Communication: Spectroscopic observation of the O-bonded T-shaped isomer of the CO-CO₂ dimer and two of its intermolecular frequencies

CO₂–CO is a fundamental dimer consisting of two very basic monomers, namely, CO and CO₂. These monomers are of chemical, planetary, and astrophysical interest. The interaction of CO and CO₂ and their ability to adapt their bonding to each other is therefore of general interest.

CO₂–CO was observed in infrared and microwave spectra in 1989 by Legon and Suckley.¹ They determined a T-shaped structure in which the CO₂ monomer forms the “top” of the T and CO forms the “stem,” with the C atoms located adjacent to each other and an intermolecular distance of 3.91 Å, as shown at the top of Fig. 1. The original infrared spectrum was a b-type band observed in the CO₂ ν₃ region (≈2350 cm⁻¹).¹ Subsequently, an a-type infrared band was also studied in the CO fundamental region (≈2150 cm⁻¹).^2,3 As well, the dimer dipole moment has been measured using radiofrequency asymmetry doubling transitions.⁴ CO₂–CO has also been observed by matrix isolation spectroscopy,⁵ but the relevance of these results to the isolated dimer is not so obvious, since there appear to be significant matrix effects.^6,7

Shortly after the gas phase studies, a number of theoretical ab initio calculations appeared which predicted the existence of two T-shaped isomers, a C-bound (the most stable isomer) and an O-bound (a higher energy structure).^8–11 As shown below, our new infrared spectra confirm the existence and stability of the T-shaped O-bonded CO₂–CO in a helium buffer. Additionally, we measure two of the four intermolecular frequencies of this isomer. Because measurements of these low-frequency intermolecular modes provide a means to probe potential surfaces at a high level of precision, we conclude that the published ab initio results are inadequate in predicting these frequencies and therefore call for modern high-level ab initio calculations which explore the complete intermolecular potential surface with results relevant to high-resolution spectroscopy. Naturally, studies of the isoelectronic dimer CO₂–N₂, both experimental^12,13 and theoretical,¹⁴ are also relevant in the present context.

Spectra were recorded at the University of Calgary as described previously,^15–17 using a pulsed supersonic slit jet apparatus and a Daylight Solutions’ cw quantum cascade laser. The expansion mixture contained about 0.2% carbon monoxide plus 0.2% carbon dioxide in helium carrier gas, and the jet backing pressure was 9 atm. Under these conditions, the CO dimer spectrum¹⁷ was only weakly observable. Wavenumber calibration was carried out by simultaneously recording signals from a fixed etalon and a reference gas cell containing N₂O. Spectral assignment and simulation were made using the PGOPHER software.¹⁸ 

The fundamental band of the CO₂–CO dimer in the carbon monoxide stretch region has been observed previously^2,3 with its origin at 2148.24 cm⁻¹ and 4.97 cm⁻¹ above that of the CO monomer. We discovered a new band at 2140.29 cm⁻¹ and 2.98 cm⁻¹ below the monomer origin, as shown in the bottom panel of Fig. 2. The new band is qualitatively similar to the original one. Both are a-type parallel (ΔK_a = 0) transitions of a near-prolate asymmetric rotor, and both have only even values of K_a″. The new band has somewhat larger (by 15%–20%) values for the B and C rotational constants. As mentioned, the original CO₂–CO dimer has the T-shaped structure shown at the top of Fig. 1, and it is natural to assume that the new band is due to the predicted^9,10 T-shaped isomer with the O atom of CO adjacent to CO₂ shown at the bottom of Fig. 1. In both cases, the structures have C_2v symmetry with the CO monomer aligned along the C₂ symmetry axis, and the absence of odd K_a″-values is explained by the nuclear spin statistics of the equivalent ¹⁶O atoms of the CO₂.

In addition to the fundamental band at 2148.24 cm⁻¹, we also searched for and detected two combination bands of the new O-bonded isomer 2. These involve transitions from the ground state of the dimer to combinations of the CO stretch fundamental plus a (relatively) low frequency intermolecular mode. As illustrated in the middle and top panels of Fig. 2, both combination bands have perpendicular (ΔK_a = ± 1) selection rules, so their appearance is quite different from the fundamental one. The first band, at 2154.48 cm⁻¹ (middle of Fig. 2), is b-type (ΔK_c = ± 1), while the second band, at 2162.97 cm⁻¹ (top of Fig. 2), is c-type (ΔK_c = 0).

We assigned 42 transitions in the fundamental band, 42 in the b-type combination band, and 37 in the c-type combination band. These were analyzed in a simultaneous least-squares fit to determine the set of 18 parameters listed in Table I. The quality of the fit was good, with an average error of 0.000 33 cm⁻¹, and the detailed assignments are given as the supplementary material.¹⁹ The simulated spectra in Fig. 2 are based on these parameters, together with an effective rotational temperature of 1.8 K and a Gaussian line width of 0.0025 cm⁻¹. There are only small changes in the parameters between the ground state and the excited fundamental vibration. However, the changes are larger for the combination bands, particularly the value of A, which decreases by 0.0383 cm⁻¹ for the b-type band and increases by the same amount for the c-type band. This equal and opposite change in A almost certainly indicates the presence of an a-type Coriolis interaction between these two excited intermolecular states, similar, for example, to that noted previously for the N₂O dimer.¹⁶ We found that inclusion of an interaction parameter equal to 0.575 cm⁻¹ effectively removes the large changes in A, while giving a fit which is equal in quality to the original one. This interaction parameter implies a value of 0.72 for the dimensionless Coriolis constant, ζ^a. Parameters from this Coriolis fit are given in the supplementary material,¹⁹ and the main change is, of course, in the A values of the excited combination states.

Each CO₂–CO dimer isomer in Fig. 1 has 4 intermolecular vibrational modes: van der Waals stretch (with a₁ symmetry in the C_2v point group); out-of-plane rock (b₁ symmetry); and two in-plane bends (b₂ symmetry). If they are relatively uncoupled, the two b₂ modes could be described as CO rock and CO₂ rock; otherwise, if they are more coupled, they could be described as geared and anti-geared bends. The observed c-type combination band at 2162.97 cm⁻¹ is clearly due to the b₁ out-of-plane rock, and we believe that the observed b-type band at 2154.48 cm⁻¹ is likely ascribed to the b₂ CO rock/geared bend. The observed intermolecular frequencies are compared with two theoretical ab initio calculations^9,10 in Table II. Agreement is fairly good for the b₂ mode (especially with the calculation by Venayagamoorthy and Ford¹⁰), but both calculations seriously overestimate the out-of-plane frequency. Note, of course, that the calculations refer to the ground state, whereas our observations are for the intermolecular modes in combination with the carbon monoxide CO stretch. But, there is good evidence that the coupling of intermolecular and intramolecular modes is small.²⁰ 

Table III lists some parameters of isomers 1 and 2 of CO₂–CO including for comparison the isoelectronic species CO₂–N₂. As noted previously, the band origin is blue-shifted for isomer 1 and red-shifted for isomer 2, relative to the CO monomer frequency (of course, there is no such parameter for CO₂–N₂). The observed shifts of +4.97 and −2.98 cm⁻¹, respectively, can be compared with calculated values of +5.3 and −0.6 cm⁻¹ (Ref. 9) or +3.6 and −0.4 cm⁻¹ (Ref. 10). These MP2 level calculations thus predict the signs of the shifts successfully for both isomers, but considerably underestimate the isomer 2 red shift. The similar and relatively small inertial defects in Table III help to confirm the planar equilibrium geometry of all three species. It is not surprising, but still interesting, that the rotational constants (and resulting intermolecular distance) of CO₂–N₂ are very close to midway between those of CO₂–CO isomers 1 and 2.

Binding energies at the MP2 level were calculated to be 429.9 and 306.3 cm⁻¹ (Ref. 9) or 410.6 and 308.9 cm⁻¹ (Ref. 10), for isomers 1 and 2 of CO₂–CO, respectively, without correction for basis set superposition error. After making the superposition correction, the latter binding energies drop to 265.6 and 100.2 cm⁻¹, respectively.¹⁰ However, a more recent and higher level ab initio calculation¹¹ gives a corrected binding energy of 409 cm⁻¹ for isomer 1. Anyway, it seems clear from theory that isomer 1 (C-bonded) is considerably more strongly bound than isomer 2 (O-bonded). In spite of this energy difference, we observed the isomer 2 fundamental to be quite strong, with roughly half the intensity of isomer 1. This could be partly due to a difference in transition probability, as indeed that of isomer 2 is calculated to be 15%–20% stronger.^9,10 But, there is obviously still a sizeable population of isomer 2 in our dilute helium supersonic expansion. Based on other cases (like N₂O dimer),²⁰ we expect the population of isomer 2 to disappear (or be significantly reduced) in a neon or an argon expansion, but this experimental verification of the binding order has not yet been carried out.

In conclusion, we have observed the O-bonded form (“isomer 2”) of the CO₂–CO dimer for the first time by high resolution infrared spectroscopy and determined two of its low-frequency intermolecular modes. A planar T-shaped structure with C_2v symmetry and 3.58 Å intermolecular distance (bottom of Fig. 1) is confirmed by good agreement of observed and calculated rotational constants and by the observed absence of transitions with odd values of K_a″. Observation of the microwave spectrum of isomer 2, and thus even more precise determination of rotational constants, should now be straightforward, guided by the parameters given here in Table I.

The financial support of the Canadian Space Agency and the Natural Sciences and Engineering Research Council of Canada is gratefully acknowledged.