Synthesis and Characterization of Dimethylbis(2-pyridyl)borate Nickel(II) Complexes: Unimolecular Square-Planar to Square-Planar Rotation around Nickel(II).

The syntheses of novel dimethylbis(2-pyridyl)borate nickel(II) complexes 4 and 6 are reported. These complexes were unambiguously characterized by X-ray analysis. In dichloromethane solvent, complex 4 undergoes a unique square-planar to square-planar rotation around the nickel(II) center, for which activation parameters of ΔH⧧ = 12.2(1) kcal mol-1 and ΔS⧧ = 0.8(5) eu were measured via NMR inversion recovery experiments. Complex 4 was also observed to isomerize via a relatively slow ring flip: ΔH⧧ = 15.0(2) kcal mol-1; and ΔS⧧ = -4.2(7) eu. DFT studies support the experimentally measured rotation activation energy (cf. calculated ΔH⧧ = 11.1 kcal mol-1) as well as the presence of a high-energy triplet intermediate (ΔH = 8.8 kcal mol-1).

Introduction

A variety of low-valent nickel n class="Chemical">complexes have important reactivity in reactions ranging from the cycloisomerization of C=C[1] and C=O[2] π systems to the reduction of CO2.[3] In line with the latter, we have recently observed that ammonia–borane dehydrogenation catalysts 1 and 2 (Figure 1A),[4] each bearing an anionic bidentate borate ligand, are capable of CO2 reduction concurrent with ammonia–borane dehydrogenation. We therefore became interested in investigating the reactivity of other nickel complexes of the anionic bidentate ligand dimethylbis(2-pyridyl)borate (3; Figure 1B). In developing the synthesis of such complexes, we found an unexpected unimolecular square-planar to square-planar mutarotation of diamagnetic (borate)nickel(II) complex 4. Rotations of this class, e.g. cis/trans isomerization of diamagnetic, square-planar late-metal complexes, are generally known not to proceed through unimolecular mechanisms.[5] Nickel(II) has a possible exception, however, in that these species can equilibrate between diamagnetic square-planar and paramagnetic tetrahedral configurations.[5a,6] Nonetheless, no case has been carefully documented wherein a diamagnetic square-planar metal undergoes facile, unimolecular ligand rotation.

Figure 1

(A) Ru(II) complexes possessing an anionic bidentate borate ligand. (B) The dimethylbis(2-pyridyl)borate ligand 3.

(A) Ru(II) n class="Chemical">complexes possessing an anionic bidentate borate ligand. (B) The dimethylbis(2-pyridyl)borate ligand 3.

Herein we report the synthesis, characterization, and solution dynamics of the novel n class="Chemical">nickel(II) complexes (dimethylbis(2-pyridyl)borate)(triphenylphosphine)nickel(II) chloride (4) and (dimethylbis(2-pyridyl)borate)nickel(II) acetylacetonate (6) (Scheme 1). Complex 4 undergoes isomerization via two different mechanisms: (1) a unimolecular square-planar to square-planar rotation around the nickel(II) center and (2) a relatively slower ring flip. The activation parameters (ΔH⧧ and ΔS⧧) for both mechanisms were measured using 1H NMR inversion recovery experiments.[7] Studies to show that rotation around the metal center is indeed unimolecular—not the usual associative or dissociative isomerization pathways that have been studied for many four-coordinate square planar metals[8]—are discussed.

Scheme 1

Syntheses of 4 and 6

Results

Synthesis and Characterization of Complexes 4 and 6

Complexes 4 and 6 were synthesized via simple metathesis reactions. The reaction of 1 equiv of n class="Chemical">bis(triphenylphosphine)nickel(II) chloride with borate 3 in dichloromethane gives complex 4 in 71% isolated yield (Scheme 1). The formation of complex 4 is accompanied by formation of a small amount of the thermodynamically favored bis(borate) nickel complex 5, which is known to form with facility when 3 is treated with nickel(II) salts.[9] The acetylacetonate-ligated nickel complex 6 was readily prepared by treating an excess of Ni(acac)2 (2.5 equiv) with sodium borate 3, which yielded complex 6 (67%, Scheme 1). In contrast, the reactions of 1 equiv of either nickel(II) acetate tetrahydrate (Ni(OAc)2·4H2O) or nickel(II) acetylacetonate (Ni(acac)2) with 3 exclusively provide bis(borate) nickel complex 5.

Both 4 and 6 were characterized by single-crystal X-ray crystallography (Figure 2). Both complexes are square planar. However, n class="Chemical">complex 4 exhibits a slight distortion. Whereas in complex 6 the sum of the bond angles around the nickel is 360.03(7)°, the sum of angles around nickel for complex 4 is 361.6(2)°, with the P–Ni–Cl plane tilted 14° relative to the N–Ni–N plane. An analogous nickel complex of the bidentate borate ligand dihydrobis(pyrazolyl)borate (Bp) has recently been reported and similarly adopts a square-planar structure.[10]

Figure 2

ORTEP diagrams of (A) complex 4 and (B) complex 6. Ellipsoids are drawn at the 50% probability level. Complex 4 cocrystallizes with 0.5 equivalent of hexanes, which is omitted for clarity.

ORTEP diagrams of (A) complex 4 and (B) n class="Chemical">complex 6. Ellipsoids are drawn at the 50% probability level. Complex 4 cocrystallizes with 0.5 equivalent of hexanes, which is omitted for clarity.

As in all previous complexes of 3, the n class="Chemical">metal–chelate six-membered ring exhibits a boat conformation in both 4 and 6.[4,11] The dihedral angle between metal and pyridine planes in complex 6 is 53.3(4)°. However, the dihedral angle in complex 4 is 47.9(2)°, indicating that the dimethylbis(2-pyridyl)borate ligand is tilted farther out of plane in complex 4. Consistent with the observation of the increased tilt is the decreased distance between nickel and the carbon in the endo-configured (methyl)boron group in complex 4 (3.143 Å versus 3.171 Å in complex 6). Complex 4 exhibits the expected trans effect with the Ni–N bond trans to PPh3 longer than that trans to the Cl– (1.945(2) and 1.893(3) Å, respectively). On the other hand, Ni–N bonds are equivalent (1.8943(8) and 1.8955(8) Å), and the Ni–O bonds are similar (1.8567(7) and 1.8615(7) Å) in complex 6. The Ni–N bond lengths in both 4 and 6 are consistent with those observed for 5 (1.906(2) and 1.902(2) Å).[9]

Low-Temperature NMR Studies on Nickel Complex 4

The 1H n class="Chemical">NMR spectrum of nickel complex 4 in CD2Cl2 at 25 °C shows that 4 is predominantly diamagnetic in solution. Further, it undergoes a dynamic process in solution, which is observed in variable-temperature (VT) NMR studies (Figure 3). Whereas the pyridyl α hydrogens of the bound dimethylbis(2-pyridyl)borate ligand give a coalesced singlet (δ 8.55 ppm) at 25 °C, two distinct singlets are observed at lower temperatures (i.e., δ 8.12 and 8.82 ppm at −40 °C). This means that complex 4 is isomerizing in solution (Figure 4A). A similar isomerization has been observed for an analogous Bp complex, 7 (Figure 4B).[10]

Figure 3

Variable-temperature 1H NMR spectra of complex 4.

Figure 4

(A) Observed isomerization in complex 4. (B) Isomerization of the dihydrobis(pyrazolyl)borate (Bp) complex 7.[10]

Variable-temperature 1H n class="Chemical">NMR spectra of complex 4.

(A) Observed isomerization in complex 4. (B) Isomerization of the n class="Chemical">dihydrobis(pyrazolyl)borate (Bp) complex 7.[10]

Isomerization of square-planar complexes usually proceeds via either an associative or a dissociative mechanism (Scheme 2, vide infra).[8] However, we were intrigued by the possibility that we were observing isomerization via a mechanism involving a unimolecular (first-order) square-planar to square-planar rotation around a n class="Chemical">metal center. Such a process could explain the observation from the 1H NMR VT studies.

Scheme 2

Possible Mechanisms of Isomerization

Proposed associative and dissociative mechanisms involving PPh3 are shown, but it is possible that the other two ligands, Cl– and 3, could also be responsible for effecting either of these mechanisms.

Proposed associative and dissociative mechanisms involving PPh3 are shown, but it is possible that the other two ligands, Cl– and 3, n class="Chemical">could also be responsible for effecting either of these mechanisms.

Kinetic Studies of the Conformational Dynamics of 4

To show that neither an associative nor dissociative mechanism is responsible for the isomerization of nickel 4, we analyzed the dependence of the rate of isomerization with respect to the three ligands present in 4: n class="Chemical">PPh3, chloride, and 3.

The 1H n class="Chemical">NMR spectrum of nickel 4 at −40 °C shows well-differentiated pyridyl α hydrogens of the borate ligand (Figure 5). This presents an ideal opportunity to use 1H NMR inversion recovery experiments to determine the rate of this isomerization. For example, when the pyridyl proton at 8.82 ppm is pulsed (selectively labeled by inversion), magnetization transfer to the pyridyl proton at 8.12 ppm can be observed. By acquisition of data with different mixing times, a rate constant for the isomerization can be obtained.[7]

Figure 5

1H NMR spectrum of nickel 4 at −40 °C. (Methyl)boron groups are assigned by 1D-NOESY spectroscopy.

1H n class="Chemical">NMR spectrum of nickel 4 at −40 °C. (Methyl)boron groups are assigned by 1D-NOESY spectroscopy.

Rate Dependence on PPh3

The rate constants for the isomerization of 4 as a function of [n class="Chemical">PPh3] at −41.1 °C were obtained by using 1H NMR inversion recovery experiments and fitting the data into CIFIT.[7] A plot of ln kobs vs ln [PPh3] (Figure 6, left) shows that the rate of isomerization is independent of [PPh3]. The rate of isomerization was unchanged upon addition of up to 10 equiv of PPh3, and the slope of the ln/ln plot comparing the concentration of PPh3 with the observed rate of rotation had a slope of 0.0, which indicates that PPh3 is of kinetic order 0.0 in the rotation mechanism.

Figure 6

ln/ln plots of the dependence of isomerization rate of 4 on [PPh3] (left) and [Cl–] (right). For the plot on the left, inversion recovery data were collected at −41.1 °C using a 12 mM CD2Cl2 solution of 4 with [PPh3] ranging from 0.1 to 10.0 equiv. Slope = 0.00(1). For the plot on the right, inversion recovery data were collected at −43.0 °C using a 12 mM CD2Cl2 solution of 4 with [(n-Bu)4NCl] ranging among 0.4, 0.7, 1.1, 1.5, and 2.0 equiv versus Ni atom. Slope = −0.01(3).

ln/ln plots of the dependence of isomerization rate of 4 on [PPh3] (left) and [Cl–] (right). For the plot on the left, inversion ren class="Chemical">covery data were collected at −41.1 °C using a 12 mM CD2Cl2 solution of 4 with [PPh3] ranging from 0.1 to 10.0 equiv. Slope = 0.00(1). For the plot on the right, inversion recovery data were collected at −43.0 °C using a 12 mM CD2Cl2 solution of 4 with [(n-Bu)4NCl] ranging among 0.4, 0.7, 1.1, 1.5, and 2.0 equiv versus Ni atom. Slope = −0.01(3).

Also, to ensure that PPh3 exchange is not occurring, we performed a n class="Chemical">31P NMR inversion recovery study to see whether magnetization transfer from coordinated PPh3 to free PPh3 would occur. No magnetization transfer was observed at −42.4 °C (Figure 7), indicating that PPh3 exchange is indeed not occurring at a temperature where rotation is fast: t1/2 = 30 ms.

Figure 7

31P NMR inversion recovery experiment at −42.4 °C. Data were collected using a 12 mM CD2Cl2 solution of 4 with 0.5 equiv of added PPh3: (○) coordinated PPh3, δ(31P) 12.60, 31P T1 = 213(3) ms; (□) free PPh3;, δ(31P) −7.23; linear slope 0.00(0).

31P n class="Chemical">NMR inversion recovery experiment at −42.4 °C. Data were collected using a 12 mM CD2Cl2 solution of 4 with 0.5 equiv of added PPh3: (○) coordinated PPh3, δ(31P) 12.60, 31P T1 = 213(3) ms; (□) free PPh3;, δ(31P) −7.23; linear slope 0.00(0).

Rate Dependence on Chloride

The rate of isomerization of 4 is also independent of the concentration of the n class="Chemical">chloride ligand (Figure 6, right). Because of the insolubility of the chloride anion in CD2Cl2, the isomerization is unlikely to be proceeding via a chloride dissociation mechanism. We nonetheless examined the rate of isomerization in the presence of tetra-n-butylammonium chloride ((n-Bu)4NCl), a soluble source of chloride. Addition of (n-Bu)4NCl leads to a reaction wherein a small portion of an unidentified paramagnetic species is formed; regardless, 4 is observed in the 1H NMR spectrum, and we recorded rate constants for the rotation in the presence of up to 2 equiv of (n-Bu)4NCl. The rate of isomerization was unchanged by the presence of excess (n-Bu)4NCl (Figure 6, right). We also examined the rate of isomerization in the presence of excess LiCl, which is insoluble in CD2Cl2 and does not react with 4. The rate of isomerization at −42.0 °C is unaffected by addition of an excess of LiCl (see the Supporting Information).

While the dissociation of the chloride ligand to form a positively charged trin class="Chemical">coordinate intermediate in a noncoordinating solvent (CD2Cl2) is unlikely, if the chloride ligand is dissociating from 4, addition of thallium triflate should lead to formation of TlCl and to the removal of chloride from the coordination sphere of 4. To test for this situation, we treated a solution of 4 with 2 mol equiv of Tl(OTf). The 1H NMR spectrum of this sample is unaltered by the addition of Tl(OTf): 4 is stable in the presence of Tl(OTf) at room temperature for days, which indicates that the Cl– ligand does not dissociate. Moreover, the presence of 2 equiv of Tl(OTf) does not alter the rate of isomerization at −42.0 °C (23.1(4) s–1 immediately after addition; 23.5(6) s–1 4 days after addition).

Consideration of Borate 3 and Observation of a Second, Relatively Slower Isomerization Pathway

A mechanism of isomerization involving (a) dissociation of one of the pyridine rings of ligand 3, followed by (b) rotation around the n class="Chemical">Ni–N bond of the bound pyridine, and then (c) recoordination of the free pyridine ring (Scheme 3) can be envisaged. This process leads to an isomerized product equivalent to an isomerized product of a ring flip. Since the two (methyl)boron groups are differentiated in the 1H NMR spectrum of 4 (singlets at 2.22 and 0.32 ppm, Figure 5), inversion recovery experiments to measure the rate of this process are possible. When an 1H NMR inversion recovery experiment is performed at −40 °C, no magnetization transfer is observed from one methyl group to the other. Even at 0 °C, the magnetization transfer is too slow to permit measurement of a rate constant. Only at 13.7 °C were we able to obtain a rate constant: kobs = 2.69(4) s–1. This ring flip is therefore a different, slower isomerization pathway for nickel 4 (Scheme 4). Since this separate process is much slower than the rotation, we find that the observed rotation behavior cannot be accounted for on the basis of a ring flip.

Scheme 3

Possible Mechanism of Isomerization Involving Dissociation of Ligand 3

Ring flip is possible with or without initial dissociation of a pyridyl nitrogen. The measured rate of ring flip is slower than the rate of rotation by greater than 2 orders of magnitude.

Scheme 4

Ring Flip Places the (Methyl)boron Groups of 4 in a Different Chemical Environment, Whereas Rotation Places Them in the Same Chemical Environment

Magnetization transfer between endo and exo (methyl)boron groups is not observed at low temperatures because rotation places the methyl groups in the same chemical environment (by symmetry). In contrast, magnetization transfer can be observed at higher temperatures via ring flip.

Magnetization transfer between endo and exo (methyl)boron groups is not observed at low temperatures because rotation places the methyl groups in the same chemical environment (by symmetry). In n class="Chemical">contrast, magnetization transfer can be observed at higher temperatures via ring flip.

Activation Parameters for Square-Planar Rotation and Ring Flip

To determine activation parameters, we used Eyring plots constructed from rate n class="Chemical">constants obtained by NMR inversion recovery experiments. For the square-planar rotation of 4, we obtained values of ΔH⧧ = 12.2(1) kcal mol–1 and ΔS⧧ = 0.8(5) eu (Figure 8, left). For the ring flip, we measured values of ΔH⧧ = 15.0(2) kcal mol–1 and ΔS⧧ = −4.2(7) eu (Figure 8, right).[12,13] Note that the rates for rotation and ring flip differ by greater than 2 orders of magnitude (ca. 3 kcal/mol), with rotation being faster.

Figure 8

Eyring plots for unimolecular rotation (left) and ring flip (right). For the plot on the left, inversion recovery data were collected using a 12 mM CD2Cl2 solution of 4 at −61.6, −61.4, −53.7, −46.8, −42.0, −32.6, −31.8, and −18.9 °C. ΔH⧧ = 12.2(1) kcal mol–1 and ΔS⧧ = 0.8(5) eu.[13] For the plot on the right, data were collected using a 12 mM CD2Cl2 solution of 4 at 13.7, 18.8, 28.1, and 38.2 °C. ΔH⧧ = 15.0(2) kcal mol–1 and ΔS⧧ = −4.2(7) eu.[13]

Eyring plots for unimolecular rotation (left) and ring flip (right). For the plot on the left, inversion recovery data were n class="Chemical">collected using a 12 mM CD2Cl2 solution of 4 at −61.6, −61.4, −53.7, −46.8, −42.0, −32.6, −31.8, and −18.9 °C. ΔH⧧ = 12.2(1) kcal mol–1 and ΔS⧧ = 0.8(5) eu.[13] For the plot on the right, data were collected using a 12 mM CD2Cl2 solution of 4 at 13.7, 18.8, 28.1, and 38.2 °C. ΔH⧧ = 15.0(2) kcal mol–1 and ΔS⧧ = −4.2(7) eu.[13]

DFT Studies

The potential energy surface for rotational isomerization of the PMe3 analogue of n class="Chemical">nickel 4 was evaluated computationally by DFT (Figure 9). The geometry-optimized structure of singlet A shows an excellent correspondence with the experimentally determined structure of 4, which reinforces the use of this model as a probe for the isomerization in 4. Diamagnetic A is the thermodynamically preferred isomer, and the pseudotetrahedral triplet B lies 8.8 kcal mol–1 above A. The computed free energy difference between A and B is 9.9 kcal mol–1, leading to a Keq value of ca. 10–8 for the triplet at room temperature, whose concentration is negligible and is in keeping with the silent EPR spectrum and magnetic susceptibility data recorded for 4 (vide infra). B is formed by a directionally specific clockwise rotation of the PMe3 and Cl ligands in A, and continued rotation of these groups furnishes the transition structure TSBB’ that exhibits C symmetry. Here the PMe3 ligand maintains a proximal orientation with respect to the BMe2 moiety throughout the isomerization; PMe3 rotation away from the BMe2 moiety and under the pyridyl rings leads to deleterious steric interactions. The computed enthalpic barrier of 11.1 kcal mol–1 closely matches the experimentally determined ΔH⧧ value for the process.

Figure 9

PBE-optimized structures for the singlet (A) and triplet (B) species [Me2B(2-py)2]NiCl(PMe3) and the potential energy surface for the low-energy rotational isomerization of the PMe3 and Cl ligands via TSBB′. Energy values (ΔH) are given in kcal mol–1.

PBE-optimized structures for the singlet (A) and triplet (B) species [n class="Chemical">Me2B(2-py)2]NiCl(PMe3) and the potential energy surface for the low-energy rotational isomerization of the PMe3 and Cl ligands via TSBB′. Energy values (ΔH) are given in kcal mol–1.

Discussion

Possible Mechanisms of Isomerization

While no example of a unimolecular square-planar to square-planar rotation around a diamagnetic late-metal n class="Chemical">complex has been documented, prior studies show that such a rotation is possible for some nickel(II) systems. In noncoordinating solvents, four-coordinate d8 nickel(II) complexes that exist in equilibrium as diamagnetic (S = 0) square planar and paramagnetic (S = 1) tetrahedral isomers have been well documented.[14] Eaton used the Woodward–Hoffmann rules to obtain the selection rules for isomerization of four-coordinate nickel(II) complexes and showed that the isomerization from square planar to tetrahedral should be a facile, thermally allowed process.[15] Likewise, using orbital correspondence analysis with maximum symmetry, Halevi and Knorr showed that tetrahedral triplet to planar singlet isomerizations of nickel(II) complexes accompanied by a spin flip are thermally allowed.[16] Indeed, the reported rates of isomerizations for several nickel(II) species are fast. For example, lifetimes of 10–5 s for each isomer have been measured for nickel(II) aminotroponeiminates,[14a] bis(salicylaldimine)nickel(II),[14b,14c] and a series of bis(n-alkyldiphenylphosphine)nickel(II) dihalides.[14k] Moreover, the racemization of optically active diastereomeric (Δ and Λ) nickel(II) species was noted to be so fast that each diastereomer could not be observed by NMR.[14g,14k] Thermodynamic studies show that this equilibrium is affected by ligand sterics, with the presence of bulkier ligands favoring formation of the usually less thermodynamically stable tetrahedral complex.[14] The effects on equilibrium by electronic factors are also important.[14a,14j−14l,14p]

Although the literature suggests that rotation around a square-planar nickel(II) center is possible, only two classes of isomerization have been mechanistically characterized by studies of a large number of diamagnetic square-planar n class="Chemical">complexes, (1) associative and (2) dissociative, with associative mechanisms being more frequently reported.[8,17] Associative isomerization can proceed through two mechanisms: Berry pseudorotation and consecutive displacement. These associative isomerizations initiate by coordination of a catalytic ligand to metal to form a five-coordinate intermediate. This ligand can be an added base, solvent, or any metal-coordinating group. In the case of the Berry pseudorotation, the five-coordinate intermediate isomerizes to a trigonal bipyramid in which ligands can be isomerized, followed by dissociation of the catalytic ligand. In the consecutive displacement mechanism, coordination of the catalytic ligand is followed by either loss of an anionic ligand to generate a cationic metal species or dissociation of a neutral ligand to generate a neutral metal species. This is followed by recoordination of the original ligand and loss of the catalytic ligand. Some authors contend that displacement of a neutral ligand would not result in isomerization, due to the stereospecific nature of ligand substitutions,[8] while others claim to have definitively proven the existence of the neutral intermediate.[18] There is also debate about the existence of complexes which isomerize via a pseudorotation mechanism and disagreement about what evidence definitively proves a pseudorotation mechanism over a consecutive displacement mechanism.

While it is difficult to distinguish among the mechanisms of associative isomerization, it is straightforward to classify an isomerization as associative: by showing first-order rate dependence on added ligand. In contrast, dissociative isomerization proceeds through loss of a ligand to generate a three-n class="Chemical">coordinate complex. Diminution of rate in the presence of excess ligand, isomerization in the absence of potential ligands, including coordinating solvents, and positive ΔS⧧ are often taken as evidence of a dissociative process, but evidence of the existence of a three-coordinate intermediate is required confidently to conclude a dissociative mechanism.

Unimolecular Rotation of Nickel 4

Previous studies on square-planar to tetrahedral equilibria of n class="Chemical">nickel(II) complexes, which include molecular orbital analyses, indicate the possibility of isomerization via a unimolecular square-planar to square-planar rotation around a nickel center (i.e., a 90° rotation from square planar to tetrahedral followed by another 90° rotation from tetrahedral to square planar).[14−16] However, no study that systematically shows—by excluding the associative or dissociative isomerization pathways—unimolecular rotation around a diamagnetic square-planar nickel complex has been reported. These cases involve equilibrium between two observable geometric isomers at nickel: a diamagnetic square-planar complex and a paramagnetic tetrahedral complex. Moreover, ligands in some of these complexes are known to be substitutionally labile. For example, in the presence of excess phosphine, bis(n-alkyldiphenylphosphine)nickel(II) dihalides undergo phosphine exchange to give a second-order ligand exchange mechanism that contributes to the rate of isomerization.[14k] Similarly, bis(salicylaldimine)nickel(II)[14d] and bis(β-ketoimine)nickel(II)[14g] complexes undergo facile mixed ligand exchange.

The studies mentioned in the Results show that the rotation of complex 4 in a nonn class="Chemical">coordinating solvent involves neither an associative nor a dissociative mechanism at the temperatures studied. The most likely ligand to be involved in associative or dissociative isomerization is PPh3. The observation that the rate of isomerization is independent of [PPh3] excludes an associative process and provides evidence against a dissociative process. Moreover, 31P NMR inversion recovery experiments show that, in the presence of excess PPh3, exchange between coordinated and free PPh3 is not occurring, thus excluding a dissociative mechanism involving PPh3.

The rate of isomerization is also unaffected by the presence of excess (n-Bu)4NCl or n class="Chemical">LiCl, which precludes an associative mechanism with respect to the chloride ligand. This result is not surprising, given the low dielectric constant of CD2Cl2, which disfavors charge separation. A dissociative process is ruled out by the fact that addition of 2 equiv of Tl(OTf) does not alter the 1H NMR spectrum of complex 4. If dissociation of the chloride ligand is occurring, the addition of thallium, which has strong affinity for halides, would sequester free chloride in solution and alter the 1H NMR spectrum. Moreover, the rate of isomerization at −42.0 °C was unaffected by the presence of 2 equiv of Tl(OTf), even after the solution was allowed to stand at room temperature for 4 days. This shows that (a) the structure and (b) the conformational dynamics of 4 are unaffected by the presence of Tl(OTf), which is inconsistent with chloride dissociation under these conditions.

Studies of the dependence of the rate of isomerization on [3] are not possible because addition of excess 3 results in fast formation of bis(borate) nickel n class="Chemical">complex 5. However, the isomerization proceeds in the absence of excess 3; thus, an associative process involving 3 is unlikely. Also, the ΔS⧧ value for isomerization (0.8(5) eu) is inconsistent with ligand association (which generally has ΔS⧧ = −10 to −15 eu).[19]

Likewise, the ΔS⧧ value is too small to be consistent with ligand dissociation (which generally has ΔS⧧ = 10 to 15 eu).[19] For example, the ΔS⧧ values for the dissociative substitution of n class="Chemical">CO in Ni(CO)4 range from +8 to +13 eu, depending on solvent.[20] Moreover, the bond enthalpy of a Ni–N(pyridine) bond (ca. 26 kcal mol–1)[21] is much higher than the measured ΔH⧧ value for the isomerization (12.2(1) kcal mol–1), which indicates that the Ni–N bond is not broken during the isomerization process. Furthermore, an isomerization involving the dissociation of 3 (Scheme 3), which leads to a product equivalent to a ring flip product, was considered. This process was found to be a separate, slower isomerization pathway: ring flip is >100-fold slower than rotation at a given temperature. The ΔS⧧ for the ring flip (−4.2(7) eu) indicates a more ordered transition state and thus argues against a dissociative mechanism (i.e., the ring flip occurs without hemidissociation of 3). If hemidissociation is not occurring at higher temperatures where the ring flip is observed (as high as 38.2 °C), then it is unlikely that hemidissociation is occurring at temperatures as low as −61.6 °C, where rotation is observed. Finally, the presence of excess PPh3 should affect the rate of isomerization if hemidissociation is occurring, but this is not observed. On the basis of these observations, we believe that association or dissociation of ligand 3 is not involved in the rotation mechanism.

A further associative rotation mechanism that fits these kinetic data is one that involves formation of an agostic n class="Chemical">Ni–HC interaction[22] to the endo (methyl)boron group to generate a transient five-coordinate nickel center. We believe that such an agostic interaction is not forming for three key reasons. (1) The 1H NMR peak width of the endo-positioned (methyl)boron group in 4 is invariant over a range of 75 °C (6.9 ± 1.0 Hz, Figure 3). (2) The 1H chemical shift of the same is invariant over the same temperature range: 2.22 ± 0.09 ppm. (3) The solid-state Ni–H distance is 2.43 Å, which is longer than those observed for agostic interactions between this ligand and ruthenium(II) (1.72 Å)[4a] or platinum(IV) (2.02 Å).[11b] Moreover, the agostic Ni–H distance in a (NacNac)Ni(κ2-C2H5) complex is 1.66 Å,[23] which is far smaller than our observed Ni–H distance (NacNac = N,N′-(2,6-(CH3)2C6H3)2CH3C(N)CHC(N)CH3 anion). Whereas none of our observations are consistent with known agostic complexes involving borate 3, we predict that no agostic intermediate is involved in the unimolecular rotation of 4.

As corron class="Chemical">borated by DFT calculations, the unimolecular rotation of 4 involves the intermediacy of a paramagnetic, tetrahedral (or pseudotetrahedral) complex, but this species is not present in an observable concentration. For example, the phosphine crossover experiment shown in Figure 7 shows that, under the isomerization conditions, the 31P NMR integrations of 4 and PPh3 in solution match the portions added to the sample. This provides evidence that there is not a large portion of phosphine associated with an NMR-invisible paramagnetic species. Furthermore, an EPR spectrum of 4 (X-band, 25 °C, toluene) showed no signals.[24] Moreover, whereas the paramagnetic susceptibility of many nickel(II) complexes which exhibit square-planar to tetrahedral equilibria can be measured using the Evans method,[14a,14j−14l] we observe no paramagnetic susceptibility of complex 4 even at high concentrations (127 mM, 25 °C; see the Supporting Information) using this method.[25]

DFT studies confirm the presence of a high-energy triplet intermediate on the potential energy surface for the rotational exchange of the n class="Chemical">phosphine and chlorine ligands in 4. These computational studies show that the enthalpic barrier for rotation in the PMe3 analogue (ΔH⧧ = 11.1 kcal mol–1) of nickel 4 is very close to that value experimentally measured for nickel 4. The presence of a high-energy triplet intermediate (ΔH = 8.8 kcal mol–1) was also corroborated by these studies.

On the basis of the forgoing evidence, we believe that the internal rotation of 4 is the first documented case of a unimolecular square-planar to square-planar rotation around a late-metal center. We believe that this mechanism proceeds through a high-energy paramagnetic n class="Species">tetrahedral (or pseudotetrahedral) nickel(II) intermediate species. We propose, however, that no ligand coordination or dissociation, including agostic coordination of a (methyl)boron group, is requisite to the mechanism.

Conclusion

The novel dimethylbis(2-pyridyl)borate n class="Chemical">nickel complexes 4 and 6 were synthesized and characterized. The solution dynamics of complex 4 were studied, and it was found to isomerize via two different mechanisms: (a) a unimolecular square-planar to square-planar rotation around the nickel center and (b) a relatively slower ring flip. The activation parameters for each isomerization pathway were measured using 1H NMR inversion recovery experiments. The barrier for rotation (ΔH⧧ = 12.2(1) kcal mol–1 and ΔS⧧ = 0.8(5) eu) is lower than the barrier for the ring flip (ΔH⧧ = 15.0(2) kcal mol–1 and ΔS⧧ = −4.2(7) eu), as expected. This study is the first to systematically show (by excluding the associative and dissociative pathways) that a 180° rotation around a square-planar nickel center is a viable mechanism of isomerization.

Experimental Section

General Procedures

All air- and water-sensitive procedures were carried out either in a Vacuum Atmospheres glovebox under n class="Chemical">nitrogen (2–10 ppm O2 for all manipulations) or using standard Schlenk techniques under nitrogen. Deuterated NMR solvents were purchased from Cambridge Isotopes Laboratories. Dichloromethane, ethyl ether, and hexanes were purchased from VWR and dried in a J. C. Meyer solvent purification system with alumina/copper(II) oxide columns, benzene (EMD) was dried by distillation over sodium–benzophenone ketyl, and bis(triphenylphosphine)nickel(II) chloride and nickel(II) acetylacetonate were purchased from Strem and used as received.

NMR spectra were ren class="Chemical">corded on a Varian VNMRS 500 or VNMRS 600 spectrometer. All chemical shifts are reported in units of ppm and referenced to the residual 1H or 13C solvent peak and line-listed according to (s) singlet, (bs) broad singlet, (d) doublet, (t) triplet, (dd) doublet of doublets, etc. 13C spectra are delimited by carbon peaks, not carbon count. 11B and 31P chemical shifts are referenced to the lock channel. Air-sensitive NMR spectra were taken in 8 in. J. Young tubes (Wilmad or Norell) with Teflon valve plugs. Inversion recovery kinetics data were fitted using CIFIT 2.0 by Alex Bain.[7b] Temperatures for NMR experiments were calibrated using an external methanol standard. Errors in Eyring parameters are calculated using the equations derived by Girolami et al.[13]

MALDI mass spectra were obtained on an Applied Biosystems Voyager spectrometer using the evaporated drop method on a coated 96-well plate. The matrix was n class="Chemical">anthracene. In a standard preparation, ca. 1 mg of analyte and ca. 10 mg of matrix were dissolved in dry benzene and spotted on the plate with a glass capillary. Infrared spectra were recorded on a Bruker OPUS FTIR spectrometer. High-resolution ESI mass spectra were recorded at the University of California, Riverside. CHN elemental analyses were collected at the University of Illinois at Urbana–Champaign at the School of Chemical Sciences Microanalysis Laboratory.

Nickel Complex 4

In the drybox under nitrogen, n class="Chemical">(PPh3)2NiCl2 (65.4 mg, 0.10 mmol) was dissolved in 10 mL of dry dichloromethane in a dry vial containing a Teflon stir bar. In another vial, [(py)2BMe2]Na[9] (22.0 mg, 0.10 mmol) was dissolved in 5 mL of dry dichloromethane and added slowly to the (PPh3)2NiCl2 solution. The [(py)2BMe2]Na vial was rinsed with 5 mL of dichloromethane and added to the (PPh3)2NiCl2 solution. The dark green solution turned brown and then orange upon addition of [(py)2BMe2]Na. The solution was stirred for 2 h and then filtered. The solvent was removed under reduced pressure. Dry diethyl ether was added to the residue, and the vial was sonicated briefly. The residue was then cooled using a cold well, and cold, dry hexanes was added. The suspension was filtered and washed with cold, dry ether. The suspension was dissolved with cold, dry benzene. The benzene was then removed by lyophilization to yield 39.5 mg (0.071 mmol, 71%) of [(py)2BMe2]Ni(PPh3)Cl as a pale pink solid. Crystallization from dichloromethane and hexanes produced crystals suitable for X-ray crystallographic analysis.

1H n class="Chemical">NMR (CD2Cl2, 600 MHz, 25 °C): δ 8.55 (br s, 2H), 7.77–7.61 (m, 6H), 7.44 (t, J = 7.3 Hz, 3H), 7.40 (d, J = 7.6 Hz, 2H), 7.33 (t, J = 7.6 Hz, 6H), 7.16 (br s, 2H), 6.41 (br s, 2H), 2.22 (s, 3H), 0.32 (s, 3H). 1H NMR (CD2Cl2, 600 MHz, −40 °C): δ 8.82 (s, 1H), 8.12 (s, 1H), 8.02–7.02 (m, 18H), 6.94 (s, 1H), 6.86 (s, 1H), 5.92 (s, 1H), 2.15 (s, 3H), 0.26 (s, 3H). 13C NMR (CD2Cl2, 150 MHz, 25 °C): δ 189.72 (br), 151.89, 135.07, 134.86, 131.08, 129.91, 128.91, 127.00, 119.75, 19.51 (br), 9.48 (br). 11B NMR (CD2Cl2, 192 MHz): δ −12.68. 31P NMR (CD2Cl2, 243 MHz, −40 °C): δ 12.42. MALDI: m/z 552.8941, calcd for C30H29BClNiN2P+ [M]+ 552.1203. FT-IR (thin film/cm–1): ν 3067.76, 2912.82, 2823.99, 1963.28, 1896.13, 1817.39, 1592.78, 1553.13, 149.62, 1435.29, 1290.62, 1217.84, 1093.77, 1012.64, 744.01. Anal. Calcd for C30H29BClNiN2P: C, 65.1; H, 5.28; N, 5.06. Found: C, 65.04; H, 5.47; N, 5.28.

Nickel Complex 6

In the drybox under nitrogen, n class="Chemical">Ni(acac)2 (64.2 mg, 0.25 mmol) was suspended in 10 mL of dry dichloromethane in a dry vial containing a Teflon stir bar. In another dry vial, [(py)2BMe2]Na[9] (22.0 mg, 0.10 mmol) was dissolved in 5 mL of dichloromethane and added slowly to the solution of Ni(acac)2. Another 5 mL of dichloromethane was used to rinse the vial and added slowly to the solution of Ni(acac)2. The green solution turned orange on addition of [(py)2BMe2]Na. The solution was stirred for 2 h at room temperature and then filtered through Celite. The solvent was removed in vacuo. Dry hexanes (10 mL) was added to the residue, and the vial was treated with sonication briefly. The suspension was then cooled using a cold well, filtered, and washed with cold, dry hexanes. The solid was dried under vacuum to yield 23.7 mg (0.067 mmol, 67%) of [(py)2BMe2]Ni(acac) as an orange solid. Crystallization from dichloromethane and hexanes produced crystals suitable for X-ray crystallographic analysis.

1H n class="Chemical">NMR (CD2Cl2, 600 MHz): δ 8.30 (d, J = 5.6 Hz, 2H), 7.40 (d, J = 7.5 Hz, 2H), 7.34 (t, J = 7.3 Hz, 2H), 6.83 (t, J = 6.3 Hz, 2H), 5.55 (s, 1H), 2.27 (br s, 2H), 1.88 (s, 6H), 0.27 (br s, 3H). 13C NMR (CD2Cl2, 150 MHz): δ 189.81 (q, J = 48 Hz) 188.11, 149.89, 135.23, 125.93, 119.71, 102.04, 26.09, 18.93 (br), 8.93 (br). 11B NMR (CD2Cl2, 192 MHz): δ −12.68. MALDI: m/z 354.2477, calcd for C17H22BN2O2Ni+ [M]+ 354.1050. FT-IR (thin film/cm–1): ν 2891.92, 2821.81, 1586.41, 1532.39, 1388.63, 1289.47, 122.57, 1159.73, 1015.71, 934.76, 788.67, 753.86, 738.46. HR-MS (+ESI): m/z 355.1129, calcd for C17H22BN2O2Ni+ [MH]+ 355.1122.

Computational Methodology and Modeling Details

The geometry optimizations reported here were performed with the DFT gradient-corrected n class="Chemical">correlation functional PBE, as implemented by the Gaussian 09 program package.[26] The Ni atom was described by a Stuttgart–Dresden effective core potential (ecp) and an SDD basis set, while the 6-31G(d′) basis set was employed for the remaining Cl, P, N, C, B, and H atoms.

All structures were fully optimized and evaluated for the correct number of imaginary frequencies through calculation of the vibrational frequencies, using the analytical Hessian (positive eigenvalues ground-state minima and one negative eigenvalue for a transition state). The n class="Chemical">computed frequencies were used to make zero-point and thermal corrections to the electronic energies, and the reported enthalpies are quoted in kcal/mol relative to singlet species A. The computed triplet species B and TSBB′ revealed no significant spin contamination from higher order spin states. The geometry-optimized structures have been drawn with the JIMP2 molecular visualization and manipulation program.[27]