A Flexible Synthetic Strategy for the Preparation of Heteroleptic Metallacycles of Porphyrins

We present a stepwise synthetic strategy for the preparation of the unprecedented heteroleptic 2+2 neutral metallacycle [{t,c,c-RuCl2(CO)2}2(4′cisDPyP)(3′cisDPyP)] (5), in which two different 5,10-meso-dipyridylporphyrins, 4′cisDPyP [i.e., 5,10-bis(4′-pyridyl)-15,20-diphenylporphyrin] and 3′cisDPyP [i.e., 5,10-bis(3′-pyridyl)-15,20-diphenylporphyrin], are joined through equal 90°-angular Ru(II) connectors. The synthesis of 5 was accomplished through the preparation of a reactive ditopic intermediate in which one of the two pyridylporphyrins is linked to two neutral ruthenium fragments, each having one residual readily available coordination site (a dmso-O). Thus, compound 5 was obtained under mild conditions through two complementary routes: either by treatment of [{t,c,c-RuCl2(CO)2(dmso-O)}2(4′cisDPyP)] (3) with 1 equiv of 3′cisDPyP or, alternatively, by treatment of [{t,c,c-RuCl2(CO)2(dmso-O)}2(3′cisDPyP)] (4) with 1 equiv of 4′cisDPyP. Heteroleptic metallacycle 5 was isolated in pure form in acceptable yield and fully characterized. Spectroscopic data and a molecular model show that 5 has an L-shaped geometry, with the two porphyrins almost orthogonal to one another. The modular approach that we established is highly flexible and opens the way to several possible exciting developments.


■ INTRODUCTION
Nature uses sophisticated arrays of tetrapyrrolic macrocycles (e.g., chlorophyll, cytochromes, etc.) to perform precise energy and electron transfer processes. In addition to the specific nature of the macrocycles, their number and relative orientations are of paramount importance for determining the properties of such assemblies. The development of simple procedures for preparing synthetic arrays of tetrapyrrole macrocycles with full stereocontrol is one of the challenges of supramolecular chemistry.
The corresponding zincated metallacycles 1Zn and 2Zn, in which each embedded metal center is capable of forming an additional axial bond, were exploited by us as two-point molecular panels (Figure 1) for the preparation of molecular sandwiches, boxes, and prisms. 25,27,28 A vast majority of the metal-mediated arrays of porphyrins described in the literature, including metallacycles 1 and 2, are homoleptic systems, as they contain a single type of porphyrin. With few exceptions, they are also homometallic, because most of them contain a single type of metal connector. In the past, we described the stepwise preparation of the heterobimetallic 2+2 metallacycle of porphyrins [Pd(dppp){t,c,c-RuCl 2 (CO) 2 (4′cisDPyP) 2 }](CF 3 SO 3 ) 2 [dppp = 1,3-bis-(diphenylphosphanyl)propane] that features an octahedral neutral Ru(II) complex at one corner and a square planar cationic Pd(II) linker at the other ( Figure 2). 29 In the solid state, the metallacycle, obtained by treatment of the reactive 90°metal-containing ligand t,c,c-[RuCl 2 (CO) 2 (4′-cisDPyP) 2 ] with the cis-protected complex [Pd(dppp)(CF 3 SO 3 ) 2 ], was found to have a folded geometry in which the two porphyrins form a dihedral angle of 41.7(1)°.
More recently, we described new stereoisomers of homoleptic molecular square 1, in which one or both {t,c,c-RuCl 2 (CO) 2 } corners are replaced by the stereoisomeric {c,c,c-RuCl 2 (CO) 2 } fragment, which can have a C or A configuration, namely, [{t,c,c-RuCl 2 (CO) 2 }(4′cisDPyP) 2 {c,c,c-RuCl 2 (CO) 2 }] and [c,c,c-RuCl 2 (CO) 2 (4′cisDPyP)] 2 . 30 The examples of metal-mediated assemblies containing different porphyrins are even rarer. In 1998, Drain and coworkers described the high-yield one-pot formation of a nonameric neutral square-shaped grid obtained by addition of 12 equiv of [PdCl 2 (NCPh) 2 ], a precursor of the linear {trans-PdCl 2 } linker, to a 1:4:4 mixture of three different meso-4′pyridylporphyrins: 16 one X-shaped 4′-TPyP unit (taking the central position; 4′-TPyP = 5,10,15,20-tetrapyridylporphyrin), four T-shaped 4′-TrPyP units [making the sides of the array; 4′-TrPyP = 5,10,15-tris(4′-pyridyl)-20-phenylporphyrin], and four angular 4′cisDPyP units (making the corners). 31 The adduct, which was not isolated, was believed to be the thermodynamic product, and the reversibility of the pyridyl− Pd(II) bond was apparently a critical feature for the high selectivity of the preparation. On the contrary, with a similar one-pot approach, Pd(II)-or Pt(II)-mediated one-dimensional (1D) and 2D "tapes" of pyridylporphyrins were obtained with much lower selectivities, as mixtures of adducts with different nuclearities. 31 The same group later described a mixed porphyrin/porphyrazine assembly obtained by postsynthetic modification. 32,33 More recently, Schmittel and co-workers, exploiting the HETTAP concept for controlling the coordination equilibrium at the metal ion, 34 prepared Cu-mediated assemblies containing two different cofacial metalloporphyrins. 35,36 It should be noted that in this case the two porphyrins have very different peripheral binding sites, one being a pyridyl ring and the other a sterically shielded phenanthroline, and the strategy exploits steric and electronic effects originating from the latter.
Given these premises, we aimed to develop a new flexible synthetic strategy that might be used for the construction of ruthenium-mediated heteroleptic systems of meso-pyridylporphyrins containing PyPs that differ in the number of peripheral pyridyl rings (from two to four) and/or in the position of the pyridyl N atom (3′ or 4′). This would open the way to new extended arrays as well as to unprecedented geometries. The formation of Ru-mediated metallacycles of PyPs typically occurs under kinetic control; therefore, a one-pot synthetic approach does not seem to be suitable for our aim. In fact, because all PyPs share the pyridyl ring as a common binding motif, no stereoelectronic discrimination upon coordination to ruthenium can be expected to occur. Thus, we decided to develop a new stepwise modular approach that requires the initial preparation of a reactive polytopic "acceptor" Figure 1. Homoleptic 2+2 metallacycles 1 (top) and 2 (bottom) with their schematic representations as molecular panels (right). Distances from the X-ray structures. 26 Color code: red for 4′cisDPyP, blue for 3′cisDPyP, gray for {t,c,c-RuCl 2 (CO) 2 }, and gold for the centroid of the porphyrin. . Distances from the X-ray structure. 29 Color code: red for 4′cisDPyP, gray for {t,c,c-RuCl 2 (CO) 2 }, green for {Pd(dppp)} 2+ , and gold for the centroid of the porphyrin.
Inorganic Chemistry pubs.acs.org/IC Article intermediate, i.e., a pyridylporphyrin bound to at least two ruthenium fragments, each having one residual readily available coordination site, that is then reacted with the second porphyrin to yield the final heteroleptic assembly. Dimetallic acceptors, not porphyrin-based and typically with a 2+ charge, have been extensively used by several groups for self-assembly purposes. For example, Stang and co-workers exploited di-Pt(II) acceptors of different shapes for the preparation of many metallacycles and metallacages. 37 Dimetallic molecular clips were used as pillars for the preparation of metallacages featuring two equal face-to-face flat organic linkers, including porphyrins. 1,37−44 As a proof of concept, this strategy was here first used for the stepwise preparation of the heteroleptic 2+2 neutral metallacycle [{t,c,c-RuCl 2 (CO) 2 } 2 (4′cisDPyP)(3′cisDPyP)] (5), in which two different cis-dipyridylporphyrins, 4′cisDPyP and 3′cisDPyP, are joined through equal 90°-angular Ru(II) connectors. The synthesis of 5 required the preparation of the reactive ditopic intermediate [{t,c,c-RuCl 2 (CO) 2 (dmso-O)} 2 (4′cisDPyP)] (3), which already contains one of the two porphyrins, that was then treated with 1 equiv of 3′cisDPyP (Scheme 1). Alternatively, compound 5 was obtained by treatment of the reactive intermediate [{t,c,c-RuCl 2 (CO) 2 (dmso-O)} 2 (3′cisDPyP)] (4) with 1 equiv of 4′cisDPyP (Scheme 1). Heteroleptic metallacycle 5 was isolated in pure form in reasonable yield and fully characterized. 45 Spectroscopic data and the molecular model are consistent with 5 having an unprecedented L-shaped geometry, with the two porphyrins almost orthogonal to one another.

■ RESULTS AND DISCUSSION
Model Systems. Because the purification of cisDPyPs is a laborious process, we first ascertained that the two dmso-O ligands in the Ru(II)-dmso carbonyl complex t,c,c- Compound 3 can be isolated in almost pure form by dissolving 3H 2 O in a chloroform/DMSO mixture followed by precipitation with diethyl ether. Also in this case, the coordinated water is not fully replaced by dmso-O. In fact, the NMR spectrum of this species recorded in CDCl 3 ( Figure  3) shows, in addition to the major set resonances of 3, a minor set of partially resolved signals, attributed to the residual aqua species 3H 2 O. The equilibrium between these two species is shifted toward 3 by the addition of DMSO-d 6 , and the resonances of 3H 2 O concomitantly disappear ( Figure 3). Because from the point of view of the further reactivity 3H 2 O and 3 behave equally, and the influence on the MW and stoichiometric ratio of the reactions is marginal, for the sake of brevity in the following no distinction between them will be made and only 3 will be used.
The corresponding ditopic intermediates with 5,10-bis(4′pyridyl)-15,20-di-p-(tolyl)porphyrin (4′cisDPyMP, 3Me) and with 3′cisDPyP (4) were prepared with the same procedure (Supporting Information). 49 Synthesis and Characterization of Heteroleptic 2+2 Metallacycle [{t,c,c-RuCl 2 (CO) 2 } 2 (4′cisDPyP)(3′cisDPyP)] (5). Treatment of 3′cisDPyP with a slight excess of 3 at 40°C in dried CH 2 Cl 2 afforded heteroleptic 2+2 metallacycle [{t,c,c-RuCl 2 (CO) 2 } 2 (4′cisDPyP)(3′cisDPyP)] (5). The reaction was monitored by TLC and stopped when depletion of 3′cisDPyP was complete (∼1 h). The crude was purified by flash chromatography, and a single fraction was collected, with an isolated yield of pure product of 26%. To the best of our knowledge, this is the first example of an isolated heteroleptic metallacycle that features two porphyrins with almost indistinguishable binding sites. In the reasonable hypothesis that each porphyrin maintains a coordination geometry similar to that found in homoleptic metallacycles 1 and 2, 25,26 i.e., either coplanar (4′cisDPyP) or perpendicular (3′cisDPyP) to the equatorial coordination plane (N, N, C, and C) of each Ru linker, 2+2 metallacycle 5 is expected to have a folded, L-shaped geometry in which the planes of the two porphyrins are almost orthogonal to one another. Even though we were unable to grow crystals of 5 suitable for X-ray analysis, its energy-minimized model ( Figure  4) is perfectly consistent with this hypothesis. The average planes of the porphyrins form a dihedral angle of 92.6°. The model also shows that the 4′cisDPyP side of the L is significantly longer than the 3′cisDPyP side; in fact, the calculated distance from the intersection between the average planes of the two porphyrins (that falls into a pyrrole ring of 3′cisDPyP) to the centroid of 4′cisDPyP, 9.36 Å, is more than double that to the centroid of 3′cisDPyP, 3.95 Å (Figure 4).
The 1 H NMR spectrum of 5 is reported in Figure 5. Assignments were made according to 2D spectra (Supporting Information) and by comparison with homoleptic metallacycles 1 and 2.
In the high-frequency region, the spectrum displays two sets of equally intense signals typical of 4′cisDPyP and 3′cisDPyP coordinated in a symmetrical fashion to two {t,c,c-RuCl 2 (CO) 2 } fragments. The two doublets at 9.70 and 9.38 ppm were assigned to H2 and H6, respectively, of 4′cisDPyP. In fact, the meso six-membered rings typically lay approximately perpendicular to the porphyrin ring and are in slow rotation about the C meso −C ring bond; therefore, when the plane of the porphyrin is not a plane of symmetry for the molecule, as in the case of 5, the protons on the two sides have distinct resonances. Consistent with this hypothesis, the two doublets are correlated by a clear exchange cross peak in the 1 H− 1 H ROESY spectrum (Supporting Information). The doublet at 9.38 ppm was attributed to H6, i.e., the proton that in our scheme lays above the average plane of 4′cisDPyP and partially falls into the shielding cone of the adjacent 3′cisDPyP. Protons H3 and H5 are more distant from the other porphyrin, so their signals [identified through the COSY spectrum ( Figure 6)] fall closer to one other, together with other phenyl signals in a multiplet centered at 8.13 ppm.
The resonances of protons H2′ and H6′ of 3′cisDPyP, a singlet and a doublet, respectively, are partially overlapped at ∼10 ppm. The resonances of the 16 pyrrole protons (singlets and doublets, 2H each) partially overlap between 8.7 and 9.2 ppm with one notable exception. One singlet falls at 7.47 ppm and was assigned to the two equivalent Hβ protons of 4′cisDPyP located between the pyridyl rings [i.e., in positions 7 and 8 (red label in Figure 5)]: they point toward 3′cisDPyP and partially fall into its shielding cone. Consistent with the model structure (see above), the corresponding β-protons on 3′cisDPyP are more removed from 4′cisDPyP and thus less shielded. Lastly, the NH protons of the two porphyrins resonate as two well-resolved sharp singlets (2H each) at −2.91 and −3.18 ppm. On the basis of the consideration of mutual shielding described above, the singlet at −3.18 ppm (which falls at a frequency significantly lower than those of both 1 and 2) was assigned to 4′cisDPyP. Consistently, in the COSY spectrum ( Figure 6) it has a cross peak with the Hβ singlet at 7.47 ppm.
The ultraviolet−visible (UV−vis) spectrum of 5 is similar to those of the component porphyrins, and consistent with the nearly orthogonal orientation between the two porphyrins, no exciton splitting is observed in the Soret band (contrary to what is found in 2). 26 In the IR spectrum, the two CO stretching bands (2074 and 2014 cm −1 ) are similar to those measured for precursor 3.
Heteroleptic metallacycle 5 was also successfully prepared (even though on only a small scale, in an NMR tube) following the complementary synthetic procedure, i.e., upon treating reactive ditopic intermediate 4 with 4′cisDPyP (Supporting Information).
Strictly similar heteroleptic 2+2 metallacycle [{t,c,c-RuCl 2 (CO) 2 } 2 (4′cisDPyMP)(3′cisDPyP)] (8), which contains 4′cisDPyMP in place of 4′cisDPyP, was obtained in 20% isolated yield with the same synthetic approach. Ditopic intermediate [{t,c,c-RuCl 2 (CO) 2 (dmso-O)} 2 (4′cisDPyMP)] (3Me) was treated with a stoichiometric amount of 3′cisDPyP in chloroform at 40°C, followed by flash chromatography purification of the crude product. The 1 H NMR spectrum of 8 (Supporting Information) is similar to that of 5, with the addition of the singlet for the methyl groups at 2.65 ppm and a better resolution of the signals in the crowded aromatic region. Treatment of 8 with an excess of zinc acetate in a CHCl 3 / MeOH mixture afforded the corresponding fully zincated metallacycle 8Zn. The insertion of zinc involves the expected decrease in the number of Q bands in the UV−vis spectrum from four to two (Supporting Information). Even though, aside from the absence of the NH signals, the 1 H NMR spectrum of 8Zn in CDCl 3 is very similar to that of 8 ( Figure  7), several resonances are better resolved, for example, those of protons H2′ and H6′ of 3′cisDPyP and of the pyrrole protons (see also the Supporting Information).
By virtue of its better resolved 1 H NMR spectrum, 8Zn was well suited for a DOSY investigation to obtain more information about the size of these heteroleptic metallacyles (Supporting Information). The measured diffusion coefficient (D t ) for 8Zn was (6.06 ± 0.10) × 10 −6 cm 2 s −1 , which Finally, we compared the stepwise synthetic approach with the one-pot preparation, which consists of treating a 1:1 mixture of 4′cisDPyMP and 3′cisDPyP with 2 equiv of Ru(II) precursor 6. A careful analysis of the TLC spots of the mixture as well as of the low-frequency region (NH resonances) of the otherwise extremely crowded 1 H NMR spectrum allowed us to establish that the crude product was a mixture of the three p o s s i b l e 2 + 2 m e t a l l a c y c l e s 8 , [ t , c , c -RuCl 2 (CO) 2 (4′cisDPyMP)] 2 (1Me), 5 0 and [t,c,c-RuCl 2 (CO) 2 (3′cisDPyP)] 2 (2) in almost equal amounts (in addition to oligomeric species, whose presence is suggested by some broad resonances), as expected for a reaction under kinetic control (Supporting Information). Given the very similar chromatographic mobility of the three metallacycles  Inorganic Chemistry pubs.acs.org/IC Article (Experimental Section), this statistical mixture was not easily amenable to separation by column chromatography. In conclusion, even for this relatively simple 2+2 heteroleptic metallacycle, the one-pot approach turned out to be less suitable than the stepwise one.

■ CONCLUSIONS
We described a modular stepwise approach for the preparation of heteroleptic metallaycles of porphyrins linked through neutral 90°-angular Ru(II) fragments. As a proof of concept, the synthetic route was successfully applied to the preparation of unprecedented 2+2 heteroleptic metallacycle [{t,c,c-RuCl 2 (CO) 2  Compound 5, which was obtained in pure form in acceptable isolated yield, fills the gap in the series of the corresponding homoleptic metallacycles with 4′cisDPyP (1) and 3′cisDPyP (2) (Figure 8). It has an L-shaped geometry, with the 4′cisDPyP side of the L being more than twice as long as the 3′cisDPyP side.
The modular approach that we established is highly flexible and opens the way to several possible exciting developments, such as the preparation of extended metallacycles. For example, treatment of reactive intermediate 3 with 0.5 equiv of a tetrapyridylporphyrin is expected to lead to the formation of 3+4 extended metallacycles (i.e., three porphyrins and four Ru linkers) featuring two different porphyrins. Combinations of  Inorganic Chemistry pubs.acs.org/IC Article 4′-and 3′PyPs will afford unprecedented geometries that, after metalation, like 1 and 2 might be further exploited as molecular panels for the hierarchical construction of 3D assemblies. In addition, because the porphyrins are introduced in a stepwise manner, it is possible, in principle, to place different metal ions in selected positions inside the metallacycle, thus opening the way to currently unavailable perspectives. For example, it might become possible to investigate how the rate of photoinduced processes (e.g., energy or electron transfer) depends not only on the relative orientation of the porphyrins (orthogonal vs parallel) but also on the specific direction (e.g., center to periphery or vice versa). Lastly, other reactive intermediates, in which the porphyrin has a different geometry (e.g., 4′transDPyP) or bears reactive metal fragments different from {t,c,c-RuCl 2 (CO) 2 (dmso-O)} (or both), might be prepared and be used as precursors of new heteroleptic metallacycles. [t,c,c-RuCl 2 (CO) 2 (4′cisDPyMP)] 2 (1Me) was prepared and purified following the same procedure described in ref 25 for 1. Crystals suitable for X-ray analysis were obtained by layering n-hexane on top of a CH 2 Cl 2 solution of 1Me. Elemental analysis for [t,c,c-RuCl 2 (CO) 2  Instrumental Methods. Mono-and bidimensional ( 1 H− 1 H COSY and 1 H− 13 C HSQC) NMR spectra were recorded at room temperature, unless stated otherwise, on a Varian 400 or 500 spectrometer ( 1 H, 400 or 500 MHz; 13 C{ 1 H}, 100.5 or 125.7 MHz). 1 H DOSY experiments were performed on the Varian 500 spectrometer at −5°C (CDCl 3 ), using the Bipolar Pulse Pair Stimulated Echo with Convection Compensation Sequence implemented in the VnmrJ software. 1 H chemical shifts were referenced to the peak of residual nondeuterated solvent (δ 7.26 for CDCl 3 , δ 5.32 for CD 2 Cl 2 ). ESI mass spectra were recorded in positive mode on a PerkinElmer APII spectrometer at 5600 eV. However, as is typical for such neutral systems, 45 the ESI-MS spectra of the model complexes and metallacycles of porphyrins showed only peaks deriving from the fragmentation. The UV−vis spectra were recorded on an Agilent Cary 60 spectrophotometer, using 1.0 cm path-length quartz cuvettes (3.0 mL). Infrared spectra of chloroform solutions in the CO stretching region were recorded between CaF 2 windows (0.5 mm spacer) on a PerkinElmer Fourier-transform IR/Raman 2000 instrument in transmission mode. Elemental analyses were performed on a Thermo Flash 2000 CHNS/O analyzer in the Department of Chemistry of the University of Bologna (Bologna, Italy). 45 X-ray Diffraction. Data were collected at the X-ray diffraction beamline (XRD1) of the Elettra Synchrotron of Trieste, Italy, equipped with a Pilatus 2M image plate detector. The collection temperature was 100 K (nitrogen stream supplied through an Oxford Cryostream 700 instrument); the wavelength of the monochromatic X-ray beam was 0.700 Å, and the diffractograms were obtained with the rotating crystal method. The crystals were dipped in N-paratone and mounted on the goniometer head with a nylon loop. The diffraction data were indexed, integrated, and scaled using the XDS code. 51 The structure was determined by the dual-space algorithm implemented in the SHELXT code. 52 Fourier analysis and refinement were performed by the full-matrix least-squares methods based on F 2 implemented in SHELXL. 53 The Coot and SHELXLE programs were used for modeling. 54,55 Anisotropic thermal motion was allowed for all non-hydrogen atoms. Hydrogen atoms were placed at calculated positions with isotropic factors U = 1.2U eq , where U eq is the equivalent isotropic thermal factor of the bonded non-hydrogen atom. Crystal data and details of refinements are given in Supporting Information.
Computational Methods. The model of compound 5 was set up by connecting the corresponding fragments taken from existing published X-ray structures of 1 and 2. 26 The so obtained starting geometry was then relaxed with the Quantum-Espresso suite of codes, 56,57 in the frame of density functional theory with the Kohn− Sham orbitals expanded in plane waves and the effects of atomic core regions accounted for by projector-augmented wave (PAW) pseudopotentials. 58 The exchange-correlation part of the energy functional was modeled with the (spin-unpolarized) generalized gradient approximation (GGA), in the PBE parametrization. 59 The plane wave expansion of the crystalline orbitals was truncated at a cutoff energy of 80 Ry (as determined by convergence of the total energy with a stepwise increase in the cutoff), and a corresponding cutoff of 800 Ry was used for the expansion of charge density and potential. Due to the large unit cell used for the calculation, the first Brillouin zone was sampled at the Γ point only. A "molecule in the box" methodology was applied, where a single molecule is (periodically) simulated in a unit cell, large enough to minimize any interactions between the molecule and its periodic images. A cubic cell with a cell constant of 31.00 Å was found to give a minimum separation distance of >10 Å between nearest atoms of any two contiguous periodic images. The correction of Makov and Payne to the total energy for isolated systems when simulated with large unit cells, as implemented in the Quantum-Espresso suite, was also applied. 60 Convergence thresholds for geometry relaxation were 1 × 10 −4 Ry for total energy and 1 × 10 −3 Ry/Å for the maximum force component acting on atoms. [{t,c,c-RuCl 2 (CO) 2 (OH 2 )} 2 (4′cisDPyMP)] (3Me). The procedure is similar to that leading to 3. To a 178.8 mg (0.465 mmol) amount of 6, dissolved in 20 mL of CHCl 3 , was added 50.0 mg (0.078 mmol) of 4′cisDPyMP (6:4′cisDPyMP ratio of 6), and the purple solution was stirred for 50 min at room temperature. The solvent was removed under reduced pressure, and the purple solid was washed with water and dried under vacuum. The solid was redissolved in chloroform and dried with Na 2 SO 4 . Yield: 71.2 mg, 81%. 1  [{t,c,c-RuCl 2 (CO) 2 (OH 2 )} 2 (3′cisDPyP)] (4). To a 186.9 mg (0.486 mmol) amount of 6, dissolved in 50 mL of CHCl 3 , was added 50.0 mg (0.081 mmol) of 3′cisDPyP (6:3′cisDPyP ratio of 6), and the purple solution was stirred at room temperature for 50 min; then, the solvent was removed under reduced pressure. The recovered purple solid was washed with H 2 O to remove the unreacted complex and the free DMSO and dried under vacuum. The solid was redissolved in chloroform and dried on Na 2 SO 4 . The solution was recovered, the solvent removed by evaporation at reduced pressure, and the residue dried under vacuum. Yield: 84.7 mg, 94%. 1  One-Pot Synthesis of [{t,c,c-RuCl 2 (CO) 2 } 2 (4′cisDPyMP)-(3′cisDPyP)] (8). The reaction was carried out on a small scale in an NMR tube. A 1.2 mg amount (3.1 × 10 −3 mmol) of 6 was dissolved in 700 μL of CDCl 3 , and 1.0 mg of 4′cisDPyMP (1.5 × 10 −3 mmol) and 0.9 mg of 3′cisDPyP (1.5 × 10 −3 mmol) (4′cisD-PyMP:6:3′cisDPyP ratio of 1:2:1) were added. The purple solution was monitored by 1 H NMR during a 24 h period at room temperature. At the end of the reaction, the solvent was removed by evaporation at reduced pressure and the purple solid obtained was thoroughly washed with diethyl ether to remove DMSO (it was partially soluble in the most suitable acetone). TLC analysis (silica gel, CHCl 3 ) showed the presence of multiple spots, the most intense belonging to 1Me (R f = 0.23), 8 (R f = 0.15), and 2 (R f = 0.13). The presence of the three metallacycles in almost equal amounts was confirmed by the diagnostic signals of the NH protons in the CDCl 3 1 H NMR spectrum of the mixture: 1Me (−2.70 ppm), 8 (−2.90 and −3.17 ppm), and 2 (−3.00 ppm).

Notes
The authors declare no competing financial interest. The University of Trieste is a partner of the CRUI/ACS 2020−2023 agreement.

■ ACKNOWLEDGMENTS
The contributions to the experimental work of undergraduate students Laura Russo and Gaia Stojanovic are acknowledged. E.A. is thankful to BASF Italia Srl for a donation of hydrated ruthenium chloride. Financial support from the University of Trieste (FRA2018) is gratefully acknowledged. The Ph.D. fellowship of A.V. was supported by FSE-S3 2014/2020 and Regione FVG, Project HEaD, code FP1799034003.