Introduction

Complexes of the group 10 metals are widely used as catalysts to facilitate transformations of organic substrates. For example, palladium-catalyzed cross-coupling reactions, which form C–X bonds (X = C, N, O, S, etc.), have revolutionized synthetic chemistry [1]. While oxidation states of 0 and +2 are commonly invoked in the catalytic reactions of these metals, there is a growing interest in the chemistry of other oxidation states [2]. The +1 oxidation state of nickel is well established as playing a key role in catalysis [3], the structural aspects of Ni(I) complexes have been recently reviewed [4], and activation of small molecules by Ni(I) has been highlighted [5]. While Pd(I) and Pt(I) binuclear complexes are known [2, 6], only recently has the monomeric Pd(I) complex [Pd(PtBu3)2][PF6] been isolated and shown to catalyze the oxidative cross-coupling of arylantimony and arylboron nucleophiles [7]. With the resurgence of photoredox chemistry in organic synthesis, Ni(I) complexes are often invoked as key intermediates in catalytic cycles [8,9,10] and the gas-phase IR spectroscopy [11, 12] and chemistry [13, 14] of several ligated Ni(I) complexes have been reported.

Mass spectrometry-based techniques have been developed over the past 50 years to provide fundamental structural, mechanistic, and energetic information about intermediates and reactions associated with metal-mediated processes [15,16,17,18]. Professor Schwarz has been one of the key pioneers in this area, where he has shed light on mechanistic aspects of C–H bond activation in methane [19, 20] and developed a number of important concepts including “remote functionalization” [21] and “two-state reactivity” [22]. Gas-phase studies can provide powerful insights into catalytic cycles since they reveal details on elementary steps that can be difficult to study under typical conditions used for catalysis [23, 24]. While the gas-phase chemistry of bare monoatomic group 10 cations has been extensively studied by Schwarz [25, 26], Armentrout [27,28,29], and others [30], previous work on the reactions of [((C6H5)3P)nAu]+ with iodobenzene has shown that ligation can have a profound effect on reactivity, with the bare cation (n = 0) undergoing a range of reactions, the monoligated (n = 1) complex selectively undergoing C–I bond activation and the bisligated complex (n = 2) being unreactive [31]. Thus, we were interested in developing a gas-phase synthesis of ligated group 10 cations in the formal oxidation state of +1. We chose the [(phen)M(O2CCH3)]+ complexes as suitable precursors since the 1,10-phenanthroline (phen) ligand has been widely used in condensed phase synthetic and catalytic studies [32]; our previous work has shown that these complexes are readily generated via ESI-MS and decarboxylate upon CID [33] to generate organometallic cations, [(phen)M(CH3)]+, which undergo a number of interesting reactions of relevance to catalysis [34,35,36,37,38,39]; CID of related cations suggest that bond homolysis should occur [40]. Here, we examine the use of bond homolysis reactions of [(phen)M(CH3)]+ to generate the cations [(phen)M]+· for subsequent ion-molecule reaction (IMR) studies. This approach is complementary to that of Parker and Gronert, who generated the ligated nickel radical cations [(phen)nNi]+· (n = 1 and 2) via electron transfer dissociation (ETD) to the dication [(phen)2Ni]2+ [41]. Our aim is to compare the bimolecular reactivity of ligated M(I) cations for Ni, Pd, and Pt, which has not yet been achieved due to the challenges of generating these reactive intermediates in the condensed phase where they can be depleted via disproportionation or other reactions. The IMR we have examined include those with background water, O2 and N2 and the reagents acetone and allyl iodide. Acetone was chosen for the following reasons: (i) previous studies on its reaction with the bare atomic cations Ni+ and Pd+ have revealed that C–C bond activation occurs [42,43,44,45,46]; (ii) our previous study on the reactions of [(phen)M(CH3)]+ with acetone uncovered a catalytic cycle [47]; (iii) metal complexes in low oxidation states react with acetone to form a coordinated ketyl radical that undergoes subsequent radical-driven fragmentation [48]. Allyl iodide was chosen since its C–C bond coupling reactions often proceed via metal-mediated C–I bond activation [49, 50].

Experimental

Chemicals from the following suppliers were used without further purification: (i) Merck, methanol (HPLC grade for ESI/MSn experiments); (ii) Chem-Supply, acetone (AR grade); (iii) Cambridge Isotope Laboratories, acetone-d6; (iv) Sigma-Aldrich, allyl iodide.

While the precursors [(phen)Ni(O2CCH3)2] and [(phen)Pd(O2CCH3)2] were available from a previous study [35], [(phen)Pt(O2CCH3)2] was synthesized via a literature procedure [51,52,53].

Mass Spectrometry Experiments

Mass spectrometric experiments were conducted on a Thermo Scientific linear ion trap (LTQ) spectrometer modified to allow IMR studies to be undertaken [54, 55]. It has been demonstrated that collisions with the helium bath gas quasi-thermalize ions to room temperature under IMR conditions [56]. Since the product complexes of interest are reactive towards background gases, precursor ions were mass selected with a 1-Da isolation window. Thus, the distinction between complexes with similar mass-to-charge ratios such as [(phen)60Ni(CH3)]+ and [(phen)58Ni(OH)]+ was achieved by monoisotopic mass selection (58Ni, 106Pd, and 195Pt). While isolation of a single isotope from the isotope manifold does lead to an overall decrease in the signal, the dynamic range of the ion trap is sufficient to carry out the desired MSn experiments (n = 2–5).

The organometallic acetate cations, [(phen)M(O2CCH3)]+, were generated via direct electrospray ionization (ESI) of the acetate complexes [(phen)M(O2CCH3)2] dissolved in methanol (~ 0.1 mM). The analyte solution was injected into the ESI source at a flow rate of 10 μL/min. ESI conditions used were the following: spray voltage 4.0 kV, capillary temperature 300 °C, nitrogen sheath gas flow rate ca. 10 arbitrary units. The three-coordinate organometallic cations, [(phen)M(CH3)]+, were generated by low-energy CID of the mass-selected complexes, [(phen)M(O2CCH3)]+, in the linear ion trap. Low-energy CID on the mass-selected [(phen)M(CH3)]+ can be used to generate the two-coordinate organometallic cations, [(phen)M]+·. The helium bath gas was used as the collision gas and the CID conditions used included a Q value of 0.25 and an excitation time of 30 ms, with the normalized collisional energy varied from 13 to 20 (arbitrary units). The implementation of IMR in the modified LTQ has been previously described in detail (see references [54, 55]). In brief, IMR of the mass-selected organometallic cations, [(phen)M]+· with the neutral reagents, were carried out in a series of MS4 experiments. The liquid neutral reagent (acetone or allyl iodide) is injected at flow rates of 10–100 μL/h directly into a stainless steel line that supplies helium to the linear ion trap. The line is heated to the boiling point of the liquid to promote vaporization. The flow of helium is controlled to 230 cm3/min, and thus the composition of the liquid reagent/helium mixture can be calculated using the known molar flow rates of helium and the neutral reagent. Since the helium pressure regulator to control the flow of gas into the ion trap is bypassed, the helium pressure is instead manually controlled by matching the pressure in the vacuum chamber surrounding the ion trap, measured using the ion gauge, to that under normal operating conditions (≈ 7–8 × 10−6 Torr). This corresponds to a pressure of ≈ 2 × 10−3 Torr inside the linear ion trap. During the IMR, no CID voltage was applied (i.e., NCE = 0) and appropriate reaction time was chosen to observe the formation of product ions. Acetone was introduced in the ion trap via the helium inlet line at a flow rate of 10 μL/h for the reactions with the Ni and Pd cations and 100 μL/h for the Pt cations. IMR between [(phen)M]+· and allyl iodide were carried out with a reaction time of 30 ms. Allyl iodide was introduced via the helium inlet line at a flow rate of 50 μL/h.

Computational Methods

Structures of all reactants, transition states, and intermediates were fully optimized using the software Gaussian 09 [57]. If not stated otherwise, the M06 DFT functional [58] employing the SDD effective core potentials (ECPs) for Pt, Pd, and Ni [59, 60] and the 6-31G(d) basis set for all other atoms were used [61]. Some structures were re-optimized using Gaussian16 [62] and the B3LYP DFT functional [63] with Grimme’s D3 correction [64, 65] with the basis set SDD for Pt, Pd, and Ni and the 6-311G(d) basis set for all other atoms [66]. Vibrational frequency calculations were performed to confirm stationary points as minimal (no imaginary frequency) or transition states (one imaginary frequency) and to obtain the zero-point energy (ZPE) corrections. Each transition state was connected to two minima by intrinsic reaction coordinate (IRC) calculations. A series of single-point energy calculations were carried out using the ORCA suite of programs [67] using the semi-empirical hybrid functional ωB97X-D3 [68] and double-hybrid functionals PWPB95-D3 [69] and DSDPBEP86-D3 [70], with the RI approximation [71] on the MP2 part of the calculation, with the basis set def2-TZVPP and ECPs on Pd and Pt [72]. Semi-empirical double-hybrid functionals have been chosen over non-empirical ones because the former has been shown to be more robust [73]. The optimized structures with the B3LYP-D3/6-311G(d) level of theory were used for the single-point energy calculations.

The thermodynamic parameter used to determine the likelihood of low-pressure unimolecular and bimolecular reactions to proceed was the reaction energy, Δ(E+ZPE). The precursor ions undergo multiple collisions with the helium bath gas resulting in slow “heating” until fragmentation occurs, in the case of low energy CID [74]. Thus, the effective temperature is not known and will be different for each system. For the IMR, Dau et al. have highlighted that the relevant thermodynamic parameter to use for low-pressure bimolecular reactions is the reaction energy rather than the Gibbs free energy since the computed entropy of the reactants is not a good measure of the entropy at the entrance channel [75].

With regard to the multiplicities of the organometallic cations, [(phen)M(CH3)]+, we found that the ground states are singlets in the case of Pd and Pt, but that the ground state of [(phen)Ni(CH3)]+ is a triplet. This is consistent with the related [(bipy)M(CH3)]+ cations, where the ground state is a singlet for Pd and Pt and a triplet for Ni [76].

Results and Discussion

Bond Homolysis vs. Carbene Formation

Electrospray ionization of [(phen)M(O2CCH3)2] (M=Ni, Pd, Pt) complexes dissolved in methanol in the positive ion mode results in the formation of the mononuclear cationic complexes [(phen)M(O2CCH3)]+. CID of these mass-selected complexes results in the formation of the methyl complexes [(phen)M(CH3)]+ as previously described (Fig. S1, Eq. (1)) [34, 35]. Besides decarboxylation, acetoxyl loss is another channel for M = Ni, Pd, resulting in the formation of the cationic complexes, [(phen)M]+·, where the metal is formally in the oxidation state +1 (Eq. (2)). As noted previously [34], the remaining ions observed in the mass spectra arise from IMR between the fragment ions and the background gases H2O, N2, and O2. Thus, the complexes [(phen)M(CH3)]+ form adducts with dinitrogen for M = Pt (Eq. (3)) and with water for M = Pd and Pt (Eq. (4)). [(phen)M(CH3)]+ undergoes hydrolysis resulting in the formation of [(phen)M(OH)]+ for M = Ni and Pt (Eq. (5)). For the short reaction time associated with these CID reactions, [(phen)M]+· forms adducts with water and dioxygen only for M = Ni (Eq. (6)).

$$ {\left[\left(\mathrm{phen}\right)\mathrm{M}\left({\mathrm{O}}_2{\mathrm{CCH}}_3\right)\right]}^{+}\kern0.5em \to \kern0.5em {\left[\left(\mathrm{phen}\right)\mathrm{M}\left({\mathrm{CH}}_3\right)\right]}^{+}+{\mathrm{CO}}_2\kern0.5em \left(\mathrm{M}=\mathrm{Ni},\mathrm{Pd},\mathrm{Pt}\right) $$
(1)
$$ {\left[\left(\mathrm{phen}\right)\mathrm{M}\left({\mathrm{O}}_2{\mathrm{CCH}}_3\right)\right]}^{+}\kern0.5em \to \kern0.5em {\left[\left(\mathrm{phen}\right)\mathrm{M}\right]}^{+\bullet }+{\mathrm{CH}}_3\mathrm{C}{\mathrm{O}}_2\bullet \kern0.5em \left(\mathrm{M}=\mathrm{Ni},\mathrm{Pd}\right) $$
(2)
$$ {\left[\left(\mathrm{phen}\right)\mathrm{Pt}\left({\mathrm{CH}}_3\right)\right]}^{+}+{\mathrm{N}}_2\kern0.5em \to \kern0.5em {\left[\left(\mathrm{phen}\right)\mathrm{Pt}\left({\mathrm{CH}}_3\right)\left({\mathrm{N}}_2\right)\right]}^{+} $$
(3)
$$ {\left[\left(\mathrm{phen}\right)\mathrm{M}\left({\mathrm{CH}}_3\right)\right]}^{+}+{\mathrm{H}}_2\mathrm{O}\kern0.5em \to \kern0.5em {\left[\left(\mathrm{phen}\right)\mathrm{M}\left({\mathrm{CH}}_3\right)\left({\mathrm{OH}}_2\right)\right]}^{+}\kern0.5em \left(\mathrm{M}=\mathrm{Pd},\mathrm{Pt}\right) $$
(4)
$$ {\left[\left(\mathrm{phen}\right)\mathrm{M}\left({\mathrm{CH}}_3\right)\right]}^{+}+{\mathrm{H}}_2\mathrm{O}\kern0.5em \to \kern0.5em {\left[\left(\mathrm{phen}\right)\mathrm{M}\left(\mathrm{OH}\right)\right]}^{+}+{\mathrm{CH}}_4\kern0.5em \left(\mathrm{M}=\mathrm{Ni},\mathrm{Pt}\right) $$
(5)
$$ {\left[\left(\mathrm{phen}\right)\mathrm{M}\right]}^{+\bullet }+\mathrm{L}\kern0.5em \to \kern0.5em {\left[\left(\mathrm{phen}\right)\mathrm{M}\left(\mathrm{L}\right)\right]}^{+\bullet } $$
(6)

Upon CID, the mass-selected complexes [(phen)M(CH3)]+ lose a methyl radical to form [(phen)M]+· (Figure 1, Eq. (7)). However, for M = Pt, the loss of a methyl radical is only a minor pathway. The major pathway is the loss of Pt=CH2 to generate the protonated ligand phenanthroline [(phen)H]+ (m/z 181, Eq. (8)), which is detected in the mass spectrum (Figure 1c). Butschke and Schwarz reported a related reaction in the CID of [(bipy)Pt(CH3)]+ [76]. The organometallic cation [(phen)Pt]+· (Eq. (7)) and the carbene Pt=CH2 (Eq. (8)) are formed with branching ratios of 2% and 98% respectively. Although the thermochemistry of the neutral metal carbenes is not known for all group 10 metals, the Pt=CH2 bond dissociation energy has been determined to be 446 kJ mol−1 [77], which highlights that it is a particularly stable species.

Figure 1
figure 1

LTQ MS3 spectra of mass-selected precursor ions, [(phen)M(CH3)]+, 2, undergoing CID at a normalized collision energy of 13 for M = Ni, Pd and 20 for M = Pt (arbitrary units) in the linear ion trap: (a) M = Ni, (b) M = Pd, and (c) M = Pt. The mass-selected precursor ions are designated by an asterisk (*)

Full size image
$$ {\left[\left(\mathrm{phen}\right)\mathrm{M}\left({\mathrm{CH}}_3\right)\right]}^{+}\kern0.5em \to \kern0.5em {\left[\left(\mathrm{phen}\right)\mathrm{M}\right]}^{+\bullet }+\bullet {\mathrm{CH}}_3 $$
(7)
$$ {\left[\left(\mathrm{phen}\right)\mathrm{M}\left({\mathrm{CH}}_3\right)\right]}^{+}\kern0.5em \to \kern0.5em {\left[\left(\mathrm{phen}\right)\mathrm{H}\right]}^{+}+\mathrm{M}={\mathrm{CH}}_2\kern0.5em \left(\mathrm{M}=\mathrm{Pt}\right) $$
(8)

To better understand the competing fragmentation pathways for [(phen)M(CH3)]+, DFT calculations have been performed to compare the energetics for bond homolysis to form [(phen)M]+· (Eq. (7)) relative to the barrier heights associated with formation of the metal carbene, M=CH2 (Eq. (8)). Bond homolysis is a barrierless process, whereas the formation of M=CH2 requires surmounting a transition state, in which a hydrogen from the methyl group translocates to the nitrogen of phenanthroline to form an intermediate, [(phen)H(M=CH2)]+, where the M=CH2 moiety is bound coordinatively to the protonated ligand, before dissociation. For every metal, the singlet state has been considered, and for nickel also the triplet state was considered because the triplet precursor ion is 25 kJ mol−1 more stable than its singlet state. The respective barriers are compared in a potential energy diagram (Figure 2). The trends are consistent with the experiment: going down group 10, the bond homolysis is disfavored whereas metal carbene formation becomes favored. Furthermore, among the singlet transition states, the one for the formation of Pt=CH2 is the lowest. The transition state for the triplet nickel is the lowest, but the energy of the bond homolysis still lies substantially below. However, even if the trends are in agreement with the experiment that bond homolysis occurs for nickel and palladium, but carbene formation is preferred for platinum, both the thermodynamics and the kinetics for the competing reaction are not consistent with experiment for the case of platinum. Thus, with the B3LYP-D3/SDD 6-311G(d) level of theory, the bond homolysis requires less energy than forming the transition state and is energetically favored over the formation of protonated phenanthroline and carbene for all the metals. While the B3LYP functional is commonly used, since it generally performs well for geometries and vibrational frequencies; previous detailed benchmarking studies have shown that the B3LYP thermochemistry can often be poorly estimated [78]. Thus, a series of single-point calculations were carried out for platinum with higher levels of theory and different functionals. Single-point energy calculations have been carried out on the structures that were optimized with the B3LYP-D3/SDD 6-311G(d) level of theory with the functionals ωB97X-D3, PWPB95-D3, and DSD-PBEP86-D3 and the basis set def2-TZVPP for all atoms. These functionals were chosen since they have shown to be very robust in past studies, providing better thermochemical estimates [78]. The relevant energetics obtained with these functionals for Pt are given in Table 1 while the potential energy diagrams are given in Fig. S2. With the functionals ωB97X-D3 and PWPB95-D3, the overall energetics for the observed formation of 5 + [(phen)H]+ are thermodynamically favored over bond homolysis to give 3 + ·CH3, while for the functional DSD-PBEP86-D3, the energetics are both kinetically and thermodynamically favored for Pt=CH2 formation. The best results are given by both double-hybrid functionals PWPB95-D3 and DSD-PBP86-D3, which are expected to perform better than common hybrid DFT functionals [78]. Even at the highest level of theory, the computed surface only loosely aligns with the experimental results since Pt=CH2 loss requires a tight transition state and is thus disadvantaged relative to the simple bond cleavage. The problem is likely to be with the energies computed for the simple bond cleavage since Chen has previously noted that theory performs better for the energetics of rearrangement reactions in related ligated Pd cations than for bond homolysis [40].

Figure 2
figure 2

Results of DFT calculations at B3LYP-D3/SDD 6-311G(d) level of theory. (a) Plot of the potential energy diagram for both pathways of dissociation of [(phen)M(CH3)]+ with M = 1-Ni (blue), 3-Ni (orange), Pd (red), and Pt (black); (b) structures of key species

Full size image
Table 1 DFT Calculated Reaction Energetics (in kJ mol−1) for Fragmentation of [(phen)Pt(CH3)]+ via Competing Bond Homolysis Pathway (Giving 3 + ·CH3) and Pt=CH2 Pathway
Full size table

A key question regarding the [(phen)M]+· metal complexes is whether they are in the +1 oxidation or +2 oxidation state (i.e., where the phen ligand is a redox non-innocent ligand [79]). An examination of the DFT calculated spin density at the metal centers in the [(phen)M]+· metal complexes reveals (Table S1) that they range from 94.7% for M = Ni to 77.8% for M = Pt. This suggests that the complexes can be regarded as being in the +1 oxidation state and that the phen ligand largely behaves as a redox innocent ligand.

IMR of [(phen)M]+· with the Background Gases O2, N2, and H2O

Water and dinitrogen are always present in the ion trap and readily react with the various three-coordinate metal complexes to form adducts (Eqs. (3) and (4)) [34]. The two-coordinate metal complexes, [(phen)M]+·, react with background water, dinitrogen, and dioxygen to form adducts (Eq. (6), Figure 3). When M = Ni, mainly water and dioxygen adducts are formed, whereas only a minor peak can be observed for dinitrogen adduct formation. For M = Pd, adduct formation only occurs to a minor degree with water and dioxygen. For M = Pt, adduct formation readily occurs for the three background gases, the water and dinitrogen adduct being favored over dioxygen adduct formation.

Figure 3
figure 3

(a–c) LTQ MS4 spectra of mass-selected precursor ions, [(phen)M]+·, 3, undergoing IMR at an activation time of 100 ms with background gases present in the linear ion trap: (a) M = Ni, (b) M = Pd, and (c) M = Pt The mass-selected precursor ions are designated by an asterisk (*), (d) Structures of key species calculated at the M06/SDD 6-31G(d) level of theory

Full size image

The structures and energetics of adduct formation for each metal of the cation [(phen)M]+· have been calculated at the M06/SDD 6-31G(d) level of theory (Figure 3d, Table 2). In the case of water and dinitrogen, an end-on fashioned binding mode is preferred. For dioxygen two stable isomers have been found: one where dioxygen binds in an end-on orientation (η1 hapticity) and one where it binds in a side-on fashion (η2 hapticity), the latter being favored over the former thermodynamically. Overall, the thermodynamics for adduct formation of each background gas with each metal are favorable and follow the order Pt > Ni > Pd, which is consistent with more abundant adduct formation observed in the cases of Pt and Ni (Figure 3c and a) over that of Pd (Figure 3b). The calculated reaction energies for the nitrogen adducts reflect the strength of the orbital overlap, which will be influenced by both the σ and π interactions between the metal center and N2 [80]. These in turn are affected by factors such as electronegativity and relativistic effects [80, 81]. Thus, nickel is a much better σ-acceptor and π- donor than palladium, while relativistic effects operate on platinum to allow more metal-to-ligand back donation. The η2 binding mode for oxygen for related ligated Pd(0) complexes has been suggested to arise from oxidation to give a Pd(II) coordinated peroxide [82] and several such complexes have been isolated and structurally characterized via X-ray crystallography [83].

Table 2 DFT Calculated Reaction Energetics of Forming the Water, Dinitrogen and Dioxygen Adducts of [(phen)M]+· (M = Ni, Pd, Pt) with M06/SDD 6-31G(d) Level of Theory
Full size table

IMR of [(phen)M]+· with Acetone

The metal cations, [(phen)M]+·, form the adducts, [(phen)M(OC(CH3)2)]+·, with acetone in IMR (Eq. (9), Figure 4). In the case of Pt, a minor ion is observed at m/z 418, which corresponds to a loss of a methyl radical from the acetone adduct to form the coordinated acyl anion [(phen)Pt(OCCH3)]+ (Eq. (10)). The analogous peak is observed as a minor peak in the case of acetone-d6, likely due to the operation of a secondary kinetic isotope effect for the formation of [(phen)Pt(OCCD3)]+ [84]. The observed reactivity for the ligated metal cations, [(phen)M]+· contrast to the behavior of the bare metal ions, which activate the C–C bond of acetone to produce ethane as the major product (Eq. (12)) and CO as the minor product (Eq. (12)) [42,43,44,45,46].

$$ {\left[\left(\mathrm{phen}\right)\mathrm{M}\right]}^{+\bullet}\kern0.5em +\kern0.5em \mathrm{OC}{\left({\mathrm{CX}}_3\right)}_2\kern0.5em \to \kern0.5em {\left[\left(\mathrm{phen}\right)\mathrm{M}\left(\mathrm{OC}{\left({\mathrm{CX}}_3\right)}_2\right)\right]}^{+\bullet}\kern0.5em \left(\mathrm{M}=\mathrm{Ni},\mathrm{Pd},\mathrm{Pt}\right) $$
(9)
$$ {\left[\left(\mathrm{phen}\right)\mathrm{Pt}\right]}^{+\bullet}\kern0.5em +\kern0.5em \mathrm{OC}{\left({\mathrm{CX}}_3\right)}_2\kern0.5em \to \kern0.5em {\left[\left(\mathrm{phen}\right)\mathrm{Pt}\left({\mathrm{OCCX}}_3\right)\right]}^{+}+\bullet {\mathrm{CX}}_3 $$
(10)

where X = H or D

Figure 4
figure 4

LTQ MS4 spectra of mass-selected precursor ions, [(phen)M]+·, undergoing IMR at an activation time of 100 MS with acetone (a–c) and acetone-d6 (d) in the linear ion trap: (a) M = Ni, (b) M = Pd, and (c–d) M = Pt. The mass-selected precursor ions are designated by an asterisk (*)

Full size image
$$ {\mathrm{M}}^{+\bullet}\kern0.5em +\kern0.5em \mathrm{OC}{\left({\mathrm{C}\mathrm{H}}_3\right)}_2\kern0.5em \to \kern0.5em {\displaystyle \begin{array}{cc}{\left[\mathrm{M}\left(\mathrm{CO}\right)\right]}^{+\bullet }+{\mathrm{C}}_2{\mathrm{H}}_6& \left(\mathrm{M}=\mathrm{Ni},\mathrm{Pd}\right)\end{array}} $$
(11)
$$ \kern1.92em \to \kern0.5em {\left[\mathrm{M}{\left(\mathrm{C}{\mathrm{H}}_3\right)}_2\right]}^{+\bullet }+\mathrm{CO}\kern0.5em \left(\mathrm{M}=\mathrm{Ni},\mathrm{Pd}\right) $$
(12)

Upon CID, the acetone adducts dissociate to regenerate [(phen)M]+· in the case of M = Ni, Pd (Eq. (13)). For M = Pt, the dissociation is only a minor pathway compared to the loss of a methyl radical to form the acyl complex [(phen)Pt(OCCH3)]+ (Eq. (14), Figure 5, m/z 418). The loss of methyl radical (15 Da) was confirmed by deuterium labeling experiments with fully deuterated acetone, where a mass loss of 18 Da was observed. The differences in branching ratios for OC(CH3)2 (BR Eq. (13) = 19.5%) versus OC(CD3)2 (BR Eq. (13) = 42.5%) and for ·CH3 (BR Eq. (14) = 70.2%) versus ·CD3 (BR Eq. (14) = 45.6%) losses suggest that a secondary isotope effect (kH/kD) operates for the C–C bond cleavage step associated with methyl radical loss. An estimate of kH/kD of 1.5 is derived from BR(·CH3 loss, Eq. (14))/BR(·CD3 loss, Eq. (14)) [84]. This is much smaller than the kH/kD of 5.5 measured for the loss of ethane from the [Ni(OC(CH3)2)]+· complex, which was suggested to arise from a rate-limiting step of C–C bond coupling rather than C–C bond activation [46].

$$ {\left[\left(\mathrm{phen}\right)\mathrm{M}\left(\mathrm{OC}{\left({\mathrm{CX}}_3\right)}_2\right)\right]}^{+\bullet}\kern0.5em \to \kern0.5em {\left[\left(\mathrm{phen}\right)\mathrm{M}\right]}^{+\bullet}\kern0.5em +\mathrm{OC}{\left({\mathrm{CX}}_3\right)}_2\kern0.5em \left(\mathrm{M}=\mathrm{Ni},\mathrm{Pd}\right) $$
(13)
$$ {\left[\left(\mathrm{phen}\right)\mathrm{Pt}\left(\mathrm{OC}{\left({\mathrm{CX}}_3\right)}_2\right)\right]}^{+\bullet}\kern0.5em \to \kern0.5em {\left[\left(\mathrm{phen}\right)\mathrm{Pt}\left({\mathrm{OCCX}}_3\right)\right]}^{+}\kern0.5em +\bullet {\mathrm{CX}}_3 $$
(14)

where X = H or D

Figure 5
figure 5

LTQ MS5 spectra of mass-selected precursor ions, [(phen)M(OC(CH3)2)]+· undergoing CID at a normalized collision energy of 20 (arbitrary units) in the linear ion trap: (a) M = Ni, (b) M = Pd, and (c–d) M = Pt. The mass-selected precursor ions are designated by an asterisk (*)

Full size image

Upon further CID, the acyl complex [(phen)Pt(OCCH3)]+ (m/z 418) loses CO to regenerate the three-coordinate methyl complex [(phen)Pt(CH3)]+ (Eq. (15), m/z 390). In the deuterium labeling experiments, the same complex was generated with a fully deuterated methyl ligand (Fig. S3). Thus, in contrast to the nickel and palladium complexes, [(phen)Pt]+· formally catalyzes the decomposition of acetone into CO and two methyl radicals (Eq. (16), Scheme 1). Although photolysis or thermolysis of acetone gives a range of products, C–C bond homolysis to give methyl radicals is a key channel [85, 86].

$$ {\left[\left(\mathrm{phen}\right)\mathrm{Pt}\left({\mathrm{OCCH}}_3\right)\right]}^{+}\kern0.5em \to \kern0.5em {\left[\left(\mathrm{phen}\right)\mathrm{Pt}\left({\mathrm{CH}}_3\right)\right]}^{+}\kern0.5em +\mathrm{CO} $$
(15)
$$ {\left({\mathrm{CH}}_3\right)}_2\mathrm{CO}\kern0.5em \to \kern0.5em \mathrm{CO}+2\bullet {\mathrm{CH}}_3 $$
(16)
Scheme 1
scheme 1

Catalytic cycle for the decomposition of acetone mediated by [(phen)Pt]+·

Full size image

An examination of the catalytic cycle (Scheme 1) reveals that it suffers from the off cycle Pt=CH2 loss (Eq. (8)) being the dominant pathway, with the regeneration of the catalyst 3 being only a minor product in step 4. Nonetheless, DFT calculations were carried out to examine the energetics associated with all steps and the data is summarized in Table 3. Adduct formation is exothermic for step 1, consistent with the observation that this reaction occurs under the IMR conditions. Three different adducts (9a–9c) were located (Figure 6b). For step 2 of the cycle, the transition state TS9a-10 for activating the C–C bond of acetone to form a methyl acyl intermediate [(phen)Pt(CH3)(OCCH3)]+ (10) is favored over the dissociation of the adduct to give 3 (Figure 6, black surface). For the Pd (red surface) and the Ni complexes (which can give either the singlet (blue surface) or triplet (orange surface)), however, the formation of an acyl complex is disfavored both thermodynamically and kinetically, as the energies of the transition state and the acyl complex are higher than the energy needed for the dissociation of the adduct. The formation of the acyl complex 11 requires energy, which is delivered upon CID. However, the dissociation of the intermediate adduct(s) 9a-c via TS9a-10 requires less energy than the formation of the acyl complex. The energy difference of 8 kJ mol−1 lies within the error of the functional used [78].

Table 3 DFT Calculated Reaction Energetics of the Catalytic Cycle of the Decomposition of Acetone Mediated by [(phen)Pt]+· with M06/SDD 6-31G(d) Level of Theory
Full size table
Figure 6
figure 6

DFT calculations at M06/SDD 6-31G(d) level of theory. (a) Plot of the potential energy diagram for acetone adduct formation and subsequent methyl loss for [(phen)M(CH3)]+ with M = 2-Ni (followed by 1-Ni (to give singlet product: blue), 3-Ni (to give triplet product: orange), Pd (red) and Pt (black); (b) structures of key species

Full size image

The decarbonylation of 11 in step 3 to regenerate the methyl complex, [(phen)Pt(CH3)]+(2), and carbon monoxide requires 139 kJ mol−1 (Fig. S4), as found for the related decarbonylation of the bidentate enolate complex [(phen)Pt((OC(CH2)CH3)]+ [47]. The transition state TS11b-12 lies only 1 kJ mol−1 above the acyl complex 11b, while the intermediate 12, in which CO is still coordinatively bound to the metal center, lies below 11b (− 102 kJ mol−1). Finally, the energetics associated with regeneration of the catalyst, 3, has already been discussed in detail in “Bond Homolysis vs. Carbene Formation.”

IMR with Allyl Iodide

The IMR spectra of [(phen)M]+· with allyl iodide are given in Figure 7. For all three metals, Ni, Pd, and Pt the formation of an allyl complex via allyl abstraction are observed (Eq. (17)). A previous report noted that [(phen)Ni]+· reacts via allyl abstraction as the major reaction together with the formation of the minor product due to the addition of allyl iodide [41]. For M = Pt two further products could be observed: the iodo complex [(phen)Pt(I)]+ and the diiodoallyl complex [(phen)Pt(I)2(CH2CHCH2)]+ (Eqs. (18) and (19)). The iodo complex also reacts with background gases to form the respective adducts (Eq. (20)).

Figure 7
figure 7

(a–c) LTQ MS4 spectra of mass-selected precursor ions, [(phen)M]+·, undergoing IMR at an activation time of 30 ms with allyl iodide in the linear ion trap (infusion rate 50 μL/h): (a) M = Ni, (b) M = Pd, and (c) M = Pt. The mass-selected precursor ions are designated by an asterisk (*); (d) structures of key species

Full size image
$$ {\left[\left(\mathrm{phen}\right)\mathrm{M}\right]}^{+\bullet}\kern0.5em {\mathrm{CH}}_2{\mathrm{CH}\mathrm{CH}}_2\mathrm{I}\kern0.5em \to \kern0.5em {\left[\left(\mathrm{phen}\right)\mathrm{M}\left({\mathrm{CH}}_2{\mathrm{CH}\mathrm{CH}}_2\right)\right]}^{+}+\bullet \mathrm{I}\kern0.50em \ \left(\mathrm{M}=\mathrm{Ni},\mathrm{Pd},\mathrm{Pt}\right) $$
(17)
$$ {\left[\left(\mathrm{phen}\right)\mathrm{Pt}\right]}^{+\bullet}\kern0.5em {\mathrm{CH}}_2{\mathrm{CH}\mathrm{CH}}_2\mathrm{I}\kern0.5em \to \kern0.5em {\left[\left(\mathrm{phen}\right)\mathrm{Pt}\left(\mathrm{I}\right)\right]}^{+}+\bullet {\mathrm{CH}}_2{\mathrm{CH}\mathrm{CH}}_2 $$
(18)
$$ {\left[\left(\mathrm{phen}\right)\mathrm{Pt}\left(\mathrm{I}\right)\right]}^{+}\kern0.5em +\kern0.5em {\mathrm{CH}}_2{\mathrm{CH}\mathrm{CH}}_2\mathrm{I}\kern0.5em \to \kern0.5em {\left[\left(\mathrm{phen}\right)\mathrm{Pt}{\left(\mathrm{I}\right)}_2\left({\mathrm{CH}}_2{\mathrm{CH}\mathrm{CH}}_2\right)\right]}^{+} $$
(19)
$$ {\left[\left(\mathrm{phen}\right)\mathrm{Pt}\left(\mathrm{I}\right)\right]}^{+}\kern0.5em +\kern0.5em \mathrm{L}\kern0.5em \to \kern0.5em {\left[\left(\mathrm{phen}\right)\mathrm{Pt}\left(\mathrm{I}\right)\left(\mathrm{L}\right)\right]}^{+} $$
(20)

The energetics of the observed reactions have been calculated (Figure 7d, Table 4). The energetics of the three reactions giving 13, 14, and 15 as singlets become more favorable going down group 10. The allyl abstraction is thermodynamically favored for each metal, whereas the iodine abstraction is only favored for platinum. The formation of [(phen)M(I)2(CH2CHCH2)]+ is favored for each metal, but can only be observed for platinum. This is consistent with the calculated thermochemistry for the stepwise formation of [(phen)M(I)2(CH2CHCH2)]+, which requires the formation of [(phen)M(I)]+, which is only favored for platinum. Since the ground state of nickel complexes can vary, we also calculated the thermochemistry for the formation of the triplet states of 13, 14, and 15. Just as the parent three-coordinate complex [(phen)Ni(CH3)]+ has been shown to be more stable in the triplet state [47], so too are the triplets of the nickel complexes of 14 and 15. For 13, however, the singlet state in which the allyl ligand adopts a η3 binding mode is more stable than the triplet in which allyl ligand binds in a η1 fashion. Similar differences in the singlet versus triplet stabilities have been calculated for the formation of [(phen)Ni(CH2CHCH2)]+ and [(phen)Ni(Br)]+ in the reactions of allyl bromide with [(phen)Ni]+· [41]. Overall, the theoretical results are consistent with the experiments: only singlet formation is expected to be kinetically viable since triplet formation requires the participation of an electron from a filled d orbital on the reactant ion. Formation of singlet 13 is exothermic in all cases and is observed for all metals (Figure 7). Formation of singlet 14, the precursor to formation of singlet 15, is only exothermic for Pt, which is why formation of 14 (via Eq. (18)) and 15 (via Eq. (19)) is only observed in the IMR of [(phen)Pt]+· (Figure 7c).

Table 4 DFT Calculated Reaction Energetics for the Formation of Products 13–15 in the Reactions of [(phen)M]+· with Allyl Iodide with M06/SDD 6-31G(d) Level of Theory
Full size table

Conclusion

Gas-phase studies allow a systematic evaluation of the role of ligand and metal oxidation state on reactivity. By studying all members of a group, it is possible to relate changes in reactivity to electronic effects. In previous work, we have shown that only [(phen)Pt(CH3)]+ reacts with acetone to form the enolate [(phen)Pt(OC(CH2)CH3)]+ [47]. Here, we have shown that MSn experiments in a linear ion trap mass spectrometer readily yield the ligated group 10 metal complexes [(phen)M]+· in the formal oxidation state of +1, thereby allowing their bimolecular reactivity to be explored. Of all the metals, Pt stands out once again for two of the reactions studied. In the fragmentation of [(phen)Pt(CH3)]+, the desired bond homolysis is only a minor pathway, while the loss of the neutral carbene Pt=CH2 dominates. The acetone adduct [(phen)Pt(OC(CH3)2)]+· fragments via methyl radical loss to form the coordinated acyl complex, [(phen)Pt(OCCH3)]+, a reaction that is unique for Pt. It will be interesting to explore the gas-phase formation and reactivity of other [(L)M]+· complexes in which the ligand, L, exerts different geometrical and electronic effects to 1,10-phenanthroline. One class of ligands that would be particularly interesting are phosphine-based ligands. Previous studies suggest that alternative “synthetic” strategies to CID of carboxylate complexes are required since decarbonylation rather than decarboxylation occurs for [CH3CO2Pd((PPh2)2CH2)]+ [87]. Given the widespread success of photoactivated C–I bond homolysis [88] and that Pd-azide bonds in Pd(II) complexes are known to undergo photoactivated bond homolysis [89], we are exploring whether related M–X photoactivated bond homolysis could be used to prepare [(L)M]+· from suitable precursors.