Azulene as an anomaly in triplet-state engineering via bromination and/or carbonylation: internal conversion overrides intersystem crossing Authors: Kavya Vinod, Claire Tonnelé, Anup Rana, T. Amalnadh, P. E. Swathi Krishna, Juan Carlos Roldao, Henrik Ottosson, David Casanova, Mahesh Hariharan Published: 2026 ABSTRACT Azulene, a non-benzenoid aromatic system, exhibits a unique electronic structure that gives rise to unconventional excited-state dynamics. Here, we investigate a series of halogenated and carbonylated azulenes using ultrafast spectroscopy and quantum chemical calculations. Despite incorporating substituents that are known to enhance intersystem crossing (ISC), halogenated/carbonylated azulenes exclusively display singlet excited-state dynamics. Femtosecond and nanosecond transient absorption measurements reveal rapid internal conversion without any detectable triplet population. Computational analysis using the Marcus formalism confirms that, although spin–orbit coupling increases upon bromination/carbonylation, ISC remains orders of magnitude too slow to compete with fast internal conversion. These findings highlight the intrinsic resistance of azulene derivatives to triplet-state formation, challenging established strategies for promoting ISC and offering new insights into substitution-dependent excited-state pathways in non-benzenoid aromatics. Introduction Azulene (Az), a non-benzenoid aromatic hydrocarbon, exhibits an atypical excited-state ordering in which the optically allowed S2 state lies above a weakly emissive S1 state.2–7 This arrangement enables efficient S2 fluorescence and renders Az a prototypical chromophore for investigating non-Kasha photophysics.3,7–10 The origin of non-Kasha emission in Az has been clarified through the framework of excited-state aromaticity: the S2 state retains an aromatic character and relaxes only minimally, whereas the S1 state is antiaromatic and rapidly reorganises toward an S1/S0 conical intersection that facilitates ultrafast internal conversion (IC).11,12 This contrasting aromaticity provides a structural basis for the stabilisation of S2, the instability of S1, and the pronounced S2–S1 energy separation that facilitates the characteristic non-Kasha emission in Az. Beyond its unusual singlet-state ordering, Az possesses an intrinsically small S1–T1 energy gap (ΔEST ≈ 0.5 eV) arising from its non-alternant electronic structure, which spatially separates the HOMO and LUMO and consequently reduces the exchange integral.13 Such a narrow singlet–triplet splitting would ordinarily favour efficient triplet sensitization, provided that intersystem crossing (ISC) from S1 were competitive with IC. This prospect is of particular relevance given the increasing interest in metal-free triplet photosensitizers for photocatalysis, photodynamic therapy, and photovoltaic applications.14 Yet many organic chromophores suffer from large ΔEST values and weak spin–orbit coupling (SOC), which inherently limit ISC efficiency. A common strategy to enhance ISC in organic chromophores involves introducing heavy-atom substituents (e.g., Br, I) or heteroatom-containing carbonyl groups. These modifications perturb orbital character, introduce n–π* configurations, and increase SOC, routinely driving kISC to 108 to 1010 s−1 in conventional polyaromatic scaffolds.15–17 Beyond heavy-atom and carbonylation strategies, alternative design paradigms such as spin–orbit charge-transfer intersystem crossing (SOCT-ISC), charge separation (CS), low-lying charge-transfer (CT) state engineering, and radical pair-mediated intersystem crossing (RP-ISC), have emerged as effective, heavy-atom-free mechanisms to facilitate triplet generation in molecular architectures.18–23 However, obtaining substantial triplet populations requires kISC to outcompete all non-radiative singlet deactivation channels by roughly an order of magnitude.14 In systems where IC dominates, like photostabilizers engineered to suppress triplet populations via ultrafast non-radiative decay, the kinetic barrier for competitive ISC becomes exceptionally high.24,25 Because azulene possesses two robust, intrinsic IC channels (S2 → S1 IC on the nanosecond scale and antiaromaticity-driven S1 → S0 IC on the picosecond scale),11,12 the extent to which substituent-induced SOC enhancements can overcome these fast singlet deactivation pathways to enable triplet formation in azulene derivatives remains to be quantitatively established.26–31 Here, we investigate a series of halogenated and carbonylated azulenes through femtosecond and nanosecond transient absorption spectroscopy, supported by quantum-chemical simulations employing Marcus theory. Despite incorporating substituents that typically promote ISC, all derivatives examined display exclusively singlet-state relaxation pathways. Computed ISC rates remain orders of magnitude slower than IC, even with enhanced SOC, indicating that azulenes retain a strong intrinsic resistance to triplet-state formation. These results challenge conventional approaches to triplet-state engineering in azulenes and provide new insight into how excited-state energetics and molecular topology govern ISC in non-benzenoid chromophores. Results and discussion Az and its substituted derivatives (Fig. 1),32 6-tert-butyl azulene-1,3-dialdehyde (Az-1)32 and 1,3-dibromoazulene (Az-3)33 were synthesized following previously reported procedures. 2-Bromoazulene-1,3-dialdehyde (Az-2) and 1,2,3-tribromoazulene (Az-5) were synthesized by the Vilsmeier-Haack formylation and bromination by using N-bromosuccinimide (NBS) of 2-bromo-azulene34 (see Section 2 in SI for details), respectively. 1,3,6-Tribromoazulene (Az-4) was synthesized by the NBS bromination of 6-bromoazulene35 (see Section 2 in SI for details). All the synthesized molecules were fully characterized by using standard spectroscopic methods. Single crystals of Az-5, obtained from hexane, were characterized by X-ray diffraction and found to crystallize in a monoclinic system with a P21/c space group (Table S1). Chemical structure of azulene (Az) and the substituted azulenes (Az-1 to Az-5). The ground-state optical features of Az and Az-1 to Az-5 in toluene are shown in Fig. 2. The electronic absorption spectrum of Az displays a weak S0 → S1 transition at 600 nm and an intense, well-defined peak at 340 nm corresponding to the S0 → S2 transition, in good agreement with computed vertical excitation energies at the ADC(2)/def2-TZVP level (Table 1), and in line with previous reports.1,36–38 The lowest singlet–singlet excitation in Az arises from a HOMO → LUMO transition, while excitation to S2 involves contributions from both HOMO − 1 → LUMO and HOMO → LUMO + 1 transitions (Fig. S1).6,39–41 The substituted azulenes exhibit wavelength shifts in their electronic absorption spectra relative to Az, attributed to the electron-donating or electron-withdrawing nature of the substituents. Az-1 and Az-2 show S0 → S1 absorption maxima at 450 nm and 441 nm, respectively, consistent with the computed vertical excitation energies of 498 nm and 495 nm (Table S2). The blue-shift in the S0 → S1 absorption maxima for Az-1 and Az-2 relative to Az arises from an increased HOMO–LUMO gap, primarily due to a stronger stabilization of the HOMO (Table S3). This effect stems from the aldehyde groups being attached to carbon atoms with higher electron density in the HOMO.20 Despite the presence of bromine, which is a π-electron donating and σ-electron withdrawing substituent, the S0 → S1 transitions in the brominated derivatives Az-3, Az-4, and Az-5 are only slightly perturbed relative to Az, with absorption maxima at 614 nm, 619 nm, and 619 nm, respectively.14 These results are reproduced by electronic structure calculations, with vertical excitation energies closely matching the experimental absorption maxima. The S0 → S2 transitions of the substituted azulenes are red-shifted compared to Az, as reflected in both experiment and computed vertical energies. Az-1 and Az-2 exhibit S2 absorption maximum at 351 nm and 352 nm respectively, while Az-3, Az-4, and Az-5 show maxima at 371 nm, 362 nm, and 361 nm, respectively (Table 1). The fluorescence emission intensity of Az and Az-1 to Az-5 in toluene upon S1 excitation was too low to be reliably measured. (a) Normalized absorption spectra of Az and Az-1 to Az-5 in toluene; (b) Normalized fluorescence emission profiles of Az and Az-1 to Az-5 in toluene. Table 1Experimental and computed photoexcited properties of Az and the substituted azulenes, Az-1 to Az-5. Calculated values (calc) were obtained at the ADC(2)/def2-TZVP levelaMolecule (nm) (M−1 cm−1) (nm) (nm) (M−1 cm−1) (nm) (nm) (nm)ΔṽStokes (cm−1) ϕ Fl (%)Az[a]600348548340401132137234125302.42Az-1450713498351697535841645216020.31Az-244196849535211 20136941642915370.29Az-3[a]614366616371599334339437523232.40Az-4619306626362590535240538023850.40Az-5619299597361453935240338428100.93a = S1 absorption maximum; = Molar extinction coefficient at S1 maximum; = S1 vertical excitation energy; = S2 absorption maximum; = Molar extinction coefficient at S2 maximum; = S2 vertical excitation energy; = S2 fluorescence maximum; = S2 vertical de-excitation energy; ΔṽStokes = Stokes shift calculated between S2 absorption and emission maxima; ϕFl = Fluorescence quantum yield determined relative to anthracene; [a]in line with previous report.1 Experimental and computed photoexcited properties of Az and the substituted azulenes, Az-1 to Az-5. Calculated values (calc) were obtained at the ADC(2)/def2-TZVP levela = S1 absorption maximum; = Molar extinction coefficient at S1 maximum; = S1 vertical excitation energy; = S2 absorption maximum; = Molar extinction coefficient at S2 maximum; = S2 vertical excitation energy; = S2 fluorescence maximum; = S2 vertical de-excitation energy; ΔṽStokes = Stokes shift calculated between S2 absorption and emission maxima; ϕFl = Fluorescence quantum yield determined relative to anthracene; [a]in line with previous report.1 Upon excitation to the S2 state, Az exhibits a well-defined fluorescence profile with an emission maximum at 372 nm and a fluorescence quantum yield of 2.42%.28 The brominated derivatives Az-3, Az-4, and Az-5 display similar emission features, with maxima at 394 nm, 405 nm, and 403 nm, respectively. In contrast, Az-1 and Az-2 exhibit broader fluorescence profiles with red-shifted emission maxima at 416 nm. This bathochromic shift is consistent with the ADC(2)-calculated de-excitation energies from the S2 potential energy minima to the ground state. Due to the extremely low fluorescence quantum yields of the substituted azulenes, fluorescence lifetimes could not be estimated reliably from time-correlated single photon counting (TCSPC) measurements. The fluorescence quantum yields of the carbonylated derivatives Az-1 and Az-2 were determined to be 0.31% and 0.29%, respectively. For the brominated derivatives, the fluorescence quantum yields were 2.40% (Az-3), 0.40% (Az-4), and 0.93% (Az-5). The notably lower yields observed in Az-1 and Az-2 may stem from their reduced S1/S2 energy gaps, which are expected to promote efficient IC between the two excited singlet states. The excited-state properties of Az and its substituted derivatives were investigated using femtosecond transient absorption (fsTA) spectroscopy. The excited-state processes of Az and substituted azulenes Az-1 to Az-5 were monitored following S2 excitation using a 340 nm pump for Az, while a 370 nm pump was used for the substituted azulenes. Following S2 excitation of Az, an excited-state absorption (ESA) band emerges in the 480–660 nm region, which gradually decays within the experimental time window of ∼3.5 ns (Fig. S2), consistent with previous reports. The A → B → GS model was used for the global fitting of the fsTA spectra to obtain the evolution-associated spectra (EAS) for Az. Upon deconvoluting the fsTA spectra of Az, two components sharing spectral similarity are obtained. The initial component A decays with a time constant of 21.2 ps, attributed to vibrational relaxation, yielding a longer-lived component B with a 1.35 ns decay constant, consistent with the previously reported S2-state lifetime of 1.06 ns in ethylene glycol. This latter decay is consistent with the S2 → S1 IC, in agreement with previous studies,11,28,29,38,42,43 followed by an ultrafast nonradiative decay from S1 to the ground state12,27,43–48 occurring here within sub-ps to ps timescale. This interpretation aligns with earlier reports on azulene and its derivatives, where the non-radiative rate constants for S2 decay follow the log-linear energy gap law, suggesting S2 → S1 IC as the primary deactivation channel.26,28 In our case, the absence of observable S1 features in the fsTA spectra suggests that the populated S1 state upon S2 → S1 deactivation is extremely short-lived, undergoing rapid IC to the ground state (vide infra). The unusually long-lived S2 → S1 IC is attributed to the effective coupling of C–C accepting vibrational modes, combined with the limited participation of C–H stretching vibrations, thereby allowing the former to dominate the energy dissipation pathway.26 Previous works involving direct excitation of S1 in Az also demonstrated that S1 → S0 IC occurs on the femtosecond to picosecond timescale.44,47,49 A recent report describes that S1 state in Az is anti-aromatic and in the S1 minimum-energy geometry, Az undergoes a favourable, nonradiative transition to S0 through a conical intersection to alleviate the Baird-antiaromaticity in the S1 state.11 The S2 state of Az, on the other hand, is aromatic and does not undergo substantial geometric relaxation, resulting in a low rate of S2 → S1 IC. Thus, upon photoexcitation to the S2 state, Az follows a sequential deactivation model wherein S2 → S1 IC occurs on the nanosecond scale, followed by ultrafast decay from S1 to S0.6,50 The substituted azulenes exhibit transient absorption features broadly consistent with those of Az. Az-1 and Az-2 show broad ESA bands spanning 480–660 nm, which decay within the experimental time window of ∼3.5 ns (Fig. S3 & S4). Upon S2 excitation, both compounds initially populate higher excited states, giving rise to ESA centred near 590 nm. Global fitting of the fsTA spectra for the azulene derivatives using the A → B → GS model yielded two EAS with closely similar spectral features. For Az-1, the initially formed component A decays within 14.1 ps to form component B, which then decays with a time constant of 2.68 ns. The absence of any new spectral features during this process suggests that the initial decay corresponds to vibrational relaxation into a relaxed singlet state. A blue shift in the ESA band is also observed, further supporting ultrafast vibrational relaxation preceding singlet decay. Az-2 shows a qualitatively similar behaviour, although the decay times differ. Component A relaxes within 22.4 ps and component B decays over 2.17 ns. The substantial difference in vibrational relaxation times between Az-1 and Az-2 likely stems from differing vibrational mode densities and their coupling with the solvent environment. Similar solvent-dependent vibrational cooling effects have been observed for Az in hexane, acetonitrile, and within molecular containers.51 Interestingly, the introduction of electron-withdrawing aldehyde groups increases the singlet lifetime from 1.35 ns for Az to 2.68 ns for Az-1 and 2.17 ns for Az-2, corresponding to approximately 2-fold and 1.6-fold increases, respectively. The higher brominated analogues Az-3, Az-4, and Az-5 exhibit similar ESA maxima as Az (Fig. 3). Upon S2 excitation, Az-3 shows rapid ESA development at ∼580 nm, which decays completely within the experimental time scale of ∼3.5 ns. Sequential global analysis for Az-3 reveals two components (Fig. S5): a 139.8 ps initial decay followed by a 0.61 ns secondary decay. Az-4 and Az-5 also exhibit similar ESA profiles as Az-3. In Az-4, vibrational relaxation occurs within 18.2 ps, followed by decay of the relaxed singlet over 0.10 ns (Fig. S6). Similar ultrafast excitation dynamics behaviour has also been reported for guaiazulene.52 For Az-5, the two sequential components decay within 53.9 ps and 0.23 ns, respectively (Fig. S7). The modest shift in ESA maxima for Az-3 to Az-5 with respect to Az likely reflects substitution-induced changes in excited-state energy levels. Positional effects of substituents may further influence electronic relaxation, as evident from both steady-state absorption as well as vertical excitation energies.1 (Top) Femtosecond transient absorption spectra of (a) Az-3, (b) Az-4, and (c) Az-5 in toluene upon S2 excitation. (Middle) EAS reconstructed from global analysis of the A → B → GS model, where A is the vibrationally excited singlet state, B is the vibrationally relaxed singlet state, GS is the ground state for the azulenes. (Bottom) The relative population profiles of the excited states. Strikingly, none of the substituted azulenes explored in this investigation display evidence of triplet state formation. Following S2 excitation, a rapid decay component (<150 ps) is observed in all cases from Az-1 to Az-5, attributed to vibrational cooling/redistribution from initially populated, high-density states. This is also consistent with the known behaviour in azulene derivatives.42,45,51,52 The observed relaxation dynamics may involve solvent-induced vibrational relaxation (VR), intramolecular vibrational redistribution (IVR) as well as IC, as supported by calculations showing near-degenerate S2/S3 states in the substituted azulenes (Table S4). In solution, separating vibrational and electronic relaxation is challenging; however, under collision-free gas-phase conditions, techniques such as molecular beam spectroscopy can disentangle IVR and IC processes.47 Notably, no new spectral features emerge following singlet decay in the fsTA spectra, contrary to the expectation that heavy-atom substitution could enhance SOC and facilitate ISC. Despite this, no signatures of triplet formation are observed in the azulenes Az-1 to Az-5. Previous studies on difluorinated azulene have reported a relatively long-lived S2 state (∼6.86 ns in hexane), with no detectable ISC, further underscoring the inherent inefficiency of ISC in azulene derivatives.29 Likewise, even in tri-brominated derivatives, we observe no evidence of ISC. Through triplet energy transfer with appropriate donors, Gorman and co-workers have determined the triplet lifetime of azulene in benzene as 11 ± 1 µs.53 Following fsTA measurements, nanosecond transient absorption (nsTA) measurements in toluene confirm the negligible effect of carbonylation and heavy-atom substitution in the azulenes, where no detectable triplet absorption could be noticed within the ∼8 ns instrument response window (Figs. S8–S10). The theoretical upper bound for the triplet quantum yield (ϕT = kISC/(kISC + kIC) is estimated to be ϕT < 10−7 from S2 state. Because these predicted values fall well below the transient absorption detection limit (ΔA ∼10−4), any potential triplet population is negligible and thus remains unobservable. Furthermore, previous pulse radiolysis and sensitization studies confirm that azulene triplets are optically bright with high extinction coefficients (ε ≈ 4 × 103 M−1 cm−1), ruling out the presence of a spectroscopically inaccessible triplet state.5–7,53 Assuming negligible contributions from ISC (ϕISC ∼0) and other competing relaxation pathways, the fluorescence yields of the azulene derivatives also correlate to the efficiency of IC . The variation in ϕFl (and correspondingly ) across the derivatives is governed by substituent-induced changes in the non-radiative rate constant (). In the multibrominated derivatives Az-3, Az-4 and Az-5, the decrease in ϕFl correlates directly with a dramatic shortening of the S2 lifetime (τS2 = 0.10 ns for Az-4 vs. 1.35 ns for Az). This elevation of up to 1.0 × 1010 s−1 (Table 3) further reinforces non-radiative IC as the dominant deactivation path, leaving radiative emission as a minor decay channel. Further, to unambiguously assign the excited-state deactivation pathways and isolate S1 state kinetics, fsTA measurements were carried out under S1 excitation (λexc = 600 nm for Az, Az-3, Az-4, and Az-5; λexc = 450 nm for Az-1 and Az-2; Figs. S11–S15). Photoexcitation of Az and Az-5 yields transient ESA features that decay rapidly to the ground state within τS1 < 2 ps, confirming unhindered IC through the antiaromaticity-relieving S1/S0 conical intersection.11 Conversely, IC in the brominated analogues Az-3 and Az-4 occurs within the instrument response window of <150 fs and display no discernible transient absorption signals. Carbonylated derivatives Az-1 and Az-2 exhibit substantially extended S1 lifetimes of τS1 ≈ 0.74 ns and 0.86 ns, respectively. Formyl substitution at the 1,3-positions possibly influences the Baird antiaromatic character of S1, dictating the activation barrier required to reach the S1/S0 conical intersection. Crucially, no long-lived triplet transient absorption features are observed for any of the derivatives following S1 decay. In the brominated derivatives, S1 → Tn ISC is thwarted by ultrafast sub-ps to ps IC, whereas in the carbonylated derivatives, despite extending the S1 lifetime to hundreds of ps, ISC remains kinetically inaccessible due to negligible SOC (vide infra). To understand the absence of triplet formation despite heavy-atom substitution, we computationally evaluated the ISC rates. ISC efficiency typically depends on the singlet–triplet energy gap (ΔEST) and is mediated by SOC, which is enhanced by heavy atoms, and can be estimated using the Marcus formalism (eqn S1), and by effective time-dependent (TD) approaches considering the Franck–Condon (FC) approximation and with Herzberg–Teller (HT) corrections (see SI for details). Given that SOC increases with atomic number, bromine substitution in azulene is expected to significantly enhance SOC. Therefore, we examine the extent to which SOC must increase in Az for ISC to effectively compete with S2 → S1 IC and approach the experimentally observed non-radiative decay time of 1.36 ns. Using the ADC(2)-computed singlet–triplet adiabatic gap and reorganization energy, together with the SOC value obtained with the ωB97X-D3 functional, both Marcus and TD models (with and without HT contributions) predict a slow ISC process (Table 2), with a characteristic time on the order of 10−2 s from Marcus theory and even slower within TD approaches. This timescale is several orders of magnitude slower than the S2 → S1 IC lifetime (∼1 ns for Az) calculated at a comparable level of theory.28 To accelerate ISC to the nanosecond regime, as required to compete with S2 → S1 IC, the SOC would need to increase by at least three orders of magnitude (based on Marcus rates, and even more within TD treatments), from the computed value of 0.04 cm−1 for Az to approximately 135 cm−1 (Fig. S16). The computed singlet–triplet SOC values for the three brominated derivatives, Az-3, Az-4, and Az-5, show an increase of up to one order of magnitude, reaching values below 0.4 cm−1. However, these enhancements remain well below the threshold (∼135 cm−1) required for ISC to effectively compete with IC, thus explaining the absence of efficient triplet formation in these compounds. It should be emphasized that SOC values alone do not provide a quantitative estimate of ISC rate constants, which also depend on factors such as singlet-triplet energy gaps and vibronic coupling. Accordingly, SOC magnitudes are used here only as qualitative indicators. In addition, ISC rates are expected to be modulated by substitution-induced changes in the relative state energies. To account for this, we explicitly compute ISC rates for Az-3, Az-4, and Az-5. For completeness, we also evaluate ISC pathways originating from the lowest excited singlet state, despite the fact that S1 → S0 nonradiative decay is known to occur on an ultrafast timescale, typically in the femtosecond regime.44,47,49 Results associated with the largest ISC rate from the relaxed S2 state for each compound are gathered in Table 2 (complete data and ISC rates from the relaxed S1 state can be found in Tables S5–S10). Table 2Computed photoexcited properties of Az and the substituted azulenes, Az-3 to Az-5. Calculated singlet–triplet (S2–T4) adiabatic energy gaps (ΔEadiab in eV), reorganization energies (λ in eV), SOCs (at the S2 geometry in cm−1) values, and ISC rates (kISC in s−1) and lifetimes (τISC in parenthesis in s) within the Marcus and time-dependent approaches (FC and FC + HT) rate expressionsMoleculeΔEadiab λ SOC k ISC (τISC)MarcusTD(FC)TD(FC + HT)Az0.4860.1300.047.71 × 1012.25 × 10−42.52 × 10−1(1.30 × 10−2)(4.45 × 103)(3.97 × 100)Az-30.7600.1570.203.97 × 10−35.70 × 10−77.58 × 10−3(2.52 × 102)(1.75 × 106)(1.32 × 102)Az-40.6450.1690.311.41 × 1024.07 × 10−42.01 × 100(7.09 × 10−3)(2.46 × 103)(4.98 × 10−1)Az-50.5890.1560.192.00 × 1025.60 × 10−47.82 × 10−1(5.00 × 10−3)(1.79 × 103)(1.28 × 100) Computed photoexcited properties of Az and the substituted azulenes, Az-3 to Az-5. Calculated singlet–triplet (S2–T4) adiabatic energy gaps (ΔEadiab in eV), reorganization energies (λ in eV), SOCs (at the S2 geometry in cm−1) values, and ISC rates (kISC in s−1) and lifetimes (τISC in parenthesis in s) within the Marcus and time-dependent approaches (FC and FC + HT) rate expressions In all cases, the highest ISC rate from S1 and S2 is obtained for T2 and T4, respectively. As expected, the introduction of the bromine atoms results in a sizable increase of the SOC compared to the reference Az. This translates into a one to two order of magnitude increase of the ISC rate from the S1, depending on the associated SOC with T2 triplet state, as only small changes are predicted for both the S1–T2 energy gaps and the associated reorganization energies. In contrast, the asymmetric increase of the two latter quantities for the S2 → T4 transition leads to only a weak enhancement of the ISC rate for Az-4 and Az-5, and even a dramatic decrease for Az-3. These results suggest that population of triplet manifold via ISC cannot compete with the fast S2 → S1 IC, which has been previously predicted computationally to occur on a much shorter timescale (∼0.3 ns in Az-3).28 It is worth noting that these three bromo-derivatives exhibit a symmetric substitution pattern, with all three compounds belonging to the C2v point group, which may be detrimental to achieving sizeable singlet-triplet SOCs. To assess the impact of molecular symmetry, we computed SOCs and ISC rates for 1,2-dibromoazulene with Cs symmetry (Table S9). Although some singlet-triplet couplings are no longer symmetry-forbidden in this case, the magnitudes of the SOCs along the main ISC pathways remain comparable to those obtained for Az-3. A qualitative comparison between the experimentally derived and calculated rate constants corroborates the absence of triplet species. Because the experimental fluorescence quantum yields are extremely low, the non-radiative decay rate (knr) is effectively equal to the S2 → S1 internal conversion rate (kIC). Using the S2 lifetimes (τS2 = 0.1–2.6 ns), kIC is determined to be (7.4 × 108) – (1.0 × 1010) s−1 across the azulene derivatives (Table 3). Due to the absence of triplet transient absorption features in both fsTA and nsTA measurements, the triplet yield is negligible. As shown in Tables 2 and 3, the theoretically computed kISC values are also very low, ranging from 10−3 to 102 s−1. Because IC outcompetes ISC by several orders of magnitude, heavy-atom and carbonyl substitutions remain kinetically incompetent at promoting triplet state population in these azulene frameworks. Table 3Comparison of experimental internal conversion rate constants (kIC) and computed intersystem crossing rate constants (kISC) from the S2 state for azulene and its derivatives Az-3 to Az-5Molecule τ IC (ns) k IC (s−1) k ISC (s−1)Ratio (kIC/kISC)Az1.357.4 × 1087.71 × 1019.6 × 106Az-30.611.6 × 1093.97 × 10−34.0 × 1011Az-40.101.0 × 10101.41 × 1027.1 × 107Az-50.234.3 × 1092.00 × 1022.2 × 107 Comparison of experimental internal conversion rate constants (kIC) and computed intersystem crossing rate constants (kISC) from the S2 state for azulene and its derivatives Az-3 to Az-5 A summary of the photophysical dynamics observed across Az and its derivatives upon S2 excitation is presented in Fig. 4. Time-resolved spectroscopic measurements, supported by theoretical calculations, consistently indicate negligible triplet formation via ISC in the carbonylated and/or brominated azulenes explored in this work. These findings challenge the presumed efficacy of bromine and/or carbonyl substituents in enhancing ISC rate constants (kISC) within the azulene scaffold. To overcome this resistance and engineer efficient azulene-based triplet photosensitizers, alternative structural design principles must be pursued. Strategic donor–acceptor (D–A) substitution can reduce Hückel antiaromatic character of cyclobutadiene and pentalene in their S0 states,54,55 and we postulate that this approach also could lead to reduced Baird-antiaromatic character of azulene in its S1 state.56–59 By increasing the activation energy barrier required to reach the S1/S0 conical intersection, the rate of S1 → S0 IC can be slowed, extending the S1 lifetime to facilitate competitive ISC. Incorporating strong push–pull D–A motifs can also introduce low-lying charge-transfer states that alter state ordering and satisfy El-Sayed rules, thereby boosting the SOC and narrowing singlet–triplet gaps.60,61 Exploiting structural asymmetry and target vibrational modes to induce spin–vibronic mixing between close-lying singlet and triplet manifolds can also provide non-adiabatic pathways to accelerate kISC in non-benzenoid chromophores like azulenes beyond the limitations of direct spin–orbit operator matrix elements.62 Schematic diagram representing the major ultrafast photophysical processes in the substituted azulenes (Ex = excitation, F = fluorescence, VR = vibronic relaxation, IC = internal conversion, ISC = intersystem crossing). Conclusions In summary, this work elucidates the excited-state dynamics in Az and a series of halogenated and carbonyl-substituted derivatives through femtosecond transient absorption spectroscopy and theoretical analysis. Despite the introduction of substituents commonly used to enhance SOC and facilitate ISC, none of the derivatives exhibit detectable triplet-state formation, as confirmed by both spectroscopic measurements and computational analysis. Time-resolved measurements reveal IC from the optically bright S2 state to the S1 state, with the excited-state lifetimes confined to the nanosecond regime. The absence of long-lived triplet features across the series suggests that ISC remains kinetically inaccessible in the azulenes explored herein. Complementary computational calculations confirm that although SOC is indeed augmented by halogen and carbonyl substitution, the energy gaps and associated reorganization energies do not favour population transfer to triplet manifolds over IC. Taken together, these findings establish Az and its derivatives as anomalous within the broader landscape of triplet-state engineering, resisting the widely employed design strategies (carbonylation/bromination) that facilitate ISC in aromatic systems. These insights highlight the need for alternative design strategies that go beyond conventional substituent effects, emphasizing the importance of controlling not only SOC, but also the relative energetics and kinetics of competing decay pathways such as VR and IC in the design of efficient triplet photosensitizers. Author contributions All authors contributed to the final manuscript. Conflicts of interest The authors declare no conflicts of interest. Supplementary Material