H elimination and metastable lifetimes in the UV photoexcitation of diacetylene

Abstract

We present an experimental investigation of the UV photochemistry of diacetylene under collisionless conditions. The H loss channel is studied using DC slice ion imaging with two-color reduced-Doppler detection at 243 nm and 212 nm. The photochemistry is further studied deep in the vacuum UV, that is, at Lyman-alpha (121.6 nm). Translational energy distributions for the H + C₄H product arising from dissociation of C₄H₂ after excitation at 243, 212, and 121.6 nm show an isotropic angular distribution and characteristic translational energy profile suggesting statistical dissociation from the ground state or possibly from a low-lying triplet state. From these distributions, a two-photon dissociation process is inferred at 243 nm and 212 nm, whereas at 121.6 nm, a one-photon dissociation process prevails. The results are interpreted with the aid of ab initio calculations on the reaction pathways and statistical calculations of the dissociation rates and product branching. In a second series of experiments, nanosecond time-resolved phototionization measurements yield a direct determination of the lifetime of metastable triplet diacetylene under collisionless conditions, as well as its dependence on excitation energy. The observed submicrosecond lifetimes suggest that reactions of metastable diacetylene are likely to be less important in Titan's atmosphere than previously believed.

Saturn's moon, Titan, is the only solar system body besides Earth and Venus with a dense atmosphere (1, 2). It is widely considered as a natural laboratory on the planetary scale in understanding the prebiotic chemistry on proto-Earth. Diacetylene is believed to play a key role in the formation of polyynes and polycyclic aromatic hydrocarbons (PAHs) that partially comprise the haze layer in Titan's upper atmosphere (2–4). It is well established that the formation of diacetylene is initiated by photodissociation of acetylene below 217 nm (2, 5–8) according to the following reaction mechanism:

The importance ascribed to diacetylene arises in part because it absorbs light at longer wavelengths, where the solar flux is higher, than any other major constituents of Titan's atmosphere; moreover, experimental results suggest it is still photochemically reactive even well below the threshold for dissociation (9–12). Understanding the dynamics of diacetylene photoexcitation is thus key to revealing the factors driving the chemistry of Titan's atmosphere.

To date, no experiments on the photochemistry of diacetylene have been performed under collisionless conditions. In a pioneering study, Glicker and Okabe (9) determined a quantum yield of 2.0 ± 0.5 for diacetylene photodissociation in the wavelength region of 147–254 nm. Between 184 and 254 nm, no free-radical products were detected and polymeric material was found to coat the inside of the reaction cell. The upper limit for the quantum yield of C₄H formation was then determined to be only 0.06 at 228 nm based on experimental uncertainty. However, at the time, the published thermochemical thresholds were in error, and it is now known that the threshold for H elimination is 5.77 eV (215 nm), significantly higher than the energy assumed by Glicker and Okabe. The diacetylene dissociation quantum yield was ascribed to reactivity of a long-lived metastable form, universally assumed to be the lowest triplet state. Subsequently, Zwier and coworkers (10, 13) extensively investigated the UV photoinduced chemistry of diacetylene through reactions in a ceramic nozzle with a VUV probe of the products downstream. After excitation of the ¹Δ_u excited state, secondary reactions were found to lead to the formation of various larger hydrocarbons (12, 14, 15). The laser-based studies, principally at 231 and 243 nm, also found no evidence for radical products proceeding from primary photodissociation of diacetylene (1, 11, 16–18). Although triplet diacetylene reactions invoked to account for the observed chemistry are now often incorporated in models of Titan's atmosphere, with an assumed metastable lifetime of 1 ms or more, to clarify their role, the triplet lifetime must be measured directly and as a function of excitation energy (19). We present such measurements here.

Several other theoretical models and experiments have examined the secondary photochemistry of C₄H₂ (2, 20, 21). However, for a clear understanding of the role of diacetylene in Titan's atmosphere it is essential to have a better knowledge of its primary photochemistry (product branching and energy dependence), in addition to the electronic decay pathways and rates. In this article, we report the experimental results for primary C₄H₂ photodissociation and metastable lifetimes under collisionless conditions. The experiments are supported by a series of ab initio and Rice–Ramsperger–Kassel–Marcus (RRKM) calculations to assist in interpretation of the results (22).

Results and Discussion

To facilitate the following discussion, computed stationary points and dissociation asymptotes for ground-state diacetylene are shown in Fig. 1. In the first series of experiments, DC-sliced ion images of H atoms from diacetylene photodissociation at three different wavelengths were recorded (Fig. 2). Background signals were subtracted from the raw images and total center-of-mass translational energy distributions were derived from the refined data (Fig. 3). The images all show isotropic angular distributions. The distributions at all three wavelengths studied (243 nm, 212 nm, and 121.6 nm) have peaks ≈0.45 eV and decay to higher recoil energies, extending to 3–4 eV for the 243 and 121.6 nm results and to ≈5 eV for the 212 nm dissociation. The isotropic angular distributions and structureless translational energy distributions peaking at low energy are typical of the statistical, barrierless hydrogen elimination process (23) and suggest dissociation on the ground electronic state or possibly from the lowest triplet. As shown in Fig. 1, the C–H bond in diacetylene is very strong, with a dissociation energy of 133 kcal/mol. The threshold wavelength for single-photon dissociation of diacetylene is thus ≈215 nm, whereas at 243 nm the single-photon energy is only 118 kcal/mol. Single-photon dissociation at 243 nm is clearly not possible. However, if C₄H₂ absorbs two photons at 243 nm, dissociation to C₄H + H is possible with a total excess energy of 103 kcal/mol. Such a process is consistent with the translational energy distribution in Fig. 3, which extends nearly to this limit.

Fig. 1.

Profile of the ground-state potential energy surface of diacetylene calculated at the CCSD(T)/CBS+ ZPE(B3LYP/6-311G**) level of theory.

Fig. 2.

DC sliced H atom images from diacetylene excitation at 243 nm (Top), 121.6 nm (Middle), and 212 nm (Bottom).

Fig. 3.

Total translational energy release spectra derived from images in Fig. 2.

We can directly compare this result with that obtained at the same two-photon energy by considering dissociation (and probe) at 121.6 nm. The kinetic energy distribution recorded at this wavelength is also shown in Fig. 3 and is nearly superimposable on that obtained at 243 nm. We thus conclude that 243 nm production of H atom from diacetylene likely results from absorption of two photons, but at 121.6 nm, it is a single-photon dissociation.

We next consider 212 nm, which is several kilocalories per mole above the threshold for H loss in diacetylene, the lowest photochemical channel. The distribution in Fig. 3 shows a translational energy release similar in shape but extending to even higher energy than at 243 or 121.6 nm. However, the distribution is entirely confined within the available energy (138 kcal/mol) of the two-photon excitation at 212 nm. We thus conclude that two-photon dissociation still dominates, even though we are now certainly above the single-photon dissociation threshold.

All of the observed dissociation processes likely come from the ground state after some intramolecular electronic relaxation processes. This ground-state decomposition is only a part of the picture for diacetylene. We also need to understand the electronic decay dynamics leading to the ground state and the possible time spent as a potentially reactive metastable triplet species. In the introduction we mentioned the importance ascribed to metastable diacetylene in Titan's atmosphere. If reactive C₄H₂* is very long-lived, its contribution to the chemistry in Titan's stratosphere will clearly be much greater than if intersystem crossing (ISC) takes it to the unreactive ground state before it has an opportunity to encounter a suitable reaction partner (e.g., some other unsaturated hydrocarbon.)

To examine these issues, first, we consider the possible excited states involved. Vila et al. (22) have calculated energies and geometries for a range of low-lying excited states of diacetylene by using CASSCF and CASMP2 methods and we draw on their results for this discussion. If we consider first the linear geometry of the ground state, the initial excitation is to the second singlet state. This is the only low-lying excited state that is linear. Internal conversion (IC) may then populate the first singlet state, or ISC may take the system to one of several triplet states. IC in the triplet manifold will then result in formation of the lowest triplet (T₁), generally regarded as the identity of the long-lived metastable species. We should note that these other excited states split into cis and trans and even nonplanar isomers fairly close in energy, but with significant associated relaxation energy in some cases. Zwier and coworkers (24) have shown by linewidth analysis that the initially excited state must have a subpicosecond lifetime, and our spectra are in excellent agreement with their results. Relative rates for IC and ISC leading to the lowest triplet, such as could be obtained by femtosecond time-resolved photoelectron spectroscopy, would be very interesting for this system, but these measurements are not yet available. In any case, we may assume that these processes are very rapid relative to the final step, ISC for T₁-S₀. This is the decay rate that represents the metastable lifetime for an isolated molecule, and one that we can measure by using a UV-pump, VUV-probe strategy. In this approach, analogous to early studies by Smalley and coworkers (25, 26), the nanosecond UV laser excites C₄H₂, after which it relaxes rapidly to T₁. A 7.9-eV (157 nm, F₂ excimer) probe laser can then ionize the electronically excited states of C₄H₂ [I.P. 10.30 eV vertical, 10.17 adiabatic (27)], but not the ground state. Our own CCSD(T) calculations for the lowest triplet states, ³B_u (trans) and ³B₂ (cis), give adiabatic energies of 3.41 and 3.43 eV relative to the singlet ground state, and vertical ionization energies of 7.71 and 7.67 eV, respectively. The latter values confirm our ability to induce efficient single-photon ionization at 157 nm for the triplets. By monitoring the parent ion yield as a function of pump-probe delay, we determine the lifetime of T₁, and we can do this for any initial UV excitation energy. Results for a typical scan at 231.5 nm are shown in Fig. 4A. Single exponential decays are readily seen and fitted after accounting for the laser pulse duration. Experimental decay rates were determined at 231.5, 243.11, and 247.6 nm and plotted in Fig. 4B. Great care was taken in these measurements to ensure that fly-out effects did not contribute to the decays. This conclusion is supported by the fact that the measured decays varied strongly with pump wavelength, but were insensitive to the position of the unfocused lasers.

Fig. 4.

Lifetime measurements for triplet diacetylene. (A) Typical pump-probe decay profile obtained after excitation at 231.5 nm. Points are experimental result and line is single exponential fit after convolution over laser pulse duration. (B) Triplet decay rate plotted vs. excess energy in T₁. Solid line is Arrhenius fit to the points.

This strong dependence of lifetime on excitation energy is a manifestation of the dependence of the T₁-S₀ ISC rate on vibrational excitation in the triplet molecule. This behavior is commonly seen, and may be ascribed to a barrier on the triplet surface leading to the crossing region. The barrier may be the actual crossing seam of T₁ and S₀, or it may simply be a region of T₁ that must be passed through to access a lower-energy T₁/S₀ crossing. If we fit the experimental points to an Arrhenius rate expression based on excess energy in the lowest triplet, we can extrapolate this determination to higher excitation energies (Fig. 4B). We use our theoretical value of 3.41 eV for the origin of T₁. An experimental value of 3.27 eV was determined by Vuitton et al. (8) based on emission in the matrix; however, it is possible that poor Franck–Condon factors preclude emission from the vibrationless level. Based on the fit shown in Fig. 4B, we obtain a triplet lifetime of 36 ns at 212 nm. Given the uncertainty in the origin energy for T₁, and variations in our decay measurements, we estimate an uncertainty of, at most, an order of magnitude in this extrapolated lifetime. This result indicates that at 212 nm, just above the lowest dissociation threshold, ISC to S₀ is likely to occur for a significant fraction of the excited molecules within the duration of our laser pulse.

The only other measurement of the triplet lifetime is the study by Vuitton et al. (8) in argon and krypton matrices. They monitored phosphorescence after excitation at 249 nm and extrapolated the results to the gas phase after accounting for the dielectric effect of the matrix on the lifetime. They obtained a value of ≈70 ms. It is difficult to compare this thermalized value in a matrix at 5–30 K with our microcanonical determinations at 2–3 eV vibrational energy, but we may simply note that the trend we observe is not inconsistent with their measurement.

The most important practical issue arising from our lifetime measurement is the relevant value for Titan. In attempting to incorporate metastable reactions into atmospheric models, a lifetime of 1 ms has generally been used (10, 12, 28), and this has been cited as the lifetime of triplet acetylene determined indirectly by Klemperer and coworkers (29, 30). Recent experiments have more directly examined the triplet lifetime in acetylene after ISC from specific rovibrational levels of S₁ and obtained a value on the order of 80–100 μs that depends on excitation energy (31), just as we have seen here for diacetylene. For diacetylene, our measurement clearly shows a lifetime that is many orders of magnitude lower than what has been assumed at the energies relevant for UV photoexcitation on Titan. However, if collisions result in vibrational cooling of the triplet to the ambient temperature, this lifetime will be extended, perhaps to the point that triplet reactions can contribute significantly to the chemistry. However, in Titan's upper atmosphere where the UV solar flux is significant, the pressure is too low for vibrational cooling of the triplet to dominate over T₁-S₀ ISC given these submicrosecond lifetimes.

We now return to considering the possible H loss products and pathways on the ground state, guided by the ab initio calculations as shown in Fig. 1. H elimination from diacetylene can occur without a barrier to the product channel P1 giving linear C₄H (32, 33). This dissociation pathway can also be accessed via an intermediate (IS1). In this case, H atom migration first occurs from one terminal carbon atom to the other terminal carbon atom with a 101 kcal/mol barrier. The intermediate product will also undergo H elimination without a barrier. At high energy (e.g., Lyman-α wavelength), all three channels will be accessible. Starting from the diacetylene ground state, one way of producing P2 and P3 products is through the ring-closing reaction that occurs with a 91.8 kcal/mol barrier to form IS2. Finally, IS2 can undergo H elimination without a barrier to produce P3. Alternatively, IS2 can rearrange via ring contraction to IS3 with a small barrier of 15.1 kcal/mol. The IS3 intermediate finally undergoes barrierless H elimination to produce P2. The other alternative route of forming P2 products is through intermediate IS1 followed by cyclization with a 59.9 kcal/mol barrier to IS4. The IS4 intermediate dissociates to P2 products. Finally, it is also possible to have rearrangement of IS4 to IS3 through H migration. This reaction occurs with a barrier of 28.6 kcal/mol.

To gain a sense of the relative importance of these different pathways, we have performed RRKM calculations of the dissociation rates and branching ratios to the H loss pathways, as well as all other possible dissociation channels, at several energies of interest. The results are shown in Table 1. First, we examine the total rates for H elimination summed over all C₄H product channels. Just above threshold, at 212 nm, we see a H loss rate of 5.2 × 10⁴ s⁻¹. This very low rate readily accounts for the absence of any dissociation signal in our experiment, for which the detection window is only the 10-ns duration of the laser pulse. If the fluence at 212 nm were sufficiently low, single-photon dissociation at 212 nm should be seen. However, the time scale is such that very little will be formed within the 10-ns duration of the laser pulse, so that it drops below the sensitivity limits of our experiment. At 193 nm, the rate rises rapidly to 5.3 × 10⁷ s⁻¹. Although we have not yet studied dissociation at this energy, we see there should be reasonable probability of decomposition within the duration of the laser pulse, with some minor branching to C₂ + C₂H₂. At 157 nm, we begin to see a small contribution from C₂H + C₂H, which has the highest threshold energy of any of these channels. At a total energy of 9–10 eV, H elimination still dominates (68%) and approximately half is formed via R1 and half via IS1. It is interesting, however, that despite the high threshold, the larger A factor for the C₂H + C₂H channel causes it to win out over the other minor channels, C₄ + H₂ and C₂ + C₂H₂. Branching to the lower energy rhombic isomer of C₄ is negligible in the calculations except at the highest energy studied, where it is still 100-fold lower than that to the linear isomer.

Table 1.

Computed branching ratios (%) and dissociation rates for indicated product channels

Product	Wavelength
	212 nm	193 nm	157 nm	121.6 nm	2x212 nm
Energy, eV	5.85	6.4	7.9	10.2	11.7
HCCCC + H	100.0	88.4	79.6	74.6	68.3
C4(1Σg+)+H2	0.0	0.0	1.5	5.3	7.3
C4(1A′)+H2	0.0	0.0	0.0	0.0	0.1
C4(1Ag)+H2	0.0	0	0.0	0.0	0.0
C2+C2H2	0.0	11.6	15.8	6.4	4.7
C2H + C2H	0.0	0.0	3.2	13.7	19.6
HCCCC + H (from R1)	81.5	55.0	41.9	40.1	40.3
HCCCC + H (from IS1)	18.5	33.4	37.7	34.5	28.0
k(H loss), s−1	5.23 × 104	5.26 × 107	2.39 × 1010	6.88 × 1011	1.98 × 1012
k(C2H), s−1	0	0	8.24 × 108	1.14 × 1011	5.12 × 1011

These results suggest the following scenario for the observed H atom signals: the first photon excites the C₄H₂ molecule from ¹Σ_g ground state, which has linear D_∞h symmetry, to ¹Δ_u with the same symmetry. Very rapid electronic relaxation likely precedes absorption of a second photon, which then excites the molecule to one of many high-lying Rydberg states, likely now with considerable vibrational excitation (34). This highly excited diacetylene can then undergo efficient electronic relaxation, ultimately to the ground state where dissociation takes place. This second photon absorption may be attributed to the presence of low-lying Rydberg states of C₄H₂ ≈9.4 eV and below (34). Excited-state calculations at CIS(D) and EOM-CCSD levels of theory have confirmed the presence of a multitude of such states in the range of 8.0–10.2 eV with significant oscillator strength.

In Okabe's work, discharge lamps were used with fluences lower by many orders of magnitude. Two-photon processes are thus unlikely in that work, and in Titan as well, we should note. However, in the previous work by Zwier and coworkers, it seems laser fluences greater than ours were used, so it is interesting that no radical processes were detected. In those experiments, it may be the number density in the irradiated nozzle extension that is key, so that the rate of metastable reaction could exceed the rate of additional photoexcitation and decomposition to radicals. In comparing and reconciling all of the various experiments performed under a wide range of conditions, we see many competing factors come into play. This underscores the importance of understanding time scales and the complex interactions between electronic and vibrational relaxation, secondary photoabsorption, and metastable reaction to determine the processes truly relevant to Titan's atmosphere.

Materials and Methods

Experimental.

The detailed DC slice imaging experimental set-up has been reported elsewhere (35, 36). Here, we review the essential components of the present configuration. A pulsed supersonic molecular beam containing ≈40% diacetylene seeded in argon is expanded from 1000 torr via a piezoelectric pulsed valve into the source chamber held at ≈10⁻⁶ torr. The beam, collimated after passing through a skimmer, enters an interaction chamber (≈10⁻⁸ torr) between the repeller and extractor electrodes. The laser and molecular beam delay is adjusted to access the earlier portion of the molecular beam pulse to eliminate the contribution from diacetylene dimer or clusters. Two counterpropagating laser beams of different wavelengths are focused on the molecular beam in the interaction region. The C₄H₂ is excited and dissociated to produce H atoms, which are then probed by a two-color reduced-Doppler (TCRD) REMPI (37, 38) scheme, or 1 + 1′ ionization in the case of the Lyman-α dissociation.

In this application of the TCRD scheme, one laser beam is used to excite the diacetylene, and a second laser is used in combination with the first to detect the product. For example, laser light at 212 nm is focused onto the molecular beam, giving rise to H atom products. Counterpropagating 285-m light then combines with the 212-m light to drive the H atom 1s–2s two-photon transition with a significantly reduced Doppler width. Ionization is then achieved by absorption of a third photon from either laser beam. Furthermore, the diacetylene is transparent at 285 nm, so the photodissociation is induced by the 212-nm light alone. The 243-nm beam is produced by frequency doubling of the output of a dye laser pumped by a 308-nm XeCl excimer laser. Lyman-α radiation is generated by frequency tripling of 364.7 nm laser light in a VUV cell containing 30% xenon gas, phase matched with argon at a total pressure of 900 torr. The 364.7-nm beam is focused to the center of the VUV cell by using a tight-focusing quartz lens. The resultant Lyman-α light is then loosely focused to the center of the interaction region by a MgF₂ lens. After ionization, the resulting protons are accelerated along the 46-cm flight tube onto a position-sensitive 75-mm diameter microchannel plate (MCP) detector coupled to a P-47 phosphor screen. Application of a narrow gate (43 ns) to the MCP assembly is used to implement the experimental slicing of the equatorial region of the H atom recoiling velocity distribution in this particular experiment. The images are recorded by using a CCD camera with our IMACQ megapixel software. Negligible H⁺ signal is seen when the TCRD or Lyman-α resonant condition is not met.

In the lifetime measurements, we use a tunable, narrow-linewidth (0.07 cm⁻¹) OPO laser that is frequency doubled to provide the pump light at 230–250 nm. The probe light is an F₂ excimer laser beam at 157 nm. Both lasers are unfocused and directed at right angles mutually perpendicular to the diacetylene beam. The power in each beam is attenuated to the point that negligible signal is seen from either laser alone. For the UV beam this corresponds to 0.3 mJ in a spot of 1-cm diameter (≈150 μJ/cm²), whereas the VUV probe is estimated to be roughly half this value. Total ion yield at the parent C₄H₂ mass is then recorded as a function of delay between the two lasers.

Computational.

Molecular geometries and vibrational frequencies of various C₄H₂ and C₄H local minima and transition states were calculated at the hybrid density functional B3LYP/6-311G** level of theory (39, 40) with the only exception being the linear HCCCC(²Σ⁺) product, for which this method gives one imaginary frequency. When the C_∞v symmetry constraints were lifted for HCCCC, the B3LYP/6-311G** geometry optimization converged to a slightly bent structure, in contradiction to experiment. Alternatively, calculations at the quadratic configuration interaction QCISD/6-311G** level (41) gave a perfectly linear HCCCC geometry with all real frequencies. The B3LYP and QCISD calculations were carried out by using the GAUSSIAN 98 package (42).

Relative energies of the reactant, products, intermediates, and transition states on the C₄H₂ ground-state potential energy surface were refined at the highest theoretical level feasible by using the coupled cluster CCSD(T) method as implemented in the MOLPRO program package (43) with extrapolation to the complete basis set (CBS) limit. To achieve this, we computed CCSD(T) total energies for each stationary point with Dunning's correlation-consistent cc-pVDZ, cc-pVTZ, cc-pVQZ, and cc-pV5Z basis sets (44) and projected them to CCSD(T)/CBS total energies by fitting to the following equation (45)

where x is the cardinal number of the basis set (2, 3, 4, and 5, respectively) and E_tot(∞) is the CCSD(T)/CBS total energy.

Adiabatic excitation energies to the lowest triplet triplet electronic states of C₄H₂, ³B₂ (cis) and ³B_u (trans) were calculated at the CCSD(T)/CBS level with their geometries optimized and vibrational frequencies computed at B3LYP/6-311G**. Vertical ionization energies of the triplet structures were also calculated by using the CCSD(T)/CBS method.

To compute rate constants for individual reaction steps dependent on the energy of absorbed photons, we used conventional microcanonical RRKM theory (46) with ab initio calculated relative energies and molecular parameters. The computational procedure has been described in detail previously (47). The harmonic approximation was used in calculations of numbers and densities of state. For reaction steps occurring without distinct transition states, which include H eliminations from various C₄H₂ intermediates and the C–C bond fission to produce C₂H + C₂H, we used microcanonical variational transition state theory (VTST) (47, 48). With all rate constants in hand, we calculated product-branching ratios by solving first-order kinetic equations for unimolecular reactions on the C₄H₂ surface according to the reaction scheme shown in Fig. 1. Only a single total-energy level was considered throughout, as for collisionless conditions. We used the fourth-order Runge–Kutta method to solve the equations; the product concentrations at the time when they converged were used to compute branching ratios. The calculated concentration profile of the C₄H + H product vs. time was then fit to a first-order kinetic law to deduce the overall rate constant for H elimination from diacetylene.