Ultrafast infrared transient absorption spectroscopy of gas-phase Ni(CO)₄ photodissociation at 261 nm

Metal carbonyls—the family of molecules wherein transition metal atoms bond with CO ligands—serve as model systems for spectroscopic investigation of universal processes, such as ligand binding and catalysis.^1,2 The ultraviolet (UV) spectrum of metal carbonyls features bright metal-to-ligand charge transfer (MLCT) states arising from the promotion of a metal-centered d electron into a $πCO*$ orbital.^3,4 Upon UV excitation to a MLCT state, the molecule typically undergoes many internal conversion events within the first 100 fs, including electronic states of d ← d character—i.e., ligand-field states—which are repulsive along the metal-carbonyl bond, leading to ultrafast loss of CO ligands.^4,5 For example, the gas-phase photodissociation of iron pentacarbonyl [Fe(CO)₅] results in the loss of multiple CO groups^6–9 occurring within the singlet excited state manifold.^10,11

Nickel tetracarbonyl [Ni(CO)₄, NT] is a historically important intermediate in the production of high purity nickel.^12,13 Like other metal carbonyls, the UV spectrum of NT features several bright MLCT states^3,14 (characterized by electrons transitioning from the full d shell² to π* orbitals¹⁵), which are notoriously difficult to theoretically characterize.^14–17 High level benchmarking studies have shown that accurate predictions of metal carbonyl excitation energies require employment of a highly correlated electronic structure method, such as equation of motion coupled cluster theory (EOM-CC).^14,15,17 At the Franck–Condon geometry of NT, the density of electronic states is extremely high with ∼20 singlet states (including all components of degenerate states) within 1.2 eV of the first singlet excited state;¹⁵ symmetry dictates that the only bright states from the singlet electronic ground state of tetrahedral Ni(CO)₄ are triply degenerate states of T₂ symmetry. Many previous publications (see Refs. 3, 14, 15, and 18–22 for examples) have attempted to firmly assign the character of the NT and Ni(CO)_x electronic excited states. Given the challenges with accurate assignment of state ordering in NT and unsaturated nickel carbonyl photoproducts^14,15,20 and the lack of direct excited state characterization in the present infrared spectroscopy studies, we will refer to electronic states of photoproducts with generic S and T labels for singlet and triplet states, respectively. Furthermore, to aid in keeping track of electronic states upon symmetry breaking, the components of degenerate states will be numbered sequentially and referenced simultaneously [e.g., (S₁,S₂) for a doubly degenerate singlet state]. Electronic excited states of NT will be labeled according to Ref. 15.

Similar to other transition metal carbonyls, photoexcitation of the MLCT states of NT leads to multiple photofragments in the gas phase [henceforth collectively referred to as Ni(CO)_x, where x < 4].²³ However, unlike other metal carbonyls, the Ni in NT has a full d shell² and the 4s ← d transition may give rise to the dissociative electronic states along the Ni–CO bonds.²¹ Furthermore—though Jahn–Teller distortion along pseudorotation coordinates has been implicated in the electronic relaxation of several unsaturated metal carbonyl photoproducts²⁴—the Jahn–Teller active coordinates of Ni(CO)_x do not connect excited singlet states to the S₀ ground state,^21,24 thus resulting in long-lived, luminescent Ni(CO)_x photoproducts at room temperature.^25–27 Following photoexcitation at 308 nm, a broad luminescence band has been reported with a pressure-dependent central wavelength of ∼650 nm (Refs. 25–27) and a 400 cm⁻¹ vibronic progression on the short wavelength edge.²⁶ This luminescence band has been assigned to the nickel tricarbonyl photoproduct [Ni(CO)₃, henceforth 3] produced in an electronic excited state with a luminescence quantum yield of ∼10%.²⁵ The >10 µs luminescence lifetime²⁵ of electronically excited 3 has been tentatively attributed to a symmetry-forbidden fluorescence transition.²⁰ Dual-band luminescence is observed following photoexcitation at 248 nm, featuring maxima at 615 and 690 nm with the new, lower energy peak attributed to the production of electronically excited nickel dicarbonyl [Ni(CO)₂, henceforth 2];²⁷ to the best of our knowledge, neither the luminescence quantum yields nor lifetimes of 3 or 2 have been measured following 248 nm photoexcitation.

The first detailed study of NT photoproduct distributions at multiple photoexcitation energies was conducted by Schlenker et al.²⁸ by measuring the free CO rovibrational distributions via laser-induced fluorescence. The rovibrational distributions were fit with a statistical model of energy partitioning to determine the Ni(CO)_x photoproduct identities and branching ratios. From these results, Schlenker et al. concluded that photoexcitation at 248 nm results in the production of both ground and electronically excited 3 and 2 and proposed a branched kinetic model (see schematic in Fig. 1). However, as the authors themselves highlight, the excited electronic states of NT do not correlate with the ground state 3 + CO exit channel,^19,20 and an “indirect” coupling mechanism was instead proposed but not substantiated.²⁸ 

The only previous ultrafast study of gas-phase NT photodissociation examined the fragmentation dynamics following 267 nm photoexcitation with multi-photon 800 nm photoionization and time-of-flight mass spectrometric detection.²¹ By simultaneously fitting the time-resolved ion yields of NT⁺ and all Ni(CO)_x⁺ species, Fuss et al.²¹ determined time constants for multiple photodissociation steps shown schematically in Fig. 1. First, several sub-100 fs transitions link the initially excited, bright MLCT state of ¹T₂ character to the lowest lying optically dark repulsive singlet surface [NT (1¹T₁), the S₁ state¹⁵]. Electronically excited NT then dissociates to 3 (S₁)—i.e., the first singlet excited state of 3—with a 0.6 ps time constant, a significantly longer photodissociation timescale than other metal carbonyls,⁴ which Fuss et al. attributed to a possible excited state barrier along the dissociation coordinate. 3 (S₁) may then further dissociate to produce 2 (S₁,S₂) with a time constant of 55 ps; this longer timescale was found to decrease with decreasing pump wavelength, e.g., 42 ps at 260 nm, supporting an endothermic reaction coordinate leading from 3 (S₁) to 2 (S₁,S₂). From a persistent NiC₃O₃⁺ signal and the absence of any further observable dynamics over the 500 ps temporal window, Fuss et al. speculated that some 3 (S₁) population must undergo intersystem crossing (ISC) over nanoseconds to the nearby triplet manifold, thus producing 3 (T₁) photoproducts that ultimately luminesce. No evidence was found for a ground state dissociation channel, though the authors noted that any 3 (S₀) would be highly vibrationally excited and may undergo secondary dissociation to 2 (S₀) on a shorter—and thus unobservable—timescale than 3 (S₁).

While the studies of Schlenker et al.²⁸ and Fuss et al.²¹ provide considerable insights into the NT photodissociation mechanisms, the absence of Ni(CO)_x spectroscopic signatures leaves uncertainty in the final photoproduct assignments. Furthermore, the ultrafast dissociation times observed by Fuss et al. likely cast doubt on the use of a statistical model to interpret the CO rovibrational distributions. Unlike in iron pentacarbonyl,⁵ the potential role of a concerted dissociation mechanism for NT has not been previously considered. Finally, though ISC in transition metal complexes is common (see Ref. 29 and references therein for examples) and sub-picosecond ISC has been demonstrated in a first row transition metal³⁰ and the Ni period,³¹ the role of triplet states in NT photochemistry has not received significant attention. Thus, in the present work, we will use ultrafast transient infrared (IR) absorption spectroscopy to monitor the structural dynamics following NT photoexcitation in the gas-phase. Similar work by our group on iron pentacarbonyl demonstrated the utility of IR spectroscopy to both reproduce known dissociation time constants and observe previously unreported photoproduct spectral evolution.⁹ Herein, using a combination of theoretical tools and experimental observation of the neutral Ni(CO)_x photofragments, we seek to address the following three outstanding questions regarding NT photodissociation: Can we corroborate the reported dissociation timescales of excited NT and 3 (S₁)? Is there direct evidence for the proposed NT (1¹T₁) → 3 (S₀) → 2 (S₀) pathway²⁸ following ∼250 nm photoexcitation and, if so, with what dissociation times? What, if any, roles do triplet states or concerted dissociation mechanisms play in the overall reaction?

As NT is both pyrophoric and lethal if inhaled, all handling of bulk NT was performed in a dry nitrogen glovebox. A liquid nitrogen cold trap was placed in the vacuum foreline, while the vacuum pump itself was purged with nitrogen during the experiments. The experimental apparatus used in the present work has been previously described in detail,⁹ and the following is a brief overview; where noted, further details may be found in the supplementary material. The ambient vapor from room temperature NT (>99.9% metal purity from Strem) is passed into a 3 mm path length sample cell maintained at a total pressure of 1.5–1.7 Torr. The sample cell is housed in a small vacuum chamber, which is purged with ∼0.5 Torr of nitrogen to minimize build-up of photoproducts on the laser windows (see Ref. 9 for a schematic). These conditions resulted in ∼60% transmission of the infrared probe at the C≡O stretching resonance and a mean interval between collisions of ∼70 ns.

The 3.4 mJ pulse⁻¹, 50 fs, fundamental output from a commercial Ti:sapphire regenerative amplifier is divided into two beams: one to generate the 261 nm UV pulses used to photoexcite the NT sample (the “pump”) and the other to generate the mid-infrared light used to interrogate the UV-initiated structural dynamics (the “probe”). In the latter arm, ∼2 mJ of the fundamental beam pumps a commercial optical parametric amplifier to produce tunable near-infrared signal and idler pulses, which are difference frequency mixed in a 1 mm AgGaS₂ crystal to produce tunable mid-infrared (mid-IR) light centered around 2030 cm⁻¹ with a spectral full-width-at-half-maximum (FWHM) of ∼400 cm⁻¹, allowing for the detection of the C≡O stretching vibration from NT (2061 cm⁻¹),^32,33 all partially unsaturated nickel carbonyls (1920–2050 cm⁻¹), and free CO (2170 cm⁻¹).³⁴ To remove the contribution of rotational dephasing from the transient absorption (TA) signal, a waveplate/polarizer combination was used to hold the probe polarization at magic angle relative to that of the pump. The probe is then focused into the sample cell by a 101.6 mm effective focal length, bare gold-coated off-axis parabolic mirror (OAP); transmitted light is collimated by a matched OAP and focused by a third OAP into a 3-grating spectrometer where it is then dispersed onto a cryogenically cooled 64-element HgCdTe array detector. The pulse energy of the mid-IR probe at the sample position is ∼15 nJ. At the central probe frequency of ∼2030 cm⁻¹, the spectrometer gratings used in the present experiments yield a spectral resolution of ∼6 cm⁻¹/pixel⁻¹ (50 grooves/mm) or ∼2 cm⁻¹/pixel⁻¹ (150 grooves/mm).

The remaining ∼1 mJ of the fundamental beam is optically delayed to match the path length inside the optical parametric amplifier. This delay line includes a hollow-corner gold retroreflector mounted on a motorized translation stage to provide ∼300 ps of computer-controlled time-delay (Δt) between the UV pump and IR probe pulses. The third harmonic of the fundamental is produced after the delay line, resulting in UV pulses centered at 261 nm (4.74 eV, FWHM = 0.06 eV); an example pump spectrum is shown in Fig. S1 of the supplementary material. Following generation, the UV pump is isolated from the remaining fundamental and second harmonic by a series of high-reflecting dichroic mirrors and then focused by a 25 cm focal length lens through a hole in the first OAP, resulting in a collinear pump–probe geometry within the sample cell. The focal point of the probe was located at the approximate center of the sample cell along the propagation direction. To help provide a sufficiently large pump–probe interaction volume, the pump focus was placed a few centimeters after the probe focus. At the probe focal point, the probe had a 1/e² diameter of ∼150 μm, while the pump spot size was 280 μm.

An undoped, 0.5 mm thick Ge (100) wafer attached to the sample cell allows for the in situ determination of pump–probe temporal overlap (Δt = 0). The Ge wafer is also used to measure the instrument response function (IRF), characterized by a temporal FWHM ≈ 160 fs assuming Gaussian temporal profiles of the infrared and UV pulses. Pump–probe time-delays were randomly sampled from a set list to minimize the influence of long-term changes to the signal (namely, photoproduct accumulation on the CaF₂ windows).

The absorption cross section of NT at 261 nm is ∼2 × 10⁻¹⁷ cm².³ Pump power dependence studies (see Fig. S1 in the supplementary material) were performed using a waveplate–polarizer combination in the fundamental beam prior to UV generation. The observed transient signal had an average 1.0 ± 0.1 photon power dependence up to a pump pulse energy of 8 μJ pulse⁻¹ and over all transient features ≤2100 cm⁻¹; a power of 1 μJ pulse⁻¹ was used (∼2 × 10¹⁰ W cm⁻²) for the data reported herein. At high 261 nm pump fluences, a small transient absorption feature was observed at 2200 cm⁻¹ with a two-photon power dependence; this feature is assigned to the NT cation²² and does not contribute to the transient behavior at lower pump powers used herein (see supplementary material for further details). Thus, the reported transient signals are likely from single photon photoexcitation of NT; this assumption will be used for all subsequent analysis and discussion.

All S₀ and T₁ geometries were calculated using coupled cluster theory with single, double, and perturbative triple excitations [CCSD(T)] and the cc-pVDZ basis set^35–38 using a developer’s version of the CFOUR program package.³⁹ Anharmonic frequencies for ground and excited state species were computed with density functional theory (DFT) and time-dependent density functional theory (TD-DFT), respectively, using the ωB97X-D functional,⁴⁰ cc-pVDZ³⁶ basis, and second-order vibrational perturbation theory (VPT2)⁴¹ as implemented in the Q-Chem package.⁴² As symmetry was not enforced during geometry optimizations and frequency calculations, the resulting near-degenerate frequencies have been averaged. Minimum energy crossing points⁴³ (MECPs) between the S₀ and S₁ states of NT, 3, 2, and 1 and between the S₀ and T₁ states of 3, 2, and 1 were computed using Q-Chem using the DFT/TD-DFT methods employed in the anharmonic frequency calculations. The relative energies at each MECP were recalculated using equation of motion excitation energy coupled cluster theory with single and double excitations (EOM-EE-CCSD), see below.

One-dimensional cuts through the potential energy surfaces were calculated along a linear interpolation of internal coordinates (LIIC) designed to smoothly and asymptotically approach the minimum energy geometry of the lowest-lying singlet excited state species as the Ni–CO bond elongates. Exact reaction coordinates are given in Sec. S2.1 of the supplementary material. For NT and each Ni(CO)_x species, 12 excited singlet and 12 excited triplet state energies are calculated at the EOM-EE-CCSD level of theory with the cc-pVDZ basis used for C and O atoms and a Wachters + f basis used for Ni,^44–46 as deemed successful in benchmark studies of electronic energy levels of Ni(CO)₄.^15,16 Mean-field spin–orbit coupling constants (SOCCs)⁴⁷ were computed at minimum energy geometries and MECPs using EOM-EE-CCSD. Order of magnitude calculations of the ISC time constants are computed using a method derived from Ref. 48 and described further in the supplementary material. All EOM-EE-CCSD calculations were performed in Q-Chem. The energy difference between EOM-EE reference and excited states is set to a high accuracy benchmark NT (S₀)–NT (1¹T₂) gap of 4.76 eV,¹⁴ as previously described;^49–53 the resulting correction of −0.275 eV was applied to all NT and Ni(CO)_x electronic excited state energies.

As EOM-EE-CCSD is a single reference method, we performed a limited number of benchmark calculations of our reported energies using multi-configurational self-consistent field theory (MCSCF)⁵⁴ and second-order n-electron valence state perturbation theory (NEVPT2)⁵⁵ as implemented in MOLPRO.^56,57 The active space used in C_2v symmetry is composed of 12 electrons in the 26–32 A′ and 12–16 A″ orbitals, based off of highest occupation in MCSCF calculations. When calculating the NT (S₀) → 3 (S₀) + CO bond dissociation energy (BDE), the EOM-EE-CCSD result (∼0.8 eV, see below) was much closer to the literature value of 1.1 eV (Ref. 58), whereas NEVPT2(12,12) yielded ∼2.1 eV. Furthermore, NEVPT2(12,12) calculations return an NT (S₀)–NT (1¹T₁) energy gap of ∼5.4 eV, significantly higher than previously reported values^14,15 and the results of our present EOM-EE-CCSD calculations. Thus, despite known caveats of the method, EOM-EE-CCSD appears to adequately reproduce NT observables and is an appropriate method for the present work.

To evaluate possible photofragmentation channels in NT, we first consider the energetics of each possible mechanism. For the primary photodissociation of NT, the energy available to the photoproducts is given by

where hν is the photon energy (4.74 eV herein), E₀ is the nascent energy of the parent molecule, BDE is the bond dissociation energy, E_T is the photofragment translational energy, and $EintA$ is the internal energy (i.e., rotational, vibrational, and electronic) of species A. E_avail is the total available energy where $EintNi(CO)x$ becomes E_avail for subsequent dissociation steps. Due to the relatively large number of low frequency vibrational modes in NT (see Table S1 in the supplementary material) and the room temperature gas sample (see Sec. II), we estimate a nascent thermal energy of E₀ ≈ 0.2 eV that is available for photochemistry; see Sec. S2.2 of the supplementary material for further discussion. Although previous reports^21,28 have concluded that 2 is the smallest Ni(CO)_x species produced at this pump wavelength, the literature^28,58 and calculated BDEs given in Table I reveal that nickel monocarbonyl (NiCO, henceforth 1) is the smallest photofragment possibly accessible following excitation at 261 nm. However, from the free CO temperatures following excitation at 248 nm,²⁸ we estimate an average E_T + $EintCO$ ≈ 0.4 eV reduction in E_avail for each CO ligand lost. Thus, the production of singlet excited 1 is only possible in the limiting case of near-zero photofragment translational energy and internally cold free CO co-fragments (i.e., E_T = $EintCO$ = 0). Finally, we note that our calculated BDEs will be used for all further discussion of energetics as thermochemical measurements,^59,60 and vacuum UV photodissociation⁶¹ suggest that the sum of the experimental^28,58 ground state BDEs underestimates the complete NT decarbonylation energy extracted from thermochemistry by ∼0.6 eV.

Similar to our previous work on iron pentacarbonyl,⁹ we consider both sequential and concerted CO loss pathways from excited NT. Example potential energy cuts (PECs) are shown in Fig. 2; additional PECs are shown in Figs. S7–S11 in the supplementary material. Immediately apparent from Fig. 2 are the large number of electronic states accessible throughout any dissociation channel, particularly when considering both excited singlet and triplet states. Although further discussion of possible pathways will focus on particular electronic excited states, the high density of states suggests that the excited state dynamics will occur on multiple states. With photoexcitation initially populating the NT (1¹T₂) state, Fuss et al.²¹ propose a rapid IC process down to NT (1¹T₁) followed by dissociation over a small barrier. However, the calculated PECs reveal a completely repulsive NT (1¹T₁) → 3 (S₁) + CO surface with no evidence for the proposed excited state barrier; searches for a NT (S₁) local minimum energy geometry were unsuccessful. Similarly, the energy required to follow the 3 (S₁) → 2 (S₁,S₂) + CO PEC computed here is ∼0.7 eV; although less than the 0.95 eV predicted by Fuss et al., this channel is still endothermic and supports the overall conclusions drawn from the reported wavelength-dependent picosecond time constant. We also consider the concerted loss of two CO ligands shown in Fig. 2(c). As with the loss of a single CO group from NT, the channel representing concerted loss of two CO ligands is also an overall “downhill” dissociation, though a shallow ∼0.05 eV well is evident on the S₁ state. While there are numerous challenges in identifying and predicting the importance of reaction pathways in systems with a high density of electronic states, we conclude that both sequential and concerted CO loss mechanisms are important for short times due to the exothermic nature of both reactions. Theoretical studies of the short timescale dynamics (currently underway in our group) are thus needed for a more precise picture of the competition between these mechanisms and their timescales. The 2 → 1 + CO reaction coordinate [see Fig. 2(d)] features a ∼1.4 eV barrier for dissociation in the S₁ state, providing further evidence that 1 may only be produced in the S₀ or lowest lying triplet excited states.

As expected from previous theory^19,20 and symmetry arguments,²¹ at no point along these PECs, do we observe an intersection between the ground and a singlet excited electronic state, suggesting that the primary NT photoproducts are produced in an electronic excited state. Searches for S₁/S₀ MECPs in NT, 3, 2, and 1 found MECPs with S₀ energies of 7.2, 6.5, 5.3, and 1.6 eV, respectively, relative to each S₀ equilibrium geometry (see Fig. S13 and Table I). Thus, the S₁/S₀ MECP for each species is not energetically accessible at the current pump wavelength and will not contribute to the present dynamics. Although we cannot access the NT (S₁/S₀) MECP, the calculated geometry features elongation of one Ni–C bond as predicted Fuss et al.²¹ 

While the above discussions have focused on the singlet excited states, all PECs show a dense manifold of triplet states for NT and all Ni(CO)_x. The large number of regions with a small singlet–triplet energy gap, along with the prevalence of ISC in transition metal complexes,²⁹ strongly suggests that triplet states play an important role in the dissociation dynamics of NT. Searches for T₁/S₀ MECPs in 3, 2, and 1 found MECPs with S₀ energies of 3.3, 2.2, and 0.5 eV, respectively, relative to each S₀ equilibrium geometry (see Fig. S14); all three T₁/S₀ MECPs are energetically accessible and may contribute to the observed dynamics. Additionally, Tables S3–S5 give the calculated SOCC and estimated ISC timescales between electronic excited states of 3, 2, and 1, revealing multiple instances of sub-nanosecond ISC. We note that these ISC time constant calculations were performed for a limited number of geometries and, as the SOCC are sensitive to geometry, are only used as a guide.

The anharmonic CO stretching frequencies of NT and Ni(CO)_x are given in Table II with the full list of harmonic and anharmonic frequencies for NT given in Table S1 in the supplementary material. Descriptions of the optimized geometry for a given photoproduct and electronic state are also given; images of each geometry may be found in Fig. 2 and S12. We note that the linear and bent geometries of 2 are nearly isoenergetic (within ∼0.01 eV for S₀ and <0.2 eV for higher lying singlet states) with Renner–Teller coupling creating significant challenges for determining the energetic ordering of the two conformers. A scaling factor for the DFT and TD-DFT anharmonic frequencies was computed as 0.96 via comparisons to the experimental gas-phase C≡O stretch frequencies of NT (S₀) and 1 (S₀); these scaled anharmonic frequencies will be used throughout this paper. It is important to note that we are applying the DFT-derived scaling factor to the TD-DFT anharmonic frequencies, as there are no previous experimental measurements of vibrational frequencies in excited electronic states for derivation of an analogous TD-DFT scaling factor. Thus, instead of using scaled TD-DFT anharmonic frequencies to firmly assign experimental peaks, we will use these frequencies to guide the discussion of trends.

As was predicted in our previous work on iron pentacarbonyl,⁹ for most Ni(CO)_x, the C≡O antibonding character of the electronic excited states results in a decrease in the C≡O stretching frequencies relative to the ground electronic state. Furthermore, excepting 1, we see a decrease in frequency with decreasing saturation of the Ni center as has been observed for NT⁶² and other transition metal carbonyls (see Ref. 64 for examples). We also find that for most Ni(CO)_x, the C≡O stretching modes in the T₁ state appear at slightly lower frequencies than in the S₁ state. NT, 3, and 2 each feature two or more C≡O stretching modes, typically separated by up to ∼100 cm⁻¹, where the higher frequency mode is a symmetric C≡O stretching mode showing little to no IR activity. For photoproducts with lower symmetry such as 3 (S₁), a relatively low intensity TA feature may be experimentally observable at higher frequencies due to the weakly IR-active symmetric stretching mode. The computed C≡O stretching frequencies for 3 (S₁), 3 (T₁), 3 (T₂), and 2 (T₁,T₂) are remarkably similar and are thus difficult to distinguish experimentally. Finally, we note that accurate prediction of vibrational frequencies in 2 is further complicated by Renner–Teller coupling; while only one IR peak observed in the Ar matrix was attributed to 2, DeKock carefully notes that 2 may be bent with a weak (and unobservable) symmetric C≡O stretch mode.⁶² The large discrepancy between our calculated and the experimental C≡O stretching frequency of 2 (S₀) may also be due to the influence of the Ar matrix on the measured frequency.

Transient infrared spectra of NT following photoexcitation at 261 nm are shown in Fig. 3. The time- and frequency-dependent oscillations appearing at Δt < 0 are due to perturbed free-induction decay and have been discussed elsewhere.^65,66 To avoid interpreting these oscillations and other coherent artifacts arising from temporal overlap of the pump and probe pulses,⁶⁷ we will limit our analysis to Δt ≥ 0.2 ps. At the earliest time-delays, loss of the parent NT signal at 2061 cm⁻¹ appears as a negative absorption [ground state bleach (GSB)] alongside a broad, unstructured positive absorption feature [transient absorption (TA)] centered at 1950 cm⁻¹. Transient spectra acquired with a higher resolution grating (∼2 cm⁻¹ pixel⁻¹) do not show any structure within each broad TA peak. A small, positive TA peak that is nearly overlapped with the GSB is also evident as a small shoulder around 2040 cm⁻¹. Over the first picosecond, the center frequency of the broad TA feature shifts to slightly lower frequencies. The initial TA feature then decays over tens of picoseconds into two distinct peaks centered around 1920 and 1950 cm⁻¹, while a new peak at 2000 cm⁻¹ grows in over a similar timescale. The GSB appears to simultaneously lose intensity; however, as NT photodissociation is irreversible within the current temporal window, this apparent GSB recovery (see Fig. S3) must be due to either the radiative decay of a long-lived NT excited state, which our calculations do not find evidence for, or the growth of a “buried” TA feature centered around 2050–2060 cm⁻¹. The timescale of the apparent GSB recovery is difficult to quantify; preliminary multiexponential fits to a lane integrated transient suggest a ∼100 ps decay time constant, though acceptable fits are also achievable with $o$(10³ ps) time constants.

As highlighted in Fig. 3(b), there is a marked difference in the temporal behavior of the 1920 and 1950 cm⁻¹ features. Initial multiexponential decay fits to these single-frequency transients reveal two sets of time constants describing the temporal evolution of these spectral features. First, the sub-picosecond shift of the broad TA peak to lower frequencies is evident from a simultaneous high frequency decay and low frequency growth in signal intensity; this shift is also evident as a small increase in GSB intensity over the same temporal window (see Fig. S3 in the supplementary material) likely due to reduced overlap between the broad TA peak and the GSB. Next, the peak at 1920 cm⁻¹ follows a ∼10 ps decay to a final, constant signal intensity reproduced with a time constant of $o$(10⁴ ps). Conversely, the behavior of the 1950 cm⁻¹ feature is well reproduced by fits with decay time constants of either ∼50 and $o$(10³ ps) or ∼80 ps and $o$(10⁴ ps); see Fig. S5 for example fits. Although the former fit reproduces the dissociation time constant of 3 (S₁) reported by Fuss et al.²¹ and would imply further evolution on a few nanosecond timescale, we are unable to quantify either of the longer time constants with the present 300 ps temporal window.

Along with the calculated PECs, these preliminary observations suggest that a branched kinetic model—wherein two primary photoproducts undergo independent sequential decay dynamics—appears to be the simplest model for adequately describing the observed global transient spectral evolution. As a common $o$(10⁴ ps) time constant for all frequencies minimizes the number of fit parameters in our global kinetic model, we will continue the analysis under this assumption. Thus, the kinetic model used in the present work [shown schematically in Fig. 4(a)] consists of five time constants with the best-fit species associated difference spectra (SADS) and associated time constants displayed in Fig. 4(b); fit residuals are shown in Fig. S6 in the supplementary material. Only in the case of a correct kinetic model will the SADS correspond to species-specific IR spectra. Based on the results of Fuss et al.,²¹ which reveal multiple sub-100 fs processes immediately following photoexcitation, we have fixed the value of τ₁ to be less than the present IRF of 0.16 ps. Similarly, the final time constant τ₅ was fixed at 10⁴ ps to reproduce the long-time behavior from the preliminary fits at 1950 and 1920 cm⁻¹ discussed above. Thus, the final kinetic model comprises three free fit parameters yielding global fit time constants of 0.6, 15, and 90 ps. The 0.6 and 90 ps time constants are in reasonable agreement with the results of Fuss et al.,²¹ whereas the 15 ps time constant has not been previously reported.

As shown in Fig. 4(b), the SADS associated with τ₁ highlights the prompt appearance of the GSB, a broad TA feature peaked around 1950 cm⁻¹, and the small TA feature ∼2040 cm⁻¹. Next, the τ₂ and τ₃ SADS show the relatively unstructured feature centered around 1950 cm⁻¹, decaying with time constants of 0.6 and 90 ps. Similarly, the SADS associated with τ₄ clearly separates the 1920 and 2000 cm⁻¹ peaks from the other TA features, suggesting that these two peaks arise from a different photodissociation channel that correlates the growth of the peak at 2000 cm⁻¹ to the decay at 1920 cm⁻¹. We note that the overlapping τ₂ and τ₃ SADS intensity at both peaks implies that the broad feature at 1950 cm⁻¹ may influence the observed behavior at 1920 and 2000 cm⁻¹. Finally, only the τ₁, τ₂, and τ₅ SADS show substantial contributions to the buried TA peak at 2050–2060 cm⁻¹. With minor contributions to the buried TA peak also from both τ₃ and τ₄, we are unable to firmly correlate the apparent growth of the highest frequency TA feature to the evolution of another spectral feature. Thus, this kinetic model provides an overall satisfactory fit to these data with relatively few free fit parameters. Simpler kinetic models, such as a sequential model using only the 0.6 and ∼50 ps time constants reported by Fuss et al.,²¹ do not adequately describe the temporal behavior of the peak at 1920 cm⁻¹.

As with our previous work on iron pentacarbonyl,⁹ in the present study, we are unable to directly observe any free CO following photoexcitation of NT. The absence of the free CO signal is largely due to the significant spread of rotational and vibrational states²⁸ and low IR cross section⁶⁸ resulting in a broad, low intensity free CO transient spectrum that we cannot observe at present. Time-resolved measurements directly examining the production of CO, such as photofragment velocity map imaging or x-ray absorption spectroscopy, would provide key insights into when each CO ligand dissociates.

Although free CO internal energy distributions²⁸ and dissociative ionization products²¹ have been measured, no previous study has explicitly measured the absolute CO product quantum yield or neutral Ni(CO)_x photoproduct distribution. Indeed, the conclusions from Schlenker et al.²⁸ that 2 is produced in this pump photon energy range was based on a model of statistical energy partitioning—a questionable assumption given the ultrafast timescales of NT photodissociation. Given the PECs in Fig. 2, the observation of luminescence attributed to 3,²⁷ and the persistent NiC₃O₃⁺ signal observed by Fuss et al.,²¹ some NT must dissociate to an electronically excited 3, which persists beyond the present 300 ps temporal window. Although the previously reported dual-band luminescence could arise from multiple electronic excited states of a single photoproduct—indeed our predicted energy gap between 3 (S₁) and 3 (T₁) of ∼0.2 eV is in excellent agreement with the observed energy spacing between the luminescence bands—the appearance of a second luminescence band with increasing pump photon energy is most easily explained by the production of 2 in an electronic excited state.²⁷ Without a firm count of the free CO released in our measurements, we thus interpret the observed spectral evolution based on the reported production of 3 and 2,^21,27,28 energetic constraints, predicted PECs, and qualitative comparisons to calculated vibrational frequencies.

Based on the above considerations, we may begin to assign transient spectral features to particular Ni(CO)_x. The calculated frequencies in Table II immediately suggest that the peak at 1950 cm⁻¹ is due to the production of 3 and/or 2 with multiple electronic and vibrational excited states of each likely contributing to the broad transient spectrum at early time-delays. Indeed, considering both Fig. 2 and Table II, the present results could equally support the sequential (loss of single CO ligand) and concerted (loss of two CO ligands) dissociation pathways. Next, our calculations predict that the feature at ∼2040 cm⁻¹ is the result of a weakly IR-active symmetric C≡O stretching mode of an electronically excited 3 or 2 nearly overlapping with the GSB around 2050–2060 cm⁻¹, where the latter would only appear for bent configurations of 2. The absence of a local minimum energy geometry in NT (1¹T₁) casts significant doubt on a fluorescence mechanism, and re-population of ground state NT is thus unlikely to contribute to the transient feature at ∼2040 cm⁻¹. Therefore, despite the calculated IR intensity ratios of 3 suggesting a symmetric C≡O stretching signal of less than 0.1 ΔmOD, we predict that the shoulder at ∼2040 cm⁻¹ is due to a “buried” TA peak from an electronically excited 3. The calculated C≡O stretching frequencies of 2 (S₀), 2 (S₃,S₄), 3 (S₁), and 3 (T₂)—as well as the experimental value of 1 (S₀)—all appear around 2000 cm⁻¹, and a more detailed spectral assignment cannot be made without a detailed consideration of possible mechanisms and the associated temporal evolution; see Sec. IV B. Finally, none of the calculated anharmonic frequencies directly reproduce the experimental peak at 1920 cm⁻¹. Indeed, the calculated asymmetric C≡O stretching frequency of 3 (T₁) or 2 in a triplet excited state (see Tables II and S2) are closest to the experimental value, suggesting that the 1920 cm⁻¹ TA feature could arise from a triplet excited 3 or 2 with possible vibrational excitation of the C≡O stretching modes. Unfortunately, the large number of accessible electronic states, similarity of some Ni(CO)_x vibrational frequencies, and the uncertainty in the scaled anharmonic frequency values prevent a unique peak assignment, and we next connect the observed temporal evolution to computationally predicted photodynamic pathways to determine the best assignments of each spectral feature.

From the experimental and computational results and our initial discussion of transient spectral assignments, it is clear that the dissociation of NT presents a significantly more complex picture than has been previously considered. While the large number of overlapping photoproduct frequencies prevents a firm assignment of the observed spectral evolution to unique dynamical processes, we can still address the questions posed at the end of Sec. I. We present these insights with the caveat that while our kinetic model captures all the salient aspects of the experimental data with the smallest possible number of floating parameters, the model is likely incomplete, and the time constants we report may describe the aggregate behavior of multiple, spectrally overlapping processes. Figure 5 is first presented here as a reference to guide the ensuing discussion of the following three potential channels: (I) an excited singlet state pathway (dotted blue line), (II) a ground state pathway (dashed green line), and (III) a concerted dissociation involving the triplet manifold (orange line). As will be discussed below, parts of the channel III pathway shown in Fig. 5 are speculative with no experimental evidence to disprove a particular assignment.

We first consider the singlet excited state reaction pathway proposed by Schlenker et al. and Fuss et al. (channel I, shown as blue arrows in Fig. 5) and find that our present results could support the proposed excited state mechanism. First, though we are unable to resolve the multiple, sub-100 fs electronic relaxations proposed by Fuss et al., the τ₁ SADS reveals a broad TA feature with intensity extending beyond our present spectral window. Such a broad spectral response qualitatively supports one or more short-lived intermediates (arising from the significant number of excited states¹⁵) that decay on timescales shorter than τ₁. Next, the 0.6 ps time constant—appearing as a small shift in the TA center to a lower frequency—is consistent with the dissociation of NT (1¹T₁) into the 3 (S₁) photoproduct; the largely overlapping vibrational spectra are rationalized by the increased CO antibonding character of NT (1¹T₂), NT (1¹T₁), and 3 (S₁). Thus, the prompt appearance of the shoulder at ∼2040 cm⁻¹, the broad TA signal around 1950 cm⁻¹, and the dissociative PEC all strongly support the conclusion that 3 (S₁) is one of the earliest photoproducts produced. While our PECs along the NT → 3 + CO coordinate do not reveal any evidence for a barrier on the S₁ state, there is evidence for barriers along the NT → 2 + 2CO coordinate, which show shallow ∼0.05 to 0.1 eV wells along the S₁ and S₃ excited states; the 0.6 ps time constant may arise from excited NT sampling these regions of the potential energy surface during the initial IC and prior to dissociation. The high density of NT excited states¹⁵ may also delay dissociation, as numerous IC events may transiently populate nearby excited states that are not purely dissociative in nature. Ab initio molecular dynamics simulations of the initial NT dissociation dynamics—currently under way in our group—will provide significant insights into the pathways and timescales of excited NT evolution within this sub-picosecond time frame.

The penultimate step of the proposed²¹ excited singlet dissociation pathway, the dissociation of 3 (S₁) to produce 2 (S₁,S₂) with a time constant of ∼50 ps, is qualitatively supported by our present results. The current kinetic model reveals that the 1950 cm⁻¹ TA peak we have assigned to 3 (S₁) and 2 (S₁,S₂) evolves with a 90 ps time constant. Although the present time constant is longer than the 42 ps time constant reported by Fuss et al.²¹ at 260 nm, we again note that the behavior of the 1950 cm⁻¹ feature is also well described by exponential decays of ∼50 and $o$(10³ ps) (see Fig. S5), suggesting that the discrepancy between reported decay times may be due to multiple overlapping dynamics that we are unable to resolve at present. PECs along the 3 (S₁) → 2 (S₁,S₂) reaction pathway support the endothermic reaction proposed by Fuss et al. to explain their observed pump wavelength dependence of this time constant. Overlapping photoproduct spectra are qualitatively supported by the anharmonic frequency calculations, though the decay in TA signal intensity is in contrast with the calculated IR intensities. While a corresponding evolution of the ∼2040 cm⁻¹ shoulder assigned to 3 (S₁) is visible (see Fig. S3), the weak feature, overall small changes in signal intensity, and ambiguous global fitting results make a detailed assessment untenable at present.

Finally, to explain the long luminescence lifetime,²⁵ Fuss et al.²¹ speculated that any remaining 3 (S₁) photoproduct will undergo ISC to the triplet manifold over many nanoseconds (i.e., beyond their 500 ps temporal window). While a long 3 (S₁) → 3 (T₁) ISC timescale is supported by a SOCC ≤ 1 cm⁻¹, the multitude of crossings between singlet and triplet excited states with large SOCC implies that ISC to the triplet manifold in 3 may be faster than proposed by Fuss et al.²¹ and that the long luminescence lifetime²⁵ is due to phosphorescence. Furthermore, though the ISC time constants in Table S4 suggest that ISC from 2 (S₁,S₂)→2 (T₁,T₂) would occur with a $o$(10³ ps) time constant—which could be observable within our present temporal window—we do not see any evidence for a rise in intensity at lower frequencies as predicted for triplet-excited 2 (see Table II). Ultimately, due to a limited temporal window of ∼300 ps, we refrain from attempts to assign any possible $o$(10³ ps) dynamics and, thus, omit these final processes from Fig. 5. Thus, while the excited state model from Fuss et al. is generally supported by our data—and we can find no proof against the model—channel I does not explain all of our observations and we will need to consider other parallel pathways.

Given the evidence for the singlet excited state pathway discussed above, we next consider the possible ground state dissociation pathway (channel II in Fig. 5) proposed by Schlenker et al.²⁸ Based on the relative ordering of vibrational frequencies from matrix isolation studies,⁶² the two remaining peaks at 1920 and 2000 cm⁻¹ could correspond to 2 (S₀) and 3 (S₀), respectively, with the large available energies potentially leading to C≡O vibrational excitation and accounting for the shift to lower frequencies (see Tables I and II). However, this peak assignment is inconsistent with the observed temporal evolution as (i) the growth of 3 (S₀) should not correlate with the decay of 2 (S₀) and (ii) any 3 (S₀) produced by this mechanism should appear promptly. An alternate peak assignment—assigning the 1920 cm⁻¹ feature as a highly vibrationally excited 3 (S₀) and the feature at 2000 cm⁻¹ to 2 (S₀)—would require nearly all of the 3 (S₀) available energy to be partitioned into the C≡O stretching mode and would result in a significantly higher 2 (S₀) frequency than was observed experimentally.⁶² Finally, despite the careful consideration and speculation by Fuss et al.,²¹ we find no evidence for any energetically accessible MECPs that connect the S₁ and S₀ surfaces, suggesting that the only ultrafast mechanism for producing 3 (S₀) at the present pump wavelength is via a 3 (T₁) intermediate. Although the numerous crossing points between the singlet and triplet manifolds shown in Fig. 2(a) could—assuming rapid IC within the triplet states—lead to the prompt production of 3 (T₁), further dissociation to 2 (T₁,T₂) is much more energetically favorable than accessing the 3 (T₁/S₀) MECP [see Figs. 2(b) and S14]. Thus, we do not find any evidence to support channel II [the previously proposed²⁸ NT → 3 (S₀) → 2 (S₀) dissociation pathway] and must consider alternate mechanisms to explain the observed transient spectral behavior.

As the two transient features at 1920 and 2000 cm⁻¹ have correlated spectral evolution, we will attempt to assign these peaks using a single dynamical process. Although multiple electronic states of 2 and 3 may be assigned to the peak at 2000 cm⁻¹, many of these assignments may be discredited due to the absence of evolution within other spectral regions, energetic constraints, or the inability to simultaneously explain the peak at 1920 cm⁻¹; a detailed discussion of each possibility may be found in Sec. S3 of the supplementary material. Thus, we instead consider potential assignments to the 1920 cm⁻¹ feature first. The anharmonic C≡O stretching frequencies of either 3 (T₁) or 2 in a triplet excited state (multiple triplet states of 2 have similar frequencies, see Tables II and S2) are the calculated frequencies closest to the observed 1920 cm⁻¹ peak, and we first consider the assignment of the 1920 cm⁻¹ peak to 3 (T₁). From the previous paragraph, the ultrafast production of 3 (T₁) from NT (1¹T₁) is plausible though the pathway remains speculative; as such, this pathway is shown in light orange in Fig. 5 with the “fast” label indicating an unknown, but prompt, timescale. Population in 3 (T₁) may access the T₁/S₀ MECP or undergo further dissociation to either produce 2 (T₁,T₂) or 1 into the three closely spaced, lowest lying triplet states [collectively referred to as 1 (T_1–3)]; see Figs. S14, 2(b), and S9, respectively. Similar to channel II,²⁸ the assignment of the 2000 cm⁻¹ feature to either 2 (S₀) or 2 (T₁,T₂) is inconsistent with the experimental frequency and the trend of decreasing frequency upon electronic excitation. With an excited state barrier of 1.6–1.9 eV, the concerted 3 (T₁) → 1 (T_1–3) + 2CO channel shown in Fig. S9 of the supplementary material would require nearly all of the available energy, casting doubt on the importance of this pathway as a source of the 15 ps spectral evolution. Thus, though the dissociation of 3 (T₁) seems unlikely to explain the transient behavior of both spectral features, the persistent signal at 1920 cm⁻¹ is tentatively assigned as 3 (T₁) that does not undergo further reaction and instead luminesces on longer timescales (not shown in Fig. 5).

Based on the overlapping singlet and triplet states in Fig. 2(c), the large SOCC for multiple states of 2 in Table S4, and again assuming rapid IC within the triplet manifold, the prompt appearance of the peak at 1920 cm⁻¹ is plausibly explained by the concerted production of 2 (T₁,T₂) from excited NT. In the absence of a firm mechanism or timescale, this triplet/concerted pathway remains speculative and is shown in light orange in Fig. 5; further ultrafast electronic spectroscopy or detailed dynamics simulations, including spin–orbit coupling, would be of significant interest to explore this potential mechanism. If a majority of E_avail has remained with 2 (T₁,T₂), as is qualitatively suggested by the low peak frequency of 1920 cm⁻¹, two further reactions remain accessible: ISC to 2 (S₀) via the 2 (T₁/S₀) MECP or further dissociation to 1 (T_1–3) with the latter process, shown as channel III in Fig. 5, being the lower energy channel [see Figs. 2(d) and S10]. Although either mechanism could reasonably lead to the production of 1 (S₀), an attractive assignment of the observed feature at 2000 cm⁻¹,⁶³ the absence of an intermediate spectral signature suggests that escape from the 2 (T₁,T₂) minimum is the rate-limiting step (i.e., either ISC or dissociation must occur with a 15 ps time constant). However, rapid ISC to 2 (S₀) is not supported by the calculated ISC time constants (shown in Table S4), thus casting doubt on a potential sub-nanosecond crossing to 2 (S₀), though we again note that the ISC time constants were only calculated for select geometries. Regarding the possible dissociation of 2 (T₁,T₂) to 1 (T_1–3), Figs. 2(d) and S10 both show a ∼1 eV barrier to dissociation on the lowest triplet states. Although a 15 ps time constant to surmount a ∼1 eV barrier is not intuitive, many of the MECPs involve elongation of the Ni–C bonds (see Figs. S13 and S14), and the production of 2 (T₁,T₂) may include significant vibrational excitation in the reactive coordinate. Additionally, the excited state energies of 1 are particularly sensitive to molecular geometry—compare Figs. 2(d), S10, and S11—and a barrier along the true minimum energy path from 2 (T₁,T₂) to 1 (T_1–3) may be smaller than 1 eV. The absence of any spectral evidence for an intermediate state suggests that ISC from 1 (T_1–3) to 1 (S₀) occurs with a time constant shorter than 15 ps, as qualitatively supported by the relatively large SOCC in 1 (see Table S5).

Thus, the production of 1 (S₀) via channel III is energetically accessible at 261 nm, supported by the PECs and estimated ISC time constants, and satisfactorily assigns and explains the correlated evolution at 1920 and 2000 cm⁻¹. Finally, we note that neither Schlenker et al.²⁸ nor Fuss et al.²¹ reported the production of 1 following photoexcitation around 5 eV. The Ni(CO)_x photoproduct branching ratios of Schlenker et al. are based on a statistical model of energy partitioning rather than direct evidence for each Ni(CO)_x. In the latter study, Fuss et al. did not observe any further increases in the NiCO⁺ transient signal as would be expected from the production of 1 (S₀). Three non-exclusive possibilities may explain this discrepancy: (i) the previous multiphoton photoionization scheme could not ionize 1 (S₀), (ii) the production of this species is only possible given the short pump wavelength and room temperature sample used herein, or (iii) 1 (S₀) is a relatively minor photoproduct. Further work exploring the detailed pump wavelength dependence of specific Ni(CO)_x transient features would help elucidate the excited state branching mechanisms.

While the above discussion focuses on a few key aspects of the NT photodynamics, the complexity of the PECs and strong dependence of excited state energies and SOCC on molecular geometry all suggest that many other photodynamical pathways probably exist. Within a relatively narrow photoexcitation window, several channels are likely accessed with excited state population able to follow multiple avenues to the final photoproducts; the overlapping Ni(CO)_x spectra unfortunately make other possibilities difficult to either support or disprove. Regardless of the remaining questions raised herein, these possibilities both emphasize the complexity of NT dissociation and highlight the number of possible reaction mechanisms that have not yet been considered.

Herein, we have addressed three outstanding questions surrounding the photodissociation dynamics of NT. Our present results support a previously proposed²¹ excited state mechanism whereby photoexcited NT undergoes sequential dissociation to produce both 3 (S₁) and 2 (S₁). Although our present results do not support the proposed ground state dissociation mechanism,²⁸ the signal observed at 2000 cm⁻¹ is best explained as the production of 1 (S₀), a previously unreported product from NT photodissociation around 5 eV. Furthermore, the production of 1 (S₀) must invoke triplet electronic excited states and concerted dissociation—two mechanisms that have not received much prior consideration to explain NT photochemistry. Despite the challenging and congested transient IR spectra, we have nonetheless drawn several key conclusions and highlighted specific future research avenues to test our hypotheses and answer the remaining questions. Most importantly, these results demonstrate that the ultrafast photodynamics of NT is far more complex than the relatively simple frameworks previously considered.^21,28 Current efforts in our group are focused on interpreting similar NT photodissociation results following excitation at 197 nm as well as ab initio molecular dynamics calculations; both efforts promise to further aid our understanding of these metal carbonyl systems. Furthermore, as with our previous study on iron pentacarbonyl,⁹ the present work continues to demonstrate that transient IR spectroscopy is an essential complement to more commonly used destructive techniques (such as multiphoton photoionization) for elucidating the full photodissociation mechanisms of gas-phase species, warranting further theoretical and modeling work to link the infrared experimental observables and molecular photodissociation dynamics.

See the supplementary material for additional experimental results, additional computational results, further discussion of the 1920 and 2000 cm⁻¹ features, and details of the Zenodo data archive.

The authors would like to thank David Osborn for helpful discussions and Tim Zwier for manuscript feedback. The work presented here is supported by the Division of Chemical Sciences, Geosciences and Biosciences, Office of Basic Energy Sciences (BES), U.S. Department of Energy (USDOE). Sandia National Laboratories is a multi-mission laboratory managed and operated by National Technology and Engineering Solutions of Sandia, LLC., a wholly owned subsidiary of Honeywell International, Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under Contract No. DE-NA-0003525. This paper describes objective technical results and analysis. Any subjective views or opinions that might be expressed in the paper do not necessarily represent the views of the U.S. Department of Energy or the United States Government.

The authors have no conflicts to disclose.

The data that support the findings of this study are openly available in Zenodo data archive at https://doi.org/10.5281/zenodo.5750749.

BDE =

bond dissociation energy

EOM-CC =

equation of motion coupled cluster theory

GSB =

ground state bleach

IC =

internal conversion

IR =

infrared

ISC =

intersystem crossing

MECP =

minimum energy crossing point

MLCT =

metal-to-ligand charge transfer

NT =

nickel tetracarbonyl

PEC =

potential energy cut

SADS =

species associated difference spectrum

SOCC =

spin–orbit coupling constant

TA =

transient absorption

TD-DFT =

time-dependent density functional theory

UV =

ultraviolet