Photodissociation of acetaldehyde as a second example of the roaming mechanism

Abstract

Product state distributions of the CO produced in the 308-nm photolysis of acetaldehyde show clear evidence of two dissociation mechanisms. One is attributed to the conventional transition state mechanism predicted by theory, with high rotational and translational energy of the CO and a pronounced v⊥J vector correlation. However, as much as 15% of the reaction flux proceeds via another pathway that produces low CO rotational and translational energy, very high CH₄ internal energy, and no correlation between the CO velocity and angular momentum vectors. The attributes of this channel are dynamically similar to the recently reported “roaming atom” mechanism in formaldehyde. We therefore speculate that the second pathway in acetaldehyde also occurs via a roaming mechanism in the CH₃ + HCO exit channel that decays into the CH₄ + CO channel.

The concept of the transition state (TS) is central to chemistry. It is the transient structure at the highest point of the minimum energy pathway (reaction coordinate) between reactants and products. Reaction mechanisms are often described with reference to this structure. For example, an S_N2 reaction is defined by the transient 5-center carbon atom at the TS, or a 3-center elimination reaction refers to the transient 3-membered ring at the TS. Interpretation of the kinetics and thermodynamics of reactions is also based on the energy and entropy of the TS—from the simplest Arrhenius model to more sophisticated TS theories, including their variational forms.

What would happen if reactions were found to bypass the TS? The result would be a new class of reaction mechanisms, with reaction products and kinetics that cannot be predicted by current TS theories. Recently, a mechanism of this type was reported, in which the photodissociation of H₂CO to H₂ + CO was observed to occur via a second, non-TS mechanism (1). The conventional 3-center elimination of H₂ from H₂CO has long been known to produce very high relative translational energy of the departing fragments, highly rotationally excited CO, and modest vibrational energy of the H₂ fragment (2). The new mechanism revealed a very different signature: rotationally cold CO, coupled with very high vibrational excitation of H₂ and low relative translational energy. The mechanism was interpreted by reference to quasiclassical trajectory calculations on a high-level ab initio potential energy surface (1). These calculations revealed that all of the reactions that produced the unusual signature circumvented the TS. The reaction starts out looking like a conventional C–H bond cleavage, which would produce the radical products of H₂CO photolysis: H + HCO. In a significant number of trajectories, just enough energy is tied up in internal motion of the HCO moiety so that the H-atom cannot quite escape; it turns back at very long distance (several angstroms) from the HCO but is still loosely bound. At this range, vibrational energy exchange is very inefficient and so the H-atom orbits the HCO. At some stage in the orbit it encounters the other H-atom and abstracts it from the HCO to produce very highly vibrationally excited H₂ and little vibration or rotation of the CO. This mechanism was referred to as the “roaming atom mechanism.” What was surprising in the formaldehyde study was the relative importance of this mechanism in the overall reaction flux. Theory and experiment both demonstrated that the flux for the “roaming” mechanism increased with increasing excitation energy (1). At 25 kJ/mol above the threshold for “roaming,” as much as 30% of the CO + H₂ reaction flux was found to proceed via the roaming mechanism rather than over the TS.

Formaldehyde is the only molecule for which the roaming mechanism has been shown to have such a significant effect in the overall photodissociation. Formaldehyde holds a fairly unique place in photochemistry: it is among the smallest of systems that can produce polyatomic fragments, there is a moderately low density of states at the TS because of its small size and high vibrational frequencies, and also H-atoms would seem to be the most likely moieties to “roam” because of the large amplitude motion exhibited by vibrations involving hydrogen.

In this article, we test whether the importance of roaming mechanisms in chemistry is limited to formaldehyde, or other photochemically simple systems, by investigating the photodissociation of its larger cousin, acetaldehyde. Specifically, we have measured both the CO product state distributions at high resolution and their vector correlations. We explore whether there is any evidence for the signature of a roaming mechanism.

The photochemistry of acetaldehyde is far more complex than that of formaldehyde. Fig. 1 presents an energy level diagram showing the energetically accessible dissociation pathways and electronic states for excitation energies below ≈400 kJ/mol. The figure is constructed so that the “heavy” product channels: CH₃ + HCO and CH₄ + CO, both of which have been observed experimentally, are on the right. The two “light” fragment channels, H + CH₃CO and H₂ + CH₂CO, neither of which has been observed experimentally, are shown on the left. Two of these mechanisms, one on each side, are bond cleavage channels that produce radical products with no barrier to the reverse reaction. The other two mechanisms involve nuclear rearrangements to produce stable molecular products via a significant barrier. An observation that may prove to be crucial in later discussion is that the critical energy for all four channels is very similar: 360 ± 20 kJ/mol. It is also important to note that triple fragmentation, producing H + CO + CH₃, is energetically inaccessible at the energies used in this study.

Fig. 1.

Energy-level schematic showing the relevant chemical pathways for acetaldehyde dissociation relative to the photon energy, 388 kJ/mol. Four distinct channels are available on the S₀ surface—two correlating asymptotically with radical products, and two correlating with molecular products over significant barriers. The critical energies for all four mechanisms lie in the narrow range from 340 to 377 kJ/mol. The product energies are obtained from standard heats of formation (11). ^a, energies from Kono et al. (12); ^b, energies from Liu et al. (13); ^c, energies from Martell et al. (6).

Results

Laser-induced fluorescence spectra of the 0-0, 1-1, and 3-2 bands of CO were measured, probing population in v = 0, 1, and 2 with full rotational resolution. The intensity of each line was measured, scaled by the line strength, and corrected for electronic perturbations by using the data of Field (3) (R. W. Field, personal communication). Some spectral lines from v = 1 and 2 were found to be in the same region of the spectrum as v = 0. Measurement of these lines and use of the known Franck–Condon factors for CO (4) allowed us self-consistently to scale the population of all of rovibrational levels of CO.

The rotational distributions are plotted in Fig. 2. Immediately apparent is that v = 0 is the most highly populated level with population reducing significantly for v = 1 and further for v = 2. The v = 0 and 1 rotational distributions are well fit by the sum of two Gaussian functions. These two distinct components of the J distributions is the first indication that there might be more than one chemical pathway in the acetaldehyde reaction, and for present labeling we refer to the high J distributions as Mechanism A and the low J distributions as Mechanism B. The vibrational state distribution, obtained by integrating the rotational population for each vibrational state and each mechanism, is shown in Fig. 2 Lower.

Fig. 2.

CO product state distributions (Upper) and branching ratios (Lower) for the dissociation of CH₃CHO at 308 nm. Two distinct distributions are evident in the v = 0 and v = 1 populations, which are attributed to the normal transition state (TS) mechanism and to a “roaming” mechanism (roam).

We have measured sub-Doppler profiles of many peaks in the CO laser-induced fluorescence spectrum to determine the relative CO translational energy produced in this reaction. Several of these are shown in Fig. 3a. Each is a Q-branch transition, with the lowest J at the bottom and successively higher J shown displaced vertically above for clarity. The peak widths clearly increase as J increases, indicating increasing recoil energy for higher J values.

Fig. 3.

Doppler profiles of a variety of rotational lines in the 0–0 band of CO. Vertical displacement of some lines is for clarity. (a) Lines broaden with increasing rotational excitation. (b) Flattening of the Q profiles in comparison with the P or R profiles. (c) No difference with polarization of the probe laser with respect to the photolysis laser. The laser linewidth is shown as a dashed line.

Before commenting on the CO translational energy, there are two subtle features of the Doppler profiles, shown in Fig. 3 b and c, that require explanation. First, the profiles flatten on top as the peak broadens. The corresponding P or R-branch profiles are narrower as shown in Fig. 3b. This difference between Q and P, R-branch Doppler profiles is clear evidence that the angular momentum vector of the recoiling CO is aligned preferentially perpendicular to the velocity vector (a “v-J” vector correlation) (5). The narrow peaks at low J do not show such a difference, although it would be more difficult to discern by eye because of their narrower width. Second, the Doppler profiles have been measured with the polarization of the probe laser aligned both vertical and horizontal with respect to the direction of the photolysis laser. These experiments probe whether there is any spatial anisotropy in the recoiling CO—a “μ-v” vector correlation. Fig. 3c shows the results for the Q (31) line. The identical nature of the two Doppler profiles shows that there is no significant anisotropy in the reaction (5), thereby indicating that the reaction of acetaldehyde takes longer than the rotational period of the molecule.

The relative translational energy of the recoiling CO fragment has been analyzed for each Doppler profile in Fig. 3a using a forward convolution technique, assuming no μ-v correlation and assuming various degrees of v⊥J correlation. The average total translational energy rises from 33.6 kJ/mol at J = 1 to 123 kJ/mol at J = 46. The degree of v⊥J correlation also depends strongly on rotational level, increasing from 0 at J = 1 to 0.7 ± 0.3 at J = 46, where a value of unity would indicate complete correlation. A summary of these findings is shown in Table 1. Note that in the low rotational energy mechanism (Mechanism B) significant rotational excitation of CH₄ would be precluded for reasonable impact parameters by the low rotation of the CO; hence, most of the 87% of the available energy must be released as CH₄ vibration.

Table 1.

Characteristics of the two identified mechanisms for CO production

Mechanism	Energy (kJ/mol)*				Alignment deg. v⊥J	Fraction of flux
	Translation	CO rotation	CO vibration	CH4 vibration/rotation
A	High (101)	High (39)	Low (10)	High (253)	High
	25%	10%	3%	63%	0.7 ± 0.3	85%
Theory†	32%	13%	3%	52%	1.0	—
B	Low (43)	Very low (3)	Low (5)	Very high (352)	Low	15%
	11%	0.5%	1.5%	87%	0

Mechanism A is identified as the regular TS mechanism by comparison with theoretical calculations (10). Mechanism B is attributed to a roaming mechanism. E_avail = 403 kJ/mol.

*Percentages are relative to the total available energy.

^†Ref. 10.

The experimental CO product state distributions from photolysis of CH₃CHO show all of the hallmarks of two competing mechanisms. Table 1 summarizes the average energy in each product degree of freedom. One mechanism, A, produces CO with high rotational energy, coupled with low vibrational energy, significant translational energy, and substantial v⊥J correlation. Mechanism B is characterized by CO with low rotational energy, low translational energy, no v-J correlation, and very high internal energy of the CH₄ fragment. The relative flux for each mechanism can be obtained by integrating the high-J Gaussian fits to the data in Fig. 2, which yields 85% for Mechanism A and 15% for Mechanism B.

Discussion

There have been several ab initio calculations of the potential energy surface leading to CH₄ + CO production from acetaldehyde; each shows a 3-center TS with the aldehyde H-atom bridging the two C-atoms (see Fig. 1) (6–10). Quasiclassical trajectory calculations have been carried out on one of these potential energy surfaces (9, 10), albeit with reaction energy 94 kJ/mol (7,800 cm⁻¹) above the reaction energy in these experiments. These calculations show that CO is produced with a symmetric, Gaussian distribution peaked at J ≈ 57 (10). CO vibrational excitation is modest because the CO bond length at the TS is similar to that of free CO. There is considerable translational energy in the products. However, the internal energy of the CH₄ is most highly populated with ≈52% of the available energy. Finally, the v-J vector correlation was also reported from the quasiclassical trajectory calculations (9), which showed almost perfect v⊥J alignment for dissociation via the 3-center TS.

All aspects of the trajectory study are in at least qualitative agreement with the distributions we have labeled as Mechanism A: moderately high CO rotation, low CO vibration, high relative translation, high CH₄ internal energy, and a high v⊥J alignment. To better facilitate comparison of the experiment and theory at quite different initial energies, we report percent disposal into each degree of freedom in Table 1. The average disposal of energy in all degrees of freedom as well as the alignment data are found to be in semiquantitative agreement. Not only are the averages in good agreement, but the shape of the experimental and theoretical distributions, where available, also agree. We therefore attribute Mechanism A to the conventional reaction over the 3-center TS, as modeled by the theory.

The other mechanism, B, exhibits a set of product state distributions that are qualitatively and quantitatively different. Table 1 reports the average energy disposal into each degree of freedom for this mechanism, again as a percentage of the available energy to facilitate comparison with Mechanism A and the trajectory calculations. The percent energy disposal shows low CO vibration, very low CO rotation, low relative translation, and, by conservation of energy, very large internal energy in the CH₄ sibling fragment. There is also effectively no correlation between the velocity and angular momentum vectors. These features are not found in the theoretical studies of TS dynamics and so we conclude that CO products associated with this distribution of energy must arise from a different mechanism. Fig. 1 shows all other known or postulated pathways for acetaldehyde photolysis at energies below 400 kJ/mol. None of these includes CO as a product. The only other known channel involving CO is the triple fragmentation into CO + CH₃ + H, also shown in Fig. 1, but this channel is inaccessible in these experiments by ≈25 kJ/mol.

The CO rotational distributions reported here for Mechanism B are quite similar to the CO rotational distributions observed in the photodissociation of formaldehyde (2). In that study, when the photolysis energy was below the threshold for the radical H + HCO channel, the CO distribution was also Gaussian, and peaked at high J. As the photolysis energy was raised above the onset of the radical channel, a new, low J feature emerged in the CO rotational distribution. This has now been interpreted as the opening up of the “roaming atom” mechanism (1).

A critical factor in the elucidation of the roaming mechanism in H₂CO was the observation that CO fragments in the low J (roaming) part of the distribution were also recoiling slowly (1). This is the same as we observe here for the low J CO molecules in Mechanism B. In formaldehyde, the roaming mechanism was associated with very high vibrational energy of the partner H₂ fragment. Table 1 shows that the partner CH₄ fragment from acetaldehyde has ≈87% of the available energy, which should be mostly vibrational energy as discussed above. The only possible discrepancy is that the CO rotational distribution from acetaldehyde, peaking at J = 30, is somewhat colder than that for formaldehyde, which peaks at J = 45–50. From an impulsive model, one might reasonably expect the CO rotational distribution for acetaldehyde to be hotter than that for formaldehyde, but the fact that CH₄ can absorb substantial internal energy makes a simple impulsive model less applicable to acetaldehyde. Indeed, calculations show that the CH₄ internal energy, 52% of the total in acetaldehyde (10), is much higher than the H₂ internal energy, ≈25% of the total in formaldehyde (1).

The Mechanism B distributions therefore appear to be similar in all regards to product state distributions for the “roaming atom” mechanism in H₂CO (1, 2). Because of these similarities, we explore the possibility that a similar mechanism might be at play here. Using H₂CO as a guide, it appears that what is needed is a simple bond cleavage channel to radicals at an energy similar to that of the top of the TS. A similar energy barrier is required because, if the interacting channel were too high or low in energy, the reaction flux through the lower energy channel would dominate. Fig. 1 shows that there are two such simple bond cleavage pathways producing radicals in acetaldehyde. Both have no exit barriers and are within 15 kJ/mol of the calculated height of the TS (6, 9), and therefore both are good candidates on the basis of energy.

We can now speculate about which roaming mechanism might lead to the observed results. The higher-energy pathway (CH₃CO + H) would lead to a roaming mechanism that is conceptually similar to the H₂CO case, with a roaming H-atom abstracting a methyl group from the acetyl fragment. This mechanism has some attraction in that it is the light H-atom that is roaming the periphery of the complex, but there are also difficulties. The methyl group is sp³ hybridized in acetyl, and therefore the H-atom has no ready site of attack to abstract it from CO.

The lower-energy pathway (CH₃ + HCO) would involve the methyl and formyl groups orbiting each other, with the CH₃ abstracting an H-atom from HCO. The attraction of this pathway is that both fragments can tie up some of the excess reaction energy to prevent simple dissociation. Also, at such long distances as required for a weakly bound complex, the methyl would be planar and sp² hybridized, and would readily abstract the exposed H-atom on the formyl group. We favor this mechanism, but clearly theoretical support would be needed to decide between the two mechanisms.

Conclusions

In conclusion, we have presented detailed product state distributions of the CO produced in the 308-nm photolysis of CH₃CHO. These distributions show clear evidence of two mechanisms for CO production. One of these mechanisms is the conventional TS mechanism. However, as much as 15% of the reaction flux proceeds via another pathway that is dynamically very distinct. We attribute this other mechanism to a roaming mechanism in the CH₃ + HCO channel that decays into the CH₄ + CO exit channel. It is dynamically similar to the recently reported “roaming atom” mechanism in formaldehyde.

Experimental Procedures

A 5% mixture of acetaldehyde vapor in He was expanded via a pulsed nozzle into a vacuum chamber. Supersonically cooled acetaldehyde molecules (15 K) were photolysed ≈10 mm downstream by an excimer laser at 308 nm. The nascent CO reaction products were detected 50–200 ns later by laser-induced fluorescence via the A–X% transition in the region 140–160 nm. The vacuum UV radiation was generated in Mg vapor by 4-wave mixing, the full details of which have been published previously (14). CO fluorescence was imaged directly onto a solar blind photomultiplier attached to the chamber.

Abbreviation

TS

transition state.