Detection of Intermediates in Dual Gold Catalysis Using High

11 mins ago - We have probed for reaction intermediates involved in the dual-gold-catalyzed activation of a conjugated 1,5-diyne substrate and its fur...
1 downloads 6 Views 2MB Size
Article Cite This: Organometallics XXXX, XXX, XXX−XXX

Detection of Intermediates in Dual Gold Catalysis Using HighResolution Ion Mobility Mass Spectrometry Jean-François Greisch,†,‡ Patrick Weis,† Katrina Brendle,† Manfred M. Kappes,*,†,‡ Jean R. N. Haler,§ Johann Far,§ Edwin De Pauw,§ Christian Albers,∥ Sarah Bay,⊥ Thomas Wurm,⊥ Matthias Rudolph,⊥ Jürgen Schulmeister,⊥ and A. Stephen K. Hashmi⊥ †

Institute of Physical Chemistry, KIT, D-76131 Karlsruhe, Germany Institute of Nanotechnology, KIT, D-76344 Eggenstein-Leopoldshafen, Germany § Mass Spectrometry Laboratory, University of Liège, MolSys Research Unit, B-4000 Liège, Belgium ∥ Bruker Daltonic GmbH, Bremen, Germany ⊥ Organic Chemistry Institute, Heidelberg University, D-69120 Heidelberg, Germany ‡

S Supporting Information *

ABSTRACT: We have probed for reaction intermediates involved in the dual-gold-catalyzed activation of a conjugated 1,5-diyne substrate and its further coupling to benzene in the liquid phase. This was done by sampling the reaction mixture by electrospray ionization followed by high-resolution ion mobility mass spectrometryunder conditions allowing for the resolution of structural isomers differing in their collision cross sections by less than 0.5%. For the cationic mass corresponding to catalyst + diyne (activation stage) we resolve four isomers. At the mass corresponding to catalyst + diyne + benzene, two isomers are observed. By comparing the experimentally obtained cross sections to those inferred for model structures derived from density functional computations, we find our measurements to be consistent with the proposed solution mechanism. This constitutes the first direct observation of intermediates in dual gold catalysis and supports the previous inference that the mechanism involves cooperative interactions between two gold centers.



INTRODUCTION Homogeneous gold catalysis1−21 has evolved into a highly innovative tool for organic synthesis.22−24 The classical reactivity patterns of mononuclear gold catalysis, involving typical intermediates such as π complexes of gold, vinylgold intermediates, and (mononuclear) gold carbene complexes, are quite well understood.16 On the other hand, in the field of dual gold catalysis,25 which is thought to involve cooperative interactions between two gold centers, the structures of most of the corresponding binuclear gold complexes have only been inferred from indirect experimental evidence26 and computational investigation.27−31 This reflects the limited dynamic range of the analytical methods which have so far been used to probe the reaction mixtures. Reactive and therefore short-lived intermediates generally have concentrations which are too low for in situ detection: e.g., by solution-phase NMR. Similarly, efforts to isolate early intermediates of the catalytic cycles from these reaction mixtures have so far failed. Whereas the electronic properties of gold cannot be influenced by ligand variation sufficiently to stabilize the dual gold intermediates for isolation in macroscopic amounts, this is not the case if one of the gold centers is substituted by another transition metal such as platinum or ruthenium. Then, bimetallic intermediate species analogous to those proposed © XXXX American Chemical Society

for dual gold catalysis become accessible. This has allowed reactivity patterns proposed for dual gold catalysis to be explored and observed in stoichiometric reactions involving the combination of gold and ruthenium.32 While this proves the principal existence of cooperative catalytic reactivity patterns involving two transition-metal centers, analogous interactions between two gold centers (and the substrate(s)) have remained unproven. Taking into account the increasing possibilities and the growing number of applications of dual gold catalysis,33−35 further mechanistic insight into such processes is clearly important. One method for the direct isolation/detection of short-lived intermediates in homogeneous catalysis is mass spectrometry (MS).36 Correspondingly, MS has already been applied to various aspects of gold catalysis.37−40 Apart from determination of the mass to charge ratios of gold-containing species electrosprayed from reactive solutions, hyphenated mass spectrometric methods are also increasingly being used. In particular, threshold collision induced dissociation (CID), IRUV double resonance spectroscopy, and IR multiphoton dissociation spectroscopy (IRMPD), each combined with a Received: March 2, 2018

A

DOI: 10.1021/acs.organomet.8b00128 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics

predicted collision cross sections) derived from quantum chemical computations at the level of density functional theory. These are in turn based on mechanistic concepts and associated reaction intermediates proposed for this system in previous solution studies. The high level of agreement between experiment and the proposed mechanism indicates that ESITIMS-TOFMS is a useful new tool for mechanistic studies in dual gold catalysis.

prior mass selection step, have been applied. In combination with quantum chemical calculations typically at the DFT (density functional theory) level, such experiments allow obtaining useful structural and/or thermochemical information on key reaction intermediates. The Roithová group has pioneered this approach also toward exploring the structures of mass-selected binuclear gold complexes related to gold catalysis.41−43 Both σ,π-dicoordinated and gem-diaurated structures have been inferred in separate studies. Hyphenated mass spectrometry experiments, as applied to catalytically active solutions, are particularly technically demanding because they require isolation of key intermediates present in only very low concentrations. Consequently, they are limited by the volatilization efficiency (electrospray ionization (ESI)), mass resolution, structure selectivity, sensitivity, and dynamic range available for the technique combination used. Here, we have applied a combination of ion mobility and mass spectrometry (IM-MS) to elucidate the mechanism of dual gold catalysis. IM-MS allows determination of the mobility and thus the collision cross section of an ion of interest. This in turn allows a structure assignment in combination with structural models from theory. Trapped ion mobility time of flight mass spectrometry (TIMS-TOFMS) is a recently introduced variant of IM-MS which hyphenates ion mobility and mass spectrometry to achieve unprecedented mobility resolving power and dynamic range. We have used this method for the first IM-MS study of intermediates in dual gold catalysisas obtained by electrospray monitoring of the reaction mixture. As a model reaction, we have selected a gold-catalyzed hydroarylating cyclization of a 1,5-diyne substrate26 run in a tetrahydrofuran solution in the presence of a cationic dual gold activation catalyst (bearing a triflimide counteranion in the solid state).44 Benzene as nucleophile was added until the solutions reached 1% v/v. The cationic catalyst mediates the addition of a 1,5-diyne substrate to benzene to generate a β-naphthalene derivative. The reaction is activated by initial ring closure (indicated in Scheme 1).



METHODS

Synthesis and Sample Preparation. The synthesis of the precatalyst was carried out according to the literature.44 Stock tetrahydrofuran (THF) solutions of the precatalyst and the 1,5-diyne substrate (1,2-diethynyl-4,5-dimethylbenzene) were mixed directly prior to the TIMS-TOFMS measurements to yield a final solution of 10−4 M in the precatalyst and of 10−3 M in the diyne typically solubilized in THF. This generates the active catalyst complexed with the diyne substrate. For the TIMS-TOFMS (timsTOF, Bruker, Bremen) experiments involving coupling of benzene to the preactivated diyne substrate, benzene was added to the solution directly before the measurements to reach 1% v/v. Due to safety and regulatory reasons, TIMS experiments, which involve a standard electrospray source, using pure benzene (as solvent as well as reagent) were not possible. A comparison of the mass spectra of the (pre)catalyst and substrate mixture in pure benzene, toluene, and THF solutions was separately performed using a nanoelectrospray source on a SYNAPT G2S HDMS (Waters, Manchester). Trapped Ion Mobility Spectrometry. Trapped ion mobility measurements were carried out on two Bruker timsTOF instruments, one in Bremen, Germany, and the other in Liège, Belgium, using nitrogen as carrier gas. Measurement conditions are described in the Supporting Information. Briefly, electrosprayed ions are transferred into a TIMS cell, where they are first trapped and then sequentially released into a mass spectrometer according to their different mobilities. Identical separations were obtained on the two setups. As described in our study of coordination complexes,45 the experimental mobilities K = νg/E = A/(Urelease − Uout) of ions sequentially released from the TIMS cell after trapping depend on the corresponding trapping electric field window E ± ΔE and the velocity of the carrier gas vg. This is more conveniently expressed by a single calibration constant A, the release voltage Urelease, and the voltage Uout applied to the tunnel exit. Experimental ion mobilities can be converted into (uncorrected) collision cross sections (Ω) using the Mason−Schamp equation:

Scheme 1. Overview of the Dual-Gold-Catalyzed Reaction of Interest

Ω=

=

1/2 (18π )1/2 ze ⎡ 1 1 ⎤ 1 T 101325 1 + ⎢ ⎥ 16 (kBT )1/2 ⎣ m1 m2 ⎦ K 273.15 P N0 1/2 (18π )1/2 ze ⎡ 1 1 ⎤ 1 1 + ⎢ ⎥ 16 (kBT )1/2 ⎣ m1 m2 ⎦ K 0 N0

where ze is the charge of the ion, kB is the Boltzmann constant, N0 = P0/kBT0 is the number density, and m1 and m2 refer to the masses of the ion and bath gas, respectively. In the present work we have used the approach described by Haler and co-workers in ref 46 to extract N2 collision cross sections. This relies on the use of several singly charged sodium adducts of poly(ethylene oxide) monomethyl ether (CH3O-PEO-H) as calibrants for both collision cross sections and reduced ion mobilities, K0 = (P/ 101325)(273.15/T)K. The choice of sodium adducts of CH3O-PEOH as calibrant is motivated by (1) their predictable, largely temperature independent (near-spherical) shape,46−49 (2) the broad range of collision cross-section values (as provided by typical polydisperse samples) which scale monotonically with the number of monomers,46,49 and (3) the straightforward, robust, and reproducible interpolation and extrapolation of their calibration curves.46 Figure S1 in the Supporting Information shows a calibration curve obtained by contrasting TIMS-TOFMS measurements of inverse

Previous work in solution indicates that the reaction proceeds via either a 5-endo (Scheme 1, top) or 6-endo (Scheme 1, bottom) ring closure which gives rise to a gold vinylidene or diaurated phenyl cation, respectively. Possibly both types of activation reaction occur. Our measurements resolve four different ionic species at the mass corresponding to the activation stage (diyne substrate + catalyst). We observe two further isomers at the “preproduct” formation stage (mass of β-naphthalene product + catalyst). We rationalize the corresponding species in terms of structural models (and B

DOI: 10.1021/acs.organomet.8b00128 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics ion mobility with analogous results obtained using a N2 drift tube IMMS systemfor the same sodium adducts of various CH3O-PEO-H oligomers. The effect of in source collisional activation on the abundance of the observed species was also studied. Using CH3O-PEO-H as a reference substance, no shift in the mobilograms was observed. The only difference lies in the abundance of some of the species, as discussed later. Density Functional Computations. All quantum chemical calculations were conducted by employing the Gaussian 09 program package.50 The computational approach is based on density functional methodologies using the B3LYP51 functional. Dispersion effects were treated according to Grimme.52 The correlation-consistent double-ζ basis (cc-pVDZ)53 was used as the basis set. For gold, additional relativistic effective core potentials from the the Stuttgart group were employed.54 The gas-phase local minima, optimized at the B3LYP-d3/ cc-pVDZ level of theory, were verified to exhibit no imaginary frequency. Computations of the N2 Collision Cross Sections and Comparison to Experiments. Comparison of experiment with DFT computations and corresponding structural assignments were achieved via the calculation of N2 collision cross sections for DFT structures using a trajectory method with uncut Lennard−Jones potentials as implemented in IMoS.55,56 The temperature parameter was varied to assess its influence on the computed cross sections. A temperature of 298 K (room temperature of 293 K) for the buffer gas was found to yield optimum agreement between computed and experimental cross sectionson comparison of the most stable (lowest energy) computed structures with the most abundant isomers for intermediates observed at m/z 1323.576 and 1401.623or m/z 1407.660 in case of deuterated benzene (see Figure 1). Additionally, global deviation was found to be minimal for the assignment of the other peaks of the mobilograms when this temperature was assumed (see Tables S1 and S2 in the Supporting Information). Before we enter into a thorough discussion of the experimental results, it is worth mentioning that (1) for the intermediates studied,

direct nitrogen−gold interactions are generally avoided due to the bulkiness of the 1,3-bis[2,6-bis(propan-2-yl)phenyl]-1,3-dihydro-2Himidazol-2-ylidene ligands attached to the gold centers (as can be readily seen from their CPK molecular representations, Figure S2) and (2) the species studied here are only weakly polar (see Table S3, which gives the computed dipole moments of the species studied). Therefore, we expect that the trajectory method approach used to determine N2 collision cross sections from DFT structures should be quite accurate.



RESULTS We are interested in the mechanism of the homogeneous catalytic process by which the cationic catalyst (CAT) mediates the addition of benzene (SB) to a 1,5-diyne (SA···H) substrate in room-temperature solution to generate a β-naphthalene derivative (SA−SB···H). This is initialized using a precatalyst (P-CAT with PC3H3, m/z 1209.532; see structure in Figure S2), which in a first stage reacts with the 1,5-diyne substrate (SA···H) to generate species with m/z 1323.576 corresponding to CAT + SAthereby displacing the “protecting” C3H3 group via release of C3H4. This is followed in a second stage by the addition of benzene to generate species with m/z 1401.623 (CAT + SA + SB). Presumably, in a third stage, the actual producta β-naphthalene derivative (SA−SBH), which we are not sensitive to because it is neutralis displaced from a charged “pre-product” of mass m/z 1401.623 by insertion of a new SA···H unit and the catalytic cycle starts over. While the m/z ratios given above correspond to the lowest isotopomer in each case, they are hereafter reduced to four-digit unrounded numbers for convenience. In the following section we first present electrospray mass spectra obtained for the precatalyst dissolved in benzene, toluene, and THF solutions in the presence of the 1,5-diyne substrate (SA···H), which document that the reaction of interest is associated with two detectable sets of singly positively charged intermediates (CAT + SA at m/z 1323 and CAT + SA + SB at m/z 1401 or 1407 in the case of deuterated benzene). We then present trapped ion mobility measurements which indicate that each of these detectable masses comprises multiple isomers. The corresponding experimental ion mobilities are then converted to N2 collision cross sections via calibration against singly charged sodium adducts of poly(ethylene oxide) monomethyl ether. Finally we compare these experimental values with collision cross sections calculated using IMoS,55,56 on the basis of candidate structures computed at the DFT level. MS Measurements. Normally, the liquid medium for the dual gold catalysis studied here is pure benzene, which also acts as a substrate (SB) in the second stage of the catalytic process. Figure 1 shows that under these conditions the CAT + SA intermediates react so quickly that only precatalyst (m/z 1209) and CAT + SA + SB (m/z 1401 or 1407 in the case of deuterated benzene) are observable. Using toluene instead of benzene appears to slow the overall reaction down slightly, and as a result a signal at the mass of CAT + SA (m/z 1323) emerges. Substituting toluene by THF leads to a further enhancement of the abundance of these m/z 1323 intermediates. In contrast, adding benzene to the THF solution (not shown) depopulates the intermediates to again yield CAT + SA + SB. By using deuterated benzene we could unambiguously show that under these conditions a hydrogen atom migrates from the benzene ring to the rearranged substrate (the mass to charge ratio 1406, instead of 1407, would

Figure 1. Nanoelectrospray mass spectrum of a solution of the precatalyst (m/z 1209; P-CAT) upon addition of the diyne substrate (forming the CAT + SA complex at m/z 1323) in (a) deuterated benzene (C6D6), (b) toluene, (c) THF (with traces of water), and (d) THF at approximately 10 times lower precatalyst concentration (leading to enhanced CA + SA relative to the P-CAT (as well as enhanced side products with THF)). The m/z 1323 intermediate reacts with the solvents to yield the m/z 1407 (addition of deuterated benzene), m/z 1415 (addition of toluene), m/z 1395 (addition of THF), and m/z 1413 (addition of THF and H2O) species, respectively. C

DOI: 10.1021/acs.organomet.8b00128 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics be observed if THF solvent molecules were somehow involved). The fact that the same charged species are observed for a pure benzene solution and a THF solution with benzene added suggests that the same catalytic pathway is active in both cases. In comparison to pure benzene, THF slows down the reaction, making the intermediates more easily observable. This could be due to its dipole moment of 1.63 D (ref 57), which may contribute to stabilize some of the latter CAT + SA intermediates via nonbonding interactions. Overall, all three solutionsbenzene, toluene, and THFare catalytically active with the diyne substrate being consumed fastest in benzene. A detailed kinetic study of the reaction system using mass spectrometric methods will be left for a subsequent paper. Instead, the focus here is on the isolation and structural characterization of possible reaction intermediates as well as the implications of our observations for the catalytic pathway of the diauryl catalyst studied (strictly speaking for the case of a solution comprising 1% v/v benzene in predominantly THF). TIMS Measurements. Typical mobilograms (plots of normalized ion intensity versus inverse reduced mobility) of the charged species pertinent to this study are displayed in Figure 2 (large TIMS ramp measurement with resolution of

Table 1. Experimental and Computed N2 Collision Cross Sections of the Reaction Intermediates with m/z 1323 with Relative Energies in Parentheses exptl. resolved species

av. 1/K0 (cm2 V−1 s−1)

I1 I2 I4 I3

1.6340 1.66(0) 1.7899 1.7637

av. exptl. CCS (Å2) 336.4 341.6 367.7 362.5

± ± ± ±

0.5 1.0 0.6 0.6

comput. species no. (rel. en. (kJ/mol))

CCS@298 K (Å2)

1a (0)/1b 2 (40) 3 (57) 4 (45) 5 (32)

336.7/329.8 342.2 366.3 361.8 359.7

Table 2. Experimental and Computed N2 Collision Cross Sections of the Reaction Intermediates at m/z 1401 with Relative Energies in Parenthesesa exptl. resolved species

av. 1/K0 (cm2 V−1 s−1)

av. exptl. CCS (Å2)

P2 P1

1.6759 1.6449

344.8 ± 0.6 338.5 ± 0.6

comput. species no. (rel. en. (kJ/mol))

CCS@298 K (Å2)

A (0) B (−173) C (−373) D (−128) E (−414)

365.4 346.0 339.7 361.8 331.1

Note that as for the intermediates (I1−I4, P1, and P2), the cross section determined for the precatalyst (P-CAT) is consistent with our DFT calculations (see the Supporting Information). However, the comparatively small propynyl protecting group leads to multiple rotamer structures lying close in energy. It is unclear which are present in experiment. a

charginga problem which can be recognized and eliminated by systematically reducing the corresponding ion abundance. The resulting calibrated experimental cross section for P-CAT is 316.2 ± 0.6 Å2 (see the Supporting Information). Computed Structures and Collision Cross Sections. Reactants, products, and proposed intermediates for the catalytic reaction studied are displayed in Scheme 2 (see Scheme 2. Proposed Detailed Mechanistic Pathways for the Hydroarylating Aromatization of a 1,5-Bis-Terminal Diyne As Catalyzed by the Synergistic Interplay of Two Active Gold Centers

Figure 2. Mobilograms of the species investigated in the present work: (a) the m/z 1209 precatalyst (formally [Au2C54N4H72···C3H3]+ (black; P-CAT)), (b) the m/z 1323 intermediate of reaction with diyne ([Au2C54N4H72···C12H9]+ (red; I1−I4)), and (c) the m/z 1401 adduct with benzene ([C6H6···Au2C54N4H72···C12H9]+ (blue; P1 and P2)). Note: the signals for each m/z are normalized to the most intense isomers.

150). The extracted inverse ion mobilities and corresponding collision cross sections obtained by calibration against singly charged sodium adducts of poly(ethylene oxide) monomethyl ether (see Figure S1) are given in Table 1 for the reaction intermediates at m/z 1323 (CAT + SA) and in Table 2 for the “product” species at m/z 1401 (CAT + SA + SB). We have analyzed all resolvable isomeric species detected with signal to noise ratios better than 10:1 (see also reproducibility control measurements of mobilograms on two adjacent isotopomer masses as well as logarithmic representations of ion intensity vs inverse reduced mobility in Figures S3−S5 in the Supporting Information). Note that due to its much larger ion abundance in comparison to the intermediates, the precatalyst trace also shown in Figure 2 is subject to space D

DOI: 10.1021/acs.organomet.8b00128 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics Discussion for details). The early CAT + SA intermediates corresponding to catalyst + conjugated 1,5-diyne substrate (minus one hydrogen) are numbered (red), while the later CAT + SA + SB intermediates (or “products”) are alphabetically labeled (blue). The possible intermediates are labeled 1−5 (red, CAT + SA) and A−E (blue, CAT + SA + SB). The reaction is initiated by P-CAT (gray). L refers to 1,3-bis[2,6-bis(propan-2-yl)phenyl]1,3-dihydro-2H-imidazol-2-ylidene ligands. We have performed DFT calculations as described above to model all intermediates potentially accessed in the proposed two-part mechanism 1−5 for m/z 1323 and A−E for m/z 1401. The resulting optimized structures were then used to obtain N2 collision cross sections via the trajectory method as described in Methods. The results are shown in Figures 3 and 4, respectively.

Figure 4. Structures, total energies (relative to the most stable isomer) at the DFT B3LYP-D3/cc-pVTZ level of theory, and predicted collision cross sections computed for N2 buffer gas using IMoS (Lennard−Jones interaction potentials) of possible intermediates comprising catalyst, diyne substrate, and benzene adduct (m/z 1401 for C6H6 and 1407 for C6D6 adduct) according to the proposed mechanism.

suggesting the involvement of isomer interconversion processes. Therefore, while we assume in the following that all detected isomers were present as such in solution, we do not discuss the implications of their relative intensities for solution kinetics. In future work it will be interesting to study the activation energy for possible isomer interconversion in the gas phase by systematically varying experimental conditions (and thus internal excitation levels) concomitantly with the estimation of the interconversion barriers using appropriate calculations. Previous Mechanistic Inferences from CondensedPhase Studies. In order to put the present ESI-IM-MS measurements into context, it is useful to consider further what is known about the mechanism of related digold catalysis from previous solution studies. A fast formation of acetylides is crucial for a selective reaction toward the β-naphthalene product. This was inferred by the analysis of the reaction selectivity at different conversion rates which, in the absence of basic additives, gives rise to the α-benzene addition product in the early stage of the reaction, while at later stages of the reaction only the β-product was formed.26 This led not only to the conclusion that a synergistic interplay of σ- and π-activation is mechanistically relevant but also to the design of the σ,πdicoordinated propyne complexes as precatalysts (as used here).58 By transfer of the two gold fragments onto the diyne substrates at the beginning of the reaction (simultaneously), initiation phases could be completely suppressed and perfect selectivity toward the β-product was obtained. Further proof for the participation of acetylide complexes was obtained by the conversion of nonsymmetrically substituted monoaurated dialkyne systems that led to the selective incorporation of the benzene moiety at the former acetylide moiety.26 Nevertheless, due to the high reactivity of the system, none of these earlystage intermediates had yet been directly detected in the condensed phase and the predictions of these intermediates were based only on the aforementioned experiments and on the conversion of mixed metallic Ru/Au species that showed related reactivity patterns.32

Figure 3. Structures, total energies (relative to the most stable isomer) at the DFT B3LYP-D3/cc-pVDZ level of theory, and predicted collision cross sections computed for N2 buffer gas using IMoS (Lennard−Jones interaction potentials) of possible intermediates comprising catalyst and diyne substrate (m/z 1323) in the proposed mechanism. In addition to renditions of the corresponding threedimensional structures we also show reduced schematics of the molecules without their stabilizing IPR ligands to aid visualization (see also Scheme 2). Species 1 is characterized by the existence of two rotamers (rotation of the ligands relative to the substrate).



DISCUSSION Probing Solution Compositions by ESI-MS. While it is relatively easy to transfer large molecules from solution into the gas phase using electrospray ionization, this may be associated with significant collisional excitation. In particular for noncovalent complexes, significant changes in gas-phase structures may be induced relative to their solution-phase equilibrium forms. In the experiments reported here, conditions were generally optimized to obtain all of the discussed intermediates with a signal to noise ratio better than 10:1. In order to assess the effect of the experimental parameters on the observed species, in-source collisional activation was performed. While the ion mobilities of the species studied did not change, the relative intensities of some of the contributing isomers were found to depend on these parameters to some degree. In particular, the relative abundance of the peak labeled I3 was found to increase upon in-source collisional activation, thereby E

DOI: 10.1021/acs.organomet.8b00128 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics

cross section, species 5 is tentatively ruled out because it fits slightly less well to I3 than does species 4. We can resolve two species, P1 and P2, with m/z 1401 distinguishable on the basis of their mobilities (Figure 2 and Figure S5). Their relative intensities are comparable. On the basis of DFT calculations and corresponding IMoS predictions of collision cross sections for the proposed intermediates A−E, we tentatively assign the experimentally observed features to CAT + SA + SB isomers as follows: P1 to B and P2 to C. At present, we refrain from assigning minor ion signals recorded at larger inverse mobilities because their S/N levels were below 10 (they may be related to species A and/or D). Interestingly, we find no evidence for the presence of species E, possibly because the product molecules are too rapidly displaced by the diyne substrate to re-form 1. It is noteworthy that this species was isolated only if stoichiometric amounts of gold were present. This scenario is not comparable with the catalytic cycle as studied herebecause the complete starting diyne is rapidly consumed, preventing further turnover.

The next steps of the reaction are thought to involve either an intramolecular 6-endo-dig or an alternative 5-endo cyclization (see Scheme 2, which shows early intermediates numbered in red (1−5) and the later intermediates alphabetically labeled in blue (A−E)). Hints for a 5-endo cyclization are derived from the obtained benzofulvene products that are obtained if one of the terminal alkyne units in the starting material is replaced by an arylalkyne. In that case, the arene unit is believed to intramolecularly insert into a species resembling vinylidene 3. The obtained fulvene structure in that case underlies the possibility of a 5-endo cyclization, at least for closely related diyne systems.59 In addition a 6-endo pathway is also reasonable. This assumption is based on the connectivity of the products obtained by the intramolecular insertion of CH bonds into gold-stabilized benzene cations as reported by the groups of Zhang60 and Hashmi,61 which were observed for nonaromatic or thiophene-tethered diyne backbones. Both possible cyclization modes must be triggered by intermediate 2, which is formed after the shift of one activated (cationic) gold fragment from starting complex 1 onto the terminal alkyne, allowing for π-activation. After this internal rearrangement a direct nucleophilic attack of the gold acetylide onto the second alkyne unit takes place. This gives rise to the mentioned 6endo-dig (4/5) and/or 5-endo-dig cyclization (3). In the case of a 5-endo cyclization, a possible pathway consists of a benzene addition onto gold vinylidene 3 under formation of complex A, which then rearranges to carbene intermediate B after a proton shift. Ring expansion and subsequent aromatization of intermediate C should then deliver the isolable and completely characterized species E.26 If a 6-endo pathway would be favored, the initial cyclization would lead to the 1,3-diaurated phenyl cation 4, which rearranges to its more stable 1,4-substituted derivative 5 by a 1,2-gold shiftfor details upon these potential pathways, including DFT calculations, see ref 61. Benzene addition onto this highly electrophilic species via complex D and subsequent aromatization/protodemetalation would then deliver the same gemdiaurated species E as in the former case. A catalyst transfer from this species onto a new substrate molecule, which is believed to take place in a stepwise manner via the dinuclear cluster, then releases the naphthalene product. For a detailed computational analysis of this process see ref 28. Mechanistic Implications of TIMS-TOFMS Measurements. Overall Scheme 2 comprises two, initially congruent catalytic cycles which bifurcate at the ring closure stage and reconnect at the last step E prior to displacement of the product molecule by a new SA···H substrate to re-form 1thus beginning a new catalytic cycle. Thus, within this picture there are two possible pathways: 1→2→3→A→B→C→E



CONCLUSION We have used electrospray ionization coupled with highresolution trapped ion mobility spectrometry and mass spectrometry to study the cationic species present under reaction conditions in a prototypical homogeneous dual gold catalysis system. Specifically, we probed the gold-catalyzed hydroarylating aromatization of a 1,5-bis-terminal diyne to form a β-naphthalene derivative. The reaction is catalyzed by a dual activation catalyst in a room-temperature solution under ambient conditions. For the first time we have succeeded in clearly resolving several charged species which can be assigned as intermediates. In particular we identify four isomers at m/z 1323 (corresponding to catalyst + diyne substrate), all of which can be reconciled with the previously proposed reaction mechanism. In addition to the initially formed σ,π-gemdiaurated alkyne complex 1, our measurements imply the presence of species 2, in which the gold catalyst is transferred to the second alkyne. Furthermore, we have obtained evidence for an aurated phenyl cation as well as a gold vinylidene, which suggests that both potential pathways can operate simultaneously due to flat energy surfaces.61 Two other species are detected at m/z 1401 (presumably corresponding to the further addition of benzene onto vinylidene 3). Overall, the high level of agreement with a mechanism which has been validated in condensed phase by systematic educt variation and neutral product analysis suggests that the gas-phase ions resolved here were in fact present in the active solution and correspond to intermediates of the dual gold catalytic cycle. In further ESI-TIMS-TOFMS studies of this reaction system it will be interesting to pulse inject isotopically labeled substrates62 in order to probe the kinetics of individual reaction steps by monitoring the time required to generate specific isomers of deuterated analogues (e.g., of CAT + SA or CAT + SA + SB) following the reagent pulse. Complementary information would also be obtainable by resonant IR-laser irradiation of specific intermediates (also perhaps in solution while monitoring the corresponding isomer signals by TIMSTOFMS). Note that a full kinetic description requires measurements of solvent dependences (and possibly blocking/deblocking of reactive intermediates) and systematic probes of how gas-phase ion signals depend on volatilization and detection conditions as well as ultimately molecular dynamics simulations of structure changes associated with

(i)

and 1→2→4→5→D→E

(ii)

We can resolve four species with m/z 1323distinguishable on the basis of their mobilities (Figure 2 and Figure S3). On the basis of DFT calculations and corresponding IMoS predictions of collision cross sections for the proposed intermediates 1−5, we tentatively assign the experimentally observed features to CAT + SA isomers as follows: I1 to 1a, I2 to 2, I3 to 4, and I4 to 3 (see also Scheme 1 and Table 1). Deviating by 2.8 Å2 from the nearest experimentally inferred F

DOI: 10.1021/acs.organomet.8b00128 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics

(11) Arcadi, A. Chem. Rev. 2008, 108, 3266−3325. (12) Gorin, D. J.; Sherry, B. D.; Toste, F. D. Chem. Rev. 2008, 108, 3351−3378. (13) Skouta, R.; Li, C. J. Tetrahedron 2008, 64, 4917−4938. (14) Shen, H. C. Tetrahedron 2008, 64, 3885−3903. (15) Shen, H. C. Tetrahedron 2008, 64, 7847−7870. (16) Hashmi, A. S. K. Angew. Chem., Int. Ed. 2010, 49, 5232−5241. (17) Nevado, C. Chimia 2010, 64, 247−251. (18) Sengupta, S.; Shi, X. ChemCatChem 2010, 2, 609−619. (19) Corma, A.; Leyva-Pérez, A.; Sabater, M. J. Chem. Rev. 2011, 111, 1657−1712. (20) Xiao, J.; Li, X. Angew. Chem., Int. Ed. 2011, 50, 7226−7236. (21) Obradors, C.; Echavarren, A. M. Chem. Commun. 2014, 50, 16− 18. (22) Hashmi, A. S. K.; Rudolph, M. Chem. Soc. Rev. 2008, 37, 1766− 1775. (23) Rudolph, M.; Hashmi, A. S. K. Chem. Soc. Rev. 2012, 41, 2448− 2462. (24) Pflästerer, D.; Hashmi, A. S. K. Chem. Soc. Rev. 2016, 45, 1331− 1367. (25) Hashmi, A. S. K. Acc. Chem. Res. 2014, 47, 864−876. (26) Hashmi, A. S. K.; Braun, I.; Rudolph, M.; Rominger, F. Organometallics 2012, 31, 644−661. (27) Vilhelmsen, M. H.; Hashmi, A. S. K. Chem. - Eur. J. 2014, 20, 1901−1908. (28) Larsen, M. H.; Houk, K. N.; Hashmi, A. S. K. J. Am. Chem. Soc. 2015, 137, 10668−10676. (29) Nunes dos Santos Comprido, L.; Klein, J. E. M. N.; Knizia, G.; Kästner, J.; Hashmi, A. S. K. Chem. - Eur. J. 2016, 22, 2892−2895. (30) Villegas-Escobar, N.; Vilhelmsen, M. H.; Gutíerrez-Oliva, S.; Hashmi, A. S. K.; Toro-Labbé, A. Chem. - Eur. J. 2017, 23, 13360− 13368. (31) Klein, J. E. M. N.; Knizia, G.; Nunes dos Santos Comprido, L.; Kästner, J.; Hashmi, A. S. K. Chem. - Eur. J. 2017, 23, 16097−16103. (32) Wieteck, M.; Larsen, M. H.; Nösel, P.; Schulmeister, J.; Rominger, F.; Rudolph, M.; Pernpointner, M.; Hashmi, A. S. K. Adv. Synth. Catal. 2016, 358, 1449−1462. (33) Tšupova, S.; Cadu, A.; Stuck, F.; Rominger, F.; Rudolph, M.; Samec, J. S. M.; Hashmi, A. S. K. ChemCatChem 2017, 9, 1915−1920. (34) Tšupova, S.; Hansmann, M. M.; Rudolph, M.; Rominger, F.; Hashmi, A. S. K. Chem. - Eur. J. 2016, 22, 16286−16291. (35) Tsupova, S.; Rudolph, M.; Rominger, F.; Hashmi, A. S. K. Chem. - Eur. J. 2017, 23, 12259−12263. (36) Fiebig, L.; Kuttner, J.; Hilt, G.; Schwarzer, M. C.; Frenking, G.; Schmalz, H. G.; Schäfer, M. J. Org. Chem. 2013, 78, 10485−10493. (37) Fedorov, A.; Couzijn, E. P. A.; Nagornova, N. S.; Boyarkin, O. V.; Rizzo, T. R.; Chen, P. J. Am. Chem. Soc. 2010, 132, 13789−13798. (38) Ringger, D. H.; Kobylianskii, I. J.; Serra, D.; Chen, P. Chem. Eur. J. 2014, 20, 14270−14281. (39) Fedorov, A.; Moret, M. E.; Chen, P. J. Am. Chem. Soc. 2008, 130, 8880−8881. (40) Schulz, J.; Jasiková, L.; Skriba, A.; Roithová, J. J. Am. Chem. Soc. 2014, 136, 11513−11523. (41) Roithová, J.; Janková, S.; Jasiková, L.; Vana, J.; Hybelbauerová, S. Angew. Chem., Int. Ed. 2012, 51, 8378−8382. (42) Jasiková, L.; Roithová, J. Organometallics 2013, 32, 7025−7033. (43) Schulz, J.; Shcherbachenka, E.; Roithová, J. Organometallics 2015, 34, 3979−3987. (44) Hashmi, A. S. K.; Lauterbach, T.; Nösel, P.; Vilhelmsen, M. H.; Rudolph, M.; Rominger, F. Chem. - Eur. J. 2013, 19 (3), 1058−1065. (45) Greisch, J. F.; Chmela, J.; Harding, M. E.; Wunderlich, B.; Schäfer, B.; Ruben, M.; Klopper, W.; Schooss, D.; Kappes, M. M. Phys. Chem. Chem. Phys. 2017, 19, 6105−6112. (46) Haler, J. R. N.; Kune, C.; Massonnet, P.; Comby-Zerbino, C.; Jordens, J.; Honing, M.; Mengerink, Y.; Far, J.; De Pauw, E. Anal. Chem. 2017, 89, 12076−12086. (47) Wyttenbach, T.; Helden, G.; Batka, J. J.; Carlat, D.; Bowers, M. T. J. Am. Soc. Mass Spectrom. 1997, 8, 275−282.

solvation and reaction. All of this looks to be within reach within the next few years.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.organomet.8b00128. TIMS measurement conditions, calibration curve, precatalyst structure, mobilograms of m/z 1323 species, mobilograms of m/z 1401 species, overview of mobilograms on adjacent isotopomer masses to illustrate reproducibility, experimental and computed cross sections of m/z 1323 species, experimental and computed cross sections of m/z 1401 species, calculated dipole moments, experimental cross sections versus cross sections of assigned structures, proposed mechanism with identified intermediates highlighted, TIMS measurements of precatalyst and their analysis, ion intensity dependence of the mobilograms of the m/z 1209 precatalyst, and rotamers calculated for the precatalyst (P-CAT 1−3, respectively) (PDF)



AUTHOR INFORMATION

Corresponding Author

*E-mail for M.M.K.: [email protected]. ORCID

Jean-François Greisch: 0000-0002-0761-8191 Patrick Weis: 0000-0001-7006-6759 Manfred M. Kappes: 0000-0002-1199-1730 Jean R. N. Haler: 0000-0002-0315-1846 A. Stephen K. Hashmi: 0000-0002-6720-8602 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS S.B., J.-F.G., A.S.K.H. and M.M.K. gratefully acknowledge the Hector Fellow Academy and the “Hector Stiftung II” for generous provision of funding. Additionally, M.M.K. and P.W. acknowledge support by the DFG (as administered by the Collaborative Research Center, TRR 88 “3MET” (Teilprojekt C6) for IMS method development work leading to this study. We are also grateful to the Karlsruhe Nano Micro Facility (KNMF) for generous access to its travelling wave IMS-MS apparatus. J.R.N.H. acknowledges the FRS-FNRS for the financial support (J.R.N.H. is an FRIA doctorate fellow).



REFERENCES

(1) Dyker, G. Angew. Chem., Int. Ed. 2000, 39, 4237−4239. (2) Hashmi, A. S. K. Gold Bull. 2003, 36, 3−9. (3) Hashmi, A. S. K. Gold Bull. 2004, 37, 51−65. (4) Hoffmann-Röder, A.; Krause, N. Org. Biomol. Chem. 2005, 3, 387−391. (5) Hashmi, A. S. K. Angew. Chem., Int. Ed. 2005, 44, 6990−6993. (6) Hashmi, A. S. K.; Hutchings, G. Angew. Chem., Int. Ed. 2006, 45, 7896−7936. (7) Fürstner, A.; Davies, P. W. Angew. Chem., Int. Ed. 2007, 46, 3410−3449. (8) Jimenez-Nunez, E.; Echavarren, A. M. Chem. Commun. 2007, 333−346. (9) Hashmi, A. S. K. Chem. Rev. 2007, 107, 3180−3211. (10) Li, Z.; Brouwer, C.; He, C. Chem. Rev. 2008, 108, 3239−3265. G

DOI: 10.1021/acs.organomet.8b00128 Organometallics XXXX, XXX, XXX−XXX

Article

Organometallics (48) Von Helden, G.; Wyttenbach, T.; Bowers, M. T. Int. J. Mass Spectrom. Ion Processes 1995, 146−147, 349−364. (49) Haler, J. R. N.; Massonnet, P.; Chirot, F.; Kune, C.; CombyZerbino, C.; Jordens, J.; Honing, M.; Mengerink, Y.; Far, J.; Dugourd, P.; De Pauw, E. J. Am. Soc. Mass Spectrom. 2018, 29, 114−120. (50) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A., et al. Gaussian 09, Revision D01; Gaussian, Inc.: Wallingford, CT, 2009. (51) Becke, A. D. J. Chem. Phys. 1993, 98, 5648−5652. (52) Grimme, S. J. Chem. Phys. 2006, 124, 034108. (53) Dunning, T. H. J. J. Chem. Phys. 1989, 90, 1007−1023. (54) Peterson, K. A.; Puzzarini, C. Theor. Chem. Acc. 2005, 114, 283− 296. (55) Larriba-Andaluz, C.; Fernández-Garcia, J.; Ewing, M. A.; Hogan, C. J. J.; Clemmer, D. E. Phys. Chem. Chem. Phys. 2015, 17, 15019− 15029. (56) Larriba-Andaluz, C.; Hogan, C. J. J. J. Chem. Phys. 2014, 141, 194107−1−9. (57) Nelson, R. D. J.; Lide, D. R.; Maryott, A. A. NSRDS-NBS10 1967. (58) Hashmi, A. S. K.; Lauterbach, T.; Nösel, P.; Vilhelmsen, M. H.; Rudolph, M.; Rominger, F. Chem. - Eur. J. 2013, 19, 1058−1065. (59) Hashmi, A. S. K.; Braun, I.; Nösel, P.; Schädlich, J.; Wieteck, M.; Rudolph, M.; Rominger, F. Angew. Chem., Int. Ed. 2012, 51, 4456− 4460. (60) Wang, Y.; Yepremyan, A.; Ghorai, S.; Todd, R.; Aue, D. H.; Zhang, L. Angew. Chem., Int. Ed. 2013, 52, 7795−7799. (61) Hansmann, M. M.; Rudolph, M.; Rominger, F.; Hashmi, A. S. K. Angew. Chem., Int. Ed. 2013, 52, 2593−2598. (62) Vana, J.; Terencio, T.; Petrovic, V.; Tischler, O.; Novak, Z.; Roithova, J. Organometallics 2017, 36, 2072−2080.

H

DOI: 10.1021/acs.organomet.8b00128 Organometallics XXXX, XXX, XXX−XXX