Observation of Light-Induced Reactions of Olefin–Ozone Complexes in Cryogenic Matrices Using Fourier-Transform Infrared Spectroscopy

Abstract

Each olefin (ethylene, trans-1,3-butadiene, isoprene, dimethyl butadiene (DMB)) and ozone molecules were codeposited on a CsI window at cryogenic temperature, and the products of photolysis with ultraviolet–visible light were observed using Fourier-transform infrared spectroscopy. The products of the C₂H₄–O₃ system could be assigned to glyoxal (CHO–CHO), ethylene oxide (c–C₂H₄O), CO, and CO₂. The formation of CHO–CHO and c–C₂H₄ and the absence of H₂CO and HCOOH indicated that the main reaction channels did not involve C–C bond breaking. Based on this simple scheme, the photoproducts of different olefin–O₃ systems were assigned, and the vibrational features predicted by density functional theory calculations were compared with the observed spectra. Regarding butadiene, spectral matches between the observations and calculations seemed reasonable, while assignments for isoprene ambiguities of and DMB remain, mainly because of the limited availability of authentic sample spectra.

1. Introduction

The products of oxidation of volatile organic compounds (VOCs) contribute to secondary organic aerosols (SOAs) in the atmosphere and influence climate and local air quality [1]. In this respect, unsaturated VOCs such as ethylene (C₂H₄) [2] and isoprene (C₅H₈) [3], have been intensively investigated. The key reactants for the oxidation processes are ozone (O₃), hydroxyl (OH), and nitrate (NO₃) radicals. Ozone forms charge–transfer (CT) complexes with various organic molecules, and exhibits strong CT bands in the ultraviolet–visible (UV–VIS) region [4,5,6,7]. The oxidation processes involve reactions in both light and dark conditions; therefore, it is important to understand how light-induced oxidation proceeds in the olefin–ozone system.

The purpose of this study is to investigate the light-induced reactions of olefin–O₃ systems spectroscopically in a cryogenic noble gas matrix to minimize the effect of thermal excitation of reactants. Numerous investigations have been conducted on reactions of olefin-O₃ systems experimentally [8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23] and theoretically [24,25,26]. Many of the gas phase studies have been focused on ozonolysis in dark conditions, and have been done at moderate temperatures [8,9,10,11,12,13]. Quite recently, these reactions have been under further scrutiny at lower temperature by using new experimental techniques [14,15]. In matrices, on the other hand, studies on low temperature reactions have been done for decades [16,17,18,19,20,21,22,23]. In situ observations of light-induced reactions of olefin-O₃ systems at cryogenic temperature are, to the best of our knowledge, not so abundant [23]. Ault and coworkers have performed systematic studies on these systems [24,25,26,27], and we will refer to them in comparison with this study.

In the present study, two samples, i.e., olefin/Ar (olefin = ethylene (C₂H₄), 1,3-butadiene (C₄H₆), isoprene (C₅H₈), dimethyl butadiene (C₆H₁₀)) and O₃/Ar premixed samples, were co-deposited on a cold CsI window via separate pulse valves and irradiated by several light sources at various wavelengths. The observed vibrational features of the photoproducts were compared with existing spectral data and the simulated features obtained using quantum chemical calculations, as performed previously [28,29].

2. Materials and Methods

2.1. Experiment

Gaseous ethylene and butadiene were purchased from GL Sciences and TCI, respectively, and used without further purification. Isoprene and dimethyl butadiene (WAKO) were degassed in freeze–pump–thaw cycles before sample preparation. Ozone was generated using a homemade ozonizer made of O₂ gas, trapped in a liquid N₂-cooled silica gel, and subjected to vacuum distillation to remove residual O₂. Each gaseous sample was diluted with Ar at various ratios and stored in a shielded glass vessel. Olefin/Ar and O₃/Ar samples were codeposited onto a cold CsI window through separate pulse valves at 9–30 K, with a repetition rate of 10 Hz and pulse width of 450 µs. Matrix-isolated species were irradiated by several light sources, including continuous wave (CW) diode lasers at 405, 670, and 800 nm, pulsed nanosecond lasers at 532 and 780 nm (LOTIS LS-2137/2 and LT-2211A), and a CW Xe discharge lamp (ENERGETIQ EQ-99FC) with sets of optical filters. The typical exposure time was 10–30 min. Infrared (IR) spectra before and after photolysis were recorded with a JASCO FT/IR-6100 spectrometer, and the results were used to obtain the difference spectra for each system. Spectra were recorded in the region of 650–4000 cm⁻¹ with a resolution of 0.5 cm⁻¹. A schematic of the experimental setup is presented in Figure 1. Details of the instrumentation can be found in our previous articles [28,29]. The obtained difference spectra for C₂H₄, C₄H₆, C₅H₈, and C₆H₁₀ are displayed in Figure 2, Figure 3, Figure 4 and Figure 5, respectively. Spectral changes upon isotopic substitution (¹⁶O→¹⁸O) are listed for the three systems in Figure 6. Ozone (¹⁸O₃) was prepared using the same procedure as used for the preparation of normal species from ¹⁸O₂ (Matheson).

2.2. Calculation

For the assignment of photoproducts, density functional theory (DFT) calculations were carried out for candidate molecules. Structure optimization and vibrational calculation for each stable species were performed at the dcp-b3lyp/6-31 + G (2d,2p) level [30]. Anharmonic vibrational frequencies for all fundamental bands of each species were obtained with vibrational second-order perturbation treatment [31,32] for direct comparison with the observed vibrational peak position without frequency scaling. All calculations were completed with Gaussian09 (Rev. 02) [33].

3. Results

3.1. C₂H₄–O₃ System

Figure 2 shows the difference spectra for the ethylene–ozone system upon irradiation; the photoproducts produce positive peaks, whereas reactants show depletion. Below 15 K, the difference spectra showed no decrease in reactants nor the emergence of photoproducts. This indicated that thermal diffusion after deposition was crucial to achieving close contact between C₂H₄ and O₃ in the matrix. Four peaks could be ascribed to the photoproducts, and two of them (at 2138 and 2343 cm⁻¹) were readily assigned to CO and CO₂, respectively. While the peaks at 867 cm⁻¹ and 1270 cm⁻¹ were assigned to the ν₅ and ν₃ bands of ethylene oxide (c–C₂H₄) [34], respectively, the assignment of the most intense band at 1726 cm⁻¹ was not straightforward, because it was not due to the aldehydes (H₂CO [35] and CH₃CHO [36]) or the carboxylic acids (HCOOH [37] and CH₃COOH [38]) as expected. Finally, we assigned this peak to trans-glyoxal (CHO–CHO), in conformation with previous matrix data [39]. The remaining peak at 1351 cm⁻¹ could not be assigned to a known species. Although it was in good agreement with ν₇ of CH₃CHO, a ν₈ band with comparable intensity could not be located. Therefore, it was left unassigned. From existing data on the IR band intensities of these molecules [40,41,42], it was found that the depletion of C₂H₄ and O₃ obeyed ~ 1:1 stoichiometry. The mass balance of C atoms in these photoreactions could be estimated to be more than 43% in total, and more than 70% for UV-photolysis. Therefore, it is reasonable to conclude that the photoreactions of the C₂H₄–O₃ system mainly proceed without C–C bond breaking.

It is known that the reaction of C₂H₄ and an oxygen atom (³P) in the gas phase at room temperature produces mainly CH₃CHO and fragmented products due to the high excess energy of the C₂H₄O adduct [43]. Conversely, in liquid N₂, the addition reaction leads to the simultaneous formation of c–C₂H₄O and CH₃CHO with a comparable branching ratio, owing to the rapid energy release of the adduct into the cold media [44]. The absence of CH₃CHO in the photoproducts in the Ar matrix suggests that the photoreaction of the C₂H₄ + O₃ system is not stepwise as follows:

$${\mathrm{C}}_2{\mathrm{H}}_4+{\mathrm{O}}_3\stackrel{\mathrm{h}\mathsf{\nu }}{\to }{\mathrm{C}}_2{\mathrm{H}}_4+\mathrm{O}+{\mathrm{O}}_2\to {\left[{\mathrm{C}}_2{\mathrm{H}}_4\mathrm{O}\right]}^{‡}\to \mathrm{products}$$

where [C₂H₄O]^‡ stands for a transition state.

It seems more reasonable to assume the following concerted reaction instead:

$${\mathrm{C}}_2{\mathrm{H}}_4+{\mathrm{O}}_3\to \left[{\mathrm{C}}_2{\mathrm{H}}_4\cdots {\mathrm{O}}_3\right]\stackrel{\mathrm{h}\mathsf{\nu }}{\to }\mathrm{products}$$

where [C₂H₄···O₃] represents a CT complex. Contrarily, the decrease in CO and CO₂ during photolysis at longer wavelengths may indicate that they originate from the UV dissociation of photoproducts (mainly CHO–CHO).

Production of CHO–CHO from this system is reasonable in view of the following energetics:

C₂H₄ + O₃ → CHO–CHO + H₂O △H₂₉₈ = −648.9 kJ mol⁻¹

C₂H₄ + O₃ → CH₃CHO + O₂ △H₂₉₈ = −365.8 kJ mol⁻¹

C₂H₄ + O₃ → c–C₂H₄O + O₂ △H₂₉₈ = −247.7 kJ mol⁻¹

Understanding the reasons behind why CH₃CHO was not observed in the photoproducts of this system is difficult. Theoretically, the isomerization of c–C₂H₄O to CH₃CHO and vice versa is hampered by the high activation energy barriers [45]; therefore, we can safely state that CH₃CHO is not a primary product in the photolysis of the C₂H₄--O₃ complex at cryogenic temperatures. In the dehydration reaction, glyoxal and H₂O molecules are produced simultaneously. The apparent absence of H₂O in the observed spectra is due to a compound effect of optical fringes and the ineffective removal of atmospheric water vapor, even with a purge system by dry N₂ flow. Since it was practically impossible to assign photoproducts in this region for larger olefins, the corresponding part was trimmed in Figure 4, Figure 5, Figure 6 and Figure 7; see each figure caption.

The ozonolysis of C₂H₄ (dark reaction) initiated at an elevated temperature, with an energy threshold of 5.2 kCal mol⁻¹ [46]. Thus, this study shows the initial stage of photoreactions followed by complicated chains of the oxidation processes, leading to secondary organic aerosol (SOA) formation at ambient temperature.

Therefore, it is assumed that the primary photolysis of olefin–O₃ systems occurs in the following ways:

Olefin + O₃ → oxirane-like product + O₂

and

Olefin + O₃ → glyoxal-like product + H₂O

as observed for the C₂H₄–O₃ system. The primary products predicted from these two schemes for each olefin are shown in Figure 7. Regarding the dehydration products from C₅H₈–O₃ and C₆H₁₀–O₃ systems, it was difficult to derive glyoxal-like molecules by removing one O₂ and two H₂ atoms symmetrically from the Criegee intermediates (see the bottom trace of Figure 7). Thereafter, other types of molecules, including an oxirane skeleton and a formyl group, were assumed. The name of each molecule was obtained from PubChem via the Add-ons of the ACD/ChemSketch software. Simulated spectra for the predicted products were obtained from anharmonic vibrational calculations, as described in Section 2.2, and compared with the experimental spectra in Figure 8, Figure 9 and Figure 10.

3.2. C₄H₆–O₃ System

Figure 3 shows the difference spectra of the 1,3-butadiene–ozone system upon irradiation similar to that of the C₂H₄–O₃ system. In addition to the peaks for CO and CO₂, four vibrational peaks were observed as the photoproducts. Notably, the peak at 1749 cm⁻¹ was broad and degraded at longer wavelengths. As mentioned above, the production of CO and CO₂ is suppressed during photolysis at longer wavelengths. The observed peaks reproduced the result of Ault [27]. Based on the reaction schemes in Figure 7, the following were assigned to the vibrational peaks of the photoproducts:

822, 923, 993 cm⁻¹: butadiene monoxide

1385, 1652, 1713, 1749 cm⁻¹: 2-oxo-3-butenal

The gas-phase spectrum for butadiene monoxide in the National Institute of Standards and Technology (NIST) database conformed to the present matrix-isolation peaks. The peaks assigned to butadiene monoxide also conformed to those observed in the Ar matrix produced from visible laser photolysis of the C₄H₆–NO₂ system [47]. Conversely, for the 2-oxo-3-butenal, no experimental spectrum is available, so direct comparison was not possible. Instead, the anharmonic vibrational calculation showed a good match with the observed peak positions (Figure 8). Ault assigned the peaks at 1385, 1652, 1749 cm⁻¹ to 3-butenal and the peak at 1713 cm⁻¹ to 2-butenal. Tanaka et al. reported a vibrational peak of 3-butenal at 1735 cm⁻¹ [47], and this position is not consistent with the assignment of ref. [27]. The spectral matches between observation and calculation in Figure 8 showed better agreement for 2-oxo-3-butenal, and it would be more conclusive to assign these peaks to the glyoxal-like 2-oxo-3-butenal. The isotopic shift (¹⁶O → ¹⁸O) of this band is 24 cm⁻¹, which is comparable to the calculated anharmonic wavenumber shift for 2-oxo-3-butenal (36 cm⁻¹). The two peaks at 1250 and 1311 cm⁻¹ could not be assigned to these two molecules and remain unidentified. Besides, we cannot rule out the possibility that peaks of other photoproducts overlap with the 1749 cm⁻¹ peak, in view of the broad and strong spectral feature.

3.3. C₅H₈–O₃ System

Figure 4 shows the difference spectra for the isoprene–ozone system upon irradiation. Due to its limited sensitivity, only two peaks at 917 and 1740 cm⁻¹ were observed in addition to the peak for CO₂. In contrast, the inequivalence of the two double bonds in isoprene provides two possible isoprene–ozone complexes, and thus, four photoproducts are predicted (see Figure 7). We carried out DFT calculations at the B3LYP/aug-cc-pVTZ level for the CT complexes, and found that both CT complexes are comparably stable (energy difference < 50 cm⁻¹). It is thus reasonable to assume that they coexist. Among these four candidate molecules, the spectral database contained the spectra for only 2-methyl-2-vinyloxirane in the form of transmission and ATR spectra for liquid films [48]. The experimental spectra showed a strong peak at 890 cm⁻¹ that does not conform to the peak at 917 cm⁻¹. Tanaka et al. observed visible laser photoproducts in the isoprene–NO₂ system and assigned the 930 cm⁻¹ peak to this molecule. Since these two peaks cannot be misidentified, the 917 cm⁻¹ peak was assigned to 3,4-epoxy-2-methyl-1-butene. Although the 917 cm⁻¹ peak was close to the O-O stretching band of the simplest Criegee intermediate CH₂OO [49], such possibility was ruled out from the very small isotope shift upon ¹⁶O → ¹⁸O substitution; see Figure 6.

Regarding the 1740 cm⁻¹ band, the two candidate molecules exhibited similar simulated patterns, and it is not clear which is more plausible. Because of the considerable energy difference between them, the peak was temporarily assigned to 3-methyl-2-oxobut-3-enal. As shown in Figure 9, the spectral match remains qualitative owing to the limited spectroscopic data available. The isotopic shift (¹⁶O → ¹⁸O) of this band is 28 cm⁻¹, which is comparable to the calculated anharmonic wavenumber shift for this molecule (34 cm⁻¹).

3.4. C₆H₁₀–O₃ System

Figure 5 shows the difference spectra for the dimethyl butadiene–ozone system upon irradiation. Five peaks were observed at 864, 1071, 1176, 1649 and 1733 cm⁻¹, in addition to the CO₂ peak. Notably, the production of CO₂ is negligible even at 405 nm, implying the absence of a glyoxal-like product. Tanaka et al. observed peaks at 865, 1070, 1175 in the visible laser photolysis of a DMB–NO₂ system in Ar and assigned them to 2-methyl-2-(1-methylethenyl) oxirane [47]. They did not report the peak at 1649 cm⁻¹, and it may be obscured by the strong peak of nitrite radical at 1657 cm⁻¹. They observed a peak at 1731 cm⁻¹ and assigned it to 2,3-dimethyl-3-butenal. Our assignments of the three low lying peaks were consistent with theirs, although it is not clear which molecule the 1733 cm⁻¹ band is assignable to; no experimental IR data are available for the authentic samples of the two candidate molecules. Anharmonic vibrational calculations for these two species provided essentially the same spectral patterns, and the qualitative match between the experiment and calculation is similar to that of the C₅H₈–O₃ system, as shown in Figure 10. The isotopic shift (¹⁶O → ¹⁸O) of this band is 33 cm⁻¹, comparable to the calculated anharmonic wavenumber shift for the molecule (38 cm⁻¹).

4. Discussions

The peak positions of the photoproducts in this study are listed in

Table 1. Observed vibrational peaks of photoproducts and their assignments. The name of each molecule was obtained from PubChem. The molecule in the last row cannot be found in the database, so its structure is presented instead.

Olefin	Peak Position (cm−1)	Assignment	Calc.
C2H4	867	Ethyleneoxide ν5 (ring deform)
	1270	Ethyleneoxide ν3 (ring stretch)
	1351	unidentified
	1726	Glyoxal ν10 (C=O stretch)
C4H6	822	butadiene monoxide (ring deform)
	923	butadiene monoxide (ring deform)
	993	butadiene monoxide (CH2=C bend)
	1250	unidentified
	1311	unidentified
	1385	2-oxo-3-butenal (CH2 bend)	1408
	1652	2-oxo-3-butenal (C=C stretch)	1648
	1713	2-oxo-3-butenal (C=O stretch)	1725
	1749	2-oxo-3-butenal (C=O stretch)	1773
C5H8	917	3,4-epoxy-2-methyl-1-butene (ring deform)	907
	1740	3-methyl-2-oxobut-3-enal (C = O stretch)	1696
C6H10	864	2-methyl-2-(1-methylethenyl) oxirane (ring deform)	864
	1071	2-methyl-2-(1-methylethenyl)oxirane (CH2=C bend)	917
	1176	2-methyl-2-(1-methylethenyl)oxirane (CH2=C bend)	1180
	1649	2-methyl-2-(1-methylethenyl)oxirane (C=C stretch)	1690
	1733	(C=O stretch) Or 2,3-dimethyl-3-butenal (C=O stretch)	1782
			1759

, together with their assignments. As described above, the vibrational peaks for the photoproducts for each of the three olefin–ozone systems can be interpreted consistently by assuming there was no C–C bond cleavage upon irradiation. It would be reasonable to conclude that the CT bands of these ozone complexes do not have significant absorption cross-sections in the longer wavelength limit of the visible region, because we could not detect products of photolysis at 780 and 800 nm. Regarding the C₂H₄–O₃ system, we could not observe photoproducts spectroscopically upon irradiation at 405 nm, and there might be a “gap” in the CT band around this wavelength, whereas no such depletion of photolysis was found for other olefins. The observation of CT bands has been reported for some organic molecule–ozone complexes [4,5,6,7]. Jonnalagadda et al. measured the CT band for the C₂H₄–O₃ complex in the Xe matrix [6]. The spectrum shows a maximum at ca. 300 nm and a decrease in the longer wavelength region. This is consistent with the photolytic efficiency observed in this study, except for the possible gap at approximately 405 nm. One possible source of this small discrepancy might be the difference in host atoms; Xe is more polarizable than Ar and, therefore, may influence the CT band profile via host–guest interactions. The enhancement of the CT band for the allene–O₃ complex in Xe was found to be the host effect [4]. We also attempted to observe the CT band of the C₂H₄–O₃ complex in the Ar matrix using a UV–VIS spectrometer (StellarNet Qmini2 WIDE-V), but the strong scattering of the host matrix did not allow for reproducible detection. Instead, we carried out two types of theoretical calculations for the CT band using time-dependent DFT (TD-DFT) and a configuration interaction including only single excitation (CIS). An optimized structure for the C₂H₄–O₃ complex in the electronic ground state was obtained at the cam-B3LYP /aug-cc-pVTZ level, and it was used as the reference geometry for the excited state calculations. The range-separated function was employed because it provides a better description for excited states of CT complexes [50,51,52]. The results shown in Figure 11 indicate that CIS predicts a bimodal absorption profile, whereas TD-DFT derives a strong absorption band in between. It seems that the CIS calculation qualitatively reproduced the observed photolytic efficiency.

In this study, the types of photoproducts derived were not dependent on the wavelength of the light source. Kamata et al. [53] and Kon et al. [54] reported the switching of photoreaction channels at visible wavelengths. It is still not clear how and why such differences arise, which presents another issue to be solved.

5. Conclusions

In this study, we experimentally observed the photolysis of four types of olefin–ozone complexes and presented the possible assignment of photoproducts. Regarding the C₂H₄–O₃ and C₄H₆–O₃ systems, the spectral match between observation and prediction seemed satisfactory, but those for the other systems remain ambiguous, mainly because of the unavailability of authentic sample spectra. We must stress that the work of Tanaka et al. on diene–NO₂ systems [47] was very helpful in narrowing down candidate molecules, even though we had to rely on the simulated spectra for the temporary assignment of many of these products. Further research is required to verify the assignments. Finally, the importance of the low carbon balance for the C₂H₄–O₃ system reaction that amounts to 70% at most, as described above, indicates that there are some other reaction channels followed by the photoexcitation of the CT band, whose products were not identified by infrared spectroscopy. The application of other experimental techniques is required in future.