Infrared Spectroscopy of Gas – phase M + ( CO 2 ) n ( M = Co , Rh , Ir ) Ion – molecule Complexes

The structures of gas-phase M+(CO2)n (M = Co, Rh, Ir; n = 2-15) ion-molecule complexes have been investigated using a combination of infrared resonance-enhanced photodissociation (IR-REPD) spectroscopy and density functional theory. The results provide insight into fundamental metal ion-CO2 interactions, highlighting the trends with increasing ligand number and with different group 9 ions. Spectra have been recorded in the region of the CO2 asymmetric stretch around 2350 cm-1 using the inert messenger technique and their interpretation has been aided by comparison with simulated infrared spectra of calculated low-energy isomeric structures. All vibrational bands in the smaller complexes are blue-shifted relative to the asymmetric stretch in free CO2, consistent with direct binding to the metal center dominated by charge-quadrupole interactions. For all three metal ions, a core [M+(CO2)2] structure is identified to which subsequent ligands are less strongly bound. No evidence is observed in this size regime for complete activation or insertion reactions.


I Introduction
There has been much interest in carbon dioxide (CO 2 ) activation in synthetic chemistry over the past two decades. [1][2][3] Most practical CO 2 transformations involve metal-based catalysts and detailed investigations are required to better understand the fundamental interactions involved. In this context, gas-phase studies of metal-ion complexes have a key role to play in elucidating important aspects of metal-ligand bonding, metal ion solvation and molecular activation. [4][5] Such studies can provide valuable insight into the salient features of reaction mechanisms, energetics, kinetics etc., free from the perturbing effects of solvents, counterions, aggregates, or surface inhomogeneities encountered in situ. [5][6][7] Number densities of gas-phase metal-ion complexes produced by laser ablation are typically too low to be probed directly by traditional direct absorption spectroscopic methods.
However, modern laboratory-based intense, tunable infrared systems can now produce IR light in the wavelength range 1.35 to 5 µm. This has, in turn, driven the development of novel action spectroscopies such as infrared resonance enhanced photodissociation (IR-REPD), which provides information on geometric structures of isolated, gas-phase metal ion-ligand complexes. IR-REPD measurements on metal-ion containing complexes were first performed by Lisy et al. who studied alkali metal-ion complexes with ligands including H 2 O, CH 3 OH and CH 4 . [8][9][10][11][12][13][14][15] CO 2 complexes with main group (Mg + , Al + and Si + ) [16][17][18] and first-row transition metal cations [19][20][21][22][23][24] have been studied by Duncan and coworkers, [4][5] while Zhou and coworkers have investigated Ti + (CO 2 ) n clusters. 24 Spectroscopic studies show a systematic blue-shift of the fundamental transition associated with the asymmetric stretching mode of complexed CO 2 4 relative to that of free CO 2 . The binding of the CO 2 ligands to the metal cation center occurs almost exclusively through a charge-quadrupole interaction via one of the oxygen atoms, with little evidence of charge transfer. 25 As the number of CO 2 ligands increases, multiple bands are observed including some that appear close to the wavenumber of free CO 2 , indicating the presence of second solvent-sphere or more weakly-bound ligands. Such observations can highlight characteristic CO 2 coordination numbers by comparison of experimental IR-REPD spectra with simulated IR spectra for low-lying structures calculated typically via density functional theory (DFT).
In providing structural information about the complexes interrogated, IR-REPD studies have also revealed evidence of various intracluster reactions such as oxide-carbonyl formation in Ni + (CO 2 ) n , Si + (CO 2 ) n and Ti + (CO 2 ) n clusters, and oxalate-type anion formation in large V + (CO 2 ) n complexes. 18,[23][24]26 5 By contrast with the cationic species, IR-REPD studies of anionic, M -(CO 2 ) n , complexes (M = Au, Ag, Co and Ni) 27-31 by Weber et al. reveal considerable CO 2 activation marked by redshifts in vibrational frequencies of several hundred cm -1 . In the case of Co -(CO 2 ) n complexes, the vibrational frequency of directly-bound ligands is red-shifted to ca. 1750 cm -1 from the free CO 2 asymmetric stretch vibrational frequency at 2349 cm -1 . This weakening of the CO 2 bond enhances the chances of bond-insertion reactions leading to inserted OCo -(CO)(CO 2 ) n complexes as putative global minima. 30 Here, we report the results of an IR-REPD spectroscopy study of gas-phase M + (CO 2 ) n complexes (M + = Co, Rh, Ir, n = 2-15) designed to elucidate trends in cluster structure down the group 9 metal ions. For the smaller complexes (n ≤ 10), the inert messenger or 'rare gas tagging' 32-36 technique has been employed whereby loss of a weakly-bound argon atom provides a mass spectrometric signature of photon absorption. This technique has been exploited previously by our group in studies of larger naked and decorated transition metal clusters. [37][38][39][40][41][42][43] For the larger M + (CO 2 ) n complexes (n > 10), however, the intensity of the Artagged species is too low and depletion of the M + (CO 2 ) n complex is monitored by CO 2 loss directly. Where it is possible to record naked and Ar-tagged complexes simultaneously, in most cases we observe no qualitative difference in the spectra of the two, lending support to the proposition that the argon tag is indeed inert and does not significantly perturb the cluster structure.  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60 6

II. Experimental
The laser ablation/time-of-flight instrument used to perform the experimental work is new and a brief introduction is warranted. Figure 1  The metal vapor generated by ablation is entrained in a pulse of argon gas seeded with a low percentage of carbon dioxide (typically 0.1-1%), delivered by a pulsed solenoid valve (Parker-Hannifin, Series 9) from a typical backing pressure of 6 bar. During experimentation, the pressure in the source (detection) chamber rises to ca. 1 × 10 -4 (8 × 10 -6 ) mbar.
Following expansion, the molecular / cluster beam is skimmed (Beam Dynamics, 2 mm diameter orifice) as it enters the detection region. In this region, the ions are extracted IR photodissociation of M + (CO 2 ) n -Ar complexes is performed using loosely-focused tunable IR light generated from an OPO/OPA laser (LaserVision -10 Hz) with the IR beam overlapped with, and counter propagating to, the cluster beam. The IR laser is calibrated using IR photoacoustic spectra of both CO and CO 2 . IR-REPD spectra are recorded by capturing mass spectra alternately with and without IR pulses for 150 laser shots at each wavelength. All spectra shown here were recorded in the range of the CO 2 asymmetric stretch (ν 3 = 2349 cm -1 ), with all data acquisition and scanning controlled via a home-built LabVIEW routine.
Whenever the incident IR light is resonant with an IR-active mode of an Ar-tagged complex, IR photons can be absorbed and, following intramolecular vibrational relaxation (IVR), the Ar atom is lost from the complex. The result is depletion in the signal of the parent M + (CO 2 ) n -Ar complex which serves as a signature of photon absorption. Recording the depletion as a function of wavelength yields an infrared action spectrum of the complex.
To aid with assignment and interpretation, experimental spectra are compared with harmonic frequency calculations for a range of low-lying calculated cluster structures. Initial geometry optimization was performed on various starting geometries of the M + (CO 2 ) n complexes generated via the stochastic KICK algorithm 45 using the B3P86 density functional 46 coupled with the Def2TZVP basis set 47 on all atoms, without symmetric constraint. Singlet, triplet and quintet spin states were considered for all complexes, with the triplet states found to be consistently lowest in energy for all complexes. This is expected for Co + and Rh + which both 8 have d 8 3 F ground terms but is, perhaps, surprising for Ir + given its 5d 7 6s 5 F lowest energy term. Such effects have, however, been seen previously. A switching of the relative energetic ordering of different Fe + electronic states upon complexation with CO 2 was observed by Armentrout et al. 48 in their study of sequential bond energies in Fe + (CO 2 ) n . In the present case the lower-spin d 8 configurations provide more symmetric structures than the d 7 s states in which hybridization is induced by incoming ligands. All spectra have been recorded in the region of the CO 2 asymmetric stretch region and for this reason computational efforts were focused on M + (CO 2 ) n complexes. We have not searched exhaustively for insertion, OM + CO(CO 2 ) n-1 , complexes, but, at least for n>2, we found the M + (CO 2 ) n complexes to be lower in energy.
Harmonic vibrational frequency calculations were performed on all unique structures identified to ensure they were indeed true minima (zero imaginary frequencies). Any found to possess higher symmetry were re-optimized within the highest possible symmetric point group. This ensures there are no significant differences between the symmetry-constrained and unconstrained energies, and aids assignment of the electronic states and vibrational modes of each relevant structure. Additional calculations were performed on all unique structures in the presence of an Ar atom in order to determine any effect the rare gas tag may have. All calculations were performed using the Gaussian 09 suite of programs 49 1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60 9 calculated asymmetric stretch of CO 2 at the B3P86/Def2TZVP level of theory (2442 cm -1 ) matches that from experiment (2349 cm -1 ).

III. Results and Discussion
A. Co + (CO 2 ) n -Ar spectra Figure 2 shows the time-of-flight mass spectrum obtained via the laser ablation of a cobalt disk in the presence of a backing gas mixture of ca. 0.5% CO 2 in Ar. A wide range of Co + (CO 2 ) n and Co + (CO 2 ) n -Ar species are formed with signal intensities gradually reducing as the number of CO 2 ligands increases. Qualitatively similar mass spectra are obtained for the Rh + (CO 2 ) n and Ir + (CO 2 ) n complexes.  25 This is much greater than the photon energy in the vicinity of the asymmetric stretch of CO 2 (2349 cm -1 ), thus necessitating the use of rare-gas tagging. [32][33][34][35][36] Even Ar binds strongly to Co + . The Brucat group determined a Co + -Ar binding energy of 4111 cm -1 , 51 and similar argon binding energies have previously been encountered in the case of Ni + (CO 2 ) n -Ar clusters studied by Duncan et al.. 23 In the present case the result is that, even at high laser fluences, no photodissociation of the Co + (CO 2 )-Ar complex is observed. Figure 3 shows the IR-REPD spectra recorded for the Co + (CO 2 ) n -Ar (n = 2-5) complexes in the region 2275 to 2450 cm -1 . The Co + (CO 2 ) 2 -Ar spectrum exhibits a single strong feature at 2385 cm -1 , i.e. blue-shifted ca. 36 cm -1 from the frequency of the asymmetric stretch in free CO 2 (2349 cm -1 ). Similar blue-shifts have previously been observed in other M + -CO 2 containing clusters. [16][17][18][19][20][21][22][23]26 The M + -CO 2 interaction is dominated by charge-quadrupole interaction, 25 with a linear geometry preferred (i.e., η 1 via the O atom, reflecting the negative  1  2  3  4  5  6  7  8  9  10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36  37  38  39  40  41  42  43  44  45  46  47  48  49  50  51  52  53  54  55  56  57  58  59  60 11 quadrupole moment of CO 2 ). The blue-shift indicates little CO 2 activation as expected given the lack of back-bonding from the metal ion towards the slightly negatively-charged O atom.
For the n = 2 complex, both in-phase and out-of-phase combinations occur but, in the case of a linear (D ∞h ) complex, only the out-of-phase combination is IR-active. The single feature in the spectrum of Co + (CO 2 ) 2 -Ar is thus indicative of a linear, D ∞h configuration. Such a linear structure is consistent with the global minimum structure from DFT calculations (Structure I - Figure 4). This is consistent with significant s-d hybridization with electron density directed away from the CO 2 ligands minimizing ion-ligand repulsion and encouraging ligand binding linearly on opposite sides of the metal cation. This rationale also accounts for the linear structure calculated for the Sr + (CO 2 ) 2 complex. 52 As the number of CO 2 ligands increases, an additional vibrational band appears in the IR spectra to the red of the original band (see Figure 3). This new band gradually red-shifts with increasing n towards the free CO 2 value at 2349 cm -1 . For the Co + (CO 2 ) 3 complex, multiple vibrational modes arise. A planar D 3h structure would have a single IR active mode and can thus be ruled out. Our DFT calculations instead suggest a global minimum structure with a near-linear Co + (CO 2 ) 2 'core' structure ( Figure 4, Structure I) and a third CO 2 weakly-bound in a non-planar structure (Structure II - Figure 4, see also S.I.). The spectral signature of the weakly bound CO 2 is still blue-shifted from free CO 2 (by ca. 20 cm -1 ), but by less than the 'core' ligands at 2385 cm -1 . This interpretation of two 'core' ligands and a more weakly bound third is borne out in the calculated M + -ligand bond lengths with M + --O bond lengths for the two 'core' CO 2 ligands of 2.01 Å, 2.16 Å for the third CO 2 (for structural details, see S.I.). The relative binding energies of successive ligands also reflects this structure with a 12 halving of the Co + (CO 2 ) n-1 --CO 2 binding energy between n = 2 and 3 (from ca. 0.91 eV to

B. Rh + (CO 2 ) n -Ar spectra
The IR-REPD spectra for the Rh + (CO 2 ) n -Ar complexes are presented in Figure 5. In contrast to the Co + (CO 2 ) 2 -Ar, spectrum, the Rh + (CO 2 ) 2 -Ar spectrum shows two peaks (at ca. 2401 and 2373 cm -1 ) blue-shifted by 52 and 24 cm -1 , respectively, from the free CO 2 stretch. This is consistent with in-phase and out-of-phase linear combinations of CO 2 stretches indicating a non-linear structure. Our calculations predict a near-linear, C 2v , structure as the global minimum (Structure III - Figure 6) but this is inconsistent with the experimental spectrum as it predicts only a single IR band in this region. At least one other isomer must be present.
We calculate a low-lying C 2v symmetry with O 2 C-Rh + -CO 2 bond angle of 109.3⁰ (Structure IV - Figure 6) around 0.21 eV above the global minimum, the simulated IR spectrum for which agrees well with the experimental spectrum. Including an argon atom in the calculations does not significantly affect the relative energies of the two isomers. Other spin multiplicities have been considered but, in each case, triplet state structures are consistently the lowest in energy. Similar bent geometries to that of structure IV have been observed previously for the Mg + (CO 2 ) 2 and Al + (CO 2 ) 2 complexes, [16][17] and predicted (for Mg + (CO 2 ) 2 ) by calculations performed by Sodupe et al.. 52 The origin of non-linearity of ML 2 molecules has received considerable computational attention in the context of hybridization and pseudo-Jahn-Teller distortion. [52][53] In some cases, hybridization of the metal ion is proposed to lead to pronounced polarization by a first ligand leading to a preference for off axis binding of a second. Rh + (at α = 34.4 a 0 3 ) is calculated to be significantly more polarizable than the Co + ion (α= 20.6 a 0 3 ), 54 which would support a more bent structure for Rh + (CO 2 ) 2 than for Co + (CO 2 ) 2 .
As the number of ligands increases, so the vibrational mode due to weakly-bound CO 2 molecules tends further towards the free CO 2 band ( Figure 5), reflecting the fact that each successive additional ligand that binds to the Rh + (CO 2 ) 2 'core' experiences weaker perturbation. Beyond n = 6, the most blue-shifted band (interpreted as the in-phase combination of the CO 2 asymmetric stretch in the core Rh + (CO 2 ) 2 cluster) becomes too weak to be observed. This may result from the core becoming increasingly linear as the number of ligands increases, rendering this mode IR-inactive. The IR-REPD spectra of the larger n > 6 complexes can thus be interpreted as a single core vibrational mode (around 2385 cm -1 ) and an intense, unresolved feature around the wavenumber of free CO 2 reflecting the outer ligands.  Figure 7 shows the IR-REPD spectra for the Ir + (CO 2 ) n -Ar complexes. The smallest complex (n = 2) exhibits one strong feature that appears at ca. 2367 cm -1 accompanied by a broader shoulder which extends to 2400 cm -1 . The similarity with the spectrum of Rh + (CO 2 ) 2 -Ar indicates a non-linear structure of Ir + (CO 2 ) 2 .

C. Ir + (CO 2 ) n -Ar spectra
Calculations predict a C 2h global minimum structure (Figure 8 -Structure VI) for Ir + (CO 2 ) 2 , but a C 2v isomer (Structure VII) some 0.79 eV higher in energy provides a better match with the experimental spectrum. In this case, the presence of the Ar tag changes the relative energetic ordering of low-lying structures with the C 2v variant of Ir + (CO 2 ) 2 -Ar calculated to be the global minimum (structure VII*). This structure is not unlike the lowest energy structure calculated for Rh + (CO 2 ) 2 (Structure IV - Figure 6).
The polarizability of Ir + (at α = 27.9 a 0 3 -Ref. 54 Figure 9). The simulated IR spectra of the two lowest energy structures (including the respective Ar-tagged counterparts) compare well with the experimental IR-REPD spectrum.

16
The Ir + (CO 2 ) n -Ar complexes show more structured IR-REPD spectra than the Co + and Rh + analogues and the changes upon addition of more ligands are more pronounced. Unlike the Co + (CO 2 ) n and Rh + (CO 2 ) n complexes, the wavenumber of the feature associated with the 'core' (ca. 2365 cm -1 ) initially red-shifts with the sequential addition of another two CO 2 ligands, appearing at ca. 2353 cm -1 for the n = 5 complex. From n = 6 onwards, however, this feature subsequently blue-shifts as further CO 2 ligands are attached reaching ca. 2365 cm -1 for the Ir + (CO 2 ) 15 complex.
It is likely that n = 6 represents a full secondary solvation shell, with an Ir + (CO 2 ) 2 core and four more weakly bound ligands, lending a degree of rigidity to the complex structure. This idea is supported by the fact that the in-phase asymmetric stretching combination of the Ir + (CO 2 ) 2 'core' follows the same trend as the out-of-phase asymmetric stretching combination. Figure 10 shows a summary of the trends in the vibrational bands identified for all species observed in this study as a function of increasing number of ligands. All spectra for n ≤ 10
The DFT simulations suggest that all three systems studied exhibit a 'core' M + (CO 2 ) 2 structure to which additional ligands bind more weakly. This is supported by the trend in binding energy as a function of CO 2 ligands, displayed in Figure 11, which shows that the third CO 2 ligand binds much more weakly than the first two in all three sets of M + (CO 2 ) n complexes. Additionally, the geometrical structures of the M + (CO 2 ) 2 complexes are different, with the bond angle apparently correlated with the polarizability of the metal ion. The weakly polarizable Co + (α = 20.6 a 0 3 ) results experimentally in a linear Co + (CO 2 ) 2 structure, which exhibits a single vibrational band in its IR-REPD spectrum. The much higher polarizability of Rh + (α = 34.4 a 0 3 ) leads to the formation of a very bent, low-lying Rh + (CO 2 ) 2 isomer, which results in two clear vibrational bands being observed. Lastly, Ir + (CO 2 ) 2 has an intermediate, weakly bent, structure reflecting the Ir + polarizability of 27.9 a 0 3 . 54 All of these structures distort upon addition of further ligands but their signature remains in markedly shorter bond lengths for two CO 2 molecules.