Ultrafast transient absorption spectroscopy of the photodecarbonylation of photo-oxadibenzocyclooctyne (photo-ODIBO)

Selectivity, high yield, and favorable reaction kinetics are some of the criteria required for efficient click-reactions.¹ Realizing this reaction class in living cells, however, demands also that the reagents are biocompatible. Traditionally, azide–alkyne cycloaddition reactions made use of exogenous metals as catalysts, but the cytotoxicity of these catalysts restricts their application in living cells.^1–4 Chemical synthesis has since shifted to cyclooctynes, dibenzocyclooctynes, and azadibenzocyclooctynes as more versatile alternatives.¹ In addition to their bio-orthogonality and fast kinetics of the strain-promoted azide–alkyne cycloaddition (SPAAC), cycloalkynes and alkynes have an edge because they can be photochemically generated from their parent cyclopropenones with high quantum yields (QYs) (Φ = 0.2–1.0).⁵ For example, yields as high as 1.00 ± 0.003 have been reported for the formation of diphenylacetylene (DPA) from diphenylcyclopropenone (DPCP) in benzene.^6–8 

Another key merit of cycloalkyne click reactions is that the parent cyclopropenones are comparatively inert. Photo-OxaDiBenzocycloOctyne (Photo-ODIBO) and photo-DIBO, for example, can survive heating up to 160 °C and shows no reactivity with thiols.^2,5,8 Because cells have an abundant supply of thiols, it is important for click reagents to remain unaltered by the contents of the cell. Upon photo-generation, the cycloalkynes such as ODIBO have been shown to exhibit cycloaddition reaction rates as high as 45 M⁻¹s⁻¹, while the parent cyclopropenones are unreactive toward azides.² Ultimately, these reagents are superior because they allow for spatial and temporal control of metal catalyst-free click reactions, which is important in applications such as labeling of biomolecules in situ.^2,9

Although photodecarbonylation is the most common reaction of cyclopropenones, there is very little agreement in the literature on the photoactivation mechanism. A multitude of studies and techniques have been employed toward understanding their photo-dissociation to cycloalkynes.^{2,6,7,10–14} Among these techniques, transient absorption spectroscopy (TAS) is the most popular because of the ability to record excited-state evolution with spectral and temporal resolution. To date, three different pathways have been proposed for this photodissociation reaction based on TAS measurements.⁶ The mechanism one starts with UV excitation from the cyclopropenone ground state to the n-th singlet state (S_n, where n > 1) followed by a decay to the cycloalkyne singlet (S₁). Subsequent intersystem crossing to the triplet (T₁) follows thereafter. The second proposed pathway suggests that the cycloalkyne is formed directly from a vibrationally hot ground state of the cyclopropenone.¹⁵ The third proposal argues that the ground state cyclooctyne is formed via a short-lived density functional theory (DFT)-predicted zwitterionic intermediate on the excited singlet surface.^6–8 This intermediate comprises a positively charged C=O group and negative charge on the corresponding acetylene.^7,15 A summary of the three possible reaction pathways as proposed by Vennekate et al. is provided in Scheme 1.⁶ 

The purpose of this study is to determine the pathway for the generation of ODIBO from photo-ODIBO using femtosecond transient absorption spectroscopy (TAS), which, to the best of the authors’ knowledge, has not been established yet.

The ultrafast measurements were taken using a 1 kHz regenerative amplifier (Coherent Inc. Legend Elite) seeded by a Ti:sapphire oscillator (MIRA Optima 900) and a custom transient absorption spectrometer. ∼0.6 mJ of the 800 nm fundamental beam is passed through an 80/20 beam splitter for the pump and Vis/NIR probe. The remainder of the fundamental beam is used to pump a traveling-wave optical parametric amplifier (TOPAS-C) to produce tunable UV pulses at 3.5 μJ–12 μJ and 135 fs at the sample position. 350 nm and 321 nm wavelengths were chosen as pump wavelengths from UV–Vis spectroscopy data of photo-ODIBO.² The TOPAS-C output pump beam is then sent through an optical chopper (Thorlabs MC2000) operating at 500 Hz to block every other pulse. This pump beam is also focused just beyond the sample position using a 750 mm lens in order to keep the beam size larger than the probe beam.

The white-light continuum (WLC) probe is generated from the 800 nm fundamental in a 3 mm sapphire window and spans a wavelength range from 440 nm–770 nm with the excess fundamental subsequently filtered out with a short pass 770 nm filter. The relative timing between the pump and probe pulses is scanned by means of an optical delay line that consists of a retroreflector mounted on a motorized translation stage (Newport Inc. ILS150PP with ESP301 motion controller). An Avantes (AvaSpec-1650F-USB2) fiber spectrometer records the transmitted WLC spectra at each pump–probe delay position. The transient absorption spectra (ΔmOD) are calculated as follows:

where $Ion, off$ are the probe intensities with and without excitation.

The liquid samples were cycled in a flow cell (Harrick Scientific TFC) with 1 mm and 2 mm CaF₂ windows for the front and back, respectively. The two windows are separated by a 2 mm inert rubber O-ring. TAS measurements were performed on 1 mM solutions of photo-ODIBO and ODIBO in methanol to help distinguish spectral signatures associated with the relaxation dynamics of photo-ODIBO. All the transient and steady-state spectroscopy experiments were conducted under ambient temperature and pressure. TAS data were analyzed using Global Lifetime Analysis (GLA) in Glotaran.¹⁶ Chirp, time-zero correction and initial kinetic fitting were done in Surface Xplorer.¹⁷ Lifetime density analysis was also carried out for the TAS data at each pump wavelength using the Optimus computer package.¹⁸ 

The quantum yield of the photo-decarbonylation reaction of photo-ODIBO in methanol solutions under 321 nm and 350 nm irradiation has been determined by chemical actinometry using 4-nitroveratrole as a standard (see Sec. 2 of the supplementary material).¹⁹ The quantum efficiency of photo-ODIBO conversion was found to be slightly higher under 350 nm (Φ = 0.18 ± 0.02) than under 321 nm irradiation (Φ = 0.14 ± 0.01). Nanosecond laser flash photolysis was also conducted using the LKS.50 kinetic spectrometer (Applied Photophysics) equipped with a Nd:YAG laser (see Sec. 1 of the supplementary material).

Quantum-mechanical calculations were then carried out in GAMESS.²⁰ Geometry optimization with DFT was undertaken using the Becke, 3 parameter, Lee-Yang-Parr hybrid functional with the 6-311G (3df, 2p) basis set. Excited-state energies were computed using the same functional and basis set combinations by employing the time-dependent density functional theory (TD-DFT) method, and the results were compared with steady-state experiments. The Raman modes were also calculated in order to explain vibrational features in the excited-state TAS data.

From previous UV–Vis studies conducted by McNitt et al., photo-ODIBO showed intense absorption bands at 343 nm, 329 nm, and 308 nm.² The predicted absorption peaks for photo-ODIBO are at 385 nm, 329 nm, 291 nm, and 284 nm, as shown in Fig. 1. These peaks can be attributed to the S₁, S₂, S₃, and S₄ singlet states of photo-ODIBO. These calculated energies suggest that the 343 nm absorption peak in the experimental steady-state spectrum arises from excitation to the S₂ state of photo-ODIBO. The lower wavelength absorption peaks at 329 nm and 308 nm, therefore, correspond to the S₃ and S₄ states, respectively. The calculations also show a very low oscillator strength transition to S₁ at 385 nm and the most likely orbitals for this transition point to an nπ^* character for this transition with n-orbitals located on the carbonyl lone pair. UV–Vis collected from a concentrated sample revealed a shoulder at 402 nm, which could be evidence of the low-lying S₁ singlet predicted by TD-DFT (Fig. S1). The absorption spectrum for the photoproduct (ODIBO) was also recorded, and it showed absorption bands at 322 nm, 308 nm, and 290 nm.² The calculated excitation energies for ODIBO are 320 nm, 296 nm, 287 nm, and 274 nm corresponding to excitation to S₁, S₂, S₃, and S₄ levels, respectively. Since ODIBO does not show any calculated nπ^* transitions due to the missing carbonyl group, the absorption peaks in the ODIBO spectrum result from ππ^* transitions. Similar molecules like diphenylcyclopropenone (DPCP) and diphenylacetylene (DPAC) have also been shown to have close-lying singlet states; thus, the bands in Fig. 1 could be S₀–S_n transitions with n = 1, 2, and 3 in the respective molecules.¹¹ The lowest-lying triplets for both photo-ODIBO and ODIBO were also calculated and are listed in Table 2 of the supplementary material.

ODIBO absorption at 350 nm is rather weak (Fig. 1), thus allowing for selective excitation of photo-ODIBO. As such, the wavelength 350 nm was chosen as one of the pump wavelengths to rule out possible contamination of the dynamics if a newly formed ground state photoproduct is present. Independent TAS measurements with 321 nm excitation on ODIBO and photo-ODIBO were also carried out to identify ODIBO excited state spectral signatures in the photo-ODIBO TAS data. TAS data following 350 nm and 321 nm excitation are compared for both the parent and product molecules in order to resolve their dynamics toward the proposal of a pathway for the photodissociation of photo-ODIBO. McNitt et al. previously reported a photodissociation probability of 0.16 using a UVA light source (λ_max = 350 nm).²¹ In the present study, quantum yield (QY) measurements were completed under steady-state irradiation conditions using monochromatic radiation (Sec. 1 of the supplementary material). The QY for photo-decarbonylation of photo-ODIBO at 350 nm is Φ₃₅₀ = 0.18 ± 0.02 and at 321 nm is Φ₃₂₁ = 0.14 ± 0.01. As such, the expected amount of photoproduct formed at both wavelengths is roughly similar. The UV–Vis spectra and a short description of QY determination are given in the supplementary material (Fig. S2).

The time-resolved TAS data for ODIBO after 321 nm excitation are depicted in Figs. 2(a), 2(d), and 2(g). From the static absorption spectra and DFT calculations presented in Table I, this pump wavelength results in excitation to the first singlet (S₁). As such, the broad positive excited-state absorption (ESA) signal with a peak near 480 nm, shown in Fig. 2(a), is ascribed to S₁–S_n transitions. The ESA signal covered the entire probe spectral window from 390–676 nm, and within 100 ps, another peak at 622 nm became more apparent in the spectrum. At the end of the experimental time window, a double peak feature (481 nm and 622 nm) with a small shoulder at 407 nm remained in the spectrum. No ground state bleaching (GSB) signal was observed as it is expected at shorter wavelengths than the 376 nm short wavelength limit for our probe. 321 nm is sufficient for excitation to the ODIBO S₁ state. In fact, the wavelength assigned to the S₁ absorption shows the highest oscillator strength. It follows, therefore, that the 481 nm peak that showed up immediately after excitation was a result of pump induced probe absorption from the first singlet. The 407 nm shoulder, 481 nm, and 622 nm peaks observed at a long delay-time could arise from the ODIBO triplet T₁ predicted by DFT to be at 466 nm. This assignment is based on the fact that T₁ is energetically below S₁ and can therefore be populated non-radiatively via intersystem crossing. A similar double-peak feature was observed after photoexcitation of a similar alkyne, DPA, and was attributed to the S₁ state.⁷ This assignment is unlikely in our spectra because our excitation directly accesses the S₁ state of ODIBO and such a spectral feature would have to be present at earlier time delays in our spectra.

Global analysis of the TAS spectra showed an adequate description of the dynamics with random residuals (see Fig. S9) using a sequential triple exponential decay model convoluted with a Gaussian. The resulting evolution associated spectra (EAS) for each of the components are shown in Fig. 2(d). Figure 2(g) shows the kinetics for each of the spectra given in the EAS plots obtained from global analysis. Select spectra normalized at the 481 nm maxima (see Fig. S5) showed that within the first 10 ps, only the intensity of the longer wavelength region decreased, and the higher wavelength edge remains unchanged leading to band narrowing with increasing delay-time. This feature is indicative of a vibrational cooling step in the early dynamics.^22,23 The first time constant, 9.93 ps, was therefore assigned to vibrational cooling to the S₁ minimum. The 622 nm peak that grew into spectra after about 90 ps suggested that the next step (83.1 ps time constant) described an S₁ – T₁ intersystem crossing. The third time constant was then assigned to trapping in the T₁ state due to an optically forbidden transition back to the ground state. This long lifetime (>1 ns) was not fully resolved within our 800 ps experimental time window.

For the parent molecule, photo-ODIBO, 350 nm excitation results in the appearance of a 466 nm absorption peak, as shown in Fig. 2(b). Using the DFT and UV–Vis results as a reference, 350 nm lies on the red-side of the first peak in the absorption spectrum depicted in Fig. 1. This wavelength is also slightly red-shifted from the calculated position of the first allowed transition (329 nm). The positive absorptive signal (ESA) is, therefore, ascribed to S₂–S_n transitions following excitation by the pump pulses. In fact, the ratio of the TD-DFT oscillator strengths for S₀ – S₂ (⁠$fS2$⁠) transition is 48 times greater than that for S₀ – S₁ (⁠$fS1$⁠), i.e., (⁠$fS2fS1≈48$⁠). This ESA band showed up immediately after excitation and extended to the long-wavelength region of the spectrum. Within the first 100 fs of evolution, a negative band from 380 nm to 414 nm appeared in the spectra. The intensity of this band was about 14% of the ESA signal observed at the same time scale, and at about 7 ps, the negative signal is overwhelmed by the positive ESA band tail. The negative band observed at small delay-times is most likely due to the stimulated emission (SE) from the S₂ singlet. This band cannot be assigned to S₁ on the grounds that S₁ is a dark state (nπ^*, low oscillator strength), and as such, it would be less probable to observe emissions from this state, as compared to the bright ππ^* S₂ state. This SE feature also showed a progressive redshift with delay-time, which could arise from solvent interactions with the molecule in the S₂ state. Such a solvent interaction will result in a lower energy S₂-solvent equilibrium state. In TAS spectra, this shift will cause the SE band to progressively move toward lower energy (higher wavelength) regions in the probe.^7,24 Ground state bleaching (GSB) can be excluded as a possible assignment for this negative feature because the SE band was red-shifted with respect to the expected location for the bleach (∼350 nm). At 500 ps, a small negative band and a broad positive feature from 474 nm to 676 nm remained.

For the 321 nm excitation of the parent molecule, photo-ODIBO, a 466 nm band also appeared immediately after excitation, as shown in Fig. 2(c). The excitation wavelength used was sufficient for excitation to photo-ODIBO S₂ as it is higher in energy than the S₂ energy level predicted by DFT calculations and shown in the UV–Vis (Fig. 1) spectrum. This 466 nm band, similar to 350 nm excitation TAS, was accordingly assigned to ESA from S₂ – S_n. For delay-times as early as 300 fs, the growth of a second peak at 620 nm was observed and can clearly be seen in the spectra normalized at the 466 nm maximum (Fig. S4). With this peak, a shoulder at 407 nm also developed and both the 407 nm and 622 nm features remained at the maximum delay-time. With the higher energy excitation, no negative SE or GSB signal was observed, which could be a result of overlap with photoproduct spectral features in the spectra. It should also be noted that although photodecarbonylation results in CO formation, no overlapping carbon monoxide bands are observable in the visible wavelength probe window.²⁵ Previous studies on similar cyclopropenone, however, were able to observe CO bands in the IR region (2100 cm⁻¹–2200 cm⁻¹).⁷ 

In order to deduce the photodecarbonylation mechanism, the transient absorption characteristics of ODIBO and photo-ODIBO are now compared. 321 nm TAS spectra for photo-ODIBO and ODIBO showed striking similarities. At earlier delay-times (0.98 ps and 10.1 ps), the two molecules also shared the same spectral shape. The common features seen in the two molecules’ spectra confirm that 321 nm excitation of photo-ODIBO results in the formation of ODIBO, as was observed by the depletion of the 343 nm absorption peak in the UV–Vis spectra (Fig. S2). At long delay-times, both 407 nm shoulder and 622 nm peaks were observed in the parent molecule (photo-ODIBO) and product (ODIBO) TAS spectra and are shown in Fig. 3. Notable, however, is that photo-ODIBO has a maximum at 466 nm while ODIBO has a maximum at 486 nm, a difference that was seen in the early delay-time spectra and was consistent for all delay-times shown in Fig. 3. Since both 466 nm and 486 nm maxima occurred immediately after excitation, they were accordingly assigned to parent photo-ODIBO S₂ and ODIBO S₁, respectively. 350 nm excitation, on the other hand, gave a broad ESA feature that is very similar to the 321 nm spectra at short delay-times (100 fs spectra in Fig. 3), except for the negative SE band. Such a similarity confirms that both wavelengths result in S₀–S₂ transitions. The 350 nm excitation, however, is lower than ODIBO S₁, which means that at this wavelength, a different spectral profile attributable to a different pathway should be observed upon photodecarbonylation. In fact, within 1 ps after excitation, the 350 nm spectra began to deviate from the trends observed at λ_pump = 321 nm. At 10.1 ps (Fig. 3), two peaks located around 555 nm and 640 nm are observed, and at 250 ps, the 466 nm peak disappeared completely and was replaced by an SE feature from 425 nm to 475 nm. The 250 ps spectrum (Fig. 3, red) showed a striking resemblance to the spectra obtained from 400 nm photolysis of a similar cyclopropenone (number 5) in the study by Poloukhtine et al.⁷ In their transient absorption study, they ascribed the spectra to an excited state of a photodecarbonylation intermediate. Hence, the SE signal and double-peak ESA observed with 350 nm excitation could result from the formation of an intermediate that is otherwise not formed after 321 nm excitation, thereby also accounting for the missing SE in the 321 nm TAS data. The existence of an ultrafast intermediate accessible from the excited singlet was also observed by Poloukhtine et al., following 320 nm photolysis of a similarly conjugated cyclopropenone, 1-naphthyl, denoted as 1(n) in their study.²⁶ 

Also present in both photo-ODIBO and ODIBO TAS spectra are relatively sharp bands superimposed on the ESA signals [Figs. 2(e) and 2(f), Fig. S5]. The separation of these bands, i.e., the difference in frequency position of two nearby bands are 693 cm⁻¹, 857 cm⁻¹, 1139 cm⁻¹, and 1343 cm⁻¹ in photo-ODIBO and 517 cm⁻¹, 773 cm⁻¹, 957 cm⁻¹, 1213 cm⁻¹, and 1396 cm⁻¹ in ODIBO. To determine the nature of these vibrational modes, Raman calculations were carried out and the resulting Raman spectra are shown in the supplementary material (Fig. S6). Modes below 1000 cm⁻¹ involve ring vibrations perpendicular to the plane of the rings. The rest of the modes involve motion in the plane of the rings. Representative modes are also shown in the supplementary material (Fig. S7). The sharp peaks in TAS spectra were observed after excitation to the S₂ state in photo-ODIBO and S₁ in ODIBO. As such, the vibrations could be assigned to tuning motion responsible for inducing a degeneracy between these parent and product excited states.^27,28 This could then suggest a possible dissociation pathway between the connected photo-ODIBO S₂ and ODIBO S₁ states as a result of this degeneracy.

Global analysis and Singular Value Decomposition (SVD) of the 321 nm and 350 nm photo-ODIBO TAS showed that four component sequential exponential decay models were needed to adequately describe the dynamics (see Sec. 7 of the supplementary material). The residuals demonstrating the quality of the global fit are shown in Fig. 9 of the supplementary material. Figures 2(e) and 2(f) show the resulting EAS with the time constants indicated in the respective legends, and in Figs. 2(h) and 2(i), the global fit kinetics of the sequential growth and decay of each component spectra in the EAS are depicted. To allow further model-independent analysis of the lifetimes for each TAS experiment, lifetime density maps (LDMs) were also calculated.¹⁸ This complimentary analysis technique has been previously used to disentangle complicated dynamics for photo-isomerization mechanisms in photoswitches, charge transfer mechanisms in photosystem 1, and fluorescence decay dynamics.^29–31 The results from the density analysis on photo-ODIBO and ODIBO time-resolved spectra are presented in Fig. 4. ODIBO LDM³²¹ shows high-density lifetimes in the 10¹ ps–10² ps region across the whole probe wavelength range. The double peak feature above 10³ ps matching the long-lived (>1 ns) EAS in Fig. 3 can also be identified. For λ_pump = 350 nm (LDM³⁵⁰), the dominant lifetime is around the 0.2 ps–0.3 ps region, followed by contributions between 20 ps–80 ps. An additional peak around 4 ps lifetime can also be identified in the LDM³⁵⁰ [Fig. 4(b)]. Notably, photo-ODIBO LDM³²¹ for λ_pump = 321 nm [Fig. 4(c)] resembles the ODIBO LDM³²¹ closely, especially the positive distribution in the range 2 ps–200 ps and > 10³ ps. This photo-ODIBO LDM³²¹ also shows a small lifetime contribution around 0.3 ps similar to that observed in the same region in LDM³⁵⁰ for photo-ODIBO.

Since the photoproduct, ODIBO, does not undergo additional photodissociation upon 321 nm excitation, the time constants observed describe electronic relaxation from higher excited states to the ground state of ODIBO as described earlier.¹⁴ For the 321 nm excitation of photo-ODIBO, the analysis of spectra normalized at the 466 nm maximum revealed growth of the 622 nm peak within the first 1 ps of spectral evolution, which pointed to the presence of the product at those early time scales (Fig. S4). Normalizing spectra at 466 nm was done to highlight the growth of other spectral features that would otherwise be overshadowed by the intense main peak. As seen in Fig. S3, the 407 nm and 620 nm peaks develop into the spectrum as early as 500 fs in the photo-ODIBO spectra. Because of the similarity of this double peak feature to the ODIBO spectra, it can be concluded that the 294 fs lifetime describes the formation of an excited state of the photoproduct, ODIBO. The 294 fs EAS, therefore, corresponds to the excited-state (S₂) dissociation of the parent photo-ODIBO involving cleavage of the two C—C bonds releasing C=C. The individual cleavage of these bonds did not result in any observable spectral changes within the measured 294 fs lifetime and, therefore, could not be fully resolved in the TAS spectra. This dissociation time matches the characteristic photodecarbonylation time obtained for cyclopropenone DCPC (⁠$≤$200 fs).^14,32 Vibrational relaxation to S₂ minimum in photo-ODIBO could also be occurring on the same time scale since the shape of the 466 nm peak remained largely unchanged within this decay-time after excitation with higher energy than the S₀ – S₂ minimum energy gap. Competing vibrational relaxation could also account for the less than unity quantum yield (ϕ ∼ 14%–18%) measured from the steady-state spectra. Unproductive non-radiative internal conversion (IC) in the unconverted photo-ODIBO molecules would then follow this vibrational cooling, providing a loss mechanism leading to the low quantum yield.³³ Additionally, the LDM³²¹ for ODIBO and photo-ODIBO with pump = 321 nm revealed striking similarities for lifetimes beyond 10⁰ ps, which suggests that the newly formed excited-state photoproduct could be responsible for the subsequent dynamics observed. The formation of ODIBO and its excitation with subsequent 321 nm pulses cannot, however, be completely ruled out as a possible explanation for the similarities between 321 nm TAS for ODIBO and photo-ODIBO.³² Flowing the sample through the pumped volume cannot entirely remove previously excited molecules, which could lead to ODIBO contamination of the spectra. This contamination, however, is estimated to be insignificant on the basis of focal volume calculations and low conversion efficiency.

For λ_pump = 350 nm, there is very little similarity observed in both the EAS and LDMs, as compared to the 321 nm excitation. Beyond 980 fs, as depicted in Fig. 3, spectral features for photo-ODIBO deviate from those obtained with the lower wavelength (321 nm) excitation. In addition, in the LDM³⁵⁰, the characteristic lifetime distribution beyond 1 ps does not resemble either the 321 nm photo-ODIBO or ODIBO dynamics. These differences likely arise from an alternative photodecarbonylation mechanism at this higher wavelength. The 350 nm excitation wavelength cannot result in photodecarbonylation to the S₁ state of ODIBO as it is energetically insufficient to reach the higher ODIBO S₁ energy located by DFT and UV–Vis spectra. Dissociation to an ODIBO triplet (T₁ – 466 nm) is an energetically possible pathway, but the long-lived EAS and LDM components do not match the corresponding characteristics obtained from independent excitation of ODIBO, which rules out this possibility. However, given that the long-lived EAS component closely matched the spectra obtained for a similar cyclopropenone by Poloukhtine et al., the same mechanism involving an intermediate can also be proposed here.⁷ The presence of an intermediate in the dissociation pathway for cyclopropenones was also observed in several theoretical studies in the gas and solution phases.^7,8,15,26 Specifically, the 130 fs time constant is ascribed to vibrational relaxation immediately after excitation followed by the 1.15 ps internal conversion to the S₁ state of photo-ODIBO. The next time constant (31.05 ps) would then describe the formation of this excited-state intermediate, which would further relax to form the ground state photoproduct beyond the maximum delay-time for our experiment. The slightly higher quantum yield obtained with 350 nm photodecarbonylation (ϕ ∼ 18%) suggests that the mechanism via an intermediate is more efficient, but with more excess energy added in the 321 nm excitation, the conversion via ODIBO S₁ is favored.

The observation of the T₁ state of ODIBO was conducted using nanosecond laser flash photolysis (Sec. 3 of the supplementary material). Upon excitation of ODIBO (57 µM in MeOH) with 266 nm laser pulses, a transient with the lifetime of τ = 230 ± 15 µs is observed. The lifetime of this transient is inversely proportional to the oxygen concentration in solution, allowing for identification of this transient as the (T₁) exited state of ODIBO (Sec. 3 of the supplementary material). This observation suggests that the initially formed S₁ state of ODIBO undergoes rapid (τ < 10 ns) intersystem crossing to ODIBO T₁.

A transient with the same lifetime, albeit with a weaker absorbance, was observed in the 266 nm laser flash photolysis of photo-ODIBO. The transient observed in 266 nm photolysis of photo-ODIBO was also quenched by oxygen, suggesting that it can be assigned ODIBO T₁. The excitation of photo-ODIBO and ODIBO solutions with 352 nm pulses does not produce any detectable transients with lifetime τ > 10 ns. Since ODIBO has virtually no absorbance at this wavelength, it is just not being excited by the laser pulse. Irradiation of photo-ODIBO with 355 nm, on the other hand, results in the decarbonylation and the quantitative formation of ODIBO (Scheme 1). The latter observation allows for the conclusion that 355 nm, nanosecond irradiation-induced photodecarbonylation of photo-ODIBO produces ODIBO in its electronically ground state. The observation of the excited state of ODIBO in 266 nm photolysis of photo-ODIBO is probably due to the formation and subsequent excitation of ODIBO during the 4 ns–5 ns Nd: YAG laser pulse.

A summary of the photodecarbonylation mechanism of photo-ODIBO based on the TAS spectra and excited-state dynamics is shown in Fig. 5. The two excitation wavelengths are also indicated on the left and right sides of Fig. 5. IM^* indicates the intermediate proposed to explain the spectral features obtained with λ_pump = 350 nm. This intermediate results in a ground state of ODIBO, as observed in the nanosecond flash photolysis. While photodecarbonylation via an intermediate species cannot be entirely ruled out after 321 nm excitation, there are no obvious features in the EAS that would indicate any relevance.

The photochemical mechanism of conversion for photo-ODIBO to ODIBO following 321 nm and 350 nm excitation was established using transient absorption spectroscopy. Global lifetime and lifetime density analysis of the transient absorption data revealed the excitation wavelength-dependent mechanisms for the photodecarbonylation process. The 350 nm excitation of photo-ODIBO does not result in conversion to the excited-state ODIBO observable with our TAS setup; therefore, conversion via the intermediate is too slow and eventually yields ground state ODIBO. Excitation of photo-ODIBO using a 321 nm pump does result in ultrafast conversion to ODIBO in less than 294 fs, which is concluded from the observation of similar spectral features after 321 nm excitation of both photo-ODIBO and ODIBO. These similarities could imply the conversion pathway of photo-ODIBO results in an excited-state of the product ODIBO.

See the supplementary material for UV–Vis plots for higher concentration solution and quantum yield calculation, nanosecond flash photolysis, calculated triplets and orbitals for the first four singlets of each molecule, normalized TAS spectra, calculated vibrational modes, and lifetime analysis procedures (GLA and LDA).

This work was partially supported by the National Science Foundation (Grant Nos. CHE-1362237 and CHE-1800050).

The data that support the findings of this study are available from the corresponding author upon reasonable request.