temperature. A Hewlett-Packard 432A power meter was used for microwave power calibration, with measurement conditions as follows: for 2 – 9.3762 GHz, 4 G modulation amplitude, 2500 G center field, 5000 G sweep width, 5.024 mW power, 55 dB gain, 327.68 ms time constant, 10 ms conversion time, and 15 K; for 4 – 9.3765 GHz, 4 G modulation amplitude, 2500 G center field, 5000 G sweep width, 5.024 mW power, 60 dB gain, 655.36 ms time constant, 10 ms conversion time, and 10 K; for 4-hν/NO₂^• – 9.3765 GHz, 4 G modulation amplitude, 2500 G center field, 5000 G sweep width, 5.024 mW power, 60 dB gain, 655.36 ms time constant, 10 ms conversion time, and 10 K; for 5 – 9.3832 GHz, 4 G modulation amplitude, 2500 G center field, 5000 G sweep width, 5.024 mW power, 55 dB gain, 655.36 ms time constant, 10 ms conversion time, and 10 K. Spectral simulations were performed using the program EasySpin.³⁷

X-ray Crystallographic Data Collection and Structure Determination

Crystallographic data were measured at the Molecular Structure Laboratory of the Chemistry Department of the University of Wisconsin–Madison. Suitable crystals of 4 and 5 were selected under oil and ambient conditions. For 4, a purple block shaped crystal with dimensions 0.728 × 0.374 × 0.216 mm³ was selected, and for 5 a purple plate crystal with dimensions 0.166 × 0.155 × 0.094 mm³ was chosen. The crystals were attached to the tip of a MiTeGen MicroMount©, mounted in a stream of cold nitrogen at 100(1) K, and centered in the X-ray beam using a video monitoring system. The crystal evaluation and data collection were performed on a Bruker Quazar SMART APEX-II diffractometer with Mo-Kα (λ = 0.71073 Å) radiation. The data were collected using a routine to survey the reciprocal space to the extent of a full sphere to a resolution of 0.70 Å for 4 and 0.80 Å for 5 and were indexed by the APEX program.³⁸ The structures were solved via direct methods and refined by iterative cycles of least-squares refinement on F² followed by difference Fourier synthesis. All hydrogen atoms were included in the final structure factor calculation at idealized positions and were allowed to ride on the neighboring atoms with relative isotropic displacement coefficients. Absorption corrections were based on a fitted function to the empirical transmission surface as sampled by multiple equivalent measurements.³⁹ The systematic absences in the diffraction data were uniquely consistent with the space groups P2₁/c for 4 and P2₁/n for 5, yielding chemically reasonable and computationally stable results of refinement, and both structures were solved using direct methods using XS software.⁴⁰ Compound 4 is a pseudo-merohedral twin with a twin component ratio of 54:46. The twin components are related by a 180° rotation about [1 0 0]. Compound 5 is a non-merohedral twin with a twin component ratio of 60:40. The twin components are related by 180° a rotation about [0 0 1].

Figure 6.

EPR spectrum and simulation of 4 recorded at 10 K.