Insights into the κ-P,N Coordination of 1,3,5-Triaza-7-phosphaadamantane and Derivatives: κ-P,N-Heterometallic Complexes and a 15N Nuclear Magnetic Resonance Survey.

Complexes {[(PTA)2CpRu-μ-CN-1κC:2κ2N-RuCp(PTA)2-ZnCl3]}·2DMSO (13) {[ZnCl2(H2O)]-(PTA-1κP:2κ2N)(PTA)CpRu-μ-CN-1κC:2κ2N-RuCp(PTA)(PTA-1κP:2κ2N)-[ZnCl2(H2O)]}Cl (14), [RuCp(HdmoPTA)(PPh3)(PTA)](CF3SO3)2 (20), [RuCp(HdmoPTA)(HPTA)(PPh3)](CF3SO3)3 (21), and [RuCp(dmoPTA)(PPh3)(PTA)](CF3SO3) (22) were obtained and characterized, and their crystal structure together with that of the previously published complex 18 is reported. The behavior of the 1,3,5-triaza-7-phosphatricyclo[3.3.1.13,7]decane (PTA) and 3,7-dimethyl-1,3,7-triaza-5-phosphabicyclo[3.3.1]nonane (dmoPTA) ligands against protonation and κN-coordination is discussed, on the basis of 15N nuclear magnetic resonance data collected on 22 different compounds, including PTA (1), HdmoPTA (7H), and some common derivatives as free ligands (2-6 and 8), along with mono- and polymetallic complexes containing PTA and/or HdmoPTA (9-22). 15N detection via 1H-15N heteronuclear multiple bond correlation allowed the construction of a small library of 15N chemical shifts that shed light on important features regarding κN-coordination in PTA and its derivatives.

Introduction

Today, hydrophilic phosphines are very common ligands in organometallic and coordination chemistry.[1,2] In this class of compounds, monodentate m-monosulfonated PPh3 (m-TPPMS) and tris-m-sulfonated PPh3 (m-TPPTS) are among the most popular examples, but bidentate diphosphines and tridentate tripodal phosphines are also known and have been used.[3] There are also examples of hydrosoluble cage-like phosphines such as Verkade-type phospha-amides[4] and 1,3,5-triaza-7-phosphaadamantane (1), which was first reported in 1974 by Daigle et al. (usually abbreviated as PTA or pta; the acronym TPA and the name “monophosphaurotropine” have been also used to indicate the ligand; the IUPAC name is rarely used in the scientific literature).[5] This ligand contains a soft phosphorus atom and three hard nitrogen atoms, which can be functionalized providing a large variety of derivatives (some examples are shown in Figure ),[6] useful for obtaining catalysts,[7−13] bioactive agents,[14−25] luminescent compounds,[26,27] and new materials.[28−35]

Figure 1

Structures of 1,3,5-triaza-7-phosphaadamantane (PTA) (1) and the derivatives studied in this work.

During the past several years, we have devoted a great deal of effort to synthesizing mono- and polymetallic complexes containing PTA and its derivatives, affording good homogeneous catalysts for the isomerization of allylic alcohols in water[36−42] like complex 9,[43] highly antiproliferative compounds[44−47] such as compounds 17–19,[48,49] and heterometallic polymers like 15 built via assembly of dimetallic complex 12 through metallic moieties (Figure ).[50] The simplest functionalizations of the PTA cage are mono- and bis-N-methylation to afford the cationic ligands N-methyl-PTA (mPTA) (5) and N,N′-dimethyl-1,3,5-triaza-7-phosphaadamantane (dmPTA) (6).[46] While the first is very stable, the latter decomposes under mild conditions, providing the ligand 3,7-dimethyl-1,3,7-triaza-5-phosphabicyclo[3.3.1]nonane (dmoPTA) (7). Half-sandwich Ru(II) complexes containing 7 and the protonated ligand 3,7-H-3,7-dimethyl-1,3,7-triaza-5-phosphabicyclo[3.3.1]nonane (HdmoPTA) (7H+), such as 17 and 18 (Figure ), exhibit great antiproliferative activities[48] and the ability to chelate a second metallic moiety through the methylated nitrogen atoms. It was shown that the chelation of a second metal, such as in the bimetallic complexes [RuCp(PPh3)2-μ-dmoPTA-1κP:2κ2N,N′-ZnCl2](CF3SO3) (19) and [RuCp(PPh3)2-μ-dmoPTA-1κP:2κ2N,N′-CoCl2](CF3SO3), improved the antiproliferative activity that was found to be 200 times higher than that of cisplatin for T-47D and WiDr human solid tumor cell lines.[47,49]

Figure 2

Structures of the complexes studied in this work.

Most of the time, the coordination of PTA and its derivatives to one metal through the phosphorus atom can be proven by 31P nuclear magnetic resonance (NMR). Nevertheless, spectroscopic characterization of κN-coordination by infrared (IR), Raman, ultraviolet–visible (UV–vis), and 1H and 13C NMR is not straightforward, and only single-crystal X-ray diffraction can provide the needed confirmation,[51−54] which only ensures the N coordination only in the solid state. Thus, to obtain more information about whether and how PTA and its derivatives are coordinated by the N atoms, we thought to lean on nitrogen NMR.

The most abundant isotope of nitrogen, 14N, is quadrupolar; thus, most of the time, its detection is not practical. On the contrary, 15N has a spin of 1/2 but its natural abundance is only 0.365%; therefore, the duration for data collection for non-enriched samples is usually very long. Nevertheless, this issue can be bypassed by the detection of 1N via 1H through long-range correlation experiments, today run by using routine and robust pulse sequences as heteronuclear multiple-bond correlation (HMBC) or heteronuclear multiple quantum correlation.[55] It is well established that 1H–15N HMBC is a very useful tool for the structural assignments in protein sequences[56] and was also used for the characterization of natural products[57−61] and the structural resolution of isomers.[62] Also, various studies were published, including 15N-enriched coordination compounds such as bimetallic clusters,[63] and in recent years, organometallic complexes were also characterized by this technique.[64−76] In this work, we used the 1H–15N HMBC pulse sequence to investigate in solution some noncoordinated PTA derivatives (Figure ) as well as some previously published representative Ru half-sandwich complexes displaying κP- and κP,Ncoordination (Figure ). Additionally, the new monometallic (20–22) and polymetallic (13 and 14) complexes containing dmoPTA and/or PTA were synthesized and characterized by single-crystal X-ray diffraction, expanding the family of PTA-κP,N-complexes, illuminating new aspects of the coordination behavior of the PTA and dmoPTA ligand in the solid state and solution.

Experimental Procedures

Synthesis and Characterization of 13 and 14

Trying to obtain an analogue of polymer 15 with Zn instead of Cd, we reacted bimetallic complex 12 with 3 equiv of ZnCl2 in water. Immediately, a light brown precipitated formed, which was redissolved in dimethyl sulfoxide (DMSO) and water. When the DMSO solution is cooled, complex [(PTA)2CpRu-μ-CN-1κC:2κ2N-RuCp(PTA)2-ZnCl3] (13) crystallizes as the DMSO solvate, while upon evaporation of the water solution, tetrametallic complex {[ZnCl2(H2O)]-(PTA-1κP:2κ2N)(PTA)CpRu-μ-CN-1κC:2κ2N-RuCp(PTA)(PTA-1κP:2κ2N)-[ZnCl2(H2O)]}Cl (14) was obtained (Scheme ). Structures of these complexes were characterized by single-crystal X-ray diffraction, as described below, but the first assessment of their different composition was first allowed by comparison of their IR spectra. The cyanide vibration frequency is significantly different for 13 (2131 cm–1) and 14 (2109 cm–1), and the IR spectrum of 14 shows H2O stretching and bending bands (3479 and 1623 cm–1, respectively), which are absent in 13.

Scheme 1

Synthesis of 13 and 14

The 31P{1H} NMR spectra in D2O of both complexes 13 and 14 display two singlets in a 1:1 ratio, at −19.8 and −22.4 ppm, respectively. This behavior agrees with the cleavage of the PTA-Zn bond upon dissolution, as previously reported for other PTA-k-P,N complexes.[50,77,78]

Synthesis and Characterization of 20–22

Complex [RuCp(HdmoPTA)(PPh3)(PTA)](CF3SO3)2 (20) was obtained in 81% yield by treatment of [RuClCp(PPh3)(PTA)] with AgOTf and further reaction with dmPTA (6) at 50 °C (Scheme ). The presence of the CF3SO3– anion in the complex composition was confirmed by both its 13C{1H} NMR and infrared spectra. Except for the resonances assigned to PTA, the 1H and 13C chemical shifts are similar to those of complex 17, supporting that the ligand dmoPTA is protonated.[48] Its 31P{1H} NMR spectrum in CD3OD displays the signals relative to PTA (−39.40 ppm), protonated dmoPTA (−3.99 ppm), and PPh3 (46.33 ppm) in a 1:1:1 ratio as an AMX pattern and agrees with those of previously reported compounds containing these ligands.[43,48] Addition of 1.9 equiv of triflic acid to a solution of 20 in CD3OD shifts the phosphine signals to higher fields, arising at −27.09 ppm (HPTA+), −5.91 ppm (HdmoPTA+), and 44.81 ppm (PPh3). The large variation of the chemical shift for PTA (Δδ31P = 12.31 ppm) suggests its protonation to give complex [RuCp(HdmoPTA)(PPh3)(HPTA)](CF3SO3)3 (21). The 1H multiplets of the NCHN atoms (4.82 ppm) are shifted to a higher field in 20 (Δδ1H = 0.32 ppm), as found after elucidating its 1H NMR spectrum by 1H COSY and 1H–13C HSQC NMR. The values observed for 31P, 1H, and 13C resonances agree with those of the similar previously reported complex. Finally, when 20 is reacted with 1.7 equiv of BuOK, deprotonated complex [RuCp(dmoPTA)(PPh3)(PTA)](CF3SO3) (22) forms. Its 31P{1H} NMR spectrum displays the signals of PTA, dmoPTA, and PPh3 at −37.54, 5.48, and 50.03 ppm, respectively, which are similar to what was observed for 18. It is also worth mentioning that the carbons of the methyl groups are inequivalent, arising at 44.19 and 44.27 ppm in the 13C{1H} NMR spectrum, as evidenced by its 1H–13C HSQC spectrum.[48]

Scheme 2

Synthesis of 20–22

Crystal Structure of 13 and 14

Complexes 13 and 14 crystallize in the P21/n and P1̅ space groups, respectively. Selected bond lengths and angles are listed in Table , while a complete list can be found in Tables S7, S8, S12, and S13. In terms of 13, the asymmetric unit contains one molecule of trimetallic complex [(PTA)2CpRu-μ-CN-1κC:2κ2N-RuCp(PTA)-μ-PTA-1κP:2κ2N-ZnCl3] (Figure ) and two molecules of DMSO. The complex unit consists of two cyanide-bridged piano-stool {RuCp(PTA)2}+ moieties, which coordinate a {ZnCl3}− anion through one nitrogen of a PTA ligand. In the case of 14, the asymmetric unit contains a {ZnCl2(H2O)-μ-(PTA-1κP:2κ2N)(PTA)CpRu-C(N)}1/2+ moiety (Figure ) and 1/2Cl–, which upon growing around the inversion center affords a tetrametallic complex with a formula of {ZnCl2(H2O)-μ-(PTA-1κP:2κ2N)(PTA)CpRu-μ-CN-1κC:2κ2N-RuCp(PTA)-μ-(PTA-1κP:2κ2N)-ZnCl2(H2O)}Cl. The bond lengths between the ruthenium and the phosphorus of the bidentate κP,N-PTA ligands are almost identical in both complexes 13 and 14 [13, Ru1–P1 = 2.256(2) Å; 14, Ru1–P1 = 2.2515(10) Å], while the same distances from monodentate κP-PTA are longer in 13 and shorter in 14 [13, Ru1–P2 = 2.277(2) Å, Ru1–P3 = 2.264(2) Å, and Ru1–P4 = 2.272(2) Å; 14, Ru1–P2 = 2.2477(11) Å]. The C≡N bond in 13 is slightly longer than that in 14 [13, CCNA–NCNA = 1.156(11) Å; 14, CCN–NCN = 1.138(6) Å], which agrees with their C≡N vibration energies, as mentioned above, and the lengths are in the range found for complex 12, polymer 15, and its analogue polymers with Au, Co, and Ni, where the PTA acts as a bidentate κP,N ligand [12, 1.137(9) Å; 15, 1.158(7) Å; trans-(12-CoCl3), 1.147(4) Å; cis-(12-CoCl3), 1.140(7) Å; trans-(12-NiCl3), 1.141(10) Å; trans-(12-Au(CN)4), 1.152(6) Å].[28,30,31,79] In both complexes, Ru–C(N) distances are in the range found for similar compounds, as well as the bond lengths found for the PTA ligand.[35,80−83]

Table 1

Selected Bond Lengths and Angles for 13 and 14

bond lengths (Å)						bond angles (deg)
13			14			13				14
Ru1	P1	2.256(2)	Ru1	P1	2.2515(10)	P1	Ru1	P2	95.60(8)	P1	Ru1	P2	96.89(4)
Ru1	P2	2.277(2)				P3	Ru2	P4	97.60(8)
Ru2	P3	2.264(2)	Ru1	P2	2.2477(11)	N(C)CN	Ru1	P1	88.8(2)	NCN1	Ru1	P1	85.50(11)
Ru2	P4	2.272(2)				N(C)CN	Ru1	P2	86.5(2)
Ru1	NCNA	2.021(8)	Ru1	N(C)CN	2.027(4)	C(N)CN	Ru2	P3	86.7(2)	NCN1	Ru1	P2	88.69(11)
Ru2	CCNA	2.023(8)				C(N)CN	Ru2	P4	89.0(2)
NCNA	CCNA	1.156(11)	NCN	CCN	1.138(6)	Cl1	Zn1	Cl2	115.51(10)	Cl2	Zn1	Cl1	115.91(4)
Zn1	N1	2.123(7)	Zn1	N1	2.114(3)	Cl1	Zn1	Cl3	117.91(10)	N1	Zn1	Cl1	104.93(9)
Zn1	Cl1	2.235(3)	Zn1	O1	1.940(3)	Cl2	Zn1	Cl3	112.85(10)	N1	Zn1	Cl2	102.40(10)
Zn1	Cl2	2.246(2)	Zn1	Cl1	2.2455(12)	N1	Zn1	Cl1	99.34(19)	O1	Zn1	Cl1	110.90(10)
Zn1	Cl3	2.251(3)	Zn1	Cl2	2.2380(11)	N1	Zn1	Cl2	104.60(19)	O1	Zn1	Cl2	114.99(10)
						N1	Zn1	Cl3	103.8(2)	O1	Zn1	N1	106.32(13)

Figure 3

Thermal ellipsoid representation of the complex unit in the crystal structures of 13 and 14. Their relevant bond lengths, plane angles, and torsion angles are listed in Tables S7, S8, S12, and S13. Anions, solvent molecules, and hydrogen atoms connected to the carbon atoms have been omitted for the sake of clarity.

The zinc atoms in 13 and 14 display a distorted tetrahedral geometry and complete the coordination spheres with three chlorides in 13 and two chlorides and one H2O in 14. The Zn–Cl bond lengths are comparable in both complexes (13, Zn–Cl = 2.244 Å; 14, Zn–Cl = 2.242 Å), similar to the typical bond lengths in the structure of zinc chloride complexes in aqueous solution {[ZnCl2(PTA)2], Zn–Cl = 2.231 Å; [ZnCl2(μ-O=PTA)], Zn–Cl = 2.218 Å; [ZnCl2(O=PTA)(H2O)], Zn–Cl = 2.218 Å; [ZnCl2(S=PTAH)(S=PTAZnCl3)], Zn–Cl = 2.245 Å},[77,78,84] while the Zn–N bond in 13 is slightly longer than that in 14 [13, Zn1–N1 = 2.123(7) Å; 14, Zn1–N1 = 2.114(3) Å] as well as other examples of Zn–NPTA bonds {[ZnCl2(PTA)2], Zn1–N1 = 2.055(3) Å; [ZnCl2(μ-O=PTA)], Zn1–N1 = 2.108(16) Å; [ZnCl2(O=PTA)(H2O)], Zn1–N1 = 2.093(10) Å; [ZnCl2(S=PTAH)(S=PTAZnCl3)], Zn1–N1 = 2.085(15) Å}.[77,78,85]

With regard to the intermolecular contacts, in the crystal packing of 13, there is no interaction worth mentioning, while in 14, each tetrametallic moiety of 14 is connected to four neighboring complexes via the O1 and N6 atoms, forming hydrogen-bonded layers along the reciprocal b*–c* plane (Figure ).

Figure 4

Hydrogen bonds around 14 forming a layered structure.

Crystal Structure of 18, 21, and 22

Complex [RuCp(PPh3)2(dmoPTA-κP)](CF3SO3)·CHCl3 (18) was previously prepared and published by us,[49] but single crystals suitable for X-ray diffraction were obtained only recently by diffusion of Et2O vapor into a solution of the complex in chloroform, giving chloroform-solvated (18·CHCl3). Single crystals of complexes [RuCp(HdmoPTA)(PPh3)(HPTA)](CF3SO3)3·3H2O (21·H2O) and [RuCp(dmoPTA)(PPh3)(PTA)](CF3SO3)·MeOH (22·MeOH) were obtained by slow evaporation of a solution of the corresponding complex in water and methanol, respectively. Selected bond distances and angles for the three complexes are listed in Table , and complete lists of bond lengths and angles are provided in Tables S9–S11 and S14–S16. The asymmetric unit of complex 18·CHCl3 contains a [RuCp(PPh3)2(dmoPTA-κP)]+ molecule, one CF3SO3–, and one CHCl3. The coordination sphere of the ruthenium atom displays a piano-stool geometry and is constituted by a η5-Cp, two PPh3 ligands, and one dmoPTA ligand (Figure ). Bond distances between the metal and the Cp centroid and phosphine P atoms are similar to those found for similar complexes (Table ).[46,86−88] The dmoPTA ligand is distorted (C1–P1–C2 = 102.15°; C1–P1–C3 = 95.86°; C2–P1–C3 = 94.84°), and the distance between the CH3N atoms [N1–N2 = 3.609(4) Å] is significantly larger than the distance for the equivalent protonated complex (2.80 Å). Comparison of the crystal structure of 18·CHCl3 with its protonated form also shows that protonation of dmoPTA causes an extension of the three Ru–P bonds [Δ(Ru1–P1) = +0.0023 Å, Δ(Ru1–P2) = +0.0119 Å, and Δ(Ru1–P3) = +0.0326 Å].[48]

Table 2

Selected Bond Lengths and Angles for 18, 21, and 22a

		bond length (Å)
atoms		17b	18	21	22
Ru1	P1	2.321(1)	2.3187(9)	2.2821(14)	2.3066(8)
Ru1	P2	2.366(1)	2.3541(8)	2.2838(13)	2.3009(8)
Ru1	P3	2.389(1)	2.3564(9)	2.3433(12)	2.3116(8)
N1	N2	2.800(6)	3.609(4)	2.670(8)	3.565(4)

For the sake of comparison, the values obtained for 17 were also added.

From ref (48).

Figure 5

Thermal ellipsoid representations of the cationic portion of the crystal structures of 18, 21, and 22. Their relevant bond lengths, plane angles, and torsion and angles are listed in Table , which also includes the data of 17(48) for the sake of comparison (structure not shown). Anions, solvent molecules, and hydrogen atoms connected to carbons have been omitted for the sake of clarity.

Complexes 21 and 22 are constituted by the same complex moiety: a η5-Cp bonded to the Ru, which completes its coordination geometry by one κP-PTA, one PPh3, and one κP-dmoPTA. Nevertheless, in 21, PTA and dmoPTA are monoprotonated, its composition being [RuCp(PPh3)(HPTA-κP)(HdmoPTA-κP)]3+, while in 22, these ligands are deprotonated, its composition being [RuCp(PPh3)(PTA-κP)(dmoPTA-κP)]+ (Figure ). In both 21 and 22, the counterions are CF3SO3–, which are disordered in the lattice. The bond lengths and angles in complexes 21 and 22 are similar and fall in the ranges found for related complexes.[46−49,86,89] Nevertheless, it is important to point out that the Ru–PdmoPTA bond in 21 [Ru–P1 = 2.2821(14) Å] is significantly shorter than that in 22 [Ru–P1 = 2.3066(8) Å] and 18 [Ru–P1 = 2.3187(9) Å]. The same tendency is also observed for the Ru–PTA bond in 21 [Ru–P2 = 2.2838(13) Å], which is shorter than that in 22 [Ru–P2 = 2.3009(8) Å] and similar to the Ru–PdmoPTA bond for this complex. Also, the Ru–PPh3 bond length is somewhat different in the three complexes [18, 2.3541(8) and 2.3564(9) Å; 21, 2.3433(12) Å; 22, 2.3116(8) Å]. The longest Ru–P bond observed in 18 is reasonable, due to the steric effect exerted by the PPh3 ligands, which also provoke a large distortion of the coordination geometry, as reflected by the angles between ligands [18, P1–Ru1–P2 = 95.44(3)°; P2–Ru1–P3 = 105.94(3)°; P1–Ru1–P3 = 94.27(3)°]. Nevertheless, compared to what found in dmoPTA–ZnCl2 complex 19, the Ru–PPh3 bond length is only somewhat shorter in the protonated complex 21 and significantly shorter in deprotonated complex 22. The cone angles[90−92] of the HdmoPTA and dmoPTA ligands calculated from the crystal structure are only slightly different, ∼103° for 21 and ∼106° for 22, leading to a similar angle between ligands [21, P1–Ru1–P3 = 97.09(5)°, P1–Ru1–P2 = 95.65(5)°, and P2–Ru1–P3 = 97.14(5)°; 22, P1–Ru–P3 = 95.95(3)°, P2–Ru1–P1 = 94.12(3)°, and P2–Ru1–P3 = 98.64(3)°]. The distance between the CH3N atoms is 2.670(8) Å for 21, where the ligand is protonated, whereas for the deprotonated ligand in 22, it is 3.565(4) Å, which is like that found in 18. The protonation of PTA in 21 is localized on the N4 atom, which is part of a network of hydrogen bonds with water molecules and triflate anions (Figure ).

Figure 6

Hydrogen bonds network between adjacent molecules in the crystal structure of 21.

It is important to additionally consider the effect of the deprotonation of the HdmoPTA ligand in complexes 18 and 22. When the CH3NdmoPTA atoms are not H-bridged, the bottom rim of the triazaphosphaadamantane-like cage opens through an inversion about one of the methylated nitrogen atoms, which moves away from their lone pairs from each other. Given that in 18 and 22 the molecules are not connected by a network of hydrogen bonds, the greater N1–N2 length in 18 and 22 may suggest that their distance is susceptible to the steric hindrance of the surrounding ligands, which in 18 imposes a larger separation between the methyl groups (Figure ). Intending to predict the N′–N″ chelation behavior of the dmoPTA ligand, we find these data shed light on the possibility of chelating also metals with a van der Waals radius larger than those of Zn, Ni, and Co, which are the only examples of dmoPTA-κN′,N″-coordinated metals obtained to date.[89,93]

Figure 7

Highlights of ligands 7 and 7H+ in 17, 18, 21, and 22. In red are shown distances between N1 and N2. In green are shown torsion angles among C6, N1, C4, and N3. For the sake of clarity, only hydrogen atoms bonded to nitrogen atoms are represented.

Results and Discussion

15N Chemical Shifts of 1–22 via 1H–15N NMR Long-Range Correlations

δ15N of the Free Ligands

The adamantane-like phosphines and derivatives studied by 1H–15N long-range NMR correlations are displayed in Figure . Considering that selected metal-free triazaphosphines 1–3 display C3 symmetry, while for ligands 4–8 is C, in the first group only the cross peak due to three magnetically equivalent nitrogen atoms is expected, while for the second, two different signals should be found; the respective nitrogen atoms are numbered according to Figure . The obtained δ15N values for compounds 1–8 are summarized in Figure and Table S1. As expected, only a 15N signal was observed for 1 in D2O that arises at 24.6 ppm, which is the expected region for a tertiary amine.[94] When the phosphorus atom is functionalized, the 15N resonance of the PTA suffers inductive deshielding, as seen for 2 and 3. Upon methylation of the phosphorus atom, compound 2 is obtained, which displays a singlet at 42.9 ppm, while oxidation of the phosphorus, which gives 3, shifts the signal to 64.3 ppm.

Figure 8

δ15N values for compounds 1–8.

Simple protonation of 1 affords compound 4 that displays a singlet due to fast proton exchange. This peak is shifted by only 0.1 ppm in D2O with respect to 1, which is the usual behavior for sp3 nitrogens.[95] Nevertheless, this fact indicates the large difference in the chemical shift between the phosphorus in 31P NMR, which is approximately 1 order of magnitude larger.[96] Methylation of one of the nitrogen atoms of 1 affords the ligand mPTA (5), which displays the δ15N signal corresponding to quaternary nitrogen N1 that moved by only 0.2 ppm to a higher field with respect to 1. On the contrary, the tertiary nitrogens N2 suffer a dramatic inductive effect, arising at 34.8 ppm in D2O. Therefore, the nonsubstituted PTA nitrogen atoms in 5 are shifted by almost +10 ppm with respect to 1 but −9 ppm with respect to P-methylated regioisomer 2. These results show that the functionalization of the PTA at the phosphorus atom leads to a Δδ15N much more pronounced than that at the nitrogen. Further N methylation of 5 gives rise to N,N′-dimethylated derivative 6, usually known as dmPTA, whose CH3N1- and N2-δ15N are found at 41.2 and 47.1 ppm, respectively, in acetone-d6, showing how these atoms are markedly deshielded by the second methylation. The Δδ15N1 and Δδ15N2 between 6 and 5 are larger (Δδ15N1 = +18.1 ppm, and Δδ15N2 = +12.9 ppm) than those between 5 and 1, due to the inductive effect produced by the quaternary nitrogen atoms, which is doubled in 6 and reciprocally exercised by both N1. It is interesting to point out that the mono- and dimethylation of 1 produce a similar effect on the δ31P chemical shift.

Derivatives obtained by functionalization of PTA at the two nitrogen atoms suffer, under the appropriate conditions, the lysis of the CH2 group bridging the functionalized nitrogens.[97] The simplest example may be compound 6, which loses the methylene between the CH3N atoms, giving the neutral compound 7, where the δ15N1 and δ15N2 in D2O are found at a higher field concerning 6 and also 1 (δ15N1 = 33.1 ppm; δ15N2 40.6 ppm). The difference of ∼10 ppm between the signals of 1 and 7 could be caused by the opening of the adamantane-like cage. A similar shielding effect on the nonfunctionalized atom N2 is observed also for ligand 8, usually known as DAPTA. For 8, the δ15N2 arises at 31 ppm in DMSO-d6, while the acylated nitrogen atoms are found at 107.9–109.1 ppm, in the expected region for amides.[98]

It is important to evidence that the solvent polarity also has a slight influence on the chemical shift of the N atoms of the studied compounds. In terms of 2, the 15N signal is linearly shielded with a decrease in solvent polarity, while compound 1 displays a different behavior: it suffers shielding when the more polar DMSO-d6 rather than acetone-d6 is used, but in D2O, the signal is deshielded by 0.5 ppm concerning acetone-d6. This behavior, which is shown also by 5, can be tentatively addressed considering the possible involvement of the phosphorus atom in hydrogen bonding, which could induce the deshielding of the NPTA atoms in a manner similar to but less intense than that caused by the methylation of the PPTA atom.

δ15N of Metal Complexes Containing Ligands 1, 4, and 5

The 1H–15N HMBC of complex [RuClCp(PTA)2] (9), which contains two equivalent PTA ligands, is characterized by a correlation at 40.2 ppm in D2O that is relative to its six equivalent nitrogen atoms, together with another set of cross-peaks at 39.3 ppm. This additional signal can be assigned to complex [RuCp(PTA)2(D2O)]+ (10) that is in equilibrium with 9 in water.[99] The δ15N1 of 11 was determined to be shifted by 3.5 ppm to 9 (Figure ). As observed in 4, only one 15N resonance is observed, due to fast proton exchange.

Figure 9

δ15N values for compounds 9–16.

A new step in complexity is represented by diruthenium complex 12, which was synthesized by the reaction of 9 with a half-equivalent of KCN.[79] The 15N resonances for 12 arise in the same chemical shift range as 9. Additionally, two singlets are observed at 40.8 and 41.8 ppm corresponding with N1 and N2 atoms, respectively (see Figure ), which are due to the asymmetry of the cyanide bridge making the {RuCp(PTA)2}+ moieties inequivalent. N1 corresponds to the nitrogen atoms of the PTA bonded to the Ru-CN fragment, and N2 to those of the PTA bonded to the Ru-NC.

The coordination of one {ZnCl3}− moiety or two {ZnCl2(H2O)} moieties to the nitrogen atoms of 12 leads to trimetallic complex 13 and tetrametallic complex 14, respectively. The absence of multiplicity in their 31P{1H} NMR resonances suggests that the N–Zn bond is cleaved upon dissolution. This assumption can be also supported by their identical δ15N values (41.1 and 42.7 ppm in D2O and 41.4 and 43.5 ppm in DMSO-d6), very similar to those of 12.

Under adequate reaction conditions, polymeric complexes such as 15 can be obtained from 12, in which two PTA-N atoms are coordinated to two different {CdCl3}− units. The 1H–15N HMBC of this polymer provides identical correlations with respect to 12–14, supporting previous evidence that indicates that upon dissolution in water the Cd–N bonds are cleaved.[50]

Finally, complex 16, which contains the methylated ligand mPTA (5), shows the resonances relative to methylated atom N1 and nonmethylated N2 in DMSO-d6, arising at 37.7 and 44.7 ppm, respectively. The differences in the chemical shift between the coordinated and free ligand (Δδ15N1 = +14.6 ppm, and Δδ15N2 = +10.9 ppm) are in the range found for the complexes containing 1. Also, it is interesting to point out that N1 resonates at a frequency similar to that of the protonated species 11.

In general, the 15N resonances for complexes 9–16 (Table S2) and compound 2 appear in a very narrow chemical shift range, suggesting that P methylation of the PTA and the κP-coordination to the ruthenium produce a similar deshielding effect.

δ15N of Metal Complexes Containing Ligand dmoPTA (7)

Generally, to avoid possible side reactions caused by the chelating nitrogen atoms, the complexes containing dmoPTA (7) are obtained employing 6 as a proligand. Under reaction conditions, once 6 is κP-coordinated, it usually undergoes solvolysis of the CH2 bridging the ammonium groups and converts into phosphine 7 in its protonated form, 7H+, which occurs when [RuCp(HdmoPTA)(PPh3)2](CF3SO3)2 (17) is synthesized. The 1H–15N HMBC spectrum of 17 displays two correlations at 45.3 and 47.5 ppm (CD3OD), which are slightly susceptible to solvent changes (Figure ). The presence of just two sets of correlations in the 1H–15N HMBC spectrum supports the idea that the proton is shared between the two methylated nitrogen atoms N1. The deprotonation of 17 affords 18 and displaces the N1 chemical shift to 33.7 ppm in CD3OD (Table S3). Also, N2 is slightly shielded and resonates at 42.1 ppm. Surprisingly, after coordination to the Ru center, the δ15N values of 7 do not vary as much as observed for ligands 1 and 5.

Figure 10

δ15N values for compounds 17–19.

Ligand 7 can coordinate a variety of metallic centers through its methylated nitrogen atoms, affording bimetallic complexes whose antiproliferative activity is usually much higher than those of the monometallic parent compounds and cisplatin.[19,47−49,100] This is the case for complex [RuCp(PPh3)2-μ-dmoPTA-1κP:2κ2N,N′-ZnCl2](CF3SO3) (19), which is 5 times more potent than 18 and 425 times more potent than cisplatin on WiDr colon cancer cells. The chelation of the {ZnCl2} moiety closes the bottom rim of the dmoPTA ligand and deshields both N1 and N2, which in CDCl3 appear at 41.1 and 47.8 ppm, respectively, near the observed signals for 6 in acetone-d6. The trend shown by the δ15N values of compounds 17–19 is revealed to be very significant to assess the coordination of a second metallic unit to CH3NdmoPTA. The Δδ31P of the singlets found for 17 and 19 is only 1.2 ppm, while the δ15N of their methylated nitrogen differs by 4.21 ppm, making it easier to distinguish whether dmoPTA is protonated or coordinated to a metal.

Complexes 20–22 (Figure ) contain the ligand PPh3 and the neutral or protonated PTA and dmoPTA, providing an ideal platform for studying the effect of selective protonation on the δ15N of metal complexes containing these aminophosphines. The synthesis of these complexes starts from [RuCpCl(PPh3)(PTA)] by abstraction of the chloride with AgCF3SO3 and subsequent reaction with dmPTA (6). The resulting complex [RuCp(HdmoPTA)(PPh3)(PTA)](CF3SO3)2 (20) displays three sets of 15N atoms that are enumerated in Figure . Their chemical shifts show how the presence of the PTA shields the nitrogen atoms of HdmoPTA, producing in CD3OD differences in chemical shifts to 17 of Δδ15N1 = 0.8 ppm and Δδ15N2 = −1 ppm (Table S4). The nitrogen atom (N3) of the PTA appears at 41.9 ppm (Figure ), which is close to those obtained for complexes 9, 10, and 12 in D2O. Complex 20 is susceptible to additional protonation on the PTA, but its HdmoPTA can be deprotonated into dmoPTA. The complex containing the protonated PTA ligand (21) was obtained by the addition of 1 equiv of CF3SO3H to a solution of 20 in CD3OD. The 1H–15N HMBC spectrum of the resulting complex shows how the protonation shifts the 31P multiplet corresponding to PTA to −27.09 ppm (Δδ31P = +12.3 ppm) and shields the N3 resonance by a magnitude similar to ∼31.9 ppm (Δδ15N3 = −10.0 ppm) with regard to 20. Also, the two 15N signals corresponding to HdmoPTA appear at higher fields than in 20: that for δ15N1 at 41.5 ppm and that for δ15N2 at 44.1 ppm. Complete deprotonation of 20 employing BuOK affords complex 22 that shows a slight deshielding of N3 to 42.7 ppm, while dmoPTA-N1 and N2 appear at 31.7 and 39.7 ppm, respectively, being shielded with respect to the corresponding signals in both 20 and 21, which follow the observed trend for 17 and 18, respectively.

Figure 11

15N chemical shifts for 20–22.

Conclusions

To shed light on the behavior of PTA and dmoPTA ligands upon κN-coordination and N protonation, ruthenium half-sandwich complexes 13, 14, 21, and 22 were synthesized and characterized by multinuclear NMR, IR, and single-crystal X-ray diffraction. Complexes 13 and 14 are nice examples of the complexes containing the PTA ligand as the linker between metals, providing heterometallic complexes. Both complexes, trimetallic 13 and tetrametallic 14, possess terminal tetrahedral zinc centers κN-coordinated to one or two PTA ligands, being new examples of κP,Nmultidentation of the PTA in the solid state. It was shown that, like other previously published complexes containing PTA-κP,N ligands, the Zn–N bond is not stable in solution. Complexes 21 and 22 are monometallic species containing PPh3, PTA, and dmoPTA in different protonation states. Their characterization by single-crystal X-ray diffraction confirmed that, upon deprotonation, dmoPTA undergoes a deep conformational change that leads to the separation of the methylated amino groups. The 15N chemical shifts of PTA and a representative variety of its derivatives as well as complexes (compounds 1–22) were studied by 1H–15N HMBC NMR in various solvents. This study can be of general help to chemists working with ligands 1–8 and was performed to obtain more information about the behavior of the coordination sites of PTA and derivatives in solution. The studies supported the instability of PTA-κN multimetallic complexes in solution because of the cleavage of the PTA-κN-M bond, such as observed for 13–15, and revealed it to be important and complementary to 31P{1H} NMR in assessing the dmoPTA-κN,N′ coordination in complexes, such as shown for 17–22.